diff --git "a/raw/parallel_extract/chart.jsonl" "b/raw/parallel_extract/chart.jsonl" new file mode 100644--- /dev/null +++ "b/raw/parallel_extract/chart.jsonl" @@ -0,0 +1,568 @@ +{"completed_at":"2026-11-04T19:45:40.596071","latency_in_ms":1449,"output":{"example_id":"chart/(Web_version)_E-Government_Survey_2024_1392024_p101","grounded_pages":[],"layout_pages":[],"markdown":"2024 UN E-GovErNmENt SUrvEy\n\n76\n\n**Chapter 2**\n\nFigure 2.33 Average OSI subindex values for groups of countries in special situations relative to global averages, 2024\n\nOverall, the legislative framework for advancing e-government is better developed in llDcs, including those that are least developed, than in SIDS or lDcs (see figure 2.35). Most llDcs have an e-government or digital government strategy (78 per cent), legislation or policy documents on cybersecurity (91 per cent), legal provisions for data protection (88 per cent), legislation governing freedom of information (78 per cent), and digital ID regulations (72 per cent). Fewer SIDS and lDcs (51 to 65 per cent) have these types of legislation in place, and the proportions are even lower for lDc/SIDS (between 13 and 65 per cent, depending on the type of legislation).\n\n_Source_ : 2024 United Nations E-Government Surveys.\n\nAmericas\n\nAfrica\n\nAsia\n\nEurope\n\nOceania\n\nVery high EGDI High EGDI Middle EGDI Low EGDI\n\n0\\.2791\n\n0\\.3261\n\n0\\.3672\n\n0\\.4444\n\n0\\.6875\n\n0\\.3598 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nSP\n\nTEC\n\nCP\n\nIF\n\nLDC/LLDCs\n\nLinear (average OSI for LDC/LLDCs) Linear (global average OSI)\n\n0\\.4919\n\n0\\.5363\n\n0\\.5606\n\n0\\.6644\n\n0\\.8079\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\n193 UN Member States Linear (global average OSI)\n\nYes No Metrix Yes No Metrix Yes No Metrix Yes No Metrix Yes No Metrix\n\n**SIDS LDCs LLDCs LDC/LLDCs LDC/SIDS**\n\nLDC/LLDCs\n\nLinear (average OSI for LDC/LLDC) F\n\nLinear (global average OSI)\n\n0\\.1969\n\n0\\.2169\n\n0\\.2344\n\n0\\.3333\n\n0\\.4750\n\n0\\.2407 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nSP\n\nTEC\n\nCP\n\nIF\n\nLDC/SIDS\n\nLinear (average OSI for LDC/SIDS) Linear (global average OSI)\n\n0\\.3081\n\n0\\.3902\n\n0\\.4038\n\n0\\.5075\n\n0\\.6746\n\n0\\.4076 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nSIDS\n\nLinear (average OSI for SIDS) Linear (global average OSI)\n\n0\\.2569\n\n0\\.2878\n\n0\\.3667\n\n0\\.4272\n\n0\\.6284\n\n0\\.3264 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nSP\n\nTEC\n\nCP\n\nIF\n\nLDCs\n\nLinear (average OSI for LDCs) Linear (global average OSI)\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nLLDCs\n\nLinear (average OSI for LLDCs) Linear (global average OSI)\n\n0\\.3960\n\n0\\.4668\n\n0\\.4680\n\n0\\.5938\n\n0\\.7725\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\n0\\.4885\n\nAmericas\n\nAfrica\n\nAsia\n\nEurope\n\nOceania\n\nVery high EGDI High EGDI Middle EGDI Low EGDI\n\n0\\.2791\n\n0\\.3261\n\n0\\.3672\n\n0\\.4444\n\n0\\.6875\n\n0\\.3598 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nSP\n\nTEC\n\nCP\n\nIF\n\nLDC/LLDCs\n\nLinear (average OSI for LDC/LLDCs) Linear (global average OSI)\n\n0\\.4919\n\n0\\.5363\n\n0\\.5606\n\n0\\.6644\n\n0\\.8079\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\n193 UN Member States Linear (global average OSI)\n\nYes No Metrix Yes No Metrix Yes No Metrix Yes No Metrix Yes No Metrix\n\n**SIDS LDCs LLDCs LDC/LLDCs LDC/SIDS**\n\nDCs\n\nLinear (average OSI for LDC/LLDC) F\n\nLinear (global average OSI)\n\n0\\.1969\n\n0\\.2169\n\n0\\.2344\n\n0\\.3333\n\n0\\.4750\n\n0\\.2407 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nSP\n\nTEC\n\nCP\n\nIF\n\nLDC/SIDS\n\nLinear (average OSI for LDC/SIDS) Linear (global average OSI)\n\n0\\.3081\n\n0\\.3902\n\n0\\.4038\n\n0\\.5075\n\n0\\.6746\n\n0\\.4076 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nSIDS\n\nLinear (average OSI for SIDS) Linear (global average OSI)\n\n0\\.2569\n\n0\\.2878\n\n0\\.3667\n\n0\\.4272\n\n0\\.6284\n\n0\\.3264 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nSP\n\nTEC\n\nCP\n\nIF\n\nLDCs\n\nLinear (average OSI for LDCs) Linear (global average OSI)\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nLLDCs\n\nLinear (average OSI for LLDCs) Linear (global average OSI)\n\n0\\.3960\n\n0\\.4668\n\n0\\.4680\n\n0\\.5938\n\n0\\.7725\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\n0\\.4885","pages":[{"line_numbers":[],"markdown":"2024 UN E-GovErNmENt SUrvEy\n\n76\n\n**Chapter 2**\n\nFigure 2.33 Average OSI subindex values for groups of countries in special situations relative to global averages, 2024\n\nOverall, the legislative framework for advancing e-government is better developed in llDcs, including those that are least developed, than in SIDS or lDcs (see figure 2.35). Most llDcs have an e-government or digital government strategy (78 per cent), legislation or policy documents on cybersecurity (91 per cent), legal provisions for data protection (88 per cent), legislation governing freedom of information (78 per cent), and digital ID regulations (72 per cent). Fewer SIDS and lDcs (51 to 65 per cent) have these types of legislation in place, and the proportions are even lower for lDc/SIDS (between 13 and 65 per cent, depending on the type of legislation).\n\n_Source_ : 2024 United Nations E-Government Surveys.\n\nAmericas\n\nAfrica\n\nAsia\n\nEurope\n\nOceania\n\nVery high EGDI High EGDI Middle EGDI Low EGDI\n\n0\\.2791\n\n0\\.3261\n\n0\\.3672\n\n0\\.4444\n\n0\\.6875\n\n0\\.3598 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nSP\n\nTEC\n\nCP\n\nIF\n\nLDC/LLDCs\n\nLinear (average OSI for LDC/LLDCs) Linear (global average OSI)\n\n0\\.4919\n\n0\\.5363\n\n0\\.5606\n\n0\\.6644\n\n0\\.8079\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\n193 UN Member States Linear (global average OSI)\n\nYes No Metrix Yes No Metrix Yes No Metrix Yes No Metrix Yes No Metrix\n\n**SIDS LDCs LLDCs LDC/LLDCs LDC/SIDS**\n\nLDC/LLDCs\n\nLinear (average OSI for LDC/LLDC) F\n\nLinear (global average OSI)\n\n0\\.1969\n\n0\\.2169\n\n0\\.2344\n\n0\\.3333\n\n0\\.4750\n\n0\\.2407 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nSP\n\nTEC\n\nCP\n\nIF\n\nLDC/SIDS\n\nLinear (average OSI for LDC/SIDS) Linear (global average OSI)\n\n0\\.3081\n\n0\\.3902\n\n0\\.4038\n\n0\\.5075\n\n0\\.6746\n\n0\\.4076 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nSIDS\n\nLinear (average OSI for SIDS) Linear (global average OSI)\n\n0\\.2569\n\n0\\.2878\n\n0\\.3667\n\n0\\.4272\n\n0\\.6284\n\n0\\.3264 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nSP\n\nTEC\n\nCP\n\nIF\n\nLDCs\n\nLinear (average OSI for LDCs) Linear (global average OSI)\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nLLDCs\n\nLinear (average OSI for LLDCs) Linear (global average OSI)\n\n0\\.3960\n\n0\\.4668\n\n0\\.4680\n\n0\\.5938\n\n0\\.7725\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\n0\\.4885\n\nAmericas\n\nAfrica\n\nAsia\n\nEurope\n\nOceania\n\nVery high EGDI High EGDI Middle EGDI Low EGDI\n\n0\\.2791\n\n0\\.3261\n\n0\\.3672\n\n0\\.4444\n\n0\\.6875\n\n0\\.3598 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nSP\n\nTEC\n\nCP\n\nIF\n\nLDC/LLDCs\n\nLinear (average OSI for LDC/LLDCs) Linear (global average OSI)\n\n0\\.4919\n\n0\\.5363\n\n0\\.5606\n\n0\\.6644\n\n0\\.8079\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\n193 UN Member States Linear (global average OSI)\n\nYes No Metrix Yes No Metrix Yes No Metrix Yes No Metrix Yes No Metrix\n\n**SIDS LDCs LLDCs LDC/LLDCs LDC/SIDS**\n\nDCs\n\nLinear (average OSI for LDC/LLDC) F\n\nLinear (global average OSI)\n\n0\\.1969\n\n0\\.2169\n\n0\\.2344\n\n0\\.3333\n\n0\\.4750\n\n0\\.2407 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nSP\n\nTEC\n\nCP\n\nIF\n\nLDC/SIDS\n\nLinear (average OSI for LDC/SIDS) Linear (global average OSI)\n\n0\\.3081\n\n0\\.3902\n\n0\\.4038\n\n0\\.5075\n\n0\\.6746\n\n0\\.4076 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nSIDS\n\nLinear (average OSI for SIDS) Linear (global average OSI)\n\n0\\.2569\n\n0\\.2878\n\n0\\.3667\n\n0\\.4272\n\n0\\.6284\n\n0\\.3264 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nSP\n\nTEC\n\nCP\n\nIF\n\nLDCs\n\nLinear (average OSI for LDCs) Linear (global average OSI)\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nLLDCs\n\nLinear (average OSI for LLDCs) Linear (global average OSI)\n\n0\\.3960\n\n0\\.4668\n\n0\\.4680\n\n0\\.5938\n\n0\\.7725\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\n0\\.4885","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"2024 UN E-GovErNmENt SUrvEy\n\n76\n\n**Chapter 2**\n\nFigure 2.33 Average OSI subindex values for groups of countries in special situations relative to global averages, 2024\n\nOverall, the legislative framework for advancing e-government is better developed in llDcs, including those that are least developed, than in SIDS or lDcs (see figure 2.35). Most llDcs have an e-government or digital government strategy (78 per cent), legislation or policy documents on cybersecurity (91 per cent), legal provisions for data protection (88 per cent), legislation governing freedom of information (78 per cent), and digital ID regulations (72 per cent). Fewer SIDS and lDcs (51 to 65 per cent) have these types of legislation in place, and the proportions are even lower for lDc/SIDS (between 13 and 65 per cent, depending on the type of legislation).\n\n_Source_ : 2024 United Nations E-Government Surveys.\n\nAmericas\n\nAfrica\n\nAsia\n\nEurope\n\nOceania\n\nVery high EGDI High EGDI Middle EGDI Low EGDI\n\n0\\.2791\n\n0\\.3261\n\n0\\.3672\n\n0\\.4444\n\n0\\.6875\n\n0\\.3598 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nSP\n\nTEC\n\nCP\n\nIF\n\nLDC/LLDCs\n\nLinear (average OSI for LDC/LLDCs) Linear (global average OSI)\n\n0\\.4919\n\n0\\.5363\n\n0\\.5606\n\n0\\.6644\n\n0\\.8079\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\n193 UN Member States Linear (global average OSI)\n\nYes No Metrix Yes No Metrix Yes No Metrix Yes No Metrix Yes No Metrix\n\n**SIDS LDCs LLDCs LDC/LLDCs LDC/SIDS**\n\nLDC/LLDCs\n\nLinear (average OSI for LDC/LLDC) F\n\nLinear (global average OSI)\n\n0\\.1969\n\n0\\.2169\n\n0\\.2344\n\n0\\.3333\n\n0\\.4750\n\n0\\.2407 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nSP\n\nTEC\n\nCP\n\nIF\n\nLDC/SIDS\n\nLinear (average OSI for LDC/SIDS) Linear (global average OSI)\n\n0\\.3081\n\n0\\.3902\n\n0\\.4038\n\n0\\.5075\n\n0\\.6746\n\n0\\.4076 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nSIDS\n\nLinear (average OSI for SIDS) Linear (global average OSI)\n\n0\\.2569\n\n0\\.2878\n\n0\\.3667\n\n0\\.4272\n\n0\\.6284\n\n0\\.3264 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nSP\n\nTEC\n\nCP\n\nIF\n\nLDCs\n\nLinear (average OSI for LDCs) Linear (global average OSI)\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nLLDCs\n\nLinear (average OSI for LLDCs) Linear (global average OSI)\n\n0\\.3960\n\n0\\.4668\n\n0\\.4680\n\n0\\.5938\n\n0\\.7725\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\n0\\.4885\n\nAmericas\n\nAfrica\n\nAsia\n\nEurope\n\nOceania\n\nVery high EGDI High EGDI Middle EGDI Low EGDI\n\n0\\.2791\n\n0\\.3261\n\n0\\.3672\n\n0\\.4444\n\n0\\.6875\n\n0\\.3598 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nSP\n\nTEC\n\nCP\n\nIF\n\nLDC/LLDCs\n\nLinear (average OSI for LDC/LLDCs) Linear (global average OSI)\n\n0\\.4919\n\n0\\.5363\n\n0\\.5606\n\n0\\.6644\n\n0\\.8079\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\n193 UN Member States Linear (global average OSI)\n\nYes No Metrix Yes No Metrix Yes No Metrix Yes No Metrix Yes No Metrix\n\n**SIDS LDCs LLDCs LDC/LLDCs LDC/SIDS**\n\nDCs\n\nLinear (average OSI for LDC/LLDC) F\n\nLinear (global average OSI)\n\n0\\.1969\n\n0\\.2169\n\n0\\.2344\n\n0\\.3333\n\n0\\.4750\n\n0\\.2407 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nSP\n\nTEC\n\nCP\n\nIF\n\nLDC/SIDS\n\nLinear (average OSI for LDC/SIDS) Linear (global average OSI)\n\n0\\.3081\n\n0\\.3902\n\n0\\.4038\n\n0\\.5075\n\n0\\.6746\n\n0\\.4076 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nSIDS\n\nLinear (average OSI for SIDS) Linear (global average OSI)\n\n0\\.2569\n\n0\\.2878\n\n0\\.3667\n\n0\\.4272\n\n0\\.6284\n\n0\\.3264 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nSP\n\nTEC\n\nCP\n\nIF\n\nLDCs\n\nLinear (average OSI for LDCs) Linear (global average OSI)\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nLLDCs\n\nLinear (average OSI for LLDCs) Linear (global average OSI)\n\n0\\.3960\n\n0\\.4668\n\n0\\.4680\n\n0\\.5938\n\n0\\.7725\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\n0\\.4885","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/%28Web_version%29_E-Government_Survey_2024_1392024_p101.pdf"]},"response":{"errors":[],"extract_id":"extract_8d665101814c5994a2ba88e412a90077","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["2024 UN E-GovErNmENt SUrvEy\n\n76\n\n**Chapter 2**\n\nFigure 2.33 Average OSI subindex values for groups of countries in special situations relative to global averages, 2024\n\nOverall, the legislative framework for advancing e-government is better developed in llDcs, including those that are least developed, than in SIDS or lDcs (see figure 2.35). Most llDcs have an e-government or digital government strategy (78 per cent), legislation or policy documents on cybersecurity (91 per cent), legal provisions for data protection (88 per cent), legislation governing freedom of information (78 per cent), and digital ID regulations (72 per cent). Fewer SIDS and lDcs (51 to 65 per cent) have these types of legislation in place, and the proportions are even lower for lDc/SIDS (between 13 and 65 per cent, depending on the type of legislation).\n\n_Source_ : 2024 United Nations E-Government Surveys.\n\nAmericas\n\nAfrica\n\nAsia\n\nEurope\n\nOceania\n\nVery high EGDI High EGDI Middle EGDI Low EGDI\n\n0\\.2791\n\n0\\.3261\n\n0\\.3672\n\n0\\.4444\n\n0\\.6875\n\n0\\.3598 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nSP\n\nTEC\n\nCP\n\nIF\n\nLDC/LLDCs\n\nLinear (average OSI for LDC/LLDCs) Linear (global average OSI)\n\n0\\.4919\n\n0\\.5363\n\n0\\.5606\n\n0\\.6644\n\n0\\.8079\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\n193 UN Member States Linear (global average OSI)\n\nYes No Metrix Yes No Metrix Yes No Metrix Yes No Metrix Yes No Metrix\n\n**SIDS LDCs LLDCs LDC/LLDCs LDC/SIDS**\n\nLDC/LLDCs\n\nLinear (average OSI for LDC/LLDC) F\n\nLinear (global average OSI)\n\n0\\.1969\n\n0\\.2169\n\n0\\.2344\n\n0\\.3333\n\n0\\.4750\n\n0\\.2407 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nSP\n\nTEC\n\nCP\n\nIF\n\nLDC/SIDS\n\nLinear (average OSI for LDC/SIDS) Linear (global average OSI)\n\n0\\.3081\n\n0\\.3902\n\n0\\.4038\n\n0\\.5075\n\n0\\.6746\n\n0\\.4076 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nSIDS\n\nLinear (average OSI for SIDS) Linear (global average OSI)\n\n0\\.2569\n\n0\\.2878\n\n0\\.3667\n\n0\\.4272\n\n0\\.6284\n\n0\\.3264 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nSP\n\nTEC\n\nCP\n\nIF\n\nLDCs\n\nLinear (average OSI for LDCs) Linear (global average OSI)\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nLLDCs\n\nLinear (average OSI for LLDCs) Linear (global average OSI)\n\n0\\.3960\n\n0\\.4668\n\n0\\.4680\n\n0\\.5938\n\n0\\.7725\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\n0\\.4885\n\nAmericas\n\nAfrica\n\nAsia\n\nEurope\n\nOceania\n\nVery high EGDI High EGDI Middle EGDI Low EGDI\n\n0\\.2791\n\n0\\.3261\n\n0\\.3672\n\n0\\.4444\n\n0\\.6875\n\n0\\.3598 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nSP\n\nTEC\n\nCP\n\nIF\n\nLDC/LLDCs\n\nLinear (average OSI for LDC/LLDCs) Linear (global average OSI)\n\n0\\.4919\n\n0\\.5363\n\n0\\.5606\n\n0\\.6644\n\n0\\.8079\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\n193 UN Member States Linear (global average OSI)\n\nYes No Metrix Yes No Metrix Yes No Metrix Yes No Metrix Yes No Metrix\n\n**SIDS LDCs LLDCs LDC/LLDCs LDC/SIDS**\n\nDCs\n\nLinear (average OSI for LDC/LLDC) F\n\nLinear (global average OSI)\n\n0\\.1969\n\n0\\.2169\n\n0\\.2344\n\n0\\.3333\n\n0\\.4750\n\n0\\.2407 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nSP\n\nTEC\n\nCP\n\nIF\n\nLDC/SIDS\n\nLinear (average OSI for LDC/SIDS) Linear (global average OSI)\n\n0\\.3081\n\n0\\.3902\n\n0\\.4038\n\n0\\.5075\n\n0\\.6746\n\n0\\.4076 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nSIDS\n\nLinear (average OSI for SIDS) Linear (global average OSI)\n\n0\\.2569\n\n0\\.2878\n\n0\\.3667\n\n0\\.4272\n\n0\\.6284\n\n0\\.3264 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nSP\n\nTEC\n\nCP\n\nIF\n\nLDCs\n\nLinear (average OSI for LDCs) Linear (global average OSI)\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nLLDCs\n\nLinear (average OSI for LLDCs) Linear (global average OSI)\n\n0\\.3960\n\n0\\.4668\n\n0\\.4680\n\n0\\.5938\n\n0\\.7725\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\n0\\.4885"],"full_content":"2024 UN E-GovErNmENt SUrvEy\n\n76\n\n**Chapter 2**\n\nFigure 2.33 Average OSI subindex values for groups of countries in special situations relative to global averages, 2024\n\nOverall, the legislative framework for advancing e-government is better developed in llDcs, including those that are least developed, than in SIDS or lDcs (see figure 2.35). Most llDcs have an e-government or digital government strategy (78 per cent), legislation or policy documents on cybersecurity (91 per cent), legal provisions for data protection (88 per cent), legislation governing freedom of information (78 per cent), and digital ID regulations (72 per cent). Fewer SIDS and lDcs (51 to 65 per cent) have these types of legislation in place, and the proportions are even lower for lDc/SIDS (between 13 and 65 per cent, depending on the type of legislation).\n\n_Source_ : 2024 United Nations E-Government Surveys.\n\nAmericas\n\nAfrica\n\nAsia\n\nEurope\n\nOceania\n\nVery high EGDI High EGDI Middle EGDI Low EGDI\n\n0\\.2791\n\n0\\.3261\n\n0\\.3672\n\n0\\.4444\n\n0\\.6875\n\n0\\.3598 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nSP\n\nTEC\n\nCP\n\nIF\n\nLDC/LLDCs\n\nLinear (average OSI for LDC/LLDCs) Linear (global average OSI)\n\n0\\.4919\n\n0\\.5363\n\n0\\.5606\n\n0\\.6644\n\n0\\.8079\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\n193 UN Member States Linear (global average OSI)\n\nYes No Metrix Yes No Metrix Yes No Metrix Yes No Metrix Yes No Metrix\n\n**SIDS LDCs LLDCs LDC/LLDCs LDC/SIDS**\n\nLDC/LLDCs\n\nLinear (average OSI for LDC/LLDC) F\n\nLinear (global average OSI)\n\n0\\.1969\n\n0\\.2169\n\n0\\.2344\n\n0\\.3333\n\n0\\.4750\n\n0\\.2407 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nSP\n\nTEC\n\nCP\n\nIF\n\nLDC/SIDS\n\nLinear (average OSI for LDC/SIDS) Linear (global average OSI)\n\n0\\.3081\n\n0\\.3902\n\n0\\.4038\n\n0\\.5075\n\n0\\.6746\n\n0\\.4076 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nSIDS\n\nLinear (average OSI for SIDS) Linear (global average OSI)\n\n0\\.2569\n\n0\\.2878\n\n0\\.3667\n\n0\\.4272\n\n0\\.6284\n\n0\\.3264 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nSP\n\nTEC\n\nCP\n\nIF\n\nLDCs\n\nLinear (average OSI for LDCs) Linear (global average OSI)\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nLLDCs\n\nLinear (average OSI for LLDCs) Linear (global average OSI)\n\n0\\.3960\n\n0\\.4668\n\n0\\.4680\n\n0\\.5938\n\n0\\.7725\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\n0\\.4885\n\nAmericas\n\nAfrica\n\nAsia\n\nEurope\n\nOceania\n\nVery high EGDI High EGDI Middle EGDI Low EGDI\n\n0\\.2791\n\n0\\.3261\n\n0\\.3672\n\n0\\.4444\n\n0\\.6875\n\n0\\.3598 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nSP\n\nTEC\n\nCP\n\nIF\n\nLDC/LLDCs\n\nLinear (average OSI for LDC/LLDCs) Linear (global average OSI)\n\n0\\.4919\n\n0\\.5363\n\n0\\.5606\n\n0\\.6644\n\n0\\.8079\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\n193 UN Member States Linear (global average OSI)\n\nYes No Metrix Yes No Metrix Yes No Metrix Yes No Metrix Yes No Metrix\n\n**SIDS LDCs LLDCs LDC/LLDCs LDC/SIDS**\n\nDCs\n\nLinear (average OSI for LDC/LLDC) F\n\nLinear (global average OSI)\n\n0\\.1969\n\n0\\.2169\n\n0\\.2344\n\n0\\.3333\n\n0\\.4750\n\n0\\.2407 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nSP\n\nTEC\n\nCP\n\nIF\n\nLDC/SIDS\n\nLinear (average OSI for LDC/SIDS) Linear (global average OSI)\n\n0\\.3081\n\n0\\.3902\n\n0\\.4038\n\n0\\.5075\n\n0\\.6746\n\n0\\.4076 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nSIDS\n\nLinear (average OSI for SIDS) Linear (global average OSI)\n\n0\\.2569\n\n0\\.2878\n\n0\\.3667\n\n0\\.4272\n\n0\\.6284\n\n0\\.3264 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nSP\n\nTEC\n\nCP\n\nIF\n\nLDCs\n\nLinear (average OSI for LDCs) Linear (global average OSI)\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nLLDCs\n\nLinear (average OSI for LLDCs) Linear (global average OSI)\n\n0\\.3960\n\n0\\.4668\n\n0\\.4680\n\n0\\.5938\n\n0\\.7725\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\n0\\.4885","publish_date":null,"title":"Very high EGDI Very high EGDI High EGDI High EGDI Middle EGDI Middle EGDI Low EGDI Low EGDI","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/%28Web_version%29_E-Government_Survey_2024_1392024_p101.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/%28Web_version%29_E-Government_Survey_2024_1392024_p101.pdf"},"request":{"example_id":"chart/(Web_version)_E-Government_Survey_2024_1392024_p101","product_type":"parse","source_file_path":"(Web_version)_E-Government_Survey_2024_1392024_p101.pdf"},"started_at":"2026-11-04T19:45:39.146841"} +{"completed_at":"2026-11-04T19:45:42.320422","latency_in_ms":1461,"output":{"example_id":"chart/(Web_version)_E-Government_Survey_2024_1392024_p102","grounded_pages":[],"layout_pages":[],"markdown":"77\n\n**Chapter 2**\n\nChapter 2 • Global trends in e-Government\n\nFigure 2.34 Percentage of countries in special situations that have implemented organizational features of the OSI institutional framework subindex, 2024\n\nFigure 2.35 Percentage of countries in special situations with legislative frameworks relevant to e-government development, 2024\n\n_Source_ : 2024 United Nations E-Government Surveys.\n\n_Source_ : 2024 United Nations E-Government Surveys.\n\n193 UN Member States\n\nAmericas\n\nAfrica\n\nAsia\n\nEurope\n\nOceania\n\n## Very high EGDI High EGDI Middle EGDI Low EGDI\n\nYes No Metrix Yes No MetrixMetrix Yes No\n\n**SIDS LDCs LLDCs**\n\n97%\n\n97%\n\n68%\n\n97%\n\n100%\n\n89%\n\n67%\n\n84%\n\n100%\n\n94%\n\n75%\n\n84%\n\n100%\n\n88%\n\n69%\n\n88%\n\n100%\n\n100%\n\n50%\n\n100%\n\nLinks on the national portal to any subnational/local government agencies\n\nAvailability of organizational structure and/or chart of the government\n\nInformation on national portal about a national Chief Information Offcer (CIO) or equivalent\n\nAvailability of names/titles of heads of government agencies/departments/ministries\n\nYes No Yes No Yes No Yes No Yes No\n\nYes No MetrixMetrix Yes No MetrixMet Yes No MetrixMetr Yes No\n\n**SIDS LDCs LLDCs LDC/LLDCs LDC/SIDS**\n\nOrganizational aspects of the institutional framework\n\n### Very h High E\n\n193 UN Member States\n\nAmericas\n\nAfrica\n\nAsia\n\nEurope\n\nOceania\n\n### Very high EGDI High EGDI Middle EGDI Low EGDI\n\n**SIDS LDCs LLDCs LDC/LLDCs LDC/SIDS**\n\n19%\n\n32%\n\n19%\n\n27%\n\n22%\n\n54%\n\n65%\n\n65%\n\n62%\n\n54%\n\n11%\n\n13%\n\n18%\n\n20%\n\n33%\n\n53%\n\n62%\n\n64%\n\n60%\n\n51%\n\n22%\n\n25%\n\n34%\n\n34%\n\n56%\n\n78%\n\n78%\n\n91%\n\n88%\n\n72%\n\n6%\n\n13%\n\n19%\n\n19%\n\n38%\n\n75%\n\n69%\n\n81%\n\n75%\n\n56%\n\n25%\n\n13%\n\n13%\n\n25%\n\n63%\n\n38%\n\n38%\n\n13%\n\nAI\n\nCloud\n\nMisinformaiton, disinformation or fake news\n\nE-participation\n\nOpen government data\n\nFreedom of or access to information act\n\nE-government or digital government\n\nCybersecurity\n\nData privacy and data protection\n\nDigital ID\n\nYes No Yes No Yes No Yes No Yes No\n\n**SIDS LDCs LLDCs LDC/LLDCs LDC/SIDS**\n\nAvailability of legislation, policy or strategy documents on:\n\nPolicies on open government data, e-participation, artificial intelligence, cloud computing, and protecting the public against misinformation, disinformation and/or fake news are found in about 3 out of 10 llDcs, 2 out of 10 SIDS, and 1 out 10 lDcs.","pages":[{"line_numbers":[],"markdown":"77\n\n**Chapter 2**\n\nChapter 2 • Global trends in e-Government\n\nFigure 2.34 Percentage of countries in special situations that have implemented organizational features of the OSI institutional framework subindex, 2024\n\nFigure 2.35 Percentage of countries in special situations with legislative frameworks relevant to e-government development, 2024\n\n_Source_ : 2024 United Nations E-Government Surveys.\n\n_Source_ : 2024 United Nations E-Government Surveys.\n\n193 UN Member States\n\nAmericas\n\nAfrica\n\nAsia\n\nEurope\n\nOceania\n\n## Very high EGDI High EGDI Middle EGDI Low EGDI\n\nYes No Metrix Yes No MetrixMetrix Yes No\n\n**SIDS LDCs LLDCs**\n\n97%\n\n97%\n\n68%\n\n97%\n\n100%\n\n89%\n\n67%\n\n84%\n\n100%\n\n94%\n\n75%\n\n84%\n\n100%\n\n88%\n\n69%\n\n88%\n\n100%\n\n100%\n\n50%\n\n100%\n\nLinks on the national portal to any subnational/local government agencies\n\nAvailability of organizational structure and/or chart of the government\n\nInformation on national portal about a national Chief Information Offcer (CIO) or equivalent\n\nAvailability of names/titles of heads of government agencies/departments/ministries\n\nYes No Yes No Yes No Yes No Yes No\n\nYes No MetrixMetrix Yes No MetrixMet Yes No MetrixMetr Yes No\n\n**SIDS LDCs LLDCs LDC/LLDCs LDC/SIDS**\n\nOrganizational aspects of the institutional framework\n\n### Very h High E\n\n193 UN Member States\n\nAmericas\n\nAfrica\n\nAsia\n\nEurope\n\nOceania\n\n### Very high EGDI High EGDI Middle EGDI Low EGDI\n\n**SIDS LDCs LLDCs LDC/LLDCs LDC/SIDS**\n\n19%\n\n32%\n\n19%\n\n27%\n\n22%\n\n54%\n\n65%\n\n65%\n\n62%\n\n54%\n\n11%\n\n13%\n\n18%\n\n20%\n\n33%\n\n53%\n\n62%\n\n64%\n\n60%\n\n51%\n\n22%\n\n25%\n\n34%\n\n34%\n\n56%\n\n78%\n\n78%\n\n91%\n\n88%\n\n72%\n\n6%\n\n13%\n\n19%\n\n19%\n\n38%\n\n75%\n\n69%\n\n81%\n\n75%\n\n56%\n\n25%\n\n13%\n\n13%\n\n25%\n\n63%\n\n38%\n\n38%\n\n13%\n\nAI\n\nCloud\n\nMisinformaiton, disinformation or fake news\n\nE-participation\n\nOpen government data\n\nFreedom of or access to information act\n\nE-government or digital government\n\nCybersecurity\n\nData privacy and data protection\n\nDigital ID\n\nYes No Yes No Yes No Yes No Yes No\n\n**SIDS LDCs LLDCs LDC/LLDCs LDC/SIDS**\n\nAvailability of legislation, policy or strategy documents on:\n\nPolicies on open government data, e-participation, artificial intelligence, cloud computing, and protecting the public against misinformation, disinformation and/or fake news are found in about 3 out of 10 llDcs, 2 out of 10 SIDS, and 1 out 10 lDcs.","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"77\n\n**Chapter 2**\n\nChapter 2 • Global trends in e-Government\n\nFigure 2.34 Percentage of countries in special situations that have implemented organizational features of the OSI institutional framework subindex, 2024\n\nFigure 2.35 Percentage of countries in special situations with legislative frameworks relevant to e-government development, 2024\n\n_Source_ : 2024 United Nations E-Government Surveys.\n\n_Source_ : 2024 United Nations E-Government Surveys.\n\n193 UN Member States\n\nAmericas\n\nAfrica\n\nAsia\n\nEurope\n\nOceania\n\n## Very high EGDI High EGDI Middle EGDI Low EGDI\n\nYes No Metrix Yes No MetrixMetrix Yes No\n\n**SIDS LDCs LLDCs**\n\n97%\n\n97%\n\n68%\n\n97%\n\n100%\n\n89%\n\n67%\n\n84%\n\n100%\n\n94%\n\n75%\n\n84%\n\n100%\n\n88%\n\n69%\n\n88%\n\n100%\n\n100%\n\n50%\n\n100%\n\nLinks on the national portal to any subnational/local government agencies\n\nAvailability of organizational structure and/or chart of the government\n\nInformation on national portal about a national Chief Information Offcer (CIO) or equivalent\n\nAvailability of names/titles of heads of government agencies/departments/ministries\n\nYes No Yes No Yes No Yes No Yes No\n\nYes No MetrixMetrix Yes No MetrixMet Yes No MetrixMetr Yes No\n\n**SIDS LDCs LLDCs LDC/LLDCs LDC/SIDS**\n\nOrganizational aspects of the institutional framework\n\n### Very h High E\n\n193 UN Member States\n\nAmericas\n\nAfrica\n\nAsia\n\nEurope\n\nOceania\n\n### Very high EGDI High EGDI Middle EGDI Low EGDI\n\n**SIDS LDCs LLDCs LDC/LLDCs LDC/SIDS**\n\n19%\n\n32%\n\n19%\n\n27%\n\n22%\n\n54%\n\n65%\n\n65%\n\n62%\n\n54%\n\n11%\n\n13%\n\n18%\n\n20%\n\n33%\n\n53%\n\n62%\n\n64%\n\n60%\n\n51%\n\n22%\n\n25%\n\n34%\n\n34%\n\n56%\n\n78%\n\n78%\n\n91%\n\n88%\n\n72%\n\n6%\n\n13%\n\n19%\n\n19%\n\n38%\n\n75%\n\n69%\n\n81%\n\n75%\n\n56%\n\n25%\n\n13%\n\n13%\n\n25%\n\n63%\n\n38%\n\n38%\n\n13%\n\nAI\n\nCloud\n\nMisinformaiton, disinformation or fake news\n\nE-participation\n\nOpen government data\n\nFreedom of or access to information act\n\nE-government or digital government\n\nCybersecurity\n\nData privacy and data protection\n\nDigital ID\n\nYes No Yes No Yes No Yes No Yes No\n\n**SIDS LDCs LLDCs LDC/LLDCs LDC/SIDS**\n\nAvailability of legislation, policy or strategy documents on:\n\nPolicies on open government data, e-participation, artificial intelligence, cloud computing, and protecting the public against misinformation, disinformation and/or fake news are found in about 3 out of 10 llDcs, 2 out of 10 SIDS, and 1 out 10 lDcs.","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/%28Web_version%29_E-Government_Survey_2024_1392024_p102.pdf"]},"response":{"errors":[],"extract_id":"extract_d2da11acf54c5d0094cf8ae2bf259948","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["77\n\n**Chapter 2**\n\nChapter 2 • Global trends in e-Government\n\nFigure 2.34 Percentage of countries in special situations that have implemented organizational features of the OSI institutional framework subindex, 2024\n\nFigure 2.35 Percentage of countries in special situations with legislative frameworks relevant to e-government development, 2024\n\n_Source_ : 2024 United Nations E-Government Surveys.\n\n_Source_ : 2024 United Nations E-Government Surveys.\n\n193 UN Member States\n\nAmericas\n\nAfrica\n\nAsia\n\nEurope\n\nOceania\n\n## Very high EGDI High EGDI Middle EGDI Low EGDI\n\nYes No Metrix Yes No MetrixMetrix Yes No\n\n**SIDS LDCs LLDCs**\n\n97%\n\n97%\n\n68%\n\n97%\n\n100%\n\n89%\n\n67%\n\n84%\n\n100%\n\n94%\n\n75%\n\n84%\n\n100%\n\n88%\n\n69%\n\n88%\n\n100%\n\n100%\n\n50%\n\n100%\n\nLinks on the national portal to any subnational/local government agencies\n\nAvailability of organizational structure and/or chart of the government\n\nInformation on national portal about a national Chief Information Offcer (CIO) or equivalent\n\nAvailability of names/titles of heads of government agencies/departments/ministries\n\nYes No Yes No Yes No Yes No Yes No\n\nYes No MetrixMetrix Yes No MetrixMet Yes No MetrixMetr Yes No\n\n**SIDS LDCs LLDCs LDC/LLDCs LDC/SIDS**\n\nOrganizational aspects of the institutional framework\n\n### Very h High E\n\n193 UN Member States\n\nAmericas\n\nAfrica\n\nAsia\n\nEurope\n\nOceania\n\n### Very high EGDI High EGDI Middle EGDI Low EGDI\n\n**SIDS LDCs LLDCs LDC/LLDCs LDC/SIDS**\n\n19%\n\n32%\n\n19%\n\n27%\n\n22%\n\n54%\n\n65%\n\n65%\n\n62%\n\n54%\n\n11%\n\n13%\n\n18%\n\n20%\n\n33%\n\n53%\n\n62%\n\n64%\n\n60%\n\n51%\n\n22%\n\n25%\n\n34%\n\n34%\n\n56%\n\n78%\n\n78%\n\n91%\n\n88%\n\n72%\n\n6%\n\n13%\n\n19%\n\n19%\n\n38%\n\n75%\n\n69%\n\n81%\n\n75%\n\n56%\n\n25%\n\n13%\n\n13%\n\n25%\n\n63%\n\n38%\n\n38%\n\n13%\n\nAI\n\nCloud\n\nMisinformaiton, disinformation or fake news\n\nE-participation\n\nOpen government data\n\nFreedom of or access to information act\n\nE-government or digital government\n\nCybersecurity\n\nData privacy and data protection\n\nDigital ID\n\nYes No Yes No Yes No Yes No Yes No\n\n**SIDS LDCs LLDCs LDC/LLDCs LDC/SIDS**\n\nAvailability of legislation, policy or strategy documents on:\n\nPolicies on open government data, e-participation, artificial intelligence, cloud computing, and protecting the public against misinformation, disinformation and/or fake news are found in about 3 out of 10 llDcs, 2 out of 10 SIDS, and 1 out 10 lDcs."],"full_content":"77\n\n**Chapter 2**\n\nChapter 2 • Global trends in e-Government\n\nFigure 2.34 Percentage of countries in special situations that have implemented organizational features of the OSI institutional framework subindex, 2024\n\nFigure 2.35 Percentage of countries in special situations with legislative frameworks relevant to e-government development, 2024\n\n_Source_ : 2024 United Nations E-Government Surveys.\n\n_Source_ : 2024 United Nations E-Government Surveys.\n\n193 UN Member States\n\nAmericas\n\nAfrica\n\nAsia\n\nEurope\n\nOceania\n\n## Very high EGDI High EGDI Middle EGDI Low EGDI\n\nYes No Metrix Yes No MetrixMetrix Yes No\n\n**SIDS LDCs LLDCs**\n\n97%\n\n97%\n\n68%\n\n97%\n\n100%\n\n89%\n\n67%\n\n84%\n\n100%\n\n94%\n\n75%\n\n84%\n\n100%\n\n88%\n\n69%\n\n88%\n\n100%\n\n100%\n\n50%\n\n100%\n\nLinks on the national portal to any subnational/local government agencies\n\nAvailability of organizational structure and/or chart of the government\n\nInformation on national portal about a national Chief Information Offcer (CIO) or equivalent\n\nAvailability of names/titles of heads of government agencies/departments/ministries\n\nYes No Yes No Yes No Yes No Yes No\n\nYes No MetrixMetrix Yes No MetrixMet Yes No MetrixMetr Yes No\n\n**SIDS LDCs LLDCs LDC/LLDCs LDC/SIDS**\n\nOrganizational aspects of the institutional framework\n\n### Very h High E\n\n193 UN Member States\n\nAmericas\n\nAfrica\n\nAsia\n\nEurope\n\nOceania\n\n### Very high EGDI High EGDI Middle EGDI Low EGDI\n\n**SIDS LDCs LLDCs LDC/LLDCs LDC/SIDS**\n\n19%\n\n32%\n\n19%\n\n27%\n\n22%\n\n54%\n\n65%\n\n65%\n\n62%\n\n54%\n\n11%\n\n13%\n\n18%\n\n20%\n\n33%\n\n53%\n\n62%\n\n64%\n\n60%\n\n51%\n\n22%\n\n25%\n\n34%\n\n34%\n\n56%\n\n78%\n\n78%\n\n91%\n\n88%\n\n72%\n\n6%\n\n13%\n\n19%\n\n19%\n\n38%\n\n75%\n\n69%\n\n81%\n\n75%\n\n56%\n\n25%\n\n13%\n\n13%\n\n25%\n\n63%\n\n38%\n\n38%\n\n13%\n\nAI\n\nCloud\n\nMisinformaiton, disinformation or fake news\n\nE-participation\n\nOpen government data\n\nFreedom of or access to information act\n\nE-government or digital government\n\nCybersecurity\n\nData privacy and data protection\n\nDigital ID\n\nYes No Yes No Yes No Yes No Yes No\n\n**SIDS LDCs LLDCs LDC/LLDCs LDC/SIDS**\n\nAvailability of legislation, policy or strategy documents on:\n\nPolicies on open government data, e-participation, artificial intelligence, cloud computing, and protecting the public against misinformation, disinformation and/or fake news are found in about 3 out of 10 llDcs, 2 out of 10 SIDS, and 1 out 10 lDcs.","publish_date":null,"title":"Very high EGDI High EGDI Middle EGDI Low EGDI","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/%28Web_version%29_E-Government_Survey_2024_1392024_p102.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/%28Web_version%29_E-Government_Survey_2024_1392024_p102.pdf"},"request":{"example_id":"chart/(Web_version)_E-Government_Survey_2024_1392024_p102","product_type":"parse","source_file_path":"(Web_version)_E-Government_Survey_2024_1392024_p102.pdf"},"started_at":"2026-11-04T19:45:40.859224"} +{"completed_at":"2026-11-04T19:45:45.268037","latency_in_ms":3098,"output":{"example_id":"chart/(Web_version)_E-Government_Survey_2024_1392024_p103","grounded_pages":[],"layout_pages":[],"markdown":"2024 UN E-GovErNmENt SUrvEy\n\n78\n\n**Chapter 2**\n\nE-participation in countries in special situations\n\nconsistent with global trends, e-participation values remain the lowest among the five OSI subindices for countries in special situations. This is a clear indication that these countries are facing broad challenges in engaging citizens through digital platforms, which is crucial for inclusive and participatory governance. Enhancing e-participation in these regions will require not only investments in technology but also significant cultural and policy shifts to encourage and facilitate citizen engagement. Over the years UN DESA implemented and continues to support countries with various capacity building initiatives in the areas of digital government, participation and accountability, innovation and delivery of public services that contribute to the achievement of sustainable development goal 16 ⁹ .\n\n_Open government data (OGD)_\n\nGroup averages indicating the proportion of countries in special situations that share information through dedicated OGD portals are comparable to the corresponding global average; between 63 and 81 per cent of the countries in the respective groups maintain OGD portals, and 50 to 68 per cent provide geospatial data. Most countries in these groups also share information about government expenditures, often in open formats. However, only 3 out of 10 countries in special situations allow members of the public to request or propose new open data sets, compared to about 5 in 10 countries at the global level. More than 70 per cent of lDcs, llDcs and SIDS have not yet adopted open data licensing, and events such as hackathons around open data use are rare. Most OGD are not available in real time, and fewer than 2 in 10 countries have participatory budgeting mechanisms (see figure 2.36).\n\nFigure 2.36 Percentage of countries in special situations with OGD portals and various aspects of open data governance\n\n_Source_ : 2024 United Nations E-Government Survey.\n\n_Note_ : The availability of OGD in machine-readable formats is indicated by the lighter shades of the same colours in the regional performance graphs.\n\n\\* The availability of OGD in real time for both machine-readable and non-machine-readable data sets.\n\nEDGI 2024 levels\n\nVery high EGDI High EGDI Middle EGDI\n\n### Very hig High EG Middle Low EG\n\n193 UN Member States\n\nAmericas\n\nAfrica\n\nAsia\n\nEurope\n\nOceania\n\n### Very high EGDI High EGDI Middle EGDI Low EGDI\n\nY N M t i Y N M t i M t i Y N M t i M t Y N M t i M t Y N\n\n**SIDS LDCs LLDCs LDC/LLDCs LDC/SIDS**\n\n11%\n\n68%\n\n14%\n\n11%\n\n24%\n\n35%\n\n68%\n\n73%\n\n19%\n\n5%\n\n13%\n\n62%\n\n16%\n\n11%\n\n18%\n\n31%\n\n51%\n\n67%\n\n18%\n\n11%\n\n16%\n\n47%\n\n19%\n\n31%\n\n31%\n\n34%\n\n53%\n\n81%\n\n41%\n\n6%\n\n13%\n\n63%\n\n19%\n\n25%\n\n25%\n\n25%\n\n50%\n\n75%\n\n19%\n\n6%\n\n38%\n\n38%\n\n63%\n\n63%\n\n25%\n\n13%\n\nThere are participatory budgeting mechanisms on national portals\n\nThere are OGD available on national government expenditures or budget\n\nOGD is available in real time\n\nInformation is available about competitions, hackathons, or events around the use of OGD\n\nThere is open data licensing for OGD\n\nPublic can request or propose new OGD data sets\n\nNational portals provide GIS or other geospatial data\n\nThere is an OGD portal\n\nYes Machine-readable No Yes Machine-readable No Yes Machine-readable No Yes Machine-readable No Yes Machine-readable No\n\nYes Machine-readable No Yes Machine-readable No Yes Machine-readable No Yes Machine-readable No\n\n**SIDS LDCs LLDCs LDC/LLDCs LDC/SIDS**","pages":[{"line_numbers":[],"markdown":"2024 UN E-GovErNmENt SUrvEy\n\n78\n\n**Chapter 2**\n\nE-participation in countries in special situations\n\nconsistent with global trends, e-participation values remain the lowest among the five OSI subindices for countries in special situations. This is a clear indication that these countries are facing broad challenges in engaging citizens through digital platforms, which is crucial for inclusive and participatory governance. Enhancing e-participation in these regions will require not only investments in technology but also significant cultural and policy shifts to encourage and facilitate citizen engagement. Over the years UN DESA implemented and continues to support countries with various capacity building initiatives in the areas of digital government, participation and accountability, innovation and delivery of public services that contribute to the achievement of sustainable development goal 16 ⁹ .\n\n_Open government data (OGD)_\n\nGroup averages indicating the proportion of countries in special situations that share information through dedicated OGD portals are comparable to the corresponding global average; between 63 and 81 per cent of the countries in the respective groups maintain OGD portals, and 50 to 68 per cent provide geospatial data. Most countries in these groups also share information about government expenditures, often in open formats. However, only 3 out of 10 countries in special situations allow members of the public to request or propose new open data sets, compared to about 5 in 10 countries at the global level. More than 70 per cent of lDcs, llDcs and SIDS have not yet adopted open data licensing, and events such as hackathons around open data use are rare. Most OGD are not available in real time, and fewer than 2 in 10 countries have participatory budgeting mechanisms (see figure 2.36).\n\nFigure 2.36 Percentage of countries in special situations with OGD portals and various aspects of open data governance\n\n_Source_ : 2024 United Nations E-Government Survey.\n\n_Note_ : The availability of OGD in machine-readable formats is indicated by the lighter shades of the same colours in the regional performance graphs.\n\n\\* The availability of OGD in real time for both machine-readable and non-machine-readable data sets.\n\nEDGI 2024 levels\n\nVery high EGDI High EGDI Middle EGDI\n\n### Very hig High EG Middle Low EG\n\n193 UN Member States\n\nAmericas\n\nAfrica\n\nAsia\n\nEurope\n\nOceania\n\n### Very high EGDI High EGDI Middle EGDI Low EGDI\n\nY N M t i Y N M t i M t i Y N M t i M t Y N M t i M t Y N\n\n**SIDS LDCs LLDCs LDC/LLDCs LDC/SIDS**\n\n11%\n\n68%\n\n14%\n\n11%\n\n24%\n\n35%\n\n68%\n\n73%\n\n19%\n\n5%\n\n13%\n\n62%\n\n16%\n\n11%\n\n18%\n\n31%\n\n51%\n\n67%\n\n18%\n\n11%\n\n16%\n\n47%\n\n19%\n\n31%\n\n31%\n\n34%\n\n53%\n\n81%\n\n41%\n\n6%\n\n13%\n\n63%\n\n19%\n\n25%\n\n25%\n\n25%\n\n50%\n\n75%\n\n19%\n\n6%\n\n38%\n\n38%\n\n63%\n\n63%\n\n25%\n\n13%\n\nThere are participatory budgeting mechanisms on national portals\n\nThere are OGD available on national government expenditures or budget\n\nOGD is available in real time\n\nInformation is available about competitions, hackathons, or events around the use of OGD\n\nThere is open data licensing for OGD\n\nPublic can request or propose new OGD data sets\n\nNational portals provide GIS or other geospatial data\n\nThere is an OGD portal\n\nYes Machine-readable No Yes Machine-readable No Yes Machine-readable No Yes Machine-readable No Yes Machine-readable No\n\nYes Machine-readable No Yes Machine-readable No Yes Machine-readable No Yes Machine-readable No\n\n**SIDS LDCs LLDCs LDC/LLDCs LDC/SIDS**","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"2024 UN E-GovErNmENt SUrvEy\n\n78\n\n**Chapter 2**\n\nE-participation in countries in special situations\n\nconsistent with global trends, e-participation values remain the lowest among the five OSI subindices for countries in special situations. This is a clear indication that these countries are facing broad challenges in engaging citizens through digital platforms, which is crucial for inclusive and participatory governance. Enhancing e-participation in these regions will require not only investments in technology but also significant cultural and policy shifts to encourage and facilitate citizen engagement. Over the years UN DESA implemented and continues to support countries with various capacity building initiatives in the areas of digital government, participation and accountability, innovation and delivery of public services that contribute to the achievement of sustainable development goal 16 ⁹ .\n\n_Open government data (OGD)_\n\nGroup averages indicating the proportion of countries in special situations that share information through dedicated OGD portals are comparable to the corresponding global average; between 63 and 81 per cent of the countries in the respective groups maintain OGD portals, and 50 to 68 per cent provide geospatial data. Most countries in these groups also share information about government expenditures, often in open formats. However, only 3 out of 10 countries in special situations allow members of the public to request or propose new open data sets, compared to about 5 in 10 countries at the global level. More than 70 per cent of lDcs, llDcs and SIDS have not yet adopted open data licensing, and events such as hackathons around open data use are rare. Most OGD are not available in real time, and fewer than 2 in 10 countries have participatory budgeting mechanisms (see figure 2.36).\n\nFigure 2.36 Percentage of countries in special situations with OGD portals and various aspects of open data governance\n\n_Source_ : 2024 United Nations E-Government Survey.\n\n_Note_ : The availability of OGD in machine-readable formats is indicated by the lighter shades of the same colours in the regional performance graphs.\n\n\\* The availability of OGD in real time for both machine-readable and non-machine-readable data sets.\n\nEDGI 2024 levels\n\nVery high EGDI High EGDI Middle EGDI\n\n### Very hig High EG Middle Low EG\n\n193 UN Member States\n\nAmericas\n\nAfrica\n\nAsia\n\nEurope\n\nOceania\n\n### Very high EGDI High EGDI Middle EGDI Low EGDI\n\nY N M t i Y N M t i M t i Y N M t i M t Y N M t i M t Y N\n\n**SIDS LDCs LLDCs LDC/LLDCs LDC/SIDS**\n\n11%\n\n68%\n\n14%\n\n11%\n\n24%\n\n35%\n\n68%\n\n73%\n\n19%\n\n5%\n\n13%\n\n62%\n\n16%\n\n11%\n\n18%\n\n31%\n\n51%\n\n67%\n\n18%\n\n11%\n\n16%\n\n47%\n\n19%\n\n31%\n\n31%\n\n34%\n\n53%\n\n81%\n\n41%\n\n6%\n\n13%\n\n63%\n\n19%\n\n25%\n\n25%\n\n25%\n\n50%\n\n75%\n\n19%\n\n6%\n\n38%\n\n38%\n\n63%\n\n63%\n\n25%\n\n13%\n\nThere are participatory budgeting mechanisms on national portals\n\nThere are OGD available on national government expenditures or budget\n\nOGD is available in real time\n\nInformation is available about competitions, hackathons, or events around the use of OGD\n\nThere is open data licensing for OGD\n\nPublic can request or propose new OGD data sets\n\nNational portals provide GIS or other geospatial data\n\nThere is an OGD portal\n\nYes Machine-readable No Yes Machine-readable No Yes Machine-readable No Yes Machine-readable No Yes Machine-readable No\n\nYes Machine-readable No Yes Machine-readable No Yes Machine-readable No Yes Machine-readable No\n\n**SIDS LDCs LLDCs LDC/LLDCs LDC/SIDS**","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/%28Web_version%29_E-Government_Survey_2024_1392024_p103.pdf"]},"response":{"errors":[],"extract_id":"extract_154082c7e942c687eeae313f77849a2d","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["2024 UN E-GovErNmENt SUrvEy\n\n78\n\n**Chapter 2**\n\nE-participation in countries in special situations\n\nconsistent with global trends, e-participation values remain the lowest among the five OSI subindices for countries in special situations. This is a clear indication that these countries are facing broad challenges in engaging citizens through digital platforms, which is crucial for inclusive and participatory governance. Enhancing e-participation in these regions will require not only investments in technology but also significant cultural and policy shifts to encourage and facilitate citizen engagement. Over the years UN DESA implemented and continues to support countries with various capacity building initiatives in the areas of digital government, participation and accountability, innovation and delivery of public services that contribute to the achievement of sustainable development goal 16 ⁹ .\n\n_Open government data (OGD)_\n\nGroup averages indicating the proportion of countries in special situations that share information through dedicated OGD portals are comparable to the corresponding global average; between 63 and 81 per cent of the countries in the respective groups maintain OGD portals, and 50 to 68 per cent provide geospatial data. Most countries in these groups also share information about government expenditures, often in open formats. However, only 3 out of 10 countries in special situations allow members of the public to request or propose new open data sets, compared to about 5 in 10 countries at the global level. More than 70 per cent of lDcs, llDcs and SIDS have not yet adopted open data licensing, and events such as hackathons around open data use are rare. Most OGD are not available in real time, and fewer than 2 in 10 countries have participatory budgeting mechanisms (see figure 2.36).\n\nFigure 2.36 Percentage of countries in special situations with OGD portals and various aspects of open data governance\n\n_Source_ : 2024 United Nations E-Government Survey.\n\n_Note_ : The availability of OGD in machine-readable formats is indicated by the lighter shades of the same colours in the regional performance graphs.\n\n\\* The availability of OGD in real time for both machine-readable and non-machine-readable data sets.\n\nEDGI 2024 levels\n\nVery high EGDI High EGDI Middle EGDI\n\n### Very hig High EG Middle Low EG\n\n193 UN Member States\n\nAmericas\n\nAfrica\n\nAsia\n\nEurope\n\nOceania\n\n### Very high EGDI High EGDI Middle EGDI Low EGDI\n\nY N M t i Y N M t i M t i Y N M t i M t Y N M t i M t Y N\n\n**SIDS LDCs LLDCs LDC/LLDCs LDC/SIDS**\n\n11%\n\n68%\n\n14%\n\n11%\n\n24%\n\n35%\n\n68%\n\n73%\n\n19%\n\n5%\n\n13%\n\n62%\n\n16%\n\n11%\n\n18%\n\n31%\n\n51%\n\n67%\n\n18%\n\n11%\n\n16%\n\n47%\n\n19%\n\n31%\n\n31%\n\n34%\n\n53%\n\n81%\n\n41%\n\n6%\n\n13%\n\n63%\n\n19%\n\n25%\n\n25%\n\n25%\n\n50%\n\n75%\n\n19%\n\n6%\n\n38%\n\n38%\n\n63%\n\n63%\n\n25%\n\n13%\n\nThere are participatory budgeting mechanisms on national portals\n\nThere are OGD available on national government expenditures or budget\n\nOGD is available in real time\n\nInformation is available about competitions, hackathons, or events around the use of OGD\n\nThere is open data licensing for OGD\n\nPublic can request or propose new OGD data sets\n\nNational portals provide GIS or other geospatial data\n\nThere is an OGD portal\n\nYes Machine-readable No Yes Machine-readable No Yes Machine-readable No Yes Machine-readable No Yes Machine-readable No\n\nYes Machine-readable No Yes Machine-readable No Yes Machine-readable No Yes Machine-readable No\n\n**SIDS LDCs LLDCs LDC/LLDCs LDC/SIDS**"],"full_content":"2024 UN E-GovErNmENt SUrvEy\n\n78\n\n**Chapter 2**\n\nE-participation in countries in special situations\n\nconsistent with global trends, e-participation values remain the lowest among the five OSI subindices for countries in special situations. This is a clear indication that these countries are facing broad challenges in engaging citizens through digital platforms, which is crucial for inclusive and participatory governance. Enhancing e-participation in these regions will require not only investments in technology but also significant cultural and policy shifts to encourage and facilitate citizen engagement. Over the years UN DESA implemented and continues to support countries with various capacity building initiatives in the areas of digital government, participation and accountability, innovation and delivery of public services that contribute to the achievement of sustainable development goal 16 ⁹ .\n\n_Open government data (OGD)_\n\nGroup averages indicating the proportion of countries in special situations that share information through dedicated OGD portals are comparable to the corresponding global average; between 63 and 81 per cent of the countries in the respective groups maintain OGD portals, and 50 to 68 per cent provide geospatial data. Most countries in these groups also share information about government expenditures, often in open formats. However, only 3 out of 10 countries in special situations allow members of the public to request or propose new open data sets, compared to about 5 in 10 countries at the global level. More than 70 per cent of lDcs, llDcs and SIDS have not yet adopted open data licensing, and events such as hackathons around open data use are rare. Most OGD are not available in real time, and fewer than 2 in 10 countries have participatory budgeting mechanisms (see figure 2.36).\n\nFigure 2.36 Percentage of countries in special situations with OGD portals and various aspects of open data governance\n\n_Source_ : 2024 United Nations E-Government Survey.\n\n_Note_ : The availability of OGD in machine-readable formats is indicated by the lighter shades of the same colours in the regional performance graphs.\n\n\\* The availability of OGD in real time for both machine-readable and non-machine-readable data sets.\n\nEDGI 2024 levels\n\nVery high EGDI High EGDI Middle EGDI\n\n### Very hig High EG Middle Low EG\n\n193 UN Member States\n\nAmericas\n\nAfrica\n\nAsia\n\nEurope\n\nOceania\n\n### Very high EGDI High EGDI Middle EGDI Low EGDI\n\nY N M t i Y N M t i M t i Y N M t i M t Y N M t i M t Y N\n\n**SIDS LDCs LLDCs LDC/LLDCs LDC/SIDS**\n\n11%\n\n68%\n\n14%\n\n11%\n\n24%\n\n35%\n\n68%\n\n73%\n\n19%\n\n5%\n\n13%\n\n62%\n\n16%\n\n11%\n\n18%\n\n31%\n\n51%\n\n67%\n\n18%\n\n11%\n\n16%\n\n47%\n\n19%\n\n31%\n\n31%\n\n34%\n\n53%\n\n81%\n\n41%\n\n6%\n\n13%\n\n63%\n\n19%\n\n25%\n\n25%\n\n25%\n\n50%\n\n75%\n\n19%\n\n6%\n\n38%\n\n38%\n\n63%\n\n63%\n\n25%\n\n13%\n\nThere are participatory budgeting mechanisms on national portals\n\nThere are OGD available on national government expenditures or budget\n\nOGD is available in real time\n\nInformation is available about competitions, hackathons, or events around the use of OGD\n\nThere is open data licensing for OGD\n\nPublic can request or propose new OGD data sets\n\nNational portals provide GIS or other geospatial data\n\nThere is an OGD portal\n\nYes Machine-readable No Yes Machine-readable No Yes Machine-readable No Yes Machine-readable No Yes Machine-readable No\n\nYes Machine-readable No Yes Machine-readable No Yes Machine-readable No Yes Machine-readable No\n\n**SIDS LDCs LLDCs LDC/LLDCs LDC/SIDS**","publish_date":null,"title":"Very high EGDI High EGDI Middle EGDI Low EGDI Very high EGDI Very high EGDI High EGDI High EGDI Middle","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/%28Web_version%29_E-Government_Survey_2024_1392024_p103.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/%28Web_version%29_E-Government_Survey_2024_1392024_p103.pdf"},"request":{"example_id":"chart/(Web_version)_E-Government_Survey_2024_1392024_p103","product_type":"parse","source_file_path":"(Web_version)_E-Government_Survey_2024_1392024_p103.pdf"},"started_at":"2026-11-04T19:45:42.169185"} +{"completed_at":"2026-11-04T19:45:45.702992","latency_in_ms":2179,"output":{"example_id":"chart/(Web_version)_E-Government_Survey_2024_1392024_p170","grounded_pages":[],"layout_pages":[],"markdown":"145\n\n**Chapter 4**\n\nChapter 4 • LoCaL e-Government DeveLopment\n\nof the assessed portals, signaling an area in need of improvement. The availability of municipality budget information has improved but remains suboptimal at just under 58 per cent. More than two thirds of the city portals assessed now provide access to sports and cultural information and resources. The Government of Singapore ¹² offers a range of online services to support immigrants, fostering community integration and cultural acclimatization. The portal includes resources such as Your _Neighbourly Welcome Guide_ , which offers new residents insights into local culture and encourages community bonding by sharing stories from established residents. The suite of available programmes includes home visits, community learning activities, and the Singapore citizenship Journey, culminating in a formal citizenship ceremony. Additionally, the portal features forums for residents to share opinions and facilitates annual meetings, allowing new immigrants to actively contribute to and benefit from community activities.\n\nFigure 4.7 Implementation of content provision indicators in city portals: sectoral information\n\nFigure 4.8 Implementation of content provision indicators in city portals: addressing everyday needs\n\n35%\n\n48%\n\n50%\n\n58%\n\n67%\n\n70%\n\n74%\n\n76%\n\n30%\n\n45%\n\n49%\n\n54%\n\n63%\n\n68%\n\n70%\n\n74%\n\n0% 10% 20% 30% 40% 50% 60% 70% 80%\n\nServices in partnership with civil society\n\nHelp desk access\n\nInformation for vulnerable groups\n\nMunicipality budget information\n\nSports and cultural information\n\nInformation about provided services\n\nMunicipality information\n\nMunicipality contact details\n\n2022 2024\n\n46%\n\n46%\n\n53%\n\n54%\n\n61%\n\n63%\n\n38%\n\n39%\n\n53%\n\n52%\n\n55%\n\n56%\n\n0% 10% 20% 30% 40% 50% 60% 70%\n\nJustice information\n\nEmployment information\n\nEducation information\n\nSocial welfare information\n\nEnvironmental information\n\nHealth information\n\n2022 2024","pages":[{"line_numbers":[],"markdown":"145\n\n**Chapter 4**\n\nChapter 4 • LoCaL e-Government DeveLopment\n\nof the assessed portals, signaling an area in need of improvement. The availability of municipality budget information has improved but remains suboptimal at just under 58 per cent. More than two thirds of the city portals assessed now provide access to sports and cultural information and resources. The Government of Singapore ¹² offers a range of online services to support immigrants, fostering community integration and cultural acclimatization. The portal includes resources such as Your _Neighbourly Welcome Guide_ , which offers new residents insights into local culture and encourages community bonding by sharing stories from established residents. The suite of available programmes includes home visits, community learning activities, and the Singapore citizenship Journey, culminating in a formal citizenship ceremony. Additionally, the portal features forums for residents to share opinions and facilitates annual meetings, allowing new immigrants to actively contribute to and benefit from community activities.\n\nFigure 4.7 Implementation of content provision indicators in city portals: sectoral information\n\nFigure 4.8 Implementation of content provision indicators in city portals: addressing everyday needs\n\n35%\n\n48%\n\n50%\n\n58%\n\n67%\n\n70%\n\n74%\n\n76%\n\n30%\n\n45%\n\n49%\n\n54%\n\n63%\n\n68%\n\n70%\n\n74%\n\n0% 10% 20% 30% 40% 50% 60% 70% 80%\n\nServices in partnership with civil society\n\nHelp desk access\n\nInformation for vulnerable groups\n\nMunicipality budget information\n\nSports and cultural information\n\nInformation about provided services\n\nMunicipality information\n\nMunicipality contact details\n\n2022 2024\n\n46%\n\n46%\n\n53%\n\n54%\n\n61%\n\n63%\n\n38%\n\n39%\n\n53%\n\n52%\n\n55%\n\n56%\n\n0% 10% 20% 30% 40% 50% 60% 70%\n\nJustice information\n\nEmployment information\n\nEducation information\n\nSocial welfare information\n\nEnvironmental information\n\nHealth information\n\n2022 2024","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"145\n\n**Chapter 4**\n\nChapter 4 • LoCaL e-Government DeveLopment\n\nof the assessed portals, signaling an area in need of improvement. The availability of municipality budget information has improved but remains suboptimal at just under 58 per cent. More than two thirds of the city portals assessed now provide access to sports and cultural information and resources. The Government of Singapore ¹² offers a range of online services to support immigrants, fostering community integration and cultural acclimatization. The portal includes resources such as Your _Neighbourly Welcome Guide_ , which offers new residents insights into local culture and encourages community bonding by sharing stories from established residents. The suite of available programmes includes home visits, community learning activities, and the Singapore citizenship Journey, culminating in a formal citizenship ceremony. Additionally, the portal features forums for residents to share opinions and facilitates annual meetings, allowing new immigrants to actively contribute to and benefit from community activities.\n\nFigure 4.7 Implementation of content provision indicators in city portals: sectoral information\n\nFigure 4.8 Implementation of content provision indicators in city portals: addressing everyday needs\n\n35%\n\n48%\n\n50%\n\n58%\n\n67%\n\n70%\n\n74%\n\n76%\n\n30%\n\n45%\n\n49%\n\n54%\n\n63%\n\n68%\n\n70%\n\n74%\n\n0% 10% 20% 30% 40% 50% 60% 70% 80%\n\nServices in partnership with civil society\n\nHelp desk access\n\nInformation for vulnerable groups\n\nMunicipality budget information\n\nSports and cultural information\n\nInformation about provided services\n\nMunicipality information\n\nMunicipality contact details\n\n2022 2024\n\n46%\n\n46%\n\n53%\n\n54%\n\n61%\n\n63%\n\n38%\n\n39%\n\n53%\n\n52%\n\n55%\n\n56%\n\n0% 10% 20% 30% 40% 50% 60% 70%\n\nJustice information\n\nEmployment information\n\nEducation information\n\nSocial welfare information\n\nEnvironmental information\n\nHealth information\n\n2022 2024","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/%28Web_version%29_E-Government_Survey_2024_1392024_p170.pdf"]},"response":{"errors":[],"extract_id":"extract_ba022159a1870d9893eee7cc89af86df","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["145\n\n**Chapter 4**\n\nChapter 4 • LoCaL e-Government DeveLopment\n\nof the assessed portals, signaling an area in need of improvement. The availability of municipality budget information has improved but remains suboptimal at just under 58 per cent. More than two thirds of the city portals assessed now provide access to sports and cultural information and resources. The Government of Singapore ¹² offers a range of online services to support immigrants, fostering community integration and cultural acclimatization. The portal includes resources such as Your _Neighbourly Welcome Guide_ , which offers new residents insights into local culture and encourages community bonding by sharing stories from established residents. The suite of available programmes includes home visits, community learning activities, and the Singapore citizenship Journey, culminating in a formal citizenship ceremony. Additionally, the portal features forums for residents to share opinions and facilitates annual meetings, allowing new immigrants to actively contribute to and benefit from community activities.\n\nFigure 4.7 Implementation of content provision indicators in city portals: sectoral information\n\nFigure 4.8 Implementation of content provision indicators in city portals: addressing everyday needs\n\n35%\n\n48%\n\n50%\n\n58%\n\n67%\n\n70%\n\n74%\n\n76%\n\n30%\n\n45%\n\n49%\n\n54%\n\n63%\n\n68%\n\n70%\n\n74%\n\n0% 10% 20% 30% 40% 50% 60% 70% 80%\n\nServices in partnership with civil society\n\nHelp desk access\n\nInformation for vulnerable groups\n\nMunicipality budget information\n\nSports and cultural information\n\nInformation about provided services\n\nMunicipality information\n\nMunicipality contact details\n\n2022 2024\n\n46%\n\n46%\n\n53%\n\n54%\n\n61%\n\n63%\n\n38%\n\n39%\n\n53%\n\n52%\n\n55%\n\n56%\n\n0% 10% 20% 30% 40% 50% 60% 70%\n\nJustice information\n\nEmployment information\n\nEducation information\n\nSocial welfare information\n\nEnvironmental information\n\nHealth information\n\n2022 2024"],"full_content":"145\n\n**Chapter 4**\n\nChapter 4 • LoCaL e-Government DeveLopment\n\nof the assessed portals, signaling an area in need of improvement. The availability of municipality budget information has improved but remains suboptimal at just under 58 per cent. More than two thirds of the city portals assessed now provide access to sports and cultural information and resources. The Government of Singapore ¹² offers a range of online services to support immigrants, fostering community integration and cultural acclimatization. The portal includes resources such as Your _Neighbourly Welcome Guide_ , which offers new residents insights into local culture and encourages community bonding by sharing stories from established residents. The suite of available programmes includes home visits, community learning activities, and the Singapore citizenship Journey, culminating in a formal citizenship ceremony. Additionally, the portal features forums for residents to share opinions and facilitates annual meetings, allowing new immigrants to actively contribute to and benefit from community activities.\n\nFigure 4.7 Implementation of content provision indicators in city portals: sectoral information\n\nFigure 4.8 Implementation of content provision indicators in city portals: addressing everyday needs\n\n35%\n\n48%\n\n50%\n\n58%\n\n67%\n\n70%\n\n74%\n\n76%\n\n30%\n\n45%\n\n49%\n\n54%\n\n63%\n\n68%\n\n70%\n\n74%\n\n0% 10% 20% 30% 40% 50% 60% 70% 80%\n\nServices in partnership with civil society\n\nHelp desk access\n\nInformation for vulnerable groups\n\nMunicipality budget information\n\nSports and cultural information\n\nInformation about provided services\n\nMunicipality information\n\nMunicipality contact details\n\n2022 2024\n\n46%\n\n46%\n\n53%\n\n54%\n\n61%\n\n63%\n\n38%\n\n39%\n\n53%\n\n52%\n\n55%\n\n56%\n\n0% 10% 20% 30% 40% 50% 60% 70%\n\nJustice information\n\nEmployment information\n\nEducation information\n\nSocial welfare information\n\nEnvironmental information\n\nHealth information\n\n2022 2024","publish_date":null,"title":"Figure 4.7 Implementation of content provision indicators in city portals: sectoral information of the assessed portals, signaling an area in","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/%28Web_version%29_E-Government_Survey_2024_1392024_p170.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/%28Web_version%29_E-Government_Survey_2024_1392024_p170.pdf"},"request":{"example_id":"chart/(Web_version)_E-Government_Survey_2024_1392024_p170","product_type":"parse","source_file_path":"(Web_version)_E-Government_Survey_2024_1392024_p170.pdf"},"started_at":"2026-11-04T19:45:43.523039"} +{"completed_at":"2026-11-04T19:45:48.117817","latency_in_ms":2649,"output":{"example_id":"chart/(Web_version)_E-Government_Survey_2024_1392024_p171","grounded_pages":[],"layout_pages":[],"markdown":"2024 UN E-GovErNmENt SUrvEy\n\n146\n\n**Chapter 4**\n\nThe 2024 analysis related to public procurement reveals that 59 per cent of the city portals assessed share procurement announcements – a notable increase from 53 per cent in 2022. There has been a more modest rise, from 40 to 42 per cent, in the publication of procurement results. The adoption of e-procurement platforms has seen significant growth, rising from 36 per cent in 2022 to 42 per cent in 2024. Those cities that utilize national portals are awarded a point here as long as they link to a procurement portal with clear guidelines. For example, the Dublin city portal, in its Doing Business with the council section, ¹³ clearly explains public procurement guidelines and provides a link to the national portal (etenders.gov.ie). conversely, the Tokyo e-procurement portal ¹⁴ serves as a centralized platform for managing procurement processes electronically, facilitating transparent and efficient procurement activities for the city’s government agencies and vendors.\n\n_Services provision_\n\nAlthough services provision continues to have the lowest level of compliance among the six criteria assessed (see figure 4.5), the data reveal a notable increase in the availability of online services since 2022, with several indicators showing improvement (see figure 4.10). Among the most frequently met indicators are those relating to waste and recycling information and public transportation services, followed by online payment options for fines, utility bills and services.\n\nIt is encouraging to see the increased presence of waste and recycling information on city portals, as this signals a growing commitment to environmental sustainability and the involvement of residents in waste management efforts. cities recognize the importance of offering people access to resources on waste reduction, recycling and reuse to encourage environmentally friendly behaviours and promote widespread engagement in tackling environmental challenges. These trends are promising and have the potential to contribute meaningfully to environmental sustainability. The Berlin city portal ¹⁵ offers comprehensive guidance on waste management and disposal in the city, providing information on bulky waste disposal, clothing donations, e-waste recycling, and household waste segregation. Through its portal, the city government emphasizes the importance of environmental responsibility, offering information on recycling centres, donation organizations, and pollutant collection points to facilitate proper disposal practices and promote sustainability within the community.\n\nFigure 4.9 Procurement information on city portals\n\n59%\n\n42%\n\n42%\n\n53%\n\n40%\n\n36%\n\n0% 10% 20% 30% 40% 50% 60% 70%\n\nProcurement announcements\n\nProcurement results\n\nProcurement platform\n\n2022 2024","pages":[{"line_numbers":[],"markdown":"2024 UN E-GovErNmENt SUrvEy\n\n146\n\n**Chapter 4**\n\nThe 2024 analysis related to public procurement reveals that 59 per cent of the city portals assessed share procurement announcements – a notable increase from 53 per cent in 2022. There has been a more modest rise, from 40 to 42 per cent, in the publication of procurement results. The adoption of e-procurement platforms has seen significant growth, rising from 36 per cent in 2022 to 42 per cent in 2024. Those cities that utilize national portals are awarded a point here as long as they link to a procurement portal with clear guidelines. For example, the Dublin city portal, in its Doing Business with the council section, ¹³ clearly explains public procurement guidelines and provides a link to the national portal (etenders.gov.ie). conversely, the Tokyo e-procurement portal ¹⁴ serves as a centralized platform for managing procurement processes electronically, facilitating transparent and efficient procurement activities for the city’s government agencies and vendors.\n\n_Services provision_\n\nAlthough services provision continues to have the lowest level of compliance among the six criteria assessed (see figure 4.5), the data reveal a notable increase in the availability of online services since 2022, with several indicators showing improvement (see figure 4.10). Among the most frequently met indicators are those relating to waste and recycling information and public transportation services, followed by online payment options for fines, utility bills and services.\n\nIt is encouraging to see the increased presence of waste and recycling information on city portals, as this signals a growing commitment to environmental sustainability and the involvement of residents in waste management efforts. cities recognize the importance of offering people access to resources on waste reduction, recycling and reuse to encourage environmentally friendly behaviours and promote widespread engagement in tackling environmental challenges. These trends are promising and have the potential to contribute meaningfully to environmental sustainability. The Berlin city portal ¹⁵ offers comprehensive guidance on waste management and disposal in the city, providing information on bulky waste disposal, clothing donations, e-waste recycling, and household waste segregation. Through its portal, the city government emphasizes the importance of environmental responsibility, offering information on recycling centres, donation organizations, and pollutant collection points to facilitate proper disposal practices and promote sustainability within the community.\n\nFigure 4.9 Procurement information on city portals\n\n59%\n\n42%\n\n42%\n\n53%\n\n40%\n\n36%\n\n0% 10% 20% 30% 40% 50% 60% 70%\n\nProcurement announcements\n\nProcurement results\n\nProcurement platform\n\n2022 2024","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"2024 UN E-GovErNmENt SUrvEy\n\n146\n\n**Chapter 4**\n\nThe 2024 analysis related to public procurement reveals that 59 per cent of the city portals assessed share procurement announcements – a notable increase from 53 per cent in 2022. There has been a more modest rise, from 40 to 42 per cent, in the publication of procurement results. The adoption of e-procurement platforms has seen significant growth, rising from 36 per cent in 2022 to 42 per cent in 2024. Those cities that utilize national portals are awarded a point here as long as they link to a procurement portal with clear guidelines. For example, the Dublin city portal, in its Doing Business with the council section, ¹³ clearly explains public procurement guidelines and provides a link to the national portal (etenders.gov.ie). conversely, the Tokyo e-procurement portal ¹⁴ serves as a centralized platform for managing procurement processes electronically, facilitating transparent and efficient procurement activities for the city’s government agencies and vendors.\n\n_Services provision_\n\nAlthough services provision continues to have the lowest level of compliance among the six criteria assessed (see figure 4.5), the data reveal a notable increase in the availability of online services since 2022, with several indicators showing improvement (see figure 4.10). Among the most frequently met indicators are those relating to waste and recycling information and public transportation services, followed by online payment options for fines, utility bills and services.\n\nIt is encouraging to see the increased presence of waste and recycling information on city portals, as this signals a growing commitment to environmental sustainability and the involvement of residents in waste management efforts. cities recognize the importance of offering people access to resources on waste reduction, recycling and reuse to encourage environmentally friendly behaviours and promote widespread engagement in tackling environmental challenges. These trends are promising and have the potential to contribute meaningfully to environmental sustainability. The Berlin city portal ¹⁵ offers comprehensive guidance on waste management and disposal in the city, providing information on bulky waste disposal, clothing donations, e-waste recycling, and household waste segregation. Through its portal, the city government emphasizes the importance of environmental responsibility, offering information on recycling centres, donation organizations, and pollutant collection points to facilitate proper disposal practices and promote sustainability within the community.\n\nFigure 4.9 Procurement information on city portals\n\n59%\n\n42%\n\n42%\n\n53%\n\n40%\n\n36%\n\n0% 10% 20% 30% 40% 50% 60% 70%\n\nProcurement announcements\n\nProcurement results\n\nProcurement platform\n\n2022 2024","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/%28Web_version%29_E-Government_Survey_2024_1392024_p171.pdf"]},"response":{"errors":[],"extract_id":"extract_15ade1e6d6128129c85e0006037ab36f","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["2024 UN E-GovErNmENt SUrvEy\n\n146\n\n**Chapter 4**\n\nThe 2024 analysis related to public procurement reveals that 59 per cent of the city portals assessed share procurement announcements – a notable increase from 53 per cent in 2022. There has been a more modest rise, from 40 to 42 per cent, in the publication of procurement results. The adoption of e-procurement platforms has seen significant growth, rising from 36 per cent in 2022 to 42 per cent in 2024. Those cities that utilize national portals are awarded a point here as long as they link to a procurement portal with clear guidelines. For example, the Dublin city portal, in its Doing Business with the council section, ¹³ clearly explains public procurement guidelines and provides a link to the national portal (etenders.gov.ie). conversely, the Tokyo e-procurement portal ¹⁴ serves as a centralized platform for managing procurement processes electronically, facilitating transparent and efficient procurement activities for the city’s government agencies and vendors.\n\n_Services provision_\n\nAlthough services provision continues to have the lowest level of compliance among the six criteria assessed (see figure 4.5), the data reveal a notable increase in the availability of online services since 2022, with several indicators showing improvement (see figure 4.10). Among the most frequently met indicators are those relating to waste and recycling information and public transportation services, followed by online payment options for fines, utility bills and services.\n\nIt is encouraging to see the increased presence of waste and recycling information on city portals, as this signals a growing commitment to environmental sustainability and the involvement of residents in waste management efforts. cities recognize the importance of offering people access to resources on waste reduction, recycling and reuse to encourage environmentally friendly behaviours and promote widespread engagement in tackling environmental challenges. These trends are promising and have the potential to contribute meaningfully to environmental sustainability. The Berlin city portal ¹⁵ offers comprehensive guidance on waste management and disposal in the city, providing information on bulky waste disposal, clothing donations, e-waste recycling, and household waste segregation. Through its portal, the city government emphasizes the importance of environmental responsibility, offering information on recycling centres, donation organizations, and pollutant collection points to facilitate proper disposal practices and promote sustainability within the community.\n\nFigure 4.9 Procurement information on city portals\n\n59%\n\n42%\n\n42%\n\n53%\n\n40%\n\n36%\n\n0% 10% 20% 30% 40% 50% 60% 70%\n\nProcurement announcements\n\nProcurement results\n\nProcurement platform\n\n2022 2024"],"full_content":"2024 UN E-GovErNmENt SUrvEy\n\n146\n\n**Chapter 4**\n\nThe 2024 analysis related to public procurement reveals that 59 per cent of the city portals assessed share procurement announcements – a notable increase from 53 per cent in 2022. There has been a more modest rise, from 40 to 42 per cent, in the publication of procurement results. The adoption of e-procurement platforms has seen significant growth, rising from 36 per cent in 2022 to 42 per cent in 2024. Those cities that utilize national portals are awarded a point here as long as they link to a procurement portal with clear guidelines. For example, the Dublin city portal, in its Doing Business with the council section, ¹³ clearly explains public procurement guidelines and provides a link to the national portal (etenders.gov.ie). conversely, the Tokyo e-procurement portal ¹⁴ serves as a centralized platform for managing procurement processes electronically, facilitating transparent and efficient procurement activities for the city’s government agencies and vendors.\n\n_Services provision_\n\nAlthough services provision continues to have the lowest level of compliance among the six criteria assessed (see figure 4.5), the data reveal a notable increase in the availability of online services since 2022, with several indicators showing improvement (see figure 4.10). Among the most frequently met indicators are those relating to waste and recycling information and public transportation services, followed by online payment options for fines, utility bills and services.\n\nIt is encouraging to see the increased presence of waste and recycling information on city portals, as this signals a growing commitment to environmental sustainability and the involvement of residents in waste management efforts. cities recognize the importance of offering people access to resources on waste reduction, recycling and reuse to encourage environmentally friendly behaviours and promote widespread engagement in tackling environmental challenges. These trends are promising and have the potential to contribute meaningfully to environmental sustainability. The Berlin city portal ¹⁵ offers comprehensive guidance on waste management and disposal in the city, providing information on bulky waste disposal, clothing donations, e-waste recycling, and household waste segregation. Through its portal, the city government emphasizes the importance of environmental responsibility, offering information on recycling centres, donation organizations, and pollutant collection points to facilitate proper disposal practices and promote sustainability within the community.\n\nFigure 4.9 Procurement information on city portals\n\n59%\n\n42%\n\n42%\n\n53%\n\n40%\n\n36%\n\n0% 10% 20% 30% 40% 50% 60% 70%\n\nProcurement announcements\n\nProcurement results\n\nProcurement platform\n\n2022 2024","publish_date":null,"title":"The 2024 analysis related to public procurement reveals that 59 per cent of the city portals assessed share procurement announcements","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/%28Web_version%29_E-Government_Survey_2024_1392024_p171.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/%28Web_version%29_E-Government_Survey_2024_1392024_p171.pdf"},"request":{"example_id":"chart/(Web_version)_E-Government_Survey_2024_1392024_p171","product_type":"parse","source_file_path":"(Web_version)_E-Government_Survey_2024_1392024_p171.pdf"},"started_at":"2026-11-04T19:45:45.467899"} +{"completed_at":"2026-11-04T19:45:48.868792","latency_in_ms":2024,"output":{"example_id":"chart/(Web_version)_E-Government_Survey_2024_1392024_p172","grounded_pages":[],"layout_pages":[],"markdown":"147\n\n**Chapter 4**\n\nChapter 4 • LoCaL e-Government DeveLopment\n\nFigure 4.10 Implementation of services provision indicators in city portals\n\nThe availability of public transportation information and services on city portals reflects a commitment to enhancing urban mobility and reducing traffic congestion. cities recognize the importance of providing residents with convenient access to public transportation timetables, route maps, fare information, and payment options. By offering these services online, cities aim to improve the overall efficiency and accessibility of public transportation systems, encourage the use of sustainable transportation modes, and reduce reliance on private vehicles.\n\n_Participation and engagement_\n\nAs illustrated in figure 4.11, compliance rates for most of the lOSI participation and engagement indicators increased between 2022 and 2024. There has been notable progress made in feedback/ complaint submission, information on the public meetings of the municipal council, and participatory budgeting. Other indicators, such as e-voting, reporting incidents in public spaces, participatory land use planning, online deliberation processes, feedback on consultation processes, and announcing upcoming e-participation activities, have seen slight to moderate increases or have remained stable over the two-year period.\n\n12%\n\n17%\n\n19%\n\n20%\n\n21%\n\n22%\n\n24%\n\n27%\n\n27%\n\n31%\n\n31%\n\n38%\n\n38%\n\n40%\n\n41%\n\n42%\n\n42%\n\n44%\n\n46%\n\n50%\n\n51%\n\n10%\n\n13%\n\n16%\n\n17%\n\n20%\n\n19%\n\n21%\n\n22%\n\n21%\n\n24%\n\n27%\n\n33%\n\n32%\n\n37%\n\n35%\n\n33%\n\n36%\n\n41%\n\n44%\n\n47%\n\n48%\n\n0% 10% 20% 30% 40% 50% 60%\n\nOnline vehicle registration\n\nOnline driver's licence\n\nOnline land title registration\n\nOnline residentship\n\nOnline death certifcate\n\nAddress change notifcation\n\nOnline marriage certifcate\n\nOnline environment-related permit\n\nOnline birth certifcate\n\nOnline police declaration\n\nElectricity/gas payment\n\nOnline business licence\n\nWater payment\n\nOnline vacancies\n\nOnline building permit\n\nOnline tax declaration\n\nProcurement platform\n\nEvidence of mobile phone apps\n\nOnline fee payment\n\nPublic tranportation information\n\nWaste and recycling information\n\n2022 2024","pages":[{"line_numbers":[],"markdown":"147\n\n**Chapter 4**\n\nChapter 4 • LoCaL e-Government DeveLopment\n\nFigure 4.10 Implementation of services provision indicators in city portals\n\nThe availability of public transportation information and services on city portals reflects a commitment to enhancing urban mobility and reducing traffic congestion. cities recognize the importance of providing residents with convenient access to public transportation timetables, route maps, fare information, and payment options. By offering these services online, cities aim to improve the overall efficiency and accessibility of public transportation systems, encourage the use of sustainable transportation modes, and reduce reliance on private vehicles.\n\n_Participation and engagement_\n\nAs illustrated in figure 4.11, compliance rates for most of the lOSI participation and engagement indicators increased between 2022 and 2024. There has been notable progress made in feedback/ complaint submission, information on the public meetings of the municipal council, and participatory budgeting. Other indicators, such as e-voting, reporting incidents in public spaces, participatory land use planning, online deliberation processes, feedback on consultation processes, and announcing upcoming e-participation activities, have seen slight to moderate increases or have remained stable over the two-year period.\n\n12%\n\n17%\n\n19%\n\n20%\n\n21%\n\n22%\n\n24%\n\n27%\n\n27%\n\n31%\n\n31%\n\n38%\n\n38%\n\n40%\n\n41%\n\n42%\n\n42%\n\n44%\n\n46%\n\n50%\n\n51%\n\n10%\n\n13%\n\n16%\n\n17%\n\n20%\n\n19%\n\n21%\n\n22%\n\n21%\n\n24%\n\n27%\n\n33%\n\n32%\n\n37%\n\n35%\n\n33%\n\n36%\n\n41%\n\n44%\n\n47%\n\n48%\n\n0% 10% 20% 30% 40% 50% 60%\n\nOnline vehicle registration\n\nOnline driver's licence\n\nOnline land title registration\n\nOnline residentship\n\nOnline death certifcate\n\nAddress change notifcation\n\nOnline marriage certifcate\n\nOnline environment-related permit\n\nOnline birth certifcate\n\nOnline police declaration\n\nElectricity/gas payment\n\nOnline business licence\n\nWater payment\n\nOnline vacancies\n\nOnline building permit\n\nOnline tax declaration\n\nProcurement platform\n\nEvidence of mobile phone apps\n\nOnline fee payment\n\nPublic tranportation information\n\nWaste and recycling information\n\n2022 2024","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"147\n\n**Chapter 4**\n\nChapter 4 • LoCaL e-Government DeveLopment\n\nFigure 4.10 Implementation of services provision indicators in city portals\n\nThe availability of public transportation information and services on city portals reflects a commitment to enhancing urban mobility and reducing traffic congestion. cities recognize the importance of providing residents with convenient access to public transportation timetables, route maps, fare information, and payment options. By offering these services online, cities aim to improve the overall efficiency and accessibility of public transportation systems, encourage the use of sustainable transportation modes, and reduce reliance on private vehicles.\n\n_Participation and engagement_\n\nAs illustrated in figure 4.11, compliance rates for most of the lOSI participation and engagement indicators increased between 2022 and 2024. There has been notable progress made in feedback/ complaint submission, information on the public meetings of the municipal council, and participatory budgeting. Other indicators, such as e-voting, reporting incidents in public spaces, participatory land use planning, online deliberation processes, feedback on consultation processes, and announcing upcoming e-participation activities, have seen slight to moderate increases or have remained stable over the two-year period.\n\n12%\n\n17%\n\n19%\n\n20%\n\n21%\n\n22%\n\n24%\n\n27%\n\n27%\n\n31%\n\n31%\n\n38%\n\n38%\n\n40%\n\n41%\n\n42%\n\n42%\n\n44%\n\n46%\n\n50%\n\n51%\n\n10%\n\n13%\n\n16%\n\n17%\n\n20%\n\n19%\n\n21%\n\n22%\n\n21%\n\n24%\n\n27%\n\n33%\n\n32%\n\n37%\n\n35%\n\n33%\n\n36%\n\n41%\n\n44%\n\n47%\n\n48%\n\n0% 10% 20% 30% 40% 50% 60%\n\nOnline vehicle registration\n\nOnline driver's licence\n\nOnline land title registration\n\nOnline residentship\n\nOnline death certifcate\n\nAddress change notifcation\n\nOnline marriage certifcate\n\nOnline environment-related permit\n\nOnline birth certifcate\n\nOnline police declaration\n\nElectricity/gas payment\n\nOnline business licence\n\nWater payment\n\nOnline vacancies\n\nOnline building permit\n\nOnline tax declaration\n\nProcurement platform\n\nEvidence of mobile phone apps\n\nOnline fee payment\n\nPublic tranportation information\n\nWaste and recycling information\n\n2022 2024","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/%28Web_version%29_E-Government_Survey_2024_1392024_p172.pdf"]},"response":{"errors":[],"extract_id":"extract_d6dce0cfa2a55764f8164b3c4eb54c35","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["147\n\n**Chapter 4**\n\nChapter 4 • LoCaL e-Government DeveLopment\n\nFigure 4.10 Implementation of services provision indicators in city portals\n\nThe availability of public transportation information and services on city portals reflects a commitment to enhancing urban mobility and reducing traffic congestion. cities recognize the importance of providing residents with convenient access to public transportation timetables, route maps, fare information, and payment options. By offering these services online, cities aim to improve the overall efficiency and accessibility of public transportation systems, encourage the use of sustainable transportation modes, and reduce reliance on private vehicles.\n\n_Participation and engagement_\n\nAs illustrated in figure 4.11, compliance rates for most of the lOSI participation and engagement indicators increased between 2022 and 2024. There has been notable progress made in feedback/ complaint submission, information on the public meetings of the municipal council, and participatory budgeting. Other indicators, such as e-voting, reporting incidents in public spaces, participatory land use planning, online deliberation processes, feedback on consultation processes, and announcing upcoming e-participation activities, have seen slight to moderate increases or have remained stable over the two-year period.\n\n12%\n\n17%\n\n19%\n\n20%\n\n21%\n\n22%\n\n24%\n\n27%\n\n27%\n\n31%\n\n31%\n\n38%\n\n38%\n\n40%\n\n41%\n\n42%\n\n42%\n\n44%\n\n46%\n\n50%\n\n51%\n\n10%\n\n13%\n\n16%\n\n17%\n\n20%\n\n19%\n\n21%\n\n22%\n\n21%\n\n24%\n\n27%\n\n33%\n\n32%\n\n37%\n\n35%\n\n33%\n\n36%\n\n41%\n\n44%\n\n47%\n\n48%\n\n0% 10% 20% 30% 40% 50% 60%\n\nOnline vehicle registration\n\nOnline driver's licence\n\nOnline land title registration\n\nOnline residentship\n\nOnline death certifcate\n\nAddress change notifcation\n\nOnline marriage certifcate\n\nOnline environment-related permit\n\nOnline birth certifcate\n\nOnline police declaration\n\nElectricity/gas payment\n\nOnline business licence\n\nWater payment\n\nOnline vacancies\n\nOnline building permit\n\nOnline tax declaration\n\nProcurement platform\n\nEvidence of mobile phone apps\n\nOnline fee payment\n\nPublic tranportation information\n\nWaste and recycling information\n\n2022 2024"],"full_content":"147\n\n**Chapter 4**\n\nChapter 4 • LoCaL e-Government DeveLopment\n\nFigure 4.10 Implementation of services provision indicators in city portals\n\nThe availability of public transportation information and services on city portals reflects a commitment to enhancing urban mobility and reducing traffic congestion. cities recognize the importance of providing residents with convenient access to public transportation timetables, route maps, fare information, and payment options. By offering these services online, cities aim to improve the overall efficiency and accessibility of public transportation systems, encourage the use of sustainable transportation modes, and reduce reliance on private vehicles.\n\n_Participation and engagement_\n\nAs illustrated in figure 4.11, compliance rates for most of the lOSI participation and engagement indicators increased between 2022 and 2024. There has been notable progress made in feedback/ complaint submission, information on the public meetings of the municipal council, and participatory budgeting. Other indicators, such as e-voting, reporting incidents in public spaces, participatory land use planning, online deliberation processes, feedback on consultation processes, and announcing upcoming e-participation activities, have seen slight to moderate increases or have remained stable over the two-year period.\n\n12%\n\n17%\n\n19%\n\n20%\n\n21%\n\n22%\n\n24%\n\n27%\n\n27%\n\n31%\n\n31%\n\n38%\n\n38%\n\n40%\n\n41%\n\n42%\n\n42%\n\n44%\n\n46%\n\n50%\n\n51%\n\n10%\n\n13%\n\n16%\n\n17%\n\n20%\n\n19%\n\n21%\n\n22%\n\n21%\n\n24%\n\n27%\n\n33%\n\n32%\n\n37%\n\n35%\n\n33%\n\n36%\n\n41%\n\n44%\n\n47%\n\n48%\n\n0% 10% 20% 30% 40% 50% 60%\n\nOnline vehicle registration\n\nOnline driver's licence\n\nOnline land title registration\n\nOnline residentship\n\nOnline death certifcate\n\nAddress change notifcation\n\nOnline marriage certifcate\n\nOnline environment-related permit\n\nOnline birth certifcate\n\nOnline police declaration\n\nElectricity/gas payment\n\nOnline business licence\n\nWater payment\n\nOnline vacancies\n\nOnline building permit\n\nOnline tax declaration\n\nProcurement platform\n\nEvidence of mobile phone apps\n\nOnline fee payment\n\nPublic tranportation information\n\nWaste and recycling information\n\n2022 2024","publish_date":null,"title":"The availability of public transportation information and services on city portals reflects a commitment to enhancing urban mobility and reducing","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/%28Web_version%29_E-Government_Survey_2024_1392024_p172.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/%28Web_version%29_E-Government_Survey_2024_1392024_p172.pdf"},"request":{"example_id":"chart/(Web_version)_E-Government_Survey_2024_1392024_p172","product_type":"parse","source_file_path":"(Web_version)_E-Government_Survey_2024_1392024_p172.pdf"},"started_at":"2026-11-04T19:45:46.844588"} +{"completed_at":"2026-11-04T19:31:05.404436","latency_in_ms":1032,"output":{"example_id":"chart/(Web_version)_E-Government_Survey_2024_1392024_p173","grounded_pages":[],"layout_pages":[],"markdown":"2024 UN E-GovErNmENt SUrvEy\n\n148\n\n**Chapter 4**\n\nFigure 4.11 Implementation of participation and engagement indicators in city portals\n\nMost city portals incorporate feedback mechanisms to promote and facilitate resident engagement. Malé, ¹⁶ the capital of the Maldives and its most populous city, pledges to process all duly completed feedback submissions within five business days. Rome allows residents to submit suggestions, comments, complaints, and even reports relating to local government activities or offices and is committed to responding within thirty days, identifying the relevant office and personnel overseeing the investigation and providing information on actions taken. The city of Dublin informs its residents of the avenues available to report various issues to designated service areas without requiring adherence to formal complaint procedures. Formal complaints directed towards the Dublin city council ¹⁷ can be lodged via post, email or online through the citizen Hub, with assurances of a formal acknowledgment within three working days and an expected response from the relevant department within twenty- one days. Dissatisfied individuals can escalate their complaint to the executive manager within fifteen working days and further to the Office of the Ombudsman if the problem persists.\n\nE-voting is not strictly limited to electoral processes, often extending to decision-making on various initiatives, projects or ideas within city portals. The availability and utility of this option varies. Some platforms showcase e-voting functionality for specific projects or competitions, with evidence of past engagement but limited current activity. Others require registration to access e-voting results or offer opportunities for online voting on specific topics. In some instances, e-voting is integrated into participatory budgeting schemes. While these examples are encouraging, widespread implementation remains uncommon, with sporadic instances of e-voting observed across different municipalities. Although some platforms provide clear avenues for citizens to participate in e-voting opportunities, the policy impact of these contributions remains unclear in certain cases. Overall, while e-voting services are present in various forms, they often represent a minority among the participation mechanisms available within city portals.\n\n_E-government literacy_\n\nIn the 2024 edition of the E-Government Survey, UN DESA introduces the concept of e-government literacy to measure the ability of residents – especially vulnerable groups – to utilize e-government services and take advantage of e-participation opportunities. In the present analysis, e-government literacy is based on the assessment of key features within government portals, including support services, privacy protection, digital identity authentication, and access to online resources, as these indicators are aligned with enhancing digital literacy and engagement across diverse domains. Although e-government literacy was not a specific category in 2022, many of its indicators were\n\n18%\n\n25%\n\n31%\n\n31%\n\n32%\n\n40%\n\n42%\n\n47%\n\n66%\n\n15%\n\n24%\n\n26%\n\n31%\n\n28%\n\n37%\n\n40%\n\n40%\n\n61%\n\n0% 10% 20% 30% 40% 50% 60% 70%\n\nE-voting\n\nFeedback about consultation processes\n\nParticipatory budgeting\n\nAnnouncement of upcoming e-participation activities\n\nParticipatory land use planning\n\nOnline deliberation processes\n\nReporting of incidents in public spaces\n\nInformation on the public meetings of the municipal council\n\nFeedback/complaint submission\n\n2022 2024","pages":[{"line_numbers":[],"markdown":"2024 UN E-GovErNmENt SUrvEy\n\n148\n\n**Chapter 4**\n\nFigure 4.11 Implementation of participation and engagement indicators in city portals\n\nMost city portals incorporate feedback mechanisms to promote and facilitate resident engagement. Malé, ¹⁶ the capital of the Maldives and its most populous city, pledges to process all duly completed feedback submissions within five business days. Rome allows residents to submit suggestions, comments, complaints, and even reports relating to local government activities or offices and is committed to responding within thirty days, identifying the relevant office and personnel overseeing the investigation and providing information on actions taken. The city of Dublin informs its residents of the avenues available to report various issues to designated service areas without requiring adherence to formal complaint procedures. Formal complaints directed towards the Dublin city council ¹⁷ can be lodged via post, email or online through the citizen Hub, with assurances of a formal acknowledgment within three working days and an expected response from the relevant department within twenty- one days. Dissatisfied individuals can escalate their complaint to the executive manager within fifteen working days and further to the Office of the Ombudsman if the problem persists.\n\nE-voting is not strictly limited to electoral processes, often extending to decision-making on various initiatives, projects or ideas within city portals. The availability and utility of this option varies. Some platforms showcase e-voting functionality for specific projects or competitions, with evidence of past engagement but limited current activity. Others require registration to access e-voting results or offer opportunities for online voting on specific topics. In some instances, e-voting is integrated into participatory budgeting schemes. While these examples are encouraging, widespread implementation remains uncommon, with sporadic instances of e-voting observed across different municipalities. Although some platforms provide clear avenues for citizens to participate in e-voting opportunities, the policy impact of these contributions remains unclear in certain cases. Overall, while e-voting services are present in various forms, they often represent a minority among the participation mechanisms available within city portals.\n\n_E-government literacy_\n\nIn the 2024 edition of the E-Government Survey, UN DESA introduces the concept of e-government literacy to measure the ability of residents – especially vulnerable groups – to utilize e-government services and take advantage of e-participation opportunities. In the present analysis, e-government literacy is based on the assessment of key features within government portals, including support services, privacy protection, digital identity authentication, and access to online resources, as these indicators are aligned with enhancing digital literacy and engagement across diverse domains. Although e-government literacy was not a specific category in 2022, many of its indicators were\n\n18%\n\n25%\n\n31%\n\n31%\n\n32%\n\n40%\n\n42%\n\n47%\n\n66%\n\n15%\n\n24%\n\n26%\n\n31%\n\n28%\n\n37%\n\n40%\n\n40%\n\n61%\n\n0% 10% 20% 30% 40% 50% 60% 70%\n\nE-voting\n\nFeedback about consultation processes\n\nParticipatory budgeting\n\nAnnouncement of upcoming e-participation activities\n\nParticipatory land use planning\n\nOnline deliberation processes\n\nReporting of incidents in public spaces\n\nInformation on the public meetings of the municipal council\n\nFeedback/complaint submission\n\n2022 2024","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"2024 UN E-GovErNmENt SUrvEy\n\n148\n\n**Chapter 4**\n\nFigure 4.11 Implementation of participation and engagement indicators in city portals\n\nMost city portals incorporate feedback mechanisms to promote and facilitate resident engagement. Malé, ¹⁶ the capital of the Maldives and its most populous city, pledges to process all duly completed feedback submissions within five business days. Rome allows residents to submit suggestions, comments, complaints, and even reports relating to local government activities or offices and is committed to responding within thirty days, identifying the relevant office and personnel overseeing the investigation and providing information on actions taken. The city of Dublin informs its residents of the avenues available to report various issues to designated service areas without requiring adherence to formal complaint procedures. Formal complaints directed towards the Dublin city council ¹⁷ can be lodged via post, email or online through the citizen Hub, with assurances of a formal acknowledgment within three working days and an expected response from the relevant department within twenty- one days. Dissatisfied individuals can escalate their complaint to the executive manager within fifteen working days and further to the Office of the Ombudsman if the problem persists.\n\nE-voting is not strictly limited to electoral processes, often extending to decision-making on various initiatives, projects or ideas within city portals. The availability and utility of this option varies. Some platforms showcase e-voting functionality for specific projects or competitions, with evidence of past engagement but limited current activity. Others require registration to access e-voting results or offer opportunities for online voting on specific topics. In some instances, e-voting is integrated into participatory budgeting schemes. While these examples are encouraging, widespread implementation remains uncommon, with sporadic instances of e-voting observed across different municipalities. Although some platforms provide clear avenues for citizens to participate in e-voting opportunities, the policy impact of these contributions remains unclear in certain cases. Overall, while e-voting services are present in various forms, they often represent a minority among the participation mechanisms available within city portals.\n\n_E-government literacy_\n\nIn the 2024 edition of the E-Government Survey, UN DESA introduces the concept of e-government literacy to measure the ability of residents – especially vulnerable groups – to utilize e-government services and take advantage of e-participation opportunities. In the present analysis, e-government literacy is based on the assessment of key features within government portals, including support services, privacy protection, digital identity authentication, and access to online resources, as these indicators are aligned with enhancing digital literacy and engagement across diverse domains. Although e-government literacy was not a specific category in 2022, many of its indicators were\n\n18%\n\n25%\n\n31%\n\n31%\n\n32%\n\n40%\n\n42%\n\n47%\n\n66%\n\n15%\n\n24%\n\n26%\n\n31%\n\n28%\n\n37%\n\n40%\n\n40%\n\n61%\n\n0% 10% 20% 30% 40% 50% 60% 70%\n\nE-voting\n\nFeedback about consultation processes\n\nParticipatory budgeting\n\nAnnouncement of upcoming e-participation activities\n\nParticipatory land use planning\n\nOnline deliberation processes\n\nReporting of incidents in public spaces\n\nInformation on the public meetings of the municipal council\n\nFeedback/complaint submission\n\n2022 2024","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/%28Web_version%29_E-Government_Survey_2024_1392024_p173.pdf"]},"response":{"errors":[],"extract_id":"extract_567745c17d398746ccde16ee5e562a90","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["2024 UN E-GovErNmENt SUrvEy\n\n148\n\n**Chapter 4**\n\nFigure 4.11 Implementation of participation and engagement indicators in city portals\n\nMost city portals incorporate feedback mechanisms to promote and facilitate resident engagement. Malé, ¹⁶ the capital of the Maldives and its most populous city, pledges to process all duly completed feedback submissions within five business days. Rome allows residents to submit suggestions, comments, complaints, and even reports relating to local government activities or offices and is committed to responding within thirty days, identifying the relevant office and personnel overseeing the investigation and providing information on actions taken. The city of Dublin informs its residents of the avenues available to report various issues to designated service areas without requiring adherence to formal complaint procedures. Formal complaints directed towards the Dublin city council ¹⁷ can be lodged via post, email or online through the citizen Hub, with assurances of a formal acknowledgment within three working days and an expected response from the relevant department within twenty- one days. Dissatisfied individuals can escalate their complaint to the executive manager within fifteen working days and further to the Office of the Ombudsman if the problem persists.\n\nE-voting is not strictly limited to electoral processes, often extending to decision-making on various initiatives, projects or ideas within city portals. The availability and utility of this option varies. Some platforms showcase e-voting functionality for specific projects or competitions, with evidence of past engagement but limited current activity. Others require registration to access e-voting results or offer opportunities for online voting on specific topics. In some instances, e-voting is integrated into participatory budgeting schemes. While these examples are encouraging, widespread implementation remains uncommon, with sporadic instances of e-voting observed across different municipalities. Although some platforms provide clear avenues for citizens to participate in e-voting opportunities, the policy impact of these contributions remains unclear in certain cases. Overall, while e-voting services are present in various forms, they often represent a minority among the participation mechanisms available within city portals.\n\n_E-government literacy_\n\nIn the 2024 edition of the E-Government Survey, UN DESA introduces the concept of e-government literacy to measure the ability of residents – especially vulnerable groups – to utilize e-government services and take advantage of e-participation opportunities. In the present analysis, e-government literacy is based on the assessment of key features within government portals, including support services, privacy protection, digital identity authentication, and access to online resources, as these indicators are aligned with enhancing digital literacy and engagement across diverse domains. Although e-government literacy was not a specific category in 2022, many of its indicators were\n\n18%\n\n25%\n\n31%\n\n31%\n\n32%\n\n40%\n\n42%\n\n47%\n\n66%\n\n15%\n\n24%\n\n26%\n\n31%\n\n28%\n\n37%\n\n40%\n\n40%\n\n61%\n\n0% 10% 20% 30% 40% 50% 60% 70%\n\nE-voting\n\nFeedback about consultation processes\n\nParticipatory budgeting\n\nAnnouncement of upcoming e-participation activities\n\nParticipatory land use planning\n\nOnline deliberation processes\n\nReporting of incidents in public spaces\n\nInformation on the public meetings of the municipal council\n\nFeedback/complaint submission\n\n2022 2024"],"full_content":"2024 UN E-GovErNmENt SUrvEy\n\n148\n\n**Chapter 4**\n\nFigure 4.11 Implementation of participation and engagement indicators in city portals\n\nMost city portals incorporate feedback mechanisms to promote and facilitate resident engagement. Malé, ¹⁶ the capital of the Maldives and its most populous city, pledges to process all duly completed feedback submissions within five business days. Rome allows residents to submit suggestions, comments, complaints, and even reports relating to local government activities or offices and is committed to responding within thirty days, identifying the relevant office and personnel overseeing the investigation and providing information on actions taken. The city of Dublin informs its residents of the avenues available to report various issues to designated service areas without requiring adherence to formal complaint procedures. Formal complaints directed towards the Dublin city council ¹⁷ can be lodged via post, email or online through the citizen Hub, with assurances of a formal acknowledgment within three working days and an expected response from the relevant department within twenty- one days. Dissatisfied individuals can escalate their complaint to the executive manager within fifteen working days and further to the Office of the Ombudsman if the problem persists.\n\nE-voting is not strictly limited to electoral processes, often extending to decision-making on various initiatives, projects or ideas within city portals. The availability and utility of this option varies. Some platforms showcase e-voting functionality for specific projects or competitions, with evidence of past engagement but limited current activity. Others require registration to access e-voting results or offer opportunities for online voting on specific topics. In some instances, e-voting is integrated into participatory budgeting schemes. While these examples are encouraging, widespread implementation remains uncommon, with sporadic instances of e-voting observed across different municipalities. Although some platforms provide clear avenues for citizens to participate in e-voting opportunities, the policy impact of these contributions remains unclear in certain cases. Overall, while e-voting services are present in various forms, they often represent a minority among the participation mechanisms available within city portals.\n\n_E-government literacy_\n\nIn the 2024 edition of the E-Government Survey, UN DESA introduces the concept of e-government literacy to measure the ability of residents – especially vulnerable groups – to utilize e-government services and take advantage of e-participation opportunities. In the present analysis, e-government literacy is based on the assessment of key features within government portals, including support services, privacy protection, digital identity authentication, and access to online resources, as these indicators are aligned with enhancing digital literacy and engagement across diverse domains. Although e-government literacy was not a specific category in 2022, many of its indicators were\n\n18%\n\n25%\n\n31%\n\n31%\n\n32%\n\n40%\n\n42%\n\n47%\n\n66%\n\n15%\n\n24%\n\n26%\n\n31%\n\n28%\n\n37%\n\n40%\n\n40%\n\n61%\n\n0% 10% 20% 30% 40% 50% 60% 70%\n\nE-voting\n\nFeedback about consultation processes\n\nParticipatory budgeting\n\nAnnouncement of upcoming e-participation activities\n\nParticipatory land use planning\n\nOnline deliberation processes\n\nReporting of incidents in public spaces\n\nInformation on the public meetings of the municipal council\n\nFeedback/complaint submission\n\n2022 2024","publish_date":null,"title":"Figure 4.11 Implementation of participation and engagement indicators in city portals Most city portals incorporate feedback mechanisms to promote and","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/%28Web_version%29_E-Government_Survey_2024_1392024_p173.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/%28Web_version%29_E-Government_Survey_2024_1392024_p173.pdf"},"request":{"example_id":"chart/(Web_version)_E-Government_Survey_2024_1392024_p173","product_type":"parse","source_file_path":"(Web_version)_E-Government_Survey_2024_1392024_p173.pdf"},"started_at":"2026-11-04T19:31:04.372043"} +{"completed_at":"2026-11-04T19:45:50.420521","latency_in_ms":1188,"output":{"example_id":"chart/(Web_version)_E-Government_Survey_2024_1392024_p174","grounded_pages":[],"layout_pages":[],"markdown":"149\n\n**Chapter 4**\n\nChapter 4 • LoCaL e-Government DeveLopment\n\nFigure 4.12 Implementation of e-government literacy indicators in city portals\n\nincluded in the lOSI assessment, allowing for feasible comparisons between 2022 and 2024 data. As seen in figure 4.12, progress has been made in several areas, with rates of compliance increasing for indicators such as real-time communication functionality, the availability of information on online services use, privacy policies, online user support features, portal authentication, internal search mechanisms, and the integration of social networking features in government portals. These advances reflect a commitment to enhancing accessibility and engagement for the users of online services.\n\nMany members of society remain digitally disconnected and are in danger of being left behind in a world that is rapidly moving online. Providing free Internet access at the local level is crucial for ensuring inclusion. The availability of physical spaces for accessing online services varies across cities. Some municipalities offer service centres or “digital islands” within government offices, facilitating access to online services and often providing personal assistance. Vienna is one city that offers digital islands and provides support for users of online services. Belmopan, in Belize, boasts numerous free mobile Wi-Fi hotspots. Public libraries often serve as hubs for accessing online services, with free Wi-Fi widely available in many cities. While some cities have dedicated spaces for digital access, such as community centres or libraries, others lack clear evidence of facilities for public Internet access. Notably, some cities provide free Wi-Fi in public spaces such as parks, squares and metro stations, further enhancing connectivity. These varied efforts to ensure public Internet access underscore the importance assigned to digital inclusion in modern governance.\n\n_Technology_\n\nAs illustrated in figure 4.13, there have been improvements in most of the lOSI technology indicators since 2022. Notably, compliance rates have increased for mobile device accessibility (from 70 to 76 per cent), the ease with which portals can be found (from 73 to 77 per cent), and browser compatibility (from 75 to 78 per cent). Indicators such as the availability of internal advanced search mechanisms, foreign language support, and navigability have also registered improvement. However, compliance rates have dropped slightly for alignment with display standards, markup validation standards, and accessibility standards. As an example of the technological advancements being made, the Beijing Municipal Government Service centre website has introduced easy-access and senior-friendly features such as text-to-speech functionality, large fonts, and customizable colours, along with a one-stop service zone to simplify access to various services for individuals with disabilities and the elderly. These enhancements are intended to make government services more accessible and user-friendly for special groups, ensuring a more satisfying online experience for a wider range of consumers.\n\n19%\n\n31%\n\n37%\n\n38%\n\n46%\n\n46%\n\n52%\n\n67%\n\n69%\n\n17%\n\n31%\n\n34%\n\n38%\n\n40%\n\n43%\n\n49%\n\n64%\n\n65%\n\n0% 10% 20% 30% 40% 50% 60% 70% 80%\n\nReal-time communication\n\nOpen data policy\n\nInformation on online services use\n\nFacilitation of free Internet access\n\nPrivacy policy\n\nOnline user support\n\nPortal authentication\n\nInternal search mechanism\n\nSocial networking features\n\n2022 2024","pages":[{"line_numbers":[],"markdown":"149\n\n**Chapter 4**\n\nChapter 4 • LoCaL e-Government DeveLopment\n\nFigure 4.12 Implementation of e-government literacy indicators in city portals\n\nincluded in the lOSI assessment, allowing for feasible comparisons between 2022 and 2024 data. As seen in figure 4.12, progress has been made in several areas, with rates of compliance increasing for indicators such as real-time communication functionality, the availability of information on online services use, privacy policies, online user support features, portal authentication, internal search mechanisms, and the integration of social networking features in government portals. These advances reflect a commitment to enhancing accessibility and engagement for the users of online services.\n\nMany members of society remain digitally disconnected and are in danger of being left behind in a world that is rapidly moving online. Providing free Internet access at the local level is crucial for ensuring inclusion. The availability of physical spaces for accessing online services varies across cities. Some municipalities offer service centres or “digital islands” within government offices, facilitating access to online services and often providing personal assistance. Vienna is one city that offers digital islands and provides support for users of online services. Belmopan, in Belize, boasts numerous free mobile Wi-Fi hotspots. Public libraries often serve as hubs for accessing online services, with free Wi-Fi widely available in many cities. While some cities have dedicated spaces for digital access, such as community centres or libraries, others lack clear evidence of facilities for public Internet access. Notably, some cities provide free Wi-Fi in public spaces such as parks, squares and metro stations, further enhancing connectivity. These varied efforts to ensure public Internet access underscore the importance assigned to digital inclusion in modern governance.\n\n_Technology_\n\nAs illustrated in figure 4.13, there have been improvements in most of the lOSI technology indicators since 2022. Notably, compliance rates have increased for mobile device accessibility (from 70 to 76 per cent), the ease with which portals can be found (from 73 to 77 per cent), and browser compatibility (from 75 to 78 per cent). Indicators such as the availability of internal advanced search mechanisms, foreign language support, and navigability have also registered improvement. However, compliance rates have dropped slightly for alignment with display standards, markup validation standards, and accessibility standards. As an example of the technological advancements being made, the Beijing Municipal Government Service centre website has introduced easy-access and senior-friendly features such as text-to-speech functionality, large fonts, and customizable colours, along with a one-stop service zone to simplify access to various services for individuals with disabilities and the elderly. These enhancements are intended to make government services more accessible and user-friendly for special groups, ensuring a more satisfying online experience for a wider range of consumers.\n\n19%\n\n31%\n\n37%\n\n38%\n\n46%\n\n46%\n\n52%\n\n67%\n\n69%\n\n17%\n\n31%\n\n34%\n\n38%\n\n40%\n\n43%\n\n49%\n\n64%\n\n65%\n\n0% 10% 20% 30% 40% 50% 60% 70% 80%\n\nReal-time communication\n\nOpen data policy\n\nInformation on online services use\n\nFacilitation of free Internet access\n\nPrivacy policy\n\nOnline user support\n\nPortal authentication\n\nInternal search mechanism\n\nSocial networking features\n\n2022 2024","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"149\n\n**Chapter 4**\n\nChapter 4 • LoCaL e-Government DeveLopment\n\nFigure 4.12 Implementation of e-government literacy indicators in city portals\n\nincluded in the lOSI assessment, allowing for feasible comparisons between 2022 and 2024 data. As seen in figure 4.12, progress has been made in several areas, with rates of compliance increasing for indicators such as real-time communication functionality, the availability of information on online services use, privacy policies, online user support features, portal authentication, internal search mechanisms, and the integration of social networking features in government portals. These advances reflect a commitment to enhancing accessibility and engagement for the users of online services.\n\nMany members of society remain digitally disconnected and are in danger of being left behind in a world that is rapidly moving online. Providing free Internet access at the local level is crucial for ensuring inclusion. The availability of physical spaces for accessing online services varies across cities. Some municipalities offer service centres or “digital islands” within government offices, facilitating access to online services and often providing personal assistance. Vienna is one city that offers digital islands and provides support for users of online services. Belmopan, in Belize, boasts numerous free mobile Wi-Fi hotspots. Public libraries often serve as hubs for accessing online services, with free Wi-Fi widely available in many cities. While some cities have dedicated spaces for digital access, such as community centres or libraries, others lack clear evidence of facilities for public Internet access. Notably, some cities provide free Wi-Fi in public spaces such as parks, squares and metro stations, further enhancing connectivity. These varied efforts to ensure public Internet access underscore the importance assigned to digital inclusion in modern governance.\n\n_Technology_\n\nAs illustrated in figure 4.13, there have been improvements in most of the lOSI technology indicators since 2022. Notably, compliance rates have increased for mobile device accessibility (from 70 to 76 per cent), the ease with which portals can be found (from 73 to 77 per cent), and browser compatibility (from 75 to 78 per cent). Indicators such as the availability of internal advanced search mechanisms, foreign language support, and navigability have also registered improvement. However, compliance rates have dropped slightly for alignment with display standards, markup validation standards, and accessibility standards. As an example of the technological advancements being made, the Beijing Municipal Government Service centre website has introduced easy-access and senior-friendly features such as text-to-speech functionality, large fonts, and customizable colours, along with a one-stop service zone to simplify access to various services for individuals with disabilities and the elderly. These enhancements are intended to make government services more accessible and user-friendly for special groups, ensuring a more satisfying online experience for a wider range of consumers.\n\n19%\n\n31%\n\n37%\n\n38%\n\n46%\n\n46%\n\n52%\n\n67%\n\n69%\n\n17%\n\n31%\n\n34%\n\n38%\n\n40%\n\n43%\n\n49%\n\n64%\n\n65%\n\n0% 10% 20% 30% 40% 50% 60% 70% 80%\n\nReal-time communication\n\nOpen data policy\n\nInformation on online services use\n\nFacilitation of free Internet access\n\nPrivacy policy\n\nOnline user support\n\nPortal authentication\n\nInternal search mechanism\n\nSocial networking features\n\n2022 2024","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/%28Web_version%29_E-Government_Survey_2024_1392024_p174.pdf"]},"response":{"errors":[],"extract_id":"extract_009314eb70bb67642f15b69e13ce7b19","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["149\n\n**Chapter 4**\n\nChapter 4 • LoCaL e-Government DeveLopment\n\nFigure 4.12 Implementation of e-government literacy indicators in city portals\n\nincluded in the lOSI assessment, allowing for feasible comparisons between 2022 and 2024 data. As seen in figure 4.12, progress has been made in several areas, with rates of compliance increasing for indicators such as real-time communication functionality, the availability of information on online services use, privacy policies, online user support features, portal authentication, internal search mechanisms, and the integration of social networking features in government portals. These advances reflect a commitment to enhancing accessibility and engagement for the users of online services.\n\nMany members of society remain digitally disconnected and are in danger of being left behind in a world that is rapidly moving online. Providing free Internet access at the local level is crucial for ensuring inclusion. The availability of physical spaces for accessing online services varies across cities. Some municipalities offer service centres or “digital islands” within government offices, facilitating access to online services and often providing personal assistance. Vienna is one city that offers digital islands and provides support for users of online services. Belmopan, in Belize, boasts numerous free mobile Wi-Fi hotspots. Public libraries often serve as hubs for accessing online services, with free Wi-Fi widely available in many cities. While some cities have dedicated spaces for digital access, such as community centres or libraries, others lack clear evidence of facilities for public Internet access. Notably, some cities provide free Wi-Fi in public spaces such as parks, squares and metro stations, further enhancing connectivity. These varied efforts to ensure public Internet access underscore the importance assigned to digital inclusion in modern governance.\n\n_Technology_\n\nAs illustrated in figure 4.13, there have been improvements in most of the lOSI technology indicators since 2022. Notably, compliance rates have increased for mobile device accessibility (from 70 to 76 per cent), the ease with which portals can be found (from 73 to 77 per cent), and browser compatibility (from 75 to 78 per cent). Indicators such as the availability of internal advanced search mechanisms, foreign language support, and navigability have also registered improvement. However, compliance rates have dropped slightly for alignment with display standards, markup validation standards, and accessibility standards. As an example of the technological advancements being made, the Beijing Municipal Government Service centre website has introduced easy-access and senior-friendly features such as text-to-speech functionality, large fonts, and customizable colours, along with a one-stop service zone to simplify access to various services for individuals with disabilities and the elderly. These enhancements are intended to make government services more accessible and user-friendly for special groups, ensuring a more satisfying online experience for a wider range of consumers.\n\n19%\n\n31%\n\n37%\n\n38%\n\n46%\n\n46%\n\n52%\n\n67%\n\n69%\n\n17%\n\n31%\n\n34%\n\n38%\n\n40%\n\n43%\n\n49%\n\n64%\n\n65%\n\n0% 10% 20% 30% 40% 50% 60% 70% 80%\n\nReal-time communication\n\nOpen data policy\n\nInformation on online services use\n\nFacilitation of free Internet access\n\nPrivacy policy\n\nOnline user support\n\nPortal authentication\n\nInternal search mechanism\n\nSocial networking features\n\n2022 2024"],"full_content":"149\n\n**Chapter 4**\n\nChapter 4 • LoCaL e-Government DeveLopment\n\nFigure 4.12 Implementation of e-government literacy indicators in city portals\n\nincluded in the lOSI assessment, allowing for feasible comparisons between 2022 and 2024 data. As seen in figure 4.12, progress has been made in several areas, with rates of compliance increasing for indicators such as real-time communication functionality, the availability of information on online services use, privacy policies, online user support features, portal authentication, internal search mechanisms, and the integration of social networking features in government portals. These advances reflect a commitment to enhancing accessibility and engagement for the users of online services.\n\nMany members of society remain digitally disconnected and are in danger of being left behind in a world that is rapidly moving online. Providing free Internet access at the local level is crucial for ensuring inclusion. The availability of physical spaces for accessing online services varies across cities. Some municipalities offer service centres or “digital islands” within government offices, facilitating access to online services and often providing personal assistance. Vienna is one city that offers digital islands and provides support for users of online services. Belmopan, in Belize, boasts numerous free mobile Wi-Fi hotspots. Public libraries often serve as hubs for accessing online services, with free Wi-Fi widely available in many cities. While some cities have dedicated spaces for digital access, such as community centres or libraries, others lack clear evidence of facilities for public Internet access. Notably, some cities provide free Wi-Fi in public spaces such as parks, squares and metro stations, further enhancing connectivity. These varied efforts to ensure public Internet access underscore the importance assigned to digital inclusion in modern governance.\n\n_Technology_\n\nAs illustrated in figure 4.13, there have been improvements in most of the lOSI technology indicators since 2022. Notably, compliance rates have increased for mobile device accessibility (from 70 to 76 per cent), the ease with which portals can be found (from 73 to 77 per cent), and browser compatibility (from 75 to 78 per cent). Indicators such as the availability of internal advanced search mechanisms, foreign language support, and navigability have also registered improvement. However, compliance rates have dropped slightly for alignment with display standards, markup validation standards, and accessibility standards. As an example of the technological advancements being made, the Beijing Municipal Government Service centre website has introduced easy-access and senior-friendly features such as text-to-speech functionality, large fonts, and customizable colours, along with a one-stop service zone to simplify access to various services for individuals with disabilities and the elderly. These enhancements are intended to make government services more accessible and user-friendly for special groups, ensuring a more satisfying online experience for a wider range of consumers.\n\n19%\n\n31%\n\n37%\n\n38%\n\n46%\n\n46%\n\n52%\n\n67%\n\n69%\n\n17%\n\n31%\n\n34%\n\n38%\n\n40%\n\n43%\n\n49%\n\n64%\n\n65%\n\n0% 10% 20% 30% 40% 50% 60% 70% 80%\n\nReal-time communication\n\nOpen data policy\n\nInformation on online services use\n\nFacilitation of free Internet access\n\nPrivacy policy\n\nOnline user support\n\nPortal authentication\n\nInternal search mechanism\n\nSocial networking features\n\n2022 2024","publish_date":null,"title":"included in the lOSI assessment, allowing for feasible comparisons between 2022 and 2024 data. As seen in figure 4.12, progress","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/%28Web_version%29_E-Government_Survey_2024_1392024_p174.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/%28Web_version%29_E-Government_Survey_2024_1392024_p174.pdf"},"request":{"example_id":"chart/(Web_version)_E-Government_Survey_2024_1392024_p174","product_type":"parse","source_file_path":"(Web_version)_E-Government_Survey_2024_1392024_p174.pdf"},"started_at":"2026-11-04T19:45:49.232245"} +{"completed_at":"2026-11-04T19:45:52.608325","latency_in_ms":2386,"output":{"example_id":"chart/(Web_version)_E-Government_Survey_2024_1392024_p175","grounded_pages":[],"layout_pages":[],"markdown":"2024 UN E-GovErNmENt SUrvEy\n\n150\n\n**Chapter 4**\n\nFigure 4.13 Implementation of technology indicators in city portals\n\nThe importance of search engines cannot be overstated. Growing numbers of people needing services or information are turning to search engines and generative AI tools such as chatGPT and copilot for assistance. In the present context, the ability to find city portals easily on the first result page produced by a search engine is crucial. This ensures that residents and visitors can swiftly access important information and services provided by municipal governments. It is worth noting that, across a wide range of countries, city portal links consistently appear among the top results from popular search engines such as Google, Bing and Yahoo. Search engine visibility is key to ensuring seamless access to municipal services and information.\n\n### **4\\.3 Smart cities for sustainable development**\n\nA smart city for sustainable development embodies a vision of urban progress aligned with the principles of inclusivity, safety, resilience, and sustainability articulated in SDG 11. cities that offer numerous opportunities and promote prosperity can also face challenges such as poor health conditions and environmental degradation. The integration of digital technologies in urban infrastructure and services is essential for addressing these and other challenges. By leveraging digital innovations, cities can better meet the needs of residents, enhancing overall liveability, workability and sustainability. The lOSI underscores the importance of digital technologies in empowering local officials to improve urban conditions and meet the evolving needs of communities. This emphasis on digital empowerment aligns with the broader goal of creating cities that are not only technologically advanced but also inclusive, resilient and environmentally sustainable. Towards that end, the lOSI can serve as a tool for measuring e-government progress within cities.\n\nA number of smart city initiatives have been undertaken worldwide. It is important to note that there is no universally agreed-upon definition of a smart city. The digital transformation of urban spaces is a dynamic process or journey rather than a static destination; with the rapid advances in technology, cities can continuously evolve and improve their smart capabilities. UN DESA, through its lOSI assessments since 2018, has been instrumental in highlighting various smart city initiatives but has not provided definitive guidelines on what constitutes a smart city. ITU offers the following definition: “A smart sustainable city is an innovative city that uses information and communication technologies (IcTs) and other means to improve quality of life, efficiency of urban operation and services, and\n\n4%\n\n5%\n\n5%\n\n24%\n\n43%\n\n46%\n\n76%\n\n77%\n\n78%\n\n5%\n\n4%\n\n6%\n\n22%\n\n44%\n\n42%\n\n70%\n\n73%\n\n75%\n\n0% 10% 20% 30% 40% 50% 60% 70% 80% 90%\n\nAlignment with display standards\n\nAlignment with markup validation standards\n\nAlignment with accessibility standards\n\nInternal advanced search mechanism\n\nForeign language support\n\nNavigability\n\nMobile device accessibility\n\nEase of portal fnding\n\nBrowser compatibility\n\n2022 2024","pages":[{"line_numbers":[],"markdown":"2024 UN E-GovErNmENt SUrvEy\n\n150\n\n**Chapter 4**\n\nFigure 4.13 Implementation of technology indicators in city portals\n\nThe importance of search engines cannot be overstated. Growing numbers of people needing services or information are turning to search engines and generative AI tools such as chatGPT and copilot for assistance. In the present context, the ability to find city portals easily on the first result page produced by a search engine is crucial. This ensures that residents and visitors can swiftly access important information and services provided by municipal governments. It is worth noting that, across a wide range of countries, city portal links consistently appear among the top results from popular search engines such as Google, Bing and Yahoo. Search engine visibility is key to ensuring seamless access to municipal services and information.\n\n### **4\\.3 Smart cities for sustainable development**\n\nA smart city for sustainable development embodies a vision of urban progress aligned with the principles of inclusivity, safety, resilience, and sustainability articulated in SDG 11. cities that offer numerous opportunities and promote prosperity can also face challenges such as poor health conditions and environmental degradation. The integration of digital technologies in urban infrastructure and services is essential for addressing these and other challenges. By leveraging digital innovations, cities can better meet the needs of residents, enhancing overall liveability, workability and sustainability. The lOSI underscores the importance of digital technologies in empowering local officials to improve urban conditions and meet the evolving needs of communities. This emphasis on digital empowerment aligns with the broader goal of creating cities that are not only technologically advanced but also inclusive, resilient and environmentally sustainable. Towards that end, the lOSI can serve as a tool for measuring e-government progress within cities.\n\nA number of smart city initiatives have been undertaken worldwide. It is important to note that there is no universally agreed-upon definition of a smart city. The digital transformation of urban spaces is a dynamic process or journey rather than a static destination; with the rapid advances in technology, cities can continuously evolve and improve their smart capabilities. UN DESA, through its lOSI assessments since 2018, has been instrumental in highlighting various smart city initiatives but has not provided definitive guidelines on what constitutes a smart city. ITU offers the following definition: “A smart sustainable city is an innovative city that uses information and communication technologies (IcTs) and other means to improve quality of life, efficiency of urban operation and services, and\n\n4%\n\n5%\n\n5%\n\n24%\n\n43%\n\n46%\n\n76%\n\n77%\n\n78%\n\n5%\n\n4%\n\n6%\n\n22%\n\n44%\n\n42%\n\n70%\n\n73%\n\n75%\n\n0% 10% 20% 30% 40% 50% 60% 70% 80% 90%\n\nAlignment with display standards\n\nAlignment with markup validation standards\n\nAlignment with accessibility standards\n\nInternal advanced search mechanism\n\nForeign language support\n\nNavigability\n\nMobile device accessibility\n\nEase of portal fnding\n\nBrowser compatibility\n\n2022 2024","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"2024 UN E-GovErNmENt SUrvEy\n\n150\n\n**Chapter 4**\n\nFigure 4.13 Implementation of technology indicators in city portals\n\nThe importance of search engines cannot be overstated. Growing numbers of people needing services or information are turning to search engines and generative AI tools such as chatGPT and copilot for assistance. In the present context, the ability to find city portals easily on the first result page produced by a search engine is crucial. This ensures that residents and visitors can swiftly access important information and services provided by municipal governments. It is worth noting that, across a wide range of countries, city portal links consistently appear among the top results from popular search engines such as Google, Bing and Yahoo. Search engine visibility is key to ensuring seamless access to municipal services and information.\n\n### **4\\.3 Smart cities for sustainable development**\n\nA smart city for sustainable development embodies a vision of urban progress aligned with the principles of inclusivity, safety, resilience, and sustainability articulated in SDG 11. cities that offer numerous opportunities and promote prosperity can also face challenges such as poor health conditions and environmental degradation. The integration of digital technologies in urban infrastructure and services is essential for addressing these and other challenges. By leveraging digital innovations, cities can better meet the needs of residents, enhancing overall liveability, workability and sustainability. The lOSI underscores the importance of digital technologies in empowering local officials to improve urban conditions and meet the evolving needs of communities. This emphasis on digital empowerment aligns with the broader goal of creating cities that are not only technologically advanced but also inclusive, resilient and environmentally sustainable. Towards that end, the lOSI can serve as a tool for measuring e-government progress within cities.\n\nA number of smart city initiatives have been undertaken worldwide. It is important to note that there is no universally agreed-upon definition of a smart city. The digital transformation of urban spaces is a dynamic process or journey rather than a static destination; with the rapid advances in technology, cities can continuously evolve and improve their smart capabilities. UN DESA, through its lOSI assessments since 2018, has been instrumental in highlighting various smart city initiatives but has not provided definitive guidelines on what constitutes a smart city. ITU offers the following definition: “A smart sustainable city is an innovative city that uses information and communication technologies (IcTs) and other means to improve quality of life, efficiency of urban operation and services, and\n\n4%\n\n5%\n\n5%\n\n24%\n\n43%\n\n46%\n\n76%\n\n77%\n\n78%\n\n5%\n\n4%\n\n6%\n\n22%\n\n44%\n\n42%\n\n70%\n\n73%\n\n75%\n\n0% 10% 20% 30% 40% 50% 60% 70% 80% 90%\n\nAlignment with display standards\n\nAlignment with markup validation standards\n\nAlignment with accessibility standards\n\nInternal advanced search mechanism\n\nForeign language support\n\nNavigability\n\nMobile device accessibility\n\nEase of portal fnding\n\nBrowser compatibility\n\n2022 2024","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/%28Web_version%29_E-Government_Survey_2024_1392024_p175.pdf"]},"response":{"errors":[],"extract_id":"extract_4de19df9f5c66f7df0ad9dd8baf8af5b","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["2024 UN E-GovErNmENt SUrvEy\n\n150\n\n**Chapter 4**\n\nFigure 4.13 Implementation of technology indicators in city portals\n\nThe importance of search engines cannot be overstated. Growing numbers of people needing services or information are turning to search engines and generative AI tools such as chatGPT and copilot for assistance. In the present context, the ability to find city portals easily on the first result page produced by a search engine is crucial. This ensures that residents and visitors can swiftly access important information and services provided by municipal governments. It is worth noting that, across a wide range of countries, city portal links consistently appear among the top results from popular search engines such as Google, Bing and Yahoo. Search engine visibility is key to ensuring seamless access to municipal services and information.\n\n### **4\\.3 Smart cities for sustainable development**\n\nA smart city for sustainable development embodies a vision of urban progress aligned with the principles of inclusivity, safety, resilience, and sustainability articulated in SDG 11. cities that offer numerous opportunities and promote prosperity can also face challenges such as poor health conditions and environmental degradation. The integration of digital technologies in urban infrastructure and services is essential for addressing these and other challenges. By leveraging digital innovations, cities can better meet the needs of residents, enhancing overall liveability, workability and sustainability. The lOSI underscores the importance of digital technologies in empowering local officials to improve urban conditions and meet the evolving needs of communities. This emphasis on digital empowerment aligns with the broader goal of creating cities that are not only technologically advanced but also inclusive, resilient and environmentally sustainable. Towards that end, the lOSI can serve as a tool for measuring e-government progress within cities.\n\nA number of smart city initiatives have been undertaken worldwide. It is important to note that there is no universally agreed-upon definition of a smart city. The digital transformation of urban spaces is a dynamic process or journey rather than a static destination; with the rapid advances in technology, cities can continuously evolve and improve their smart capabilities. UN DESA, through its lOSI assessments since 2018, has been instrumental in highlighting various smart city initiatives but has not provided definitive guidelines on what constitutes a smart city. ITU offers the following definition: “A smart sustainable city is an innovative city that uses information and communication technologies (IcTs) and other means to improve quality of life, efficiency of urban operation and services, and\n\n4%\n\n5%\n\n5%\n\n24%\n\n43%\n\n46%\n\n76%\n\n77%\n\n78%\n\n5%\n\n4%\n\n6%\n\n22%\n\n44%\n\n42%\n\n70%\n\n73%\n\n75%\n\n0% 10% 20% 30% 40% 50% 60% 70% 80% 90%\n\nAlignment with display standards\n\nAlignment with markup validation standards\n\nAlignment with accessibility standards\n\nInternal advanced search mechanism\n\nForeign language support\n\nNavigability\n\nMobile device accessibility\n\nEase of portal fnding\n\nBrowser compatibility\n\n2022 2024"],"full_content":"2024 UN E-GovErNmENt SUrvEy\n\n150\n\n**Chapter 4**\n\nFigure 4.13 Implementation of technology indicators in city portals\n\nThe importance of search engines cannot be overstated. Growing numbers of people needing services or information are turning to search engines and generative AI tools such as chatGPT and copilot for assistance. In the present context, the ability to find city portals easily on the first result page produced by a search engine is crucial. This ensures that residents and visitors can swiftly access important information and services provided by municipal governments. It is worth noting that, across a wide range of countries, city portal links consistently appear among the top results from popular search engines such as Google, Bing and Yahoo. Search engine visibility is key to ensuring seamless access to municipal services and information.\n\n### **4\\.3 Smart cities for sustainable development**\n\nA smart city for sustainable development embodies a vision of urban progress aligned with the principles of inclusivity, safety, resilience, and sustainability articulated in SDG 11. cities that offer numerous opportunities and promote prosperity can also face challenges such as poor health conditions and environmental degradation. The integration of digital technologies in urban infrastructure and services is essential for addressing these and other challenges. By leveraging digital innovations, cities can better meet the needs of residents, enhancing overall liveability, workability and sustainability. The lOSI underscores the importance of digital technologies in empowering local officials to improve urban conditions and meet the evolving needs of communities. This emphasis on digital empowerment aligns with the broader goal of creating cities that are not only technologically advanced but also inclusive, resilient and environmentally sustainable. Towards that end, the lOSI can serve as a tool for measuring e-government progress within cities.\n\nA number of smart city initiatives have been undertaken worldwide. It is important to note that there is no universally agreed-upon definition of a smart city. The digital transformation of urban spaces is a dynamic process or journey rather than a static destination; with the rapid advances in technology, cities can continuously evolve and improve their smart capabilities. UN DESA, through its lOSI assessments since 2018, has been instrumental in highlighting various smart city initiatives but has not provided definitive guidelines on what constitutes a smart city. ITU offers the following definition: “A smart sustainable city is an innovative city that uses information and communication technologies (IcTs) and other means to improve quality of life, efficiency of urban operation and services, and\n\n4%\n\n5%\n\n5%\n\n24%\n\n43%\n\n46%\n\n76%\n\n77%\n\n78%\n\n5%\n\n4%\n\n6%\n\n22%\n\n44%\n\n42%\n\n70%\n\n73%\n\n75%\n\n0% 10% 20% 30% 40% 50% 60% 70% 80% 90%\n\nAlignment with display standards\n\nAlignment with markup validation standards\n\nAlignment with accessibility standards\n\nInternal advanced search mechanism\n\nForeign language support\n\nNavigability\n\nMobile device accessibility\n\nEase of portal fnding\n\nBrowser compatibility\n\n2022 2024","publish_date":null,"title":"4.3 Smart cities for sustainable development","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/%28Web_version%29_E-Government_Survey_2024_1392024_p175.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/%28Web_version%29_E-Government_Survey_2024_1392024_p175.pdf"},"request":{"example_id":"chart/(Web_version)_E-Government_Survey_2024_1392024_p175","product_type":"parse","source_file_path":"(Web_version)_E-Government_Survey_2024_1392024_p175.pdf"},"started_at":"2026-11-04T19:45:50.221961"} +{"completed_at":"2026-11-04T19:45:54.436009","latency_in_ms":2453,"output":{"example_id":"chart/(Web_version)_E-Government_Survey_2024_1392024_p177","grounded_pages":[],"layout_pages":[],"markdown":"2024 UN E-GovErNmENt SUrvEy\n\n152\n\n**Chapter 4**\n\nFigure 4.15 presents proxy indicators from the lOSI assessment that may offer some insight into the extent of digital readiness for smart city development. These indicators serve as a valuable tool for gauging the level of digitalization and innovation within urban environments and underscore the importance of data in driving smart city initiatives forward. Data not only inform decision-making processes but also serve as the foundation for AI applications and other smart city endeavours. In the lOSI assessments, indicators such as the existence of smart city initiatives, open government data sharing, the presence of online policies for data accessibility, and indications of emergent technology utilization provide valuable insight into the progress cities are making towards becoming smarter and more responsive to the needs of residents.\n\nAn analysis of smart city initiatives around the world reveals varying levels and stages of engagement, implementation and visibility. Many cities, including Dubai, Amsterdam, Riyadh, Seoul, Singapore, and New York city, are actively engaged in smart city development. cities such as Berlin, ²⁰ copenhagen and Istanbul have dedicated departments, units or projects focused solely on smart city development – an indication of their strong commitment to digital urban development. Smart city Berlin is a collaborative effort involving various stakeholders, including government agencies, private companies, academic institutions, and the residents themselves, all working together to create a more liveable, resilient and prosperous urban environment. The copenhagen Solutions lab ²¹ is a development unit under the Technical and Environmental Administration for the Municipality of copenhagen. The lab acts as an incubator for smart city development and leverages the opportunities provided by urban laboratories, which are real-world testing grounds within the city itself. These laboratories provide the ideal scale for systematic and targeted experiments and the exploration of new solutions. By developing, testing, and operationalizing innovative solutions in urban laboratories, copenhagen Solutions lab gains valuable insight into which ones can be effectively scaled to encompass larger areas of the city, ultimately benefiting all of copenhagen. This approach also provides a critical foundation for informed decisions regarding investments in new technologies for the Municipality’s smart city initiatives.\n\nFigure 4.15 LOSI indicators as a proxy for progress towards smart city development\n\n49%\n\n37%\n\n33%\n\n33%\n\n46%\n\n35%\n\n31%\n\n30%\n\n0% 10% 20% 30% 40% 50% 60%\n\nExistence of smart city initiatives\n\nOpen government data sharing\n\nOpen government data policy online\n\nActual or intended use of emergent technologies\n\n2022 2024","pages":[{"line_numbers":[],"markdown":"2024 UN E-GovErNmENt SUrvEy\n\n152\n\n**Chapter 4**\n\nFigure 4.15 presents proxy indicators from the lOSI assessment that may offer some insight into the extent of digital readiness for smart city development. These indicators serve as a valuable tool for gauging the level of digitalization and innovation within urban environments and underscore the importance of data in driving smart city initiatives forward. Data not only inform decision-making processes but also serve as the foundation for AI applications and other smart city endeavours. In the lOSI assessments, indicators such as the existence of smart city initiatives, open government data sharing, the presence of online policies for data accessibility, and indications of emergent technology utilization provide valuable insight into the progress cities are making towards becoming smarter and more responsive to the needs of residents.\n\nAn analysis of smart city initiatives around the world reveals varying levels and stages of engagement, implementation and visibility. Many cities, including Dubai, Amsterdam, Riyadh, Seoul, Singapore, and New York city, are actively engaged in smart city development. cities such as Berlin, ²⁰ copenhagen and Istanbul have dedicated departments, units or projects focused solely on smart city development – an indication of their strong commitment to digital urban development. Smart city Berlin is a collaborative effort involving various stakeholders, including government agencies, private companies, academic institutions, and the residents themselves, all working together to create a more liveable, resilient and prosperous urban environment. The copenhagen Solutions lab ²¹ is a development unit under the Technical and Environmental Administration for the Municipality of copenhagen. The lab acts as an incubator for smart city development and leverages the opportunities provided by urban laboratories, which are real-world testing grounds within the city itself. These laboratories provide the ideal scale for systematic and targeted experiments and the exploration of new solutions. By developing, testing, and operationalizing innovative solutions in urban laboratories, copenhagen Solutions lab gains valuable insight into which ones can be effectively scaled to encompass larger areas of the city, ultimately benefiting all of copenhagen. This approach also provides a critical foundation for informed decisions regarding investments in new technologies for the Municipality’s smart city initiatives.\n\nFigure 4.15 LOSI indicators as a proxy for progress towards smart city development\n\n49%\n\n37%\n\n33%\n\n33%\n\n46%\n\n35%\n\n31%\n\n30%\n\n0% 10% 20% 30% 40% 50% 60%\n\nExistence of smart city initiatives\n\nOpen government data sharing\n\nOpen government data policy online\n\nActual or intended use of emergent technologies\n\n2022 2024","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"2024 UN E-GovErNmENt SUrvEy\n\n152\n\n**Chapter 4**\n\nFigure 4.15 presents proxy indicators from the lOSI assessment that may offer some insight into the extent of digital readiness for smart city development. These indicators serve as a valuable tool for gauging the level of digitalization and innovation within urban environments and underscore the importance of data in driving smart city initiatives forward. Data not only inform decision-making processes but also serve as the foundation for AI applications and other smart city endeavours. In the lOSI assessments, indicators such as the existence of smart city initiatives, open government data sharing, the presence of online policies for data accessibility, and indications of emergent technology utilization provide valuable insight into the progress cities are making towards becoming smarter and more responsive to the needs of residents.\n\nAn analysis of smart city initiatives around the world reveals varying levels and stages of engagement, implementation and visibility. Many cities, including Dubai, Amsterdam, Riyadh, Seoul, Singapore, and New York city, are actively engaged in smart city development. cities such as Berlin, ²⁰ copenhagen and Istanbul have dedicated departments, units or projects focused solely on smart city development – an indication of their strong commitment to digital urban development. Smart city Berlin is a collaborative effort involving various stakeholders, including government agencies, private companies, academic institutions, and the residents themselves, all working together to create a more liveable, resilient and prosperous urban environment. The copenhagen Solutions lab ²¹ is a development unit under the Technical and Environmental Administration for the Municipality of copenhagen. The lab acts as an incubator for smart city development and leverages the opportunities provided by urban laboratories, which are real-world testing grounds within the city itself. These laboratories provide the ideal scale for systematic and targeted experiments and the exploration of new solutions. By developing, testing, and operationalizing innovative solutions in urban laboratories, copenhagen Solutions lab gains valuable insight into which ones can be effectively scaled to encompass larger areas of the city, ultimately benefiting all of copenhagen. This approach also provides a critical foundation for informed decisions regarding investments in new technologies for the Municipality’s smart city initiatives.\n\nFigure 4.15 LOSI indicators as a proxy for progress towards smart city development\n\n49%\n\n37%\n\n33%\n\n33%\n\n46%\n\n35%\n\n31%\n\n30%\n\n0% 10% 20% 30% 40% 50% 60%\n\nExistence of smart city initiatives\n\nOpen government data sharing\n\nOpen government data policy online\n\nActual or intended use of emergent technologies\n\n2022 2024","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/%28Web_version%29_E-Government_Survey_2024_1392024_p177.pdf"]},"response":{"errors":[],"extract_id":"extract_c0859df34c8509d6758aa0ab69c7b622","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["2024 UN E-GovErNmENt SUrvEy\n\n152\n\n**Chapter 4**\n\nFigure 4.15 presents proxy indicators from the lOSI assessment that may offer some insight into the extent of digital readiness for smart city development. These indicators serve as a valuable tool for gauging the level of digitalization and innovation within urban environments and underscore the importance of data in driving smart city initiatives forward. Data not only inform decision-making processes but also serve as the foundation for AI applications and other smart city endeavours. In the lOSI assessments, indicators such as the existence of smart city initiatives, open government data sharing, the presence of online policies for data accessibility, and indications of emergent technology utilization provide valuable insight into the progress cities are making towards becoming smarter and more responsive to the needs of residents.\n\nAn analysis of smart city initiatives around the world reveals varying levels and stages of engagement, implementation and visibility. Many cities, including Dubai, Amsterdam, Riyadh, Seoul, Singapore, and New York city, are actively engaged in smart city development. cities such as Berlin, ²⁰ copenhagen and Istanbul have dedicated departments, units or projects focused solely on smart city development – an indication of their strong commitment to digital urban development. Smart city Berlin is a collaborative effort involving various stakeholders, including government agencies, private companies, academic institutions, and the residents themselves, all working together to create a more liveable, resilient and prosperous urban environment. The copenhagen Solutions lab ²¹ is a development unit under the Technical and Environmental Administration for the Municipality of copenhagen. The lab acts as an incubator for smart city development and leverages the opportunities provided by urban laboratories, which are real-world testing grounds within the city itself. These laboratories provide the ideal scale for systematic and targeted experiments and the exploration of new solutions. By developing, testing, and operationalizing innovative solutions in urban laboratories, copenhagen Solutions lab gains valuable insight into which ones can be effectively scaled to encompass larger areas of the city, ultimately benefiting all of copenhagen. This approach also provides a critical foundation for informed decisions regarding investments in new technologies for the Municipality’s smart city initiatives.\n\nFigure 4.15 LOSI indicators as a proxy for progress towards smart city development\n\n49%\n\n37%\n\n33%\n\n33%\n\n46%\n\n35%\n\n31%\n\n30%\n\n0% 10% 20% 30% 40% 50% 60%\n\nExistence of smart city initiatives\n\nOpen government data sharing\n\nOpen government data policy online\n\nActual or intended use of emergent technologies\n\n2022 2024"],"full_content":"2024 UN E-GovErNmENt SUrvEy\n\n152\n\n**Chapter 4**\n\nFigure 4.15 presents proxy indicators from the lOSI assessment that may offer some insight into the extent of digital readiness for smart city development. These indicators serve as a valuable tool for gauging the level of digitalization and innovation within urban environments and underscore the importance of data in driving smart city initiatives forward. Data not only inform decision-making processes but also serve as the foundation for AI applications and other smart city endeavours. In the lOSI assessments, indicators such as the existence of smart city initiatives, open government data sharing, the presence of online policies for data accessibility, and indications of emergent technology utilization provide valuable insight into the progress cities are making towards becoming smarter and more responsive to the needs of residents.\n\nAn analysis of smart city initiatives around the world reveals varying levels and stages of engagement, implementation and visibility. Many cities, including Dubai, Amsterdam, Riyadh, Seoul, Singapore, and New York city, are actively engaged in smart city development. cities such as Berlin, ²⁰ copenhagen and Istanbul have dedicated departments, units or projects focused solely on smart city development – an indication of their strong commitment to digital urban development. Smart city Berlin is a collaborative effort involving various stakeholders, including government agencies, private companies, academic institutions, and the residents themselves, all working together to create a more liveable, resilient and prosperous urban environment. The copenhagen Solutions lab ²¹ is a development unit under the Technical and Environmental Administration for the Municipality of copenhagen. The lab acts as an incubator for smart city development and leverages the opportunities provided by urban laboratories, which are real-world testing grounds within the city itself. These laboratories provide the ideal scale for systematic and targeted experiments and the exploration of new solutions. By developing, testing, and operationalizing innovative solutions in urban laboratories, copenhagen Solutions lab gains valuable insight into which ones can be effectively scaled to encompass larger areas of the city, ultimately benefiting all of copenhagen. This approach also provides a critical foundation for informed decisions regarding investments in new technologies for the Municipality’s smart city initiatives.\n\nFigure 4.15 LOSI indicators as a proxy for progress towards smart city development\n\n49%\n\n37%\n\n33%\n\n33%\n\n46%\n\n35%\n\n31%\n\n30%\n\n0% 10% 20% 30% 40% 50% 60%\n\nExistence of smart city initiatives\n\nOpen government data sharing\n\nOpen government data policy online\n\nActual or intended use of emergent technologies\n\n2022 2024","publish_date":null,"title":"Figure 4.15 presents proxy indicators from the lOSI assessment that may offer some insight into the extent of digital readiness","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/%28Web_version%29_E-Government_Survey_2024_1392024_p177.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/%28Web_version%29_E-Government_Survey_2024_1392024_p177.pdf"},"request":{"example_id":"chart/(Web_version)_E-Government_Survey_2024_1392024_p177","product_type":"parse","source_file_path":"(Web_version)_E-Government_Survey_2024_1392024_p177.pdf"},"started_at":"2026-11-04T19:45:51.982161"} +{"completed_at":"2026-11-04T19:45:55.270577","latency_in_ms":1453,"output":{"example_id":"chart/(Web_version)_E-Government_Survey_2024_1392024_p62","grounded_pages":[],"layout_pages":[],"markdown":"37\n\n**Chapter 2**\n\nChapter 2 • Global trends in e-Government\n\nFigure 2.2 Number and percentage of countries in each EGDI group, 2014 to 2024\n\nFor the first time, Member States with very high EGDI values (above 0.75) comprise the largest share, accounting for 39 per cent of the total (76 of the 193 countries assessed). This group is followed by countries with high EGDI values (ranging from 0.50 to 0.75), which make up 32 per cent (62 countries). The number of countries with middle EGDI values (ranging from 0.25 to 0.50) has declined from 53 in 2022 to 44 (23 per cent) in 2024. However, the number of countries with low EGDI values has increased from 7 to 11 (6 per cent) since 2022, primarily due to geopolitical conflicts and post-conflict situations that have hindered their digital development. The overall progress achieved is consistent with the positive e-government development trend observed over the past decade, as illustrated in figure 2.2.\n\nThe number of countries with very high EGDI values has more than tripled over the past ten years, rising from 25 in 2014 to 76 in 2024. The combined number of countries with very high and high EGDI values has increased from 87 in 2014 to 138 in 2024. This underscores the growing importance and priority Governments have placed on digital transformation over the past decade. It highlights the significant strides made in enhancing e-government services and infrastructure, reflecting a global commitment to leveraging technology for improved governance and public services delivery. As a result of these gains, the number of countries with middle and low EGDI values has decreased significantly, falling from 106 in 2014 to 55 in 2024. This shift is broadening the base of nations that are strengthening their digital capabilities, ensuring that more citizens can benefit from the efficiencies and conveniences of digital government services.\n\nFigure 2.3 presents the 2024 global and regional average EGDI values and the percentage increases in these averages since 2022. Europe has the highest average EGDI value (0.8493), followed by Asia (0.6990), the Americas (0.6701), Oceania (0.5289), and Africa (0.4247). Asia has seen the sharpest increase in its average EGDI value (7.7 per cent), followed by Africa, (4.8 per cent), the Americas and Oceania (4.1 per cent), and Europe (2.3 per cent).\n\n32 (16.6%) 32 (16.6%) 15 (7.8%) 8 (4.1%) 7 (3.6%) 11 (5.7%)\n\n74 (38.3%)\n\n57 (29.5%) 65 (33.7%) 59 (30.6%) 53 (27.5%) 44 (22.8%)\n\n62 (32.1%) 75 (38.9%) 73 (37.8%)\n\n69 (35.8%) 73 (37.8%) 62 (32.1%)\n\n25 (13.0%) 29 (15.0%) 40 (20.7%) 57 (29.5%) 60 (31.1%) 76 (39.4%)\n\n0\n\n20\n\n40\n\n60\n\n80\n\n100\n\n120\n\n140\n\n160\n\n180\n\n200\n\n2014 2016 2018 2020 2022 2024\n\nLow EGDI Middle EGDI High EGDI Very high EGDI\n\n## Very high EGDI High EGDI Middle EGDI Low EGDI\n\n_Sources_ : 2014-2024 United Nations E-Government Surveys.","pages":[{"line_numbers":[],"markdown":"37\n\n**Chapter 2**\n\nChapter 2 • Global trends in e-Government\n\nFigure 2.2 Number and percentage of countries in each EGDI group, 2014 to 2024\n\nFor the first time, Member States with very high EGDI values (above 0.75) comprise the largest share, accounting for 39 per cent of the total (76 of the 193 countries assessed). This group is followed by countries with high EGDI values (ranging from 0.50 to 0.75), which make up 32 per cent (62 countries). The number of countries with middle EGDI values (ranging from 0.25 to 0.50) has declined from 53 in 2022 to 44 (23 per cent) in 2024. However, the number of countries with low EGDI values has increased from 7 to 11 (6 per cent) since 2022, primarily due to geopolitical conflicts and post-conflict situations that have hindered their digital development. The overall progress achieved is consistent with the positive e-government development trend observed over the past decade, as illustrated in figure 2.2.\n\nThe number of countries with very high EGDI values has more than tripled over the past ten years, rising from 25 in 2014 to 76 in 2024. The combined number of countries with very high and high EGDI values has increased from 87 in 2014 to 138 in 2024. This underscores the growing importance and priority Governments have placed on digital transformation over the past decade. It highlights the significant strides made in enhancing e-government services and infrastructure, reflecting a global commitment to leveraging technology for improved governance and public services delivery. As a result of these gains, the number of countries with middle and low EGDI values has decreased significantly, falling from 106 in 2014 to 55 in 2024. This shift is broadening the base of nations that are strengthening their digital capabilities, ensuring that more citizens can benefit from the efficiencies and conveniences of digital government services.\n\nFigure 2.3 presents the 2024 global and regional average EGDI values and the percentage increases in these averages since 2022. Europe has the highest average EGDI value (0.8493), followed by Asia (0.6990), the Americas (0.6701), Oceania (0.5289), and Africa (0.4247). Asia has seen the sharpest increase in its average EGDI value (7.7 per cent), followed by Africa, (4.8 per cent), the Americas and Oceania (4.1 per cent), and Europe (2.3 per cent).\n\n32 (16.6%) 32 (16.6%) 15 (7.8%) 8 (4.1%) 7 (3.6%) 11 (5.7%)\n\n74 (38.3%)\n\n57 (29.5%) 65 (33.7%) 59 (30.6%) 53 (27.5%) 44 (22.8%)\n\n62 (32.1%) 75 (38.9%) 73 (37.8%)\n\n69 (35.8%) 73 (37.8%) 62 (32.1%)\n\n25 (13.0%) 29 (15.0%) 40 (20.7%) 57 (29.5%) 60 (31.1%) 76 (39.4%)\n\n0\n\n20\n\n40\n\n60\n\n80\n\n100\n\n120\n\n140\n\n160\n\n180\n\n200\n\n2014 2016 2018 2020 2022 2024\n\nLow EGDI Middle EGDI High EGDI Very high EGDI\n\n## Very high EGDI High EGDI Middle EGDI Low EGDI\n\n_Sources_ : 2014-2024 United Nations E-Government Surveys.","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"37\n\n**Chapter 2**\n\nChapter 2 • Global trends in e-Government\n\nFigure 2.2 Number and percentage of countries in each EGDI group, 2014 to 2024\n\nFor the first time, Member States with very high EGDI values (above 0.75) comprise the largest share, accounting for 39 per cent of the total (76 of the 193 countries assessed). This group is followed by countries with high EGDI values (ranging from 0.50 to 0.75), which make up 32 per cent (62 countries). The number of countries with middle EGDI values (ranging from 0.25 to 0.50) has declined from 53 in 2022 to 44 (23 per cent) in 2024. However, the number of countries with low EGDI values has increased from 7 to 11 (6 per cent) since 2022, primarily due to geopolitical conflicts and post-conflict situations that have hindered their digital development. The overall progress achieved is consistent with the positive e-government development trend observed over the past decade, as illustrated in figure 2.2.\n\nThe number of countries with very high EGDI values has more than tripled over the past ten years, rising from 25 in 2014 to 76 in 2024. The combined number of countries with very high and high EGDI values has increased from 87 in 2014 to 138 in 2024. This underscores the growing importance and priority Governments have placed on digital transformation over the past decade. It highlights the significant strides made in enhancing e-government services and infrastructure, reflecting a global commitment to leveraging technology for improved governance and public services delivery. As a result of these gains, the number of countries with middle and low EGDI values has decreased significantly, falling from 106 in 2014 to 55 in 2024. This shift is broadening the base of nations that are strengthening their digital capabilities, ensuring that more citizens can benefit from the efficiencies and conveniences of digital government services.\n\nFigure 2.3 presents the 2024 global and regional average EGDI values and the percentage increases in these averages since 2022. Europe has the highest average EGDI value (0.8493), followed by Asia (0.6990), the Americas (0.6701), Oceania (0.5289), and Africa (0.4247). Asia has seen the sharpest increase in its average EGDI value (7.7 per cent), followed by Africa, (4.8 per cent), the Americas and Oceania (4.1 per cent), and Europe (2.3 per cent).\n\n32 (16.6%) 32 (16.6%) 15 (7.8%) 8 (4.1%) 7 (3.6%) 11 (5.7%)\n\n74 (38.3%)\n\n57 (29.5%) 65 (33.7%) 59 (30.6%) 53 (27.5%) 44 (22.8%)\n\n62 (32.1%) 75 (38.9%) 73 (37.8%)\n\n69 (35.8%) 73 (37.8%) 62 (32.1%)\n\n25 (13.0%) 29 (15.0%) 40 (20.7%) 57 (29.5%) 60 (31.1%) 76 (39.4%)\n\n0\n\n20\n\n40\n\n60\n\n80\n\n100\n\n120\n\n140\n\n160\n\n180\n\n200\n\n2014 2016 2018 2020 2022 2024\n\nLow EGDI Middle EGDI High EGDI Very high EGDI\n\n## Very high EGDI High EGDI Middle EGDI Low EGDI\n\n_Sources_ : 2014-2024 United Nations E-Government Surveys.","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/%28Web_version%29_E-Government_Survey_2024_1392024_p62.pdf"]},"response":{"errors":[],"extract_id":"extract_6fc6e4bba4555f11150bcd4908c430ab","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["37\n\n**Chapter 2**\n\nChapter 2 • Global trends in e-Government\n\nFigure 2.2 Number and percentage of countries in each EGDI group, 2014 to 2024\n\nFor the first time, Member States with very high EGDI values (above 0.75) comprise the largest share, accounting for 39 per cent of the total (76 of the 193 countries assessed). This group is followed by countries with high EGDI values (ranging from 0.50 to 0.75), which make up 32 per cent (62 countries). The number of countries with middle EGDI values (ranging from 0.25 to 0.50) has declined from 53 in 2022 to 44 (23 per cent) in 2024. However, the number of countries with low EGDI values has increased from 7 to 11 (6 per cent) since 2022, primarily due to geopolitical conflicts and post-conflict situations that have hindered their digital development. The overall progress achieved is consistent with the positive e-government development trend observed over the past decade, as illustrated in figure 2.2.\n\nThe number of countries with very high EGDI values has more than tripled over the past ten years, rising from 25 in 2014 to 76 in 2024. The combined number of countries with very high and high EGDI values has increased from 87 in 2014 to 138 in 2024. This underscores the growing importance and priority Governments have placed on digital transformation over the past decade. It highlights the significant strides made in enhancing e-government services and infrastructure, reflecting a global commitment to leveraging technology for improved governance and public services delivery. As a result of these gains, the number of countries with middle and low EGDI values has decreased significantly, falling from 106 in 2014 to 55 in 2024. This shift is broadening the base of nations that are strengthening their digital capabilities, ensuring that more citizens can benefit from the efficiencies and conveniences of digital government services.\n\nFigure 2.3 presents the 2024 global and regional average EGDI values and the percentage increases in these averages since 2022. Europe has the highest average EGDI value (0.8493), followed by Asia (0.6990), the Americas (0.6701), Oceania (0.5289), and Africa (0.4247). Asia has seen the sharpest increase in its average EGDI value (7.7 per cent), followed by Africa, (4.8 per cent), the Americas and Oceania (4.1 per cent), and Europe (2.3 per cent).\n\n32 (16.6%) 32 (16.6%) 15 (7.8%) 8 (4.1%) 7 (3.6%) 11 (5.7%)\n\n74 (38.3%)\n\n57 (29.5%) 65 (33.7%) 59 (30.6%) 53 (27.5%) 44 (22.8%)\n\n62 (32.1%) 75 (38.9%) 73 (37.8%)\n\n69 (35.8%) 73 (37.8%) 62 (32.1%)\n\n25 (13.0%) 29 (15.0%) 40 (20.7%) 57 (29.5%) 60 (31.1%) 76 (39.4%)\n\n0\n\n20\n\n40\n\n60\n\n80\n\n100\n\n120\n\n140\n\n160\n\n180\n\n200\n\n2014 2016 2018 2020 2022 2024\n\nLow EGDI Middle EGDI High EGDI Very high EGDI\n\n## Very high EGDI High EGDI Middle EGDI Low EGDI\n\n_Sources_ : 2014-2024 United Nations E-Government Surveys."],"full_content":"37\n\n**Chapter 2**\n\nChapter 2 • Global trends in e-Government\n\nFigure 2.2 Number and percentage of countries in each EGDI group, 2014 to 2024\n\nFor the first time, Member States with very high EGDI values (above 0.75) comprise the largest share, accounting for 39 per cent of the total (76 of the 193 countries assessed). This group is followed by countries with high EGDI values (ranging from 0.50 to 0.75), which make up 32 per cent (62 countries). The number of countries with middle EGDI values (ranging from 0.25 to 0.50) has declined from 53 in 2022 to 44 (23 per cent) in 2024. However, the number of countries with low EGDI values has increased from 7 to 11 (6 per cent) since 2022, primarily due to geopolitical conflicts and post-conflict situations that have hindered their digital development. The overall progress achieved is consistent with the positive e-government development trend observed over the past decade, as illustrated in figure 2.2.\n\nThe number of countries with very high EGDI values has more than tripled over the past ten years, rising from 25 in 2014 to 76 in 2024. The combined number of countries with very high and high EGDI values has increased from 87 in 2014 to 138 in 2024. This underscores the growing importance and priority Governments have placed on digital transformation over the past decade. It highlights the significant strides made in enhancing e-government services and infrastructure, reflecting a global commitment to leveraging technology for improved governance and public services delivery. As a result of these gains, the number of countries with middle and low EGDI values has decreased significantly, falling from 106 in 2014 to 55 in 2024. This shift is broadening the base of nations that are strengthening their digital capabilities, ensuring that more citizens can benefit from the efficiencies and conveniences of digital government services.\n\nFigure 2.3 presents the 2024 global and regional average EGDI values and the percentage increases in these averages since 2022. Europe has the highest average EGDI value (0.8493), followed by Asia (0.6990), the Americas (0.6701), Oceania (0.5289), and Africa (0.4247). Asia has seen the sharpest increase in its average EGDI value (7.7 per cent), followed by Africa, (4.8 per cent), the Americas and Oceania (4.1 per cent), and Europe (2.3 per cent).\n\n32 (16.6%) 32 (16.6%) 15 (7.8%) 8 (4.1%) 7 (3.6%) 11 (5.7%)\n\n74 (38.3%)\n\n57 (29.5%) 65 (33.7%) 59 (30.6%) 53 (27.5%) 44 (22.8%)\n\n62 (32.1%) 75 (38.9%) 73 (37.8%)\n\n69 (35.8%) 73 (37.8%) 62 (32.1%)\n\n25 (13.0%) 29 (15.0%) 40 (20.7%) 57 (29.5%) 60 (31.1%) 76 (39.4%)\n\n0\n\n20\n\n40\n\n60\n\n80\n\n100\n\n120\n\n140\n\n160\n\n180\n\n200\n\n2014 2016 2018 2020 2022 2024\n\nLow EGDI Middle EGDI High EGDI Very high EGDI\n\n## Very high EGDI High EGDI Middle EGDI Low EGDI\n\n_Sources_ : 2014-2024 United Nations E-Government Surveys.","publish_date":null,"title":"Very high EGDI High EGDI Middle EGDI Low EGDI","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/%28Web_version%29_E-Government_Survey_2024_1392024_p62.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/%28Web_version%29_E-Government_Survey_2024_1392024_p62.pdf"},"request":{"example_id":"chart/(Web_version)_E-Government_Survey_2024_1392024_p62","product_type":"parse","source_file_path":"(Web_version)_E-Government_Survey_2024_1392024_p62.pdf"},"started_at":"2026-11-04T19:45:53.816818"} +{"completed_at":"2026-11-04T19:45:56.856360","latency_in_ms":1446,"output":{"example_id":"chart/(Web_version)_E-Government_Survey_2024_1392024_p63","grounded_pages":[],"layout_pages":[],"markdown":"2024 UN E-GovErNmENt SUrvEy\n\n38\n\n**Chapter 2**\n\nIt is noteworthy that the TII is the highest component index contributing to average EGDI values globally and regionally. This reflects the increased investment in digital infrastructure during the post- cOVID-19 pandemic recovery process. Over the past two years, the average TII value has increased by 19.9 per cent globally (see table 2.1). The most significant regional increase is observed in Oceania (29.4 per cent), followed by Africa (27.8 per cent), Asia (25.5 per cent), the Americas (19.6 per cent), and Europe (9.9 per cent). These increases highlight the global emphasis on establishing a strong telecommunications infrastructure as a foundation for digital growth.\n\nFigure 2.3 Global and regional EGDI averages, 2022 and 2024\n\n0\\.6102 0.4054 0.6438 0.6493 0.8305 0.5081\n\n0\\.6382\n\n0\\.4247\n\n0\\.6701\n\n0\\.8493\n\n0\\.5289\n\n4\\.6%\n\n4\\.8%\n\n4\\.1% 7.7%\n\n2\\.3%\n\n4\\.1%\n\n0\n\n0\\.25\n\n0\\.5\n\n0\\.75\n\n1\n\n193 UN Member States Africa Americas Asia Europe Oceania\n\n2022\n\n2024 for 193 UN Member States, Africa, Americas, Asia, Europe and Oceania Linear (global EGDI average 2024)\n\n0\\.6990\n\n_Sources_ : 2022 and 2024 United Nations E-Government Surveys.\n\nTable 2.1 Average global and regional values for the EGDI and its component indices, 2022 and 2024\n\nAverage values for: EGDI OSI TII HCI\n\n**193 UN Member**\n\n**States**\n\n**2024 0.6382 4.59% 0.5754 3.6% 0.6896 19.9% 0.6494 -7.2%**\n\n**2022 0.6102 0.5554 0.5751 0.7001**\n\nAfrica 2024 0.4247 4.8% 0.3862 5.2% 0.4534 27.8% 0.4346 -12.1%\n\n2022 0.4054 0.3670 0.3548 0.4945\n\nAmericas 2024 0.6701 4.1% 0.5797 3.8% 0.7345 19.6% 0.6962 -8.3%\n\n2022 0.6438 0.5585 0.6139 0.7590\n\nAsia 2024 0.6990 7.7% 0.6401 4.3% 0.7740 25.5% 0.6828 -4.8%\n\n2022 0.6493 0.6137 0.6166 0.7175\n\nEurope 2024 0.8493 2.3% 0.7836 1.8% 0.9227 9.9% 0.8418 -4.6%\n\n2022 0.8305 0.7699 0.8392 0.8825\n\nOceania 2024 0.5289 4.1% 0.4378 4.2% 0.4885 29.4% 0.6603 -9.5%\n\n2022 0.5081 0.4201 0.3775 0.7298\n\n_Sources_ : 2022 and 2024 United Nations E-Government Surveys.","pages":[{"line_numbers":[],"markdown":"2024 UN E-GovErNmENt SUrvEy\n\n38\n\n**Chapter 2**\n\nIt is noteworthy that the TII is the highest component index contributing to average EGDI values globally and regionally. This reflects the increased investment in digital infrastructure during the post- cOVID-19 pandemic recovery process. Over the past two years, the average TII value has increased by 19.9 per cent globally (see table 2.1). The most significant regional increase is observed in Oceania (29.4 per cent), followed by Africa (27.8 per cent), Asia (25.5 per cent), the Americas (19.6 per cent), and Europe (9.9 per cent). These increases highlight the global emphasis on establishing a strong telecommunications infrastructure as a foundation for digital growth.\n\nFigure 2.3 Global and regional EGDI averages, 2022 and 2024\n\n0\\.6102 0.4054 0.6438 0.6493 0.8305 0.5081\n\n0\\.6382\n\n0\\.4247\n\n0\\.6701\n\n0\\.8493\n\n0\\.5289\n\n4\\.6%\n\n4\\.8%\n\n4\\.1% 7.7%\n\n2\\.3%\n\n4\\.1%\n\n0\n\n0\\.25\n\n0\\.5\n\n0\\.75\n\n1\n\n193 UN Member States Africa Americas Asia Europe Oceania\n\n2022\n\n2024 for 193 UN Member States, Africa, Americas, Asia, Europe and Oceania Linear (global EGDI average 2024)\n\n0\\.6990\n\n_Sources_ : 2022 and 2024 United Nations E-Government Surveys.\n\nTable 2.1 Average global and regional values for the EGDI and its component indices, 2022 and 2024\n\nAverage values for: EGDI OSI TII HCI\n\n**193 UN Member**\n\n**States**\n\n**2024 0.6382 4.59% 0.5754 3.6% 0.6896 19.9% 0.6494 -7.2%**\n\n**2022 0.6102 0.5554 0.5751 0.7001**\n\nAfrica 2024 0.4247 4.8% 0.3862 5.2% 0.4534 27.8% 0.4346 -12.1%\n\n2022 0.4054 0.3670 0.3548 0.4945\n\nAmericas 2024 0.6701 4.1% 0.5797 3.8% 0.7345 19.6% 0.6962 -8.3%\n\n2022 0.6438 0.5585 0.6139 0.7590\n\nAsia 2024 0.6990 7.7% 0.6401 4.3% 0.7740 25.5% 0.6828 -4.8%\n\n2022 0.6493 0.6137 0.6166 0.7175\n\nEurope 2024 0.8493 2.3% 0.7836 1.8% 0.9227 9.9% 0.8418 -4.6%\n\n2022 0.8305 0.7699 0.8392 0.8825\n\nOceania 2024 0.5289 4.1% 0.4378 4.2% 0.4885 29.4% 0.6603 -9.5%\n\n2022 0.5081 0.4201 0.3775 0.7298\n\n_Sources_ : 2022 and 2024 United Nations E-Government Surveys.","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"2024 UN E-GovErNmENt SUrvEy\n\n38\n\n**Chapter 2**\n\nIt is noteworthy that the TII is the highest component index contributing to average EGDI values globally and regionally. This reflects the increased investment in digital infrastructure during the post- cOVID-19 pandemic recovery process. Over the past two years, the average TII value has increased by 19.9 per cent globally (see table 2.1). The most significant regional increase is observed in Oceania (29.4 per cent), followed by Africa (27.8 per cent), Asia (25.5 per cent), the Americas (19.6 per cent), and Europe (9.9 per cent). These increases highlight the global emphasis on establishing a strong telecommunications infrastructure as a foundation for digital growth.\n\nFigure 2.3 Global and regional EGDI averages, 2022 and 2024\n\n0\\.6102 0.4054 0.6438 0.6493 0.8305 0.5081\n\n0\\.6382\n\n0\\.4247\n\n0\\.6701\n\n0\\.8493\n\n0\\.5289\n\n4\\.6%\n\n4\\.8%\n\n4\\.1% 7.7%\n\n2\\.3%\n\n4\\.1%\n\n0\n\n0\\.25\n\n0\\.5\n\n0\\.75\n\n1\n\n193 UN Member States Africa Americas Asia Europe Oceania\n\n2022\n\n2024 for 193 UN Member States, Africa, Americas, Asia, Europe and Oceania Linear (global EGDI average 2024)\n\n0\\.6990\n\n_Sources_ : 2022 and 2024 United Nations E-Government Surveys.\n\nTable 2.1 Average global and regional values for the EGDI and its component indices, 2022 and 2024\n\nAverage values for: EGDI OSI TII HCI\n\n**193 UN Member**\n\n**States**\n\n**2024 0.6382 4.59% 0.5754 3.6% 0.6896 19.9% 0.6494 -7.2%**\n\n**2022 0.6102 0.5554 0.5751 0.7001**\n\nAfrica 2024 0.4247 4.8% 0.3862 5.2% 0.4534 27.8% 0.4346 -12.1%\n\n2022 0.4054 0.3670 0.3548 0.4945\n\nAmericas 2024 0.6701 4.1% 0.5797 3.8% 0.7345 19.6% 0.6962 -8.3%\n\n2022 0.6438 0.5585 0.6139 0.7590\n\nAsia 2024 0.6990 7.7% 0.6401 4.3% 0.7740 25.5% 0.6828 -4.8%\n\n2022 0.6493 0.6137 0.6166 0.7175\n\nEurope 2024 0.8493 2.3% 0.7836 1.8% 0.9227 9.9% 0.8418 -4.6%\n\n2022 0.8305 0.7699 0.8392 0.8825\n\nOceania 2024 0.5289 4.1% 0.4378 4.2% 0.4885 29.4% 0.6603 -9.5%\n\n2022 0.5081 0.4201 0.3775 0.7298\n\n_Sources_ : 2022 and 2024 United Nations E-Government Surveys.","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/%28Web_version%29_E-Government_Survey_2024_1392024_p63.pdf"]},"response":{"errors":[],"extract_id":"extract_ece0147d2cbddaff9d6158b0995e20d9","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["2024 UN E-GovErNmENt SUrvEy\n\n38\n\n**Chapter 2**\n\nIt is noteworthy that the TII is the highest component index contributing to average EGDI values globally and regionally. This reflects the increased investment in digital infrastructure during the post- cOVID-19 pandemic recovery process. Over the past two years, the average TII value has increased by 19.9 per cent globally (see table 2.1). The most significant regional increase is observed in Oceania (29.4 per cent), followed by Africa (27.8 per cent), Asia (25.5 per cent), the Americas (19.6 per cent), and Europe (9.9 per cent). These increases highlight the global emphasis on establishing a strong telecommunications infrastructure as a foundation for digital growth.\n\nFigure 2.3 Global and regional EGDI averages, 2022 and 2024\n\n0\\.6102 0.4054 0.6438 0.6493 0.8305 0.5081\n\n0\\.6382\n\n0\\.4247\n\n0\\.6701\n\n0\\.8493\n\n0\\.5289\n\n4\\.6%\n\n4\\.8%\n\n4\\.1% 7.7%\n\n2\\.3%\n\n4\\.1%\n\n0\n\n0\\.25\n\n0\\.5\n\n0\\.75\n\n1\n\n193 UN Member States Africa Americas Asia Europe Oceania\n\n2022\n\n2024 for 193 UN Member States, Africa, Americas, Asia, Europe and Oceania Linear (global EGDI average 2024)\n\n0\\.6990\n\n_Sources_ : 2022 and 2024 United Nations E-Government Surveys.\n\nTable 2.1 Average global and regional values for the EGDI and its component indices, 2022 and 2024\n\nAverage values for: EGDI OSI TII HCI\n\n**193 UN Member**\n\n**States**\n\n**2024 0.6382 4.59% 0.5754 3.6% 0.6896 19.9% 0.6494 -7.2%**\n\n**2022 0.6102 0.5554 0.5751 0.7001**\n\nAfrica 2024 0.4247 4.8% 0.3862 5.2% 0.4534 27.8% 0.4346 -12.1%\n\n2022 0.4054 0.3670 0.3548 0.4945\n\nAmericas 2024 0.6701 4.1% 0.5797 3.8% 0.7345 19.6% 0.6962 -8.3%\n\n2022 0.6438 0.5585 0.6139 0.7590\n\nAsia 2024 0.6990 7.7% 0.6401 4.3% 0.7740 25.5% 0.6828 -4.8%\n\n2022 0.6493 0.6137 0.6166 0.7175\n\nEurope 2024 0.8493 2.3% 0.7836 1.8% 0.9227 9.9% 0.8418 -4.6%\n\n2022 0.8305 0.7699 0.8392 0.8825\n\nOceania 2024 0.5289 4.1% 0.4378 4.2% 0.4885 29.4% 0.6603 -9.5%\n\n2022 0.5081 0.4201 0.3775 0.7298\n\n_Sources_ : 2022 and 2024 United Nations E-Government Surveys."],"full_content":"2024 UN E-GovErNmENt SUrvEy\n\n38\n\n**Chapter 2**\n\nIt is noteworthy that the TII is the highest component index contributing to average EGDI values globally and regionally. This reflects the increased investment in digital infrastructure during the post- cOVID-19 pandemic recovery process. Over the past two years, the average TII value has increased by 19.9 per cent globally (see table 2.1). The most significant regional increase is observed in Oceania (29.4 per cent), followed by Africa (27.8 per cent), Asia (25.5 per cent), the Americas (19.6 per cent), and Europe (9.9 per cent). These increases highlight the global emphasis on establishing a strong telecommunications infrastructure as a foundation for digital growth.\n\nFigure 2.3 Global and regional EGDI averages, 2022 and 2024\n\n0\\.6102 0.4054 0.6438 0.6493 0.8305 0.5081\n\n0\\.6382\n\n0\\.4247\n\n0\\.6701\n\n0\\.8493\n\n0\\.5289\n\n4\\.6%\n\n4\\.8%\n\n4\\.1% 7.7%\n\n2\\.3%\n\n4\\.1%\n\n0\n\n0\\.25\n\n0\\.5\n\n0\\.75\n\n1\n\n193 UN Member States Africa Americas Asia Europe Oceania\n\n2022\n\n2024 for 193 UN Member States, Africa, Americas, Asia, Europe and Oceania Linear (global EGDI average 2024)\n\n0\\.6990\n\n_Sources_ : 2022 and 2024 United Nations E-Government Surveys.\n\nTable 2.1 Average global and regional values for the EGDI and its component indices, 2022 and 2024\n\nAverage values for: EGDI OSI TII HCI\n\n**193 UN Member**\n\n**States**\n\n**2024 0.6382 4.59% 0.5754 3.6% 0.6896 19.9% 0.6494 -7.2%**\n\n**2022 0.6102 0.5554 0.5751 0.7001**\n\nAfrica 2024 0.4247 4.8% 0.3862 5.2% 0.4534 27.8% 0.4346 -12.1%\n\n2022 0.4054 0.3670 0.3548 0.4945\n\nAmericas 2024 0.6701 4.1% 0.5797 3.8% 0.7345 19.6% 0.6962 -8.3%\n\n2022 0.6438 0.5585 0.6139 0.7590\n\nAsia 2024 0.6990 7.7% 0.6401 4.3% 0.7740 25.5% 0.6828 -4.8%\n\n2022 0.6493 0.6137 0.6166 0.7175\n\nEurope 2024 0.8493 2.3% 0.7836 1.8% 0.9227 9.9% 0.8418 -4.6%\n\n2022 0.8305 0.7699 0.8392 0.8825\n\nOceania 2024 0.5289 4.1% 0.4378 4.2% 0.4885 29.4% 0.6603 -9.5%\n\n2022 0.5081 0.4201 0.3775 0.7298\n\n_Sources_ : 2022 and 2024 United Nations E-Government Surveys.","publish_date":null,"title":"Very high EGDI High EGDI Middle EGDI Low EGDI","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/%28Web_version%29_E-Government_Survey_2024_1392024_p63.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/%28Web_version%29_E-Government_Survey_2024_1392024_p63.pdf"},"request":{"example_id":"chart/(Web_version)_E-Government_Survey_2024_1392024_p63","product_type":"parse","source_file_path":"(Web_version)_E-Government_Survey_2024_1392024_p63.pdf"},"started_at":"2026-11-04T19:45:55.409449"} +{"completed_at":"2026-11-04T19:45:59.482592","latency_in_ms":2529,"output":{"example_id":"chart/(Web_version)_E-Government_Survey_2024_1392024_p65","grounded_pages":[],"layout_pages":[],"markdown":"2024 UN E-GovErNmENt SUrvEy\n\n40\n\n**Chapter 2**\n\nAmong the countries with high EGDI values, 20 are in the Americas, 17 are in Africa, 14 are in Asia, 7 are in Europe, and 4 are in Oceania. The majority of countries with middle EGDI values are in Africa (28), followed by Oceania (8), Asia (5), and the Americas (3). Among the 11 countries with low EGDI values, 7 are in Africa, 3 are in Asia, and 1 is in the Americas.\n\nAll of the countries with low EGDI values in 2022 are still in the same group in 2024, and four other countries have moved from the middle to the low EGDI group.\n\nA more detailed analysis of e-government development at the regional level is provided in chapter 3. Section 2.6 of this chapter includes an analysis of countries in special situations.\n\nMovement between EGDI groups and rating classes\n\nTwenty-three countries have moved to a higher EGDI group. Ten countries in Asia (Armenia, Azerbaijan, Brunei Darussalam, Indonesia, Kuwait, Mongolia, Philippines, Qatar, Uzbekistan, and Viet Nam), three countries in the Americas (colombia, Ecuador, and Mexico), two countries in Africa (Mauritius and South Africa), and two countries in Europe (Albania and the Republic of Moldova) have moved from the high to the very high EGDI group (see figure 2.5). Three countries in Africa (Eswatini, libya, and Senegal), two in Asia (Myanmar and Pakistan), and one in Oceania (Vanuatu) have moved from the middle to the high EGDI group.\n\nFigure 2.5 Movement of countries between EGDI groups during the period 2022-2024, by region\n\n_(Number of countries)_\n\n3\n\n2\n\n1\n\n2\n\n3\n\n10\n\n2 -1\n\n\\-1\n\n\\-1\n\n\\-3\n\n\\-4 -2 0 2 4 6 8 10 12 14\n\nAfrica\n\nAmericas\n\nAsia\n\nEurope\n\nOceania\n\nMiddle to high High to very high Very high to high High to middle Middle to low\n\n_Source_ : 2024 United Nations E-Government Survey.\n\nThese countries have made significant progress in their digital transformation journeys, focusing on enhancing their telecommunications infrastructure, improving online services, and investing in human capital development. They have implemented various strategies and initiatives to boost their digital capabilities, such as expanding broadband access, developing e-government platforms, and promoting digital literacy among their populations. These efforts have been instrumental in their advancement and upward movement in the EGDI rankings.","pages":[{"line_numbers":[],"markdown":"2024 UN E-GovErNmENt SUrvEy\n\n40\n\n**Chapter 2**\n\nAmong the countries with high EGDI values, 20 are in the Americas, 17 are in Africa, 14 are in Asia, 7 are in Europe, and 4 are in Oceania. The majority of countries with middle EGDI values are in Africa (28), followed by Oceania (8), Asia (5), and the Americas (3). Among the 11 countries with low EGDI values, 7 are in Africa, 3 are in Asia, and 1 is in the Americas.\n\nAll of the countries with low EGDI values in 2022 are still in the same group in 2024, and four other countries have moved from the middle to the low EGDI group.\n\nA more detailed analysis of e-government development at the regional level is provided in chapter 3. Section 2.6 of this chapter includes an analysis of countries in special situations.\n\nMovement between EGDI groups and rating classes\n\nTwenty-three countries have moved to a higher EGDI group. Ten countries in Asia (Armenia, Azerbaijan, Brunei Darussalam, Indonesia, Kuwait, Mongolia, Philippines, Qatar, Uzbekistan, and Viet Nam), three countries in the Americas (colombia, Ecuador, and Mexico), two countries in Africa (Mauritius and South Africa), and two countries in Europe (Albania and the Republic of Moldova) have moved from the high to the very high EGDI group (see figure 2.5). Three countries in Africa (Eswatini, libya, and Senegal), two in Asia (Myanmar and Pakistan), and one in Oceania (Vanuatu) have moved from the middle to the high EGDI group.\n\nFigure 2.5 Movement of countries between EGDI groups during the period 2022-2024, by region\n\n_(Number of countries)_\n\n3\n\n2\n\n1\n\n2\n\n3\n\n10\n\n2 -1\n\n\\-1\n\n\\-1\n\n\\-3\n\n\\-4 -2 0 2 4 6 8 10 12 14\n\nAfrica\n\nAmericas\n\nAsia\n\nEurope\n\nOceania\n\nMiddle to high High to very high Very high to high High to middle Middle to low\n\n_Source_ : 2024 United Nations E-Government Survey.\n\nThese countries have made significant progress in their digital transformation journeys, focusing on enhancing their telecommunications infrastructure, improving online services, and investing in human capital development. They have implemented various strategies and initiatives to boost their digital capabilities, such as expanding broadband access, developing e-government platforms, and promoting digital literacy among their populations. These efforts have been instrumental in their advancement and upward movement in the EGDI rankings.","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"2024 UN E-GovErNmENt SUrvEy\n\n40\n\n**Chapter 2**\n\nAmong the countries with high EGDI values, 20 are in the Americas, 17 are in Africa, 14 are in Asia, 7 are in Europe, and 4 are in Oceania. The majority of countries with middle EGDI values are in Africa (28), followed by Oceania (8), Asia (5), and the Americas (3). Among the 11 countries with low EGDI values, 7 are in Africa, 3 are in Asia, and 1 is in the Americas.\n\nAll of the countries with low EGDI values in 2022 are still in the same group in 2024, and four other countries have moved from the middle to the low EGDI group.\n\nA more detailed analysis of e-government development at the regional level is provided in chapter 3. Section 2.6 of this chapter includes an analysis of countries in special situations.\n\nMovement between EGDI groups and rating classes\n\nTwenty-three countries have moved to a higher EGDI group. Ten countries in Asia (Armenia, Azerbaijan, Brunei Darussalam, Indonesia, Kuwait, Mongolia, Philippines, Qatar, Uzbekistan, and Viet Nam), three countries in the Americas (colombia, Ecuador, and Mexico), two countries in Africa (Mauritius and South Africa), and two countries in Europe (Albania and the Republic of Moldova) have moved from the high to the very high EGDI group (see figure 2.5). Three countries in Africa (Eswatini, libya, and Senegal), two in Asia (Myanmar and Pakistan), and one in Oceania (Vanuatu) have moved from the middle to the high EGDI group.\n\nFigure 2.5 Movement of countries between EGDI groups during the period 2022-2024, by region\n\n_(Number of countries)_\n\n3\n\n2\n\n1\n\n2\n\n3\n\n10\n\n2 -1\n\n\\-1\n\n\\-1\n\n\\-3\n\n\\-4 -2 0 2 4 6 8 10 12 14\n\nAfrica\n\nAmericas\n\nAsia\n\nEurope\n\nOceania\n\nMiddle to high High to very high Very high to high High to middle Middle to low\n\n_Source_ : 2024 United Nations E-Government Survey.\n\nThese countries have made significant progress in their digital transformation journeys, focusing on enhancing their telecommunications infrastructure, improving online services, and investing in human capital development. They have implemented various strategies and initiatives to boost their digital capabilities, such as expanding broadband access, developing e-government platforms, and promoting digital literacy among their populations. These efforts have been instrumental in their advancement and upward movement in the EGDI rankings.","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/%28Web_version%29_E-Government_Survey_2024_1392024_p65.pdf"]},"response":{"errors":[],"extract_id":"extract_4186c0f9462e6919044fb4a22bb90fee","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["2024 UN E-GovErNmENt SUrvEy\n\n40\n\n**Chapter 2**\n\nAmong the countries with high EGDI values, 20 are in the Americas, 17 are in Africa, 14 are in Asia, 7 are in Europe, and 4 are in Oceania. The majority of countries with middle EGDI values are in Africa (28), followed by Oceania (8), Asia (5), and the Americas (3). Among the 11 countries with low EGDI values, 7 are in Africa, 3 are in Asia, and 1 is in the Americas.\n\nAll of the countries with low EGDI values in 2022 are still in the same group in 2024, and four other countries have moved from the middle to the low EGDI group.\n\nA more detailed analysis of e-government development at the regional level is provided in chapter 3. Section 2.6 of this chapter includes an analysis of countries in special situations.\n\nMovement between EGDI groups and rating classes\n\nTwenty-three countries have moved to a higher EGDI group. Ten countries in Asia (Armenia, Azerbaijan, Brunei Darussalam, Indonesia, Kuwait, Mongolia, Philippines, Qatar, Uzbekistan, and Viet Nam), three countries in the Americas (colombia, Ecuador, and Mexico), two countries in Africa (Mauritius and South Africa), and two countries in Europe (Albania and the Republic of Moldova) have moved from the high to the very high EGDI group (see figure 2.5). Three countries in Africa (Eswatini, libya, and Senegal), two in Asia (Myanmar and Pakistan), and one in Oceania (Vanuatu) have moved from the middle to the high EGDI group.\n\nFigure 2.5 Movement of countries between EGDI groups during the period 2022-2024, by region\n\n_(Number of countries)_\n\n3\n\n2\n\n1\n\n2\n\n3\n\n10\n\n2 -1\n\n\\-1\n\n\\-1\n\n\\-3\n\n\\-4 -2 0 2 4 6 8 10 12 14\n\nAfrica\n\nAmericas\n\nAsia\n\nEurope\n\nOceania\n\nMiddle to high High to very high Very high to high High to middle Middle to low\n\n_Source_ : 2024 United Nations E-Government Survey.\n\nThese countries have made significant progress in their digital transformation journeys, focusing on enhancing their telecommunications infrastructure, improving online services, and investing in human capital development. They have implemented various strategies and initiatives to boost their digital capabilities, such as expanding broadband access, developing e-government platforms, and promoting digital literacy among their populations. These efforts have been instrumental in their advancement and upward movement in the EGDI rankings."],"full_content":"2024 UN E-GovErNmENt SUrvEy\n\n40\n\n**Chapter 2**\n\nAmong the countries with high EGDI values, 20 are in the Americas, 17 are in Africa, 14 are in Asia, 7 are in Europe, and 4 are in Oceania. The majority of countries with middle EGDI values are in Africa (28), followed by Oceania (8), Asia (5), and the Americas (3). Among the 11 countries with low EGDI values, 7 are in Africa, 3 are in Asia, and 1 is in the Americas.\n\nAll of the countries with low EGDI values in 2022 are still in the same group in 2024, and four other countries have moved from the middle to the low EGDI group.\n\nA more detailed analysis of e-government development at the regional level is provided in chapter 3. Section 2.6 of this chapter includes an analysis of countries in special situations.\n\nMovement between EGDI groups and rating classes\n\nTwenty-three countries have moved to a higher EGDI group. Ten countries in Asia (Armenia, Azerbaijan, Brunei Darussalam, Indonesia, Kuwait, Mongolia, Philippines, Qatar, Uzbekistan, and Viet Nam), three countries in the Americas (colombia, Ecuador, and Mexico), two countries in Africa (Mauritius and South Africa), and two countries in Europe (Albania and the Republic of Moldova) have moved from the high to the very high EGDI group (see figure 2.5). Three countries in Africa (Eswatini, libya, and Senegal), two in Asia (Myanmar and Pakistan), and one in Oceania (Vanuatu) have moved from the middle to the high EGDI group.\n\nFigure 2.5 Movement of countries between EGDI groups during the period 2022-2024, by region\n\n_(Number of countries)_\n\n3\n\n2\n\n1\n\n2\n\n3\n\n10\n\n2 -1\n\n\\-1\n\n\\-1\n\n\\-3\n\n\\-4 -2 0 2 4 6 8 10 12 14\n\nAfrica\n\nAmericas\n\nAsia\n\nEurope\n\nOceania\n\nMiddle to high High to very high Very high to high High to middle Middle to low\n\n_Source_ : 2024 United Nations E-Government Survey.\n\nThese countries have made significant progress in their digital transformation journeys, focusing on enhancing their telecommunications infrastructure, improving online services, and investing in human capital development. They have implemented various strategies and initiatives to boost their digital capabilities, such as expanding broadband access, developing e-government platforms, and promoting digital literacy among their populations. These efforts have been instrumental in their advancement and upward movement in the EGDI rankings.","publish_date":null,"title":"Very high EGDI High EGDI Middle EGDI Low EGDI","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/%28Web_version%29_E-Government_Survey_2024_1392024_p65.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/%28Web_version%29_E-Government_Survey_2024_1392024_p65.pdf"},"request":{"example_id":"chart/(Web_version)_E-Government_Survey_2024_1392024_p65","product_type":"parse","source_file_path":"(Web_version)_E-Government_Survey_2024_1392024_p65.pdf"},"started_at":"2026-11-04T19:45:56.952957"} +{"completed_at":"2026-11-04T19:46:04.277459","latency_in_ms":6002,"output":{"example_id":"chart/(Web_version)_E-Government_Survey_2024_1392024_p77","grounded_pages":[],"layout_pages":[],"markdown":"2024 UN E-GovErNmENt SUrvEy\n\n52\n\n**Chapter 2**\n\n### **2\\.5.2 OSI subindices**\n\nThe subsections above have highlighted areas that require further attention if countries wish to strengthen overall e-government development (as reflected in their respective EGDI levels). In cases where improvements in online services provision (OSI values) are necessary, it is important to understand which specific aspects of this component require attention. The main Survey findings for each of the five OSI subindices are presented below.\n\nAs shown in figure 2.9, the institutional framework (IF) is better developed than other aspects of online services provision in all regions. content provision (cP) has the next highest subindex value, followed by services provision (SP), technical characteristics (TEc) and e-participation (EPI). At the regional level, Europe is leading in terms of average OSI and subindex values, followed by Asia, the Americas, Oceania, and Africa – the one exception being that the technical aspects of the OSI are slightly better developed in Africa than in Oceania.\n\nFigure 2.9 Values for OSI subindices (IF, CP, SP, TEC and EPI) at the regional and global levels, 2024\n\nLegend:\n\nLinear (global OSI average)\n\nLinear (average OSI for Africa)\n\nLinear (average OSI for Asia) Linear (average OSI for Europe) Linear (average OSI for Oaceania)\n\n193 UN Member States\n\nAfrica\n\nAsia Europe Oceania\n\nLinear (average OSI for the Americas) Americas\n\n193 UN Member States Linear (global average OSI) Africa\n\nLegend:\n\nLinear (global OSI average)\n\nLinear (average OSI for Africa)\n\nLinear (average OSI for Asia) Linear (average OSI for Europe) Linear (average OSI for Oaceania)\n\n193 UN Member States\n\nAfrica\n\nAsia Europe Oceania\n\nLinear (average OSI for the Americas) Americas\n\n0 6644\n\n0\\.8079\n\nCP\n\nIF\n\n193 UN Member States Linear (global average OSI)\n\n_Source_ : 2024 United Nations E-Government Survey.\n\nLegend:\n\nLinear (global OSI average)\n\nLinear (average OSI for Africa)\n\nLinear (average OSI for Asia) Linear (average OSI for Europe) Linear (average OSI for Oaceania)\n\n193 UN Member States\n\nAfrica\n\nAsia Europe Oceania\n\nLinear (average OSI for the Americas) Americas\n\n0\\.4919\n\n0\\.5363\n\n0\\.5606\n\n0\\.6644\n\n0\\.8079\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\n193 UN Member States Linear (global average OSI)\n\n0\\.4736\n\n0\\.5411\n\n0\\.5797\n\n0\\.6730\n\n0\\.8171\n\n0\\.5797\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nAmericas\n\n0\\.6642\n\n0\\.7263\n\n0\\.7688\n\n0\\.8501\n\n0\\.9395\n\n0\\.7836 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nEurope\n\n0\\.3001\n\n0\\.4167\n\n0\\.3583\n\n0\\.4815\n\n0\\.6852\n\n0\\.3862 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nAfrica\n\n0\\.5462\n\n0\\.5984\n\n0\\.6311\n\n0\\.7471\n\n0\\.8562\n\n0\\.6401\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nAsia\n\n0\\.3748\n\n0\\.3839\n\n0\\.4174\n\n0\\.5000\n\n0\\.6914\n\n0\\.4378 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nOceania\n\nLegend:\n\nLinear (global OSI average)\n\nLinear (average OSI for Africa)\n\nLinear (average OSI for Asia) Linear (average OSI for Europe) Linear (average OSI for Oaceania)\n\n193 UN Member States\n\nAfrica\n\nAsia Europe Oceania\n\nLinear (average OSI for the Americas) Americas\n\n0\\.4919\n\n0\\.5363\n\n0\\.5606\n\n0\\.6644\n\n0\\.8079\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\n193 UN Member States Linear (global average OSI)\n\n0\\.4736\n\n0\\.5411\n\n0\\.5797\n\n0\\.6730\n\n0\\.8171\n\n0\\.5797\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nAmericas\n\n0\\.6642\n\n0\\.7263\n\n0\\.7688\n\n0\\.8501\n\n0\\.9395\n\n0\\.7836 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nEurope\n\n0\\.3001\n\n0\\.4167\n\n0\\.3583\n\n0\\.4815\n\n0\\.6852\n\n0\\.3862 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nAfrica\n\n0\\.5462\n\n0\\.5984\n\n0\\.6311\n\n0\\.7471\n\n0\\.8562\n\n0\\.6401\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nAsia\n\n0\\.3748\n\n0\\.3839\n\n0\\.4174\n\n0\\.5000\n\n0\\.6914\n\n0\\.4378 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nOceania\n\nLegend:\n\nLinear (global OSI average)\n\nLinear (average OSI for Africa)\n\nLinear (average OSI for Asia) Linear (average OSI for Europe) Linear (average OSI for Oaceania)\n\n193 UN Member States\n\nAfrica\n\nAsia Europe Oceania\n\nLinear (average OSI for the Americas) Americas\n\n0\\.4919\n\n0\\.5363\n\n0\\.5606\n\n0\\.6644\n\n0\\.8079\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\n193 UN Member States Linear (global average OSI)\n\n0\\.4736\n\n0\\.5411\n\n0\\.5797\n\n0\\.6730\n\n0\\.8171\n\n0\\.5797\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nAmericas\n\n0\\.6642\n\n0\\.7263\n\n0\\.7688\n\n0\\.8501\n\n0\\.9395\n\n0\\.7836 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nEurope\n\n0\\.3001\n\n0\\.4167\n\n0\\.3583\n\n0\\.4815\n\n0\\.6852\n\n0\\.3862 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nAfrica\n\n0\\.5462\n\n0\\.5984\n\n0\\.6311\n\n0\\.7471\n\n0\\.8562\n\n0\\.6401\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nAsia\n\n0\\.3748\n\n0\\.3839\n\n0\\.4174\n\n0\\.5000\n\n0\\.6914\n\n0\\.4378 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nOceania\n\nLegend:\n\nLinear (global OSI average)\n\nLinear (average OSI for Africa)\n\nLinear (average OSI for Asia) Linear (average OSI for Europe) Linear (average OSI for Oaceania)\n\n193 UN Member States\n\nAfrica\n\nAsia Europe Oceania\n\nLinear (average OSI for the Americas) Americas\n\n0\\.4919\n\n0\\.5363\n\n0\\.5606\n\n0\\.6644\n\n0\\.8079\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\n193 UN Member States Linear (global average OSI)\n\n0\\.4736\n\n0\\.5411\n\n0\\.5797\n\n0\\.6730\n\n0\\.8171\n\n0\\.5797\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nAmericas\n\n0\\.6642\n\n0\\.7263\n\n0\\.7688\n\n0\\.8501\n\n0\\.9395\n\n0\\.7836 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nEurope\n\n0\\.3001\n\n0\\.4167\n\n0\\.3583\n\n0\\.4815\n\n0\\.6852\n\n0\\.3862 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nAfrica\n\n0\\.5462\n\n0\\.5984\n\n0\\.6311\n\n0\\.7471\n\n0\\.8562\n\n0\\.6401\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nAsia\n\n0\\.3748\n\n0\\.3839\n\n0\\.4174\n\n0\\.5000\n\n0\\.6914\n\n0\\.4378 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nOceania\n\nLegend:\n\nLinear (global OSI average)\n\nLinear (average OSI for Africa)\n\nLinear (average OSI for Asia) Linear (average OSI for Europe) Linear (average OSI for Oaceania)\n\n193 UN Member States\n\nAfrica\n\nAsia Europe Oceania\n\nLinear (average OSI for the Americas) Americas\n\n0\\.4919\n\n0\\.5363\n\n0\\.5606\n\n0\\.6644\n\n0\\.8079\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\n193 UN Member States Linear (global average OSI)\n\n0\\.4736\n\n0\\.5411\n\n0\\.5797\n\n0\\.6730\n\n0\\.8171\n\n0\\.5797\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nAmericas\n\n0\\.6642\n\n0\\.7263\n\n0\\.7688\n\n0\\.8501\n\n0\\.9395\n\n0\\.7836 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nEurope\n\n0\\.3001\n\n0\\.4167\n\n0\\.3583\n\n0\\.4815\n\n0\\.6852\n\n0\\.3862 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nAfrica\n\n0\\.5462\n\n0\\.5984\n\n0\\.6311\n\n0\\.7471\n\n0\\.8562\n\n0\\.6401\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nAsia\n\n0\\.3748\n\n0\\.3839\n\n0\\.4174\n\n0\\.5000\n\n0\\.6914\n\n0\\.4378 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nOceania\n\nLegend:\n\nLinear (global OSI average)\n\nLinear (average OSI for Africa)\n\nLinear (average OSI for Asia) Linear (average OSI for Europe) Linear (average OSI for Oaceania)\n\n193 UN Member States\n\nAfrica\n\nAsia Europe Oceania\n\nLinear (average OSI for the Americas) Americas\n\n0\\.4919\n\n0\\.5363\n\n0\\.5606\n\n0\\.6644\n\n0\\.8079\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\n193 UN Member States Linear (global average OSI)\n\n0\\.4736\n\n0\\.5411\n\n0\\.5797\n\n0\\.6730\n\n0\\.8171\n\n0\\.5797\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nAmericas\n\n0\\.6642\n\n0\\.7263\n\n0\\.7688\n\n0\\.8501\n\n0\\.9395\n\n0\\.7836 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nEurope\n\n0\\.3001\n\n0\\.4167\n\n0\\.3583\n\n0\\.4815\n\n0\\.6852\n\n0\\.3862 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nAfrica\n\n0\\.5462\n\n0\\.5984\n\n0\\.6311\n\n0\\.7471\n\n0\\.8562\n\n0\\.6401\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nAsia\n\n0\\.3748\n\n0\\.3839\n\n0\\.4174\n\n0\\.5000\n\n0\\.6914\n\n0\\.4378 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nOceania","pages":[{"line_numbers":[],"markdown":"2024 UN E-GovErNmENt SUrvEy\n\n52\n\n**Chapter 2**\n\n### **2\\.5.2 OSI subindices**\n\nThe subsections above have highlighted areas that require further attention if countries wish to strengthen overall e-government development (as reflected in their respective EGDI levels). In cases where improvements in online services provision (OSI values) are necessary, it is important to understand which specific aspects of this component require attention. The main Survey findings for each of the five OSI subindices are presented below.\n\nAs shown in figure 2.9, the institutional framework (IF) is better developed than other aspects of online services provision in all regions. content provision (cP) has the next highest subindex value, followed by services provision (SP), technical characteristics (TEc) and e-participation (EPI). At the regional level, Europe is leading in terms of average OSI and subindex values, followed by Asia, the Americas, Oceania, and Africa – the one exception being that the technical aspects of the OSI are slightly better developed in Africa than in Oceania.\n\nFigure 2.9 Values for OSI subindices (IF, CP, SP, TEC and EPI) at the regional and global levels, 2024\n\nLegend:\n\nLinear (global OSI average)\n\nLinear (average OSI for Africa)\n\nLinear (average OSI for Asia) Linear (average OSI for Europe) Linear (average OSI for Oaceania)\n\n193 UN Member States\n\nAfrica\n\nAsia Europe Oceania\n\nLinear (average OSI for the Americas) Americas\n\n193 UN Member States Linear (global average OSI) Africa\n\nLegend:\n\nLinear (global OSI average)\n\nLinear (average OSI for Africa)\n\nLinear (average OSI for Asia) Linear (average OSI for Europe) Linear (average OSI for Oaceania)\n\n193 UN Member States\n\nAfrica\n\nAsia Europe Oceania\n\nLinear (average OSI for the Americas) Americas\n\n0 6644\n\n0\\.8079\n\nCP\n\nIF\n\n193 UN Member States Linear (global average OSI)\n\n_Source_ : 2024 United Nations E-Government Survey.\n\nLegend:\n\nLinear (global OSI average)\n\nLinear (average OSI for Africa)\n\nLinear (average OSI for Asia) Linear (average OSI for Europe) Linear (average OSI for Oaceania)\n\n193 UN Member States\n\nAfrica\n\nAsia Europe Oceania\n\nLinear (average OSI for the Americas) Americas\n\n0\\.4919\n\n0\\.5363\n\n0\\.5606\n\n0\\.6644\n\n0\\.8079\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\n193 UN Member States Linear (global average OSI)\n\n0\\.4736\n\n0\\.5411\n\n0\\.5797\n\n0\\.6730\n\n0\\.8171\n\n0\\.5797\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nAmericas\n\n0\\.6642\n\n0\\.7263\n\n0\\.7688\n\n0\\.8501\n\n0\\.9395\n\n0\\.7836 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nEurope\n\n0\\.3001\n\n0\\.4167\n\n0\\.3583\n\n0\\.4815\n\n0\\.6852\n\n0\\.3862 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nAfrica\n\n0\\.5462\n\n0\\.5984\n\n0\\.6311\n\n0\\.7471\n\n0\\.8562\n\n0\\.6401\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nAsia\n\n0\\.3748\n\n0\\.3839\n\n0\\.4174\n\n0\\.5000\n\n0\\.6914\n\n0\\.4378 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nOceania\n\nLegend:\n\nLinear (global OSI average)\n\nLinear (average OSI for Africa)\n\nLinear (average OSI for Asia) Linear (average OSI for Europe) Linear (average OSI for Oaceania)\n\n193 UN Member States\n\nAfrica\n\nAsia Europe Oceania\n\nLinear (average OSI for the Americas) Americas\n\n0\\.4919\n\n0\\.5363\n\n0\\.5606\n\n0\\.6644\n\n0\\.8079\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\n193 UN Member States Linear (global average OSI)\n\n0\\.4736\n\n0\\.5411\n\n0\\.5797\n\n0\\.6730\n\n0\\.8171\n\n0\\.5797\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nAmericas\n\n0\\.6642\n\n0\\.7263\n\n0\\.7688\n\n0\\.8501\n\n0\\.9395\n\n0\\.7836 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nEurope\n\n0\\.3001\n\n0\\.4167\n\n0\\.3583\n\n0\\.4815\n\n0\\.6852\n\n0\\.3862 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nAfrica\n\n0\\.5462\n\n0\\.5984\n\n0\\.6311\n\n0\\.7471\n\n0\\.8562\n\n0\\.6401\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nAsia\n\n0\\.3748\n\n0\\.3839\n\n0\\.4174\n\n0\\.5000\n\n0\\.6914\n\n0\\.4378 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nOceania\n\nLegend:\n\nLinear (global OSI average)\n\nLinear (average OSI for Africa)\n\nLinear (average OSI for Asia) Linear (average OSI for Europe) Linear (average OSI for Oaceania)\n\n193 UN Member States\n\nAfrica\n\nAsia Europe Oceania\n\nLinear (average OSI for the Americas) Americas\n\n0\\.4919\n\n0\\.5363\n\n0\\.5606\n\n0\\.6644\n\n0\\.8079\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\n193 UN Member States Linear (global average OSI)\n\n0\\.4736\n\n0\\.5411\n\n0\\.5797\n\n0\\.6730\n\n0\\.8171\n\n0\\.5797\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nAmericas\n\n0\\.6642\n\n0\\.7263\n\n0\\.7688\n\n0\\.8501\n\n0\\.9395\n\n0\\.7836 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nEurope\n\n0\\.3001\n\n0\\.4167\n\n0\\.3583\n\n0\\.4815\n\n0\\.6852\n\n0\\.3862 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nAfrica\n\n0\\.5462\n\n0\\.5984\n\n0\\.6311\n\n0\\.7471\n\n0\\.8562\n\n0\\.6401\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nAsia\n\n0\\.3748\n\n0\\.3839\n\n0\\.4174\n\n0\\.5000\n\n0\\.6914\n\n0\\.4378 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nOceania\n\nLegend:\n\nLinear (global OSI average)\n\nLinear (average OSI for Africa)\n\nLinear (average OSI for Asia) Linear (average OSI for Europe) Linear (average OSI for Oaceania)\n\n193 UN Member States\n\nAfrica\n\nAsia Europe Oceania\n\nLinear (average OSI for the Americas) Americas\n\n0\\.4919\n\n0\\.5363\n\n0\\.5606\n\n0\\.6644\n\n0\\.8079\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\n193 UN Member States Linear (global average OSI)\n\n0\\.4736\n\n0\\.5411\n\n0\\.5797\n\n0\\.6730\n\n0\\.8171\n\n0\\.5797\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nAmericas\n\n0\\.6642\n\n0\\.7263\n\n0\\.7688\n\n0\\.8501\n\n0\\.9395\n\n0\\.7836 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nEurope\n\n0\\.3001\n\n0\\.4167\n\n0\\.3583\n\n0\\.4815\n\n0\\.6852\n\n0\\.3862 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nAfrica\n\n0\\.5462\n\n0\\.5984\n\n0\\.6311\n\n0\\.7471\n\n0\\.8562\n\n0\\.6401\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nAsia\n\n0\\.3748\n\n0\\.3839\n\n0\\.4174\n\n0\\.5000\n\n0\\.6914\n\n0\\.4378 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nOceania\n\nLegend:\n\nLinear (global OSI average)\n\nLinear (average OSI for Africa)\n\nLinear (average OSI for Asia) Linear (average OSI for Europe) Linear (average OSI for Oaceania)\n\n193 UN Member States\n\nAfrica\n\nAsia Europe Oceania\n\nLinear (average OSI for the Americas) Americas\n\n0\\.4919\n\n0\\.5363\n\n0\\.5606\n\n0\\.6644\n\n0\\.8079\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\n193 UN Member States Linear (global average OSI)\n\n0\\.4736\n\n0\\.5411\n\n0\\.5797\n\n0\\.6730\n\n0\\.8171\n\n0\\.5797\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nAmericas\n\n0\\.6642\n\n0\\.7263\n\n0\\.7688\n\n0\\.8501\n\n0\\.9395\n\n0\\.7836 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nEurope\n\n0\\.3001\n\n0\\.4167\n\n0\\.3583\n\n0\\.4815\n\n0\\.6852\n\n0\\.3862 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nAfrica\n\n0\\.5462\n\n0\\.5984\n\n0\\.6311\n\n0\\.7471\n\n0\\.8562\n\n0\\.6401\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nAsia\n\n0\\.3748\n\n0\\.3839\n\n0\\.4174\n\n0\\.5000\n\n0\\.6914\n\n0\\.4378 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nOceania\n\nLegend:\n\nLinear (global OSI average)\n\nLinear (average OSI for Africa)\n\nLinear (average OSI for Asia) Linear (average OSI for Europe) Linear (average OSI for Oaceania)\n\n193 UN Member States\n\nAfrica\n\nAsia Europe Oceania\n\nLinear (average OSI for the Americas) Americas\n\n0\\.4919\n\n0\\.5363\n\n0\\.5606\n\n0\\.6644\n\n0\\.8079\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\n193 UN Member States Linear (global average OSI)\n\n0\\.4736\n\n0\\.5411\n\n0\\.5797\n\n0\\.6730\n\n0\\.8171\n\n0\\.5797\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nAmericas\n\n0\\.6642\n\n0\\.7263\n\n0\\.7688\n\n0\\.8501\n\n0\\.9395\n\n0\\.7836 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nEurope\n\n0\\.3001\n\n0\\.4167\n\n0\\.3583\n\n0\\.4815\n\n0\\.6852\n\n0\\.3862 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nAfrica\n\n0\\.5462\n\n0\\.5984\n\n0\\.6311\n\n0\\.7471\n\n0\\.8562\n\n0\\.6401\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nAsia\n\n0\\.3748\n\n0\\.3839\n\n0\\.4174\n\n0\\.5000\n\n0\\.6914\n\n0\\.4378 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nOceania","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"2024 UN E-GovErNmENt SUrvEy\n\n52\n\n**Chapter 2**\n\n### **2\\.5.2 OSI subindices**\n\nThe subsections above have highlighted areas that require further attention if countries wish to strengthen overall e-government development (as reflected in their respective EGDI levels). In cases where improvements in online services provision (OSI values) are necessary, it is important to understand which specific aspects of this component require attention. The main Survey findings for each of the five OSI subindices are presented below.\n\nAs shown in figure 2.9, the institutional framework (IF) is better developed than other aspects of online services provision in all regions. content provision (cP) has the next highest subindex value, followed by services provision (SP), technical characteristics (TEc) and e-participation (EPI). At the regional level, Europe is leading in terms of average OSI and subindex values, followed by Asia, the Americas, Oceania, and Africa – the one exception being that the technical aspects of the OSI are slightly better developed in Africa than in Oceania.\n\nFigure 2.9 Values for OSI subindices (IF, CP, SP, TEC and EPI) at the regional and global levels, 2024\n\nLegend:\n\nLinear (global OSI average)\n\nLinear (average OSI for Africa)\n\nLinear (average OSI for Asia) Linear (average OSI for Europe) Linear (average OSI for Oaceania)\n\n193 UN Member States\n\nAfrica\n\nAsia Europe Oceania\n\nLinear (average OSI for the Americas) Americas\n\n193 UN Member States Linear (global average OSI) Africa\n\nLegend:\n\nLinear (global OSI average)\n\nLinear (average OSI for Africa)\n\nLinear (average OSI for Asia) Linear (average OSI for Europe) Linear (average OSI for Oaceania)\n\n193 UN Member States\n\nAfrica\n\nAsia Europe Oceania\n\nLinear (average OSI for the Americas) Americas\n\n0 6644\n\n0\\.8079\n\nCP\n\nIF\n\n193 UN Member States Linear (global average OSI)\n\n_Source_ : 2024 United Nations E-Government Survey.\n\nLegend:\n\nLinear (global OSI average)\n\nLinear (average OSI for Africa)\n\nLinear (average OSI for Asia) Linear (average OSI for Europe) Linear (average OSI for Oaceania)\n\n193 UN Member States\n\nAfrica\n\nAsia Europe Oceania\n\nLinear (average OSI for the Americas) Americas\n\n0\\.4919\n\n0\\.5363\n\n0\\.5606\n\n0\\.6644\n\n0\\.8079\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\n193 UN Member States Linear (global average OSI)\n\n0\\.4736\n\n0\\.5411\n\n0\\.5797\n\n0\\.6730\n\n0\\.8171\n\n0\\.5797\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nAmericas\n\n0\\.6642\n\n0\\.7263\n\n0\\.7688\n\n0\\.8501\n\n0\\.9395\n\n0\\.7836 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nEurope\n\n0\\.3001\n\n0\\.4167\n\n0\\.3583\n\n0\\.4815\n\n0\\.6852\n\n0\\.3862 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nAfrica\n\n0\\.5462\n\n0\\.5984\n\n0\\.6311\n\n0\\.7471\n\n0\\.8562\n\n0\\.6401\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nAsia\n\n0\\.3748\n\n0\\.3839\n\n0\\.4174\n\n0\\.5000\n\n0\\.6914\n\n0\\.4378 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nOceania\n\nLegend:\n\nLinear (global OSI average)\n\nLinear (average OSI for Africa)\n\nLinear (average OSI for Asia) Linear (average OSI for Europe) Linear (average OSI for Oaceania)\n\n193 UN Member States\n\nAfrica\n\nAsia Europe Oceania\n\nLinear (average OSI for the Americas) Americas\n\n0\\.4919\n\n0\\.5363\n\n0\\.5606\n\n0\\.6644\n\n0\\.8079\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\n193 UN Member States Linear (global average OSI)\n\n0\\.4736\n\n0\\.5411\n\n0\\.5797\n\n0\\.6730\n\n0\\.8171\n\n0\\.5797\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nAmericas\n\n0\\.6642\n\n0\\.7263\n\n0\\.7688\n\n0\\.8501\n\n0\\.9395\n\n0\\.7836 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nEurope\n\n0\\.3001\n\n0\\.4167\n\n0\\.3583\n\n0\\.4815\n\n0\\.6852\n\n0\\.3862 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nAfrica\n\n0\\.5462\n\n0\\.5984\n\n0\\.6311\n\n0\\.7471\n\n0\\.8562\n\n0\\.6401\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nAsia\n\n0\\.3748\n\n0\\.3839\n\n0\\.4174\n\n0\\.5000\n\n0\\.6914\n\n0\\.4378 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nOceania\n\nLegend:\n\nLinear (global OSI average)\n\nLinear (average OSI for Africa)\n\nLinear (average OSI for Asia) Linear (average OSI for Europe) Linear (average OSI for Oaceania)\n\n193 UN Member States\n\nAfrica\n\nAsia Europe Oceania\n\nLinear (average OSI for the Americas) Americas\n\n0\\.4919\n\n0\\.5363\n\n0\\.5606\n\n0\\.6644\n\n0\\.8079\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\n193 UN Member States Linear (global average OSI)\n\n0\\.4736\n\n0\\.5411\n\n0\\.5797\n\n0\\.6730\n\n0\\.8171\n\n0\\.5797\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nAmericas\n\n0\\.6642\n\n0\\.7263\n\n0\\.7688\n\n0\\.8501\n\n0\\.9395\n\n0\\.7836 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nEurope\n\n0\\.3001\n\n0\\.4167\n\n0\\.3583\n\n0\\.4815\n\n0\\.6852\n\n0\\.3862 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nAfrica\n\n0\\.5462\n\n0\\.5984\n\n0\\.6311\n\n0\\.7471\n\n0\\.8562\n\n0\\.6401\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nAsia\n\n0\\.3748\n\n0\\.3839\n\n0\\.4174\n\n0\\.5000\n\n0\\.6914\n\n0\\.4378 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nOceania\n\nLegend:\n\nLinear (global OSI average)\n\nLinear (average OSI for Africa)\n\nLinear (average OSI for Asia) Linear (average OSI for Europe) Linear (average OSI for Oaceania)\n\n193 UN Member States\n\nAfrica\n\nAsia Europe Oceania\n\nLinear (average OSI for the Americas) Americas\n\n0\\.4919\n\n0\\.5363\n\n0\\.5606\n\n0\\.6644\n\n0\\.8079\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\n193 UN Member States Linear (global average OSI)\n\n0\\.4736\n\n0\\.5411\n\n0\\.5797\n\n0\\.6730\n\n0\\.8171\n\n0\\.5797\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nAmericas\n\n0\\.6642\n\n0\\.7263\n\n0\\.7688\n\n0\\.8501\n\n0\\.9395\n\n0\\.7836 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nEurope\n\n0\\.3001\n\n0\\.4167\n\n0\\.3583\n\n0\\.4815\n\n0\\.6852\n\n0\\.3862 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nAfrica\n\n0\\.5462\n\n0\\.5984\n\n0\\.6311\n\n0\\.7471\n\n0\\.8562\n\n0\\.6401\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nAsia\n\n0\\.3748\n\n0\\.3839\n\n0\\.4174\n\n0\\.5000\n\n0\\.6914\n\n0\\.4378 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nOceania\n\nLegend:\n\nLinear (global OSI average)\n\nLinear (average OSI for Africa)\n\nLinear (average OSI for Asia) Linear (average OSI for Europe) Linear (average OSI for Oaceania)\n\n193 UN Member States\n\nAfrica\n\nAsia Europe Oceania\n\nLinear (average OSI for the Americas) Americas\n\n0\\.4919\n\n0\\.5363\n\n0\\.5606\n\n0\\.6644\n\n0\\.8079\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\n193 UN Member States Linear (global average OSI)\n\n0\\.4736\n\n0\\.5411\n\n0\\.5797\n\n0\\.6730\n\n0\\.8171\n\n0\\.5797\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nAmericas\n\n0\\.6642\n\n0\\.7263\n\n0\\.7688\n\n0\\.8501\n\n0\\.9395\n\n0\\.7836 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nEurope\n\n0\\.3001\n\n0\\.4167\n\n0\\.3583\n\n0\\.4815\n\n0\\.6852\n\n0\\.3862 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nAfrica\n\n0\\.5462\n\n0\\.5984\n\n0\\.6311\n\n0\\.7471\n\n0\\.8562\n\n0\\.6401\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nAsia\n\n0\\.3748\n\n0\\.3839\n\n0\\.4174\n\n0\\.5000\n\n0\\.6914\n\n0\\.4378 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nOceania\n\nLegend:\n\nLinear (global OSI average)\n\nLinear (average OSI for Africa)\n\nLinear (average OSI for Asia) Linear (average OSI for Europe) Linear (average OSI for Oaceania)\n\n193 UN Member States\n\nAfrica\n\nAsia Europe Oceania\n\nLinear (average OSI for the Americas) Americas\n\n0\\.4919\n\n0\\.5363\n\n0\\.5606\n\n0\\.6644\n\n0\\.8079\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\n193 UN Member States Linear (global average OSI)\n\n0\\.4736\n\n0\\.5411\n\n0\\.5797\n\n0\\.6730\n\n0\\.8171\n\n0\\.5797\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nAmericas\n\n0\\.6642\n\n0\\.7263\n\n0\\.7688\n\n0\\.8501\n\n0\\.9395\n\n0\\.7836 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nEurope\n\n0\\.3001\n\n0\\.4167\n\n0\\.3583\n\n0\\.4815\n\n0\\.6852\n\n0\\.3862 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nAfrica\n\n0\\.5462\n\n0\\.5984\n\n0\\.6311\n\n0\\.7471\n\n0\\.8562\n\n0\\.6401\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nAsia\n\n0\\.3748\n\n0\\.3839\n\n0\\.4174\n\n0\\.5000\n\n0\\.6914\n\n0\\.4378 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nOceania","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/%28Web_version%29_E-Government_Survey_2024_1392024_p77.pdf"]},"response":{"errors":[],"extract_id":"extract_8eeb48dd2dc689dce3f12e9c3d49e6fd","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["2024 UN E-GovErNmENt SUrvEy\n\n52\n\n**Chapter 2**\n\n### **2\\.5.2 OSI subindices**\n\nThe subsections above have highlighted areas that require further attention if countries wish to strengthen overall e-government development (as reflected in their respective EGDI levels). In cases where improvements in online services provision (OSI values) are necessary, it is important to understand which specific aspects of this component require attention. The main Survey findings for each of the five OSI subindices are presented below.\n\nAs shown in figure 2.9, the institutional framework (IF) is better developed than other aspects of online services provision in all regions. content provision (cP) has the next highest subindex value, followed by services provision (SP), technical characteristics (TEc) and e-participation (EPI). At the regional level, Europe is leading in terms of average OSI and subindex values, followed by Asia, the Americas, Oceania, and Africa – the one exception being that the technical aspects of the OSI are slightly better developed in Africa than in Oceania.\n\nFigure 2.9 Values for OSI subindices (IF, CP, SP, TEC and EPI) at the regional and global levels, 2024\n\nLegend:\n\nLinear (global OSI average)\n\nLinear (average OSI for Africa)\n\nLinear (average OSI for Asia) Linear (average OSI for Europe) Linear (average OSI for Oaceania)\n\n193 UN Member States\n\nAfrica\n\nAsia Europe Oceania\n\nLinear (average OSI for the Americas) Americas\n\n193 UN Member States Linear (global average OSI) Africa\n\nLegend:\n\nLinear (global OSI average)\n\nLinear (average OSI for Africa)\n\nLinear (average OSI for Asia) Linear (average OSI for Europe) Linear (average OSI for Oaceania)\n\n193 UN Member States\n\nAfrica\n\nAsia Europe Oceania\n\nLinear (average OSI for the Americas) Americas\n\n0 6644\n\n0\\.8079\n\nCP\n\nIF\n\n193 UN Member States Linear (global average OSI)\n\n_Source_ : 2024 United Nations E-Government Survey.\n\nLegend:\n\nLinear (global OSI average)\n\nLinear (average OSI for Africa)\n\nLinear (average OSI for Asia) Linear (average OSI for Europe) Linear (average OSI for Oaceania)\n\n193 UN Member States\n\nAfrica\n\nAsia Europe Oceania\n\nLinear (average OSI for the Americas) Americas\n\n0\\.4919\n\n0\\.5363\n\n0\\.5606\n\n0\\.6644\n\n0\\.8079\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\n193 UN Member States Linear (global average OSI)\n\n0\\.4736\n\n0\\.5411\n\n0\\.5797\n\n0\\.6730\n\n0\\.8171\n\n0\\.5797\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nAmericas\n\n0\\.6642\n\n0\\.7263\n\n0\\.7688\n\n0\\.8501\n\n0\\.9395\n\n0\\.7836 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nEurope\n\n0\\.3001\n\n0\\.4167\n\n0\\.3583\n\n0\\.4815\n\n0\\.6852\n\n0\\.3862 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nAfrica\n\n0\\.5462\n\n0\\.5984\n\n0\\.6311\n\n0\\.7471\n\n0\\.8562\n\n0\\.6401\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nAsia\n\n0\\.3748\n\n0\\.3839\n\n0\\.4174\n\n0\\.5000\n\n0\\.6914\n\n0\\.4378 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nOceania\n\nLegend:\n\nLinear (global OSI average)\n\nLinear (average OSI for Africa)\n\nLinear (average OSI for Asia) Linear (average OSI for Europe) Linear (average OSI for Oaceania)\n\n193 UN Member States\n\nAfrica\n\nAsia Europe Oceania\n\nLinear (average OSI for the Americas) Americas\n\n0\\.4919\n\n0\\.5363\n\n0\\.5606\n\n0\\.6644\n\n0\\.8079\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\n193 UN Member States Linear (global average OSI)\n\n0\\.4736\n\n0\\.5411\n\n0\\.5797\n\n0\\.6730\n\n0\\.8171\n\n0\\.5797\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nAmericas\n\n0\\.6642\n\n0\\.7263\n\n0\\.7688\n\n0\\.8501\n\n0\\.9395\n\n0\\.7836 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nEurope\n\n0\\.3001\n\n0\\.4167\n\n0\\.3583\n\n0\\.4815\n\n0\\.6852\n\n0\\.3862 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nAfrica\n\n0\\.5462\n\n0\\.5984\n\n0\\.6311\n\n0\\.7471\n\n0\\.8562\n\n0\\.6401\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nAsia\n\n0\\.3748\n\n0\\.3839\n\n0\\.4174\n\n0\\.5000\n\n0\\.6914\n\n0\\.4378 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nOceania\n\nLegend:\n\nLinear (global OSI average)\n\nLinear (average OSI for Africa)\n\nLinear (average OSI for Asia) Linear (average OSI for Europe) Linear (average OSI for Oaceania)\n\n193 UN Member States\n\nAfrica\n\nAsia Europe Oceania\n\nLinear (average OSI for the Americas) Americas\n\n0\\.4919\n\n0\\.5363\n\n0\\.5606\n\n0\\.6644\n\n0\\.8079\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\n193 UN Member States Linear (global average OSI)\n\n0\\.4736\n\n0\\.5411\n\n0\\.5797\n\n0\\.6730\n\n0\\.8171\n\n0\\.5797\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nAmericas\n\n0\\.6642\n\n0\\.7263\n\n0\\.7688\n\n0\\.8501\n\n0\\.9395\n\n0\\.7836 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nEurope\n\n0\\.3001\n\n0\\.4167\n\n0\\.3583\n\n0\\.4815\n\n0\\.6852\n\n0\\.3862 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nAfrica\n\n0\\.5462\n\n0\\.5984\n\n0\\.6311\n\n0\\.7471\n\n0\\.8562\n\n0\\.6401\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nAsia\n\n0\\.3748\n\n0\\.3839\n\n0\\.4174\n\n0\\.5000\n\n0\\.6914\n\n0\\.4378 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nOceania\n\nLegend:\n\nLinear (global OSI average)\n\nLinear (average OSI for Africa)\n\nLinear (average OSI for Asia) Linear (average OSI for Europe) Linear (average OSI for Oaceania)\n\n193 UN Member States\n\nAfrica\n\nAsia Europe Oceania\n\nLinear (average OSI for the Americas) Americas\n\n0\\.4919\n\n0\\.5363\n\n0\\.5606\n\n0\\.6644\n\n0\\.8079\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\n193 UN Member States Linear (global average OSI)\n\n0\\.4736\n\n0\\.5411\n\n0\\.5797\n\n0\\.6730\n\n0\\.8171\n\n0\\.5797\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nAmericas\n\n0\\.6642\n\n0\\.7263\n\n0\\.7688\n\n0\\.8501\n\n0\\.9395\n\n0\\.7836 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nEurope\n\n0\\.3001\n\n0\\.4167\n\n0\\.3583\n\n0\\.4815\n\n0\\.6852\n\n0\\.3862 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nAfrica\n\n0\\.5462\n\n0\\.5984\n\n0\\.6311\n\n0\\.7471\n\n0\\.8562\n\n0\\.6401\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nAsia\n\n0\\.3748\n\n0\\.3839\n\n0\\.4174\n\n0\\.5000\n\n0\\.6914\n\n0\\.4378 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nOceania\n\nLegend:\n\nLinear (global OSI average)\n\nLinear (average OSI for Africa)\n\nLinear (average OSI for Asia) Linear (average OSI for Europe) Linear (average OSI for Oaceania)\n\n193 UN Member States\n\nAfrica\n\nAsia Europe Oceania\n\nLinear (average OSI for the Americas) Americas\n\n0\\.4919\n\n0\\.5363\n\n0\\.5606\n\n0\\.6644\n\n0\\.8079\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\n193 UN Member States Linear (global average OSI)\n\n0\\.4736\n\n0\\.5411\n\n0\\.5797\n\n0\\.6730\n\n0\\.8171\n\n0\\.5797\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nAmericas\n\n0\\.6642\n\n0\\.7263\n\n0\\.7688\n\n0\\.8501\n\n0\\.9395\n\n0\\.7836 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nEurope\n\n0\\.3001\n\n0\\.4167\n\n0\\.3583\n\n0\\.4815\n\n0\\.6852\n\n0\\.3862 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nAfrica\n\n0\\.5462\n\n0\\.5984\n\n0\\.6311\n\n0\\.7471\n\n0\\.8562\n\n0\\.6401\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nAsia\n\n0\\.3748\n\n0\\.3839\n\n0\\.4174\n\n0\\.5000\n\n0\\.6914\n\n0\\.4378 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nOceania\n\nLegend:\n\nLinear (global OSI average)\n\nLinear (average OSI for Africa)\n\nLinear (average OSI for Asia) Linear (average OSI for Europe) Linear (average OSI for Oaceania)\n\n193 UN Member States\n\nAfrica\n\nAsia Europe Oceania\n\nLinear (average OSI for the Americas) Americas\n\n0\\.4919\n\n0\\.5363\n\n0\\.5606\n\n0\\.6644\n\n0\\.8079\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\n193 UN Member States Linear (global average OSI)\n\n0\\.4736\n\n0\\.5411\n\n0\\.5797\n\n0\\.6730\n\n0\\.8171\n\n0\\.5797\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nAmericas\n\n0\\.6642\n\n0\\.7263\n\n0\\.7688\n\n0\\.8501\n\n0\\.9395\n\n0\\.7836 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nEurope\n\n0\\.3001\n\n0\\.4167\n\n0\\.3583\n\n0\\.4815\n\n0\\.6852\n\n0\\.3862 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nAfrica\n\n0\\.5462\n\n0\\.5984\n\n0\\.6311\n\n0\\.7471\n\n0\\.8562\n\n0\\.6401\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nAsia\n\n0\\.3748\n\n0\\.3839\n\n0\\.4174\n\n0\\.5000\n\n0\\.6914\n\n0\\.4378 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nOceania"],"full_content":"2024 UN E-GovErNmENt SUrvEy\n\n52\n\n**Chapter 2**\n\n### **2\\.5.2 OSI subindices**\n\nThe subsections above have highlighted areas that require further attention if countries wish to strengthen overall e-government development (as reflected in their respective EGDI levels). In cases where improvements in online services provision (OSI values) are necessary, it is important to understand which specific aspects of this component require attention. The main Survey findings for each of the five OSI subindices are presented below.\n\nAs shown in figure 2.9, the institutional framework (IF) is better developed than other aspects of online services provision in all regions. content provision (cP) has the next highest subindex value, followed by services provision (SP), technical characteristics (TEc) and e-participation (EPI). At the regional level, Europe is leading in terms of average OSI and subindex values, followed by Asia, the Americas, Oceania, and Africa – the one exception being that the technical aspects of the OSI are slightly better developed in Africa than in Oceania.\n\nFigure 2.9 Values for OSI subindices (IF, CP, SP, TEC and EPI) at the regional and global levels, 2024\n\nLegend:\n\nLinear (global OSI average)\n\nLinear (average OSI for Africa)\n\nLinear (average OSI for Asia) Linear (average OSI for Europe) Linear (average OSI for Oaceania)\n\n193 UN Member States\n\nAfrica\n\nAsia Europe Oceania\n\nLinear (average OSI for the Americas) Americas\n\n193 UN Member States Linear (global average OSI) Africa\n\nLegend:\n\nLinear (global OSI average)\n\nLinear (average OSI for Africa)\n\nLinear (average OSI for Asia) Linear (average OSI for Europe) Linear (average OSI for Oaceania)\n\n193 UN Member States\n\nAfrica\n\nAsia Europe Oceania\n\nLinear (average OSI for the Americas) Americas\n\n0 6644\n\n0\\.8079\n\nCP\n\nIF\n\n193 UN Member States Linear (global average OSI)\n\n_Source_ : 2024 United Nations E-Government Survey.\n\nLegend:\n\nLinear (global OSI average)\n\nLinear (average OSI for Africa)\n\nLinear (average OSI for Asia) Linear (average OSI for Europe) Linear (average OSI for Oaceania)\n\n193 UN Member States\n\nAfrica\n\nAsia Europe Oceania\n\nLinear (average OSI for the Americas) Americas\n\n0\\.4919\n\n0\\.5363\n\n0\\.5606\n\n0\\.6644\n\n0\\.8079\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\n193 UN Member States Linear (global average OSI)\n\n0\\.4736\n\n0\\.5411\n\n0\\.5797\n\n0\\.6730\n\n0\\.8171\n\n0\\.5797\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nAmericas\n\n0\\.6642\n\n0\\.7263\n\n0\\.7688\n\n0\\.8501\n\n0\\.9395\n\n0\\.7836 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nEurope\n\n0\\.3001\n\n0\\.4167\n\n0\\.3583\n\n0\\.4815\n\n0\\.6852\n\n0\\.3862 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nAfrica\n\n0\\.5462\n\n0\\.5984\n\n0\\.6311\n\n0\\.7471\n\n0\\.8562\n\n0\\.6401\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nAsia\n\n0\\.3748\n\n0\\.3839\n\n0\\.4174\n\n0\\.5000\n\n0\\.6914\n\n0\\.4378 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nOceania\n\nLegend:\n\nLinear (global OSI average)\n\nLinear (average OSI for Africa)\n\nLinear (average OSI for Asia) Linear (average OSI for Europe) Linear (average OSI for Oaceania)\n\n193 UN Member States\n\nAfrica\n\nAsia Europe Oceania\n\nLinear (average OSI for the Americas) Americas\n\n0\\.4919\n\n0\\.5363\n\n0\\.5606\n\n0\\.6644\n\n0\\.8079\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\n193 UN Member States Linear (global average OSI)\n\n0\\.4736\n\n0\\.5411\n\n0\\.5797\n\n0\\.6730\n\n0\\.8171\n\n0\\.5797\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nAmericas\n\n0\\.6642\n\n0\\.7263\n\n0\\.7688\n\n0\\.8501\n\n0\\.9395\n\n0\\.7836 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nEurope\n\n0\\.3001\n\n0\\.4167\n\n0\\.3583\n\n0\\.4815\n\n0\\.6852\n\n0\\.3862 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nAfrica\n\n0\\.5462\n\n0\\.5984\n\n0\\.6311\n\n0\\.7471\n\n0\\.8562\n\n0\\.6401\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nAsia\n\n0\\.3748\n\n0\\.3839\n\n0\\.4174\n\n0\\.5000\n\n0\\.6914\n\n0\\.4378 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nOceania\n\nLegend:\n\nLinear (global OSI average)\n\nLinear (average OSI for Africa)\n\nLinear (average OSI for Asia) Linear (average OSI for Europe) Linear (average OSI for Oaceania)\n\n193 UN Member States\n\nAfrica\n\nAsia Europe Oceania\n\nLinear (average OSI for the Americas) Americas\n\n0\\.4919\n\n0\\.5363\n\n0\\.5606\n\n0\\.6644\n\n0\\.8079\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\n193 UN Member States Linear (global average OSI)\n\n0\\.4736\n\n0\\.5411\n\n0\\.5797\n\n0\\.6730\n\n0\\.8171\n\n0\\.5797\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nAmericas\n\n0\\.6642\n\n0\\.7263\n\n0\\.7688\n\n0\\.8501\n\n0\\.9395\n\n0\\.7836 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nEurope\n\n0\\.3001\n\n0\\.4167\n\n0\\.3583\n\n0\\.4815\n\n0\\.6852\n\n0\\.3862 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nAfrica\n\n0\\.5462\n\n0\\.5984\n\n0\\.6311\n\n0\\.7471\n\n0\\.8562\n\n0\\.6401\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nAsia\n\n0\\.3748\n\n0\\.3839\n\n0\\.4174\n\n0\\.5000\n\n0\\.6914\n\n0\\.4378 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nOceania\n\nLegend:\n\nLinear (global OSI average)\n\nLinear (average OSI for Africa)\n\nLinear (average OSI for Asia) Linear (average OSI for Europe) Linear (average OSI for Oaceania)\n\n193 UN Member States\n\nAfrica\n\nAsia Europe Oceania\n\nLinear (average OSI for the Americas) Americas\n\n0\\.4919\n\n0\\.5363\n\n0\\.5606\n\n0\\.6644\n\n0\\.8079\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\n193 UN Member States Linear (global average OSI)\n\n0\\.4736\n\n0\\.5411\n\n0\\.5797\n\n0\\.6730\n\n0\\.8171\n\n0\\.5797\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nAmericas\n\n0\\.6642\n\n0\\.7263\n\n0\\.7688\n\n0\\.8501\n\n0\\.9395\n\n0\\.7836 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nEurope\n\n0\\.3001\n\n0\\.4167\n\n0\\.3583\n\n0\\.4815\n\n0\\.6852\n\n0\\.3862 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nAfrica\n\n0\\.5462\n\n0\\.5984\n\n0\\.6311\n\n0\\.7471\n\n0\\.8562\n\n0\\.6401\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nAsia\n\n0\\.3748\n\n0\\.3839\n\n0\\.4174\n\n0\\.5000\n\n0\\.6914\n\n0\\.4378 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nOceania\n\nLegend:\n\nLinear (global OSI average)\n\nLinear (average OSI for Africa)\n\nLinear (average OSI for Asia) Linear (average OSI for Europe) Linear (average OSI for Oaceania)\n\n193 UN Member States\n\nAfrica\n\nAsia Europe Oceania\n\nLinear (average OSI for the Americas) Americas\n\n0\\.4919\n\n0\\.5363\n\n0\\.5606\n\n0\\.6644\n\n0\\.8079\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\n193 UN Member States Linear (global average OSI)\n\n0\\.4736\n\n0\\.5411\n\n0\\.5797\n\n0\\.6730\n\n0\\.8171\n\n0\\.5797\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nAmericas\n\n0\\.6642\n\n0\\.7263\n\n0\\.7688\n\n0\\.8501\n\n0\\.9395\n\n0\\.7836 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nEurope\n\n0\\.3001\n\n0\\.4167\n\n0\\.3583\n\n0\\.4815\n\n0\\.6852\n\n0\\.3862 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nAfrica\n\n0\\.5462\n\n0\\.5984\n\n0\\.6311\n\n0\\.7471\n\n0\\.8562\n\n0\\.6401\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nAsia\n\n0\\.3748\n\n0\\.3839\n\n0\\.4174\n\n0\\.5000\n\n0\\.6914\n\n0\\.4378 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nOceania\n\nLegend:\n\nLinear (global OSI average)\n\nLinear (average OSI for Africa)\n\nLinear (average OSI for Asia) Linear (average OSI for Europe) Linear (average OSI for Oaceania)\n\n193 UN Member States\n\nAfrica\n\nAsia Europe Oceania\n\nLinear (average OSI for the Americas) Americas\n\n0\\.4919\n\n0\\.5363\n\n0\\.5606\n\n0\\.6644\n\n0\\.8079\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\n193 UN Member States Linear (global average OSI)\n\n0\\.4736\n\n0\\.5411\n\n0\\.5797\n\n0\\.6730\n\n0\\.8171\n\n0\\.5797\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nAmericas\n\n0\\.6642\n\n0\\.7263\n\n0\\.7688\n\n0\\.8501\n\n0\\.9395\n\n0\\.7836 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nEurope\n\n0\\.3001\n\n0\\.4167\n\n0\\.3583\n\n0\\.4815\n\n0\\.6852\n\n0\\.3862 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nAfrica\n\n0\\.5462\n\n0\\.5984\n\n0\\.6311\n\n0\\.7471\n\n0\\.8562\n\n0\\.6401\n\n0\\.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nAsia\n\n0\\.3748\n\n0\\.3839\n\n0\\.4174\n\n0\\.5000\n\n0\\.6914\n\n0\\.4378 0.5754\n\n0\\.00 0.25 0.50 0.75 1.00\n\nEPI\n\nTEC\n\nSP\n\nCP\n\nIF\n\nOceania","publish_date":null,"title":"Very high EGDI Very high EGDI High EGDI High EGDI Very high EGDI Very high EGDI Middle EGDI Middle EGDI","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/%28Web_version%29_E-Government_Survey_2024_1392024_p77.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/%28Web_version%29_E-Government_Survey_2024_1392024_p77.pdf"},"request":{"example_id":"chart/(Web_version)_E-Government_Survey_2024_1392024_p77","product_type":"parse","source_file_path":"(Web_version)_E-Government_Survey_2024_1392024_p77.pdf"},"started_at":"2026-11-04T19:45:58.275452"} +{"completed_at":"2026-11-04T19:46:01.496399","latency_in_ms":2055,"output":{"example_id":"chart/(Web_version)_E-Government_Survey_2024_1392024_p78","grounded_pages":[],"layout_pages":[],"markdown":"53\n\n**Chapter 2**\n\nChapter 2 • Global trends in e-Government\n\n### **2\\.5.3 OSI institutional framework subindex**\n\nThe organizational aspects of the institutional framework, which orient users on engaging government agencies through the online platforms, are well developed across the board (see figure 2.10). All Member States except Belize\\* have fully operational national portals. The vast majority of countries (95 per cent) make the government organizational chart and information on the government structure available on these portals, 91 per cent provide the names and titles of the heads of government agencies, departments and ministries, 81 per cent furnish information on the national chief information officer (cIO) or the equivalent, and 75 per cent share links to subnational or local government agencies. More than 90 per cent of the countries have national portals that provide links to ministerial websites and offer sources of information on sector-specific policies.\n\nFigure 2.10 Percentage of countries addressing various aspects of the institutional framework, 2024\n\n_Sources_ : 2024 United Nations E-Government Survey.\n\nEDGI 2024 levels\n\nVery high EGDI High EGDI Middle EGDI Low EGDI\n\n## Very high EGDI High EGDI Middle EGDI Low EGDI\n\n75%\n\n81%\n\n91%\n\n95%\n\nLinks to subnational/local government agencies\n\nInformation about Chief Information Offcer (CIO)\n\nNames and titles of heads of government agencies/departments/ministries\n\nOrganizational structure and/or chart of the government\n\nOrganizational framework\n\n89%\n\n93%\n\n95%\n\n94%\n\n90%\n\n95%\n\nSocial protection\n\nJustice\n\nHealth\n\nEnvironment\n\nEmployment\n\nEducation\n\nLinks to sectoral or ministerial websites\n\n90%\n\n91%\n\n94%\n\n92%\n\n90%\n\n92%\n\nSocial protection\n\nJustice\n\nHealth\n\nEnvironment\n\nEmployment\n\nEducation\n\nInformation on sectoral policies\n\nThe legislative, policy and strategy aspects of the institutional framework are less consistently developed (see figure 2.11). Between 81 and 83 per cent of countries have national e-government strategies, policies or legislation on cybersecurity, data privacy, and data protection, 78 per cent require citizens to use a digital ID when accessing public services, and 76 per cent offer public access to legislation on the right of citizens to access government information. The shares are lower for legislation or policies on open government data (63 per cent), e-participation (51 per cent), protecting the public against misinformation, disinformation, and/or fake news (47 per cent), and frontier technologies such as cloud computing (44 per cent) and artificial intelligence (42 per cent). Regional disparities are more pronounced for these last five indicators; more than half of the countries in Asia and Europe have already adopted relevant legislation, policies or strategies, the average for the Americas is slightly above 30 per cent and that for Africa slightly below 30 per cent, and in Oceania the compliance rates range from 4 to 11 per cent.\n\n\\* At the time of the assessment for the Survey, the national portal of Belize was under review and inaccessible. To the extent possible,\n\nthe assessment was conducted through various ministerial website","pages":[{"line_numbers":[],"markdown":"53\n\n**Chapter 2**\n\nChapter 2 • Global trends in e-Government\n\n### **2\\.5.3 OSI institutional framework subindex**\n\nThe organizational aspects of the institutional framework, which orient users on engaging government agencies through the online platforms, are well developed across the board (see figure 2.10). All Member States except Belize\\* have fully operational national portals. The vast majority of countries (95 per cent) make the government organizational chart and information on the government structure available on these portals, 91 per cent provide the names and titles of the heads of government agencies, departments and ministries, 81 per cent furnish information on the national chief information officer (cIO) or the equivalent, and 75 per cent share links to subnational or local government agencies. More than 90 per cent of the countries have national portals that provide links to ministerial websites and offer sources of information on sector-specific policies.\n\nFigure 2.10 Percentage of countries addressing various aspects of the institutional framework, 2024\n\n_Sources_ : 2024 United Nations E-Government Survey.\n\nEDGI 2024 levels\n\nVery high EGDI High EGDI Middle EGDI Low EGDI\n\n## Very high EGDI High EGDI Middle EGDI Low EGDI\n\n75%\n\n81%\n\n91%\n\n95%\n\nLinks to subnational/local government agencies\n\nInformation about Chief Information Offcer (CIO)\n\nNames and titles of heads of government agencies/departments/ministries\n\nOrganizational structure and/or chart of the government\n\nOrganizational framework\n\n89%\n\n93%\n\n95%\n\n94%\n\n90%\n\n95%\n\nSocial protection\n\nJustice\n\nHealth\n\nEnvironment\n\nEmployment\n\nEducation\n\nLinks to sectoral or ministerial websites\n\n90%\n\n91%\n\n94%\n\n92%\n\n90%\n\n92%\n\nSocial protection\n\nJustice\n\nHealth\n\nEnvironment\n\nEmployment\n\nEducation\n\nInformation on sectoral policies\n\nThe legislative, policy and strategy aspects of the institutional framework are less consistently developed (see figure 2.11). Between 81 and 83 per cent of countries have national e-government strategies, policies or legislation on cybersecurity, data privacy, and data protection, 78 per cent require citizens to use a digital ID when accessing public services, and 76 per cent offer public access to legislation on the right of citizens to access government information. The shares are lower for legislation or policies on open government data (63 per cent), e-participation (51 per cent), protecting the public against misinformation, disinformation, and/or fake news (47 per cent), and frontier technologies such as cloud computing (44 per cent) and artificial intelligence (42 per cent). Regional disparities are more pronounced for these last five indicators; more than half of the countries in Asia and Europe have already adopted relevant legislation, policies or strategies, the average for the Americas is slightly above 30 per cent and that for Africa slightly below 30 per cent, and in Oceania the compliance rates range from 4 to 11 per cent.\n\n\\* At the time of the assessment for the Survey, the national portal of Belize was under review and inaccessible. To the extent possible,\n\nthe assessment was conducted through various ministerial website","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"53\n\n**Chapter 2**\n\nChapter 2 • Global trends in e-Government\n\n### **2\\.5.3 OSI institutional framework subindex**\n\nThe organizational aspects of the institutional framework, which orient users on engaging government agencies through the online platforms, are well developed across the board (see figure 2.10). All Member States except Belize\\* have fully operational national portals. The vast majority of countries (95 per cent) make the government organizational chart and information on the government structure available on these portals, 91 per cent provide the names and titles of the heads of government agencies, departments and ministries, 81 per cent furnish information on the national chief information officer (cIO) or the equivalent, and 75 per cent share links to subnational or local government agencies. More than 90 per cent of the countries have national portals that provide links to ministerial websites and offer sources of information on sector-specific policies.\n\nFigure 2.10 Percentage of countries addressing various aspects of the institutional framework, 2024\n\n_Sources_ : 2024 United Nations E-Government Survey.\n\nEDGI 2024 levels\n\nVery high EGDI High EGDI Middle EGDI Low EGDI\n\n## Very high EGDI High EGDI Middle EGDI Low EGDI\n\n75%\n\n81%\n\n91%\n\n95%\n\nLinks to subnational/local government agencies\n\nInformation about Chief Information Offcer (CIO)\n\nNames and titles of heads of government agencies/departments/ministries\n\nOrganizational structure and/or chart of the government\n\nOrganizational framework\n\n89%\n\n93%\n\n95%\n\n94%\n\n90%\n\n95%\n\nSocial protection\n\nJustice\n\nHealth\n\nEnvironment\n\nEmployment\n\nEducation\n\nLinks to sectoral or ministerial websites\n\n90%\n\n91%\n\n94%\n\n92%\n\n90%\n\n92%\n\nSocial protection\n\nJustice\n\nHealth\n\nEnvironment\n\nEmployment\n\nEducation\n\nInformation on sectoral policies\n\nThe legislative, policy and strategy aspects of the institutional framework are less consistently developed (see figure 2.11). Between 81 and 83 per cent of countries have national e-government strategies, policies or legislation on cybersecurity, data privacy, and data protection, 78 per cent require citizens to use a digital ID when accessing public services, and 76 per cent offer public access to legislation on the right of citizens to access government information. The shares are lower for legislation or policies on open government data (63 per cent), e-participation (51 per cent), protecting the public against misinformation, disinformation, and/or fake news (47 per cent), and frontier technologies such as cloud computing (44 per cent) and artificial intelligence (42 per cent). Regional disparities are more pronounced for these last five indicators; more than half of the countries in Asia and Europe have already adopted relevant legislation, policies or strategies, the average for the Americas is slightly above 30 per cent and that for Africa slightly below 30 per cent, and in Oceania the compliance rates range from 4 to 11 per cent.\n\n\\* At the time of the assessment for the Survey, the national portal of Belize was under review and inaccessible. To the extent possible,\n\nthe assessment was conducted through various ministerial website","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/%28Web_version%29_E-Government_Survey_2024_1392024_p78.pdf"]},"response":{"errors":[],"extract_id":"extract_4af94691c099fdf56abfdf0b351d4631","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["53\n\n**Chapter 2**\n\nChapter 2 • Global trends in e-Government\n\n### **2\\.5.3 OSI institutional framework subindex**\n\nThe organizational aspects of the institutional framework, which orient users on engaging government agencies through the online platforms, are well developed across the board (see figure 2.10). All Member States except Belize\\* have fully operational national portals. The vast majority of countries (95 per cent) make the government organizational chart and information on the government structure available on these portals, 91 per cent provide the names and titles of the heads of government agencies, departments and ministries, 81 per cent furnish information on the national chief information officer (cIO) or the equivalent, and 75 per cent share links to subnational or local government agencies. More than 90 per cent of the countries have national portals that provide links to ministerial websites and offer sources of information on sector-specific policies.\n\nFigure 2.10 Percentage of countries addressing various aspects of the institutional framework, 2024\n\n_Sources_ : 2024 United Nations E-Government Survey.\n\nEDGI 2024 levels\n\nVery high EGDI High EGDI Middle EGDI Low EGDI\n\n## Very high EGDI High EGDI Middle EGDI Low EGDI\n\n75%\n\n81%\n\n91%\n\n95%\n\nLinks to subnational/local government agencies\n\nInformation about Chief Information Offcer (CIO)\n\nNames and titles of heads of government agencies/departments/ministries\n\nOrganizational structure and/or chart of the government\n\nOrganizational framework\n\n89%\n\n93%\n\n95%\n\n94%\n\n90%\n\n95%\n\nSocial protection\n\nJustice\n\nHealth\n\nEnvironment\n\nEmployment\n\nEducation\n\nLinks to sectoral or ministerial websites\n\n90%\n\n91%\n\n94%\n\n92%\n\n90%\n\n92%\n\nSocial protection\n\nJustice\n\nHealth\n\nEnvironment\n\nEmployment\n\nEducation\n\nInformation on sectoral policies\n\nThe legislative, policy and strategy aspects of the institutional framework are less consistently developed (see figure 2.11). Between 81 and 83 per cent of countries have national e-government strategies, policies or legislation on cybersecurity, data privacy, and data protection, 78 per cent require citizens to use a digital ID when accessing public services, and 76 per cent offer public access to legislation on the right of citizens to access government information. The shares are lower for legislation or policies on open government data (63 per cent), e-participation (51 per cent), protecting the public against misinformation, disinformation, and/or fake news (47 per cent), and frontier technologies such as cloud computing (44 per cent) and artificial intelligence (42 per cent). Regional disparities are more pronounced for these last five indicators; more than half of the countries in Asia and Europe have already adopted relevant legislation, policies or strategies, the average for the Americas is slightly above 30 per cent and that for Africa slightly below 30 per cent, and in Oceania the compliance rates range from 4 to 11 per cent.\n\n\\* At the time of the assessment for the Survey, the national portal of Belize was under review and inaccessible. To the extent possible,\n\nthe assessment was conducted through various ministerial website"],"full_content":"53\n\n**Chapter 2**\n\nChapter 2 • Global trends in e-Government\n\n### **2\\.5.3 OSI institutional framework subindex**\n\nThe organizational aspects of the institutional framework, which orient users on engaging government agencies through the online platforms, are well developed across the board (see figure 2.10). All Member States except Belize\\* have fully operational national portals. The vast majority of countries (95 per cent) make the government organizational chart and information on the government structure available on these portals, 91 per cent provide the names and titles of the heads of government agencies, departments and ministries, 81 per cent furnish information on the national chief information officer (cIO) or the equivalent, and 75 per cent share links to subnational or local government agencies. More than 90 per cent of the countries have national portals that provide links to ministerial websites and offer sources of information on sector-specific policies.\n\nFigure 2.10 Percentage of countries addressing various aspects of the institutional framework, 2024\n\n_Sources_ : 2024 United Nations E-Government Survey.\n\nEDGI 2024 levels\n\nVery high EGDI High EGDI Middle EGDI Low EGDI\n\n## Very high EGDI High EGDI Middle EGDI Low EGDI\n\n75%\n\n81%\n\n91%\n\n95%\n\nLinks to subnational/local government agencies\n\nInformation about Chief Information Offcer (CIO)\n\nNames and titles of heads of government agencies/departments/ministries\n\nOrganizational structure and/or chart of the government\n\nOrganizational framework\n\n89%\n\n93%\n\n95%\n\n94%\n\n90%\n\n95%\n\nSocial protection\n\nJustice\n\nHealth\n\nEnvironment\n\nEmployment\n\nEducation\n\nLinks to sectoral or ministerial websites\n\n90%\n\n91%\n\n94%\n\n92%\n\n90%\n\n92%\n\nSocial protection\n\nJustice\n\nHealth\n\nEnvironment\n\nEmployment\n\nEducation\n\nInformation on sectoral policies\n\nThe legislative, policy and strategy aspects of the institutional framework are less consistently developed (see figure 2.11). Between 81 and 83 per cent of countries have national e-government strategies, policies or legislation on cybersecurity, data privacy, and data protection, 78 per cent require citizens to use a digital ID when accessing public services, and 76 per cent offer public access to legislation on the right of citizens to access government information. The shares are lower for legislation or policies on open government data (63 per cent), e-participation (51 per cent), protecting the public against misinformation, disinformation, and/or fake news (47 per cent), and frontier technologies such as cloud computing (44 per cent) and artificial intelligence (42 per cent). Regional disparities are more pronounced for these last five indicators; more than half of the countries in Asia and Europe have already adopted relevant legislation, policies or strategies, the average for the Americas is slightly above 30 per cent and that for Africa slightly below 30 per cent, and in Oceania the compliance rates range from 4 to 11 per cent.\n\n\\* At the time of the assessment for the Survey, the national portal of Belize was under review and inaccessible. To the extent possible,\n\nthe assessment was conducted through various ministerial website","publish_date":null,"title":"Very high EGDI High EGDI Middle EGDI Low EGDI","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/%28Web_version%29_E-Government_Survey_2024_1392024_p78.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/%28Web_version%29_E-Government_Survey_2024_1392024_p78.pdf"},"request":{"example_id":"chart/(Web_version)_E-Government_Survey_2024_1392024_p78","product_type":"parse","source_file_path":"(Web_version)_E-Government_Survey_2024_1392024_p78.pdf"},"started_at":"2026-11-04T19:45:59.441294"} +{"completed_at":"2026-11-04T19:46:02.845371","latency_in_ms":1543,"output":{"example_id":"chart/(Web_version)_E-Government_Survey_2024_1392024_p79","grounded_pages":[],"layout_pages":[],"markdown":"2024 UN E-GovErNmENt SUrvEy\n\n54\n\n**Chapter 2**\n\n### **2\\.5.4 OSI content provision subindex**\n\nIn most countries (86 per cent), the Government provides information and services in multiple languages, which strengthens inclusiveness and facilitates access to information and online services in multilingual societies (see figure 2.12). However, fewer than half (47 per cent) proactively share web statistics on usage such as the number of new visits, total page views, or average time spent on site on their national portals.\n\nFigure 2.11 Percentage of countries with legislative frameworks relevant to e-government development, 2024\n\nEDGI 2024 levels\n\nVery high EGDI High EGDI Middle EGDI Low EGDI\n\n## Very high EGDI High EGDI Middle EGDI Low EGDI\n\nCloud, 44%\n\nAI, 42%\n\nE-participation, 51%\n\nMisinformation, disinformation or fake news, 47%\n\nOpen government data, 63% Digital ID, 78% E-government or digital government, 81% Privacy statement on government website, 83%\n\nFreedom of or access to information act, 76%\n\nNational data strategy or policy, 81%\n\nPersonal data protection, 83%\n\nCybersecurity, 82%\n\nAvailability of legislation, policy or strategy documents on:\n\n**193 UN Member States**\n\n193 UN Member States\n\nAmericas\n\nAfrica\n\nAsia\n\nEurope\n\nOceania\n\n15%\n\n22%\n\n26%\n\n28%\n\n48%\n\n85%\n\n78%\n\n74%\n\n72%\n\n52%\n\nCloud\n\nAI\n\nE-participation\n\nMisinformation, disinformation or fake news\n\nOpen government data\n\nAfrica (54 countries)\n\n24%\n\n24%\n\n33%\n\n28%\n\n35%\n\nAmericas (35 countries)\n\n48%\n\n39%\n\n57%\n\n57%\n\n59%\n\n52%\n\n61%\n\n43%\n\n43%\n\n41%\n\nAsia (47 countries)\n\n57%\n\n59%\n\n61%\n\n52%\n\n74%\n\n43%\n\n41%\n\n39%\n\n48%\n\n26%\n\nEurope (43 countries)\n\n11%\n\n7%\n\n6%\n\n4%\n\n7%\n\n89%\n\n93%\n\n94%\n\n96%\n\n93%\n\nOceania (14 countries)\n\n76%\n\n76%\n\n67%\n\n72%\n\n65%\n\n15%\n\n22%\n\n26%\n\n28%\n\n48%\n\nCloud\n\nAI\n\nE-participation\n\nsinformation, disinformation or fake news\n\nOpen government data\n\n### Africa (54 countries)\n\n24%\n\n24%\n\n33%\n\n28%\n\n35%\n\n### Americas (35 countries)\n\n48%\n\n39%\n\n57%\n\n57%\n\n59%\n\n### Asia (47 countries)\n\n57%\n\n59%\n\n61%\n\n52%\n\n74%\n\n### Europe countr\n\nSource: 2024 United Nations E-Government Survey.\n\nFigure 2.12 Content provision on national portals, 2024\n\n_(Percentage of countries, by region)_\n\n193 UN Member States\n\nAfrica\n\n86%\n\n47%\n\n89%\n\n78%\n\nAvailability of content in more than one offcial language of the country\n\nAvailability of web statistics on usage\n\nAnnouncements of forthcoming procurement or bidding processes published on the portal\n\nInformation provided about the results of government procurement or bidding online\n\n193 UN Member States\n\n100%\n\n70%\n\n94%\n\n83%\n\nAsia\n\n83%\n\n40%\n\n89%\n\n80%\n\nAmericas\n\n100%\n\n51%\n\n95%\n\n98%\n\nEurope\n\n50%\n\n36%\n\n93%\n\n57%\n\nOceania\n\n74%\n\n30%\n\n78%\n\n61%\n\nAvailability of content in more than one offcial language of the country\n\nAvailability of web statistics on usage\n\nAnnouncements of forthcoming procurement or bidding processes published on the portal\n\nInformation provided about the results of government procurement or bidding online\n\nAfrica\n\n_Source_ : 2024 United Nations E-Government Survey.","pages":[{"line_numbers":[],"markdown":"2024 UN E-GovErNmENt SUrvEy\n\n54\n\n**Chapter 2**\n\n### **2\\.5.4 OSI content provision subindex**\n\nIn most countries (86 per cent), the Government provides information and services in multiple languages, which strengthens inclusiveness and facilitates access to information and online services in multilingual societies (see figure 2.12). However, fewer than half (47 per cent) proactively share web statistics on usage such as the number of new visits, total page views, or average time spent on site on their national portals.\n\nFigure 2.11 Percentage of countries with legislative frameworks relevant to e-government development, 2024\n\nEDGI 2024 levels\n\nVery high EGDI High EGDI Middle EGDI Low EGDI\n\n## Very high EGDI High EGDI Middle EGDI Low EGDI\n\nCloud, 44%\n\nAI, 42%\n\nE-participation, 51%\n\nMisinformation, disinformation or fake news, 47%\n\nOpen government data, 63% Digital ID, 78% E-government or digital government, 81% Privacy statement on government website, 83%\n\nFreedom of or access to information act, 76%\n\nNational data strategy or policy, 81%\n\nPersonal data protection, 83%\n\nCybersecurity, 82%\n\nAvailability of legislation, policy or strategy documents on:\n\n**193 UN Member States**\n\n193 UN Member States\n\nAmericas\n\nAfrica\n\nAsia\n\nEurope\n\nOceania\n\n15%\n\n22%\n\n26%\n\n28%\n\n48%\n\n85%\n\n78%\n\n74%\n\n72%\n\n52%\n\nCloud\n\nAI\n\nE-participation\n\nMisinformation, disinformation or fake news\n\nOpen government data\n\nAfrica (54 countries)\n\n24%\n\n24%\n\n33%\n\n28%\n\n35%\n\nAmericas (35 countries)\n\n48%\n\n39%\n\n57%\n\n57%\n\n59%\n\n52%\n\n61%\n\n43%\n\n43%\n\n41%\n\nAsia (47 countries)\n\n57%\n\n59%\n\n61%\n\n52%\n\n74%\n\n43%\n\n41%\n\n39%\n\n48%\n\n26%\n\nEurope (43 countries)\n\n11%\n\n7%\n\n6%\n\n4%\n\n7%\n\n89%\n\n93%\n\n94%\n\n96%\n\n93%\n\nOceania (14 countries)\n\n76%\n\n76%\n\n67%\n\n72%\n\n65%\n\n15%\n\n22%\n\n26%\n\n28%\n\n48%\n\nCloud\n\nAI\n\nE-participation\n\nsinformation, disinformation or fake news\n\nOpen government data\n\n### Africa (54 countries)\n\n24%\n\n24%\n\n33%\n\n28%\n\n35%\n\n### Americas (35 countries)\n\n48%\n\n39%\n\n57%\n\n57%\n\n59%\n\n### Asia (47 countries)\n\n57%\n\n59%\n\n61%\n\n52%\n\n74%\n\n### Europe countr\n\nSource: 2024 United Nations E-Government Survey.\n\nFigure 2.12 Content provision on national portals, 2024\n\n_(Percentage of countries, by region)_\n\n193 UN Member States\n\nAfrica\n\n86%\n\n47%\n\n89%\n\n78%\n\nAvailability of content in more than one offcial language of the country\n\nAvailability of web statistics on usage\n\nAnnouncements of forthcoming procurement or bidding processes published on the portal\n\nInformation provided about the results of government procurement or bidding online\n\n193 UN Member States\n\n100%\n\n70%\n\n94%\n\n83%\n\nAsia\n\n83%\n\n40%\n\n89%\n\n80%\n\nAmericas\n\n100%\n\n51%\n\n95%\n\n98%\n\nEurope\n\n50%\n\n36%\n\n93%\n\n57%\n\nOceania\n\n74%\n\n30%\n\n78%\n\n61%\n\nAvailability of content in more than one offcial language of the country\n\nAvailability of web statistics on usage\n\nAnnouncements of forthcoming procurement or bidding processes published on the portal\n\nInformation provided about the results of government procurement or bidding online\n\nAfrica\n\n_Source_ : 2024 United Nations E-Government Survey.","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"2024 UN E-GovErNmENt SUrvEy\n\n54\n\n**Chapter 2**\n\n### **2\\.5.4 OSI content provision subindex**\n\nIn most countries (86 per cent), the Government provides information and services in multiple languages, which strengthens inclusiveness and facilitates access to information and online services in multilingual societies (see figure 2.12). However, fewer than half (47 per cent) proactively share web statistics on usage such as the number of new visits, total page views, or average time spent on site on their national portals.\n\nFigure 2.11 Percentage of countries with legislative frameworks relevant to e-government development, 2024\n\nEDGI 2024 levels\n\nVery high EGDI High EGDI Middle EGDI Low EGDI\n\n## Very high EGDI High EGDI Middle EGDI Low EGDI\n\nCloud, 44%\n\nAI, 42%\n\nE-participation, 51%\n\nMisinformation, disinformation or fake news, 47%\n\nOpen government data, 63% Digital ID, 78% E-government or digital government, 81% Privacy statement on government website, 83%\n\nFreedom of or access to information act, 76%\n\nNational data strategy or policy, 81%\n\nPersonal data protection, 83%\n\nCybersecurity, 82%\n\nAvailability of legislation, policy or strategy documents on:\n\n**193 UN Member States**\n\n193 UN Member States\n\nAmericas\n\nAfrica\n\nAsia\n\nEurope\n\nOceania\n\n15%\n\n22%\n\n26%\n\n28%\n\n48%\n\n85%\n\n78%\n\n74%\n\n72%\n\n52%\n\nCloud\n\nAI\n\nE-participation\n\nMisinformation, disinformation or fake news\n\nOpen government data\n\nAfrica (54 countries)\n\n24%\n\n24%\n\n33%\n\n28%\n\n35%\n\nAmericas (35 countries)\n\n48%\n\n39%\n\n57%\n\n57%\n\n59%\n\n52%\n\n61%\n\n43%\n\n43%\n\n41%\n\nAsia (47 countries)\n\n57%\n\n59%\n\n61%\n\n52%\n\n74%\n\n43%\n\n41%\n\n39%\n\n48%\n\n26%\n\nEurope (43 countries)\n\n11%\n\n7%\n\n6%\n\n4%\n\n7%\n\n89%\n\n93%\n\n94%\n\n96%\n\n93%\n\nOceania (14 countries)\n\n76%\n\n76%\n\n67%\n\n72%\n\n65%\n\n15%\n\n22%\n\n26%\n\n28%\n\n48%\n\nCloud\n\nAI\n\nE-participation\n\nsinformation, disinformation or fake news\n\nOpen government data\n\n### Africa (54 countries)\n\n24%\n\n24%\n\n33%\n\n28%\n\n35%\n\n### Americas (35 countries)\n\n48%\n\n39%\n\n57%\n\n57%\n\n59%\n\n### Asia (47 countries)\n\n57%\n\n59%\n\n61%\n\n52%\n\n74%\n\n### Europe countr\n\nSource: 2024 United Nations E-Government Survey.\n\nFigure 2.12 Content provision on national portals, 2024\n\n_(Percentage of countries, by region)_\n\n193 UN Member States\n\nAfrica\n\n86%\n\n47%\n\n89%\n\n78%\n\nAvailability of content in more than one offcial language of the country\n\nAvailability of web statistics on usage\n\nAnnouncements of forthcoming procurement or bidding processes published on the portal\n\nInformation provided about the results of government procurement or bidding online\n\n193 UN Member States\n\n100%\n\n70%\n\n94%\n\n83%\n\nAsia\n\n83%\n\n40%\n\n89%\n\n80%\n\nAmericas\n\n100%\n\n51%\n\n95%\n\n98%\n\nEurope\n\n50%\n\n36%\n\n93%\n\n57%\n\nOceania\n\n74%\n\n30%\n\n78%\n\n61%\n\nAvailability of content in more than one offcial language of the country\n\nAvailability of web statistics on usage\n\nAnnouncements of forthcoming procurement or bidding processes published on the portal\n\nInformation provided about the results of government procurement or bidding online\n\nAfrica\n\n_Source_ : 2024 United Nations E-Government Survey.","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/%28Web_version%29_E-Government_Survey_2024_1392024_p79.pdf"]},"response":{"errors":[],"extract_id":"extract_71d128b115521a110413c87ac7c4b6ba","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["2024 UN E-GovErNmENt SUrvEy\n\n54\n\n**Chapter 2**\n\n### **2\\.5.4 OSI content provision subindex**\n\nIn most countries (86 per cent), the Government provides information and services in multiple languages, which strengthens inclusiveness and facilitates access to information and online services in multilingual societies (see figure 2.12). However, fewer than half (47 per cent) proactively share web statistics on usage such as the number of new visits, total page views, or average time spent on site on their national portals.\n\nFigure 2.11 Percentage of countries with legislative frameworks relevant to e-government development, 2024\n\nEDGI 2024 levels\n\nVery high EGDI High EGDI Middle EGDI Low EGDI\n\n## Very high EGDI High EGDI Middle EGDI Low EGDI\n\nCloud, 44%\n\nAI, 42%\n\nE-participation, 51%\n\nMisinformation, disinformation or fake news, 47%\n\nOpen government data, 63% Digital ID, 78% E-government or digital government, 81% Privacy statement on government website, 83%\n\nFreedom of or access to information act, 76%\n\nNational data strategy or policy, 81%\n\nPersonal data protection, 83%\n\nCybersecurity, 82%\n\nAvailability of legislation, policy or strategy documents on:\n\n**193 UN Member States**\n\n193 UN Member States\n\nAmericas\n\nAfrica\n\nAsia\n\nEurope\n\nOceania\n\n15%\n\n22%\n\n26%\n\n28%\n\n48%\n\n85%\n\n78%\n\n74%\n\n72%\n\n52%\n\nCloud\n\nAI\n\nE-participation\n\nMisinformation, disinformation or fake news\n\nOpen government data\n\nAfrica (54 countries)\n\n24%\n\n24%\n\n33%\n\n28%\n\n35%\n\nAmericas (35 countries)\n\n48%\n\n39%\n\n57%\n\n57%\n\n59%\n\n52%\n\n61%\n\n43%\n\n43%\n\n41%\n\nAsia (47 countries)\n\n57%\n\n59%\n\n61%\n\n52%\n\n74%\n\n43%\n\n41%\n\n39%\n\n48%\n\n26%\n\nEurope (43 countries)\n\n11%\n\n7%\n\n6%\n\n4%\n\n7%\n\n89%\n\n93%\n\n94%\n\n96%\n\n93%\n\nOceania (14 countries)\n\n76%\n\n76%\n\n67%\n\n72%\n\n65%\n\n15%\n\n22%\n\n26%\n\n28%\n\n48%\n\nCloud\n\nAI\n\nE-participation\n\nsinformation, disinformation or fake news\n\nOpen government data\n\n### Africa (54 countries)\n\n24%\n\n24%\n\n33%\n\n28%\n\n35%\n\n### Americas (35 countries)\n\n48%\n\n39%\n\n57%\n\n57%\n\n59%\n\n### Asia (47 countries)\n\n57%\n\n59%\n\n61%\n\n52%\n\n74%\n\n### Europe countr\n\nSource: 2024 United Nations E-Government Survey.\n\nFigure 2.12 Content provision on national portals, 2024\n\n_(Percentage of countries, by region)_\n\n193 UN Member States\n\nAfrica\n\n86%\n\n47%\n\n89%\n\n78%\n\nAvailability of content in more than one offcial language of the country\n\nAvailability of web statistics on usage\n\nAnnouncements of forthcoming procurement or bidding processes published on the portal\n\nInformation provided about the results of government procurement or bidding online\n\n193 UN Member States\n\n100%\n\n70%\n\n94%\n\n83%\n\nAsia\n\n83%\n\n40%\n\n89%\n\n80%\n\nAmericas\n\n100%\n\n51%\n\n95%\n\n98%\n\nEurope\n\n50%\n\n36%\n\n93%\n\n57%\n\nOceania\n\n74%\n\n30%\n\n78%\n\n61%\n\nAvailability of content in more than one offcial language of the country\n\nAvailability of web statistics on usage\n\nAnnouncements of forthcoming procurement or bidding processes published on the portal\n\nInformation provided about the results of government procurement or bidding online\n\nAfrica\n\n_Source_ : 2024 United Nations E-Government Survey."],"full_content":"2024 UN E-GovErNmENt SUrvEy\n\n54\n\n**Chapter 2**\n\n### **2\\.5.4 OSI content provision subindex**\n\nIn most countries (86 per cent), the Government provides information and services in multiple languages, which strengthens inclusiveness and facilitates access to information and online services in multilingual societies (see figure 2.12). However, fewer than half (47 per cent) proactively share web statistics on usage such as the number of new visits, total page views, or average time spent on site on their national portals.\n\nFigure 2.11 Percentage of countries with legislative frameworks relevant to e-government development, 2024\n\nEDGI 2024 levels\n\nVery high EGDI High EGDI Middle EGDI Low EGDI\n\n## Very high EGDI High EGDI Middle EGDI Low EGDI\n\nCloud, 44%\n\nAI, 42%\n\nE-participation, 51%\n\nMisinformation, disinformation or fake news, 47%\n\nOpen government data, 63% Digital ID, 78% E-government or digital government, 81% Privacy statement on government website, 83%\n\nFreedom of or access to information act, 76%\n\nNational data strategy or policy, 81%\n\nPersonal data protection, 83%\n\nCybersecurity, 82%\n\nAvailability of legislation, policy or strategy documents on:\n\n**193 UN Member States**\n\n193 UN Member States\n\nAmericas\n\nAfrica\n\nAsia\n\nEurope\n\nOceania\n\n15%\n\n22%\n\n26%\n\n28%\n\n48%\n\n85%\n\n78%\n\n74%\n\n72%\n\n52%\n\nCloud\n\nAI\n\nE-participation\n\nMisinformation, disinformation or fake news\n\nOpen government data\n\nAfrica (54 countries)\n\n24%\n\n24%\n\n33%\n\n28%\n\n35%\n\nAmericas (35 countries)\n\n48%\n\n39%\n\n57%\n\n57%\n\n59%\n\n52%\n\n61%\n\n43%\n\n43%\n\n41%\n\nAsia (47 countries)\n\n57%\n\n59%\n\n61%\n\n52%\n\n74%\n\n43%\n\n41%\n\n39%\n\n48%\n\n26%\n\nEurope (43 countries)\n\n11%\n\n7%\n\n6%\n\n4%\n\n7%\n\n89%\n\n93%\n\n94%\n\n96%\n\n93%\n\nOceania (14 countries)\n\n76%\n\n76%\n\n67%\n\n72%\n\n65%\n\n15%\n\n22%\n\n26%\n\n28%\n\n48%\n\nCloud\n\nAI\n\nE-participation\n\nsinformation, disinformation or fake news\n\nOpen government data\n\n### Africa (54 countries)\n\n24%\n\n24%\n\n33%\n\n28%\n\n35%\n\n### Americas (35 countries)\n\n48%\n\n39%\n\n57%\n\n57%\n\n59%\n\n### Asia (47 countries)\n\n57%\n\n59%\n\n61%\n\n52%\n\n74%\n\n### Europe countr\n\nSource: 2024 United Nations E-Government Survey.\n\nFigure 2.12 Content provision on national portals, 2024\n\n_(Percentage of countries, by region)_\n\n193 UN Member States\n\nAfrica\n\n86%\n\n47%\n\n89%\n\n78%\n\nAvailability of content in more than one offcial language of the country\n\nAvailability of web statistics on usage\n\nAnnouncements of forthcoming procurement or bidding processes published on the portal\n\nInformation provided about the results of government procurement or bidding online\n\n193 UN Member States\n\n100%\n\n70%\n\n94%\n\n83%\n\nAsia\n\n83%\n\n40%\n\n89%\n\n80%\n\nAmericas\n\n100%\n\n51%\n\n95%\n\n98%\n\nEurope\n\n50%\n\n36%\n\n93%\n\n57%\n\nOceania\n\n74%\n\n30%\n\n78%\n\n61%\n\nAvailability of content in more than one offcial language of the country\n\nAvailability of web statistics on usage\n\nAnnouncements of forthcoming procurement or bidding processes published on the portal\n\nInformation provided about the results of government procurement or bidding online\n\nAfrica\n\n_Source_ : 2024 United Nations E-Government Survey.","publish_date":null,"title":"Oceania (14 countries)","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/%28Web_version%29_E-Government_Survey_2024_1392024_p79.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/%28Web_version%29_E-Government_Survey_2024_1392024_p79.pdf"},"request":{"example_id":"chart/(Web_version)_E-Government_Survey_2024_1392024_p79","product_type":"parse","source_file_path":"(Web_version)_E-Government_Survey_2024_1392024_p79.pdf"},"started_at":"2026-11-04T19:46:01.301837"} +{"completed_at":"2026-11-04T19:46:05.455816","latency_in_ms":3081,"output":{"example_id":"chart/(Web_version)_E-Government_Survey_2024_1392024_p81","grounded_pages":[],"layout_pages":[],"markdown":"2024 UN E-GovErNmENt SUrvEy\n\n56\n\n**Chapter 2**\n\nFigure 2.13 Distribution of EGDI levels relative to EPI levels for United Nations Member States, 2024\n\nAlthough e-participation is the least advanced among the five OSI subindices (see figure 2.9), the global average EPI value has increased by 9 per cent, from 0.4450 to 0.4893, since 2022 (see figure 2.14).\n\n_Source_ : 2024 United Nations E-Government Survey.\n\nFigure 2.14 Average EPI values by region and percentage change between 2022 and 2024\n\n_Sources_ : 2022 and 2024 United Nations E-Government Surveys.\n\nAfrica Americas Asia Europe Oceania 193 UN member states\n\n0\\.00\n\n0\\.25\n\n0\\.50\n\n0\\.75\n\n1\\.00\n\n0\\.2973 0.2595\n\n12\\.73%\n\n0\\.4736 0.4253\n\n10\\.19%\n\n0\\.5404 0.5024\n\n7\\.02%\n\n0\\.7247 0.6631\n\n8\\.51%\n\n0\\.3748 0.3474\n\n7\\.30%\n\n0\\.4893 0.4450\n\n9\\.06%\n\nEPI average 2022\n\nEPI 2024\n\nEPI average 2024 Africa\n\nEPI average 2024 Africa Americas Asia Europe Oceania 193 UN member states\n\nEDGI 2024 levels Very high EGDI High EGDI Middle EGDI Low EGDI\n\nALB\n\nARE\n\nARG\n\nARM AUS\n\nAUT\n\nAZE BEL\n\nBGR\n\nBRA\n\nBRN\n\nCAN\n\nCHN DEU\n\nDNK\n\nECU\n\nFIN\n\nFRA\n\nGEO\n\nGRC\n\nHRV\n\nHUN\n\nIDN\n\nKWT\n\nLTU\n\nMDA\n\nMUS\n\nNLD\n\nPOL\n\nQAT\n\nROU\n\nVNM\n\nAND\n\nATG\n\nBGD\n\nBHS\n\nBIH\n\nBRB\n\nBTN\n\nBWA\n\nCIV\n\nDMA\n\nDOM\n\nDZA\n\nGAB\n\nGRD\n\nGTM\n\nGUY\n\nIND\n\nIRN\n\nKHM\n\nKNA\n\nLBN\n\nLCA\n\nLKA MAR\n\nMCO\n\nMDV\n\nMKD\n\nNIC\n\nNPL\n\nPAK\n\nRWA\n\nSEN\n\nSMR\n\nTJK\n\nTON\n\nAGO\n\nBEN\n\nBLZ\n\nCOD\n\nCOG CUB\n\nDJI\n\nETH\n\nGIN\n\nGMB\n\nGNB\n\nHND\n\nIRQ\n\nKIR\n\nLBR\n\nLSO\n\nMDG\n\nMLI\n\nMRT\n\nMWI\n\nNRU\n\nSLB\n\nSTP\n\nSYR\n\nTLS\n\nTUV\n\nUGA\n\nCAF HTI\n\nNER\n\nSOM\n\nTCD\n\ny = 1.04443x-0.176424 R-Squared = 0\\.731916 p-value < 0.0001\n\n0\\.000.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.45 0.50 0.55 0.60 0.65 0.70 0.75 0.80 0.85 0.90 0.951.00\n\n_EGDI 2024_\n\n0\\.0\n\n0\\.1\n\n0\\.2\n\n0\\.3\n\n0\\.4\n\n0\\.5\n\n0\\.6\n\n0\\.7\n\n0\\.8\n\n0\\.9\n\n1\\.0\n\n_EPI 2024_\n\ncorrelation line","pages":[{"line_numbers":[],"markdown":"2024 UN E-GovErNmENt SUrvEy\n\n56\n\n**Chapter 2**\n\nFigure 2.13 Distribution of EGDI levels relative to EPI levels for United Nations Member States, 2024\n\nAlthough e-participation is the least advanced among the five OSI subindices (see figure 2.9), the global average EPI value has increased by 9 per cent, from 0.4450 to 0.4893, since 2022 (see figure 2.14).\n\n_Source_ : 2024 United Nations E-Government Survey.\n\nFigure 2.14 Average EPI values by region and percentage change between 2022 and 2024\n\n_Sources_ : 2022 and 2024 United Nations E-Government Surveys.\n\nAfrica Americas Asia Europe Oceania 193 UN member states\n\n0\\.00\n\n0\\.25\n\n0\\.50\n\n0\\.75\n\n1\\.00\n\n0\\.2973 0.2595\n\n12\\.73%\n\n0\\.4736 0.4253\n\n10\\.19%\n\n0\\.5404 0.5024\n\n7\\.02%\n\n0\\.7247 0.6631\n\n8\\.51%\n\n0\\.3748 0.3474\n\n7\\.30%\n\n0\\.4893 0.4450\n\n9\\.06%\n\nEPI average 2022\n\nEPI 2024\n\nEPI average 2024 Africa\n\nEPI average 2024 Africa Americas Asia Europe Oceania 193 UN member states\n\nEDGI 2024 levels Very high EGDI High EGDI Middle EGDI Low EGDI\n\nALB\n\nARE\n\nARG\n\nARM AUS\n\nAUT\n\nAZE BEL\n\nBGR\n\nBRA\n\nBRN\n\nCAN\n\nCHN DEU\n\nDNK\n\nECU\n\nFIN\n\nFRA\n\nGEO\n\nGRC\n\nHRV\n\nHUN\n\nIDN\n\nKWT\n\nLTU\n\nMDA\n\nMUS\n\nNLD\n\nPOL\n\nQAT\n\nROU\n\nVNM\n\nAND\n\nATG\n\nBGD\n\nBHS\n\nBIH\n\nBRB\n\nBTN\n\nBWA\n\nCIV\n\nDMA\n\nDOM\n\nDZA\n\nGAB\n\nGRD\n\nGTM\n\nGUY\n\nIND\n\nIRN\n\nKHM\n\nKNA\n\nLBN\n\nLCA\n\nLKA MAR\n\nMCO\n\nMDV\n\nMKD\n\nNIC\n\nNPL\n\nPAK\n\nRWA\n\nSEN\n\nSMR\n\nTJK\n\nTON\n\nAGO\n\nBEN\n\nBLZ\n\nCOD\n\nCOG CUB\n\nDJI\n\nETH\n\nGIN\n\nGMB\n\nGNB\n\nHND\n\nIRQ\n\nKIR\n\nLBR\n\nLSO\n\nMDG\n\nMLI\n\nMRT\n\nMWI\n\nNRU\n\nSLB\n\nSTP\n\nSYR\n\nTLS\n\nTUV\n\nUGA\n\nCAF HTI\n\nNER\n\nSOM\n\nTCD\n\ny = 1.04443x-0.176424 R-Squared = 0\\.731916 p-value < 0.0001\n\n0\\.000.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.45 0.50 0.55 0.60 0.65 0.70 0.75 0.80 0.85 0.90 0.951.00\n\n_EGDI 2024_\n\n0\\.0\n\n0\\.1\n\n0\\.2\n\n0\\.3\n\n0\\.4\n\n0\\.5\n\n0\\.6\n\n0\\.7\n\n0\\.8\n\n0\\.9\n\n1\\.0\n\n_EPI 2024_\n\ncorrelation line","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"2024 UN E-GovErNmENt SUrvEy\n\n56\n\n**Chapter 2**\n\nFigure 2.13 Distribution of EGDI levels relative to EPI levels for United Nations Member States, 2024\n\nAlthough e-participation is the least advanced among the five OSI subindices (see figure 2.9), the global average EPI value has increased by 9 per cent, from 0.4450 to 0.4893, since 2022 (see figure 2.14).\n\n_Source_ : 2024 United Nations E-Government Survey.\n\nFigure 2.14 Average EPI values by region and percentage change between 2022 and 2024\n\n_Sources_ : 2022 and 2024 United Nations E-Government Surveys.\n\nAfrica Americas Asia Europe Oceania 193 UN member states\n\n0\\.00\n\n0\\.25\n\n0\\.50\n\n0\\.75\n\n1\\.00\n\n0\\.2973 0.2595\n\n12\\.73%\n\n0\\.4736 0.4253\n\n10\\.19%\n\n0\\.5404 0.5024\n\n7\\.02%\n\n0\\.7247 0.6631\n\n8\\.51%\n\n0\\.3748 0.3474\n\n7\\.30%\n\n0\\.4893 0.4450\n\n9\\.06%\n\nEPI average 2022\n\nEPI 2024\n\nEPI average 2024 Africa\n\nEPI average 2024 Africa Americas Asia Europe Oceania 193 UN member states\n\nEDGI 2024 levels Very high EGDI High EGDI Middle EGDI Low EGDI\n\nALB\n\nARE\n\nARG\n\nARM AUS\n\nAUT\n\nAZE BEL\n\nBGR\n\nBRA\n\nBRN\n\nCAN\n\nCHN DEU\n\nDNK\n\nECU\n\nFIN\n\nFRA\n\nGEO\n\nGRC\n\nHRV\n\nHUN\n\nIDN\n\nKWT\n\nLTU\n\nMDA\n\nMUS\n\nNLD\n\nPOL\n\nQAT\n\nROU\n\nVNM\n\nAND\n\nATG\n\nBGD\n\nBHS\n\nBIH\n\nBRB\n\nBTN\n\nBWA\n\nCIV\n\nDMA\n\nDOM\n\nDZA\n\nGAB\n\nGRD\n\nGTM\n\nGUY\n\nIND\n\nIRN\n\nKHM\n\nKNA\n\nLBN\n\nLCA\n\nLKA MAR\n\nMCO\n\nMDV\n\nMKD\n\nNIC\n\nNPL\n\nPAK\n\nRWA\n\nSEN\n\nSMR\n\nTJK\n\nTON\n\nAGO\n\nBEN\n\nBLZ\n\nCOD\n\nCOG CUB\n\nDJI\n\nETH\n\nGIN\n\nGMB\n\nGNB\n\nHND\n\nIRQ\n\nKIR\n\nLBR\n\nLSO\n\nMDG\n\nMLI\n\nMRT\n\nMWI\n\nNRU\n\nSLB\n\nSTP\n\nSYR\n\nTLS\n\nTUV\n\nUGA\n\nCAF HTI\n\nNER\n\nSOM\n\nTCD\n\ny = 1.04443x-0.176424 R-Squared = 0\\.731916 p-value < 0.0001\n\n0\\.000.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.45 0.50 0.55 0.60 0.65 0.70 0.75 0.80 0.85 0.90 0.951.00\n\n_EGDI 2024_\n\n0\\.0\n\n0\\.1\n\n0\\.2\n\n0\\.3\n\n0\\.4\n\n0\\.5\n\n0\\.6\n\n0\\.7\n\n0\\.8\n\n0\\.9\n\n1\\.0\n\n_EPI 2024_\n\ncorrelation line","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/%28Web_version%29_E-Government_Survey_2024_1392024_p81.pdf"]},"response":{"errors":[],"extract_id":"extract_0de0ecaec49710ddff87cbae037bc82b","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["2024 UN E-GovErNmENt SUrvEy\n\n56\n\n**Chapter 2**\n\nFigure 2.13 Distribution of EGDI levels relative to EPI levels for United Nations Member States, 2024\n\nAlthough e-participation is the least advanced among the five OSI subindices (see figure 2.9), the global average EPI value has increased by 9 per cent, from 0.4450 to 0.4893, since 2022 (see figure 2.14).\n\n_Source_ : 2024 United Nations E-Government Survey.\n\nFigure 2.14 Average EPI values by region and percentage change between 2022 and 2024\n\n_Sources_ : 2022 and 2024 United Nations E-Government Surveys.\n\nAfrica Americas Asia Europe Oceania 193 UN member states\n\n0\\.00\n\n0\\.25\n\n0\\.50\n\n0\\.75\n\n1\\.00\n\n0\\.2973 0.2595\n\n12\\.73%\n\n0\\.4736 0.4253\n\n10\\.19%\n\n0\\.5404 0.5024\n\n7\\.02%\n\n0\\.7247 0.6631\n\n8\\.51%\n\n0\\.3748 0.3474\n\n7\\.30%\n\n0\\.4893 0.4450\n\n9\\.06%\n\nEPI average 2022\n\nEPI 2024\n\nEPI average 2024 Africa\n\nEPI average 2024 Africa Americas Asia Europe Oceania 193 UN member states\n\nEDGI 2024 levels Very high EGDI High EGDI Middle EGDI Low EGDI\n\nALB\n\nARE\n\nARG\n\nARM AUS\n\nAUT\n\nAZE BEL\n\nBGR\n\nBRA\n\nBRN\n\nCAN\n\nCHN DEU\n\nDNK\n\nECU\n\nFIN\n\nFRA\n\nGEO\n\nGRC\n\nHRV\n\nHUN\n\nIDN\n\nKWT\n\nLTU\n\nMDA\n\nMUS\n\nNLD\n\nPOL\n\nQAT\n\nROU\n\nVNM\n\nAND\n\nATG\n\nBGD\n\nBHS\n\nBIH\n\nBRB\n\nBTN\n\nBWA\n\nCIV\n\nDMA\n\nDOM\n\nDZA\n\nGAB\n\nGRD\n\nGTM\n\nGUY\n\nIND\n\nIRN\n\nKHM\n\nKNA\n\nLBN\n\nLCA\n\nLKA MAR\n\nMCO\n\nMDV\n\nMKD\n\nNIC\n\nNPL\n\nPAK\n\nRWA\n\nSEN\n\nSMR\n\nTJK\n\nTON\n\nAGO\n\nBEN\n\nBLZ\n\nCOD\n\nCOG CUB\n\nDJI\n\nETH\n\nGIN\n\nGMB\n\nGNB\n\nHND\n\nIRQ\n\nKIR\n\nLBR\n\nLSO\n\nMDG\n\nMLI\n\nMRT\n\nMWI\n\nNRU\n\nSLB\n\nSTP\n\nSYR\n\nTLS\n\nTUV\n\nUGA\n\nCAF HTI\n\nNER\n\nSOM\n\nTCD\n\ny = 1.04443x-0.176424 R-Squared = 0\\.731916 p-value < 0.0001\n\n0\\.000.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.45 0.50 0.55 0.60 0.65 0.70 0.75 0.80 0.85 0.90 0.951.00\n\n_EGDI 2024_\n\n0\\.0\n\n0\\.1\n\n0\\.2\n\n0\\.3\n\n0\\.4\n\n0\\.5\n\n0\\.6\n\n0\\.7\n\n0\\.8\n\n0\\.9\n\n1\\.0\n\n_EPI 2024_\n\ncorrelation line"],"full_content":"2024 UN E-GovErNmENt SUrvEy\n\n56\n\n**Chapter 2**\n\nFigure 2.13 Distribution of EGDI levels relative to EPI levels for United Nations Member States, 2024\n\nAlthough e-participation is the least advanced among the five OSI subindices (see figure 2.9), the global average EPI value has increased by 9 per cent, from 0.4450 to 0.4893, since 2022 (see figure 2.14).\n\n_Source_ : 2024 United Nations E-Government Survey.\n\nFigure 2.14 Average EPI values by region and percentage change between 2022 and 2024\n\n_Sources_ : 2022 and 2024 United Nations E-Government Surveys.\n\nAfrica Americas Asia Europe Oceania 193 UN member states\n\n0\\.00\n\n0\\.25\n\n0\\.50\n\n0\\.75\n\n1\\.00\n\n0\\.2973 0.2595\n\n12\\.73%\n\n0\\.4736 0.4253\n\n10\\.19%\n\n0\\.5404 0.5024\n\n7\\.02%\n\n0\\.7247 0.6631\n\n8\\.51%\n\n0\\.3748 0.3474\n\n7\\.30%\n\n0\\.4893 0.4450\n\n9\\.06%\n\nEPI average 2022\n\nEPI 2024\n\nEPI average 2024 Africa\n\nEPI average 2024 Africa Americas Asia Europe Oceania 193 UN member states\n\nEDGI 2024 levels Very high EGDI High EGDI Middle EGDI Low EGDI\n\nALB\n\nARE\n\nARG\n\nARM AUS\n\nAUT\n\nAZE BEL\n\nBGR\n\nBRA\n\nBRN\n\nCAN\n\nCHN DEU\n\nDNK\n\nECU\n\nFIN\n\nFRA\n\nGEO\n\nGRC\n\nHRV\n\nHUN\n\nIDN\n\nKWT\n\nLTU\n\nMDA\n\nMUS\n\nNLD\n\nPOL\n\nQAT\n\nROU\n\nVNM\n\nAND\n\nATG\n\nBGD\n\nBHS\n\nBIH\n\nBRB\n\nBTN\n\nBWA\n\nCIV\n\nDMA\n\nDOM\n\nDZA\n\nGAB\n\nGRD\n\nGTM\n\nGUY\n\nIND\n\nIRN\n\nKHM\n\nKNA\n\nLBN\n\nLCA\n\nLKA MAR\n\nMCO\n\nMDV\n\nMKD\n\nNIC\n\nNPL\n\nPAK\n\nRWA\n\nSEN\n\nSMR\n\nTJK\n\nTON\n\nAGO\n\nBEN\n\nBLZ\n\nCOD\n\nCOG CUB\n\nDJI\n\nETH\n\nGIN\n\nGMB\n\nGNB\n\nHND\n\nIRQ\n\nKIR\n\nLBR\n\nLSO\n\nMDG\n\nMLI\n\nMRT\n\nMWI\n\nNRU\n\nSLB\n\nSTP\n\nSYR\n\nTLS\n\nTUV\n\nUGA\n\nCAF HTI\n\nNER\n\nSOM\n\nTCD\n\ny = 1.04443x-0.176424 R-Squared = 0\\.731916 p-value < 0.0001\n\n0\\.000.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.45 0.50 0.55 0.60 0.65 0.70 0.75 0.80 0.85 0.90 0.951.00\n\n_EGDI 2024_\n\n0\\.0\n\n0\\.1\n\n0\\.2\n\n0\\.3\n\n0\\.4\n\n0\\.5\n\n0\\.6\n\n0\\.7\n\n0\\.8\n\n0\\.9\n\n1\\.0\n\n_EPI 2024_\n\ncorrelation line","publish_date":null,"title":"Very high EGDI High EGDI Middle EGDI Low EGDI","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/%28Web_version%29_E-Government_Survey_2024_1392024_p81.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/%28Web_version%29_E-Government_Survey_2024_1392024_p81.pdf"},"request":{"example_id":"chart/(Web_version)_E-Government_Survey_2024_1392024_p81","product_type":"parse","source_file_path":"(Web_version)_E-Government_Survey_2024_1392024_p81.pdf"},"started_at":"2026-11-04T19:46:02.374533"} +{"completed_at":"2026-11-04T19:46:06.434801","latency_in_ms":2541,"output":{"example_id":"chart/(Web_version)_E-Government_Survey_2024_1392024_p86","grounded_pages":[],"layout_pages":[],"markdown":"61\n\n**Chapter 2**\n\nChapter 2 • Global trends in e-Government\n\nFigure 2.20 Percentage of countries offering various e-participation mechanisms and tools, 2024\n\n_Source_ : 2024 United Nations E-Government Survey.\n\nAround 50 per cent of the Member States have a dedicated e-participation portal, 55 per cent publish calendar announcements about upcoming consultations and other participatory activities, and about 40 per cent of use e-petitions or similar mechanisms to engage the population in policy deliberations. The region with the highest proportion of countries providing evidence of having conducted at least one e-consultation in the 12 months preceding the administration of the Survey is Europe (91 per cent), followed by Asia (70 per cent), the Americas (60 per cent), Africa (24 per cent) and Oceania (14 per cent).\n\nInformation on the results of such deliberations or on how people’s voices are included in policy- related decision-making is published by an average of only 31 per cent of the countries surveyed, though regional disparities are quite pronounced; nearly half of the European countries, slightly over one third of countries in Asia and Oceania, just under a third of countries in the Americas, and only a tenth of the countries in Africa publish the results of public deliberations on their portals.\n\nIn terms of the proportion of countries in each region proactively offering e-participation channels, tools or mechanisms, Europe is the leader (78 per cent), followed by Asia (62 per cent), the Americas (44 per cent), Oceania (33 per cent), and Africa (27 per cent).\n\n### **2\\.5.6 OSI services provision subindex: progress in online services delivery**\n\nThe services provision subindex of the OSI assesses the availability of various online transactional services, how government services are accessed (through one main portal or multiple channels), the existence and functionality of e-procurement platforms and digital invoicing, the integration of GIS or geospatial data and technologies in online services provision, and the availability of sector- specific services and services for people in vulnerable situations. The data analysis and key findings are presented below.\n\nFor the 2024 Survey, 25 online services have been assessed, up from 22 for the last Survey cycle. The number of countries offering at least one of these online services has not changed, remaining at 189 (98 per cent) of the 193 Member States. Figure 2.21 illustrates the overall expansion in the range of services provided, which is reflected in the increased number of countries providing each type of service.\n\nEDGI 2024 levels\n\nVery high EGDI High EGDI Middle EGDI Low EGDI\n\n### Very high EGDI High EGDI Middle EGDI Low EGDI\n\n193 UN Member States\n\nAmericas\n\nAfrica\n\nAsia\n\nEurope\n\nOceania\n\n### Very high EGDI High EGDI Middle EGDI Low EGDI\n\nPortal provides a way for people to report labour law violations, 53.4%\n\nPortal provides a way for people to report corruption among public servants or institutions, 67.4%\n\nGovernment publishes information on how people's voices, including those of vulnerable groups, are included in policy decision-making, 30.6%\n\nPortal offers e-petition or similar mechanisms, 39.9% Portal integrates e-tools for public consultation/deliberation, 57.0%\n\nPortal publishes calendar/announcements about e-participation activities, 54.9%\n\nThere is an e-participation portal, 50.3%\n\n48%\n\n24%\n\n9%\n\n19%\n\n30%\n\n30%\n\n31%\n\n57%\n\n46%\n\n29%\n\n34%\n\n51%\n\n49%\n\n40%\n\n81%\n\n72%\n\n38%\n\n45%\n\n68%\n\n72%\n\n57%\n\n88%\n\n84%\n\n49%\n\n74%\n\n79%\n\n88%\n\n86%\n\n57%\n\n29%\n\n36%\n\n14%\n\n43%\n\n36%\n\n14%\n\nPortal provides a way for people to report corruption among public servants or institutions\n\nPortal provides a way for people to report labour law violations\n\nGovernment publishes information on how people's voices, including those of vulnerable groups, are included in policy decision-making\n\nPortal offers e-petition or similar mechanisms\n\nPortal publishes calendar/announcements about e-participation activities\n\nPortal integrates e-tools for public consultation/deliberation\n\nThere is an e-participation portal\n\nAfrica Americas Asia Europe Oceania\n\nEDGI 2024 levels\n\nVery high EGDI High EGDI Middle EGDI Low EGDI\n\n### Very high EG High EGDI Middle EGDI Low EGDI\n\n193 UN Member States\n\nAmericas\n\nAfrica\n\nAsia\n\nEurope\n\nOceania\n\n### Very high EGDI High EGDI Middle EGDI Low EGDI\n\nPortal provides a way for people to report labour law violations, 53.4%\n\nPortal provides a way for people to report corruption among public servants or institutions, 67.4%\n\nGovernment publishes information on how people's voices, including those of vulnerable groups, are included in policy decision-making, 30.6%\n\nPortal offers e-petition or similar mechanisms, 39.9% Portal integrates e-tools for public consultation/deliberation, 57.0%\n\nPortal publishes calendar/announcements about e-participation activities, 54.9%\n\nThere is an e-participation portal, 50.3%\n\n48%\n\n24%\n\n9%\n\n19%\n\n30%\n\n30%\n\n31%\n\n57%\n\n46%\n\n29%\n\n34%\n\n51%\n\n49%\n\n40%\n\n81%\n\n72%\n\n38%\n\n45%\n\n68%\n\n72%\n\n57%\n\n88%\n\n84%\n\n49%\n\n74%\n\n79%\n\n88%\n\n86%\n\n57%\n\n29%\n\n36%\n\n14%\n\n43%\n\n36%\n\n14%\n\nPortal provides a way for people to report corruption among public servants or institutions\n\nPortal provides a way for people to report labour law violations\n\nGovernment publishes information on how people's voices, including those of vulnerable groups, are included in policy decision-making\n\nPortal offers e-petition or similar mechanisms\n\nPortal publishes calendar/announcements about e-participation activities\n\nPortal integrates e-tools for public consultation/deliberation\n\nThere is an e-participation portal\n\nAfrica Americas Asia Europe Oceania","pages":[{"line_numbers":[],"markdown":"61\n\n**Chapter 2**\n\nChapter 2 • Global trends in e-Government\n\nFigure 2.20 Percentage of countries offering various e-participation mechanisms and tools, 2024\n\n_Source_ : 2024 United Nations E-Government Survey.\n\nAround 50 per cent of the Member States have a dedicated e-participation portal, 55 per cent publish calendar announcements about upcoming consultations and other participatory activities, and about 40 per cent of use e-petitions or similar mechanisms to engage the population in policy deliberations. The region with the highest proportion of countries providing evidence of having conducted at least one e-consultation in the 12 months preceding the administration of the Survey is Europe (91 per cent), followed by Asia (70 per cent), the Americas (60 per cent), Africa (24 per cent) and Oceania (14 per cent).\n\nInformation on the results of such deliberations or on how people’s voices are included in policy- related decision-making is published by an average of only 31 per cent of the countries surveyed, though regional disparities are quite pronounced; nearly half of the European countries, slightly over one third of countries in Asia and Oceania, just under a third of countries in the Americas, and only a tenth of the countries in Africa publish the results of public deliberations on their portals.\n\nIn terms of the proportion of countries in each region proactively offering e-participation channels, tools or mechanisms, Europe is the leader (78 per cent), followed by Asia (62 per cent), the Americas (44 per cent), Oceania (33 per cent), and Africa (27 per cent).\n\n### **2\\.5.6 OSI services provision subindex: progress in online services delivery**\n\nThe services provision subindex of the OSI assesses the availability of various online transactional services, how government services are accessed (through one main portal or multiple channels), the existence and functionality of e-procurement platforms and digital invoicing, the integration of GIS or geospatial data and technologies in online services provision, and the availability of sector- specific services and services for people in vulnerable situations. The data analysis and key findings are presented below.\n\nFor the 2024 Survey, 25 online services have been assessed, up from 22 for the last Survey cycle. The number of countries offering at least one of these online services has not changed, remaining at 189 (98 per cent) of the 193 Member States. Figure 2.21 illustrates the overall expansion in the range of services provided, which is reflected in the increased number of countries providing each type of service.\n\nEDGI 2024 levels\n\nVery high EGDI High EGDI Middle EGDI Low EGDI\n\n### Very high EGDI High EGDI Middle EGDI Low EGDI\n\n193 UN Member States\n\nAmericas\n\nAfrica\n\nAsia\n\nEurope\n\nOceania\n\n### Very high EGDI High EGDI Middle EGDI Low EGDI\n\nPortal provides a way for people to report labour law violations, 53.4%\n\nPortal provides a way for people to report corruption among public servants or institutions, 67.4%\n\nGovernment publishes information on how people's voices, including those of vulnerable groups, are included in policy decision-making, 30.6%\n\nPortal offers e-petition or similar mechanisms, 39.9% Portal integrates e-tools for public consultation/deliberation, 57.0%\n\nPortal publishes calendar/announcements about e-participation activities, 54.9%\n\nThere is an e-participation portal, 50.3%\n\n48%\n\n24%\n\n9%\n\n19%\n\n30%\n\n30%\n\n31%\n\n57%\n\n46%\n\n29%\n\n34%\n\n51%\n\n49%\n\n40%\n\n81%\n\n72%\n\n38%\n\n45%\n\n68%\n\n72%\n\n57%\n\n88%\n\n84%\n\n49%\n\n74%\n\n79%\n\n88%\n\n86%\n\n57%\n\n29%\n\n36%\n\n14%\n\n43%\n\n36%\n\n14%\n\nPortal provides a way for people to report corruption among public servants or institutions\n\nPortal provides a way for people to report labour law violations\n\nGovernment publishes information on how people's voices, including those of vulnerable groups, are included in policy decision-making\n\nPortal offers e-petition or similar mechanisms\n\nPortal publishes calendar/announcements about e-participation activities\n\nPortal integrates e-tools for public consultation/deliberation\n\nThere is an e-participation portal\n\nAfrica Americas Asia Europe Oceania\n\nEDGI 2024 levels\n\nVery high EGDI High EGDI Middle EGDI Low EGDI\n\n### Very high EG High EGDI Middle EGDI Low EGDI\n\n193 UN Member States\n\nAmericas\n\nAfrica\n\nAsia\n\nEurope\n\nOceania\n\n### Very high EGDI High EGDI Middle EGDI Low EGDI\n\nPortal provides a way for people to report labour law violations, 53.4%\n\nPortal provides a way for people to report corruption among public servants or institutions, 67.4%\n\nGovernment publishes information on how people's voices, including those of vulnerable groups, are included in policy decision-making, 30.6%\n\nPortal offers e-petition or similar mechanisms, 39.9% Portal integrates e-tools for public consultation/deliberation, 57.0%\n\nPortal publishes calendar/announcements about e-participation activities, 54.9%\n\nThere is an e-participation portal, 50.3%\n\n48%\n\n24%\n\n9%\n\n19%\n\n30%\n\n30%\n\n31%\n\n57%\n\n46%\n\n29%\n\n34%\n\n51%\n\n49%\n\n40%\n\n81%\n\n72%\n\n38%\n\n45%\n\n68%\n\n72%\n\n57%\n\n88%\n\n84%\n\n49%\n\n74%\n\n79%\n\n88%\n\n86%\n\n57%\n\n29%\n\n36%\n\n14%\n\n43%\n\n36%\n\n14%\n\nPortal provides a way for people to report corruption among public servants or institutions\n\nPortal provides a way for people to report labour law violations\n\nGovernment publishes information on how people's voices, including those of vulnerable groups, are included in policy decision-making\n\nPortal offers e-petition or similar mechanisms\n\nPortal publishes calendar/announcements about e-participation activities\n\nPortal integrates e-tools for public consultation/deliberation\n\nThere is an e-participation portal\n\nAfrica Americas Asia Europe Oceania","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"61\n\n**Chapter 2**\n\nChapter 2 • Global trends in e-Government\n\nFigure 2.20 Percentage of countries offering various e-participation mechanisms and tools, 2024\n\n_Source_ : 2024 United Nations E-Government Survey.\n\nAround 50 per cent of the Member States have a dedicated e-participation portal, 55 per cent publish calendar announcements about upcoming consultations and other participatory activities, and about 40 per cent of use e-petitions or similar mechanisms to engage the population in policy deliberations. The region with the highest proportion of countries providing evidence of having conducted at least one e-consultation in the 12 months preceding the administration of the Survey is Europe (91 per cent), followed by Asia (70 per cent), the Americas (60 per cent), Africa (24 per cent) and Oceania (14 per cent).\n\nInformation on the results of such deliberations or on how people’s voices are included in policy- related decision-making is published by an average of only 31 per cent of the countries surveyed, though regional disparities are quite pronounced; nearly half of the European countries, slightly over one third of countries in Asia and Oceania, just under a third of countries in the Americas, and only a tenth of the countries in Africa publish the results of public deliberations on their portals.\n\nIn terms of the proportion of countries in each region proactively offering e-participation channels, tools or mechanisms, Europe is the leader (78 per cent), followed by Asia (62 per cent), the Americas (44 per cent), Oceania (33 per cent), and Africa (27 per cent).\n\n### **2\\.5.6 OSI services provision subindex: progress in online services delivery**\n\nThe services provision subindex of the OSI assesses the availability of various online transactional services, how government services are accessed (through one main portal or multiple channels), the existence and functionality of e-procurement platforms and digital invoicing, the integration of GIS or geospatial data and technologies in online services provision, and the availability of sector- specific services and services for people in vulnerable situations. The data analysis and key findings are presented below.\n\nFor the 2024 Survey, 25 online services have been assessed, up from 22 for the last Survey cycle. The number of countries offering at least one of these online services has not changed, remaining at 189 (98 per cent) of the 193 Member States. Figure 2.21 illustrates the overall expansion in the range of services provided, which is reflected in the increased number of countries providing each type of service.\n\nEDGI 2024 levels\n\nVery high EGDI High EGDI Middle EGDI Low EGDI\n\n### Very high EGDI High EGDI Middle EGDI Low EGDI\n\n193 UN Member States\n\nAmericas\n\nAfrica\n\nAsia\n\nEurope\n\nOceania\n\n### Very high EGDI High EGDI Middle EGDI Low EGDI\n\nPortal provides a way for people to report labour law violations, 53.4%\n\nPortal provides a way for people to report corruption among public servants or institutions, 67.4%\n\nGovernment publishes information on how people's voices, including those of vulnerable groups, are included in policy decision-making, 30.6%\n\nPortal offers e-petition or similar mechanisms, 39.9% Portal integrates e-tools for public consultation/deliberation, 57.0%\n\nPortal publishes calendar/announcements about e-participation activities, 54.9%\n\nThere is an e-participation portal, 50.3%\n\n48%\n\n24%\n\n9%\n\n19%\n\n30%\n\n30%\n\n31%\n\n57%\n\n46%\n\n29%\n\n34%\n\n51%\n\n49%\n\n40%\n\n81%\n\n72%\n\n38%\n\n45%\n\n68%\n\n72%\n\n57%\n\n88%\n\n84%\n\n49%\n\n74%\n\n79%\n\n88%\n\n86%\n\n57%\n\n29%\n\n36%\n\n14%\n\n43%\n\n36%\n\n14%\n\nPortal provides a way for people to report corruption among public servants or institutions\n\nPortal provides a way for people to report labour law violations\n\nGovernment publishes information on how people's voices, including those of vulnerable groups, are included in policy decision-making\n\nPortal offers e-petition or similar mechanisms\n\nPortal publishes calendar/announcements about e-participation activities\n\nPortal integrates e-tools for public consultation/deliberation\n\nThere is an e-participation portal\n\nAfrica Americas Asia Europe Oceania\n\nEDGI 2024 levels\n\nVery high EGDI High EGDI Middle EGDI Low EGDI\n\n### Very high EG High EGDI Middle EGDI Low EGDI\n\n193 UN Member States\n\nAmericas\n\nAfrica\n\nAsia\n\nEurope\n\nOceania\n\n### Very high EGDI High EGDI Middle EGDI Low EGDI\n\nPortal provides a way for people to report labour law violations, 53.4%\n\nPortal provides a way for people to report corruption among public servants or institutions, 67.4%\n\nGovernment publishes information on how people's voices, including those of vulnerable groups, are included in policy decision-making, 30.6%\n\nPortal offers e-petition or similar mechanisms, 39.9% Portal integrates e-tools for public consultation/deliberation, 57.0%\n\nPortal publishes calendar/announcements about e-participation activities, 54.9%\n\nThere is an e-participation portal, 50.3%\n\n48%\n\n24%\n\n9%\n\n19%\n\n30%\n\n30%\n\n31%\n\n57%\n\n46%\n\n29%\n\n34%\n\n51%\n\n49%\n\n40%\n\n81%\n\n72%\n\n38%\n\n45%\n\n68%\n\n72%\n\n57%\n\n88%\n\n84%\n\n49%\n\n74%\n\n79%\n\n88%\n\n86%\n\n57%\n\n29%\n\n36%\n\n14%\n\n43%\n\n36%\n\n14%\n\nPortal provides a way for people to report corruption among public servants or institutions\n\nPortal provides a way for people to report labour law violations\n\nGovernment publishes information on how people's voices, including those of vulnerable groups, are included in policy decision-making\n\nPortal offers e-petition or similar mechanisms\n\nPortal publishes calendar/announcements about e-participation activities\n\nPortal integrates e-tools for public consultation/deliberation\n\nThere is an e-participation portal\n\nAfrica Americas Asia Europe Oceania","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/%28Web_version%29_E-Government_Survey_2024_1392024_p86.pdf"]},"response":{"errors":[],"extract_id":"extract_48f2978f5fe3197e575cde0bd45253d4","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["61\n\n**Chapter 2**\n\nChapter 2 • Global trends in e-Government\n\nFigure 2.20 Percentage of countries offering various e-participation mechanisms and tools, 2024\n\n_Source_ : 2024 United Nations E-Government Survey.\n\nAround 50 per cent of the Member States have a dedicated e-participation portal, 55 per cent publish calendar announcements about upcoming consultations and other participatory activities, and about 40 per cent of use e-petitions or similar mechanisms to engage the population in policy deliberations. The region with the highest proportion of countries providing evidence of having conducted at least one e-consultation in the 12 months preceding the administration of the Survey is Europe (91 per cent), followed by Asia (70 per cent), the Americas (60 per cent), Africa (24 per cent) and Oceania (14 per cent).\n\nInformation on the results of such deliberations or on how people’s voices are included in policy- related decision-making is published by an average of only 31 per cent of the countries surveyed, though regional disparities are quite pronounced; nearly half of the European countries, slightly over one third of countries in Asia and Oceania, just under a third of countries in the Americas, and only a tenth of the countries in Africa publish the results of public deliberations on their portals.\n\nIn terms of the proportion of countries in each region proactively offering e-participation channels, tools or mechanisms, Europe is the leader (78 per cent), followed by Asia (62 per cent), the Americas (44 per cent), Oceania (33 per cent), and Africa (27 per cent).\n\n### **2\\.5.6 OSI services provision subindex: progress in online services delivery**\n\nThe services provision subindex of the OSI assesses the availability of various online transactional services, how government services are accessed (through one main portal or multiple channels), the existence and functionality of e-procurement platforms and digital invoicing, the integration of GIS or geospatial data and technologies in online services provision, and the availability of sector- specific services and services for people in vulnerable situations. The data analysis and key findings are presented below.\n\nFor the 2024 Survey, 25 online services have been assessed, up from 22 for the last Survey cycle. The number of countries offering at least one of these online services has not changed, remaining at 189 (98 per cent) of the 193 Member States. Figure 2.21 illustrates the overall expansion in the range of services provided, which is reflected in the increased number of countries providing each type of service.\n\nEDGI 2024 levels\n\nVery high EGDI High EGDI Middle EGDI Low EGDI\n\n### Very high EGDI High EGDI Middle EGDI Low EGDI\n\n193 UN Member States\n\nAmericas\n\nAfrica\n\nAsia\n\nEurope\n\nOceania\n\n### Very high EGDI High EGDI Middle EGDI Low EGDI\n\nPortal provides a way for people to report labour law violations, 53.4%\n\nPortal provides a way for people to report corruption among public servants or institutions, 67.4%\n\nGovernment publishes information on how people's voices, including those of vulnerable groups, are included in policy decision-making, 30.6%\n\nPortal offers e-petition or similar mechanisms, 39.9% Portal integrates e-tools for public consultation/deliberation, 57.0%\n\nPortal publishes calendar/announcements about e-participation activities, 54.9%\n\nThere is an e-participation portal, 50.3%\n\n48%\n\n24%\n\n9%\n\n19%\n\n30%\n\n30%\n\n31%\n\n57%\n\n46%\n\n29%\n\n34%\n\n51%\n\n49%\n\n40%\n\n81%\n\n72%\n\n38%\n\n45%\n\n68%\n\n72%\n\n57%\n\n88%\n\n84%\n\n49%\n\n74%\n\n79%\n\n88%\n\n86%\n\n57%\n\n29%\n\n36%\n\n14%\n\n43%\n\n36%\n\n14%\n\nPortal provides a way for people to report corruption among public servants or institutions\n\nPortal provides a way for people to report labour law violations\n\nGovernment publishes information on how people's voices, including those of vulnerable groups, are included in policy decision-making\n\nPortal offers e-petition or similar mechanisms\n\nPortal publishes calendar/announcements about e-participation activities\n\nPortal integrates e-tools for public consultation/deliberation\n\nThere is an e-participation portal\n\nAfrica Americas Asia Europe Oceania\n\nEDGI 2024 levels\n\nVery high EGDI High EGDI Middle EGDI Low EGDI\n\n### Very high EG High EGDI Middle EGDI Low EGDI\n\n193 UN Member States\n\nAmericas\n\nAfrica\n\nAsia\n\nEurope\n\nOceania\n\n### Very high EGDI High EGDI Middle EGDI Low EGDI\n\nPortal provides a way for people to report labour law violations, 53.4%\n\nPortal provides a way for people to report corruption among public servants or institutions, 67.4%\n\nGovernment publishes information on how people's voices, including those of vulnerable groups, are included in policy decision-making, 30.6%\n\nPortal offers e-petition or similar mechanisms, 39.9% Portal integrates e-tools for public consultation/deliberation, 57.0%\n\nPortal publishes calendar/announcements about e-participation activities, 54.9%\n\nThere is an e-participation portal, 50.3%\n\n48%\n\n24%\n\n9%\n\n19%\n\n30%\n\n30%\n\n31%\n\n57%\n\n46%\n\n29%\n\n34%\n\n51%\n\n49%\n\n40%\n\n81%\n\n72%\n\n38%\n\n45%\n\n68%\n\n72%\n\n57%\n\n88%\n\n84%\n\n49%\n\n74%\n\n79%\n\n88%\n\n86%\n\n57%\n\n29%\n\n36%\n\n14%\n\n43%\n\n36%\n\n14%\n\nPortal provides a way for people to report corruption among public servants or institutions\n\nPortal provides a way for people to report labour law violations\n\nGovernment publishes information on how people's voices, including those of vulnerable groups, are included in policy decision-making\n\nPortal offers e-petition or similar mechanisms\n\nPortal publishes calendar/announcements about e-participation activities\n\nPortal integrates e-tools for public consultation/deliberation\n\nThere is an e-participation portal\n\nAfrica Americas Asia Europe Oceania"],"full_content":"61\n\n**Chapter 2**\n\nChapter 2 • Global trends in e-Government\n\nFigure 2.20 Percentage of countries offering various e-participation mechanisms and tools, 2024\n\n_Source_ : 2024 United Nations E-Government Survey.\n\nAround 50 per cent of the Member States have a dedicated e-participation portal, 55 per cent publish calendar announcements about upcoming consultations and other participatory activities, and about 40 per cent of use e-petitions or similar mechanisms to engage the population in policy deliberations. The region with the highest proportion of countries providing evidence of having conducted at least one e-consultation in the 12 months preceding the administration of the Survey is Europe (91 per cent), followed by Asia (70 per cent), the Americas (60 per cent), Africa (24 per cent) and Oceania (14 per cent).\n\nInformation on the results of such deliberations or on how people’s voices are included in policy- related decision-making is published by an average of only 31 per cent of the countries surveyed, though regional disparities are quite pronounced; nearly half of the European countries, slightly over one third of countries in Asia and Oceania, just under a third of countries in the Americas, and only a tenth of the countries in Africa publish the results of public deliberations on their portals.\n\nIn terms of the proportion of countries in each region proactively offering e-participation channels, tools or mechanisms, Europe is the leader (78 per cent), followed by Asia (62 per cent), the Americas (44 per cent), Oceania (33 per cent), and Africa (27 per cent).\n\n### **2\\.5.6 OSI services provision subindex: progress in online services delivery**\n\nThe services provision subindex of the OSI assesses the availability of various online transactional services, how government services are accessed (through one main portal or multiple channels), the existence and functionality of e-procurement platforms and digital invoicing, the integration of GIS or geospatial data and technologies in online services provision, and the availability of sector- specific services and services for people in vulnerable situations. The data analysis and key findings are presented below.\n\nFor the 2024 Survey, 25 online services have been assessed, up from 22 for the last Survey cycle. The number of countries offering at least one of these online services has not changed, remaining at 189 (98 per cent) of the 193 Member States. Figure 2.21 illustrates the overall expansion in the range of services provided, which is reflected in the increased number of countries providing each type of service.\n\nEDGI 2024 levels\n\nVery high EGDI High EGDI Middle EGDI Low EGDI\n\n### Very high EGDI High EGDI Middle EGDI Low EGDI\n\n193 UN Member States\n\nAmericas\n\nAfrica\n\nAsia\n\nEurope\n\nOceania\n\n### Very high EGDI High EGDI Middle EGDI Low EGDI\n\nPortal provides a way for people to report labour law violations, 53.4%\n\nPortal provides a way for people to report corruption among public servants or institutions, 67.4%\n\nGovernment publishes information on how people's voices, including those of vulnerable groups, are included in policy decision-making, 30.6%\n\nPortal offers e-petition or similar mechanisms, 39.9% Portal integrates e-tools for public consultation/deliberation, 57.0%\n\nPortal publishes calendar/announcements about e-participation activities, 54.9%\n\nThere is an e-participation portal, 50.3%\n\n48%\n\n24%\n\n9%\n\n19%\n\n30%\n\n30%\n\n31%\n\n57%\n\n46%\n\n29%\n\n34%\n\n51%\n\n49%\n\n40%\n\n81%\n\n72%\n\n38%\n\n45%\n\n68%\n\n72%\n\n57%\n\n88%\n\n84%\n\n49%\n\n74%\n\n79%\n\n88%\n\n86%\n\n57%\n\n29%\n\n36%\n\n14%\n\n43%\n\n36%\n\n14%\n\nPortal provides a way for people to report corruption among public servants or institutions\n\nPortal provides a way for people to report labour law violations\n\nGovernment publishes information on how people's voices, including those of vulnerable groups, are included in policy decision-making\n\nPortal offers e-petition or similar mechanisms\n\nPortal publishes calendar/announcements about e-participation activities\n\nPortal integrates e-tools for public consultation/deliberation\n\nThere is an e-participation portal\n\nAfrica Americas Asia Europe Oceania\n\nEDGI 2024 levels\n\nVery high EGDI High EGDI Middle EGDI Low EGDI\n\n### Very high EG High EGDI Middle EGDI Low EGDI\n\n193 UN Member States\n\nAmericas\n\nAfrica\n\nAsia\n\nEurope\n\nOceania\n\n### Very high EGDI High EGDI Middle EGDI Low EGDI\n\nPortal provides a way for people to report labour law violations, 53.4%\n\nPortal provides a way for people to report corruption among public servants or institutions, 67.4%\n\nGovernment publishes information on how people's voices, including those of vulnerable groups, are included in policy decision-making, 30.6%\n\nPortal offers e-petition or similar mechanisms, 39.9% Portal integrates e-tools for public consultation/deliberation, 57.0%\n\nPortal publishes calendar/announcements about e-participation activities, 54.9%\n\nThere is an e-participation portal, 50.3%\n\n48%\n\n24%\n\n9%\n\n19%\n\n30%\n\n30%\n\n31%\n\n57%\n\n46%\n\n29%\n\n34%\n\n51%\n\n49%\n\n40%\n\n81%\n\n72%\n\n38%\n\n45%\n\n68%\n\n72%\n\n57%\n\n88%\n\n84%\n\n49%\n\n74%\n\n79%\n\n88%\n\n86%\n\n57%\n\n29%\n\n36%\n\n14%\n\n43%\n\n36%\n\n14%\n\nPortal provides a way for people to report corruption among public servants or institutions\n\nPortal provides a way for people to report labour law violations\n\nGovernment publishes information on how people's voices, including those of vulnerable groups, are included in policy decision-making\n\nPortal offers e-petition or similar mechanisms\n\nPortal publishes calendar/announcements about e-participation activities\n\nPortal integrates e-tools for public consultation/deliberation\n\nThere is an e-participation portal\n\nAfrica Americas Asia Europe Oceania","publish_date":null,"title":"Very high EGDI High EGDI Middle EGDI Low EGDI Very high EGDI High EGDI Middle EGDI Low EGDI","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/%28Web_version%29_E-Government_Survey_2024_1392024_p86.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/%28Web_version%29_E-Government_Survey_2024_1392024_p86.pdf"},"request":{"example_id":"chart/(Web_version)_E-Government_Survey_2024_1392024_p86","product_type":"parse","source_file_path":"(Web_version)_E-Government_Survey_2024_1392024_p86.pdf"},"started_at":"2026-11-04T19:46:03.893320"} +{"completed_at":"2026-11-04T19:46:07.614085","latency_in_ms":1453,"output":{"example_id":"chart/(Web_version)_E-Government_Survey_2024_1392024_p90","grounded_pages":[],"layout_pages":[],"markdown":"65\n\n**Chapter 2**\n\nChapter 2 • Global trends in e-Government\n\nPublic procurement services\n\nA total of 135 countries (70 per cent) have dedicated e-procurement portals, an increase of 4 per cent since 2022. The number of countries that issue digital invoices through their e-procurement portals has increased by almost 7 per cent and now stands at 101 (52 per cent). Nearly all countries in Europe (93 per cent) have e-procurement portals, and most (86 per cent) offer digital invoicing. In Asia and the Americas, about 80 per cent of countries have portals, but fewer (60 and 54 per cent, respectively) issue digital invoices. The gap is wider in Oceania and Africa, where the corresponding proportions are 57 versus 21 per cent and 39 versus 26 per cent, respectively. It is worth mentioning that once e-procurement portals are set up, it may take some time for countries to integrate digital invoicing. In the Americas, for instance, the number of countries with e-procurement portals has risen by 5.7 per cent since 2022, whereas the number of countries offering digital invoicing has increased by 14.3 per cent (see figure 2.24).\n\n### **2\\.5.7 Targeted services for people in vulnerable situations**\n\nThe number of countries providing information and services that target specific vulnerable populations decreased by an average of 5 per cent between 2022 and 2024. As shown in figure 2.25, the sharpest decline (13.5 per cent) is recorded for immigrants; this is an alarming shift given that in 2022 the services aimed at supporting immigrants were provided by the highest number of countries (163, compared with 141 countries in 2024). The provision of services for women (148 countries) and people living below poverty line (132 countries) has also declined (by 8.6 and 8.3 per cent, respectively). The same is true for youth services (a drop from 155 to 152 countries, representing a 1.9 per cent decline), and for services supporting persons with disabilities (a drop from 157 to 155\n\nFigure 2.24 Number and percentage of countries with e-procurement platforms and digital invoicing capabilities, global and regional data, 2022 and 2024\n\n_Sources_ : 2022 and 2024 United Nations E-Government Surveys.\n\n193 UN Member States\n\nAmericas\n\nAfrica\n\nAsia\n\nEurope\n\nOceania\n\n### Very high EGDI High EGDI Middle EGDI Low EGDI\n\n127 countries, 66%\n\n88 countries, 46%\n\n19 countries, 35%\n\n10 countries, 19%\n\n26 countries, 74%\n\n14 countries, 40%\n\n36 countries, 77%\n\n26 countries, 55%\n\n39 countries, 91%\n\n35 countries, 81%\n\n7 countries, 50%\n\n3 countries, 21%\n\n4\\.1%\n\n6\\.7%\n\n3\\.7%\n\n7\\.4%\n\n5\\.7%\n\n14\\.3%\n\n4\\.3%\n\n4\\.3%\n\n2\\.3%\n\n4\\.7%\n\n7\\.1%\n\n135 countries, 70% 101 countries, 52%\n\n21 countries, 39% 14 countries, 26%\n\n28 countries, 80% 19 countries, 54%\n\n38 countries, 81% 28 countries, 60%\n\n40 countries, 93% 37 countries, 86%\n\n8 countries, 57% 3 countries, 21%\n\n0%\n\n20%\n\n40%\n\n60%\n\n80%\n\n100%\n\nE-procurement platform Digital invoicing E-procurement platform Digital invoicing E-procurement platform Digital invoicing E-procurement platform Digital Invoicing E-procurement platform Digital invoicing E-procurement platform Digital invoicing\n\n193 UN Member States Africa Americas Asia Europe Oceania\n\nNumber and percentage of countries globally and in each region, 2022\n\n2024\n\n2022\n\nPercentage increase since 2022 Number and percentage of countries globally and in each region, 2022","pages":[{"line_numbers":[],"markdown":"65\n\n**Chapter 2**\n\nChapter 2 • Global trends in e-Government\n\nPublic procurement services\n\nA total of 135 countries (70 per cent) have dedicated e-procurement portals, an increase of 4 per cent since 2022. The number of countries that issue digital invoices through their e-procurement portals has increased by almost 7 per cent and now stands at 101 (52 per cent). Nearly all countries in Europe (93 per cent) have e-procurement portals, and most (86 per cent) offer digital invoicing. In Asia and the Americas, about 80 per cent of countries have portals, but fewer (60 and 54 per cent, respectively) issue digital invoices. The gap is wider in Oceania and Africa, where the corresponding proportions are 57 versus 21 per cent and 39 versus 26 per cent, respectively. It is worth mentioning that once e-procurement portals are set up, it may take some time for countries to integrate digital invoicing. In the Americas, for instance, the number of countries with e-procurement portals has risen by 5.7 per cent since 2022, whereas the number of countries offering digital invoicing has increased by 14.3 per cent (see figure 2.24).\n\n### **2\\.5.7 Targeted services for people in vulnerable situations**\n\nThe number of countries providing information and services that target specific vulnerable populations decreased by an average of 5 per cent between 2022 and 2024. As shown in figure 2.25, the sharpest decline (13.5 per cent) is recorded for immigrants; this is an alarming shift given that in 2022 the services aimed at supporting immigrants were provided by the highest number of countries (163, compared with 141 countries in 2024). The provision of services for women (148 countries) and people living below poverty line (132 countries) has also declined (by 8.6 and 8.3 per cent, respectively). The same is true for youth services (a drop from 155 to 152 countries, representing a 1.9 per cent decline), and for services supporting persons with disabilities (a drop from 157 to 155\n\nFigure 2.24 Number and percentage of countries with e-procurement platforms and digital invoicing capabilities, global and regional data, 2022 and 2024\n\n_Sources_ : 2022 and 2024 United Nations E-Government Surveys.\n\n193 UN Member States\n\nAmericas\n\nAfrica\n\nAsia\n\nEurope\n\nOceania\n\n### Very high EGDI High EGDI Middle EGDI Low EGDI\n\n127 countries, 66%\n\n88 countries, 46%\n\n19 countries, 35%\n\n10 countries, 19%\n\n26 countries, 74%\n\n14 countries, 40%\n\n36 countries, 77%\n\n26 countries, 55%\n\n39 countries, 91%\n\n35 countries, 81%\n\n7 countries, 50%\n\n3 countries, 21%\n\n4\\.1%\n\n6\\.7%\n\n3\\.7%\n\n7\\.4%\n\n5\\.7%\n\n14\\.3%\n\n4\\.3%\n\n4\\.3%\n\n2\\.3%\n\n4\\.7%\n\n7\\.1%\n\n135 countries, 70% 101 countries, 52%\n\n21 countries, 39% 14 countries, 26%\n\n28 countries, 80% 19 countries, 54%\n\n38 countries, 81% 28 countries, 60%\n\n40 countries, 93% 37 countries, 86%\n\n8 countries, 57% 3 countries, 21%\n\n0%\n\n20%\n\n40%\n\n60%\n\n80%\n\n100%\n\nE-procurement platform Digital invoicing E-procurement platform Digital invoicing E-procurement platform Digital invoicing E-procurement platform Digital Invoicing E-procurement platform Digital invoicing E-procurement platform Digital invoicing\n\n193 UN Member States Africa Americas Asia Europe Oceania\n\nNumber and percentage of countries globally and in each region, 2022\n\n2024\n\n2022\n\nPercentage increase since 2022 Number and percentage of countries globally and in each region, 2022","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"65\n\n**Chapter 2**\n\nChapter 2 • Global trends in e-Government\n\nPublic procurement services\n\nA total of 135 countries (70 per cent) have dedicated e-procurement portals, an increase of 4 per cent since 2022. The number of countries that issue digital invoices through their e-procurement portals has increased by almost 7 per cent and now stands at 101 (52 per cent). Nearly all countries in Europe (93 per cent) have e-procurement portals, and most (86 per cent) offer digital invoicing. In Asia and the Americas, about 80 per cent of countries have portals, but fewer (60 and 54 per cent, respectively) issue digital invoices. The gap is wider in Oceania and Africa, where the corresponding proportions are 57 versus 21 per cent and 39 versus 26 per cent, respectively. It is worth mentioning that once e-procurement portals are set up, it may take some time for countries to integrate digital invoicing. In the Americas, for instance, the number of countries with e-procurement portals has risen by 5.7 per cent since 2022, whereas the number of countries offering digital invoicing has increased by 14.3 per cent (see figure 2.24).\n\n### **2\\.5.7 Targeted services for people in vulnerable situations**\n\nThe number of countries providing information and services that target specific vulnerable populations decreased by an average of 5 per cent between 2022 and 2024. As shown in figure 2.25, the sharpest decline (13.5 per cent) is recorded for immigrants; this is an alarming shift given that in 2022 the services aimed at supporting immigrants were provided by the highest number of countries (163, compared with 141 countries in 2024). The provision of services for women (148 countries) and people living below poverty line (132 countries) has also declined (by 8.6 and 8.3 per cent, respectively). The same is true for youth services (a drop from 155 to 152 countries, representing a 1.9 per cent decline), and for services supporting persons with disabilities (a drop from 157 to 155\n\nFigure 2.24 Number and percentage of countries with e-procurement platforms and digital invoicing capabilities, global and regional data, 2022 and 2024\n\n_Sources_ : 2022 and 2024 United Nations E-Government Surveys.\n\n193 UN Member States\n\nAmericas\n\nAfrica\n\nAsia\n\nEurope\n\nOceania\n\n### Very high EGDI High EGDI Middle EGDI Low EGDI\n\n127 countries, 66%\n\n88 countries, 46%\n\n19 countries, 35%\n\n10 countries, 19%\n\n26 countries, 74%\n\n14 countries, 40%\n\n36 countries, 77%\n\n26 countries, 55%\n\n39 countries, 91%\n\n35 countries, 81%\n\n7 countries, 50%\n\n3 countries, 21%\n\n4\\.1%\n\n6\\.7%\n\n3\\.7%\n\n7\\.4%\n\n5\\.7%\n\n14\\.3%\n\n4\\.3%\n\n4\\.3%\n\n2\\.3%\n\n4\\.7%\n\n7\\.1%\n\n135 countries, 70% 101 countries, 52%\n\n21 countries, 39% 14 countries, 26%\n\n28 countries, 80% 19 countries, 54%\n\n38 countries, 81% 28 countries, 60%\n\n40 countries, 93% 37 countries, 86%\n\n8 countries, 57% 3 countries, 21%\n\n0%\n\n20%\n\n40%\n\n60%\n\n80%\n\n100%\n\nE-procurement platform Digital invoicing E-procurement platform Digital invoicing E-procurement platform Digital invoicing E-procurement platform Digital Invoicing E-procurement platform Digital invoicing E-procurement platform Digital invoicing\n\n193 UN Member States Africa Americas Asia Europe Oceania\n\nNumber and percentage of countries globally and in each region, 2022\n\n2024\n\n2022\n\nPercentage increase since 2022 Number and percentage of countries globally and in each region, 2022","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/%28Web_version%29_E-Government_Survey_2024_1392024_p90.pdf"]},"response":{"errors":[],"extract_id":"extract_874dc1d85bd40fd72a8f9904d1b0eb31","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["65\n\n**Chapter 2**\n\nChapter 2 • Global trends in e-Government\n\nPublic procurement services\n\nA total of 135 countries (70 per cent) have dedicated e-procurement portals, an increase of 4 per cent since 2022. The number of countries that issue digital invoices through their e-procurement portals has increased by almost 7 per cent and now stands at 101 (52 per cent). Nearly all countries in Europe (93 per cent) have e-procurement portals, and most (86 per cent) offer digital invoicing. In Asia and the Americas, about 80 per cent of countries have portals, but fewer (60 and 54 per cent, respectively) issue digital invoices. The gap is wider in Oceania and Africa, where the corresponding proportions are 57 versus 21 per cent and 39 versus 26 per cent, respectively. It is worth mentioning that once e-procurement portals are set up, it may take some time for countries to integrate digital invoicing. In the Americas, for instance, the number of countries with e-procurement portals has risen by 5.7 per cent since 2022, whereas the number of countries offering digital invoicing has increased by 14.3 per cent (see figure 2.24).\n\n### **2\\.5.7 Targeted services for people in vulnerable situations**\n\nThe number of countries providing information and services that target specific vulnerable populations decreased by an average of 5 per cent between 2022 and 2024. As shown in figure 2.25, the sharpest decline (13.5 per cent) is recorded for immigrants; this is an alarming shift given that in 2022 the services aimed at supporting immigrants were provided by the highest number of countries (163, compared with 141 countries in 2024). The provision of services for women (148 countries) and people living below poverty line (132 countries) has also declined (by 8.6 and 8.3 per cent, respectively). The same is true for youth services (a drop from 155 to 152 countries, representing a 1.9 per cent decline), and for services supporting persons with disabilities (a drop from 157 to 155\n\nFigure 2.24 Number and percentage of countries with e-procurement platforms and digital invoicing capabilities, global and regional data, 2022 and 2024\n\n_Sources_ : 2022 and 2024 United Nations E-Government Surveys.\n\n193 UN Member States\n\nAmericas\n\nAfrica\n\nAsia\n\nEurope\n\nOceania\n\n### Very high EGDI High EGDI Middle EGDI Low EGDI\n\n127 countries, 66%\n\n88 countries, 46%\n\n19 countries, 35%\n\n10 countries, 19%\n\n26 countries, 74%\n\n14 countries, 40%\n\n36 countries, 77%\n\n26 countries, 55%\n\n39 countries, 91%\n\n35 countries, 81%\n\n7 countries, 50%\n\n3 countries, 21%\n\n4\\.1%\n\n6\\.7%\n\n3\\.7%\n\n7\\.4%\n\n5\\.7%\n\n14\\.3%\n\n4\\.3%\n\n4\\.3%\n\n2\\.3%\n\n4\\.7%\n\n7\\.1%\n\n135 countries, 70% 101 countries, 52%\n\n21 countries, 39% 14 countries, 26%\n\n28 countries, 80% 19 countries, 54%\n\n38 countries, 81% 28 countries, 60%\n\n40 countries, 93% 37 countries, 86%\n\n8 countries, 57% 3 countries, 21%\n\n0%\n\n20%\n\n40%\n\n60%\n\n80%\n\n100%\n\nE-procurement platform Digital invoicing E-procurement platform Digital invoicing E-procurement platform Digital invoicing E-procurement platform Digital Invoicing E-procurement platform Digital invoicing E-procurement platform Digital invoicing\n\n193 UN Member States Africa Americas Asia Europe Oceania\n\nNumber and percentage of countries globally and in each region, 2022\n\n2024\n\n2022\n\nPercentage increase since 2022 Number and percentage of countries globally and in each region, 2022"],"full_content":"65\n\n**Chapter 2**\n\nChapter 2 • Global trends in e-Government\n\nPublic procurement services\n\nA total of 135 countries (70 per cent) have dedicated e-procurement portals, an increase of 4 per cent since 2022. The number of countries that issue digital invoices through their e-procurement portals has increased by almost 7 per cent and now stands at 101 (52 per cent). Nearly all countries in Europe (93 per cent) have e-procurement portals, and most (86 per cent) offer digital invoicing. In Asia and the Americas, about 80 per cent of countries have portals, but fewer (60 and 54 per cent, respectively) issue digital invoices. The gap is wider in Oceania and Africa, where the corresponding proportions are 57 versus 21 per cent and 39 versus 26 per cent, respectively. It is worth mentioning that once e-procurement portals are set up, it may take some time for countries to integrate digital invoicing. In the Americas, for instance, the number of countries with e-procurement portals has risen by 5.7 per cent since 2022, whereas the number of countries offering digital invoicing has increased by 14.3 per cent (see figure 2.24).\n\n### **2\\.5.7 Targeted services for people in vulnerable situations**\n\nThe number of countries providing information and services that target specific vulnerable populations decreased by an average of 5 per cent between 2022 and 2024. As shown in figure 2.25, the sharpest decline (13.5 per cent) is recorded for immigrants; this is an alarming shift given that in 2022 the services aimed at supporting immigrants were provided by the highest number of countries (163, compared with 141 countries in 2024). The provision of services for women (148 countries) and people living below poverty line (132 countries) has also declined (by 8.6 and 8.3 per cent, respectively). The same is true for youth services (a drop from 155 to 152 countries, representing a 1.9 per cent decline), and for services supporting persons with disabilities (a drop from 157 to 155\n\nFigure 2.24 Number and percentage of countries with e-procurement platforms and digital invoicing capabilities, global and regional data, 2022 and 2024\n\n_Sources_ : 2022 and 2024 United Nations E-Government Surveys.\n\n193 UN Member States\n\nAmericas\n\nAfrica\n\nAsia\n\nEurope\n\nOceania\n\n### Very high EGDI High EGDI Middle EGDI Low EGDI\n\n127 countries, 66%\n\n88 countries, 46%\n\n19 countries, 35%\n\n10 countries, 19%\n\n26 countries, 74%\n\n14 countries, 40%\n\n36 countries, 77%\n\n26 countries, 55%\n\n39 countries, 91%\n\n35 countries, 81%\n\n7 countries, 50%\n\n3 countries, 21%\n\n4\\.1%\n\n6\\.7%\n\n3\\.7%\n\n7\\.4%\n\n5\\.7%\n\n14\\.3%\n\n4\\.3%\n\n4\\.3%\n\n2\\.3%\n\n4\\.7%\n\n7\\.1%\n\n135 countries, 70% 101 countries, 52%\n\n21 countries, 39% 14 countries, 26%\n\n28 countries, 80% 19 countries, 54%\n\n38 countries, 81% 28 countries, 60%\n\n40 countries, 93% 37 countries, 86%\n\n8 countries, 57% 3 countries, 21%\n\n0%\n\n20%\n\n40%\n\n60%\n\n80%\n\n100%\n\nE-procurement platform Digital invoicing E-procurement platform Digital invoicing E-procurement platform Digital invoicing E-procurement platform Digital Invoicing E-procurement platform Digital invoicing E-procurement platform Digital invoicing\n\n193 UN Member States Africa Americas Asia Europe Oceania\n\nNumber and percentage of countries globally and in each region, 2022\n\n2024\n\n2022\n\nPercentage increase since 2022 Number and percentage of countries globally and in each region, 2022","publish_date":null,"title":"Very high EGDI High EGDI Middle EGDI Low EGDI Very high EGDI High EGDI Middle EGDI Low EGDI","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/%28Web_version%29_E-Government_Survey_2024_1392024_p90.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/%28Web_version%29_E-Government_Survey_2024_1392024_p90.pdf"},"request":{"example_id":"chart/(Web_version)_E-Government_Survey_2024_1392024_p90","product_type":"parse","source_file_path":"(Web_version)_E-Government_Survey_2024_1392024_p90.pdf"},"started_at":"2026-11-04T19:46:06.161001"} +{"completed_at":"2026-11-04T19:46:08.767698","latency_in_ms":1151,"output":{"example_id":"chart/(Web_version)_E-Government_Survey_2024_1392024_p91","grounded_pages":[],"layout_pages":[],"markdown":"2024 UN E-GovErNmENt SUrvEy\n\n66\n\n**Chapter 2**\n\ncountries, representing a 1.3 per cent decline). The only upturn has been a 2.7 per cent increase in the number of countries providing services for older people (148 countries in 2024 versus 144 in 2022). Further studies are required to examine the possible reasons for such deline. On a positive note, the provision of services that can be transacted fully online has increased by an average of 1 to 7 per cent for all groups except immigrants. countries are moving steadily towards improving the convenience and efficiency of services for users.\n\nEurope remains the most homogeneous region in terms of the provision of online services for people in vulnerable situations (94 per cent of countries), and for 45 per cent of those services (the highest proportion among the regions), transactions can be completed online. While there has been a downturn in the provision of online services to vulnerable populations, the majority of the UN Member States (more than 80 per cent) still offer such services; in regional terms, Europe accounts for the highest proportion, followed the Americas (80 per cent), Asia (79 per cent), Oceania (70 per cent), and Africa (56 per cent). The Americas and Asia have comparable shares of countries offering services to people in vulnerable situations; however, fully digitalized services are offered by more countries in Asia (30 per cent) than in the Americas (23 per cent).\n\nFigure 2.25 Percentage of countries offering services for people in vulnerable situations that can be completed partially or fully online, 2022 and 2024\n\n_Sources_ : 2022 and 2024 United Nations E-Government Surveys.\n\n193 UN Member States\n\nAmericas\n\nAfrica\n\nAsia\n\nEurope\n\nOceania\n\n## Very high EGDI High EGDI Middle EGDI Low EGDI\n\n21%\n\n26%\n\n23%\n\n26%\n\n23%\n\n26%\n\n52%\n\n50%\n\n46%\n\n54%\n\n54%\n\n53%\n\n\\-13.5%\n\n2\\.7%\n\n\\-1.3%\n\n\\-8.6%\n\n\\-1.9%\n\n22%\n\n19%\n\n20%\n\n22%\n\n19%\n\n25%\n\n63%\n\n55%\n\n55%\n\n60%\n\n65%\n\n55%\n\nImmigrants\n\nOlder people\n\nPersons living below poverty line\n\nPersons with disabilities\n\nWomen\n\nYouth\n\n193 UN Member States\n\nAvailable Fully digitalized Available Fully digitalized\n\n**2024 2022** Percentage change since 2022\n\n6%\n\n6%\n\n4%\n\n6%\n\n4%\n\n6%\n\n41%\n\n48%\n\n39%\n\n54%\n\n59%\n\n67%\n\n23%\n\n26%\n\n20%\n\n23%\n\n26%\n\n23%\n\n51%\n\n60%\n\n51%\n\n69%\n\n54%\n\n57%\n\n26%\n\n36%\n\n28%\n\n30%\n\n32%\n\n30%\n\n57%\n\n43%\n\n49%\n\n49%\n\n51%\n\n45%\n\n37%\n\n44%\n\n44%\n\n53%\n\n42%\n\n51%\n\n60%\n\n51%\n\n51%\n\n47%\n\n44%\n\n40%\n\n14%\n\n21%\n\n21%\n\n21%\n\n21%\n\n50%\n\n57%\n\n29%\n\n57%\n\n71%\n\n57%\n\nImmigrants\n\nOlder people\n\nPersons living below poverty line\n\nPersons with disabilities\n\nWomen\n\nYouth\n\n**Africa Americas Asia Europe Oceania**\n\nAvailable\n\nFully digitalized Available\n\nFully digitalized Available\n\nFully digitalized Available\n\nFully digitalized Available\n\nFully digitalized\n\n\\-8.3%","pages":[{"line_numbers":[],"markdown":"2024 UN E-GovErNmENt SUrvEy\n\n66\n\n**Chapter 2**\n\ncountries, representing a 1.3 per cent decline). The only upturn has been a 2.7 per cent increase in the number of countries providing services for older people (148 countries in 2024 versus 144 in 2022). Further studies are required to examine the possible reasons for such deline. On a positive note, the provision of services that can be transacted fully online has increased by an average of 1 to 7 per cent for all groups except immigrants. countries are moving steadily towards improving the convenience and efficiency of services for users.\n\nEurope remains the most homogeneous region in terms of the provision of online services for people in vulnerable situations (94 per cent of countries), and for 45 per cent of those services (the highest proportion among the regions), transactions can be completed online. While there has been a downturn in the provision of online services to vulnerable populations, the majority of the UN Member States (more than 80 per cent) still offer such services; in regional terms, Europe accounts for the highest proportion, followed the Americas (80 per cent), Asia (79 per cent), Oceania (70 per cent), and Africa (56 per cent). The Americas and Asia have comparable shares of countries offering services to people in vulnerable situations; however, fully digitalized services are offered by more countries in Asia (30 per cent) than in the Americas (23 per cent).\n\nFigure 2.25 Percentage of countries offering services for people in vulnerable situations that can be completed partially or fully online, 2022 and 2024\n\n_Sources_ : 2022 and 2024 United Nations E-Government Surveys.\n\n193 UN Member States\n\nAmericas\n\nAfrica\n\nAsia\n\nEurope\n\nOceania\n\n## Very high EGDI High EGDI Middle EGDI Low EGDI\n\n21%\n\n26%\n\n23%\n\n26%\n\n23%\n\n26%\n\n52%\n\n50%\n\n46%\n\n54%\n\n54%\n\n53%\n\n\\-13.5%\n\n2\\.7%\n\n\\-1.3%\n\n\\-8.6%\n\n\\-1.9%\n\n22%\n\n19%\n\n20%\n\n22%\n\n19%\n\n25%\n\n63%\n\n55%\n\n55%\n\n60%\n\n65%\n\n55%\n\nImmigrants\n\nOlder people\n\nPersons living below poverty line\n\nPersons with disabilities\n\nWomen\n\nYouth\n\n193 UN Member States\n\nAvailable Fully digitalized Available Fully digitalized\n\n**2024 2022** Percentage change since 2022\n\n6%\n\n6%\n\n4%\n\n6%\n\n4%\n\n6%\n\n41%\n\n48%\n\n39%\n\n54%\n\n59%\n\n67%\n\n23%\n\n26%\n\n20%\n\n23%\n\n26%\n\n23%\n\n51%\n\n60%\n\n51%\n\n69%\n\n54%\n\n57%\n\n26%\n\n36%\n\n28%\n\n30%\n\n32%\n\n30%\n\n57%\n\n43%\n\n49%\n\n49%\n\n51%\n\n45%\n\n37%\n\n44%\n\n44%\n\n53%\n\n42%\n\n51%\n\n60%\n\n51%\n\n51%\n\n47%\n\n44%\n\n40%\n\n14%\n\n21%\n\n21%\n\n21%\n\n21%\n\n50%\n\n57%\n\n29%\n\n57%\n\n71%\n\n57%\n\nImmigrants\n\nOlder people\n\nPersons living below poverty line\n\nPersons with disabilities\n\nWomen\n\nYouth\n\n**Africa Americas Asia Europe Oceania**\n\nAvailable\n\nFully digitalized Available\n\nFully digitalized Available\n\nFully digitalized Available\n\nFully digitalized Available\n\nFully digitalized\n\n\\-8.3%","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"2024 UN E-GovErNmENt SUrvEy\n\n66\n\n**Chapter 2**\n\ncountries, representing a 1.3 per cent decline). The only upturn has been a 2.7 per cent increase in the number of countries providing services for older people (148 countries in 2024 versus 144 in 2022). Further studies are required to examine the possible reasons for such deline. On a positive note, the provision of services that can be transacted fully online has increased by an average of 1 to 7 per cent for all groups except immigrants. countries are moving steadily towards improving the convenience and efficiency of services for users.\n\nEurope remains the most homogeneous region in terms of the provision of online services for people in vulnerable situations (94 per cent of countries), and for 45 per cent of those services (the highest proportion among the regions), transactions can be completed online. While there has been a downturn in the provision of online services to vulnerable populations, the majority of the UN Member States (more than 80 per cent) still offer such services; in regional terms, Europe accounts for the highest proportion, followed the Americas (80 per cent), Asia (79 per cent), Oceania (70 per cent), and Africa (56 per cent). The Americas and Asia have comparable shares of countries offering services to people in vulnerable situations; however, fully digitalized services are offered by more countries in Asia (30 per cent) than in the Americas (23 per cent).\n\nFigure 2.25 Percentage of countries offering services for people in vulnerable situations that can be completed partially or fully online, 2022 and 2024\n\n_Sources_ : 2022 and 2024 United Nations E-Government Surveys.\n\n193 UN Member States\n\nAmericas\n\nAfrica\n\nAsia\n\nEurope\n\nOceania\n\n## Very high EGDI High EGDI Middle EGDI Low EGDI\n\n21%\n\n26%\n\n23%\n\n26%\n\n23%\n\n26%\n\n52%\n\n50%\n\n46%\n\n54%\n\n54%\n\n53%\n\n\\-13.5%\n\n2\\.7%\n\n\\-1.3%\n\n\\-8.6%\n\n\\-1.9%\n\n22%\n\n19%\n\n20%\n\n22%\n\n19%\n\n25%\n\n63%\n\n55%\n\n55%\n\n60%\n\n65%\n\n55%\n\nImmigrants\n\nOlder people\n\nPersons living below poverty line\n\nPersons with disabilities\n\nWomen\n\nYouth\n\n193 UN Member States\n\nAvailable Fully digitalized Available Fully digitalized\n\n**2024 2022** Percentage change since 2022\n\n6%\n\n6%\n\n4%\n\n6%\n\n4%\n\n6%\n\n41%\n\n48%\n\n39%\n\n54%\n\n59%\n\n67%\n\n23%\n\n26%\n\n20%\n\n23%\n\n26%\n\n23%\n\n51%\n\n60%\n\n51%\n\n69%\n\n54%\n\n57%\n\n26%\n\n36%\n\n28%\n\n30%\n\n32%\n\n30%\n\n57%\n\n43%\n\n49%\n\n49%\n\n51%\n\n45%\n\n37%\n\n44%\n\n44%\n\n53%\n\n42%\n\n51%\n\n60%\n\n51%\n\n51%\n\n47%\n\n44%\n\n40%\n\n14%\n\n21%\n\n21%\n\n21%\n\n21%\n\n50%\n\n57%\n\n29%\n\n57%\n\n71%\n\n57%\n\nImmigrants\n\nOlder people\n\nPersons living below poverty line\n\nPersons with disabilities\n\nWomen\n\nYouth\n\n**Africa Americas Asia Europe Oceania**\n\nAvailable\n\nFully digitalized Available\n\nFully digitalized Available\n\nFully digitalized Available\n\nFully digitalized Available\n\nFully digitalized\n\n\\-8.3%","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/%28Web_version%29_E-Government_Survey_2024_1392024_p91.pdf"]},"response":{"errors":[],"extract_id":"extract_6711c85abf66c40973c82fe5b365a937","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["2024 UN E-GovErNmENt SUrvEy\n\n66\n\n**Chapter 2**\n\ncountries, representing a 1.3 per cent decline). The only upturn has been a 2.7 per cent increase in the number of countries providing services for older people (148 countries in 2024 versus 144 in 2022). Further studies are required to examine the possible reasons for such deline. On a positive note, the provision of services that can be transacted fully online has increased by an average of 1 to 7 per cent for all groups except immigrants. countries are moving steadily towards improving the convenience and efficiency of services for users.\n\nEurope remains the most homogeneous region in terms of the provision of online services for people in vulnerable situations (94 per cent of countries), and for 45 per cent of those services (the highest proportion among the regions), transactions can be completed online. While there has been a downturn in the provision of online services to vulnerable populations, the majority of the UN Member States (more than 80 per cent) still offer such services; in regional terms, Europe accounts for the highest proportion, followed the Americas (80 per cent), Asia (79 per cent), Oceania (70 per cent), and Africa (56 per cent). The Americas and Asia have comparable shares of countries offering services to people in vulnerable situations; however, fully digitalized services are offered by more countries in Asia (30 per cent) than in the Americas (23 per cent).\n\nFigure 2.25 Percentage of countries offering services for people in vulnerable situations that can be completed partially or fully online, 2022 and 2024\n\n_Sources_ : 2022 and 2024 United Nations E-Government Surveys.\n\n193 UN Member States\n\nAmericas\n\nAfrica\n\nAsia\n\nEurope\n\nOceania\n\n## Very high EGDI High EGDI Middle EGDI Low EGDI\n\n21%\n\n26%\n\n23%\n\n26%\n\n23%\n\n26%\n\n52%\n\n50%\n\n46%\n\n54%\n\n54%\n\n53%\n\n\\-13.5%\n\n2\\.7%\n\n\\-1.3%\n\n\\-8.6%\n\n\\-1.9%\n\n22%\n\n19%\n\n20%\n\n22%\n\n19%\n\n25%\n\n63%\n\n55%\n\n55%\n\n60%\n\n65%\n\n55%\n\nImmigrants\n\nOlder people\n\nPersons living below poverty line\n\nPersons with disabilities\n\nWomen\n\nYouth\n\n193 UN Member States\n\nAvailable Fully digitalized Available Fully digitalized\n\n**2024 2022** Percentage change since 2022\n\n6%\n\n6%\n\n4%\n\n6%\n\n4%\n\n6%\n\n41%\n\n48%\n\n39%\n\n54%\n\n59%\n\n67%\n\n23%\n\n26%\n\n20%\n\n23%\n\n26%\n\n23%\n\n51%\n\n60%\n\n51%\n\n69%\n\n54%\n\n57%\n\n26%\n\n36%\n\n28%\n\n30%\n\n32%\n\n30%\n\n57%\n\n43%\n\n49%\n\n49%\n\n51%\n\n45%\n\n37%\n\n44%\n\n44%\n\n53%\n\n42%\n\n51%\n\n60%\n\n51%\n\n51%\n\n47%\n\n44%\n\n40%\n\n14%\n\n21%\n\n21%\n\n21%\n\n21%\n\n50%\n\n57%\n\n29%\n\n57%\n\n71%\n\n57%\n\nImmigrants\n\nOlder people\n\nPersons living below poverty line\n\nPersons with disabilities\n\nWomen\n\nYouth\n\n**Africa Americas Asia Europe Oceania**\n\nAvailable\n\nFully digitalized Available\n\nFully digitalized Available\n\nFully digitalized Available\n\nFully digitalized Available\n\nFully digitalized\n\n\\-8.3%"],"full_content":"2024 UN E-GovErNmENt SUrvEy\n\n66\n\n**Chapter 2**\n\ncountries, representing a 1.3 per cent decline). The only upturn has been a 2.7 per cent increase in the number of countries providing services for older people (148 countries in 2024 versus 144 in 2022). Further studies are required to examine the possible reasons for such deline. On a positive note, the provision of services that can be transacted fully online has increased by an average of 1 to 7 per cent for all groups except immigrants. countries are moving steadily towards improving the convenience and efficiency of services for users.\n\nEurope remains the most homogeneous region in terms of the provision of online services for people in vulnerable situations (94 per cent of countries), and for 45 per cent of those services (the highest proportion among the regions), transactions can be completed online. While there has been a downturn in the provision of online services to vulnerable populations, the majority of the UN Member States (more than 80 per cent) still offer such services; in regional terms, Europe accounts for the highest proportion, followed the Americas (80 per cent), Asia (79 per cent), Oceania (70 per cent), and Africa (56 per cent). The Americas and Asia have comparable shares of countries offering services to people in vulnerable situations; however, fully digitalized services are offered by more countries in Asia (30 per cent) than in the Americas (23 per cent).\n\nFigure 2.25 Percentage of countries offering services for people in vulnerable situations that can be completed partially or fully online, 2022 and 2024\n\n_Sources_ : 2022 and 2024 United Nations E-Government Surveys.\n\n193 UN Member States\n\nAmericas\n\nAfrica\n\nAsia\n\nEurope\n\nOceania\n\n## Very high EGDI High EGDI Middle EGDI Low EGDI\n\n21%\n\n26%\n\n23%\n\n26%\n\n23%\n\n26%\n\n52%\n\n50%\n\n46%\n\n54%\n\n54%\n\n53%\n\n\\-13.5%\n\n2\\.7%\n\n\\-1.3%\n\n\\-8.6%\n\n\\-1.9%\n\n22%\n\n19%\n\n20%\n\n22%\n\n19%\n\n25%\n\n63%\n\n55%\n\n55%\n\n60%\n\n65%\n\n55%\n\nImmigrants\n\nOlder people\n\nPersons living below poverty line\n\nPersons with disabilities\n\nWomen\n\nYouth\n\n193 UN Member States\n\nAvailable Fully digitalized Available Fully digitalized\n\n**2024 2022** Percentage change since 2022\n\n6%\n\n6%\n\n4%\n\n6%\n\n4%\n\n6%\n\n41%\n\n48%\n\n39%\n\n54%\n\n59%\n\n67%\n\n23%\n\n26%\n\n20%\n\n23%\n\n26%\n\n23%\n\n51%\n\n60%\n\n51%\n\n69%\n\n54%\n\n57%\n\n26%\n\n36%\n\n28%\n\n30%\n\n32%\n\n30%\n\n57%\n\n43%\n\n49%\n\n49%\n\n51%\n\n45%\n\n37%\n\n44%\n\n44%\n\n53%\n\n42%\n\n51%\n\n60%\n\n51%\n\n51%\n\n47%\n\n44%\n\n40%\n\n14%\n\n21%\n\n21%\n\n21%\n\n21%\n\n50%\n\n57%\n\n29%\n\n57%\n\n71%\n\n57%\n\nImmigrants\n\nOlder people\n\nPersons living below poverty line\n\nPersons with disabilities\n\nWomen\n\nYouth\n\n**Africa Americas Asia Europe Oceania**\n\nAvailable\n\nFully digitalized Available\n\nFully digitalized Available\n\nFully digitalized Available\n\nFully digitalized Available\n\nFully digitalized\n\n\\-8.3%","publish_date":null,"title":"Very high EGDI High EGDI Middle EGDI Low EGDI","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/%28Web_version%29_E-Government_Survey_2024_1392024_p91.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/%28Web_version%29_E-Government_Survey_2024_1392024_p91.pdf"},"request":{"example_id":"chart/(Web_version)_E-Government_Survey_2024_1392024_p91","product_type":"parse","source_file_path":"(Web_version)_E-Government_Survey_2024_1392024_p91.pdf"},"started_at":"2026-11-04T19:46:07.616395"} +{"completed_at":"2026-11-04T19:46:11.669715","latency_in_ms":1320,"output":{"example_id":"chart/(Web_version)_E-Government_Survey_2024_1392024_p92","grounded_pages":[],"layout_pages":[],"markdown":"67\n\n**Chapter 2**\n\nChapter 2 • Global trends in e-Government\n\nFigure 2.26 Percentage of countries providing sector-based information and services online and through mobile channels and SMS alerts, 2024\n\n_Source_ : 2024 United Nations E-Government Survey.\n\nThe proportion of countries providing services for people in vulnerable situations has risen from 45 to 70 per cent in Oceania and from 44 to 56 per cent in Africa. However, the share of countries that offer fully digitalized services is only 21 per cent in Oceania and 5 per cent in Africa.\n\n### **2\\.5.8 Sector-specific online information and services: sharing via mobile technologies**\n\nOnline services specifically relating to health, education, employment, environment, justice, and social protection can be found on the portals of 58 to 74 per cent of the countries surveyed (see figure 2.26), and most of these services are easily accessible through mobile technologies and applications (apps). Services relating to education, health, and employment are the most prevalent and are available in almost three quarters of the countries surveyed. Services relating to justice (63 per cent), social protection (63 per cent), and environment (58 per cent) are found on the portals of fewer countries. Although Governments still use SMS alerts to inform people about sector-specific services or important issues, this practice is not very common; 44 per cent of countries still utilize SMS for health sector alerts, but for the other five sectors the corresponding proportions range from 19 to 28 per cent.\n\n### **2\\.5.9 Technology subindex**\n\nAt the time the 2024 Survey assessments were carried out, all countries except Belize had accessible national portals, and 98 per cent of those portals could be found by search engines. Most government portals (97 per cent) have a “contact us” page and utilize Hypertext Transfer Protocol Secure (HTTPS) to provide a safe experience for users (see figure 2.27). In 91 per cent of the countries assessed, the portals have been developed using responsive web design (a 3 per cent increase since 2022), and in 89 per cent the portals are updated at least once a month (a 6 per cent increase). Most portals have a sitemap (74 per cent) and offer “advanced search” options (56 per cent); it should be noted that the proportion of countries offering the latter has declined (from 58 per cent) since 2022.\n\nEDGI 2024 levels\n\nVery high EGDI High EGDI Middle EGDI Low EGDI\n\nVery high EGDI\n\nHigh EGDI\n\nMiddle EGDI\n\nLow EGDI\n\n193 UN Member States\n\nAmericas\n\nAfrica\n\nAsia\n\nEurope\n\nOceania\n\n## Very high EGDI High EGDI Middle EGDI Low EGDI\n\n74%\n\n72%\n\n58%\n\n72%\n\n63%\n\n63%\n\n77%\n\n70%\n\n61%\n\n72%\n\n65%\n\n63%\n\n14%\n\n16%\n\n13%\n\n2%\n\n20%\n\n11%\n\n28%\n\n24%\n\n22%\n\n44%\n\n19%\n\n25%\n\nEducation\n\nEmployment\n\nEnvironment\n\nHealth\n\nJustice\n\nSocial protection\n\nAvailability of services Accessibility of services via mobile device (browser or app) Availability of SMS alerts","pages":[{"line_numbers":[],"markdown":"67\n\n**Chapter 2**\n\nChapter 2 • Global trends in e-Government\n\nFigure 2.26 Percentage of countries providing sector-based information and services online and through mobile channels and SMS alerts, 2024\n\n_Source_ : 2024 United Nations E-Government Survey.\n\nThe proportion of countries providing services for people in vulnerable situations has risen from 45 to 70 per cent in Oceania and from 44 to 56 per cent in Africa. However, the share of countries that offer fully digitalized services is only 21 per cent in Oceania and 5 per cent in Africa.\n\n### **2\\.5.8 Sector-specific online information and services: sharing via mobile technologies**\n\nOnline services specifically relating to health, education, employment, environment, justice, and social protection can be found on the portals of 58 to 74 per cent of the countries surveyed (see figure 2.26), and most of these services are easily accessible through mobile technologies and applications (apps). Services relating to education, health, and employment are the most prevalent and are available in almost three quarters of the countries surveyed. Services relating to justice (63 per cent), social protection (63 per cent), and environment (58 per cent) are found on the portals of fewer countries. Although Governments still use SMS alerts to inform people about sector-specific services or important issues, this practice is not very common; 44 per cent of countries still utilize SMS for health sector alerts, but for the other five sectors the corresponding proportions range from 19 to 28 per cent.\n\n### **2\\.5.9 Technology subindex**\n\nAt the time the 2024 Survey assessments were carried out, all countries except Belize had accessible national portals, and 98 per cent of those portals could be found by search engines. Most government portals (97 per cent) have a “contact us” page and utilize Hypertext Transfer Protocol Secure (HTTPS) to provide a safe experience for users (see figure 2.27). In 91 per cent of the countries assessed, the portals have been developed using responsive web design (a 3 per cent increase since 2022), and in 89 per cent the portals are updated at least once a month (a 6 per cent increase). Most portals have a sitemap (74 per cent) and offer “advanced search” options (56 per cent); it should be noted that the proportion of countries offering the latter has declined (from 58 per cent) since 2022.\n\nEDGI 2024 levels\n\nVery high EGDI High EGDI Middle EGDI Low EGDI\n\nVery high EGDI\n\nHigh EGDI\n\nMiddle EGDI\n\nLow EGDI\n\n193 UN Member States\n\nAmericas\n\nAfrica\n\nAsia\n\nEurope\n\nOceania\n\n## Very high EGDI High EGDI Middle EGDI Low EGDI\n\n74%\n\n72%\n\n58%\n\n72%\n\n63%\n\n63%\n\n77%\n\n70%\n\n61%\n\n72%\n\n65%\n\n63%\n\n14%\n\n16%\n\n13%\n\n2%\n\n20%\n\n11%\n\n28%\n\n24%\n\n22%\n\n44%\n\n19%\n\n25%\n\nEducation\n\nEmployment\n\nEnvironment\n\nHealth\n\nJustice\n\nSocial protection\n\nAvailability of services Accessibility of services via mobile device (browser or app) Availability of SMS alerts","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"67\n\n**Chapter 2**\n\nChapter 2 • Global trends in e-Government\n\nFigure 2.26 Percentage of countries providing sector-based information and services online and through mobile channels and SMS alerts, 2024\n\n_Source_ : 2024 United Nations E-Government Survey.\n\nThe proportion of countries providing services for people in vulnerable situations has risen from 45 to 70 per cent in Oceania and from 44 to 56 per cent in Africa. However, the share of countries that offer fully digitalized services is only 21 per cent in Oceania and 5 per cent in Africa.\n\n### **2\\.5.8 Sector-specific online information and services: sharing via mobile technologies**\n\nOnline services specifically relating to health, education, employment, environment, justice, and social protection can be found on the portals of 58 to 74 per cent of the countries surveyed (see figure 2.26), and most of these services are easily accessible through mobile technologies and applications (apps). Services relating to education, health, and employment are the most prevalent and are available in almost three quarters of the countries surveyed. Services relating to justice (63 per cent), social protection (63 per cent), and environment (58 per cent) are found on the portals of fewer countries. Although Governments still use SMS alerts to inform people about sector-specific services or important issues, this practice is not very common; 44 per cent of countries still utilize SMS for health sector alerts, but for the other five sectors the corresponding proportions range from 19 to 28 per cent.\n\n### **2\\.5.9 Technology subindex**\n\nAt the time the 2024 Survey assessments were carried out, all countries except Belize had accessible national portals, and 98 per cent of those portals could be found by search engines. Most government portals (97 per cent) have a “contact us” page and utilize Hypertext Transfer Protocol Secure (HTTPS) to provide a safe experience for users (see figure 2.27). In 91 per cent of the countries assessed, the portals have been developed using responsive web design (a 3 per cent increase since 2022), and in 89 per cent the portals are updated at least once a month (a 6 per cent increase). Most portals have a sitemap (74 per cent) and offer “advanced search” options (56 per cent); it should be noted that the proportion of countries offering the latter has declined (from 58 per cent) since 2022.\n\nEDGI 2024 levels\n\nVery high EGDI High EGDI Middle EGDI Low EGDI\n\nVery high EGDI\n\nHigh EGDI\n\nMiddle EGDI\n\nLow EGDI\n\n193 UN Member States\n\nAmericas\n\nAfrica\n\nAsia\n\nEurope\n\nOceania\n\n## Very high EGDI High EGDI Middle EGDI Low EGDI\n\n74%\n\n72%\n\n58%\n\n72%\n\n63%\n\n63%\n\n77%\n\n70%\n\n61%\n\n72%\n\n65%\n\n63%\n\n14%\n\n16%\n\n13%\n\n2%\n\n20%\n\n11%\n\n28%\n\n24%\n\n22%\n\n44%\n\n19%\n\n25%\n\nEducation\n\nEmployment\n\nEnvironment\n\nHealth\n\nJustice\n\nSocial protection\n\nAvailability of services Accessibility of services via mobile device (browser or app) Availability of SMS alerts","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/%28Web_version%29_E-Government_Survey_2024_1392024_p92.pdf"]},"response":{"errors":[],"extract_id":"extract_71f590c246cd46c807ec7a340a7671d5","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["67\n\n**Chapter 2**\n\nChapter 2 • Global trends in e-Government\n\nFigure 2.26 Percentage of countries providing sector-based information and services online and through mobile channels and SMS alerts, 2024\n\n_Source_ : 2024 United Nations E-Government Survey.\n\nThe proportion of countries providing services for people in vulnerable situations has risen from 45 to 70 per cent in Oceania and from 44 to 56 per cent in Africa. However, the share of countries that offer fully digitalized services is only 21 per cent in Oceania and 5 per cent in Africa.\n\n### **2\\.5.8 Sector-specific online information and services: sharing via mobile technologies**\n\nOnline services specifically relating to health, education, employment, environment, justice, and social protection can be found on the portals of 58 to 74 per cent of the countries surveyed (see figure 2.26), and most of these services are easily accessible through mobile technologies and applications (apps). Services relating to education, health, and employment are the most prevalent and are available in almost three quarters of the countries surveyed. Services relating to justice (63 per cent), social protection (63 per cent), and environment (58 per cent) are found on the portals of fewer countries. Although Governments still use SMS alerts to inform people about sector-specific services or important issues, this practice is not very common; 44 per cent of countries still utilize SMS for health sector alerts, but for the other five sectors the corresponding proportions range from 19 to 28 per cent.\n\n### **2\\.5.9 Technology subindex**\n\nAt the time the 2024 Survey assessments were carried out, all countries except Belize had accessible national portals, and 98 per cent of those portals could be found by search engines. Most government portals (97 per cent) have a “contact us” page and utilize Hypertext Transfer Protocol Secure (HTTPS) to provide a safe experience for users (see figure 2.27). In 91 per cent of the countries assessed, the portals have been developed using responsive web design (a 3 per cent increase since 2022), and in 89 per cent the portals are updated at least once a month (a 6 per cent increase). Most portals have a sitemap (74 per cent) and offer “advanced search” options (56 per cent); it should be noted that the proportion of countries offering the latter has declined (from 58 per cent) since 2022.\n\nEDGI 2024 levels\n\nVery high EGDI High EGDI Middle EGDI Low EGDI\n\nVery high EGDI\n\nHigh EGDI\n\nMiddle EGDI\n\nLow EGDI\n\n193 UN Member States\n\nAmericas\n\nAfrica\n\nAsia\n\nEurope\n\nOceania\n\n## Very high EGDI High EGDI Middle EGDI Low EGDI\n\n74%\n\n72%\n\n58%\n\n72%\n\n63%\n\n63%\n\n77%\n\n70%\n\n61%\n\n72%\n\n65%\n\n63%\n\n14%\n\n16%\n\n13%\n\n2%\n\n20%\n\n11%\n\n28%\n\n24%\n\n22%\n\n44%\n\n19%\n\n25%\n\nEducation\n\nEmployment\n\nEnvironment\n\nHealth\n\nJustice\n\nSocial protection\n\nAvailability of services Accessibility of services via mobile device (browser or app) Availability of SMS alerts"],"full_content":"67\n\n**Chapter 2**\n\nChapter 2 • Global trends in e-Government\n\nFigure 2.26 Percentage of countries providing sector-based information and services online and through mobile channels and SMS alerts, 2024\n\n_Source_ : 2024 United Nations E-Government Survey.\n\nThe proportion of countries providing services for people in vulnerable situations has risen from 45 to 70 per cent in Oceania and from 44 to 56 per cent in Africa. However, the share of countries that offer fully digitalized services is only 21 per cent in Oceania and 5 per cent in Africa.\n\n### **2\\.5.8 Sector-specific online information and services: sharing via mobile technologies**\n\nOnline services specifically relating to health, education, employment, environment, justice, and social protection can be found on the portals of 58 to 74 per cent of the countries surveyed (see figure 2.26), and most of these services are easily accessible through mobile technologies and applications (apps). Services relating to education, health, and employment are the most prevalent and are available in almost three quarters of the countries surveyed. Services relating to justice (63 per cent), social protection (63 per cent), and environment (58 per cent) are found on the portals of fewer countries. Although Governments still use SMS alerts to inform people about sector-specific services or important issues, this practice is not very common; 44 per cent of countries still utilize SMS for health sector alerts, but for the other five sectors the corresponding proportions range from 19 to 28 per cent.\n\n### **2\\.5.9 Technology subindex**\n\nAt the time the 2024 Survey assessments were carried out, all countries except Belize had accessible national portals, and 98 per cent of those portals could be found by search engines. Most government portals (97 per cent) have a “contact us” page and utilize Hypertext Transfer Protocol Secure (HTTPS) to provide a safe experience for users (see figure 2.27). In 91 per cent of the countries assessed, the portals have been developed using responsive web design (a 3 per cent increase since 2022), and in 89 per cent the portals are updated at least once a month (a 6 per cent increase). Most portals have a sitemap (74 per cent) and offer “advanced search” options (56 per cent); it should be noted that the proportion of countries offering the latter has declined (from 58 per cent) since 2022.\n\nEDGI 2024 levels\n\nVery high EGDI High EGDI Middle EGDI Low EGDI\n\nVery high EGDI\n\nHigh EGDI\n\nMiddle EGDI\n\nLow EGDI\n\n193 UN Member States\n\nAmericas\n\nAfrica\n\nAsia\n\nEurope\n\nOceania\n\n## Very high EGDI High EGDI Middle EGDI Low EGDI\n\n74%\n\n72%\n\n58%\n\n72%\n\n63%\n\n63%\n\n77%\n\n70%\n\n61%\n\n72%\n\n65%\n\n63%\n\n14%\n\n16%\n\n13%\n\n2%\n\n20%\n\n11%\n\n28%\n\n24%\n\n22%\n\n44%\n\n19%\n\n25%\n\nEducation\n\nEmployment\n\nEnvironment\n\nHealth\n\nJustice\n\nSocial protection\n\nAvailability of services Accessibility of services via mobile device (browser or app) Availability of SMS alerts","publish_date":null,"title":"Very high EGDI High EGDI Middle EGDI Low EGDI","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/%28Web_version%29_E-Government_Survey_2024_1392024_p92.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/%28Web_version%29_E-Government_Survey_2024_1392024_p92.pdf"},"request":{"example_id":"chart/(Web_version)_E-Government_Survey_2024_1392024_p92","product_type":"parse","source_file_path":"(Web_version)_E-Government_Survey_2024_1392024_p92.pdf"},"started_at":"2026-11-04T19:46:10.348821"} +{"completed_at":"2026-11-04T19:46:14.290942","latency_in_ms":2891,"output":{"example_id":"chart/(Web_version)_E-Government_Survey_2024_1392024_p93","grounded_pages":[],"layout_pages":[],"markdown":"2024 UN E-GovErNmENt SUrvEy\n\n68\n\n**Chapter 2**\n\nThe number of countries that allow individuals and businesses to access or modify their data on government portals has increased by an average of 6 per cent since 2022. Globally, more countries allow users to access their data (71 per cent for businesses and 68 per cent for individuals) than to modify their data (64 per cent for businesses and 56 per cent for individuals).\n\nIn 118 countries (a 3 per cent increase from 2022), it is possible for users to save specific service transactions initiated on the portal and access them later. The proportion of countries that allow users to access a list of their previous transactions or rely on AI-enabled chatbot functionality is nearly the same in 2024 as in 2022.\n\nFor the 2024 Survey, a new “affordability” indicator was added to the TII and the use of “the fixed (wired) broadband subscriptions” indicator was discontinued (see the technical appendix for details on the methodology used). While the technology subindex of the OSI does not include TII component data, it is important to highlight the changes in TII indicators as part of the evolving landscape of technological advancement. Table 2.8 summarizes the global and regional findings for indicators relating to mobile service pricing, broadband and cellular subscriptions, and Internet use for 2022 and 2024. Figure 2.28 highlights the changes in the latter three indicators between 2022 and 2024.\n\nInternet use and subscriptions for mobile broadband and cellular services are on the rise. The sharpest increases in Internet use have been in Oceania (49 per cent) and Africa (31 per cent). The active mobile broadband subscription rate has increased by 27 per cent in Africa, 10 per cent in Asia and the Americas, and 8 per cent in Europe, but has declined by almost 2 per cent in Oceania (see figure 2.28). Europe remains the leader in active mobile broadband use, with 105 subscriptions per 100 inhabitants, followed by Asia (89) and the Americas (73).\n\nFigure 2.27 Number of Member States with portals incorporating the assessed technology features, 2022 and 2024\n\n_Sources_ : 2022 and 2024 United Nations E-Government Surveys.\n\n193 UN Member States\n\nAmericas\n\nAfrica\n\nAsia\n\nEurope\n\nOceania\n\n## Very high EGD High EGDI Middle EGDI Low EGDI\n\n26%\n\n56%\n\n58%\n\n50%\n\n64%\n\n59%\n\n65%\n\n58%\n\n71%\n\n83%\n\n88%\n\n97%\n\n98%\n\n98%\n\n27%\n\n58%\n\n61%\n\n56%\n\n68%\n\n64%\n\n71%\n\n56%\n\n74%\n\n89%\n\n91%\n\n97%\n\n97%\n\n98%\n\n0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Portal offers AI-enabled chatbot functionality\n\nUsers can access a list of their previous interactions or transactions in the portal\n\nSpecifc service transactions initiated through the portal can be saved and accessed later\n\nIndividuals can modify their own data online\n\nIndividuals can access their own data online\n\nBusinesses can modify their own data online\n\nBusinesses can access their own data online\n\nThere is an \"advanced search\" option\n\nThere is a sitemap (or index)\n\nPortal has been updated in the last month\n\nPortal is developed using responsive web design\n\nPortal utilizes HTTPS\n\nThere is a \"contact us\" feature\n\nPortal can be found on search engine\n\n2024 2022","pages":[{"line_numbers":[],"markdown":"2024 UN E-GovErNmENt SUrvEy\n\n68\n\n**Chapter 2**\n\nThe number of countries that allow individuals and businesses to access or modify their data on government portals has increased by an average of 6 per cent since 2022. Globally, more countries allow users to access their data (71 per cent for businesses and 68 per cent for individuals) than to modify their data (64 per cent for businesses and 56 per cent for individuals).\n\nIn 118 countries (a 3 per cent increase from 2022), it is possible for users to save specific service transactions initiated on the portal and access them later. The proportion of countries that allow users to access a list of their previous transactions or rely on AI-enabled chatbot functionality is nearly the same in 2024 as in 2022.\n\nFor the 2024 Survey, a new “affordability” indicator was added to the TII and the use of “the fixed (wired) broadband subscriptions” indicator was discontinued (see the technical appendix for details on the methodology used). While the technology subindex of the OSI does not include TII component data, it is important to highlight the changes in TII indicators as part of the evolving landscape of technological advancement. Table 2.8 summarizes the global and regional findings for indicators relating to mobile service pricing, broadband and cellular subscriptions, and Internet use for 2022 and 2024. Figure 2.28 highlights the changes in the latter three indicators between 2022 and 2024.\n\nInternet use and subscriptions for mobile broadband and cellular services are on the rise. The sharpest increases in Internet use have been in Oceania (49 per cent) and Africa (31 per cent). The active mobile broadband subscription rate has increased by 27 per cent in Africa, 10 per cent in Asia and the Americas, and 8 per cent in Europe, but has declined by almost 2 per cent in Oceania (see figure 2.28). Europe remains the leader in active mobile broadband use, with 105 subscriptions per 100 inhabitants, followed by Asia (89) and the Americas (73).\n\nFigure 2.27 Number of Member States with portals incorporating the assessed technology features, 2022 and 2024\n\n_Sources_ : 2022 and 2024 United Nations E-Government Surveys.\n\n193 UN Member States\n\nAmericas\n\nAfrica\n\nAsia\n\nEurope\n\nOceania\n\n## Very high EGD High EGDI Middle EGDI Low EGDI\n\n26%\n\n56%\n\n58%\n\n50%\n\n64%\n\n59%\n\n65%\n\n58%\n\n71%\n\n83%\n\n88%\n\n97%\n\n98%\n\n98%\n\n27%\n\n58%\n\n61%\n\n56%\n\n68%\n\n64%\n\n71%\n\n56%\n\n74%\n\n89%\n\n91%\n\n97%\n\n97%\n\n98%\n\n0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Portal offers AI-enabled chatbot functionality\n\nUsers can access a list of their previous interactions or transactions in the portal\n\nSpecifc service transactions initiated through the portal can be saved and accessed later\n\nIndividuals can modify their own data online\n\nIndividuals can access their own data online\n\nBusinesses can modify their own data online\n\nBusinesses can access their own data online\n\nThere is an \"advanced search\" option\n\nThere is a sitemap (or index)\n\nPortal has been updated in the last month\n\nPortal is developed using responsive web design\n\nPortal utilizes HTTPS\n\nThere is a \"contact us\" feature\n\nPortal can be found on search engine\n\n2024 2022","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"2024 UN E-GovErNmENt SUrvEy\n\n68\n\n**Chapter 2**\n\nThe number of countries that allow individuals and businesses to access or modify their data on government portals has increased by an average of 6 per cent since 2022. Globally, more countries allow users to access their data (71 per cent for businesses and 68 per cent for individuals) than to modify their data (64 per cent for businesses and 56 per cent for individuals).\n\nIn 118 countries (a 3 per cent increase from 2022), it is possible for users to save specific service transactions initiated on the portal and access them later. The proportion of countries that allow users to access a list of their previous transactions or rely on AI-enabled chatbot functionality is nearly the same in 2024 as in 2022.\n\nFor the 2024 Survey, a new “affordability” indicator was added to the TII and the use of “the fixed (wired) broadband subscriptions” indicator was discontinued (see the technical appendix for details on the methodology used). While the technology subindex of the OSI does not include TII component data, it is important to highlight the changes in TII indicators as part of the evolving landscape of technological advancement. Table 2.8 summarizes the global and regional findings for indicators relating to mobile service pricing, broadband and cellular subscriptions, and Internet use for 2022 and 2024. Figure 2.28 highlights the changes in the latter three indicators between 2022 and 2024.\n\nInternet use and subscriptions for mobile broadband and cellular services are on the rise. The sharpest increases in Internet use have been in Oceania (49 per cent) and Africa (31 per cent). The active mobile broadband subscription rate has increased by 27 per cent in Africa, 10 per cent in Asia and the Americas, and 8 per cent in Europe, but has declined by almost 2 per cent in Oceania (see figure 2.28). Europe remains the leader in active mobile broadband use, with 105 subscriptions per 100 inhabitants, followed by Asia (89) and the Americas (73).\n\nFigure 2.27 Number of Member States with portals incorporating the assessed technology features, 2022 and 2024\n\n_Sources_ : 2022 and 2024 United Nations E-Government Surveys.\n\n193 UN Member States\n\nAmericas\n\nAfrica\n\nAsia\n\nEurope\n\nOceania\n\n## Very high EGD High EGDI Middle EGDI Low EGDI\n\n26%\n\n56%\n\n58%\n\n50%\n\n64%\n\n59%\n\n65%\n\n58%\n\n71%\n\n83%\n\n88%\n\n97%\n\n98%\n\n98%\n\n27%\n\n58%\n\n61%\n\n56%\n\n68%\n\n64%\n\n71%\n\n56%\n\n74%\n\n89%\n\n91%\n\n97%\n\n97%\n\n98%\n\n0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Portal offers AI-enabled chatbot functionality\n\nUsers can access a list of their previous interactions or transactions in the portal\n\nSpecifc service transactions initiated through the portal can be saved and accessed later\n\nIndividuals can modify their own data online\n\nIndividuals can access their own data online\n\nBusinesses can modify their own data online\n\nBusinesses can access their own data online\n\nThere is an \"advanced search\" option\n\nThere is a sitemap (or index)\n\nPortal has been updated in the last month\n\nPortal is developed using responsive web design\n\nPortal utilizes HTTPS\n\nThere is a \"contact us\" feature\n\nPortal can be found on search engine\n\n2024 2022","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/%28Web_version%29_E-Government_Survey_2024_1392024_p93.pdf"]},"response":{"errors":[],"extract_id":"extract_b0016a043eaaec3a0d6e9c64687f307d","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["2024 UN E-GovErNmENt SUrvEy\n\n68\n\n**Chapter 2**\n\nThe number of countries that allow individuals and businesses to access or modify their data on government portals has increased by an average of 6 per cent since 2022. Globally, more countries allow users to access their data (71 per cent for businesses and 68 per cent for individuals) than to modify their data (64 per cent for businesses and 56 per cent for individuals).\n\nIn 118 countries (a 3 per cent increase from 2022), it is possible for users to save specific service transactions initiated on the portal and access them later. The proportion of countries that allow users to access a list of their previous transactions or rely on AI-enabled chatbot functionality is nearly the same in 2024 as in 2022.\n\nFor the 2024 Survey, a new “affordability” indicator was added to the TII and the use of “the fixed (wired) broadband subscriptions” indicator was discontinued (see the technical appendix for details on the methodology used). While the technology subindex of the OSI does not include TII component data, it is important to highlight the changes in TII indicators as part of the evolving landscape of technological advancement. Table 2.8 summarizes the global and regional findings for indicators relating to mobile service pricing, broadband and cellular subscriptions, and Internet use for 2022 and 2024. Figure 2.28 highlights the changes in the latter three indicators between 2022 and 2024.\n\nInternet use and subscriptions for mobile broadband and cellular services are on the rise. The sharpest increases in Internet use have been in Oceania (49 per cent) and Africa (31 per cent). The active mobile broadband subscription rate has increased by 27 per cent in Africa, 10 per cent in Asia and the Americas, and 8 per cent in Europe, but has declined by almost 2 per cent in Oceania (see figure 2.28). Europe remains the leader in active mobile broadband use, with 105 subscriptions per 100 inhabitants, followed by Asia (89) and the Americas (73).\n\nFigure 2.27 Number of Member States with portals incorporating the assessed technology features, 2022 and 2024\n\n_Sources_ : 2022 and 2024 United Nations E-Government Surveys.\n\n193 UN Member States\n\nAmericas\n\nAfrica\n\nAsia\n\nEurope\n\nOceania\n\n## Very high EGD High EGDI Middle EGDI Low EGDI\n\n26%\n\n56%\n\n58%\n\n50%\n\n64%\n\n59%\n\n65%\n\n58%\n\n71%\n\n83%\n\n88%\n\n97%\n\n98%\n\n98%\n\n27%\n\n58%\n\n61%\n\n56%\n\n68%\n\n64%\n\n71%\n\n56%\n\n74%\n\n89%\n\n91%\n\n97%\n\n97%\n\n98%\n\n0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Portal offers AI-enabled chatbot functionality\n\nUsers can access a list of their previous interactions or transactions in the portal\n\nSpecifc service transactions initiated through the portal can be saved and accessed later\n\nIndividuals can modify their own data online\n\nIndividuals can access their own data online\n\nBusinesses can modify their own data online\n\nBusinesses can access their own data online\n\nThere is an \"advanced search\" option\n\nThere is a sitemap (or index)\n\nPortal has been updated in the last month\n\nPortal is developed using responsive web design\n\nPortal utilizes HTTPS\n\nThere is a \"contact us\" feature\n\nPortal can be found on search engine\n\n2024 2022"],"full_content":"2024 UN E-GovErNmENt SUrvEy\n\n68\n\n**Chapter 2**\n\nThe number of countries that allow individuals and businesses to access or modify their data on government portals has increased by an average of 6 per cent since 2022. Globally, more countries allow users to access their data (71 per cent for businesses and 68 per cent for individuals) than to modify their data (64 per cent for businesses and 56 per cent for individuals).\n\nIn 118 countries (a 3 per cent increase from 2022), it is possible for users to save specific service transactions initiated on the portal and access them later. The proportion of countries that allow users to access a list of their previous transactions or rely on AI-enabled chatbot functionality is nearly the same in 2024 as in 2022.\n\nFor the 2024 Survey, a new “affordability” indicator was added to the TII and the use of “the fixed (wired) broadband subscriptions” indicator was discontinued (see the technical appendix for details on the methodology used). While the technology subindex of the OSI does not include TII component data, it is important to highlight the changes in TII indicators as part of the evolving landscape of technological advancement. Table 2.8 summarizes the global and regional findings for indicators relating to mobile service pricing, broadband and cellular subscriptions, and Internet use for 2022 and 2024. Figure 2.28 highlights the changes in the latter three indicators between 2022 and 2024.\n\nInternet use and subscriptions for mobile broadband and cellular services are on the rise. The sharpest increases in Internet use have been in Oceania (49 per cent) and Africa (31 per cent). The active mobile broadband subscription rate has increased by 27 per cent in Africa, 10 per cent in Asia and the Americas, and 8 per cent in Europe, but has declined by almost 2 per cent in Oceania (see figure 2.28). Europe remains the leader in active mobile broadband use, with 105 subscriptions per 100 inhabitants, followed by Asia (89) and the Americas (73).\n\nFigure 2.27 Number of Member States with portals incorporating the assessed technology features, 2022 and 2024\n\n_Sources_ : 2022 and 2024 United Nations E-Government Surveys.\n\n193 UN Member States\n\nAmericas\n\nAfrica\n\nAsia\n\nEurope\n\nOceania\n\n## Very high EGD High EGDI Middle EGDI Low EGDI\n\n26%\n\n56%\n\n58%\n\n50%\n\n64%\n\n59%\n\n65%\n\n58%\n\n71%\n\n83%\n\n88%\n\n97%\n\n98%\n\n98%\n\n27%\n\n58%\n\n61%\n\n56%\n\n68%\n\n64%\n\n71%\n\n56%\n\n74%\n\n89%\n\n91%\n\n97%\n\n97%\n\n98%\n\n0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Portal offers AI-enabled chatbot functionality\n\nUsers can access a list of their previous interactions or transactions in the portal\n\nSpecifc service transactions initiated through the portal can be saved and accessed later\n\nIndividuals can modify their own data online\n\nIndividuals can access their own data online\n\nBusinesses can modify their own data online\n\nBusinesses can access their own data online\n\nThere is an \"advanced search\" option\n\nThere is a sitemap (or index)\n\nPortal has been updated in the last month\n\nPortal is developed using responsive web design\n\nPortal utilizes HTTPS\n\nThere is a \"contact us\" feature\n\nPortal can be found on search engine\n\n2024 2022","publish_date":null,"title":"Very high EGDI High EGDI Middle EGDI Low EGDI","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/%28Web_version%29_E-Government_Survey_2024_1392024_p93.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/%28Web_version%29_E-Government_Survey_2024_1392024_p93.pdf"},"request":{"example_id":"chart/(Web_version)_E-Government_Survey_2024_1392024_p93","product_type":"parse","source_file_path":"(Web_version)_E-Government_Survey_2024_1392024_p93.pdf"},"started_at":"2026-11-04T19:46:11.399764"} +{"completed_at":"2026-11-04T19:46:16.444653","latency_in_ms":3142,"output":{"example_id":"chart/(Web_version)_E-Government_Survey_2024_1392024_p94","grounded_pages":[],"layout_pages":[],"markdown":"69\n\n**Chapter 2**\n\nChapter 2 • Global trends in e-Government\n\nAffordability plays a key role in digital access and engagement. The cost of Internet and active mobile broadband subscriptions as a percentage of gross national income (GNI) per capita remains significantly higher in Africa (10.4 per cent) and Oceania (7.6 per cent) than in other parts of the world, contributing to the digital divide (see figure 2.29).\n\nTable 2.8 Affordability of mobile data and voice services, mobile broadband and cellular subscriptions\n\nper 100 inhabitants, and percentage of individuals using the Internet, by region, 2022 and 2024\n\nAffordability\n\nActive mobile broadband\n\nsubscriptions per 100\n\ninhabitants\n\nMobile cellular\n\ntelephone subscriptions\n\nper 100 inhabitants\n\nPercentage of individuals\n\nusing the Internet\n\nMobile data and voice\n\nhigh consumption\n\nbasket price (as a\n\npercentage of GNI per\n\ncapita)\n\n2024 2022 2024 2022 2024 2022 2024 2022\n\nAfrica 10.4 - 54.4 42.77 88.95 83.68 43.4 33.01\n\nAmericas 3.8 - 72.8 65.96 102.83 101.92 77.1 67.81\n\nAsia 2.8 - 88.5 80.5 106.05 105.93 71.9 63.21\n\nEurope 1.1 - 105.2 97.9 116.75 113.86 89.5 85.52\n\nOceania 7.6 - 42.4 43.15 73.76 72.53 65.0 43.59\n\n193 UN Member States 5.2 - 76.5 68.47 100.73 98.32 68.3 59.14\n\n_Sources_ : ITU, Statistics for individuals using the Internet (2022 and 2024), available at https://www.itu.int/en/ITU-D/Statistics/Pages/ stat/default.aspx; ITU, “Mobile data and voice high-consumption basket”, DataHub, available at https://datahub.itu.int/data/?i=34619.\n\nFigure 2.28 Percentage change at the global and regional levels in Internet usage and in active mobile broadband and mobile cellular subscriptions per 100 inhabitants, 2022-2024\n\n_Sources_ : 2022 and 2024 United Nations E-Government Surveys.\n\n193 UN Member States\n\nAmericas\n\nAfrica\n\nAsia\n\nEurope\n\nOceania\n\n### Very high EGDI High EGDI Middle EGDI Low EGDI\n\n27\\.3\n\n6\\.3\n\n31\\.4\n\n10\\.4\n\n0\\.9\n\n13\\.6 9.9\n\n0\\.1\n\n13\\.8\n\n7\\.5\n\n2\\.5 4.6\n\n\\-1.8\n\n1\\.7\n\n49\\.0\n\n11\\.7\n\n2\\.5\n\n15\\.4\n\n\\-10\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\nPercentage change in active mobile broadband subscriptions per 100 inhabitants\n\nPercentage change in mobile cellular telephone subscriptions per 100 inhabitants\n\nPercentage change in Internet usage among individuals\n\nAfrica Americas Asia Europe Oceania 193 UN Member States\n\n### Very high EGDI High EGDI","pages":[{"line_numbers":[],"markdown":"69\n\n**Chapter 2**\n\nChapter 2 • Global trends in e-Government\n\nAffordability plays a key role in digital access and engagement. The cost of Internet and active mobile broadband subscriptions as a percentage of gross national income (GNI) per capita remains significantly higher in Africa (10.4 per cent) and Oceania (7.6 per cent) than in other parts of the world, contributing to the digital divide (see figure 2.29).\n\nTable 2.8 Affordability of mobile data and voice services, mobile broadband and cellular subscriptions\n\nper 100 inhabitants, and percentage of individuals using the Internet, by region, 2022 and 2024\n\nAffordability\n\nActive mobile broadband\n\nsubscriptions per 100\n\ninhabitants\n\nMobile cellular\n\ntelephone subscriptions\n\nper 100 inhabitants\n\nPercentage of individuals\n\nusing the Internet\n\nMobile data and voice\n\nhigh consumption\n\nbasket price (as a\n\npercentage of GNI per\n\ncapita)\n\n2024 2022 2024 2022 2024 2022 2024 2022\n\nAfrica 10.4 - 54.4 42.77 88.95 83.68 43.4 33.01\n\nAmericas 3.8 - 72.8 65.96 102.83 101.92 77.1 67.81\n\nAsia 2.8 - 88.5 80.5 106.05 105.93 71.9 63.21\n\nEurope 1.1 - 105.2 97.9 116.75 113.86 89.5 85.52\n\nOceania 7.6 - 42.4 43.15 73.76 72.53 65.0 43.59\n\n193 UN Member States 5.2 - 76.5 68.47 100.73 98.32 68.3 59.14\n\n_Sources_ : ITU, Statistics for individuals using the Internet (2022 and 2024), available at https://www.itu.int/en/ITU-D/Statistics/Pages/ stat/default.aspx; ITU, “Mobile data and voice high-consumption basket”, DataHub, available at https://datahub.itu.int/data/?i=34619.\n\nFigure 2.28 Percentage change at the global and regional levels in Internet usage and in active mobile broadband and mobile cellular subscriptions per 100 inhabitants, 2022-2024\n\n_Sources_ : 2022 and 2024 United Nations E-Government Surveys.\n\n193 UN Member States\n\nAmericas\n\nAfrica\n\nAsia\n\nEurope\n\nOceania\n\n### Very high EGDI High EGDI Middle EGDI Low EGDI\n\n27\\.3\n\n6\\.3\n\n31\\.4\n\n10\\.4\n\n0\\.9\n\n13\\.6 9.9\n\n0\\.1\n\n13\\.8\n\n7\\.5\n\n2\\.5 4.6\n\n\\-1.8\n\n1\\.7\n\n49\\.0\n\n11\\.7\n\n2\\.5\n\n15\\.4\n\n\\-10\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\nPercentage change in active mobile broadband subscriptions per 100 inhabitants\n\nPercentage change in mobile cellular telephone subscriptions per 100 inhabitants\n\nPercentage change in Internet usage among individuals\n\nAfrica Americas Asia Europe Oceania 193 UN Member States\n\n### Very high EGDI High EGDI","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"69\n\n**Chapter 2**\n\nChapter 2 • Global trends in e-Government\n\nAffordability plays a key role in digital access and engagement. The cost of Internet and active mobile broadband subscriptions as a percentage of gross national income (GNI) per capita remains significantly higher in Africa (10.4 per cent) and Oceania (7.6 per cent) than in other parts of the world, contributing to the digital divide (see figure 2.29).\n\nTable 2.8 Affordability of mobile data and voice services, mobile broadband and cellular subscriptions\n\nper 100 inhabitants, and percentage of individuals using the Internet, by region, 2022 and 2024\n\nAffordability\n\nActive mobile broadband\n\nsubscriptions per 100\n\ninhabitants\n\nMobile cellular\n\ntelephone subscriptions\n\nper 100 inhabitants\n\nPercentage of individuals\n\nusing the Internet\n\nMobile data and voice\n\nhigh consumption\n\nbasket price (as a\n\npercentage of GNI per\n\ncapita)\n\n2024 2022 2024 2022 2024 2022 2024 2022\n\nAfrica 10.4 - 54.4 42.77 88.95 83.68 43.4 33.01\n\nAmericas 3.8 - 72.8 65.96 102.83 101.92 77.1 67.81\n\nAsia 2.8 - 88.5 80.5 106.05 105.93 71.9 63.21\n\nEurope 1.1 - 105.2 97.9 116.75 113.86 89.5 85.52\n\nOceania 7.6 - 42.4 43.15 73.76 72.53 65.0 43.59\n\n193 UN Member States 5.2 - 76.5 68.47 100.73 98.32 68.3 59.14\n\n_Sources_ : ITU, Statistics for individuals using the Internet (2022 and 2024), available at https://www.itu.int/en/ITU-D/Statistics/Pages/ stat/default.aspx; ITU, “Mobile data and voice high-consumption basket”, DataHub, available at https://datahub.itu.int/data/?i=34619.\n\nFigure 2.28 Percentage change at the global and regional levels in Internet usage and in active mobile broadband and mobile cellular subscriptions per 100 inhabitants, 2022-2024\n\n_Sources_ : 2022 and 2024 United Nations E-Government Surveys.\n\n193 UN Member States\n\nAmericas\n\nAfrica\n\nAsia\n\nEurope\n\nOceania\n\n### Very high EGDI High EGDI Middle EGDI Low EGDI\n\n27\\.3\n\n6\\.3\n\n31\\.4\n\n10\\.4\n\n0\\.9\n\n13\\.6 9.9\n\n0\\.1\n\n13\\.8\n\n7\\.5\n\n2\\.5 4.6\n\n\\-1.8\n\n1\\.7\n\n49\\.0\n\n11\\.7\n\n2\\.5\n\n15\\.4\n\n\\-10\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\nPercentage change in active mobile broadband subscriptions per 100 inhabitants\n\nPercentage change in mobile cellular telephone subscriptions per 100 inhabitants\n\nPercentage change in Internet usage among individuals\n\nAfrica Americas Asia Europe Oceania 193 UN Member States\n\n### Very high EGDI High EGDI","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/%28Web_version%29_E-Government_Survey_2024_1392024_p94.pdf"]},"response":{"errors":[],"extract_id":"extract_44eff927c162c82ede0561df78fc1cd9","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["69\n\n**Chapter 2**\n\nChapter 2 • Global trends in e-Government\n\nAffordability plays a key role in digital access and engagement. The cost of Internet and active mobile broadband subscriptions as a percentage of gross national income (GNI) per capita remains significantly higher in Africa (10.4 per cent) and Oceania (7.6 per cent) than in other parts of the world, contributing to the digital divide (see figure 2.29).\n\nTable 2.8 Affordability of mobile data and voice services, mobile broadband and cellular subscriptions\n\nper 100 inhabitants, and percentage of individuals using the Internet, by region, 2022 and 2024\n\nAffordability\n\nActive mobile broadband\n\nsubscriptions per 100\n\ninhabitants\n\nMobile cellular\n\ntelephone subscriptions\n\nper 100 inhabitants\n\nPercentage of individuals\n\nusing the Internet\n\nMobile data and voice\n\nhigh consumption\n\nbasket price (as a\n\npercentage of GNI per\n\ncapita)\n\n2024 2022 2024 2022 2024 2022 2024 2022\n\nAfrica 10.4 - 54.4 42.77 88.95 83.68 43.4 33.01\n\nAmericas 3.8 - 72.8 65.96 102.83 101.92 77.1 67.81\n\nAsia 2.8 - 88.5 80.5 106.05 105.93 71.9 63.21\n\nEurope 1.1 - 105.2 97.9 116.75 113.86 89.5 85.52\n\nOceania 7.6 - 42.4 43.15 73.76 72.53 65.0 43.59\n\n193 UN Member States 5.2 - 76.5 68.47 100.73 98.32 68.3 59.14\n\n_Sources_ : ITU, Statistics for individuals using the Internet (2022 and 2024), available at https://www.itu.int/en/ITU-D/Statistics/Pages/ stat/default.aspx; ITU, “Mobile data and voice high-consumption basket”, DataHub, available at https://datahub.itu.int/data/?i=34619.\n\nFigure 2.28 Percentage change at the global and regional levels in Internet usage and in active mobile broadband and mobile cellular subscriptions per 100 inhabitants, 2022-2024\n\n_Sources_ : 2022 and 2024 United Nations E-Government Surveys.\n\n193 UN Member States\n\nAmericas\n\nAfrica\n\nAsia\n\nEurope\n\nOceania\n\n### Very high EGDI High EGDI Middle EGDI Low EGDI\n\n27\\.3\n\n6\\.3\n\n31\\.4\n\n10\\.4\n\n0\\.9\n\n13\\.6 9.9\n\n0\\.1\n\n13\\.8\n\n7\\.5\n\n2\\.5 4.6\n\n\\-1.8\n\n1\\.7\n\n49\\.0\n\n11\\.7\n\n2\\.5\n\n15\\.4\n\n\\-10\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\nPercentage change in active mobile broadband subscriptions per 100 inhabitants\n\nPercentage change in mobile cellular telephone subscriptions per 100 inhabitants\n\nPercentage change in Internet usage among individuals\n\nAfrica Americas Asia Europe Oceania 193 UN Member States\n\n### Very high EGDI High EGDI"],"full_content":"69\n\n**Chapter 2**\n\nChapter 2 • Global trends in e-Government\n\nAffordability plays a key role in digital access and engagement. The cost of Internet and active mobile broadband subscriptions as a percentage of gross national income (GNI) per capita remains significantly higher in Africa (10.4 per cent) and Oceania (7.6 per cent) than in other parts of the world, contributing to the digital divide (see figure 2.29).\n\nTable 2.8 Affordability of mobile data and voice services, mobile broadband and cellular subscriptions\n\nper 100 inhabitants, and percentage of individuals using the Internet, by region, 2022 and 2024\n\nAffordability\n\nActive mobile broadband\n\nsubscriptions per 100\n\ninhabitants\n\nMobile cellular\n\ntelephone subscriptions\n\nper 100 inhabitants\n\nPercentage of individuals\n\nusing the Internet\n\nMobile data and voice\n\nhigh consumption\n\nbasket price (as a\n\npercentage of GNI per\n\ncapita)\n\n2024 2022 2024 2022 2024 2022 2024 2022\n\nAfrica 10.4 - 54.4 42.77 88.95 83.68 43.4 33.01\n\nAmericas 3.8 - 72.8 65.96 102.83 101.92 77.1 67.81\n\nAsia 2.8 - 88.5 80.5 106.05 105.93 71.9 63.21\n\nEurope 1.1 - 105.2 97.9 116.75 113.86 89.5 85.52\n\nOceania 7.6 - 42.4 43.15 73.76 72.53 65.0 43.59\n\n193 UN Member States 5.2 - 76.5 68.47 100.73 98.32 68.3 59.14\n\n_Sources_ : ITU, Statistics for individuals using the Internet (2022 and 2024), available at https://www.itu.int/en/ITU-D/Statistics/Pages/ stat/default.aspx; ITU, “Mobile data and voice high-consumption basket”, DataHub, available at https://datahub.itu.int/data/?i=34619.\n\nFigure 2.28 Percentage change at the global and regional levels in Internet usage and in active mobile broadband and mobile cellular subscriptions per 100 inhabitants, 2022-2024\n\n_Sources_ : 2022 and 2024 United Nations E-Government Surveys.\n\n193 UN Member States\n\nAmericas\n\nAfrica\n\nAsia\n\nEurope\n\nOceania\n\n### Very high EGDI High EGDI Middle EGDI Low EGDI\n\n27\\.3\n\n6\\.3\n\n31\\.4\n\n10\\.4\n\n0\\.9\n\n13\\.6 9.9\n\n0\\.1\n\n13\\.8\n\n7\\.5\n\n2\\.5 4.6\n\n\\-1.8\n\n1\\.7\n\n49\\.0\n\n11\\.7\n\n2\\.5\n\n15\\.4\n\n\\-10\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\nPercentage change in active mobile broadband subscriptions per 100 inhabitants\n\nPercentage change in mobile cellular telephone subscriptions per 100 inhabitants\n\nPercentage change in Internet usage among individuals\n\nAfrica Americas Asia Europe Oceania 193 UN Member States\n\n### Very high EGDI High EGDI","publish_date":null,"title":"Very high EGDI High EGDI Middle EGDI Low EGDI Very high EGDI Very high EGDI High EGDI High EGDI Middle","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/%28Web_version%29_E-Government_Survey_2024_1392024_p94.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/%28Web_version%29_E-Government_Survey_2024_1392024_p94.pdf"},"request":{"example_id":"chart/(Web_version)_E-Government_Survey_2024_1392024_p94","product_type":"parse","source_file_path":"(Web_version)_E-Government_Survey_2024_1392024_p94.pdf"},"started_at":"2026-11-04T19:46:13.302564"} +{"completed_at":"2026-11-04T19:46:16.316898","latency_in_ms":1188,"output":{"example_id":"chart/(Web_version)_E-Government_Survey_2024_1392024_p97","grounded_pages":[],"layout_pages":[],"markdown":"2024 UN E-GovErNmENt SUrvEy\n\n72\n\n**Chapter 2**\n\nFigure 2.30 The distribution of countries in special situations among the four EGDI levels, 2022 and 2024\n\n_Sources_ : 2022 and 2024 United Nations E-Government Surveys.\n\n_Notes_ : countries in special situations include least developed countries (lDcs), landlocked developing countries (llDcs), and small island developing States (SIDS). The list of lDcs and SIDS have changed from 2022 due to Bhutan graduating from the lDc category and Bahrain graduating from the SIDS category. Percentages may not add to 100 due to rounding.\n\nEDGI 2024 levels\n\nVery high EGDI High EGDI Middle EGDI Low EGDI\n\n## Very high High EGD Middle E Low EGD\n\n193 UN Member States\n\nAmericas\n\nAfrica\n\nAsia\n\nEurope\n\nOceania\n\n## Very high EGDI High EGDI Middle EGDI Low EGDI\n\n22%\n\n19%\n\n3%\n\n62%\n\n28%\n\n38%\n\n16%\n\n34%\n\n54%\n\n19%\n\n5%\n\nLDCs (46 countries)\n\nLLDCs (32 countries)\n\nSIDS (38 countries)\n\n15%\n\n13%\n\n3%\n\n72%\n\n38%\n\n37%\n\n13%\n\n47%\n\n55%\n\n3%\n\n5%\n\nLDCs (45 countries)\n\nLLDCs (32 countries)\n\nSIDS (37 countries)\n\nLow EGDI Middle EGDI High EGDI Very high EGDI 2022 label Low EGDI Middle EGDI High EGDI Very high EGDI\n\n2022 2024\n\nLow EGDI Middle EGDI High EGDI Very high EGDI 2022 label 2024 label\n\nLow EGDI Middle EGDI High EGDI Very high EGDI\n\nThe overall distribution of SIDS across the different EGDI levels has remained relatively stable (see figure 2.30). The majority of SIDS (54 per cent) are in the high EGDI group, followed by 38 per cent in the middle group, 5 per cent in the very high group, and 3 per cent in the low group. Notable changes include Mauritius moving from the high to the very high EGDI group, Vanuatu advancing from the middle to the high EGDI group, and Belize dropping from the high to the middle EGDI level. These shifts illustrate both progress and setbacks, highlighting the diverse digital development trajectories within SIDS. While disparities remain, the general stability in the EGDI distribution suggests a consistent focus on enhancing digital services and public engagement.\n\nThe llDcs have experienced more dynamic changes in e-government development since 2022. Armenia, Azerbaijan, Mongolia, the Republic of Moldova, and Uzbekistan have all moved from the high to the very high EGDI group, increasing the proportion of llDcs in the latter group from 3 to 19 per cent. Eswatini has also achieved notable progress and has moved from the middle to the high EGDI group. conversely, Afghanistan and Burundi have moved from the middle to the low EGDI group, raising the proportion of llDcs in the low EGDI group from 13 to 19 per cent. These shifts indicate both significant progress and ongoing challenges within the llDcs.\n\nThe majority of lDcs (62 per cent) remain in the middle EGDI group. However, there have been notable shifts, with two countries moving up to the high EGDI level and three countries dropping to the low EGDI level; this has increased the proportions of countries in the high and low EGDI groups by 3 and 7 per cent, respectively. These trends highlight widening digital divides within the lDcs, underscoring the need for targeted interventions to support countries lagging in digital development.","pages":[{"line_numbers":[],"markdown":"2024 UN E-GovErNmENt SUrvEy\n\n72\n\n**Chapter 2**\n\nFigure 2.30 The distribution of countries in special situations among the four EGDI levels, 2022 and 2024\n\n_Sources_ : 2022 and 2024 United Nations E-Government Surveys.\n\n_Notes_ : countries in special situations include least developed countries (lDcs), landlocked developing countries (llDcs), and small island developing States (SIDS). The list of lDcs and SIDS have changed from 2022 due to Bhutan graduating from the lDc category and Bahrain graduating from the SIDS category. Percentages may not add to 100 due to rounding.\n\nEDGI 2024 levels\n\nVery high EGDI High EGDI Middle EGDI Low EGDI\n\n## Very high High EGD Middle E Low EGD\n\n193 UN Member States\n\nAmericas\n\nAfrica\n\nAsia\n\nEurope\n\nOceania\n\n## Very high EGDI High EGDI Middle EGDI Low EGDI\n\n22%\n\n19%\n\n3%\n\n62%\n\n28%\n\n38%\n\n16%\n\n34%\n\n54%\n\n19%\n\n5%\n\nLDCs (46 countries)\n\nLLDCs (32 countries)\n\nSIDS (38 countries)\n\n15%\n\n13%\n\n3%\n\n72%\n\n38%\n\n37%\n\n13%\n\n47%\n\n55%\n\n3%\n\n5%\n\nLDCs (45 countries)\n\nLLDCs (32 countries)\n\nSIDS (37 countries)\n\nLow EGDI Middle EGDI High EGDI Very high EGDI 2022 label Low EGDI Middle EGDI High EGDI Very high EGDI\n\n2022 2024\n\nLow EGDI Middle EGDI High EGDI Very high EGDI 2022 label 2024 label\n\nLow EGDI Middle EGDI High EGDI Very high EGDI\n\nThe overall distribution of SIDS across the different EGDI levels has remained relatively stable (see figure 2.30). The majority of SIDS (54 per cent) are in the high EGDI group, followed by 38 per cent in the middle group, 5 per cent in the very high group, and 3 per cent in the low group. Notable changes include Mauritius moving from the high to the very high EGDI group, Vanuatu advancing from the middle to the high EGDI group, and Belize dropping from the high to the middle EGDI level. These shifts illustrate both progress and setbacks, highlighting the diverse digital development trajectories within SIDS. While disparities remain, the general stability in the EGDI distribution suggests a consistent focus on enhancing digital services and public engagement.\n\nThe llDcs have experienced more dynamic changes in e-government development since 2022. Armenia, Azerbaijan, Mongolia, the Republic of Moldova, and Uzbekistan have all moved from the high to the very high EGDI group, increasing the proportion of llDcs in the latter group from 3 to 19 per cent. Eswatini has also achieved notable progress and has moved from the middle to the high EGDI group. conversely, Afghanistan and Burundi have moved from the middle to the low EGDI group, raising the proportion of llDcs in the low EGDI group from 13 to 19 per cent. These shifts indicate both significant progress and ongoing challenges within the llDcs.\n\nThe majority of lDcs (62 per cent) remain in the middle EGDI group. However, there have been notable shifts, with two countries moving up to the high EGDI level and three countries dropping to the low EGDI level; this has increased the proportions of countries in the high and low EGDI groups by 3 and 7 per cent, respectively. These trends highlight widening digital divides within the lDcs, underscoring the need for targeted interventions to support countries lagging in digital development.","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"2024 UN E-GovErNmENt SUrvEy\n\n72\n\n**Chapter 2**\n\nFigure 2.30 The distribution of countries in special situations among the four EGDI levels, 2022 and 2024\n\n_Sources_ : 2022 and 2024 United Nations E-Government Surveys.\n\n_Notes_ : countries in special situations include least developed countries (lDcs), landlocked developing countries (llDcs), and small island developing States (SIDS). The list of lDcs and SIDS have changed from 2022 due to Bhutan graduating from the lDc category and Bahrain graduating from the SIDS category. Percentages may not add to 100 due to rounding.\n\nEDGI 2024 levels\n\nVery high EGDI High EGDI Middle EGDI Low EGDI\n\n## Very high High EGD Middle E Low EGD\n\n193 UN Member States\n\nAmericas\n\nAfrica\n\nAsia\n\nEurope\n\nOceania\n\n## Very high EGDI High EGDI Middle EGDI Low EGDI\n\n22%\n\n19%\n\n3%\n\n62%\n\n28%\n\n38%\n\n16%\n\n34%\n\n54%\n\n19%\n\n5%\n\nLDCs (46 countries)\n\nLLDCs (32 countries)\n\nSIDS (38 countries)\n\n15%\n\n13%\n\n3%\n\n72%\n\n38%\n\n37%\n\n13%\n\n47%\n\n55%\n\n3%\n\n5%\n\nLDCs (45 countries)\n\nLLDCs (32 countries)\n\nSIDS (37 countries)\n\nLow EGDI Middle EGDI High EGDI Very high EGDI 2022 label Low EGDI Middle EGDI High EGDI Very high EGDI\n\n2022 2024\n\nLow EGDI Middle EGDI High EGDI Very high EGDI 2022 label 2024 label\n\nLow EGDI Middle EGDI High EGDI Very high EGDI\n\nThe overall distribution of SIDS across the different EGDI levels has remained relatively stable (see figure 2.30). The majority of SIDS (54 per cent) are in the high EGDI group, followed by 38 per cent in the middle group, 5 per cent in the very high group, and 3 per cent in the low group. Notable changes include Mauritius moving from the high to the very high EGDI group, Vanuatu advancing from the middle to the high EGDI group, and Belize dropping from the high to the middle EGDI level. These shifts illustrate both progress and setbacks, highlighting the diverse digital development trajectories within SIDS. While disparities remain, the general stability in the EGDI distribution suggests a consistent focus on enhancing digital services and public engagement.\n\nThe llDcs have experienced more dynamic changes in e-government development since 2022. Armenia, Azerbaijan, Mongolia, the Republic of Moldova, and Uzbekistan have all moved from the high to the very high EGDI group, increasing the proportion of llDcs in the latter group from 3 to 19 per cent. Eswatini has also achieved notable progress and has moved from the middle to the high EGDI group. conversely, Afghanistan and Burundi have moved from the middle to the low EGDI group, raising the proportion of llDcs in the low EGDI group from 13 to 19 per cent. These shifts indicate both significant progress and ongoing challenges within the llDcs.\n\nThe majority of lDcs (62 per cent) remain in the middle EGDI group. However, there have been notable shifts, with two countries moving up to the high EGDI level and three countries dropping to the low EGDI level; this has increased the proportions of countries in the high and low EGDI groups by 3 and 7 per cent, respectively. These trends highlight widening digital divides within the lDcs, underscoring the need for targeted interventions to support countries lagging in digital development.","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/%28Web_version%29_E-Government_Survey_2024_1392024_p97.pdf"]},"response":{"errors":[],"extract_id":"extract_9f8e96eac09989f96aab2f23a399c8fa","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["2024 UN E-GovErNmENt SUrvEy\n\n72\n\n**Chapter 2**\n\nFigure 2.30 The distribution of countries in special situations among the four EGDI levels, 2022 and 2024\n\n_Sources_ : 2022 and 2024 United Nations E-Government Surveys.\n\n_Notes_ : countries in special situations include least developed countries (lDcs), landlocked developing countries (llDcs), and small island developing States (SIDS). The list of lDcs and SIDS have changed from 2022 due to Bhutan graduating from the lDc category and Bahrain graduating from the SIDS category. Percentages may not add to 100 due to rounding.\n\nEDGI 2024 levels\n\nVery high EGDI High EGDI Middle EGDI Low EGDI\n\n## Very high High EGD Middle E Low EGD\n\n193 UN Member States\n\nAmericas\n\nAfrica\n\nAsia\n\nEurope\n\nOceania\n\n## Very high EGDI High EGDI Middle EGDI Low EGDI\n\n22%\n\n19%\n\n3%\n\n62%\n\n28%\n\n38%\n\n16%\n\n34%\n\n54%\n\n19%\n\n5%\n\nLDCs (46 countries)\n\nLLDCs (32 countries)\n\nSIDS (38 countries)\n\n15%\n\n13%\n\n3%\n\n72%\n\n38%\n\n37%\n\n13%\n\n47%\n\n55%\n\n3%\n\n5%\n\nLDCs (45 countries)\n\nLLDCs (32 countries)\n\nSIDS (37 countries)\n\nLow EGDI Middle EGDI High EGDI Very high EGDI 2022 label Low EGDI Middle EGDI High EGDI Very high EGDI\n\n2022 2024\n\nLow EGDI Middle EGDI High EGDI Very high EGDI 2022 label 2024 label\n\nLow EGDI Middle EGDI High EGDI Very high EGDI\n\nThe overall distribution of SIDS across the different EGDI levels has remained relatively stable (see figure 2.30). The majority of SIDS (54 per cent) are in the high EGDI group, followed by 38 per cent in the middle group, 5 per cent in the very high group, and 3 per cent in the low group. Notable changes include Mauritius moving from the high to the very high EGDI group, Vanuatu advancing from the middle to the high EGDI group, and Belize dropping from the high to the middle EGDI level. These shifts illustrate both progress and setbacks, highlighting the diverse digital development trajectories within SIDS. While disparities remain, the general stability in the EGDI distribution suggests a consistent focus on enhancing digital services and public engagement.\n\nThe llDcs have experienced more dynamic changes in e-government development since 2022. Armenia, Azerbaijan, Mongolia, the Republic of Moldova, and Uzbekistan have all moved from the high to the very high EGDI group, increasing the proportion of llDcs in the latter group from 3 to 19 per cent. Eswatini has also achieved notable progress and has moved from the middle to the high EGDI group. conversely, Afghanistan and Burundi have moved from the middle to the low EGDI group, raising the proportion of llDcs in the low EGDI group from 13 to 19 per cent. These shifts indicate both significant progress and ongoing challenges within the llDcs.\n\nThe majority of lDcs (62 per cent) remain in the middle EGDI group. However, there have been notable shifts, with two countries moving up to the high EGDI level and three countries dropping to the low EGDI level; this has increased the proportions of countries in the high and low EGDI groups by 3 and 7 per cent, respectively. These trends highlight widening digital divides within the lDcs, underscoring the need for targeted interventions to support countries lagging in digital development."],"full_content":"2024 UN E-GovErNmENt SUrvEy\n\n72\n\n**Chapter 2**\n\nFigure 2.30 The distribution of countries in special situations among the four EGDI levels, 2022 and 2024\n\n_Sources_ : 2022 and 2024 United Nations E-Government Surveys.\n\n_Notes_ : countries in special situations include least developed countries (lDcs), landlocked developing countries (llDcs), and small island developing States (SIDS). The list of lDcs and SIDS have changed from 2022 due to Bhutan graduating from the lDc category and Bahrain graduating from the SIDS category. Percentages may not add to 100 due to rounding.\n\nEDGI 2024 levels\n\nVery high EGDI High EGDI Middle EGDI Low EGDI\n\n## Very high High EGD Middle E Low EGD\n\n193 UN Member States\n\nAmericas\n\nAfrica\n\nAsia\n\nEurope\n\nOceania\n\n## Very high EGDI High EGDI Middle EGDI Low EGDI\n\n22%\n\n19%\n\n3%\n\n62%\n\n28%\n\n38%\n\n16%\n\n34%\n\n54%\n\n19%\n\n5%\n\nLDCs (46 countries)\n\nLLDCs (32 countries)\n\nSIDS (38 countries)\n\n15%\n\n13%\n\n3%\n\n72%\n\n38%\n\n37%\n\n13%\n\n47%\n\n55%\n\n3%\n\n5%\n\nLDCs (45 countries)\n\nLLDCs (32 countries)\n\nSIDS (37 countries)\n\nLow EGDI Middle EGDI High EGDI Very high EGDI 2022 label Low EGDI Middle EGDI High EGDI Very high EGDI\n\n2022 2024\n\nLow EGDI Middle EGDI High EGDI Very high EGDI 2022 label 2024 label\n\nLow EGDI Middle EGDI High EGDI Very high EGDI\n\nThe overall distribution of SIDS across the different EGDI levels has remained relatively stable (see figure 2.30). The majority of SIDS (54 per cent) are in the high EGDI group, followed by 38 per cent in the middle group, 5 per cent in the very high group, and 3 per cent in the low group. Notable changes include Mauritius moving from the high to the very high EGDI group, Vanuatu advancing from the middle to the high EGDI group, and Belize dropping from the high to the middle EGDI level. These shifts illustrate both progress and setbacks, highlighting the diverse digital development trajectories within SIDS. While disparities remain, the general stability in the EGDI distribution suggests a consistent focus on enhancing digital services and public engagement.\n\nThe llDcs have experienced more dynamic changes in e-government development since 2022. Armenia, Azerbaijan, Mongolia, the Republic of Moldova, and Uzbekistan have all moved from the high to the very high EGDI group, increasing the proportion of llDcs in the latter group from 3 to 19 per cent. Eswatini has also achieved notable progress and has moved from the middle to the high EGDI group. conversely, Afghanistan and Burundi have moved from the middle to the low EGDI group, raising the proportion of llDcs in the low EGDI group from 13 to 19 per cent. These shifts indicate both significant progress and ongoing challenges within the llDcs.\n\nThe majority of lDcs (62 per cent) remain in the middle EGDI group. However, there have been notable shifts, with two countries moving up to the high EGDI level and three countries dropping to the low EGDI level; this has increased the proportions of countries in the high and low EGDI groups by 3 and 7 per cent, respectively. These trends highlight widening digital divides within the lDcs, underscoring the need for targeted interventions to support countries lagging in digital development.","publish_date":null,"title":"Very high EGDI Very high EGDI High EGDI High EGDI Middle EGDI Middle EGDI Low EGDI Low EGDI","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/%28Web_version%29_E-Government_Survey_2024_1392024_p97.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/%28Web_version%29_E-Government_Survey_2024_1392024_p97.pdf"},"request":{"example_id":"chart/(Web_version)_E-Government_Survey_2024_1392024_p97","product_type":"parse","source_file_path":"(Web_version)_E-Government_Survey_2024_1392024_p97.pdf"},"started_at":"2026-11-04T19:46:15.128152"} +{"completed_at":"2026-11-04T19:46:18.960633","latency_in_ms":2514,"output":{"example_id":"chart/(Web_version)_E-Government_Survey_2024_1392024_p99","grounded_pages":[],"layout_pages":[],"markdown":"2024 UN E-GovErNmENt SUrvEy\n\n74\n\n**Chapter 2**\n\nFigure 2.32 Average EGDI composite and component values for countries in special situations, 2022 and 2024\n\n_Sources_ : 2022 and 2024 United Nations E-Government Surveys.\n\n_Notes_ : countries in special situations include least developed countries (lDcs), landlocked developing countries (llDcs), and small island developing States (SIDS). The list of lDcs and SIDSs have changed since 2022 with the graduation of Bhutan from the lDc category and the graduation of Bahrain from the SIDS category; the slight changes in the number of countries assessed and the necessary recalculations have resulted in small variations between the numbers reported here and those reported in the previous Survey.\n\nPrivate consumption Government spending Fixed investment Net exports\n\n0\\.4885\n\n0\\.3598\n\n0\\.3321\n\n0\\.2407\n\n0\\.4076\n\n0\\.4191\n\n(0.4642)\n\n(0.3597)\n\n(0.3308)\n\n(0.2226)\n\n(0.3931)\n\n(0.4054)\n\n0\\.00 0.25 0.50 0.75\n\nLLDCs\n\nLDC/LLDCs\n\nLDCs\n\nLDC/SIDS\n\nSIDS\n\nALL CSS\n\n### OSI\n\n2022 2024 Global average (2024), 0.5754\n\n0\\.5138\n\n0\\.3335\n\n0\\.3547\n\n0\\.3551\n\n0\\.5412\n\n0\\.4884\n\n(0.4847)\n\n(0.3368)\n\n(0.3498)\n\n(0.3498)\n\n(0.5262)\n\n(0.4703)\n\n0\\.00 0.25 0.50 0.75\n\nLLDCs\n\nLDC/LLDCs\n\nLDCs\n\nLDC/SIDS\n\nSIDS\n\nALL CSS\n\n### EGDI\n\n2022 2024 Global average (2024), 0.6382\n\n0\\.5332\n\n0\\.2949\n\n0\\.3539\n\n0\\.3624\n\n0\\.5932\n\n0\\.5206\n\n(0.4121)\n\n(0.2481)\n\n(0.2848)\n\n(0.2925)\n\n(0.4870)\n\n(0.4163)\n\n0\\.00 0.25 0.50 0.75\n\nLLDCs\n\nLDC/LLDCs\n\nLDCs\n\nLDC/SIDS\n\nSIDS\n\nALL CSS\n\n### TII\n\n2022 2024 Global average (2024), 0.6896\n\n0\\.5196\n\n0\\.3458\n\n0\\.3781\n\n0\\.4622\n\n0\\.6229\n\n0\\.5255\n\n(0.5778)\n\n(0.4026)\n\n(0.4337)\n\n(0.5342)\n\n(0.6985)\n\n(0.5893)\n\n0\\.00 0.25 0.50 0.75\n\nLLDCs\n\nLDC/LLDCs\n\nLDCs\n\nLDC/SIDS\n\nSIDS\n\nALL CSS\n\n### HCI\n\n2022 2024 Global average (2024), 0.6496\n\nPrivate consumption Government spending Fixed investment Net exports\n\n0\\.4885\n\n0\\.3598\n\n0\\.3321\n\n0\\.2407\n\n0\\.4076\n\n0\\.4191\n\n(0.4642)\n\n(0.3597)\n\n(0.3308)\n\n(0.2226)\n\n(0.3931)\n\n(0.4054)\n\n0\\.00 0.25 0.50 0.75\n\nLLDCs\n\nLDC/LLDCs\n\nLDCs\n\nLDC/SIDS\n\nSIDS\n\nALL CSS\n\n### OSI\n\n2022 2024 Global average (2024), 0.5754\n\n0\\.5138\n\n0\\.3335\n\n0\\.3547\n\n0\\.3551\n\n0\\.5412\n\n0\\.4884\n\n(0.4847)\n\n(0.3368)\n\n(0.3498)\n\n(0.3498)\n\n(0.5262)\n\n(0.4703)\n\n0\\.00 0.25 0.50 0.75\n\nLLDCs\n\nLDC/LLDCs\n\nLDCs\n\nLDC/SIDS\n\nSIDS\n\nALL CSS\n\n### EGDI\n\n2022 2024 Global average (2024), 0.6382\n\n0\\.5332\n\n0\\.2949\n\n0\\.3539\n\n0\\.3624\n\n0\\.5932\n\n0\\.5206\n\n(0.4121)\n\n(0.2481)\n\n(0.2848)\n\n(0.2925)\n\n(0.4870)\n\n(0.4163)\n\n0\\.00 0.25 0.50 0.75\n\nLLDCs\n\nLDC/LLDCs\n\nLDCs\n\nLDC/SIDS\n\nSIDS\n\nALL CSS\n\n### TII\n\n2022 2024 Global average (2024), 0.6896\n\n0\\.5196\n\n0\\.3458\n\n0\\.3781\n\n0\\.4622\n\n0\\.6229\n\n0\\.5255\n\n(0.5778)\n\n(0.4026)\n\n(0.4337)\n\n(0.5342)\n\n(0.6985)\n\n(0.5893)\n\n0\\.00 0.25 0.50 0.75\n\nLLDCs\n\nLDC/LLDCs\n\nLDCs\n\nLDC/SIDS\n\nSIDS\n\nALL CSS\n\n### HCI\n\n2022 2024 Global average (2024), 0.6496\n\nWith a 6 per cent increase in their average EGDI value, the llDcs have achieved the most significant gains among three groups. This notable improvement reflects the success of these countries in overcoming geographical barriers to strengthen e-government. Such progress contrasts with the stagnant or declining trends observed for lDc/llDcs, highlighting intragroup disparities and the need for targeted interventions.\n\nThe average EGDI value for SIDS has increased by 3 per cent, reflecting steady progress in digital government development. The lDc/SIDS have shown significant improvement, particularly in online services provision; their average OSI value rose by 8 per cent – the highest increase for this component among the countries in special situations. The progress made by this group suggests that access to marine transportation routes and other infrastructure-related advantages play a critical role in advancing the development and provision of digital services.","pages":[{"line_numbers":[],"markdown":"2024 UN E-GovErNmENt SUrvEy\n\n74\n\n**Chapter 2**\n\nFigure 2.32 Average EGDI composite and component values for countries in special situations, 2022 and 2024\n\n_Sources_ : 2022 and 2024 United Nations E-Government Surveys.\n\n_Notes_ : countries in special situations include least developed countries (lDcs), landlocked developing countries (llDcs), and small island developing States (SIDS). The list of lDcs and SIDSs have changed since 2022 with the graduation of Bhutan from the lDc category and the graduation of Bahrain from the SIDS category; the slight changes in the number of countries assessed and the necessary recalculations have resulted in small variations between the numbers reported here and those reported in the previous Survey.\n\nPrivate consumption Government spending Fixed investment Net exports\n\n0\\.4885\n\n0\\.3598\n\n0\\.3321\n\n0\\.2407\n\n0\\.4076\n\n0\\.4191\n\n(0.4642)\n\n(0.3597)\n\n(0.3308)\n\n(0.2226)\n\n(0.3931)\n\n(0.4054)\n\n0\\.00 0.25 0.50 0.75\n\nLLDCs\n\nLDC/LLDCs\n\nLDCs\n\nLDC/SIDS\n\nSIDS\n\nALL CSS\n\n### OSI\n\n2022 2024 Global average (2024), 0.5754\n\n0\\.5138\n\n0\\.3335\n\n0\\.3547\n\n0\\.3551\n\n0\\.5412\n\n0\\.4884\n\n(0.4847)\n\n(0.3368)\n\n(0.3498)\n\n(0.3498)\n\n(0.5262)\n\n(0.4703)\n\n0\\.00 0.25 0.50 0.75\n\nLLDCs\n\nLDC/LLDCs\n\nLDCs\n\nLDC/SIDS\n\nSIDS\n\nALL CSS\n\n### EGDI\n\n2022 2024 Global average (2024), 0.6382\n\n0\\.5332\n\n0\\.2949\n\n0\\.3539\n\n0\\.3624\n\n0\\.5932\n\n0\\.5206\n\n(0.4121)\n\n(0.2481)\n\n(0.2848)\n\n(0.2925)\n\n(0.4870)\n\n(0.4163)\n\n0\\.00 0.25 0.50 0.75\n\nLLDCs\n\nLDC/LLDCs\n\nLDCs\n\nLDC/SIDS\n\nSIDS\n\nALL CSS\n\n### TII\n\n2022 2024 Global average (2024), 0.6896\n\n0\\.5196\n\n0\\.3458\n\n0\\.3781\n\n0\\.4622\n\n0\\.6229\n\n0\\.5255\n\n(0.5778)\n\n(0.4026)\n\n(0.4337)\n\n(0.5342)\n\n(0.6985)\n\n(0.5893)\n\n0\\.00 0.25 0.50 0.75\n\nLLDCs\n\nLDC/LLDCs\n\nLDCs\n\nLDC/SIDS\n\nSIDS\n\nALL CSS\n\n### HCI\n\n2022 2024 Global average (2024), 0.6496\n\nPrivate consumption Government spending Fixed investment Net exports\n\n0\\.4885\n\n0\\.3598\n\n0\\.3321\n\n0\\.2407\n\n0\\.4076\n\n0\\.4191\n\n(0.4642)\n\n(0.3597)\n\n(0.3308)\n\n(0.2226)\n\n(0.3931)\n\n(0.4054)\n\n0\\.00 0.25 0.50 0.75\n\nLLDCs\n\nLDC/LLDCs\n\nLDCs\n\nLDC/SIDS\n\nSIDS\n\nALL CSS\n\n### OSI\n\n2022 2024 Global average (2024), 0.5754\n\n0\\.5138\n\n0\\.3335\n\n0\\.3547\n\n0\\.3551\n\n0\\.5412\n\n0\\.4884\n\n(0.4847)\n\n(0.3368)\n\n(0.3498)\n\n(0.3498)\n\n(0.5262)\n\n(0.4703)\n\n0\\.00 0.25 0.50 0.75\n\nLLDCs\n\nLDC/LLDCs\n\nLDCs\n\nLDC/SIDS\n\nSIDS\n\nALL CSS\n\n### EGDI\n\n2022 2024 Global average (2024), 0.6382\n\n0\\.5332\n\n0\\.2949\n\n0\\.3539\n\n0\\.3624\n\n0\\.5932\n\n0\\.5206\n\n(0.4121)\n\n(0.2481)\n\n(0.2848)\n\n(0.2925)\n\n(0.4870)\n\n(0.4163)\n\n0\\.00 0.25 0.50 0.75\n\nLLDCs\n\nLDC/LLDCs\n\nLDCs\n\nLDC/SIDS\n\nSIDS\n\nALL CSS\n\n### TII\n\n2022 2024 Global average (2024), 0.6896\n\n0\\.5196\n\n0\\.3458\n\n0\\.3781\n\n0\\.4622\n\n0\\.6229\n\n0\\.5255\n\n(0.5778)\n\n(0.4026)\n\n(0.4337)\n\n(0.5342)\n\n(0.6985)\n\n(0.5893)\n\n0\\.00 0.25 0.50 0.75\n\nLLDCs\n\nLDC/LLDCs\n\nLDCs\n\nLDC/SIDS\n\nSIDS\n\nALL CSS\n\n### HCI\n\n2022 2024 Global average (2024), 0.6496\n\nWith a 6 per cent increase in their average EGDI value, the llDcs have achieved the most significant gains among three groups. This notable improvement reflects the success of these countries in overcoming geographical barriers to strengthen e-government. Such progress contrasts with the stagnant or declining trends observed for lDc/llDcs, highlighting intragroup disparities and the need for targeted interventions.\n\nThe average EGDI value for SIDS has increased by 3 per cent, reflecting steady progress in digital government development. The lDc/SIDS have shown significant improvement, particularly in online services provision; their average OSI value rose by 8 per cent – the highest increase for this component among the countries in special situations. The progress made by this group suggests that access to marine transportation routes and other infrastructure-related advantages play a critical role in advancing the development and provision of digital services.","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"2024 UN E-GovErNmENt SUrvEy\n\n74\n\n**Chapter 2**\n\nFigure 2.32 Average EGDI composite and component values for countries in special situations, 2022 and 2024\n\n_Sources_ : 2022 and 2024 United Nations E-Government Surveys.\n\n_Notes_ : countries in special situations include least developed countries (lDcs), landlocked developing countries (llDcs), and small island developing States (SIDS). The list of lDcs and SIDSs have changed since 2022 with the graduation of Bhutan from the lDc category and the graduation of Bahrain from the SIDS category; the slight changes in the number of countries assessed and the necessary recalculations have resulted in small variations between the numbers reported here and those reported in the previous Survey.\n\nPrivate consumption Government spending Fixed investment Net exports\n\n0\\.4885\n\n0\\.3598\n\n0\\.3321\n\n0\\.2407\n\n0\\.4076\n\n0\\.4191\n\n(0.4642)\n\n(0.3597)\n\n(0.3308)\n\n(0.2226)\n\n(0.3931)\n\n(0.4054)\n\n0\\.00 0.25 0.50 0.75\n\nLLDCs\n\nLDC/LLDCs\n\nLDCs\n\nLDC/SIDS\n\nSIDS\n\nALL CSS\n\n### OSI\n\n2022 2024 Global average (2024), 0.5754\n\n0\\.5138\n\n0\\.3335\n\n0\\.3547\n\n0\\.3551\n\n0\\.5412\n\n0\\.4884\n\n(0.4847)\n\n(0.3368)\n\n(0.3498)\n\n(0.3498)\n\n(0.5262)\n\n(0.4703)\n\n0\\.00 0.25 0.50 0.75\n\nLLDCs\n\nLDC/LLDCs\n\nLDCs\n\nLDC/SIDS\n\nSIDS\n\nALL CSS\n\n### EGDI\n\n2022 2024 Global average (2024), 0.6382\n\n0\\.5332\n\n0\\.2949\n\n0\\.3539\n\n0\\.3624\n\n0\\.5932\n\n0\\.5206\n\n(0.4121)\n\n(0.2481)\n\n(0.2848)\n\n(0.2925)\n\n(0.4870)\n\n(0.4163)\n\n0\\.00 0.25 0.50 0.75\n\nLLDCs\n\nLDC/LLDCs\n\nLDCs\n\nLDC/SIDS\n\nSIDS\n\nALL CSS\n\n### TII\n\n2022 2024 Global average (2024), 0.6896\n\n0\\.5196\n\n0\\.3458\n\n0\\.3781\n\n0\\.4622\n\n0\\.6229\n\n0\\.5255\n\n(0.5778)\n\n(0.4026)\n\n(0.4337)\n\n(0.5342)\n\n(0.6985)\n\n(0.5893)\n\n0\\.00 0.25 0.50 0.75\n\nLLDCs\n\nLDC/LLDCs\n\nLDCs\n\nLDC/SIDS\n\nSIDS\n\nALL CSS\n\n### HCI\n\n2022 2024 Global average (2024), 0.6496\n\nPrivate consumption Government spending Fixed investment Net exports\n\n0\\.4885\n\n0\\.3598\n\n0\\.3321\n\n0\\.2407\n\n0\\.4076\n\n0\\.4191\n\n(0.4642)\n\n(0.3597)\n\n(0.3308)\n\n(0.2226)\n\n(0.3931)\n\n(0.4054)\n\n0\\.00 0.25 0.50 0.75\n\nLLDCs\n\nLDC/LLDCs\n\nLDCs\n\nLDC/SIDS\n\nSIDS\n\nALL CSS\n\n### OSI\n\n2022 2024 Global average (2024), 0.5754\n\n0\\.5138\n\n0\\.3335\n\n0\\.3547\n\n0\\.3551\n\n0\\.5412\n\n0\\.4884\n\n(0.4847)\n\n(0.3368)\n\n(0.3498)\n\n(0.3498)\n\n(0.5262)\n\n(0.4703)\n\n0\\.00 0.25 0.50 0.75\n\nLLDCs\n\nLDC/LLDCs\n\nLDCs\n\nLDC/SIDS\n\nSIDS\n\nALL CSS\n\n### EGDI\n\n2022 2024 Global average (2024), 0.6382\n\n0\\.5332\n\n0\\.2949\n\n0\\.3539\n\n0\\.3624\n\n0\\.5932\n\n0\\.5206\n\n(0.4121)\n\n(0.2481)\n\n(0.2848)\n\n(0.2925)\n\n(0.4870)\n\n(0.4163)\n\n0\\.00 0.25 0.50 0.75\n\nLLDCs\n\nLDC/LLDCs\n\nLDCs\n\nLDC/SIDS\n\nSIDS\n\nALL CSS\n\n### TII\n\n2022 2024 Global average (2024), 0.6896\n\n0\\.5196\n\n0\\.3458\n\n0\\.3781\n\n0\\.4622\n\n0\\.6229\n\n0\\.5255\n\n(0.5778)\n\n(0.4026)\n\n(0.4337)\n\n(0.5342)\n\n(0.6985)\n\n(0.5893)\n\n0\\.00 0.25 0.50 0.75\n\nLLDCs\n\nLDC/LLDCs\n\nLDCs\n\nLDC/SIDS\n\nSIDS\n\nALL CSS\n\n### HCI\n\n2022 2024 Global average (2024), 0.6496\n\nWith a 6 per cent increase in their average EGDI value, the llDcs have achieved the most significant gains among three groups. This notable improvement reflects the success of these countries in overcoming geographical barriers to strengthen e-government. Such progress contrasts with the stagnant or declining trends observed for lDc/llDcs, highlighting intragroup disparities and the need for targeted interventions.\n\nThe average EGDI value for SIDS has increased by 3 per cent, reflecting steady progress in digital government development. The lDc/SIDS have shown significant improvement, particularly in online services provision; their average OSI value rose by 8 per cent – the highest increase for this component among the countries in special situations. The progress made by this group suggests that access to marine transportation routes and other infrastructure-related advantages play a critical role in advancing the development and provision of digital services.","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/%28Web_version%29_E-Government_Survey_2024_1392024_p99.pdf"]},"response":{"errors":[],"extract_id":"extract_fd35a74f6fb1b498407cc26729f32646","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["2024 UN E-GovErNmENt SUrvEy\n\n74\n\n**Chapter 2**\n\nFigure 2.32 Average EGDI composite and component values for countries in special situations, 2022 and 2024\n\n_Sources_ : 2022 and 2024 United Nations E-Government Surveys.\n\n_Notes_ : countries in special situations include least developed countries (lDcs), landlocked developing countries (llDcs), and small island developing States (SIDS). The list of lDcs and SIDSs have changed since 2022 with the graduation of Bhutan from the lDc category and the graduation of Bahrain from the SIDS category; the slight changes in the number of countries assessed and the necessary recalculations have resulted in small variations between the numbers reported here and those reported in the previous Survey.\n\nPrivate consumption Government spending Fixed investment Net exports\n\n0\\.4885\n\n0\\.3598\n\n0\\.3321\n\n0\\.2407\n\n0\\.4076\n\n0\\.4191\n\n(0.4642)\n\n(0.3597)\n\n(0.3308)\n\n(0.2226)\n\n(0.3931)\n\n(0.4054)\n\n0\\.00 0.25 0.50 0.75\n\nLLDCs\n\nLDC/LLDCs\n\nLDCs\n\nLDC/SIDS\n\nSIDS\n\nALL CSS\n\n### OSI\n\n2022 2024 Global average (2024), 0.5754\n\n0\\.5138\n\n0\\.3335\n\n0\\.3547\n\n0\\.3551\n\n0\\.5412\n\n0\\.4884\n\n(0.4847)\n\n(0.3368)\n\n(0.3498)\n\n(0.3498)\n\n(0.5262)\n\n(0.4703)\n\n0\\.00 0.25 0.50 0.75\n\nLLDCs\n\nLDC/LLDCs\n\nLDCs\n\nLDC/SIDS\n\nSIDS\n\nALL CSS\n\n### EGDI\n\n2022 2024 Global average (2024), 0.6382\n\n0\\.5332\n\n0\\.2949\n\n0\\.3539\n\n0\\.3624\n\n0\\.5932\n\n0\\.5206\n\n(0.4121)\n\n(0.2481)\n\n(0.2848)\n\n(0.2925)\n\n(0.4870)\n\n(0.4163)\n\n0\\.00 0.25 0.50 0.75\n\nLLDCs\n\nLDC/LLDCs\n\nLDCs\n\nLDC/SIDS\n\nSIDS\n\nALL CSS\n\n### TII\n\n2022 2024 Global average (2024), 0.6896\n\n0\\.5196\n\n0\\.3458\n\n0\\.3781\n\n0\\.4622\n\n0\\.6229\n\n0\\.5255\n\n(0.5778)\n\n(0.4026)\n\n(0.4337)\n\n(0.5342)\n\n(0.6985)\n\n(0.5893)\n\n0\\.00 0.25 0.50 0.75\n\nLLDCs\n\nLDC/LLDCs\n\nLDCs\n\nLDC/SIDS\n\nSIDS\n\nALL CSS\n\n### HCI\n\n2022 2024 Global average (2024), 0.6496\n\nPrivate consumption Government spending Fixed investment Net exports\n\n0\\.4885\n\n0\\.3598\n\n0\\.3321\n\n0\\.2407\n\n0\\.4076\n\n0\\.4191\n\n(0.4642)\n\n(0.3597)\n\n(0.3308)\n\n(0.2226)\n\n(0.3931)\n\n(0.4054)\n\n0\\.00 0.25 0.50 0.75\n\nLLDCs\n\nLDC/LLDCs\n\nLDCs\n\nLDC/SIDS\n\nSIDS\n\nALL CSS\n\n### OSI\n\n2022 2024 Global average (2024), 0.5754\n\n0\\.5138\n\n0\\.3335\n\n0\\.3547\n\n0\\.3551\n\n0\\.5412\n\n0\\.4884\n\n(0.4847)\n\n(0.3368)\n\n(0.3498)\n\n(0.3498)\n\n(0.5262)\n\n(0.4703)\n\n0\\.00 0.25 0.50 0.75\n\nLLDCs\n\nLDC/LLDCs\n\nLDCs\n\nLDC/SIDS\n\nSIDS\n\nALL CSS\n\n### EGDI\n\n2022 2024 Global average (2024), 0.6382\n\n0\\.5332\n\n0\\.2949\n\n0\\.3539\n\n0\\.3624\n\n0\\.5932\n\n0\\.5206\n\n(0.4121)\n\n(0.2481)\n\n(0.2848)\n\n(0.2925)\n\n(0.4870)\n\n(0.4163)\n\n0\\.00 0.25 0.50 0.75\n\nLLDCs\n\nLDC/LLDCs\n\nLDCs\n\nLDC/SIDS\n\nSIDS\n\nALL CSS\n\n### TII\n\n2022 2024 Global average (2024), 0.6896\n\n0\\.5196\n\n0\\.3458\n\n0\\.3781\n\n0\\.4622\n\n0\\.6229\n\n0\\.5255\n\n(0.5778)\n\n(0.4026)\n\n(0.4337)\n\n(0.5342)\n\n(0.6985)\n\n(0.5893)\n\n0\\.00 0.25 0.50 0.75\n\nLLDCs\n\nLDC/LLDCs\n\nLDCs\n\nLDC/SIDS\n\nSIDS\n\nALL CSS\n\n### HCI\n\n2022 2024 Global average (2024), 0.6496\n\nWith a 6 per cent increase in their average EGDI value, the llDcs have achieved the most significant gains among three groups. This notable improvement reflects the success of these countries in overcoming geographical barriers to strengthen e-government. Such progress contrasts with the stagnant or declining trends observed for lDc/llDcs, highlighting intragroup disparities and the need for targeted interventions.\n\nThe average EGDI value for SIDS has increased by 3 per cent, reflecting steady progress in digital government development. The lDc/SIDS have shown significant improvement, particularly in online services provision; their average OSI value rose by 8 per cent – the highest increase for this component among the countries in special situations. The progress made by this group suggests that access to marine transportation routes and other infrastructure-related advantages play a critical role in advancing the development and provision of digital services."],"full_content":"2024 UN E-GovErNmENt SUrvEy\n\n74\n\n**Chapter 2**\n\nFigure 2.32 Average EGDI composite and component values for countries in special situations, 2022 and 2024\n\n_Sources_ : 2022 and 2024 United Nations E-Government Surveys.\n\n_Notes_ : countries in special situations include least developed countries (lDcs), landlocked developing countries (llDcs), and small island developing States (SIDS). The list of lDcs and SIDSs have changed since 2022 with the graduation of Bhutan from the lDc category and the graduation of Bahrain from the SIDS category; the slight changes in the number of countries assessed and the necessary recalculations have resulted in small variations between the numbers reported here and those reported in the previous Survey.\n\nPrivate consumption Government spending Fixed investment Net exports\n\n0\\.4885\n\n0\\.3598\n\n0\\.3321\n\n0\\.2407\n\n0\\.4076\n\n0\\.4191\n\n(0.4642)\n\n(0.3597)\n\n(0.3308)\n\n(0.2226)\n\n(0.3931)\n\n(0.4054)\n\n0\\.00 0.25 0.50 0.75\n\nLLDCs\n\nLDC/LLDCs\n\nLDCs\n\nLDC/SIDS\n\nSIDS\n\nALL CSS\n\n### OSI\n\n2022 2024 Global average (2024), 0.5754\n\n0\\.5138\n\n0\\.3335\n\n0\\.3547\n\n0\\.3551\n\n0\\.5412\n\n0\\.4884\n\n(0.4847)\n\n(0.3368)\n\n(0.3498)\n\n(0.3498)\n\n(0.5262)\n\n(0.4703)\n\n0\\.00 0.25 0.50 0.75\n\nLLDCs\n\nLDC/LLDCs\n\nLDCs\n\nLDC/SIDS\n\nSIDS\n\nALL CSS\n\n### EGDI\n\n2022 2024 Global average (2024), 0.6382\n\n0\\.5332\n\n0\\.2949\n\n0\\.3539\n\n0\\.3624\n\n0\\.5932\n\n0\\.5206\n\n(0.4121)\n\n(0.2481)\n\n(0.2848)\n\n(0.2925)\n\n(0.4870)\n\n(0.4163)\n\n0\\.00 0.25 0.50 0.75\n\nLLDCs\n\nLDC/LLDCs\n\nLDCs\n\nLDC/SIDS\n\nSIDS\n\nALL CSS\n\n### TII\n\n2022 2024 Global average (2024), 0.6896\n\n0\\.5196\n\n0\\.3458\n\n0\\.3781\n\n0\\.4622\n\n0\\.6229\n\n0\\.5255\n\n(0.5778)\n\n(0.4026)\n\n(0.4337)\n\n(0.5342)\n\n(0.6985)\n\n(0.5893)\n\n0\\.00 0.25 0.50 0.75\n\nLLDCs\n\nLDC/LLDCs\n\nLDCs\n\nLDC/SIDS\n\nSIDS\n\nALL CSS\n\n### HCI\n\n2022 2024 Global average (2024), 0.6496\n\nPrivate consumption Government spending Fixed investment Net exports\n\n0\\.4885\n\n0\\.3598\n\n0\\.3321\n\n0\\.2407\n\n0\\.4076\n\n0\\.4191\n\n(0.4642)\n\n(0.3597)\n\n(0.3308)\n\n(0.2226)\n\n(0.3931)\n\n(0.4054)\n\n0\\.00 0.25 0.50 0.75\n\nLLDCs\n\nLDC/LLDCs\n\nLDCs\n\nLDC/SIDS\n\nSIDS\n\nALL CSS\n\n### OSI\n\n2022 2024 Global average (2024), 0.5754\n\n0\\.5138\n\n0\\.3335\n\n0\\.3547\n\n0\\.3551\n\n0\\.5412\n\n0\\.4884\n\n(0.4847)\n\n(0.3368)\n\n(0.3498)\n\n(0.3498)\n\n(0.5262)\n\n(0.4703)\n\n0\\.00 0.25 0.50 0.75\n\nLLDCs\n\nLDC/LLDCs\n\nLDCs\n\nLDC/SIDS\n\nSIDS\n\nALL CSS\n\n### EGDI\n\n2022 2024 Global average (2024), 0.6382\n\n0\\.5332\n\n0\\.2949\n\n0\\.3539\n\n0\\.3624\n\n0\\.5932\n\n0\\.5206\n\n(0.4121)\n\n(0.2481)\n\n(0.2848)\n\n(0.2925)\n\n(0.4870)\n\n(0.4163)\n\n0\\.00 0.25 0.50 0.75\n\nLLDCs\n\nLDC/LLDCs\n\nLDCs\n\nLDC/SIDS\n\nSIDS\n\nALL CSS\n\n### TII\n\n2022 2024 Global average (2024), 0.6896\n\n0\\.5196\n\n0\\.3458\n\n0\\.3781\n\n0\\.4622\n\n0\\.6229\n\n0\\.5255\n\n(0.5778)\n\n(0.4026)\n\n(0.4337)\n\n(0.5342)\n\n(0.6985)\n\n(0.5893)\n\n0\\.00 0.25 0.50 0.75\n\nLLDCs\n\nLDC/LLDCs\n\nLDCs\n\nLDC/SIDS\n\nSIDS\n\nALL CSS\n\n### HCI\n\n2022 2024 Global average (2024), 0.6496\n\nWith a 6 per cent increase in their average EGDI value, the llDcs have achieved the most significant gains among three groups. This notable improvement reflects the success of these countries in overcoming geographical barriers to strengthen e-government. Such progress contrasts with the stagnant or declining trends observed for lDc/llDcs, highlighting intragroup disparities and the need for targeted interventions.\n\nThe average EGDI value for SIDS has increased by 3 per cent, reflecting steady progress in digital government development. The lDc/SIDS have shown significant improvement, particularly in online services provision; their average OSI value rose by 8 per cent – the highest increase for this component among the countries in special situations. The progress made by this group suggests that access to marine transportation routes and other infrastructure-related advantages play a critical role in advancing the development and provision of digital services.","publish_date":null,"title":"EGDI EGDI OSI OSI TII TII HCI HCI","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/%28Web_version%29_E-Government_Survey_2024_1392024_p99.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/%28Web_version%29_E-Government_Survey_2024_1392024_p99.pdf"},"request":{"example_id":"chart/(Web_version)_E-Government_Survey_2024_1392024_p99","product_type":"parse","source_file_path":"(Web_version)_E-Government_Survey_2024_1392024_p99.pdf"},"started_at":"2026-11-04T19:46:16.446567"} +{"completed_at":"2026-11-04T19:31:05.592546","latency_in_ms":2596,"output":{"example_id":"chart/05021ff2-en_p19","grounded_pages":[],"layout_pages":[],"markdown":" **17**\n\nA DECADE OF OECD COMPETITION TRENDS, DATA AND INSIGHTS © OECD 2025\n\n### **Figure 2.2. Total merger notifications (left) and the average annual growth rate (right) for OECD and Non-OECD jurisdictions, 2015-2024**\n\nNote: Data based on the 59 jurisdictions in the OECD CompStats database that provided data for each of the variables for the period 2015- 2024. Source: OECD CompStats database\n\nGlobal trends in mergers and acquisitions in the past ten years have been impacted by macroeconomic dynamics and the correlation between merger notifications and variables such as GDP growth, inflation and interest rates is strong. Periods of strong GDP growth tend to spur deals, while periods of recession or slower growth dampen them (see Figure 2.3). The most impactful shock in the last decade was the COVID-19 pandemic, where OECD jurisdictions observed a decrease of -4.75% in their GDP (and OECD estimates for the decrease in the World’s GDP were around 3% (OECD, 2020[14]; 2021[15]) that correlated with a drop in M&A activity in 2020 compared to 2019. This represented on average 10.3% fewer notifications, while the post pandemic rebound saw record deal volumes in 2021. This initial decline was observed in 70% of the jurisdictions in the sample, with Latin American jurisdictions being the ones that observed the biggest drop – as notifications fell more than 40% in Barbados, Ecuador, Panama and Chile. The subsequent recovery, which on average increased notifications by 40% in 2021 compared to 2020, was observed in 80% of those jurisdictions. In Latvia, the United States, Malta, the Slovak Republic and Greece, the increase was higher than 100% with respect to notifications in 2020.\n\nAfter the two years with significant changes, 2022 presented a mixed picture. In OECD jurisdictions notifications decreased, although they remained higher than the pre COVID-19 period, while in non-OECD jurisdictions they increased further. A sudden increase in the number of mergers notified in 2021, which directly impact merger reviews, may have impacted the resources that authorities had to dedicate to merger reviews. ⁴ More mergers to assess require more time and, often, a more flexible allocation of the staff from other enforcement areas. Challenges on resource allocation, case prioritisation and the interaction between increased activity in one of the areas of work and the resources dedicated to others have raised concerns from competition authorities recently. ⁵\n\n0\n\n2 000\n\n4 000\n\n6 000\n\n8 000\n\n10 000\n\n12 000\n\n14 000\n\n2015 2016 2017 2018 2019 2020 2021 2022 2023 2024\n\nOECD Non-OECD\n\nTotal number of merger notifications\n\n\\-30% -20% -10% 0% 10% 20% 30% 40% 50%\n\n2015/2016\n\n2016/2017\n\n2017/2018\n\n2018/2019\n\n2019/2020\n\n2020/2021\n\n2021/2022\n\n2022/2023\n\n2023/2024\n\nAnnual growth rate OECD Non-OECD","pages":[{"line_numbers":[],"markdown":" **17**\n\nA DECADE OF OECD COMPETITION TRENDS, DATA AND INSIGHTS © OECD 2025\n\n### **Figure 2.2. Total merger notifications (left) and the average annual growth rate (right) for OECD and Non-OECD jurisdictions, 2015-2024**\n\nNote: Data based on the 59 jurisdictions in the OECD CompStats database that provided data for each of the variables for the period 2015- 2024. Source: OECD CompStats database\n\nGlobal trends in mergers and acquisitions in the past ten years have been impacted by macroeconomic dynamics and the correlation between merger notifications and variables such as GDP growth, inflation and interest rates is strong. Periods of strong GDP growth tend to spur deals, while periods of recession or slower growth dampen them (see Figure 2.3). The most impactful shock in the last decade was the COVID-19 pandemic, where OECD jurisdictions observed a decrease of -4.75% in their GDP (and OECD estimates for the decrease in the World’s GDP were around 3% (OECD, 2020[14]; 2021[15]) that correlated with a drop in M&A activity in 2020 compared to 2019. This represented on average 10.3% fewer notifications, while the post pandemic rebound saw record deal volumes in 2021. This initial decline was observed in 70% of the jurisdictions in the sample, with Latin American jurisdictions being the ones that observed the biggest drop – as notifications fell more than 40% in Barbados, Ecuador, Panama and Chile. The subsequent recovery, which on average increased notifications by 40% in 2021 compared to 2020, was observed in 80% of those jurisdictions. In Latvia, the United States, Malta, the Slovak Republic and Greece, the increase was higher than 100% with respect to notifications in 2020.\n\nAfter the two years with significant changes, 2022 presented a mixed picture. In OECD jurisdictions notifications decreased, although they remained higher than the pre COVID-19 period, while in non-OECD jurisdictions they increased further. A sudden increase in the number of mergers notified in 2021, which directly impact merger reviews, may have impacted the resources that authorities had to dedicate to merger reviews. ⁴ More mergers to assess require more time and, often, a more flexible allocation of the staff from other enforcement areas. Challenges on resource allocation, case prioritisation and the interaction between increased activity in one of the areas of work and the resources dedicated to others have raised concerns from competition authorities recently. ⁵\n\n0\n\n2 000\n\n4 000\n\n6 000\n\n8 000\n\n10 000\n\n12 000\n\n14 000\n\n2015 2016 2017 2018 2019 2020 2021 2022 2023 2024\n\nOECD Non-OECD\n\nTotal number of merger notifications\n\n\\-30% -20% -10% 0% 10% 20% 30% 40% 50%\n\n2015/2016\n\n2016/2017\n\n2017/2018\n\n2018/2019\n\n2019/2020\n\n2020/2021\n\n2021/2022\n\n2022/2023\n\n2023/2024\n\nAnnual growth rate OECD Non-OECD","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":" **17**\n\nA DECADE OF OECD COMPETITION TRENDS, DATA AND INSIGHTS © OECD 2025\n\n### **Figure 2.2. Total merger notifications (left) and the average annual growth rate (right) for OECD and Non-OECD jurisdictions, 2015-2024**\n\nNote: Data based on the 59 jurisdictions in the OECD CompStats database that provided data for each of the variables for the period 2015- 2024. Source: OECD CompStats database\n\nGlobal trends in mergers and acquisitions in the past ten years have been impacted by macroeconomic dynamics and the correlation between merger notifications and variables such as GDP growth, inflation and interest rates is strong. Periods of strong GDP growth tend to spur deals, while periods of recession or slower growth dampen them (see Figure 2.3). The most impactful shock in the last decade was the COVID-19 pandemic, where OECD jurisdictions observed a decrease of -4.75% in their GDP (and OECD estimates for the decrease in the World’s GDP were around 3% (OECD, 2020[14]; 2021[15]) that correlated with a drop in M&A activity in 2020 compared to 2019. This represented on average 10.3% fewer notifications, while the post pandemic rebound saw record deal volumes in 2021. This initial decline was observed in 70% of the jurisdictions in the sample, with Latin American jurisdictions being the ones that observed the biggest drop – as notifications fell more than 40% in Barbados, Ecuador, Panama and Chile. The subsequent recovery, which on average increased notifications by 40% in 2021 compared to 2020, was observed in 80% of those jurisdictions. In Latvia, the United States, Malta, the Slovak Republic and Greece, the increase was higher than 100% with respect to notifications in 2020.\n\nAfter the two years with significant changes, 2022 presented a mixed picture. In OECD jurisdictions notifications decreased, although they remained higher than the pre COVID-19 period, while in non-OECD jurisdictions they increased further. A sudden increase in the number of mergers notified in 2021, which directly impact merger reviews, may have impacted the resources that authorities had to dedicate to merger reviews. ⁴ More mergers to assess require more time and, often, a more flexible allocation of the staff from other enforcement areas. Challenges on resource allocation, case prioritisation and the interaction between increased activity in one of the areas of work and the resources dedicated to others have raised concerns from competition authorities recently. ⁵\n\n0\n\n2 000\n\n4 000\n\n6 000\n\n8 000\n\n10 000\n\n12 000\n\n14 000\n\n2015 2016 2017 2018 2019 2020 2021 2022 2023 2024\n\nOECD Non-OECD\n\nTotal number of merger notifications\n\n\\-30% -20% -10% 0% 10% 20% 30% 40% 50%\n\n2015/2016\n\n2016/2017\n\n2017/2018\n\n2018/2019\n\n2019/2020\n\n2020/2021\n\n2021/2022\n\n2022/2023\n\n2023/2024\n\nAnnual growth rate OECD Non-OECD","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/05021ff2-en_p19.pdf"]},"response":{"errors":[],"extract_id":"extract_31dfe08226f1db1ea0a420cfa9fd3387","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":[" **17**\n\nA DECADE OF OECD COMPETITION TRENDS, DATA AND INSIGHTS © OECD 2025\n\n### **Figure 2.2. Total merger notifications (left) and the average annual growth rate (right) for OECD and Non-OECD jurisdictions, 2015-2024**\n\nNote: Data based on the 59 jurisdictions in the OECD CompStats database that provided data for each of the variables for the period 2015- 2024. Source: OECD CompStats database\n\nGlobal trends in mergers and acquisitions in the past ten years have been impacted by macroeconomic dynamics and the correlation between merger notifications and variables such as GDP growth, inflation and interest rates is strong. Periods of strong GDP growth tend to spur deals, while periods of recession or slower growth dampen them (see Figure 2.3). The most impactful shock in the last decade was the COVID-19 pandemic, where OECD jurisdictions observed a decrease of -4.75% in their GDP (and OECD estimates for the decrease in the World’s GDP were around 3% (OECD, 2020[14]; 2021[15]) that correlated with a drop in M&A activity in 2020 compared to 2019. This represented on average 10.3% fewer notifications, while the post pandemic rebound saw record deal volumes in 2021. This initial decline was observed in 70% of the jurisdictions in the sample, with Latin American jurisdictions being the ones that observed the biggest drop – as notifications fell more than 40% in Barbados, Ecuador, Panama and Chile. The subsequent recovery, which on average increased notifications by 40% in 2021 compared to 2020, was observed in 80% of those jurisdictions. In Latvia, the United States, Malta, the Slovak Republic and Greece, the increase was higher than 100% with respect to notifications in 2020.\n\nAfter the two years with significant changes, 2022 presented a mixed picture. In OECD jurisdictions notifications decreased, although they remained higher than the pre COVID-19 period, while in non-OECD jurisdictions they increased further. A sudden increase in the number of mergers notified in 2021, which directly impact merger reviews, may have impacted the resources that authorities had to dedicate to merger reviews. ⁴ More mergers to assess require more time and, often, a more flexible allocation of the staff from other enforcement areas. Challenges on resource allocation, case prioritisation and the interaction between increased activity in one of the areas of work and the resources dedicated to others have raised concerns from competition authorities recently. ⁵\n\n0\n\n2 000\n\n4 000\n\n6 000\n\n8 000\n\n10 000\n\n12 000\n\n14 000\n\n2015 2016 2017 2018 2019 2020 2021 2022 2023 2024\n\nOECD Non-OECD\n\nTotal number of merger notifications\n\n\\-30% -20% -10% 0% 10% 20% 30% 40% 50%\n\n2015/2016\n\n2016/2017\n\n2017/2018\n\n2018/2019\n\n2019/2020\n\n2020/2021\n\n2021/2022\n\n2022/2023\n\n2023/2024\n\nAnnual growth rate OECD Non-OECD"],"full_content":" **17**\n\nA DECADE OF OECD COMPETITION TRENDS, DATA AND INSIGHTS © OECD 2025\n\n### **Figure 2.2. Total merger notifications (left) and the average annual growth rate (right) for OECD and Non-OECD jurisdictions, 2015-2024**\n\nNote: Data based on the 59 jurisdictions in the OECD CompStats database that provided data for each of the variables for the period 2015- 2024. Source: OECD CompStats database\n\nGlobal trends in mergers and acquisitions in the past ten years have been impacted by macroeconomic dynamics and the correlation between merger notifications and variables such as GDP growth, inflation and interest rates is strong. Periods of strong GDP growth tend to spur deals, while periods of recession or slower growth dampen them (see Figure 2.3). The most impactful shock in the last decade was the COVID-19 pandemic, where OECD jurisdictions observed a decrease of -4.75% in their GDP (and OECD estimates for the decrease in the World’s GDP were around 3% (OECD, 2020[14]; 2021[15]) that correlated with a drop in M&A activity in 2020 compared to 2019. This represented on average 10.3% fewer notifications, while the post pandemic rebound saw record deal volumes in 2021. This initial decline was observed in 70% of the jurisdictions in the sample, with Latin American jurisdictions being the ones that observed the biggest drop – as notifications fell more than 40% in Barbados, Ecuador, Panama and Chile. The subsequent recovery, which on average increased notifications by 40% in 2021 compared to 2020, was observed in 80% of those jurisdictions. In Latvia, the United States, Malta, the Slovak Republic and Greece, the increase was higher than 100% with respect to notifications in 2020.\n\nAfter the two years with significant changes, 2022 presented a mixed picture. In OECD jurisdictions notifications decreased, although they remained higher than the pre COVID-19 period, while in non-OECD jurisdictions they increased further. A sudden increase in the number of mergers notified in 2021, which directly impact merger reviews, may have impacted the resources that authorities had to dedicate to merger reviews. ⁴ More mergers to assess require more time and, often, a more flexible allocation of the staff from other enforcement areas. Challenges on resource allocation, case prioritisation and the interaction between increased activity in one of the areas of work and the resources dedicated to others have raised concerns from competition authorities recently. ⁵\n\n0\n\n2 000\n\n4 000\n\n6 000\n\n8 000\n\n10 000\n\n12 000\n\n14 000\n\n2015 2016 2017 2018 2019 2020 2021 2022 2023 2024\n\nOECD Non-OECD\n\nTotal number of merger notifications\n\n\\-30% -20% -10% 0% 10% 20% 30% 40% 50%\n\n2015/2016\n\n2016/2017\n\n2017/2018\n\n2018/2019\n\n2019/2020\n\n2020/2021\n\n2021/2022\n\n2022/2023\n\n2023/2024\n\nAnnual growth rate OECD Non-OECD","publish_date":null,"title":"Figure 2.2. Total merger notifications (left) and the average annual growth rate (right) for OECD and Non-OECD jurisdictions, 2015-2024","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/05021ff2-en_p19.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/05021ff2-en_p19.pdf"},"request":{"example_id":"chart/05021ff2-en_p19","product_type":"parse","source_file_path":"05021ff2-en_p19.pdf"},"started_at":"2026-11-04T19:31:02.996483"} +{"completed_at":"2026-11-04T19:46:22.792199","latency_in_ms":2201,"output":{"example_id":"chart/05021ff2-en_p20","grounded_pages":[],"layout_pages":[],"markdown":"**18** \n\nA DECADE OF OECD COMPETITION TRENDS, DATA AND INSIGHTS © OECD 2025\n\n### **Figure 2.3. Annual growth rates of CompStats jurisdictions** **’ GDP (in** **PPP) and total merger notifications, 2015-2024**\n\nNote: A three-year average is used to smooth out volatility caused by random fluctuations. Data based on the 59 jurisdictions in the OECD CompStats database that provided data for merger notifications for the period 2015-2024. Source: OECD CompStats database and World Bank\n\nRecent literature suggests that lower interest rates stimulate transactions by reducing borrowing costs and that high inflation, particularly when prolonged, disincentivises acquisitions as profit margins are reduced (Lakra, 2024[16]; Chiriac, 2021[17]). In line with these findings, moderate inflation and low interest rates may have also contributed to the expansion of M&A activities observed both between 2015 and 2019 and in 2021. However, the interest rate hikes since 2022, which reduced liquidity and made debt more expensive, may have contributed to the decline of activity observed in 2022 and 2023. Together with more elevated inflation rates, slower global economic activity, increased macroeconomic uncertainty and heightened geopolitical tensions, negative growth rates were observed for mergers between 2022 and 2024.\n\nTrade policies and geopolitical dynamics also have an impact on deal flows. Conflicts, tensions and even elections generate wait-and-see dynamics where firms pause deals amid tariff and regulatory swings (OECD, 2025[18]). Should geopolitical tensions ease and interest rates fall, the world could see a new wave of merger activity, pushing competition authorities to gear up for a potential flood of filings.\n\n### **2\\.2. Competition authorities have maintained stable merger intervention levels**\n\nThe majority of mergers assessed by competition authorities do not have a significant impact on competition. This is reflected in the percentage of mergers being cleared without any remedy (94.6% on average, per jurisdiction, per year). However, despite increases in the levels of mergers notified in the past decade, competition authorities have maintained relatively stable intervention levels, which include conditionally approving mergers, challenging or prohibiting them (according to the applicable legal regime). The average intervention rate during the decade was 5.39% with a peak of 6.87% in 2017 and values for the beginning and the end of the decade close to 4.8% (see Figure 2.4). When looking at jurisdiction-level intervention rates and their evolution during the decade, it is possible to see that in 23 jurisdictions the intervention rate showed an increase while in the other 25 it decreased. Six jurisdictions observed stable levels of intervention, as in the aggregate case (see Figure A B.3).\n\n\\-30%\n\n\\-20%\n\n\\-10%\n\n0%\n\n10%\n\n20%\n\n30%\n\n40%\n\n50%\n\nGrowth rate\n\nGDP in PPP Merger notifications","pages":[{"line_numbers":[],"markdown":"**18** \n\nA DECADE OF OECD COMPETITION TRENDS, DATA AND INSIGHTS © OECD 2025\n\n### **Figure 2.3. Annual growth rates of CompStats jurisdictions** **’ GDP (in** **PPP) and total merger notifications, 2015-2024**\n\nNote: A three-year average is used to smooth out volatility caused by random fluctuations. Data based on the 59 jurisdictions in the OECD CompStats database that provided data for merger notifications for the period 2015-2024. Source: OECD CompStats database and World Bank\n\nRecent literature suggests that lower interest rates stimulate transactions by reducing borrowing costs and that high inflation, particularly when prolonged, disincentivises acquisitions as profit margins are reduced (Lakra, 2024[16]; Chiriac, 2021[17]). In line with these findings, moderate inflation and low interest rates may have also contributed to the expansion of M&A activities observed both between 2015 and 2019 and in 2021. However, the interest rate hikes since 2022, which reduced liquidity and made debt more expensive, may have contributed to the decline of activity observed in 2022 and 2023. Together with more elevated inflation rates, slower global economic activity, increased macroeconomic uncertainty and heightened geopolitical tensions, negative growth rates were observed for mergers between 2022 and 2024.\n\nTrade policies and geopolitical dynamics also have an impact on deal flows. Conflicts, tensions and even elections generate wait-and-see dynamics where firms pause deals amid tariff and regulatory swings (OECD, 2025[18]). Should geopolitical tensions ease and interest rates fall, the world could see a new wave of merger activity, pushing competition authorities to gear up for a potential flood of filings.\n\n### **2\\.2. Competition authorities have maintained stable merger intervention levels**\n\nThe majority of mergers assessed by competition authorities do not have a significant impact on competition. This is reflected in the percentage of mergers being cleared without any remedy (94.6% on average, per jurisdiction, per year). However, despite increases in the levels of mergers notified in the past decade, competition authorities have maintained relatively stable intervention levels, which include conditionally approving mergers, challenging or prohibiting them (according to the applicable legal regime). The average intervention rate during the decade was 5.39% with a peak of 6.87% in 2017 and values for the beginning and the end of the decade close to 4.8% (see Figure 2.4). When looking at jurisdiction-level intervention rates and their evolution during the decade, it is possible to see that in 23 jurisdictions the intervention rate showed an increase while in the other 25 it decreased. Six jurisdictions observed stable levels of intervention, as in the aggregate case (see Figure A B.3).\n\n\\-30%\n\n\\-20%\n\n\\-10%\n\n0%\n\n10%\n\n20%\n\n30%\n\n40%\n\n50%\n\nGrowth rate\n\nGDP in PPP Merger notifications","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**18** \n\nA DECADE OF OECD COMPETITION TRENDS, DATA AND INSIGHTS © OECD 2025\n\n### **Figure 2.3. Annual growth rates of CompStats jurisdictions** **’ GDP (in** **PPP) and total merger notifications, 2015-2024**\n\nNote: A three-year average is used to smooth out volatility caused by random fluctuations. Data based on the 59 jurisdictions in the OECD CompStats database that provided data for merger notifications for the period 2015-2024. Source: OECD CompStats database and World Bank\n\nRecent literature suggests that lower interest rates stimulate transactions by reducing borrowing costs and that high inflation, particularly when prolonged, disincentivises acquisitions as profit margins are reduced (Lakra, 2024[16]; Chiriac, 2021[17]). In line with these findings, moderate inflation and low interest rates may have also contributed to the expansion of M&A activities observed both between 2015 and 2019 and in 2021. However, the interest rate hikes since 2022, which reduced liquidity and made debt more expensive, may have contributed to the decline of activity observed in 2022 and 2023. Together with more elevated inflation rates, slower global economic activity, increased macroeconomic uncertainty and heightened geopolitical tensions, negative growth rates were observed for mergers between 2022 and 2024.\n\nTrade policies and geopolitical dynamics also have an impact on deal flows. Conflicts, tensions and even elections generate wait-and-see dynamics where firms pause deals amid tariff and regulatory swings (OECD, 2025[18]). Should geopolitical tensions ease and interest rates fall, the world could see a new wave of merger activity, pushing competition authorities to gear up for a potential flood of filings.\n\n### **2\\.2. Competition authorities have maintained stable merger intervention levels**\n\nThe majority of mergers assessed by competition authorities do not have a significant impact on competition. This is reflected in the percentage of mergers being cleared without any remedy (94.6% on average, per jurisdiction, per year). However, despite increases in the levels of mergers notified in the past decade, competition authorities have maintained relatively stable intervention levels, which include conditionally approving mergers, challenging or prohibiting them (according to the applicable legal regime). The average intervention rate during the decade was 5.39% with a peak of 6.87% in 2017 and values for the beginning and the end of the decade close to 4.8% (see Figure 2.4). When looking at jurisdiction-level intervention rates and their evolution during the decade, it is possible to see that in 23 jurisdictions the intervention rate showed an increase while in the other 25 it decreased. Six jurisdictions observed stable levels of intervention, as in the aggregate case (see Figure A B.3).\n\n\\-30%\n\n\\-20%\n\n\\-10%\n\n0%\n\n10%\n\n20%\n\n30%\n\n40%\n\n50%\n\nGrowth rate\n\nGDP in PPP Merger notifications","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/05021ff2-en_p20.pdf"]},"response":{"errors":[],"extract_id":"extract_8040184d786df750ac5e3fd9db03a2c6","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**18** \n\nA DECADE OF OECD COMPETITION TRENDS, DATA AND INSIGHTS © OECD 2025\n\n### **Figure 2.3. Annual growth rates of CompStats jurisdictions** **’ GDP (in** **PPP) and total merger notifications, 2015-2024**\n\nNote: A three-year average is used to smooth out volatility caused by random fluctuations. Data based on the 59 jurisdictions in the OECD CompStats database that provided data for merger notifications for the period 2015-2024. Source: OECD CompStats database and World Bank\n\nRecent literature suggests that lower interest rates stimulate transactions by reducing borrowing costs and that high inflation, particularly when prolonged, disincentivises acquisitions as profit margins are reduced (Lakra, 2024[16]; Chiriac, 2021[17]). In line with these findings, moderate inflation and low interest rates may have also contributed to the expansion of M&A activities observed both between 2015 and 2019 and in 2021. However, the interest rate hikes since 2022, which reduced liquidity and made debt more expensive, may have contributed to the decline of activity observed in 2022 and 2023. Together with more elevated inflation rates, slower global economic activity, increased macroeconomic uncertainty and heightened geopolitical tensions, negative growth rates were observed for mergers between 2022 and 2024.\n\nTrade policies and geopolitical dynamics also have an impact on deal flows. Conflicts, tensions and even elections generate wait-and-see dynamics where firms pause deals amid tariff and regulatory swings (OECD, 2025[18]). Should geopolitical tensions ease and interest rates fall, the world could see a new wave of merger activity, pushing competition authorities to gear up for a potential flood of filings.\n\n### **2\\.2. Competition authorities have maintained stable merger intervention levels**\n\nThe majority of mergers assessed by competition authorities do not have a significant impact on competition. This is reflected in the percentage of mergers being cleared without any remedy (94.6% on average, per jurisdiction, per year). However, despite increases in the levels of mergers notified in the past decade, competition authorities have maintained relatively stable intervention levels, which include conditionally approving mergers, challenging or prohibiting them (according to the applicable legal regime). The average intervention rate during the decade was 5.39% with a peak of 6.87% in 2017 and values for the beginning and the end of the decade close to 4.8% (see Figure 2.4). When looking at jurisdiction-level intervention rates and their evolution during the decade, it is possible to see that in 23 jurisdictions the intervention rate showed an increase while in the other 25 it decreased. Six jurisdictions observed stable levels of intervention, as in the aggregate case (see Figure A B.3).\n\n\\-30%\n\n\\-20%\n\n\\-10%\n\n0%\n\n10%\n\n20%\n\n30%\n\n40%\n\n50%\n\nGrowth rate\n\nGDP in PPP Merger notifications"],"full_content":"**18** \n\nA DECADE OF OECD COMPETITION TRENDS, DATA AND INSIGHTS © OECD 2025\n\n### **Figure 2.3. Annual growth rates of CompStats jurisdictions** **’ GDP (in** **PPP) and total merger notifications, 2015-2024**\n\nNote: A three-year average is used to smooth out volatility caused by random fluctuations. Data based on the 59 jurisdictions in the OECD CompStats database that provided data for merger notifications for the period 2015-2024. Source: OECD CompStats database and World Bank\n\nRecent literature suggests that lower interest rates stimulate transactions by reducing borrowing costs and that high inflation, particularly when prolonged, disincentivises acquisitions as profit margins are reduced (Lakra, 2024[16]; Chiriac, 2021[17]). In line with these findings, moderate inflation and low interest rates may have also contributed to the expansion of M&A activities observed both between 2015 and 2019 and in 2021. However, the interest rate hikes since 2022, which reduced liquidity and made debt more expensive, may have contributed to the decline of activity observed in 2022 and 2023. Together with more elevated inflation rates, slower global economic activity, increased macroeconomic uncertainty and heightened geopolitical tensions, negative growth rates were observed for mergers between 2022 and 2024.\n\nTrade policies and geopolitical dynamics also have an impact on deal flows. Conflicts, tensions and even elections generate wait-and-see dynamics where firms pause deals amid tariff and regulatory swings (OECD, 2025[18]). Should geopolitical tensions ease and interest rates fall, the world could see a new wave of merger activity, pushing competition authorities to gear up for a potential flood of filings.\n\n### **2\\.2. Competition authorities have maintained stable merger intervention levels**\n\nThe majority of mergers assessed by competition authorities do not have a significant impact on competition. This is reflected in the percentage of mergers being cleared without any remedy (94.6% on average, per jurisdiction, per year). However, despite increases in the levels of mergers notified in the past decade, competition authorities have maintained relatively stable intervention levels, which include conditionally approving mergers, challenging or prohibiting them (according to the applicable legal regime). The average intervention rate during the decade was 5.39% with a peak of 6.87% in 2017 and values for the beginning and the end of the decade close to 4.8% (see Figure 2.4). When looking at jurisdiction-level intervention rates and their evolution during the decade, it is possible to see that in 23 jurisdictions the intervention rate showed an increase while in the other 25 it decreased. Six jurisdictions observed stable levels of intervention, as in the aggregate case (see Figure A B.3).\n\n\\-30%\n\n\\-20%\n\n\\-10%\n\n0%\n\n10%\n\n20%\n\n30%\n\n40%\n\n50%\n\nGrowth rate\n\nGDP in PPP Merger notifications","publish_date":null,"title":"Figure 2.3. Annual growth rates of CompStats jurisdictions’ GDP (in PPP) and total merger notifications, 2015-2024 2.2. Competition authorities have","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/05021ff2-en_p20.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/05021ff2-en_p20.pdf"},"request":{"example_id":"chart/05021ff2-en_p20","product_type":"parse","source_file_path":"05021ff2-en_p20.pdf"},"started_at":"2026-11-04T19:46:20.590822"} +{"completed_at":"2026-11-04T19:31:08.954433","latency_in_ms":1619,"output":{"example_id":"chart/05021ff2-en_p21","grounded_pages":[],"layout_pages":[],"markdown":" **19**\n\nA DECADE OF OECD COMPETITION TRENDS, DATA AND INSIGHTS © OECD 2025\n\n### **Figure 2.4. Average merger intervention rate, 2015-2024**\n\nNote: The intervention rate is the share of mergers approved with remedies, prohibited or challenged over total merger decisions. A three-year average is used to smooth out volatility caused by random fluctuations. Data based on the 54 jurisdictions in the OECD CompStats database that provided data for merger remedies and merger prohibitions or challenges for the period 2015-2024. Source: OECD CompStats database\n\nNotably, the number of mergers prohibited or challenged increased as a proportion of total decisions, particularly in OECD jurisdictions, where most prohibition decisions are concentrated (see Figure 2.5). 2024 was the year with the most prohibited or challenged mergers as a proportion of total merger decisions. As observed in the 2024 version of the Competition Trends report, an analysis of 216 prohibition or challenge decisions between 2015 and 2022 revealed that theories of harm based on unilateral, horizontal effects make up the majority of these decisions (OECD, 2024[19]). Prohibition decisions in non-OECD jurisdictions are relatively rare, representing less than 0.1% of merger decisions in 2024.\n\n### **Figure 2.5. Proportion of mergers challenged or prohibited over total merger decisions for OECD and Non-OECD jurisdictions, 2015-2024**\n\nNote: A three-year average is used to smooth out volatility caused by random fluctuations. Data based on the 58 jurisdictions in the OECD CompStats database that provided data for merger prohibitions or challenges for the period 2015-2024. Source: OECD CompStats database\n\n0%\n\n1%\n\n2%\n\n3%\n\n4%\n\n5%\n\n6%\n\n7%\n\nAverage merger intervention rate Trendline\n\nPercentage of decisions\n\n0\\.0%\n\n0\\.1%\n\n0\\.1%\n\n0\\.2%\n\n0\\.2%\n\n0\\.3%\n\n0\\.3%\n\n0\\.4%\n\n0\\.4%\n\n0\\.5%\n\nPercentage\n\nAll OECD Non-OECD","pages":[{"line_numbers":[],"markdown":" **19**\n\nA DECADE OF OECD COMPETITION TRENDS, DATA AND INSIGHTS © OECD 2025\n\n### **Figure 2.4. Average merger intervention rate, 2015-2024**\n\nNote: The intervention rate is the share of mergers approved with remedies, prohibited or challenged over total merger decisions. A three-year average is used to smooth out volatility caused by random fluctuations. Data based on the 54 jurisdictions in the OECD CompStats database that provided data for merger remedies and merger prohibitions or challenges for the period 2015-2024. Source: OECD CompStats database\n\nNotably, the number of mergers prohibited or challenged increased as a proportion of total decisions, particularly in OECD jurisdictions, where most prohibition decisions are concentrated (see Figure 2.5). 2024 was the year with the most prohibited or challenged mergers as a proportion of total merger decisions. As observed in the 2024 version of the Competition Trends report, an analysis of 216 prohibition or challenge decisions between 2015 and 2022 revealed that theories of harm based on unilateral, horizontal effects make up the majority of these decisions (OECD, 2024[19]). Prohibition decisions in non-OECD jurisdictions are relatively rare, representing less than 0.1% of merger decisions in 2024.\n\n### **Figure 2.5. Proportion of mergers challenged or prohibited over total merger decisions for OECD and Non-OECD jurisdictions, 2015-2024**\n\nNote: A three-year average is used to smooth out volatility caused by random fluctuations. Data based on the 58 jurisdictions in the OECD CompStats database that provided data for merger prohibitions or challenges for the period 2015-2024. Source: OECD CompStats database\n\n0%\n\n1%\n\n2%\n\n3%\n\n4%\n\n5%\n\n6%\n\n7%\n\nAverage merger intervention rate Trendline\n\nPercentage of decisions\n\n0\\.0%\n\n0\\.1%\n\n0\\.1%\n\n0\\.2%\n\n0\\.2%\n\n0\\.3%\n\n0\\.3%\n\n0\\.4%\n\n0\\.4%\n\n0\\.5%\n\nPercentage\n\nAll OECD Non-OECD","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":" **19**\n\nA DECADE OF OECD COMPETITION TRENDS, DATA AND INSIGHTS © OECD 2025\n\n### **Figure 2.4. Average merger intervention rate, 2015-2024**\n\nNote: The intervention rate is the share of mergers approved with remedies, prohibited or challenged over total merger decisions. A three-year average is used to smooth out volatility caused by random fluctuations. Data based on the 54 jurisdictions in the OECD CompStats database that provided data for merger remedies and merger prohibitions or challenges for the period 2015-2024. Source: OECD CompStats database\n\nNotably, the number of mergers prohibited or challenged increased as a proportion of total decisions, particularly in OECD jurisdictions, where most prohibition decisions are concentrated (see Figure 2.5). 2024 was the year with the most prohibited or challenged mergers as a proportion of total merger decisions. As observed in the 2024 version of the Competition Trends report, an analysis of 216 prohibition or challenge decisions between 2015 and 2022 revealed that theories of harm based on unilateral, horizontal effects make up the majority of these decisions (OECD, 2024[19]). Prohibition decisions in non-OECD jurisdictions are relatively rare, representing less than 0.1% of merger decisions in 2024.\n\n### **Figure 2.5. Proportion of mergers challenged or prohibited over total merger decisions for OECD and Non-OECD jurisdictions, 2015-2024**\n\nNote: A three-year average is used to smooth out volatility caused by random fluctuations. Data based on the 58 jurisdictions in the OECD CompStats database that provided data for merger prohibitions or challenges for the period 2015-2024. Source: OECD CompStats database\n\n0%\n\n1%\n\n2%\n\n3%\n\n4%\n\n5%\n\n6%\n\n7%\n\nAverage merger intervention rate Trendline\n\nPercentage of decisions\n\n0\\.0%\n\n0\\.1%\n\n0\\.1%\n\n0\\.2%\n\n0\\.2%\n\n0\\.3%\n\n0\\.3%\n\n0\\.4%\n\n0\\.4%\n\n0\\.5%\n\nPercentage\n\nAll OECD Non-OECD","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/05021ff2-en_p21.pdf"]},"response":{"errors":[],"extract_id":"extract_da1436099095d360ae0abe7f9b9b1d60","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":[" **19**\n\nA DECADE OF OECD COMPETITION TRENDS, DATA AND INSIGHTS © OECD 2025\n\n### **Figure 2.4. Average merger intervention rate, 2015-2024**\n\nNote: The intervention rate is the share of mergers approved with remedies, prohibited or challenged over total merger decisions. A three-year average is used to smooth out volatility caused by random fluctuations. Data based on the 54 jurisdictions in the OECD CompStats database that provided data for merger remedies and merger prohibitions or challenges for the period 2015-2024. Source: OECD CompStats database\n\nNotably, the number of mergers prohibited or challenged increased as a proportion of total decisions, particularly in OECD jurisdictions, where most prohibition decisions are concentrated (see Figure 2.5). 2024 was the year with the most prohibited or challenged mergers as a proportion of total merger decisions. As observed in the 2024 version of the Competition Trends report, an analysis of 216 prohibition or challenge decisions between 2015 and 2022 revealed that theories of harm based on unilateral, horizontal effects make up the majority of these decisions (OECD, 2024[19]). Prohibition decisions in non-OECD jurisdictions are relatively rare, representing less than 0.1% of merger decisions in 2024.\n\n### **Figure 2.5. Proportion of mergers challenged or prohibited over total merger decisions for OECD and Non-OECD jurisdictions, 2015-2024**\n\nNote: A three-year average is used to smooth out volatility caused by random fluctuations. Data based on the 58 jurisdictions in the OECD CompStats database that provided data for merger prohibitions or challenges for the period 2015-2024. Source: OECD CompStats database\n\n0%\n\n1%\n\n2%\n\n3%\n\n4%\n\n5%\n\n6%\n\n7%\n\nAverage merger intervention rate Trendline\n\nPercentage of decisions\n\n0\\.0%\n\n0\\.1%\n\n0\\.1%\n\n0\\.2%\n\n0\\.2%\n\n0\\.3%\n\n0\\.3%\n\n0\\.4%\n\n0\\.4%\n\n0\\.5%\n\nPercentage\n\nAll OECD Non-OECD"],"full_content":" **19**\n\nA DECADE OF OECD COMPETITION TRENDS, DATA AND INSIGHTS © OECD 2025\n\n### **Figure 2.4. Average merger intervention rate, 2015-2024**\n\nNote: The intervention rate is the share of mergers approved with remedies, prohibited or challenged over total merger decisions. A three-year average is used to smooth out volatility caused by random fluctuations. Data based on the 54 jurisdictions in the OECD CompStats database that provided data for merger remedies and merger prohibitions or challenges for the period 2015-2024. Source: OECD CompStats database\n\nNotably, the number of mergers prohibited or challenged increased as a proportion of total decisions, particularly in OECD jurisdictions, where most prohibition decisions are concentrated (see Figure 2.5). 2024 was the year with the most prohibited or challenged mergers as a proportion of total merger decisions. As observed in the 2024 version of the Competition Trends report, an analysis of 216 prohibition or challenge decisions between 2015 and 2022 revealed that theories of harm based on unilateral, horizontal effects make up the majority of these decisions (OECD, 2024[19]). Prohibition decisions in non-OECD jurisdictions are relatively rare, representing less than 0.1% of merger decisions in 2024.\n\n### **Figure 2.5. Proportion of mergers challenged or prohibited over total merger decisions for OECD and Non-OECD jurisdictions, 2015-2024**\n\nNote: A three-year average is used to smooth out volatility caused by random fluctuations. Data based on the 58 jurisdictions in the OECD CompStats database that provided data for merger prohibitions or challenges for the period 2015-2024. Source: OECD CompStats database\n\n0%\n\n1%\n\n2%\n\n3%\n\n4%\n\n5%\n\n6%\n\n7%\n\nAverage merger intervention rate Trendline\n\nPercentage of decisions\n\n0\\.0%\n\n0\\.1%\n\n0\\.1%\n\n0\\.2%\n\n0\\.2%\n\n0\\.3%\n\n0\\.3%\n\n0\\.4%\n\n0\\.4%\n\n0\\.5%\n\nPercentage\n\nAll OECD Non-OECD","publish_date":null,"title":"Figure 2.4. Average merger intervention rate, 2015-2024 Figure 2.5. Proportion of mergers challenged or prohibited over total merger decisions for","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/05021ff2-en_p21.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/05021ff2-en_p21.pdf"},"request":{"example_id":"chart/05021ff2-en_p21","product_type":"parse","source_file_path":"05021ff2-en_p21.pdf"},"started_at":"2026-11-04T19:31:07.334509"} +{"completed_at":"2026-11-04T19:46:25.033032","latency_in_ms":1378,"output":{"example_id":"chart/05021ff2-en_p27","grounded_pages":[],"layout_pages":[],"markdown":" **25**\n\nA DECADE OF OECD COMPETITION TRENDS, DATA AND INSIGHTS © OECD 2025\n\n### **3\\.1. Market studies remain a valuable tool for competition authorities**\n\nMarket studies are a useful tool for competition authorities to assess whether competition in a market is working effectively and identify recommendations to address any issues if not. They are also key for the advocacy activities of the authorities, as they may include key messages and recommendations for relevant actors in the markets. A total of 1 561 market studies were conducted in the past decade, with the average number of market studies carried out per year remained relatively constant (three per year per jurisdiction). It is important to note that these numbers do not consider the different definitions that each authority may have of what constitutes a market study, as well as any changes in content, quality and depth of the studies through time.\n\nAs it can be seen in Figure 3.1 there was a surge in the number of studies observed for OECD jurisdictions between 2020 and 2022 and despite the level of studies dropping off in the past two years, there remain more market studies per year in OECD jurisdictions than at the beginning of the decade. This shows their continued relevance for competition authorities, although there has been a fall in non-OECD jurisdictions. Reflecting their relevance, the decade has seen several pieces of OECD work related to market studies. Box 3.1 summarises this work.\n\n### **Figure 3.1. Total number of market studies conducted, 2015-2024**\n\nNote: A three-year average is used to smooth out volatility caused by random fluctuations. Data based on the 60 jurisdictions in the OECD CompStats database that provided data for market studies for the period 2015-2024. Source: OECD CompStats database.\n\n0\n\n20\n\n40\n\n60\n\n80\n\n100\n\n120\n\n140\n\n160\n\n180\n\n200\n\nAll OECD Non-OECD","pages":[{"line_numbers":[],"markdown":" **25**\n\nA DECADE OF OECD COMPETITION TRENDS, DATA AND INSIGHTS © OECD 2025\n\n### **3\\.1. Market studies remain a valuable tool for competition authorities**\n\nMarket studies are a useful tool for competition authorities to assess whether competition in a market is working effectively and identify recommendations to address any issues if not. They are also key for the advocacy activities of the authorities, as they may include key messages and recommendations for relevant actors in the markets. A total of 1 561 market studies were conducted in the past decade, with the average number of market studies carried out per year remained relatively constant (three per year per jurisdiction). It is important to note that these numbers do not consider the different definitions that each authority may have of what constitutes a market study, as well as any changes in content, quality and depth of the studies through time.\n\nAs it can be seen in Figure 3.1 there was a surge in the number of studies observed for OECD jurisdictions between 2020 and 2022 and despite the level of studies dropping off in the past two years, there remain more market studies per year in OECD jurisdictions than at the beginning of the decade. This shows their continued relevance for competition authorities, although there has been a fall in non-OECD jurisdictions. Reflecting their relevance, the decade has seen several pieces of OECD work related to market studies. Box 3.1 summarises this work.\n\n### **Figure 3.1. Total number of market studies conducted, 2015-2024**\n\nNote: A three-year average is used to smooth out volatility caused by random fluctuations. Data based on the 60 jurisdictions in the OECD CompStats database that provided data for market studies for the period 2015-2024. Source: OECD CompStats database.\n\n0\n\n20\n\n40\n\n60\n\n80\n\n100\n\n120\n\n140\n\n160\n\n180\n\n200\n\nAll OECD Non-OECD","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":" **25**\n\nA DECADE OF OECD COMPETITION TRENDS, DATA AND INSIGHTS © OECD 2025\n\n### **3\\.1. Market studies remain a valuable tool for competition authorities**\n\nMarket studies are a useful tool for competition authorities to assess whether competition in a market is working effectively and identify recommendations to address any issues if not. They are also key for the advocacy activities of the authorities, as they may include key messages and recommendations for relevant actors in the markets. A total of 1 561 market studies were conducted in the past decade, with the average number of market studies carried out per year remained relatively constant (three per year per jurisdiction). It is important to note that these numbers do not consider the different definitions that each authority may have of what constitutes a market study, as well as any changes in content, quality and depth of the studies through time.\n\nAs it can be seen in Figure 3.1 there was a surge in the number of studies observed for OECD jurisdictions between 2020 and 2022 and despite the level of studies dropping off in the past two years, there remain more market studies per year in OECD jurisdictions than at the beginning of the decade. This shows their continued relevance for competition authorities, although there has been a fall in non-OECD jurisdictions. Reflecting their relevance, the decade has seen several pieces of OECD work related to market studies. Box 3.1 summarises this work.\n\n### **Figure 3.1. Total number of market studies conducted, 2015-2024**\n\nNote: A three-year average is used to smooth out volatility caused by random fluctuations. Data based on the 60 jurisdictions in the OECD CompStats database that provided data for market studies for the period 2015-2024. Source: OECD CompStats database.\n\n0\n\n20\n\n40\n\n60\n\n80\n\n100\n\n120\n\n140\n\n160\n\n180\n\n200\n\nAll OECD Non-OECD","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/05021ff2-en_p27.pdf"]},"response":{"errors":[],"extract_id":"extract_796f223a1b29263790efdcd16c2327f8","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":[" **25**\n\nA DECADE OF OECD COMPETITION TRENDS, DATA AND INSIGHTS © OECD 2025\n\n### **3\\.1. Market studies remain a valuable tool for competition authorities**\n\nMarket studies are a useful tool for competition authorities to assess whether competition in a market is working effectively and identify recommendations to address any issues if not. They are also key for the advocacy activities of the authorities, as they may include key messages and recommendations for relevant actors in the markets. A total of 1 561 market studies were conducted in the past decade, with the average number of market studies carried out per year remained relatively constant (three per year per jurisdiction). It is important to note that these numbers do not consider the different definitions that each authority may have of what constitutes a market study, as well as any changes in content, quality and depth of the studies through time.\n\nAs it can be seen in Figure 3.1 there was a surge in the number of studies observed for OECD jurisdictions between 2020 and 2022 and despite the level of studies dropping off in the past two years, there remain more market studies per year in OECD jurisdictions than at the beginning of the decade. This shows their continued relevance for competition authorities, although there has been a fall in non-OECD jurisdictions. Reflecting their relevance, the decade has seen several pieces of OECD work related to market studies. Box 3.1 summarises this work.\n\n### **Figure 3.1. Total number of market studies conducted, 2015-2024**\n\nNote: A three-year average is used to smooth out volatility caused by random fluctuations. Data based on the 60 jurisdictions in the OECD CompStats database that provided data for market studies for the period 2015-2024. Source: OECD CompStats database.\n\n0\n\n20\n\n40\n\n60\n\n80\n\n100\n\n120\n\n140\n\n160\n\n180\n\n200\n\nAll OECD Non-OECD"],"full_content":" **25**\n\nA DECADE OF OECD COMPETITION TRENDS, DATA AND INSIGHTS © OECD 2025\n\n### **3\\.1. Market studies remain a valuable tool for competition authorities**\n\nMarket studies are a useful tool for competition authorities to assess whether competition in a market is working effectively and identify recommendations to address any issues if not. They are also key for the advocacy activities of the authorities, as they may include key messages and recommendations for relevant actors in the markets. A total of 1 561 market studies were conducted in the past decade, with the average number of market studies carried out per year remained relatively constant (three per year per jurisdiction). It is important to note that these numbers do not consider the different definitions that each authority may have of what constitutes a market study, as well as any changes in content, quality and depth of the studies through time.\n\nAs it can be seen in Figure 3.1 there was a surge in the number of studies observed for OECD jurisdictions between 2020 and 2022 and despite the level of studies dropping off in the past two years, there remain more market studies per year in OECD jurisdictions than at the beginning of the decade. This shows their continued relevance for competition authorities, although there has been a fall in non-OECD jurisdictions. Reflecting their relevance, the decade has seen several pieces of OECD work related to market studies. Box 3.1 summarises this work.\n\n### **Figure 3.1. Total number of market studies conducted, 2015-2024**\n\nNote: A three-year average is used to smooth out volatility caused by random fluctuations. Data based on the 60 jurisdictions in the OECD CompStats database that provided data for market studies for the period 2015-2024. Source: OECD CompStats database.\n\n0\n\n20\n\n40\n\n60\n\n80\n\n100\n\n120\n\n140\n\n160\n\n180\n\n200\n\nAll OECD Non-OECD","publish_date":null,"title":"3.1. Market studies remain a valuable tool for competition authorities Figure 3.1. Total number of market studies conducted, 2015-2024","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/05021ff2-en_p27.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/05021ff2-en_p27.pdf"},"request":{"example_id":"chart/05021ff2-en_p27","product_type":"parse","source_file_path":"05021ff2-en_p27.pdf"},"started_at":"2026-11-04T19:46:23.654231"} +{"completed_at":"2026-11-04T19:46:27.137686","latency_in_ms":2536,"output":{"example_id":"chart/2022_SPI_Benchmark_Executive_Summary_p16","grounded_pages":[],"layout_pages":[],"markdown":"### Continuing the last three years’ trend of steady climbing, projects in 2021 averagely had a contribution margin of 36.5%.\n\n### Drivers for increasing profit margins within the industry are the companies who manage to implement real-time visibility in their daily operations.\n\n### IT solutions such as Business Intelligence and Professional Services Automation (PSA) software help companies facilitate this implementation.\n\n# Project profitability The steady climb continues - with a nice bump upwardsprojects\n\n### 2017 2018 2020 2021 2019\n\n### Project contribution ratio 5 year trend:\n\nProjects don’t provide you with the expected\n\nROI?\n\nAre your prices based on actual time\n\nconsumption, or do you price your fixed\n\nprice projects too cheap?\n\nDo you remember to staff your projects’\n\njunior tasks with junior resources only?\n\nInitiated Piloted Deployed Institutionalised Optimised\n\n### 36\\.5%\n\n### 31\\.7%\n\n### 34\\.7% 35.4% 35.6%\n\nProject contribution ratio across maturity levels\n\nMature companies see higher profits on their projects\n\n15\\.4% 26.4% 37.7% 47.2% 55.5%\n\n### 16","pages":[{"line_numbers":[],"markdown":"### Continuing the last three years’ trend of steady climbing, projects in 2021 averagely had a contribution margin of 36.5%.\n\n### Drivers for increasing profit margins within the industry are the companies who manage to implement real-time visibility in their daily operations.\n\n### IT solutions such as Business Intelligence and Professional Services Automation (PSA) software help companies facilitate this implementation.\n\n# Project profitability The steady climb continues - with a nice bump upwardsprojects\n\n### 2017 2018 2020 2021 2019\n\n### Project contribution ratio 5 year trend:\n\nProjects don’t provide you with the expected\n\nROI?\n\nAre your prices based on actual time\n\nconsumption, or do you price your fixed\n\nprice projects too cheap?\n\nDo you remember to staff your projects’\n\njunior tasks with junior resources only?\n\nInitiated Piloted Deployed Institutionalised Optimised\n\n### 36\\.5%\n\n### 31\\.7%\n\n### 34\\.7% 35.4% 35.6%\n\nProject contribution ratio across maturity levels\n\nMature companies see higher profits on their projects\n\n15\\.4% 26.4% 37.7% 47.2% 55.5%\n\n### 16","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"### Continuing the last three years’ trend of steady climbing, projects in 2021 averagely had a contribution margin of 36.5%.\n\n### Drivers for increasing profit margins within the industry are the companies who manage to implement real-time visibility in their daily operations.\n\n### IT solutions such as Business Intelligence and Professional Services Automation (PSA) software help companies facilitate this implementation.\n\n# Project profitability The steady climb continues - with a nice bump upwardsprojects\n\n### 2017 2018 2020 2021 2019\n\n### Project contribution ratio 5 year trend:\n\nProjects don’t provide you with the expected\n\nROI?\n\nAre your prices based on actual time\n\nconsumption, or do you price your fixed\n\nprice projects too cheap?\n\nDo you remember to staff your projects’\n\njunior tasks with junior resources only?\n\nInitiated Piloted Deployed Institutionalised Optimised\n\n### 36\\.5%\n\n### 31\\.7%\n\n### 34\\.7% 35.4% 35.6%\n\nProject contribution ratio across maturity levels\n\nMature companies see higher profits on their projects\n\n15\\.4% 26.4% 37.7% 47.2% 55.5%\n\n### 16","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2022_SPI_Benchmark_Executive_Summary_p16.pdf"]},"response":{"errors":[],"extract_id":"extract_1dd86ce6a94cbbcf2d63192170ab97d1","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["### Continuing the last three years’ trend of steady climbing, projects in 2021 averagely had a contribution margin of 36.5%.\n\n### Drivers for increasing profit margins within the industry are the companies who manage to implement real-time visibility in their daily operations.\n\n### IT solutions such as Business Intelligence and Professional Services Automation (PSA) software help companies facilitate this implementation.\n\n# Project profitability The steady climb continues - with a nice bump upwardsprojects\n\n### 2017 2018 2020 2021 2019\n\n### Project contribution ratio 5 year trend:\n\nProjects don’t provide you with the expected\n\nROI?\n\nAre your prices based on actual time\n\nconsumption, or do you price your fixed\n\nprice projects too cheap?\n\nDo you remember to staff your projects’\n\njunior tasks with junior resources only?\n\nInitiated Piloted Deployed Institutionalised Optimised\n\n### 36\\.5%\n\n### 31\\.7%\n\n### 34\\.7% 35.4% 35.6%\n\nProject contribution ratio across maturity levels\n\nMature companies see higher profits on their projects\n\n15\\.4% 26.4% 37.7% 47.2% 55.5%\n\n### 16"],"full_content":"### Continuing the last three years’ trend of steady climbing, projects in 2021 averagely had a contribution margin of 36.5%.\n\n### Drivers for increasing profit margins within the industry are the companies who manage to implement real-time visibility in their daily operations.\n\n### IT solutions such as Business Intelligence and Professional Services Automation (PSA) software help companies facilitate this implementation.\n\n# Project profitability The steady climb continues - with a nice bump upwardsprojects\n\n### 2017 2018 2020 2021 2019\n\n### Project contribution ratio 5 year trend:\n\nProjects don’t provide you with the expected\n\nROI?\n\nAre your prices based on actual time\n\nconsumption, or do you price your fixed\n\nprice projects too cheap?\n\nDo you remember to staff your projects’\n\njunior tasks with junior resources only?\n\nInitiated Piloted Deployed Institutionalised Optimised\n\n### 36\\.5%\n\n### 31\\.7%\n\n### 34\\.7% 35.4% 35.6%\n\nProject contribution ratio across maturity levels\n\nMature companies see higher profits on their projects\n\n15\\.4% 26.4% 37.7% 47.2% 55.5%\n\n### 16","publish_date":null,"title":"Project profitability","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2022_SPI_Benchmark_Executive_Summary_p16.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2022_SPI_Benchmark_Executive_Summary_p16.pdf"},"request":{"example_id":"chart/2022_SPI_Benchmark_Executive_Summary_p16","product_type":"parse","source_file_path":"2022_SPI_Benchmark_Executive_Summary_p16.pdf"},"started_at":"2026-11-04T19:46:24.601348"} +{"completed_at":"2026-11-04T19:31:09.677328","latency_in_ms":1294,"output":{"example_id":"chart/2022_SPI_Benchmark_Executive_Summary_p18","grounded_pages":[],"layout_pages":[],"markdown":"### The hourly rate companies invoice their clients is a poor indicator of how profitable their businesses are.\n\n### Internal tasks, fixed price projects exceeding budgets, discounts and lack of clarity as to how many hours should be billed always lower the realised hourly rate.\n\n### Looking at the discrepancy between the billed and the realised hourly rate gives you crucial insights into your ability to budget and execute your projects.\n\n# Realised hourly rate at €92 What is your time worth?projects\n\n### 2017 2018 2020 2021 2019\n\n### Realised hourly rate the past 5 years\\*:\n\nNot achieving your desired realised hourly rate?\n\nDo you remember to always round hours\n\nto nearest quarter or half an hour on your\n\ninvoices?\n\nDo you remember to include all internal\n\nproject management costs in fixed price\n\nprojects when giving potential clients a\n\nquote?\n\nInitiated Piloted Deployed Institutionalised Optimised\n\nRealised hourly rate across maturity levels\n\nMature companies see higher revenue on their consultants’ working hours\n\n€37.5 €67.0 €92.9 €113.0 €123.2\n\n### €87.1\n\n### €92.4 €90.6 €92 €92\n\n\\* Calculated on a yearly work time of 2,000 hours\n\n### 18","pages":[{"line_numbers":[],"markdown":"### The hourly rate companies invoice their clients is a poor indicator of how profitable their businesses are.\n\n### Internal tasks, fixed price projects exceeding budgets, discounts and lack of clarity as to how many hours should be billed always lower the realised hourly rate.\n\n### Looking at the discrepancy between the billed and the realised hourly rate gives you crucial insights into your ability to budget and execute your projects.\n\n# Realised hourly rate at €92 What is your time worth?projects\n\n### 2017 2018 2020 2021 2019\n\n### Realised hourly rate the past 5 years\\*:\n\nNot achieving your desired realised hourly rate?\n\nDo you remember to always round hours\n\nto nearest quarter or half an hour on your\n\ninvoices?\n\nDo you remember to include all internal\n\nproject management costs in fixed price\n\nprojects when giving potential clients a\n\nquote?\n\nInitiated Piloted Deployed Institutionalised Optimised\n\nRealised hourly rate across maturity levels\n\nMature companies see higher revenue on their consultants’ working hours\n\n€37.5 €67.0 €92.9 €113.0 €123.2\n\n### €87.1\n\n### €92.4 €90.6 €92 €92\n\n\\* Calculated on a yearly work time of 2,000 hours\n\n### 18","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"### The hourly rate companies invoice their clients is a poor indicator of how profitable their businesses are.\n\n### Internal tasks, fixed price projects exceeding budgets, discounts and lack of clarity as to how many hours should be billed always lower the realised hourly rate.\n\n### Looking at the discrepancy between the billed and the realised hourly rate gives you crucial insights into your ability to budget and execute your projects.\n\n# Realised hourly rate at €92 What is your time worth?projects\n\n### 2017 2018 2020 2021 2019\n\n### Realised hourly rate the past 5 years\\*:\n\nNot achieving your desired realised hourly rate?\n\nDo you remember to always round hours\n\nto nearest quarter or half an hour on your\n\ninvoices?\n\nDo you remember to include all internal\n\nproject management costs in fixed price\n\nprojects when giving potential clients a\n\nquote?\n\nInitiated Piloted Deployed Institutionalised Optimised\n\nRealised hourly rate across maturity levels\n\nMature companies see higher revenue on their consultants’ working hours\n\n€37.5 €67.0 €92.9 €113.0 €123.2\n\n### €87.1\n\n### €92.4 €90.6 €92 €92\n\n\\* Calculated on a yearly work time of 2,000 hours\n\n### 18","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2022_SPI_Benchmark_Executive_Summary_p18.pdf"]},"response":{"errors":[],"extract_id":"extract_fb9c4738ae79c9f7b81608c1c8e3aab8","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["### The hourly rate companies invoice their clients is a poor indicator of how profitable their businesses are.\n\n### Internal tasks, fixed price projects exceeding budgets, discounts and lack of clarity as to how many hours should be billed always lower the realised hourly rate.\n\n### Looking at the discrepancy between the billed and the realised hourly rate gives you crucial insights into your ability to budget and execute your projects.\n\n# Realised hourly rate at €92 What is your time worth?projects\n\n### 2017 2018 2020 2021 2019\n\n### Realised hourly rate the past 5 years\\*:\n\nNot achieving your desired realised hourly rate?\n\nDo you remember to always round hours\n\nto nearest quarter or half an hour on your\n\ninvoices?\n\nDo you remember to include all internal\n\nproject management costs in fixed price\n\nprojects when giving potential clients a\n\nquote?\n\nInitiated Piloted Deployed Institutionalised Optimised\n\nRealised hourly rate across maturity levels\n\nMature companies see higher revenue on their consultants’ working hours\n\n€37.5 €67.0 €92.9 €113.0 €123.2\n\n### €87.1\n\n### €92.4 €90.6 €92 €92\n\n\\* Calculated on a yearly work time of 2,000 hours\n\n### 18"],"full_content":"### The hourly rate companies invoice their clients is a poor indicator of how profitable their businesses are.\n\n### Internal tasks, fixed price projects exceeding budgets, discounts and lack of clarity as to how many hours should be billed always lower the realised hourly rate.\n\n### Looking at the discrepancy between the billed and the realised hourly rate gives you crucial insights into your ability to budget and execute your projects.\n\n# Realised hourly rate at €92 What is your time worth?projects\n\n### 2017 2018 2020 2021 2019\n\n### Realised hourly rate the past 5 years\\*:\n\nNot achieving your desired realised hourly rate?\n\nDo you remember to always round hours\n\nto nearest quarter or half an hour on your\n\ninvoices?\n\nDo you remember to include all internal\n\nproject management costs in fixed price\n\nprojects when giving potential clients a\n\nquote?\n\nInitiated Piloted Deployed Institutionalised Optimised\n\nRealised hourly rate across maturity levels\n\nMature companies see higher revenue on their consultants’ working hours\n\n€37.5 €67.0 €92.9 €113.0 €123.2\n\n### €87.1\n\n### €92.4 €90.6 €92 €92\n\n\\* Calculated on a yearly work time of 2,000 hours\n\n### 18","publish_date":null,"title":"Realised hourly rate at €92","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2022_SPI_Benchmark_Executive_Summary_p18.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2022_SPI_Benchmark_Executive_Summary_p18.pdf"},"request":{"example_id":"chart/2022_SPI_Benchmark_Executive_Summary_p18","product_type":"parse","source_file_path":"2022_SPI_Benchmark_Executive_Summary_p18.pdf"},"started_at":"2026-11-04T19:31:08.382734"} +{"completed_at":"2026-11-04T19:46:31.753269","latency_in_ms":4257,"output":{"example_id":"chart/2022_SPI_Benchmark_Executive_Summary_p19","grounded_pages":[],"layout_pages":[],"markdown":"### The job market was hotter than ever in the professional services industry in 2021.\n\n### While headcount growth took a big jump from 5.5% in 2020 to 9.1% in 2021, this figure could be much higher, if talent attrition didn’t also skyrocket.\n\n### While the good news is that layoffs were down by 11% compared to 2020, annual, voluntary talent attrition went up by 42% to 9.8%.\n\n# Headcount growth Hiring was heating back up in 2021 - but so was attrition\n\n### 2017 2018 2020 2021 2019\n\n### 5 years of bumpy headcount growth:\n\nHaving trouble attracting talent?\n\nDo you have a clearly defined and marketed\n\ncareer ladder, talent and mentoring\n\nprogram within your organisation?\n\nInitiated Piloted Deployed Institutionalised Optimised\n\nHeadcount growth across maturity levels\n\nIn the hunt for the best talent, mature companies manage to find and attract the people they need\n\n7\\.0% 7.8% 9.0% 13.4% 15.4%\n\n### 9\\.1% 9.2%\n\n### 7\\.7%\n\n### 9\\.0%\n\n### 5\\.5%\n\n### 19","pages":[{"line_numbers":[],"markdown":"### The job market was hotter than ever in the professional services industry in 2021.\n\n### While headcount growth took a big jump from 5.5% in 2020 to 9.1% in 2021, this figure could be much higher, if talent attrition didn’t also skyrocket.\n\n### While the good news is that layoffs were down by 11% compared to 2020, annual, voluntary talent attrition went up by 42% to 9.8%.\n\n# Headcount growth Hiring was heating back up in 2021 - but so was attrition\n\n### 2017 2018 2020 2021 2019\n\n### 5 years of bumpy headcount growth:\n\nHaving trouble attracting talent?\n\nDo you have a clearly defined and marketed\n\ncareer ladder, talent and mentoring\n\nprogram within your organisation?\n\nInitiated Piloted Deployed Institutionalised Optimised\n\nHeadcount growth across maturity levels\n\nIn the hunt for the best talent, mature companies manage to find and attract the people they need\n\n7\\.0% 7.8% 9.0% 13.4% 15.4%\n\n### 9\\.1% 9.2%\n\n### 7\\.7%\n\n### 9\\.0%\n\n### 5\\.5%\n\n### 19","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"### The job market was hotter than ever in the professional services industry in 2021.\n\n### While headcount growth took a big jump from 5.5% in 2020 to 9.1% in 2021, this figure could be much higher, if talent attrition didn’t also skyrocket.\n\n### While the good news is that layoffs were down by 11% compared to 2020, annual, voluntary talent attrition went up by 42% to 9.8%.\n\n# Headcount growth Hiring was heating back up in 2021 - but so was attrition\n\n### 2017 2018 2020 2021 2019\n\n### 5 years of bumpy headcount growth:\n\nHaving trouble attracting talent?\n\nDo you have a clearly defined and marketed\n\ncareer ladder, talent and mentoring\n\nprogram within your organisation?\n\nInitiated Piloted Deployed Institutionalised Optimised\n\nHeadcount growth across maturity levels\n\nIn the hunt for the best talent, mature companies manage to find and attract the people they need\n\n7\\.0% 7.8% 9.0% 13.4% 15.4%\n\n### 9\\.1% 9.2%\n\n### 7\\.7%\n\n### 9\\.0%\n\n### 5\\.5%\n\n### 19","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2022_SPI_Benchmark_Executive_Summary_p19.pdf"]},"response":{"errors":[],"extract_id":"extract_a0a8dea830168ff99c2b9f4b8c3da3f2","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["### The job market was hotter than ever in the professional services industry in 2021.\n\n### While headcount growth took a big jump from 5.5% in 2020 to 9.1% in 2021, this figure could be much higher, if talent attrition didn’t also skyrocket.\n\n### While the good news is that layoffs were down by 11% compared to 2020, annual, voluntary talent attrition went up by 42% to 9.8%.\n\n# Headcount growth Hiring was heating back up in 2021 - but so was attrition\n\n### 2017 2018 2020 2021 2019\n\n### 5 years of bumpy headcount growth:\n\nHaving trouble attracting talent?\n\nDo you have a clearly defined and marketed\n\ncareer ladder, talent and mentoring\n\nprogram within your organisation?\n\nInitiated Piloted Deployed Institutionalised Optimised\n\nHeadcount growth across maturity levels\n\nIn the hunt for the best talent, mature companies manage to find and attract the people they need\n\n7\\.0% 7.8% 9.0% 13.4% 15.4%\n\n### 9\\.1% 9.2%\n\n### 7\\.7%\n\n### 9\\.0%\n\n### 5\\.5%\n\n### 19"],"full_content":"### The job market was hotter than ever in the professional services industry in 2021.\n\n### While headcount growth took a big jump from 5.5% in 2020 to 9.1% in 2021, this figure could be much higher, if talent attrition didn’t also skyrocket.\n\n### While the good news is that layoffs were down by 11% compared to 2020, annual, voluntary talent attrition went up by 42% to 9.8%.\n\n# Headcount growth Hiring was heating back up in 2021 - but so was attrition\n\n### 2017 2018 2020 2021 2019\n\n### 5 years of bumpy headcount growth:\n\nHaving trouble attracting talent?\n\nDo you have a clearly defined and marketed\n\ncareer ladder, talent and mentoring\n\nprogram within your organisation?\n\nInitiated Piloted Deployed Institutionalised Optimised\n\nHeadcount growth across maturity levels\n\nIn the hunt for the best talent, mature companies manage to find and attract the people they need\n\n7\\.0% 7.8% 9.0% 13.4% 15.4%\n\n### 9\\.1% 9.2%\n\n### 7\\.7%\n\n### 9\\.0%\n\n### 5\\.5%\n\n### 19","publish_date":null,"title":"Headcount growth","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2022_SPI_Benchmark_Executive_Summary_p19.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2022_SPI_Benchmark_Executive_Summary_p19.pdf"},"request":{"example_id":"chart/2022_SPI_Benchmark_Executive_Summary_p19","product_type":"parse","source_file_path":"2022_SPI_Benchmark_Executive_Summary_p19.pdf"},"started_at":"2026-11-04T19:46:27.495834"} +{"completed_at":"2026-11-04T19:31:11.163777","latency_in_ms":1955,"output":{"example_id":"chart/2022_SPI_Benchmark_Executive_Summary_p20","grounded_pages":[],"layout_pages":[],"markdown":"### Utilisation is consistently the most measured KPI in professional services companies. It is also an indicator of the balance between workload and opportunity.\n\n### This year the industry surpassed the past 15 years of billable utilisation levels with a record-breaking 73.2%.\n\n### One of the main drivers for a high billable utilisation is the adoption of PSA solutions, as companies using PSA display 75.3% billable utilisation. In contrast, companies that have not yet adopted a PSA solution could only muster 68.1% in 2021.\n\n# Billable utilisation Consultants’ time was highly billable in 2021\n\n### 2017 2018 2020 2021 2019\n\n### 5 years of billable utilisation:\n\nDo your consultants not spend enough time on\n\nbillable work?\n\nDo you and your employees have access to\n\nreal time dashboards that let you\n\nkeep track of your billable utilisation?\n\nHave you identified the non-billable\n\nactivities your employees spend most time\n\non?\n\nInitiated Piloted Deployed Institutionalised Optimised\n\n### 73\\.2%\n\n### 71\\.5%\n\n### 69\\.7%\n\n### 71\\.7% 71.4%\n\nBillable utilisation across maturity levels\n\nMature companies outperform industry average utilisation\n\nby 13%\n\n57\\.3% 68.2% 77.2% 77.5% 82.5%\n\n### 20","pages":[{"line_numbers":[],"markdown":"### Utilisation is consistently the most measured KPI in professional services companies. It is also an indicator of the balance between workload and opportunity.\n\n### This year the industry surpassed the past 15 years of billable utilisation levels with a record-breaking 73.2%.\n\n### One of the main drivers for a high billable utilisation is the adoption of PSA solutions, as companies using PSA display 75.3% billable utilisation. In contrast, companies that have not yet adopted a PSA solution could only muster 68.1% in 2021.\n\n# Billable utilisation Consultants’ time was highly billable in 2021\n\n### 2017 2018 2020 2021 2019\n\n### 5 years of billable utilisation:\n\nDo your consultants not spend enough time on\n\nbillable work?\n\nDo you and your employees have access to\n\nreal time dashboards that let you\n\nkeep track of your billable utilisation?\n\nHave you identified the non-billable\n\nactivities your employees spend most time\n\non?\n\nInitiated Piloted Deployed Institutionalised Optimised\n\n### 73\\.2%\n\n### 71\\.5%\n\n### 69\\.7%\n\n### 71\\.7% 71.4%\n\nBillable utilisation across maturity levels\n\nMature companies outperform industry average utilisation\n\nby 13%\n\n57\\.3% 68.2% 77.2% 77.5% 82.5%\n\n### 20","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"### Utilisation is consistently the most measured KPI in professional services companies. It is also an indicator of the balance between workload and opportunity.\n\n### This year the industry surpassed the past 15 years of billable utilisation levels with a record-breaking 73.2%.\n\n### One of the main drivers for a high billable utilisation is the adoption of PSA solutions, as companies using PSA display 75.3% billable utilisation. In contrast, companies that have not yet adopted a PSA solution could only muster 68.1% in 2021.\n\n# Billable utilisation Consultants’ time was highly billable in 2021\n\n### 2017 2018 2020 2021 2019\n\n### 5 years of billable utilisation:\n\nDo your consultants not spend enough time on\n\nbillable work?\n\nDo you and your employees have access to\n\nreal time dashboards that let you\n\nkeep track of your billable utilisation?\n\nHave you identified the non-billable\n\nactivities your employees spend most time\n\non?\n\nInitiated Piloted Deployed Institutionalised Optimised\n\n### 73\\.2%\n\n### 71\\.5%\n\n### 69\\.7%\n\n### 71\\.7% 71.4%\n\nBillable utilisation across maturity levels\n\nMature companies outperform industry average utilisation\n\nby 13%\n\n57\\.3% 68.2% 77.2% 77.5% 82.5%\n\n### 20","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2022_SPI_Benchmark_Executive_Summary_p20.pdf"]},"response":{"errors":[],"extract_id":"extract_44ac64dd11de7c79ff3fce3517d8a5a9","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["### Utilisation is consistently the most measured KPI in professional services companies. It is also an indicator of the balance between workload and opportunity.\n\n### This year the industry surpassed the past 15 years of billable utilisation levels with a record-breaking 73.2%.\n\n### One of the main drivers for a high billable utilisation is the adoption of PSA solutions, as companies using PSA display 75.3% billable utilisation. In contrast, companies that have not yet adopted a PSA solution could only muster 68.1% in 2021.\n\n# Billable utilisation Consultants’ time was highly billable in 2021\n\n### 2017 2018 2020 2021 2019\n\n### 5 years of billable utilisation:\n\nDo your consultants not spend enough time on\n\nbillable work?\n\nDo you and your employees have access to\n\nreal time dashboards that let you\n\nkeep track of your billable utilisation?\n\nHave you identified the non-billable\n\nactivities your employees spend most time\n\non?\n\nInitiated Piloted Deployed Institutionalised Optimised\n\n### 73\\.2%\n\n### 71\\.5%\n\n### 69\\.7%\n\n### 71\\.7% 71.4%\n\nBillable utilisation across maturity levels\n\nMature companies outperform industry average utilisation\n\nby 13%\n\n57\\.3% 68.2% 77.2% 77.5% 82.5%\n\n### 20"],"full_content":"### Utilisation is consistently the most measured KPI in professional services companies. It is also an indicator of the balance between workload and opportunity.\n\n### This year the industry surpassed the past 15 years of billable utilisation levels with a record-breaking 73.2%.\n\n### One of the main drivers for a high billable utilisation is the adoption of PSA solutions, as companies using PSA display 75.3% billable utilisation. In contrast, companies that have not yet adopted a PSA solution could only muster 68.1% in 2021.\n\n# Billable utilisation Consultants’ time was highly billable in 2021\n\n### 2017 2018 2020 2021 2019\n\n### 5 years of billable utilisation:\n\nDo your consultants not spend enough time on\n\nbillable work?\n\nDo you and your employees have access to\n\nreal time dashboards that let you\n\nkeep track of your billable utilisation?\n\nHave you identified the non-billable\n\nactivities your employees spend most time\n\non?\n\nInitiated Piloted Deployed Institutionalised Optimised\n\n### 73\\.2%\n\n### 71\\.5%\n\n### 69\\.7%\n\n### 71\\.7% 71.4%\n\nBillable utilisation across maturity levels\n\nMature companies outperform industry average utilisation\n\nby 13%\n\n57\\.3% 68.2% 77.2% 77.5% 82.5%\n\n### 20","publish_date":null,"title":"Billable utilisation","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2022_SPI_Benchmark_Executive_Summary_p20.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2022_SPI_Benchmark_Executive_Summary_p20.pdf"},"request":{"example_id":"chart/2022_SPI_Benchmark_Executive_Summary_p20","product_type":"parse","source_file_path":"2022_SPI_Benchmark_Executive_Summary_p20.pdf"},"started_at":"2026-11-04T19:31:09.208234"} +{"completed_at":"2026-11-04T19:46:47.686085","latency_in_ms":16772,"output":{"example_id":"chart/2023-05-sigma-01-english_p10","grounded_pages":[],"layout_pages":[],"markdown":"10 **Swiss Re Institute** _sigma_ 1/2023 In perspective: 2022 and longer-term trends\n\n### Up-trend reaffirmed, and set to continue\n\nInsured losses from natural catastrophes have been on a long-term upward trajectory for much longer than the last six years. Since 1992, the average annual trend growth of losses has been 5‒7% (see Figure 6). In the period 2012‒2016, losses were at a lower annual mean, but there has been a return to said trend growth over the last six years and we expect this to continue. We project that irrespective of year-on-year volatility, insured losses will continue to grow at trend, even when real-time amplifying factors such as high inflation subdue.\n\nStarting from 2017, average annual insured losses from natural catastrophes have been above USD 110 billion, more than double the average of USD 52 billion over the previous 5-year period. This marks a notable step-up in the scale of losses after the 2012‒2016 period of benign losses. Asset value accumulation in an area struck by an extreme weather or other natural catastrophe can spark heavy financial losses. Another factor has been changes in construction costs. Of late, aging infrastructure vulnerabilities and inflationary pressures have boosted repair costs. And, with demand-supply imbalances of materials and labour still in play in the post COVID-19 era, we expect construction costs to remain elevated into 2023, further inflating repair and reconstruction costs. This will impact attritional losses and further augment property losses over the next two years at least. Hazard intensification will likely play a bigger contributory role to rising losses in the coming decade also. With the world getting warmer, findings from scientific research infer that climate change effects on loss frequency and severity will intensify.\n\nWe also expect, irrespective of below-average loss years, that annual insured losses will average more than USD 100 billion from hereon. Our expectation is supported by other parties: for example, Verisk recently modeled the global insured average annual loss as USD 123 billion. ¹⁴ In any one year losses can be higher or lower depending on whether natural catastrophe events do or do not strike urban and more populated areas. A case in point, had Hurricane Ian last year made landfall in the Tampa Bay area as many predictions were showing, the resulting losses would have been much higher. There is no reason to anticipate that this, nor peak-loss disasters like Hurricane Katrina in 2005, will not happen again in the future. The takeaway is to not underestimate loss potential on account of a year or period of below trend growth.\n\n14 _Global Modelled Catastrophe Losses_ , Verisk, 2022 _._\n\nWe expect that average insured losses from natural catastrophes will continue to grow by 5‒7% annually.\n\n**Figure 6** Growth in global natural catastrophe insured losses in USD billion (2022 prices)\n\nSource: Swiss Re Institute\n\n0\n\n25\n\n50\n\n75\n\n100\n\n125\n\n150\n\n175\n\n200\n\n2022 2020 2018 2016 2014 2012 2010 2008 2006 2004 2002 2000 1998 1996 1994 1992\n\nTrend Earthquakes Weather-related\n\nSince 2017, average annual insured losses have exceeded USD 110 billion.\n\nLosses of more that USD 100 billion annually are here to stay.","pages":[{"line_numbers":[],"markdown":"10 **Swiss Re Institute** _sigma_ 1/2023 In perspective: 2022 and longer-term trends\n\n### Up-trend reaffirmed, and set to continue\n\nInsured losses from natural catastrophes have been on a long-term upward trajectory for much longer than the last six years. Since 1992, the average annual trend growth of losses has been 5‒7% (see Figure 6). In the period 2012‒2016, losses were at a lower annual mean, but there has been a return to said trend growth over the last six years and we expect this to continue. We project that irrespective of year-on-year volatility, insured losses will continue to grow at trend, even when real-time amplifying factors such as high inflation subdue.\n\nStarting from 2017, average annual insured losses from natural catastrophes have been above USD 110 billion, more than double the average of USD 52 billion over the previous 5-year period. This marks a notable step-up in the scale of losses after the 2012‒2016 period of benign losses. Asset value accumulation in an area struck by an extreme weather or other natural catastrophe can spark heavy financial losses. Another factor has been changes in construction costs. Of late, aging infrastructure vulnerabilities and inflationary pressures have boosted repair costs. And, with demand-supply imbalances of materials and labour still in play in the post COVID-19 era, we expect construction costs to remain elevated into 2023, further inflating repair and reconstruction costs. This will impact attritional losses and further augment property losses over the next two years at least. Hazard intensification will likely play a bigger contributory role to rising losses in the coming decade also. With the world getting warmer, findings from scientific research infer that climate change effects on loss frequency and severity will intensify.\n\nWe also expect, irrespective of below-average loss years, that annual insured losses will average more than USD 100 billion from hereon. Our expectation is supported by other parties: for example, Verisk recently modeled the global insured average annual loss as USD 123 billion. ¹⁴ In any one year losses can be higher or lower depending on whether natural catastrophe events do or do not strike urban and more populated areas. A case in point, had Hurricane Ian last year made landfall in the Tampa Bay area as many predictions were showing, the resulting losses would have been much higher. There is no reason to anticipate that this, nor peak-loss disasters like Hurricane Katrina in 2005, will not happen again in the future. The takeaway is to not underestimate loss potential on account of a year or period of below trend growth.\n\n14 _Global Modelled Catastrophe Losses_ , Verisk, 2022 _._\n\nWe expect that average insured losses from natural catastrophes will continue to grow by 5‒7% annually.\n\n**Figure 6** Growth in global natural catastrophe insured losses in USD billion (2022 prices)\n\nSource: Swiss Re Institute\n\n0\n\n25\n\n50\n\n75\n\n100\n\n125\n\n150\n\n175\n\n200\n\n2022 2020 2018 2016 2014 2012 2010 2008 2006 2004 2002 2000 1998 1996 1994 1992\n\nTrend Earthquakes Weather-related\n\nSince 2017, average annual insured losses have exceeded USD 110 billion.\n\nLosses of more that USD 100 billion annually are here to stay.","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"10 **Swiss Re Institute** _sigma_ 1/2023 In perspective: 2022 and longer-term trends\n\n### Up-trend reaffirmed, and set to continue\n\nInsured losses from natural catastrophes have been on a long-term upward trajectory for much longer than the last six years. Since 1992, the average annual trend growth of losses has been 5‒7% (see Figure 6). In the period 2012‒2016, losses were at a lower annual mean, but there has been a return to said trend growth over the last six years and we expect this to continue. We project that irrespective of year-on-year volatility, insured losses will continue to grow at trend, even when real-time amplifying factors such as high inflation subdue.\n\nStarting from 2017, average annual insured losses from natural catastrophes have been above USD 110 billion, more than double the average of USD 52 billion over the previous 5-year period. This marks a notable step-up in the scale of losses after the 2012‒2016 period of benign losses. Asset value accumulation in an area struck by an extreme weather or other natural catastrophe can spark heavy financial losses. Another factor has been changes in construction costs. Of late, aging infrastructure vulnerabilities and inflationary pressures have boosted repair costs. And, with demand-supply imbalances of materials and labour still in play in the post COVID-19 era, we expect construction costs to remain elevated into 2023, further inflating repair and reconstruction costs. This will impact attritional losses and further augment property losses over the next two years at least. Hazard intensification will likely play a bigger contributory role to rising losses in the coming decade also. With the world getting warmer, findings from scientific research infer that climate change effects on loss frequency and severity will intensify.\n\nWe also expect, irrespective of below-average loss years, that annual insured losses will average more than USD 100 billion from hereon. Our expectation is supported by other parties: for example, Verisk recently modeled the global insured average annual loss as USD 123 billion. ¹⁴ In any one year losses can be higher or lower depending on whether natural catastrophe events do or do not strike urban and more populated areas. A case in point, had Hurricane Ian last year made landfall in the Tampa Bay area as many predictions were showing, the resulting losses would have been much higher. There is no reason to anticipate that this, nor peak-loss disasters like Hurricane Katrina in 2005, will not happen again in the future. The takeaway is to not underestimate loss potential on account of a year or period of below trend growth.\n\n14 _Global Modelled Catastrophe Losses_ , Verisk, 2022 _._\n\nWe expect that average insured losses from natural catastrophes will continue to grow by 5‒7% annually.\n\n**Figure 6** Growth in global natural catastrophe insured losses in USD billion (2022 prices)\n\nSource: Swiss Re Institute\n\n0\n\n25\n\n50\n\n75\n\n100\n\n125\n\n150\n\n175\n\n200\n\n2022 2020 2018 2016 2014 2012 2010 2008 2006 2004 2002 2000 1998 1996 1994 1992\n\nTrend Earthquakes Weather-related\n\nSince 2017, average annual insured losses have exceeded USD 110 billion.\n\nLosses of more that USD 100 billion annually are here to stay.","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2023-05-sigma-01-english_p10.pdf"]},"response":{"errors":[],"extract_id":"extract_ea67c39f9e46fec7bbafb7c383b240b5","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["10 **Swiss Re Institute** _sigma_ 1/2023 In perspective: 2022 and longer-term trends\n\n### Up-trend reaffirmed, and set to continue\n\nInsured losses from natural catastrophes have been on a long-term upward trajectory for much longer than the last six years. Since 1992, the average annual trend growth of losses has been 5‒7% (see Figure 6). In the period 2012‒2016, losses were at a lower annual mean, but there has been a return to said trend growth over the last six years and we expect this to continue. We project that irrespective of year-on-year volatility, insured losses will continue to grow at trend, even when real-time amplifying factors such as high inflation subdue.\n\nStarting from 2017, average annual insured losses from natural catastrophes have been above USD 110 billion, more than double the average of USD 52 billion over the previous 5-year period. This marks a notable step-up in the scale of losses after the 2012‒2016 period of benign losses. Asset value accumulation in an area struck by an extreme weather or other natural catastrophe can spark heavy financial losses. Another factor has been changes in construction costs. Of late, aging infrastructure vulnerabilities and inflationary pressures have boosted repair costs. And, with demand-supply imbalances of materials and labour still in play in the post COVID-19 era, we expect construction costs to remain elevated into 2023, further inflating repair and reconstruction costs. This will impact attritional losses and further augment property losses over the next two years at least. Hazard intensification will likely play a bigger contributory role to rising losses in the coming decade also. With the world getting warmer, findings from scientific research infer that climate change effects on loss frequency and severity will intensify.\n\nWe also expect, irrespective of below-average loss years, that annual insured losses will average more than USD 100 billion from hereon. Our expectation is supported by other parties: for example, Verisk recently modeled the global insured average annual loss as USD 123 billion. ¹⁴ In any one year losses can be higher or lower depending on whether natural catastrophe events do or do not strike urban and more populated areas. A case in point, had Hurricane Ian last year made landfall in the Tampa Bay area as many predictions were showing, the resulting losses would have been much higher. There is no reason to anticipate that this, nor peak-loss disasters like Hurricane Katrina in 2005, will not happen again in the future. The takeaway is to not underestimate loss potential on account of a year or period of below trend growth.\n\n14 _Global Modelled Catastrophe Losses_ , Verisk, 2022 _._\n\nWe expect that average insured losses from natural catastrophes will continue to grow by 5‒7% annually.\n\n**Figure 6** Growth in global natural catastrophe insured losses in USD billion (2022 prices)\n\nSource: Swiss Re Institute\n\n0\n\n25\n\n50\n\n75\n\n100\n\n125\n\n150\n\n175\n\n200\n\n2022 2020 2018 2016 2014 2012 2010 2008 2006 2004 2002 2000 1998 1996 1994 1992\n\nTrend Earthquakes Weather-related\n\nSince 2017, average annual insured losses have exceeded USD 110 billion.\n\nLosses of more that USD 100 billion annually are here to stay."],"full_content":"10 **Swiss Re Institute** _sigma_ 1/2023 In perspective: 2022 and longer-term trends\n\n### Up-trend reaffirmed, and set to continue\n\nInsured losses from natural catastrophes have been on a long-term upward trajectory for much longer than the last six years. Since 1992, the average annual trend growth of losses has been 5‒7% (see Figure 6). In the period 2012‒2016, losses were at a lower annual mean, but there has been a return to said trend growth over the last six years and we expect this to continue. We project that irrespective of year-on-year volatility, insured losses will continue to grow at trend, even when real-time amplifying factors such as high inflation subdue.\n\nStarting from 2017, average annual insured losses from natural catastrophes have been above USD 110 billion, more than double the average of USD 52 billion over the previous 5-year period. This marks a notable step-up in the scale of losses after the 2012‒2016 period of benign losses. Asset value accumulation in an area struck by an extreme weather or other natural catastrophe can spark heavy financial losses. Another factor has been changes in construction costs. Of late, aging infrastructure vulnerabilities and inflationary pressures have boosted repair costs. And, with demand-supply imbalances of materials and labour still in play in the post COVID-19 era, we expect construction costs to remain elevated into 2023, further inflating repair and reconstruction costs. This will impact attritional losses and further augment property losses over the next two years at least. Hazard intensification will likely play a bigger contributory role to rising losses in the coming decade also. With the world getting warmer, findings from scientific research infer that climate change effects on loss frequency and severity will intensify.\n\nWe also expect, irrespective of below-average loss years, that annual insured losses will average more than USD 100 billion from hereon. Our expectation is supported by other parties: for example, Verisk recently modeled the global insured average annual loss as USD 123 billion. ¹⁴ In any one year losses can be higher or lower depending on whether natural catastrophe events do or do not strike urban and more populated areas. A case in point, had Hurricane Ian last year made landfall in the Tampa Bay area as many predictions were showing, the resulting losses would have been much higher. There is no reason to anticipate that this, nor peak-loss disasters like Hurricane Katrina in 2005, will not happen again in the future. The takeaway is to not underestimate loss potential on account of a year or period of below trend growth.\n\n14 _Global Modelled Catastrophe Losses_ , Verisk, 2022 _._\n\nWe expect that average insured losses from natural catastrophes will continue to grow by 5‒7% annually.\n\n**Figure 6** Growth in global natural catastrophe insured losses in USD billion (2022 prices)\n\nSource: Swiss Re Institute\n\n0\n\n25\n\n50\n\n75\n\n100\n\n125\n\n150\n\n175\n\n200\n\n2022 2020 2018 2016 2014 2012 2010 2008 2006 2004 2002 2000 1998 1996 1994 1992\n\nTrend Earthquakes Weather-related\n\nSince 2017, average annual insured losses have exceeded USD 110 billion.\n\nLosses of more that USD 100 billion annually are here to stay.","publish_date":null,"title":"Up-trend reaffirmed, and set to continue","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2023-05-sigma-01-english_p10.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2023-05-sigma-01-english_p10.pdf"},"request":{"example_id":"chart/2023-05-sigma-01-english_p10","product_type":"parse","source_file_path":"2023-05-sigma-01-english_p10.pdf"},"started_at":"2026-11-04T19:46:30.913281"} +{"completed_at":"2026-11-04T19:46:35.331898","latency_in_ms":3211,"output":{"example_id":"chart/2023-05-sigma-01-english_p23","grounded_pages":[],"layout_pages":[],"markdown":"Market dislocation: Hurricane Ian, inflation andinterest rates _sigma_ 1/2023 **Swiss Re Institute** 23\n\n**US property insurance exposure, claims and premiums** US property insurance business is under earnings pressure. Annual natural catastrophe property claims averaged USD 83 billion in 2017‒2022, a more-than-doubling of average annual pay outs in the previous six years. At the same time, the average share of natural catastrophe-related losses of all property claims rose from 20% to 35%, ³⁹ mainly driven by a step-up in annual loss severity since 2017. Since 2019, so too has been an increase in the replacement value of buildings and equipment sparked by the pandemic-induced surge in inflation. The replacement value of the net stock of private structures in the US was an estimated USD 53.5 trillion in 2022, ⁴⁰ up around 40% from 2019. Since 2011 the replacement cost of private structures has grown by 6% annually compared to 5% for property lines premiums (see Figure 15). The outcome is that even with market hardening in primary commercial property since 2019, premiums have lagged replacement cost increases.\n\nUnderwriting experience indicates that pricing has also lagged natural catastrophe losses and replacement values. Between 2011 and 2016, the combined ratios for US property lines averaged 91%, while from 2017‒2022 they averaged 105% (see Figure 16). Based on 2022 US property premiums of USD 200 billion, the profitability gap amounts to extra annual losses of nearly USD 30 billion, or 3‒5% of industry ROE. Returning to underwriting profitability in US property will require continued underwriting discipline alongside a reassessment of the underlying risks.\n\n39 We classify allied lines (including crop and flood), home- and farm-owners, commercial multi-peril (non- liability), ocean marine, inland marine, and motor physical damage as weather catastrophe-exposed lines.\n\n40 To estimate the replacement cost of structures at year-end 2022, we adjust the BEA current-cost net stock of private fixed assets at year-end 2021 (USD 47.2 trillion) by the YTD growth through 3Q22 of the current cost basis of residential and nonresidential structures owned by households and non-financial noncorporate business.\n\nThe pandemic-induced surge in inflation has increased replacement costs in US property.\n\n**Figure 15** Exposure and premium growth, US property, 2011 = 100\n\nSource: US Bureau of Economic Analysis, S&P Global Capital IQ Pro, Swiss Re Institute\n\n0\n\n50\n\n100\n\n150\n\n200\n\n2022 2021 2020 2019 2018 2017 2016 2015 2014 2013 2012 2011\n\nPersonal property (homeowner) premiums Commercial property (fire & allied) premiums\n\nReplacement cost of structures\n\nRe/insurance rate gains have lagged increases in insured losses and replacement costs.","pages":[{"line_numbers":[],"markdown":"Market dislocation: Hurricane Ian, inflation andinterest rates _sigma_ 1/2023 **Swiss Re Institute** 23\n\n**US property insurance exposure, claims and premiums** US property insurance business is under earnings pressure. Annual natural catastrophe property claims averaged USD 83 billion in 2017‒2022, a more-than-doubling of average annual pay outs in the previous six years. At the same time, the average share of natural catastrophe-related losses of all property claims rose from 20% to 35%, ³⁹ mainly driven by a step-up in annual loss severity since 2017. Since 2019, so too has been an increase in the replacement value of buildings and equipment sparked by the pandemic-induced surge in inflation. The replacement value of the net stock of private structures in the US was an estimated USD 53.5 trillion in 2022, ⁴⁰ up around 40% from 2019. Since 2011 the replacement cost of private structures has grown by 6% annually compared to 5% for property lines premiums (see Figure 15). The outcome is that even with market hardening in primary commercial property since 2019, premiums have lagged replacement cost increases.\n\nUnderwriting experience indicates that pricing has also lagged natural catastrophe losses and replacement values. Between 2011 and 2016, the combined ratios for US property lines averaged 91%, while from 2017‒2022 they averaged 105% (see Figure 16). Based on 2022 US property premiums of USD 200 billion, the profitability gap amounts to extra annual losses of nearly USD 30 billion, or 3‒5% of industry ROE. Returning to underwriting profitability in US property will require continued underwriting discipline alongside a reassessment of the underlying risks.\n\n39 We classify allied lines (including crop and flood), home- and farm-owners, commercial multi-peril (non- liability), ocean marine, inland marine, and motor physical damage as weather catastrophe-exposed lines.\n\n40 To estimate the replacement cost of structures at year-end 2022, we adjust the BEA current-cost net stock of private fixed assets at year-end 2021 (USD 47.2 trillion) by the YTD growth through 3Q22 of the current cost basis of residential and nonresidential structures owned by households and non-financial noncorporate business.\n\nThe pandemic-induced surge in inflation has increased replacement costs in US property.\n\n**Figure 15** Exposure and premium growth, US property, 2011 = 100\n\nSource: US Bureau of Economic Analysis, S&P Global Capital IQ Pro, Swiss Re Institute\n\n0\n\n50\n\n100\n\n150\n\n200\n\n2022 2021 2020 2019 2018 2017 2016 2015 2014 2013 2012 2011\n\nPersonal property (homeowner) premiums Commercial property (fire & allied) premiums\n\nReplacement cost of structures\n\nRe/insurance rate gains have lagged increases in insured losses and replacement costs.","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Market dislocation: Hurricane Ian, inflation andinterest rates _sigma_ 1/2023 **Swiss Re Institute** 23\n\n**US property insurance exposure, claims and premiums** US property insurance business is under earnings pressure. Annual natural catastrophe property claims averaged USD 83 billion in 2017‒2022, a more-than-doubling of average annual pay outs in the previous six years. At the same time, the average share of natural catastrophe-related losses of all property claims rose from 20% to 35%, ³⁹ mainly driven by a step-up in annual loss severity since 2017. Since 2019, so too has been an increase in the replacement value of buildings and equipment sparked by the pandemic-induced surge in inflation. The replacement value of the net stock of private structures in the US was an estimated USD 53.5 trillion in 2022, ⁴⁰ up around 40% from 2019. Since 2011 the replacement cost of private structures has grown by 6% annually compared to 5% for property lines premiums (see Figure 15). The outcome is that even with market hardening in primary commercial property since 2019, premiums have lagged replacement cost increases.\n\nUnderwriting experience indicates that pricing has also lagged natural catastrophe losses and replacement values. Between 2011 and 2016, the combined ratios for US property lines averaged 91%, while from 2017‒2022 they averaged 105% (see Figure 16). Based on 2022 US property premiums of USD 200 billion, the profitability gap amounts to extra annual losses of nearly USD 30 billion, or 3‒5% of industry ROE. Returning to underwriting profitability in US property will require continued underwriting discipline alongside a reassessment of the underlying risks.\n\n39 We classify allied lines (including crop and flood), home- and farm-owners, commercial multi-peril (non- liability), ocean marine, inland marine, and motor physical damage as weather catastrophe-exposed lines.\n\n40 To estimate the replacement cost of structures at year-end 2022, we adjust the BEA current-cost net stock of private fixed assets at year-end 2021 (USD 47.2 trillion) by the YTD growth through 3Q22 of the current cost basis of residential and nonresidential structures owned by households and non-financial noncorporate business.\n\nThe pandemic-induced surge in inflation has increased replacement costs in US property.\n\n**Figure 15** Exposure and premium growth, US property, 2011 = 100\n\nSource: US Bureau of Economic Analysis, S&P Global Capital IQ Pro, Swiss Re Institute\n\n0\n\n50\n\n100\n\n150\n\n200\n\n2022 2021 2020 2019 2018 2017 2016 2015 2014 2013 2012 2011\n\nPersonal property (homeowner) premiums Commercial property (fire & allied) premiums\n\nReplacement cost of structures\n\nRe/insurance rate gains have lagged increases in insured losses and replacement costs.","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2023-05-sigma-01-english_p23.pdf"]},"response":{"errors":[],"extract_id":"extract_a4b100e985f6c474de6e94dc340c751d","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Market dislocation: Hurricane Ian, inflation andinterest rates _sigma_ 1/2023 **Swiss Re Institute** 23\n\n**US property insurance exposure, claims and premiums** US property insurance business is under earnings pressure. Annual natural catastrophe property claims averaged USD 83 billion in 2017‒2022, a more-than-doubling of average annual pay outs in the previous six years. At the same time, the average share of natural catastrophe-related losses of all property claims rose from 20% to 35%, ³⁹ mainly driven by a step-up in annual loss severity since 2017. Since 2019, so too has been an increase in the replacement value of buildings and equipment sparked by the pandemic-induced surge in inflation. The replacement value of the net stock of private structures in the US was an estimated USD 53.5 trillion in 2022, ⁴⁰ up around 40% from 2019. Since 2011 the replacement cost of private structures has grown by 6% annually compared to 5% for property lines premiums (see Figure 15). The outcome is that even with market hardening in primary commercial property since 2019, premiums have lagged replacement cost increases.\n\nUnderwriting experience indicates that pricing has also lagged natural catastrophe losses and replacement values. Between 2011 and 2016, the combined ratios for US property lines averaged 91%, while from 2017‒2022 they averaged 105% (see Figure 16). Based on 2022 US property premiums of USD 200 billion, the profitability gap amounts to extra annual losses of nearly USD 30 billion, or 3‒5% of industry ROE. Returning to underwriting profitability in US property will require continued underwriting discipline alongside a reassessment of the underlying risks.\n\n39 We classify allied lines (including crop and flood), home- and farm-owners, commercial multi-peril (non- liability), ocean marine, inland marine, and motor physical damage as weather catastrophe-exposed lines.\n\n40 To estimate the replacement cost of structures at year-end 2022, we adjust the BEA current-cost net stock of private fixed assets at year-end 2021 (USD 47.2 trillion) by the YTD growth through 3Q22 of the current cost basis of residential and nonresidential structures owned by households and non-financial noncorporate business.\n\nThe pandemic-induced surge in inflation has increased replacement costs in US property.\n\n**Figure 15** Exposure and premium growth, US property, 2011 = 100\n\nSource: US Bureau of Economic Analysis, S&P Global Capital IQ Pro, Swiss Re Institute\n\n0\n\n50\n\n100\n\n150\n\n200\n\n2022 2021 2020 2019 2018 2017 2016 2015 2014 2013 2012 2011\n\nPersonal property (homeowner) premiums Commercial property (fire & allied) premiums\n\nReplacement cost of structures\n\nRe/insurance rate gains have lagged increases in insured losses and replacement costs."],"full_content":"Market dislocation: Hurricane Ian, inflation andinterest rates _sigma_ 1/2023 **Swiss Re Institute** 23\n\n**US property insurance exposure, claims and premiums** US property insurance business is under earnings pressure. Annual natural catastrophe property claims averaged USD 83 billion in 2017‒2022, a more-than-doubling of average annual pay outs in the previous six years. At the same time, the average share of natural catastrophe-related losses of all property claims rose from 20% to 35%, ³⁹ mainly driven by a step-up in annual loss severity since 2017. Since 2019, so too has been an increase in the replacement value of buildings and equipment sparked by the pandemic-induced surge in inflation. The replacement value of the net stock of private structures in the US was an estimated USD 53.5 trillion in 2022, ⁴⁰ up around 40% from 2019. Since 2011 the replacement cost of private structures has grown by 6% annually compared to 5% for property lines premiums (see Figure 15). The outcome is that even with market hardening in primary commercial property since 2019, premiums have lagged replacement cost increases.\n\nUnderwriting experience indicates that pricing has also lagged natural catastrophe losses and replacement values. Between 2011 and 2016, the combined ratios for US property lines averaged 91%, while from 2017‒2022 they averaged 105% (see Figure 16). Based on 2022 US property premiums of USD 200 billion, the profitability gap amounts to extra annual losses of nearly USD 30 billion, or 3‒5% of industry ROE. Returning to underwriting profitability in US property will require continued underwriting discipline alongside a reassessment of the underlying risks.\n\n39 We classify allied lines (including crop and flood), home- and farm-owners, commercial multi-peril (non- liability), ocean marine, inland marine, and motor physical damage as weather catastrophe-exposed lines.\n\n40 To estimate the replacement cost of structures at year-end 2022, we adjust the BEA current-cost net stock of private fixed assets at year-end 2021 (USD 47.2 trillion) by the YTD growth through 3Q22 of the current cost basis of residential and nonresidential structures owned by households and non-financial noncorporate business.\n\nThe pandemic-induced surge in inflation has increased replacement costs in US property.\n\n**Figure 15** Exposure and premium growth, US property, 2011 = 100\n\nSource: US Bureau of Economic Analysis, S&P Global Capital IQ Pro, Swiss Re Institute\n\n0\n\n50\n\n100\n\n150\n\n200\n\n2022 2021 2020 2019 2018 2017 2016 2015 2014 2013 2012 2011\n\nPersonal property (homeowner) premiums Commercial property (fire & allied) premiums\n\nReplacement cost of structures\n\nRe/insurance rate gains have lagged increases in insured losses and replacement costs.","publish_date":null,"title":"US property insurance exposure, claims and premiums US property insurance business is under earnings pressure. Annual natural catastrophe property claims","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2023-05-sigma-01-english_p23.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2023-05-sigma-01-english_p23.pdf"},"request":{"example_id":"chart/2023-05-sigma-01-english_p23","product_type":"parse","source_file_path":"2023-05-sigma-01-english_p23.pdf"},"started_at":"2026-11-04T19:46:32.120759"} +{"completed_at":"2026-11-04T19:46:36.201240","latency_in_ms":1923,"output":{"example_id":"chart/2023-05-sigma-01-english_p24","grounded_pages":[],"layout_pages":[],"markdown":"24 **Swiss Re Institute** _sigma_ 1/2023 Market dislocation: Hurricane Ian, inflation andinterest rates\n\nHigh economic inflation has impacted exposures and demand for coverage directly. The supply-side impact has been indirect. Rising prices led to decisive monetary policy action by the US Federal Reserve (Fed) and many other central banks. In 2022, the Fed, Bank of England (BoE), European Central Bank (ECB) and other central banks raised short-term policy rates from zero (or near-zero) to 4.375% (Fed), 3.5% (BoE) and 2.5% (ECB). Long-term interest rates also moved up considerably.\n\nOne effect of the higher interest rates has been a decline in financial asset values and more specifically shareholder equity. This has had immediate impact on re/insurers, whose fixed income portfolios have suffered significant mark-to-market losses. The global bond index was down 16% in 2022, and global equity markets also declined (the S&P Dow Jones Broad Market Index was down 18% year-on-year). The combined effect on reinsurer balance sheets – where invested assets are typically 3-4 times equity – was significant. By the end of 2022, reinsurance capital (traditional and alternative) had declined by around 20‒25% from year-end 2021. After adjusting for the interest rate\n\n**Figure 16** Combined ratios, US property lines (Fire, Allied, Homeowners), 2011‒2022 (est.)\n\nSource: S&P Global Capital IQ Pro, Swiss Re Institute\n\n0\n\n20\n\n40\n\n60\n\n80\n\n100\n\n120\n\n2021 2022 (est.) 2020 2019 2018 2017 2016 2015 2014 2013 2012 2011\n\nDirect average: 2011–2016, 2017–2022 Direct Net\n\nHigh inflation has sparked interest rates hikes...\n\n**Figure 17** Inflation and interest rates in the US and Germany, 2019 – 2022, monthly data\n\nSource: Swiss Re Institute\n\n0%\n\n2%\n\n4%\n\n6%\n\n8%\n\n10%\n\n2022:12\n\n2022:09\n\n2022:06\n\n2022:03\n\n2021:12\n\n2021:09\n\n2021:06\n\n2021:03\n\n2020:12\n\n2020:09\n\n2020:06\n\n2020:03\n\n2019:12\n\n2019:09\n\n2019:06\n\n2019:03\n\n2018:12\n\n**US**\n\n–2%\n\n0%\n\n2%\n\n4%\n\n6%\n\n8%\n\n10%\n\n12%\n\n2022:12\n\n2022:09\n\n2022:06\n\n2022:03\n\n2021:12\n\n2021:09\n\n2021:06\n\n2021:03\n\n2020:12\n\n2020:09\n\n2020:06\n\n2020:03\n\n2019:12\n\n2019:09\n\n2019:06\n\n2019:03\n\n2018:12\n\n**Germany**\n\nInflation Short-term interest rate Long-term interest rate Inflation Short-term interest rate Long-term interest rate\n\n…materially affecting re/insurance capital by decreasing asset valuations.","pages":[{"line_numbers":[],"markdown":"24 **Swiss Re Institute** _sigma_ 1/2023 Market dislocation: Hurricane Ian, inflation andinterest rates\n\nHigh economic inflation has impacted exposures and demand for coverage directly. The supply-side impact has been indirect. Rising prices led to decisive monetary policy action by the US Federal Reserve (Fed) and many other central banks. In 2022, the Fed, Bank of England (BoE), European Central Bank (ECB) and other central banks raised short-term policy rates from zero (or near-zero) to 4.375% (Fed), 3.5% (BoE) and 2.5% (ECB). Long-term interest rates also moved up considerably.\n\nOne effect of the higher interest rates has been a decline in financial asset values and more specifically shareholder equity. This has had immediate impact on re/insurers, whose fixed income portfolios have suffered significant mark-to-market losses. The global bond index was down 16% in 2022, and global equity markets also declined (the S&P Dow Jones Broad Market Index was down 18% year-on-year). The combined effect on reinsurer balance sheets – where invested assets are typically 3-4 times equity – was significant. By the end of 2022, reinsurance capital (traditional and alternative) had declined by around 20‒25% from year-end 2021. After adjusting for the interest rate\n\n**Figure 16** Combined ratios, US property lines (Fire, Allied, Homeowners), 2011‒2022 (est.)\n\nSource: S&P Global Capital IQ Pro, Swiss Re Institute\n\n0\n\n20\n\n40\n\n60\n\n80\n\n100\n\n120\n\n2021 2022 (est.) 2020 2019 2018 2017 2016 2015 2014 2013 2012 2011\n\nDirect average: 2011–2016, 2017–2022 Direct Net\n\nHigh inflation has sparked interest rates hikes...\n\n**Figure 17** Inflation and interest rates in the US and Germany, 2019 – 2022, monthly data\n\nSource: Swiss Re Institute\n\n0%\n\n2%\n\n4%\n\n6%\n\n8%\n\n10%\n\n2022:12\n\n2022:09\n\n2022:06\n\n2022:03\n\n2021:12\n\n2021:09\n\n2021:06\n\n2021:03\n\n2020:12\n\n2020:09\n\n2020:06\n\n2020:03\n\n2019:12\n\n2019:09\n\n2019:06\n\n2019:03\n\n2018:12\n\n**US**\n\n–2%\n\n0%\n\n2%\n\n4%\n\n6%\n\n8%\n\n10%\n\n12%\n\n2022:12\n\n2022:09\n\n2022:06\n\n2022:03\n\n2021:12\n\n2021:09\n\n2021:06\n\n2021:03\n\n2020:12\n\n2020:09\n\n2020:06\n\n2020:03\n\n2019:12\n\n2019:09\n\n2019:06\n\n2019:03\n\n2018:12\n\n**Germany**\n\nInflation Short-term interest rate Long-term interest rate Inflation Short-term interest rate Long-term interest rate\n\n…materially affecting re/insurance capital by decreasing asset valuations.","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"24 **Swiss Re Institute** _sigma_ 1/2023 Market dislocation: Hurricane Ian, inflation andinterest rates\n\nHigh economic inflation has impacted exposures and demand for coverage directly. The supply-side impact has been indirect. Rising prices led to decisive monetary policy action by the US Federal Reserve (Fed) and many other central banks. In 2022, the Fed, Bank of England (BoE), European Central Bank (ECB) and other central banks raised short-term policy rates from zero (or near-zero) to 4.375% (Fed), 3.5% (BoE) and 2.5% (ECB). Long-term interest rates also moved up considerably.\n\nOne effect of the higher interest rates has been a decline in financial asset values and more specifically shareholder equity. This has had immediate impact on re/insurers, whose fixed income portfolios have suffered significant mark-to-market losses. The global bond index was down 16% in 2022, and global equity markets also declined (the S&P Dow Jones Broad Market Index was down 18% year-on-year). The combined effect on reinsurer balance sheets – where invested assets are typically 3-4 times equity – was significant. By the end of 2022, reinsurance capital (traditional and alternative) had declined by around 20‒25% from year-end 2021. After adjusting for the interest rate\n\n**Figure 16** Combined ratios, US property lines (Fire, Allied, Homeowners), 2011‒2022 (est.)\n\nSource: S&P Global Capital IQ Pro, Swiss Re Institute\n\n0\n\n20\n\n40\n\n60\n\n80\n\n100\n\n120\n\n2021 2022 (est.) 2020 2019 2018 2017 2016 2015 2014 2013 2012 2011\n\nDirect average: 2011–2016, 2017–2022 Direct Net\n\nHigh inflation has sparked interest rates hikes...\n\n**Figure 17** Inflation and interest rates in the US and Germany, 2019 – 2022, monthly data\n\nSource: Swiss Re Institute\n\n0%\n\n2%\n\n4%\n\n6%\n\n8%\n\n10%\n\n2022:12\n\n2022:09\n\n2022:06\n\n2022:03\n\n2021:12\n\n2021:09\n\n2021:06\n\n2021:03\n\n2020:12\n\n2020:09\n\n2020:06\n\n2020:03\n\n2019:12\n\n2019:09\n\n2019:06\n\n2019:03\n\n2018:12\n\n**US**\n\n–2%\n\n0%\n\n2%\n\n4%\n\n6%\n\n8%\n\n10%\n\n12%\n\n2022:12\n\n2022:09\n\n2022:06\n\n2022:03\n\n2021:12\n\n2021:09\n\n2021:06\n\n2021:03\n\n2020:12\n\n2020:09\n\n2020:06\n\n2020:03\n\n2019:12\n\n2019:09\n\n2019:06\n\n2019:03\n\n2018:12\n\n**Germany**\n\nInflation Short-term interest rate Long-term interest rate Inflation Short-term interest rate Long-term interest rate\n\n…materially affecting re/insurance capital by decreasing asset valuations.","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2023-05-sigma-01-english_p24.pdf"]},"response":{"errors":[],"extract_id":"extract_e148356859b73628cfe53e1b2ca82c01","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["24 **Swiss Re Institute** _sigma_ 1/2023 Market dislocation: Hurricane Ian, inflation andinterest rates\n\nHigh economic inflation has impacted exposures and demand for coverage directly. The supply-side impact has been indirect. Rising prices led to decisive monetary policy action by the US Federal Reserve (Fed) and many other central banks. In 2022, the Fed, Bank of England (BoE), European Central Bank (ECB) and other central banks raised short-term policy rates from zero (or near-zero) to 4.375% (Fed), 3.5% (BoE) and 2.5% (ECB). Long-term interest rates also moved up considerably.\n\nOne effect of the higher interest rates has been a decline in financial asset values and more specifically shareholder equity. This has had immediate impact on re/insurers, whose fixed income portfolios have suffered significant mark-to-market losses. The global bond index was down 16% in 2022, and global equity markets also declined (the S&P Dow Jones Broad Market Index was down 18% year-on-year). The combined effect on reinsurer balance sheets – where invested assets are typically 3-4 times equity – was significant. By the end of 2022, reinsurance capital (traditional and alternative) had declined by around 20‒25% from year-end 2021. After adjusting for the interest rate\n\n**Figure 16** Combined ratios, US property lines (Fire, Allied, Homeowners), 2011‒2022 (est.)\n\nSource: S&P Global Capital IQ Pro, Swiss Re Institute\n\n0\n\n20\n\n40\n\n60\n\n80\n\n100\n\n120\n\n2021 2022 (est.) 2020 2019 2018 2017 2016 2015 2014 2013 2012 2011\n\nDirect average: 2011–2016, 2017–2022 Direct Net\n\nHigh inflation has sparked interest rates hikes...\n\n**Figure 17** Inflation and interest rates in the US and Germany, 2019 – 2022, monthly data\n\nSource: Swiss Re Institute\n\n0%\n\n2%\n\n4%\n\n6%\n\n8%\n\n10%\n\n2022:12\n\n2022:09\n\n2022:06\n\n2022:03\n\n2021:12\n\n2021:09\n\n2021:06\n\n2021:03\n\n2020:12\n\n2020:09\n\n2020:06\n\n2020:03\n\n2019:12\n\n2019:09\n\n2019:06\n\n2019:03\n\n2018:12\n\n**US**\n\n–2%\n\n0%\n\n2%\n\n4%\n\n6%\n\n8%\n\n10%\n\n12%\n\n2022:12\n\n2022:09\n\n2022:06\n\n2022:03\n\n2021:12\n\n2021:09\n\n2021:06\n\n2021:03\n\n2020:12\n\n2020:09\n\n2020:06\n\n2020:03\n\n2019:12\n\n2019:09\n\n2019:06\n\n2019:03\n\n2018:12\n\n**Germany**\n\nInflation Short-term interest rate Long-term interest rate Inflation Short-term interest rate Long-term interest rate\n\n…materially affecting re/insurance capital by decreasing asset valuations."],"full_content":"24 **Swiss Re Institute** _sigma_ 1/2023 Market dislocation: Hurricane Ian, inflation andinterest rates\n\nHigh economic inflation has impacted exposures and demand for coverage directly. The supply-side impact has been indirect. Rising prices led to decisive monetary policy action by the US Federal Reserve (Fed) and many other central banks. In 2022, the Fed, Bank of England (BoE), European Central Bank (ECB) and other central banks raised short-term policy rates from zero (or near-zero) to 4.375% (Fed), 3.5% (BoE) and 2.5% (ECB). Long-term interest rates also moved up considerably.\n\nOne effect of the higher interest rates has been a decline in financial asset values and more specifically shareholder equity. This has had immediate impact on re/insurers, whose fixed income portfolios have suffered significant mark-to-market losses. The global bond index was down 16% in 2022, and global equity markets also declined (the S&P Dow Jones Broad Market Index was down 18% year-on-year). The combined effect on reinsurer balance sheets – where invested assets are typically 3-4 times equity – was significant. By the end of 2022, reinsurance capital (traditional and alternative) had declined by around 20‒25% from year-end 2021. After adjusting for the interest rate\n\n**Figure 16** Combined ratios, US property lines (Fire, Allied, Homeowners), 2011‒2022 (est.)\n\nSource: S&P Global Capital IQ Pro, Swiss Re Institute\n\n0\n\n20\n\n40\n\n60\n\n80\n\n100\n\n120\n\n2021 2022 (est.) 2020 2019 2018 2017 2016 2015 2014 2013 2012 2011\n\nDirect average: 2011–2016, 2017–2022 Direct Net\n\nHigh inflation has sparked interest rates hikes...\n\n**Figure 17** Inflation and interest rates in the US and Germany, 2019 – 2022, monthly data\n\nSource: Swiss Re Institute\n\n0%\n\n2%\n\n4%\n\n6%\n\n8%\n\n10%\n\n2022:12\n\n2022:09\n\n2022:06\n\n2022:03\n\n2021:12\n\n2021:09\n\n2021:06\n\n2021:03\n\n2020:12\n\n2020:09\n\n2020:06\n\n2020:03\n\n2019:12\n\n2019:09\n\n2019:06\n\n2019:03\n\n2018:12\n\n**US**\n\n–2%\n\n0%\n\n2%\n\n4%\n\n6%\n\n8%\n\n10%\n\n12%\n\n2022:12\n\n2022:09\n\n2022:06\n\n2022:03\n\n2021:12\n\n2021:09\n\n2021:06\n\n2021:03\n\n2020:12\n\n2020:09\n\n2020:06\n\n2020:03\n\n2019:12\n\n2019:09\n\n2019:06\n\n2019:03\n\n2018:12\n\n**Germany**\n\nInflation Short-term interest rate Long-term interest rate Inflation Short-term interest rate Long-term interest rate\n\n…materially affecting re/insurance capital by decreasing asset valuations.","publish_date":null,"title":"High economic inflation has impacted exposures and demand for coverage directly. The supply-side impact has been indirect. Rising prices led","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2023-05-sigma-01-english_p24.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2023-05-sigma-01-english_p24.pdf"},"request":{"example_id":"chart/2023-05-sigma-01-english_p24","product_type":"parse","source_file_path":"2023-05-sigma-01-english_p24.pdf"},"started_at":"2026-11-04T19:46:34.277615"} +{"completed_at":"2026-11-04T19:46:38.325487","latency_in_ms":2526,"output":{"example_id":"chart/2023-05-sigma-01-english_p27","grounded_pages":[],"layout_pages":[],"markdown":"_sigma_ 1/2023 **Swiss Re Institute** 27\n\n### Facts and figures\n\n**Number of catastrophic events: 285** Based on _sigma_ criteria, there were 285 catastrophes worldwide in 2022, down from 306 in 2021. There were 187 natural catastrophes, up from the 186 in 2021, and 98 man-made disasters (down from 120 in 2021).\n\n**Number of victims: more than 35 000** Worldwide, 35 157 people are believed to have died or gone missing in disaster events in 2022. Natural catastrophes claimed over 32 600 victims, and man-made disasters over 2500.\n\n# Appendix 1: 2022 – the year in review\n\n**Figure 19** Number of catastrophic events, 1970‒2022\n\nSource: Swiss Re Institute\n\n0\n\n50\n\n100\n\n150\n\n200\n\n250\n\n300\n\n2021 2018 2015 2012 2009 2006 2003 2000 1997 1994 1991 1988 1985 1982 1979 1976 1973 1970\n\nMan-made disasters Natural catastrophes\n\n**Figure 20** Number of victims, 1970‒2022\n\nNote: Scale is logarithmic: the number of victims increases tenfold per band. Source: Swiss Re Institute\n\n1 000\n\n10 000\n\n100 000\n\n1 000 000\n\n10 000 000\n\n2021 2018 2015 2012 2009 2006 2003 2000 1997 1994 1991 1988 1985 1982 1979 1976 1973 1970\n\nMan-made disasters Natural catastrophes\n\n1991: Cyclone Gorky, Bangladesh\n\n1970: Bangladesh Storm\n\n2004: Indian Ocean Earthquake and tsunami\n\n2010: Haiti Earthquake\n\n2008: Cyclone Nargis, Myanmar\n\n2013: Typhoon Haiyan, Philippines\n\n2015: Earthquake in Nepal\n\n1976: Tangshan Earthquake, China\n\n2022: Heatwaves in Europe","pages":[{"line_numbers":[],"markdown":"_sigma_ 1/2023 **Swiss Re Institute** 27\n\n### Facts and figures\n\n**Number of catastrophic events: 285** Based on _sigma_ criteria, there were 285 catastrophes worldwide in 2022, down from 306 in 2021. There were 187 natural catastrophes, up from the 186 in 2021, and 98 man-made disasters (down from 120 in 2021).\n\n**Number of victims: more than 35 000** Worldwide, 35 157 people are believed to have died or gone missing in disaster events in 2022. Natural catastrophes claimed over 32 600 victims, and man-made disasters over 2500.\n\n# Appendix 1: 2022 – the year in review\n\n**Figure 19** Number of catastrophic events, 1970‒2022\n\nSource: Swiss Re Institute\n\n0\n\n50\n\n100\n\n150\n\n200\n\n250\n\n300\n\n2021 2018 2015 2012 2009 2006 2003 2000 1997 1994 1991 1988 1985 1982 1979 1976 1973 1970\n\nMan-made disasters Natural catastrophes\n\n**Figure 20** Number of victims, 1970‒2022\n\nNote: Scale is logarithmic: the number of victims increases tenfold per band. Source: Swiss Re Institute\n\n1 000\n\n10 000\n\n100 000\n\n1 000 000\n\n10 000 000\n\n2021 2018 2015 2012 2009 2006 2003 2000 1997 1994 1991 1988 1985 1982 1979 1976 1973 1970\n\nMan-made disasters Natural catastrophes\n\n1991: Cyclone Gorky, Bangladesh\n\n1970: Bangladesh Storm\n\n2004: Indian Ocean Earthquake and tsunami\n\n2010: Haiti Earthquake\n\n2008: Cyclone Nargis, Myanmar\n\n2013: Typhoon Haiyan, Philippines\n\n2015: Earthquake in Nepal\n\n1976: Tangshan Earthquake, China\n\n2022: Heatwaves in Europe","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"_sigma_ 1/2023 **Swiss Re Institute** 27\n\n### Facts and figures\n\n**Number of catastrophic events: 285** Based on _sigma_ criteria, there were 285 catastrophes worldwide in 2022, down from 306 in 2021. There were 187 natural catastrophes, up from the 186 in 2021, and 98 man-made disasters (down from 120 in 2021).\n\n**Number of victims: more than 35 000** Worldwide, 35 157 people are believed to have died or gone missing in disaster events in 2022. Natural catastrophes claimed over 32 600 victims, and man-made disasters over 2500.\n\n# Appendix 1: 2022 – the year in review\n\n**Figure 19** Number of catastrophic events, 1970‒2022\n\nSource: Swiss Re Institute\n\n0\n\n50\n\n100\n\n150\n\n200\n\n250\n\n300\n\n2021 2018 2015 2012 2009 2006 2003 2000 1997 1994 1991 1988 1985 1982 1979 1976 1973 1970\n\nMan-made disasters Natural catastrophes\n\n**Figure 20** Number of victims, 1970‒2022\n\nNote: Scale is logarithmic: the number of victims increases tenfold per band. Source: Swiss Re Institute\n\n1 000\n\n10 000\n\n100 000\n\n1 000 000\n\n10 000 000\n\n2021 2018 2015 2012 2009 2006 2003 2000 1997 1994 1991 1988 1985 1982 1979 1976 1973 1970\n\nMan-made disasters Natural catastrophes\n\n1991: Cyclone Gorky, Bangladesh\n\n1970: Bangladesh Storm\n\n2004: Indian Ocean Earthquake and tsunami\n\n2010: Haiti Earthquake\n\n2008: Cyclone Nargis, Myanmar\n\n2013: Typhoon Haiyan, Philippines\n\n2015: Earthquake in Nepal\n\n1976: Tangshan Earthquake, China\n\n2022: Heatwaves in Europe","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2023-05-sigma-01-english_p27.pdf"]},"response":{"errors":[],"extract_id":"extract_0338be6a373e77cd7ca649cfdda3cf2b","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["_sigma_ 1/2023 **Swiss Re Institute** 27\n\n### Facts and figures\n\n**Number of catastrophic events: 285** Based on _sigma_ criteria, there were 285 catastrophes worldwide in 2022, down from 306 in 2021. There were 187 natural catastrophes, up from the 186 in 2021, and 98 man-made disasters (down from 120 in 2021).\n\n**Number of victims: more than 35 000** Worldwide, 35 157 people are believed to have died or gone missing in disaster events in 2022. Natural catastrophes claimed over 32 600 victims, and man-made disasters over 2500.\n\n# Appendix 1: 2022 – the year in review\n\n**Figure 19** Number of catastrophic events, 1970‒2022\n\nSource: Swiss Re Institute\n\n0\n\n50\n\n100\n\n150\n\n200\n\n250\n\n300\n\n2021 2018 2015 2012 2009 2006 2003 2000 1997 1994 1991 1988 1985 1982 1979 1976 1973 1970\n\nMan-made disasters Natural catastrophes\n\n**Figure 20** Number of victims, 1970‒2022\n\nNote: Scale is logarithmic: the number of victims increases tenfold per band. Source: Swiss Re Institute\n\n1 000\n\n10 000\n\n100 000\n\n1 000 000\n\n10 000 000\n\n2021 2018 2015 2012 2009 2006 2003 2000 1997 1994 1991 1988 1985 1982 1979 1976 1973 1970\n\nMan-made disasters Natural catastrophes\n\n1991: Cyclone Gorky, Bangladesh\n\n1970: Bangladesh Storm\n\n2004: Indian Ocean Earthquake and tsunami\n\n2010: Haiti Earthquake\n\n2008: Cyclone Nargis, Myanmar\n\n2013: Typhoon Haiyan, Philippines\n\n2015: Earthquake in Nepal\n\n1976: Tangshan Earthquake, China\n\n2022: Heatwaves in Europe"],"full_content":"_sigma_ 1/2023 **Swiss Re Institute** 27\n\n### Facts and figures\n\n**Number of catastrophic events: 285** Based on _sigma_ criteria, there were 285 catastrophes worldwide in 2022, down from 306 in 2021. There were 187 natural catastrophes, up from the 186 in 2021, and 98 man-made disasters (down from 120 in 2021).\n\n**Number of victims: more than 35 000** Worldwide, 35 157 people are believed to have died or gone missing in disaster events in 2022. Natural catastrophes claimed over 32 600 victims, and man-made disasters over 2500.\n\n# Appendix 1: 2022 – the year in review\n\n**Figure 19** Number of catastrophic events, 1970‒2022\n\nSource: Swiss Re Institute\n\n0\n\n50\n\n100\n\n150\n\n200\n\n250\n\n300\n\n2021 2018 2015 2012 2009 2006 2003 2000 1997 1994 1991 1988 1985 1982 1979 1976 1973 1970\n\nMan-made disasters Natural catastrophes\n\n**Figure 20** Number of victims, 1970‒2022\n\nNote: Scale is logarithmic: the number of victims increases tenfold per band. Source: Swiss Re Institute\n\n1 000\n\n10 000\n\n100 000\n\n1 000 000\n\n10 000 000\n\n2021 2018 2015 2012 2009 2006 2003 2000 1997 1994 1991 1988 1985 1982 1979 1976 1973 1970\n\nMan-made disasters Natural catastrophes\n\n1991: Cyclone Gorky, Bangladesh\n\n1970: Bangladesh Storm\n\n2004: Indian Ocean Earthquake and tsunami\n\n2010: Haiti Earthquake\n\n2008: Cyclone Nargis, Myanmar\n\n2013: Typhoon Haiyan, Philippines\n\n2015: Earthquake in Nepal\n\n1976: Tangshan Earthquake, China\n\n2022: Heatwaves in Europe","publish_date":null,"title":"Appendix 1: 2022 – the year in review","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2023-05-sigma-01-english_p27.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2023-05-sigma-01-english_p27.pdf"},"request":{"example_id":"chart/2023-05-sigma-01-english_p27","product_type":"parse","source_file_path":"2023-05-sigma-01-english_p27.pdf"},"started_at":"2026-11-04T19:46:35.798634"} +{"completed_at":"2026-11-04T19:46:38.740961","latency_in_ms":1336,"output":{"example_id":"chart/2023-05-sigma-01-english_p28","grounded_pages":[],"layout_pages":[],"markdown":"28 **Swiss Re Institute** _sigma_ 1/2023 Appendix 1: 2022 – the year in review\n\n**Primary and secondary perils** Hurricane Ian brought the share of insured losses from primary perils to 57% in 2022, from a previous 10-year average of 37%.\n\n**Figure 21** Insured catastrophe losses, 1970–2022, in USD billion at 2022 prices\n\nSource: Swiss Re Institute\n\n0\n\n20\n\n40\n\n60\n\n80\n\n100\n\n120\n\n140\n\n160\n\n180\n\n200\n\n2021 2018 2015 2012 2009 2006 2003 2000 1997 1994 1991 1988 1985 1982 1979 1976 1973 1970\n\nEarthquake/tsunami Weather-related catastrophes Man-made disasters\n\nHurricanes Katrina, Rita, Wilma\n\nHurricanes Ike, Gustav\n\nJapan, NZ earthquakes, Thailand flood\n\nHurricane Sandy\n\nHurricanes Harvey, Irma, Maria\n\nHurricane Ian\n\nHurricanes Ivan, Charley, Frances\n\nHurricane Andrew\n\nWinter Storm Lothar WTC\n\nHurricane Ida\n\n**Figure 22** Global insured losses from primary and secondary perils in USD billion at 2022 prices\n\nSource: Swiss Re Institute\n\n0\n\n25\n\n50\n\n75\n\n100\n\n125\n\n150\n\n175\n\n2021 2018 2015 2012 2009 2006 2003 2000 1997 1994 1991 1988 1985 1982 1979 1976 1973 1970\n\nPrimary perils (incl secondary effects) Secondary","pages":[{"line_numbers":[],"markdown":"28 **Swiss Re Institute** _sigma_ 1/2023 Appendix 1: 2022 – the year in review\n\n**Primary and secondary perils** Hurricane Ian brought the share of insured losses from primary perils to 57% in 2022, from a previous 10-year average of 37%.\n\n**Figure 21** Insured catastrophe losses, 1970–2022, in USD billion at 2022 prices\n\nSource: Swiss Re Institute\n\n0\n\n20\n\n40\n\n60\n\n80\n\n100\n\n120\n\n140\n\n160\n\n180\n\n200\n\n2021 2018 2015 2012 2009 2006 2003 2000 1997 1994 1991 1988 1985 1982 1979 1976 1973 1970\n\nEarthquake/tsunami Weather-related catastrophes Man-made disasters\n\nHurricanes Katrina, Rita, Wilma\n\nHurricanes Ike, Gustav\n\nJapan, NZ earthquakes, Thailand flood\n\nHurricane Sandy\n\nHurricanes Harvey, Irma, Maria\n\nHurricane Ian\n\nHurricanes Ivan, Charley, Frances\n\nHurricane Andrew\n\nWinter Storm Lothar WTC\n\nHurricane Ida\n\n**Figure 22** Global insured losses from primary and secondary perils in USD billion at 2022 prices\n\nSource: Swiss Re Institute\n\n0\n\n25\n\n50\n\n75\n\n100\n\n125\n\n150\n\n175\n\n2021 2018 2015 2012 2009 2006 2003 2000 1997 1994 1991 1988 1985 1982 1979 1976 1973 1970\n\nPrimary perils (incl secondary effects) Secondary","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"28 **Swiss Re Institute** _sigma_ 1/2023 Appendix 1: 2022 – the year in review\n\n**Primary and secondary perils** Hurricane Ian brought the share of insured losses from primary perils to 57% in 2022, from a previous 10-year average of 37%.\n\n**Figure 21** Insured catastrophe losses, 1970–2022, in USD billion at 2022 prices\n\nSource: Swiss Re Institute\n\n0\n\n20\n\n40\n\n60\n\n80\n\n100\n\n120\n\n140\n\n160\n\n180\n\n200\n\n2021 2018 2015 2012 2009 2006 2003 2000 1997 1994 1991 1988 1985 1982 1979 1976 1973 1970\n\nEarthquake/tsunami Weather-related catastrophes Man-made disasters\n\nHurricanes Katrina, Rita, Wilma\n\nHurricanes Ike, Gustav\n\nJapan, NZ earthquakes, Thailand flood\n\nHurricane Sandy\n\nHurricanes Harvey, Irma, Maria\n\nHurricane Ian\n\nHurricanes Ivan, Charley, Frances\n\nHurricane Andrew\n\nWinter Storm Lothar WTC\n\nHurricane Ida\n\n**Figure 22** Global insured losses from primary and secondary perils in USD billion at 2022 prices\n\nSource: Swiss Re Institute\n\n0\n\n25\n\n50\n\n75\n\n100\n\n125\n\n150\n\n175\n\n2021 2018 2015 2012 2009 2006 2003 2000 1997 1994 1991 1988 1985 1982 1979 1976 1973 1970\n\nPrimary perils (incl secondary effects) Secondary","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2023-05-sigma-01-english_p28.pdf"]},"response":{"errors":[],"extract_id":"extract_09e61b4e1a1334aa734280df22062de2","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["28 **Swiss Re Institute** _sigma_ 1/2023 Appendix 1: 2022 – the year in review\n\n**Primary and secondary perils** Hurricane Ian brought the share of insured losses from primary perils to 57% in 2022, from a previous 10-year average of 37%.\n\n**Figure 21** Insured catastrophe losses, 1970–2022, in USD billion at 2022 prices\n\nSource: Swiss Re Institute\n\n0\n\n20\n\n40\n\n60\n\n80\n\n100\n\n120\n\n140\n\n160\n\n180\n\n200\n\n2021 2018 2015 2012 2009 2006 2003 2000 1997 1994 1991 1988 1985 1982 1979 1976 1973 1970\n\nEarthquake/tsunami Weather-related catastrophes Man-made disasters\n\nHurricanes Katrina, Rita, Wilma\n\nHurricanes Ike, Gustav\n\nJapan, NZ earthquakes, Thailand flood\n\nHurricane Sandy\n\nHurricanes Harvey, Irma, Maria\n\nHurricane Ian\n\nHurricanes Ivan, Charley, Frances\n\nHurricane Andrew\n\nWinter Storm Lothar WTC\n\nHurricane Ida\n\n**Figure 22** Global insured losses from primary and secondary perils in USD billion at 2022 prices\n\nSource: Swiss Re Institute\n\n0\n\n25\n\n50\n\n75\n\n100\n\n125\n\n150\n\n175\n\n2021 2018 2015 2012 2009 2006 2003 2000 1997 1994 1991 1988 1985 1982 1979 1976 1973 1970\n\nPrimary perils (incl secondary effects) Secondary"],"full_content":"28 **Swiss Re Institute** _sigma_ 1/2023 Appendix 1: 2022 – the year in review\n\n**Primary and secondary perils** Hurricane Ian brought the share of insured losses from primary perils to 57% in 2022, from a previous 10-year average of 37%.\n\n**Figure 21** Insured catastrophe losses, 1970–2022, in USD billion at 2022 prices\n\nSource: Swiss Re Institute\n\n0\n\n20\n\n40\n\n60\n\n80\n\n100\n\n120\n\n140\n\n160\n\n180\n\n200\n\n2021 2018 2015 2012 2009 2006 2003 2000 1997 1994 1991 1988 1985 1982 1979 1976 1973 1970\n\nEarthquake/tsunami Weather-related catastrophes Man-made disasters\n\nHurricanes Katrina, Rita, Wilma\n\nHurricanes Ike, Gustav\n\nJapan, NZ earthquakes, Thailand flood\n\nHurricane Sandy\n\nHurricanes Harvey, Irma, Maria\n\nHurricane Ian\n\nHurricanes Ivan, Charley, Frances\n\nHurricane Andrew\n\nWinter Storm Lothar WTC\n\nHurricane Ida\n\n**Figure 22** Global insured losses from primary and secondary perils in USD billion at 2022 prices\n\nSource: Swiss Re Institute\n\n0\n\n25\n\n50\n\n75\n\n100\n\n125\n\n150\n\n175\n\n2021 2018 2015 2012 2009 2006 2003 2000 1997 1994 1991 1988 1985 1982 1979 1976 1973 1970\n\nPrimary perils (incl secondary effects) Secondary","publish_date":null,"title":"Primary and secondary perils Hurricane Ian brought the share of insured losses from primary perils to 57% in 2022, from","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2023-05-sigma-01-english_p28.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2023-05-sigma-01-english_p28.pdf"},"request":{"example_id":"chart/2023-05-sigma-01-english_p28","product_type":"parse","source_file_path":"2023-05-sigma-01-english_p28.pdf"},"started_at":"2026-11-04T19:46:37.404190"} +{"completed_at":"2026-11-04T19:46:40.177113","latency_in_ms":1432,"output":{"example_id":"chart/2023-05-sigma-01-english_p6","grounded_pages":[],"layout_pages":[],"markdown":"6 **Swiss Re Institute** _sigma_ 1/2023 In perspective: 2022 and longer-term trends\n\nAt over USD 33 billion, global insured losses from severe convective storms (SCS) were also above prior-period averages (see Figure 2), driven by thunderstorms with hail and tornadoes in the US. Another contributor was the highest-ever annual loss (USD 5 billion) from hailstorms in France. Global losses from floods were just above average, the main event being flooding in eastern Australia in February-March. The flooding resulted in insured losses of USD 4.3 billion, the biggest natural catastrophe claims event ever in Australia. Another national “costliest ever” event was flooding in Durban in South Africa in April, leading to estimated insured losses of USD 1.5 billion.\n\nAfter a few years with low loss-severity winter storms, in February 2022 a cluster of storms (Eunice, Dudley, Franklin) in northwestern Europe triggered combined claims of an estimated USD 4.1 billion, bringing the total for this category to almost double the previous 10-year average. Winds in winter storms are less severe than in tropical cyclones, but large parts of Europe can be impacted by a single storm, and damage in different locations can see combined losses accumulate to multi-billion levels.\n\nAt the opposite end of the temperature spectrum, weather variability and anomalous atmospheric circulation conditions caused severe drought and heatwaves across the world. Heat and drought impacted crop yields in many regions, adding to global food inflation pressures and elevated agriculture insurance losses. Table 1 shows the crop loss impacts in US dollar terms in select markets. In Europe, the summer of 2022 was the hottest on record. ² , ³ In Morocco, heat and dry conditions ⁴ coincided with a North Atlantic Oscillation phase of rainfall deficit. ⁵ In Brazil, monsoon rains were below average. ⁶ Crop yields, particularly for soybean and maize, suffered most. ⁷ Total precipitation across contiguous US in 2022 made the year the third driest on record, ⁸ and crops yield were lower than in 2021. ⁹ And in China, extreme heat and dry conditions in the Yangtze River and weak monsoon rains impacted summer crops. ¹⁰\n\n2 _Summer 2022 Europe’s hottest on record_ , Copernicus, 8 September 2022.\n\n3 _Trockenheit in Europa 2022_ , Deutscher Wetterdienst, July 2022\n\n4 _Drought in western Mediterranean February 2022_ , European Commission, 22 March 2022.\n\n5 See _Climate Prediction Center_ , National Weather Service.\n\n6 See “Phases of the South American Monsoon System”, _Climate Prediction Center_ , Natl Weather Service\n\n7 See _Crop Explorer - World Agricultural Production Briefs: Brazil_ , Foreign Agriculture Service, USDA.\n\n8 _Record drought gripped much of the US in 2022_ , National Oceanic and Atmospheric Administration, 10 January 2023.\n\n9 _Corn and soybean production down in 2022, USDA reports Corn stocks down, soybean stocks down from year earlier Winter Wheat Seedings up for 2023_ , USDA, 10 January 2023.\n\n10 “Scientific interpretation of severe drought in the Yangtze River Basin”, _Journal of Arid Meteorology_ , 2022.\n\nFlood and hailstorm events across the world also resulted in significant losses.\n\n**Figure 2** Global insured losses from natural catastrophes in 2022 by category, in USD billion at 2022 prices\n\nSource: Swiss Re Institute\n\n0 20 40 60 80 100 120 140\n\n2022\n\n5-year average (2017–2021)\n\n10-year average (2012–2021)\n\n15-year average (2007–2021)\n\n30-year average (1992–2021)\n\n14%\n\n54%\n\n62%\n\n115%\n\nEarthquakes Tropical cyclones Floods Severe convective storms Wildfires\n\nEuropean winter storms Other secondary perils Droughts\n\n2022 insured loss over previous year averages\n\nLosses from winter storms in Europe were above average.\n\nHeatwaves and droughts led to crop yield losses in many regions.","pages":[{"line_numbers":[],"markdown":"6 **Swiss Re Institute** _sigma_ 1/2023 In perspective: 2022 and longer-term trends\n\nAt over USD 33 billion, global insured losses from severe convective storms (SCS) were also above prior-period averages (see Figure 2), driven by thunderstorms with hail and tornadoes in the US. Another contributor was the highest-ever annual loss (USD 5 billion) from hailstorms in France. Global losses from floods were just above average, the main event being flooding in eastern Australia in February-March. The flooding resulted in insured losses of USD 4.3 billion, the biggest natural catastrophe claims event ever in Australia. Another national “costliest ever” event was flooding in Durban in South Africa in April, leading to estimated insured losses of USD 1.5 billion.\n\nAfter a few years with low loss-severity winter storms, in February 2022 a cluster of storms (Eunice, Dudley, Franklin) in northwestern Europe triggered combined claims of an estimated USD 4.1 billion, bringing the total for this category to almost double the previous 10-year average. Winds in winter storms are less severe than in tropical cyclones, but large parts of Europe can be impacted by a single storm, and damage in different locations can see combined losses accumulate to multi-billion levels.\n\nAt the opposite end of the temperature spectrum, weather variability and anomalous atmospheric circulation conditions caused severe drought and heatwaves across the world. Heat and drought impacted crop yields in many regions, adding to global food inflation pressures and elevated agriculture insurance losses. Table 1 shows the crop loss impacts in US dollar terms in select markets. In Europe, the summer of 2022 was the hottest on record. ² , ³ In Morocco, heat and dry conditions ⁴ coincided with a North Atlantic Oscillation phase of rainfall deficit. ⁵ In Brazil, monsoon rains were below average. ⁶ Crop yields, particularly for soybean and maize, suffered most. ⁷ Total precipitation across contiguous US in 2022 made the year the third driest on record, ⁸ and crops yield were lower than in 2021. ⁹ And in China, extreme heat and dry conditions in the Yangtze River and weak monsoon rains impacted summer crops. ¹⁰\n\n2 _Summer 2022 Europe’s hottest on record_ , Copernicus, 8 September 2022.\n\n3 _Trockenheit in Europa 2022_ , Deutscher Wetterdienst, July 2022\n\n4 _Drought in western Mediterranean February 2022_ , European Commission, 22 March 2022.\n\n5 See _Climate Prediction Center_ , National Weather Service.\n\n6 See “Phases of the South American Monsoon System”, _Climate Prediction Center_ , Natl Weather Service\n\n7 See _Crop Explorer - World Agricultural Production Briefs: Brazil_ , Foreign Agriculture Service, USDA.\n\n8 _Record drought gripped much of the US in 2022_ , National Oceanic and Atmospheric Administration, 10 January 2023.\n\n9 _Corn and soybean production down in 2022, USDA reports Corn stocks down, soybean stocks down from year earlier Winter Wheat Seedings up for 2023_ , USDA, 10 January 2023.\n\n10 “Scientific interpretation of severe drought in the Yangtze River Basin”, _Journal of Arid Meteorology_ , 2022.\n\nFlood and hailstorm events across the world also resulted in significant losses.\n\n**Figure 2** Global insured losses from natural catastrophes in 2022 by category, in USD billion at 2022 prices\n\nSource: Swiss Re Institute\n\n0 20 40 60 80 100 120 140\n\n2022\n\n5-year average (2017–2021)\n\n10-year average (2012–2021)\n\n15-year average (2007–2021)\n\n30-year average (1992–2021)\n\n14%\n\n54%\n\n62%\n\n115%\n\nEarthquakes Tropical cyclones Floods Severe convective storms Wildfires\n\nEuropean winter storms Other secondary perils Droughts\n\n2022 insured loss over previous year averages\n\nLosses from winter storms in Europe were above average.\n\nHeatwaves and droughts led to crop yield losses in many regions.","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"6 **Swiss Re Institute** _sigma_ 1/2023 In perspective: 2022 and longer-term trends\n\nAt over USD 33 billion, global insured losses from severe convective storms (SCS) were also above prior-period averages (see Figure 2), driven by thunderstorms with hail and tornadoes in the US. Another contributor was the highest-ever annual loss (USD 5 billion) from hailstorms in France. Global losses from floods were just above average, the main event being flooding in eastern Australia in February-March. The flooding resulted in insured losses of USD 4.3 billion, the biggest natural catastrophe claims event ever in Australia. Another national “costliest ever” event was flooding in Durban in South Africa in April, leading to estimated insured losses of USD 1.5 billion.\n\nAfter a few years with low loss-severity winter storms, in February 2022 a cluster of storms (Eunice, Dudley, Franklin) in northwestern Europe triggered combined claims of an estimated USD 4.1 billion, bringing the total for this category to almost double the previous 10-year average. Winds in winter storms are less severe than in tropical cyclones, but large parts of Europe can be impacted by a single storm, and damage in different locations can see combined losses accumulate to multi-billion levels.\n\nAt the opposite end of the temperature spectrum, weather variability and anomalous atmospheric circulation conditions caused severe drought and heatwaves across the world. Heat and drought impacted crop yields in many regions, adding to global food inflation pressures and elevated agriculture insurance losses. Table 1 shows the crop loss impacts in US dollar terms in select markets. In Europe, the summer of 2022 was the hottest on record. ² , ³ In Morocco, heat and dry conditions ⁴ coincided with a North Atlantic Oscillation phase of rainfall deficit. ⁵ In Brazil, monsoon rains were below average. ⁶ Crop yields, particularly for soybean and maize, suffered most. ⁷ Total precipitation across contiguous US in 2022 made the year the third driest on record, ⁸ and crops yield were lower than in 2021. ⁹ And in China, extreme heat and dry conditions in the Yangtze River and weak monsoon rains impacted summer crops. ¹⁰\n\n2 _Summer 2022 Europe’s hottest on record_ , Copernicus, 8 September 2022.\n\n3 _Trockenheit in Europa 2022_ , Deutscher Wetterdienst, July 2022\n\n4 _Drought in western Mediterranean February 2022_ , European Commission, 22 March 2022.\n\n5 See _Climate Prediction Center_ , National Weather Service.\n\n6 See “Phases of the South American Monsoon System”, _Climate Prediction Center_ , Natl Weather Service\n\n7 See _Crop Explorer - World Agricultural Production Briefs: Brazil_ , Foreign Agriculture Service, USDA.\n\n8 _Record drought gripped much of the US in 2022_ , National Oceanic and Atmospheric Administration, 10 January 2023.\n\n9 _Corn and soybean production down in 2022, USDA reports Corn stocks down, soybean stocks down from year earlier Winter Wheat Seedings up for 2023_ , USDA, 10 January 2023.\n\n10 “Scientific interpretation of severe drought in the Yangtze River Basin”, _Journal of Arid Meteorology_ , 2022.\n\nFlood and hailstorm events across the world also resulted in significant losses.\n\n**Figure 2** Global insured losses from natural catastrophes in 2022 by category, in USD billion at 2022 prices\n\nSource: Swiss Re Institute\n\n0 20 40 60 80 100 120 140\n\n2022\n\n5-year average (2017–2021)\n\n10-year average (2012–2021)\n\n15-year average (2007–2021)\n\n30-year average (1992–2021)\n\n14%\n\n54%\n\n62%\n\n115%\n\nEarthquakes Tropical cyclones Floods Severe convective storms Wildfires\n\nEuropean winter storms Other secondary perils Droughts\n\n2022 insured loss over previous year averages\n\nLosses from winter storms in Europe were above average.\n\nHeatwaves and droughts led to crop yield losses in many regions.","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2023-05-sigma-01-english_p6.pdf"]},"response":{"errors":[],"extract_id":"extract_d87e71ab3feb2a470ee3ff0afd031130","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["6 **Swiss Re Institute** _sigma_ 1/2023 In perspective: 2022 and longer-term trends\n\nAt over USD 33 billion, global insured losses from severe convective storms (SCS) were also above prior-period averages (see Figure 2), driven by thunderstorms with hail and tornadoes in the US. Another contributor was the highest-ever annual loss (USD 5 billion) from hailstorms in France. Global losses from floods were just above average, the main event being flooding in eastern Australia in February-March. The flooding resulted in insured losses of USD 4.3 billion, the biggest natural catastrophe claims event ever in Australia. Another national “costliest ever” event was flooding in Durban in South Africa in April, leading to estimated insured losses of USD 1.5 billion.\n\nAfter a few years with low loss-severity winter storms, in February 2022 a cluster of storms (Eunice, Dudley, Franklin) in northwestern Europe triggered combined claims of an estimated USD 4.1 billion, bringing the total for this category to almost double the previous 10-year average. Winds in winter storms are less severe than in tropical cyclones, but large parts of Europe can be impacted by a single storm, and damage in different locations can see combined losses accumulate to multi-billion levels.\n\nAt the opposite end of the temperature spectrum, weather variability and anomalous atmospheric circulation conditions caused severe drought and heatwaves across the world. Heat and drought impacted crop yields in many regions, adding to global food inflation pressures and elevated agriculture insurance losses. Table 1 shows the crop loss impacts in US dollar terms in select markets. In Europe, the summer of 2022 was the hottest on record. ² , ³ In Morocco, heat and dry conditions ⁴ coincided with a North Atlantic Oscillation phase of rainfall deficit. ⁵ In Brazil, monsoon rains were below average. ⁶ Crop yields, particularly for soybean and maize, suffered most. ⁷ Total precipitation across contiguous US in 2022 made the year the third driest on record, ⁸ and crops yield were lower than in 2021. ⁹ And in China, extreme heat and dry conditions in the Yangtze River and weak monsoon rains impacted summer crops. ¹⁰\n\n2 _Summer 2022 Europe’s hottest on record_ , Copernicus, 8 September 2022.\n\n3 _Trockenheit in Europa 2022_ , Deutscher Wetterdienst, July 2022\n\n4 _Drought in western Mediterranean February 2022_ , European Commission, 22 March 2022.\n\n5 See _Climate Prediction Center_ , National Weather Service.\n\n6 See “Phases of the South American Monsoon System”, _Climate Prediction Center_ , Natl Weather Service\n\n7 See _Crop Explorer - World Agricultural Production Briefs: Brazil_ , Foreign Agriculture Service, USDA.\n\n8 _Record drought gripped much of the US in 2022_ , National Oceanic and Atmospheric Administration, 10 January 2023.\n\n9 _Corn and soybean production down in 2022, USDA reports Corn stocks down, soybean stocks down from year earlier Winter Wheat Seedings up for 2023_ , USDA, 10 January 2023.\n\n10 “Scientific interpretation of severe drought in the Yangtze River Basin”, _Journal of Arid Meteorology_ , 2022.\n\nFlood and hailstorm events across the world also resulted in significant losses.\n\n**Figure 2** Global insured losses from natural catastrophes in 2022 by category, in USD billion at 2022 prices\n\nSource: Swiss Re Institute\n\n0 20 40 60 80 100 120 140\n\n2022\n\n5-year average (2017–2021)\n\n10-year average (2012–2021)\n\n15-year average (2007–2021)\n\n30-year average (1992–2021)\n\n14%\n\n54%\n\n62%\n\n115%\n\nEarthquakes Tropical cyclones Floods Severe convective storms Wildfires\n\nEuropean winter storms Other secondary perils Droughts\n\n2022 insured loss over previous year averages\n\nLosses from winter storms in Europe were above average.\n\nHeatwaves and droughts led to crop yield losses in many regions."],"full_content":"6 **Swiss Re Institute** _sigma_ 1/2023 In perspective: 2022 and longer-term trends\n\nAt over USD 33 billion, global insured losses from severe convective storms (SCS) were also above prior-period averages (see Figure 2), driven by thunderstorms with hail and tornadoes in the US. Another contributor was the highest-ever annual loss (USD 5 billion) from hailstorms in France. Global losses from floods were just above average, the main event being flooding in eastern Australia in February-March. The flooding resulted in insured losses of USD 4.3 billion, the biggest natural catastrophe claims event ever in Australia. Another national “costliest ever” event was flooding in Durban in South Africa in April, leading to estimated insured losses of USD 1.5 billion.\n\nAfter a few years with low loss-severity winter storms, in February 2022 a cluster of storms (Eunice, Dudley, Franklin) in northwestern Europe triggered combined claims of an estimated USD 4.1 billion, bringing the total for this category to almost double the previous 10-year average. Winds in winter storms are less severe than in tropical cyclones, but large parts of Europe can be impacted by a single storm, and damage in different locations can see combined losses accumulate to multi-billion levels.\n\nAt the opposite end of the temperature spectrum, weather variability and anomalous atmospheric circulation conditions caused severe drought and heatwaves across the world. Heat and drought impacted crop yields in many regions, adding to global food inflation pressures and elevated agriculture insurance losses. Table 1 shows the crop loss impacts in US dollar terms in select markets. In Europe, the summer of 2022 was the hottest on record. ² , ³ In Morocco, heat and dry conditions ⁴ coincided with a North Atlantic Oscillation phase of rainfall deficit. ⁵ In Brazil, monsoon rains were below average. ⁶ Crop yields, particularly for soybean and maize, suffered most. ⁷ Total precipitation across contiguous US in 2022 made the year the third driest on record, ⁸ and crops yield were lower than in 2021. ⁹ And in China, extreme heat and dry conditions in the Yangtze River and weak monsoon rains impacted summer crops. ¹⁰\n\n2 _Summer 2022 Europe’s hottest on record_ , Copernicus, 8 September 2022.\n\n3 _Trockenheit in Europa 2022_ , Deutscher Wetterdienst, July 2022\n\n4 _Drought in western Mediterranean February 2022_ , European Commission, 22 March 2022.\n\n5 See _Climate Prediction Center_ , National Weather Service.\n\n6 See “Phases of the South American Monsoon System”, _Climate Prediction Center_ , Natl Weather Service\n\n7 See _Crop Explorer - World Agricultural Production Briefs: Brazil_ , Foreign Agriculture Service, USDA.\n\n8 _Record drought gripped much of the US in 2022_ , National Oceanic and Atmospheric Administration, 10 January 2023.\n\n9 _Corn and soybean production down in 2022, USDA reports Corn stocks down, soybean stocks down from year earlier Winter Wheat Seedings up for 2023_ , USDA, 10 January 2023.\n\n10 “Scientific interpretation of severe drought in the Yangtze River Basin”, _Journal of Arid Meteorology_ , 2022.\n\nFlood and hailstorm events across the world also resulted in significant losses.\n\n**Figure 2** Global insured losses from natural catastrophes in 2022 by category, in USD billion at 2022 prices\n\nSource: Swiss Re Institute\n\n0 20 40 60 80 100 120 140\n\n2022\n\n5-year average (2017–2021)\n\n10-year average (2012–2021)\n\n15-year average (2007–2021)\n\n30-year average (1992–2021)\n\n14%\n\n54%\n\n62%\n\n115%\n\nEarthquakes Tropical cyclones Floods Severe convective storms Wildfires\n\nEuropean winter storms Other secondary perils Droughts\n\n2022 insured loss over previous year averages\n\nLosses from winter storms in Europe were above average.\n\nHeatwaves and droughts led to crop yield losses in many regions.","publish_date":null,"title":"At over USD 33 billion, global insured losses from severe convective storms (SCS) were also above prior-period averages (see Figure","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2023-05-sigma-01-english_p6.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2023-05-sigma-01-english_p6.pdf"},"request":{"example_id":"chart/2023-05-sigma-01-english_p6","product_type":"parse","source_file_path":"2023-05-sigma-01-english_p6.pdf"},"started_at":"2026-11-04T19:46:38.744555"} +{"completed_at":"2026-11-04T19:46:43.916432","latency_in_ms":3585,"output":{"example_id":"chart/2023-05-sigma-01-english_p7","grounded_pages":[],"layout_pages":[],"markdown":"In perspective: 2022 and longer-term trends _sigma_ 1/2023 **Swiss Re Institute** 7\n\nWaves of high temperatures in Europe also inflicted a heavy human toll: according to the WHO, a series heatwaves in Europe caused at least an estimated 15 000 excess deaths. ¹¹ In spite of the heatwaves, one peril for which losses were below recent-year totals and historical averages were those for wildfire.\n\n### Longer-term loss trends\n\nLosses from individual perils fluctuate year on year. That said, at on average 30% each, the contribution to global insured losses from the two biggest peril categories – tropical cyclones and SCS – have remained largely stable over the last 40 years (see Figure 3). On the east coast of the US, hurricanes (primary perils) originating in the North Atlantic are a main threat to residents and businesses. ¹² Though rare, when a major hurricane strikes, the losses can be very severe. As in the case of Hurricane Ian, just one peak tropical cyclone event can wreak very large losses.\n\nSCS are categorised as secondary perils, occur more frequently and happen all over the world. Given the frequency, aggregated their annual loss amounts are less volatile. ¹³ Typically, losses resulting from SCS are lower than for primary perils, but there have been instances when a single SCS has resulted in insured losses of similar size to those coming in the wake of medium-sized hurricane. Noteworthy too is a marked step up in the share of all SCS to insured losses in the last decade.\n\n11 _Statement – Climate change is already killing us, but strong action now can prevent more deaths_ , World Health Organization, 2 November 2022.\n\n12 Primary perils are natural catastrophes that tend to happen less frequently, but have high loss potential. They can include secondary effects. Examples include tropical cyclones, earthquakes and European winter storms.\n\n13 Secondary perils are independent natural catastrophe events that can happen frequently, typically generating low- to medium-sized losses. Examples include severe convective storms (thunderstorms, hail and tornadoes), drought, wildfire, snow, flash floods and landslides.\n\n**Table 1** Insured crop losses due to drought in select markets, USD billion\n\nSource: CAN, PSR, Ministry of Emergency Management, Swiss Re\n\n**Country Economic losses Insured losses**\n\nBrazil 13 1\n\nEurope 6.2 0.6\n\nChina 4.7 0.8\n\nMorocco 0.25 0.04\n\nWildfire-related losses were less than in recent years.\n\nAt 30% each, the contributions of tropical cyclones…\n\n…and SCS to annual insured losses have remained largely stable.\n\n**Figure 3** Share of insured losses by peril type by decade\n\nSource: Swiss Re Institute\n\n0%\n\n5%\n\n10%\n\n15%\n\n20%\n\n25%\n\n30%\n\n35%\n\n40%\n\nOthers Droughts European winter storms Wildfires Severe convective storms Floods Tropical cyclones Earthquakes\n\n3%\n\n7%\n\n1%\n\n7%\n\n31%\n\n13%\n\n29%\n\n8%\n\n1983–1992 1993–2002 2003–2012 2013–2022 40-year average (1983–2022)","pages":[{"line_numbers":[],"markdown":"In perspective: 2022 and longer-term trends _sigma_ 1/2023 **Swiss Re Institute** 7\n\nWaves of high temperatures in Europe also inflicted a heavy human toll: according to the WHO, a series heatwaves in Europe caused at least an estimated 15 000 excess deaths. ¹¹ In spite of the heatwaves, one peril for which losses were below recent-year totals and historical averages were those for wildfire.\n\n### Longer-term loss trends\n\nLosses from individual perils fluctuate year on year. That said, at on average 30% each, the contribution to global insured losses from the two biggest peril categories – tropical cyclones and SCS – have remained largely stable over the last 40 years (see Figure 3). On the east coast of the US, hurricanes (primary perils) originating in the North Atlantic are a main threat to residents and businesses. ¹² Though rare, when a major hurricane strikes, the losses can be very severe. As in the case of Hurricane Ian, just one peak tropical cyclone event can wreak very large losses.\n\nSCS are categorised as secondary perils, occur more frequently and happen all over the world. Given the frequency, aggregated their annual loss amounts are less volatile. ¹³ Typically, losses resulting from SCS are lower than for primary perils, but there have been instances when a single SCS has resulted in insured losses of similar size to those coming in the wake of medium-sized hurricane. Noteworthy too is a marked step up in the share of all SCS to insured losses in the last decade.\n\n11 _Statement – Climate change is already killing us, but strong action now can prevent more deaths_ , World Health Organization, 2 November 2022.\n\n12 Primary perils are natural catastrophes that tend to happen less frequently, but have high loss potential. They can include secondary effects. Examples include tropical cyclones, earthquakes and European winter storms.\n\n13 Secondary perils are independent natural catastrophe events that can happen frequently, typically generating low- to medium-sized losses. Examples include severe convective storms (thunderstorms, hail and tornadoes), drought, wildfire, snow, flash floods and landslides.\n\n**Table 1** Insured crop losses due to drought in select markets, USD billion\n\nSource: CAN, PSR, Ministry of Emergency Management, Swiss Re\n\n**Country Economic losses Insured losses**\n\nBrazil 13 1\n\nEurope 6.2 0.6\n\nChina 4.7 0.8\n\nMorocco 0.25 0.04\n\nWildfire-related losses were less than in recent years.\n\nAt 30% each, the contributions of tropical cyclones…\n\n…and SCS to annual insured losses have remained largely stable.\n\n**Figure 3** Share of insured losses by peril type by decade\n\nSource: Swiss Re Institute\n\n0%\n\n5%\n\n10%\n\n15%\n\n20%\n\n25%\n\n30%\n\n35%\n\n40%\n\nOthers Droughts European winter storms Wildfires Severe convective storms Floods Tropical cyclones Earthquakes\n\n3%\n\n7%\n\n1%\n\n7%\n\n31%\n\n13%\n\n29%\n\n8%\n\n1983–1992 1993–2002 2003–2012 2013–2022 40-year average (1983–2022)","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"In perspective: 2022 and longer-term trends _sigma_ 1/2023 **Swiss Re Institute** 7\n\nWaves of high temperatures in Europe also inflicted a heavy human toll: according to the WHO, a series heatwaves in Europe caused at least an estimated 15 000 excess deaths. ¹¹ In spite of the heatwaves, one peril for which losses were below recent-year totals and historical averages were those for wildfire.\n\n### Longer-term loss trends\n\nLosses from individual perils fluctuate year on year. That said, at on average 30% each, the contribution to global insured losses from the two biggest peril categories – tropical cyclones and SCS – have remained largely stable over the last 40 years (see Figure 3). On the east coast of the US, hurricanes (primary perils) originating in the North Atlantic are a main threat to residents and businesses. ¹² Though rare, when a major hurricane strikes, the losses can be very severe. As in the case of Hurricane Ian, just one peak tropical cyclone event can wreak very large losses.\n\nSCS are categorised as secondary perils, occur more frequently and happen all over the world. Given the frequency, aggregated their annual loss amounts are less volatile. ¹³ Typically, losses resulting from SCS are lower than for primary perils, but there have been instances when a single SCS has resulted in insured losses of similar size to those coming in the wake of medium-sized hurricane. Noteworthy too is a marked step up in the share of all SCS to insured losses in the last decade.\n\n11 _Statement – Climate change is already killing us, but strong action now can prevent more deaths_ , World Health Organization, 2 November 2022.\n\n12 Primary perils are natural catastrophes that tend to happen less frequently, but have high loss potential. They can include secondary effects. Examples include tropical cyclones, earthquakes and European winter storms.\n\n13 Secondary perils are independent natural catastrophe events that can happen frequently, typically generating low- to medium-sized losses. Examples include severe convective storms (thunderstorms, hail and tornadoes), drought, wildfire, snow, flash floods and landslides.\n\n**Table 1** Insured crop losses due to drought in select markets, USD billion\n\nSource: CAN, PSR, Ministry of Emergency Management, Swiss Re\n\n**Country Economic losses Insured losses**\n\nBrazil 13 1\n\nEurope 6.2 0.6\n\nChina 4.7 0.8\n\nMorocco 0.25 0.04\n\nWildfire-related losses were less than in recent years.\n\nAt 30% each, the contributions of tropical cyclones…\n\n…and SCS to annual insured losses have remained largely stable.\n\n**Figure 3** Share of insured losses by peril type by decade\n\nSource: Swiss Re Institute\n\n0%\n\n5%\n\n10%\n\n15%\n\n20%\n\n25%\n\n30%\n\n35%\n\n40%\n\nOthers Droughts European winter storms Wildfires Severe convective storms Floods Tropical cyclones Earthquakes\n\n3%\n\n7%\n\n1%\n\n7%\n\n31%\n\n13%\n\n29%\n\n8%\n\n1983–1992 1993–2002 2003–2012 2013–2022 40-year average (1983–2022)","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2023-05-sigma-01-english_p7.pdf"]},"response":{"errors":[],"extract_id":"extract_6ab31d9deb1d34d5d6d3a89cfdf3a721","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["In perspective: 2022 and longer-term trends _sigma_ 1/2023 **Swiss Re Institute** 7\n\nWaves of high temperatures in Europe also inflicted a heavy human toll: according to the WHO, a series heatwaves in Europe caused at least an estimated 15 000 excess deaths. ¹¹ In spite of the heatwaves, one peril for which losses were below recent-year totals and historical averages were those for wildfire.\n\n### Longer-term loss trends\n\nLosses from individual perils fluctuate year on year. That said, at on average 30% each, the contribution to global insured losses from the two biggest peril categories – tropical cyclones and SCS – have remained largely stable over the last 40 years (see Figure 3). On the east coast of the US, hurricanes (primary perils) originating in the North Atlantic are a main threat to residents and businesses. ¹² Though rare, when a major hurricane strikes, the losses can be very severe. As in the case of Hurricane Ian, just one peak tropical cyclone event can wreak very large losses.\n\nSCS are categorised as secondary perils, occur more frequently and happen all over the world. Given the frequency, aggregated their annual loss amounts are less volatile. ¹³ Typically, losses resulting from SCS are lower than for primary perils, but there have been instances when a single SCS has resulted in insured losses of similar size to those coming in the wake of medium-sized hurricane. Noteworthy too is a marked step up in the share of all SCS to insured losses in the last decade.\n\n11 _Statement – Climate change is already killing us, but strong action now can prevent more deaths_ , World Health Organization, 2 November 2022.\n\n12 Primary perils are natural catastrophes that tend to happen less frequently, but have high loss potential. They can include secondary effects. Examples include tropical cyclones, earthquakes and European winter storms.\n\n13 Secondary perils are independent natural catastrophe events that can happen frequently, typically generating low- to medium-sized losses. Examples include severe convective storms (thunderstorms, hail and tornadoes), drought, wildfire, snow, flash floods and landslides.\n\n**Table 1** Insured crop losses due to drought in select markets, USD billion\n\nSource: CAN, PSR, Ministry of Emergency Management, Swiss Re\n\n**Country Economic losses Insured losses**\n\nBrazil 13 1\n\nEurope 6.2 0.6\n\nChina 4.7 0.8\n\nMorocco 0.25 0.04\n\nWildfire-related losses were less than in recent years.\n\nAt 30% each, the contributions of tropical cyclones…\n\n…and SCS to annual insured losses have remained largely stable.\n\n**Figure 3** Share of insured losses by peril type by decade\n\nSource: Swiss Re Institute\n\n0%\n\n5%\n\n10%\n\n15%\n\n20%\n\n25%\n\n30%\n\n35%\n\n40%\n\nOthers Droughts European winter storms Wildfires Severe convective storms Floods Tropical cyclones Earthquakes\n\n3%\n\n7%\n\n1%\n\n7%\n\n31%\n\n13%\n\n29%\n\n8%\n\n1983–1992 1993–2002 2003–2012 2013–2022 40-year average (1983–2022)"],"full_content":"In perspective: 2022 and longer-term trends _sigma_ 1/2023 **Swiss Re Institute** 7\n\nWaves of high temperatures in Europe also inflicted a heavy human toll: according to the WHO, a series heatwaves in Europe caused at least an estimated 15 000 excess deaths. ¹¹ In spite of the heatwaves, one peril for which losses were below recent-year totals and historical averages were those for wildfire.\n\n### Longer-term loss trends\n\nLosses from individual perils fluctuate year on year. That said, at on average 30% each, the contribution to global insured losses from the two biggest peril categories – tropical cyclones and SCS – have remained largely stable over the last 40 years (see Figure 3). On the east coast of the US, hurricanes (primary perils) originating in the North Atlantic are a main threat to residents and businesses. ¹² Though rare, when a major hurricane strikes, the losses can be very severe. As in the case of Hurricane Ian, just one peak tropical cyclone event can wreak very large losses.\n\nSCS are categorised as secondary perils, occur more frequently and happen all over the world. Given the frequency, aggregated their annual loss amounts are less volatile. ¹³ Typically, losses resulting from SCS are lower than for primary perils, but there have been instances when a single SCS has resulted in insured losses of similar size to those coming in the wake of medium-sized hurricane. Noteworthy too is a marked step up in the share of all SCS to insured losses in the last decade.\n\n11 _Statement – Climate change is already killing us, but strong action now can prevent more deaths_ , World Health Organization, 2 November 2022.\n\n12 Primary perils are natural catastrophes that tend to happen less frequently, but have high loss potential. They can include secondary effects. Examples include tropical cyclones, earthquakes and European winter storms.\n\n13 Secondary perils are independent natural catastrophe events that can happen frequently, typically generating low- to medium-sized losses. Examples include severe convective storms (thunderstorms, hail and tornadoes), drought, wildfire, snow, flash floods and landslides.\n\n**Table 1** Insured crop losses due to drought in select markets, USD billion\n\nSource: CAN, PSR, Ministry of Emergency Management, Swiss Re\n\n**Country Economic losses Insured losses**\n\nBrazil 13 1\n\nEurope 6.2 0.6\n\nChina 4.7 0.8\n\nMorocco 0.25 0.04\n\nWildfire-related losses were less than in recent years.\n\nAt 30% each, the contributions of tropical cyclones…\n\n…and SCS to annual insured losses have remained largely stable.\n\n**Figure 3** Share of insured losses by peril type by decade\n\nSource: Swiss Re Institute\n\n0%\n\n5%\n\n10%\n\n15%\n\n20%\n\n25%\n\n30%\n\n35%\n\n40%\n\nOthers Droughts European winter storms Wildfires Severe convective storms Floods Tropical cyclones Earthquakes\n\n3%\n\n7%\n\n1%\n\n7%\n\n31%\n\n13%\n\n29%\n\n8%\n\n1983–1992 1993–2002 2003–2012 2013–2022 40-year average (1983–2022)","publish_date":null,"title":"Longer-term loss trends","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2023-05-sigma-01-english_p7.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2023-05-sigma-01-english_p7.pdf"},"request":{"example_id":"chart/2023-05-sigma-01-english_p7","product_type":"parse","source_file_path":"2023-05-sigma-01-english_p7.pdf"},"started_at":"2026-11-04T19:46:40.330570"} +{"completed_at":"2026-11-04T19:46:43.554242","latency_in_ms":2311,"output":{"example_id":"chart/2024-ltra_corrected_july_2025_p14","grounded_pages":[],"layout_pages":[],"markdown":"**Capacity and Energy Assessment**\n\n**2024 Long-Term Reliability Assessment**\n\n**14**\n\nMRO-SaskPower Resource projections for Saskatchewan over the 10-year period have increased since the _2023 LTRA_ , rising to just over 1 GW and include two new 377 MW gas-fired generation facilities, expansion of existing gas-fired generation facilities, and new geothermal resources. The peak demand forecast growth rate (1.35% annually) has changed little since the _2023 LTRA_ . The ProbA for SaskPower indicates there is some risk of load loss and unserved energy that diminishes with the new resource additions that are planned for 2027 ( **Table 3** ). Saskatchewan is a winter- peaking system; however, the unserved energy risks are more closely associated with the planned outages of large generators that typically occur in the fall and spring shoulder periods. The months of September and October are the highest risk periods because of the potential for high temperatures to unexpectedly extend beyond summer and into the maintenance period.\n\n**Table 3: MRO-SaskPower ProbA Summary of Results**\n\n2026\\* 2026 2028\n\nEUE (MWh) 117.0 **75\\.64 8.55**\n\nEUE (PPM) 4.4 **2\\.81 0.30**\n\nLOLH (hours per year) 0.9 **0\\.54 0.08**\n\nOperable On-Peak Margin 24.6% **24\\.8% 30.8%**\n\n_\\* Year 2026 Results from the 2022 ProbA provided for trending_\n\nMRO-SPP New resource capacity in the interconnection queue—primarily solar, battery, and wind—is currently projected to keep up with rising peak demand forecasts. DR programs are also increasing over the next 10 years. Resource adequacy issues can arise, however, if generators retire earlier than anticipated or new resource additions connect more slowly. Currently, there are over 8 GW of coal and gas-fired generators that have indicated that they may retire within the next 10 years, making the resource outlook unclear. SPP’s summer reserve margins, based on installed generator capacity, remain above RMLs through 2029 (see **Figure 5** ).\n\n**Figure 5: MRO-SPP Five-Year Planning Reserve Margin–Summer**\n\nAll-hours probabilistic analysis performed by SPP shows that resource adequacy challenges are shifting from summer (when area demand is at its highest) to winter months (when the new resource mix is less capable of performing reliably). Annual results of the ProbA are in **Table 4** below. SPP’s 2023 loss-of-load expectation (LOLE) study results indicate that a separate RML and seasonal resource capacity contributions are needed for the summer and winter seasons to effectively provide for resource adequacy.\n\n**Table 4: MRO-SPP ProbA Summary of Results**\n\n2026\\* 2026 2028\n\nEUE (MWh) 0.84 **0\\.00 6.61**\n\nEUE (PPM) 0.00 **0\\.00 0.02**\n\nLOLH (hours per year) 0.00 **0\\.00 0.01**\n\nOperable On-Peak Margin 19.6% **19\\.5% 16.0%**\n\n_\\* Year 2026 Results from the 2022 ProbA provided for trending_","pages":[{"line_numbers":[],"markdown":"**Capacity and Energy Assessment**\n\n**2024 Long-Term Reliability Assessment**\n\n**14**\n\nMRO-SaskPower Resource projections for Saskatchewan over the 10-year period have increased since the _2023 LTRA_ , rising to just over 1 GW and include two new 377 MW gas-fired generation facilities, expansion of existing gas-fired generation facilities, and new geothermal resources. The peak demand forecast growth rate (1.35% annually) has changed little since the _2023 LTRA_ . The ProbA for SaskPower indicates there is some risk of load loss and unserved energy that diminishes with the new resource additions that are planned for 2027 ( **Table 3** ). Saskatchewan is a winter- peaking system; however, the unserved energy risks are more closely associated with the planned outages of large generators that typically occur in the fall and spring shoulder periods. The months of September and October are the highest risk periods because of the potential for high temperatures to unexpectedly extend beyond summer and into the maintenance period.\n\n**Table 3: MRO-SaskPower ProbA Summary of Results**\n\n2026\\* 2026 2028\n\nEUE (MWh) 117.0 **75\\.64 8.55**\n\nEUE (PPM) 4.4 **2\\.81 0.30**\n\nLOLH (hours per year) 0.9 **0\\.54 0.08**\n\nOperable On-Peak Margin 24.6% **24\\.8% 30.8%**\n\n_\\* Year 2026 Results from the 2022 ProbA provided for trending_\n\nMRO-SPP New resource capacity in the interconnection queue—primarily solar, battery, and wind—is currently projected to keep up with rising peak demand forecasts. DR programs are also increasing over the next 10 years. Resource adequacy issues can arise, however, if generators retire earlier than anticipated or new resource additions connect more slowly. Currently, there are over 8 GW of coal and gas-fired generators that have indicated that they may retire within the next 10 years, making the resource outlook unclear. SPP’s summer reserve margins, based on installed generator capacity, remain above RMLs through 2029 (see **Figure 5** ).\n\n**Figure 5: MRO-SPP Five-Year Planning Reserve Margin–Summer**\n\nAll-hours probabilistic analysis performed by SPP shows that resource adequacy challenges are shifting from summer (when area demand is at its highest) to winter months (when the new resource mix is less capable of performing reliably). Annual results of the ProbA are in **Table 4** below. SPP’s 2023 loss-of-load expectation (LOLE) study results indicate that a separate RML and seasonal resource capacity contributions are needed for the summer and winter seasons to effectively provide for resource adequacy.\n\n**Table 4: MRO-SPP ProbA Summary of Results**\n\n2026\\* 2026 2028\n\nEUE (MWh) 0.84 **0\\.00 6.61**\n\nEUE (PPM) 0.00 **0\\.00 0.02**\n\nLOLH (hours per year) 0.00 **0\\.00 0.01**\n\nOperable On-Peak Margin 19.6% **19\\.5% 16.0%**\n\n_\\* Year 2026 Results from the 2022 ProbA provided for trending_","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**Capacity and Energy Assessment**\n\n**2024 Long-Term Reliability Assessment**\n\n**14**\n\nMRO-SaskPower Resource projections for Saskatchewan over the 10-year period have increased since the _2023 LTRA_ , rising to just over 1 GW and include two new 377 MW gas-fired generation facilities, expansion of existing gas-fired generation facilities, and new geothermal resources. The peak demand forecast growth rate (1.35% annually) has changed little since the _2023 LTRA_ . The ProbA for SaskPower indicates there is some risk of load loss and unserved energy that diminishes with the new resource additions that are planned for 2027 ( **Table 3** ). Saskatchewan is a winter- peaking system; however, the unserved energy risks are more closely associated with the planned outages of large generators that typically occur in the fall and spring shoulder periods. The months of September and October are the highest risk periods because of the potential for high temperatures to unexpectedly extend beyond summer and into the maintenance period.\n\n**Table 3: MRO-SaskPower ProbA Summary of Results**\n\n2026\\* 2026 2028\n\nEUE (MWh) 117.0 **75\\.64 8.55**\n\nEUE (PPM) 4.4 **2\\.81 0.30**\n\nLOLH (hours per year) 0.9 **0\\.54 0.08**\n\nOperable On-Peak Margin 24.6% **24\\.8% 30.8%**\n\n_\\* Year 2026 Results from the 2022 ProbA provided for trending_\n\nMRO-SPP New resource capacity in the interconnection queue—primarily solar, battery, and wind—is currently projected to keep up with rising peak demand forecasts. DR programs are also increasing over the next 10 years. Resource adequacy issues can arise, however, if generators retire earlier than anticipated or new resource additions connect more slowly. Currently, there are over 8 GW of coal and gas-fired generators that have indicated that they may retire within the next 10 years, making the resource outlook unclear. SPP’s summer reserve margins, based on installed generator capacity, remain above RMLs through 2029 (see **Figure 5** ).\n\n**Figure 5: MRO-SPP Five-Year Planning Reserve Margin–Summer**\n\nAll-hours probabilistic analysis performed by SPP shows that resource adequacy challenges are shifting from summer (when area demand is at its highest) to winter months (when the new resource mix is less capable of performing reliably). Annual results of the ProbA are in **Table 4** below. SPP’s 2023 loss-of-load expectation (LOLE) study results indicate that a separate RML and seasonal resource capacity contributions are needed for the summer and winter seasons to effectively provide for resource adequacy.\n\n**Table 4: MRO-SPP ProbA Summary of Results**\n\n2026\\* 2026 2028\n\nEUE (MWh) 0.84 **0\\.00 6.61**\n\nEUE (PPM) 0.00 **0\\.00 0.02**\n\nLOLH (hours per year) 0.00 **0\\.00 0.01**\n\nOperable On-Peak Margin 19.6% **19\\.5% 16.0%**\n\n_\\* Year 2026 Results from the 2022 ProbA provided for trending_","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2024-ltra_corrected_july_2025_p14.pdf"]},"response":{"errors":[],"extract_id":"extract_c462256266bdf063a9b1c2a72249153e","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**Capacity and Energy Assessment**\n\n**2024 Long-Term Reliability Assessment**\n\n**14**\n\nMRO-SaskPower Resource projections for Saskatchewan over the 10-year period have increased since the _2023 LTRA_ , rising to just over 1 GW and include two new 377 MW gas-fired generation facilities, expansion of existing gas-fired generation facilities, and new geothermal resources. The peak demand forecast growth rate (1.35% annually) has changed little since the _2023 LTRA_ . The ProbA for SaskPower indicates there is some risk of load loss and unserved energy that diminishes with the new resource additions that are planned for 2027 ( **Table 3** ). Saskatchewan is a winter- peaking system; however, the unserved energy risks are more closely associated with the planned outages of large generators that typically occur in the fall and spring shoulder periods. The months of September and October are the highest risk periods because of the potential for high temperatures to unexpectedly extend beyond summer and into the maintenance period.\n\n**Table 3: MRO-SaskPower ProbA Summary of Results**\n\n2026\\* 2026 2028\n\nEUE (MWh) 117.0 **75\\.64 8.55**\n\nEUE (PPM) 4.4 **2\\.81 0.30**\n\nLOLH (hours per year) 0.9 **0\\.54 0.08**\n\nOperable On-Peak Margin 24.6% **24\\.8% 30.8%**\n\n_\\* Year 2026 Results from the 2022 ProbA provided for trending_\n\nMRO-SPP New resource capacity in the interconnection queue—primarily solar, battery, and wind—is currently projected to keep up with rising peak demand forecasts. DR programs are also increasing over the next 10 years. Resource adequacy issues can arise, however, if generators retire earlier than anticipated or new resource additions connect more slowly. Currently, there are over 8 GW of coal and gas-fired generators that have indicated that they may retire within the next 10 years, making the resource outlook unclear. SPP’s summer reserve margins, based on installed generator capacity, remain above RMLs through 2029 (see **Figure 5** ).\n\n**Figure 5: MRO-SPP Five-Year Planning Reserve Margin–Summer**\n\nAll-hours probabilistic analysis performed by SPP shows that resource adequacy challenges are shifting from summer (when area demand is at its highest) to winter months (when the new resource mix is less capable of performing reliably). Annual results of the ProbA are in **Table 4** below. SPP’s 2023 loss-of-load expectation (LOLE) study results indicate that a separate RML and seasonal resource capacity contributions are needed for the summer and winter seasons to effectively provide for resource adequacy.\n\n**Table 4: MRO-SPP ProbA Summary of Results**\n\n2026\\* 2026 2028\n\nEUE (MWh) 0.84 **0\\.00 6.61**\n\nEUE (PPM) 0.00 **0\\.00 0.02**\n\nLOLH (hours per year) 0.00 **0\\.00 0.01**\n\nOperable On-Peak Margin 19.6% **19\\.5% 16.0%**\n\n_\\* Year 2026 Results from the 2022 ProbA provided for trending_"],"full_content":"**Capacity and Energy Assessment**\n\n**2024 Long-Term Reliability Assessment**\n\n**14**\n\nMRO-SaskPower Resource projections for Saskatchewan over the 10-year period have increased since the _2023 LTRA_ , rising to just over 1 GW and include two new 377 MW gas-fired generation facilities, expansion of existing gas-fired generation facilities, and new geothermal resources. The peak demand forecast growth rate (1.35% annually) has changed little since the _2023 LTRA_ . The ProbA for SaskPower indicates there is some risk of load loss and unserved energy that diminishes with the new resource additions that are planned for 2027 ( **Table 3** ). Saskatchewan is a winter- peaking system; however, the unserved energy risks are more closely associated with the planned outages of large generators that typically occur in the fall and spring shoulder periods. The months of September and October are the highest risk periods because of the potential for high temperatures to unexpectedly extend beyond summer and into the maintenance period.\n\n**Table 3: MRO-SaskPower ProbA Summary of Results**\n\n2026\\* 2026 2028\n\nEUE (MWh) 117.0 **75\\.64 8.55**\n\nEUE (PPM) 4.4 **2\\.81 0.30**\n\nLOLH (hours per year) 0.9 **0\\.54 0.08**\n\nOperable On-Peak Margin 24.6% **24\\.8% 30.8%**\n\n_\\* Year 2026 Results from the 2022 ProbA provided for trending_\n\nMRO-SPP New resource capacity in the interconnection queue—primarily solar, battery, and wind—is currently projected to keep up with rising peak demand forecasts. DR programs are also increasing over the next 10 years. Resource adequacy issues can arise, however, if generators retire earlier than anticipated or new resource additions connect more slowly. Currently, there are over 8 GW of coal and gas-fired generators that have indicated that they may retire within the next 10 years, making the resource outlook unclear. SPP’s summer reserve margins, based on installed generator capacity, remain above RMLs through 2029 (see **Figure 5** ).\n\n**Figure 5: MRO-SPP Five-Year Planning Reserve Margin–Summer**\n\nAll-hours probabilistic analysis performed by SPP shows that resource adequacy challenges are shifting from summer (when area demand is at its highest) to winter months (when the new resource mix is less capable of performing reliably). Annual results of the ProbA are in **Table 4** below. SPP’s 2023 loss-of-load expectation (LOLE) study results indicate that a separate RML and seasonal resource capacity contributions are needed for the summer and winter seasons to effectively provide for resource adequacy.\n\n**Table 4: MRO-SPP ProbA Summary of Results**\n\n2026\\* 2026 2028\n\nEUE (MWh) 0.84 **0\\.00 6.61**\n\nEUE (PPM) 0.00 **0\\.00 0.02**\n\nLOLH (hours per year) 0.00 **0\\.00 0.01**\n\nOperable On-Peak Margin 19.6% **19\\.5% 16.0%**\n\n_\\* Year 2026 Results from the 2022 ProbA provided for trending_","publish_date":null,"title":"MRO-SaskPower 2023 LTRA Resource projections for Saskatchewan over the 10-year period have increased since the , rising to just over","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2024-ltra_corrected_july_2025_p14.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2024-ltra_corrected_july_2025_p14.pdf"},"request":{"example_id":"chart/2024-ltra_corrected_july_2025_p14","product_type":"parse","source_file_path":"2024-ltra_corrected_july_2025_p14.pdf"},"started_at":"2026-11-04T19:46:41.242596"} +{"completed_at":"2026-11-04T19:44:47.995423","latency_in_ms":1850,"output":{"example_id":"chart/2024-ltra_corrected_july_2025_p15","grounded_pages":[],"layout_pages":[],"markdown":"**Capacity and Energy Assessment**\n\n**2024 Long-Term Reliability Assessment**\n\n**15**\n\n**NPCC-New England** New England is forecasting unprecedented demand growth driven by electrification of heating and transportation. Wind, solar, and batteries make up the projected resource additions, along with growth in distributed energy resources. Since the _2023 LTRA_ , the peak demand forecast is relatively unchanged. New England has among the strongest winter demand forecast growth rates of any assessment area, rising over 7 GW in the 10-year period, or 35% from its current peak demand forecast. The 2024 ProbA for NPCC-New England reveals increasing risk of load loss and unserved energy compared to the 2022 ProbA (see **Table 5** ). Higher-demand forecast and replacement of dispatchable resources with more VERs contribute to this trend. In the ProbA modeling, supply shortfall risk is limited to the summer months and is associated with periods of above-normal demand and low resource output (e.g., low wind, solar, or high thermal outages).\n\n**Table 5: NPCC-New England ProbA Summary of Results**\n\n2026\\* 2026 2028\n\nEUE (MWh) 0.55 **10\\.69 67.40**\n\nEUE (PPM) 0.00 **0\\.09 0.06**\n\nLOLH (hours per year) 0.00 **0\\.07 0.03**\n\nOperable On-Peak Margin 27.8% **12\\.4% 13.7%**\n\n_\\* Year 2026 Results from the 2022 ProbA provided for trending_\n\nWhile NPCC’s ProbA results indicate that the risk of unserved energy in New England is concentrated in summer months, a persistent concern is whether there will be sufficient fuel available to satisfy electrical energy and operating reserve demands during an extended cold spell or a series of cold spells given the existing resource mix and regional fuel delivery infrastructure. Potential natural gas transportation constraints can affect supply to generators during extreme cold temperatures, when natural gas demand for space heating is also peaking. Most natural-gas-fired generators in the area do not have firm natural gas transportation service and can be subject to supply curtailment during peak conditions. Many natural-gas-fired generators in New England can also operate with stored oil, providing for assured electricity supply in extreme winter conditions while oil stores are procured and maintained. Scenarios of extended and extreme cold weather that affect the availability of natural gas and oil replenishments present a reliability risk but were not within the scope of the 2024 ProbA.\n\nNPCC-Ontario Near- and long-term resource adequacy challenges continue in the NPCC-Ontario assessment area as the Independent Electricity System Operator (IESO) manages lengthy refurbishment outages at nuclear generators and demand growth driven by agriculture and electrification. The beginning of projected shortfalls occurs in 2027 when some contracts for firm capacity are set to expire (see **Figure 6** ). However, a number of resources with expiring contracts between 2026 and 2029 are anticipated to be re-committed for five-year terms through the IESO’s second medium-term procurement, which is underway. As such, Ontario’s Anticipated Reserve Margins (ARM) and Prospective Reserve Margins shown in **Figure 6** (which do not include this set of expiring resources) are conservative. Recommitment of these resources is expected to raise margins above the reference level in 2029 and help maintain reliability through the end of the decade. In May 2024, the IESO finalized long-term commitments for new-build battery storage facilities (1,784 MW) and new-build natural gas and biogas facilities (411 MW). Early operation incentives anticipated to bring a portion of these resources online as early as 2027 could help alleviate the reserve margin gap. The IESO also launched procurements for new capacity and energy-producing resources to be online as early as 2029. In addition, the IESO finalized an agreement with Hydro-Québec for capacity sharing in November 2024, providing 600 MW of firm imports to Ontario starting in Summer 2025 through Summer 2031. This capacity is also expected to help alleviate shortfalls indicated in 2027.\n\n**Figure 6: NPCC-Ontario Five-Year Planning Reserve Margin–Summer**","pages":[{"line_numbers":[],"markdown":"**Capacity and Energy Assessment**\n\n**2024 Long-Term Reliability Assessment**\n\n**15**\n\n**NPCC-New England** New England is forecasting unprecedented demand growth driven by electrification of heating and transportation. Wind, solar, and batteries make up the projected resource additions, along with growth in distributed energy resources. Since the _2023 LTRA_ , the peak demand forecast is relatively unchanged. New England has among the strongest winter demand forecast growth rates of any assessment area, rising over 7 GW in the 10-year period, or 35% from its current peak demand forecast. The 2024 ProbA for NPCC-New England reveals increasing risk of load loss and unserved energy compared to the 2022 ProbA (see **Table 5** ). Higher-demand forecast and replacement of dispatchable resources with more VERs contribute to this trend. In the ProbA modeling, supply shortfall risk is limited to the summer months and is associated with periods of above-normal demand and low resource output (e.g., low wind, solar, or high thermal outages).\n\n**Table 5: NPCC-New England ProbA Summary of Results**\n\n2026\\* 2026 2028\n\nEUE (MWh) 0.55 **10\\.69 67.40**\n\nEUE (PPM) 0.00 **0\\.09 0.06**\n\nLOLH (hours per year) 0.00 **0\\.07 0.03**\n\nOperable On-Peak Margin 27.8% **12\\.4% 13.7%**\n\n_\\* Year 2026 Results from the 2022 ProbA provided for trending_\n\nWhile NPCC’s ProbA results indicate that the risk of unserved energy in New England is concentrated in summer months, a persistent concern is whether there will be sufficient fuel available to satisfy electrical energy and operating reserve demands during an extended cold spell or a series of cold spells given the existing resource mix and regional fuel delivery infrastructure. Potential natural gas transportation constraints can affect supply to generators during extreme cold temperatures, when natural gas demand for space heating is also peaking. Most natural-gas-fired generators in the area do not have firm natural gas transportation service and can be subject to supply curtailment during peak conditions. Many natural-gas-fired generators in New England can also operate with stored oil, providing for assured electricity supply in extreme winter conditions while oil stores are procured and maintained. Scenarios of extended and extreme cold weather that affect the availability of natural gas and oil replenishments present a reliability risk but were not within the scope of the 2024 ProbA.\n\nNPCC-Ontario Near- and long-term resource adequacy challenges continue in the NPCC-Ontario assessment area as the Independent Electricity System Operator (IESO) manages lengthy refurbishment outages at nuclear generators and demand growth driven by agriculture and electrification. The beginning of projected shortfalls occurs in 2027 when some contracts for firm capacity are set to expire (see **Figure 6** ). However, a number of resources with expiring contracts between 2026 and 2029 are anticipated to be re-committed for five-year terms through the IESO’s second medium-term procurement, which is underway. As such, Ontario’s Anticipated Reserve Margins (ARM) and Prospective Reserve Margins shown in **Figure 6** (which do not include this set of expiring resources) are conservative. Recommitment of these resources is expected to raise margins above the reference level in 2029 and help maintain reliability through the end of the decade. In May 2024, the IESO finalized long-term commitments for new-build battery storage facilities (1,784 MW) and new-build natural gas and biogas facilities (411 MW). Early operation incentives anticipated to bring a portion of these resources online as early as 2027 could help alleviate the reserve margin gap. The IESO also launched procurements for new capacity and energy-producing resources to be online as early as 2029. In addition, the IESO finalized an agreement with Hydro-Québec for capacity sharing in November 2024, providing 600 MW of firm imports to Ontario starting in Summer 2025 through Summer 2031. This capacity is also expected to help alleviate shortfalls indicated in 2027.\n\n**Figure 6: NPCC-Ontario Five-Year Planning Reserve Margin–Summer**","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**Capacity and Energy Assessment**\n\n**2024 Long-Term Reliability Assessment**\n\n**15**\n\n**NPCC-New England** New England is forecasting unprecedented demand growth driven by electrification of heating and transportation. Wind, solar, and batteries make up the projected resource additions, along with growth in distributed energy resources. Since the _2023 LTRA_ , the peak demand forecast is relatively unchanged. New England has among the strongest winter demand forecast growth rates of any assessment area, rising over 7 GW in the 10-year period, or 35% from its current peak demand forecast. The 2024 ProbA for NPCC-New England reveals increasing risk of load loss and unserved energy compared to the 2022 ProbA (see **Table 5** ). Higher-demand forecast and replacement of dispatchable resources with more VERs contribute to this trend. In the ProbA modeling, supply shortfall risk is limited to the summer months and is associated with periods of above-normal demand and low resource output (e.g., low wind, solar, or high thermal outages).\n\n**Table 5: NPCC-New England ProbA Summary of Results**\n\n2026\\* 2026 2028\n\nEUE (MWh) 0.55 **10\\.69 67.40**\n\nEUE (PPM) 0.00 **0\\.09 0.06**\n\nLOLH (hours per year) 0.00 **0\\.07 0.03**\n\nOperable On-Peak Margin 27.8% **12\\.4% 13.7%**\n\n_\\* Year 2026 Results from the 2022 ProbA provided for trending_\n\nWhile NPCC’s ProbA results indicate that the risk of unserved energy in New England is concentrated in summer months, a persistent concern is whether there will be sufficient fuel available to satisfy electrical energy and operating reserve demands during an extended cold spell or a series of cold spells given the existing resource mix and regional fuel delivery infrastructure. Potential natural gas transportation constraints can affect supply to generators during extreme cold temperatures, when natural gas demand for space heating is also peaking. Most natural-gas-fired generators in the area do not have firm natural gas transportation service and can be subject to supply curtailment during peak conditions. Many natural-gas-fired generators in New England can also operate with stored oil, providing for assured electricity supply in extreme winter conditions while oil stores are procured and maintained. Scenarios of extended and extreme cold weather that affect the availability of natural gas and oil replenishments present a reliability risk but were not within the scope of the 2024 ProbA.\n\nNPCC-Ontario Near- and long-term resource adequacy challenges continue in the NPCC-Ontario assessment area as the Independent Electricity System Operator (IESO) manages lengthy refurbishment outages at nuclear generators and demand growth driven by agriculture and electrification. The beginning of projected shortfalls occurs in 2027 when some contracts for firm capacity are set to expire (see **Figure 6** ). However, a number of resources with expiring contracts between 2026 and 2029 are anticipated to be re-committed for five-year terms through the IESO’s second medium-term procurement, which is underway. As such, Ontario’s Anticipated Reserve Margins (ARM) and Prospective Reserve Margins shown in **Figure 6** (which do not include this set of expiring resources) are conservative. Recommitment of these resources is expected to raise margins above the reference level in 2029 and help maintain reliability through the end of the decade. In May 2024, the IESO finalized long-term commitments for new-build battery storage facilities (1,784 MW) and new-build natural gas and biogas facilities (411 MW). Early operation incentives anticipated to bring a portion of these resources online as early as 2027 could help alleviate the reserve margin gap. The IESO also launched procurements for new capacity and energy-producing resources to be online as early as 2029. In addition, the IESO finalized an agreement with Hydro-Québec for capacity sharing in November 2024, providing 600 MW of firm imports to Ontario starting in Summer 2025 through Summer 2031. This capacity is also expected to help alleviate shortfalls indicated in 2027.\n\n**Figure 6: NPCC-Ontario Five-Year Planning Reserve Margin–Summer**","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2024-ltra_corrected_july_2025_p15.pdf"]},"response":{"errors":[],"extract_id":"extract_b1414bd75aa3f0e448e0bb4510a8e71d","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**Capacity and Energy Assessment**\n\n**2024 Long-Term Reliability Assessment**\n\n**15**\n\n**NPCC-New England** New England is forecasting unprecedented demand growth driven by electrification of heating and transportation. Wind, solar, and batteries make up the projected resource additions, along with growth in distributed energy resources. Since the _2023 LTRA_ , the peak demand forecast is relatively unchanged. New England has among the strongest winter demand forecast growth rates of any assessment area, rising over 7 GW in the 10-year period, or 35% from its current peak demand forecast. The 2024 ProbA for NPCC-New England reveals increasing risk of load loss and unserved energy compared to the 2022 ProbA (see **Table 5** ). Higher-demand forecast and replacement of dispatchable resources with more VERs contribute to this trend. In the ProbA modeling, supply shortfall risk is limited to the summer months and is associated with periods of above-normal demand and low resource output (e.g., low wind, solar, or high thermal outages).\n\n**Table 5: NPCC-New England ProbA Summary of Results**\n\n2026\\* 2026 2028\n\nEUE (MWh) 0.55 **10\\.69 67.40**\n\nEUE (PPM) 0.00 **0\\.09 0.06**\n\nLOLH (hours per year) 0.00 **0\\.07 0.03**\n\nOperable On-Peak Margin 27.8% **12\\.4% 13.7%**\n\n_\\* Year 2026 Results from the 2022 ProbA provided for trending_\n\nWhile NPCC’s ProbA results indicate that the risk of unserved energy in New England is concentrated in summer months, a persistent concern is whether there will be sufficient fuel available to satisfy electrical energy and operating reserve demands during an extended cold spell or a series of cold spells given the existing resource mix and regional fuel delivery infrastructure. Potential natural gas transportation constraints can affect supply to generators during extreme cold temperatures, when natural gas demand for space heating is also peaking. Most natural-gas-fired generators in the area do not have firm natural gas transportation service and can be subject to supply curtailment during peak conditions. Many natural-gas-fired generators in New England can also operate with stored oil, providing for assured electricity supply in extreme winter conditions while oil stores are procured and maintained. Scenarios of extended and extreme cold weather that affect the availability of natural gas and oil replenishments present a reliability risk but were not within the scope of the 2024 ProbA.\n\nNPCC-Ontario Near- and long-term resource adequacy challenges continue in the NPCC-Ontario assessment area as the Independent Electricity System Operator (IESO) manages lengthy refurbishment outages at nuclear generators and demand growth driven by agriculture and electrification. The beginning of projected shortfalls occurs in 2027 when some contracts for firm capacity are set to expire (see **Figure 6** ). However, a number of resources with expiring contracts between 2026 and 2029 are anticipated to be re-committed for five-year terms through the IESO’s second medium-term procurement, which is underway. As such, Ontario’s Anticipated Reserve Margins (ARM) and Prospective Reserve Margins shown in **Figure 6** (which do not include this set of expiring resources) are conservative. Recommitment of these resources is expected to raise margins above the reference level in 2029 and help maintain reliability through the end of the decade. In May 2024, the IESO finalized long-term commitments for new-build battery storage facilities (1,784 MW) and new-build natural gas and biogas facilities (411 MW). Early operation incentives anticipated to bring a portion of these resources online as early as 2027 could help alleviate the reserve margin gap. The IESO also launched procurements for new capacity and energy-producing resources to be online as early as 2029. In addition, the IESO finalized an agreement with Hydro-Québec for capacity sharing in November 2024, providing 600 MW of firm imports to Ontario starting in Summer 2025 through Summer 2031. This capacity is also expected to help alleviate shortfalls indicated in 2027.\n\n**Figure 6: NPCC-Ontario Five-Year Planning Reserve Margin–Summer**"],"full_content":"**Capacity and Energy Assessment**\n\n**2024 Long-Term Reliability Assessment**\n\n**15**\n\n**NPCC-New England** New England is forecasting unprecedented demand growth driven by electrification of heating and transportation. Wind, solar, and batteries make up the projected resource additions, along with growth in distributed energy resources. Since the _2023 LTRA_ , the peak demand forecast is relatively unchanged. New England has among the strongest winter demand forecast growth rates of any assessment area, rising over 7 GW in the 10-year period, or 35% from its current peak demand forecast. The 2024 ProbA for NPCC-New England reveals increasing risk of load loss and unserved energy compared to the 2022 ProbA (see **Table 5** ). Higher-demand forecast and replacement of dispatchable resources with more VERs contribute to this trend. In the ProbA modeling, supply shortfall risk is limited to the summer months and is associated with periods of above-normal demand and low resource output (e.g., low wind, solar, or high thermal outages).\n\n**Table 5: NPCC-New England ProbA Summary of Results**\n\n2026\\* 2026 2028\n\nEUE (MWh) 0.55 **10\\.69 67.40**\n\nEUE (PPM) 0.00 **0\\.09 0.06**\n\nLOLH (hours per year) 0.00 **0\\.07 0.03**\n\nOperable On-Peak Margin 27.8% **12\\.4% 13.7%**\n\n_\\* Year 2026 Results from the 2022 ProbA provided for trending_\n\nWhile NPCC’s ProbA results indicate that the risk of unserved energy in New England is concentrated in summer months, a persistent concern is whether there will be sufficient fuel available to satisfy electrical energy and operating reserve demands during an extended cold spell or a series of cold spells given the existing resource mix and regional fuel delivery infrastructure. Potential natural gas transportation constraints can affect supply to generators during extreme cold temperatures, when natural gas demand for space heating is also peaking. Most natural-gas-fired generators in the area do not have firm natural gas transportation service and can be subject to supply curtailment during peak conditions. Many natural-gas-fired generators in New England can also operate with stored oil, providing for assured electricity supply in extreme winter conditions while oil stores are procured and maintained. Scenarios of extended and extreme cold weather that affect the availability of natural gas and oil replenishments present a reliability risk but were not within the scope of the 2024 ProbA.\n\nNPCC-Ontario Near- and long-term resource adequacy challenges continue in the NPCC-Ontario assessment area as the Independent Electricity System Operator (IESO) manages lengthy refurbishment outages at nuclear generators and demand growth driven by agriculture and electrification. The beginning of projected shortfalls occurs in 2027 when some contracts for firm capacity are set to expire (see **Figure 6** ). However, a number of resources with expiring contracts between 2026 and 2029 are anticipated to be re-committed for five-year terms through the IESO’s second medium-term procurement, which is underway. As such, Ontario’s Anticipated Reserve Margins (ARM) and Prospective Reserve Margins shown in **Figure 6** (which do not include this set of expiring resources) are conservative. Recommitment of these resources is expected to raise margins above the reference level in 2029 and help maintain reliability through the end of the decade. In May 2024, the IESO finalized long-term commitments for new-build battery storage facilities (1,784 MW) and new-build natural gas and biogas facilities (411 MW). Early operation incentives anticipated to bring a portion of these resources online as early as 2027 could help alleviate the reserve margin gap. The IESO also launched procurements for new capacity and energy-producing resources to be online as early as 2029. In addition, the IESO finalized an agreement with Hydro-Québec for capacity sharing in November 2024, providing 600 MW of firm imports to Ontario starting in Summer 2025 through Summer 2031. This capacity is also expected to help alleviate shortfalls indicated in 2027.\n\n**Figure 6: NPCC-Ontario Five-Year Planning Reserve Margin–Summer**","publish_date":null,"title":"NPCC-New England NPCC-Ontario Near- and long-term resource adequacy challenges continue in the NPCC-Ontario assessment area as New England is forecasting","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2024-ltra_corrected_july_2025_p15.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2024-ltra_corrected_july_2025_p15.pdf"},"request":{"example_id":"chart/2024-ltra_corrected_july_2025_p15","product_type":"parse","source_file_path":"2024-ltra_corrected_july_2025_p15.pdf"},"started_at":"2026-11-04T19:44:46.145336"} +{"completed_at":"2026-11-04T19:44:49.005681","latency_in_ms":2856,"output":{"example_id":"chart/2024-ltra_corrected_july_2025_p25","grounded_pages":[],"layout_pages":[],"markdown":"**Resource Mix Changes**\n\n**2024 Long-Term Reliability Assessment**\n\n**25**\n\nOffshore wind plants are increasingly entering interconnection queues, located around the Northeastern United States. **Figure 11** shows the current planned capacity in the planning process through 2034 by assessment area. Currently, the existing offshore wind nameplate capacities are small in comparison: a single 29 MW plant in NPCC-New England and a 12 MW plant in PJM.\n\n**Figure 11: Offshore Wind Capacity Planned through 2034**\n\n**Battery Resources** As the BPS increases the share of energy provided by VERs, the ability to provide energy by BESS or hybrid-solar PV and wind plants is increasingly important. While currently installed capacity totals just over 17 GW, an additional 306.5 GW of BESS are in planning. **Figure 12** shows the nameplate capacity of BESS resources currently in operation and in planning for connection to the BPS through 2034.\n\n18 NERC–WECC _2022 California Battery Energy Storage System Disturbances_\n\n**Figure 12: Battery Resource Capacity Existing and Planned through 2034**\n\nBESS are improving reliability by helping to offset the variability and uncertainty of inverter-based resources (IBR). BESS are, however, a relatively new type of grid resource with unique operating characteristics. The joint _NERC-WECC Staff Report: 2022 California Battery Energy Storage System Disturbances_ ¹⁸ highlights an event when a BESS, like some other IBRs, failed to properly ride through a normal system fault. This indicates that BESS must be included in the currently underway strategies to address IBR performance issues. Because the electrical output of variable energy resources (e.g., wind, solar PV) depends on weather conditions, on-peak capacity contributions are less than nameplate capacity. **Table 14** shows the nameplate value and the capacity contribution of existing wind, solar PV, hydro, and energy storage systems resources at the peak demand hour for each assessment area. During risk periods after peak demand (e.g., U.S. assessment areas in WECC), contributions of solar PV resources output is diminished during evening periods and additional resources are needed.\n\nERCOT (Not shown) (MW) Existing: 7,335 Tier 1: 19,923\n\nTier 2: 97,960\n\nTier 3: 47,404","pages":[{"line_numbers":[],"markdown":"**Resource Mix Changes**\n\n**2024 Long-Term Reliability Assessment**\n\n**25**\n\nOffshore wind plants are increasingly entering interconnection queues, located around the Northeastern United States. **Figure 11** shows the current planned capacity in the planning process through 2034 by assessment area. Currently, the existing offshore wind nameplate capacities are small in comparison: a single 29 MW plant in NPCC-New England and a 12 MW plant in PJM.\n\n**Figure 11: Offshore Wind Capacity Planned through 2034**\n\n**Battery Resources** As the BPS increases the share of energy provided by VERs, the ability to provide energy by BESS or hybrid-solar PV and wind plants is increasingly important. While currently installed capacity totals just over 17 GW, an additional 306.5 GW of BESS are in planning. **Figure 12** shows the nameplate capacity of BESS resources currently in operation and in planning for connection to the BPS through 2034.\n\n18 NERC–WECC _2022 California Battery Energy Storage System Disturbances_\n\n**Figure 12: Battery Resource Capacity Existing and Planned through 2034**\n\nBESS are improving reliability by helping to offset the variability and uncertainty of inverter-based resources (IBR). BESS are, however, a relatively new type of grid resource with unique operating characteristics. The joint _NERC-WECC Staff Report: 2022 California Battery Energy Storage System Disturbances_ ¹⁸ highlights an event when a BESS, like some other IBRs, failed to properly ride through a normal system fault. This indicates that BESS must be included in the currently underway strategies to address IBR performance issues. Because the electrical output of variable energy resources (e.g., wind, solar PV) depends on weather conditions, on-peak capacity contributions are less than nameplate capacity. **Table 14** shows the nameplate value and the capacity contribution of existing wind, solar PV, hydro, and energy storage systems resources at the peak demand hour for each assessment area. During risk periods after peak demand (e.g., U.S. assessment areas in WECC), contributions of solar PV resources output is diminished during evening periods and additional resources are needed.\n\nERCOT (Not shown) (MW) Existing: 7,335 Tier 1: 19,923\n\nTier 2: 97,960\n\nTier 3: 47,404","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**Resource Mix Changes**\n\n**2024 Long-Term Reliability Assessment**\n\n**25**\n\nOffshore wind plants are increasingly entering interconnection queues, located around the Northeastern United States. **Figure 11** shows the current planned capacity in the planning process through 2034 by assessment area. Currently, the existing offshore wind nameplate capacities are small in comparison: a single 29 MW plant in NPCC-New England and a 12 MW plant in PJM.\n\n**Figure 11: Offshore Wind Capacity Planned through 2034**\n\n**Battery Resources** As the BPS increases the share of energy provided by VERs, the ability to provide energy by BESS or hybrid-solar PV and wind plants is increasingly important. While currently installed capacity totals just over 17 GW, an additional 306.5 GW of BESS are in planning. **Figure 12** shows the nameplate capacity of BESS resources currently in operation and in planning for connection to the BPS through 2034.\n\n18 NERC–WECC _2022 California Battery Energy Storage System Disturbances_\n\n**Figure 12: Battery Resource Capacity Existing and Planned through 2034**\n\nBESS are improving reliability by helping to offset the variability and uncertainty of inverter-based resources (IBR). BESS are, however, a relatively new type of grid resource with unique operating characteristics. The joint _NERC-WECC Staff Report: 2022 California Battery Energy Storage System Disturbances_ ¹⁸ highlights an event when a BESS, like some other IBRs, failed to properly ride through a normal system fault. This indicates that BESS must be included in the currently underway strategies to address IBR performance issues. Because the electrical output of variable energy resources (e.g., wind, solar PV) depends on weather conditions, on-peak capacity contributions are less than nameplate capacity. **Table 14** shows the nameplate value and the capacity contribution of existing wind, solar PV, hydro, and energy storage systems resources at the peak demand hour for each assessment area. During risk periods after peak demand (e.g., U.S. assessment areas in WECC), contributions of solar PV resources output is diminished during evening periods and additional resources are needed.\n\nERCOT (Not shown) (MW) Existing: 7,335 Tier 1: 19,923\n\nTier 2: 97,960\n\nTier 3: 47,404","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2024-ltra_corrected_july_2025_p25.pdf"]},"response":{"errors":[],"extract_id":"extract_532670979d941cae4cc37741e40e2bcc","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**Resource Mix Changes**\n\n**2024 Long-Term Reliability Assessment**\n\n**25**\n\nOffshore wind plants are increasingly entering interconnection queues, located around the Northeastern United States. **Figure 11** shows the current planned capacity in the planning process through 2034 by assessment area. Currently, the existing offshore wind nameplate capacities are small in comparison: a single 29 MW plant in NPCC-New England and a 12 MW plant in PJM.\n\n**Figure 11: Offshore Wind Capacity Planned through 2034**\n\n**Battery Resources** As the BPS increases the share of energy provided by VERs, the ability to provide energy by BESS or hybrid-solar PV and wind plants is increasingly important. While currently installed capacity totals just over 17 GW, an additional 306.5 GW of BESS are in planning. **Figure 12** shows the nameplate capacity of BESS resources currently in operation and in planning for connection to the BPS through 2034.\n\n18 NERC–WECC _2022 California Battery Energy Storage System Disturbances_\n\n**Figure 12: Battery Resource Capacity Existing and Planned through 2034**\n\nBESS are improving reliability by helping to offset the variability and uncertainty of inverter-based resources (IBR). BESS are, however, a relatively new type of grid resource with unique operating characteristics. The joint _NERC-WECC Staff Report: 2022 California Battery Energy Storage System Disturbances_ ¹⁸ highlights an event when a BESS, like some other IBRs, failed to properly ride through a normal system fault. This indicates that BESS must be included in the currently underway strategies to address IBR performance issues. Because the electrical output of variable energy resources (e.g., wind, solar PV) depends on weather conditions, on-peak capacity contributions are less than nameplate capacity. **Table 14** shows the nameplate value and the capacity contribution of existing wind, solar PV, hydro, and energy storage systems resources at the peak demand hour for each assessment area. During risk periods after peak demand (e.g., U.S. assessment areas in WECC), contributions of solar PV resources output is diminished during evening periods and additional resources are needed.\n\nERCOT (Not shown) (MW) Existing: 7,335 Tier 1: 19,923\n\nTier 2: 97,960\n\nTier 3: 47,404"],"full_content":"**Resource Mix Changes**\n\n**2024 Long-Term Reliability Assessment**\n\n**25**\n\nOffshore wind plants are increasingly entering interconnection queues, located around the Northeastern United States. **Figure 11** shows the current planned capacity in the planning process through 2034 by assessment area. Currently, the existing offshore wind nameplate capacities are small in comparison: a single 29 MW plant in NPCC-New England and a 12 MW plant in PJM.\n\n**Figure 11: Offshore Wind Capacity Planned through 2034**\n\n**Battery Resources** As the BPS increases the share of energy provided by VERs, the ability to provide energy by BESS or hybrid-solar PV and wind plants is increasingly important. While currently installed capacity totals just over 17 GW, an additional 306.5 GW of BESS are in planning. **Figure 12** shows the nameplate capacity of BESS resources currently in operation and in planning for connection to the BPS through 2034.\n\n18 NERC–WECC _2022 California Battery Energy Storage System Disturbances_\n\n**Figure 12: Battery Resource Capacity Existing and Planned through 2034**\n\nBESS are improving reliability by helping to offset the variability and uncertainty of inverter-based resources (IBR). BESS are, however, a relatively new type of grid resource with unique operating characteristics. The joint _NERC-WECC Staff Report: 2022 California Battery Energy Storage System Disturbances_ ¹⁸ highlights an event when a BESS, like some other IBRs, failed to properly ride through a normal system fault. This indicates that BESS must be included in the currently underway strategies to address IBR performance issues. Because the electrical output of variable energy resources (e.g., wind, solar PV) depends on weather conditions, on-peak capacity contributions are less than nameplate capacity. **Table 14** shows the nameplate value and the capacity contribution of existing wind, solar PV, hydro, and energy storage systems resources at the peak demand hour for each assessment area. During risk periods after peak demand (e.g., U.S. assessment areas in WECC), contributions of solar PV resources output is diminished during evening periods and additional resources are needed.\n\nERCOT (Not shown) (MW) Existing: 7,335 Tier 1: 19,923\n\nTier 2: 97,960\n\nTier 3: 47,404","publish_date":null,"title":"Offshore wind plants are increasingly entering interconnection queues, located around the Northeastern United States. shows the current planned capacity in","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2024-ltra_corrected_july_2025_p25.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2024-ltra_corrected_july_2025_p25.pdf"},"request":{"example_id":"chart/2024-ltra_corrected_july_2025_p25","product_type":"parse","source_file_path":"2024-ltra_corrected_july_2025_p25.pdf"},"started_at":"2026-11-04T19:44:46.148954"} +{"completed_at":"2026-11-04T19:44:50.077447","latency_in_ms":1773,"output":{"example_id":"chart/2024-ltra_corrected_july_2025_p27","grounded_pages":[],"layout_pages":[],"markdown":"**Resource Mix Changes**\n\n**2024 Long-Term Reliability Assessment**\n\n**27**\n\n**Solar PV Distributed Energy Resource Growth** Behind-the-meter (BTM) solar PV generators are solar PV resources connected on the distribution system, such as residential rooftop solar systems. The rapid growth of BTM solar PV continues with cumulative levels expected to reach over 123 GW by the end of this 10-year assessment period (up from 89 GW reported in the _2023 LTRA_ , an increase of 38%). There are currently 58.7 GW of installed BTM solar PV across the North American BPS. BTM solar PV generators, like grid-connected solar PV, are also VERs. In large penetrations, their predictable change in output from the time of day contributes to steep ramps in demand. As the sun sets and output diminishes, grid resources must make up for the decrease in solar generation and increase in demand that was being served. The opposite ramp occurs during morning hours; it may be less impactful to reliability but can be challenging for grid-connected generator scheduling and dispatch. **Figure 13** shows the current and projected BTM solar PV by area through 2034.\n\n**Figure 13: Solar PV DER Capacity Existing and Planned through 2034**\n\n### **Generation Retirements**\n\nThe total capacity of traditional baseload generation fuel types will continue to decline as older generators retire. Generators become confirmed for retirement according to various processes in place in the Interconnections, such as regional planning tariffs in the wholesale electricity market areas or the integrated resource planning process in vertically integrated states. Properly designed mechanisms can prevent generators from retiring before planners can study and address reliability issues that could occur. Currently, over 79 GW of fossil-fired and nuclear generating capacity is being retired over this assessment period (see **Figure 14** ), a small decrease from the 83 GW in retirements reported in the _2023 LTRA_ . This capacity includes generators that are confirmed for retirement through retirement planning processes or that have indicated plans to retire to an ISO/RTO or Planning Coordinator.\n\n**Figure 14: Projected Generation Retirement Capacity through 2034**\n\nAdditional fossil-fired generator retirements are expected and would result in further loss of existing capacity. Generator Owners often announce plans to retire generator units before initiating the\n\n**2023 LTRA Total Retirements**\n\n## …","pages":[{"line_numbers":[],"markdown":"**Resource Mix Changes**\n\n**2024 Long-Term Reliability Assessment**\n\n**27**\n\n**Solar PV Distributed Energy Resource Growth** Behind-the-meter (BTM) solar PV generators are solar PV resources connected on the distribution system, such as residential rooftop solar systems. The rapid growth of BTM solar PV continues with cumulative levels expected to reach over 123 GW by the end of this 10-year assessment period (up from 89 GW reported in the _2023 LTRA_ , an increase of 38%). There are currently 58.7 GW of installed BTM solar PV across the North American BPS. BTM solar PV generators, like grid-connected solar PV, are also VERs. In large penetrations, their predictable change in output from the time of day contributes to steep ramps in demand. As the sun sets and output diminishes, grid resources must make up for the decrease in solar generation and increase in demand that was being served. The opposite ramp occurs during morning hours; it may be less impactful to reliability but can be challenging for grid-connected generator scheduling and dispatch. **Figure 13** shows the current and projected BTM solar PV by area through 2034.\n\n**Figure 13: Solar PV DER Capacity Existing and Planned through 2034**\n\n### **Generation Retirements**\n\nThe total capacity of traditional baseload generation fuel types will continue to decline as older generators retire. Generators become confirmed for retirement according to various processes in place in the Interconnections, such as regional planning tariffs in the wholesale electricity market areas or the integrated resource planning process in vertically integrated states. Properly designed mechanisms can prevent generators from retiring before planners can study and address reliability issues that could occur. Currently, over 79 GW of fossil-fired and nuclear generating capacity is being retired over this assessment period (see **Figure 14** ), a small decrease from the 83 GW in retirements reported in the _2023 LTRA_ . This capacity includes generators that are confirmed for retirement through retirement planning processes or that have indicated plans to retire to an ISO/RTO or Planning Coordinator.\n\n**Figure 14: Projected Generation Retirement Capacity through 2034**\n\nAdditional fossil-fired generator retirements are expected and would result in further loss of existing capacity. Generator Owners often announce plans to retire generator units before initiating the\n\n**2023 LTRA Total Retirements**\n\n## …","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**Resource Mix Changes**\n\n**2024 Long-Term Reliability Assessment**\n\n**27**\n\n**Solar PV Distributed Energy Resource Growth** Behind-the-meter (BTM) solar PV generators are solar PV resources connected on the distribution system, such as residential rooftop solar systems. The rapid growth of BTM solar PV continues with cumulative levels expected to reach over 123 GW by the end of this 10-year assessment period (up from 89 GW reported in the _2023 LTRA_ , an increase of 38%). There are currently 58.7 GW of installed BTM solar PV across the North American BPS. BTM solar PV generators, like grid-connected solar PV, are also VERs. In large penetrations, their predictable change in output from the time of day contributes to steep ramps in demand. As the sun sets and output diminishes, grid resources must make up for the decrease in solar generation and increase in demand that was being served. The opposite ramp occurs during morning hours; it may be less impactful to reliability but can be challenging for grid-connected generator scheduling and dispatch. **Figure 13** shows the current and projected BTM solar PV by area through 2034.\n\n**Figure 13: Solar PV DER Capacity Existing and Planned through 2034**\n\n### **Generation Retirements**\n\nThe total capacity of traditional baseload generation fuel types will continue to decline as older generators retire. Generators become confirmed for retirement according to various processes in place in the Interconnections, such as regional planning tariffs in the wholesale electricity market areas or the integrated resource planning process in vertically integrated states. Properly designed mechanisms can prevent generators from retiring before planners can study and address reliability issues that could occur. Currently, over 79 GW of fossil-fired and nuclear generating capacity is being retired over this assessment period (see **Figure 14** ), a small decrease from the 83 GW in retirements reported in the _2023 LTRA_ . This capacity includes generators that are confirmed for retirement through retirement planning processes or that have indicated plans to retire to an ISO/RTO or Planning Coordinator.\n\n**Figure 14: Projected Generation Retirement Capacity through 2034**\n\nAdditional fossil-fired generator retirements are expected and would result in further loss of existing capacity. Generator Owners often announce plans to retire generator units before initiating the\n\n**2023 LTRA Total Retirements**\n\n## …","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2024-ltra_corrected_july_2025_p27.pdf"]},"response":{"errors":[],"extract_id":"extract_7fd4cd65569b1bbb9ef9117acea6eaf4","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**Resource Mix Changes**\n\n**2024 Long-Term Reliability Assessment**\n\n**27**\n\n**Solar PV Distributed Energy Resource Growth** Behind-the-meter (BTM) solar PV generators are solar PV resources connected on the distribution system, such as residential rooftop solar systems. The rapid growth of BTM solar PV continues with cumulative levels expected to reach over 123 GW by the end of this 10-year assessment period (up from 89 GW reported in the _2023 LTRA_ , an increase of 38%). There are currently 58.7 GW of installed BTM solar PV across the North American BPS. BTM solar PV generators, like grid-connected solar PV, are also VERs. In large penetrations, their predictable change in output from the time of day contributes to steep ramps in demand. As the sun sets and output diminishes, grid resources must make up for the decrease in solar generation and increase in demand that was being served. The opposite ramp occurs during morning hours; it may be less impactful to reliability but can be challenging for grid-connected generator scheduling and dispatch. **Figure 13** shows the current and projected BTM solar PV by area through 2034.\n\n**Figure 13: Solar PV DER Capacity Existing and Planned through 2034**\n\n### **Generation Retirements**\n\nThe total capacity of traditional baseload generation fuel types will continue to decline as older generators retire. Generators become confirmed for retirement according to various processes in place in the Interconnections, such as regional planning tariffs in the wholesale electricity market areas or the integrated resource planning process in vertically integrated states. Properly designed mechanisms can prevent generators from retiring before planners can study and address reliability issues that could occur. Currently, over 79 GW of fossil-fired and nuclear generating capacity is being retired over this assessment period (see **Figure 14** ), a small decrease from the 83 GW in retirements reported in the _2023 LTRA_ . This capacity includes generators that are confirmed for retirement through retirement planning processes or that have indicated plans to retire to an ISO/RTO or Planning Coordinator.\n\n**Figure 14: Projected Generation Retirement Capacity through 2034**\n\nAdditional fossil-fired generator retirements are expected and would result in further loss of existing capacity. Generator Owners often announce plans to retire generator units before initiating the\n\n**2023 LTRA Total Retirements**\n\n## …"],"full_content":"**Resource Mix Changes**\n\n**2024 Long-Term Reliability Assessment**\n\n**27**\n\n**Solar PV Distributed Energy Resource Growth** Behind-the-meter (BTM) solar PV generators are solar PV resources connected on the distribution system, such as residential rooftop solar systems. The rapid growth of BTM solar PV continues with cumulative levels expected to reach over 123 GW by the end of this 10-year assessment period (up from 89 GW reported in the _2023 LTRA_ , an increase of 38%). There are currently 58.7 GW of installed BTM solar PV across the North American BPS. BTM solar PV generators, like grid-connected solar PV, are also VERs. In large penetrations, their predictable change in output from the time of day contributes to steep ramps in demand. As the sun sets and output diminishes, grid resources must make up for the decrease in solar generation and increase in demand that was being served. The opposite ramp occurs during morning hours; it may be less impactful to reliability but can be challenging for grid-connected generator scheduling and dispatch. **Figure 13** shows the current and projected BTM solar PV by area through 2034.\n\n**Figure 13: Solar PV DER Capacity Existing and Planned through 2034**\n\n### **Generation Retirements**\n\nThe total capacity of traditional baseload generation fuel types will continue to decline as older generators retire. Generators become confirmed for retirement according to various processes in place in the Interconnections, such as regional planning tariffs in the wholesale electricity market areas or the integrated resource planning process in vertically integrated states. Properly designed mechanisms can prevent generators from retiring before planners can study and address reliability issues that could occur. Currently, over 79 GW of fossil-fired and nuclear generating capacity is being retired over this assessment period (see **Figure 14** ), a small decrease from the 83 GW in retirements reported in the _2023 LTRA_ . This capacity includes generators that are confirmed for retirement through retirement planning processes or that have indicated plans to retire to an ISO/RTO or Planning Coordinator.\n\n**Figure 14: Projected Generation Retirement Capacity through 2034**\n\nAdditional fossil-fired generator retirements are expected and would result in further loss of existing capacity. Generator Owners often announce plans to retire generator units before initiating the\n\n**2023 LTRA Total Retirements**\n\n## …","publish_date":null,"title":"…","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2024-ltra_corrected_july_2025_p27.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2024-ltra_corrected_july_2025_p27.pdf"},"request":{"example_id":"chart/2024-ltra_corrected_july_2025_p27","product_type":"parse","source_file_path":"2024-ltra_corrected_july_2025_p27.pdf"},"started_at":"2026-11-04T19:44:48.303573"} +{"completed_at":"2026-11-04T19:44:51.755707","latency_in_ms":2534,"output":{"example_id":"chart/2024-ltra_corrected_july_2025_p28","grounded_pages":[],"layout_pages":[],"markdown":"**Resource Mix Changes**\n\n**2024 Long-Term Reliability Assessment**\n\n**28**\n\ninterconnection planning process, and the announced plans or timing may be subject to change before the retirement is confirmed. Wholesale electricity market areas, where merchant electric generators make up a large part of the generating fleet, have more uncertainty around future generator retirements, making resource planning and adequacy assessment difficult. **Figure 15** shows the total capacity of reported retirements (i.e., reported to ISO/RTOs and planning entities) as well as owner-announced, unconfirmed retirements of fossil-fueled and nuclear generators across the BPS over the next 10 years in each assessment area. ¹⁹ This total of confirmed and announced-potential retirements over the next 10 years is over 115 GW (3 GW lower than 10-year projections in the _2023 LTRA_ ).\n\n**Figure 15: Projected Capacity Retirements of Nuclear and Fossil Generation 2024–2034**\n\n19 Confirmed generator retirements are reported to NERC by each assessment area in this _2024 LTRA_ development process. NERC obtained data on announced, unconfirmed generator retirements from Energy Ventures Analysis, Inc. and from each assessment area. Some sources of information on announced generator retirements include EIA 860 data, trade press, and utility integrated resource plans.\n\n**Natural Gas Fuel Reliance Trends** Natural-gas-fired generators are and will remain a critical resource for BPS reliability in many areas over the 10-year assessment period, especially during winter. These generators provide many necessary reliability attributes that are exiting the system as traditional generators retire and inverter- based renewable resources take their place in the resource mix. Natural-gas-fired generators are dispatchable and provide the ERSs of inertia, frequency response, and ramping flexibility. In winter, when peak demand in most areas occurs during early morning hours, natural-gas-fired generation is at its highest contribution to the resource mix in many areas. Severe winter weather events in 2021 and 2022 provided stark evidence of the critical nature of natural gas as a generator fuel and the importance of secure supplies during times of extreme electricity demand. As the generation resource mix evolves and becomes more weather-dependent, fuel assurance becomes increasingly critical when conducting reliability assessments. Until recently, comparing seasonally available capacity to a predetermined PRM level has sufficed as a means to assess resource adequacy. However, resource adequacy is now only the first step of a meaningful reliability assessment. Energy adequacy must now be layered on top to assess the risk that available capacity may be rendered unavailable due to phenomena including low wind or solar output and/or natural gas supply and transportation issues. This year, the Energy Information Administration (EIA) projected that more than 40% of delivered natural gas would be consumed by power generators, more than 0.6 Bcf/d higher than was consumed by power generators in 2023 and 7% higher than what the EIA had projected for 2024 in the fall of last year. Natural-gas-fired power plants generated 42% of the electrical energy consumed by end- use electricity customers in 2023 and are on track to maintain that share of generation this year. A significant percentage of natural-gas-fired power plants rely upon as-available, non-firm gas supply and transportation arrangements. Non-firm natural gas supply and transportation is generally sufficient for electric generators most of the year. However, during extreme cold weather, demand for natural gas by both generators and natural gas distribution companies (generally firm shippers) can at the same time dramatically increase. In these instances, generators that lack firm supply and transport arrangements are at risk of fuel unavailability, and when winter weather impacts gas production facilities, the resulting imbalance in pipeline injections and withdrawals can put at risk even firm pipeline customers as preparatory linepack is rapidly depleted.","pages":[{"line_numbers":[],"markdown":"**Resource Mix Changes**\n\n**2024 Long-Term Reliability Assessment**\n\n**28**\n\ninterconnection planning process, and the announced plans or timing may be subject to change before the retirement is confirmed. Wholesale electricity market areas, where merchant electric generators make up a large part of the generating fleet, have more uncertainty around future generator retirements, making resource planning and adequacy assessment difficult. **Figure 15** shows the total capacity of reported retirements (i.e., reported to ISO/RTOs and planning entities) as well as owner-announced, unconfirmed retirements of fossil-fueled and nuclear generators across the BPS over the next 10 years in each assessment area. ¹⁹ This total of confirmed and announced-potential retirements over the next 10 years is over 115 GW (3 GW lower than 10-year projections in the _2023 LTRA_ ).\n\n**Figure 15: Projected Capacity Retirements of Nuclear and Fossil Generation 2024–2034**\n\n19 Confirmed generator retirements are reported to NERC by each assessment area in this _2024 LTRA_ development process. NERC obtained data on announced, unconfirmed generator retirements from Energy Ventures Analysis, Inc. and from each assessment area. Some sources of information on announced generator retirements include EIA 860 data, trade press, and utility integrated resource plans.\n\n**Natural Gas Fuel Reliance Trends** Natural-gas-fired generators are and will remain a critical resource for BPS reliability in many areas over the 10-year assessment period, especially during winter. These generators provide many necessary reliability attributes that are exiting the system as traditional generators retire and inverter- based renewable resources take their place in the resource mix. Natural-gas-fired generators are dispatchable and provide the ERSs of inertia, frequency response, and ramping flexibility. In winter, when peak demand in most areas occurs during early morning hours, natural-gas-fired generation is at its highest contribution to the resource mix in many areas. Severe winter weather events in 2021 and 2022 provided stark evidence of the critical nature of natural gas as a generator fuel and the importance of secure supplies during times of extreme electricity demand. As the generation resource mix evolves and becomes more weather-dependent, fuel assurance becomes increasingly critical when conducting reliability assessments. Until recently, comparing seasonally available capacity to a predetermined PRM level has sufficed as a means to assess resource adequacy. However, resource adequacy is now only the first step of a meaningful reliability assessment. Energy adequacy must now be layered on top to assess the risk that available capacity may be rendered unavailable due to phenomena including low wind or solar output and/or natural gas supply and transportation issues. This year, the Energy Information Administration (EIA) projected that more than 40% of delivered natural gas would be consumed by power generators, more than 0.6 Bcf/d higher than was consumed by power generators in 2023 and 7% higher than what the EIA had projected for 2024 in the fall of last year. Natural-gas-fired power plants generated 42% of the electrical energy consumed by end- use electricity customers in 2023 and are on track to maintain that share of generation this year. A significant percentage of natural-gas-fired power plants rely upon as-available, non-firm gas supply and transportation arrangements. Non-firm natural gas supply and transportation is generally sufficient for electric generators most of the year. However, during extreme cold weather, demand for natural gas by both generators and natural gas distribution companies (generally firm shippers) can at the same time dramatically increase. In these instances, generators that lack firm supply and transport arrangements are at risk of fuel unavailability, and when winter weather impacts gas production facilities, the resulting imbalance in pipeline injections and withdrawals can put at risk even firm pipeline customers as preparatory linepack is rapidly depleted.","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**Resource Mix Changes**\n\n**2024 Long-Term Reliability Assessment**\n\n**28**\n\ninterconnection planning process, and the announced plans or timing may be subject to change before the retirement is confirmed. Wholesale electricity market areas, where merchant electric generators make up a large part of the generating fleet, have more uncertainty around future generator retirements, making resource planning and adequacy assessment difficult. **Figure 15** shows the total capacity of reported retirements (i.e., reported to ISO/RTOs and planning entities) as well as owner-announced, unconfirmed retirements of fossil-fueled and nuclear generators across the BPS over the next 10 years in each assessment area. ¹⁹ This total of confirmed and announced-potential retirements over the next 10 years is over 115 GW (3 GW lower than 10-year projections in the _2023 LTRA_ ).\n\n**Figure 15: Projected Capacity Retirements of Nuclear and Fossil Generation 2024–2034**\n\n19 Confirmed generator retirements are reported to NERC by each assessment area in this _2024 LTRA_ development process. NERC obtained data on announced, unconfirmed generator retirements from Energy Ventures Analysis, Inc. and from each assessment area. Some sources of information on announced generator retirements include EIA 860 data, trade press, and utility integrated resource plans.\n\n**Natural Gas Fuel Reliance Trends** Natural-gas-fired generators are and will remain a critical resource for BPS reliability in many areas over the 10-year assessment period, especially during winter. These generators provide many necessary reliability attributes that are exiting the system as traditional generators retire and inverter- based renewable resources take their place in the resource mix. Natural-gas-fired generators are dispatchable and provide the ERSs of inertia, frequency response, and ramping flexibility. In winter, when peak demand in most areas occurs during early morning hours, natural-gas-fired generation is at its highest contribution to the resource mix in many areas. Severe winter weather events in 2021 and 2022 provided stark evidence of the critical nature of natural gas as a generator fuel and the importance of secure supplies during times of extreme electricity demand. As the generation resource mix evolves and becomes more weather-dependent, fuel assurance becomes increasingly critical when conducting reliability assessments. Until recently, comparing seasonally available capacity to a predetermined PRM level has sufficed as a means to assess resource adequacy. However, resource adequacy is now only the first step of a meaningful reliability assessment. Energy adequacy must now be layered on top to assess the risk that available capacity may be rendered unavailable due to phenomena including low wind or solar output and/or natural gas supply and transportation issues. This year, the Energy Information Administration (EIA) projected that more than 40% of delivered natural gas would be consumed by power generators, more than 0.6 Bcf/d higher than was consumed by power generators in 2023 and 7% higher than what the EIA had projected for 2024 in the fall of last year. Natural-gas-fired power plants generated 42% of the electrical energy consumed by end- use electricity customers in 2023 and are on track to maintain that share of generation this year. A significant percentage of natural-gas-fired power plants rely upon as-available, non-firm gas supply and transportation arrangements. Non-firm natural gas supply and transportation is generally sufficient for electric generators most of the year. However, during extreme cold weather, demand for natural gas by both generators and natural gas distribution companies (generally firm shippers) can at the same time dramatically increase. In these instances, generators that lack firm supply and transport arrangements are at risk of fuel unavailability, and when winter weather impacts gas production facilities, the resulting imbalance in pipeline injections and withdrawals can put at risk even firm pipeline customers as preparatory linepack is rapidly depleted.","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2024-ltra_corrected_july_2025_p28.pdf"]},"response":{"errors":[],"extract_id":"extract_3510823c16b518778ad3d116046f0fe0","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**Resource Mix Changes**\n\n**2024 Long-Term Reliability Assessment**\n\n**28**\n\ninterconnection planning process, and the announced plans or timing may be subject to change before the retirement is confirmed. Wholesale electricity market areas, where merchant electric generators make up a large part of the generating fleet, have more uncertainty around future generator retirements, making resource planning and adequacy assessment difficult. **Figure 15** shows the total capacity of reported retirements (i.e., reported to ISO/RTOs and planning entities) as well as owner-announced, unconfirmed retirements of fossil-fueled and nuclear generators across the BPS over the next 10 years in each assessment area. ¹⁹ This total of confirmed and announced-potential retirements over the next 10 years is over 115 GW (3 GW lower than 10-year projections in the _2023 LTRA_ ).\n\n**Figure 15: Projected Capacity Retirements of Nuclear and Fossil Generation 2024–2034**\n\n19 Confirmed generator retirements are reported to NERC by each assessment area in this _2024 LTRA_ development process. NERC obtained data on announced, unconfirmed generator retirements from Energy Ventures Analysis, Inc. and from each assessment area. Some sources of information on announced generator retirements include EIA 860 data, trade press, and utility integrated resource plans.\n\n**Natural Gas Fuel Reliance Trends** Natural-gas-fired generators are and will remain a critical resource for BPS reliability in many areas over the 10-year assessment period, especially during winter. These generators provide many necessary reliability attributes that are exiting the system as traditional generators retire and inverter- based renewable resources take their place in the resource mix. Natural-gas-fired generators are dispatchable and provide the ERSs of inertia, frequency response, and ramping flexibility. In winter, when peak demand in most areas occurs during early morning hours, natural-gas-fired generation is at its highest contribution to the resource mix in many areas. Severe winter weather events in 2021 and 2022 provided stark evidence of the critical nature of natural gas as a generator fuel and the importance of secure supplies during times of extreme electricity demand. As the generation resource mix evolves and becomes more weather-dependent, fuel assurance becomes increasingly critical when conducting reliability assessments. Until recently, comparing seasonally available capacity to a predetermined PRM level has sufficed as a means to assess resource adequacy. However, resource adequacy is now only the first step of a meaningful reliability assessment. Energy adequacy must now be layered on top to assess the risk that available capacity may be rendered unavailable due to phenomena including low wind or solar output and/or natural gas supply and transportation issues. This year, the Energy Information Administration (EIA) projected that more than 40% of delivered natural gas would be consumed by power generators, more than 0.6 Bcf/d higher than was consumed by power generators in 2023 and 7% higher than what the EIA had projected for 2024 in the fall of last year. Natural-gas-fired power plants generated 42% of the electrical energy consumed by end- use electricity customers in 2023 and are on track to maintain that share of generation this year. A significant percentage of natural-gas-fired power plants rely upon as-available, non-firm gas supply and transportation arrangements. Non-firm natural gas supply and transportation is generally sufficient for electric generators most of the year. However, during extreme cold weather, demand for natural gas by both generators and natural gas distribution companies (generally firm shippers) can at the same time dramatically increase. In these instances, generators that lack firm supply and transport arrangements are at risk of fuel unavailability, and when winter weather impacts gas production facilities, the resulting imbalance in pipeline injections and withdrawals can put at risk even firm pipeline customers as preparatory linepack is rapidly depleted."],"full_content":"**Resource Mix Changes**\n\n**2024 Long-Term Reliability Assessment**\n\n**28**\n\ninterconnection planning process, and the announced plans or timing may be subject to change before the retirement is confirmed. Wholesale electricity market areas, where merchant electric generators make up a large part of the generating fleet, have more uncertainty around future generator retirements, making resource planning and adequacy assessment difficult. **Figure 15** shows the total capacity of reported retirements (i.e., reported to ISO/RTOs and planning entities) as well as owner-announced, unconfirmed retirements of fossil-fueled and nuclear generators across the BPS over the next 10 years in each assessment area. ¹⁹ This total of confirmed and announced-potential retirements over the next 10 years is over 115 GW (3 GW lower than 10-year projections in the _2023 LTRA_ ).\n\n**Figure 15: Projected Capacity Retirements of Nuclear and Fossil Generation 2024–2034**\n\n19 Confirmed generator retirements are reported to NERC by each assessment area in this _2024 LTRA_ development process. NERC obtained data on announced, unconfirmed generator retirements from Energy Ventures Analysis, Inc. and from each assessment area. Some sources of information on announced generator retirements include EIA 860 data, trade press, and utility integrated resource plans.\n\n**Natural Gas Fuel Reliance Trends** Natural-gas-fired generators are and will remain a critical resource for BPS reliability in many areas over the 10-year assessment period, especially during winter. These generators provide many necessary reliability attributes that are exiting the system as traditional generators retire and inverter- based renewable resources take their place in the resource mix. Natural-gas-fired generators are dispatchable and provide the ERSs of inertia, frequency response, and ramping flexibility. In winter, when peak demand in most areas occurs during early morning hours, natural-gas-fired generation is at its highest contribution to the resource mix in many areas. Severe winter weather events in 2021 and 2022 provided stark evidence of the critical nature of natural gas as a generator fuel and the importance of secure supplies during times of extreme electricity demand. As the generation resource mix evolves and becomes more weather-dependent, fuel assurance becomes increasingly critical when conducting reliability assessments. Until recently, comparing seasonally available capacity to a predetermined PRM level has sufficed as a means to assess resource adequacy. However, resource adequacy is now only the first step of a meaningful reliability assessment. Energy adequacy must now be layered on top to assess the risk that available capacity may be rendered unavailable due to phenomena including low wind or solar output and/or natural gas supply and transportation issues. This year, the Energy Information Administration (EIA) projected that more than 40% of delivered natural gas would be consumed by power generators, more than 0.6 Bcf/d higher than was consumed by power generators in 2023 and 7% higher than what the EIA had projected for 2024 in the fall of last year. Natural-gas-fired power plants generated 42% of the electrical energy consumed by end- use electricity customers in 2023 and are on track to maintain that share of generation this year. A significant percentage of natural-gas-fired power plants rely upon as-available, non-firm gas supply and transportation arrangements. Non-firm natural gas supply and transportation is generally sufficient for electric generators most of the year. However, during extreme cold weather, demand for natural gas by both generators and natural gas distribution companies (generally firm shippers) can at the same time dramatically increase. In these instances, generators that lack firm supply and transport arrangements are at risk of fuel unavailability, and when winter weather impacts gas production facilities, the resulting imbalance in pipeline injections and withdrawals can put at risk even firm pipeline customers as preparatory linepack is rapidly depleted.","publish_date":null,"title":"interconnection planning process, and the announced plans or timing may be subject to change Natural Gas Fuel Reliance Trends Natural-gas-fired","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2024-ltra_corrected_july_2025_p28.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2024-ltra_corrected_july_2025_p28.pdf"},"request":{"example_id":"chart/2024-ltra_corrected_july_2025_p28","product_type":"parse","source_file_path":"2024-ltra_corrected_july_2025_p28.pdf"},"started_at":"2026-11-04T19:44:49.220962"} +{"completed_at":"2026-11-04T19:44:52.341947","latency_in_ms":1192,"output":{"example_id":"chart/2024-ltra_corrected_july_2025_p29","grounded_pages":[],"layout_pages":[],"markdown":"**Resource Mix Changes**\n\n**2024 Long-Term Reliability Assessment**\n\n**29**\n\nThe _2024 LTRA_ projects that an additional 6,500 MW of new natural-gas-fired generation will be added in North America over the next five years. While some assessment areas project net decreases in natural-gas-fired capacity, other areas report increases ranging from 3% to over 50%. In the United States, 8 out of 13 assessment areas are adding capacity to their fleet of natural-gas- fired power plants over the next 10 years, amounting to over 10 GW of new natural-gas-fired power capacity. Altogether, those additions translate to an increase in natural gas use for power generation of more than 84 Bcf (245 MMcf/d). Whether this increase in natural gas generation capacity is sufficient to meet the increase in net internal demand projected for these same assessment areas remains to be seen, but based on the capacity additions alone, some areas are set to see an insufficient increase in gas pipeline capacity, as per the EIA’s database of nationwide U.S. gas pipeline projects. (see **Figure 16** ). PJM is one such area, projected to see a 3% increase in natural-gas-fired generation capacity (+2,500 MW). Net internal electricity demand in PJM is forecasted to rise by 25 GW over the planning horizon at the same time that non-variable resources are expected to shrink their share of the capacity mix from 96% to 88%. To the extent that variable resources are uncertain during times of peak stress on the grid, there is a risk of gas delivery capacity being insufficient for the amount of natural gas supply that could be required to meet the demand increase. SPP is another assessment area where natural- gas-fired power capacity is set to rise, by 5%, or roughly 1,500 MW. In both PJM and SPP, there is a risk of a shortfall in natural gas pipeline capacity that could facilitate delivery of the additional fuel that is necessary to meet their capacity expansion plans.\n\n**Figure 16: Annual U.S. Natural Gas Pipeline Capacity Additions by Type (2017– 2024) Bcf/d (Source: U.S. Energy Information Administration)**\n\n### **Reliability Implications**\n\nThe addition of variable resources, primarily wind and solar PV, and the retirement of conventional generation are fundamentally changing how the BPS is planned and operated. With electricity supplies coming increasingly from VERs and natural-gas-fired generators, there is a growing risk that supplies can fall short of demand during some periods. Geographically diverse wind and solar resources and loads can help reduce these risks, but they require robust transmission networks, comprehensive energy and transfer capability analysis, and effective operating procedures and market mechanisms. **Maintaining ERSs for Grid Support** For the grid to operate reliably, it needs resources that are not only sufficient for meeting demand and energy requirements but also capable of controlling voltages, maintaining stable frequency across the system, and flexibly ramping up or down to dispatch control at all times. Conventional generators, such as nuclear, coal, and natural gas-fired generators, provide much of the ERSs that support reliable operation today. New resources on the grid are almost all IBRs, which have different physical and operating characteristics that affect the level of ERSs that they can provide. Without the spinning mass or traditional excitation systems providing the voltage control of a hydroelectric or thermal generator,\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n2017 2018 2019 2020 2021 2022 2023 2024\n\n**Pipeline Additions**\n\nInterstate Intrastate","pages":[{"line_numbers":[],"markdown":"**Resource Mix Changes**\n\n**2024 Long-Term Reliability Assessment**\n\n**29**\n\nThe _2024 LTRA_ projects that an additional 6,500 MW of new natural-gas-fired generation will be added in North America over the next five years. While some assessment areas project net decreases in natural-gas-fired capacity, other areas report increases ranging from 3% to over 50%. In the United States, 8 out of 13 assessment areas are adding capacity to their fleet of natural-gas- fired power plants over the next 10 years, amounting to over 10 GW of new natural-gas-fired power capacity. Altogether, those additions translate to an increase in natural gas use for power generation of more than 84 Bcf (245 MMcf/d). Whether this increase in natural gas generation capacity is sufficient to meet the increase in net internal demand projected for these same assessment areas remains to be seen, but based on the capacity additions alone, some areas are set to see an insufficient increase in gas pipeline capacity, as per the EIA’s database of nationwide U.S. gas pipeline projects. (see **Figure 16** ). PJM is one such area, projected to see a 3% increase in natural-gas-fired generation capacity (+2,500 MW). Net internal electricity demand in PJM is forecasted to rise by 25 GW over the planning horizon at the same time that non-variable resources are expected to shrink their share of the capacity mix from 96% to 88%. To the extent that variable resources are uncertain during times of peak stress on the grid, there is a risk of gas delivery capacity being insufficient for the amount of natural gas supply that could be required to meet the demand increase. SPP is another assessment area where natural- gas-fired power capacity is set to rise, by 5%, or roughly 1,500 MW. In both PJM and SPP, there is a risk of a shortfall in natural gas pipeline capacity that could facilitate delivery of the additional fuel that is necessary to meet their capacity expansion plans.\n\n**Figure 16: Annual U.S. Natural Gas Pipeline Capacity Additions by Type (2017– 2024) Bcf/d (Source: U.S. Energy Information Administration)**\n\n### **Reliability Implications**\n\nThe addition of variable resources, primarily wind and solar PV, and the retirement of conventional generation are fundamentally changing how the BPS is planned and operated. With electricity supplies coming increasingly from VERs and natural-gas-fired generators, there is a growing risk that supplies can fall short of demand during some periods. Geographically diverse wind and solar resources and loads can help reduce these risks, but they require robust transmission networks, comprehensive energy and transfer capability analysis, and effective operating procedures and market mechanisms. **Maintaining ERSs for Grid Support** For the grid to operate reliably, it needs resources that are not only sufficient for meeting demand and energy requirements but also capable of controlling voltages, maintaining stable frequency across the system, and flexibly ramping up or down to dispatch control at all times. Conventional generators, such as nuclear, coal, and natural gas-fired generators, provide much of the ERSs that support reliable operation today. New resources on the grid are almost all IBRs, which have different physical and operating characteristics that affect the level of ERSs that they can provide. Without the spinning mass or traditional excitation systems providing the voltage control of a hydroelectric or thermal generator,\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n2017 2018 2019 2020 2021 2022 2023 2024\n\n**Pipeline Additions**\n\nInterstate Intrastate","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**Resource Mix Changes**\n\n**2024 Long-Term Reliability Assessment**\n\n**29**\n\nThe _2024 LTRA_ projects that an additional 6,500 MW of new natural-gas-fired generation will be added in North America over the next five years. While some assessment areas project net decreases in natural-gas-fired capacity, other areas report increases ranging from 3% to over 50%. In the United States, 8 out of 13 assessment areas are adding capacity to their fleet of natural-gas- fired power plants over the next 10 years, amounting to over 10 GW of new natural-gas-fired power capacity. Altogether, those additions translate to an increase in natural gas use for power generation of more than 84 Bcf (245 MMcf/d). Whether this increase in natural gas generation capacity is sufficient to meet the increase in net internal demand projected for these same assessment areas remains to be seen, but based on the capacity additions alone, some areas are set to see an insufficient increase in gas pipeline capacity, as per the EIA’s database of nationwide U.S. gas pipeline projects. (see **Figure 16** ). PJM is one such area, projected to see a 3% increase in natural-gas-fired generation capacity (+2,500 MW). Net internal electricity demand in PJM is forecasted to rise by 25 GW over the planning horizon at the same time that non-variable resources are expected to shrink their share of the capacity mix from 96% to 88%. To the extent that variable resources are uncertain during times of peak stress on the grid, there is a risk of gas delivery capacity being insufficient for the amount of natural gas supply that could be required to meet the demand increase. SPP is another assessment area where natural- gas-fired power capacity is set to rise, by 5%, or roughly 1,500 MW. In both PJM and SPP, there is a risk of a shortfall in natural gas pipeline capacity that could facilitate delivery of the additional fuel that is necessary to meet their capacity expansion plans.\n\n**Figure 16: Annual U.S. Natural Gas Pipeline Capacity Additions by Type (2017– 2024) Bcf/d (Source: U.S. Energy Information Administration)**\n\n### **Reliability Implications**\n\nThe addition of variable resources, primarily wind and solar PV, and the retirement of conventional generation are fundamentally changing how the BPS is planned and operated. With electricity supplies coming increasingly from VERs and natural-gas-fired generators, there is a growing risk that supplies can fall short of demand during some periods. Geographically diverse wind and solar resources and loads can help reduce these risks, but they require robust transmission networks, comprehensive energy and transfer capability analysis, and effective operating procedures and market mechanisms. **Maintaining ERSs for Grid Support** For the grid to operate reliably, it needs resources that are not only sufficient for meeting demand and energy requirements but also capable of controlling voltages, maintaining stable frequency across the system, and flexibly ramping up or down to dispatch control at all times. Conventional generators, such as nuclear, coal, and natural gas-fired generators, provide much of the ERSs that support reliable operation today. New resources on the grid are almost all IBRs, which have different physical and operating characteristics that affect the level of ERSs that they can provide. Without the spinning mass or traditional excitation systems providing the voltage control of a hydroelectric or thermal generator,\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n2017 2018 2019 2020 2021 2022 2023 2024\n\n**Pipeline Additions**\n\nInterstate Intrastate","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2024-ltra_corrected_july_2025_p29.pdf"]},"response":{"errors":[],"extract_id":"extract_d9bfdf354d714226d02cf3c64c03bf14","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**Resource Mix Changes**\n\n**2024 Long-Term Reliability Assessment**\n\n**29**\n\nThe _2024 LTRA_ projects that an additional 6,500 MW of new natural-gas-fired generation will be added in North America over the next five years. While some assessment areas project net decreases in natural-gas-fired capacity, other areas report increases ranging from 3% to over 50%. In the United States, 8 out of 13 assessment areas are adding capacity to their fleet of natural-gas- fired power plants over the next 10 years, amounting to over 10 GW of new natural-gas-fired power capacity. Altogether, those additions translate to an increase in natural gas use for power generation of more than 84 Bcf (245 MMcf/d). Whether this increase in natural gas generation capacity is sufficient to meet the increase in net internal demand projected for these same assessment areas remains to be seen, but based on the capacity additions alone, some areas are set to see an insufficient increase in gas pipeline capacity, as per the EIA’s database of nationwide U.S. gas pipeline projects. (see **Figure 16** ). PJM is one such area, projected to see a 3% increase in natural-gas-fired generation capacity (+2,500 MW). Net internal electricity demand in PJM is forecasted to rise by 25 GW over the planning horizon at the same time that non-variable resources are expected to shrink their share of the capacity mix from 96% to 88%. To the extent that variable resources are uncertain during times of peak stress on the grid, there is a risk of gas delivery capacity being insufficient for the amount of natural gas supply that could be required to meet the demand increase. SPP is another assessment area where natural- gas-fired power capacity is set to rise, by 5%, or roughly 1,500 MW. In both PJM and SPP, there is a risk of a shortfall in natural gas pipeline capacity that could facilitate delivery of the additional fuel that is necessary to meet their capacity expansion plans.\n\n**Figure 16: Annual U.S. Natural Gas Pipeline Capacity Additions by Type (2017– 2024) Bcf/d (Source: U.S. Energy Information Administration)**\n\n### **Reliability Implications**\n\nThe addition of variable resources, primarily wind and solar PV, and the retirement of conventional generation are fundamentally changing how the BPS is planned and operated. With electricity supplies coming increasingly from VERs and natural-gas-fired generators, there is a growing risk that supplies can fall short of demand during some periods. Geographically diverse wind and solar resources and loads can help reduce these risks, but they require robust transmission networks, comprehensive energy and transfer capability analysis, and effective operating procedures and market mechanisms. **Maintaining ERSs for Grid Support** For the grid to operate reliably, it needs resources that are not only sufficient for meeting demand and energy requirements but also capable of controlling voltages, maintaining stable frequency across the system, and flexibly ramping up or down to dispatch control at all times. Conventional generators, such as nuclear, coal, and natural gas-fired generators, provide much of the ERSs that support reliable operation today. New resources on the grid are almost all IBRs, which have different physical and operating characteristics that affect the level of ERSs that they can provide. Without the spinning mass or traditional excitation systems providing the voltage control of a hydroelectric or thermal generator,\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n2017 2018 2019 2020 2021 2022 2023 2024\n\n**Pipeline Additions**\n\nInterstate Intrastate"],"full_content":"**Resource Mix Changes**\n\n**2024 Long-Term Reliability Assessment**\n\n**29**\n\nThe _2024 LTRA_ projects that an additional 6,500 MW of new natural-gas-fired generation will be added in North America over the next five years. While some assessment areas project net decreases in natural-gas-fired capacity, other areas report increases ranging from 3% to over 50%. In the United States, 8 out of 13 assessment areas are adding capacity to their fleet of natural-gas- fired power plants over the next 10 years, amounting to over 10 GW of new natural-gas-fired power capacity. Altogether, those additions translate to an increase in natural gas use for power generation of more than 84 Bcf (245 MMcf/d). Whether this increase in natural gas generation capacity is sufficient to meet the increase in net internal demand projected for these same assessment areas remains to be seen, but based on the capacity additions alone, some areas are set to see an insufficient increase in gas pipeline capacity, as per the EIA’s database of nationwide U.S. gas pipeline projects. (see **Figure 16** ). PJM is one such area, projected to see a 3% increase in natural-gas-fired generation capacity (+2,500 MW). Net internal electricity demand in PJM is forecasted to rise by 25 GW over the planning horizon at the same time that non-variable resources are expected to shrink their share of the capacity mix from 96% to 88%. To the extent that variable resources are uncertain during times of peak stress on the grid, there is a risk of gas delivery capacity being insufficient for the amount of natural gas supply that could be required to meet the demand increase. SPP is another assessment area where natural- gas-fired power capacity is set to rise, by 5%, or roughly 1,500 MW. In both PJM and SPP, there is a risk of a shortfall in natural gas pipeline capacity that could facilitate delivery of the additional fuel that is necessary to meet their capacity expansion plans.\n\n**Figure 16: Annual U.S. Natural Gas Pipeline Capacity Additions by Type (2017– 2024) Bcf/d (Source: U.S. Energy Information Administration)**\n\n### **Reliability Implications**\n\nThe addition of variable resources, primarily wind and solar PV, and the retirement of conventional generation are fundamentally changing how the BPS is planned and operated. With electricity supplies coming increasingly from VERs and natural-gas-fired generators, there is a growing risk that supplies can fall short of demand during some periods. Geographically diverse wind and solar resources and loads can help reduce these risks, but they require robust transmission networks, comprehensive energy and transfer capability analysis, and effective operating procedures and market mechanisms. **Maintaining ERSs for Grid Support** For the grid to operate reliably, it needs resources that are not only sufficient for meeting demand and energy requirements but also capable of controlling voltages, maintaining stable frequency across the system, and flexibly ramping up or down to dispatch control at all times. Conventional generators, such as nuclear, coal, and natural gas-fired generators, provide much of the ERSs that support reliable operation today. New resources on the grid are almost all IBRs, which have different physical and operating characteristics that affect the level of ERSs that they can provide. Without the spinning mass or traditional excitation systems providing the voltage control of a hydroelectric or thermal generator,\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n2017 2018 2019 2020 2021 2022 2023 2024\n\n**Pipeline Additions**\n\nInterstate Intrastate","publish_date":null,"title":"Reliability Implications","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2024-ltra_corrected_july_2025_p29.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2024-ltra_corrected_july_2025_p29.pdf"},"request":{"example_id":"chart/2024-ltra_corrected_july_2025_p29","product_type":"parse","source_file_path":"2024-ltra_corrected_july_2025_p29.pdf"},"started_at":"2026-11-04T19:44:51.149113"} +{"completed_at":"2026-11-04T19:44:52.975772","latency_in_ms":1116,"output":{"example_id":"chart/2024-ltra_corrected_july_2025_p31","grounded_pages":[],"layout_pages":[],"markdown":"**2024 Long-Term Reliability Assessment**\n\n**31**\n\n## **Demand Trends and Implications**\n\n### **Demand and Energy Projections**\n\nElectricity peak demand and energy growth forecasts over the 10-year assessment period continue to climb higher than at any point in the past two decades. The aggregated assessment area summer peak demand forecast is expected to rise by over 132 GW, and aggregated winter peak demand forecasts are increasing by 149 GW. The growth rates of forecasted peak demand (see **Figure 17** ) and energy (see **Figure 18** ) continue to rise sharply since the 2022 LTRA, reversing an almost two-decade trend of falling or flat growth rates. See **Figure 17** for seasonal peak demand growth over the current and prior assessment periods and **Figure 18** for net energy growth. A map of the primary demand drivers for the North American BPS is illustrated for each assessment area in **Figure 19** . **Electrification and Demand Growth** Electrification of household appliances (e.g., heat pumps for household heating) and projections for electric vehicle growth over this assessment period are components of the demand and energy estimates provided by each assessment area. Since the _2023 LTRA_ , peak season CAGR has risen in all assessment areas except two: NPCC-Maritimes’ winter CAGR fell from 0.98% to 0.63% and WECC-BC’s winter CAGR fell from 1.05% to 0.28%. Rising peak demand forecasts are contributing to the lower reserve margins projected for nearly all assessment areas. **Large Commercial and Industrial Loads** Increasing amounts of large commercial and industrial loads are connecting rapidly to the BPS. Emerging large loads, such as data centers (including crypto and AI) and hydrogen fuel plants, present unique challenges to forecasting and planning for increased demand. Earlier this year, NERC’s RSTC established a Large Loads Task Force to better understand the reliability implications of growth in large loads and develop solutions.\n\n**Figure 17: 10-Year Summer and Winter Peak Demand Growth and Rate Trends**\n\n**Figure 18: Net Energy for Load Growth and Rate Projections**","pages":[{"line_numbers":[],"markdown":"**2024 Long-Term Reliability Assessment**\n\n**31**\n\n## **Demand Trends and Implications**\n\n### **Demand and Energy Projections**\n\nElectricity peak demand and energy growth forecasts over the 10-year assessment period continue to climb higher than at any point in the past two decades. The aggregated assessment area summer peak demand forecast is expected to rise by over 132 GW, and aggregated winter peak demand forecasts are increasing by 149 GW. The growth rates of forecasted peak demand (see **Figure 17** ) and energy (see **Figure 18** ) continue to rise sharply since the 2022 LTRA, reversing an almost two-decade trend of falling or flat growth rates. See **Figure 17** for seasonal peak demand growth over the current and prior assessment periods and **Figure 18** for net energy growth. A map of the primary demand drivers for the North American BPS is illustrated for each assessment area in **Figure 19** . **Electrification and Demand Growth** Electrification of household appliances (e.g., heat pumps for household heating) and projections for electric vehicle growth over this assessment period are components of the demand and energy estimates provided by each assessment area. Since the _2023 LTRA_ , peak season CAGR has risen in all assessment areas except two: NPCC-Maritimes’ winter CAGR fell from 0.98% to 0.63% and WECC-BC’s winter CAGR fell from 1.05% to 0.28%. Rising peak demand forecasts are contributing to the lower reserve margins projected for nearly all assessment areas. **Large Commercial and Industrial Loads** Increasing amounts of large commercial and industrial loads are connecting rapidly to the BPS. Emerging large loads, such as data centers (including crypto and AI) and hydrogen fuel plants, present unique challenges to forecasting and planning for increased demand. Earlier this year, NERC’s RSTC established a Large Loads Task Force to better understand the reliability implications of growth in large loads and develop solutions.\n\n**Figure 17: 10-Year Summer and Winter Peak Demand Growth and Rate Trends**\n\n**Figure 18: Net Energy for Load Growth and Rate Projections**","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**2024 Long-Term Reliability Assessment**\n\n**31**\n\n## **Demand Trends and Implications**\n\n### **Demand and Energy Projections**\n\nElectricity peak demand and energy growth forecasts over the 10-year assessment period continue to climb higher than at any point in the past two decades. The aggregated assessment area summer peak demand forecast is expected to rise by over 132 GW, and aggregated winter peak demand forecasts are increasing by 149 GW. The growth rates of forecasted peak demand (see **Figure 17** ) and energy (see **Figure 18** ) continue to rise sharply since the 2022 LTRA, reversing an almost two-decade trend of falling or flat growth rates. See **Figure 17** for seasonal peak demand growth over the current and prior assessment periods and **Figure 18** for net energy growth. A map of the primary demand drivers for the North American BPS is illustrated for each assessment area in **Figure 19** . **Electrification and Demand Growth** Electrification of household appliances (e.g., heat pumps for household heating) and projections for electric vehicle growth over this assessment period are components of the demand and energy estimates provided by each assessment area. Since the _2023 LTRA_ , peak season CAGR has risen in all assessment areas except two: NPCC-Maritimes’ winter CAGR fell from 0.98% to 0.63% and WECC-BC’s winter CAGR fell from 1.05% to 0.28%. Rising peak demand forecasts are contributing to the lower reserve margins projected for nearly all assessment areas. **Large Commercial and Industrial Loads** Increasing amounts of large commercial and industrial loads are connecting rapidly to the BPS. Emerging large loads, such as data centers (including crypto and AI) and hydrogen fuel plants, present unique challenges to forecasting and planning for increased demand. Earlier this year, NERC’s RSTC established a Large Loads Task Force to better understand the reliability implications of growth in large loads and develop solutions.\n\n**Figure 17: 10-Year Summer and Winter Peak Demand Growth and Rate Trends**\n\n**Figure 18: Net Energy for Load Growth and Rate Projections**","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2024-ltra_corrected_july_2025_p31.pdf"]},"response":{"errors":[],"extract_id":"extract_d1aa4fc9fbb5cb850ad171ca286d164d","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**2024 Long-Term Reliability Assessment**\n\n**31**\n\n## **Demand Trends and Implications**\n\n### **Demand and Energy Projections**\n\nElectricity peak demand and energy growth forecasts over the 10-year assessment period continue to climb higher than at any point in the past two decades. The aggregated assessment area summer peak demand forecast is expected to rise by over 132 GW, and aggregated winter peak demand forecasts are increasing by 149 GW. The growth rates of forecasted peak demand (see **Figure 17** ) and energy (see **Figure 18** ) continue to rise sharply since the 2022 LTRA, reversing an almost two-decade trend of falling or flat growth rates. See **Figure 17** for seasonal peak demand growth over the current and prior assessment periods and **Figure 18** for net energy growth. A map of the primary demand drivers for the North American BPS is illustrated for each assessment area in **Figure 19** . **Electrification and Demand Growth** Electrification of household appliances (e.g., heat pumps for household heating) and projections for electric vehicle growth over this assessment period are components of the demand and energy estimates provided by each assessment area. Since the _2023 LTRA_ , peak season CAGR has risen in all assessment areas except two: NPCC-Maritimes’ winter CAGR fell from 0.98% to 0.63% and WECC-BC’s winter CAGR fell from 1.05% to 0.28%. Rising peak demand forecasts are contributing to the lower reserve margins projected for nearly all assessment areas. **Large Commercial and Industrial Loads** Increasing amounts of large commercial and industrial loads are connecting rapidly to the BPS. Emerging large loads, such as data centers (including crypto and AI) and hydrogen fuel plants, present unique challenges to forecasting and planning for increased demand. Earlier this year, NERC’s RSTC established a Large Loads Task Force to better understand the reliability implications of growth in large loads and develop solutions.\n\n**Figure 17: 10-Year Summer and Winter Peak Demand Growth and Rate Trends**\n\n**Figure 18: Net Energy for Load Growth and Rate Projections**"],"full_content":"**2024 Long-Term Reliability Assessment**\n\n**31**\n\n## **Demand Trends and Implications**\n\n### **Demand and Energy Projections**\n\nElectricity peak demand and energy growth forecasts over the 10-year assessment period continue to climb higher than at any point in the past two decades. The aggregated assessment area summer peak demand forecast is expected to rise by over 132 GW, and aggregated winter peak demand forecasts are increasing by 149 GW. The growth rates of forecasted peak demand (see **Figure 17** ) and energy (see **Figure 18** ) continue to rise sharply since the 2022 LTRA, reversing an almost two-decade trend of falling or flat growth rates. See **Figure 17** for seasonal peak demand growth over the current and prior assessment periods and **Figure 18** for net energy growth. A map of the primary demand drivers for the North American BPS is illustrated for each assessment area in **Figure 19** . **Electrification and Demand Growth** Electrification of household appliances (e.g., heat pumps for household heating) and projections for electric vehicle growth over this assessment period are components of the demand and energy estimates provided by each assessment area. Since the _2023 LTRA_ , peak season CAGR has risen in all assessment areas except two: NPCC-Maritimes’ winter CAGR fell from 0.98% to 0.63% and WECC-BC’s winter CAGR fell from 1.05% to 0.28%. Rising peak demand forecasts are contributing to the lower reserve margins projected for nearly all assessment areas. **Large Commercial and Industrial Loads** Increasing amounts of large commercial and industrial loads are connecting rapidly to the BPS. Emerging large loads, such as data centers (including crypto and AI) and hydrogen fuel plants, present unique challenges to forecasting and planning for increased demand. Earlier this year, NERC’s RSTC established a Large Loads Task Force to better understand the reliability implications of growth in large loads and develop solutions.\n\n**Figure 17: 10-Year Summer and Winter Peak Demand Growth and Rate Trends**\n\n**Figure 18: Net Energy for Load Growth and Rate Projections**","publish_date":null,"title":"Demand Trends and Implications","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2024-ltra_corrected_july_2025_p31.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2024-ltra_corrected_july_2025_p31.pdf"},"request":{"example_id":"chart/2024-ltra_corrected_july_2025_p31","product_type":"parse","source_file_path":"2024-ltra_corrected_july_2025_p31.pdf"},"started_at":"2026-11-04T19:44:51.859598"} +{"completed_at":"2026-11-04T19:44:54.887166","latency_in_ms":1909,"output":{"example_id":"chart/2024_Annual_Financial_Review_Upstream_FINAL_p10","grounded_pages":[],"layout_pages":[],"markdown":"**Natural gas front-month futures price (Henry Hub)** real 2024 dollars per million British thermal units\n\nData source: Bloomberg L.P.\n\n### **10**\n\n_Petroleum and Liquid Fuels Markets Team | Financial Review 2024 May 2025_\n\n# In 2024, U.S. natural gas prices were 12% lower in real terms than in 2023\n\n$0\n\n$1\n\n$2\n\n$3\n\n$4\n\n$5\n\n$6\n\n$7\n\n$8\n\n2015 2016 2017 2018 2019 2020 2021 2022 2023 2024\n\nLower natural gas prices reduced cash from operations since companies earned less from selling natural gas.\n\nLower natural gas prices also led the energy companies to reduce their natural gas production.\n\nNatural gas prices in 2024 were at their lowest in real terms since at least 2000.","pages":[{"line_numbers":[],"markdown":"**Natural gas front-month futures price (Henry Hub)** real 2024 dollars per million British thermal units\n\nData source: Bloomberg L.P.\n\n### **10**\n\n_Petroleum and Liquid Fuels Markets Team | Financial Review 2024 May 2025_\n\n# In 2024, U.S. natural gas prices were 12% lower in real terms than in 2023\n\n$0\n\n$1\n\n$2\n\n$3\n\n$4\n\n$5\n\n$6\n\n$7\n\n$8\n\n2015 2016 2017 2018 2019 2020 2021 2022 2023 2024\n\nLower natural gas prices reduced cash from operations since companies earned less from selling natural gas.\n\nLower natural gas prices also led the energy companies to reduce their natural gas production.\n\nNatural gas prices in 2024 were at their lowest in real terms since at least 2000.","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**Natural gas front-month futures price (Henry Hub)** real 2024 dollars per million British thermal units\n\nData source: Bloomberg L.P.\n\n### **10**\n\n_Petroleum and Liquid Fuels Markets Team | Financial Review 2024 May 2025_\n\n# In 2024, U.S. natural gas prices were 12% lower in real terms than in 2023\n\n$0\n\n$1\n\n$2\n\n$3\n\n$4\n\n$5\n\n$6\n\n$7\n\n$8\n\n2015 2016 2017 2018 2019 2020 2021 2022 2023 2024\n\nLower natural gas prices reduced cash from operations since companies earned less from selling natural gas.\n\nLower natural gas prices also led the energy companies to reduce their natural gas production.\n\nNatural gas prices in 2024 were at their lowest in real terms since at least 2000.","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2024_Annual_Financial_Review_Upstream_FINAL_p10.pdf"]},"response":{"errors":[],"extract_id":"extract_1ebf288a3b811f2dd8835a92ea477e38","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**Natural gas front-month futures price (Henry Hub)** real 2024 dollars per million British thermal units\n\nData source: Bloomberg L.P.\n\n### **10**\n\n_Petroleum and Liquid Fuels Markets Team | Financial Review 2024 May 2025_\n\n# In 2024, U.S. natural gas prices were 12% lower in real terms than in 2023\n\n$0\n\n$1\n\n$2\n\n$3\n\n$4\n\n$5\n\n$6\n\n$7\n\n$8\n\n2015 2016 2017 2018 2019 2020 2021 2022 2023 2024\n\nLower natural gas prices reduced cash from operations since companies earned less from selling natural gas.\n\nLower natural gas prices also led the energy companies to reduce their natural gas production.\n\nNatural gas prices in 2024 were at their lowest in real terms since at least 2000."],"full_content":"**Natural gas front-month futures price (Henry Hub)** real 2024 dollars per million British thermal units\n\nData source: Bloomberg L.P.\n\n### **10**\n\n_Petroleum and Liquid Fuels Markets Team | Financial Review 2024 May 2025_\n\n# In 2024, U.S. natural gas prices were 12% lower in real terms than in 2023\n\n$0\n\n$1\n\n$2\n\n$3\n\n$4\n\n$5\n\n$6\n\n$7\n\n$8\n\n2015 2016 2017 2018 2019 2020 2021 2022 2023 2024\n\nLower natural gas prices reduced cash from operations since companies earned less from selling natural gas.\n\nLower natural gas prices also led the energy companies to reduce their natural gas production.\n\nNatural gas prices in 2024 were at their lowest in real terms since at least 2000.","publish_date":null,"title":"In 2024, U.S. natural gas prices were 12% lower in real terms than in 2023","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2024_Annual_Financial_Review_Upstream_FINAL_p10.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2024_Annual_Financial_Review_Upstream_FINAL_p10.pdf"},"request":{"example_id":"chart/2024_Annual_Financial_Review_Upstream_FINAL_p10","product_type":"parse","source_file_path":"2024_Annual_Financial_Review_Upstream_FINAL_p10.pdf"},"started_at":"2026-11-04T19:44:52.977690"} +{"completed_at":"2026-11-04T19:44:55.912207","latency_in_ms":1744,"output":{"example_id":"chart/2024_Annual_Financial_Review_Upstream_FINAL_p13","grounded_pages":[],"layout_pages":[],"markdown":"**Cash from operations for select energy companies** billion real 2024 dollars\n\n_Petroleum and Liquid Fuels Markets Team | Financial Review 2024 May 2025_\n\n# Cash from operations decreased by 9% from 2023 to 2024\n\nData source: Evaluate Energy Note: _Select energy companies_ includes 158 global oil and natural gas companies. The Appendix contains the full list of companies.\n\n### $0\n\n### $100\n\n### $200\n\n### $300\n\n### $400\n\n### $500\n\n### $600\n\n### $700\n\n### $800\n\n### $900\n\n### $1,000\n\n### 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024\n\n### **13**\n\nLower crude oil and natural gas prices contributed to lower cash from operations in 2024 compared with 2023.","pages":[{"line_numbers":[],"markdown":"**Cash from operations for select energy companies** billion real 2024 dollars\n\n_Petroleum and Liquid Fuels Markets Team | Financial Review 2024 May 2025_\n\n# Cash from operations decreased by 9% from 2023 to 2024\n\nData source: Evaluate Energy Note: _Select energy companies_ includes 158 global oil and natural gas companies. The Appendix contains the full list of companies.\n\n### $0\n\n### $100\n\n### $200\n\n### $300\n\n### $400\n\n### $500\n\n### $600\n\n### $700\n\n### $800\n\n### $900\n\n### $1,000\n\n### 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024\n\n### **13**\n\nLower crude oil and natural gas prices contributed to lower cash from operations in 2024 compared with 2023.","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**Cash from operations for select energy companies** billion real 2024 dollars\n\n_Petroleum and Liquid Fuels Markets Team | Financial Review 2024 May 2025_\n\n# Cash from operations decreased by 9% from 2023 to 2024\n\nData source: Evaluate Energy Note: _Select energy companies_ includes 158 global oil and natural gas companies. The Appendix contains the full list of companies.\n\n### $0\n\n### $100\n\n### $200\n\n### $300\n\n### $400\n\n### $500\n\n### $600\n\n### $700\n\n### $800\n\n### $900\n\n### $1,000\n\n### 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024\n\n### **13**\n\nLower crude oil and natural gas prices contributed to lower cash from operations in 2024 compared with 2023.","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2024_Annual_Financial_Review_Upstream_FINAL_p13.pdf"]},"response":{"errors":[],"extract_id":"extract_d57fdadf1ee93fdb49a777318d1a9c01","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**Cash from operations for select energy companies** billion real 2024 dollars\n\n_Petroleum and Liquid Fuels Markets Team | Financial Review 2024 May 2025_\n\n# Cash from operations decreased by 9% from 2023 to 2024\n\nData source: Evaluate Energy Note: _Select energy companies_ includes 158 global oil and natural gas companies. The Appendix contains the full list of companies.\n\n### $0\n\n### $100\n\n### $200\n\n### $300\n\n### $400\n\n### $500\n\n### $600\n\n### $700\n\n### $800\n\n### $900\n\n### $1,000\n\n### 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024\n\n### **13**\n\nLower crude oil and natural gas prices contributed to lower cash from operations in 2024 compared with 2023."],"full_content":"**Cash from operations for select energy companies** billion real 2024 dollars\n\n_Petroleum and Liquid Fuels Markets Team | Financial Review 2024 May 2025_\n\n# Cash from operations decreased by 9% from 2023 to 2024\n\nData source: Evaluate Energy Note: _Select energy companies_ includes 158 global oil and natural gas companies. The Appendix contains the full list of companies.\n\n### $0\n\n### $100\n\n### $200\n\n### $300\n\n### $400\n\n### $500\n\n### $600\n\n### $700\n\n### $800\n\n### $900\n\n### $1,000\n\n### 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024\n\n### **13**\n\nLower crude oil and natural gas prices contributed to lower cash from operations in 2024 compared with 2023.","publish_date":null,"title":"Cash from operations decreased by 9% from 2023 to 2024","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2024_Annual_Financial_Review_Upstream_FINAL_p13.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2024_Annual_Financial_Review_Upstream_FINAL_p13.pdf"},"request":{"example_id":"chart/2024_Annual_Financial_Review_Upstream_FINAL_p13","product_type":"parse","source_file_path":"2024_Annual_Financial_Review_Upstream_FINAL_p13.pdf"},"started_at":"2026-11-04T19:44:54.167908"} +{"completed_at":"2026-11-04T19:44:57.837363","latency_in_ms":2285,"output":{"example_id":"chart/2024_Annual_Financial_Review_Upstream_FINAL_p14","grounded_pages":[],"layout_pages":[],"markdown":"_Petroleum and Liquid Fuels Markets Team | Financial Review 2024 May 2025_\n\n# After net purchases and production, the energy companies’ global proved reserves were essentially unchanged in 2024\n\nData source: Evaluate Energy Note: Mergers and acquisitions between companies may affect net reserve changes for the year and comparisons between years. _Total proved reserves, start of year, extensions and discoveries, revisions,_ _sales,_ and _production_ include EIA estimates for acquired company reserves based on previous trends and company announcements. _Select energy companies_ includes 158 global oil and natural gas companies. The Appendix contains the full list of companies.\n\n**Change in proved reserves from beginning to end of year for select energy companies** billion barrels of oil equivalent\n\n220\n\n230\n\n240\n\n250\n\n260\n\n270\n\ntotal proved reserves,\n\nstart of year\n\nimproved recovery,\n\nextensions,\n\ndiscoveries, and\n\nrevisions\n\npurchases sales production total proved reserves,\n\nend of year\n\n**net -0.1 billion barrels of oil equivalent**\n\n**Asia Pacific, Russia, and Central Asia**\n\n**Europe Middle East and Africa**\n\n**Latin America Canada United States**\n\n**other world and associates**\n\n### **14**\n\n### // 0","pages":[{"line_numbers":[],"markdown":"_Petroleum and Liquid Fuels Markets Team | Financial Review 2024 May 2025_\n\n# After net purchases and production, the energy companies’ global proved reserves were essentially unchanged in 2024\n\nData source: Evaluate Energy Note: Mergers and acquisitions between companies may affect net reserve changes for the year and comparisons between years. _Total proved reserves, start of year, extensions and discoveries, revisions,_ _sales,_ and _production_ include EIA estimates for acquired company reserves based on previous trends and company announcements. _Select energy companies_ includes 158 global oil and natural gas companies. The Appendix contains the full list of companies.\n\n**Change in proved reserves from beginning to end of year for select energy companies** billion barrels of oil equivalent\n\n220\n\n230\n\n240\n\n250\n\n260\n\n270\n\ntotal proved reserves,\n\nstart of year\n\nimproved recovery,\n\nextensions,\n\ndiscoveries, and\n\nrevisions\n\npurchases sales production total proved reserves,\n\nend of year\n\n**net -0.1 billion barrels of oil equivalent**\n\n**Asia Pacific, Russia, and Central Asia**\n\n**Europe Middle East and Africa**\n\n**Latin America Canada United States**\n\n**other world and associates**\n\n### **14**\n\n### // 0","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"_Petroleum and Liquid Fuels Markets Team | Financial Review 2024 May 2025_\n\n# After net purchases and production, the energy companies’ global proved reserves were essentially unchanged in 2024\n\nData source: Evaluate Energy Note: Mergers and acquisitions between companies may affect net reserve changes for the year and comparisons between years. _Total proved reserves, start of year, extensions and discoveries, revisions,_ _sales,_ and _production_ include EIA estimates for acquired company reserves based on previous trends and company announcements. _Select energy companies_ includes 158 global oil and natural gas companies. The Appendix contains the full list of companies.\n\n**Change in proved reserves from beginning to end of year for select energy companies** billion barrels of oil equivalent\n\n220\n\n230\n\n240\n\n250\n\n260\n\n270\n\ntotal proved reserves,\n\nstart of year\n\nimproved recovery,\n\nextensions,\n\ndiscoveries, and\n\nrevisions\n\npurchases sales production total proved reserves,\n\nend of year\n\n**net -0.1 billion barrels of oil equivalent**\n\n**Asia Pacific, Russia, and Central Asia**\n\n**Europe Middle East and Africa**\n\n**Latin America Canada United States**\n\n**other world and associates**\n\n### **14**\n\n### // 0","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2024_Annual_Financial_Review_Upstream_FINAL_p14.pdf"]},"response":{"errors":[],"extract_id":"extract_751beac4c170904b45406b4fc3fd6dde","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["_Petroleum and Liquid Fuels Markets Team | Financial Review 2024 May 2025_\n\n# After net purchases and production, the energy companies’ global proved reserves were essentially unchanged in 2024\n\nData source: Evaluate Energy Note: Mergers and acquisitions between companies may affect net reserve changes for the year and comparisons between years. _Total proved reserves, start of year, extensions and discoveries, revisions,_ _sales,_ and _production_ include EIA estimates for acquired company reserves based on previous trends and company announcements. _Select energy companies_ includes 158 global oil and natural gas companies. The Appendix contains the full list of companies.\n\n**Change in proved reserves from beginning to end of year for select energy companies** billion barrels of oil equivalent\n\n220\n\n230\n\n240\n\n250\n\n260\n\n270\n\ntotal proved reserves,\n\nstart of year\n\nimproved recovery,\n\nextensions,\n\ndiscoveries, and\n\nrevisions\n\npurchases sales production total proved reserves,\n\nend of year\n\n**net -0.1 billion barrels of oil equivalent**\n\n**Asia Pacific, Russia, and Central Asia**\n\n**Europe Middle East and Africa**\n\n**Latin America Canada United States**\n\n**other world and associates**\n\n### **14**\n\n### // 0"],"full_content":"_Petroleum and Liquid Fuels Markets Team | Financial Review 2024 May 2025_\n\n# After net purchases and production, the energy companies’ global proved reserves were essentially unchanged in 2024\n\nData source: Evaluate Energy Note: Mergers and acquisitions between companies may affect net reserve changes for the year and comparisons between years. _Total proved reserves, start of year, extensions and discoveries, revisions,_ _sales,_ and _production_ include EIA estimates for acquired company reserves based on previous trends and company announcements. _Select energy companies_ includes 158 global oil and natural gas companies. The Appendix contains the full list of companies.\n\n**Change in proved reserves from beginning to end of year for select energy companies** billion barrels of oil equivalent\n\n220\n\n230\n\n240\n\n250\n\n260\n\n270\n\ntotal proved reserves,\n\nstart of year\n\nimproved recovery,\n\nextensions,\n\ndiscoveries, and\n\nrevisions\n\npurchases sales production total proved reserves,\n\nend of year\n\n**net -0.1 billion barrels of oil equivalent**\n\n**Asia Pacific, Russia, and Central Asia**\n\n**Europe Middle East and Africa**\n\n**Latin America Canada United States**\n\n**other world and associates**\n\n### **14**\n\n### // 0","publish_date":null,"title":"After net purchases and production, the energy companies’ global proved reserves were essentially unchanged in 2024","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2024_Annual_Financial_Review_Upstream_FINAL_p14.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2024_Annual_Financial_Review_Upstream_FINAL_p14.pdf"},"request":{"example_id":"chart/2024_Annual_Financial_Review_Upstream_FINAL_p14","product_type":"parse","source_file_path":"2024_Annual_Financial_Review_Upstream_FINAL_p14.pdf"},"started_at":"2026-11-04T19:44:55.551412"} +{"completed_at":"2026-11-04T19:44:57.407232","latency_in_ms":1011,"output":{"example_id":"chart/2024_Annual_Financial_Review_Upstream_FINAL_p16","grounded_pages":[],"layout_pages":[],"markdown":"**Change in proved reserves by region for select energy companies** billion barrels of oil equivalent\n\n_Petroleum and Liquid Fuels Markets Team | Financial Review 2024 May 2025_\n\n# Changes in proved reserves were relatively small across all regions in 2024\n\nData source: Evaluate Energy Note: Mergers and acquisitions between companies may affect net reserve changes and comparisons between years. Each year shows change based on end-of-year totals as reported. _Other world and associates_ includes the regions outside the ones identified in this figure and reserves held by associate companies regardless of location. _Select energy companies_ includes 158 global oil and natural gas companies. The Appendix contains the full list of companies.\n\n### \\-25 -20 -15 -10\n\n### \\-5 0 5\n\n### 10 15 20 25\n\n### 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024\n\n**other world and associates Latin America United States net change Canada Europe Middle East and Africa Asia Pacific, Russia, and Central Asia**\n\n### **16**","pages":[{"line_numbers":[],"markdown":"**Change in proved reserves by region for select energy companies** billion barrels of oil equivalent\n\n_Petroleum and Liquid Fuels Markets Team | Financial Review 2024 May 2025_\n\n# Changes in proved reserves were relatively small across all regions in 2024\n\nData source: Evaluate Energy Note: Mergers and acquisitions between companies may affect net reserve changes and comparisons between years. Each year shows change based on end-of-year totals as reported. _Other world and associates_ includes the regions outside the ones identified in this figure and reserves held by associate companies regardless of location. _Select energy companies_ includes 158 global oil and natural gas companies. The Appendix contains the full list of companies.\n\n### \\-25 -20 -15 -10\n\n### \\-5 0 5\n\n### 10 15 20 25\n\n### 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024\n\n**other world and associates Latin America United States net change Canada Europe Middle East and Africa Asia Pacific, Russia, and Central Asia**\n\n### **16**","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**Change in proved reserves by region for select energy companies** billion barrels of oil equivalent\n\n_Petroleum and Liquid Fuels Markets Team | Financial Review 2024 May 2025_\n\n# Changes in proved reserves were relatively small across all regions in 2024\n\nData source: Evaluate Energy Note: Mergers and acquisitions between companies may affect net reserve changes and comparisons between years. Each year shows change based on end-of-year totals as reported. _Other world and associates_ includes the regions outside the ones identified in this figure and reserves held by associate companies regardless of location. _Select energy companies_ includes 158 global oil and natural gas companies. The Appendix contains the full list of companies.\n\n### \\-25 -20 -15 -10\n\n### \\-5 0 5\n\n### 10 15 20 25\n\n### 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024\n\n**other world and associates Latin America United States net change Canada Europe Middle East and Africa Asia Pacific, Russia, and Central Asia**\n\n### **16**","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2024_Annual_Financial_Review_Upstream_FINAL_p16.pdf"]},"response":{"errors":[],"extract_id":"extract_ab691d26c1f05421dbd32e95830458b2","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**Change in proved reserves by region for select energy companies** billion barrels of oil equivalent\n\n_Petroleum and Liquid Fuels Markets Team | Financial Review 2024 May 2025_\n\n# Changes in proved reserves were relatively small across all regions in 2024\n\nData source: Evaluate Energy Note: Mergers and acquisitions between companies may affect net reserve changes and comparisons between years. Each year shows change based on end-of-year totals as reported. _Other world and associates_ includes the regions outside the ones identified in this figure and reserves held by associate companies regardless of location. _Select energy companies_ includes 158 global oil and natural gas companies. The Appendix contains the full list of companies.\n\n### \\-25 -20 -15 -10\n\n### \\-5 0 5\n\n### 10 15 20 25\n\n### 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024\n\n**other world and associates Latin America United States net change Canada Europe Middle East and Africa Asia Pacific, Russia, and Central Asia**\n\n### **16**"],"full_content":"**Change in proved reserves by region for select energy companies** billion barrels of oil equivalent\n\n_Petroleum and Liquid Fuels Markets Team | Financial Review 2024 May 2025_\n\n# Changes in proved reserves were relatively small across all regions in 2024\n\nData source: Evaluate Energy Note: Mergers and acquisitions between companies may affect net reserve changes and comparisons between years. Each year shows change based on end-of-year totals as reported. _Other world and associates_ includes the regions outside the ones identified in this figure and reserves held by associate companies regardless of location. _Select energy companies_ includes 158 global oil and natural gas companies. The Appendix contains the full list of companies.\n\n### \\-25 -20 -15 -10\n\n### \\-5 0 5\n\n### 10 15 20 25\n\n### 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024\n\n**other world and associates Latin America United States net change Canada Europe Middle East and Africa Asia Pacific, Russia, and Central Asia**\n\n### **16**","publish_date":null,"title":"Changes in proved reserves were relatively small across all regions in 2024","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2024_Annual_Financial_Review_Upstream_FINAL_p16.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2024_Annual_Financial_Review_Upstream_FINAL_p16.pdf"},"request":{"example_id":"chart/2024_Annual_Financial_Review_Upstream_FINAL_p16","product_type":"parse","source_file_path":"2024_Annual_Financial_Review_Upstream_FINAL_p16.pdf"},"started_at":"2026-11-04T19:44:56.396052"} +{"completed_at":"2026-11-04T19:44:59.498940","latency_in_ms":1850,"output":{"example_id":"chart/2024_Annual_Financial_Review_Upstream_FINAL_p17","grounded_pages":[],"layout_pages":[],"markdown":"_Petroleum and Liquid Fuels Markets Team | Financial Review 2024 May 2025_\n\n# Organic proved reserve additions and revisions for petroleum liquids and natural gas totaled 18 billion barrels in 2024\n\nData source: Evaluate Energy Note: _Select energy companies_ includes 158 global oil and natural gas companies. The Appendix contains the full list of companies.\n\n**Proved reserve additions for select energy companies** billion barrels of oil equivalent\n\n### \\-20\n\n### \\-15\n\n### \\-10\n\n### \\-5\n\n### 0\n\n### 5\n\n### 10\n\n### 15\n\n### 20\n\n### 25\n\n### 30\n\n### 35\n\n### 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024\n\n**total crude oil and natural gas reserve additions**\n\n**extensions and discoveries**\n\n**improved recovery**\n\n**revisions**\n\n### **17**\n\n**Organic proved reserve additions** create new proved reserves from improved recovery, extensions, and discoveries.\n\nCompanies can also increase their proved reserves by purchasing them from other companies, but these additions do not create new proved reserves.","pages":[{"line_numbers":[],"markdown":"_Petroleum and Liquid Fuels Markets Team | Financial Review 2024 May 2025_\n\n# Organic proved reserve additions and revisions for petroleum liquids and natural gas totaled 18 billion barrels in 2024\n\nData source: Evaluate Energy Note: _Select energy companies_ includes 158 global oil and natural gas companies. The Appendix contains the full list of companies.\n\n**Proved reserve additions for select energy companies** billion barrels of oil equivalent\n\n### \\-20\n\n### \\-15\n\n### \\-10\n\n### \\-5\n\n### 0\n\n### 5\n\n### 10\n\n### 15\n\n### 20\n\n### 25\n\n### 30\n\n### 35\n\n### 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024\n\n**total crude oil and natural gas reserve additions**\n\n**extensions and discoveries**\n\n**improved recovery**\n\n**revisions**\n\n### **17**\n\n**Organic proved reserve additions** create new proved reserves from improved recovery, extensions, and discoveries.\n\nCompanies can also increase their proved reserves by purchasing them from other companies, but these additions do not create new proved reserves.","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"_Petroleum and Liquid Fuels Markets Team | Financial Review 2024 May 2025_\n\n# Organic proved reserve additions and revisions for petroleum liquids and natural gas totaled 18 billion barrels in 2024\n\nData source: Evaluate Energy Note: _Select energy companies_ includes 158 global oil and natural gas companies. The Appendix contains the full list of companies.\n\n**Proved reserve additions for select energy companies** billion barrels of oil equivalent\n\n### \\-20\n\n### \\-15\n\n### \\-10\n\n### \\-5\n\n### 0\n\n### 5\n\n### 10\n\n### 15\n\n### 20\n\n### 25\n\n### 30\n\n### 35\n\n### 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024\n\n**total crude oil and natural gas reserve additions**\n\n**extensions and discoveries**\n\n**improved recovery**\n\n**revisions**\n\n### **17**\n\n**Organic proved reserve additions** create new proved reserves from improved recovery, extensions, and discoveries.\n\nCompanies can also increase their proved reserves by purchasing them from other companies, but these additions do not create new proved reserves.","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2024_Annual_Financial_Review_Upstream_FINAL_p17.pdf"]},"response":{"errors":[],"extract_id":"extract_0fc5aeb7a5c8bb7a08b173e56c527e75","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["_Petroleum and Liquid Fuels Markets Team | Financial Review 2024 May 2025_\n\n# Organic proved reserve additions and revisions for petroleum liquids and natural gas totaled 18 billion barrels in 2024\n\nData source: Evaluate Energy Note: _Select energy companies_ includes 158 global oil and natural gas companies. The Appendix contains the full list of companies.\n\n**Proved reserve additions for select energy companies** billion barrels of oil equivalent\n\n### \\-20\n\n### \\-15\n\n### \\-10\n\n### \\-5\n\n### 0\n\n### 5\n\n### 10\n\n### 15\n\n### 20\n\n### 25\n\n### 30\n\n### 35\n\n### 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024\n\n**total crude oil and natural gas reserve additions**\n\n**extensions and discoveries**\n\n**improved recovery**\n\n**revisions**\n\n### **17**\n\n**Organic proved reserve additions** create new proved reserves from improved recovery, extensions, and discoveries.\n\nCompanies can also increase their proved reserves by purchasing them from other companies, but these additions do not create new proved reserves."],"full_content":"_Petroleum and Liquid Fuels Markets Team | Financial Review 2024 May 2025_\n\n# Organic proved reserve additions and revisions for petroleum liquids and natural gas totaled 18 billion barrels in 2024\n\nData source: Evaluate Energy Note: _Select energy companies_ includes 158 global oil and natural gas companies. The Appendix contains the full list of companies.\n\n**Proved reserve additions for select energy companies** billion barrels of oil equivalent\n\n### \\-20\n\n### \\-15\n\n### \\-10\n\n### \\-5\n\n### 0\n\n### 5\n\n### 10\n\n### 15\n\n### 20\n\n### 25\n\n### 30\n\n### 35\n\n### 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024\n\n**total crude oil and natural gas reserve additions**\n\n**extensions and discoveries**\n\n**improved recovery**\n\n**revisions**\n\n### **17**\n\n**Organic proved reserve additions** create new proved reserves from improved recovery, extensions, and discoveries.\n\nCompanies can also increase their proved reserves by purchasing them from other companies, but these additions do not create new proved reserves.","publish_date":null,"title":"Organic proved reserve additions and revisions for petroleum liquids and natural gas totaled 18 billion barrels in 2024","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2024_Annual_Financial_Review_Upstream_FINAL_p17.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2024_Annual_Financial_Review_Upstream_FINAL_p17.pdf"},"request":{"example_id":"chart/2024_Annual_Financial_Review_Upstream_FINAL_p17","product_type":"parse","source_file_path":"2024_Annual_Financial_Review_Upstream_FINAL_p17.pdf"},"started_at":"2026-11-04T19:44:57.648577"} +{"completed_at":"2026-11-04T19:45:01.177055","latency_in_ms":2399,"output":{"example_id":"chart/2024_Annual_Financial_Review_Upstream_FINAL_p19","grounded_pages":[],"layout_pages":[],"markdown":"**Upstream costs incurred for select energy companies** billion 2024 dollars\n\n_Petroleum and Liquid Fuels Markets Team | Financial Review 2024 May 2025_\n\n# Ten companies accounted for nearly 90% of proved reserve and unproved reserve acquisition spending by the companies\n\nData source: Evaluate Energy\n\n### $0 $20 $40 $60 $80 $100\n\n### Exxon Mobil Corporation\n\n### Diamondback Energy Inc.\n\n### ConocoPhillips\n\n### Expand Energy Corporation\n\n### Occidental Petroleum Corporation\n\n### Canadian Natural Resources Limited\n\n### Chord Energy Corporation\n\n### Devon Energy Corporation\n\n### APA Corporation\n\n### California Resources Corporation\n\n**proved reserve acquisition**\n\n**unproved reserve acquisition**\n\n### **19**","pages":[{"line_numbers":[],"markdown":"**Upstream costs incurred for select energy companies** billion 2024 dollars\n\n_Petroleum and Liquid Fuels Markets Team | Financial Review 2024 May 2025_\n\n# Ten companies accounted for nearly 90% of proved reserve and unproved reserve acquisition spending by the companies\n\nData source: Evaluate Energy\n\n### $0 $20 $40 $60 $80 $100\n\n### Exxon Mobil Corporation\n\n### Diamondback Energy Inc.\n\n### ConocoPhillips\n\n### Expand Energy Corporation\n\n### Occidental Petroleum Corporation\n\n### Canadian Natural Resources Limited\n\n### Chord Energy Corporation\n\n### Devon Energy Corporation\n\n### APA Corporation\n\n### California Resources Corporation\n\n**proved reserve acquisition**\n\n**unproved reserve acquisition**\n\n### **19**","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**Upstream costs incurred for select energy companies** billion 2024 dollars\n\n_Petroleum and Liquid Fuels Markets Team | Financial Review 2024 May 2025_\n\n# Ten companies accounted for nearly 90% of proved reserve and unproved reserve acquisition spending by the companies\n\nData source: Evaluate Energy\n\n### $0 $20 $40 $60 $80 $100\n\n### Exxon Mobil Corporation\n\n### Diamondback Energy Inc.\n\n### ConocoPhillips\n\n### Expand Energy Corporation\n\n### Occidental Petroleum Corporation\n\n### Canadian Natural Resources Limited\n\n### Chord Energy Corporation\n\n### Devon Energy Corporation\n\n### APA Corporation\n\n### California Resources Corporation\n\n**proved reserve acquisition**\n\n**unproved reserve acquisition**\n\n### **19**","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2024_Annual_Financial_Review_Upstream_FINAL_p19.pdf"]},"response":{"errors":[],"extract_id":"extract_d44b2d4478ee9b489c005509f5ad0ac0","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**Upstream costs incurred for select energy companies** billion 2024 dollars\n\n_Petroleum and Liquid Fuels Markets Team | Financial Review 2024 May 2025_\n\n# Ten companies accounted for nearly 90% of proved reserve and unproved reserve acquisition spending by the companies\n\nData source: Evaluate Energy\n\n### $0 $20 $40 $60 $80 $100\n\n### Exxon Mobil Corporation\n\n### Diamondback Energy Inc.\n\n### ConocoPhillips\n\n### Expand Energy Corporation\n\n### Occidental Petroleum Corporation\n\n### Canadian Natural Resources Limited\n\n### Chord Energy Corporation\n\n### Devon Energy Corporation\n\n### APA Corporation\n\n### California Resources Corporation\n\n**proved reserve acquisition**\n\n**unproved reserve acquisition**\n\n### **19**"],"full_content":"**Upstream costs incurred for select energy companies** billion 2024 dollars\n\n_Petroleum and Liquid Fuels Markets Team | Financial Review 2024 May 2025_\n\n# Ten companies accounted for nearly 90% of proved reserve and unproved reserve acquisition spending by the companies\n\nData source: Evaluate Energy\n\n### $0 $20 $40 $60 $80 $100\n\n### Exxon Mobil Corporation\n\n### Diamondback Energy Inc.\n\n### ConocoPhillips\n\n### Expand Energy Corporation\n\n### Occidental Petroleum Corporation\n\n### Canadian Natural Resources Limited\n\n### Chord Energy Corporation\n\n### Devon Energy Corporation\n\n### APA Corporation\n\n### California Resources Corporation\n\n**proved reserve acquisition**\n\n**unproved reserve acquisition**\n\n### **19**","publish_date":null,"title":"Ten companies accounted for nearly 90% of proved reserve and unproved reserve acquisition spending by the companies","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2024_Annual_Financial_Review_Upstream_FINAL_p19.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2024_Annual_Financial_Review_Upstream_FINAL_p19.pdf"},"request":{"example_id":"chart/2024_Annual_Financial_Review_Upstream_FINAL_p19","product_type":"parse","source_file_path":"2024_Annual_Financial_Review_Upstream_FINAL_p19.pdf"},"started_at":"2026-11-04T19:44:58.777525"} +{"completed_at":"2026-11-04T19:45:01.960413","latency_in_ms":1645,"output":{"example_id":"chart/2024_Annual_Financial_Review_Upstream_FINAL_p28","grounded_pages":[],"layout_pages":[],"markdown":"**Long-term debt-to-equity ratio for select energy companies and U.S. manufacturing companies** percentage\n\n_Petroleum and Liquid Fuels Markets Team | Financial Review 2024 May 2025_\n\n# The long-term debt-to-equity ratio was relatively flat for the energy companies and U.S. manufacturing companies, at 42% and 56%\n\nData source: Evaluate Energy and U.S. Census Bureau Note: _Select energy companies_ includes 158 global oil and natural gas companies. The Appendix contains the full list of companies.\n\n### 0%\n\n### 10%\n\n### 20%\n\n### 30%\n\n### 40%\n\n### 50%\n\n### 60%\n\n### 70%\n\n### 80%\n\n### 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024\n\n### **energy companies**\n\n### **all U.S. manufacturing**\n\n### **28**","pages":[{"line_numbers":[],"markdown":"**Long-term debt-to-equity ratio for select energy companies and U.S. manufacturing companies** percentage\n\n_Petroleum and Liquid Fuels Markets Team | Financial Review 2024 May 2025_\n\n# The long-term debt-to-equity ratio was relatively flat for the energy companies and U.S. manufacturing companies, at 42% and 56%\n\nData source: Evaluate Energy and U.S. Census Bureau Note: _Select energy companies_ includes 158 global oil and natural gas companies. The Appendix contains the full list of companies.\n\n### 0%\n\n### 10%\n\n### 20%\n\n### 30%\n\n### 40%\n\n### 50%\n\n### 60%\n\n### 70%\n\n### 80%\n\n### 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024\n\n### **energy companies**\n\n### **all U.S. manufacturing**\n\n### **28**","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**Long-term debt-to-equity ratio for select energy companies and U.S. manufacturing companies** percentage\n\n_Petroleum and Liquid Fuels Markets Team | Financial Review 2024 May 2025_\n\n# The long-term debt-to-equity ratio was relatively flat for the energy companies and U.S. manufacturing companies, at 42% and 56%\n\nData source: Evaluate Energy and U.S. Census Bureau Note: _Select energy companies_ includes 158 global oil and natural gas companies. The Appendix contains the full list of companies.\n\n### 0%\n\n### 10%\n\n### 20%\n\n### 30%\n\n### 40%\n\n### 50%\n\n### 60%\n\n### 70%\n\n### 80%\n\n### 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024\n\n### **energy companies**\n\n### **all U.S. manufacturing**\n\n### **28**","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2024_Annual_Financial_Review_Upstream_FINAL_p28.pdf"]},"response":{"errors":[],"extract_id":"extract_8d13b7a695237d7e0e83adb7f514427d","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**Long-term debt-to-equity ratio for select energy companies and U.S. manufacturing companies** percentage\n\n_Petroleum and Liquid Fuels Markets Team | Financial Review 2024 May 2025_\n\n# The long-term debt-to-equity ratio was relatively flat for the energy companies and U.S. manufacturing companies, at 42% and 56%\n\nData source: Evaluate Energy and U.S. Census Bureau Note: _Select energy companies_ includes 158 global oil and natural gas companies. The Appendix contains the full list of companies.\n\n### 0%\n\n### 10%\n\n### 20%\n\n### 30%\n\n### 40%\n\n### 50%\n\n### 60%\n\n### 70%\n\n### 80%\n\n### 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024\n\n### **energy companies**\n\n### **all U.S. manufacturing**\n\n### **28**"],"full_content":"**Long-term debt-to-equity ratio for select energy companies and U.S. manufacturing companies** percentage\n\n_Petroleum and Liquid Fuels Markets Team | Financial Review 2024 May 2025_\n\n# The long-term debt-to-equity ratio was relatively flat for the energy companies and U.S. manufacturing companies, at 42% and 56%\n\nData source: Evaluate Energy and U.S. Census Bureau Note: _Select energy companies_ includes 158 global oil and natural gas companies. The Appendix contains the full list of companies.\n\n### 0%\n\n### 10%\n\n### 20%\n\n### 30%\n\n### 40%\n\n### 50%\n\n### 60%\n\n### 70%\n\n### 80%\n\n### 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024\n\n### **energy companies**\n\n### **all U.S. manufacturing**\n\n### **28**","publish_date":null,"title":"The long-term debt-to-equity ratio was relatively flat for the energy companies and U.S. manufacturing companies, at 42% and 56%","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2024_Annual_Financial_Review_Upstream_FINAL_p28.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2024_Annual_Financial_Review_Upstream_FINAL_p28.pdf"},"request":{"example_id":"chart/2024_Annual_Financial_Review_Upstream_FINAL_p28","product_type":"parse","source_file_path":"2024_Annual_Financial_Review_Upstream_FINAL_p28.pdf"},"started_at":"2026-11-04T19:45:00.314719"} +{"completed_at":"2026-11-04T19:45:04.490624","latency_in_ms":2674,"output":{"example_id":"chart/2024_Annual_Financial_Review_Upstream_FINAL_p9","grounded_pages":[],"layout_pages":[],"markdown":"**Brent crude oil front-month futures prices** dollars per barrel (b)\n\n_Petroleum and Liquid Fuels Markets Team | Financial Review 2024 May 2025_\n\n# Proved reserves declined in value in 2024 because of lower crude oil prices\n\nData source: Intercontinental Exchange, Bloomberg L.P. Note: SEC=U.S. Securities and Exchange Commission\n\n**$75.89**\n\n**$78.70**\n\n**$83.55**\n\n**$87.42**\n\n**$83.44 $78.36**\n\n**$86.60**\n\n**$79.52**\n\n**$77.52**\n\n**$73.56**\n\n**$73.10**\n\n**$71.83**\n\n### $0\n\n### $10\n\n### $20\n\n### $30\n\n### $40\n\n### $50\n\n### $60\n\n### $70\n\n### $80\n\n### $90\n\n### $100\n\n### Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec\n\n### **Brent 2024 first-day-of-month average = $79.12/b Brent 2023 first-day-of-month average = $82.83/b**\n\n### **first-day-of-month price**\n\n### **9**\n\nThe U.S. Securities and Exchange Commission (SEC) requires companies to value proved reserves based on an average of the first-day-of-the-month price for the preceding 12-month period.","pages":[{"line_numbers":[],"markdown":"**Brent crude oil front-month futures prices** dollars per barrel (b)\n\n_Petroleum and Liquid Fuels Markets Team | Financial Review 2024 May 2025_\n\n# Proved reserves declined in value in 2024 because of lower crude oil prices\n\nData source: Intercontinental Exchange, Bloomberg L.P. Note: SEC=U.S. Securities and Exchange Commission\n\n**$75.89**\n\n**$78.70**\n\n**$83.55**\n\n**$87.42**\n\n**$83.44 $78.36**\n\n**$86.60**\n\n**$79.52**\n\n**$77.52**\n\n**$73.56**\n\n**$73.10**\n\n**$71.83**\n\n### $0\n\n### $10\n\n### $20\n\n### $30\n\n### $40\n\n### $50\n\n### $60\n\n### $70\n\n### $80\n\n### $90\n\n### $100\n\n### Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec\n\n### **Brent 2024 first-day-of-month average = $79.12/b Brent 2023 first-day-of-month average = $82.83/b**\n\n### **first-day-of-month price**\n\n### **9**\n\nThe U.S. Securities and Exchange Commission (SEC) requires companies to value proved reserves based on an average of the first-day-of-the-month price for the preceding 12-month period.","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**Brent crude oil front-month futures prices** dollars per barrel (b)\n\n_Petroleum and Liquid Fuels Markets Team | Financial Review 2024 May 2025_\n\n# Proved reserves declined in value in 2024 because of lower crude oil prices\n\nData source: Intercontinental Exchange, Bloomberg L.P. Note: SEC=U.S. Securities and Exchange Commission\n\n**$75.89**\n\n**$78.70**\n\n**$83.55**\n\n**$87.42**\n\n**$83.44 $78.36**\n\n**$86.60**\n\n**$79.52**\n\n**$77.52**\n\n**$73.56**\n\n**$73.10**\n\n**$71.83**\n\n### $0\n\n### $10\n\n### $20\n\n### $30\n\n### $40\n\n### $50\n\n### $60\n\n### $70\n\n### $80\n\n### $90\n\n### $100\n\n### Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec\n\n### **Brent 2024 first-day-of-month average = $79.12/b Brent 2023 first-day-of-month average = $82.83/b**\n\n### **first-day-of-month price**\n\n### **9**\n\nThe U.S. Securities and Exchange Commission (SEC) requires companies to value proved reserves based on an average of the first-day-of-the-month price for the preceding 12-month period.","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2024_Annual_Financial_Review_Upstream_FINAL_p9.pdf"]},"response":{"errors":[],"extract_id":"extract_f2a76312bf4bc0c120236e9b897b7766","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**Brent crude oil front-month futures prices** dollars per barrel (b)\n\n_Petroleum and Liquid Fuels Markets Team | Financial Review 2024 May 2025_\n\n# Proved reserves declined in value in 2024 because of lower crude oil prices\n\nData source: Intercontinental Exchange, Bloomberg L.P. Note: SEC=U.S. Securities and Exchange Commission\n\n**$75.89**\n\n**$78.70**\n\n**$83.55**\n\n**$87.42**\n\n**$83.44 $78.36**\n\n**$86.60**\n\n**$79.52**\n\n**$77.52**\n\n**$73.56**\n\n**$73.10**\n\n**$71.83**\n\n### $0\n\n### $10\n\n### $20\n\n### $30\n\n### $40\n\n### $50\n\n### $60\n\n### $70\n\n### $80\n\n### $90\n\n### $100\n\n### Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec\n\n### **Brent 2024 first-day-of-month average = $79.12/b Brent 2023 first-day-of-month average = $82.83/b**\n\n### **first-day-of-month price**\n\n### **9**\n\nThe U.S. Securities and Exchange Commission (SEC) requires companies to value proved reserves based on an average of the first-day-of-the-month price for the preceding 12-month period."],"full_content":"**Brent crude oil front-month futures prices** dollars per barrel (b)\n\n_Petroleum and Liquid Fuels Markets Team | Financial Review 2024 May 2025_\n\n# Proved reserves declined in value in 2024 because of lower crude oil prices\n\nData source: Intercontinental Exchange, Bloomberg L.P. Note: SEC=U.S. Securities and Exchange Commission\n\n**$75.89**\n\n**$78.70**\n\n**$83.55**\n\n**$87.42**\n\n**$83.44 $78.36**\n\n**$86.60**\n\n**$79.52**\n\n**$77.52**\n\n**$73.56**\n\n**$73.10**\n\n**$71.83**\n\n### $0\n\n### $10\n\n### $20\n\n### $30\n\n### $40\n\n### $50\n\n### $60\n\n### $70\n\n### $80\n\n### $90\n\n### $100\n\n### Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec\n\n### **Brent 2024 first-day-of-month average = $79.12/b Brent 2023 first-day-of-month average = $82.83/b**\n\n### **first-day-of-month price**\n\n### **9**\n\nThe U.S. Securities and Exchange Commission (SEC) requires companies to value proved reserves based on an average of the first-day-of-the-month price for the preceding 12-month period.","publish_date":null,"title":"Proved reserves declined in value in 2024 because of lower crude oil prices","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2024_Annual_Financial_Review_Upstream_FINAL_p9.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2024_Annual_Financial_Review_Upstream_FINAL_p9.pdf"},"request":{"example_id":"chart/2024_Annual_Financial_Review_Upstream_FINAL_p9","product_type":"parse","source_file_path":"2024_Annual_Financial_Review_Upstream_FINAL_p9.pdf"},"started_at":"2026-11-04T19:45:01.816150"} +{"completed_at":"2026-11-04T19:45:05.479471","latency_in_ms":2907,"output":{"example_id":"chart/2024_healthatglance_rep_en_p105","grounded_pages":[],"layout_pages":[],"markdown":" **103**\n\nHEALTH AT A GLANCE: EUROPE 2024 © OECD/EUROPEAN UNION 2024\n\n**Figure 3.9. Cancer mortality in EU countries by gender, 2021 (or nearest year)**\n\nNote: The EU average is weighted. Data for the United Kingdom refer to 2020 and have been calculated by the OECD. Source: Eurostat (hlth\\_cd\\_asdr2).\n\nStatLink 2 https://stat.link/hufsd5\n\n**Figure 3.10. Change in cancer mortality by cancer site in the EU, 2011-21**\n\nNote: The red bubble signals an increase in mortality during 2011-21 while the green bubbles signal a decrease. The size of the bubbles is proportional to the mortality rate in 2021. “Lung” includes cancers of the lung, trachea, and bronchus. “Brain” refers to both brain cancers and central nervous system cancers. The data correspond to the EU27 weighted averages. Source: Eurostat (hlth\\_cd\\_asdr2).\n\nStatLink 2 https://stat.link/styxji\n\n0\n\n50\n\n100\n\n150\n\n200\n\n250\n\n300\n\n350\n\n400\n\n450\n\n500 Age-standardised mortality rate per 100 000 population\n\nTotal Men Women\n\nBladder\n\nBreast\n\nCervix uteri\n\nColorectum Leukaemia\n\nLiver\n\nOvary\n\nPancreas\n\nProstate Skin\n\nStomach\n\nBrain\n\nLung\n\n\\-40%\n\n\\-30%\n\n\\-20%\n\n\\-10%\n\n0%\n\n10%\n\n20%\n\n0 10 20 30 40 50\n\nMortality in 2021 per 100 000 population\n\nPercentage change in mortality rate 2011-21","pages":[{"line_numbers":[],"markdown":" **103**\n\nHEALTH AT A GLANCE: EUROPE 2024 © OECD/EUROPEAN UNION 2024\n\n**Figure 3.9. Cancer mortality in EU countries by gender, 2021 (or nearest year)**\n\nNote: The EU average is weighted. Data for the United Kingdom refer to 2020 and have been calculated by the OECD. Source: Eurostat (hlth\\_cd\\_asdr2).\n\nStatLink 2 https://stat.link/hufsd5\n\n**Figure 3.10. Change in cancer mortality by cancer site in the EU, 2011-21**\n\nNote: The red bubble signals an increase in mortality during 2011-21 while the green bubbles signal a decrease. The size of the bubbles is proportional to the mortality rate in 2021. “Lung” includes cancers of the lung, trachea, and bronchus. “Brain” refers to both brain cancers and central nervous system cancers. The data correspond to the EU27 weighted averages. Source: Eurostat (hlth\\_cd\\_asdr2).\n\nStatLink 2 https://stat.link/styxji\n\n0\n\n50\n\n100\n\n150\n\n200\n\n250\n\n300\n\n350\n\n400\n\n450\n\n500 Age-standardised mortality rate per 100 000 population\n\nTotal Men Women\n\nBladder\n\nBreast\n\nCervix uteri\n\nColorectum Leukaemia\n\nLiver\n\nOvary\n\nPancreas\n\nProstate Skin\n\nStomach\n\nBrain\n\nLung\n\n\\-40%\n\n\\-30%\n\n\\-20%\n\n\\-10%\n\n0%\n\n10%\n\n20%\n\n0 10 20 30 40 50\n\nMortality in 2021 per 100 000 population\n\nPercentage change in mortality rate 2011-21","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":" **103**\n\nHEALTH AT A GLANCE: EUROPE 2024 © OECD/EUROPEAN UNION 2024\n\n**Figure 3.9. Cancer mortality in EU countries by gender, 2021 (or nearest year)**\n\nNote: The EU average is weighted. Data for the United Kingdom refer to 2020 and have been calculated by the OECD. Source: Eurostat (hlth\\_cd\\_asdr2).\n\nStatLink 2 https://stat.link/hufsd5\n\n**Figure 3.10. Change in cancer mortality by cancer site in the EU, 2011-21**\n\nNote: The red bubble signals an increase in mortality during 2011-21 while the green bubbles signal a decrease. The size of the bubbles is proportional to the mortality rate in 2021. “Lung” includes cancers of the lung, trachea, and bronchus. “Brain” refers to both brain cancers and central nervous system cancers. The data correspond to the EU27 weighted averages. Source: Eurostat (hlth\\_cd\\_asdr2).\n\nStatLink 2 https://stat.link/styxji\n\n0\n\n50\n\n100\n\n150\n\n200\n\n250\n\n300\n\n350\n\n400\n\n450\n\n500 Age-standardised mortality rate per 100 000 population\n\nTotal Men Women\n\nBladder\n\nBreast\n\nCervix uteri\n\nColorectum Leukaemia\n\nLiver\n\nOvary\n\nPancreas\n\nProstate Skin\n\nStomach\n\nBrain\n\nLung\n\n\\-40%\n\n\\-30%\n\n\\-20%\n\n\\-10%\n\n0%\n\n10%\n\n20%\n\n0 10 20 30 40 50\n\nMortality in 2021 per 100 000 population\n\nPercentage change in mortality rate 2011-21","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2024_healthatglance_rep_en_p105.pdf"]},"response":{"errors":[],"extract_id":"extract_6c0dd5d941c65acc28c2147b2c08e579","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":[" **103**\n\nHEALTH AT A GLANCE: EUROPE 2024 © OECD/EUROPEAN UNION 2024\n\n**Figure 3.9. Cancer mortality in EU countries by gender, 2021 (or nearest year)**\n\nNote: The EU average is weighted. Data for the United Kingdom refer to 2020 and have been calculated by the OECD. Source: Eurostat (hlth\\_cd\\_asdr2).\n\nStatLink 2 https://stat.link/hufsd5\n\n**Figure 3.10. Change in cancer mortality by cancer site in the EU, 2011-21**\n\nNote: The red bubble signals an increase in mortality during 2011-21 while the green bubbles signal a decrease. The size of the bubbles is proportional to the mortality rate in 2021. “Lung” includes cancers of the lung, trachea, and bronchus. “Brain” refers to both brain cancers and central nervous system cancers. The data correspond to the EU27 weighted averages. Source: Eurostat (hlth\\_cd\\_asdr2).\n\nStatLink 2 https://stat.link/styxji\n\n0\n\n50\n\n100\n\n150\n\n200\n\n250\n\n300\n\n350\n\n400\n\n450\n\n500 Age-standardised mortality rate per 100 000 population\n\nTotal Men Women\n\nBladder\n\nBreast\n\nCervix uteri\n\nColorectum Leukaemia\n\nLiver\n\nOvary\n\nPancreas\n\nProstate Skin\n\nStomach\n\nBrain\n\nLung\n\n\\-40%\n\n\\-30%\n\n\\-20%\n\n\\-10%\n\n0%\n\n10%\n\n20%\n\n0 10 20 30 40 50\n\nMortality in 2021 per 100 000 population\n\nPercentage change in mortality rate 2011-21"],"full_content":" **103**\n\nHEALTH AT A GLANCE: EUROPE 2024 © OECD/EUROPEAN UNION 2024\n\n**Figure 3.9. Cancer mortality in EU countries by gender, 2021 (or nearest year)**\n\nNote: The EU average is weighted. Data for the United Kingdom refer to 2020 and have been calculated by the OECD. Source: Eurostat (hlth\\_cd\\_asdr2).\n\nStatLink 2 https://stat.link/hufsd5\n\n**Figure 3.10. Change in cancer mortality by cancer site in the EU, 2011-21**\n\nNote: The red bubble signals an increase in mortality during 2011-21 while the green bubbles signal a decrease. The size of the bubbles is proportional to the mortality rate in 2021. “Lung” includes cancers of the lung, trachea, and bronchus. “Brain” refers to both brain cancers and central nervous system cancers. The data correspond to the EU27 weighted averages. Source: Eurostat (hlth\\_cd\\_asdr2).\n\nStatLink 2 https://stat.link/styxji\n\n0\n\n50\n\n100\n\n150\n\n200\n\n250\n\n300\n\n350\n\n400\n\n450\n\n500 Age-standardised mortality rate per 100 000 population\n\nTotal Men Women\n\nBladder\n\nBreast\n\nCervix uteri\n\nColorectum Leukaemia\n\nLiver\n\nOvary\n\nPancreas\n\nProstate Skin\n\nStomach\n\nBrain\n\nLung\n\n\\-40%\n\n\\-30%\n\n\\-20%\n\n\\-10%\n\n0%\n\n10%\n\n20%\n\n0 10 20 30 40 50\n\nMortality in 2021 per 100 000 population\n\nPercentage change in mortality rate 2011-21","publish_date":null,"title":"103  Figure 3.9. Cancer mortality in EU countries by gender, 2021 (or nearest year) Figure 3.10. Change in cancer","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2024_healthatglance_rep_en_p105.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2024_healthatglance_rep_en_p105.pdf"},"request":{"example_id":"chart/2024_healthatglance_rep_en_p105","product_type":"parse","source_file_path":"2024_healthatglance_rep_en_p105.pdf"},"started_at":"2026-11-04T19:45:02.572414"} +{"completed_at":"2026-11-04T19:45:05.889690","latency_in_ms":1468,"output":{"example_id":"chart/2024_healthatglance_rep_en_p115","grounded_pages":[],"layout_pages":[],"markdown":" **113**\n\nHEALTH AT A GLANCE: EUROPE 2024 © OECD/EUROPEAN UNION 2024\n\n**Figure 3.19. Suicide rate, by sex, 2021**\n\nNote: The EU average is weighted. Differences across countries may reflect, at least in part, differences in recording practices. 1. The data for the United Kingdom relate to 2020 (the rates have been calculated by the OECD based on the European population structure). Source: Eurostat (hlth\\_cd\\_asdr2).\n\nStatLink 2 https://stat.link/x8j1n2\n\n**Figure 3.20. Trends in suicide rates, selected European countries, 2004 to 2021**\n\nNote: The EU average is weighted. Differences across countries may reflect, at least in part, differences in recording practices. Source: Eurostat (hlth\\_cd\\_asdr2).\n\nStatLink 2 https://stat.link/odgpvz\n\n**Figure 3.21. Symptoms of depression among adults aged over 50, by sex, 2021-22**\n\nNote: The EU average is unweighted. Ireland does not participate in the survey. Symptoms of depression are measured using the EURO-D score, with a cut-off point of at least 4 out of 12 symptoms. Source: SHARE survey (wave 9).\n\nStatLink 2 https://stat.link/1s9j6g\n\n0 5 10 15 20 25 30 35 40\n\nAge-standardised rate per 100 000 population\n\nWomen Men Total\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n35\n\n40\n\n45\n\n2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021\n\nAge-standardised rates per 100 000 population\n\nCyprus EU27 Slovenia Hungary Lithuania\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\nMen Women Total\n\n% of population aged 50 and over with a score of 4 or higher","pages":[{"line_numbers":[],"markdown":" **113**\n\nHEALTH AT A GLANCE: EUROPE 2024 © OECD/EUROPEAN UNION 2024\n\n**Figure 3.19. Suicide rate, by sex, 2021**\n\nNote: The EU average is weighted. Differences across countries may reflect, at least in part, differences in recording practices. 1. The data for the United Kingdom relate to 2020 (the rates have been calculated by the OECD based on the European population structure). Source: Eurostat (hlth\\_cd\\_asdr2).\n\nStatLink 2 https://stat.link/x8j1n2\n\n**Figure 3.20. Trends in suicide rates, selected European countries, 2004 to 2021**\n\nNote: The EU average is weighted. Differences across countries may reflect, at least in part, differences in recording practices. Source: Eurostat (hlth\\_cd\\_asdr2).\n\nStatLink 2 https://stat.link/odgpvz\n\n**Figure 3.21. Symptoms of depression among adults aged over 50, by sex, 2021-22**\n\nNote: The EU average is unweighted. Ireland does not participate in the survey. Symptoms of depression are measured using the EURO-D score, with a cut-off point of at least 4 out of 12 symptoms. Source: SHARE survey (wave 9).\n\nStatLink 2 https://stat.link/1s9j6g\n\n0 5 10 15 20 25 30 35 40\n\nAge-standardised rate per 100 000 population\n\nWomen Men Total\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n35\n\n40\n\n45\n\n2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021\n\nAge-standardised rates per 100 000 population\n\nCyprus EU27 Slovenia Hungary Lithuania\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\nMen Women Total\n\n% of population aged 50 and over with a score of 4 or higher","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":" **113**\n\nHEALTH AT A GLANCE: EUROPE 2024 © OECD/EUROPEAN UNION 2024\n\n**Figure 3.19. Suicide rate, by sex, 2021**\n\nNote: The EU average is weighted. Differences across countries may reflect, at least in part, differences in recording practices. 1. The data for the United Kingdom relate to 2020 (the rates have been calculated by the OECD based on the European population structure). Source: Eurostat (hlth\\_cd\\_asdr2).\n\nStatLink 2 https://stat.link/x8j1n2\n\n**Figure 3.20. Trends in suicide rates, selected European countries, 2004 to 2021**\n\nNote: The EU average is weighted. Differences across countries may reflect, at least in part, differences in recording practices. Source: Eurostat (hlth\\_cd\\_asdr2).\n\nStatLink 2 https://stat.link/odgpvz\n\n**Figure 3.21. Symptoms of depression among adults aged over 50, by sex, 2021-22**\n\nNote: The EU average is unweighted. Ireland does not participate in the survey. Symptoms of depression are measured using the EURO-D score, with a cut-off point of at least 4 out of 12 symptoms. Source: SHARE survey (wave 9).\n\nStatLink 2 https://stat.link/1s9j6g\n\n0 5 10 15 20 25 30 35 40\n\nAge-standardised rate per 100 000 population\n\nWomen Men Total\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n35\n\n40\n\n45\n\n2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021\n\nAge-standardised rates per 100 000 population\n\nCyprus EU27 Slovenia Hungary Lithuania\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\nMen Women Total\n\n% of population aged 50 and over with a score of 4 or higher","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2024_healthatglance_rep_en_p115.pdf"]},"response":{"errors":[],"extract_id":"extract_49a573f584e979c60e7d60b61e35b863","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":[" **113**\n\nHEALTH AT A GLANCE: EUROPE 2024 © OECD/EUROPEAN UNION 2024\n\n**Figure 3.19. Suicide rate, by sex, 2021**\n\nNote: The EU average is weighted. Differences across countries may reflect, at least in part, differences in recording practices. 1. The data for the United Kingdom relate to 2020 (the rates have been calculated by the OECD based on the European population structure). Source: Eurostat (hlth\\_cd\\_asdr2).\n\nStatLink 2 https://stat.link/x8j1n2\n\n**Figure 3.20. Trends in suicide rates, selected European countries, 2004 to 2021**\n\nNote: The EU average is weighted. Differences across countries may reflect, at least in part, differences in recording practices. Source: Eurostat (hlth\\_cd\\_asdr2).\n\nStatLink 2 https://stat.link/odgpvz\n\n**Figure 3.21. Symptoms of depression among adults aged over 50, by sex, 2021-22**\n\nNote: The EU average is unweighted. Ireland does not participate in the survey. Symptoms of depression are measured using the EURO-D score, with a cut-off point of at least 4 out of 12 symptoms. Source: SHARE survey (wave 9).\n\nStatLink 2 https://stat.link/1s9j6g\n\n0 5 10 15 20 25 30 35 40\n\nAge-standardised rate per 100 000 population\n\nWomen Men Total\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n35\n\n40\n\n45\n\n2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021\n\nAge-standardised rates per 100 000 population\n\nCyprus EU27 Slovenia Hungary Lithuania\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\nMen Women Total\n\n% of population aged 50 and over with a score of 4 or higher"],"full_content":" **113**\n\nHEALTH AT A GLANCE: EUROPE 2024 © OECD/EUROPEAN UNION 2024\n\n**Figure 3.19. Suicide rate, by sex, 2021**\n\nNote: The EU average is weighted. Differences across countries may reflect, at least in part, differences in recording practices. 1. The data for the United Kingdom relate to 2020 (the rates have been calculated by the OECD based on the European population structure). Source: Eurostat (hlth\\_cd\\_asdr2).\n\nStatLink 2 https://stat.link/x8j1n2\n\n**Figure 3.20. Trends in suicide rates, selected European countries, 2004 to 2021**\n\nNote: The EU average is weighted. Differences across countries may reflect, at least in part, differences in recording practices. Source: Eurostat (hlth\\_cd\\_asdr2).\n\nStatLink 2 https://stat.link/odgpvz\n\n**Figure 3.21. Symptoms of depression among adults aged over 50, by sex, 2021-22**\n\nNote: The EU average is unweighted. Ireland does not participate in the survey. Symptoms of depression are measured using the EURO-D score, with a cut-off point of at least 4 out of 12 symptoms. Source: SHARE survey (wave 9).\n\nStatLink 2 https://stat.link/1s9j6g\n\n0 5 10 15 20 25 30 35 40\n\nAge-standardised rate per 100 000 population\n\nWomen Men Total\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n35\n\n40\n\n45\n\n2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021\n\nAge-standardised rates per 100 000 population\n\nCyprus EU27 Slovenia Hungary Lithuania\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\nMen Women Total\n\n% of population aged 50 and over with a score of 4 or higher","publish_date":null,"title":"113  Figure 3.19. Suicide rate, by sex, 2021 Figure 3.20. Trends in suicide rates, selected European countries, 2004 to","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2024_healthatglance_rep_en_p115.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2024_healthatglance_rep_en_p115.pdf"},"request":{"example_id":"chart/2024_healthatglance_rep_en_p115","product_type":"parse","source_file_path":"2024_healthatglance_rep_en_p115.pdf"},"started_at":"2026-11-04T19:45:04.420870"} +{"completed_at":"2026-11-04T19:45:07.653631","latency_in_ms":1762,"output":{"example_id":"chart/2024_healthatglance_rep_en_p119","grounded_pages":[],"layout_pages":[],"markdown":" **117**\n\nHEALTH AT A GLANCE: EUROPE 2024 © OECD/EUROPEAN UNION 2024\n\n**Figure 4.1. Tobacco smoking rates among 15-year-olds, 2014 and 2022**\n\nNote: The EU average is unweighted. 1. Data for Belgium refer to the unweighted average of Flanders and Wallonia/Brussels. Source: HBSC Data Browser, https://data-browser.hbsc.org/.\n\nStatLink 2 https://stat.link/431xgk\n\n**Figure 4.2. Cannabis smoking rates among 15-year-olds, 2014 and 2022**\n\nNote: The EU average is unweighted. 1. Data for Belgium refer to the unweighted average of Flanders and Wallonia/Brussels. the unweighted average of Flanders and Wallonia/Brussels. Source: HBSC Data Browser, https://data-browser.hbsc.org/.\n\nStatLink 2 https://stat.link/yq9al0\n\n7\n\n9 10\n\n11 12 12 12 12 13 13 14 14 15 15\n\n17 17 17\n\n18 18 19 19 20\n\n23 23\n\n26 27\n\n29\n\n32\n\n4\n\n9\n\n12\n\n12\n\n13\n\n15\n\n18\n\n20\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n35\n\n40\n\n2014 2022\n\n% of 15-year-olds who reported smoking a cigarette at least once in the last 30 days\n\n3\n\n4 4 4 4 4 5\n\n5\n\n5\n\n6 6 6\n\n7\n\n7\n\n7\n\n7\n\n7 8\n\n8 8\n\n8\n\n9 9 9\n\n10 12\n\n12\n\n13\n\n2\n\n3 4\n\n5 5\n\n10\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\n16\n\n2014 2022\n\n% of 15-year-olds who reported using cannabis at least once in the last 30 days","pages":[{"line_numbers":[],"markdown":" **117**\n\nHEALTH AT A GLANCE: EUROPE 2024 © OECD/EUROPEAN UNION 2024\n\n**Figure 4.1. Tobacco smoking rates among 15-year-olds, 2014 and 2022**\n\nNote: The EU average is unweighted. 1. Data for Belgium refer to the unweighted average of Flanders and Wallonia/Brussels. Source: HBSC Data Browser, https://data-browser.hbsc.org/.\n\nStatLink 2 https://stat.link/431xgk\n\n**Figure 4.2. Cannabis smoking rates among 15-year-olds, 2014 and 2022**\n\nNote: The EU average is unweighted. 1. Data for Belgium refer to the unweighted average of Flanders and Wallonia/Brussels. the unweighted average of Flanders and Wallonia/Brussels. Source: HBSC Data Browser, https://data-browser.hbsc.org/.\n\nStatLink 2 https://stat.link/yq9al0\n\n7\n\n9 10\n\n11 12 12 12 12 13 13 14 14 15 15\n\n17 17 17\n\n18 18 19 19 20\n\n23 23\n\n26 27\n\n29\n\n32\n\n4\n\n9\n\n12\n\n12\n\n13\n\n15\n\n18\n\n20\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n35\n\n40\n\n2014 2022\n\n% of 15-year-olds who reported smoking a cigarette at least once in the last 30 days\n\n3\n\n4 4 4 4 4 5\n\n5\n\n5\n\n6 6 6\n\n7\n\n7\n\n7\n\n7\n\n7 8\n\n8 8\n\n8\n\n9 9 9\n\n10 12\n\n12\n\n13\n\n2\n\n3 4\n\n5 5\n\n10\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\n16\n\n2014 2022\n\n% of 15-year-olds who reported using cannabis at least once in the last 30 days","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":" **117**\n\nHEALTH AT A GLANCE: EUROPE 2024 © OECD/EUROPEAN UNION 2024\n\n**Figure 4.1. Tobacco smoking rates among 15-year-olds, 2014 and 2022**\n\nNote: The EU average is unweighted. 1. Data for Belgium refer to the unweighted average of Flanders and Wallonia/Brussels. Source: HBSC Data Browser, https://data-browser.hbsc.org/.\n\nStatLink 2 https://stat.link/431xgk\n\n**Figure 4.2. Cannabis smoking rates among 15-year-olds, 2014 and 2022**\n\nNote: The EU average is unweighted. 1. Data for Belgium refer to the unweighted average of Flanders and Wallonia/Brussels. the unweighted average of Flanders and Wallonia/Brussels. Source: HBSC Data Browser, https://data-browser.hbsc.org/.\n\nStatLink 2 https://stat.link/yq9al0\n\n7\n\n9 10\n\n11 12 12 12 12 13 13 14 14 15 15\n\n17 17 17\n\n18 18 19 19 20\n\n23 23\n\n26 27\n\n29\n\n32\n\n4\n\n9\n\n12\n\n12\n\n13\n\n15\n\n18\n\n20\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n35\n\n40\n\n2014 2022\n\n% of 15-year-olds who reported smoking a cigarette at least once in the last 30 days\n\n3\n\n4 4 4 4 4 5\n\n5\n\n5\n\n6 6 6\n\n7\n\n7\n\n7\n\n7\n\n7 8\n\n8 8\n\n8\n\n9 9 9\n\n10 12\n\n12\n\n13\n\n2\n\n3 4\n\n5 5\n\n10\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\n16\n\n2014 2022\n\n% of 15-year-olds who reported using cannabis at least once in the last 30 days","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2024_healthatglance_rep_en_p119.pdf"]},"response":{"errors":[],"extract_id":"extract_b3b019cd24661fe107cb41f885bcf04f","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":[" **117**\n\nHEALTH AT A GLANCE: EUROPE 2024 © OECD/EUROPEAN UNION 2024\n\n**Figure 4.1. Tobacco smoking rates among 15-year-olds, 2014 and 2022**\n\nNote: The EU average is unweighted. 1. Data for Belgium refer to the unweighted average of Flanders and Wallonia/Brussels. Source: HBSC Data Browser, https://data-browser.hbsc.org/.\n\nStatLink 2 https://stat.link/431xgk\n\n**Figure 4.2. Cannabis smoking rates among 15-year-olds, 2014 and 2022**\n\nNote: The EU average is unweighted. 1. Data for Belgium refer to the unweighted average of Flanders and Wallonia/Brussels. the unweighted average of Flanders and Wallonia/Brussels. Source: HBSC Data Browser, https://data-browser.hbsc.org/.\n\nStatLink 2 https://stat.link/yq9al0\n\n7\n\n9 10\n\n11 12 12 12 12 13 13 14 14 15 15\n\n17 17 17\n\n18 18 19 19 20\n\n23 23\n\n26 27\n\n29\n\n32\n\n4\n\n9\n\n12\n\n12\n\n13\n\n15\n\n18\n\n20\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n35\n\n40\n\n2014 2022\n\n% of 15-year-olds who reported smoking a cigarette at least once in the last 30 days\n\n3\n\n4 4 4 4 4 5\n\n5\n\n5\n\n6 6 6\n\n7\n\n7\n\n7\n\n7\n\n7 8\n\n8 8\n\n8\n\n9 9 9\n\n10 12\n\n12\n\n13\n\n2\n\n3 4\n\n5 5\n\n10\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\n16\n\n2014 2022\n\n% of 15-year-olds who reported using cannabis at least once in the last 30 days"],"full_content":" **117**\n\nHEALTH AT A GLANCE: EUROPE 2024 © OECD/EUROPEAN UNION 2024\n\n**Figure 4.1. Tobacco smoking rates among 15-year-olds, 2014 and 2022**\n\nNote: The EU average is unweighted. 1. Data for Belgium refer to the unweighted average of Flanders and Wallonia/Brussels. Source: HBSC Data Browser, https://data-browser.hbsc.org/.\n\nStatLink 2 https://stat.link/431xgk\n\n**Figure 4.2. Cannabis smoking rates among 15-year-olds, 2014 and 2022**\n\nNote: The EU average is unweighted. 1. Data for Belgium refer to the unweighted average of Flanders and Wallonia/Brussels. the unweighted average of Flanders and Wallonia/Brussels. Source: HBSC Data Browser, https://data-browser.hbsc.org/.\n\nStatLink 2 https://stat.link/yq9al0\n\n7\n\n9 10\n\n11 12 12 12 12 13 13 14 14 15 15\n\n17 17 17\n\n18 18 19 19 20\n\n23 23\n\n26 27\n\n29\n\n32\n\n4\n\n9\n\n12\n\n12\n\n13\n\n15\n\n18\n\n20\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n35\n\n40\n\n2014 2022\n\n% of 15-year-olds who reported smoking a cigarette at least once in the last 30 days\n\n3\n\n4 4 4 4 4 5\n\n5\n\n5\n\n6 6 6\n\n7\n\n7\n\n7\n\n7\n\n7 8\n\n8 8\n\n8\n\n9 9 9\n\n10 12\n\n12\n\n13\n\n2\n\n3 4\n\n5 5\n\n10\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\n16\n\n2014 2022\n\n% of 15-year-olds who reported using cannabis at least once in the last 30 days","publish_date":null,"title":"117  Figure 4.1. Tobacco smoking rates among 15-year-olds, 2014 and 2022 Figure 4.2. Cannabis smoking rates among 15-year-olds, 2014","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2024_healthatglance_rep_en_p119.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2024_healthatglance_rep_en_p119.pdf"},"request":{"example_id":"chart/2024_healthatglance_rep_en_p119","product_type":"parse","source_file_path":"2024_healthatglance_rep_en_p119.pdf"},"started_at":"2026-11-04T19:45:05.891456"} +{"completed_at":"2026-11-04T19:45:09.689722","latency_in_ms":2583,"output":{"example_id":"chart/2024_healthatglance_rep_en_p187","grounded_pages":[],"layout_pages":[],"markdown":" **185**\n\nHEALTH AT A GLANCE: EUROPE 2024 © OECD/EUROPEAN UNION 2024\n\n**Figure 7.5. Healthcare coverage for selected services, 2022 (or nearest year)**\n\nGovernment and compulsory insurance spending as proportion of total health spending by type of services\n\nNote: Outpatient medical services mainly refer to services provided by generalists and specialists in the outpatient sector. Pharmaceuticals include prescribed and over-the-counter medicines as well as medical non-durables. Therapeutic appliances refer to vision products, hearing aids, wheelchairs and other medical devices. N/A means data not available. The EU average is unweighted. Source: OECD Health Statistics 2024.\n\nStatLink 2 https://stat.link/1ubzf4\n\n37%\n\n44%\n\n45%\n\n17%\n\n45%\n\n15%\n\n41%\n\n44%\n\n34%\n\nN/A\n\n69%\n\n64%\n\nN/A\n\n47%\n\n31%\n\n15%\n\n11%\n\n16%\n\n46%\n\n69%\n\n46%\n\n39%\n\n22%\n\n24%\n\n29%\n\n33%\n\n3%\n\n36%\n\n32%\n\n82%\n\n41%\n\n30%\n\n53%\n\n15%\n\n**Therapeutic appliances**\n\n59%\n\n67%\n\n66%\n\n23%\n\n75%\n\n90%\n\n56%\n\n40%\n\n55%\n\n59%\n\n82%\n\n81%\n\n51%\n\n45%\n\n83%\n\n63%\n\n41%\n\n48%\n\n74%\n\n55%\n\n68%\n\n34%\n\n55%\n\n45%\n\n68%\n\n51%\n\n71%\n\n56%\n\n36%\n\n42%\n\n53%\n\n24%\n\n71%\n\n67%\n\n**Pharmaceuticals**\n\n35%\n\n48%\n\n33%\n\n50%\n\n62%\n\n13%\n\n43%\n\n35%\n\n27%\n\n41%\n\n67%\n\n72%\n\nN/A\n\n34%\n\nN/A\n\nN/A\n\n12%\n\n13%\n\n48%\n\n12%\n\n12%\n\n35%\n\nN/A\n\n7%\n\n46%\n\n44%\n\n2%\n\n43%\n\n34%\n\n32%\n\n27%\n\nN/A\n\n8%\n\n43%\n\n**Dental care**\n\n77%\n\n84%\n\n67%\n\n61%\n\n88%\n\n76%\n\n92%\n\n93%\n\n83%\n\n87%\n\n85%\n\n90%\n\n64%\n\n67%\n\n76%\n\n59%\n\n59%\n\n69%\n\n89%\n\n58%\n\n87%\n\n71%\n\n56%\n\n80%\n\n98%\n\n78%\n\n76%\n\n92%\n\n83%\n\n84%\n\n86%\n\n66%\n\n72%\n\n91%\n\n**Outpatient medical care**\n\n90%\n\n88%\n\n77%\n\n92%\n\n93%\n\n91%\n\n95%\n\n91%\n\n99%\n\n96%\n\n95%\n\n97%\n\n65%\n\n95%\n\n79%\n\n97%\n\n91%\n\n92%\n\n94%\n\n86%\n\n91%\n\n94%\n\n80%\n\n99%\n\n87%\n\n88%\n\n89%\n\n99%\n\n82%\n\n99%\n\n99%\n\n92%\n\n83%\n\n95%\n\nEU27\n\nAustria\n\nBelgium\n\nBulgaria\n\nCroatia\n\nCyprus\n\nCzechia\n\nDenmark\n\nEstonia\n\nFinland\n\nFrance\n\nGermany\n\nGreece\n\nHungary\n\nIreland\n\nItaly\n\nLatvia\n\nLithuania\n\nLuxembourg\n\nMalta\n\nNetherlands\n\nPoland\n\nPortugal\n\nRomania\n\nSlovak Republic\n\nSlovenia\n\nSpain\n\nSweden\n\nBosnia and Herzegovina\n\nIceland\n\nNorway\n\nSerbia\n\nSwitzerland\n\nUnited Kingdom\n\n**Inpatient care**","pages":[{"line_numbers":[],"markdown":" **185**\n\nHEALTH AT A GLANCE: EUROPE 2024 © OECD/EUROPEAN UNION 2024\n\n**Figure 7.5. Healthcare coverage for selected services, 2022 (or nearest year)**\n\nGovernment and compulsory insurance spending as proportion of total health spending by type of services\n\nNote: Outpatient medical services mainly refer to services provided by generalists and specialists in the outpatient sector. Pharmaceuticals include prescribed and over-the-counter medicines as well as medical non-durables. Therapeutic appliances refer to vision products, hearing aids, wheelchairs and other medical devices. N/A means data not available. The EU average is unweighted. Source: OECD Health Statistics 2024.\n\nStatLink 2 https://stat.link/1ubzf4\n\n37%\n\n44%\n\n45%\n\n17%\n\n45%\n\n15%\n\n41%\n\n44%\n\n34%\n\nN/A\n\n69%\n\n64%\n\nN/A\n\n47%\n\n31%\n\n15%\n\n11%\n\n16%\n\n46%\n\n69%\n\n46%\n\n39%\n\n22%\n\n24%\n\n29%\n\n33%\n\n3%\n\n36%\n\n32%\n\n82%\n\n41%\n\n30%\n\n53%\n\n15%\n\n**Therapeutic appliances**\n\n59%\n\n67%\n\n66%\n\n23%\n\n75%\n\n90%\n\n56%\n\n40%\n\n55%\n\n59%\n\n82%\n\n81%\n\n51%\n\n45%\n\n83%\n\n63%\n\n41%\n\n48%\n\n74%\n\n55%\n\n68%\n\n34%\n\n55%\n\n45%\n\n68%\n\n51%\n\n71%\n\n56%\n\n36%\n\n42%\n\n53%\n\n24%\n\n71%\n\n67%\n\n**Pharmaceuticals**\n\n35%\n\n48%\n\n33%\n\n50%\n\n62%\n\n13%\n\n43%\n\n35%\n\n27%\n\n41%\n\n67%\n\n72%\n\nN/A\n\n34%\n\nN/A\n\nN/A\n\n12%\n\n13%\n\n48%\n\n12%\n\n12%\n\n35%\n\nN/A\n\n7%\n\n46%\n\n44%\n\n2%\n\n43%\n\n34%\n\n32%\n\n27%\n\nN/A\n\n8%\n\n43%\n\n**Dental care**\n\n77%\n\n84%\n\n67%\n\n61%\n\n88%\n\n76%\n\n92%\n\n93%\n\n83%\n\n87%\n\n85%\n\n90%\n\n64%\n\n67%\n\n76%\n\n59%\n\n59%\n\n69%\n\n89%\n\n58%\n\n87%\n\n71%\n\n56%\n\n80%\n\n98%\n\n78%\n\n76%\n\n92%\n\n83%\n\n84%\n\n86%\n\n66%\n\n72%\n\n91%\n\n**Outpatient medical care**\n\n90%\n\n88%\n\n77%\n\n92%\n\n93%\n\n91%\n\n95%\n\n91%\n\n99%\n\n96%\n\n95%\n\n97%\n\n65%\n\n95%\n\n79%\n\n97%\n\n91%\n\n92%\n\n94%\n\n86%\n\n91%\n\n94%\n\n80%\n\n99%\n\n87%\n\n88%\n\n89%\n\n99%\n\n82%\n\n99%\n\n99%\n\n92%\n\n83%\n\n95%\n\nEU27\n\nAustria\n\nBelgium\n\nBulgaria\n\nCroatia\n\nCyprus\n\nCzechia\n\nDenmark\n\nEstonia\n\nFinland\n\nFrance\n\nGermany\n\nGreece\n\nHungary\n\nIreland\n\nItaly\n\nLatvia\n\nLithuania\n\nLuxembourg\n\nMalta\n\nNetherlands\n\nPoland\n\nPortugal\n\nRomania\n\nSlovak Republic\n\nSlovenia\n\nSpain\n\nSweden\n\nBosnia and Herzegovina\n\nIceland\n\nNorway\n\nSerbia\n\nSwitzerland\n\nUnited Kingdom\n\n**Inpatient care**","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":" **185**\n\nHEALTH AT A GLANCE: EUROPE 2024 © OECD/EUROPEAN UNION 2024\n\n**Figure 7.5. Healthcare coverage for selected services, 2022 (or nearest year)**\n\nGovernment and compulsory insurance spending as proportion of total health spending by type of services\n\nNote: Outpatient medical services mainly refer to services provided by generalists and specialists in the outpatient sector. Pharmaceuticals include prescribed and over-the-counter medicines as well as medical non-durables. Therapeutic appliances refer to vision products, hearing aids, wheelchairs and other medical devices. N/A means data not available. The EU average is unweighted. Source: OECD Health Statistics 2024.\n\nStatLink 2 https://stat.link/1ubzf4\n\n37%\n\n44%\n\n45%\n\n17%\n\n45%\n\n15%\n\n41%\n\n44%\n\n34%\n\nN/A\n\n69%\n\n64%\n\nN/A\n\n47%\n\n31%\n\n15%\n\n11%\n\n16%\n\n46%\n\n69%\n\n46%\n\n39%\n\n22%\n\n24%\n\n29%\n\n33%\n\n3%\n\n36%\n\n32%\n\n82%\n\n41%\n\n30%\n\n53%\n\n15%\n\n**Therapeutic appliances**\n\n59%\n\n67%\n\n66%\n\n23%\n\n75%\n\n90%\n\n56%\n\n40%\n\n55%\n\n59%\n\n82%\n\n81%\n\n51%\n\n45%\n\n83%\n\n63%\n\n41%\n\n48%\n\n74%\n\n55%\n\n68%\n\n34%\n\n55%\n\n45%\n\n68%\n\n51%\n\n71%\n\n56%\n\n36%\n\n42%\n\n53%\n\n24%\n\n71%\n\n67%\n\n**Pharmaceuticals**\n\n35%\n\n48%\n\n33%\n\n50%\n\n62%\n\n13%\n\n43%\n\n35%\n\n27%\n\n41%\n\n67%\n\n72%\n\nN/A\n\n34%\n\nN/A\n\nN/A\n\n12%\n\n13%\n\n48%\n\n12%\n\n12%\n\n35%\n\nN/A\n\n7%\n\n46%\n\n44%\n\n2%\n\n43%\n\n34%\n\n32%\n\n27%\n\nN/A\n\n8%\n\n43%\n\n**Dental care**\n\n77%\n\n84%\n\n67%\n\n61%\n\n88%\n\n76%\n\n92%\n\n93%\n\n83%\n\n87%\n\n85%\n\n90%\n\n64%\n\n67%\n\n76%\n\n59%\n\n59%\n\n69%\n\n89%\n\n58%\n\n87%\n\n71%\n\n56%\n\n80%\n\n98%\n\n78%\n\n76%\n\n92%\n\n83%\n\n84%\n\n86%\n\n66%\n\n72%\n\n91%\n\n**Outpatient medical care**\n\n90%\n\n88%\n\n77%\n\n92%\n\n93%\n\n91%\n\n95%\n\n91%\n\n99%\n\n96%\n\n95%\n\n97%\n\n65%\n\n95%\n\n79%\n\n97%\n\n91%\n\n92%\n\n94%\n\n86%\n\n91%\n\n94%\n\n80%\n\n99%\n\n87%\n\n88%\n\n89%\n\n99%\n\n82%\n\n99%\n\n99%\n\n92%\n\n83%\n\n95%\n\nEU27\n\nAustria\n\nBelgium\n\nBulgaria\n\nCroatia\n\nCyprus\n\nCzechia\n\nDenmark\n\nEstonia\n\nFinland\n\nFrance\n\nGermany\n\nGreece\n\nHungary\n\nIreland\n\nItaly\n\nLatvia\n\nLithuania\n\nLuxembourg\n\nMalta\n\nNetherlands\n\nPoland\n\nPortugal\n\nRomania\n\nSlovak Republic\n\nSlovenia\n\nSpain\n\nSweden\n\nBosnia and Herzegovina\n\nIceland\n\nNorway\n\nSerbia\n\nSwitzerland\n\nUnited Kingdom\n\n**Inpatient care**","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2024_healthatglance_rep_en_p187.pdf"]},"response":{"errors":[],"extract_id":"extract_beed4466a85ba4ed7d1284c16f81971e","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":[" **185**\n\nHEALTH AT A GLANCE: EUROPE 2024 © OECD/EUROPEAN UNION 2024\n\n**Figure 7.5. Healthcare coverage for selected services, 2022 (or nearest year)**\n\nGovernment and compulsory insurance spending as proportion of total health spending by type of services\n\nNote: Outpatient medical services mainly refer to services provided by generalists and specialists in the outpatient sector. Pharmaceuticals include prescribed and over-the-counter medicines as well as medical non-durables. Therapeutic appliances refer to vision products, hearing aids, wheelchairs and other medical devices. N/A means data not available. The EU average is unweighted. Source: OECD Health Statistics 2024.\n\nStatLink 2 https://stat.link/1ubzf4\n\n37%\n\n44%\n\n45%\n\n17%\n\n45%\n\n15%\n\n41%\n\n44%\n\n34%\n\nN/A\n\n69%\n\n64%\n\nN/A\n\n47%\n\n31%\n\n15%\n\n11%\n\n16%\n\n46%\n\n69%\n\n46%\n\n39%\n\n22%\n\n24%\n\n29%\n\n33%\n\n3%\n\n36%\n\n32%\n\n82%\n\n41%\n\n30%\n\n53%\n\n15%\n\n**Therapeutic appliances**\n\n59%\n\n67%\n\n66%\n\n23%\n\n75%\n\n90%\n\n56%\n\n40%\n\n55%\n\n59%\n\n82%\n\n81%\n\n51%\n\n45%\n\n83%\n\n63%\n\n41%\n\n48%\n\n74%\n\n55%\n\n68%\n\n34%\n\n55%\n\n45%\n\n68%\n\n51%\n\n71%\n\n56%\n\n36%\n\n42%\n\n53%\n\n24%\n\n71%\n\n67%\n\n**Pharmaceuticals**\n\n35%\n\n48%\n\n33%\n\n50%\n\n62%\n\n13%\n\n43%\n\n35%\n\n27%\n\n41%\n\n67%\n\n72%\n\nN/A\n\n34%\n\nN/A\n\nN/A\n\n12%\n\n13%\n\n48%\n\n12%\n\n12%\n\n35%\n\nN/A\n\n7%\n\n46%\n\n44%\n\n2%\n\n43%\n\n34%\n\n32%\n\n27%\n\nN/A\n\n8%\n\n43%\n\n**Dental care**\n\n77%\n\n84%\n\n67%\n\n61%\n\n88%\n\n76%\n\n92%\n\n93%\n\n83%\n\n87%\n\n85%\n\n90%\n\n64%\n\n67%\n\n76%\n\n59%\n\n59%\n\n69%\n\n89%\n\n58%\n\n87%\n\n71%\n\n56%\n\n80%\n\n98%\n\n78%\n\n76%\n\n92%\n\n83%\n\n84%\n\n86%\n\n66%\n\n72%\n\n91%\n\n**Outpatient medical care**\n\n90%\n\n88%\n\n77%\n\n92%\n\n93%\n\n91%\n\n95%\n\n91%\n\n99%\n\n96%\n\n95%\n\n97%\n\n65%\n\n95%\n\n79%\n\n97%\n\n91%\n\n92%\n\n94%\n\n86%\n\n91%\n\n94%\n\n80%\n\n99%\n\n87%\n\n88%\n\n89%\n\n99%\n\n82%\n\n99%\n\n99%\n\n92%\n\n83%\n\n95%\n\nEU27\n\nAustria\n\nBelgium\n\nBulgaria\n\nCroatia\n\nCyprus\n\nCzechia\n\nDenmark\n\nEstonia\n\nFinland\n\nFrance\n\nGermany\n\nGreece\n\nHungary\n\nIreland\n\nItaly\n\nLatvia\n\nLithuania\n\nLuxembourg\n\nMalta\n\nNetherlands\n\nPoland\n\nPortugal\n\nRomania\n\nSlovak Republic\n\nSlovenia\n\nSpain\n\nSweden\n\nBosnia and Herzegovina\n\nIceland\n\nNorway\n\nSerbia\n\nSwitzerland\n\nUnited Kingdom\n\n**Inpatient care**"],"full_content":" **185**\n\nHEALTH AT A GLANCE: EUROPE 2024 © OECD/EUROPEAN UNION 2024\n\n**Figure 7.5. Healthcare coverage for selected services, 2022 (or nearest year)**\n\nGovernment and compulsory insurance spending as proportion of total health spending by type of services\n\nNote: Outpatient medical services mainly refer to services provided by generalists and specialists in the outpatient sector. Pharmaceuticals include prescribed and over-the-counter medicines as well as medical non-durables. Therapeutic appliances refer to vision products, hearing aids, wheelchairs and other medical devices. N/A means data not available. The EU average is unweighted. Source: OECD Health Statistics 2024.\n\nStatLink 2 https://stat.link/1ubzf4\n\n37%\n\n44%\n\n45%\n\n17%\n\n45%\n\n15%\n\n41%\n\n44%\n\n34%\n\nN/A\n\n69%\n\n64%\n\nN/A\n\n47%\n\n31%\n\n15%\n\n11%\n\n16%\n\n46%\n\n69%\n\n46%\n\n39%\n\n22%\n\n24%\n\n29%\n\n33%\n\n3%\n\n36%\n\n32%\n\n82%\n\n41%\n\n30%\n\n53%\n\n15%\n\n**Therapeutic appliances**\n\n59%\n\n67%\n\n66%\n\n23%\n\n75%\n\n90%\n\n56%\n\n40%\n\n55%\n\n59%\n\n82%\n\n81%\n\n51%\n\n45%\n\n83%\n\n63%\n\n41%\n\n48%\n\n74%\n\n55%\n\n68%\n\n34%\n\n55%\n\n45%\n\n68%\n\n51%\n\n71%\n\n56%\n\n36%\n\n42%\n\n53%\n\n24%\n\n71%\n\n67%\n\n**Pharmaceuticals**\n\n35%\n\n48%\n\n33%\n\n50%\n\n62%\n\n13%\n\n43%\n\n35%\n\n27%\n\n41%\n\n67%\n\n72%\n\nN/A\n\n34%\n\nN/A\n\nN/A\n\n12%\n\n13%\n\n48%\n\n12%\n\n12%\n\n35%\n\nN/A\n\n7%\n\n46%\n\n44%\n\n2%\n\n43%\n\n34%\n\n32%\n\n27%\n\nN/A\n\n8%\n\n43%\n\n**Dental care**\n\n77%\n\n84%\n\n67%\n\n61%\n\n88%\n\n76%\n\n92%\n\n93%\n\n83%\n\n87%\n\n85%\n\n90%\n\n64%\n\n67%\n\n76%\n\n59%\n\n59%\n\n69%\n\n89%\n\n58%\n\n87%\n\n71%\n\n56%\n\n80%\n\n98%\n\n78%\n\n76%\n\n92%\n\n83%\n\n84%\n\n86%\n\n66%\n\n72%\n\n91%\n\n**Outpatient medical care**\n\n90%\n\n88%\n\n77%\n\n92%\n\n93%\n\n91%\n\n95%\n\n91%\n\n99%\n\n96%\n\n95%\n\n97%\n\n65%\n\n95%\n\n79%\n\n97%\n\n91%\n\n92%\n\n94%\n\n86%\n\n91%\n\n94%\n\n80%\n\n99%\n\n87%\n\n88%\n\n89%\n\n99%\n\n82%\n\n99%\n\n99%\n\n92%\n\n83%\n\n95%\n\nEU27\n\nAustria\n\nBelgium\n\nBulgaria\n\nCroatia\n\nCyprus\n\nCzechia\n\nDenmark\n\nEstonia\n\nFinland\n\nFrance\n\nGermany\n\nGreece\n\nHungary\n\nIreland\n\nItaly\n\nLatvia\n\nLithuania\n\nLuxembourg\n\nMalta\n\nNetherlands\n\nPoland\n\nPortugal\n\nRomania\n\nSlovak Republic\n\nSlovenia\n\nSpain\n\nSweden\n\nBosnia and Herzegovina\n\nIceland\n\nNorway\n\nSerbia\n\nSwitzerland\n\nUnited Kingdom\n\n**Inpatient care**","publish_date":null,"title":"185  Figure 7.5. Healthcare coverage for selected services, 2022 (or nearest year)","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2024_healthatglance_rep_en_p187.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2024_healthatglance_rep_en_p187.pdf"},"request":{"example_id":"chart/2024_healthatglance_rep_en_p187","product_type":"parse","source_file_path":"2024_healthatglance_rep_en_p187.pdf"},"started_at":"2026-11-04T19:45:07.105841"} +{"completed_at":"2026-11-04T19:45:12.925336","latency_in_ms":4446,"output":{"example_id":"chart/2024_healthatglance_rep_en_p197","grounded_pages":[],"layout_pages":[],"markdown":" **195**\n\nHEALTH AT A GLANCE: EUROPE 2024 © OECD/EUROPEAN UNION 2024\n\n**Figure 7.15. Practising nurses per 1 000 population, 2010 and 2022 (or nearest year)**\n\nNote: The EU average is unweighted. The values shown are for 2022. Professional nurses refer to nurses with an education level meeting the EC Directive on the recognition of professional qualifications (minimum of 3 years of postsecondary education for general nurses and at least an equivalent level of education for specialist nurses). Associate professional nurses include other categories of nurses that are recognised in some countries. Some countries are not able to make a clear distinction between these two main categories. 1. The latest data refer to 2017. 2. The data for France and Poland starts in 2013, for the Netherlands in 2014 and for Hungary in 2017. Source: OECD Health Statistics 2024; Eurostat (hlth\\_rs\\_prs2); WHO National Health Workforce Accounts for Moldova and Ukraine.\n\nStatLink 2 https://stat.link/d7te05\n\n**Figure 7.16. Ratio of nurses to doctors, 2022 (or nearest year)**\n\nNote: The EU average is unweighted. 1. For countries that have not provided data for practising nurses and/or practising doctors, the numbers relate to “professionally active” nurses and doctors. 2. The ratio for Portugal is underestimated (professionally active nurses/all doctors licensed to practise). 3. For Greece, the data refer to nurses and doctors employed in hospitals. 4. The data for Luxembourg refers to 2017 only. 5. The data for Ireland refers to 2023 to avoid a break in the series for doctors in 2022. Source: OECD Health Statistics 2024; Eurostat (hlth\\_rs\\_prs2); WHO National Health Workforce Accounts for Moldova.\n\nStatLink 2 https://stat.link/7vnosp\n\n14\\.1\n\n13\\.3\n\n12\\.0 11.7 11.5 11.010.9 10.4 10.4\n\n9\\.0 8.8 8.4 8.2 7.9 7.7 7.5 7.5\n\n6\\.6 6.5 6.2 5.7 5.5 5.3 4.3 4.2\n\n3\\.9\n\n18\\.4\n\n15\\.6\n\n14\\.9\n\n8\\.7\n\n6\\.6\n\n5\\.0 4.5\n\n2\\.9\n\n11\\.6\n\n5\\.7 5.9\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\n16\n\n18\n\n20\n\nPer 1 000 population\n\nProfessional nurses Associate professional nurses All nurses (no distinction) (2010) All nurses (no distinction)\n\n3\\.9 3.9\n\n3\\.6\n\n3\\.2 3.1\n\n2\\.9\n\n2\\.8 2.6\n\n2\\.5\n\n2\\.3 2.2 2.2 2.1 2.0 1.9 1.9\n\n1\\.7 1.7 1.6 1.6 1.6 1.5 1.5 1.4 1.3 1.2\n\n1\\.0 1.0\n\n4\\.1\n\n3\\.4\n\n3\\.2\n\n2\\.7\n\n2\\.1 2.1\n\n1\\.9\n\n1\\.5 1.5\n\n1\\.3\n\n0\n\n1\n\n2\n\n3\n\n4\n\n5\n\nRatio","pages":[{"line_numbers":[],"markdown":" **195**\n\nHEALTH AT A GLANCE: EUROPE 2024 © OECD/EUROPEAN UNION 2024\n\n**Figure 7.15. Practising nurses per 1 000 population, 2010 and 2022 (or nearest year)**\n\nNote: The EU average is unweighted. The values shown are for 2022. Professional nurses refer to nurses with an education level meeting the EC Directive on the recognition of professional qualifications (minimum of 3 years of postsecondary education for general nurses and at least an equivalent level of education for specialist nurses). Associate professional nurses include other categories of nurses that are recognised in some countries. Some countries are not able to make a clear distinction between these two main categories. 1. The latest data refer to 2017. 2. The data for France and Poland starts in 2013, for the Netherlands in 2014 and for Hungary in 2017. Source: OECD Health Statistics 2024; Eurostat (hlth\\_rs\\_prs2); WHO National Health Workforce Accounts for Moldova and Ukraine.\n\nStatLink 2 https://stat.link/d7te05\n\n**Figure 7.16. Ratio of nurses to doctors, 2022 (or nearest year)**\n\nNote: The EU average is unweighted. 1. For countries that have not provided data for practising nurses and/or practising doctors, the numbers relate to “professionally active” nurses and doctors. 2. The ratio for Portugal is underestimated (professionally active nurses/all doctors licensed to practise). 3. For Greece, the data refer to nurses and doctors employed in hospitals. 4. The data for Luxembourg refers to 2017 only. 5. The data for Ireland refers to 2023 to avoid a break in the series for doctors in 2022. Source: OECD Health Statistics 2024; Eurostat (hlth\\_rs\\_prs2); WHO National Health Workforce Accounts for Moldova.\n\nStatLink 2 https://stat.link/7vnosp\n\n14\\.1\n\n13\\.3\n\n12\\.0 11.7 11.5 11.010.9 10.4 10.4\n\n9\\.0 8.8 8.4 8.2 7.9 7.7 7.5 7.5\n\n6\\.6 6.5 6.2 5.7 5.5 5.3 4.3 4.2\n\n3\\.9\n\n18\\.4\n\n15\\.6\n\n14\\.9\n\n8\\.7\n\n6\\.6\n\n5\\.0 4.5\n\n2\\.9\n\n11\\.6\n\n5\\.7 5.9\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\n16\n\n18\n\n20\n\nPer 1 000 population\n\nProfessional nurses Associate professional nurses All nurses (no distinction) (2010) All nurses (no distinction)\n\n3\\.9 3.9\n\n3\\.6\n\n3\\.2 3.1\n\n2\\.9\n\n2\\.8 2.6\n\n2\\.5\n\n2\\.3 2.2 2.2 2.1 2.0 1.9 1.9\n\n1\\.7 1.7 1.6 1.6 1.6 1.5 1.5 1.4 1.3 1.2\n\n1\\.0 1.0\n\n4\\.1\n\n3\\.4\n\n3\\.2\n\n2\\.7\n\n2\\.1 2.1\n\n1\\.9\n\n1\\.5 1.5\n\n1\\.3\n\n0\n\n1\n\n2\n\n3\n\n4\n\n5\n\nRatio","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":" **195**\n\nHEALTH AT A GLANCE: EUROPE 2024 © OECD/EUROPEAN UNION 2024\n\n**Figure 7.15. Practising nurses per 1 000 population, 2010 and 2022 (or nearest year)**\n\nNote: The EU average is unweighted. The values shown are for 2022. Professional nurses refer to nurses with an education level meeting the EC Directive on the recognition of professional qualifications (minimum of 3 years of postsecondary education for general nurses and at least an equivalent level of education for specialist nurses). Associate professional nurses include other categories of nurses that are recognised in some countries. Some countries are not able to make a clear distinction between these two main categories. 1. The latest data refer to 2017. 2. The data for France and Poland starts in 2013, for the Netherlands in 2014 and for Hungary in 2017. Source: OECD Health Statistics 2024; Eurostat (hlth\\_rs\\_prs2); WHO National Health Workforce Accounts for Moldova and Ukraine.\n\nStatLink 2 https://stat.link/d7te05\n\n**Figure 7.16. Ratio of nurses to doctors, 2022 (or nearest year)**\n\nNote: The EU average is unweighted. 1. For countries that have not provided data for practising nurses and/or practising doctors, the numbers relate to “professionally active” nurses and doctors. 2. The ratio for Portugal is underestimated (professionally active nurses/all doctors licensed to practise). 3. For Greece, the data refer to nurses and doctors employed in hospitals. 4. The data for Luxembourg refers to 2017 only. 5. The data for Ireland refers to 2023 to avoid a break in the series for doctors in 2022. Source: OECD Health Statistics 2024; Eurostat (hlth\\_rs\\_prs2); WHO National Health Workforce Accounts for Moldova.\n\nStatLink 2 https://stat.link/7vnosp\n\n14\\.1\n\n13\\.3\n\n12\\.0 11.7 11.5 11.010.9 10.4 10.4\n\n9\\.0 8.8 8.4 8.2 7.9 7.7 7.5 7.5\n\n6\\.6 6.5 6.2 5.7 5.5 5.3 4.3 4.2\n\n3\\.9\n\n18\\.4\n\n15\\.6\n\n14\\.9\n\n8\\.7\n\n6\\.6\n\n5\\.0 4.5\n\n2\\.9\n\n11\\.6\n\n5\\.7 5.9\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\n16\n\n18\n\n20\n\nPer 1 000 population\n\nProfessional nurses Associate professional nurses All nurses (no distinction) (2010) All nurses (no distinction)\n\n3\\.9 3.9\n\n3\\.6\n\n3\\.2 3.1\n\n2\\.9\n\n2\\.8 2.6\n\n2\\.5\n\n2\\.3 2.2 2.2 2.1 2.0 1.9 1.9\n\n1\\.7 1.7 1.6 1.6 1.6 1.5 1.5 1.4 1.3 1.2\n\n1\\.0 1.0\n\n4\\.1\n\n3\\.4\n\n3\\.2\n\n2\\.7\n\n2\\.1 2.1\n\n1\\.9\n\n1\\.5 1.5\n\n1\\.3\n\n0\n\n1\n\n2\n\n3\n\n4\n\n5\n\nRatio","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2024_healthatglance_rep_en_p197.pdf"]},"response":{"errors":[],"extract_id":"extract_0f33e9724fc4143aeb90196a311dc461","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":[" **195**\n\nHEALTH AT A GLANCE: EUROPE 2024 © OECD/EUROPEAN UNION 2024\n\n**Figure 7.15. Practising nurses per 1 000 population, 2010 and 2022 (or nearest year)**\n\nNote: The EU average is unweighted. The values shown are for 2022. Professional nurses refer to nurses with an education level meeting the EC Directive on the recognition of professional qualifications (minimum of 3 years of postsecondary education for general nurses and at least an equivalent level of education for specialist nurses). Associate professional nurses include other categories of nurses that are recognised in some countries. Some countries are not able to make a clear distinction between these two main categories. 1. The latest data refer to 2017. 2. The data for France and Poland starts in 2013, for the Netherlands in 2014 and for Hungary in 2017. Source: OECD Health Statistics 2024; Eurostat (hlth\\_rs\\_prs2); WHO National Health Workforce Accounts for Moldova and Ukraine.\n\nStatLink 2 https://stat.link/d7te05\n\n**Figure 7.16. Ratio of nurses to doctors, 2022 (or nearest year)**\n\nNote: The EU average is unweighted. 1. For countries that have not provided data for practising nurses and/or practising doctors, the numbers relate to “professionally active” nurses and doctors. 2. The ratio for Portugal is underestimated (professionally active nurses/all doctors licensed to practise). 3. For Greece, the data refer to nurses and doctors employed in hospitals. 4. The data for Luxembourg refers to 2017 only. 5. The data for Ireland refers to 2023 to avoid a break in the series for doctors in 2022. Source: OECD Health Statistics 2024; Eurostat (hlth\\_rs\\_prs2); WHO National Health Workforce Accounts for Moldova.\n\nStatLink 2 https://stat.link/7vnosp\n\n14\\.1\n\n13\\.3\n\n12\\.0 11.7 11.5 11.010.9 10.4 10.4\n\n9\\.0 8.8 8.4 8.2 7.9 7.7 7.5 7.5\n\n6\\.6 6.5 6.2 5.7 5.5 5.3 4.3 4.2\n\n3\\.9\n\n18\\.4\n\n15\\.6\n\n14\\.9\n\n8\\.7\n\n6\\.6\n\n5\\.0 4.5\n\n2\\.9\n\n11\\.6\n\n5\\.7 5.9\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\n16\n\n18\n\n20\n\nPer 1 000 population\n\nProfessional nurses Associate professional nurses All nurses (no distinction) (2010) All nurses (no distinction)\n\n3\\.9 3.9\n\n3\\.6\n\n3\\.2 3.1\n\n2\\.9\n\n2\\.8 2.6\n\n2\\.5\n\n2\\.3 2.2 2.2 2.1 2.0 1.9 1.9\n\n1\\.7 1.7 1.6 1.6 1.6 1.5 1.5 1.4 1.3 1.2\n\n1\\.0 1.0\n\n4\\.1\n\n3\\.4\n\n3\\.2\n\n2\\.7\n\n2\\.1 2.1\n\n1\\.9\n\n1\\.5 1.5\n\n1\\.3\n\n0\n\n1\n\n2\n\n3\n\n4\n\n5\n\nRatio"],"full_content":" **195**\n\nHEALTH AT A GLANCE: EUROPE 2024 © OECD/EUROPEAN UNION 2024\n\n**Figure 7.15. Practising nurses per 1 000 population, 2010 and 2022 (or nearest year)**\n\nNote: The EU average is unweighted. The values shown are for 2022. Professional nurses refer to nurses with an education level meeting the EC Directive on the recognition of professional qualifications (minimum of 3 years of postsecondary education for general nurses and at least an equivalent level of education for specialist nurses). Associate professional nurses include other categories of nurses that are recognised in some countries. Some countries are not able to make a clear distinction between these two main categories. 1. The latest data refer to 2017. 2. The data for France and Poland starts in 2013, for the Netherlands in 2014 and for Hungary in 2017. Source: OECD Health Statistics 2024; Eurostat (hlth\\_rs\\_prs2); WHO National Health Workforce Accounts for Moldova and Ukraine.\n\nStatLink 2 https://stat.link/d7te05\n\n**Figure 7.16. Ratio of nurses to doctors, 2022 (or nearest year)**\n\nNote: The EU average is unweighted. 1. For countries that have not provided data for practising nurses and/or practising doctors, the numbers relate to “professionally active” nurses and doctors. 2. The ratio for Portugal is underestimated (professionally active nurses/all doctors licensed to practise). 3. For Greece, the data refer to nurses and doctors employed in hospitals. 4. The data for Luxembourg refers to 2017 only. 5. The data for Ireland refers to 2023 to avoid a break in the series for doctors in 2022. Source: OECD Health Statistics 2024; Eurostat (hlth\\_rs\\_prs2); WHO National Health Workforce Accounts for Moldova.\n\nStatLink 2 https://stat.link/7vnosp\n\n14\\.1\n\n13\\.3\n\n12\\.0 11.7 11.5 11.010.9 10.4 10.4\n\n9\\.0 8.8 8.4 8.2 7.9 7.7 7.5 7.5\n\n6\\.6 6.5 6.2 5.7 5.5 5.3 4.3 4.2\n\n3\\.9\n\n18\\.4\n\n15\\.6\n\n14\\.9\n\n8\\.7\n\n6\\.6\n\n5\\.0 4.5\n\n2\\.9\n\n11\\.6\n\n5\\.7 5.9\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\n16\n\n18\n\n20\n\nPer 1 000 population\n\nProfessional nurses Associate professional nurses All nurses (no distinction) (2010) All nurses (no distinction)\n\n3\\.9 3.9\n\n3\\.6\n\n3\\.2 3.1\n\n2\\.9\n\n2\\.8 2.6\n\n2\\.5\n\n2\\.3 2.2 2.2 2.1 2.0 1.9 1.9\n\n1\\.7 1.7 1.6 1.6 1.6 1.5 1.5 1.4 1.3 1.2\n\n1\\.0 1.0\n\n4\\.1\n\n3\\.4\n\n3\\.2\n\n2\\.7\n\n2\\.1 2.1\n\n1\\.9\n\n1\\.5 1.5\n\n1\\.3\n\n0\n\n1\n\n2\n\n3\n\n4\n\n5\n\nRatio","publish_date":null,"title":"195  Figure 7.15. Practising nurses per 1 000 population, 2010 and 2022 (or nearest year) Figure 7.16. Ratio of","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2024_healthatglance_rep_en_p197.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2024_healthatglance_rep_en_p197.pdf"},"request":{"example_id":"chart/2024_healthatglance_rep_en_p197","product_type":"parse","source_file_path":"2024_healthatglance_rep_en_p197.pdf"},"started_at":"2026-11-04T19:45:08.479034"} +{"completed_at":"2026-11-04T19:45:11.096605","latency_in_ms":1436,"output":{"example_id":"chart/2024_healthatglance_rep_en_p45","grounded_pages":[],"layout_pages":[],"markdown":" **43**\n\nHEALTH AT A GLANCE: EUROPE 2024 © OECD OECD/EUROPEAN UNION 2024\n\nOver the past decade, the decision of many doctors to continue working beyond the standard retirement age has helped to avoid an exacerbation of shortages in many EU countries. This trend is evident in the increasing proportion of doctors aged over 65 between 2012 and 2022, which has been particularly significant in countries such as Italy, the Slovak Republic, Germany, France and Belgium (Figure 1.18).\n\n**Figure 1.18. The share of doctors aged 65 and over has increased over the past decade in several EU countries**\n\nSource: Eurostat (hlth\\_rs\\_phys).\n\nStatLink 2 https://stat.link/k5rgpc\n\nItaly stands out with 27% of its medical workforce aged 65 and over, foreshadowing substantial doctor attrition in the coming years. The outflow of doctors due to retirement in the country is expected to peak in 2025 with the retirement rate projected to return to normal levels only by 2030 (Sumai-ASSOPROF, 2023[33]). To mitigate the impact of this anticipated retirement wave, Italy has implemented measures to retain older physicians longer in the system while simultaneously boosting the training of new doctors in recent years (Figure 1.13). In 2024, legislation was passed to temporarily suspend the obligation for doctors employed by public facilities (NHS) to retire at age 70, allowing them to remain in the ward until they are 72. This legislation also enables NHS local authorities to keep medical personnel in service beyond retirement age on a voluntary basis, addressing both the serious staff shortage and the training and tutoring needs of newly hired staff. Additionally, those who have retired since September 2023 can be readmitted to service within these limits (Consiglio dei ministri, 2023[34]).\n\nPension reforms and a potential increase in doctors’ capacity and willingness to work longer may have a significant impact on the supply of doctors and replacement needs in the short to medium term. Many EU countries besides Italy have combined efforts to increase statutory retirement ages, curb early retirement, and offer incentives for longer working lives along with initiatives to foster job flexibility for older workers. Normal retirement ages have increased or are set to increase in most EU countries, reaching an average of over 65 years for both men and women starting their careers today. In Denmark, Estonia, the Netherlands and Sweden, the normal retirement age is projected to rise to 70 years or more if life expectancy gains materialise as projected and legislated links with life expectancy are applied (OECD, 2023[35]). Although there are few studies examining the specific impact of these pension reforms on doctors and other health workers, they can be expected to prolong the working lives and expand the supply of doctors, nurses and other health workers after age 65 in the coming years.\n\nAccording to the 2023 OECD Health System Characteristics Survey, six EU countries (Czechia, Greece, Ireland, Latvia, Portugal and Spain) indicated that part of their strategies to maintain or increase the supply of doctors is to prolong their working lives. This may be achieved through general pension reforms that increase the retirement age for all workers or through more specific incentives targeting doctors to prolong their working lives.\n\nItaly\n\nSlovak Republic\n\nGermany\n\nFrance Belgium\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 .\n\nShare of doctors aged 65 and over","pages":[{"line_numbers":[],"markdown":" **43**\n\nHEALTH AT A GLANCE: EUROPE 2024 © OECD OECD/EUROPEAN UNION 2024\n\nOver the past decade, the decision of many doctors to continue working beyond the standard retirement age has helped to avoid an exacerbation of shortages in many EU countries. This trend is evident in the increasing proportion of doctors aged over 65 between 2012 and 2022, which has been particularly significant in countries such as Italy, the Slovak Republic, Germany, France and Belgium (Figure 1.18).\n\n**Figure 1.18. The share of doctors aged 65 and over has increased over the past decade in several EU countries**\n\nSource: Eurostat (hlth\\_rs\\_phys).\n\nStatLink 2 https://stat.link/k5rgpc\n\nItaly stands out with 27% of its medical workforce aged 65 and over, foreshadowing substantial doctor attrition in the coming years. The outflow of doctors due to retirement in the country is expected to peak in 2025 with the retirement rate projected to return to normal levels only by 2030 (Sumai-ASSOPROF, 2023[33]). To mitigate the impact of this anticipated retirement wave, Italy has implemented measures to retain older physicians longer in the system while simultaneously boosting the training of new doctors in recent years (Figure 1.13). In 2024, legislation was passed to temporarily suspend the obligation for doctors employed by public facilities (NHS) to retire at age 70, allowing them to remain in the ward until they are 72. This legislation also enables NHS local authorities to keep medical personnel in service beyond retirement age on a voluntary basis, addressing both the serious staff shortage and the training and tutoring needs of newly hired staff. Additionally, those who have retired since September 2023 can be readmitted to service within these limits (Consiglio dei ministri, 2023[34]).\n\nPension reforms and a potential increase in doctors’ capacity and willingness to work longer may have a significant impact on the supply of doctors and replacement needs in the short to medium term. Many EU countries besides Italy have combined efforts to increase statutory retirement ages, curb early retirement, and offer incentives for longer working lives along with initiatives to foster job flexibility for older workers. Normal retirement ages have increased or are set to increase in most EU countries, reaching an average of over 65 years for both men and women starting their careers today. In Denmark, Estonia, the Netherlands and Sweden, the normal retirement age is projected to rise to 70 years or more if life expectancy gains materialise as projected and legislated links with life expectancy are applied (OECD, 2023[35]). Although there are few studies examining the specific impact of these pension reforms on doctors and other health workers, they can be expected to prolong the working lives and expand the supply of doctors, nurses and other health workers after age 65 in the coming years.\n\nAccording to the 2023 OECD Health System Characteristics Survey, six EU countries (Czechia, Greece, Ireland, Latvia, Portugal and Spain) indicated that part of their strategies to maintain or increase the supply of doctors is to prolong their working lives. This may be achieved through general pension reforms that increase the retirement age for all workers or through more specific incentives targeting doctors to prolong their working lives.\n\nItaly\n\nSlovak Republic\n\nGermany\n\nFrance Belgium\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 .\n\nShare of doctors aged 65 and over","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":" **43**\n\nHEALTH AT A GLANCE: EUROPE 2024 © OECD OECD/EUROPEAN UNION 2024\n\nOver the past decade, the decision of many doctors to continue working beyond the standard retirement age has helped to avoid an exacerbation of shortages in many EU countries. This trend is evident in the increasing proportion of doctors aged over 65 between 2012 and 2022, which has been particularly significant in countries such as Italy, the Slovak Republic, Germany, France and Belgium (Figure 1.18).\n\n**Figure 1.18. The share of doctors aged 65 and over has increased over the past decade in several EU countries**\n\nSource: Eurostat (hlth\\_rs\\_phys).\n\nStatLink 2 https://stat.link/k5rgpc\n\nItaly stands out with 27% of its medical workforce aged 65 and over, foreshadowing substantial doctor attrition in the coming years. The outflow of doctors due to retirement in the country is expected to peak in 2025 with the retirement rate projected to return to normal levels only by 2030 (Sumai-ASSOPROF, 2023[33]). To mitigate the impact of this anticipated retirement wave, Italy has implemented measures to retain older physicians longer in the system while simultaneously boosting the training of new doctors in recent years (Figure 1.13). In 2024, legislation was passed to temporarily suspend the obligation for doctors employed by public facilities (NHS) to retire at age 70, allowing them to remain in the ward until they are 72. This legislation also enables NHS local authorities to keep medical personnel in service beyond retirement age on a voluntary basis, addressing both the serious staff shortage and the training and tutoring needs of newly hired staff. Additionally, those who have retired since September 2023 can be readmitted to service within these limits (Consiglio dei ministri, 2023[34]).\n\nPension reforms and a potential increase in doctors’ capacity and willingness to work longer may have a significant impact on the supply of doctors and replacement needs in the short to medium term. Many EU countries besides Italy have combined efforts to increase statutory retirement ages, curb early retirement, and offer incentives for longer working lives along with initiatives to foster job flexibility for older workers. Normal retirement ages have increased or are set to increase in most EU countries, reaching an average of over 65 years for both men and women starting their careers today. In Denmark, Estonia, the Netherlands and Sweden, the normal retirement age is projected to rise to 70 years or more if life expectancy gains materialise as projected and legislated links with life expectancy are applied (OECD, 2023[35]). Although there are few studies examining the specific impact of these pension reforms on doctors and other health workers, they can be expected to prolong the working lives and expand the supply of doctors, nurses and other health workers after age 65 in the coming years.\n\nAccording to the 2023 OECD Health System Characteristics Survey, six EU countries (Czechia, Greece, Ireland, Latvia, Portugal and Spain) indicated that part of their strategies to maintain or increase the supply of doctors is to prolong their working lives. This may be achieved through general pension reforms that increase the retirement age for all workers or through more specific incentives targeting doctors to prolong their working lives.\n\nItaly\n\nSlovak Republic\n\nGermany\n\nFrance Belgium\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 .\n\nShare of doctors aged 65 and over","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2024_healthatglance_rep_en_p45.pdf"]},"response":{"errors":[],"extract_id":"extract_f0a1b1608cc99721c654e41370443cc9","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":[" **43**\n\nHEALTH AT A GLANCE: EUROPE 2024 © OECD OECD/EUROPEAN UNION 2024\n\nOver the past decade, the decision of many doctors to continue working beyond the standard retirement age has helped to avoid an exacerbation of shortages in many EU countries. This trend is evident in the increasing proportion of doctors aged over 65 between 2012 and 2022, which has been particularly significant in countries such as Italy, the Slovak Republic, Germany, France and Belgium (Figure 1.18).\n\n**Figure 1.18. The share of doctors aged 65 and over has increased over the past decade in several EU countries**\n\nSource: Eurostat (hlth\\_rs\\_phys).\n\nStatLink 2 https://stat.link/k5rgpc\n\nItaly stands out with 27% of its medical workforce aged 65 and over, foreshadowing substantial doctor attrition in the coming years. The outflow of doctors due to retirement in the country is expected to peak in 2025 with the retirement rate projected to return to normal levels only by 2030 (Sumai-ASSOPROF, 2023[33]). To mitigate the impact of this anticipated retirement wave, Italy has implemented measures to retain older physicians longer in the system while simultaneously boosting the training of new doctors in recent years (Figure 1.13). In 2024, legislation was passed to temporarily suspend the obligation for doctors employed by public facilities (NHS) to retire at age 70, allowing them to remain in the ward until they are 72. This legislation also enables NHS local authorities to keep medical personnel in service beyond retirement age on a voluntary basis, addressing both the serious staff shortage and the training and tutoring needs of newly hired staff. Additionally, those who have retired since September 2023 can be readmitted to service within these limits (Consiglio dei ministri, 2023[34]).\n\nPension reforms and a potential increase in doctors’ capacity and willingness to work longer may have a significant impact on the supply of doctors and replacement needs in the short to medium term. Many EU countries besides Italy have combined efforts to increase statutory retirement ages, curb early retirement, and offer incentives for longer working lives along with initiatives to foster job flexibility for older workers. Normal retirement ages have increased or are set to increase in most EU countries, reaching an average of over 65 years for both men and women starting their careers today. In Denmark, Estonia, the Netherlands and Sweden, the normal retirement age is projected to rise to 70 years or more if life expectancy gains materialise as projected and legislated links with life expectancy are applied (OECD, 2023[35]). Although there are few studies examining the specific impact of these pension reforms on doctors and other health workers, they can be expected to prolong the working lives and expand the supply of doctors, nurses and other health workers after age 65 in the coming years.\n\nAccording to the 2023 OECD Health System Characteristics Survey, six EU countries (Czechia, Greece, Ireland, Latvia, Portugal and Spain) indicated that part of their strategies to maintain or increase the supply of doctors is to prolong their working lives. This may be achieved through general pension reforms that increase the retirement age for all workers or through more specific incentives targeting doctors to prolong their working lives.\n\nItaly\n\nSlovak Republic\n\nGermany\n\nFrance Belgium\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 .\n\nShare of doctors aged 65 and over"],"full_content":" **43**\n\nHEALTH AT A GLANCE: EUROPE 2024 © OECD OECD/EUROPEAN UNION 2024\n\nOver the past decade, the decision of many doctors to continue working beyond the standard retirement age has helped to avoid an exacerbation of shortages in many EU countries. This trend is evident in the increasing proportion of doctors aged over 65 between 2012 and 2022, which has been particularly significant in countries such as Italy, the Slovak Republic, Germany, France and Belgium (Figure 1.18).\n\n**Figure 1.18. The share of doctors aged 65 and over has increased over the past decade in several EU countries**\n\nSource: Eurostat (hlth\\_rs\\_phys).\n\nStatLink 2 https://stat.link/k5rgpc\n\nItaly stands out with 27% of its medical workforce aged 65 and over, foreshadowing substantial doctor attrition in the coming years. The outflow of doctors due to retirement in the country is expected to peak in 2025 with the retirement rate projected to return to normal levels only by 2030 (Sumai-ASSOPROF, 2023[33]). To mitigate the impact of this anticipated retirement wave, Italy has implemented measures to retain older physicians longer in the system while simultaneously boosting the training of new doctors in recent years (Figure 1.13). In 2024, legislation was passed to temporarily suspend the obligation for doctors employed by public facilities (NHS) to retire at age 70, allowing them to remain in the ward until they are 72. This legislation also enables NHS local authorities to keep medical personnel in service beyond retirement age on a voluntary basis, addressing both the serious staff shortage and the training and tutoring needs of newly hired staff. Additionally, those who have retired since September 2023 can be readmitted to service within these limits (Consiglio dei ministri, 2023[34]).\n\nPension reforms and a potential increase in doctors’ capacity and willingness to work longer may have a significant impact on the supply of doctors and replacement needs in the short to medium term. Many EU countries besides Italy have combined efforts to increase statutory retirement ages, curb early retirement, and offer incentives for longer working lives along with initiatives to foster job flexibility for older workers. Normal retirement ages have increased or are set to increase in most EU countries, reaching an average of over 65 years for both men and women starting their careers today. In Denmark, Estonia, the Netherlands and Sweden, the normal retirement age is projected to rise to 70 years or more if life expectancy gains materialise as projected and legislated links with life expectancy are applied (OECD, 2023[35]). Although there are few studies examining the specific impact of these pension reforms on doctors and other health workers, they can be expected to prolong the working lives and expand the supply of doctors, nurses and other health workers after age 65 in the coming years.\n\nAccording to the 2023 OECD Health System Characteristics Survey, six EU countries (Czechia, Greece, Ireland, Latvia, Portugal and Spain) indicated that part of their strategies to maintain or increase the supply of doctors is to prolong their working lives. This may be achieved through general pension reforms that increase the retirement age for all workers or through more specific incentives targeting doctors to prolong their working lives.\n\nItaly\n\nSlovak Republic\n\nGermany\n\nFrance Belgium\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 .\n\nShare of doctors aged 65 and over","publish_date":null,"title":"43  Figure 1.18. The share of doctors aged 65 and over has increased over the past decade in several","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2024_healthatglance_rep_en_p45.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2024_healthatglance_rep_en_p45.pdf"},"request":{"example_id":"chart/2024_healthatglance_rep_en_p45","product_type":"parse","source_file_path":"2024_healthatglance_rep_en_p45.pdf"},"started_at":"2026-11-04T19:45:09.660200"} +{"completed_at":"2026-11-04T19:45:13.337545","latency_in_ms":1619,"output":{"example_id":"chart/2024_healthatglance_rep_en_p46","grounded_pages":[],"layout_pages":[],"markdown":"**44** \n\nHEALTH AT A GLANCE: EUROPE 2024 © OECD OECD/EUROPEAN UNION 2024\n\n**_1\\.6.3. Improving the remuneration of some categories of health workers can increase attractiveness_**\n\n**_and retention_**\n\nAn important aspect of job quality relates to earnings, as explicitly recognised under the OECD Job Quality Framework. Based on standard economic theory, wage adjustments should address any imbalance in the supply and demand for different categories of health workers: if there are shortages of some categories of workers, wages should go up to increase supply (either in terms of the number of health workers and/or their working hours) while the demand should decrease, thereby reducing the shortage. However, this supposes that wages are free to adjust to labour market conditions and that the supply of health workers is fairly responsive (or “elastic”) to wage changes. In practice, both of these conditions often do not apply in the health sector. ⁵\n\nThe remuneration of health workers varies widely by categories of workers depending on qualification levels, years of experience (seniority), working time (or activity rates) and negotiating power. In general across EU countries, the remuneration of doctors is several times higher than the average wage of workers in all occupations reflecting their higher qualifications and longer working hours. GPs earn two to four times more than the average wage, while specialists earn two to five times more (see indicator on _“_ Remuneration of doctors” in Chapter 7). The remuneration of nurses is substantially lower than that of doctors. On average across EU countries, it is about 20% higher than the average wage of workers in all occupations, although in some countries nurses do not earn more than the average wage (see indicator on “Remuneration of nurses” in Chapter 7). The remuneration of personal care workers is about 30% lower than the economy-wide average wage on average across EU countries (OECD, 2023[36]).\n\nBased on data collected through the Eurofound’s EWCTS survey in 2021, over 40% of personal care workers on average across EU countries reported having financial difficulties (almost the same proportion as cleaners and cooks and waiters). Over 20% of nurses also reported having difficulties in making ends meet, slightly lower than the average across all occupations considered to be in shortage, but nonetheless a significant proportion. Only about 5% of doctors reported facing such financial difficulties on average across EU countries (Figure 1.19).\n\nWorkers in certain health occupations also believe that they were not fairly rewarded. This is notably the case for nurses: only 40% of nurses felt in 2021 that they were paid fairly in relation to their efforts and achievements on average across EU countries (19 percentage points less than the average across all occupations).\n\n**Figure 1.19. Almost a quarter of nurses and half of personal care workers report financial difficulties across EU countries**\n\nSource: European Commission (2023[18]), _Employment and social developments in Europe 2023_ , https://data.europa.eu/doi/10.2767/089698.\n\nStatLink 2 https://stat.link/jf50p8\n\n43\n\n22\n\n5\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n35\n\n40\n\n45\n\n50\n\nCleaners Cooks and\n\nwaiters\n\nPersonal care\n\nworkers\n\nSalespeople Building workers\n\nTruck and bus\n\ndrivers\n\nMetal workers EU average Electronic workers\n\nNurses Software developers\n\nDoctors\n\n% of workers in shortage occupations reporting difficulty in making ends meet","pages":[{"line_numbers":[],"markdown":"**44** \n\nHEALTH AT A GLANCE: EUROPE 2024 © OECD OECD/EUROPEAN UNION 2024\n\n**_1\\.6.3. Improving the remuneration of some categories of health workers can increase attractiveness_**\n\n**_and retention_**\n\nAn important aspect of job quality relates to earnings, as explicitly recognised under the OECD Job Quality Framework. Based on standard economic theory, wage adjustments should address any imbalance in the supply and demand for different categories of health workers: if there are shortages of some categories of workers, wages should go up to increase supply (either in terms of the number of health workers and/or their working hours) while the demand should decrease, thereby reducing the shortage. However, this supposes that wages are free to adjust to labour market conditions and that the supply of health workers is fairly responsive (or “elastic”) to wage changes. In practice, both of these conditions often do not apply in the health sector. ⁵\n\nThe remuneration of health workers varies widely by categories of workers depending on qualification levels, years of experience (seniority), working time (or activity rates) and negotiating power. In general across EU countries, the remuneration of doctors is several times higher than the average wage of workers in all occupations reflecting their higher qualifications and longer working hours. GPs earn two to four times more than the average wage, while specialists earn two to five times more (see indicator on _“_ Remuneration of doctors” in Chapter 7). The remuneration of nurses is substantially lower than that of doctors. On average across EU countries, it is about 20% higher than the average wage of workers in all occupations, although in some countries nurses do not earn more than the average wage (see indicator on “Remuneration of nurses” in Chapter 7). The remuneration of personal care workers is about 30% lower than the economy-wide average wage on average across EU countries (OECD, 2023[36]).\n\nBased on data collected through the Eurofound’s EWCTS survey in 2021, over 40% of personal care workers on average across EU countries reported having financial difficulties (almost the same proportion as cleaners and cooks and waiters). Over 20% of nurses also reported having difficulties in making ends meet, slightly lower than the average across all occupations considered to be in shortage, but nonetheless a significant proportion. Only about 5% of doctors reported facing such financial difficulties on average across EU countries (Figure 1.19).\n\nWorkers in certain health occupations also believe that they were not fairly rewarded. This is notably the case for nurses: only 40% of nurses felt in 2021 that they were paid fairly in relation to their efforts and achievements on average across EU countries (19 percentage points less than the average across all occupations).\n\n**Figure 1.19. Almost a quarter of nurses and half of personal care workers report financial difficulties across EU countries**\n\nSource: European Commission (2023[18]), _Employment and social developments in Europe 2023_ , https://data.europa.eu/doi/10.2767/089698.\n\nStatLink 2 https://stat.link/jf50p8\n\n43\n\n22\n\n5\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n35\n\n40\n\n45\n\n50\n\nCleaners Cooks and\n\nwaiters\n\nPersonal care\n\nworkers\n\nSalespeople Building workers\n\nTruck and bus\n\ndrivers\n\nMetal workers EU average Electronic workers\n\nNurses Software developers\n\nDoctors\n\n% of workers in shortage occupations reporting difficulty in making ends meet","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**44** \n\nHEALTH AT A GLANCE: EUROPE 2024 © OECD OECD/EUROPEAN UNION 2024\n\n**_1\\.6.3. Improving the remuneration of some categories of health workers can increase attractiveness_**\n\n**_and retention_**\n\nAn important aspect of job quality relates to earnings, as explicitly recognised under the OECD Job Quality Framework. Based on standard economic theory, wage adjustments should address any imbalance in the supply and demand for different categories of health workers: if there are shortages of some categories of workers, wages should go up to increase supply (either in terms of the number of health workers and/or their working hours) while the demand should decrease, thereby reducing the shortage. However, this supposes that wages are free to adjust to labour market conditions and that the supply of health workers is fairly responsive (or “elastic”) to wage changes. In practice, both of these conditions often do not apply in the health sector. ⁵\n\nThe remuneration of health workers varies widely by categories of workers depending on qualification levels, years of experience (seniority), working time (or activity rates) and negotiating power. In general across EU countries, the remuneration of doctors is several times higher than the average wage of workers in all occupations reflecting their higher qualifications and longer working hours. GPs earn two to four times more than the average wage, while specialists earn two to five times more (see indicator on _“_ Remuneration of doctors” in Chapter 7). The remuneration of nurses is substantially lower than that of doctors. On average across EU countries, it is about 20% higher than the average wage of workers in all occupations, although in some countries nurses do not earn more than the average wage (see indicator on “Remuneration of nurses” in Chapter 7). The remuneration of personal care workers is about 30% lower than the economy-wide average wage on average across EU countries (OECD, 2023[36]).\n\nBased on data collected through the Eurofound’s EWCTS survey in 2021, over 40% of personal care workers on average across EU countries reported having financial difficulties (almost the same proportion as cleaners and cooks and waiters). Over 20% of nurses also reported having difficulties in making ends meet, slightly lower than the average across all occupations considered to be in shortage, but nonetheless a significant proportion. Only about 5% of doctors reported facing such financial difficulties on average across EU countries (Figure 1.19).\n\nWorkers in certain health occupations also believe that they were not fairly rewarded. This is notably the case for nurses: only 40% of nurses felt in 2021 that they were paid fairly in relation to their efforts and achievements on average across EU countries (19 percentage points less than the average across all occupations).\n\n**Figure 1.19. Almost a quarter of nurses and half of personal care workers report financial difficulties across EU countries**\n\nSource: European Commission (2023[18]), _Employment and social developments in Europe 2023_ , https://data.europa.eu/doi/10.2767/089698.\n\nStatLink 2 https://stat.link/jf50p8\n\n43\n\n22\n\n5\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n35\n\n40\n\n45\n\n50\n\nCleaners Cooks and\n\nwaiters\n\nPersonal care\n\nworkers\n\nSalespeople Building workers\n\nTruck and bus\n\ndrivers\n\nMetal workers EU average Electronic workers\n\nNurses Software developers\n\nDoctors\n\n% of workers in shortage occupations reporting difficulty in making ends meet","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2024_healthatglance_rep_en_p46.pdf"]},"response":{"errors":[],"extract_id":"extract_482f56ecc7066251579ec6f1c1af09ad","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**44** \n\nHEALTH AT A GLANCE: EUROPE 2024 © OECD OECD/EUROPEAN UNION 2024\n\n**_1\\.6.3. Improving the remuneration of some categories of health workers can increase attractiveness_**\n\n**_and retention_**\n\nAn important aspect of job quality relates to earnings, as explicitly recognised under the OECD Job Quality Framework. Based on standard economic theory, wage adjustments should address any imbalance in the supply and demand for different categories of health workers: if there are shortages of some categories of workers, wages should go up to increase supply (either in terms of the number of health workers and/or their working hours) while the demand should decrease, thereby reducing the shortage. However, this supposes that wages are free to adjust to labour market conditions and that the supply of health workers is fairly responsive (or “elastic”) to wage changes. In practice, both of these conditions often do not apply in the health sector. ⁵\n\nThe remuneration of health workers varies widely by categories of workers depending on qualification levels, years of experience (seniority), working time (or activity rates) and negotiating power. In general across EU countries, the remuneration of doctors is several times higher than the average wage of workers in all occupations reflecting their higher qualifications and longer working hours. GPs earn two to four times more than the average wage, while specialists earn two to five times more (see indicator on _“_ Remuneration of doctors” in Chapter 7). The remuneration of nurses is substantially lower than that of doctors. On average across EU countries, it is about 20% higher than the average wage of workers in all occupations, although in some countries nurses do not earn more than the average wage (see indicator on “Remuneration of nurses” in Chapter 7). The remuneration of personal care workers is about 30% lower than the economy-wide average wage on average across EU countries (OECD, 2023[36]).\n\nBased on data collected through the Eurofound’s EWCTS survey in 2021, over 40% of personal care workers on average across EU countries reported having financial difficulties (almost the same proportion as cleaners and cooks and waiters). Over 20% of nurses also reported having difficulties in making ends meet, slightly lower than the average across all occupations considered to be in shortage, but nonetheless a significant proportion. Only about 5% of doctors reported facing such financial difficulties on average across EU countries (Figure 1.19).\n\nWorkers in certain health occupations also believe that they were not fairly rewarded. This is notably the case for nurses: only 40% of nurses felt in 2021 that they were paid fairly in relation to their efforts and achievements on average across EU countries (19 percentage points less than the average across all occupations).\n\n**Figure 1.19. Almost a quarter of nurses and half of personal care workers report financial difficulties across EU countries**\n\nSource: European Commission (2023[18]), _Employment and social developments in Europe 2023_ , https://data.europa.eu/doi/10.2767/089698.\n\nStatLink 2 https://stat.link/jf50p8\n\n43\n\n22\n\n5\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n35\n\n40\n\n45\n\n50\n\nCleaners Cooks and\n\nwaiters\n\nPersonal care\n\nworkers\n\nSalespeople Building workers\n\nTruck and bus\n\ndrivers\n\nMetal workers EU average Electronic workers\n\nNurses Software developers\n\nDoctors\n\n% of workers in shortage occupations reporting difficulty in making ends meet"],"full_content":"**44** \n\nHEALTH AT A GLANCE: EUROPE 2024 © OECD OECD/EUROPEAN UNION 2024\n\n**_1\\.6.3. Improving the remuneration of some categories of health workers can increase attractiveness_**\n\n**_and retention_**\n\nAn important aspect of job quality relates to earnings, as explicitly recognised under the OECD Job Quality Framework. Based on standard economic theory, wage adjustments should address any imbalance in the supply and demand for different categories of health workers: if there are shortages of some categories of workers, wages should go up to increase supply (either in terms of the number of health workers and/or their working hours) while the demand should decrease, thereby reducing the shortage. However, this supposes that wages are free to adjust to labour market conditions and that the supply of health workers is fairly responsive (or “elastic”) to wage changes. In practice, both of these conditions often do not apply in the health sector. ⁵\n\nThe remuneration of health workers varies widely by categories of workers depending on qualification levels, years of experience (seniority), working time (or activity rates) and negotiating power. In general across EU countries, the remuneration of doctors is several times higher than the average wage of workers in all occupations reflecting their higher qualifications and longer working hours. GPs earn two to four times more than the average wage, while specialists earn two to five times more (see indicator on _“_ Remuneration of doctors” in Chapter 7). The remuneration of nurses is substantially lower than that of doctors. On average across EU countries, it is about 20% higher than the average wage of workers in all occupations, although in some countries nurses do not earn more than the average wage (see indicator on “Remuneration of nurses” in Chapter 7). The remuneration of personal care workers is about 30% lower than the economy-wide average wage on average across EU countries (OECD, 2023[36]).\n\nBased on data collected through the Eurofound’s EWCTS survey in 2021, over 40% of personal care workers on average across EU countries reported having financial difficulties (almost the same proportion as cleaners and cooks and waiters). Over 20% of nurses also reported having difficulties in making ends meet, slightly lower than the average across all occupations considered to be in shortage, but nonetheless a significant proportion. Only about 5% of doctors reported facing such financial difficulties on average across EU countries (Figure 1.19).\n\nWorkers in certain health occupations also believe that they were not fairly rewarded. This is notably the case for nurses: only 40% of nurses felt in 2021 that they were paid fairly in relation to their efforts and achievements on average across EU countries (19 percentage points less than the average across all occupations).\n\n**Figure 1.19. Almost a quarter of nurses and half of personal care workers report financial difficulties across EU countries**\n\nSource: European Commission (2023[18]), _Employment and social developments in Europe 2023_ , https://data.europa.eu/doi/10.2767/089698.\n\nStatLink 2 https://stat.link/jf50p8\n\n43\n\n22\n\n5\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n35\n\n40\n\n45\n\n50\n\nCleaners Cooks and\n\nwaiters\n\nPersonal care\n\nworkers\n\nSalespeople Building workers\n\nTruck and bus\n\ndrivers\n\nMetal workers EU average Electronic workers\n\nNurses Software developers\n\nDoctors\n\n% of workers in shortage occupations reporting difficulty in making ends meet","publish_date":null,"title":"Figure 1.19. Almost a quarter of nurses and half of personal care workers report financial difficulties across EU countries","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2024_healthatglance_rep_en_p46.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2024_healthatglance_rep_en_p46.pdf"},"request":{"example_id":"chart/2024_healthatglance_rep_en_p46","product_type":"parse","source_file_path":"2024_healthatglance_rep_en_p46.pdf"},"started_at":"2026-11-04T19:45:11.717765"} +{"completed_at":"2026-11-04T19:45:14.919252","latency_in_ms":1466,"output":{"example_id":"chart/2024_healthatglance_rep_en_p77","grounded_pages":[],"layout_pages":[],"markdown":" **75**\n\nHEALTH AT A GLANCE: EUROPE 2024 © OECD/EUROPEAN UNION 2024\n\n**Table 2.2. Overview of behavioural and environmental risk factors among older people compared to younger people**\n\nPeople aged 18-64 People aged 65 or over\n\n**Behavioural risk factors** Physical inactivity (% reporting to spend less than 150 minutes per week) 65% 78%\n\nPhysical inactivity (% reporting to do physical activity less than once a week) 43% 52%\n\nObesity rate 15% 18%\n\nNutrition (% not eating any vegetable or fruit a day) 35% 25%\n\nSmoking rate (% smoking daily) 22% 9%\n\nAlcohol consumption (% reporting heavy drinking at least once a month) 21% 11%\n\n**Environmental risk factors (mortality)** Air pollution (% of deaths attributable to air pollution) 3% (people aged < 70) 4% (people aged 70+)\n\nExtreme temperature (% deaths attributable to heat or cold wave) 2% (people aged < 70) 4% (people aged 70+)\n\nNote: Data refer to 2019 for physical activity (duration), nutrition, smoking, alcohol consumption; 2021 for air pollution and extreme temperature;2022 for obesity and physical inactivity (frequency). Source: Eurostat (hlth\\_ehis\\_pe2e, ilc\\_hch07b, ilc\\_hch10, hlth\\_ehis\\_al3e, hlth\\_ehis\\_sk3e, hlth\\_ehis\\_fv3e) and IHME GBD for environmental risks.\n\n**_2\\.5.2. Physical activity decreases sharply with age, despite being crucial for mitigating the negative_**\n\n**_effects of ageing_**\n\nPhysical activity is one of the most important activities that people can do to avoid or slowdown the negative consequences of ageing. It reduces the risks of many chronic conditions and falls, promotes mental health, and reduces cognitive functioning decline. WHO released in 2020 new guidelines on physical activity and sedentary behaviour for different age groups, including people aged over 65. These guidelines recommend at least 150-300 minutes of moderate intensity physical activity or 75-150 minutes of vigorous intensity physical activity per week for older adults. As part of their weekly physical activity, it is also recommended that older adults do varied physical activity that emphasises functional balance at least three days per week. Older adults should also limit the amount of time spent being sedentary (WHO, 2020[14]).\n\nThe share of people meeting the WHO recommendation on physical activity declines with age in all EU countries (Figure 2.17). On average across the EU, it drops from 35% among those aged 18 to 64 years to 22% among those aged over 65. In 11 EU countries, less than 10% of people aged over 65 meet these guidelines. Only in Norway, the Netherlands and Sweden did at least half of the older population meet the WHO recommendations. This highlights the need for policies to promote appropriate physical activity among older adults, but also among young adults, as the shares of people reporting sufficient physical activity are low across all age groups.\n\n**Figure 2.17. Only 22% of people aged over 65 meet the WHO recommendation on minimum physical activity per week**\n\nNote: The EU average is weighted. Source: Eurostat (hlth\\_ehis\\_pe2e) (the data refer to 2019).\n\nStatLink 2 https://stat.link/79bcw8\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\n70\n\n80\n\nFrom 18 to 64 years 65 years or over\n\n% of people reporting to spend 150 minutes or more per week of physical activity","pages":[{"line_numbers":[],"markdown":" **75**\n\nHEALTH AT A GLANCE: EUROPE 2024 © OECD/EUROPEAN UNION 2024\n\n**Table 2.2. Overview of behavioural and environmental risk factors among older people compared to younger people**\n\nPeople aged 18-64 People aged 65 or over\n\n**Behavioural risk factors** Physical inactivity (% reporting to spend less than 150 minutes per week) 65% 78%\n\nPhysical inactivity (% reporting to do physical activity less than once a week) 43% 52%\n\nObesity rate 15% 18%\n\nNutrition (% not eating any vegetable or fruit a day) 35% 25%\n\nSmoking rate (% smoking daily) 22% 9%\n\nAlcohol consumption (% reporting heavy drinking at least once a month) 21% 11%\n\n**Environmental risk factors (mortality)** Air pollution (% of deaths attributable to air pollution) 3% (people aged < 70) 4% (people aged 70+)\n\nExtreme temperature (% deaths attributable to heat or cold wave) 2% (people aged < 70) 4% (people aged 70+)\n\nNote: Data refer to 2019 for physical activity (duration), nutrition, smoking, alcohol consumption; 2021 for air pollution and extreme temperature;2022 for obesity and physical inactivity (frequency). Source: Eurostat (hlth\\_ehis\\_pe2e, ilc\\_hch07b, ilc\\_hch10, hlth\\_ehis\\_al3e, hlth\\_ehis\\_sk3e, hlth\\_ehis\\_fv3e) and IHME GBD for environmental risks.\n\n**_2\\.5.2. Physical activity decreases sharply with age, despite being crucial for mitigating the negative_**\n\n**_effects of ageing_**\n\nPhysical activity is one of the most important activities that people can do to avoid or slowdown the negative consequences of ageing. It reduces the risks of many chronic conditions and falls, promotes mental health, and reduces cognitive functioning decline. WHO released in 2020 new guidelines on physical activity and sedentary behaviour for different age groups, including people aged over 65. These guidelines recommend at least 150-300 minutes of moderate intensity physical activity or 75-150 minutes of vigorous intensity physical activity per week for older adults. As part of their weekly physical activity, it is also recommended that older adults do varied physical activity that emphasises functional balance at least three days per week. Older adults should also limit the amount of time spent being sedentary (WHO, 2020[14]).\n\nThe share of people meeting the WHO recommendation on physical activity declines with age in all EU countries (Figure 2.17). On average across the EU, it drops from 35% among those aged 18 to 64 years to 22% among those aged over 65. In 11 EU countries, less than 10% of people aged over 65 meet these guidelines. Only in Norway, the Netherlands and Sweden did at least half of the older population meet the WHO recommendations. This highlights the need for policies to promote appropriate physical activity among older adults, but also among young adults, as the shares of people reporting sufficient physical activity are low across all age groups.\n\n**Figure 2.17. Only 22% of people aged over 65 meet the WHO recommendation on minimum physical activity per week**\n\nNote: The EU average is weighted. Source: Eurostat (hlth\\_ehis\\_pe2e) (the data refer to 2019).\n\nStatLink 2 https://stat.link/79bcw8\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\n70\n\n80\n\nFrom 18 to 64 years 65 years or over\n\n% of people reporting to spend 150 minutes or more per week of physical activity","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":" **75**\n\nHEALTH AT A GLANCE: EUROPE 2024 © OECD/EUROPEAN UNION 2024\n\n**Table 2.2. Overview of behavioural and environmental risk factors among older people compared to younger people**\n\nPeople aged 18-64 People aged 65 or over\n\n**Behavioural risk factors** Physical inactivity (% reporting to spend less than 150 minutes per week) 65% 78%\n\nPhysical inactivity (% reporting to do physical activity less than once a week) 43% 52%\n\nObesity rate 15% 18%\n\nNutrition (% not eating any vegetable or fruit a day) 35% 25%\n\nSmoking rate (% smoking daily) 22% 9%\n\nAlcohol consumption (% reporting heavy drinking at least once a month) 21% 11%\n\n**Environmental risk factors (mortality)** Air pollution (% of deaths attributable to air pollution) 3% (people aged < 70) 4% (people aged 70+)\n\nExtreme temperature (% deaths attributable to heat or cold wave) 2% (people aged < 70) 4% (people aged 70+)\n\nNote: Data refer to 2019 for physical activity (duration), nutrition, smoking, alcohol consumption; 2021 for air pollution and extreme temperature;2022 for obesity and physical inactivity (frequency). Source: Eurostat (hlth\\_ehis\\_pe2e, ilc\\_hch07b, ilc\\_hch10, hlth\\_ehis\\_al3e, hlth\\_ehis\\_sk3e, hlth\\_ehis\\_fv3e) and IHME GBD for environmental risks.\n\n**_2\\.5.2. Physical activity decreases sharply with age, despite being crucial for mitigating the negative_**\n\n**_effects of ageing_**\n\nPhysical activity is one of the most important activities that people can do to avoid or slowdown the negative consequences of ageing. It reduces the risks of many chronic conditions and falls, promotes mental health, and reduces cognitive functioning decline. WHO released in 2020 new guidelines on physical activity and sedentary behaviour for different age groups, including people aged over 65. These guidelines recommend at least 150-300 minutes of moderate intensity physical activity or 75-150 minutes of vigorous intensity physical activity per week for older adults. As part of their weekly physical activity, it is also recommended that older adults do varied physical activity that emphasises functional balance at least three days per week. Older adults should also limit the amount of time spent being sedentary (WHO, 2020[14]).\n\nThe share of people meeting the WHO recommendation on physical activity declines with age in all EU countries (Figure 2.17). On average across the EU, it drops from 35% among those aged 18 to 64 years to 22% among those aged over 65. In 11 EU countries, less than 10% of people aged over 65 meet these guidelines. Only in Norway, the Netherlands and Sweden did at least half of the older population meet the WHO recommendations. This highlights the need for policies to promote appropriate physical activity among older adults, but also among young adults, as the shares of people reporting sufficient physical activity are low across all age groups.\n\n**Figure 2.17. Only 22% of people aged over 65 meet the WHO recommendation on minimum physical activity per week**\n\nNote: The EU average is weighted. Source: Eurostat (hlth\\_ehis\\_pe2e) (the data refer to 2019).\n\nStatLink 2 https://stat.link/79bcw8\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\n70\n\n80\n\nFrom 18 to 64 years 65 years or over\n\n% of people reporting to spend 150 minutes or more per week of physical activity","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2024_healthatglance_rep_en_p77.pdf"]},"response":{"errors":[],"extract_id":"extract_21e6860c2234727d5f47629963e588f3","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":[" **75**\n\nHEALTH AT A GLANCE: EUROPE 2024 © OECD/EUROPEAN UNION 2024\n\n**Table 2.2. Overview of behavioural and environmental risk factors among older people compared to younger people**\n\nPeople aged 18-64 People aged 65 or over\n\n**Behavioural risk factors** Physical inactivity (% reporting to spend less than 150 minutes per week) 65% 78%\n\nPhysical inactivity (% reporting to do physical activity less than once a week) 43% 52%\n\nObesity rate 15% 18%\n\nNutrition (% not eating any vegetable or fruit a day) 35% 25%\n\nSmoking rate (% smoking daily) 22% 9%\n\nAlcohol consumption (% reporting heavy drinking at least once a month) 21% 11%\n\n**Environmental risk factors (mortality)** Air pollution (% of deaths attributable to air pollution) 3% (people aged < 70) 4% (people aged 70+)\n\nExtreme temperature (% deaths attributable to heat or cold wave) 2% (people aged < 70) 4% (people aged 70+)\n\nNote: Data refer to 2019 for physical activity (duration), nutrition, smoking, alcohol consumption; 2021 for air pollution and extreme temperature;2022 for obesity and physical inactivity (frequency). Source: Eurostat (hlth\\_ehis\\_pe2e, ilc\\_hch07b, ilc\\_hch10, hlth\\_ehis\\_al3e, hlth\\_ehis\\_sk3e, hlth\\_ehis\\_fv3e) and IHME GBD for environmental risks.\n\n**_2\\.5.2. Physical activity decreases sharply with age, despite being crucial for mitigating the negative_**\n\n**_effects of ageing_**\n\nPhysical activity is one of the most important activities that people can do to avoid or slowdown the negative consequences of ageing. It reduces the risks of many chronic conditions and falls, promotes mental health, and reduces cognitive functioning decline. WHO released in 2020 new guidelines on physical activity and sedentary behaviour for different age groups, including people aged over 65. These guidelines recommend at least 150-300 minutes of moderate intensity physical activity or 75-150 minutes of vigorous intensity physical activity per week for older adults. As part of their weekly physical activity, it is also recommended that older adults do varied physical activity that emphasises functional balance at least three days per week. Older adults should also limit the amount of time spent being sedentary (WHO, 2020[14]).\n\nThe share of people meeting the WHO recommendation on physical activity declines with age in all EU countries (Figure 2.17). On average across the EU, it drops from 35% among those aged 18 to 64 years to 22% among those aged over 65. In 11 EU countries, less than 10% of people aged over 65 meet these guidelines. Only in Norway, the Netherlands and Sweden did at least half of the older population meet the WHO recommendations. This highlights the need for policies to promote appropriate physical activity among older adults, but also among young adults, as the shares of people reporting sufficient physical activity are low across all age groups.\n\n**Figure 2.17. Only 22% of people aged over 65 meet the WHO recommendation on minimum physical activity per week**\n\nNote: The EU average is weighted. Source: Eurostat (hlth\\_ehis\\_pe2e) (the data refer to 2019).\n\nStatLink 2 https://stat.link/79bcw8\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\n70\n\n80\n\nFrom 18 to 64 years 65 years or over\n\n% of people reporting to spend 150 minutes or more per week of physical activity"],"full_content":" **75**\n\nHEALTH AT A GLANCE: EUROPE 2024 © OECD/EUROPEAN UNION 2024\n\n**Table 2.2. Overview of behavioural and environmental risk factors among older people compared to younger people**\n\nPeople aged 18-64 People aged 65 or over\n\n**Behavioural risk factors** Physical inactivity (% reporting to spend less than 150 minutes per week) 65% 78%\n\nPhysical inactivity (% reporting to do physical activity less than once a week) 43% 52%\n\nObesity rate 15% 18%\n\nNutrition (% not eating any vegetable or fruit a day) 35% 25%\n\nSmoking rate (% smoking daily) 22% 9%\n\nAlcohol consumption (% reporting heavy drinking at least once a month) 21% 11%\n\n**Environmental risk factors (mortality)** Air pollution (% of deaths attributable to air pollution) 3% (people aged < 70) 4% (people aged 70+)\n\nExtreme temperature (% deaths attributable to heat or cold wave) 2% (people aged < 70) 4% (people aged 70+)\n\nNote: Data refer to 2019 for physical activity (duration), nutrition, smoking, alcohol consumption; 2021 for air pollution and extreme temperature;2022 for obesity and physical inactivity (frequency). Source: Eurostat (hlth\\_ehis\\_pe2e, ilc\\_hch07b, ilc\\_hch10, hlth\\_ehis\\_al3e, hlth\\_ehis\\_sk3e, hlth\\_ehis\\_fv3e) and IHME GBD for environmental risks.\n\n**_2\\.5.2. Physical activity decreases sharply with age, despite being crucial for mitigating the negative_**\n\n**_effects of ageing_**\n\nPhysical activity is one of the most important activities that people can do to avoid or slowdown the negative consequences of ageing. It reduces the risks of many chronic conditions and falls, promotes mental health, and reduces cognitive functioning decline. WHO released in 2020 new guidelines on physical activity and sedentary behaviour for different age groups, including people aged over 65. These guidelines recommend at least 150-300 minutes of moderate intensity physical activity or 75-150 minutes of vigorous intensity physical activity per week for older adults. As part of their weekly physical activity, it is also recommended that older adults do varied physical activity that emphasises functional balance at least three days per week. Older adults should also limit the amount of time spent being sedentary (WHO, 2020[14]).\n\nThe share of people meeting the WHO recommendation on physical activity declines with age in all EU countries (Figure 2.17). On average across the EU, it drops from 35% among those aged 18 to 64 years to 22% among those aged over 65. In 11 EU countries, less than 10% of people aged over 65 meet these guidelines. Only in Norway, the Netherlands and Sweden did at least half of the older population meet the WHO recommendations. This highlights the need for policies to promote appropriate physical activity among older adults, but also among young adults, as the shares of people reporting sufficient physical activity are low across all age groups.\n\n**Figure 2.17. Only 22% of people aged over 65 meet the WHO recommendation on minimum physical activity per week**\n\nNote: The EU average is weighted. Source: Eurostat (hlth\\_ehis\\_pe2e) (the data refer to 2019).\n\nStatLink 2 https://stat.link/79bcw8\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\n70\n\n80\n\nFrom 18 to 64 years 65 years or over\n\n% of people reporting to spend 150 minutes or more per week of physical activity","publish_date":null,"title":"75  Table 2.2. Overview of behavioural and environmental risk factors among older people compared to younger people Figure 2.17.","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2024_healthatglance_rep_en_p77.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2024_healthatglance_rep_en_p77.pdf"},"request":{"example_id":"chart/2024_healthatglance_rep_en_p77","product_type":"parse","source_file_path":"2024_healthatglance_rep_en_p77.pdf"},"started_at":"2026-11-04T19:45:13.452868"} +{"completed_at":"2026-11-04T19:45:15.624955","latency_in_ms":1263,"output":{"example_id":"chart/2025-Business-Megatrends-Outlook_p25","grounded_pages":[],"layout_pages":[],"markdown":"As corporate attention wavers, consumer demand for authentic and responsible environmental action is intensifying. Climate anxiety is growing rapidly, especially in regions that are increasingly affected by extreme weather events. Sixty-one percent of consumers say they are more concerned about climate change than they were just two years ago, often because of direct personal experiences ²⁶ .\n\n## The Future of ESG: Trust, Technology, and Consumer Expectations\n\n## 6\\.3.\n\nSource : Bain & Company, The Visionary CEO’s Guide to Sustainability, 2024. ²⁶\n\n**Figure 1:** When asked, most consumers say their concerns about climate change have increased in the past two years **Q : How have your climate change concerns changed over the past two years?**\n\nPercentage of respondent 100%\n\n0 **Overall** US UK Nethelands Germany France Italy Japan Brazil China Indonesia\n\nMy concerns have intensifed/increased My concerns have not changed My concerns have subsied/decreased\n\n61 **%** 51 **%** 52 **%** 50 **%** 51 **%** 59 **%** 70 **%** 62 **%** 78 **%** 62 **%** 73 **%**\n\n31 **%** 42 **%** 40 **%** 39 **%** 37 **%** 33 **%** 24 **%** 34 **%** 17 **%**\n\n28 **%** 19 **%**\n\n8 **%** 8 **%** 8 **%** 7 **%** 6 **%** 4 **%** 4 **%** 8 **%** 11 **%** 12 **%** 10 **%**\n\nIn parallel, public trust is shifting: smaller businesses are now seen as more genuine and committed to sustainable practices than large corporations. This perception is beginning to influence behavior across sectors— for example, according to a Bain & Company survey conducted with 500 B2B buyers, 36% would stop working with suppliers that fail to meet their sustainability expectations ²⁷ . ESG is no longer just about meeting regulations; it has become a key driver of trust among increasingly skeptical and climate-conscious stakeholders.\n\n26 Bain & Company, The Visionary CEO’s Guide to Sustainability, 2024, https://www.bain.com/globalassets/noindex/2024/bain\\_report\\_the\\_visionary\\_ceos\\_ guide\\_to\\_sustainability-2024.pdf. 27Bain & Company, The Visionary CEO’s Guide to Sustainability, 2024, https://www.bain.com/globalassets/noindex/2024/bain\\_report\\_the\\_visionary\\_ceos\\_ guide\\_to\\_sustainability-2024.pdf\n\nNextcontinent\n\nnextcontinent.net\n\n2025 Business Outlook: Challenges & Opportunities for Consulting Firms\n\n**25**","pages":[{"line_numbers":[],"markdown":"As corporate attention wavers, consumer demand for authentic and responsible environmental action is intensifying. Climate anxiety is growing rapidly, especially in regions that are increasingly affected by extreme weather events. Sixty-one percent of consumers say they are more concerned about climate change than they were just two years ago, often because of direct personal experiences ²⁶ .\n\n## The Future of ESG: Trust, Technology, and Consumer Expectations\n\n## 6\\.3.\n\nSource : Bain & Company, The Visionary CEO’s Guide to Sustainability, 2024. ²⁶\n\n**Figure 1:** When asked, most consumers say their concerns about climate change have increased in the past two years **Q : How have your climate change concerns changed over the past two years?**\n\nPercentage of respondent 100%\n\n0 **Overall** US UK Nethelands Germany France Italy Japan Brazil China Indonesia\n\nMy concerns have intensifed/increased My concerns have not changed My concerns have subsied/decreased\n\n61 **%** 51 **%** 52 **%** 50 **%** 51 **%** 59 **%** 70 **%** 62 **%** 78 **%** 62 **%** 73 **%**\n\n31 **%** 42 **%** 40 **%** 39 **%** 37 **%** 33 **%** 24 **%** 34 **%** 17 **%**\n\n28 **%** 19 **%**\n\n8 **%** 8 **%** 8 **%** 7 **%** 6 **%** 4 **%** 4 **%** 8 **%** 11 **%** 12 **%** 10 **%**\n\nIn parallel, public trust is shifting: smaller businesses are now seen as more genuine and committed to sustainable practices than large corporations. This perception is beginning to influence behavior across sectors— for example, according to a Bain & Company survey conducted with 500 B2B buyers, 36% would stop working with suppliers that fail to meet their sustainability expectations ²⁷ . ESG is no longer just about meeting regulations; it has become a key driver of trust among increasingly skeptical and climate-conscious stakeholders.\n\n26 Bain & Company, The Visionary CEO’s Guide to Sustainability, 2024, https://www.bain.com/globalassets/noindex/2024/bain\\_report\\_the\\_visionary\\_ceos\\_ guide\\_to\\_sustainability-2024.pdf. 27Bain & Company, The Visionary CEO’s Guide to Sustainability, 2024, https://www.bain.com/globalassets/noindex/2024/bain\\_report\\_the\\_visionary\\_ceos\\_ guide\\_to\\_sustainability-2024.pdf\n\nNextcontinent\n\nnextcontinent.net\n\n2025 Business Outlook: Challenges & Opportunities for Consulting Firms\n\n**25**","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"As corporate attention wavers, consumer demand for authentic and responsible environmental action is intensifying. Climate anxiety is growing rapidly, especially in regions that are increasingly affected by extreme weather events. Sixty-one percent of consumers say they are more concerned about climate change than they were just two years ago, often because of direct personal experiences ²⁶ .\n\n## The Future of ESG: Trust, Technology, and Consumer Expectations\n\n## 6\\.3.\n\nSource : Bain & Company, The Visionary CEO’s Guide to Sustainability, 2024. ²⁶\n\n**Figure 1:** When asked, most consumers say their concerns about climate change have increased in the past two years **Q : How have your climate change concerns changed over the past two years?**\n\nPercentage of respondent 100%\n\n0 **Overall** US UK Nethelands Germany France Italy Japan Brazil China Indonesia\n\nMy concerns have intensifed/increased My concerns have not changed My concerns have subsied/decreased\n\n61 **%** 51 **%** 52 **%** 50 **%** 51 **%** 59 **%** 70 **%** 62 **%** 78 **%** 62 **%** 73 **%**\n\n31 **%** 42 **%** 40 **%** 39 **%** 37 **%** 33 **%** 24 **%** 34 **%** 17 **%**\n\n28 **%** 19 **%**\n\n8 **%** 8 **%** 8 **%** 7 **%** 6 **%** 4 **%** 4 **%** 8 **%** 11 **%** 12 **%** 10 **%**\n\nIn parallel, public trust is shifting: smaller businesses are now seen as more genuine and committed to sustainable practices than large corporations. This perception is beginning to influence behavior across sectors— for example, according to a Bain & Company survey conducted with 500 B2B buyers, 36% would stop working with suppliers that fail to meet their sustainability expectations ²⁷ . ESG is no longer just about meeting regulations; it has become a key driver of trust among increasingly skeptical and climate-conscious stakeholders.\n\n26 Bain & Company, The Visionary CEO’s Guide to Sustainability, 2024, https://www.bain.com/globalassets/noindex/2024/bain\\_report\\_the\\_visionary\\_ceos\\_ guide\\_to\\_sustainability-2024.pdf. 27Bain & Company, The Visionary CEO’s Guide to Sustainability, 2024, https://www.bain.com/globalassets/noindex/2024/bain\\_report\\_the\\_visionary\\_ceos\\_ guide\\_to\\_sustainability-2024.pdf\n\nNextcontinent\n\nnextcontinent.net\n\n2025 Business Outlook: Challenges & Opportunities for Consulting Firms\n\n**25**","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2025-Business-Megatrends-Outlook_p25.pdf"]},"response":{"errors":[],"extract_id":"extract_8b7a5f14d78e526b5ae25ecbe3fc77af","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["As corporate attention wavers, consumer demand for authentic and responsible environmental action is intensifying. Climate anxiety is growing rapidly, especially in regions that are increasingly affected by extreme weather events. Sixty-one percent of consumers say they are more concerned about climate change than they were just two years ago, often because of direct personal experiences ²⁶ .\n\n## The Future of ESG: Trust, Technology, and Consumer Expectations\n\n## 6\\.3.\n\nSource : Bain & Company, The Visionary CEO’s Guide to Sustainability, 2024. ²⁶\n\n**Figure 1:** When asked, most consumers say their concerns about climate change have increased in the past two years **Q : How have your climate change concerns changed over the past two years?**\n\nPercentage of respondent 100%\n\n0 **Overall** US UK Nethelands Germany France Italy Japan Brazil China Indonesia\n\nMy concerns have intensifed/increased My concerns have not changed My concerns have subsied/decreased\n\n61 **%** 51 **%** 52 **%** 50 **%** 51 **%** 59 **%** 70 **%** 62 **%** 78 **%** 62 **%** 73 **%**\n\n31 **%** 42 **%** 40 **%** 39 **%** 37 **%** 33 **%** 24 **%** 34 **%** 17 **%**\n\n28 **%** 19 **%**\n\n8 **%** 8 **%** 8 **%** 7 **%** 6 **%** 4 **%** 4 **%** 8 **%** 11 **%** 12 **%** 10 **%**\n\nIn parallel, public trust is shifting: smaller businesses are now seen as more genuine and committed to sustainable practices than large corporations. This perception is beginning to influence behavior across sectors— for example, according to a Bain & Company survey conducted with 500 B2B buyers, 36% would stop working with suppliers that fail to meet their sustainability expectations ²⁷ . ESG is no longer just about meeting regulations; it has become a key driver of trust among increasingly skeptical and climate-conscious stakeholders.\n\n26 Bain & Company, The Visionary CEO’s Guide to Sustainability, 2024, https://www.bain.com/globalassets/noindex/2024/bain\\_report\\_the\\_visionary\\_ceos\\_ guide\\_to\\_sustainability-2024.pdf. 27Bain & Company, The Visionary CEO’s Guide to Sustainability, 2024, https://www.bain.com/globalassets/noindex/2024/bain\\_report\\_the\\_visionary\\_ceos\\_ guide\\_to\\_sustainability-2024.pdf\n\nNextcontinent\n\nnextcontinent.net\n\n2025 Business Outlook: Challenges & Opportunities for Consulting Firms\n\n**25**"],"full_content":"As corporate attention wavers, consumer demand for authentic and responsible environmental action is intensifying. Climate anxiety is growing rapidly, especially in regions that are increasingly affected by extreme weather events. Sixty-one percent of consumers say they are more concerned about climate change than they were just two years ago, often because of direct personal experiences ²⁶ .\n\n## The Future of ESG: Trust, Technology, and Consumer Expectations\n\n## 6\\.3.\n\nSource : Bain & Company, The Visionary CEO’s Guide to Sustainability, 2024. ²⁶\n\n**Figure 1:** When asked, most consumers say their concerns about climate change have increased in the past two years **Q : How have your climate change concerns changed over the past two years?**\n\nPercentage of respondent 100%\n\n0 **Overall** US UK Nethelands Germany France Italy Japan Brazil China Indonesia\n\nMy concerns have intensifed/increased My concerns have not changed My concerns have subsied/decreased\n\n61 **%** 51 **%** 52 **%** 50 **%** 51 **%** 59 **%** 70 **%** 62 **%** 78 **%** 62 **%** 73 **%**\n\n31 **%** 42 **%** 40 **%** 39 **%** 37 **%** 33 **%** 24 **%** 34 **%** 17 **%**\n\n28 **%** 19 **%**\n\n8 **%** 8 **%** 8 **%** 7 **%** 6 **%** 4 **%** 4 **%** 8 **%** 11 **%** 12 **%** 10 **%**\n\nIn parallel, public trust is shifting: smaller businesses are now seen as more genuine and committed to sustainable practices than large corporations. This perception is beginning to influence behavior across sectors— for example, according to a Bain & Company survey conducted with 500 B2B buyers, 36% would stop working with suppliers that fail to meet their sustainability expectations ²⁷ . ESG is no longer just about meeting regulations; it has become a key driver of trust among increasingly skeptical and climate-conscious stakeholders.\n\n26 Bain & Company, The Visionary CEO’s Guide to Sustainability, 2024, https://www.bain.com/globalassets/noindex/2024/bain\\_report\\_the\\_visionary\\_ceos\\_ guide\\_to\\_sustainability-2024.pdf. 27Bain & Company, The Visionary CEO’s Guide to Sustainability, 2024, https://www.bain.com/globalassets/noindex/2024/bain\\_report\\_the\\_visionary\\_ceos\\_ guide\\_to\\_sustainability-2024.pdf\n\nNextcontinent\n\nnextcontinent.net\n\n2025 Business Outlook: Challenges & Opportunities for Consulting Firms\n\n**25**","publish_date":null,"title":"6.3. The Future of ESG: Trust, Technology, and Consumer Expectations","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2025-Business-Megatrends-Outlook_p25.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2025-Business-Megatrends-Outlook_p25.pdf"},"request":{"example_id":"chart/2025-Business-Megatrends-Outlook_p25","product_type":"parse","source_file_path":"2025-Business-Megatrends-Outlook_p25.pdf"},"started_at":"2026-11-04T19:45:14.361117"} +{"completed_at":"2026-11-04T19:45:16.920363","latency_in_ms":1293,"output":{"example_id":"chart/2025-EIS_p108","grounded_pages":[],"layout_pages":[],"markdown":"106 European Innovation Scoreboard 2025\n\nFigure 42: Shares and concentration of non-EU imports in EU and neighbouring countries.\n\n_Source: Eurostat FIGARO_\n\nI I I I I I I I I I I I I I I I I I I I\n\nI I I I I I I I I I\n\n0\\.1 0.2 0.3 0.4 0.5 0.6\n\nCroatia\n\nLithuania\n\nRomania\n\nFrance\n\nDenmark\n\nItaly\n\nPortugal\n\nSweden\n\nBelgium\n\nBulgaria\n\nUnited Kingdom\n\nLatvia\n\nAustria\n\nFinland\n\nGreece\n\nLuxembourg\n\nSpain\n\nNorway\n\nGermany\n\nMalta\n\nNetherlands\n\nPoland\n\nHungary\n\nEstonia\n\nCzechia\n\nSlovakia\n\nSwitzerland\n\nTürkiye\n\nSlovenia\n\nCyprus\n\nIreland\n\n0 10 20 30 40 50 60\n\nHHI of non-EU high-tech imports of GOODS\n\n% of total high-tech use\n\nI HHI of non-EU high-tech imports of GOODS\n\nNon-EU high-tech imports of SERVICES as % of total high-tech use\n\nNon-EU high-tech imports of GOODS as % of total high-tech use","pages":[{"line_numbers":[],"markdown":"106 European Innovation Scoreboard 2025\n\nFigure 42: Shares and concentration of non-EU imports in EU and neighbouring countries.\n\n_Source: Eurostat FIGARO_\n\nI I I I I I I I I I I I I I I I I I I I\n\nI I I I I I I I I I\n\n0\\.1 0.2 0.3 0.4 0.5 0.6\n\nCroatia\n\nLithuania\n\nRomania\n\nFrance\n\nDenmark\n\nItaly\n\nPortugal\n\nSweden\n\nBelgium\n\nBulgaria\n\nUnited Kingdom\n\nLatvia\n\nAustria\n\nFinland\n\nGreece\n\nLuxembourg\n\nSpain\n\nNorway\n\nGermany\n\nMalta\n\nNetherlands\n\nPoland\n\nHungary\n\nEstonia\n\nCzechia\n\nSlovakia\n\nSwitzerland\n\nTürkiye\n\nSlovenia\n\nCyprus\n\nIreland\n\n0 10 20 30 40 50 60\n\nHHI of non-EU high-tech imports of GOODS\n\n% of total high-tech use\n\nI HHI of non-EU high-tech imports of GOODS\n\nNon-EU high-tech imports of SERVICES as % of total high-tech use\n\nNon-EU high-tech imports of GOODS as % of total high-tech use","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"106 European Innovation Scoreboard 2025\n\nFigure 42: Shares and concentration of non-EU imports in EU and neighbouring countries.\n\n_Source: Eurostat FIGARO_\n\nI I I I I I I I I I I I I I I I I I I I\n\nI I I I I I I I I I\n\n0\\.1 0.2 0.3 0.4 0.5 0.6\n\nCroatia\n\nLithuania\n\nRomania\n\nFrance\n\nDenmark\n\nItaly\n\nPortugal\n\nSweden\n\nBelgium\n\nBulgaria\n\nUnited Kingdom\n\nLatvia\n\nAustria\n\nFinland\n\nGreece\n\nLuxembourg\n\nSpain\n\nNorway\n\nGermany\n\nMalta\n\nNetherlands\n\nPoland\n\nHungary\n\nEstonia\n\nCzechia\n\nSlovakia\n\nSwitzerland\n\nTürkiye\n\nSlovenia\n\nCyprus\n\nIreland\n\n0 10 20 30 40 50 60\n\nHHI of non-EU high-tech imports of GOODS\n\n% of total high-tech use\n\nI HHI of non-EU high-tech imports of GOODS\n\nNon-EU high-tech imports of SERVICES as % of total high-tech use\n\nNon-EU high-tech imports of GOODS as % of total high-tech use","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2025-EIS_p108.pdf"]},"response":{"errors":[],"extract_id":"extract_3f5fcdc2744ba50c0dc7ea7138428ac1","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["106 European Innovation Scoreboard 2025\n\nFigure 42: Shares and concentration of non-EU imports in EU and neighbouring countries.\n\n_Source: Eurostat FIGARO_\n\nI I I I I I I I I I I I I I I I I I I I\n\nI I I I I I I I I I\n\n0\\.1 0.2 0.3 0.4 0.5 0.6\n\nCroatia\n\nLithuania\n\nRomania\n\nFrance\n\nDenmark\n\nItaly\n\nPortugal\n\nSweden\n\nBelgium\n\nBulgaria\n\nUnited Kingdom\n\nLatvia\n\nAustria\n\nFinland\n\nGreece\n\nLuxembourg\n\nSpain\n\nNorway\n\nGermany\n\nMalta\n\nNetherlands\n\nPoland\n\nHungary\n\nEstonia\n\nCzechia\n\nSlovakia\n\nSwitzerland\n\nTürkiye\n\nSlovenia\n\nCyprus\n\nIreland\n\n0 10 20 30 40 50 60\n\nHHI of non-EU high-tech imports of GOODS\n\n% of total high-tech use\n\nI HHI of non-EU high-tech imports of GOODS\n\nNon-EU high-tech imports of SERVICES as % of total high-tech use\n\nNon-EU high-tech imports of GOODS as % of total high-tech use"],"full_content":"106 European Innovation Scoreboard 2025\n\nFigure 42: Shares and concentration of non-EU imports in EU and neighbouring countries.\n\n_Source: Eurostat FIGARO_\n\nI I I I I I I I I I I I I I I I I I I I\n\nI I I I I I I I I I\n\n0\\.1 0.2 0.3 0.4 0.5 0.6\n\nCroatia\n\nLithuania\n\nRomania\n\nFrance\n\nDenmark\n\nItaly\n\nPortugal\n\nSweden\n\nBelgium\n\nBulgaria\n\nUnited Kingdom\n\nLatvia\n\nAustria\n\nFinland\n\nGreece\n\nLuxembourg\n\nSpain\n\nNorway\n\nGermany\n\nMalta\n\nNetherlands\n\nPoland\n\nHungary\n\nEstonia\n\nCzechia\n\nSlovakia\n\nSwitzerland\n\nTürkiye\n\nSlovenia\n\nCyprus\n\nIreland\n\n0 10 20 30 40 50 60\n\nHHI of non-EU high-tech imports of GOODS\n\n% of total high-tech use\n\nI HHI of non-EU high-tech imports of GOODS\n\nNon-EU high-tech imports of SERVICES as % of total high-tech use\n\nNon-EU high-tech imports of GOODS as % of total high-tech use","publish_date":null,"title":"Figure 42: Shares and concentration of non-EU imports in EU and neighbouring countries. II II II II II II II","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2025-EIS_p108.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2025-EIS_p108.pdf"},"request":{"example_id":"chart/2025-EIS_p108","product_type":"parse","source_file_path":"2025-EIS_p108.pdf"},"started_at":"2026-11-04T19:45:15.626412"} +{"completed_at":"2026-11-04T19:45:18.481992","latency_in_ms":1940,"output":{"example_id":"chart/2025-EIS_p12","grounded_pages":[],"layout_pages":[],"markdown":"10 European Innovation Scoreboard 2025\n\nFigure 2: Innovation performance of the EU27 Member States in 2025, indexed to the EU in 2018\n\nI I I I I I I I I I I I I I I I I I I I I I I I I I I I Romania\n\nBulgaria\n\nLatvia\n\nSlovakia\n\nPoland\n\nHungary\n\nCroatia\n\nGreece\n\nCzechia\n\nLithuania\n\nCyprus\n\nPortugal\n\nSpain\n\nItaly\n\nSlovenia\n\nMalta\n\nEU\n\nEstonia\n\nFrance\n\nGermany\n\nAustria\n\nLuxembourg\n\nBelgium\n\nIreland\n\nFinland\n\nNetherlands\n\nDenmark\n\nSweden\n\n0 20 40 60 80 100 120 140 160\n\nSummary innovation index in 2025 (indexed to the EU in 2018)\n\nEmerging innovators\n\nModerate innovators\n\nStrong innovators\n\nInnovation leaders\n\nI Score in 2024","pages":[{"line_numbers":[],"markdown":"10 European Innovation Scoreboard 2025\n\nFigure 2: Innovation performance of the EU27 Member States in 2025, indexed to the EU in 2018\n\nI I I I I I I I I I I I I I I I I I I I I I I I I I I I Romania\n\nBulgaria\n\nLatvia\n\nSlovakia\n\nPoland\n\nHungary\n\nCroatia\n\nGreece\n\nCzechia\n\nLithuania\n\nCyprus\n\nPortugal\n\nSpain\n\nItaly\n\nSlovenia\n\nMalta\n\nEU\n\nEstonia\n\nFrance\n\nGermany\n\nAustria\n\nLuxembourg\n\nBelgium\n\nIreland\n\nFinland\n\nNetherlands\n\nDenmark\n\nSweden\n\n0 20 40 60 80 100 120 140 160\n\nSummary innovation index in 2025 (indexed to the EU in 2018)\n\nEmerging innovators\n\nModerate innovators\n\nStrong innovators\n\nInnovation leaders\n\nI Score in 2024","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"10 European Innovation Scoreboard 2025\n\nFigure 2: Innovation performance of the EU27 Member States in 2025, indexed to the EU in 2018\n\nI I I I I I I I I I I I I I I I I I I I I I I I I I I I Romania\n\nBulgaria\n\nLatvia\n\nSlovakia\n\nPoland\n\nHungary\n\nCroatia\n\nGreece\n\nCzechia\n\nLithuania\n\nCyprus\n\nPortugal\n\nSpain\n\nItaly\n\nSlovenia\n\nMalta\n\nEU\n\nEstonia\n\nFrance\n\nGermany\n\nAustria\n\nLuxembourg\n\nBelgium\n\nIreland\n\nFinland\n\nNetherlands\n\nDenmark\n\nSweden\n\n0 20 40 60 80 100 120 140 160\n\nSummary innovation index in 2025 (indexed to the EU in 2018)\n\nEmerging innovators\n\nModerate innovators\n\nStrong innovators\n\nInnovation leaders\n\nI Score in 2024","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2025-EIS_p12.pdf"]},"response":{"errors":[],"extract_id":"extract_47dee7d21f24d4c32e60224f8cb0ceff","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["10 European Innovation Scoreboard 2025\n\nFigure 2: Innovation performance of the EU27 Member States in 2025, indexed to the EU in 2018\n\nI I I I I I I I I I I I I I I I I I I I I I I I I I I I Romania\n\nBulgaria\n\nLatvia\n\nSlovakia\n\nPoland\n\nHungary\n\nCroatia\n\nGreece\n\nCzechia\n\nLithuania\n\nCyprus\n\nPortugal\n\nSpain\n\nItaly\n\nSlovenia\n\nMalta\n\nEU\n\nEstonia\n\nFrance\n\nGermany\n\nAustria\n\nLuxembourg\n\nBelgium\n\nIreland\n\nFinland\n\nNetherlands\n\nDenmark\n\nSweden\n\n0 20 40 60 80 100 120 140 160\n\nSummary innovation index in 2025 (indexed to the EU in 2018)\n\nEmerging innovators\n\nModerate innovators\n\nStrong innovators\n\nInnovation leaders\n\nI Score in 2024"],"full_content":"10 European Innovation Scoreboard 2025\n\nFigure 2: Innovation performance of the EU27 Member States in 2025, indexed to the EU in 2018\n\nI I I I I I I I I I I I I I I I I I I I I I I I I I I I Romania\n\nBulgaria\n\nLatvia\n\nSlovakia\n\nPoland\n\nHungary\n\nCroatia\n\nGreece\n\nCzechia\n\nLithuania\n\nCyprus\n\nPortugal\n\nSpain\n\nItaly\n\nSlovenia\n\nMalta\n\nEU\n\nEstonia\n\nFrance\n\nGermany\n\nAustria\n\nLuxembourg\n\nBelgium\n\nIreland\n\nFinland\n\nNetherlands\n\nDenmark\n\nSweden\n\n0 20 40 60 80 100 120 140 160\n\nSummary innovation index in 2025 (indexed to the EU in 2018)\n\nEmerging innovators\n\nModerate innovators\n\nStrong innovators\n\nInnovation leaders\n\nI Score in 2024","publish_date":null,"title":"Figure 2: Innovation performance of the EU27 Member States in 2025, indexed to the EU in 2018 I I I","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2025-EIS_p12.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2025-EIS_p12.pdf"},"request":{"example_id":"chart/2025-EIS_p12","product_type":"parse","source_file_path":"2025-EIS_p12.pdf"},"started_at":"2026-11-04T19:45:16.541386"} +{"completed_at":"2026-11-04T19:45:19.647283","latency_in_ms":1163,"output":{"example_id":"chart/2025-EIS_p33","grounded_pages":[],"layout_pages":[],"markdown":"31 European Innovation Scoreboard 2025\n\nFigure 9: Evolution of the innovation performance of the EU27 Member States between 2018 and 2025\n\n_Note: Performance change is measured as the difference between 2025 and 2018 scores, relative to that of the EU in 2018._\n\nEU average in 2025\n\nRomania\n\nBulgaria\n\nLatvia\n\nSlovakia\n\nPoland\n\nHungary\n\nCroatia\n\nGreece\n\nCzechia\n\nLithuania\n\nCyprus\n\nPortugal\n\nSpain\n\nItaly\n\nSlovenia\n\nMalta\n\nEU\n\nEstonia\n\nFrance\n\nGermany\n\nAustria\n\nLuxembourg\n\nBelgium\n\nIreland\n\nFinland\n\nNetherlands\n\nDenmark\n\nSweden\n\n0 20 40 60 80 100 120 140 160\n\nChange in summary innovation index between 2018 and 2025\n\nPositive change","pages":[{"line_numbers":[],"markdown":"31 European Innovation Scoreboard 2025\n\nFigure 9: Evolution of the innovation performance of the EU27 Member States between 2018 and 2025\n\n_Note: Performance change is measured as the difference between 2025 and 2018 scores, relative to that of the EU in 2018._\n\nEU average in 2025\n\nRomania\n\nBulgaria\n\nLatvia\n\nSlovakia\n\nPoland\n\nHungary\n\nCroatia\n\nGreece\n\nCzechia\n\nLithuania\n\nCyprus\n\nPortugal\n\nSpain\n\nItaly\n\nSlovenia\n\nMalta\n\nEU\n\nEstonia\n\nFrance\n\nGermany\n\nAustria\n\nLuxembourg\n\nBelgium\n\nIreland\n\nFinland\n\nNetherlands\n\nDenmark\n\nSweden\n\n0 20 40 60 80 100 120 140 160\n\nChange in summary innovation index between 2018 and 2025\n\nPositive change","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"31 European Innovation Scoreboard 2025\n\nFigure 9: Evolution of the innovation performance of the EU27 Member States between 2018 and 2025\n\n_Note: Performance change is measured as the difference between 2025 and 2018 scores, relative to that of the EU in 2018._\n\nEU average in 2025\n\nRomania\n\nBulgaria\n\nLatvia\n\nSlovakia\n\nPoland\n\nHungary\n\nCroatia\n\nGreece\n\nCzechia\n\nLithuania\n\nCyprus\n\nPortugal\n\nSpain\n\nItaly\n\nSlovenia\n\nMalta\n\nEU\n\nEstonia\n\nFrance\n\nGermany\n\nAustria\n\nLuxembourg\n\nBelgium\n\nIreland\n\nFinland\n\nNetherlands\n\nDenmark\n\nSweden\n\n0 20 40 60 80 100 120 140 160\n\nChange in summary innovation index between 2018 and 2025\n\nPositive change","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2025-EIS_p33.pdf"]},"response":{"errors":[],"extract_id":"extract_651022412df0690347524a1850ad7f50","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["31 European Innovation Scoreboard 2025\n\nFigure 9: Evolution of the innovation performance of the EU27 Member States between 2018 and 2025\n\n_Note: Performance change is measured as the difference between 2025 and 2018 scores, relative to that of the EU in 2018._\n\nEU average in 2025\n\nRomania\n\nBulgaria\n\nLatvia\n\nSlovakia\n\nPoland\n\nHungary\n\nCroatia\n\nGreece\n\nCzechia\n\nLithuania\n\nCyprus\n\nPortugal\n\nSpain\n\nItaly\n\nSlovenia\n\nMalta\n\nEU\n\nEstonia\n\nFrance\n\nGermany\n\nAustria\n\nLuxembourg\n\nBelgium\n\nIreland\n\nFinland\n\nNetherlands\n\nDenmark\n\nSweden\n\n0 20 40 60 80 100 120 140 160\n\nChange in summary innovation index between 2018 and 2025\n\nPositive change"],"full_content":"31 European Innovation Scoreboard 2025\n\nFigure 9: Evolution of the innovation performance of the EU27 Member States between 2018 and 2025\n\n_Note: Performance change is measured as the difference between 2025 and 2018 scores, relative to that of the EU in 2018._\n\nEU average in 2025\n\nRomania\n\nBulgaria\n\nLatvia\n\nSlovakia\n\nPoland\n\nHungary\n\nCroatia\n\nGreece\n\nCzechia\n\nLithuania\n\nCyprus\n\nPortugal\n\nSpain\n\nItaly\n\nSlovenia\n\nMalta\n\nEU\n\nEstonia\n\nFrance\n\nGermany\n\nAustria\n\nLuxembourg\n\nBelgium\n\nIreland\n\nFinland\n\nNetherlands\n\nDenmark\n\nSweden\n\n0 20 40 60 80 100 120 140 160\n\nChange in summary innovation index between 2018 and 2025\n\nPositive change","publish_date":null,"title":"Figure 9: Evolution of the innovation performance of the EU27 Member States between 2018 and 2025","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2025-EIS_p33.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2025-EIS_p33.pdf"},"request":{"example_id":"chart/2025-EIS_p33","product_type":"parse","source_file_path":"2025-EIS_p33.pdf"},"started_at":"2026-11-04T19:45:18.483468"} +{"completed_at":"2026-11-04T19:45:20.229151","latency_in_ms":1120,"output":{"example_id":"chart/2025-EIS_p35","grounded_pages":[],"layout_pages":[],"markdown":"33 European Innovation Scoreboard 2025\n\nFigure 10: Innovation performance of the EU27 Member States in the _Human resources_ dimension\n\n_Note: All performance scores are relative to that of the EU in 2018 for each dimension. Horizontal coloured bars represent countries’ performance in 2025, while diamonds and vertical bars indicate their performance in 2018 and 2024, respectively. The colours denote each country’s overall performance group based on the 2025 SII._","pages":[{"line_numbers":[],"markdown":"33 European Innovation Scoreboard 2025\n\nFigure 10: Innovation performance of the EU27 Member States in the _Human resources_ dimension\n\n_Note: All performance scores are relative to that of the EU in 2018 for each dimension. Horizontal coloured bars represent countries’ performance in 2025, while diamonds and vertical bars indicate their performance in 2018 and 2024, respectively. The colours denote each country’s overall performance group based on the 2025 SII._","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"33 European Innovation Scoreboard 2025\n\nFigure 10: Innovation performance of the EU27 Member States in the _Human resources_ dimension\n\n_Note: All performance scores are relative to that of the EU in 2018 for each dimension. Horizontal coloured bars represent countries’ performance in 2025, while diamonds and vertical bars indicate their performance in 2018 and 2024, respectively. The colours denote each country’s overall performance group based on the 2025 SII._","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2025-EIS_p35.pdf"]},"response":{"errors":[],"extract_id":"extract_849147dd3009f5910a23c5cf44537638","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["33 European Innovation Scoreboard 2025\n\nFigure 10: Innovation performance of the EU27 Member States in the _Human resources_ dimension\n\n_Note: All performance scores are relative to that of the EU in 2018 for each dimension. Horizontal coloured bars represent countries’ performance in 2025, while diamonds and vertical bars indicate their performance in 2018 and 2024, respectively. The colours denote each country’s overall performance group based on the 2025 SII._"],"full_content":"33 European Innovation Scoreboard 2025\n\nFigure 10: Innovation performance of the EU27 Member States in the _Human resources_ dimension\n\n_Note: All performance scores are relative to that of the EU in 2018 for each dimension. Horizontal coloured bars represent countries’ performance in 2025, while diamonds and vertical bars indicate their performance in 2018 and 2024, respectively. The colours denote each country’s overall performance group based on the 2025 SII._","publish_date":null,"title":"Figure 10: Innovation performance of the EU27 Member States in the Human resources dimension","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2025-EIS_p35.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2025-EIS_p35.pdf"},"request":{"example_id":"chart/2025-EIS_p35","product_type":"parse","source_file_path":"2025-EIS_p35.pdf"},"started_at":"2026-11-04T19:45:19.109097"} +{"completed_at":"2026-11-04T19:45:21.252689","latency_in_ms":1150,"output":{"example_id":"chart/2025-EIS_p37","grounded_pages":[],"layout_pages":[],"markdown":"35 European Innovation Scoreboard 2025\n\nFigure 11: Innovation performance of the EU27 Member States in the _Attractive research systems_ dimension\n\n_Note: All performance scores are relative to that of the EU in 2018 for each dimension. Horizontal coloured bars represent countries’ performance in 2025, while diamonds and vertical bars indicate their performance in 2018 and 2024, respectively. The colours denote each country’s overall performance group based on the 2025 SII._","pages":[{"line_numbers":[],"markdown":"35 European Innovation Scoreboard 2025\n\nFigure 11: Innovation performance of the EU27 Member States in the _Attractive research systems_ dimension\n\n_Note: All performance scores are relative to that of the EU in 2018 for each dimension. Horizontal coloured bars represent countries’ performance in 2025, while diamonds and vertical bars indicate their performance in 2018 and 2024, respectively. The colours denote each country’s overall performance group based on the 2025 SII._","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"35 European Innovation Scoreboard 2025\n\nFigure 11: Innovation performance of the EU27 Member States in the _Attractive research systems_ dimension\n\n_Note: All performance scores are relative to that of the EU in 2018 for each dimension. Horizontal coloured bars represent countries’ performance in 2025, while diamonds and vertical bars indicate their performance in 2018 and 2024, respectively. The colours denote each country’s overall performance group based on the 2025 SII._","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2025-EIS_p37.pdf"]},"response":{"errors":[],"extract_id":"extract_b7039048ae9750c4fce5646ab51f70b5","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["35 European Innovation Scoreboard 2025\n\nFigure 11: Innovation performance of the EU27 Member States in the _Attractive research systems_ dimension\n\n_Note: All performance scores are relative to that of the EU in 2018 for each dimension. Horizontal coloured bars represent countries’ performance in 2025, while diamonds and vertical bars indicate their performance in 2018 and 2024, respectively. The colours denote each country’s overall performance group based on the 2025 SII._"],"full_content":"35 European Innovation Scoreboard 2025\n\nFigure 11: Innovation performance of the EU27 Member States in the _Attractive research systems_ dimension\n\n_Note: All performance scores are relative to that of the EU in 2018 for each dimension. Horizontal coloured bars represent countries’ performance in 2025, while diamonds and vertical bars indicate their performance in 2018 and 2024, respectively. The colours denote each country’s overall performance group based on the 2025 SII._","publish_date":null,"title":"Figure 11: Innovation performance of the EU27 Member States in the Attractive research systems dimension","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2025-EIS_p37.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2025-EIS_p37.pdf"},"request":{"example_id":"chart/2025-EIS_p37","product_type":"parse","source_file_path":"2025-EIS_p37.pdf"},"started_at":"2026-11-04T19:45:20.102497"} +{"completed_at":"2026-11-04T19:45:24.312004","latency_in_ms":2622,"output":{"example_id":"chart/2025-EIS_p39","grounded_pages":[],"layout_pages":[],"markdown":"37 European Innovation Scoreboard 2025\n\nFigure 12: Innovation performance of the EU27 Member States in the _Digitalisation_ dimension\n\n_Note: All performance scores are relative to that of the EU in 2018 for each dimension. Horizontal coloured bars represent countries’ performance in 2025, while diamonds and vertical bars indicate their performance in 2018 and 2024, respectively. The colours denote each country’s overall performance group based on the 2025 SII._","pages":[{"line_numbers":[],"markdown":"37 European Innovation Scoreboard 2025\n\nFigure 12: Innovation performance of the EU27 Member States in the _Digitalisation_ dimension\n\n_Note: All performance scores are relative to that of the EU in 2018 for each dimension. Horizontal coloured bars represent countries’ performance in 2025, while diamonds and vertical bars indicate their performance in 2018 and 2024, respectively. The colours denote each country’s overall performance group based on the 2025 SII._","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"37 European Innovation Scoreboard 2025\n\nFigure 12: Innovation performance of the EU27 Member States in the _Digitalisation_ dimension\n\n_Note: All performance scores are relative to that of the EU in 2018 for each dimension. Horizontal coloured bars represent countries’ performance in 2025, while diamonds and vertical bars indicate their performance in 2018 and 2024, respectively. The colours denote each country’s overall performance group based on the 2025 SII._","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2025-EIS_p39.pdf"]},"response":{"errors":[],"extract_id":"extract_575ae2b3966722ebd33acf5a458986d4","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["37 European Innovation Scoreboard 2025\n\nFigure 12: Innovation performance of the EU27 Member States in the _Digitalisation_ dimension\n\n_Note: All performance scores are relative to that of the EU in 2018 for each dimension. Horizontal coloured bars represent countries’ performance in 2025, while diamonds and vertical bars indicate their performance in 2018 and 2024, respectively. The colours denote each country’s overall performance group based on the 2025 SII._"],"full_content":"37 European Innovation Scoreboard 2025\n\nFigure 12: Innovation performance of the EU27 Member States in the _Digitalisation_ dimension\n\n_Note: All performance scores are relative to that of the EU in 2018 for each dimension. Horizontal coloured bars represent countries’ performance in 2025, while diamonds and vertical bars indicate their performance in 2018 and 2024, respectively. The colours denote each country’s overall performance group based on the 2025 SII._","publish_date":null,"title":"Figure 12: Innovation performance of the EU27 Member States in the Digitalisation dimension","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2025-EIS_p39.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2025-EIS_p39.pdf"},"request":{"example_id":"chart/2025-EIS_p39","product_type":"parse","source_file_path":"2025-EIS_p39.pdf"},"started_at":"2026-11-04T19:45:21.689598"} +{"completed_at":"2026-11-04T19:45:25.091466","latency_in_ms":2256,"output":{"example_id":"chart/2025-EIS_p41","grounded_pages":[],"layout_pages":[],"markdown":"39 European Innovation Scoreboard 2025\n\nFigure 13: Innovation performance of the EU27 Member States in the _Finance and support_ dimension\n\n_Note: All performance scores are relative to that of the EU in 2018 for each dimension. Horizontal coloured bars represent countries’ performance in 2025, while diamonds and vertical bars indicate their performance in 2018 and 2024, respectively. The colours denote each country’s overall performance group based on the 2025 SII._","pages":[{"line_numbers":[],"markdown":"39 European Innovation Scoreboard 2025\n\nFigure 13: Innovation performance of the EU27 Member States in the _Finance and support_ dimension\n\n_Note: All performance scores are relative to that of the EU in 2018 for each dimension. Horizontal coloured bars represent countries’ performance in 2025, while diamonds and vertical bars indicate their performance in 2018 and 2024, respectively. The colours denote each country’s overall performance group based on the 2025 SII._","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"39 European Innovation Scoreboard 2025\n\nFigure 13: Innovation performance of the EU27 Member States in the _Finance and support_ dimension\n\n_Note: All performance scores are relative to that of the EU in 2018 for each dimension. Horizontal coloured bars represent countries’ performance in 2025, while diamonds and vertical bars indicate their performance in 2018 and 2024, respectively. The colours denote each country’s overall performance group based on the 2025 SII._","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2025-EIS_p41.pdf"]},"response":{"errors":[],"extract_id":"extract_c8258283e857ce31f69b0990452cd1e3","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["39 European Innovation Scoreboard 2025\n\nFigure 13: Innovation performance of the EU27 Member States in the _Finance and support_ dimension\n\n_Note: All performance scores are relative to that of the EU in 2018 for each dimension. Horizontal coloured bars represent countries’ performance in 2025, while diamonds and vertical bars indicate their performance in 2018 and 2024, respectively. The colours denote each country’s overall performance group based on the 2025 SII._"],"full_content":"39 European Innovation Scoreboard 2025\n\nFigure 13: Innovation performance of the EU27 Member States in the _Finance and support_ dimension\n\n_Note: All performance scores are relative to that of the EU in 2018 for each dimension. Horizontal coloured bars represent countries’ performance in 2025, while diamonds and vertical bars indicate their performance in 2018 and 2024, respectively. The colours denote each country’s overall performance group based on the 2025 SII._","publish_date":null,"title":"Figure 13: Innovation performance of the EU27 Member States in the Finance and support dimension","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2025-EIS_p41.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2025-EIS_p41.pdf"},"request":{"example_id":"chart/2025-EIS_p41","product_type":"parse","source_file_path":"2025-EIS_p41.pdf"},"started_at":"2026-11-04T19:45:22.835450"} +{"completed_at":"2026-11-04T19:45:26.640251","latency_in_ms":1892,"output":{"example_id":"chart/2025-EIS_p43","grounded_pages":[],"layout_pages":[],"markdown":"41 European Innovation Scoreboard 2025\n\nFigure 14: Innovation performance of the EU27 Member States in the _Firm investments_ dimension\n\n_Note: All performance scores are relative to that of the EU in 2018 for each dimension. Horizontal coloured bars represent countries’ performance in 2025, while diamonds and vertical bars indicate their performance in 2018 and 2024, respectively. The colours denote each country’s overall performance group based on the 2025 SII._","pages":[{"line_numbers":[],"markdown":"41 European Innovation Scoreboard 2025\n\nFigure 14: Innovation performance of the EU27 Member States in the _Firm investments_ dimension\n\n_Note: All performance scores are relative to that of the EU in 2018 for each dimension. Horizontal coloured bars represent countries’ performance in 2025, while diamonds and vertical bars indicate their performance in 2018 and 2024, respectively. The colours denote each country’s overall performance group based on the 2025 SII._","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"41 European Innovation Scoreboard 2025\n\nFigure 14: Innovation performance of the EU27 Member States in the _Firm investments_ dimension\n\n_Note: All performance scores are relative to that of the EU in 2018 for each dimension. Horizontal coloured bars represent countries’ performance in 2025, while diamonds and vertical bars indicate their performance in 2018 and 2024, respectively. The colours denote each country’s overall performance group based on the 2025 SII._","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2025-EIS_p43.pdf"]},"response":{"errors":[],"extract_id":"extract_7d2f69d63836512ac33cdc98387109ca","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["41 European Innovation Scoreboard 2025\n\nFigure 14: Innovation performance of the EU27 Member States in the _Firm investments_ dimension\n\n_Note: All performance scores are relative to that of the EU in 2018 for each dimension. Horizontal coloured bars represent countries’ performance in 2025, while diamonds and vertical bars indicate their performance in 2018 and 2024, respectively. The colours denote each country’s overall performance group based on the 2025 SII._"],"full_content":"41 European Innovation Scoreboard 2025\n\nFigure 14: Innovation performance of the EU27 Member States in the _Firm investments_ dimension\n\n_Note: All performance scores are relative to that of the EU in 2018 for each dimension. Horizontal coloured bars represent countries’ performance in 2025, while diamonds and vertical bars indicate their performance in 2018 and 2024, respectively. The colours denote each country’s overall performance group based on the 2025 SII._","publish_date":null,"title":"Figure 14: Innovation performance of the EU27 Member States in the Firm investments dimension","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2025-EIS_p43.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2025-EIS_p43.pdf"},"request":{"example_id":"chart/2025-EIS_p43","product_type":"parse","source_file_path":"2025-EIS_p43.pdf"},"started_at":"2026-11-04T19:45:24.748078"} +{"completed_at":"2026-11-04T19:45:27.468648","latency_in_ms":1948,"output":{"example_id":"chart/2025-EIS_p45","grounded_pages":[],"layout_pages":[],"markdown":"43 European Innovation Scoreboard 2025\n\nFigure 15: Innovation performance of the EU27 Member States in the _Investments in information technologies_ dimension\n\n_Note: All performance scores are relative to that of the EU in 2018 for each dimension. Horizontal coloured bars represent countries’ performance in 2025, while diamonds and vertical bars indicate their performance in 2018 and 2024, respectively. The colours denote each country’s overall performance group based on the 2025 SII._","pages":[{"line_numbers":[],"markdown":"43 European Innovation Scoreboard 2025\n\nFigure 15: Innovation performance of the EU27 Member States in the _Investments in information technologies_ dimension\n\n_Note: All performance scores are relative to that of the EU in 2018 for each dimension. Horizontal coloured bars represent countries’ performance in 2025, while diamonds and vertical bars indicate their performance in 2018 and 2024, respectively. The colours denote each country’s overall performance group based on the 2025 SII._","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"43 European Innovation Scoreboard 2025\n\nFigure 15: Innovation performance of the EU27 Member States in the _Investments in information technologies_ dimension\n\n_Note: All performance scores are relative to that of the EU in 2018 for each dimension. Horizontal coloured bars represent countries’ performance in 2025, while diamonds and vertical bars indicate their performance in 2018 and 2024, respectively. The colours denote each country’s overall performance group based on the 2025 SII._","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2025-EIS_p45.pdf"]},"response":{"errors":[],"extract_id":"extract_1a386558663f0a413e5d186bcba6eef5","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["43 European Innovation Scoreboard 2025\n\nFigure 15: Innovation performance of the EU27 Member States in the _Investments in information technologies_ dimension\n\n_Note: All performance scores are relative to that of the EU in 2018 for each dimension. Horizontal coloured bars represent countries’ performance in 2025, while diamonds and vertical bars indicate their performance in 2018 and 2024, respectively. The colours denote each country’s overall performance group based on the 2025 SII._"],"full_content":"43 European Innovation Scoreboard 2025\n\nFigure 15: Innovation performance of the EU27 Member States in the _Investments in information technologies_ dimension\n\n_Note: All performance scores are relative to that of the EU in 2018 for each dimension. Horizontal coloured bars represent countries’ performance in 2025, while diamonds and vertical bars indicate their performance in 2018 and 2024, respectively. The colours denote each country’s overall performance group based on the 2025 SII._","publish_date":null,"title":"Figure 15: Innovation performance of the EU27 Member States in the Investments in information technologies dimension","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2025-EIS_p45.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2025-EIS_p45.pdf"},"request":{"example_id":"chart/2025-EIS_p45","product_type":"parse","source_file_path":"2025-EIS_p45.pdf"},"started_at":"2026-11-04T19:45:25.519719"} +{"completed_at":"2026-11-04T19:45:29.437779","latency_in_ms":2696,"output":{"example_id":"chart/2025-EIS_p47","grounded_pages":[],"layout_pages":[],"markdown":"45 European Innovation Scoreboard 2025\n\nFigure 16: Innovation performance of the EU27 Member States in the _Innovators_ dimension\n\n_Note: All performance scores are relative to that of the EU in 2018 for each dimension. Horizontal coloured bars represent countries’ performance in 2025, while diamonds and vertical bars indicate their performance in 2018 and 2024, respectively. The colours denote each country’s overall performance group based on the 2025 SII._","pages":[{"line_numbers":[],"markdown":"45 European Innovation Scoreboard 2025\n\nFigure 16: Innovation performance of the EU27 Member States in the _Innovators_ dimension\n\n_Note: All performance scores are relative to that of the EU in 2018 for each dimension. Horizontal coloured bars represent countries’ performance in 2025, while diamonds and vertical bars indicate their performance in 2018 and 2024, respectively. The colours denote each country’s overall performance group based on the 2025 SII._","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"45 European Innovation Scoreboard 2025\n\nFigure 16: Innovation performance of the EU27 Member States in the _Innovators_ dimension\n\n_Note: All performance scores are relative to that of the EU in 2018 for each dimension. Horizontal coloured bars represent countries’ performance in 2025, while diamonds and vertical bars indicate their performance in 2018 and 2024, respectively. The colours denote each country’s overall performance group based on the 2025 SII._","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2025-EIS_p47.pdf"]},"response":{"errors":[],"extract_id":"extract_2625254fd526f023434e619cc8abe9ed","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["45 European Innovation Scoreboard 2025\n\nFigure 16: Innovation performance of the EU27 Member States in the _Innovators_ dimension\n\n_Note: All performance scores are relative to that of the EU in 2018 for each dimension. Horizontal coloured bars represent countries’ performance in 2025, while diamonds and vertical bars indicate their performance in 2018 and 2024, respectively. The colours denote each country’s overall performance group based on the 2025 SII._"],"full_content":"45 European Innovation Scoreboard 2025\n\nFigure 16: Innovation performance of the EU27 Member States in the _Innovators_ dimension\n\n_Note: All performance scores are relative to that of the EU in 2018 for each dimension. Horizontal coloured bars represent countries’ performance in 2025, while diamonds and vertical bars indicate their performance in 2018 and 2024, respectively. The colours denote each country’s overall performance group based on the 2025 SII._","publish_date":null,"title":"Figure 16: Innovation performance of the EU27 Member States in the Innovators dimension","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2025-EIS_p47.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2025-EIS_p47.pdf"},"request":{"example_id":"chart/2025-EIS_p47","product_type":"parse","source_file_path":"2025-EIS_p47.pdf"},"started_at":"2026-11-04T19:45:26.741275"} +{"completed_at":"2026-11-04T19:45:29.274251","latency_in_ms":1128,"output":{"example_id":"chart/2025-EIS_p49","grounded_pages":[],"layout_pages":[],"markdown":"47 European Innovation Scoreboard 2025\n\nFigure 17: Innovation performance of the EU27 Member States in the _Linkages_ dimension\n\n_Note: All performance scores are relative to that of the EU in 2018 for each dimension. Horizontal coloured bars represent countries’ performance in 2025, while diamonds and vertical bars indicate their performance in 2018 and 2024, respectively. The colours denote each country’s overall performance group based on the 2025 SII._","pages":[{"line_numbers":[],"markdown":"47 European Innovation Scoreboard 2025\n\nFigure 17: Innovation performance of the EU27 Member States in the _Linkages_ dimension\n\n_Note: All performance scores are relative to that of the EU in 2018 for each dimension. Horizontal coloured bars represent countries’ performance in 2025, while diamonds and vertical bars indicate their performance in 2018 and 2024, respectively. The colours denote each country’s overall performance group based on the 2025 SII._","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"47 European Innovation Scoreboard 2025\n\nFigure 17: Innovation performance of the EU27 Member States in the _Linkages_ dimension\n\n_Note: All performance scores are relative to that of the EU in 2018 for each dimension. Horizontal coloured bars represent countries’ performance in 2025, while diamonds and vertical bars indicate their performance in 2018 and 2024, respectively. The colours denote each country’s overall performance group based on the 2025 SII._","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2025-EIS_p49.pdf"]},"response":{"errors":[],"extract_id":"extract_ed86998ccf34c88bb87055bb267f8d0e","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["47 European Innovation Scoreboard 2025\n\nFigure 17: Innovation performance of the EU27 Member States in the _Linkages_ dimension\n\n_Note: All performance scores are relative to that of the EU in 2018 for each dimension. Horizontal coloured bars represent countries’ performance in 2025, while diamonds and vertical bars indicate their performance in 2018 and 2024, respectively. The colours denote each country’s overall performance group based on the 2025 SII._"],"full_content":"47 European Innovation Scoreboard 2025\n\nFigure 17: Innovation performance of the EU27 Member States in the _Linkages_ dimension\n\n_Note: All performance scores are relative to that of the EU in 2018 for each dimension. Horizontal coloured bars represent countries’ performance in 2025, while diamonds and vertical bars indicate their performance in 2018 and 2024, respectively. The colours denote each country’s overall performance group based on the 2025 SII._","publish_date":null,"title":"Figure 17: Innovation performance of the EU27 Member States in the Linkages dimension","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2025-EIS_p49.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2025-EIS_p49.pdf"},"request":{"example_id":"chart/2025-EIS_p49","product_type":"parse","source_file_path":"2025-EIS_p49.pdf"},"started_at":"2026-11-04T19:45:28.145595"} +{"completed_at":"2026-11-04T19:45:30.632857","latency_in_ms":1193,"output":{"example_id":"chart/2025-EIS_p51","grounded_pages":[],"layout_pages":[],"markdown":"49 European Innovation Scoreboard 2025\n\nFigure 18: Innovation performance of the EU27 Member States in the _Intellectual assets_ dimension\n\n_Note: All performance scores are relative to that of the EU in 2018 for each dimension. Horizontal coloured bars represent countries’ performance in 2025, while diamonds and vertical bars indicate their performance in 2018 and 2024, respectively. The colours denote each country’s overall performance group based on the 2025 SII._","pages":[{"line_numbers":[],"markdown":"49 European Innovation Scoreboard 2025\n\nFigure 18: Innovation performance of the EU27 Member States in the _Intellectual assets_ dimension\n\n_Note: All performance scores are relative to that of the EU in 2018 for each dimension. Horizontal coloured bars represent countries’ performance in 2025, while diamonds and vertical bars indicate their performance in 2018 and 2024, respectively. The colours denote each country’s overall performance group based on the 2025 SII._","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"49 European Innovation Scoreboard 2025\n\nFigure 18: Innovation performance of the EU27 Member States in the _Intellectual assets_ dimension\n\n_Note: All performance scores are relative to that of the EU in 2018 for each dimension. Horizontal coloured bars represent countries’ performance in 2025, while diamonds and vertical bars indicate their performance in 2018 and 2024, respectively. The colours denote each country’s overall performance group based on the 2025 SII._","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2025-EIS_p51.pdf"]},"response":{"errors":[],"extract_id":"extract_98bca2891c3127d4acdbe58d826b5bd7","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["49 European Innovation Scoreboard 2025\n\nFigure 18: Innovation performance of the EU27 Member States in the _Intellectual assets_ dimension\n\n_Note: All performance scores are relative to that of the EU in 2018 for each dimension. Horizontal coloured bars represent countries’ performance in 2025, while diamonds and vertical bars indicate their performance in 2018 and 2024, respectively. The colours denote each country’s overall performance group based on the 2025 SII._"],"full_content":"49 European Innovation Scoreboard 2025\n\nFigure 18: Innovation performance of the EU27 Member States in the _Intellectual assets_ dimension\n\n_Note: All performance scores are relative to that of the EU in 2018 for each dimension. Horizontal coloured bars represent countries’ performance in 2025, while diamonds and vertical bars indicate their performance in 2018 and 2024, respectively. The colours denote each country’s overall performance group based on the 2025 SII._","publish_date":null,"title":"Figure 18: Innovation performance of the EU27 Member States in the Intellectual assets dimension","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2025-EIS_p51.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2025-EIS_p51.pdf"},"request":{"example_id":"chart/2025-EIS_p51","product_type":"parse","source_file_path":"2025-EIS_p51.pdf"},"started_at":"2026-11-04T19:45:29.439638"} +{"completed_at":"2026-11-04T19:45:32.265382","latency_in_ms":2135,"output":{"example_id":"chart/2025-EIS_p53","grounded_pages":[],"layout_pages":[],"markdown":"51 European Innovation Scoreboard 2025\n\nFigure 19: Innovation performance of the EU27 Member States in the _Sales and employment impacts_ dimension\n\n_Note: All performance scores are relative to that of the EU in 2018 for each dimension. Horizontal coloured bars represent countries’ performance in 2025, while diamonds and vertical bars indicate their performance in 2018 and 2024, respectively. The colours denote each country’s overall performance group based on the 2025 SII._","pages":[{"line_numbers":[],"markdown":"51 European Innovation Scoreboard 2025\n\nFigure 19: Innovation performance of the EU27 Member States in the _Sales and employment impacts_ dimension\n\n_Note: All performance scores are relative to that of the EU in 2018 for each dimension. Horizontal coloured bars represent countries’ performance in 2025, while diamonds and vertical bars indicate their performance in 2018 and 2024, respectively. The colours denote each country’s overall performance group based on the 2025 SII._","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"51 European Innovation Scoreboard 2025\n\nFigure 19: Innovation performance of the EU27 Member States in the _Sales and employment impacts_ dimension\n\n_Note: All performance scores are relative to that of the EU in 2018 for each dimension. Horizontal coloured bars represent countries’ performance in 2025, while diamonds and vertical bars indicate their performance in 2018 and 2024, respectively. The colours denote each country’s overall performance group based on the 2025 SII._","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2025-EIS_p53.pdf"]},"response":{"errors":[],"extract_id":"extract_c126aaf291b8d3f53d8370478dcceb03","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["51 European Innovation Scoreboard 2025\n\nFigure 19: Innovation performance of the EU27 Member States in the _Sales and employment impacts_ dimension\n\n_Note: All performance scores are relative to that of the EU in 2018 for each dimension. Horizontal coloured bars represent countries’ performance in 2025, while diamonds and vertical bars indicate their performance in 2018 and 2024, respectively. The colours denote each country’s overall performance group based on the 2025 SII._"],"full_content":"51 European Innovation Scoreboard 2025\n\nFigure 19: Innovation performance of the EU27 Member States in the _Sales and employment impacts_ dimension\n\n_Note: All performance scores are relative to that of the EU in 2018 for each dimension. Horizontal coloured bars represent countries’ performance in 2025, while diamonds and vertical bars indicate their performance in 2018 and 2024, respectively. The colours denote each country’s overall performance group based on the 2025 SII._","publish_date":null,"title":"Figure 19: Innovation performance of the EU27 Member States in the Sales and employment impacts dimension","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2025-EIS_p53.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2025-EIS_p53.pdf"},"request":{"example_id":"chart/2025-EIS_p53","product_type":"parse","source_file_path":"2025-EIS_p53.pdf"},"started_at":"2026-11-04T19:45:30.129395"} +{"completed_at":"2026-11-04T19:45:32.280249","latency_in_ms":1109,"output":{"example_id":"chart/2025-EIS_p55","grounded_pages":[],"layout_pages":[],"markdown":"53 European Innovation Scoreboard 2025\n\nFigure 20: Innovation performance of the EU27 Member States in the _Trade impacts_ dimension\n\n_Note: All performance scores are relative to that of the EU in 2018 for each dimension. Horizontal coloured bars represent countries’ performance in 2025, while diamonds and vertical bars indicate their performance in 2018 and 2024, respectively. The colours denote each country’s overall performance group based on the 2025 SII._","pages":[{"line_numbers":[],"markdown":"53 European Innovation Scoreboard 2025\n\nFigure 20: Innovation performance of the EU27 Member States in the _Trade impacts_ dimension\n\n_Note: All performance scores are relative to that of the EU in 2018 for each dimension. Horizontal coloured bars represent countries’ performance in 2025, while diamonds and vertical bars indicate their performance in 2018 and 2024, respectively. The colours denote each country’s overall performance group based on the 2025 SII._","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"53 European Innovation Scoreboard 2025\n\nFigure 20: Innovation performance of the EU27 Member States in the _Trade impacts_ dimension\n\n_Note: All performance scores are relative to that of the EU in 2018 for each dimension. Horizontal coloured bars represent countries’ performance in 2025, while diamonds and vertical bars indicate their performance in 2018 and 2024, respectively. The colours denote each country’s overall performance group based on the 2025 SII._","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2025-EIS_p55.pdf"]},"response":{"errors":[],"extract_id":"extract_48530a1e2137341ac023c1a42ad314b5","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["53 European Innovation Scoreboard 2025\n\nFigure 20: Innovation performance of the EU27 Member States in the _Trade impacts_ dimension\n\n_Note: All performance scores are relative to that of the EU in 2018 for each dimension. Horizontal coloured bars represent countries’ performance in 2025, while diamonds and vertical bars indicate their performance in 2018 and 2024, respectively. The colours denote each country’s overall performance group based on the 2025 SII._"],"full_content":"53 European Innovation Scoreboard 2025\n\nFigure 20: Innovation performance of the EU27 Member States in the _Trade impacts_ dimension\n\n_Note: All performance scores are relative to that of the EU in 2018 for each dimension. Horizontal coloured bars represent countries’ performance in 2025, while diamonds and vertical bars indicate their performance in 2018 and 2024, respectively. The colours denote each country’s overall performance group based on the 2025 SII._","publish_date":null,"title":"Figure 20: Innovation performance of the EU27 Member States in the Trade impacts dimension","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2025-EIS_p55.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2025-EIS_p55.pdf"},"request":{"example_id":"chart/2025-EIS_p55","product_type":"parse","source_file_path":"2025-EIS_p55.pdf"},"started_at":"2026-11-04T19:45:31.170307"} +{"completed_at":"2026-11-04T19:45:33.077671","latency_in_ms":1052,"output":{"example_id":"chart/2025-EIS_p57","grounded_pages":[],"layout_pages":[],"markdown":"55 European Innovation Scoreboard 2025\n\nFigure 21: Innovation performance of the EU27 Member States in the _Resource and Labour productivity_ dimension\n\n_Note: All performance scores are relative to that of the EU in 2018 for each dimension. Horizontal coloured bars represent countries’ performance in 2025, while diamonds and vertical bars indicate their performance in 2018 and 2024, respectively. The colours denote each country’s overall performance group based on the 2025 SII._","pages":[{"line_numbers":[],"markdown":"55 European Innovation Scoreboard 2025\n\nFigure 21: Innovation performance of the EU27 Member States in the _Resource and Labour productivity_ dimension\n\n_Note: All performance scores are relative to that of the EU in 2018 for each dimension. Horizontal coloured bars represent countries’ performance in 2025, while diamonds and vertical bars indicate their performance in 2018 and 2024, respectively. The colours denote each country’s overall performance group based on the 2025 SII._","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"55 European Innovation Scoreboard 2025\n\nFigure 21: Innovation performance of the EU27 Member States in the _Resource and Labour productivity_ dimension\n\n_Note: All performance scores are relative to that of the EU in 2018 for each dimension. Horizontal coloured bars represent countries’ performance in 2025, while diamonds and vertical bars indicate their performance in 2018 and 2024, respectively. The colours denote each country’s overall performance group based on the 2025 SII._","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2025-EIS_p57.pdf"]},"response":{"errors":[],"extract_id":"extract_023980d0db806c865706eed6eadb046c","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["55 European Innovation Scoreboard 2025\n\nFigure 21: Innovation performance of the EU27 Member States in the _Resource and Labour productivity_ dimension\n\n_Note: All performance scores are relative to that of the EU in 2018 for each dimension. Horizontal coloured bars represent countries’ performance in 2025, while diamonds and vertical bars indicate their performance in 2018 and 2024, respectively. The colours denote each country’s overall performance group based on the 2025 SII._"],"full_content":"55 European Innovation Scoreboard 2025\n\nFigure 21: Innovation performance of the EU27 Member States in the _Resource and Labour productivity_ dimension\n\n_Note: All performance scores are relative to that of the EU in 2018 for each dimension. Horizontal coloured bars represent countries’ performance in 2025, while diamonds and vertical bars indicate their performance in 2018 and 2024, respectively. The colours denote each country’s overall performance group based on the 2025 SII._","publish_date":null,"title":"Figure 21: Innovation performance of the EU27 Member States in the Resource and Labour productivity dimension","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2025-EIS_p57.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2025-EIS_p57.pdf"},"request":{"example_id":"chart/2025-EIS_p57","product_type":"parse","source_file_path":"2025-EIS_p57.pdf"},"started_at":"2026-11-04T19:45:32.025611"} +{"completed_at":"2026-11-04T19:45:33.946426","latency_in_ms":947,"output":{"example_id":"chart/2025-EIS_p59","grounded_pages":[],"layout_pages":[],"markdown":"57 European Innovation Scoreboard 2025\n\n(15%-points), and _Intellectual assets_ (15%-points), still favour Moderate Innovators. The largest gaps are seen in _Innovators_ (73%-points), and _Sales and employment impacts_ (69%-points), again in favour of Moderate Innovators.\n\nFigure 22: Innovation performance of the innovation groups per dimension\n\n_Note: Average scores for each performance group are defined as the unweighted average of the relative-to-EU scores of the Member States within that group. As these unweighted averages do not consider differences in country size, results are not directly comparable. For this reason, average scores for the performance groups have been adjusted such that the unweighted average of the four groups for each dimension equals 100._\n\n54\n\n141\n\n85 119\n\n42\n\n151\n\n83 124\n\n79\n\n131\n\n89 102\n\n45\n\n148\n\n77 130\n\n50\n\n144\n\n84 122\n\n70\n\n140\n\n85 105\n\n38\n\n127\n\n111 123\n\n44\n\n145\n\n85 126\n\n72\n\n132\n\n87 109\n\n45\n\n122\n\n114 119\n\n87\n\n122\n\n86 105\n\n57\n\n123\n\n90 130 Resource and labour productivity\n\nTrade impacts\n\nSales and employment impacts\n\nIntellectual assets\n\nLinkages\n\nInnovators\n\nInvestments in information technologies\n\nFirm investments\n\nFinance and support\n\nDigitalisation\n\nAttractive research systems\n\nHuman resources\n\n0 20 40 60 80 100 120 140 160\n\nAdjusted score per dimension in 2025\n\nEmerging innovators Moderate innovators Strong innovators Innovation leaders","pages":[{"line_numbers":[],"markdown":"57 European Innovation Scoreboard 2025\n\n(15%-points), and _Intellectual assets_ (15%-points), still favour Moderate Innovators. The largest gaps are seen in _Innovators_ (73%-points), and _Sales and employment impacts_ (69%-points), again in favour of Moderate Innovators.\n\nFigure 22: Innovation performance of the innovation groups per dimension\n\n_Note: Average scores for each performance group are defined as the unweighted average of the relative-to-EU scores of the Member States within that group. As these unweighted averages do not consider differences in country size, results are not directly comparable. For this reason, average scores for the performance groups have been adjusted such that the unweighted average of the four groups for each dimension equals 100._\n\n54\n\n141\n\n85 119\n\n42\n\n151\n\n83 124\n\n79\n\n131\n\n89 102\n\n45\n\n148\n\n77 130\n\n50\n\n144\n\n84 122\n\n70\n\n140\n\n85 105\n\n38\n\n127\n\n111 123\n\n44\n\n145\n\n85 126\n\n72\n\n132\n\n87 109\n\n45\n\n122\n\n114 119\n\n87\n\n122\n\n86 105\n\n57\n\n123\n\n90 130 Resource and labour productivity\n\nTrade impacts\n\nSales and employment impacts\n\nIntellectual assets\n\nLinkages\n\nInnovators\n\nInvestments in information technologies\n\nFirm investments\n\nFinance and support\n\nDigitalisation\n\nAttractive research systems\n\nHuman resources\n\n0 20 40 60 80 100 120 140 160\n\nAdjusted score per dimension in 2025\n\nEmerging innovators Moderate innovators Strong innovators Innovation leaders","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"57 European Innovation Scoreboard 2025\n\n(15%-points), and _Intellectual assets_ (15%-points), still favour Moderate Innovators. The largest gaps are seen in _Innovators_ (73%-points), and _Sales and employment impacts_ (69%-points), again in favour of Moderate Innovators.\n\nFigure 22: Innovation performance of the innovation groups per dimension\n\n_Note: Average scores for each performance group are defined as the unweighted average of the relative-to-EU scores of the Member States within that group. As these unweighted averages do not consider differences in country size, results are not directly comparable. For this reason, average scores for the performance groups have been adjusted such that the unweighted average of the four groups for each dimension equals 100._\n\n54\n\n141\n\n85 119\n\n42\n\n151\n\n83 124\n\n79\n\n131\n\n89 102\n\n45\n\n148\n\n77 130\n\n50\n\n144\n\n84 122\n\n70\n\n140\n\n85 105\n\n38\n\n127\n\n111 123\n\n44\n\n145\n\n85 126\n\n72\n\n132\n\n87 109\n\n45\n\n122\n\n114 119\n\n87\n\n122\n\n86 105\n\n57\n\n123\n\n90 130 Resource and labour productivity\n\nTrade impacts\n\nSales and employment impacts\n\nIntellectual assets\n\nLinkages\n\nInnovators\n\nInvestments in information technologies\n\nFirm investments\n\nFinance and support\n\nDigitalisation\n\nAttractive research systems\n\nHuman resources\n\n0 20 40 60 80 100 120 140 160\n\nAdjusted score per dimension in 2025\n\nEmerging innovators Moderate innovators Strong innovators Innovation leaders","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2025-EIS_p59.pdf"]},"response":{"errors":[],"extract_id":"extract_9af0bfba2c97490557608cb79d45a32a","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["57 European Innovation Scoreboard 2025\n\n(15%-points), and _Intellectual assets_ (15%-points), still favour Moderate Innovators. The largest gaps are seen in _Innovators_ (73%-points), and _Sales and employment impacts_ (69%-points), again in favour of Moderate Innovators.\n\nFigure 22: Innovation performance of the innovation groups per dimension\n\n_Note: Average scores for each performance group are defined as the unweighted average of the relative-to-EU scores of the Member States within that group. As these unweighted averages do not consider differences in country size, results are not directly comparable. For this reason, average scores for the performance groups have been adjusted such that the unweighted average of the four groups for each dimension equals 100._\n\n54\n\n141\n\n85 119\n\n42\n\n151\n\n83 124\n\n79\n\n131\n\n89 102\n\n45\n\n148\n\n77 130\n\n50\n\n144\n\n84 122\n\n70\n\n140\n\n85 105\n\n38\n\n127\n\n111 123\n\n44\n\n145\n\n85 126\n\n72\n\n132\n\n87 109\n\n45\n\n122\n\n114 119\n\n87\n\n122\n\n86 105\n\n57\n\n123\n\n90 130 Resource and labour productivity\n\nTrade impacts\n\nSales and employment impacts\n\nIntellectual assets\n\nLinkages\n\nInnovators\n\nInvestments in information technologies\n\nFirm investments\n\nFinance and support\n\nDigitalisation\n\nAttractive research systems\n\nHuman resources\n\n0 20 40 60 80 100 120 140 160\n\nAdjusted score per dimension in 2025\n\nEmerging innovators Moderate innovators Strong innovators Innovation leaders"],"full_content":"57 European Innovation Scoreboard 2025\n\n(15%-points), and _Intellectual assets_ (15%-points), still favour Moderate Innovators. The largest gaps are seen in _Innovators_ (73%-points), and _Sales and employment impacts_ (69%-points), again in favour of Moderate Innovators.\n\nFigure 22: Innovation performance of the innovation groups per dimension\n\n_Note: Average scores for each performance group are defined as the unweighted average of the relative-to-EU scores of the Member States within that group. As these unweighted averages do not consider differences in country size, results are not directly comparable. For this reason, average scores for the performance groups have been adjusted such that the unweighted average of the four groups for each dimension equals 100._\n\n54\n\n141\n\n85 119\n\n42\n\n151\n\n83 124\n\n79\n\n131\n\n89 102\n\n45\n\n148\n\n77 130\n\n50\n\n144\n\n84 122\n\n70\n\n140\n\n85 105\n\n38\n\n127\n\n111 123\n\n44\n\n145\n\n85 126\n\n72\n\n132\n\n87 109\n\n45\n\n122\n\n114 119\n\n87\n\n122\n\n86 105\n\n57\n\n123\n\n90 130 Resource and labour productivity\n\nTrade impacts\n\nSales and employment impacts\n\nIntellectual assets\n\nLinkages\n\nInnovators\n\nInvestments in information technologies\n\nFirm investments\n\nFinance and support\n\nDigitalisation\n\nAttractive research systems\n\nHuman resources\n\n0 20 40 60 80 100 120 140 160\n\nAdjusted score per dimension in 2025\n\nEmerging innovators Moderate innovators Strong innovators Innovation leaders","publish_date":null,"title":"(15%-points), and Intellectual assets (15%-points), still favour Moderate Innovators. The largest gaps are seen in Innovators (73%-points), and Sales and","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2025-EIS_p59.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2025-EIS_p59.pdf"},"request":{"example_id":"chart/2025-EIS_p59","product_type":"parse","source_file_path":"2025-EIS_p59.pdf"},"started_at":"2026-11-04T19:45:32.998448"} +{"completed_at":"2026-11-04T19:45:36.299681","latency_in_ms":1869,"output":{"example_id":"chart/2025-EIS_p62","grounded_pages":[],"layout_pages":[],"markdown":"60 European Innovation Scoreboard 2025\n\nFigure 23: Performance Innovation Leaders\n\n_Note: Performance is relative to that of the EU in 2018. The graph on the left shows the average performance of the Innovation Leaders calculated as the unweighted average of the respective Member States. All vertical scales in Figure 23-Figure 26 span a range of 70.0%-points to allow an easy comparison of the results between the four performance groups._\n\n**Strong Innovators**\n\nFor the Strong Innovators, Ireland (138.6% of the EU average in 2018) now leads the group, while Belgium (138.1%) dropped to second in 2025, closely followed by Luxembourg (128.9%) and Austria (128.3%). Germany (125.1%) and France (122.3%) also remained among the Strong Innovators, maintaining stable high-level performance. Estonia (118.0%) stands out for its upward trajectory, significantly narrowing the gap with leading countries through sustained innovation gains since 2018.\n\n**Strong Innovators have seen notable progress in digital and collaborative domains, though performance in** **_Intellectual assets_** **has lagged, particularly in design-related outputs.**\n\nSince the 2018 baseline, Strong Innovators have made steady progress in adopting digital technologies, with most countries recording notable improvements in _Investments in information technologies_ . Estonia leads the group with a 103.6%-point increase, followed by Austria (+85.0%-points), Germany (+79.4%-points), and Luxembourg (+63.2%-points), Ireland (+62.7%-points) and Belgium (+55.7%-points) also posted solid gains, while France (+6.1%-points) showed only marginal improvement over the baseline.\n\nUnder the _Digitalisation_ dimension, nearly all Strong Innovators recorded substantial advances. Ireland stands out with the highest increase (+84.9%-points), followed by Austria (+66.1%-points), Germany (+60.6%-points), and France (+52.9%-points). Belgium\n\n100\n\n110\n\n120\n\n130\n\n140\n\n150\n\n160\n\n170\n\n'18 '19 '20 '21 '22 '23 '24 '25\n\nSummary innovation index\n\nDenmark\n\nFinland\n\nNetherlands\n\nSweden\n\n100\n\n110\n\n120\n\n130\n\n140\n\n150\n\n160\n\n170\n\n'18 '19 '20 '21 '22 '23 '24 '25\n\nSummary innovation index","pages":[{"line_numbers":[],"markdown":"60 European Innovation Scoreboard 2025\n\nFigure 23: Performance Innovation Leaders\n\n_Note: Performance is relative to that of the EU in 2018. The graph on the left shows the average performance of the Innovation Leaders calculated as the unweighted average of the respective Member States. All vertical scales in Figure 23-Figure 26 span a range of 70.0%-points to allow an easy comparison of the results between the four performance groups._\n\n**Strong Innovators**\n\nFor the Strong Innovators, Ireland (138.6% of the EU average in 2018) now leads the group, while Belgium (138.1%) dropped to second in 2025, closely followed by Luxembourg (128.9%) and Austria (128.3%). Germany (125.1%) and France (122.3%) also remained among the Strong Innovators, maintaining stable high-level performance. Estonia (118.0%) stands out for its upward trajectory, significantly narrowing the gap with leading countries through sustained innovation gains since 2018.\n\n**Strong Innovators have seen notable progress in digital and collaborative domains, though performance in** **_Intellectual assets_** **has lagged, particularly in design-related outputs.**\n\nSince the 2018 baseline, Strong Innovators have made steady progress in adopting digital technologies, with most countries recording notable improvements in _Investments in information technologies_ . Estonia leads the group with a 103.6%-point increase, followed by Austria (+85.0%-points), Germany (+79.4%-points), and Luxembourg (+63.2%-points), Ireland (+62.7%-points) and Belgium (+55.7%-points) also posted solid gains, while France (+6.1%-points) showed only marginal improvement over the baseline.\n\nUnder the _Digitalisation_ dimension, nearly all Strong Innovators recorded substantial advances. Ireland stands out with the highest increase (+84.9%-points), followed by Austria (+66.1%-points), Germany (+60.6%-points), and France (+52.9%-points). Belgium\n\n100\n\n110\n\n120\n\n130\n\n140\n\n150\n\n160\n\n170\n\n'18 '19 '20 '21 '22 '23 '24 '25\n\nSummary innovation index\n\nDenmark\n\nFinland\n\nNetherlands\n\nSweden\n\n100\n\n110\n\n120\n\n130\n\n140\n\n150\n\n160\n\n170\n\n'18 '19 '20 '21 '22 '23 '24 '25\n\nSummary innovation index","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"60 European Innovation Scoreboard 2025\n\nFigure 23: Performance Innovation Leaders\n\n_Note: Performance is relative to that of the EU in 2018. The graph on the left shows the average performance of the Innovation Leaders calculated as the unweighted average of the respective Member States. All vertical scales in Figure 23-Figure 26 span a range of 70.0%-points to allow an easy comparison of the results between the four performance groups._\n\n**Strong Innovators**\n\nFor the Strong Innovators, Ireland (138.6% of the EU average in 2018) now leads the group, while Belgium (138.1%) dropped to second in 2025, closely followed by Luxembourg (128.9%) and Austria (128.3%). Germany (125.1%) and France (122.3%) also remained among the Strong Innovators, maintaining stable high-level performance. Estonia (118.0%) stands out for its upward trajectory, significantly narrowing the gap with leading countries through sustained innovation gains since 2018.\n\n**Strong Innovators have seen notable progress in digital and collaborative domains, though performance in** **_Intellectual assets_** **has lagged, particularly in design-related outputs.**\n\nSince the 2018 baseline, Strong Innovators have made steady progress in adopting digital technologies, with most countries recording notable improvements in _Investments in information technologies_ . Estonia leads the group with a 103.6%-point increase, followed by Austria (+85.0%-points), Germany (+79.4%-points), and Luxembourg (+63.2%-points), Ireland (+62.7%-points) and Belgium (+55.7%-points) also posted solid gains, while France (+6.1%-points) showed only marginal improvement over the baseline.\n\nUnder the _Digitalisation_ dimension, nearly all Strong Innovators recorded substantial advances. Ireland stands out with the highest increase (+84.9%-points), followed by Austria (+66.1%-points), Germany (+60.6%-points), and France (+52.9%-points). Belgium\n\n100\n\n110\n\n120\n\n130\n\n140\n\n150\n\n160\n\n170\n\n'18 '19 '20 '21 '22 '23 '24 '25\n\nSummary innovation index\n\nDenmark\n\nFinland\n\nNetherlands\n\nSweden\n\n100\n\n110\n\n120\n\n130\n\n140\n\n150\n\n160\n\n170\n\n'18 '19 '20 '21 '22 '23 '24 '25\n\nSummary innovation index","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2025-EIS_p62.pdf"]},"response":{"errors":[],"extract_id":"extract_390d7a093baeccd52768cdf4394c1499","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["60 European Innovation Scoreboard 2025\n\nFigure 23: Performance Innovation Leaders\n\n_Note: Performance is relative to that of the EU in 2018. The graph on the left shows the average performance of the Innovation Leaders calculated as the unweighted average of the respective Member States. All vertical scales in Figure 23-Figure 26 span a range of 70.0%-points to allow an easy comparison of the results between the four performance groups._\n\n**Strong Innovators**\n\nFor the Strong Innovators, Ireland (138.6% of the EU average in 2018) now leads the group, while Belgium (138.1%) dropped to second in 2025, closely followed by Luxembourg (128.9%) and Austria (128.3%). Germany (125.1%) and France (122.3%) also remained among the Strong Innovators, maintaining stable high-level performance. Estonia (118.0%) stands out for its upward trajectory, significantly narrowing the gap with leading countries through sustained innovation gains since 2018.\n\n**Strong Innovators have seen notable progress in digital and collaborative domains, though performance in** **_Intellectual assets_** **has lagged, particularly in design-related outputs.**\n\nSince the 2018 baseline, Strong Innovators have made steady progress in adopting digital technologies, with most countries recording notable improvements in _Investments in information technologies_ . Estonia leads the group with a 103.6%-point increase, followed by Austria (+85.0%-points), Germany (+79.4%-points), and Luxembourg (+63.2%-points), Ireland (+62.7%-points) and Belgium (+55.7%-points) also posted solid gains, while France (+6.1%-points) showed only marginal improvement over the baseline.\n\nUnder the _Digitalisation_ dimension, nearly all Strong Innovators recorded substantial advances. Ireland stands out with the highest increase (+84.9%-points), followed by Austria (+66.1%-points), Germany (+60.6%-points), and France (+52.9%-points). Belgium\n\n100\n\n110\n\n120\n\n130\n\n140\n\n150\n\n160\n\n170\n\n'18 '19 '20 '21 '22 '23 '24 '25\n\nSummary innovation index\n\nDenmark\n\nFinland\n\nNetherlands\n\nSweden\n\n100\n\n110\n\n120\n\n130\n\n140\n\n150\n\n160\n\n170\n\n'18 '19 '20 '21 '22 '23 '24 '25\n\nSummary innovation index"],"full_content":"60 European Innovation Scoreboard 2025\n\nFigure 23: Performance Innovation Leaders\n\n_Note: Performance is relative to that of the EU in 2018. The graph on the left shows the average performance of the Innovation Leaders calculated as the unweighted average of the respective Member States. All vertical scales in Figure 23-Figure 26 span a range of 70.0%-points to allow an easy comparison of the results between the four performance groups._\n\n**Strong Innovators**\n\nFor the Strong Innovators, Ireland (138.6% of the EU average in 2018) now leads the group, while Belgium (138.1%) dropped to second in 2025, closely followed by Luxembourg (128.9%) and Austria (128.3%). Germany (125.1%) and France (122.3%) also remained among the Strong Innovators, maintaining stable high-level performance. Estonia (118.0%) stands out for its upward trajectory, significantly narrowing the gap with leading countries through sustained innovation gains since 2018.\n\n**Strong Innovators have seen notable progress in digital and collaborative domains, though performance in** **_Intellectual assets_** **has lagged, particularly in design-related outputs.**\n\nSince the 2018 baseline, Strong Innovators have made steady progress in adopting digital technologies, with most countries recording notable improvements in _Investments in information technologies_ . Estonia leads the group with a 103.6%-point increase, followed by Austria (+85.0%-points), Germany (+79.4%-points), and Luxembourg (+63.2%-points), Ireland (+62.7%-points) and Belgium (+55.7%-points) also posted solid gains, while France (+6.1%-points) showed only marginal improvement over the baseline.\n\nUnder the _Digitalisation_ dimension, nearly all Strong Innovators recorded substantial advances. Ireland stands out with the highest increase (+84.9%-points), followed by Austria (+66.1%-points), Germany (+60.6%-points), and France (+52.9%-points). Belgium\n\n100\n\n110\n\n120\n\n130\n\n140\n\n150\n\n160\n\n170\n\n'18 '19 '20 '21 '22 '23 '24 '25\n\nSummary innovation index\n\nDenmark\n\nFinland\n\nNetherlands\n\nSweden\n\n100\n\n110\n\n120\n\n130\n\n140\n\n150\n\n160\n\n170\n\n'18 '19 '20 '21 '22 '23 '24 '25\n\nSummary innovation index","publish_date":null,"title":"Figure 23: Performance Innovation Leaders Strong Innovators For the Strong Innovators, Ireland (138.6% of the EU average in 2018) now","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2025-EIS_p62.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2025-EIS_p62.pdf"},"request":{"example_id":"chart/2025-EIS_p62","product_type":"parse","source_file_path":"2025-EIS_p62.pdf"},"started_at":"2026-11-04T19:45:34.430045"} +{"completed_at":"2026-11-04T19:45:37.364072","latency_in_ms":2123,"output":{"example_id":"chart/2025-EIS_p64","grounded_pages":[],"layout_pages":[],"markdown":"62 European Innovation Scoreboard 2025\n\nBox 3: Country level observations of Strong Innovators\n\n**Ireland** demonstrated strong upward momentum, moving from 125.3% of the EU average in 2018 to 138.6% in 2025, with sustained growth particularly after 2020.\n\n**Belgium** consistently improved its score from 126.4 in 2018 to a high of 138.9 in 2024, with only a minor dip in 2025 (reaching 138.1).\n\n**Luxembourg** fluctuated over the period, peaking in 2019 (130.1), declining through 2024, and recovering to 128.9 in 2025.\n\n**Austria** saw gradual improvements in innovation performance from 2018 (120.1) to a peak in 2024 (132.3), followed by a slight decline in 2025 (128.3).\n\n**Germany** rose from 116.4 in 2018 to 127.8 in 2022, then stabilised around 127 before a slight drop to 125.1 in 2025.\n\n**France** experienced modest but steady progress, increasing from 114.9 in 2018 to 122.3 in 2025.\n\n**Estonia** recorded the strongest relative growth, rising from 88.0 in 2018 to 118.0 in 2025, with its largest gains occurring between 2019 and 2022.\n\nFigure 24: Performance of Strong Innovators\n\n_Note: Performance is relative to that of the EU in 2018. The graph on the left shows the average performance of the Strong Innovators calculated as the unweighted average of the respective Member States. All vertical scales in Figure 23-Figure 26 span a range of 70.0%-points to allow an easy comparison of the results between the four performance groups._\n\n80\n\n90\n\n100\n\n110\n\n120\n\n130\n\n140\n\n150\n\n'18 '19 '20 '21 '22 '23 '24 '25\n\nSummary innovation index\n\nAustria\n\nBelgium\n\nGermany\n\nEstonia\n\nFrance\n\nIreland\n\nLuxembourg\n\n80\n\n90\n\n100\n\n110\n\n120\n\n130\n\n140\n\n150\n\n'18 '19 '20 '21 '22 '23 '24 '25\n\nSummary innovation index","pages":[{"line_numbers":[],"markdown":"62 European Innovation Scoreboard 2025\n\nBox 3: Country level observations of Strong Innovators\n\n**Ireland** demonstrated strong upward momentum, moving from 125.3% of the EU average in 2018 to 138.6% in 2025, with sustained growth particularly after 2020.\n\n**Belgium** consistently improved its score from 126.4 in 2018 to a high of 138.9 in 2024, with only a minor dip in 2025 (reaching 138.1).\n\n**Luxembourg** fluctuated over the period, peaking in 2019 (130.1), declining through 2024, and recovering to 128.9 in 2025.\n\n**Austria** saw gradual improvements in innovation performance from 2018 (120.1) to a peak in 2024 (132.3), followed by a slight decline in 2025 (128.3).\n\n**Germany** rose from 116.4 in 2018 to 127.8 in 2022, then stabilised around 127 before a slight drop to 125.1 in 2025.\n\n**France** experienced modest but steady progress, increasing from 114.9 in 2018 to 122.3 in 2025.\n\n**Estonia** recorded the strongest relative growth, rising from 88.0 in 2018 to 118.0 in 2025, with its largest gains occurring between 2019 and 2022.\n\nFigure 24: Performance of Strong Innovators\n\n_Note: Performance is relative to that of the EU in 2018. The graph on the left shows the average performance of the Strong Innovators calculated as the unweighted average of the respective Member States. All vertical scales in Figure 23-Figure 26 span a range of 70.0%-points to allow an easy comparison of the results between the four performance groups._\n\n80\n\n90\n\n100\n\n110\n\n120\n\n130\n\n140\n\n150\n\n'18 '19 '20 '21 '22 '23 '24 '25\n\nSummary innovation index\n\nAustria\n\nBelgium\n\nGermany\n\nEstonia\n\nFrance\n\nIreland\n\nLuxembourg\n\n80\n\n90\n\n100\n\n110\n\n120\n\n130\n\n140\n\n150\n\n'18 '19 '20 '21 '22 '23 '24 '25\n\nSummary innovation index","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"62 European Innovation Scoreboard 2025\n\nBox 3: Country level observations of Strong Innovators\n\n**Ireland** demonstrated strong upward momentum, moving from 125.3% of the EU average in 2018 to 138.6% in 2025, with sustained growth particularly after 2020.\n\n**Belgium** consistently improved its score from 126.4 in 2018 to a high of 138.9 in 2024, with only a minor dip in 2025 (reaching 138.1).\n\n**Luxembourg** fluctuated over the period, peaking in 2019 (130.1), declining through 2024, and recovering to 128.9 in 2025.\n\n**Austria** saw gradual improvements in innovation performance from 2018 (120.1) to a peak in 2024 (132.3), followed by a slight decline in 2025 (128.3).\n\n**Germany** rose from 116.4 in 2018 to 127.8 in 2022, then stabilised around 127 before a slight drop to 125.1 in 2025.\n\n**France** experienced modest but steady progress, increasing from 114.9 in 2018 to 122.3 in 2025.\n\n**Estonia** recorded the strongest relative growth, rising from 88.0 in 2018 to 118.0 in 2025, with its largest gains occurring between 2019 and 2022.\n\nFigure 24: Performance of Strong Innovators\n\n_Note: Performance is relative to that of the EU in 2018. The graph on the left shows the average performance of the Strong Innovators calculated as the unweighted average of the respective Member States. All vertical scales in Figure 23-Figure 26 span a range of 70.0%-points to allow an easy comparison of the results between the four performance groups._\n\n80\n\n90\n\n100\n\n110\n\n120\n\n130\n\n140\n\n150\n\n'18 '19 '20 '21 '22 '23 '24 '25\n\nSummary innovation index\n\nAustria\n\nBelgium\n\nGermany\n\nEstonia\n\nFrance\n\nIreland\n\nLuxembourg\n\n80\n\n90\n\n100\n\n110\n\n120\n\n130\n\n140\n\n150\n\n'18 '19 '20 '21 '22 '23 '24 '25\n\nSummary innovation index","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2025-EIS_p64.pdf"]},"response":{"errors":[],"extract_id":"extract_129dc46ead21673f251f795d67ac6c55","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["62 European Innovation Scoreboard 2025\n\nBox 3: Country level observations of Strong Innovators\n\n**Ireland** demonstrated strong upward momentum, moving from 125.3% of the EU average in 2018 to 138.6% in 2025, with sustained growth particularly after 2020.\n\n**Belgium** consistently improved its score from 126.4 in 2018 to a high of 138.9 in 2024, with only a minor dip in 2025 (reaching 138.1).\n\n**Luxembourg** fluctuated over the period, peaking in 2019 (130.1), declining through 2024, and recovering to 128.9 in 2025.\n\n**Austria** saw gradual improvements in innovation performance from 2018 (120.1) to a peak in 2024 (132.3), followed by a slight decline in 2025 (128.3).\n\n**Germany** rose from 116.4 in 2018 to 127.8 in 2022, then stabilised around 127 before a slight drop to 125.1 in 2025.\n\n**France** experienced modest but steady progress, increasing from 114.9 in 2018 to 122.3 in 2025.\n\n**Estonia** recorded the strongest relative growth, rising from 88.0 in 2018 to 118.0 in 2025, with its largest gains occurring between 2019 and 2022.\n\nFigure 24: Performance of Strong Innovators\n\n_Note: Performance is relative to that of the EU in 2018. The graph on the left shows the average performance of the Strong Innovators calculated as the unweighted average of the respective Member States. All vertical scales in Figure 23-Figure 26 span a range of 70.0%-points to allow an easy comparison of the results between the four performance groups._\n\n80\n\n90\n\n100\n\n110\n\n120\n\n130\n\n140\n\n150\n\n'18 '19 '20 '21 '22 '23 '24 '25\n\nSummary innovation index\n\nAustria\n\nBelgium\n\nGermany\n\nEstonia\n\nFrance\n\nIreland\n\nLuxembourg\n\n80\n\n90\n\n100\n\n110\n\n120\n\n130\n\n140\n\n150\n\n'18 '19 '20 '21 '22 '23 '24 '25\n\nSummary innovation index"],"full_content":"62 European Innovation Scoreboard 2025\n\nBox 3: Country level observations of Strong Innovators\n\n**Ireland** demonstrated strong upward momentum, moving from 125.3% of the EU average in 2018 to 138.6% in 2025, with sustained growth particularly after 2020.\n\n**Belgium** consistently improved its score from 126.4 in 2018 to a high of 138.9 in 2024, with only a minor dip in 2025 (reaching 138.1).\n\n**Luxembourg** fluctuated over the period, peaking in 2019 (130.1), declining through 2024, and recovering to 128.9 in 2025.\n\n**Austria** saw gradual improvements in innovation performance from 2018 (120.1) to a peak in 2024 (132.3), followed by a slight decline in 2025 (128.3).\n\n**Germany** rose from 116.4 in 2018 to 127.8 in 2022, then stabilised around 127 before a slight drop to 125.1 in 2025.\n\n**France** experienced modest but steady progress, increasing from 114.9 in 2018 to 122.3 in 2025.\n\n**Estonia** recorded the strongest relative growth, rising from 88.0 in 2018 to 118.0 in 2025, with its largest gains occurring between 2019 and 2022.\n\nFigure 24: Performance of Strong Innovators\n\n_Note: Performance is relative to that of the EU in 2018. The graph on the left shows the average performance of the Strong Innovators calculated as the unweighted average of the respective Member States. All vertical scales in Figure 23-Figure 26 span a range of 70.0%-points to allow an easy comparison of the results between the four performance groups._\n\n80\n\n90\n\n100\n\n110\n\n120\n\n130\n\n140\n\n150\n\n'18 '19 '20 '21 '22 '23 '24 '25\n\nSummary innovation index\n\nAustria\n\nBelgium\n\nGermany\n\nEstonia\n\nFrance\n\nIreland\n\nLuxembourg\n\n80\n\n90\n\n100\n\n110\n\n120\n\n130\n\n140\n\n150\n\n'18 '19 '20 '21 '22 '23 '24 '25\n\nSummary innovation index","publish_date":null,"title":"Box 3: Country level observations of Strong Innovators demonstrated strong upward momentum, moving from 125.3% of the EU Ireland average","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2025-EIS_p64.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2025-EIS_p64.pdf"},"request":{"example_id":"chart/2025-EIS_p64","product_type":"parse","source_file_path":"2025-EIS_p64.pdf"},"started_at":"2026-11-04T19:45:35.241069"} +{"completed_at":"2026-11-04T19:45:37.457910","latency_in_ms":1156,"output":{"example_id":"chart/2025-EIS_p67","grounded_pages":[],"layout_pages":[],"markdown":"65 European Innovation Scoreboard 2025\n\n(reaching 89.5) before rebounding.\n\n**Cyprus** recorded a sharp rise between 2020 and 2021 (from 80.3 to 104.6), peaking in 2023 (109.9), before falling to 94.7 in 2025.\n\n**Lithuania** improved from 73.8 in 2018 to 94.7 in 2024, with a small decline to 91.2 in 2025.\n\n**Czechia** increased from 74.4 in 2018 to 99.1 by 2023–2024, followed by a drop to 90.8 in 2025.\n\n**Greece** experienced consistent growth between 2018 (70.0) and 2024 (88.2), with a slight dip to 85.3 in 2025.\n\n**Croatia** steadily improved its score from 61.2 in 2018 to a high of 81.6 in 2024, followed by a slight decline in 2025 (reaching 80.6).\n\nFigure 25: Performance Moderate Innovators\n\n_Note: Performance is relative to that of the EU in 2018. The graph on the left shows the average performance of the Moderate Innovators calculated as the unweighted average of the respective Member States. All vertical scales in Figure 23-Figure 26 span a range of 70.0%-points to allow an easy comparison of the results between the four performance groups._\n\n**Emerging Innovators**\n\nIn 2025, Hungary (at 78.3% of the EU average in 2018) was the highest-performing emerging innovator and only marginally below the threshold of the moderate innovators, reflecting strong recent momentum. Poland (74.2%) and Slovakia (70.5%) followed closely,\n\n50\n\n60\n\n70\n\n80\n\n90\n\n100\n\n110\n\n120\n\n'18 '19 '20 '21 '22 '23 '24 '25\n\nSummary innovation index\n\nCyprus\n\nCzechia Greece\n\nSpain\n\nCroatia\n\nItaly\n\nLithuania\n\nMalta\n\nPortugal\n\nSlovenia\n\n50\n\n60\n\n70\n\n80\n\n90\n\n100\n\n110\n\n120\n\n'18 '19 '20 '21 '22 '23 '24 '25\n\nSummary innovation index","pages":[{"line_numbers":[],"markdown":"65 European Innovation Scoreboard 2025\n\n(reaching 89.5) before rebounding.\n\n**Cyprus** recorded a sharp rise between 2020 and 2021 (from 80.3 to 104.6), peaking in 2023 (109.9), before falling to 94.7 in 2025.\n\n**Lithuania** improved from 73.8 in 2018 to 94.7 in 2024, with a small decline to 91.2 in 2025.\n\n**Czechia** increased from 74.4 in 2018 to 99.1 by 2023–2024, followed by a drop to 90.8 in 2025.\n\n**Greece** experienced consistent growth between 2018 (70.0) and 2024 (88.2), with a slight dip to 85.3 in 2025.\n\n**Croatia** steadily improved its score from 61.2 in 2018 to a high of 81.6 in 2024, followed by a slight decline in 2025 (reaching 80.6).\n\nFigure 25: Performance Moderate Innovators\n\n_Note: Performance is relative to that of the EU in 2018. The graph on the left shows the average performance of the Moderate Innovators calculated as the unweighted average of the respective Member States. All vertical scales in Figure 23-Figure 26 span a range of 70.0%-points to allow an easy comparison of the results between the four performance groups._\n\n**Emerging Innovators**\n\nIn 2025, Hungary (at 78.3% of the EU average in 2018) was the highest-performing emerging innovator and only marginally below the threshold of the moderate innovators, reflecting strong recent momentum. Poland (74.2%) and Slovakia (70.5%) followed closely,\n\n50\n\n60\n\n70\n\n80\n\n90\n\n100\n\n110\n\n120\n\n'18 '19 '20 '21 '22 '23 '24 '25\n\nSummary innovation index\n\nCyprus\n\nCzechia Greece\n\nSpain\n\nCroatia\n\nItaly\n\nLithuania\n\nMalta\n\nPortugal\n\nSlovenia\n\n50\n\n60\n\n70\n\n80\n\n90\n\n100\n\n110\n\n120\n\n'18 '19 '20 '21 '22 '23 '24 '25\n\nSummary innovation index","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"65 European Innovation Scoreboard 2025\n\n(reaching 89.5) before rebounding.\n\n**Cyprus** recorded a sharp rise between 2020 and 2021 (from 80.3 to 104.6), peaking in 2023 (109.9), before falling to 94.7 in 2025.\n\n**Lithuania** improved from 73.8 in 2018 to 94.7 in 2024, with a small decline to 91.2 in 2025.\n\n**Czechia** increased from 74.4 in 2018 to 99.1 by 2023–2024, followed by a drop to 90.8 in 2025.\n\n**Greece** experienced consistent growth between 2018 (70.0) and 2024 (88.2), with a slight dip to 85.3 in 2025.\n\n**Croatia** steadily improved its score from 61.2 in 2018 to a high of 81.6 in 2024, followed by a slight decline in 2025 (reaching 80.6).\n\nFigure 25: Performance Moderate Innovators\n\n_Note: Performance is relative to that of the EU in 2018. The graph on the left shows the average performance of the Moderate Innovators calculated as the unweighted average of the respective Member States. All vertical scales in Figure 23-Figure 26 span a range of 70.0%-points to allow an easy comparison of the results between the four performance groups._\n\n**Emerging Innovators**\n\nIn 2025, Hungary (at 78.3% of the EU average in 2018) was the highest-performing emerging innovator and only marginally below the threshold of the moderate innovators, reflecting strong recent momentum. Poland (74.2%) and Slovakia (70.5%) followed closely,\n\n50\n\n60\n\n70\n\n80\n\n90\n\n100\n\n110\n\n120\n\n'18 '19 '20 '21 '22 '23 '24 '25\n\nSummary innovation index\n\nCyprus\n\nCzechia Greece\n\nSpain\n\nCroatia\n\nItaly\n\nLithuania\n\nMalta\n\nPortugal\n\nSlovenia\n\n50\n\n60\n\n70\n\n80\n\n90\n\n100\n\n110\n\n120\n\n'18 '19 '20 '21 '22 '23 '24 '25\n\nSummary innovation index","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2025-EIS_p67.pdf"]},"response":{"errors":[],"extract_id":"extract_ecd23d06ced85241db721db0136916e3","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["65 European Innovation Scoreboard 2025\n\n(reaching 89.5) before rebounding.\n\n**Cyprus** recorded a sharp rise between 2020 and 2021 (from 80.3 to 104.6), peaking in 2023 (109.9), before falling to 94.7 in 2025.\n\n**Lithuania** improved from 73.8 in 2018 to 94.7 in 2024, with a small decline to 91.2 in 2025.\n\n**Czechia** increased from 74.4 in 2018 to 99.1 by 2023–2024, followed by a drop to 90.8 in 2025.\n\n**Greece** experienced consistent growth between 2018 (70.0) and 2024 (88.2), with a slight dip to 85.3 in 2025.\n\n**Croatia** steadily improved its score from 61.2 in 2018 to a high of 81.6 in 2024, followed by a slight decline in 2025 (reaching 80.6).\n\nFigure 25: Performance Moderate Innovators\n\n_Note: Performance is relative to that of the EU in 2018. The graph on the left shows the average performance of the Moderate Innovators calculated as the unweighted average of the respective Member States. All vertical scales in Figure 23-Figure 26 span a range of 70.0%-points to allow an easy comparison of the results between the four performance groups._\n\n**Emerging Innovators**\n\nIn 2025, Hungary (at 78.3% of the EU average in 2018) was the highest-performing emerging innovator and only marginally below the threshold of the moderate innovators, reflecting strong recent momentum. Poland (74.2%) and Slovakia (70.5%) followed closely,\n\n50\n\n60\n\n70\n\n80\n\n90\n\n100\n\n110\n\n120\n\n'18 '19 '20 '21 '22 '23 '24 '25\n\nSummary innovation index\n\nCyprus\n\nCzechia Greece\n\nSpain\n\nCroatia\n\nItaly\n\nLithuania\n\nMalta\n\nPortugal\n\nSlovenia\n\n50\n\n60\n\n70\n\n80\n\n90\n\n100\n\n110\n\n120\n\n'18 '19 '20 '21 '22 '23 '24 '25\n\nSummary innovation index"],"full_content":"65 European Innovation Scoreboard 2025\n\n(reaching 89.5) before rebounding.\n\n**Cyprus** recorded a sharp rise between 2020 and 2021 (from 80.3 to 104.6), peaking in 2023 (109.9), before falling to 94.7 in 2025.\n\n**Lithuania** improved from 73.8 in 2018 to 94.7 in 2024, with a small decline to 91.2 in 2025.\n\n**Czechia** increased from 74.4 in 2018 to 99.1 by 2023–2024, followed by a drop to 90.8 in 2025.\n\n**Greece** experienced consistent growth between 2018 (70.0) and 2024 (88.2), with a slight dip to 85.3 in 2025.\n\n**Croatia** steadily improved its score from 61.2 in 2018 to a high of 81.6 in 2024, followed by a slight decline in 2025 (reaching 80.6).\n\nFigure 25: Performance Moderate Innovators\n\n_Note: Performance is relative to that of the EU in 2018. The graph on the left shows the average performance of the Moderate Innovators calculated as the unweighted average of the respective Member States. All vertical scales in Figure 23-Figure 26 span a range of 70.0%-points to allow an easy comparison of the results between the four performance groups._\n\n**Emerging Innovators**\n\nIn 2025, Hungary (at 78.3% of the EU average in 2018) was the highest-performing emerging innovator and only marginally below the threshold of the moderate innovators, reflecting strong recent momentum. Poland (74.2%) and Slovakia (70.5%) followed closely,\n\n50\n\n60\n\n70\n\n80\n\n90\n\n100\n\n110\n\n120\n\n'18 '19 '20 '21 '22 '23 '24 '25\n\nSummary innovation index\n\nCyprus\n\nCzechia Greece\n\nSpain\n\nCroatia\n\nItaly\n\nLithuania\n\nMalta\n\nPortugal\n\nSlovenia\n\n50\n\n60\n\n70\n\n80\n\n90\n\n100\n\n110\n\n120\n\n'18 '19 '20 '21 '22 '23 '24 '25\n\nSummary innovation index","publish_date":null,"title":"(reaching 89.5) before rebounding. recorded a sharp rise between 2020 and 2021 (from 80.3 to 104.6), peaking in Cyprus 2023","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2025-EIS_p67.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2025-EIS_p67.pdf"},"request":{"example_id":"chart/2025-EIS_p67","product_type":"parse","source_file_path":"2025-EIS_p67.pdf"},"started_at":"2026-11-04T19:45:36.301756"} +{"completed_at":"2026-11-04T19:45:42.023298","latency_in_ms":4025,"output":{"example_id":"chart/2025-EIS_p69","grounded_pages":[],"layout_pages":[],"markdown":"67 European Innovation Scoreboard 2025\n\nSlovakia (-17.8%-points), and Romania (-14.2%-points) all saw more moderate but consistent drop.\n\nBox 5: Country level observations for Emerging Innovators\n\n**Hungary** showed strong and continuous progress, rising from 62.1%-of the EU average in 2018 to 78.3% in 2025, with marked gains after 2021.\n\n**Poland** recorded consistent growth over the period, improving from 56.1 in 2018 to 74.2 in 2025, with notable acceleration after 2022.\n\n**Slovakia** moved from 62.2 in 2018 to a high of 71.6 in 2024, before a modest drop to 70.5 in 2025.\n\n**Latvia** remained relatively stable, increasing gradually from 59.0 in 2018 to 63.9 in 2025, despite a small dip in 2021.\n\n**Bulgaria** improved steadily from 45.2 in 2018 to a peak of 53.8 in 2024, before a slight decline to 51.6 in 2025.\n\n**Romania** progressed slowly, rising from 34.2 in 2018 to 42.4 in 2025, despite a slight dip in 2023.\n\nFigure 26: Performance Emerging Innovators\n\n_Note: Performance is relative to that of the EU in 2018. The graph on the left shows the average performance of the Emerging Innovators calculated as the unweighted average of the respective Member States. All vertical scales in Figure 23-Figure 26 span a range of 70.0%-points to allow an easy comparison of the results between the four performance groups._\n\n20\n\n30\n\n40\n\n50\n\n60\n\n70\n\n80\n\n90\n\n'18 '19 '20 '21 '22 '23 '24 '25\n\nSummary innovation index\n\nBulgaria\n\nHungary\n\nLatvia\n\nPoland\n\nRomania\n\nSlovakia\n\n20\n\n30\n\n40\n\n50\n\n60\n\n70\n\n80\n\n90\n\n'18 '19 '20 '21 '22 '23 '24 '25\n\nSummary innovation index","pages":[{"line_numbers":[],"markdown":"67 European Innovation Scoreboard 2025\n\nSlovakia (-17.8%-points), and Romania (-14.2%-points) all saw more moderate but consistent drop.\n\nBox 5: Country level observations for Emerging Innovators\n\n**Hungary** showed strong and continuous progress, rising from 62.1%-of the EU average in 2018 to 78.3% in 2025, with marked gains after 2021.\n\n**Poland** recorded consistent growth over the period, improving from 56.1 in 2018 to 74.2 in 2025, with notable acceleration after 2022.\n\n**Slovakia** moved from 62.2 in 2018 to a high of 71.6 in 2024, before a modest drop to 70.5 in 2025.\n\n**Latvia** remained relatively stable, increasing gradually from 59.0 in 2018 to 63.9 in 2025, despite a small dip in 2021.\n\n**Bulgaria** improved steadily from 45.2 in 2018 to a peak of 53.8 in 2024, before a slight decline to 51.6 in 2025.\n\n**Romania** progressed slowly, rising from 34.2 in 2018 to 42.4 in 2025, despite a slight dip in 2023.\n\nFigure 26: Performance Emerging Innovators\n\n_Note: Performance is relative to that of the EU in 2018. The graph on the left shows the average performance of the Emerging Innovators calculated as the unweighted average of the respective Member States. All vertical scales in Figure 23-Figure 26 span a range of 70.0%-points to allow an easy comparison of the results between the four performance groups._\n\n20\n\n30\n\n40\n\n50\n\n60\n\n70\n\n80\n\n90\n\n'18 '19 '20 '21 '22 '23 '24 '25\n\nSummary innovation index\n\nBulgaria\n\nHungary\n\nLatvia\n\nPoland\n\nRomania\n\nSlovakia\n\n20\n\n30\n\n40\n\n50\n\n60\n\n70\n\n80\n\n90\n\n'18 '19 '20 '21 '22 '23 '24 '25\n\nSummary innovation index","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"67 European Innovation Scoreboard 2025\n\nSlovakia (-17.8%-points), and Romania (-14.2%-points) all saw more moderate but consistent drop.\n\nBox 5: Country level observations for Emerging Innovators\n\n**Hungary** showed strong and continuous progress, rising from 62.1%-of the EU average in 2018 to 78.3% in 2025, with marked gains after 2021.\n\n**Poland** recorded consistent growth over the period, improving from 56.1 in 2018 to 74.2 in 2025, with notable acceleration after 2022.\n\n**Slovakia** moved from 62.2 in 2018 to a high of 71.6 in 2024, before a modest drop to 70.5 in 2025.\n\n**Latvia** remained relatively stable, increasing gradually from 59.0 in 2018 to 63.9 in 2025, despite a small dip in 2021.\n\n**Bulgaria** improved steadily from 45.2 in 2018 to a peak of 53.8 in 2024, before a slight decline to 51.6 in 2025.\n\n**Romania** progressed slowly, rising from 34.2 in 2018 to 42.4 in 2025, despite a slight dip in 2023.\n\nFigure 26: Performance Emerging Innovators\n\n_Note: Performance is relative to that of the EU in 2018. The graph on the left shows the average performance of the Emerging Innovators calculated as the unweighted average of the respective Member States. All vertical scales in Figure 23-Figure 26 span a range of 70.0%-points to allow an easy comparison of the results between the four performance groups._\n\n20\n\n30\n\n40\n\n50\n\n60\n\n70\n\n80\n\n90\n\n'18 '19 '20 '21 '22 '23 '24 '25\n\nSummary innovation index\n\nBulgaria\n\nHungary\n\nLatvia\n\nPoland\n\nRomania\n\nSlovakia\n\n20\n\n30\n\n40\n\n50\n\n60\n\n70\n\n80\n\n90\n\n'18 '19 '20 '21 '22 '23 '24 '25\n\nSummary innovation index","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2025-EIS_p69.pdf"]},"response":{"errors":[],"extract_id":"extract_5bc93dc7128116e017f73b23bee35d83","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["67 European Innovation Scoreboard 2025\n\nSlovakia (-17.8%-points), and Romania (-14.2%-points) all saw more moderate but consistent drop.\n\nBox 5: Country level observations for Emerging Innovators\n\n**Hungary** showed strong and continuous progress, rising from 62.1%-of the EU average in 2018 to 78.3% in 2025, with marked gains after 2021.\n\n**Poland** recorded consistent growth over the period, improving from 56.1 in 2018 to 74.2 in 2025, with notable acceleration after 2022.\n\n**Slovakia** moved from 62.2 in 2018 to a high of 71.6 in 2024, before a modest drop to 70.5 in 2025.\n\n**Latvia** remained relatively stable, increasing gradually from 59.0 in 2018 to 63.9 in 2025, despite a small dip in 2021.\n\n**Bulgaria** improved steadily from 45.2 in 2018 to a peak of 53.8 in 2024, before a slight decline to 51.6 in 2025.\n\n**Romania** progressed slowly, rising from 34.2 in 2018 to 42.4 in 2025, despite a slight dip in 2023.\n\nFigure 26: Performance Emerging Innovators\n\n_Note: Performance is relative to that of the EU in 2018. The graph on the left shows the average performance of the Emerging Innovators calculated as the unweighted average of the respective Member States. All vertical scales in Figure 23-Figure 26 span a range of 70.0%-points to allow an easy comparison of the results between the four performance groups._\n\n20\n\n30\n\n40\n\n50\n\n60\n\n70\n\n80\n\n90\n\n'18 '19 '20 '21 '22 '23 '24 '25\n\nSummary innovation index\n\nBulgaria\n\nHungary\n\nLatvia\n\nPoland\n\nRomania\n\nSlovakia\n\n20\n\n30\n\n40\n\n50\n\n60\n\n70\n\n80\n\n90\n\n'18 '19 '20 '21 '22 '23 '24 '25\n\nSummary innovation index"],"full_content":"67 European Innovation Scoreboard 2025\n\nSlovakia (-17.8%-points), and Romania (-14.2%-points) all saw more moderate but consistent drop.\n\nBox 5: Country level observations for Emerging Innovators\n\n**Hungary** showed strong and continuous progress, rising from 62.1%-of the EU average in 2018 to 78.3% in 2025, with marked gains after 2021.\n\n**Poland** recorded consistent growth over the period, improving from 56.1 in 2018 to 74.2 in 2025, with notable acceleration after 2022.\n\n**Slovakia** moved from 62.2 in 2018 to a high of 71.6 in 2024, before a modest drop to 70.5 in 2025.\n\n**Latvia** remained relatively stable, increasing gradually from 59.0 in 2018 to 63.9 in 2025, despite a small dip in 2021.\n\n**Bulgaria** improved steadily from 45.2 in 2018 to a peak of 53.8 in 2024, before a slight decline to 51.6 in 2025.\n\n**Romania** progressed slowly, rising from 34.2 in 2018 to 42.4 in 2025, despite a slight dip in 2023.\n\nFigure 26: Performance Emerging Innovators\n\n_Note: Performance is relative to that of the EU in 2018. The graph on the left shows the average performance of the Emerging Innovators calculated as the unweighted average of the respective Member States. All vertical scales in Figure 23-Figure 26 span a range of 70.0%-points to allow an easy comparison of the results between the four performance groups._\n\n20\n\n30\n\n40\n\n50\n\n60\n\n70\n\n80\n\n90\n\n'18 '19 '20 '21 '22 '23 '24 '25\n\nSummary innovation index\n\nBulgaria\n\nHungary\n\nLatvia\n\nPoland\n\nRomania\n\nSlovakia\n\n20\n\n30\n\n40\n\n50\n\n60\n\n70\n\n80\n\n90\n\n'18 '19 '20 '21 '22 '23 '24 '25\n\nSummary innovation index","publish_date":null,"title":"Slovakia (-17.8%-points), and Romania (-14.2%-points) all saw more moderate but consistent drop. Box 5: Country level observations for Emerging Innovators","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2025-EIS_p69.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2025-EIS_p69.pdf"},"request":{"example_id":"chart/2025-EIS_p69","product_type":"parse","source_file_path":"2025-EIS_p69.pdf"},"started_at":"2026-11-04T19:45:37.998066"} +{"completed_at":"2026-11-04T19:45:41.532616","latency_in_ms":2222,"output":{"example_id":"chart/2025-EIS_p71","grounded_pages":[],"layout_pages":[],"markdown":"69 European Innovation Scoreboard 2025\n\nFigure 27: Convergence in innovation performance\n\n_Note: Lines show the coefficient of variation of the summary innovation index. defined as the ratio of the standard deviation to the mean, indexed to 2018._\n\n3\\.5 How do EU27 Member States perform compared to neighbouring countries? The innovation performance of the 12 neighbouring (non-EU) European countries (Albania, Bosnia and Herzegovina, Iceland, Moldova, Montenegro, North Macedonia, Norway, Serbia, Switzerland, Türkiye, Ukraine and the United Kingdom) compared with that of the EU27 Member States is examined in this section. The performance progress concerns only the indicators for which data is available as data availability varies across countries. The neighbouring countries are strategic partners for the EU, with a mix of highly advanced economies and economies in transition, some of which are on the path to EU accession.\n\nFour of the neighbouring countries score above the EU average in 2025, namely Switzerland, the UK, Norway, and Iceland. Switzerland ranks first in the enlarged ranking outperforming all EU27 Member States in 2025. However, its innovation performance has seen a slight decline, the country recording a 1.6%-point decrease compared to 2024 and an overall 1.7%-point decrease in performance since 2018. The United Kingdom outperforms 24 EU27 Member States with an improvement of 1.2%-point in 2025 marking a breakthrough into the Innovation Leaders group. The UK’s performance reflects a long-standing positive trend with a 12.2%-point increase in performance since 2018 ³¹ . Norway ranks third among Strong Innovators group but registers a 0.9%-point decrease compared to 2024. Performing at 121.6% of the EU average in 2025, it builds on a positive trend since 2018, with a strong 13%-point increase over the period. Finally, Iceland ranks sixth in the Strong Innovators\n\n31 For the United Kingdom, data from 2020 is used, based on an earlier edition of the Community Innovation Survey (CIS).\n\n0\\.0\n\n0\\.2\n\n0\\.4\n\n0\\.6\n\n0\\.8\n\n1\\.0\n\n1\\.2\n\n1\\.4\n\n1\\.6\n\n1\\.8\n\n2018 2019 2020 2021 2022 2023 2024 2025\n\nCoefficient of variation\n\nAll EU Member States\n\nEmerging innovators\n\nModerate innovators\n\nStrong innovators\n\nInnovation leaders","pages":[{"line_numbers":[],"markdown":"69 European Innovation Scoreboard 2025\n\nFigure 27: Convergence in innovation performance\n\n_Note: Lines show the coefficient of variation of the summary innovation index. defined as the ratio of the standard deviation to the mean, indexed to 2018._\n\n3\\.5 How do EU27 Member States perform compared to neighbouring countries? The innovation performance of the 12 neighbouring (non-EU) European countries (Albania, Bosnia and Herzegovina, Iceland, Moldova, Montenegro, North Macedonia, Norway, Serbia, Switzerland, Türkiye, Ukraine and the United Kingdom) compared with that of the EU27 Member States is examined in this section. The performance progress concerns only the indicators for which data is available as data availability varies across countries. The neighbouring countries are strategic partners for the EU, with a mix of highly advanced economies and economies in transition, some of which are on the path to EU accession.\n\nFour of the neighbouring countries score above the EU average in 2025, namely Switzerland, the UK, Norway, and Iceland. Switzerland ranks first in the enlarged ranking outperforming all EU27 Member States in 2025. However, its innovation performance has seen a slight decline, the country recording a 1.6%-point decrease compared to 2024 and an overall 1.7%-point decrease in performance since 2018. The United Kingdom outperforms 24 EU27 Member States with an improvement of 1.2%-point in 2025 marking a breakthrough into the Innovation Leaders group. The UK’s performance reflects a long-standing positive trend with a 12.2%-point increase in performance since 2018 ³¹ . Norway ranks third among Strong Innovators group but registers a 0.9%-point decrease compared to 2024. Performing at 121.6% of the EU average in 2025, it builds on a positive trend since 2018, with a strong 13%-point increase over the period. Finally, Iceland ranks sixth in the Strong Innovators\n\n31 For the United Kingdom, data from 2020 is used, based on an earlier edition of the Community Innovation Survey (CIS).\n\n0\\.0\n\n0\\.2\n\n0\\.4\n\n0\\.6\n\n0\\.8\n\n1\\.0\n\n1\\.2\n\n1\\.4\n\n1\\.6\n\n1\\.8\n\n2018 2019 2020 2021 2022 2023 2024 2025\n\nCoefficient of variation\n\nAll EU Member States\n\nEmerging innovators\n\nModerate innovators\n\nStrong innovators\n\nInnovation leaders","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"69 European Innovation Scoreboard 2025\n\nFigure 27: Convergence in innovation performance\n\n_Note: Lines show the coefficient of variation of the summary innovation index. defined as the ratio of the standard deviation to the mean, indexed to 2018._\n\n3\\.5 How do EU27 Member States perform compared to neighbouring countries? The innovation performance of the 12 neighbouring (non-EU) European countries (Albania, Bosnia and Herzegovina, Iceland, Moldova, Montenegro, North Macedonia, Norway, Serbia, Switzerland, Türkiye, Ukraine and the United Kingdom) compared with that of the EU27 Member States is examined in this section. The performance progress concerns only the indicators for which data is available as data availability varies across countries. The neighbouring countries are strategic partners for the EU, with a mix of highly advanced economies and economies in transition, some of which are on the path to EU accession.\n\nFour of the neighbouring countries score above the EU average in 2025, namely Switzerland, the UK, Norway, and Iceland. Switzerland ranks first in the enlarged ranking outperforming all EU27 Member States in 2025. However, its innovation performance has seen a slight decline, the country recording a 1.6%-point decrease compared to 2024 and an overall 1.7%-point decrease in performance since 2018. The United Kingdom outperforms 24 EU27 Member States with an improvement of 1.2%-point in 2025 marking a breakthrough into the Innovation Leaders group. The UK’s performance reflects a long-standing positive trend with a 12.2%-point increase in performance since 2018 ³¹ . Norway ranks third among Strong Innovators group but registers a 0.9%-point decrease compared to 2024. Performing at 121.6% of the EU average in 2025, it builds on a positive trend since 2018, with a strong 13%-point increase over the period. Finally, Iceland ranks sixth in the Strong Innovators\n\n31 For the United Kingdom, data from 2020 is used, based on an earlier edition of the Community Innovation Survey (CIS).\n\n0\\.0\n\n0\\.2\n\n0\\.4\n\n0\\.6\n\n0\\.8\n\n1\\.0\n\n1\\.2\n\n1\\.4\n\n1\\.6\n\n1\\.8\n\n2018 2019 2020 2021 2022 2023 2024 2025\n\nCoefficient of variation\n\nAll EU Member States\n\nEmerging innovators\n\nModerate innovators\n\nStrong innovators\n\nInnovation leaders","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2025-EIS_p71.pdf"]},"response":{"errors":[],"extract_id":"extract_e7e233aa024a02ad655101796ec5b5e5","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["69 European Innovation Scoreboard 2025\n\nFigure 27: Convergence in innovation performance\n\n_Note: Lines show the coefficient of variation of the summary innovation index. defined as the ratio of the standard deviation to the mean, indexed to 2018._\n\n3\\.5 How do EU27 Member States perform compared to neighbouring countries? The innovation performance of the 12 neighbouring (non-EU) European countries (Albania, Bosnia and Herzegovina, Iceland, Moldova, Montenegro, North Macedonia, Norway, Serbia, Switzerland, Türkiye, Ukraine and the United Kingdom) compared with that of the EU27 Member States is examined in this section. The performance progress concerns only the indicators for which data is available as data availability varies across countries. The neighbouring countries are strategic partners for the EU, with a mix of highly advanced economies and economies in transition, some of which are on the path to EU accession.\n\nFour of the neighbouring countries score above the EU average in 2025, namely Switzerland, the UK, Norway, and Iceland. Switzerland ranks first in the enlarged ranking outperforming all EU27 Member States in 2025. However, its innovation performance has seen a slight decline, the country recording a 1.6%-point decrease compared to 2024 and an overall 1.7%-point decrease in performance since 2018. The United Kingdom outperforms 24 EU27 Member States with an improvement of 1.2%-point in 2025 marking a breakthrough into the Innovation Leaders group. The UK’s performance reflects a long-standing positive trend with a 12.2%-point increase in performance since 2018 ³¹ . Norway ranks third among Strong Innovators group but registers a 0.9%-point decrease compared to 2024. Performing at 121.6% of the EU average in 2025, it builds on a positive trend since 2018, with a strong 13%-point increase over the period. Finally, Iceland ranks sixth in the Strong Innovators\n\n31 For the United Kingdom, data from 2020 is used, based on an earlier edition of the Community Innovation Survey (CIS).\n\n0\\.0\n\n0\\.2\n\n0\\.4\n\n0\\.6\n\n0\\.8\n\n1\\.0\n\n1\\.2\n\n1\\.4\n\n1\\.6\n\n1\\.8\n\n2018 2019 2020 2021 2022 2023 2024 2025\n\nCoefficient of variation\n\nAll EU Member States\n\nEmerging innovators\n\nModerate innovators\n\nStrong innovators\n\nInnovation leaders"],"full_content":"69 European Innovation Scoreboard 2025\n\nFigure 27: Convergence in innovation performance\n\n_Note: Lines show the coefficient of variation of the summary innovation index. defined as the ratio of the standard deviation to the mean, indexed to 2018._\n\n3\\.5 How do EU27 Member States perform compared to neighbouring countries? The innovation performance of the 12 neighbouring (non-EU) European countries (Albania, Bosnia and Herzegovina, Iceland, Moldova, Montenegro, North Macedonia, Norway, Serbia, Switzerland, Türkiye, Ukraine and the United Kingdom) compared with that of the EU27 Member States is examined in this section. The performance progress concerns only the indicators for which data is available as data availability varies across countries. The neighbouring countries are strategic partners for the EU, with a mix of highly advanced economies and economies in transition, some of which are on the path to EU accession.\n\nFour of the neighbouring countries score above the EU average in 2025, namely Switzerland, the UK, Norway, and Iceland. Switzerland ranks first in the enlarged ranking outperforming all EU27 Member States in 2025. However, its innovation performance has seen a slight decline, the country recording a 1.6%-point decrease compared to 2024 and an overall 1.7%-point decrease in performance since 2018. The United Kingdom outperforms 24 EU27 Member States with an improvement of 1.2%-point in 2025 marking a breakthrough into the Innovation Leaders group. The UK’s performance reflects a long-standing positive trend with a 12.2%-point increase in performance since 2018 ³¹ . Norway ranks third among Strong Innovators group but registers a 0.9%-point decrease compared to 2024. Performing at 121.6% of the EU average in 2025, it builds on a positive trend since 2018, with a strong 13%-point increase over the period. Finally, Iceland ranks sixth in the Strong Innovators\n\n31 For the United Kingdom, data from 2020 is used, based on an earlier edition of the Community Innovation Survey (CIS).\n\n0\\.0\n\n0\\.2\n\n0\\.4\n\n0\\.6\n\n0\\.8\n\n1\\.0\n\n1\\.2\n\n1\\.4\n\n1\\.6\n\n1\\.8\n\n2018 2019 2020 2021 2022 2023 2024 2025\n\nCoefficient of variation\n\nAll EU Member States\n\nEmerging innovators\n\nModerate innovators\n\nStrong innovators\n\nInnovation leaders","publish_date":null,"title":"Figure 27: Convergence in innovation performance 3.5 How do EU27 Member States perform compared to neighbouring countries? The innovation performance","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2025-EIS_p71.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2025-EIS_p71.pdf"},"request":{"example_id":"chart/2025-EIS_p71","product_type":"parse","source_file_path":"2025-EIS_p71.pdf"},"started_at":"2026-11-04T19:45:39.310010"} +{"completed_at":"2026-11-04T19:45:42.166077","latency_in_ms":1025,"output":{"example_id":"chart/2025-EIS_p86","grounded_pages":[],"layout_pages":[],"markdown":"84 European Innovation Scoreboard 2025\n\nFigure 33: Innovation performance of the EU and its global competitors\n\n_Note: All performance scores are relative to that of the EU in 2018. Coloured bars show countries’ performance in 2025 while the vertical bars show performance in 2024. Due to limited data availability for global competitors, scores are calculated using a smaller set of indicators._\n\nCompared to 2024, most global competitors have improved their innovation performance. Only Canada, the EU, and South Africa recorded a decline. South Korea shows the largest annual improvement (+6.3%-points), followed by China (+6.0%-points) and Australia (+3.8%-points). The gap between Australia and the United States is also narrowing.\n\nOver the longer term (Figure 34), all global competitors have improved their innovation performance between 2018 and 2025. Notably, the five countries that now outperform the EU have all recorded faster growth than the EU itself, thereby widening the performance gap. Among the six countries still lagging the EU, only Chile has increased its performance at a higher rate than the EU since 2018, with a gain of 15.5%-points.\n\nI\n\nI\n\nI\n\nI\n\nI\n\nI\n\nI\n\nI\n\nI\n\nI\n\nI\n\nI Mexico\n\nSouth Africa\n\nChile\n\nIndia\n\nBrazil\n\nJapan\n\nEU\n\nAustralia\n\nUnited States\n\nChina\n\nCanada\n\nSouth Korea\n\n0 20 40 60 80 100 120 140 160\n\nSummary innovation index in 2025 (indexed to the EU in 2018)\n\nEmerging innovators\n\nModerate innovators\n\nStrong innovators\n\nInnovation leaders\n\nI Score in 2024","pages":[{"line_numbers":[],"markdown":"84 European Innovation Scoreboard 2025\n\nFigure 33: Innovation performance of the EU and its global competitors\n\n_Note: All performance scores are relative to that of the EU in 2018. Coloured bars show countries’ performance in 2025 while the vertical bars show performance in 2024. Due to limited data availability for global competitors, scores are calculated using a smaller set of indicators._\n\nCompared to 2024, most global competitors have improved their innovation performance. Only Canada, the EU, and South Africa recorded a decline. South Korea shows the largest annual improvement (+6.3%-points), followed by China (+6.0%-points) and Australia (+3.8%-points). The gap between Australia and the United States is also narrowing.\n\nOver the longer term (Figure 34), all global competitors have improved their innovation performance between 2018 and 2025. Notably, the five countries that now outperform the EU have all recorded faster growth than the EU itself, thereby widening the performance gap. Among the six countries still lagging the EU, only Chile has increased its performance at a higher rate than the EU since 2018, with a gain of 15.5%-points.\n\nI\n\nI\n\nI\n\nI\n\nI\n\nI\n\nI\n\nI\n\nI\n\nI\n\nI\n\nI Mexico\n\nSouth Africa\n\nChile\n\nIndia\n\nBrazil\n\nJapan\n\nEU\n\nAustralia\n\nUnited States\n\nChina\n\nCanada\n\nSouth Korea\n\n0 20 40 60 80 100 120 140 160\n\nSummary innovation index in 2025 (indexed to the EU in 2018)\n\nEmerging innovators\n\nModerate innovators\n\nStrong innovators\n\nInnovation leaders\n\nI Score in 2024","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"84 European Innovation Scoreboard 2025\n\nFigure 33: Innovation performance of the EU and its global competitors\n\n_Note: All performance scores are relative to that of the EU in 2018. Coloured bars show countries’ performance in 2025 while the vertical bars show performance in 2024. Due to limited data availability for global competitors, scores are calculated using a smaller set of indicators._\n\nCompared to 2024, most global competitors have improved their innovation performance. Only Canada, the EU, and South Africa recorded a decline. South Korea shows the largest annual improvement (+6.3%-points), followed by China (+6.0%-points) and Australia (+3.8%-points). The gap between Australia and the United States is also narrowing.\n\nOver the longer term (Figure 34), all global competitors have improved their innovation performance between 2018 and 2025. Notably, the five countries that now outperform the EU have all recorded faster growth than the EU itself, thereby widening the performance gap. Among the six countries still lagging the EU, only Chile has increased its performance at a higher rate than the EU since 2018, with a gain of 15.5%-points.\n\nI\n\nI\n\nI\n\nI\n\nI\n\nI\n\nI\n\nI\n\nI\n\nI\n\nI\n\nI Mexico\n\nSouth Africa\n\nChile\n\nIndia\n\nBrazil\n\nJapan\n\nEU\n\nAustralia\n\nUnited States\n\nChina\n\nCanada\n\nSouth Korea\n\n0 20 40 60 80 100 120 140 160\n\nSummary innovation index in 2025 (indexed to the EU in 2018)\n\nEmerging innovators\n\nModerate innovators\n\nStrong innovators\n\nInnovation leaders\n\nI Score in 2024","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2025-EIS_p86.pdf"]},"response":{"errors":[],"extract_id":"extract_219bbf8450cbd30d851594d05041f7bc","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["84 European Innovation Scoreboard 2025\n\nFigure 33: Innovation performance of the EU and its global competitors\n\n_Note: All performance scores are relative to that of the EU in 2018. Coloured bars show countries’ performance in 2025 while the vertical bars show performance in 2024. Due to limited data availability for global competitors, scores are calculated using a smaller set of indicators._\n\nCompared to 2024, most global competitors have improved their innovation performance. Only Canada, the EU, and South Africa recorded a decline. South Korea shows the largest annual improvement (+6.3%-points), followed by China (+6.0%-points) and Australia (+3.8%-points). The gap between Australia and the United States is also narrowing.\n\nOver the longer term (Figure 34), all global competitors have improved their innovation performance between 2018 and 2025. Notably, the five countries that now outperform the EU have all recorded faster growth than the EU itself, thereby widening the performance gap. Among the six countries still lagging the EU, only Chile has increased its performance at a higher rate than the EU since 2018, with a gain of 15.5%-points.\n\nI\n\nI\n\nI\n\nI\n\nI\n\nI\n\nI\n\nI\n\nI\n\nI\n\nI\n\nI Mexico\n\nSouth Africa\n\nChile\n\nIndia\n\nBrazil\n\nJapan\n\nEU\n\nAustralia\n\nUnited States\n\nChina\n\nCanada\n\nSouth Korea\n\n0 20 40 60 80 100 120 140 160\n\nSummary innovation index in 2025 (indexed to the EU in 2018)\n\nEmerging innovators\n\nModerate innovators\n\nStrong innovators\n\nInnovation leaders\n\nI Score in 2024"],"full_content":"84 European Innovation Scoreboard 2025\n\nFigure 33: Innovation performance of the EU and its global competitors\n\n_Note: All performance scores are relative to that of the EU in 2018. Coloured bars show countries’ performance in 2025 while the vertical bars show performance in 2024. Due to limited data availability for global competitors, scores are calculated using a smaller set of indicators._\n\nCompared to 2024, most global competitors have improved their innovation performance. Only Canada, the EU, and South Africa recorded a decline. South Korea shows the largest annual improvement (+6.3%-points), followed by China (+6.0%-points) and Australia (+3.8%-points). The gap between Australia and the United States is also narrowing.\n\nOver the longer term (Figure 34), all global competitors have improved their innovation performance between 2018 and 2025. Notably, the five countries that now outperform the EU have all recorded faster growth than the EU itself, thereby widening the performance gap. Among the six countries still lagging the EU, only Chile has increased its performance at a higher rate than the EU since 2018, with a gain of 15.5%-points.\n\nI\n\nI\n\nI\n\nI\n\nI\n\nI\n\nI\n\nI\n\nI\n\nI\n\nI\n\nI Mexico\n\nSouth Africa\n\nChile\n\nIndia\n\nBrazil\n\nJapan\n\nEU\n\nAustralia\n\nUnited States\n\nChina\n\nCanada\n\nSouth Korea\n\n0 20 40 60 80 100 120 140 160\n\nSummary innovation index in 2025 (indexed to the EU in 2018)\n\nEmerging innovators\n\nModerate innovators\n\nStrong innovators\n\nInnovation leaders\n\nI Score in 2024","publish_date":null,"title":"Figure 33: Innovation performance of the EU and its global competitors I I I I I I I I I","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2025-EIS_p86.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/2025-EIS_p86.pdf"},"request":{"example_id":"chart/2025-EIS_p86","product_type":"parse","source_file_path":"2025-EIS_p86.pdf"},"started_at":"2026-11-04T19:45:41.140363"} +{"completed_at":"2026-11-04T19:45:45.602319","latency_in_ms":3278,"output":{"example_id":"chart/5._ESM_Staff_Report_on_Comprehensive_Review_p21","grounded_pages":[],"layout_pages":[],"markdown":"21\n\n**_Figure 5_** . **Comparison of yields at issuance of SURE bond issuances with equivalent ESM benchmark yields**\n\nSource: ESM, European Commission **Having considered the differing uptake between the PCS and SURE instrument, a follow-up workstream could be initiated to address the stigma concern related to the ESM role.** Important elements of this work would include possible refinements of the ESM precautionary toolkit and processes, which are analysed in the following section.\n\n_2\\.1.5 Possible refinements of the ESM precautionary instruments_\n\n**The emerging challenges for the euro area as well as lessons learned from the design of PCS provide a starting point for considering refinements to precautionary instruments.** Vulnerabilities that could manifest from risks such as climate change, geopolitics, and geoeconomic fragmentation pose a particularly uncertain outlook for the euro area. At the same time, the stigma associated with ESM support hampers access to its instruments by its Members even in case of severe crises. Against this backdrop, the ability of the ESM toolkit to tailor responses to a multitude of contexts remains a vital safeguard. To ensure these instruments remain effective, their design could be refined in two areas:\n\n• By making good use of the confidential consultation procedure, including as a platform for group requests.\n\n• By analysing the merits of designing a new precautionary instrument.\n\nUse of confidential consultations to improve the effectiveness of the ESM precautionary instruments\n\n**The process through which precautionary assistance is requested and granted is of crucial importance for its effectiveness.** It ensures that the formal requirements of the ESM Treaty are respected. It can also provide prospective beneficiary ESM Members with more concrete information on potential ESM assistance before submitting an official request. The procedure to request the precautionary instruments under the revised Treaty can be summarised in Figure 6.\n\n0\\.0%\n\n0\\.5%\n\n1\\.0%\n\n1\\.5%\n\n2\\.0%\n\n\\-1.0% -0.5% 0.0% 0.5% 1.0% 1.5% 2.0%\n\n10yr 20yr 5yr 30yr 15yr 7yr 30yr 15yr 5yr 25yr 8yr 25.6yr 15yr\n\nEuropean Commission: SURE ESM Benchmark Yield of Equivalent Maturity\n\n**SURE #1 20/10/20**\n\n**SURE #2 10/11/20**\n\n**SURE #3 24/11/20**\n\n**SURE #4 26/01/21**\n\n**SURE #5 09/03/21**\n\n**SURE #6 23/03/21**\n\n**SURE #7 18/05/21**\n\n**SURE #8 29/03/22**","pages":[{"line_numbers":[],"markdown":"21\n\n**_Figure 5_** . **Comparison of yields at issuance of SURE bond issuances with equivalent ESM benchmark yields**\n\nSource: ESM, European Commission **Having considered the differing uptake between the PCS and SURE instrument, a follow-up workstream could be initiated to address the stigma concern related to the ESM role.** Important elements of this work would include possible refinements of the ESM precautionary toolkit and processes, which are analysed in the following section.\n\n_2\\.1.5 Possible refinements of the ESM precautionary instruments_\n\n**The emerging challenges for the euro area as well as lessons learned from the design of PCS provide a starting point for considering refinements to precautionary instruments.** Vulnerabilities that could manifest from risks such as climate change, geopolitics, and geoeconomic fragmentation pose a particularly uncertain outlook for the euro area. At the same time, the stigma associated with ESM support hampers access to its instruments by its Members even in case of severe crises. Against this backdrop, the ability of the ESM toolkit to tailor responses to a multitude of contexts remains a vital safeguard. To ensure these instruments remain effective, their design could be refined in two areas:\n\n• By making good use of the confidential consultation procedure, including as a platform for group requests.\n\n• By analysing the merits of designing a new precautionary instrument.\n\nUse of confidential consultations to improve the effectiveness of the ESM precautionary instruments\n\n**The process through which precautionary assistance is requested and granted is of crucial importance for its effectiveness.** It ensures that the formal requirements of the ESM Treaty are respected. It can also provide prospective beneficiary ESM Members with more concrete information on potential ESM assistance before submitting an official request. The procedure to request the precautionary instruments under the revised Treaty can be summarised in Figure 6.\n\n0\\.0%\n\n0\\.5%\n\n1\\.0%\n\n1\\.5%\n\n2\\.0%\n\n\\-1.0% -0.5% 0.0% 0.5% 1.0% 1.5% 2.0%\n\n10yr 20yr 5yr 30yr 15yr 7yr 30yr 15yr 5yr 25yr 8yr 25.6yr 15yr\n\nEuropean Commission: SURE ESM Benchmark Yield of Equivalent Maturity\n\n**SURE #1 20/10/20**\n\n**SURE #2 10/11/20**\n\n**SURE #3 24/11/20**\n\n**SURE #4 26/01/21**\n\n**SURE #5 09/03/21**\n\n**SURE #6 23/03/21**\n\n**SURE #7 18/05/21**\n\n**SURE #8 29/03/22**","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"21\n\n**_Figure 5_** . **Comparison of yields at issuance of SURE bond issuances with equivalent ESM benchmark yields**\n\nSource: ESM, European Commission **Having considered the differing uptake between the PCS and SURE instrument, a follow-up workstream could be initiated to address the stigma concern related to the ESM role.** Important elements of this work would include possible refinements of the ESM precautionary toolkit and processes, which are analysed in the following section.\n\n_2\\.1.5 Possible refinements of the ESM precautionary instruments_\n\n**The emerging challenges for the euro area as well as lessons learned from the design of PCS provide a starting point for considering refinements to precautionary instruments.** Vulnerabilities that could manifest from risks such as climate change, geopolitics, and geoeconomic fragmentation pose a particularly uncertain outlook for the euro area. At the same time, the stigma associated with ESM support hampers access to its instruments by its Members even in case of severe crises. Against this backdrop, the ability of the ESM toolkit to tailor responses to a multitude of contexts remains a vital safeguard. To ensure these instruments remain effective, their design could be refined in two areas:\n\n• By making good use of the confidential consultation procedure, including as a platform for group requests.\n\n• By analysing the merits of designing a new precautionary instrument.\n\nUse of confidential consultations to improve the effectiveness of the ESM precautionary instruments\n\n**The process through which precautionary assistance is requested and granted is of crucial importance for its effectiveness.** It ensures that the formal requirements of the ESM Treaty are respected. It can also provide prospective beneficiary ESM Members with more concrete information on potential ESM assistance before submitting an official request. The procedure to request the precautionary instruments under the revised Treaty can be summarised in Figure 6.\n\n0\\.0%\n\n0\\.5%\n\n1\\.0%\n\n1\\.5%\n\n2\\.0%\n\n\\-1.0% -0.5% 0.0% 0.5% 1.0% 1.5% 2.0%\n\n10yr 20yr 5yr 30yr 15yr 7yr 30yr 15yr 5yr 25yr 8yr 25.6yr 15yr\n\nEuropean Commission: SURE ESM Benchmark Yield of Equivalent Maturity\n\n**SURE #1 20/10/20**\n\n**SURE #2 10/11/20**\n\n**SURE #3 24/11/20**\n\n**SURE #4 26/01/21**\n\n**SURE #5 09/03/21**\n\n**SURE #6 23/03/21**\n\n**SURE #7 18/05/21**\n\n**SURE #8 29/03/22**","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/5._ESM_Staff_Report_on_Comprehensive_Review_p21.pdf"]},"response":{"errors":[],"extract_id":"extract_4fc6bd079cc23ed6c7a19933c3e192ec","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["21\n\n**_Figure 5_** . **Comparison of yields at issuance of SURE bond issuances with equivalent ESM benchmark yields**\n\nSource: ESM, European Commission **Having considered the differing uptake between the PCS and SURE instrument, a follow-up workstream could be initiated to address the stigma concern related to the ESM role.** Important elements of this work would include possible refinements of the ESM precautionary toolkit and processes, which are analysed in the following section.\n\n_2\\.1.5 Possible refinements of the ESM precautionary instruments_\n\n**The emerging challenges for the euro area as well as lessons learned from the design of PCS provide a starting point for considering refinements to precautionary instruments.** Vulnerabilities that could manifest from risks such as climate change, geopolitics, and geoeconomic fragmentation pose a particularly uncertain outlook for the euro area. At the same time, the stigma associated with ESM support hampers access to its instruments by its Members even in case of severe crises. Against this backdrop, the ability of the ESM toolkit to tailor responses to a multitude of contexts remains a vital safeguard. To ensure these instruments remain effective, their design could be refined in two areas:\n\n• By making good use of the confidential consultation procedure, including as a platform for group requests.\n\n• By analysing the merits of designing a new precautionary instrument.\n\nUse of confidential consultations to improve the effectiveness of the ESM precautionary instruments\n\n**The process through which precautionary assistance is requested and granted is of crucial importance for its effectiveness.** It ensures that the formal requirements of the ESM Treaty are respected. It can also provide prospective beneficiary ESM Members with more concrete information on potential ESM assistance before submitting an official request. The procedure to request the precautionary instruments under the revised Treaty can be summarised in Figure 6.\n\n0\\.0%\n\n0\\.5%\n\n1\\.0%\n\n1\\.5%\n\n2\\.0%\n\n\\-1.0% -0.5% 0.0% 0.5% 1.0% 1.5% 2.0%\n\n10yr 20yr 5yr 30yr 15yr 7yr 30yr 15yr 5yr 25yr 8yr 25.6yr 15yr\n\nEuropean Commission: SURE ESM Benchmark Yield of Equivalent Maturity\n\n**SURE #1 20/10/20**\n\n**SURE #2 10/11/20**\n\n**SURE #3 24/11/20**\n\n**SURE #4 26/01/21**\n\n**SURE #5 09/03/21**\n\n**SURE #6 23/03/21**\n\n**SURE #7 18/05/21**\n\n**SURE #8 29/03/22**"],"full_content":"21\n\n**_Figure 5_** . **Comparison of yields at issuance of SURE bond issuances with equivalent ESM benchmark yields**\n\nSource: ESM, European Commission **Having considered the differing uptake between the PCS and SURE instrument, a follow-up workstream could be initiated to address the stigma concern related to the ESM role.** Important elements of this work would include possible refinements of the ESM precautionary toolkit and processes, which are analysed in the following section.\n\n_2\\.1.5 Possible refinements of the ESM precautionary instruments_\n\n**The emerging challenges for the euro area as well as lessons learned from the design of PCS provide a starting point for considering refinements to precautionary instruments.** Vulnerabilities that could manifest from risks such as climate change, geopolitics, and geoeconomic fragmentation pose a particularly uncertain outlook for the euro area. At the same time, the stigma associated with ESM support hampers access to its instruments by its Members even in case of severe crises. Against this backdrop, the ability of the ESM toolkit to tailor responses to a multitude of contexts remains a vital safeguard. To ensure these instruments remain effective, their design could be refined in two areas:\n\n• By making good use of the confidential consultation procedure, including as a platform for group requests.\n\n• By analysing the merits of designing a new precautionary instrument.\n\nUse of confidential consultations to improve the effectiveness of the ESM precautionary instruments\n\n**The process through which precautionary assistance is requested and granted is of crucial importance for its effectiveness.** It ensures that the formal requirements of the ESM Treaty are respected. It can also provide prospective beneficiary ESM Members with more concrete information on potential ESM assistance before submitting an official request. The procedure to request the precautionary instruments under the revised Treaty can be summarised in Figure 6.\n\n0\\.0%\n\n0\\.5%\n\n1\\.0%\n\n1\\.5%\n\n2\\.0%\n\n\\-1.0% -0.5% 0.0% 0.5% 1.0% 1.5% 2.0%\n\n10yr 20yr 5yr 30yr 15yr 7yr 30yr 15yr 5yr 25yr 8yr 25.6yr 15yr\n\nEuropean Commission: SURE ESM Benchmark Yield of Equivalent Maturity\n\n**SURE #1 20/10/20**\n\n**SURE #2 10/11/20**\n\n**SURE #3 24/11/20**\n\n**SURE #4 26/01/21**\n\n**SURE #5 09/03/21**\n\n**SURE #6 23/03/21**\n\n**SURE #7 18/05/21**\n\n**SURE #8 29/03/22**","publish_date":null,"title":"2.0% 1.5% 1.0% 0.5% 0.0% -0.5% -1.0% 10yr 20yr 5yr 30yr 15yr 7yr 30yr 15yr 5yr 25yr 8yr 25.6yr 15yr","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/5._ESM_Staff_Report_on_Comprehensive_Review_p21.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/5._ESM_Staff_Report_on_Comprehensive_Review_p21.pdf"},"request":{"example_id":"chart/5._ESM_Staff_Report_on_Comprehensive_Review_p21","product_type":"parse","source_file_path":"5._ESM_Staff_Report_on_Comprehensive_Review_p21.pdf"},"started_at":"2026-11-04T19:45:42.323438"} +{"completed_at":"2026-11-04T19:45:47.137850","latency_in_ms":3220,"output":{"example_id":"chart/5._ESM_Staff_Report_on_Comprehensive_Review_p36","grounded_pages":[],"layout_pages":[],"markdown":"36\n\n## 3\\. Adequacy of the lending volume and authorised capital stock\n\n_The evolving economic environment in the euro area and the evolution in the ESM instruments will determine future needs for ESM resources. Based on the following assessment, and in line with the latest review completed by the BoG in June 2022, a sufficient level of the ESM lending capacity should be maintained to uphold the confidence in the euro area financial architecture._\n\n**A large part of the total ESM lending capacity remains available to provide support for preserving the euro area financial stability, if required.** The ESM available lending capacity out of the €500 billion maximum amount has decreased from €461 billion in 2012 to €422 billion in April 2024, reflecting the financing provided under the financial assistance to Cyprus, Greece and Spain. The remaining lending capacity is set to fall, assuming the full ratification of the revised ESM Treaty where it is foreseen that €68 billion will be earmarked for the backstop to the SRF (Figure 7). Absent of any new programmes, under the current repayment schedules of the beneficiary Member States, a gradual increase in the available lending capacity is then projected until 2060 (€432 billion), at which point the only committed portion of the ESM lending capacity would be under the common backstop, assuming that the facility is extended by then. ⁴¹\n\n**Figure 7. ESM remaining lending capacity** (in € billion)\n\nSource: ESM Note: The maximum lending volume of the ESM is €500 billion. From 2023, the data points are projections based on current repayment schedules of beneficiaries of ESM financing. The distance between the curves from 2023 reflects the possible introduction of an up to €68 billion backstop of the SRF by the ESM, which is subject to full ratification of the revised Treaty. The backstop is added under the assumption that the current size of its nominal cap (€68 billion) is not revised in the future. **The combined lending capacity of the ESM and EFSF managed to support euro area financial stability in the past while the maximum lending capacity has slightly declined against macroeconomic aggregates over the period.** At the ESM’s inception in 2012, the BoG set the maximum lending volume at €500 billion and confirmed the combined lending ceiling of the ESM and the EFSF at €700 billion (Figure 8). As of July 2013, the EFSF may no longer engage in new financing programmes, leaving the remaining ESM lending capacity as the main component of the euro area safety net. In the absence of\n\n41 The initial availability period of the backstop facility shall be ten years, and will be extended automatically for further ten- year periods, except under the circumstances envisaged in the _European Stability Mechanism Guideline on the backstop facility to the SRB for the SRF_ , Article 12.\n\n250\n\n300\n\n350\n\n400\n\n450\n\n500\n\n2012 2016 2020 2024 2028 2032 2036 2040 2044 2048 2052 2056 2060 2064 2068\n\nUnder Current Treaty Ratification of Revised Treaty in 2024","pages":[{"line_numbers":[],"markdown":"36\n\n## 3\\. Adequacy of the lending volume and authorised capital stock\n\n_The evolving economic environment in the euro area and the evolution in the ESM instruments will determine future needs for ESM resources. Based on the following assessment, and in line with the latest review completed by the BoG in June 2022, a sufficient level of the ESM lending capacity should be maintained to uphold the confidence in the euro area financial architecture._\n\n**A large part of the total ESM lending capacity remains available to provide support for preserving the euro area financial stability, if required.** The ESM available lending capacity out of the €500 billion maximum amount has decreased from €461 billion in 2012 to €422 billion in April 2024, reflecting the financing provided under the financial assistance to Cyprus, Greece and Spain. The remaining lending capacity is set to fall, assuming the full ratification of the revised ESM Treaty where it is foreseen that €68 billion will be earmarked for the backstop to the SRF (Figure 7). Absent of any new programmes, under the current repayment schedules of the beneficiary Member States, a gradual increase in the available lending capacity is then projected until 2060 (€432 billion), at which point the only committed portion of the ESM lending capacity would be under the common backstop, assuming that the facility is extended by then. ⁴¹\n\n**Figure 7. ESM remaining lending capacity** (in € billion)\n\nSource: ESM Note: The maximum lending volume of the ESM is €500 billion. From 2023, the data points are projections based on current repayment schedules of beneficiaries of ESM financing. The distance between the curves from 2023 reflects the possible introduction of an up to €68 billion backstop of the SRF by the ESM, which is subject to full ratification of the revised Treaty. The backstop is added under the assumption that the current size of its nominal cap (€68 billion) is not revised in the future. **The combined lending capacity of the ESM and EFSF managed to support euro area financial stability in the past while the maximum lending capacity has slightly declined against macroeconomic aggregates over the period.** At the ESM’s inception in 2012, the BoG set the maximum lending volume at €500 billion and confirmed the combined lending ceiling of the ESM and the EFSF at €700 billion (Figure 8). As of July 2013, the EFSF may no longer engage in new financing programmes, leaving the remaining ESM lending capacity as the main component of the euro area safety net. In the absence of\n\n41 The initial availability period of the backstop facility shall be ten years, and will be extended automatically for further ten- year periods, except under the circumstances envisaged in the _European Stability Mechanism Guideline on the backstop facility to the SRB for the SRF_ , Article 12.\n\n250\n\n300\n\n350\n\n400\n\n450\n\n500\n\n2012 2016 2020 2024 2028 2032 2036 2040 2044 2048 2052 2056 2060 2064 2068\n\nUnder Current Treaty Ratification of Revised Treaty in 2024","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"36\n\n## 3\\. Adequacy of the lending volume and authorised capital stock\n\n_The evolving economic environment in the euro area and the evolution in the ESM instruments will determine future needs for ESM resources. Based on the following assessment, and in line with the latest review completed by the BoG in June 2022, a sufficient level of the ESM lending capacity should be maintained to uphold the confidence in the euro area financial architecture._\n\n**A large part of the total ESM lending capacity remains available to provide support for preserving the euro area financial stability, if required.** The ESM available lending capacity out of the €500 billion maximum amount has decreased from €461 billion in 2012 to €422 billion in April 2024, reflecting the financing provided under the financial assistance to Cyprus, Greece and Spain. The remaining lending capacity is set to fall, assuming the full ratification of the revised ESM Treaty where it is foreseen that €68 billion will be earmarked for the backstop to the SRF (Figure 7). Absent of any new programmes, under the current repayment schedules of the beneficiary Member States, a gradual increase in the available lending capacity is then projected until 2060 (€432 billion), at which point the only committed portion of the ESM lending capacity would be under the common backstop, assuming that the facility is extended by then. ⁴¹\n\n**Figure 7. ESM remaining lending capacity** (in € billion)\n\nSource: ESM Note: The maximum lending volume of the ESM is €500 billion. From 2023, the data points are projections based on current repayment schedules of beneficiaries of ESM financing. The distance between the curves from 2023 reflects the possible introduction of an up to €68 billion backstop of the SRF by the ESM, which is subject to full ratification of the revised Treaty. The backstop is added under the assumption that the current size of its nominal cap (€68 billion) is not revised in the future. **The combined lending capacity of the ESM and EFSF managed to support euro area financial stability in the past while the maximum lending capacity has slightly declined against macroeconomic aggregates over the period.** At the ESM’s inception in 2012, the BoG set the maximum lending volume at €500 billion and confirmed the combined lending ceiling of the ESM and the EFSF at €700 billion (Figure 8). As of July 2013, the EFSF may no longer engage in new financing programmes, leaving the remaining ESM lending capacity as the main component of the euro area safety net. In the absence of\n\n41 The initial availability period of the backstop facility shall be ten years, and will be extended automatically for further ten- year periods, except under the circumstances envisaged in the _European Stability Mechanism Guideline on the backstop facility to the SRB for the SRF_ , Article 12.\n\n250\n\n300\n\n350\n\n400\n\n450\n\n500\n\n2012 2016 2020 2024 2028 2032 2036 2040 2044 2048 2052 2056 2060 2064 2068\n\nUnder Current Treaty Ratification of Revised Treaty in 2024","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/5._ESM_Staff_Report_on_Comprehensive_Review_p36.pdf"]},"response":{"errors":[],"extract_id":"extract_2565aeca8d94855fc705328a5f7c6e95","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["36\n\n## 3\\. Adequacy of the lending volume and authorised capital stock\n\n_The evolving economic environment in the euro area and the evolution in the ESM instruments will determine future needs for ESM resources. Based on the following assessment, and in line with the latest review completed by the BoG in June 2022, a sufficient level of the ESM lending capacity should be maintained to uphold the confidence in the euro area financial architecture._\n\n**A large part of the total ESM lending capacity remains available to provide support for preserving the euro area financial stability, if required.** The ESM available lending capacity out of the €500 billion maximum amount has decreased from €461 billion in 2012 to €422 billion in April 2024, reflecting the financing provided under the financial assistance to Cyprus, Greece and Spain. The remaining lending capacity is set to fall, assuming the full ratification of the revised ESM Treaty where it is foreseen that €68 billion will be earmarked for the backstop to the SRF (Figure 7). Absent of any new programmes, under the current repayment schedules of the beneficiary Member States, a gradual increase in the available lending capacity is then projected until 2060 (€432 billion), at which point the only committed portion of the ESM lending capacity would be under the common backstop, assuming that the facility is extended by then. ⁴¹\n\n**Figure 7. ESM remaining lending capacity** (in € billion)\n\nSource: ESM Note: The maximum lending volume of the ESM is €500 billion. From 2023, the data points are projections based on current repayment schedules of beneficiaries of ESM financing. The distance between the curves from 2023 reflects the possible introduction of an up to €68 billion backstop of the SRF by the ESM, which is subject to full ratification of the revised Treaty. The backstop is added under the assumption that the current size of its nominal cap (€68 billion) is not revised in the future. **The combined lending capacity of the ESM and EFSF managed to support euro area financial stability in the past while the maximum lending capacity has slightly declined against macroeconomic aggregates over the period.** At the ESM’s inception in 2012, the BoG set the maximum lending volume at €500 billion and confirmed the combined lending ceiling of the ESM and the EFSF at €700 billion (Figure 8). As of July 2013, the EFSF may no longer engage in new financing programmes, leaving the remaining ESM lending capacity as the main component of the euro area safety net. In the absence of\n\n41 The initial availability period of the backstop facility shall be ten years, and will be extended automatically for further ten- year periods, except under the circumstances envisaged in the _European Stability Mechanism Guideline on the backstop facility to the SRB for the SRF_ , Article 12.\n\n250\n\n300\n\n350\n\n400\n\n450\n\n500\n\n2012 2016 2020 2024 2028 2032 2036 2040 2044 2048 2052 2056 2060 2064 2068\n\nUnder Current Treaty Ratification of Revised Treaty in 2024"],"full_content":"36\n\n## 3\\. Adequacy of the lending volume and authorised capital stock\n\n_The evolving economic environment in the euro area and the evolution in the ESM instruments will determine future needs for ESM resources. Based on the following assessment, and in line with the latest review completed by the BoG in June 2022, a sufficient level of the ESM lending capacity should be maintained to uphold the confidence in the euro area financial architecture._\n\n**A large part of the total ESM lending capacity remains available to provide support for preserving the euro area financial stability, if required.** The ESM available lending capacity out of the €500 billion maximum amount has decreased from €461 billion in 2012 to €422 billion in April 2024, reflecting the financing provided under the financial assistance to Cyprus, Greece and Spain. The remaining lending capacity is set to fall, assuming the full ratification of the revised ESM Treaty where it is foreseen that €68 billion will be earmarked for the backstop to the SRF (Figure 7). Absent of any new programmes, under the current repayment schedules of the beneficiary Member States, a gradual increase in the available lending capacity is then projected until 2060 (€432 billion), at which point the only committed portion of the ESM lending capacity would be under the common backstop, assuming that the facility is extended by then. ⁴¹\n\n**Figure 7. ESM remaining lending capacity** (in € billion)\n\nSource: ESM Note: The maximum lending volume of the ESM is €500 billion. From 2023, the data points are projections based on current repayment schedules of beneficiaries of ESM financing. The distance between the curves from 2023 reflects the possible introduction of an up to €68 billion backstop of the SRF by the ESM, which is subject to full ratification of the revised Treaty. The backstop is added under the assumption that the current size of its nominal cap (€68 billion) is not revised in the future. **The combined lending capacity of the ESM and EFSF managed to support euro area financial stability in the past while the maximum lending capacity has slightly declined against macroeconomic aggregates over the period.** At the ESM’s inception in 2012, the BoG set the maximum lending volume at €500 billion and confirmed the combined lending ceiling of the ESM and the EFSF at €700 billion (Figure 8). As of July 2013, the EFSF may no longer engage in new financing programmes, leaving the remaining ESM lending capacity as the main component of the euro area safety net. In the absence of\n\n41 The initial availability period of the backstop facility shall be ten years, and will be extended automatically for further ten- year periods, except under the circumstances envisaged in the _European Stability Mechanism Guideline on the backstop facility to the SRB for the SRF_ , Article 12.\n\n250\n\n300\n\n350\n\n400\n\n450\n\n500\n\n2012 2016 2020 2024 2028 2032 2036 2040 2044 2048 2052 2056 2060 2064 2068\n\nUnder Current Treaty Ratification of Revised Treaty in 2024","publish_date":null,"title":"3. Adequacy of the lending volume and authorised capital stock","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/5._ESM_Staff_Report_on_Comprehensive_Review_p36.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/5._ESM_Staff_Report_on_Comprehensive_Review_p36.pdf"},"request":{"example_id":"chart/5._ESM_Staff_Report_on_Comprehensive_Review_p36","product_type":"parse","source_file_path":"5._ESM_Staff_Report_on_Comprehensive_Review_p36.pdf"},"started_at":"2026-11-04T19:45:43.917523"} +{"completed_at":"2026-11-04T19:45:48.315196","latency_in_ms":2711,"output":{"example_id":"chart/5._ESM_Staff_Report_on_Comprehensive_Review_p37","grounded_pages":[],"layout_pages":[],"markdown":"37\n\nforecasted future demand for ESM financing, a backward-looking approach may provide some insights for the adequacy of the lending capacity. The ESM lending capacity is smaller than the total past programme amounts committed (€625 billion) and greater than the total past programme amounts disbursed (€483 billion) over the 2010-2018 period by arrangements dedicated to the euro area (Greek Loan Facility, EFSM, EFSF, ESM, and the IMF) (Figure 8). Having available lending capacity at the onset of the Covid-19 pandemic enabled the ESM to contribute to the common European response with a significant envelope. These past commitments may not be fully indicative of future needs for ESM support considering the evolving economic aggregates, the prevailing economic uncertainty and the existence of potential risks to financial stability on the one hand, nor can they reflect the advancements in the economic governance and crisis resilience since the sovereign debt crisis on the other. As the pivotal part of the euro area financial safety net, the ESM lending volume should be maintained at a sufficiently high level considering the evolution of main macroeconomic and fiscal aggregates, such as GDP, debt, and revenues (Figure 9).\n\n**Figure 8. ESM and other financial assistance sources** (in € billion)\n\nSource: ESM, European Commission, IMF, ECA Note: Amounts committed and disbursed are part of financial assistance programmes to euro area countries in 2010- 2018. The COVID-19 response envelope is comprised of the ESM's PCS, and of the European Commission's SURE. Under the PCS, up to 2% of euro area GDP as of end-2019 (€240 billion) was made available by the ESM, though no financing requests were made. The SURE envelope size of €100 billion was available to all EU27 Member States.\n\n0\n\n100 200 300 400 500 600 700 800\n\nESM-EFSF\n\nCombined Maximum\n\nLending Capacity\n\nCommitted 2010-2018\n\nDisbursed 2010-2018\n\nCOVID-19 Response\n\n(PCS+SURE) 2020-2022\n\nESM EFSF IMF GLF EFSM SURE","pages":[{"line_numbers":[],"markdown":"37\n\nforecasted future demand for ESM financing, a backward-looking approach may provide some insights for the adequacy of the lending capacity. The ESM lending capacity is smaller than the total past programme amounts committed (€625 billion) and greater than the total past programme amounts disbursed (€483 billion) over the 2010-2018 period by arrangements dedicated to the euro area (Greek Loan Facility, EFSM, EFSF, ESM, and the IMF) (Figure 8). Having available lending capacity at the onset of the Covid-19 pandemic enabled the ESM to contribute to the common European response with a significant envelope. These past commitments may not be fully indicative of future needs for ESM support considering the evolving economic aggregates, the prevailing economic uncertainty and the existence of potential risks to financial stability on the one hand, nor can they reflect the advancements in the economic governance and crisis resilience since the sovereign debt crisis on the other. As the pivotal part of the euro area financial safety net, the ESM lending volume should be maintained at a sufficiently high level considering the evolution of main macroeconomic and fiscal aggregates, such as GDP, debt, and revenues (Figure 9).\n\n**Figure 8. ESM and other financial assistance sources** (in € billion)\n\nSource: ESM, European Commission, IMF, ECA Note: Amounts committed and disbursed are part of financial assistance programmes to euro area countries in 2010- 2018. The COVID-19 response envelope is comprised of the ESM's PCS, and of the European Commission's SURE. Under the PCS, up to 2% of euro area GDP as of end-2019 (€240 billion) was made available by the ESM, though no financing requests were made. The SURE envelope size of €100 billion was available to all EU27 Member States.\n\n0\n\n100 200 300 400 500 600 700 800\n\nESM-EFSF\n\nCombined Maximum\n\nLending Capacity\n\nCommitted 2010-2018\n\nDisbursed 2010-2018\n\nCOVID-19 Response\n\n(PCS+SURE) 2020-2022\n\nESM EFSF IMF GLF EFSM SURE","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"37\n\nforecasted future demand for ESM financing, a backward-looking approach may provide some insights for the adequacy of the lending capacity. The ESM lending capacity is smaller than the total past programme amounts committed (€625 billion) and greater than the total past programme amounts disbursed (€483 billion) over the 2010-2018 period by arrangements dedicated to the euro area (Greek Loan Facility, EFSM, EFSF, ESM, and the IMF) (Figure 8). Having available lending capacity at the onset of the Covid-19 pandemic enabled the ESM to contribute to the common European response with a significant envelope. These past commitments may not be fully indicative of future needs for ESM support considering the evolving economic aggregates, the prevailing economic uncertainty and the existence of potential risks to financial stability on the one hand, nor can they reflect the advancements in the economic governance and crisis resilience since the sovereign debt crisis on the other. As the pivotal part of the euro area financial safety net, the ESM lending volume should be maintained at a sufficiently high level considering the evolution of main macroeconomic and fiscal aggregates, such as GDP, debt, and revenues (Figure 9).\n\n**Figure 8. ESM and other financial assistance sources** (in € billion)\n\nSource: ESM, European Commission, IMF, ECA Note: Amounts committed and disbursed are part of financial assistance programmes to euro area countries in 2010- 2018. The COVID-19 response envelope is comprised of the ESM's PCS, and of the European Commission's SURE. Under the PCS, up to 2% of euro area GDP as of end-2019 (€240 billion) was made available by the ESM, though no financing requests were made. The SURE envelope size of €100 billion was available to all EU27 Member States.\n\n0\n\n100 200 300 400 500 600 700 800\n\nESM-EFSF\n\nCombined Maximum\n\nLending Capacity\n\nCommitted 2010-2018\n\nDisbursed 2010-2018\n\nCOVID-19 Response\n\n(PCS+SURE) 2020-2022\n\nESM EFSF IMF GLF EFSM SURE","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/5._ESM_Staff_Report_on_Comprehensive_Review_p37.pdf"]},"response":{"errors":[],"extract_id":"extract_33eb9df95618fd3fc75e325063f454b7","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["37\n\nforecasted future demand for ESM financing, a backward-looking approach may provide some insights for the adequacy of the lending capacity. The ESM lending capacity is smaller than the total past programme amounts committed (€625 billion) and greater than the total past programme amounts disbursed (€483 billion) over the 2010-2018 period by arrangements dedicated to the euro area (Greek Loan Facility, EFSM, EFSF, ESM, and the IMF) (Figure 8). Having available lending capacity at the onset of the Covid-19 pandemic enabled the ESM to contribute to the common European response with a significant envelope. These past commitments may not be fully indicative of future needs for ESM support considering the evolving economic aggregates, the prevailing economic uncertainty and the existence of potential risks to financial stability on the one hand, nor can they reflect the advancements in the economic governance and crisis resilience since the sovereign debt crisis on the other. As the pivotal part of the euro area financial safety net, the ESM lending volume should be maintained at a sufficiently high level considering the evolution of main macroeconomic and fiscal aggregates, such as GDP, debt, and revenues (Figure 9).\n\n**Figure 8. ESM and other financial assistance sources** (in € billion)\n\nSource: ESM, European Commission, IMF, ECA Note: Amounts committed and disbursed are part of financial assistance programmes to euro area countries in 2010- 2018. The COVID-19 response envelope is comprised of the ESM's PCS, and of the European Commission's SURE. Under the PCS, up to 2% of euro area GDP as of end-2019 (€240 billion) was made available by the ESM, though no financing requests were made. The SURE envelope size of €100 billion was available to all EU27 Member States.\n\n0\n\n100 200 300 400 500 600 700 800\n\nESM-EFSF\n\nCombined Maximum\n\nLending Capacity\n\nCommitted 2010-2018\n\nDisbursed 2010-2018\n\nCOVID-19 Response\n\n(PCS+SURE) 2020-2022\n\nESM EFSF IMF GLF EFSM SURE"],"full_content":"37\n\nforecasted future demand for ESM financing, a backward-looking approach may provide some insights for the adequacy of the lending capacity. The ESM lending capacity is smaller than the total past programme amounts committed (€625 billion) and greater than the total past programme amounts disbursed (€483 billion) over the 2010-2018 period by arrangements dedicated to the euro area (Greek Loan Facility, EFSM, EFSF, ESM, and the IMF) (Figure 8). Having available lending capacity at the onset of the Covid-19 pandemic enabled the ESM to contribute to the common European response with a significant envelope. These past commitments may not be fully indicative of future needs for ESM support considering the evolving economic aggregates, the prevailing economic uncertainty and the existence of potential risks to financial stability on the one hand, nor can they reflect the advancements in the economic governance and crisis resilience since the sovereign debt crisis on the other. As the pivotal part of the euro area financial safety net, the ESM lending volume should be maintained at a sufficiently high level considering the evolution of main macroeconomic and fiscal aggregates, such as GDP, debt, and revenues (Figure 9).\n\n**Figure 8. ESM and other financial assistance sources** (in € billion)\n\nSource: ESM, European Commission, IMF, ECA Note: Amounts committed and disbursed are part of financial assistance programmes to euro area countries in 2010- 2018. The COVID-19 response envelope is comprised of the ESM's PCS, and of the European Commission's SURE. Under the PCS, up to 2% of euro area GDP as of end-2019 (€240 billion) was made available by the ESM, though no financing requests were made. The SURE envelope size of €100 billion was available to all EU27 Member States.\n\n0\n\n100 200 300 400 500 600 700 800\n\nESM-EFSF\n\nCombined Maximum\n\nLending Capacity\n\nCommitted 2010-2018\n\nDisbursed 2010-2018\n\nCOVID-19 Response\n\n(PCS+SURE) 2020-2022\n\nESM EFSF IMF GLF EFSM SURE","publish_date":null,"title":"forecasted future demand for ESM financing, a backward-looking approach may provide some insights for the adequacy of the lending capacity.","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/5._ESM_Staff_Report_on_Comprehensive_Review_p37.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/5._ESM_Staff_Report_on_Comprehensive_Review_p37.pdf"},"request":{"example_id":"chart/5._ESM_Staff_Report_on_Comprehensive_Review_p37","product_type":"parse","source_file_path":"5._ESM_Staff_Report_on_Comprehensive_Review_p37.pdf"},"started_at":"2026-11-04T19:45:45.603617"} +{"completed_at":"2026-11-04T19:31:10.997451","latency_in_ms":1332,"output":{"example_id":"chart/5._ESM_Staff_Report_on_Comprehensive_Review_p38","grounded_pages":[],"layout_pages":[],"markdown":"38\n\n**Figure 9 ESM maximum lending capacity expressed as share of selected euro area macroeconomic aggregates** (2012 vs. 2023)\n\nSource: ESM, European Commission, ECB Note: Euro area in rolling composition **Against the metric of gross financing needs, both the ESM maximum and remaining lending capacity have declined in recent years.** Debt sustainability is intrinsically linked to both stock and flow features of public debt, from which vulnerabilities can arise from term-to-maturity structures, roll-over risks, amortisation schedules and varying interest rates. As highlighted by ESM staff research (Gabriele et al., 2017), gross financing needs (GFNs) is a comprehensive flow metric for identifying such mismatches. Since the creation of the ESM in 2012, GFNs have declined within the prevailing low- interest environment (Figure 10). Within this context, the ESM maximum and remaining lending capacity reached their peak as a share of euro area GFNs in 2019 at 29% and 24% respectively. Since 2019 however, GFNs have increased considerably given the onset of the Covid pandemic and the subsequent energy crisis, reversing this trend. In 2023, only 20% of euro area GFNs could be covered by the ESM maximum lending capacity (16% for the remaining capacity), below the levels prevailing in 2012.\n\n**Figure 10. ESM lending capacity as a share of euro area gross financing needs, 2012 - 2022** (left hand scale: EA GFN in € billion, right hand scale: ESM lending capacity in % of GFN)\n\nSource: ESM, European Commission Note: Given their volatile nature, changes in methodology, and incomplete data-series, a full time-series of gross financing needs for the 2012-2023 period is not presented.\n\n0\\.0%\n\n2\\.0%\n\n4\\.0%\n\n6\\.0%\n\n8\\.0%\n\n10\\.0%\n\n12\\.0%\n\nRevenue Debt GDP\n\n2012 2023\n\n0%\n\n10%\n\n20%\n\n30%\n\n40%\n\n0\n\n500\n\n1000 1500 2000 2500 3000\n\n2012 2016 2017 2018 2019 2020 2021 2022 2023\n\nEA GFN ESM - Remaining ESM - Maximum","pages":[{"line_numbers":[],"markdown":"38\n\n**Figure 9 ESM maximum lending capacity expressed as share of selected euro area macroeconomic aggregates** (2012 vs. 2023)\n\nSource: ESM, European Commission, ECB Note: Euro area in rolling composition **Against the metric of gross financing needs, both the ESM maximum and remaining lending capacity have declined in recent years.** Debt sustainability is intrinsically linked to both stock and flow features of public debt, from which vulnerabilities can arise from term-to-maturity structures, roll-over risks, amortisation schedules and varying interest rates. As highlighted by ESM staff research (Gabriele et al., 2017), gross financing needs (GFNs) is a comprehensive flow metric for identifying such mismatches. Since the creation of the ESM in 2012, GFNs have declined within the prevailing low- interest environment (Figure 10). Within this context, the ESM maximum and remaining lending capacity reached their peak as a share of euro area GFNs in 2019 at 29% and 24% respectively. Since 2019 however, GFNs have increased considerably given the onset of the Covid pandemic and the subsequent energy crisis, reversing this trend. In 2023, only 20% of euro area GFNs could be covered by the ESM maximum lending capacity (16% for the remaining capacity), below the levels prevailing in 2012.\n\n**Figure 10. ESM lending capacity as a share of euro area gross financing needs, 2012 - 2022** (left hand scale: EA GFN in € billion, right hand scale: ESM lending capacity in % of GFN)\n\nSource: ESM, European Commission Note: Given their volatile nature, changes in methodology, and incomplete data-series, a full time-series of gross financing needs for the 2012-2023 period is not presented.\n\n0\\.0%\n\n2\\.0%\n\n4\\.0%\n\n6\\.0%\n\n8\\.0%\n\n10\\.0%\n\n12\\.0%\n\nRevenue Debt GDP\n\n2012 2023\n\n0%\n\n10%\n\n20%\n\n30%\n\n40%\n\n0\n\n500\n\n1000 1500 2000 2500 3000\n\n2012 2016 2017 2018 2019 2020 2021 2022 2023\n\nEA GFN ESM - Remaining ESM - Maximum","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"38\n\n**Figure 9 ESM maximum lending capacity expressed as share of selected euro area macroeconomic aggregates** (2012 vs. 2023)\n\nSource: ESM, European Commission, ECB Note: Euro area in rolling composition **Against the metric of gross financing needs, both the ESM maximum and remaining lending capacity have declined in recent years.** Debt sustainability is intrinsically linked to both stock and flow features of public debt, from which vulnerabilities can arise from term-to-maturity structures, roll-over risks, amortisation schedules and varying interest rates. As highlighted by ESM staff research (Gabriele et al., 2017), gross financing needs (GFNs) is a comprehensive flow metric for identifying such mismatches. Since the creation of the ESM in 2012, GFNs have declined within the prevailing low- interest environment (Figure 10). Within this context, the ESM maximum and remaining lending capacity reached their peak as a share of euro area GFNs in 2019 at 29% and 24% respectively. Since 2019 however, GFNs have increased considerably given the onset of the Covid pandemic and the subsequent energy crisis, reversing this trend. In 2023, only 20% of euro area GFNs could be covered by the ESM maximum lending capacity (16% for the remaining capacity), below the levels prevailing in 2012.\n\n**Figure 10. ESM lending capacity as a share of euro area gross financing needs, 2012 - 2022** (left hand scale: EA GFN in € billion, right hand scale: ESM lending capacity in % of GFN)\n\nSource: ESM, European Commission Note: Given their volatile nature, changes in methodology, and incomplete data-series, a full time-series of gross financing needs for the 2012-2023 period is not presented.\n\n0\\.0%\n\n2\\.0%\n\n4\\.0%\n\n6\\.0%\n\n8\\.0%\n\n10\\.0%\n\n12\\.0%\n\nRevenue Debt GDP\n\n2012 2023\n\n0%\n\n10%\n\n20%\n\n30%\n\n40%\n\n0\n\n500\n\n1000 1500 2000 2500 3000\n\n2012 2016 2017 2018 2019 2020 2021 2022 2023\n\nEA GFN ESM - Remaining ESM - Maximum","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/5._ESM_Staff_Report_on_Comprehensive_Review_p38.pdf"]},"response":{"errors":[],"extract_id":"extract_d18c4d4aa2aa1e4bfe586d0ae2907053","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["38\n\n**Figure 9 ESM maximum lending capacity expressed as share of selected euro area macroeconomic aggregates** (2012 vs. 2023)\n\nSource: ESM, European Commission, ECB Note: Euro area in rolling composition **Against the metric of gross financing needs, both the ESM maximum and remaining lending capacity have declined in recent years.** Debt sustainability is intrinsically linked to both stock and flow features of public debt, from which vulnerabilities can arise from term-to-maturity structures, roll-over risks, amortisation schedules and varying interest rates. As highlighted by ESM staff research (Gabriele et al., 2017), gross financing needs (GFNs) is a comprehensive flow metric for identifying such mismatches. Since the creation of the ESM in 2012, GFNs have declined within the prevailing low- interest environment (Figure 10). Within this context, the ESM maximum and remaining lending capacity reached their peak as a share of euro area GFNs in 2019 at 29% and 24% respectively. Since 2019 however, GFNs have increased considerably given the onset of the Covid pandemic and the subsequent energy crisis, reversing this trend. In 2023, only 20% of euro area GFNs could be covered by the ESM maximum lending capacity (16% for the remaining capacity), below the levels prevailing in 2012.\n\n**Figure 10. ESM lending capacity as a share of euro area gross financing needs, 2012 - 2022** (left hand scale: EA GFN in € billion, right hand scale: ESM lending capacity in % of GFN)\n\nSource: ESM, European Commission Note: Given their volatile nature, changes in methodology, and incomplete data-series, a full time-series of gross financing needs for the 2012-2023 period is not presented.\n\n0\\.0%\n\n2\\.0%\n\n4\\.0%\n\n6\\.0%\n\n8\\.0%\n\n10\\.0%\n\n12\\.0%\n\nRevenue Debt GDP\n\n2012 2023\n\n0%\n\n10%\n\n20%\n\n30%\n\n40%\n\n0\n\n500\n\n1000 1500 2000 2500 3000\n\n2012 2016 2017 2018 2019 2020 2021 2022 2023\n\nEA GFN ESM - Remaining ESM - Maximum"],"full_content":"38\n\n**Figure 9 ESM maximum lending capacity expressed as share of selected euro area macroeconomic aggregates** (2012 vs. 2023)\n\nSource: ESM, European Commission, ECB Note: Euro area in rolling composition **Against the metric of gross financing needs, both the ESM maximum and remaining lending capacity have declined in recent years.** Debt sustainability is intrinsically linked to both stock and flow features of public debt, from which vulnerabilities can arise from term-to-maturity structures, roll-over risks, amortisation schedules and varying interest rates. As highlighted by ESM staff research (Gabriele et al., 2017), gross financing needs (GFNs) is a comprehensive flow metric for identifying such mismatches. Since the creation of the ESM in 2012, GFNs have declined within the prevailing low- interest environment (Figure 10). Within this context, the ESM maximum and remaining lending capacity reached their peak as a share of euro area GFNs in 2019 at 29% and 24% respectively. Since 2019 however, GFNs have increased considerably given the onset of the Covid pandemic and the subsequent energy crisis, reversing this trend. In 2023, only 20% of euro area GFNs could be covered by the ESM maximum lending capacity (16% for the remaining capacity), below the levels prevailing in 2012.\n\n**Figure 10. ESM lending capacity as a share of euro area gross financing needs, 2012 - 2022** (left hand scale: EA GFN in € billion, right hand scale: ESM lending capacity in % of GFN)\n\nSource: ESM, European Commission Note: Given their volatile nature, changes in methodology, and incomplete data-series, a full time-series of gross financing needs for the 2012-2023 period is not presented.\n\n0\\.0%\n\n2\\.0%\n\n4\\.0%\n\n6\\.0%\n\n8\\.0%\n\n10\\.0%\n\n12\\.0%\n\nRevenue Debt GDP\n\n2012 2023\n\n0%\n\n10%\n\n20%\n\n30%\n\n40%\n\n0\n\n500\n\n1000 1500 2000 2500 3000\n\n2012 2016 2017 2018 2019 2020 2021 2022 2023\n\nEA GFN ESM - Remaining ESM - Maximum","publish_date":null,"title":"12.0% 10.0% 8.0% 6.0% 4.0% 2.0% 0.0% Revenue Debt GDP 2012 2023 Against the metric of gross financing needs, both","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/5._ESM_Staff_Report_on_Comprehensive_Review_p38.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/5._ESM_Staff_Report_on_Comprehensive_Review_p38.pdf"},"request":{"example_id":"chart/5._ESM_Staff_Report_on_Comprehensive_Review_p38","product_type":"parse","source_file_path":"5._ESM_Staff_Report_on_Comprehensive_Review_p38.pdf"},"started_at":"2026-11-04T19:31:09.664761"} +{"completed_at":"2026-11-04T19:45:49.637274","latency_in_ms":1533,"output":{"example_id":"chart/5._ESM_Staff_Report_on_Comprehensive_Review_p8","grounded_pages":[],"layout_pages":[],"markdown":"8\n\n**Figure 2.** Distribution of euro area member states' public debt ratios\n\nSource: European Commission, AMECO database Note: Euro area in rolling composition **Today, national budgets and euro area crisis management are again being tested by various types of crises.** Following the Covid-19 pandemic, the recovery of European economies was cut short. Russia’s war on Ukraine drove up prices for energy, metals, and agricultural commodities. The EU and its Member States responded with efforts aimed at stabilising demand and reducing supply-chain dependencies. At the same time, monetary policy became restrictive as inflation had reached levels not seen in decades. Rapid increases in market interest rates and the uncertain growth outlook raised concerns over the stability of public and private sector debt. Most recently, this insecurity has been reinforced by growing hostilities in the Middle East and military attacks on important shipping lanes, which could further aggravate geopolitical tensions and add to price volatility. **Economic crises are reminders that despite its increased resilience, the euro area, to some extent, remains vulnerable to future large shocks** . The ongoing fragmentation of the world into multiple and sometimes rival powers raises new risks to financial stability associated with a higher likelihood of conflicts, elevated security spending, and a reduction in trade and capital flows, spurred also by protectionist impulses. These risks come on top of precarious secular trends and a climate crisis, exposures to which are likely to become highly asymmetric (Alogoskoufis et al., 2021). Population ageing, which reduces the number of contributors to the welfare state and raises health expenditure (Figure 3), the rise of new technologies, including the rapid advances in artificial intelligence posing social challenges potentially also for employment, and further geopolitical risks may all be sources of important shocks in the near future.\n\n0%\n\n20%\n\n40%\n\n60%\n\n80%\n\n100%\n\n120%\n\n2001 2003 2005 2007 2009 2011 2013 2015 2017 2019 2021 2023\n\n25th percentile Median 75th percentile","pages":[{"line_numbers":[],"markdown":"8\n\n**Figure 2.** Distribution of euro area member states' public debt ratios\n\nSource: European Commission, AMECO database Note: Euro area in rolling composition **Today, national budgets and euro area crisis management are again being tested by various types of crises.** Following the Covid-19 pandemic, the recovery of European economies was cut short. Russia’s war on Ukraine drove up prices for energy, metals, and agricultural commodities. The EU and its Member States responded with efforts aimed at stabilising demand and reducing supply-chain dependencies. At the same time, monetary policy became restrictive as inflation had reached levels not seen in decades. Rapid increases in market interest rates and the uncertain growth outlook raised concerns over the stability of public and private sector debt. Most recently, this insecurity has been reinforced by growing hostilities in the Middle East and military attacks on important shipping lanes, which could further aggravate geopolitical tensions and add to price volatility. **Economic crises are reminders that despite its increased resilience, the euro area, to some extent, remains vulnerable to future large shocks** . The ongoing fragmentation of the world into multiple and sometimes rival powers raises new risks to financial stability associated with a higher likelihood of conflicts, elevated security spending, and a reduction in trade and capital flows, spurred also by protectionist impulses. These risks come on top of precarious secular trends and a climate crisis, exposures to which are likely to become highly asymmetric (Alogoskoufis et al., 2021). Population ageing, which reduces the number of contributors to the welfare state and raises health expenditure (Figure 3), the rise of new technologies, including the rapid advances in artificial intelligence posing social challenges potentially also for employment, and further geopolitical risks may all be sources of important shocks in the near future.\n\n0%\n\n20%\n\n40%\n\n60%\n\n80%\n\n100%\n\n120%\n\n2001 2003 2005 2007 2009 2011 2013 2015 2017 2019 2021 2023\n\n25th percentile Median 75th percentile","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"8\n\n**Figure 2.** Distribution of euro area member states' public debt ratios\n\nSource: European Commission, AMECO database Note: Euro area in rolling composition **Today, national budgets and euro area crisis management are again being tested by various types of crises.** Following the Covid-19 pandemic, the recovery of European economies was cut short. Russia’s war on Ukraine drove up prices for energy, metals, and agricultural commodities. The EU and its Member States responded with efforts aimed at stabilising demand and reducing supply-chain dependencies. At the same time, monetary policy became restrictive as inflation had reached levels not seen in decades. Rapid increases in market interest rates and the uncertain growth outlook raised concerns over the stability of public and private sector debt. Most recently, this insecurity has been reinforced by growing hostilities in the Middle East and military attacks on important shipping lanes, which could further aggravate geopolitical tensions and add to price volatility. **Economic crises are reminders that despite its increased resilience, the euro area, to some extent, remains vulnerable to future large shocks** . The ongoing fragmentation of the world into multiple and sometimes rival powers raises new risks to financial stability associated with a higher likelihood of conflicts, elevated security spending, and a reduction in trade and capital flows, spurred also by protectionist impulses. These risks come on top of precarious secular trends and a climate crisis, exposures to which are likely to become highly asymmetric (Alogoskoufis et al., 2021). Population ageing, which reduces the number of contributors to the welfare state and raises health expenditure (Figure 3), the rise of new technologies, including the rapid advances in artificial intelligence posing social challenges potentially also for employment, and further geopolitical risks may all be sources of important shocks in the near future.\n\n0%\n\n20%\n\n40%\n\n60%\n\n80%\n\n100%\n\n120%\n\n2001 2003 2005 2007 2009 2011 2013 2015 2017 2019 2021 2023\n\n25th percentile Median 75th percentile","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/5._ESM_Staff_Report_on_Comprehensive_Review_p8.pdf"]},"response":{"errors":[],"extract_id":"extract_9a07de6a67f463fc1315a20ecaab00ea","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["8\n\n**Figure 2.** Distribution of euro area member states' public debt ratios\n\nSource: European Commission, AMECO database Note: Euro area in rolling composition **Today, national budgets and euro area crisis management are again being tested by various types of crises.** Following the Covid-19 pandemic, the recovery of European economies was cut short. Russia’s war on Ukraine drove up prices for energy, metals, and agricultural commodities. The EU and its Member States responded with efforts aimed at stabilising demand and reducing supply-chain dependencies. At the same time, monetary policy became restrictive as inflation had reached levels not seen in decades. Rapid increases in market interest rates and the uncertain growth outlook raised concerns over the stability of public and private sector debt. Most recently, this insecurity has been reinforced by growing hostilities in the Middle East and military attacks on important shipping lanes, which could further aggravate geopolitical tensions and add to price volatility. **Economic crises are reminders that despite its increased resilience, the euro area, to some extent, remains vulnerable to future large shocks** . The ongoing fragmentation of the world into multiple and sometimes rival powers raises new risks to financial stability associated with a higher likelihood of conflicts, elevated security spending, and a reduction in trade and capital flows, spurred also by protectionist impulses. These risks come on top of precarious secular trends and a climate crisis, exposures to which are likely to become highly asymmetric (Alogoskoufis et al., 2021). Population ageing, which reduces the number of contributors to the welfare state and raises health expenditure (Figure 3), the rise of new technologies, including the rapid advances in artificial intelligence posing social challenges potentially also for employment, and further geopolitical risks may all be sources of important shocks in the near future.\n\n0%\n\n20%\n\n40%\n\n60%\n\n80%\n\n100%\n\n120%\n\n2001 2003 2005 2007 2009 2011 2013 2015 2017 2019 2021 2023\n\n25th percentile Median 75th percentile"],"full_content":"8\n\n**Figure 2.** Distribution of euro area member states' public debt ratios\n\nSource: European Commission, AMECO database Note: Euro area in rolling composition **Today, national budgets and euro area crisis management are again being tested by various types of crises.** Following the Covid-19 pandemic, the recovery of European economies was cut short. Russia’s war on Ukraine drove up prices for energy, metals, and agricultural commodities. The EU and its Member States responded with efforts aimed at stabilising demand and reducing supply-chain dependencies. At the same time, monetary policy became restrictive as inflation had reached levels not seen in decades. Rapid increases in market interest rates and the uncertain growth outlook raised concerns over the stability of public and private sector debt. Most recently, this insecurity has been reinforced by growing hostilities in the Middle East and military attacks on important shipping lanes, which could further aggravate geopolitical tensions and add to price volatility. **Economic crises are reminders that despite its increased resilience, the euro area, to some extent, remains vulnerable to future large shocks** . The ongoing fragmentation of the world into multiple and sometimes rival powers raises new risks to financial stability associated with a higher likelihood of conflicts, elevated security spending, and a reduction in trade and capital flows, spurred also by protectionist impulses. These risks come on top of precarious secular trends and a climate crisis, exposures to which are likely to become highly asymmetric (Alogoskoufis et al., 2021). Population ageing, which reduces the number of contributors to the welfare state and raises health expenditure (Figure 3), the rise of new technologies, including the rapid advances in artificial intelligence posing social challenges potentially also for employment, and further geopolitical risks may all be sources of important shocks in the near future.\n\n0%\n\n20%\n\n40%\n\n60%\n\n80%\n\n100%\n\n120%\n\n2001 2003 2005 2007 2009 2011 2013 2015 2017 2019 2021 2023\n\n25th percentile Median 75th percentile","publish_date":null,"title":"Distribution of euro area member states' public debt ratios 120% 100% 80% 60% 40% 20% 0% 2001 2003 2005 2007","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/5._ESM_Staff_Report_on_Comprehensive_Review_p8.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/5._ESM_Staff_Report_on_Comprehensive_Review_p8.pdf"},"request":{"example_id":"chart/5._ESM_Staff_Report_on_Comprehensive_Review_p8","product_type":"parse","source_file_path":"5._ESM_Staff_Report_on_Comprehensive_Review_p8.pdf"},"started_at":"2026-11-04T19:45:48.103718"} +{"completed_at":"2026-11-04T19:45:50.657301","latency_in_ms":1424,"output":{"example_id":"chart/5._ESM_Staff_Report_on_Comprehensive_Review_p9","grounded_pages":[],"layout_pages":[],"markdown":"9\n\n**Figure 3.** **Euro area demographic dynamics**\n\nSource: Eurostat Note: age dependency is measured as the ratio of population aged 0-14 and 65+ to population aged 15-64.\n\n**Acute climate-related risks may impact many sectors, including notably the financial system** . Both physical risks and uncertainty tied to transition pose the potential for enhanced volatility within the euro area financial infrastructure, with possible effects propagating via numerous channels. ⁵ ESM staff are continuing to contribute to the collective understanding of how these risks are likely to manifest and pass through to financial stability. ⁶ These are the main takeaways:\n\n• **Output losses and reduced growth prospects:** More frequent and extreme weather events could generate large economic losses and reduce production capacity by destroying human and physical capital and reducing productivity growth.\n\n• **Worsening public sector balance sheets:** Large expenditures could emerge from disaster relief, reconstruction, and recovery, as well as the materialisation of contingent liabilities. Beyond the climate mitigation expenditure burden, foregone tax revenue from activity disruptions and diminishing carbon-intensive sources could generate an additional strain on public finances.\n\n• **Deteriorating corporate and financial sectors balance sheets** : Both physical and transition risks could cause large and sudden re-evaluations of the assets of financial and non-financial firms. Notably, the insurance sector will be tested as claims rise, given the increased frequency and magnitude of natural catastrophes. ⁷\n\n5 From Basel Committee on Banking Supervision (2021): Certain economic sectors will have greater sensitivities to acute climate related physical risks or to the transition to a low-carbon economy. These climate-related events and risks are uncertain and may be subject to non-linearities. Credit risks may manifest through both income (reduced ability to repay debts) and wealth effects (capital value erosion affecting recovery rates). Market risks could propagate through sharp reductions in financial asset values and liquidity risks through reduced market conditions, such as via credit line and deposit drawdowns from counterparties. Operational, and reputational risks from increased scrutiny, expectations and market sentiments associated with the transition may equally contribute to further volatility. 6 See ESM 2023 for a discussion on how Regional Financial Arrangements contribute to fight climate change consequences.\n\n7 Swiss Re (2021) estimates that natural catastrophes caused USD 105 billion in global insured losses in 2021, the fourth highest since 1970.\n\n44%\n\n46%\n\n48%\n\n50%\n\n52%\n\n54%\n\n56%\n\n58%\n\n32\n\n34\n\n36\n\n38\n\n40\n\n42\n\n44\n\n46\n\n1984 1987 1990 1993 1996 1999 2002 2005 2008 2011 2014 2017 2020 2023\n\nMedian age of population (LHS) Age dependency ratio (RHS)","pages":[{"line_numbers":[],"markdown":"9\n\n**Figure 3.** **Euro area demographic dynamics**\n\nSource: Eurostat Note: age dependency is measured as the ratio of population aged 0-14 and 65+ to population aged 15-64.\n\n**Acute climate-related risks may impact many sectors, including notably the financial system** . Both physical risks and uncertainty tied to transition pose the potential for enhanced volatility within the euro area financial infrastructure, with possible effects propagating via numerous channels. ⁵ ESM staff are continuing to contribute to the collective understanding of how these risks are likely to manifest and pass through to financial stability. ⁶ These are the main takeaways:\n\n• **Output losses and reduced growth prospects:** More frequent and extreme weather events could generate large economic losses and reduce production capacity by destroying human and physical capital and reducing productivity growth.\n\n• **Worsening public sector balance sheets:** Large expenditures could emerge from disaster relief, reconstruction, and recovery, as well as the materialisation of contingent liabilities. Beyond the climate mitigation expenditure burden, foregone tax revenue from activity disruptions and diminishing carbon-intensive sources could generate an additional strain on public finances.\n\n• **Deteriorating corporate and financial sectors balance sheets** : Both physical and transition risks could cause large and sudden re-evaluations of the assets of financial and non-financial firms. Notably, the insurance sector will be tested as claims rise, given the increased frequency and magnitude of natural catastrophes. ⁷\n\n5 From Basel Committee on Banking Supervision (2021): Certain economic sectors will have greater sensitivities to acute climate related physical risks or to the transition to a low-carbon economy. These climate-related events and risks are uncertain and may be subject to non-linearities. Credit risks may manifest through both income (reduced ability to repay debts) and wealth effects (capital value erosion affecting recovery rates). Market risks could propagate through sharp reductions in financial asset values and liquidity risks through reduced market conditions, such as via credit line and deposit drawdowns from counterparties. Operational, and reputational risks from increased scrutiny, expectations and market sentiments associated with the transition may equally contribute to further volatility. 6 See ESM 2023 for a discussion on how Regional Financial Arrangements contribute to fight climate change consequences.\n\n7 Swiss Re (2021) estimates that natural catastrophes caused USD 105 billion in global insured losses in 2021, the fourth highest since 1970.\n\n44%\n\n46%\n\n48%\n\n50%\n\n52%\n\n54%\n\n56%\n\n58%\n\n32\n\n34\n\n36\n\n38\n\n40\n\n42\n\n44\n\n46\n\n1984 1987 1990 1993 1996 1999 2002 2005 2008 2011 2014 2017 2020 2023\n\nMedian age of population (LHS) Age dependency ratio (RHS)","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"9\n\n**Figure 3.** **Euro area demographic dynamics**\n\nSource: Eurostat Note: age dependency is measured as the ratio of population aged 0-14 and 65+ to population aged 15-64.\n\n**Acute climate-related risks may impact many sectors, including notably the financial system** . Both physical risks and uncertainty tied to transition pose the potential for enhanced volatility within the euro area financial infrastructure, with possible effects propagating via numerous channels. ⁵ ESM staff are continuing to contribute to the collective understanding of how these risks are likely to manifest and pass through to financial stability. ⁶ These are the main takeaways:\n\n• **Output losses and reduced growth prospects:** More frequent and extreme weather events could generate large economic losses and reduce production capacity by destroying human and physical capital and reducing productivity growth.\n\n• **Worsening public sector balance sheets:** Large expenditures could emerge from disaster relief, reconstruction, and recovery, as well as the materialisation of contingent liabilities. Beyond the climate mitigation expenditure burden, foregone tax revenue from activity disruptions and diminishing carbon-intensive sources could generate an additional strain on public finances.\n\n• **Deteriorating corporate and financial sectors balance sheets** : Both physical and transition risks could cause large and sudden re-evaluations of the assets of financial and non-financial firms. Notably, the insurance sector will be tested as claims rise, given the increased frequency and magnitude of natural catastrophes. ⁷\n\n5 From Basel Committee on Banking Supervision (2021): Certain economic sectors will have greater sensitivities to acute climate related physical risks or to the transition to a low-carbon economy. These climate-related events and risks are uncertain and may be subject to non-linearities. Credit risks may manifest through both income (reduced ability to repay debts) and wealth effects (capital value erosion affecting recovery rates). Market risks could propagate through sharp reductions in financial asset values and liquidity risks through reduced market conditions, such as via credit line and deposit drawdowns from counterparties. Operational, and reputational risks from increased scrutiny, expectations and market sentiments associated with the transition may equally contribute to further volatility. 6 See ESM 2023 for a discussion on how Regional Financial Arrangements contribute to fight climate change consequences.\n\n7 Swiss Re (2021) estimates that natural catastrophes caused USD 105 billion in global insured losses in 2021, the fourth highest since 1970.\n\n44%\n\n46%\n\n48%\n\n50%\n\n52%\n\n54%\n\n56%\n\n58%\n\n32\n\n34\n\n36\n\n38\n\n40\n\n42\n\n44\n\n46\n\n1984 1987 1990 1993 1996 1999 2002 2005 2008 2011 2014 2017 2020 2023\n\nMedian age of population (LHS) Age dependency ratio (RHS)","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/5._ESM_Staff_Report_on_Comprehensive_Review_p9.pdf"]},"response":{"errors":[],"extract_id":"extract_af87c12393424f39e7acc1998fe31f10","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["9\n\n**Figure 3.** **Euro area demographic dynamics**\n\nSource: Eurostat Note: age dependency is measured as the ratio of population aged 0-14 and 65+ to population aged 15-64.\n\n**Acute climate-related risks may impact many sectors, including notably the financial system** . Both physical risks and uncertainty tied to transition pose the potential for enhanced volatility within the euro area financial infrastructure, with possible effects propagating via numerous channels. ⁵ ESM staff are continuing to contribute to the collective understanding of how these risks are likely to manifest and pass through to financial stability. ⁶ These are the main takeaways:\n\n• **Output losses and reduced growth prospects:** More frequent and extreme weather events could generate large economic losses and reduce production capacity by destroying human and physical capital and reducing productivity growth.\n\n• **Worsening public sector balance sheets:** Large expenditures could emerge from disaster relief, reconstruction, and recovery, as well as the materialisation of contingent liabilities. Beyond the climate mitigation expenditure burden, foregone tax revenue from activity disruptions and diminishing carbon-intensive sources could generate an additional strain on public finances.\n\n• **Deteriorating corporate and financial sectors balance sheets** : Both physical and transition risks could cause large and sudden re-evaluations of the assets of financial and non-financial firms. Notably, the insurance sector will be tested as claims rise, given the increased frequency and magnitude of natural catastrophes. ⁷\n\n5 From Basel Committee on Banking Supervision (2021): Certain economic sectors will have greater sensitivities to acute climate related physical risks or to the transition to a low-carbon economy. These climate-related events and risks are uncertain and may be subject to non-linearities. Credit risks may manifest through both income (reduced ability to repay debts) and wealth effects (capital value erosion affecting recovery rates). Market risks could propagate through sharp reductions in financial asset values and liquidity risks through reduced market conditions, such as via credit line and deposit drawdowns from counterparties. Operational, and reputational risks from increased scrutiny, expectations and market sentiments associated with the transition may equally contribute to further volatility. 6 See ESM 2023 for a discussion on how Regional Financial Arrangements contribute to fight climate change consequences.\n\n7 Swiss Re (2021) estimates that natural catastrophes caused USD 105 billion in global insured losses in 2021, the fourth highest since 1970.\n\n44%\n\n46%\n\n48%\n\n50%\n\n52%\n\n54%\n\n56%\n\n58%\n\n32\n\n34\n\n36\n\n38\n\n40\n\n42\n\n44\n\n46\n\n1984 1987 1990 1993 1996 1999 2002 2005 2008 2011 2014 2017 2020 2023\n\nMedian age of population (LHS) Age dependency ratio (RHS)"],"full_content":"9\n\n**Figure 3.** **Euro area demographic dynamics**\n\nSource: Eurostat Note: age dependency is measured as the ratio of population aged 0-14 and 65+ to population aged 15-64.\n\n**Acute climate-related risks may impact many sectors, including notably the financial system** . Both physical risks and uncertainty tied to transition pose the potential for enhanced volatility within the euro area financial infrastructure, with possible effects propagating via numerous channels. ⁵ ESM staff are continuing to contribute to the collective understanding of how these risks are likely to manifest and pass through to financial stability. ⁶ These are the main takeaways:\n\n• **Output losses and reduced growth prospects:** More frequent and extreme weather events could generate large economic losses and reduce production capacity by destroying human and physical capital and reducing productivity growth.\n\n• **Worsening public sector balance sheets:** Large expenditures could emerge from disaster relief, reconstruction, and recovery, as well as the materialisation of contingent liabilities. Beyond the climate mitigation expenditure burden, foregone tax revenue from activity disruptions and diminishing carbon-intensive sources could generate an additional strain on public finances.\n\n• **Deteriorating corporate and financial sectors balance sheets** : Both physical and transition risks could cause large and sudden re-evaluations of the assets of financial and non-financial firms. Notably, the insurance sector will be tested as claims rise, given the increased frequency and magnitude of natural catastrophes. ⁷\n\n5 From Basel Committee on Banking Supervision (2021): Certain economic sectors will have greater sensitivities to acute climate related physical risks or to the transition to a low-carbon economy. These climate-related events and risks are uncertain and may be subject to non-linearities. Credit risks may manifest through both income (reduced ability to repay debts) and wealth effects (capital value erosion affecting recovery rates). Market risks could propagate through sharp reductions in financial asset values and liquidity risks through reduced market conditions, such as via credit line and deposit drawdowns from counterparties. Operational, and reputational risks from increased scrutiny, expectations and market sentiments associated with the transition may equally contribute to further volatility. 6 See ESM 2023 for a discussion on how Regional Financial Arrangements contribute to fight climate change consequences.\n\n7 Swiss Re (2021) estimates that natural catastrophes caused USD 105 billion in global insured losses in 2021, the fourth highest since 1970.\n\n44%\n\n46%\n\n48%\n\n50%\n\n52%\n\n54%\n\n56%\n\n58%\n\n32\n\n34\n\n36\n\n38\n\n40\n\n42\n\n44\n\n46\n\n1984 1987 1990 1993 1996 1999 2002 2005 2008 2011 2014 2017 2020 2023\n\nMedian age of population (LHS) Age dependency ratio (RHS)","publish_date":null,"title":"46 44 42 40 38 36 34 32 1984 1987 1990 1993 1996 1999 2002 2005 2008 2011 2014 2017","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/5._ESM_Staff_Report_on_Comprehensive_Review_p9.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/5._ESM_Staff_Report_on_Comprehensive_Review_p9.pdf"},"request":{"example_id":"chart/5._ESM_Staff_Report_on_Comprehensive_Review_p9","product_type":"parse","source_file_path":"5._ESM_Staff_Report_on_Comprehensive_Review_p9.pdf"},"started_at":"2026-11-04T19:45:49.232351"} +{"completed_at":"2026-11-04T19:45:52.508203","latency_in_ms":2084,"output":{"example_id":"chart/67c07d7f417ce_p17","grounded_pages":[],"layout_pages":[],"markdown":"GIE\\_AXA\\_Internal\n\nFY23\n\n2\\.5\n\nFY24\n\nCommercial\n\nlines\n\nAXA XL\n\nReinsurance\n\nRetail\n\nlines\n\n53\\.0\n\n**56\\.5**\n\n**\\+7%**\n\n# **P&C** | Remaining disciplined on pricing while accelerating volume growth\n\nIn Euro billion\n\n## GWP & Other Revenues\n\n17\n\n### **\\+6%**\n\n### Change o/w\n\npricing ¹\n\n### o/w\n\nvolume ²\n\n### \\+3% +3%\n\n### **\\+10%** \\+5% +5%\n\n### **\\+7%** \\+10% -3%\n\n▶ Continued favorable pricing and higher volumes reflecting\n\nbetter customer retention and attractive margins\n\n▶ Successful diversification at AXA XL Reinsurance towards\n\nSpecialty and Casualty against a favorable industry pricing back-drop\n\n▶ Pricing actions in the UK & Ireland and Germany to restore\n\nprofitability completed\n\n▶ +3% volume growth excluding UK & Ireland and Germany\n\nFull Year 2024 Earnings Change at constant scope and FX. 1. Price effect. 2. Includes exposure adjustments and mix & other effects.\n\n34\\.9\n\n19\\.1","pages":[{"line_numbers":[],"markdown":"GIE\\_AXA\\_Internal\n\nFY23\n\n2\\.5\n\nFY24\n\nCommercial\n\nlines\n\nAXA XL\n\nReinsurance\n\nRetail\n\nlines\n\n53\\.0\n\n**56\\.5**\n\n**\\+7%**\n\n# **P&C** | Remaining disciplined on pricing while accelerating volume growth\n\nIn Euro billion\n\n## GWP & Other Revenues\n\n17\n\n### **\\+6%**\n\n### Change o/w\n\npricing ¹\n\n### o/w\n\nvolume ²\n\n### \\+3% +3%\n\n### **\\+10%** \\+5% +5%\n\n### **\\+7%** \\+10% -3%\n\n▶ Continued favorable pricing and higher volumes reflecting\n\nbetter customer retention and attractive margins\n\n▶ Successful diversification at AXA XL Reinsurance towards\n\nSpecialty and Casualty against a favorable industry pricing back-drop\n\n▶ Pricing actions in the UK & Ireland and Germany to restore\n\nprofitability completed\n\n▶ +3% volume growth excluding UK & Ireland and Germany\n\nFull Year 2024 Earnings Change at constant scope and FX. 1. Price effect. 2. Includes exposure adjustments and mix & other effects.\n\n34\\.9\n\n19\\.1","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"GIE\\_AXA\\_Internal\n\nFY23\n\n2\\.5\n\nFY24\n\nCommercial\n\nlines\n\nAXA XL\n\nReinsurance\n\nRetail\n\nlines\n\n53\\.0\n\n**56\\.5**\n\n**\\+7%**\n\n# **P&C** | Remaining disciplined on pricing while accelerating volume growth\n\nIn Euro billion\n\n## GWP & Other Revenues\n\n17\n\n### **\\+6%**\n\n### Change o/w\n\npricing ¹\n\n### o/w\n\nvolume ²\n\n### \\+3% +3%\n\n### **\\+10%** \\+5% +5%\n\n### **\\+7%** \\+10% -3%\n\n▶ Continued favorable pricing and higher volumes reflecting\n\nbetter customer retention and attractive margins\n\n▶ Successful diversification at AXA XL Reinsurance towards\n\nSpecialty and Casualty against a favorable industry pricing back-drop\n\n▶ Pricing actions in the UK & Ireland and Germany to restore\n\nprofitability completed\n\n▶ +3% volume growth excluding UK & Ireland and Germany\n\nFull Year 2024 Earnings Change at constant scope and FX. 1. Price effect. 2. Includes exposure adjustments and mix & other effects.\n\n34\\.9\n\n19\\.1","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/67c07d7f417ce_p17.pdf"]},"response":{"errors":[],"extract_id":"extract_f2993e028ea9ddf2b8e6e302dc180246","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["GIE\\_AXA\\_Internal\n\nFY23\n\n2\\.5\n\nFY24\n\nCommercial\n\nlines\n\nAXA XL\n\nReinsurance\n\nRetail\n\nlines\n\n53\\.0\n\n**56\\.5**\n\n**\\+7%**\n\n# **P&C** | Remaining disciplined on pricing while accelerating volume growth\n\nIn Euro billion\n\n## GWP & Other Revenues\n\n17\n\n### **\\+6%**\n\n### Change o/w\n\npricing ¹\n\n### o/w\n\nvolume ²\n\n### \\+3% +3%\n\n### **\\+10%** \\+5% +5%\n\n### **\\+7%** \\+10% -3%\n\n▶ Continued favorable pricing and higher volumes reflecting\n\nbetter customer retention and attractive margins\n\n▶ Successful diversification at AXA XL Reinsurance towards\n\nSpecialty and Casualty against a favorable industry pricing back-drop\n\n▶ Pricing actions in the UK & Ireland and Germany to restore\n\nprofitability completed\n\n▶ +3% volume growth excluding UK & Ireland and Germany\n\nFull Year 2024 Earnings Change at constant scope and FX. 1. Price effect. 2. Includes exposure adjustments and mix & other effects.\n\n34\\.9\n\n19\\.1"],"full_content":"GIE\\_AXA\\_Internal\n\nFY23\n\n2\\.5\n\nFY24\n\nCommercial\n\nlines\n\nAXA XL\n\nReinsurance\n\nRetail\n\nlines\n\n53\\.0\n\n**56\\.5**\n\n**\\+7%**\n\n# **P&C** | Remaining disciplined on pricing while accelerating volume growth\n\nIn Euro billion\n\n## GWP & Other Revenues\n\n17\n\n### **\\+6%**\n\n### Change o/w\n\npricing ¹\n\n### o/w\n\nvolume ²\n\n### \\+3% +3%\n\n### **\\+10%** \\+5% +5%\n\n### **\\+7%** \\+10% -3%\n\n▶ Continued favorable pricing and higher volumes reflecting\n\nbetter customer retention and attractive margins\n\n▶ Successful diversification at AXA XL Reinsurance towards\n\nSpecialty and Casualty against a favorable industry pricing back-drop\n\n▶ Pricing actions in the UK & Ireland and Germany to restore\n\nprofitability completed\n\n▶ +3% volume growth excluding UK & Ireland and Germany\n\nFull Year 2024 Earnings Change at constant scope and FX. 1. Price effect. 2. Includes exposure adjustments and mix & other effects.\n\n34\\.9\n\n19\\.1","publish_date":null,"title":"P&C | Remaining disciplined on pricing while accelerating volume growth","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/67c07d7f417ce_p17.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/67c07d7f417ce_p17.pdf"},"request":{"example_id":"chart/67c07d7f417ce_p17","product_type":"parse","source_file_path":"67c07d7f417ce_p17.pdf"},"started_at":"2026-11-04T19:45:50.424024"} +{"completed_at":"2026-11-04T19:45:53.815243","latency_in_ms":1302,"output":{"example_id":"chart/67c07d7f417ce_p28","grounded_pages":[],"layout_pages":[],"markdown":"GIE\\_AXA\\_Internal\n\n# Delivering compelling returns to shareholders\n\n28 Full Year 2024 Earnings\n\n3\\.8 4.4\n\n1\\.1\n\n1\\.1\n\n0\\.7\n\n### FY23 FY24\n\n**1\\.2**\n\n**4\\.7**\n\n### **FY25E**\n\n### 4\\.9\n\n### 6\\.2\n\nDividend Annual share buy-back Anti-dilutive share buy-back\n\n## **Total capital return**\n\n▶ **\\+9% dividend per share growth** **1**\n\n### FY24 vs FY23\n\n▶ **€1.2bn annual share buy** **\\-back** **2**\n\n### to be executed in 2025\n\n### ▶ **Additional ca. €3.8bn share buy** **\\-back** expected to be executed\n\n### in 2025 & 2026 following the expected sale of AXA IM for €5.4bn\n\n### ▶ Attractive payout while **retaining earnings to drive organic**\n\n### **growth and compound book value per share over time**\n\nIn Euro billion\n\n1\\. Dividend to be paid in 2025 related to FY24 Underlying earnings. Based on the dividend proposed by AXA’s Board of Directors on February 26, 2025 and subject to approval by the Shareholders’ Annual General Meeting to be held on April 24, 2025. 2. Following AXA’s Board of Directors’ approval on February 26, 2025, and expected to commence as soon as reasonably practicable, subject to market conditions.","pages":[{"line_numbers":[],"markdown":"GIE\\_AXA\\_Internal\n\n# Delivering compelling returns to shareholders\n\n28 Full Year 2024 Earnings\n\n3\\.8 4.4\n\n1\\.1\n\n1\\.1\n\n0\\.7\n\n### FY23 FY24\n\n**1\\.2**\n\n**4\\.7**\n\n### **FY25E**\n\n### 4\\.9\n\n### 6\\.2\n\nDividend Annual share buy-back Anti-dilutive share buy-back\n\n## **Total capital return**\n\n▶ **\\+9% dividend per share growth** **1**\n\n### FY24 vs FY23\n\n▶ **€1.2bn annual share buy** **\\-back** **2**\n\n### to be executed in 2025\n\n### ▶ **Additional ca. €3.8bn share buy** **\\-back** expected to be executed\n\n### in 2025 & 2026 following the expected sale of AXA IM for €5.4bn\n\n### ▶ Attractive payout while **retaining earnings to drive organic**\n\n### **growth and compound book value per share over time**\n\nIn Euro billion\n\n1\\. Dividend to be paid in 2025 related to FY24 Underlying earnings. Based on the dividend proposed by AXA’s Board of Directors on February 26, 2025 and subject to approval by the Shareholders’ Annual General Meeting to be held on April 24, 2025. 2. Following AXA’s Board of Directors’ approval on February 26, 2025, and expected to commence as soon as reasonably practicable, subject to market conditions.","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"GIE\\_AXA\\_Internal\n\n# Delivering compelling returns to shareholders\n\n28 Full Year 2024 Earnings\n\n3\\.8 4.4\n\n1\\.1\n\n1\\.1\n\n0\\.7\n\n### FY23 FY24\n\n**1\\.2**\n\n**4\\.7**\n\n### **FY25E**\n\n### 4\\.9\n\n### 6\\.2\n\nDividend Annual share buy-back Anti-dilutive share buy-back\n\n## **Total capital return**\n\n▶ **\\+9% dividend per share growth** **1**\n\n### FY24 vs FY23\n\n▶ **€1.2bn annual share buy** **\\-back** **2**\n\n### to be executed in 2025\n\n### ▶ **Additional ca. €3.8bn share buy** **\\-back** expected to be executed\n\n### in 2025 & 2026 following the expected sale of AXA IM for €5.4bn\n\n### ▶ Attractive payout while **retaining earnings to drive organic**\n\n### **growth and compound book value per share over time**\n\nIn Euro billion\n\n1\\. Dividend to be paid in 2025 related to FY24 Underlying earnings. Based on the dividend proposed by AXA’s Board of Directors on February 26, 2025 and subject to approval by the Shareholders’ Annual General Meeting to be held on April 24, 2025. 2. Following AXA’s Board of Directors’ approval on February 26, 2025, and expected to commence as soon as reasonably practicable, subject to market conditions.","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/67c07d7f417ce_p28.pdf"]},"response":{"errors":[],"extract_id":"extract_5af4615a8ac99580f70f8f0771c4d9f7","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["GIE\\_AXA\\_Internal\n\n# Delivering compelling returns to shareholders\n\n28 Full Year 2024 Earnings\n\n3\\.8 4.4\n\n1\\.1\n\n1\\.1\n\n0\\.7\n\n### FY23 FY24\n\n**1\\.2**\n\n**4\\.7**\n\n### **FY25E**\n\n### 4\\.9\n\n### 6\\.2\n\nDividend Annual share buy-back Anti-dilutive share buy-back\n\n## **Total capital return**\n\n▶ **\\+9% dividend per share growth** **1**\n\n### FY24 vs FY23\n\n▶ **€1.2bn annual share buy** **\\-back** **2**\n\n### to be executed in 2025\n\n### ▶ **Additional ca. €3.8bn share buy** **\\-back** expected to be executed\n\n### in 2025 & 2026 following the expected sale of AXA IM for €5.4bn\n\n### ▶ Attractive payout while **retaining earnings to drive organic**\n\n### **growth and compound book value per share over time**\n\nIn Euro billion\n\n1\\. Dividend to be paid in 2025 related to FY24 Underlying earnings. Based on the dividend proposed by AXA’s Board of Directors on February 26, 2025 and subject to approval by the Shareholders’ Annual General Meeting to be held on April 24, 2025. 2. Following AXA’s Board of Directors’ approval on February 26, 2025, and expected to commence as soon as reasonably practicable, subject to market conditions."],"full_content":"GIE\\_AXA\\_Internal\n\n# Delivering compelling returns to shareholders\n\n28 Full Year 2024 Earnings\n\n3\\.8 4.4\n\n1\\.1\n\n1\\.1\n\n0\\.7\n\n### FY23 FY24\n\n**1\\.2**\n\n**4\\.7**\n\n### **FY25E**\n\n### 4\\.9\n\n### 6\\.2\n\nDividend Annual share buy-back Anti-dilutive share buy-back\n\n## **Total capital return**\n\n▶ **\\+9% dividend per share growth** **1**\n\n### FY24 vs FY23\n\n▶ **€1.2bn annual share buy** **\\-back** **2**\n\n### to be executed in 2025\n\n### ▶ **Additional ca. €3.8bn share buy** **\\-back** expected to be executed\n\n### in 2025 & 2026 following the expected sale of AXA IM for €5.4bn\n\n### ▶ Attractive payout while **retaining earnings to drive organic**\n\n### **growth and compound book value per share over time**\n\nIn Euro billion\n\n1\\. Dividend to be paid in 2025 related to FY24 Underlying earnings. Based on the dividend proposed by AXA’s Board of Directors on February 26, 2025 and subject to approval by the Shareholders’ Annual General Meeting to be held on April 24, 2025. 2. Following AXA’s Board of Directors’ approval on February 26, 2025, and expected to commence as soon as reasonably practicable, subject to market conditions.","publish_date":null,"title":"Delivering compelling returns to shareholders","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/67c07d7f417ce_p28.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/67c07d7f417ce_p28.pdf"},"request":{"example_id":"chart/67c07d7f417ce_p28","product_type":"parse","source_file_path":"67c07d7f417ce_p28.pdf"},"started_at":"2026-11-04T19:45:52.512609"} +{"completed_at":"2026-11-04T19:45:55.208376","latency_in_ms":1105,"output":{"example_id":"chart/67c07d7f417ce_p43","grounded_pages":[],"layout_pages":[],"markdown":"## P&C | 2025 Simplified Group Nat Cat Reinsurance Program ¹\n\n## Stable reinsurance program for 2025\n\nIn Euro\n\n### **Insurance segment**\n\n### (occurrence protection)\n\nAlternative Capital & Cat Bonds\n\n### **Reinsurance**\n\n### **segment**\n\n(illustrative)\n\n43 Full Year 2024 Earnings\n\n3\\.9bn 1.2bn 1.2bn\n\n450m 400m 600m 600m 600m\n\n4\\.0bn\n\n2\\.1bn 2.1bn 1.2bn\n\n1\\.0bn\n\n1\\.2bn\n\nEU\n\nWindstorm\n\nEurope Flood\n\nEurope\n\nEarthquake\n\nPer other\n\nperils ²\n\nNA\n\nEarthquake\n\nNA\n\nHurricane\n\n1\\. Excludes local reinsurance covers 2. Other perils include Türkiye earthquake, NA floods and wildfires, as well as a series of other secondary perils. Capacity varies by peril type.\n\n400m **Retention**\n\n**Capacity**","pages":[{"line_numbers":[],"markdown":"## P&C | 2025 Simplified Group Nat Cat Reinsurance Program ¹\n\n## Stable reinsurance program for 2025\n\nIn Euro\n\n### **Insurance segment**\n\n### (occurrence protection)\n\nAlternative Capital & Cat Bonds\n\n### **Reinsurance**\n\n### **segment**\n\n(illustrative)\n\n43 Full Year 2024 Earnings\n\n3\\.9bn 1.2bn 1.2bn\n\n450m 400m 600m 600m 600m\n\n4\\.0bn\n\n2\\.1bn 2.1bn 1.2bn\n\n1\\.0bn\n\n1\\.2bn\n\nEU\n\nWindstorm\n\nEurope Flood\n\nEurope\n\nEarthquake\n\nPer other\n\nperils ²\n\nNA\n\nEarthquake\n\nNA\n\nHurricane\n\n1\\. Excludes local reinsurance covers 2. Other perils include Türkiye earthquake, NA floods and wildfires, as well as a series of other secondary perils. Capacity varies by peril type.\n\n400m **Retention**\n\n**Capacity**","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"## P&C | 2025 Simplified Group Nat Cat Reinsurance Program ¹\n\n## Stable reinsurance program for 2025\n\nIn Euro\n\n### **Insurance segment**\n\n### (occurrence protection)\n\nAlternative Capital & Cat Bonds\n\n### **Reinsurance**\n\n### **segment**\n\n(illustrative)\n\n43 Full Year 2024 Earnings\n\n3\\.9bn 1.2bn 1.2bn\n\n450m 400m 600m 600m 600m\n\n4\\.0bn\n\n2\\.1bn 2.1bn 1.2bn\n\n1\\.0bn\n\n1\\.2bn\n\nEU\n\nWindstorm\n\nEurope Flood\n\nEurope\n\nEarthquake\n\nPer other\n\nperils ²\n\nNA\n\nEarthquake\n\nNA\n\nHurricane\n\n1\\. Excludes local reinsurance covers 2. Other perils include Türkiye earthquake, NA floods and wildfires, as well as a series of other secondary perils. Capacity varies by peril type.\n\n400m **Retention**\n\n**Capacity**","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/67c07d7f417ce_p43.pdf"]},"response":{"errors":[],"extract_id":"extract_622947a25f10f4af76b1aa104c8469cb","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["## P&C | 2025 Simplified Group Nat Cat Reinsurance Program ¹\n\n## Stable reinsurance program for 2025\n\nIn Euro\n\n### **Insurance segment**\n\n### (occurrence protection)\n\nAlternative Capital & Cat Bonds\n\n### **Reinsurance**\n\n### **segment**\n\n(illustrative)\n\n43 Full Year 2024 Earnings\n\n3\\.9bn 1.2bn 1.2bn\n\n450m 400m 600m 600m 600m\n\n4\\.0bn\n\n2\\.1bn 2.1bn 1.2bn\n\n1\\.0bn\n\n1\\.2bn\n\nEU\n\nWindstorm\n\nEurope Flood\n\nEurope\n\nEarthquake\n\nPer other\n\nperils ²\n\nNA\n\nEarthquake\n\nNA\n\nHurricane\n\n1\\. Excludes local reinsurance covers 2. Other perils include Türkiye earthquake, NA floods and wildfires, as well as a series of other secondary perils. Capacity varies by peril type.\n\n400m **Retention**\n\n**Capacity**"],"full_content":"## P&C | 2025 Simplified Group Nat Cat Reinsurance Program ¹\n\n## Stable reinsurance program for 2025\n\nIn Euro\n\n### **Insurance segment**\n\n### (occurrence protection)\n\nAlternative Capital & Cat Bonds\n\n### **Reinsurance**\n\n### **segment**\n\n(illustrative)\n\n43 Full Year 2024 Earnings\n\n3\\.9bn 1.2bn 1.2bn\n\n450m 400m 600m 600m 600m\n\n4\\.0bn\n\n2\\.1bn 2.1bn 1.2bn\n\n1\\.0bn\n\n1\\.2bn\n\nEU\n\nWindstorm\n\nEurope Flood\n\nEurope\n\nEarthquake\n\nPer other\n\nperils ²\n\nNA\n\nEarthquake\n\nNA\n\nHurricane\n\n1\\. Excludes local reinsurance covers 2. Other perils include Türkiye earthquake, NA floods and wildfires, as well as a series of other secondary perils. Capacity varies by peril type.\n\n400m **Retention**\n\n**Capacity**","publish_date":null,"title":"P&C | 2025 Simplified Group Nat Cat Reinsurance Program","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/67c07d7f417ce_p43.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/67c07d7f417ce_p43.pdf"},"request":{"example_id":"chart/67c07d7f417ce_p43","product_type":"parse","source_file_path":"67c07d7f417ce_p43.pdf"},"started_at":"2026-11-04T19:45:54.102923"} +{"completed_at":"2026-11-04T19:31:12.872888","latency_in_ms":1568,"output":{"example_id":"chart/67c07d7f417ce_p44","grounded_pages":[],"layout_pages":[],"markdown":"## P&C | AXA Group earnings deviation with different levels of Nat Cat cost ¹ in 2025\n\n44\n\n### Average expected Nat Cat charges\n\n2\\.5\n\n**2\\.7**\n\n2024 2025\n\n### Group underlying earnings deviation to average Nat Cat charges in 2025\n\nIn Euro billion\n\nnet of reinsurance, post-tax net of reinsurance, pre-tax\n\nMedian\n\n(50 th )\n\n€+0.1bn\n\n€+0.8bn\n\n€+0.7bn\n\n€+0.6bn\n\n1/10y (10 th )\n\n1/5y (20 th )\n\n1/20y\n\n(5 th )\n\n€-1.3bn\n\n€-0.8bn\n\n€-0.4n\n\n1/10y (90 th )\n\n1/5y (80 th )\n\n1/20y (95 th )\n\n**_More severe years Negative deviation in ca. 40% of cases_**\n\n1\\. Natural catastrophe cost defined as Aggregate Exceedance Probability (AEP) of all natural perils worldwide, net of tax and reinsurance. Deviation is compared to a normalized level, which represents costs associated with natural catastrophes expected in an average year (ca. 4.5 points of estimated FY25 GEP, undiscounted and net of reinsurance). Full Year 2024 Earnings\n\nEstimated\n\nimpact on GEP\n\nca.\n\n**4\\.5%**\n\nca.\n\n**4\\.5%**\n\n**_Less severe years Positive deviation in ca. 60% of cases_**","pages":[{"line_numbers":[],"markdown":"## P&C | AXA Group earnings deviation with different levels of Nat Cat cost ¹ in 2025\n\n44\n\n### Average expected Nat Cat charges\n\n2\\.5\n\n**2\\.7**\n\n2024 2025\n\n### Group underlying earnings deviation to average Nat Cat charges in 2025\n\nIn Euro billion\n\nnet of reinsurance, post-tax net of reinsurance, pre-tax\n\nMedian\n\n(50 th )\n\n€+0.1bn\n\n€+0.8bn\n\n€+0.7bn\n\n€+0.6bn\n\n1/10y (10 th )\n\n1/5y (20 th )\n\n1/20y\n\n(5 th )\n\n€-1.3bn\n\n€-0.8bn\n\n€-0.4n\n\n1/10y (90 th )\n\n1/5y (80 th )\n\n1/20y (95 th )\n\n**_More severe years Negative deviation in ca. 40% of cases_**\n\n1\\. Natural catastrophe cost defined as Aggregate Exceedance Probability (AEP) of all natural perils worldwide, net of tax and reinsurance. Deviation is compared to a normalized level, which represents costs associated with natural catastrophes expected in an average year (ca. 4.5 points of estimated FY25 GEP, undiscounted and net of reinsurance). Full Year 2024 Earnings\n\nEstimated\n\nimpact on GEP\n\nca.\n\n**4\\.5%**\n\nca.\n\n**4\\.5%**\n\n**_Less severe years Positive deviation in ca. 60% of cases_**","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"## P&C | AXA Group earnings deviation with different levels of Nat Cat cost ¹ in 2025\n\n44\n\n### Average expected Nat Cat charges\n\n2\\.5\n\n**2\\.7**\n\n2024 2025\n\n### Group underlying earnings deviation to average Nat Cat charges in 2025\n\nIn Euro billion\n\nnet of reinsurance, post-tax net of reinsurance, pre-tax\n\nMedian\n\n(50 th )\n\n€+0.1bn\n\n€+0.8bn\n\n€+0.7bn\n\n€+0.6bn\n\n1/10y (10 th )\n\n1/5y (20 th )\n\n1/20y\n\n(5 th )\n\n€-1.3bn\n\n€-0.8bn\n\n€-0.4n\n\n1/10y (90 th )\n\n1/5y (80 th )\n\n1/20y (95 th )\n\n**_More severe years Negative deviation in ca. 40% of cases_**\n\n1\\. Natural catastrophe cost defined as Aggregate Exceedance Probability (AEP) of all natural perils worldwide, net of tax and reinsurance. Deviation is compared to a normalized level, which represents costs associated with natural catastrophes expected in an average year (ca. 4.5 points of estimated FY25 GEP, undiscounted and net of reinsurance). Full Year 2024 Earnings\n\nEstimated\n\nimpact on GEP\n\nca.\n\n**4\\.5%**\n\nca.\n\n**4\\.5%**\n\n**_Less severe years Positive deviation in ca. 60% of cases_**","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/67c07d7f417ce_p44.pdf"]},"response":{"errors":[],"extract_id":"extract_038d9fa7ab1ebfba25b2f31208d7c64a","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["## P&C | AXA Group earnings deviation with different levels of Nat Cat cost ¹ in 2025\n\n44\n\n### Average expected Nat Cat charges\n\n2\\.5\n\n**2\\.7**\n\n2024 2025\n\n### Group underlying earnings deviation to average Nat Cat charges in 2025\n\nIn Euro billion\n\nnet of reinsurance, post-tax net of reinsurance, pre-tax\n\nMedian\n\n(50 th )\n\n€+0.1bn\n\n€+0.8bn\n\n€+0.7bn\n\n€+0.6bn\n\n1/10y (10 th )\n\n1/5y (20 th )\n\n1/20y\n\n(5 th )\n\n€-1.3bn\n\n€-0.8bn\n\n€-0.4n\n\n1/10y (90 th )\n\n1/5y (80 th )\n\n1/20y (95 th )\n\n**_More severe years Negative deviation in ca. 40% of cases_**\n\n1\\. Natural catastrophe cost defined as Aggregate Exceedance Probability (AEP) of all natural perils worldwide, net of tax and reinsurance. Deviation is compared to a normalized level, which represents costs associated with natural catastrophes expected in an average year (ca. 4.5 points of estimated FY25 GEP, undiscounted and net of reinsurance). Full Year 2024 Earnings\n\nEstimated\n\nimpact on GEP\n\nca.\n\n**4\\.5%**\n\nca.\n\n**4\\.5%**\n\n**_Less severe years Positive deviation in ca. 60% of cases_**"],"full_content":"## P&C | AXA Group earnings deviation with different levels of Nat Cat cost ¹ in 2025\n\n44\n\n### Average expected Nat Cat charges\n\n2\\.5\n\n**2\\.7**\n\n2024 2025\n\n### Group underlying earnings deviation to average Nat Cat charges in 2025\n\nIn Euro billion\n\nnet of reinsurance, post-tax net of reinsurance, pre-tax\n\nMedian\n\n(50 th )\n\n€+0.1bn\n\n€+0.8bn\n\n€+0.7bn\n\n€+0.6bn\n\n1/10y (10 th )\n\n1/5y (20 th )\n\n1/20y\n\n(5 th )\n\n€-1.3bn\n\n€-0.8bn\n\n€-0.4n\n\n1/10y (90 th )\n\n1/5y (80 th )\n\n1/20y (95 th )\n\n**_More severe years Negative deviation in ca. 40% of cases_**\n\n1\\. Natural catastrophe cost defined as Aggregate Exceedance Probability (AEP) of all natural perils worldwide, net of tax and reinsurance. Deviation is compared to a normalized level, which represents costs associated with natural catastrophes expected in an average year (ca. 4.5 points of estimated FY25 GEP, undiscounted and net of reinsurance). Full Year 2024 Earnings\n\nEstimated\n\nimpact on GEP\n\nca.\n\n**4\\.5%**\n\nca.\n\n**4\\.5%**\n\n**_Less severe years Positive deviation in ca. 60% of cases_**","publish_date":null,"title":"P&C | AXA Group earnings deviation with different levels of Nat Cat cost 1 in 2025","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/67c07d7f417ce_p44.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/67c07d7f417ce_p44.pdf"},"request":{"example_id":"chart/67c07d7f417ce_p44","product_type":"parse","source_file_path":"67c07d7f417ce_p44.pdf"},"started_at":"2026-11-04T19:31:11.304715"} +{"completed_at":"2026-11-04T19:45:58.698740","latency_in_ms":1111,"output":{"example_id":"chart/7c6b23db-en_p48","grounded_pages":[],"layout_pages":[],"markdown":"Energy Technology Perspectives 2023 Chapter 1. Energy supply chains in transition\n\nPAGE | 48\n\nI EA. CC BY 4.0.\n\nto be upgraded to accommodate heat pumps (i.e. enlarging radiators or underfloor exchangers), this can add cost. This matters substantially for more efficient ground-source heat pumps, where installation can take up to several weeks and requires drilling and underground piping, making their total cost much higher than other options. Installation time and costs could decline as heat pumps become more common, and offer greater opportunities than in manufacturing. The most expensive components in heat pumps (e.g. heat exchangers, compressors) have already been mass manufactured for a long time, making further manufacturing cost reductions more limited than for other clean energy technologies.\n\n**Global deployment of selected clean energy technologies in the NZE Scenario**\n\nIEA. CC BY 4.0.\n\nNotes: HF = hydrocarbon fuels\n\n**The scale and speed of deployment of clean energy technologies and their associated supply chains accelerate dramatically in the NZE Scenario.**","pages":[{"line_numbers":[],"markdown":"Energy Technology Perspectives 2023 Chapter 1. Energy supply chains in transition\n\nPAGE | 48\n\nI EA. CC BY 4.0.\n\nto be upgraded to accommodate heat pumps (i.e. enlarging radiators or underfloor exchangers), this can add cost. This matters substantially for more efficient ground-source heat pumps, where installation can take up to several weeks and requires drilling and underground piping, making their total cost much higher than other options. Installation time and costs could decline as heat pumps become more common, and offer greater opportunities than in manufacturing. The most expensive components in heat pumps (e.g. heat exchangers, compressors) have already been mass manufactured for a long time, making further manufacturing cost reductions more limited than for other clean energy technologies.\n\n**Global deployment of selected clean energy technologies in the NZE Scenario**\n\nIEA. CC BY 4.0.\n\nNotes: HF = hydrocarbon fuels\n\n**The scale and speed of deployment of clean energy technologies and their associated supply chains accelerate dramatically in the NZE Scenario.**","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Energy Technology Perspectives 2023 Chapter 1. Energy supply chains in transition\n\nPAGE | 48\n\nI EA. CC BY 4.0.\n\nto be upgraded to accommodate heat pumps (i.e. enlarging radiators or underfloor exchangers), this can add cost. This matters substantially for more efficient ground-source heat pumps, where installation can take up to several weeks and requires drilling and underground piping, making their total cost much higher than other options. Installation time and costs could decline as heat pumps become more common, and offer greater opportunities than in manufacturing. The most expensive components in heat pumps (e.g. heat exchangers, compressors) have already been mass manufactured for a long time, making further manufacturing cost reductions more limited than for other clean energy technologies.\n\n**Global deployment of selected clean energy technologies in the NZE Scenario**\n\nIEA. CC BY 4.0.\n\nNotes: HF = hydrocarbon fuels\n\n**The scale and speed of deployment of clean energy technologies and their associated supply chains accelerate dramatically in the NZE Scenario.**","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/7c6b23db-en_p48.pdf"]},"response":{"errors":[],"extract_id":"extract_2bd6a604bd1fa83cd198ef8b811e0b64","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Energy Technology Perspectives 2023 Chapter 1. Energy supply chains in transition\n\nPAGE | 48\n\nI EA. CC BY 4.0.\n\nto be upgraded to accommodate heat pumps (i.e. enlarging radiators or underfloor exchangers), this can add cost. This matters substantially for more efficient ground-source heat pumps, where installation can take up to several weeks and requires drilling and underground piping, making their total cost much higher than other options. Installation time and costs could decline as heat pumps become more common, and offer greater opportunities than in manufacturing. The most expensive components in heat pumps (e.g. heat exchangers, compressors) have already been mass manufactured for a long time, making further manufacturing cost reductions more limited than for other clean energy technologies.\n\n**Global deployment of selected clean energy technologies in the NZE Scenario**\n\nIEA. CC BY 4.0.\n\nNotes: HF = hydrocarbon fuels\n\n**The scale and speed of deployment of clean energy technologies and their associated supply chains accelerate dramatically in the NZE Scenario.**"],"full_content":"Energy Technology Perspectives 2023 Chapter 1. Energy supply chains in transition\n\nPAGE | 48\n\nI EA. CC BY 4.0.\n\nto be upgraded to accommodate heat pumps (i.e. enlarging radiators or underfloor exchangers), this can add cost. This matters substantially for more efficient ground-source heat pumps, where installation can take up to several weeks and requires drilling and underground piping, making their total cost much higher than other options. Installation time and costs could decline as heat pumps become more common, and offer greater opportunities than in manufacturing. The most expensive components in heat pumps (e.g. heat exchangers, compressors) have already been mass manufactured for a long time, making further manufacturing cost reductions more limited than for other clean energy technologies.\n\n**Global deployment of selected clean energy technologies in the NZE Scenario**\n\nIEA. CC BY 4.0.\n\nNotes: HF = hydrocarbon fuels\n\n**The scale and speed of deployment of clean energy technologies and their associated supply chains accelerate dramatically in the NZE Scenario.**","publish_date":null,"title":"to be upgraded to accommodate heat pumps (i.e. enlarging radiators or underfloor exchangers), this can add cost. This matters substantially","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/7c6b23db-en_p48.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/7c6b23db-en_p48.pdf"},"request":{"example_id":"chart/7c6b23db-en_p48","product_type":"parse","source_file_path":"7c6b23db-en_p48.pdf"},"started_at":"2026-11-04T19:45:57.587346"} +{"completed_at":"2026-11-04T19:46:01.380143","latency_in_ms":3074,"output":{"example_id":"chart/7c6b23db-en_p57","grounded_pages":[],"layout_pages":[],"markdown":"Energy Technology Perspectives 2023 Chapter 1. Energy supply chains in transition\n\nPAGE | 57\n\nI EA. CC BY 4.0.\n\n### More investments are needed urgently to avoid bottlenecks\n\nBridging the production gaps described above would require enormous additional investments in the coming years. The total cumulative investment in mining, critical material production and manufacturing of the clean energy technologies selected in this report required to bring the necessary capacity online by 2030 amounts to around USD 1.2 trillion, ⁷ in 2021 dollars (Figure 1.11). Investments associated with projects already announced reach around60% of this, with large gaps in critical mineral mining and for some technology manufacturing.\n\nWhile mobilising such levels of investment is achievable, the timeline is extremely tight, especially given the long lead times to bring production to market. Most of the investment (including for projects already announced) needs to occur over 2023-2025, implying an average of more than USD 270 billion per year over the period. To compare, this is about two-thirds of current annual capital spending of the oil and gas industry, and nearly seven times higher than the average annual investment seen in clean energy supply chains over 2016-2021.\n\n**Global investment in selected clean energy supply chains needed to bring**\n\n**online enough capacity in 2030 in the NZE Scenario, by supply chain step**\n\nIEA. CC BY 4.0.\n\nNotes: CAPEX = capital expenditures. Only refers to the investments needed to bring online enough capacity in 2030 – not counting what would be needed to further scale up in subsequent years – for the subset of clean energy technologies selected in _ETP_ \\- _2023_ . Assuming construction times of five years for mining, three for processing plants and two for technology manufacturing. Investments are assumed uniform over the available period. Excludes site-tailored technology installation. Sources: IEA analysis based on company announcement; Bartholomeusz (2022); S&P Capital (2022).\n\n**Most of the supply chain investments needed to meet NZE Scenario targets in 2030 are made over 2023-2025, at an annual average seven times higher than over 2016-2021.**\n\n7 This refers only to the investments needed to bring enough capacity online in 2030 for the technologies selected in _ETP-2023_ , excluding the investments that would also be needed by 2030 to further scale up deployment in subsequent years.\n\n0\n\n50\n\n100\n\n150\n\n200\n\n2016-2021 2023-2030\n\nBillion USD\n\nCopper mining Other Critical mineral mining Critical mineral processing Clean technology manufacuring\n\n**Average yearly CAPEX**\n\n0\n\n500\n\n1 000\n\n1 500\n\n2023-2030\n\nBillion USD\n\n**Cumulative CAPEX**\n\nFinal investment decision year\n\n2023 2024 2025 2026 2027 2028 2029 2030\n\n**Time available for final investment decision**","pages":[{"line_numbers":[],"markdown":"Energy Technology Perspectives 2023 Chapter 1. Energy supply chains in transition\n\nPAGE | 57\n\nI EA. CC BY 4.0.\n\n### More investments are needed urgently to avoid bottlenecks\n\nBridging the production gaps described above would require enormous additional investments in the coming years. The total cumulative investment in mining, critical material production and manufacturing of the clean energy technologies selected in this report required to bring the necessary capacity online by 2030 amounts to around USD 1.2 trillion, ⁷ in 2021 dollars (Figure 1.11). Investments associated with projects already announced reach around60% of this, with large gaps in critical mineral mining and for some technology manufacturing.\n\nWhile mobilising such levels of investment is achievable, the timeline is extremely tight, especially given the long lead times to bring production to market. Most of the investment (including for projects already announced) needs to occur over 2023-2025, implying an average of more than USD 270 billion per year over the period. To compare, this is about two-thirds of current annual capital spending of the oil and gas industry, and nearly seven times higher than the average annual investment seen in clean energy supply chains over 2016-2021.\n\n**Global investment in selected clean energy supply chains needed to bring**\n\n**online enough capacity in 2030 in the NZE Scenario, by supply chain step**\n\nIEA. CC BY 4.0.\n\nNotes: CAPEX = capital expenditures. Only refers to the investments needed to bring online enough capacity in 2030 – not counting what would be needed to further scale up in subsequent years – for the subset of clean energy technologies selected in _ETP_ \\- _2023_ . Assuming construction times of five years for mining, three for processing plants and two for technology manufacturing. Investments are assumed uniform over the available period. Excludes site-tailored technology installation. Sources: IEA analysis based on company announcement; Bartholomeusz (2022); S&P Capital (2022).\n\n**Most of the supply chain investments needed to meet NZE Scenario targets in 2030 are made over 2023-2025, at an annual average seven times higher than over 2016-2021.**\n\n7 This refers only to the investments needed to bring enough capacity online in 2030 for the technologies selected in _ETP-2023_ , excluding the investments that would also be needed by 2030 to further scale up deployment in subsequent years.\n\n0\n\n50\n\n100\n\n150\n\n200\n\n2016-2021 2023-2030\n\nBillion USD\n\nCopper mining Other Critical mineral mining Critical mineral processing Clean technology manufacuring\n\n**Average yearly CAPEX**\n\n0\n\n500\n\n1 000\n\n1 500\n\n2023-2030\n\nBillion USD\n\n**Cumulative CAPEX**\n\nFinal investment decision year\n\n2023 2024 2025 2026 2027 2028 2029 2030\n\n**Time available for final investment decision**","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Energy Technology Perspectives 2023 Chapter 1. Energy supply chains in transition\n\nPAGE | 57\n\nI EA. CC BY 4.0.\n\n### More investments are needed urgently to avoid bottlenecks\n\nBridging the production gaps described above would require enormous additional investments in the coming years. The total cumulative investment in mining, critical material production and manufacturing of the clean energy technologies selected in this report required to bring the necessary capacity online by 2030 amounts to around USD 1.2 trillion, ⁷ in 2021 dollars (Figure 1.11). Investments associated with projects already announced reach around60% of this, with large gaps in critical mineral mining and for some technology manufacturing.\n\nWhile mobilising such levels of investment is achievable, the timeline is extremely tight, especially given the long lead times to bring production to market. Most of the investment (including for projects already announced) needs to occur over 2023-2025, implying an average of more than USD 270 billion per year over the period. To compare, this is about two-thirds of current annual capital spending of the oil and gas industry, and nearly seven times higher than the average annual investment seen in clean energy supply chains over 2016-2021.\n\n**Global investment in selected clean energy supply chains needed to bring**\n\n**online enough capacity in 2030 in the NZE Scenario, by supply chain step**\n\nIEA. CC BY 4.0.\n\nNotes: CAPEX = capital expenditures. Only refers to the investments needed to bring online enough capacity in 2030 – not counting what would be needed to further scale up in subsequent years – for the subset of clean energy technologies selected in _ETP_ \\- _2023_ . Assuming construction times of five years for mining, three for processing plants and two for technology manufacturing. Investments are assumed uniform over the available period. Excludes site-tailored technology installation. Sources: IEA analysis based on company announcement; Bartholomeusz (2022); S&P Capital (2022).\n\n**Most of the supply chain investments needed to meet NZE Scenario targets in 2030 are made over 2023-2025, at an annual average seven times higher than over 2016-2021.**\n\n7 This refers only to the investments needed to bring enough capacity online in 2030 for the technologies selected in _ETP-2023_ , excluding the investments that would also be needed by 2030 to further scale up deployment in subsequent years.\n\n0\n\n50\n\n100\n\n150\n\n200\n\n2016-2021 2023-2030\n\nBillion USD\n\nCopper mining Other Critical mineral mining Critical mineral processing Clean technology manufacuring\n\n**Average yearly CAPEX**\n\n0\n\n500\n\n1 000\n\n1 500\n\n2023-2030\n\nBillion USD\n\n**Cumulative CAPEX**\n\nFinal investment decision year\n\n2023 2024 2025 2026 2027 2028 2029 2030\n\n**Time available for final investment decision**","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/7c6b23db-en_p57.pdf"]},"response":{"errors":[],"extract_id":"extract_aa0749afdaaeeae3a1865a32898bac54","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Energy Technology Perspectives 2023 Chapter 1. Energy supply chains in transition\n\nPAGE | 57\n\nI EA. CC BY 4.0.\n\n### More investments are needed urgently to avoid bottlenecks\n\nBridging the production gaps described above would require enormous additional investments in the coming years. The total cumulative investment in mining, critical material production and manufacturing of the clean energy technologies selected in this report required to bring the necessary capacity online by 2030 amounts to around USD 1.2 trillion, ⁷ in 2021 dollars (Figure 1.11). Investments associated with projects already announced reach around60% of this, with large gaps in critical mineral mining and for some technology manufacturing.\n\nWhile mobilising such levels of investment is achievable, the timeline is extremely tight, especially given the long lead times to bring production to market. Most of the investment (including for projects already announced) needs to occur over 2023-2025, implying an average of more than USD 270 billion per year over the period. To compare, this is about two-thirds of current annual capital spending of the oil and gas industry, and nearly seven times higher than the average annual investment seen in clean energy supply chains over 2016-2021.\n\n**Global investment in selected clean energy supply chains needed to bring**\n\n**online enough capacity in 2030 in the NZE Scenario, by supply chain step**\n\nIEA. CC BY 4.0.\n\nNotes: CAPEX = capital expenditures. Only refers to the investments needed to bring online enough capacity in 2030 – not counting what would be needed to further scale up in subsequent years – for the subset of clean energy technologies selected in _ETP_ \\- _2023_ . Assuming construction times of five years for mining, three for processing plants and two for technology manufacturing. Investments are assumed uniform over the available period. Excludes site-tailored technology installation. Sources: IEA analysis based on company announcement; Bartholomeusz (2022); S&P Capital (2022).\n\n**Most of the supply chain investments needed to meet NZE Scenario targets in 2030 are made over 2023-2025, at an annual average seven times higher than over 2016-2021.**\n\n7 This refers only to the investments needed to bring enough capacity online in 2030 for the technologies selected in _ETP-2023_ , excluding the investments that would also be needed by 2030 to further scale up deployment in subsequent years.\n\n0\n\n50\n\n100\n\n150\n\n200\n\n2016-2021 2023-2030\n\nBillion USD\n\nCopper mining Other Critical mineral mining Critical mineral processing Clean technology manufacuring\n\n**Average yearly CAPEX**\n\n0\n\n500\n\n1 000\n\n1 500\n\n2023-2030\n\nBillion USD\n\n**Cumulative CAPEX**\n\nFinal investment decision year\n\n2023 2024 2025 2026 2027 2028 2029 2030\n\n**Time available for final investment decision**"],"full_content":"Energy Technology Perspectives 2023 Chapter 1. Energy supply chains in transition\n\nPAGE | 57\n\nI EA. CC BY 4.0.\n\n### More investments are needed urgently to avoid bottlenecks\n\nBridging the production gaps described above would require enormous additional investments in the coming years. The total cumulative investment in mining, critical material production and manufacturing of the clean energy technologies selected in this report required to bring the necessary capacity online by 2030 amounts to around USD 1.2 trillion, ⁷ in 2021 dollars (Figure 1.11). Investments associated with projects already announced reach around60% of this, with large gaps in critical mineral mining and for some technology manufacturing.\n\nWhile mobilising such levels of investment is achievable, the timeline is extremely tight, especially given the long lead times to bring production to market. Most of the investment (including for projects already announced) needs to occur over 2023-2025, implying an average of more than USD 270 billion per year over the period. To compare, this is about two-thirds of current annual capital spending of the oil and gas industry, and nearly seven times higher than the average annual investment seen in clean energy supply chains over 2016-2021.\n\n**Global investment in selected clean energy supply chains needed to bring**\n\n**online enough capacity in 2030 in the NZE Scenario, by supply chain step**\n\nIEA. CC BY 4.0.\n\nNotes: CAPEX = capital expenditures. Only refers to the investments needed to bring online enough capacity in 2030 – not counting what would be needed to further scale up in subsequent years – for the subset of clean energy technologies selected in _ETP_ \\- _2023_ . Assuming construction times of five years for mining, three for processing plants and two for technology manufacturing. Investments are assumed uniform over the available period. Excludes site-tailored technology installation. Sources: IEA analysis based on company announcement; Bartholomeusz (2022); S&P Capital (2022).\n\n**Most of the supply chain investments needed to meet NZE Scenario targets in 2030 are made over 2023-2025, at an annual average seven times higher than over 2016-2021.**\n\n7 This refers only to the investments needed to bring enough capacity online in 2030 for the technologies selected in _ETP-2023_ , excluding the investments that would also be needed by 2030 to further scale up deployment in subsequent years.\n\n0\n\n50\n\n100\n\n150\n\n200\n\n2016-2021 2023-2030\n\nBillion USD\n\nCopper mining Other Critical mineral mining Critical mineral processing Clean technology manufacuring\n\n**Average yearly CAPEX**\n\n0\n\n500\n\n1 000\n\n1 500\n\n2023-2030\n\nBillion USD\n\n**Cumulative CAPEX**\n\nFinal investment decision year\n\n2023 2024 2025 2026 2027 2028 2029 2030\n\n**Time available for final investment decision**","publish_date":null,"title":"More investments are needed urgently to avoid bottlenecks","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/7c6b23db-en_p57.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/7c6b23db-en_p57.pdf"},"request":{"example_id":"chart/7c6b23db-en_p57","product_type":"parse","source_file_path":"7c6b23db-en_p57.pdf"},"started_at":"2026-11-04T19:45:58.305949"} +{"completed_at":"2026-11-04T19:46:01.220074","latency_in_ms":1735,"output":{"example_id":"chart/7c6b23db-en_p58","grounded_pages":[],"layout_pages":[],"markdown":"Energy Technology Perspectives 2023 Chapter 1. Energy supply chains in transition\n\nPAGE | 58\n\nI EA. CC BY 4.0.\n\nFor the mining of copper, lithium, nickel and cobalt alone, investment of around USD 130 billion per year over 2023-2025 is required, which would be directed to any mining project already in the permitting pipeline. This is equivalent to the entire metal mining industry investments of 2021 (S&P Global, 2022e), implying a doubling of overall investments in the sector in 2023. The copper mining market, which accounts for over half of the required mining investments, may be in part balanced through demand reduction in other sectors and recycling. For the refining of those minerals, announced projects cover about 80% of investment needs for copper, but only over 40% for other metals. In terms of clean technology manufacturing capacity, announced projects cover two-thirds of investment needs to 2030.\n\nThe overall investment needed in clean energy supply chains is much higher than the investment required to develop those under study in _ETP_ \\- _2023_ and bring sufficient capacity online by 2030. In total, investment in clean energy technologies and infrastructure reaches over USD 4.5 trillion in 2030 in the NZE Scenario (IEA, 2022f). This scale of investment is unprecedented. Mobilising it across all regions, technologies and supply chains is an enormous task. Bottlenecks can occur as a result of policy and regulatory risks, a lack of confidence in demonstration and first- of-a-kind projects, uncertainty about project pipelines, wider macroeconomic factors such as currency stability, and geopolitical events.\n\nThe risk of underinvestment in clean energy and supply bottlenecks is particularly acute in the emerging economies. One hurdle to investment in those countries is their higher cost of capital, which has a significant impact on the financial viability of clean energy projects. For example, financing costs for material production projects can be more than twice that in the advanced economies, due a greater perception of risk (Figure 1.12). Costs in China tend to be lower due to the importance of state- owned enterprises and their ability to raise cheap public finance.\n\n**Cost of capital for bulk material production industries by country/regional**\n\n**grouping, 2020**\n\nIEA. CC BY 4.0.\n\nSource: Adapted from IEA (2021c).\n\n**Financing costs for bulk material production in the emerging economies can be more than twice those in advanced economies, driven by a greater perception of risk.**\n\n0% 2% 4% 6% 8% 10%\n\nCement\n\nChemicals\n\nIron & steel Other emerging market and developing economies\n\nChina\n\nAdvanced economies","pages":[{"line_numbers":[],"markdown":"Energy Technology Perspectives 2023 Chapter 1. Energy supply chains in transition\n\nPAGE | 58\n\nI EA. CC BY 4.0.\n\nFor the mining of copper, lithium, nickel and cobalt alone, investment of around USD 130 billion per year over 2023-2025 is required, which would be directed to any mining project already in the permitting pipeline. This is equivalent to the entire metal mining industry investments of 2021 (S&P Global, 2022e), implying a doubling of overall investments in the sector in 2023. The copper mining market, which accounts for over half of the required mining investments, may be in part balanced through demand reduction in other sectors and recycling. For the refining of those minerals, announced projects cover about 80% of investment needs for copper, but only over 40% for other metals. In terms of clean technology manufacturing capacity, announced projects cover two-thirds of investment needs to 2030.\n\nThe overall investment needed in clean energy supply chains is much higher than the investment required to develop those under study in _ETP_ \\- _2023_ and bring sufficient capacity online by 2030. In total, investment in clean energy technologies and infrastructure reaches over USD 4.5 trillion in 2030 in the NZE Scenario (IEA, 2022f). This scale of investment is unprecedented. Mobilising it across all regions, technologies and supply chains is an enormous task. Bottlenecks can occur as a result of policy and regulatory risks, a lack of confidence in demonstration and first- of-a-kind projects, uncertainty about project pipelines, wider macroeconomic factors such as currency stability, and geopolitical events.\n\nThe risk of underinvestment in clean energy and supply bottlenecks is particularly acute in the emerging economies. One hurdle to investment in those countries is their higher cost of capital, which has a significant impact on the financial viability of clean energy projects. For example, financing costs for material production projects can be more than twice that in the advanced economies, due a greater perception of risk (Figure 1.12). Costs in China tend to be lower due to the importance of state- owned enterprises and their ability to raise cheap public finance.\n\n**Cost of capital for bulk material production industries by country/regional**\n\n**grouping, 2020**\n\nIEA. CC BY 4.0.\n\nSource: Adapted from IEA (2021c).\n\n**Financing costs for bulk material production in the emerging economies can be more than twice those in advanced economies, driven by a greater perception of risk.**\n\n0% 2% 4% 6% 8% 10%\n\nCement\n\nChemicals\n\nIron & steel Other emerging market and developing economies\n\nChina\n\nAdvanced economies","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Energy Technology Perspectives 2023 Chapter 1. Energy supply chains in transition\n\nPAGE | 58\n\nI EA. CC BY 4.0.\n\nFor the mining of copper, lithium, nickel and cobalt alone, investment of around USD 130 billion per year over 2023-2025 is required, which would be directed to any mining project already in the permitting pipeline. This is equivalent to the entire metal mining industry investments of 2021 (S&P Global, 2022e), implying a doubling of overall investments in the sector in 2023. The copper mining market, which accounts for over half of the required mining investments, may be in part balanced through demand reduction in other sectors and recycling. For the refining of those minerals, announced projects cover about 80% of investment needs for copper, but only over 40% for other metals. In terms of clean technology manufacturing capacity, announced projects cover two-thirds of investment needs to 2030.\n\nThe overall investment needed in clean energy supply chains is much higher than the investment required to develop those under study in _ETP_ \\- _2023_ and bring sufficient capacity online by 2030. In total, investment in clean energy technologies and infrastructure reaches over USD 4.5 trillion in 2030 in the NZE Scenario (IEA, 2022f). This scale of investment is unprecedented. Mobilising it across all regions, technologies and supply chains is an enormous task. Bottlenecks can occur as a result of policy and regulatory risks, a lack of confidence in demonstration and first- of-a-kind projects, uncertainty about project pipelines, wider macroeconomic factors such as currency stability, and geopolitical events.\n\nThe risk of underinvestment in clean energy and supply bottlenecks is particularly acute in the emerging economies. One hurdle to investment in those countries is their higher cost of capital, which has a significant impact on the financial viability of clean energy projects. For example, financing costs for material production projects can be more than twice that in the advanced economies, due a greater perception of risk (Figure 1.12). Costs in China tend to be lower due to the importance of state- owned enterprises and their ability to raise cheap public finance.\n\n**Cost of capital for bulk material production industries by country/regional**\n\n**grouping, 2020**\n\nIEA. CC BY 4.0.\n\nSource: Adapted from IEA (2021c).\n\n**Financing costs for bulk material production in the emerging economies can be more than twice those in advanced economies, driven by a greater perception of risk.**\n\n0% 2% 4% 6% 8% 10%\n\nCement\n\nChemicals\n\nIron & steel Other emerging market and developing economies\n\nChina\n\nAdvanced economies","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/7c6b23db-en_p58.pdf"]},"response":{"errors":[],"extract_id":"extract_3b49dff36e04fdbfcf2e9dabca14e88a","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Energy Technology Perspectives 2023 Chapter 1. Energy supply chains in transition\n\nPAGE | 58\n\nI EA. CC BY 4.0.\n\nFor the mining of copper, lithium, nickel and cobalt alone, investment of around USD 130 billion per year over 2023-2025 is required, which would be directed to any mining project already in the permitting pipeline. This is equivalent to the entire metal mining industry investments of 2021 (S&P Global, 2022e), implying a doubling of overall investments in the sector in 2023. The copper mining market, which accounts for over half of the required mining investments, may be in part balanced through demand reduction in other sectors and recycling. For the refining of those minerals, announced projects cover about 80% of investment needs for copper, but only over 40% for other metals. In terms of clean technology manufacturing capacity, announced projects cover two-thirds of investment needs to 2030.\n\nThe overall investment needed in clean energy supply chains is much higher than the investment required to develop those under study in _ETP_ \\- _2023_ and bring sufficient capacity online by 2030. In total, investment in clean energy technologies and infrastructure reaches over USD 4.5 trillion in 2030 in the NZE Scenario (IEA, 2022f). This scale of investment is unprecedented. Mobilising it across all regions, technologies and supply chains is an enormous task. Bottlenecks can occur as a result of policy and regulatory risks, a lack of confidence in demonstration and first- of-a-kind projects, uncertainty about project pipelines, wider macroeconomic factors such as currency stability, and geopolitical events.\n\nThe risk of underinvestment in clean energy and supply bottlenecks is particularly acute in the emerging economies. One hurdle to investment in those countries is their higher cost of capital, which has a significant impact on the financial viability of clean energy projects. For example, financing costs for material production projects can be more than twice that in the advanced economies, due a greater perception of risk (Figure 1.12). Costs in China tend to be lower due to the importance of state- owned enterprises and their ability to raise cheap public finance.\n\n**Cost of capital for bulk material production industries by country/regional**\n\n**grouping, 2020**\n\nIEA. CC BY 4.0.\n\nSource: Adapted from IEA (2021c).\n\n**Financing costs for bulk material production in the emerging economies can be more than twice those in advanced economies, driven by a greater perception of risk.**\n\n0% 2% 4% 6% 8% 10%\n\nCement\n\nChemicals\n\nIron & steel Other emerging market and developing economies\n\nChina\n\nAdvanced economies"],"full_content":"Energy Technology Perspectives 2023 Chapter 1. Energy supply chains in transition\n\nPAGE | 58\n\nI EA. CC BY 4.0.\n\nFor the mining of copper, lithium, nickel and cobalt alone, investment of around USD 130 billion per year over 2023-2025 is required, which would be directed to any mining project already in the permitting pipeline. This is equivalent to the entire metal mining industry investments of 2021 (S&P Global, 2022e), implying a doubling of overall investments in the sector in 2023. The copper mining market, which accounts for over half of the required mining investments, may be in part balanced through demand reduction in other sectors and recycling. For the refining of those minerals, announced projects cover about 80% of investment needs for copper, but only over 40% for other metals. In terms of clean technology manufacturing capacity, announced projects cover two-thirds of investment needs to 2030.\n\nThe overall investment needed in clean energy supply chains is much higher than the investment required to develop those under study in _ETP_ \\- _2023_ and bring sufficient capacity online by 2030. In total, investment in clean energy technologies and infrastructure reaches over USD 4.5 trillion in 2030 in the NZE Scenario (IEA, 2022f). This scale of investment is unprecedented. Mobilising it across all regions, technologies and supply chains is an enormous task. Bottlenecks can occur as a result of policy and regulatory risks, a lack of confidence in demonstration and first- of-a-kind projects, uncertainty about project pipelines, wider macroeconomic factors such as currency stability, and geopolitical events.\n\nThe risk of underinvestment in clean energy and supply bottlenecks is particularly acute in the emerging economies. One hurdle to investment in those countries is their higher cost of capital, which has a significant impact on the financial viability of clean energy projects. For example, financing costs for material production projects can be more than twice that in the advanced economies, due a greater perception of risk (Figure 1.12). Costs in China tend to be lower due to the importance of state- owned enterprises and their ability to raise cheap public finance.\n\n**Cost of capital for bulk material production industries by country/regional**\n\n**grouping, 2020**\n\nIEA. CC BY 4.0.\n\nSource: Adapted from IEA (2021c).\n\n**Financing costs for bulk material production in the emerging economies can be more than twice those in advanced economies, driven by a greater perception of risk.**\n\n0% 2% 4% 6% 8% 10%\n\nCement\n\nChemicals\n\nIron & steel Other emerging market and developing economies\n\nChina\n\nAdvanced economies","publish_date":null,"title":"For the mining of copper, lithium, nickel and cobalt alone, investment of around USD 130 billion per year over 2023-2025","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/7c6b23db-en_p58.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/7c6b23db-en_p58.pdf"},"request":{"example_id":"chart/7c6b23db-en_p58","product_type":"parse","source_file_path":"7c6b23db-en_p58.pdf"},"started_at":"2026-11-04T19:45:59.484499"} +{"completed_at":"2026-11-04T19:46:03.891640","latency_in_ms":2575,"output":{"example_id":"chart/89af4857-en_p11","grounded_pages":[],"layout_pages":[],"markdown":" **9**\n\nOECD ECONOMIC OUTLOOK, INTERIM REPORT MARCH 2025 © OECD 2025\n\n4\\. Trade policy uncertainty has increased markedly in recent months, with the introduction of new\n\ntrade barriers by a number of countries. This has translated into a substantial increase in newspaper-based measures of economic policy uncertainty (Figure 2, Panel C). Across the economies with available data, the biggest rise in policy uncertainty has been in Canada and Mexico (Figure 2, Panel D), both countries that are subject to increased tariff rates on bilateral exports to the United States. Greater policy uncertainty can be expected to hold back spending decisions by companies and households, particularly on longer-term items such as fixed capital investment and durable goods.\n\n### **Figure 2. Consumer confidence remains weak and economic policy uncertainty has spiked in advanced economies**\n\nNote: Panel B shows a PPP-based weighted mean and is based on G20 economies except Argentina and Saudi Arabia. In Panel C, ‘Advanced economies’ includes Australia, Canada, Chile, France, Germany, Ireland, Italy, Japan, Korea, the Netherlands, Spain, Sweden and the United Kingdom, and ‘Emerging-market economies’ includes Brazil, India and Russia. Source: policyuncertainty.com; OECD, Main Economic Indicators database; and OECD calculations.","pages":[{"line_numbers":[],"markdown":" **9**\n\nOECD ECONOMIC OUTLOOK, INTERIM REPORT MARCH 2025 © OECD 2025\n\n4\\. Trade policy uncertainty has increased markedly in recent months, with the introduction of new\n\ntrade barriers by a number of countries. This has translated into a substantial increase in newspaper-based measures of economic policy uncertainty (Figure 2, Panel C). Across the economies with available data, the biggest rise in policy uncertainty has been in Canada and Mexico (Figure 2, Panel D), both countries that are subject to increased tariff rates on bilateral exports to the United States. Greater policy uncertainty can be expected to hold back spending decisions by companies and households, particularly on longer-term items such as fixed capital investment and durable goods.\n\n### **Figure 2. Consumer confidence remains weak and economic policy uncertainty has spiked in advanced economies**\n\nNote: Panel B shows a PPP-based weighted mean and is based on G20 economies except Argentina and Saudi Arabia. In Panel C, ‘Advanced economies’ includes Australia, Canada, Chile, France, Germany, Ireland, Italy, Japan, Korea, the Netherlands, Spain, Sweden and the United Kingdom, and ‘Emerging-market economies’ includes Brazil, India and Russia. Source: policyuncertainty.com; OECD, Main Economic Indicators database; and OECD calculations.","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":" **9**\n\nOECD ECONOMIC OUTLOOK, INTERIM REPORT MARCH 2025 © OECD 2025\n\n4\\. Trade policy uncertainty has increased markedly in recent months, with the introduction of new\n\ntrade barriers by a number of countries. This has translated into a substantial increase in newspaper-based measures of economic policy uncertainty (Figure 2, Panel C). Across the economies with available data, the biggest rise in policy uncertainty has been in Canada and Mexico (Figure 2, Panel D), both countries that are subject to increased tariff rates on bilateral exports to the United States. Greater policy uncertainty can be expected to hold back spending decisions by companies and households, particularly on longer-term items such as fixed capital investment and durable goods.\n\n### **Figure 2. Consumer confidence remains weak and economic policy uncertainty has spiked in advanced economies**\n\nNote: Panel B shows a PPP-based weighted mean and is based on G20 economies except Argentina and Saudi Arabia. In Panel C, ‘Advanced economies’ includes Australia, Canada, Chile, France, Germany, Ireland, Italy, Japan, Korea, the Netherlands, Spain, Sweden and the United Kingdom, and ‘Emerging-market economies’ includes Brazil, India and Russia. Source: policyuncertainty.com; OECD, Main Economic Indicators database; and OECD calculations.","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/89af4857-en_p11.pdf"]},"response":{"errors":[],"extract_id":"extract_8a46a223c16546db0b48697ed983ce2d","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":[" **9**\n\nOECD ECONOMIC OUTLOOK, INTERIM REPORT MARCH 2025 © OECD 2025\n\n4\\. Trade policy uncertainty has increased markedly in recent months, with the introduction of new\n\ntrade barriers by a number of countries. This has translated into a substantial increase in newspaper-based measures of economic policy uncertainty (Figure 2, Panel C). Across the economies with available data, the biggest rise in policy uncertainty has been in Canada and Mexico (Figure 2, Panel D), both countries that are subject to increased tariff rates on bilateral exports to the United States. Greater policy uncertainty can be expected to hold back spending decisions by companies and households, particularly on longer-term items such as fixed capital investment and durable goods.\n\n### **Figure 2. Consumer confidence remains weak and economic policy uncertainty has spiked in advanced economies**\n\nNote: Panel B shows a PPP-based weighted mean and is based on G20 economies except Argentina and Saudi Arabia. In Panel C, ‘Advanced economies’ includes Australia, Canada, Chile, France, Germany, Ireland, Italy, Japan, Korea, the Netherlands, Spain, Sweden and the United Kingdom, and ‘Emerging-market economies’ includes Brazil, India and Russia. Source: policyuncertainty.com; OECD, Main Economic Indicators database; and OECD calculations."],"full_content":" **9**\n\nOECD ECONOMIC OUTLOOK, INTERIM REPORT MARCH 2025 © OECD 2025\n\n4\\. Trade policy uncertainty has increased markedly in recent months, with the introduction of new\n\ntrade barriers by a number of countries. This has translated into a substantial increase in newspaper-based measures of economic policy uncertainty (Figure 2, Panel C). Across the economies with available data, the biggest rise in policy uncertainty has been in Canada and Mexico (Figure 2, Panel D), both countries that are subject to increased tariff rates on bilateral exports to the United States. Greater policy uncertainty can be expected to hold back spending decisions by companies and households, particularly on longer-term items such as fixed capital investment and durable goods.\n\n### **Figure 2. Consumer confidence remains weak and economic policy uncertainty has spiked in advanced economies**\n\nNote: Panel B shows a PPP-based weighted mean and is based on G20 economies except Argentina and Saudi Arabia. In Panel C, ‘Advanced economies’ includes Australia, Canada, Chile, France, Germany, Ireland, Italy, Japan, Korea, the Netherlands, Spain, Sweden and the United Kingdom, and ‘Emerging-market economies’ includes Brazil, India and Russia. Source: policyuncertainty.com; OECD, Main Economic Indicators database; and OECD calculations.","publish_date":null,"title":"Figure 2. Consumer confidence remains weak and economic policy uncertainty has spiked in advanced economies","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/89af4857-en_p11.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/89af4857-en_p11.pdf"},"request":{"example_id":"chart/89af4857-en_p11","product_type":"parse","source_file_path":"89af4857-en_p11.pdf"},"started_at":"2026-11-04T19:46:01.316396"} +{"completed_at":"2026-11-04T19:46:04.380157","latency_in_ms":1533,"output":{"example_id":"chart/89af4857-en_p12","grounded_pages":[],"layout_pages":[],"markdown":"**10** \n\nOECD ECONOMIC OUTLOOK, INTERIM REPORT MARCH 2025 © OECD 2025\n\n5\\. A series of recently announced trade policy measures will have implications for the economic\n\noutlook if sustained. The United States has raised tariff rates on merchandise imports from China by 20 percentage points, with China taking targeted retaliatory action, and increased tariffs on imports of steel and aluminium from all countries, some of whom have introduced targeted retaliatory measures. The United States has also announced an increase of 25 percentage points in bilateral tariff rates on merchandise imports from Canada and Mexico (alongside a lower 10 percentage point tariff rate increase on imports of energy products from Canada, and potash from both Canada and Mexico). Currently, these higher tariffs apply only to imports that are not compliant with the United States-Mexico-Canada Agreement (USMCA) but they are set to apply to all imports from early April. Canada has raised tariffs on selected imports from the United States in response and announced plans to broaden the range of products affected from April. Mexico has not introduced any specific trade policy changes, but has stated a general intention to retaliate if the broader tariff increases planned by the United States go ahead as announced. Such tariffs are likely to be particularly costly for Canada and Mexico due to their greater openness to trade, and high proportion of trade with the United States (Figure 3). Broader trade policy reviews are also under way in the United States and due to be completed by early April.\n\n6\\. If the announced trade policy actions persist, as assumed in the projections, the new bilateral tariff\n\nrates will raise revenues for the governments imposing them but will be a drag on global activity, incomes and regular tax revenues. They also add to trade costs, raising the price of covered imported final goods for consumers and intermediate inputs for businesses. The impact of higher costs will be amplified where inputs cross borders several times and duties are incurred at each stage, as is the case in the integrated North American market. An agreement that would ease trade tensions, and potentially even lower existing trade barriers, would be welcome and help to improve policy certainty and growth prospects.\n\n### **Figure 3. The direct impact of merchandise trade tensions varies across countries**\n\nDistribution of merchandise export volumes in 2024\n\nNote: Merchandise export volumes as a share of GDP volumes based on 2024 data, except for Argentina, India, Mexico and Saudi Arabia where data are for 2023, and Türkiye where data are for 2021. Bilateral trade shares based on OECD estimates for 2021. Source: OECD Interim Economic Outlook 117 database; OECD Balanced International Merchandise Trade Statistics; and OECD calculations.","pages":[{"line_numbers":[],"markdown":"**10** \n\nOECD ECONOMIC OUTLOOK, INTERIM REPORT MARCH 2025 © OECD 2025\n\n5\\. A series of recently announced trade policy measures will have implications for the economic\n\noutlook if sustained. The United States has raised tariff rates on merchandise imports from China by 20 percentage points, with China taking targeted retaliatory action, and increased tariffs on imports of steel and aluminium from all countries, some of whom have introduced targeted retaliatory measures. The United States has also announced an increase of 25 percentage points in bilateral tariff rates on merchandise imports from Canada and Mexico (alongside a lower 10 percentage point tariff rate increase on imports of energy products from Canada, and potash from both Canada and Mexico). Currently, these higher tariffs apply only to imports that are not compliant with the United States-Mexico-Canada Agreement (USMCA) but they are set to apply to all imports from early April. Canada has raised tariffs on selected imports from the United States in response and announced plans to broaden the range of products affected from April. Mexico has not introduced any specific trade policy changes, but has stated a general intention to retaliate if the broader tariff increases planned by the United States go ahead as announced. Such tariffs are likely to be particularly costly for Canada and Mexico due to their greater openness to trade, and high proportion of trade with the United States (Figure 3). Broader trade policy reviews are also under way in the United States and due to be completed by early April.\n\n6\\. If the announced trade policy actions persist, as assumed in the projections, the new bilateral tariff\n\nrates will raise revenues for the governments imposing them but will be a drag on global activity, incomes and regular tax revenues. They also add to trade costs, raising the price of covered imported final goods for consumers and intermediate inputs for businesses. The impact of higher costs will be amplified where inputs cross borders several times and duties are incurred at each stage, as is the case in the integrated North American market. An agreement that would ease trade tensions, and potentially even lower existing trade barriers, would be welcome and help to improve policy certainty and growth prospects.\n\n### **Figure 3. The direct impact of merchandise trade tensions varies across countries**\n\nDistribution of merchandise export volumes in 2024\n\nNote: Merchandise export volumes as a share of GDP volumes based on 2024 data, except for Argentina, India, Mexico and Saudi Arabia where data are for 2023, and Türkiye where data are for 2021. Bilateral trade shares based on OECD estimates for 2021. Source: OECD Interim Economic Outlook 117 database; OECD Balanced International Merchandise Trade Statistics; and OECD calculations.","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**10** \n\nOECD ECONOMIC OUTLOOK, INTERIM REPORT MARCH 2025 © OECD 2025\n\n5\\. A series of recently announced trade policy measures will have implications for the economic\n\noutlook if sustained. The United States has raised tariff rates on merchandise imports from China by 20 percentage points, with China taking targeted retaliatory action, and increased tariffs on imports of steel and aluminium from all countries, some of whom have introduced targeted retaliatory measures. The United States has also announced an increase of 25 percentage points in bilateral tariff rates on merchandise imports from Canada and Mexico (alongside a lower 10 percentage point tariff rate increase on imports of energy products from Canada, and potash from both Canada and Mexico). Currently, these higher tariffs apply only to imports that are not compliant with the United States-Mexico-Canada Agreement (USMCA) but they are set to apply to all imports from early April. Canada has raised tariffs on selected imports from the United States in response and announced plans to broaden the range of products affected from April. Mexico has not introduced any specific trade policy changes, but has stated a general intention to retaliate if the broader tariff increases planned by the United States go ahead as announced. Such tariffs are likely to be particularly costly for Canada and Mexico due to their greater openness to trade, and high proportion of trade with the United States (Figure 3). Broader trade policy reviews are also under way in the United States and due to be completed by early April.\n\n6\\. If the announced trade policy actions persist, as assumed in the projections, the new bilateral tariff\n\nrates will raise revenues for the governments imposing them but will be a drag on global activity, incomes and regular tax revenues. They also add to trade costs, raising the price of covered imported final goods for consumers and intermediate inputs for businesses. The impact of higher costs will be amplified where inputs cross borders several times and duties are incurred at each stage, as is the case in the integrated North American market. An agreement that would ease trade tensions, and potentially even lower existing trade barriers, would be welcome and help to improve policy certainty and growth prospects.\n\n### **Figure 3. The direct impact of merchandise trade tensions varies across countries**\n\nDistribution of merchandise export volumes in 2024\n\nNote: Merchandise export volumes as a share of GDP volumes based on 2024 data, except for Argentina, India, Mexico and Saudi Arabia where data are for 2023, and Türkiye where data are for 2021. Bilateral trade shares based on OECD estimates for 2021. Source: OECD Interim Economic Outlook 117 database; OECD Balanced International Merchandise Trade Statistics; and OECD calculations.","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/89af4857-en_p12.pdf"]},"response":{"errors":[],"extract_id":"extract_6fbfdf420c49f641429a437b7e6fbab7","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**10** \n\nOECD ECONOMIC OUTLOOK, INTERIM REPORT MARCH 2025 © OECD 2025\n\n5\\. A series of recently announced trade policy measures will have implications for the economic\n\noutlook if sustained. The United States has raised tariff rates on merchandise imports from China by 20 percentage points, with China taking targeted retaliatory action, and increased tariffs on imports of steel and aluminium from all countries, some of whom have introduced targeted retaliatory measures. The United States has also announced an increase of 25 percentage points in bilateral tariff rates on merchandise imports from Canada and Mexico (alongside a lower 10 percentage point tariff rate increase on imports of energy products from Canada, and potash from both Canada and Mexico). Currently, these higher tariffs apply only to imports that are not compliant with the United States-Mexico-Canada Agreement (USMCA) but they are set to apply to all imports from early April. Canada has raised tariffs on selected imports from the United States in response and announced plans to broaden the range of products affected from April. Mexico has not introduced any specific trade policy changes, but has stated a general intention to retaliate if the broader tariff increases planned by the United States go ahead as announced. Such tariffs are likely to be particularly costly for Canada and Mexico due to their greater openness to trade, and high proportion of trade with the United States (Figure 3). Broader trade policy reviews are also under way in the United States and due to be completed by early April.\n\n6\\. If the announced trade policy actions persist, as assumed in the projections, the new bilateral tariff\n\nrates will raise revenues for the governments imposing them but will be a drag on global activity, incomes and regular tax revenues. They also add to trade costs, raising the price of covered imported final goods for consumers and intermediate inputs for businesses. The impact of higher costs will be amplified where inputs cross borders several times and duties are incurred at each stage, as is the case in the integrated North American market. An agreement that would ease trade tensions, and potentially even lower existing trade barriers, would be welcome and help to improve policy certainty and growth prospects.\n\n### **Figure 3. The direct impact of merchandise trade tensions varies across countries**\n\nDistribution of merchandise export volumes in 2024\n\nNote: Merchandise export volumes as a share of GDP volumes based on 2024 data, except for Argentina, India, Mexico and Saudi Arabia where data are for 2023, and Türkiye where data are for 2021. Bilateral trade shares based on OECD estimates for 2021. Source: OECD Interim Economic Outlook 117 database; OECD Balanced International Merchandise Trade Statistics; and OECD calculations."],"full_content":"**10** \n\nOECD ECONOMIC OUTLOOK, INTERIM REPORT MARCH 2025 © OECD 2025\n\n5\\. A series of recently announced trade policy measures will have implications for the economic\n\noutlook if sustained. The United States has raised tariff rates on merchandise imports from China by 20 percentage points, with China taking targeted retaliatory action, and increased tariffs on imports of steel and aluminium from all countries, some of whom have introduced targeted retaliatory measures. The United States has also announced an increase of 25 percentage points in bilateral tariff rates on merchandise imports from Canada and Mexico (alongside a lower 10 percentage point tariff rate increase on imports of energy products from Canada, and potash from both Canada and Mexico). Currently, these higher tariffs apply only to imports that are not compliant with the United States-Mexico-Canada Agreement (USMCA) but they are set to apply to all imports from early April. Canada has raised tariffs on selected imports from the United States in response and announced plans to broaden the range of products affected from April. Mexico has not introduced any specific trade policy changes, but has stated a general intention to retaliate if the broader tariff increases planned by the United States go ahead as announced. Such tariffs are likely to be particularly costly for Canada and Mexico due to their greater openness to trade, and high proportion of trade with the United States (Figure 3). Broader trade policy reviews are also under way in the United States and due to be completed by early April.\n\n6\\. If the announced trade policy actions persist, as assumed in the projections, the new bilateral tariff\n\nrates will raise revenues for the governments imposing them but will be a drag on global activity, incomes and regular tax revenues. They also add to trade costs, raising the price of covered imported final goods for consumers and intermediate inputs for businesses. The impact of higher costs will be amplified where inputs cross borders several times and duties are incurred at each stage, as is the case in the integrated North American market. An agreement that would ease trade tensions, and potentially even lower existing trade barriers, would be welcome and help to improve policy certainty and growth prospects.\n\n### **Figure 3. The direct impact of merchandise trade tensions varies across countries**\n\nDistribution of merchandise export volumes in 2024\n\nNote: Merchandise export volumes as a share of GDP volumes based on 2024 data, except for Argentina, India, Mexico and Saudi Arabia where data are for 2023, and Türkiye where data are for 2021. Bilateral trade shares based on OECD estimates for 2021. Source: OECD Interim Economic Outlook 117 database; OECD Balanced International Merchandise Trade Statistics; and OECD calculations.","publish_date":null,"title":"Figure 3. The direct impact of merchandise trade tensions varies across countries","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/89af4857-en_p12.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/89af4857-en_p12.pdf"},"request":{"example_id":"chart/89af4857-en_p12","product_type":"parse","source_file_path":"89af4857-en_p12.pdf"},"started_at":"2026-11-04T19:46:02.846866"} +{"completed_at":"2026-11-04T19:46:06.960515","latency_in_ms":2680,"output":{"example_id":"chart/89af4857-en_p15","grounded_pages":[],"layout_pages":[],"markdown":" **13**\n\nOECD ECONOMIC OUTLOOK, INTERIM REPORT MARCH 2025 © OECD 2025\n\n### **Figure 6. Household inflation expectations have recently risen in some countries**\n\nSource: European Central Bank; Bank of Korea; Office for National Statistics; University of Michigan; and OECD calculations.\n\n**_Global financial conditions have tightened slightly and volatility has risen recently_**\n\n10\\. Global financial conditions have tightened slightly since late 2024, with market volatility increasing\n\nrecently. 10-year government bond yields have risen in Europe, particularly after the recent announcement of substantial planned additional expenditures on defence and infrastructure in Germany and at the EU level. However, nominal bond yields in the United States have fallen since mid-January, unwinding the rise observed through the final months of 2024. Indicators suggest that the recent movement reflects a fall in real yields as economic growth slows, rather than lower inflation expectations. The US dollar has also fluctuated, with a depreciation against most currencies this year after a sustained rise through the final quarter of 2024 that contributed to tighter financial conditions in emerging market economies.\n\n11\\. Financial market stress has so far remained contained. Bank credit growth has gradually\n\nrecovered from low levels in advanced economies, as the impact of easier monetary policy has transmitted to lending interest rates. Despite already-stretched valuations in some sectors and some recent price corrections, equity prices are still higher than in November 2024, particularly in Germany, Spain and Italy. Robust growth in corporate bond and leveraged loan issuance has signalled investor confidence and increased risk appetite, with the equity risk premium having fallen in the United States, the United Kingdom and in Europe.\n\n### **Projections**\n\n**_Global growth is expected to soften and inflation to ease further_**\n\n12\\. Global GDP growth is projected to slow from 3.2% in 2024, to 3.1% in 2025 and 3.0% in 2026.\n\nNew data and recent developments result in a downward growth revision from the December 2024 _OECD Economic Outlook_ (Figure 7, Panel A). In quarterly terms, growth is expected to soften from the first quarter of 2025 and to remain subdued thereafter. The imposition of new bilateral tariff rates and the associated increase in policy and geopolitical uncertainty will act as a drag, particularly on business investment and trade. In addition, increased trade costs are expected to feed through gradually to final goods prices, putting additional upward pressure on inflation in many countries and requiring monetary policy to remain restrictive for longer than previously expected.","pages":[{"line_numbers":[],"markdown":" **13**\n\nOECD ECONOMIC OUTLOOK, INTERIM REPORT MARCH 2025 © OECD 2025\n\n### **Figure 6. Household inflation expectations have recently risen in some countries**\n\nSource: European Central Bank; Bank of Korea; Office for National Statistics; University of Michigan; and OECD calculations.\n\n**_Global financial conditions have tightened slightly and volatility has risen recently_**\n\n10\\. Global financial conditions have tightened slightly since late 2024, with market volatility increasing\n\nrecently. 10-year government bond yields have risen in Europe, particularly after the recent announcement of substantial planned additional expenditures on defence and infrastructure in Germany and at the EU level. However, nominal bond yields in the United States have fallen since mid-January, unwinding the rise observed through the final months of 2024. Indicators suggest that the recent movement reflects a fall in real yields as economic growth slows, rather than lower inflation expectations. The US dollar has also fluctuated, with a depreciation against most currencies this year after a sustained rise through the final quarter of 2024 that contributed to tighter financial conditions in emerging market economies.\n\n11\\. Financial market stress has so far remained contained. Bank credit growth has gradually\n\nrecovered from low levels in advanced economies, as the impact of easier monetary policy has transmitted to lending interest rates. Despite already-stretched valuations in some sectors and some recent price corrections, equity prices are still higher than in November 2024, particularly in Germany, Spain and Italy. Robust growth in corporate bond and leveraged loan issuance has signalled investor confidence and increased risk appetite, with the equity risk premium having fallen in the United States, the United Kingdom and in Europe.\n\n### **Projections**\n\n**_Global growth is expected to soften and inflation to ease further_**\n\n12\\. Global GDP growth is projected to slow from 3.2% in 2024, to 3.1% in 2025 and 3.0% in 2026.\n\nNew data and recent developments result in a downward growth revision from the December 2024 _OECD Economic Outlook_ (Figure 7, Panel A). In quarterly terms, growth is expected to soften from the first quarter of 2025 and to remain subdued thereafter. The imposition of new bilateral tariff rates and the associated increase in policy and geopolitical uncertainty will act as a drag, particularly on business investment and trade. In addition, increased trade costs are expected to feed through gradually to final goods prices, putting additional upward pressure on inflation in many countries and requiring monetary policy to remain restrictive for longer than previously expected.","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":" **13**\n\nOECD ECONOMIC OUTLOOK, INTERIM REPORT MARCH 2025 © OECD 2025\n\n### **Figure 6. Household inflation expectations have recently risen in some countries**\n\nSource: European Central Bank; Bank of Korea; Office for National Statistics; University of Michigan; and OECD calculations.\n\n**_Global financial conditions have tightened slightly and volatility has risen recently_**\n\n10\\. Global financial conditions have tightened slightly since late 2024, with market volatility increasing\n\nrecently. 10-year government bond yields have risen in Europe, particularly after the recent announcement of substantial planned additional expenditures on defence and infrastructure in Germany and at the EU level. However, nominal bond yields in the United States have fallen since mid-January, unwinding the rise observed through the final months of 2024. Indicators suggest that the recent movement reflects a fall in real yields as economic growth slows, rather than lower inflation expectations. The US dollar has also fluctuated, with a depreciation against most currencies this year after a sustained rise through the final quarter of 2024 that contributed to tighter financial conditions in emerging market economies.\n\n11\\. Financial market stress has so far remained contained. Bank credit growth has gradually\n\nrecovered from low levels in advanced economies, as the impact of easier monetary policy has transmitted to lending interest rates. Despite already-stretched valuations in some sectors and some recent price corrections, equity prices are still higher than in November 2024, particularly in Germany, Spain and Italy. Robust growth in corporate bond and leveraged loan issuance has signalled investor confidence and increased risk appetite, with the equity risk premium having fallen in the United States, the United Kingdom and in Europe.\n\n### **Projections**\n\n**_Global growth is expected to soften and inflation to ease further_**\n\n12\\. Global GDP growth is projected to slow from 3.2% in 2024, to 3.1% in 2025 and 3.0% in 2026.\n\nNew data and recent developments result in a downward growth revision from the December 2024 _OECD Economic Outlook_ (Figure 7, Panel A). In quarterly terms, growth is expected to soften from the first quarter of 2025 and to remain subdued thereafter. The imposition of new bilateral tariff rates and the associated increase in policy and geopolitical uncertainty will act as a drag, particularly on business investment and trade. In addition, increased trade costs are expected to feed through gradually to final goods prices, putting additional upward pressure on inflation in many countries and requiring monetary policy to remain restrictive for longer than previously expected.","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/89af4857-en_p15.pdf"]},"response":{"errors":[],"extract_id":"extract_d7484e5963123767777a8c94f39b9c7a","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":[" **13**\n\nOECD ECONOMIC OUTLOOK, INTERIM REPORT MARCH 2025 © OECD 2025\n\n### **Figure 6. Household inflation expectations have recently risen in some countries**\n\nSource: European Central Bank; Bank of Korea; Office for National Statistics; University of Michigan; and OECD calculations.\n\n**_Global financial conditions have tightened slightly and volatility has risen recently_**\n\n10\\. Global financial conditions have tightened slightly since late 2024, with market volatility increasing\n\nrecently. 10-year government bond yields have risen in Europe, particularly after the recent announcement of substantial planned additional expenditures on defence and infrastructure in Germany and at the EU level. However, nominal bond yields in the United States have fallen since mid-January, unwinding the rise observed through the final months of 2024. Indicators suggest that the recent movement reflects a fall in real yields as economic growth slows, rather than lower inflation expectations. The US dollar has also fluctuated, with a depreciation against most currencies this year after a sustained rise through the final quarter of 2024 that contributed to tighter financial conditions in emerging market economies.\n\n11\\. Financial market stress has so far remained contained. Bank credit growth has gradually\n\nrecovered from low levels in advanced economies, as the impact of easier monetary policy has transmitted to lending interest rates. Despite already-stretched valuations in some sectors and some recent price corrections, equity prices are still higher than in November 2024, particularly in Germany, Spain and Italy. Robust growth in corporate bond and leveraged loan issuance has signalled investor confidence and increased risk appetite, with the equity risk premium having fallen in the United States, the United Kingdom and in Europe.\n\n### **Projections**\n\n**_Global growth is expected to soften and inflation to ease further_**\n\n12\\. Global GDP growth is projected to slow from 3.2% in 2024, to 3.1% in 2025 and 3.0% in 2026.\n\nNew data and recent developments result in a downward growth revision from the December 2024 _OECD Economic Outlook_ (Figure 7, Panel A). In quarterly terms, growth is expected to soften from the first quarter of 2025 and to remain subdued thereafter. The imposition of new bilateral tariff rates and the associated increase in policy and geopolitical uncertainty will act as a drag, particularly on business investment and trade. In addition, increased trade costs are expected to feed through gradually to final goods prices, putting additional upward pressure on inflation in many countries and requiring monetary policy to remain restrictive for longer than previously expected."],"full_content":" **13**\n\nOECD ECONOMIC OUTLOOK, INTERIM REPORT MARCH 2025 © OECD 2025\n\n### **Figure 6. Household inflation expectations have recently risen in some countries**\n\nSource: European Central Bank; Bank of Korea; Office for National Statistics; University of Michigan; and OECD calculations.\n\n**_Global financial conditions have tightened slightly and volatility has risen recently_**\n\n10\\. Global financial conditions have tightened slightly since late 2024, with market volatility increasing\n\nrecently. 10-year government bond yields have risen in Europe, particularly after the recent announcement of substantial planned additional expenditures on defence and infrastructure in Germany and at the EU level. However, nominal bond yields in the United States have fallen since mid-January, unwinding the rise observed through the final months of 2024. Indicators suggest that the recent movement reflects a fall in real yields as economic growth slows, rather than lower inflation expectations. The US dollar has also fluctuated, with a depreciation against most currencies this year after a sustained rise through the final quarter of 2024 that contributed to tighter financial conditions in emerging market economies.\n\n11\\. Financial market stress has so far remained contained. Bank credit growth has gradually\n\nrecovered from low levels in advanced economies, as the impact of easier monetary policy has transmitted to lending interest rates. Despite already-stretched valuations in some sectors and some recent price corrections, equity prices are still higher than in November 2024, particularly in Germany, Spain and Italy. Robust growth in corporate bond and leveraged loan issuance has signalled investor confidence and increased risk appetite, with the equity risk premium having fallen in the United States, the United Kingdom and in Europe.\n\n### **Projections**\n\n**_Global growth is expected to soften and inflation to ease further_**\n\n12\\. Global GDP growth is projected to slow from 3.2% in 2024, to 3.1% in 2025 and 3.0% in 2026.\n\nNew data and recent developments result in a downward growth revision from the December 2024 _OECD Economic Outlook_ (Figure 7, Panel A). In quarterly terms, growth is expected to soften from the first quarter of 2025 and to remain subdued thereafter. The imposition of new bilateral tariff rates and the associated increase in policy and geopolitical uncertainty will act as a drag, particularly on business investment and trade. In addition, increased trade costs are expected to feed through gradually to final goods prices, putting additional upward pressure on inflation in many countries and requiring monetary policy to remain restrictive for longer than previously expected.","publish_date":null,"title":"Figure 6. Household inflation expectations have recently risen in some countries Projections","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/89af4857-en_p15.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/89af4857-en_p15.pdf"},"request":{"example_id":"chart/89af4857-en_p15","product_type":"parse","source_file_path":"89af4857-en_p15.pdf"},"started_at":"2026-11-04T19:46:04.279529"} +{"completed_at":"2026-11-04T19:46:07.852023","latency_in_ms":1573,"output":{"example_id":"chart/89af4857-en_p19","grounded_pages":[],"layout_pages":[],"markdown":" **17**\n\nOECD ECONOMIC OUTLOOK, INTERIM REPORT MARCH 2025 © OECD 2025\n\n### **Figure 9. Further trade fragmentation would harm global growth prospects**\n\nSimulation of a rise of 10 percentage points in US tariffs on non-commodity imports from all countries and 10 percentage points higher tariffs on non-commodity imports from the United States by all countries\n\nNote: Illustrative scenario of the impact of 10% US tariffs on non-commodity imports and retaliation from all countries. See technical appendix for details of the shocks considered. Source: OECD calculations using the NiGEM global macroeconomic model and the OECD METRO model.\n\n22\\. A second, and related, downside risk is that inflation remains stickier than otherwise anticipated,\n\nprompting more restrictive monetary policy. A catalyst may be increased tariff rates having more persistent effects on goods inflation by pushing up inflation expectations. Currently tight labour markets in many economies may also make it more likely that a spike in prices feeds higher nominal wage demands. However, such risks may be tempered by the fact that short-term household inflation expectations generally re-anchored through the past few years. Nonetheless, as discussed earlier, the extent of this re- anchoring has varied across countries and the recent turn upwards in household inflation expectations in a few economies bears close monitoring.\n\n23\\. A resurgence of inflation or downside surprises to economic growth could trigger a rapid repricing\n\nin financial markets and a further rise in market volatility. Asset valuations are still stretched in some segments despite recent moderation and low risk premia suggest that shocks could translate rapidly into falling asset prices with broader systemic effects. The latter could be exacerbated in some markets with high equity market concentration and large-scale passive portfolio management (a strategy which tracks the returns of an established market benchmark). Any increase in risk premia could also cause stress in bond markets and private credit markets amid high debt levels and substantial refinancing needs over the next two years.\n\n24\\. There are also various upside risks to the forecasts that would push economic growth higher if\n\nthey were to transpire. The baseline scenario assumes that significantly higher bilateral tariff rates are sustained throughout the projection period in Canada, China, Mexico and the United States. A reversal of these restrictions, or steps towards a broader agreement to lower tariff barriers from current levels would push economic growth higher and reduce inflation relative to the baseline. A peaceful resolution of conflicts in Europe and the Middle East could also reduce geopolitical uncertainty and improve confidence, providing a stronger tailwind for spending and investment. One other potential catalyst for improved sentiment could be a further decline in global energy prices, with an excess supply of oil on global markets expected this year. Recently announced plans in several economies and at the EU level implying an increase in debt-financed government spending, including on defence, are not included in the baseline projections, but would also contribute to stronger near-term economic growth if realised.","pages":[{"line_numbers":[],"markdown":" **17**\n\nOECD ECONOMIC OUTLOOK, INTERIM REPORT MARCH 2025 © OECD 2025\n\n### **Figure 9. Further trade fragmentation would harm global growth prospects**\n\nSimulation of a rise of 10 percentage points in US tariffs on non-commodity imports from all countries and 10 percentage points higher tariffs on non-commodity imports from the United States by all countries\n\nNote: Illustrative scenario of the impact of 10% US tariffs on non-commodity imports and retaliation from all countries. See technical appendix for details of the shocks considered. Source: OECD calculations using the NiGEM global macroeconomic model and the OECD METRO model.\n\n22\\. A second, and related, downside risk is that inflation remains stickier than otherwise anticipated,\n\nprompting more restrictive monetary policy. A catalyst may be increased tariff rates having more persistent effects on goods inflation by pushing up inflation expectations. Currently tight labour markets in many economies may also make it more likely that a spike in prices feeds higher nominal wage demands. However, such risks may be tempered by the fact that short-term household inflation expectations generally re-anchored through the past few years. Nonetheless, as discussed earlier, the extent of this re- anchoring has varied across countries and the recent turn upwards in household inflation expectations in a few economies bears close monitoring.\n\n23\\. A resurgence of inflation or downside surprises to economic growth could trigger a rapid repricing\n\nin financial markets and a further rise in market volatility. Asset valuations are still stretched in some segments despite recent moderation and low risk premia suggest that shocks could translate rapidly into falling asset prices with broader systemic effects. The latter could be exacerbated in some markets with high equity market concentration and large-scale passive portfolio management (a strategy which tracks the returns of an established market benchmark). Any increase in risk premia could also cause stress in bond markets and private credit markets amid high debt levels and substantial refinancing needs over the next two years.\n\n24\\. There are also various upside risks to the forecasts that would push economic growth higher if\n\nthey were to transpire. The baseline scenario assumes that significantly higher bilateral tariff rates are sustained throughout the projection period in Canada, China, Mexico and the United States. A reversal of these restrictions, or steps towards a broader agreement to lower tariff barriers from current levels would push economic growth higher and reduce inflation relative to the baseline. A peaceful resolution of conflicts in Europe and the Middle East could also reduce geopolitical uncertainty and improve confidence, providing a stronger tailwind for spending and investment. One other potential catalyst for improved sentiment could be a further decline in global energy prices, with an excess supply of oil on global markets expected this year. Recently announced plans in several economies and at the EU level implying an increase in debt-financed government spending, including on defence, are not included in the baseline projections, but would also contribute to stronger near-term economic growth if realised.","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":" **17**\n\nOECD ECONOMIC OUTLOOK, INTERIM REPORT MARCH 2025 © OECD 2025\n\n### **Figure 9. Further trade fragmentation would harm global growth prospects**\n\nSimulation of a rise of 10 percentage points in US tariffs on non-commodity imports from all countries and 10 percentage points higher tariffs on non-commodity imports from the United States by all countries\n\nNote: Illustrative scenario of the impact of 10% US tariffs on non-commodity imports and retaliation from all countries. See technical appendix for details of the shocks considered. Source: OECD calculations using the NiGEM global macroeconomic model and the OECD METRO model.\n\n22\\. A second, and related, downside risk is that inflation remains stickier than otherwise anticipated,\n\nprompting more restrictive monetary policy. A catalyst may be increased tariff rates having more persistent effects on goods inflation by pushing up inflation expectations. Currently tight labour markets in many economies may also make it more likely that a spike in prices feeds higher nominal wage demands. However, such risks may be tempered by the fact that short-term household inflation expectations generally re-anchored through the past few years. Nonetheless, as discussed earlier, the extent of this re- anchoring has varied across countries and the recent turn upwards in household inflation expectations in a few economies bears close monitoring.\n\n23\\. A resurgence of inflation or downside surprises to economic growth could trigger a rapid repricing\n\nin financial markets and a further rise in market volatility. Asset valuations are still stretched in some segments despite recent moderation and low risk premia suggest that shocks could translate rapidly into falling asset prices with broader systemic effects. The latter could be exacerbated in some markets with high equity market concentration and large-scale passive portfolio management (a strategy which tracks the returns of an established market benchmark). Any increase in risk premia could also cause stress in bond markets and private credit markets amid high debt levels and substantial refinancing needs over the next two years.\n\n24\\. There are also various upside risks to the forecasts that would push economic growth higher if\n\nthey were to transpire. The baseline scenario assumes that significantly higher bilateral tariff rates are sustained throughout the projection period in Canada, China, Mexico and the United States. A reversal of these restrictions, or steps towards a broader agreement to lower tariff barriers from current levels would push economic growth higher and reduce inflation relative to the baseline. A peaceful resolution of conflicts in Europe and the Middle East could also reduce geopolitical uncertainty and improve confidence, providing a stronger tailwind for spending and investment. One other potential catalyst for improved sentiment could be a further decline in global energy prices, with an excess supply of oil on global markets expected this year. Recently announced plans in several economies and at the EU level implying an increase in debt-financed government spending, including on defence, are not included in the baseline projections, but would also contribute to stronger near-term economic growth if realised.","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/89af4857-en_p19.pdf"]},"response":{"errors":[],"extract_id":"extract_9dfbd9d2598a6f3341586d2e0ec1f95a","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":[" **17**\n\nOECD ECONOMIC OUTLOOK, INTERIM REPORT MARCH 2025 © OECD 2025\n\n### **Figure 9. Further trade fragmentation would harm global growth prospects**\n\nSimulation of a rise of 10 percentage points in US tariffs on non-commodity imports from all countries and 10 percentage points higher tariffs on non-commodity imports from the United States by all countries\n\nNote: Illustrative scenario of the impact of 10% US tariffs on non-commodity imports and retaliation from all countries. See technical appendix for details of the shocks considered. Source: OECD calculations using the NiGEM global macroeconomic model and the OECD METRO model.\n\n22\\. A second, and related, downside risk is that inflation remains stickier than otherwise anticipated,\n\nprompting more restrictive monetary policy. A catalyst may be increased tariff rates having more persistent effects on goods inflation by pushing up inflation expectations. Currently tight labour markets in many economies may also make it more likely that a spike in prices feeds higher nominal wage demands. However, such risks may be tempered by the fact that short-term household inflation expectations generally re-anchored through the past few years. Nonetheless, as discussed earlier, the extent of this re- anchoring has varied across countries and the recent turn upwards in household inflation expectations in a few economies bears close monitoring.\n\n23\\. A resurgence of inflation or downside surprises to economic growth could trigger a rapid repricing\n\nin financial markets and a further rise in market volatility. Asset valuations are still stretched in some segments despite recent moderation and low risk premia suggest that shocks could translate rapidly into falling asset prices with broader systemic effects. The latter could be exacerbated in some markets with high equity market concentration and large-scale passive portfolio management (a strategy which tracks the returns of an established market benchmark). Any increase in risk premia could also cause stress in bond markets and private credit markets amid high debt levels and substantial refinancing needs over the next two years.\n\n24\\. There are also various upside risks to the forecasts that would push economic growth higher if\n\nthey were to transpire. The baseline scenario assumes that significantly higher bilateral tariff rates are sustained throughout the projection period in Canada, China, Mexico and the United States. A reversal of these restrictions, or steps towards a broader agreement to lower tariff barriers from current levels would push economic growth higher and reduce inflation relative to the baseline. A peaceful resolution of conflicts in Europe and the Middle East could also reduce geopolitical uncertainty and improve confidence, providing a stronger tailwind for spending and investment. One other potential catalyst for improved sentiment could be a further decline in global energy prices, with an excess supply of oil on global markets expected this year. Recently announced plans in several economies and at the EU level implying an increase in debt-financed government spending, including on defence, are not included in the baseline projections, but would also contribute to stronger near-term economic growth if realised."],"full_content":" **17**\n\nOECD ECONOMIC OUTLOOK, INTERIM REPORT MARCH 2025 © OECD 2025\n\n### **Figure 9. Further trade fragmentation would harm global growth prospects**\n\nSimulation of a rise of 10 percentage points in US tariffs on non-commodity imports from all countries and 10 percentage points higher tariffs on non-commodity imports from the United States by all countries\n\nNote: Illustrative scenario of the impact of 10% US tariffs on non-commodity imports and retaliation from all countries. See technical appendix for details of the shocks considered. Source: OECD calculations using the NiGEM global macroeconomic model and the OECD METRO model.\n\n22\\. A second, and related, downside risk is that inflation remains stickier than otherwise anticipated,\n\nprompting more restrictive monetary policy. A catalyst may be increased tariff rates having more persistent effects on goods inflation by pushing up inflation expectations. Currently tight labour markets in many economies may also make it more likely that a spike in prices feeds higher nominal wage demands. However, such risks may be tempered by the fact that short-term household inflation expectations generally re-anchored through the past few years. Nonetheless, as discussed earlier, the extent of this re- anchoring has varied across countries and the recent turn upwards in household inflation expectations in a few economies bears close monitoring.\n\n23\\. A resurgence of inflation or downside surprises to economic growth could trigger a rapid repricing\n\nin financial markets and a further rise in market volatility. Asset valuations are still stretched in some segments despite recent moderation and low risk premia suggest that shocks could translate rapidly into falling asset prices with broader systemic effects. The latter could be exacerbated in some markets with high equity market concentration and large-scale passive portfolio management (a strategy which tracks the returns of an established market benchmark). Any increase in risk premia could also cause stress in bond markets and private credit markets amid high debt levels and substantial refinancing needs over the next two years.\n\n24\\. There are also various upside risks to the forecasts that would push economic growth higher if\n\nthey were to transpire. The baseline scenario assumes that significantly higher bilateral tariff rates are sustained throughout the projection period in Canada, China, Mexico and the United States. A reversal of these restrictions, or steps towards a broader agreement to lower tariff barriers from current levels would push economic growth higher and reduce inflation relative to the baseline. A peaceful resolution of conflicts in Europe and the Middle East could also reduce geopolitical uncertainty and improve confidence, providing a stronger tailwind for spending and investment. One other potential catalyst for improved sentiment could be a further decline in global energy prices, with an excess supply of oil on global markets expected this year. Recently announced plans in several economies and at the EU level implying an increase in debt-financed government spending, including on defence, are not included in the baseline projections, but would also contribute to stronger near-term economic growth if realised.","publish_date":null,"title":"Figure 9. Further trade fragmentation would harm global growth prospects","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/89af4857-en_p19.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/89af4857-en_p19.pdf"},"request":{"example_id":"chart/89af4857-en_p19","product_type":"parse","source_file_path":"89af4857-en_p19.pdf"},"started_at":"2026-11-04T19:46:06.278794"} +{"completed_at":"2026-11-04T19:46:09.374188","latency_in_ms":1642,"output":{"example_id":"chart/89af4857-en_p22","grounded_pages":[],"layout_pages":[],"markdown":"**20** \n\nOECD ECONOMIC OUTLOOK, INTERIM REPORT MARCH 2025 © OECD 2025\n\n### **Figure 11. Public debt levels have increased**\n\nGross debt\n\nNote: The chart shows general government financial liabilities for Australia, Canada, Japan, Korea and the United States, and general government gross debt in percent of GDP (Maastricht definition for euro area countries). Latest data point for Argentina, Brazil, China, India, Indonesia, Mexico, South Africa and Türkiye is 2024Q2; Australia, Japan and Korea is 2023Q4. For Korea the earliest available data refers to 2008Q4. Source: Eurostat; IMF Sovereign Debt Investor Base database; OECD Economic Outlook 116 database; Office for National Statistics; and OECD calculations.\n\n**_International cooperation and structural policy reforms are especially important at the current juncture_**\n\n33\\. Governments need to find ways of addressing their concerns together within the global trading\n\nsystem to avoid a significant ratcheting up of retaliatory trade barriers between countries. As already highlighted, a broad-based further increase in trade restrictions would have significant negative impacts on living standards. Efforts to avoid further trade fragmentation should be coupled with reforms that strengthen the resilience of supply chains, including by encouraging firms to diversify both suppliers and buyers. Diversification would be aided by common or shared regulatory standards on key intermediate production inputs between countries. Countries should also not lose sight of the opportunities for potential benefits from collectively agreeing to lower the current tariff and non-tariff barriers on goods and services. In an illustrative scenario in which all countries act to lower their average effective tariff rates by 1½ percentage points (relative to those assumed in the baseline projections), global output would be raised by close to 0.3% by the third year and global inflation reduced by close to ¼ percentage point on average in the first three years (Figure 12).","pages":[{"line_numbers":[],"markdown":"**20** \n\nOECD ECONOMIC OUTLOOK, INTERIM REPORT MARCH 2025 © OECD 2025\n\n### **Figure 11. Public debt levels have increased**\n\nGross debt\n\nNote: The chart shows general government financial liabilities for Australia, Canada, Japan, Korea and the United States, and general government gross debt in percent of GDP (Maastricht definition for euro area countries). Latest data point for Argentina, Brazil, China, India, Indonesia, Mexico, South Africa and Türkiye is 2024Q2; Australia, Japan and Korea is 2023Q4. For Korea the earliest available data refers to 2008Q4. Source: Eurostat; IMF Sovereign Debt Investor Base database; OECD Economic Outlook 116 database; Office for National Statistics; and OECD calculations.\n\n**_International cooperation and structural policy reforms are especially important at the current juncture_**\n\n33\\. Governments need to find ways of addressing their concerns together within the global trading\n\nsystem to avoid a significant ratcheting up of retaliatory trade barriers between countries. As already highlighted, a broad-based further increase in trade restrictions would have significant negative impacts on living standards. Efforts to avoid further trade fragmentation should be coupled with reforms that strengthen the resilience of supply chains, including by encouraging firms to diversify both suppliers and buyers. Diversification would be aided by common or shared regulatory standards on key intermediate production inputs between countries. Countries should also not lose sight of the opportunities for potential benefits from collectively agreeing to lower the current tariff and non-tariff barriers on goods and services. In an illustrative scenario in which all countries act to lower their average effective tariff rates by 1½ percentage points (relative to those assumed in the baseline projections), global output would be raised by close to 0.3% by the third year and global inflation reduced by close to ¼ percentage point on average in the first three years (Figure 12).","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**20** \n\nOECD ECONOMIC OUTLOOK, INTERIM REPORT MARCH 2025 © OECD 2025\n\n### **Figure 11. Public debt levels have increased**\n\nGross debt\n\nNote: The chart shows general government financial liabilities for Australia, Canada, Japan, Korea and the United States, and general government gross debt in percent of GDP (Maastricht definition for euro area countries). Latest data point for Argentina, Brazil, China, India, Indonesia, Mexico, South Africa and Türkiye is 2024Q2; Australia, Japan and Korea is 2023Q4. For Korea the earliest available data refers to 2008Q4. Source: Eurostat; IMF Sovereign Debt Investor Base database; OECD Economic Outlook 116 database; Office for National Statistics; and OECD calculations.\n\n**_International cooperation and structural policy reforms are especially important at the current juncture_**\n\n33\\. Governments need to find ways of addressing their concerns together within the global trading\n\nsystem to avoid a significant ratcheting up of retaliatory trade barriers between countries. As already highlighted, a broad-based further increase in trade restrictions would have significant negative impacts on living standards. Efforts to avoid further trade fragmentation should be coupled with reforms that strengthen the resilience of supply chains, including by encouraging firms to diversify both suppliers and buyers. Diversification would be aided by common or shared regulatory standards on key intermediate production inputs between countries. Countries should also not lose sight of the opportunities for potential benefits from collectively agreeing to lower the current tariff and non-tariff barriers on goods and services. In an illustrative scenario in which all countries act to lower their average effective tariff rates by 1½ percentage points (relative to those assumed in the baseline projections), global output would be raised by close to 0.3% by the third year and global inflation reduced by close to ¼ percentage point on average in the first three years (Figure 12).","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/89af4857-en_p22.pdf"]},"response":{"errors":[],"extract_id":"extract_4eb761376399c65bc952709eeba56820","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**20** \n\nOECD ECONOMIC OUTLOOK, INTERIM REPORT MARCH 2025 © OECD 2025\n\n### **Figure 11. Public debt levels have increased**\n\nGross debt\n\nNote: The chart shows general government financial liabilities for Australia, Canada, Japan, Korea and the United States, and general government gross debt in percent of GDP (Maastricht definition for euro area countries). Latest data point for Argentina, Brazil, China, India, Indonesia, Mexico, South Africa and Türkiye is 2024Q2; Australia, Japan and Korea is 2023Q4. For Korea the earliest available data refers to 2008Q4. Source: Eurostat; IMF Sovereign Debt Investor Base database; OECD Economic Outlook 116 database; Office for National Statistics; and OECD calculations.\n\n**_International cooperation and structural policy reforms are especially important at the current juncture_**\n\n33\\. Governments need to find ways of addressing their concerns together within the global trading\n\nsystem to avoid a significant ratcheting up of retaliatory trade barriers between countries. As already highlighted, a broad-based further increase in trade restrictions would have significant negative impacts on living standards. Efforts to avoid further trade fragmentation should be coupled with reforms that strengthen the resilience of supply chains, including by encouraging firms to diversify both suppliers and buyers. Diversification would be aided by common or shared regulatory standards on key intermediate production inputs between countries. Countries should also not lose sight of the opportunities for potential benefits from collectively agreeing to lower the current tariff and non-tariff barriers on goods and services. In an illustrative scenario in which all countries act to lower their average effective tariff rates by 1½ percentage points (relative to those assumed in the baseline projections), global output would be raised by close to 0.3% by the third year and global inflation reduced by close to ¼ percentage point on average in the first three years (Figure 12)."],"full_content":"**20** \n\nOECD ECONOMIC OUTLOOK, INTERIM REPORT MARCH 2025 © OECD 2025\n\n### **Figure 11. Public debt levels have increased**\n\nGross debt\n\nNote: The chart shows general government financial liabilities for Australia, Canada, Japan, Korea and the United States, and general government gross debt in percent of GDP (Maastricht definition for euro area countries). Latest data point for Argentina, Brazil, China, India, Indonesia, Mexico, South Africa and Türkiye is 2024Q2; Australia, Japan and Korea is 2023Q4. For Korea the earliest available data refers to 2008Q4. Source: Eurostat; IMF Sovereign Debt Investor Base database; OECD Economic Outlook 116 database; Office for National Statistics; and OECD calculations.\n\n**_International cooperation and structural policy reforms are especially important at the current juncture_**\n\n33\\. Governments need to find ways of addressing their concerns together within the global trading\n\nsystem to avoid a significant ratcheting up of retaliatory trade barriers between countries. As already highlighted, a broad-based further increase in trade restrictions would have significant negative impacts on living standards. Efforts to avoid further trade fragmentation should be coupled with reforms that strengthen the resilience of supply chains, including by encouraging firms to diversify both suppliers and buyers. Diversification would be aided by common or shared regulatory standards on key intermediate production inputs between countries. Countries should also not lose sight of the opportunities for potential benefits from collectively agreeing to lower the current tariff and non-tariff barriers on goods and services. In an illustrative scenario in which all countries act to lower their average effective tariff rates by 1½ percentage points (relative to those assumed in the baseline projections), global output would be raised by close to 0.3% by the third year and global inflation reduced by close to ¼ percentage point on average in the first three years (Figure 12).","publish_date":null,"title":"Figure 11. Public debt levels have increased","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/89af4857-en_p22.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/89af4857-en_p22.pdf"},"request":{"example_id":"chart/89af4857-en_p22","product_type":"parse","source_file_path":"89af4857-en_p22.pdf"},"started_at":"2026-11-04T19:46:07.731995"} +{"completed_at":"2026-11-04T19:31:15.032695","latency_in_ms":2158,"output":{"example_id":"chart/9f653ca1-en_p17","grounded_pages":[],"layout_pages":[],"markdown":" **15**\n\nOECD ECONOMIC OUTLOOK, VOLUME 2025 ISSUE 2 © OECD 2025\n\nup about 15.5% of total world merchandise trade, with two-thirds of that originating in Asia, 3% in Europe and 2.5% in North America (WTO, 2025a).\n\n### **Figure 1.2. Trade growth has been resilient**\n\nNote: The data correspond to quarterly trade of goods and services in volume (average of exports and imports). In Panel A, “Advanced economies” include OECD member countries other than Chile, Colombia, Costa Rica, Mexico, and Türkiye. “Other emerging Asia” includes Hong Kong (China), Indonesia, Malaysia, the Philippines, Singapore, Chinese Taipei, Thailand and Viet Nam. “Other emerging market economies” are all remaining emerging-market economies. In Panel B annual growth rates are shown for 2022-2024. “North America” includes Canada, Mexico and the United States. “Other AEs” in Panel B are the 33 advanced economies who are OECD members excluding “North America”, Ireland and Switzerland. “Other EM Asia” corresponds to “Other emerging Asia” in Panel A. “Other EMEs” are the remaining emerging-market economies not shown separately. Source: OECD Economic Outlook 118 database; and OECD calculations.\n\nStatLink 2 https://stat.link/dnwhs2\n\n### **Figure 1.3. Strong tech-related demand is supporting trade, particularly in Asia**\n\nNote: Panel A is based on 2022 data. The green bar shows that almost 20% of OECD domestic value added of the manufacturing industry was consumed as a final good by the United States; the red bar shows that around 13% of G20 EME domestic value added of the information industry was consumed as a final good by China. South-East Asian economies include Hong Kong (China), Malaysia, the Philippines, Singapore, Chinese Taipei, Thailand and Viet Nam. The G20 emerging-market economies include Argentina, Brazil, India, Indonesia, Saudi Arabia and South Africa. Manufacturing includes all ISIC codes 10 to 33; business services from 45 to 82; and the information industry shows ISIC codes 25 and 58 to 63. Panel B is based on values of exports. The Chinese Taipei tech indicator includes integrated circuits; dynamic random access memory integrated circuits; bare printed circuit boards; computer storage devices; and computer parts and accessories; the Korean tech exports indicator is based on semiconductors. Source: OECD Trade in Value Added (TiVA) 2025 edition; Korean Ministry of Trade, Industry and Resources; Chinese Taipei Ministry of Finance; and OECD calculations.\n\nStatLink 2 https://stat.link/lv53mq","pages":[{"line_numbers":[],"markdown":" **15**\n\nOECD ECONOMIC OUTLOOK, VOLUME 2025 ISSUE 2 © OECD 2025\n\nup about 15.5% of total world merchandise trade, with two-thirds of that originating in Asia, 3% in Europe and 2.5% in North America (WTO, 2025a).\n\n### **Figure 1.2. Trade growth has been resilient**\n\nNote: The data correspond to quarterly trade of goods and services in volume (average of exports and imports). In Panel A, “Advanced economies” include OECD member countries other than Chile, Colombia, Costa Rica, Mexico, and Türkiye. “Other emerging Asia” includes Hong Kong (China), Indonesia, Malaysia, the Philippines, Singapore, Chinese Taipei, Thailand and Viet Nam. “Other emerging market economies” are all remaining emerging-market economies. In Panel B annual growth rates are shown for 2022-2024. “North America” includes Canada, Mexico and the United States. “Other AEs” in Panel B are the 33 advanced economies who are OECD members excluding “North America”, Ireland and Switzerland. “Other EM Asia” corresponds to “Other emerging Asia” in Panel A. “Other EMEs” are the remaining emerging-market economies not shown separately. Source: OECD Economic Outlook 118 database; and OECD calculations.\n\nStatLink 2 https://stat.link/dnwhs2\n\n### **Figure 1.3. Strong tech-related demand is supporting trade, particularly in Asia**\n\nNote: Panel A is based on 2022 data. The green bar shows that almost 20% of OECD domestic value added of the manufacturing industry was consumed as a final good by the United States; the red bar shows that around 13% of G20 EME domestic value added of the information industry was consumed as a final good by China. South-East Asian economies include Hong Kong (China), Malaysia, the Philippines, Singapore, Chinese Taipei, Thailand and Viet Nam. The G20 emerging-market economies include Argentina, Brazil, India, Indonesia, Saudi Arabia and South Africa. Manufacturing includes all ISIC codes 10 to 33; business services from 45 to 82; and the information industry shows ISIC codes 25 and 58 to 63. Panel B is based on values of exports. The Chinese Taipei tech indicator includes integrated circuits; dynamic random access memory integrated circuits; bare printed circuit boards; computer storage devices; and computer parts and accessories; the Korean tech exports indicator is based on semiconductors. Source: OECD Trade in Value Added (TiVA) 2025 edition; Korean Ministry of Trade, Industry and Resources; Chinese Taipei Ministry of Finance; and OECD calculations.\n\nStatLink 2 https://stat.link/lv53mq","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":" **15**\n\nOECD ECONOMIC OUTLOOK, VOLUME 2025 ISSUE 2 © OECD 2025\n\nup about 15.5% of total world merchandise trade, with two-thirds of that originating in Asia, 3% in Europe and 2.5% in North America (WTO, 2025a).\n\n### **Figure 1.2. Trade growth has been resilient**\n\nNote: The data correspond to quarterly trade of goods and services in volume (average of exports and imports). In Panel A, “Advanced economies” include OECD member countries other than Chile, Colombia, Costa Rica, Mexico, and Türkiye. “Other emerging Asia” includes Hong Kong (China), Indonesia, Malaysia, the Philippines, Singapore, Chinese Taipei, Thailand and Viet Nam. “Other emerging market economies” are all remaining emerging-market economies. In Panel B annual growth rates are shown for 2022-2024. “North America” includes Canada, Mexico and the United States. “Other AEs” in Panel B are the 33 advanced economies who are OECD members excluding “North America”, Ireland and Switzerland. “Other EM Asia” corresponds to “Other emerging Asia” in Panel A. “Other EMEs” are the remaining emerging-market economies not shown separately. Source: OECD Economic Outlook 118 database; and OECD calculations.\n\nStatLink 2 https://stat.link/dnwhs2\n\n### **Figure 1.3. Strong tech-related demand is supporting trade, particularly in Asia**\n\nNote: Panel A is based on 2022 data. The green bar shows that almost 20% of OECD domestic value added of the manufacturing industry was consumed as a final good by the United States; the red bar shows that around 13% of G20 EME domestic value added of the information industry was consumed as a final good by China. South-East Asian economies include Hong Kong (China), Malaysia, the Philippines, Singapore, Chinese Taipei, Thailand and Viet Nam. The G20 emerging-market economies include Argentina, Brazil, India, Indonesia, Saudi Arabia and South Africa. Manufacturing includes all ISIC codes 10 to 33; business services from 45 to 82; and the information industry shows ISIC codes 25 and 58 to 63. Panel B is based on values of exports. The Chinese Taipei tech indicator includes integrated circuits; dynamic random access memory integrated circuits; bare printed circuit boards; computer storage devices; and computer parts and accessories; the Korean tech exports indicator is based on semiconductors. Source: OECD Trade in Value Added (TiVA) 2025 edition; Korean Ministry of Trade, Industry and Resources; Chinese Taipei Ministry of Finance; and OECD calculations.\n\nStatLink 2 https://stat.link/lv53mq","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/9f653ca1-en_p17.pdf"]},"response":{"errors":[],"extract_id":"extract_f1e59b0e4132f1973794b35a4ed545e9","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":[" **15**\n\nOECD ECONOMIC OUTLOOK, VOLUME 2025 ISSUE 2 © OECD 2025\n\nup about 15.5% of total world merchandise trade, with two-thirds of that originating in Asia, 3% in Europe and 2.5% in North America (WTO, 2025a).\n\n### **Figure 1.2. Trade growth has been resilient**\n\nNote: The data correspond to quarterly trade of goods and services in volume (average of exports and imports). In Panel A, “Advanced economies” include OECD member countries other than Chile, Colombia, Costa Rica, Mexico, and Türkiye. “Other emerging Asia” includes Hong Kong (China), Indonesia, Malaysia, the Philippines, Singapore, Chinese Taipei, Thailand and Viet Nam. “Other emerging market economies” are all remaining emerging-market economies. In Panel B annual growth rates are shown for 2022-2024. “North America” includes Canada, Mexico and the United States. “Other AEs” in Panel B are the 33 advanced economies who are OECD members excluding “North America”, Ireland and Switzerland. “Other EM Asia” corresponds to “Other emerging Asia” in Panel A. “Other EMEs” are the remaining emerging-market economies not shown separately. Source: OECD Economic Outlook 118 database; and OECD calculations.\n\nStatLink 2 https://stat.link/dnwhs2\n\n### **Figure 1.3. Strong tech-related demand is supporting trade, particularly in Asia**\n\nNote: Panel A is based on 2022 data. The green bar shows that almost 20% of OECD domestic value added of the manufacturing industry was consumed as a final good by the United States; the red bar shows that around 13% of G20 EME domestic value added of the information industry was consumed as a final good by China. South-East Asian economies include Hong Kong (China), Malaysia, the Philippines, Singapore, Chinese Taipei, Thailand and Viet Nam. The G20 emerging-market economies include Argentina, Brazil, India, Indonesia, Saudi Arabia and South Africa. Manufacturing includes all ISIC codes 10 to 33; business services from 45 to 82; and the information industry shows ISIC codes 25 and 58 to 63. Panel B is based on values of exports. The Chinese Taipei tech indicator includes integrated circuits; dynamic random access memory integrated circuits; bare printed circuit boards; computer storage devices; and computer parts and accessories; the Korean tech exports indicator is based on semiconductors. Source: OECD Trade in Value Added (TiVA) 2025 edition; Korean Ministry of Trade, Industry and Resources; Chinese Taipei Ministry of Finance; and OECD calculations.\n\nStatLink 2 https://stat.link/lv53mq"],"full_content":" **15**\n\nOECD ECONOMIC OUTLOOK, VOLUME 2025 ISSUE 2 © OECD 2025\n\nup about 15.5% of total world merchandise trade, with two-thirds of that originating in Asia, 3% in Europe and 2.5% in North America (WTO, 2025a).\n\n### **Figure 1.2. Trade growth has been resilient**\n\nNote: The data correspond to quarterly trade of goods and services in volume (average of exports and imports). In Panel A, “Advanced economies” include OECD member countries other than Chile, Colombia, Costa Rica, Mexico, and Türkiye. “Other emerging Asia” includes Hong Kong (China), Indonesia, Malaysia, the Philippines, Singapore, Chinese Taipei, Thailand and Viet Nam. “Other emerging market economies” are all remaining emerging-market economies. In Panel B annual growth rates are shown for 2022-2024. “North America” includes Canada, Mexico and the United States. “Other AEs” in Panel B are the 33 advanced economies who are OECD members excluding “North America”, Ireland and Switzerland. “Other EM Asia” corresponds to “Other emerging Asia” in Panel A. “Other EMEs” are the remaining emerging-market economies not shown separately. Source: OECD Economic Outlook 118 database; and OECD calculations.\n\nStatLink 2 https://stat.link/dnwhs2\n\n### **Figure 1.3. Strong tech-related demand is supporting trade, particularly in Asia**\n\nNote: Panel A is based on 2022 data. The green bar shows that almost 20% of OECD domestic value added of the manufacturing industry was consumed as a final good by the United States; the red bar shows that around 13% of G20 EME domestic value added of the information industry was consumed as a final good by China. South-East Asian economies include Hong Kong (China), Malaysia, the Philippines, Singapore, Chinese Taipei, Thailand and Viet Nam. The G20 emerging-market economies include Argentina, Brazil, India, Indonesia, Saudi Arabia and South Africa. Manufacturing includes all ISIC codes 10 to 33; business services from 45 to 82; and the information industry shows ISIC codes 25 and 58 to 63. Panel B is based on values of exports. The Chinese Taipei tech indicator includes integrated circuits; dynamic random access memory integrated circuits; bare printed circuit boards; computer storage devices; and computer parts and accessories; the Korean tech exports indicator is based on semiconductors. Source: OECD Trade in Value Added (TiVA) 2025 edition; Korean Ministry of Trade, Industry and Resources; Chinese Taipei Ministry of Finance; and OECD calculations.\n\nStatLink 2 https://stat.link/lv53mq","publish_date":null,"title":"Figure 1.2. Trade growth has been resilient Figure 1.3. Strong tech-related demand is supporting trade, particularly in Asia","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/9f653ca1-en_p17.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/9f653ca1-en_p17.pdf"},"request":{"example_id":"chart/9f653ca1-en_p17","product_type":"parse","source_file_path":"9f653ca1-en_p17.pdf"},"started_at":"2026-11-04T19:31:12.874351"} +{"completed_at":"2026-11-04T19:46:15.247513","latency_in_ms":3812,"output":{"example_id":"chart/9f653ca1-en_p21","grounded_pages":[],"layout_pages":[],"markdown":" **19**\n\nOECD ECONOMIC OUTLOOK, VOLUME 2025 ISSUE 2 © OECD 2025\n\nPanel A), portending an increase in tariff collection in the coming months. ² The extent of the gap between revenues and estimated effective tariffs differs across types of goods. Tariff collection rates on cars and parts as well as steel and aluminium, which have been in place for some time and have few global exceptions, are closer to their effective tariff rates than is the case for other goods.\n\n### **Figure 1.7. The impact of US tariffs is increasingly being felt**\n\nNote: In Panel A, the monthly average 2025 estimated effective tariff rate is based on applicable rates to products and countries at the time they became effective, weighted by country-specific product shares of United States imports in 2024. The estimates include updated weights for compliance of imports from Canada and Mexico (see Box 1.2 for more details). Data before 2025 based on WITS average effective tariff rates, corrected for Section 301 actions on China undertaken between 2018 and 2024. 2025 tariff revenues estimated as a share of import values including cost, insurance and freight, based on monthly data from US Census Bureau. Data before this are from USITC (2025) \"U.S. imports for consumption, duties collected, and ratio of duties to value, 1891-2024\". Data in Panel B are based on import values. Goods which face no new tariffs in 2025 are all goods exempted from country-specific tariff increases as of August 2025, excluding goods imports in these categories from China (since the additional 20 percentage point tariff increase imposed on China in March 2025 is unaffected by the exemption). Source: OECD Economic Outlook 118 database; US Census Bureau; United States International Trade Commission; World Integrated Trade Solution (WITS); and OECD calculations.\n\nStatLink 2 https://stat.link/tvyw09\n\nDespite the lagged effects of higher tariffs, there is a sharp decrease in the value of US imported goods subject to tariffs compared to non-tariffed goods (Figure 1.7, Panel B). This suggests that tariffs are affecting demand, and will continue to weigh on trade volumes as announced tariffs come into full effect. This could in turn further weigh on the estimated effective tariff rate, reducing the divergence with collected duties. Continued strength in demand for technology products should still provide some support for global trade, although new export orders for manufacturing in most large economies remain relatively weak outside of the Asian emerging-market economies. Trade policy uncertainty has fallen from recent peaks, but recent announcements about further changes to bilateral tariff barriers and the possible broadening of licencing requirements for rare earth materials in China could add to uncertainty.\n\n### **Box 1.2. US tariff rates: in law and in effect**\n\nIn recent months the United States has continued to announce additional tariff increases on imports from most countries, alongside separate sector-specific tariffs on steel and aluminium, cars and parts, trucks, buses, timber, and copper. This box discusses some of the issues involved in calculating the aggregate\n\n2 The effective tariff rate in November 2025 is shown as 14.6% in Figure 1.7 rather than 14.0%. This is because a monthly average tariff rate is used in Figure 1.7 whereas the latter reflects the effective tariff rate following the policy changes that occurred by mid-November.","pages":[{"line_numbers":[],"markdown":" **19**\n\nOECD ECONOMIC OUTLOOK, VOLUME 2025 ISSUE 2 © OECD 2025\n\nPanel A), portending an increase in tariff collection in the coming months. ² The extent of the gap between revenues and estimated effective tariffs differs across types of goods. Tariff collection rates on cars and parts as well as steel and aluminium, which have been in place for some time and have few global exceptions, are closer to their effective tariff rates than is the case for other goods.\n\n### **Figure 1.7. The impact of US tariffs is increasingly being felt**\n\nNote: In Panel A, the monthly average 2025 estimated effective tariff rate is based on applicable rates to products and countries at the time they became effective, weighted by country-specific product shares of United States imports in 2024. The estimates include updated weights for compliance of imports from Canada and Mexico (see Box 1.2 for more details). Data before 2025 based on WITS average effective tariff rates, corrected for Section 301 actions on China undertaken between 2018 and 2024. 2025 tariff revenues estimated as a share of import values including cost, insurance and freight, based on monthly data from US Census Bureau. Data before this are from USITC (2025) \"U.S. imports for consumption, duties collected, and ratio of duties to value, 1891-2024\". Data in Panel B are based on import values. Goods which face no new tariffs in 2025 are all goods exempted from country-specific tariff increases as of August 2025, excluding goods imports in these categories from China (since the additional 20 percentage point tariff increase imposed on China in March 2025 is unaffected by the exemption). Source: OECD Economic Outlook 118 database; US Census Bureau; United States International Trade Commission; World Integrated Trade Solution (WITS); and OECD calculations.\n\nStatLink 2 https://stat.link/tvyw09\n\nDespite the lagged effects of higher tariffs, there is a sharp decrease in the value of US imported goods subject to tariffs compared to non-tariffed goods (Figure 1.7, Panel B). This suggests that tariffs are affecting demand, and will continue to weigh on trade volumes as announced tariffs come into full effect. This could in turn further weigh on the estimated effective tariff rate, reducing the divergence with collected duties. Continued strength in demand for technology products should still provide some support for global trade, although new export orders for manufacturing in most large economies remain relatively weak outside of the Asian emerging-market economies. Trade policy uncertainty has fallen from recent peaks, but recent announcements about further changes to bilateral tariff barriers and the possible broadening of licencing requirements for rare earth materials in China could add to uncertainty.\n\n### **Box 1.2. US tariff rates: in law and in effect**\n\nIn recent months the United States has continued to announce additional tariff increases on imports from most countries, alongside separate sector-specific tariffs on steel and aluminium, cars and parts, trucks, buses, timber, and copper. This box discusses some of the issues involved in calculating the aggregate\n\n2 The effective tariff rate in November 2025 is shown as 14.6% in Figure 1.7 rather than 14.0%. This is because a monthly average tariff rate is used in Figure 1.7 whereas the latter reflects the effective tariff rate following the policy changes that occurred by mid-November.","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":" **19**\n\nOECD ECONOMIC OUTLOOK, VOLUME 2025 ISSUE 2 © OECD 2025\n\nPanel A), portending an increase in tariff collection in the coming months. ² The extent of the gap between revenues and estimated effective tariffs differs across types of goods. Tariff collection rates on cars and parts as well as steel and aluminium, which have been in place for some time and have few global exceptions, are closer to their effective tariff rates than is the case for other goods.\n\n### **Figure 1.7. The impact of US tariffs is increasingly being felt**\n\nNote: In Panel A, the monthly average 2025 estimated effective tariff rate is based on applicable rates to products and countries at the time they became effective, weighted by country-specific product shares of United States imports in 2024. The estimates include updated weights for compliance of imports from Canada and Mexico (see Box 1.2 for more details). Data before 2025 based on WITS average effective tariff rates, corrected for Section 301 actions on China undertaken between 2018 and 2024. 2025 tariff revenues estimated as a share of import values including cost, insurance and freight, based on monthly data from US Census Bureau. Data before this are from USITC (2025) \"U.S. imports for consumption, duties collected, and ratio of duties to value, 1891-2024\". Data in Panel B are based on import values. Goods which face no new tariffs in 2025 are all goods exempted from country-specific tariff increases as of August 2025, excluding goods imports in these categories from China (since the additional 20 percentage point tariff increase imposed on China in March 2025 is unaffected by the exemption). Source: OECD Economic Outlook 118 database; US Census Bureau; United States International Trade Commission; World Integrated Trade Solution (WITS); and OECD calculations.\n\nStatLink 2 https://stat.link/tvyw09\n\nDespite the lagged effects of higher tariffs, there is a sharp decrease in the value of US imported goods subject to tariffs compared to non-tariffed goods (Figure 1.7, Panel B). This suggests that tariffs are affecting demand, and will continue to weigh on trade volumes as announced tariffs come into full effect. This could in turn further weigh on the estimated effective tariff rate, reducing the divergence with collected duties. Continued strength in demand for technology products should still provide some support for global trade, although new export orders for manufacturing in most large economies remain relatively weak outside of the Asian emerging-market economies. Trade policy uncertainty has fallen from recent peaks, but recent announcements about further changes to bilateral tariff barriers and the possible broadening of licencing requirements for rare earth materials in China could add to uncertainty.\n\n### **Box 1.2. US tariff rates: in law and in effect**\n\nIn recent months the United States has continued to announce additional tariff increases on imports from most countries, alongside separate sector-specific tariffs on steel and aluminium, cars and parts, trucks, buses, timber, and copper. This box discusses some of the issues involved in calculating the aggregate\n\n2 The effective tariff rate in November 2025 is shown as 14.6% in Figure 1.7 rather than 14.0%. This is because a monthly average tariff rate is used in Figure 1.7 whereas the latter reflects the effective tariff rate following the policy changes that occurred by mid-November.","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/9f653ca1-en_p21.pdf"]},"response":{"errors":[],"extract_id":"extract_347ccad9ba1d44bc6c6aa078527960ba","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":[" **19**\n\nOECD ECONOMIC OUTLOOK, VOLUME 2025 ISSUE 2 © OECD 2025\n\nPanel A), portending an increase in tariff collection in the coming months. ² The extent of the gap between revenues and estimated effective tariffs differs across types of goods. Tariff collection rates on cars and parts as well as steel and aluminium, which have been in place for some time and have few global exceptions, are closer to their effective tariff rates than is the case for other goods.\n\n### **Figure 1.7. The impact of US tariffs is increasingly being felt**\n\nNote: In Panel A, the monthly average 2025 estimated effective tariff rate is based on applicable rates to products and countries at the time they became effective, weighted by country-specific product shares of United States imports in 2024. The estimates include updated weights for compliance of imports from Canada and Mexico (see Box 1.2 for more details). Data before 2025 based on WITS average effective tariff rates, corrected for Section 301 actions on China undertaken between 2018 and 2024. 2025 tariff revenues estimated as a share of import values including cost, insurance and freight, based on monthly data from US Census Bureau. Data before this are from USITC (2025) \"U.S. imports for consumption, duties collected, and ratio of duties to value, 1891-2024\". Data in Panel B are based on import values. Goods which face no new tariffs in 2025 are all goods exempted from country-specific tariff increases as of August 2025, excluding goods imports in these categories from China (since the additional 20 percentage point tariff increase imposed on China in March 2025 is unaffected by the exemption). Source: OECD Economic Outlook 118 database; US Census Bureau; United States International Trade Commission; World Integrated Trade Solution (WITS); and OECD calculations.\n\nStatLink 2 https://stat.link/tvyw09\n\nDespite the lagged effects of higher tariffs, there is a sharp decrease in the value of US imported goods subject to tariffs compared to non-tariffed goods (Figure 1.7, Panel B). This suggests that tariffs are affecting demand, and will continue to weigh on trade volumes as announced tariffs come into full effect. This could in turn further weigh on the estimated effective tariff rate, reducing the divergence with collected duties. Continued strength in demand for technology products should still provide some support for global trade, although new export orders for manufacturing in most large economies remain relatively weak outside of the Asian emerging-market economies. Trade policy uncertainty has fallen from recent peaks, but recent announcements about further changes to bilateral tariff barriers and the possible broadening of licencing requirements for rare earth materials in China could add to uncertainty.\n\n### **Box 1.2. US tariff rates: in law and in effect**\n\nIn recent months the United States has continued to announce additional tariff increases on imports from most countries, alongside separate sector-specific tariffs on steel and aluminium, cars and parts, trucks, buses, timber, and copper. This box discusses some of the issues involved in calculating the aggregate\n\n2 The effective tariff rate in November 2025 is shown as 14.6% in Figure 1.7 rather than 14.0%. This is because a monthly average tariff rate is used in Figure 1.7 whereas the latter reflects the effective tariff rate following the policy changes that occurred by mid-November."],"full_content":" **19**\n\nOECD ECONOMIC OUTLOOK, VOLUME 2025 ISSUE 2 © OECD 2025\n\nPanel A), portending an increase in tariff collection in the coming months. ² The extent of the gap between revenues and estimated effective tariffs differs across types of goods. Tariff collection rates on cars and parts as well as steel and aluminium, which have been in place for some time and have few global exceptions, are closer to their effective tariff rates than is the case for other goods.\n\n### **Figure 1.7. The impact of US tariffs is increasingly being felt**\n\nNote: In Panel A, the monthly average 2025 estimated effective tariff rate is based on applicable rates to products and countries at the time they became effective, weighted by country-specific product shares of United States imports in 2024. The estimates include updated weights for compliance of imports from Canada and Mexico (see Box 1.2 for more details). Data before 2025 based on WITS average effective tariff rates, corrected for Section 301 actions on China undertaken between 2018 and 2024. 2025 tariff revenues estimated as a share of import values including cost, insurance and freight, based on monthly data from US Census Bureau. Data before this are from USITC (2025) \"U.S. imports for consumption, duties collected, and ratio of duties to value, 1891-2024\". Data in Panel B are based on import values. Goods which face no new tariffs in 2025 are all goods exempted from country-specific tariff increases as of August 2025, excluding goods imports in these categories from China (since the additional 20 percentage point tariff increase imposed on China in March 2025 is unaffected by the exemption). Source: OECD Economic Outlook 118 database; US Census Bureau; United States International Trade Commission; World Integrated Trade Solution (WITS); and OECD calculations.\n\nStatLink 2 https://stat.link/tvyw09\n\nDespite the lagged effects of higher tariffs, there is a sharp decrease in the value of US imported goods subject to tariffs compared to non-tariffed goods (Figure 1.7, Panel B). This suggests that tariffs are affecting demand, and will continue to weigh on trade volumes as announced tariffs come into full effect. This could in turn further weigh on the estimated effective tariff rate, reducing the divergence with collected duties. Continued strength in demand for technology products should still provide some support for global trade, although new export orders for manufacturing in most large economies remain relatively weak outside of the Asian emerging-market economies. Trade policy uncertainty has fallen from recent peaks, but recent announcements about further changes to bilateral tariff barriers and the possible broadening of licencing requirements for rare earth materials in China could add to uncertainty.\n\n### **Box 1.2. US tariff rates: in law and in effect**\n\nIn recent months the United States has continued to announce additional tariff increases on imports from most countries, alongside separate sector-specific tariffs on steel and aluminium, cars and parts, trucks, buses, timber, and copper. This box discusses some of the issues involved in calculating the aggregate\n\n2 The effective tariff rate in November 2025 is shown as 14.6% in Figure 1.7 rather than 14.0%. This is because a monthly average tariff rate is used in Figure 1.7 whereas the latter reflects the effective tariff rate following the policy changes that occurred by mid-November.","publish_date":null,"title":"Figure 1.7. The impact of US tariffs is increasingly being felt Box 1.2. US tariff rates: in law and in","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/9f653ca1-en_p21.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/9f653ca1-en_p21.pdf"},"request":{"example_id":"chart/9f653ca1-en_p21","product_type":"parse","source_file_path":"9f653ca1-en_p21.pdf"},"started_at":"2026-11-04T19:46:11.435063"} +{"completed_at":"2026-11-04T19:46:15.613807","latency_in_ms":1886,"output":{"example_id":"chart/9f653ca1-en_p23","grounded_pages":[],"layout_pages":[],"markdown":" **21**\n\nOECD ECONOMIC OUTLOOK, VOLUME 2025 ISSUE 2 © OECD 2025\n\nfallen elsewhere (Figure 1.9, Panel B). The updated effective estimates for Canada and Mexico now assume that 80% of goods imported into the United States will be USMCA-compliant.\n\n### **Figure 1.8. Estimated US effective tariff rates**\n\nNote: Estimates based on trade policy announcements as of 14 November 2025 and calculated using weights based on product level data for US imports by country in 2024. These take account of country-specific tariff rates and sector-specific changes due to a 25% tariff rise for listed cars, parts, trucks and buses; a 50% tariff rise on steel, aluminium and copper products; and zero-rated general goods as well as those related to semiconductors. Europe, Japan and Korea face a maximum 15% tariff on cars and parts, and timber, and 0% tariffs are applied to civil aviation products, generic pharmaceuticals and certain natural resources). The United Kingdom faces a 25% tariff rise on steel and aluminium, 7.5% for British cars within the annual quota, 10% on timber products, and duty-free civil aviation products. China, Brazil and India face additional country- specific tariffs, with varying application across products. Data classifications are based on product-level lists published by the US administration. Source: United States International Trade Commission; US Census Bureau, and OECD calculations.\n\nStatLink 2 https://stat.link/qxnj8w\n\nThe share of duty-free imports is generally higher for imports from Canada and Mexico than elsewhere across a range of goods (Figure 1.10, Panel A). Taking the share of duty-free goods as a proxy for USMCA compliance, around 60% and 53% of car imports for Mexico and Canada respectively are USMCA compliant. For steel and aluminium, however, Canada has a compliance rate of just 33% compared to 66% for Mexico. For goods excluding imports of cars and parts, steel and aluminium, and items exempted from tariffs, compliance rates are 93% and 90% for Canada and Mexico, respectively.\n\nOverall, these compliance rates result in US imports of cars and parts from Canada and Mexico currently incurring duties of 12% and 11% respectively, slightly less than half the 27% being charged on imports from many other countries (Figure 1.10, Panel B). For steel and aluminium, Mexican imports incur duties of 15% of import values, whilst those from Canada incur duties of 32%, close to the 33% incurred by imports from the rest of the world (Figure 1.10, Panel C). These rates will rise as the 50 percentage point tariff on steel and aluminium has been effective since late August 2025.\n\nAs a result of these changes in compliance across sectors, the estimated increase in the aggregate US effective tariff rate on imports from Canada and Mexico has declined. Currently, these are estimated to be 4.8% and 5.4% respectively, down from 12.5% and 14.8% based on developments up to mid-May. Incorporating these two changes, the estimated aggregate effective US tariff rate is approximately 3 percentage points lower than it otherwise would be, reflecting the importance of US trade with Canada and Mexico.\n\n\\_\\_\\_\\_\\_\\_\\_\\_\\_\\_\n\n1\\. These trade agreements also included a number of other commitments, relating to investment and regulation.","pages":[{"line_numbers":[],"markdown":" **21**\n\nOECD ECONOMIC OUTLOOK, VOLUME 2025 ISSUE 2 © OECD 2025\n\nfallen elsewhere (Figure 1.9, Panel B). The updated effective estimates for Canada and Mexico now assume that 80% of goods imported into the United States will be USMCA-compliant.\n\n### **Figure 1.8. Estimated US effective tariff rates**\n\nNote: Estimates based on trade policy announcements as of 14 November 2025 and calculated using weights based on product level data for US imports by country in 2024. These take account of country-specific tariff rates and sector-specific changes due to a 25% tariff rise for listed cars, parts, trucks and buses; a 50% tariff rise on steel, aluminium and copper products; and zero-rated general goods as well as those related to semiconductors. Europe, Japan and Korea face a maximum 15% tariff on cars and parts, and timber, and 0% tariffs are applied to civil aviation products, generic pharmaceuticals and certain natural resources). The United Kingdom faces a 25% tariff rise on steel and aluminium, 7.5% for British cars within the annual quota, 10% on timber products, and duty-free civil aviation products. China, Brazil and India face additional country- specific tariffs, with varying application across products. Data classifications are based on product-level lists published by the US administration. Source: United States International Trade Commission; US Census Bureau, and OECD calculations.\n\nStatLink 2 https://stat.link/qxnj8w\n\nThe share of duty-free imports is generally higher for imports from Canada and Mexico than elsewhere across a range of goods (Figure 1.10, Panel A). Taking the share of duty-free goods as a proxy for USMCA compliance, around 60% and 53% of car imports for Mexico and Canada respectively are USMCA compliant. For steel and aluminium, however, Canada has a compliance rate of just 33% compared to 66% for Mexico. For goods excluding imports of cars and parts, steel and aluminium, and items exempted from tariffs, compliance rates are 93% and 90% for Canada and Mexico, respectively.\n\nOverall, these compliance rates result in US imports of cars and parts from Canada and Mexico currently incurring duties of 12% and 11% respectively, slightly less than half the 27% being charged on imports from many other countries (Figure 1.10, Panel B). For steel and aluminium, Mexican imports incur duties of 15% of import values, whilst those from Canada incur duties of 32%, close to the 33% incurred by imports from the rest of the world (Figure 1.10, Panel C). These rates will rise as the 50 percentage point tariff on steel and aluminium has been effective since late August 2025.\n\nAs a result of these changes in compliance across sectors, the estimated increase in the aggregate US effective tariff rate on imports from Canada and Mexico has declined. Currently, these are estimated to be 4.8% and 5.4% respectively, down from 12.5% and 14.8% based on developments up to mid-May. Incorporating these two changes, the estimated aggregate effective US tariff rate is approximately 3 percentage points lower than it otherwise would be, reflecting the importance of US trade with Canada and Mexico.\n\n\\_\\_\\_\\_\\_\\_\\_\\_\\_\\_\n\n1\\. These trade agreements also included a number of other commitments, relating to investment and regulation.","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":" **21**\n\nOECD ECONOMIC OUTLOOK, VOLUME 2025 ISSUE 2 © OECD 2025\n\nfallen elsewhere (Figure 1.9, Panel B). The updated effective estimates for Canada and Mexico now assume that 80% of goods imported into the United States will be USMCA-compliant.\n\n### **Figure 1.8. Estimated US effective tariff rates**\n\nNote: Estimates based on trade policy announcements as of 14 November 2025 and calculated using weights based on product level data for US imports by country in 2024. These take account of country-specific tariff rates and sector-specific changes due to a 25% tariff rise for listed cars, parts, trucks and buses; a 50% tariff rise on steel, aluminium and copper products; and zero-rated general goods as well as those related to semiconductors. Europe, Japan and Korea face a maximum 15% tariff on cars and parts, and timber, and 0% tariffs are applied to civil aviation products, generic pharmaceuticals and certain natural resources). The United Kingdom faces a 25% tariff rise on steel and aluminium, 7.5% for British cars within the annual quota, 10% on timber products, and duty-free civil aviation products. China, Brazil and India face additional country- specific tariffs, with varying application across products. Data classifications are based on product-level lists published by the US administration. Source: United States International Trade Commission; US Census Bureau, and OECD calculations.\n\nStatLink 2 https://stat.link/qxnj8w\n\nThe share of duty-free imports is generally higher for imports from Canada and Mexico than elsewhere across a range of goods (Figure 1.10, Panel A). Taking the share of duty-free goods as a proxy for USMCA compliance, around 60% and 53% of car imports for Mexico and Canada respectively are USMCA compliant. For steel and aluminium, however, Canada has a compliance rate of just 33% compared to 66% for Mexico. For goods excluding imports of cars and parts, steel and aluminium, and items exempted from tariffs, compliance rates are 93% and 90% for Canada and Mexico, respectively.\n\nOverall, these compliance rates result in US imports of cars and parts from Canada and Mexico currently incurring duties of 12% and 11% respectively, slightly less than half the 27% being charged on imports from many other countries (Figure 1.10, Panel B). For steel and aluminium, Mexican imports incur duties of 15% of import values, whilst those from Canada incur duties of 32%, close to the 33% incurred by imports from the rest of the world (Figure 1.10, Panel C). These rates will rise as the 50 percentage point tariff on steel and aluminium has been effective since late August 2025.\n\nAs a result of these changes in compliance across sectors, the estimated increase in the aggregate US effective tariff rate on imports from Canada and Mexico has declined. Currently, these are estimated to be 4.8% and 5.4% respectively, down from 12.5% and 14.8% based on developments up to mid-May. Incorporating these two changes, the estimated aggregate effective US tariff rate is approximately 3 percentage points lower than it otherwise would be, reflecting the importance of US trade with Canada and Mexico.\n\n\\_\\_\\_\\_\\_\\_\\_\\_\\_\\_\n\n1\\. These trade agreements also included a number of other commitments, relating to investment and regulation.","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/9f653ca1-en_p23.pdf"]},"response":{"errors":[],"extract_id":"extract_093496195c1c79994b31f8276d55d7e6","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":[" **21**\n\nOECD ECONOMIC OUTLOOK, VOLUME 2025 ISSUE 2 © OECD 2025\n\nfallen elsewhere (Figure 1.9, Panel B). The updated effective estimates for Canada and Mexico now assume that 80% of goods imported into the United States will be USMCA-compliant.\n\n### **Figure 1.8. Estimated US effective tariff rates**\n\nNote: Estimates based on trade policy announcements as of 14 November 2025 and calculated using weights based on product level data for US imports by country in 2024. These take account of country-specific tariff rates and sector-specific changes due to a 25% tariff rise for listed cars, parts, trucks and buses; a 50% tariff rise on steel, aluminium and copper products; and zero-rated general goods as well as those related to semiconductors. Europe, Japan and Korea face a maximum 15% tariff on cars and parts, and timber, and 0% tariffs are applied to civil aviation products, generic pharmaceuticals and certain natural resources). The United Kingdom faces a 25% tariff rise on steel and aluminium, 7.5% for British cars within the annual quota, 10% on timber products, and duty-free civil aviation products. China, Brazil and India face additional country- specific tariffs, with varying application across products. Data classifications are based on product-level lists published by the US administration. Source: United States International Trade Commission; US Census Bureau, and OECD calculations.\n\nStatLink 2 https://stat.link/qxnj8w\n\nThe share of duty-free imports is generally higher for imports from Canada and Mexico than elsewhere across a range of goods (Figure 1.10, Panel A). Taking the share of duty-free goods as a proxy for USMCA compliance, around 60% and 53% of car imports for Mexico and Canada respectively are USMCA compliant. For steel and aluminium, however, Canada has a compliance rate of just 33% compared to 66% for Mexico. For goods excluding imports of cars and parts, steel and aluminium, and items exempted from tariffs, compliance rates are 93% and 90% for Canada and Mexico, respectively.\n\nOverall, these compliance rates result in US imports of cars and parts from Canada and Mexico currently incurring duties of 12% and 11% respectively, slightly less than half the 27% being charged on imports from many other countries (Figure 1.10, Panel B). For steel and aluminium, Mexican imports incur duties of 15% of import values, whilst those from Canada incur duties of 32%, close to the 33% incurred by imports from the rest of the world (Figure 1.10, Panel C). These rates will rise as the 50 percentage point tariff on steel and aluminium has been effective since late August 2025.\n\nAs a result of these changes in compliance across sectors, the estimated increase in the aggregate US effective tariff rate on imports from Canada and Mexico has declined. Currently, these are estimated to be 4.8% and 5.4% respectively, down from 12.5% and 14.8% based on developments up to mid-May. Incorporating these two changes, the estimated aggregate effective US tariff rate is approximately 3 percentage points lower than it otherwise would be, reflecting the importance of US trade with Canada and Mexico.\n\n\\_\\_\\_\\_\\_\\_\\_\\_\\_\\_\n\n1\\. These trade agreements also included a number of other commitments, relating to investment and regulation."],"full_content":" **21**\n\nOECD ECONOMIC OUTLOOK, VOLUME 2025 ISSUE 2 © OECD 2025\n\nfallen elsewhere (Figure 1.9, Panel B). The updated effective estimates for Canada and Mexico now assume that 80% of goods imported into the United States will be USMCA-compliant.\n\n### **Figure 1.8. Estimated US effective tariff rates**\n\nNote: Estimates based on trade policy announcements as of 14 November 2025 and calculated using weights based on product level data for US imports by country in 2024. These take account of country-specific tariff rates and sector-specific changes due to a 25% tariff rise for listed cars, parts, trucks and buses; a 50% tariff rise on steel, aluminium and copper products; and zero-rated general goods as well as those related to semiconductors. Europe, Japan and Korea face a maximum 15% tariff on cars and parts, and timber, and 0% tariffs are applied to civil aviation products, generic pharmaceuticals and certain natural resources). The United Kingdom faces a 25% tariff rise on steel and aluminium, 7.5% for British cars within the annual quota, 10% on timber products, and duty-free civil aviation products. China, Brazil and India face additional country- specific tariffs, with varying application across products. Data classifications are based on product-level lists published by the US administration. Source: United States International Trade Commission; US Census Bureau, and OECD calculations.\n\nStatLink 2 https://stat.link/qxnj8w\n\nThe share of duty-free imports is generally higher for imports from Canada and Mexico than elsewhere across a range of goods (Figure 1.10, Panel A). Taking the share of duty-free goods as a proxy for USMCA compliance, around 60% and 53% of car imports for Mexico and Canada respectively are USMCA compliant. For steel and aluminium, however, Canada has a compliance rate of just 33% compared to 66% for Mexico. For goods excluding imports of cars and parts, steel and aluminium, and items exempted from tariffs, compliance rates are 93% and 90% for Canada and Mexico, respectively.\n\nOverall, these compliance rates result in US imports of cars and parts from Canada and Mexico currently incurring duties of 12% and 11% respectively, slightly less than half the 27% being charged on imports from many other countries (Figure 1.10, Panel B). For steel and aluminium, Mexican imports incur duties of 15% of import values, whilst those from Canada incur duties of 32%, close to the 33% incurred by imports from the rest of the world (Figure 1.10, Panel C). These rates will rise as the 50 percentage point tariff on steel and aluminium has been effective since late August 2025.\n\nAs a result of these changes in compliance across sectors, the estimated increase in the aggregate US effective tariff rate on imports from Canada and Mexico has declined. Currently, these are estimated to be 4.8% and 5.4% respectively, down from 12.5% and 14.8% based on developments up to mid-May. Incorporating these two changes, the estimated aggregate effective US tariff rate is approximately 3 percentage points lower than it otherwise would be, reflecting the importance of US trade with Canada and Mexico.\n\n\\_\\_\\_\\_\\_\\_\\_\\_\\_\\_\n\n1\\. These trade agreements also included a number of other commitments, relating to investment and regulation.","publish_date":null,"title":"Figure 1.8. Estimated US effective tariff rates","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/9f653ca1-en_p23.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/9f653ca1-en_p23.pdf"},"request":{"example_id":"chart/9f653ca1-en_p23","product_type":"parse","source_file_path":"9f653ca1-en_p23.pdf"},"started_at":"2026-11-04T19:46:13.726946"} +{"completed_at":"2026-11-04T19:46:19.021879","latency_in_ms":3772,"output":{"example_id":"chart/9f653ca1-en_p28","grounded_pages":[],"layout_pages":[],"markdown":"**26** \n\nOECD ECONOMIC OUTLOOK, VOLUME 2025 ISSUE 2 © OECD 2025\n\n### **Figure 1.14. Underlying inflation and short-term inflation expectations have risen in some economies**\n\nNote: In Panel A, the latest data are for October 2025, except for the United States (August 2025). Trimmed mean inflation is the recomputed overall inflation rate after removing 10% of the consumer basket at the top and at the bottom of the distribution of inflation rates so that the most extreme price changes are excluded. In Panel B, the most recent monthly data are for November for Korea, New Zealand, the United Kingdom and the United States, October 2025 for Mexico, and September 2025 for other countries, except Australia and Canada (third quarter of 2025). Source: Banco de Mexico; Bank of Canada; Bank of Japan; Bank of Korea; Bureau of Economic Analysis; European Central Bank; Eurostat; Federal Reserve Bank of Atlanta; Korean Statistical Information Service; Ministry of Statistics and Program Implementation; Office of National Statistics; Reserve Bank of Australia; Statistics of Japan; Turkish Statistical Institute; University of Michigan; YouGov/Citigroup; and OECD calculations.\n\nStatLink 2 https://stat.link/0bjrsm\n\n**_Financial market conditions have improved_**\n\nFinancial conditions have eased considerably since April in both advanced and emerging-market economies, with financial market volatility and stress indicators generally remaining low, strong equity returns, compressed credit spreads and the depreciation of the US dollar against most currencies this year. However, sovereign yield curves have steepened considerably at the long end, suggesting growing concerns about fiscal risks. Soaring gold prices and large-scale hedging of US dollar-denominated exposures also point to continued uncertainty.\n\nEquity markets have risen strongly, particularly in the United States, where price-earnings valuations are approaching levels last seen in the late 1990s at the peak of the dot-com bubble (Figure 1.15, Panel A). Equity risk premia have also continued to decline, reaching levels last observed in 2005–06 prior to the global financial crisis, highlighting strong risk appetite. Despite some sell-offs in recent weeks, technology stocks remain the main driver of the surge in equity prices, reflecting investor enthusiasm for the potential gains from AI and stronger expected earnings growth than in other sectors and jurisdictions (Figure 1.15, Panel B). After soaring for most of this year, the market value of crypto-assets has declined sharply, by around 30% from the peak reached in early October. In the United States domestically focused US companies, often reliant on imported inputs, face higher costs and have continued to see more limited gains than other companies (Roulet and Tatomir, 2025). Japanese equities have also performed strongly, supported by robust earnings prospects, but equity prices have risen more slowly in Europe, reflecting softer expected earnings growth. Equity performance has remained weaker for firms heavily exposed to US tariffs in these jurisdictions.","pages":[{"line_numbers":[],"markdown":"**26** \n\nOECD ECONOMIC OUTLOOK, VOLUME 2025 ISSUE 2 © OECD 2025\n\n### **Figure 1.14. Underlying inflation and short-term inflation expectations have risen in some economies**\n\nNote: In Panel A, the latest data are for October 2025, except for the United States (August 2025). Trimmed mean inflation is the recomputed overall inflation rate after removing 10% of the consumer basket at the top and at the bottom of the distribution of inflation rates so that the most extreme price changes are excluded. In Panel B, the most recent monthly data are for November for Korea, New Zealand, the United Kingdom and the United States, October 2025 for Mexico, and September 2025 for other countries, except Australia and Canada (third quarter of 2025). Source: Banco de Mexico; Bank of Canada; Bank of Japan; Bank of Korea; Bureau of Economic Analysis; European Central Bank; Eurostat; Federal Reserve Bank of Atlanta; Korean Statistical Information Service; Ministry of Statistics and Program Implementation; Office of National Statistics; Reserve Bank of Australia; Statistics of Japan; Turkish Statistical Institute; University of Michigan; YouGov/Citigroup; and OECD calculations.\n\nStatLink 2 https://stat.link/0bjrsm\n\n**_Financial market conditions have improved_**\n\nFinancial conditions have eased considerably since April in both advanced and emerging-market economies, with financial market volatility and stress indicators generally remaining low, strong equity returns, compressed credit spreads and the depreciation of the US dollar against most currencies this year. However, sovereign yield curves have steepened considerably at the long end, suggesting growing concerns about fiscal risks. Soaring gold prices and large-scale hedging of US dollar-denominated exposures also point to continued uncertainty.\n\nEquity markets have risen strongly, particularly in the United States, where price-earnings valuations are approaching levels last seen in the late 1990s at the peak of the dot-com bubble (Figure 1.15, Panel A). Equity risk premia have also continued to decline, reaching levels last observed in 2005–06 prior to the global financial crisis, highlighting strong risk appetite. Despite some sell-offs in recent weeks, technology stocks remain the main driver of the surge in equity prices, reflecting investor enthusiasm for the potential gains from AI and stronger expected earnings growth than in other sectors and jurisdictions (Figure 1.15, Panel B). After soaring for most of this year, the market value of crypto-assets has declined sharply, by around 30% from the peak reached in early October. In the United States domestically focused US companies, often reliant on imported inputs, face higher costs and have continued to see more limited gains than other companies (Roulet and Tatomir, 2025). Japanese equities have also performed strongly, supported by robust earnings prospects, but equity prices have risen more slowly in Europe, reflecting softer expected earnings growth. Equity performance has remained weaker for firms heavily exposed to US tariffs in these jurisdictions.","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**26** \n\nOECD ECONOMIC OUTLOOK, VOLUME 2025 ISSUE 2 © OECD 2025\n\n### **Figure 1.14. Underlying inflation and short-term inflation expectations have risen in some economies**\n\nNote: In Panel A, the latest data are for October 2025, except for the United States (August 2025). Trimmed mean inflation is the recomputed overall inflation rate after removing 10% of the consumer basket at the top and at the bottom of the distribution of inflation rates so that the most extreme price changes are excluded. In Panel B, the most recent monthly data are for November for Korea, New Zealand, the United Kingdom and the United States, October 2025 for Mexico, and September 2025 for other countries, except Australia and Canada (third quarter of 2025). Source: Banco de Mexico; Bank of Canada; Bank of Japan; Bank of Korea; Bureau of Economic Analysis; European Central Bank; Eurostat; Federal Reserve Bank of Atlanta; Korean Statistical Information Service; Ministry of Statistics and Program Implementation; Office of National Statistics; Reserve Bank of Australia; Statistics of Japan; Turkish Statistical Institute; University of Michigan; YouGov/Citigroup; and OECD calculations.\n\nStatLink 2 https://stat.link/0bjrsm\n\n**_Financial market conditions have improved_**\n\nFinancial conditions have eased considerably since April in both advanced and emerging-market economies, with financial market volatility and stress indicators generally remaining low, strong equity returns, compressed credit spreads and the depreciation of the US dollar against most currencies this year. However, sovereign yield curves have steepened considerably at the long end, suggesting growing concerns about fiscal risks. Soaring gold prices and large-scale hedging of US dollar-denominated exposures also point to continued uncertainty.\n\nEquity markets have risen strongly, particularly in the United States, where price-earnings valuations are approaching levels last seen in the late 1990s at the peak of the dot-com bubble (Figure 1.15, Panel A). Equity risk premia have also continued to decline, reaching levels last observed in 2005–06 prior to the global financial crisis, highlighting strong risk appetite. Despite some sell-offs in recent weeks, technology stocks remain the main driver of the surge in equity prices, reflecting investor enthusiasm for the potential gains from AI and stronger expected earnings growth than in other sectors and jurisdictions (Figure 1.15, Panel B). After soaring for most of this year, the market value of crypto-assets has declined sharply, by around 30% from the peak reached in early October. In the United States domestically focused US companies, often reliant on imported inputs, face higher costs and have continued to see more limited gains than other companies (Roulet and Tatomir, 2025). Japanese equities have also performed strongly, supported by robust earnings prospects, but equity prices have risen more slowly in Europe, reflecting softer expected earnings growth. Equity performance has remained weaker for firms heavily exposed to US tariffs in these jurisdictions.","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/9f653ca1-en_p28.pdf"]},"response":{"errors":[],"extract_id":"extract_8dac85f13dab30cada7e849dbf061818","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**26** \n\nOECD ECONOMIC OUTLOOK, VOLUME 2025 ISSUE 2 © OECD 2025\n\n### **Figure 1.14. Underlying inflation and short-term inflation expectations have risen in some economies**\n\nNote: In Panel A, the latest data are for October 2025, except for the United States (August 2025). Trimmed mean inflation is the recomputed overall inflation rate after removing 10% of the consumer basket at the top and at the bottom of the distribution of inflation rates so that the most extreme price changes are excluded. In Panel B, the most recent monthly data are for November for Korea, New Zealand, the United Kingdom and the United States, October 2025 for Mexico, and September 2025 for other countries, except Australia and Canada (third quarter of 2025). Source: Banco de Mexico; Bank of Canada; Bank of Japan; Bank of Korea; Bureau of Economic Analysis; European Central Bank; Eurostat; Federal Reserve Bank of Atlanta; Korean Statistical Information Service; Ministry of Statistics and Program Implementation; Office of National Statistics; Reserve Bank of Australia; Statistics of Japan; Turkish Statistical Institute; University of Michigan; YouGov/Citigroup; and OECD calculations.\n\nStatLink 2 https://stat.link/0bjrsm\n\n**_Financial market conditions have improved_**\n\nFinancial conditions have eased considerably since April in both advanced and emerging-market economies, with financial market volatility and stress indicators generally remaining low, strong equity returns, compressed credit spreads and the depreciation of the US dollar against most currencies this year. However, sovereign yield curves have steepened considerably at the long end, suggesting growing concerns about fiscal risks. Soaring gold prices and large-scale hedging of US dollar-denominated exposures also point to continued uncertainty.\n\nEquity markets have risen strongly, particularly in the United States, where price-earnings valuations are approaching levels last seen in the late 1990s at the peak of the dot-com bubble (Figure 1.15, Panel A). Equity risk premia have also continued to decline, reaching levels last observed in 2005–06 prior to the global financial crisis, highlighting strong risk appetite. Despite some sell-offs in recent weeks, technology stocks remain the main driver of the surge in equity prices, reflecting investor enthusiasm for the potential gains from AI and stronger expected earnings growth than in other sectors and jurisdictions (Figure 1.15, Panel B). After soaring for most of this year, the market value of crypto-assets has declined sharply, by around 30% from the peak reached in early October. In the United States domestically focused US companies, often reliant on imported inputs, face higher costs and have continued to see more limited gains than other companies (Roulet and Tatomir, 2025). Japanese equities have also performed strongly, supported by robust earnings prospects, but equity prices have risen more slowly in Europe, reflecting softer expected earnings growth. Equity performance has remained weaker for firms heavily exposed to US tariffs in these jurisdictions."],"full_content":"**26** \n\nOECD ECONOMIC OUTLOOK, VOLUME 2025 ISSUE 2 © OECD 2025\n\n### **Figure 1.14. Underlying inflation and short-term inflation expectations have risen in some economies**\n\nNote: In Panel A, the latest data are for October 2025, except for the United States (August 2025). Trimmed mean inflation is the recomputed overall inflation rate after removing 10% of the consumer basket at the top and at the bottom of the distribution of inflation rates so that the most extreme price changes are excluded. In Panel B, the most recent monthly data are for November for Korea, New Zealand, the United Kingdom and the United States, October 2025 for Mexico, and September 2025 for other countries, except Australia and Canada (third quarter of 2025). Source: Banco de Mexico; Bank of Canada; Bank of Japan; Bank of Korea; Bureau of Economic Analysis; European Central Bank; Eurostat; Federal Reserve Bank of Atlanta; Korean Statistical Information Service; Ministry of Statistics and Program Implementation; Office of National Statistics; Reserve Bank of Australia; Statistics of Japan; Turkish Statistical Institute; University of Michigan; YouGov/Citigroup; and OECD calculations.\n\nStatLink 2 https://stat.link/0bjrsm\n\n**_Financial market conditions have improved_**\n\nFinancial conditions have eased considerably since April in both advanced and emerging-market economies, with financial market volatility and stress indicators generally remaining low, strong equity returns, compressed credit spreads and the depreciation of the US dollar against most currencies this year. However, sovereign yield curves have steepened considerably at the long end, suggesting growing concerns about fiscal risks. Soaring gold prices and large-scale hedging of US dollar-denominated exposures also point to continued uncertainty.\n\nEquity markets have risen strongly, particularly in the United States, where price-earnings valuations are approaching levels last seen in the late 1990s at the peak of the dot-com bubble (Figure 1.15, Panel A). Equity risk premia have also continued to decline, reaching levels last observed in 2005–06 prior to the global financial crisis, highlighting strong risk appetite. Despite some sell-offs in recent weeks, technology stocks remain the main driver of the surge in equity prices, reflecting investor enthusiasm for the potential gains from AI and stronger expected earnings growth than in other sectors and jurisdictions (Figure 1.15, Panel B). After soaring for most of this year, the market value of crypto-assets has declined sharply, by around 30% from the peak reached in early October. In the United States domestically focused US companies, often reliant on imported inputs, face higher costs and have continued to see more limited gains than other companies (Roulet and Tatomir, 2025). Japanese equities have also performed strongly, supported by robust earnings prospects, but equity prices have risen more slowly in Europe, reflecting softer expected earnings growth. Equity performance has remained weaker for firms heavily exposed to US tariffs in these jurisdictions.","publish_date":null,"title":"Figure 1.14. Underlying inflation and short-term inflation expectations have risen in some economies","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/9f653ca1-en_p28.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/9f653ca1-en_p28.pdf"},"request":{"example_id":"chart/9f653ca1-en_p28","product_type":"parse","source_file_path":"9f653ca1-en_p28.pdf"},"started_at":"2026-11-04T19:46:15.249430"} +{"completed_at":"2026-11-04T19:46:21.036078","latency_in_ms":4377,"output":{"example_id":"chart/9f653ca1-en_p29","grounded_pages":[],"layout_pages":[],"markdown":" **27**\n\nOECD ECONOMIC OUTLOOK, VOLUME 2025 ISSUE 2 © OECD 2025\n\n### **Figure 1.15. Elevated equity valuations are supported by the higher expected earnings of technology companies**\n\nNote: Panel A shows monthly values of the cyclically adjusted price-to-earnings ratio (CAPE), computed with 10-year average earnings adjusted for subsequent changes in consumer prices. The value for November 2025 is based on observations up to 19 November. Source: LSEG; and OECD calculations.\n\nSpreads on high-yield corporate bonds remain close to historical lows, despite a recent uptick, and investment-grade bond yields have declined in recent months. Low credit risk premia and sustained portfolio inflows in the United States and the euro area have helped to support strong corporate bond and leveraged loan issuance during the first three quarters of 2025 (Figure 1.16, Panel A). Bank credit growth has also been recovering gradually in major advanced economies, albeit from low levels, supported by an ongoing decline in lending rates (Figure 1.16, Panel B).\n\nEmerging-market economies have benefitted from higher global risk appetite. A weaker US dollar and declining exchange rate volatility have made emerging-market assets relatively more attractive. This has been accompanied by strong growth in equity prices, partly driven by a substantial rise of AI-related equities, with the main exception of India, where the stock market has been broadly flat, possibly reflecting the marked rise in the US tariff rate on imports of goods from India. Spreads on corporate bonds issued by emerging-market companies have continued to fall, reaching their lowest levels since 2007. In the sovereign bond markets of most emerging-market economies, long-term yields have broadly eased since June and spreads on US dollar-denominated bonds have declined.\n\nIn contrast, developments in sovereign debt markets in advanced economies have been less favourable. Ten-year yields have edged up in recent months in many jurisdictions and 30-year yields have increased by larger amounts, steepening the long end of the yield curve. Amid ongoing quantitative tightening and large net government debt issuance needs in the near future, that steepening could reflect growing concerns about long-run fiscal sustainability. In the United States, the 10-year sovereign yield has been underpinned by a persistently high estimated real term premium. In France, mounting political challenges have clouded the fiscal outlook, markedly widening French-German sovereign spreads since mid-August.","pages":[{"line_numbers":[],"markdown":" **27**\n\nOECD ECONOMIC OUTLOOK, VOLUME 2025 ISSUE 2 © OECD 2025\n\n### **Figure 1.15. Elevated equity valuations are supported by the higher expected earnings of technology companies**\n\nNote: Panel A shows monthly values of the cyclically adjusted price-to-earnings ratio (CAPE), computed with 10-year average earnings adjusted for subsequent changes in consumer prices. The value for November 2025 is based on observations up to 19 November. Source: LSEG; and OECD calculations.\n\nSpreads on high-yield corporate bonds remain close to historical lows, despite a recent uptick, and investment-grade bond yields have declined in recent months. Low credit risk premia and sustained portfolio inflows in the United States and the euro area have helped to support strong corporate bond and leveraged loan issuance during the first three quarters of 2025 (Figure 1.16, Panel A). Bank credit growth has also been recovering gradually in major advanced economies, albeit from low levels, supported by an ongoing decline in lending rates (Figure 1.16, Panel B).\n\nEmerging-market economies have benefitted from higher global risk appetite. A weaker US dollar and declining exchange rate volatility have made emerging-market assets relatively more attractive. This has been accompanied by strong growth in equity prices, partly driven by a substantial rise of AI-related equities, with the main exception of India, where the stock market has been broadly flat, possibly reflecting the marked rise in the US tariff rate on imports of goods from India. Spreads on corporate bonds issued by emerging-market companies have continued to fall, reaching their lowest levels since 2007. In the sovereign bond markets of most emerging-market economies, long-term yields have broadly eased since June and spreads on US dollar-denominated bonds have declined.\n\nIn contrast, developments in sovereign debt markets in advanced economies have been less favourable. Ten-year yields have edged up in recent months in many jurisdictions and 30-year yields have increased by larger amounts, steepening the long end of the yield curve. Amid ongoing quantitative tightening and large net government debt issuance needs in the near future, that steepening could reflect growing concerns about long-run fiscal sustainability. In the United States, the 10-year sovereign yield has been underpinned by a persistently high estimated real term premium. In France, mounting political challenges have clouded the fiscal outlook, markedly widening French-German sovereign spreads since mid-August.","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":" **27**\n\nOECD ECONOMIC OUTLOOK, VOLUME 2025 ISSUE 2 © OECD 2025\n\n### **Figure 1.15. Elevated equity valuations are supported by the higher expected earnings of technology companies**\n\nNote: Panel A shows monthly values of the cyclically adjusted price-to-earnings ratio (CAPE), computed with 10-year average earnings adjusted for subsequent changes in consumer prices. The value for November 2025 is based on observations up to 19 November. Source: LSEG; and OECD calculations.\n\nSpreads on high-yield corporate bonds remain close to historical lows, despite a recent uptick, and investment-grade bond yields have declined in recent months. Low credit risk premia and sustained portfolio inflows in the United States and the euro area have helped to support strong corporate bond and leveraged loan issuance during the first three quarters of 2025 (Figure 1.16, Panel A). Bank credit growth has also been recovering gradually in major advanced economies, albeit from low levels, supported by an ongoing decline in lending rates (Figure 1.16, Panel B).\n\nEmerging-market economies have benefitted from higher global risk appetite. A weaker US dollar and declining exchange rate volatility have made emerging-market assets relatively more attractive. This has been accompanied by strong growth in equity prices, partly driven by a substantial rise of AI-related equities, with the main exception of India, where the stock market has been broadly flat, possibly reflecting the marked rise in the US tariff rate on imports of goods from India. Spreads on corporate bonds issued by emerging-market companies have continued to fall, reaching their lowest levels since 2007. In the sovereign bond markets of most emerging-market economies, long-term yields have broadly eased since June and spreads on US dollar-denominated bonds have declined.\n\nIn contrast, developments in sovereign debt markets in advanced economies have been less favourable. Ten-year yields have edged up in recent months in many jurisdictions and 30-year yields have increased by larger amounts, steepening the long end of the yield curve. Amid ongoing quantitative tightening and large net government debt issuance needs in the near future, that steepening could reflect growing concerns about long-run fiscal sustainability. In the United States, the 10-year sovereign yield has been underpinned by a persistently high estimated real term premium. In France, mounting political challenges have clouded the fiscal outlook, markedly widening French-German sovereign spreads since mid-August.","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/9f653ca1-en_p29.pdf"]},"response":{"errors":[],"extract_id":"extract_2f656fbf2fbd4eb9a367cbd445ab048a","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":[" **27**\n\nOECD ECONOMIC OUTLOOK, VOLUME 2025 ISSUE 2 © OECD 2025\n\n### **Figure 1.15. Elevated equity valuations are supported by the higher expected earnings of technology companies**\n\nNote: Panel A shows monthly values of the cyclically adjusted price-to-earnings ratio (CAPE), computed with 10-year average earnings adjusted for subsequent changes in consumer prices. The value for November 2025 is based on observations up to 19 November. Source: LSEG; and OECD calculations.\n\nSpreads on high-yield corporate bonds remain close to historical lows, despite a recent uptick, and investment-grade bond yields have declined in recent months. Low credit risk premia and sustained portfolio inflows in the United States and the euro area have helped to support strong corporate bond and leveraged loan issuance during the first three quarters of 2025 (Figure 1.16, Panel A). Bank credit growth has also been recovering gradually in major advanced economies, albeit from low levels, supported by an ongoing decline in lending rates (Figure 1.16, Panel B).\n\nEmerging-market economies have benefitted from higher global risk appetite. A weaker US dollar and declining exchange rate volatility have made emerging-market assets relatively more attractive. This has been accompanied by strong growth in equity prices, partly driven by a substantial rise of AI-related equities, with the main exception of India, where the stock market has been broadly flat, possibly reflecting the marked rise in the US tariff rate on imports of goods from India. Spreads on corporate bonds issued by emerging-market companies have continued to fall, reaching their lowest levels since 2007. In the sovereign bond markets of most emerging-market economies, long-term yields have broadly eased since June and spreads on US dollar-denominated bonds have declined.\n\nIn contrast, developments in sovereign debt markets in advanced economies have been less favourable. Ten-year yields have edged up in recent months in many jurisdictions and 30-year yields have increased by larger amounts, steepening the long end of the yield curve. Amid ongoing quantitative tightening and large net government debt issuance needs in the near future, that steepening could reflect growing concerns about long-run fiscal sustainability. In the United States, the 10-year sovereign yield has been underpinned by a persistently high estimated real term premium. In France, mounting political challenges have clouded the fiscal outlook, markedly widening French-German sovereign spreads since mid-August."],"full_content":" **27**\n\nOECD ECONOMIC OUTLOOK, VOLUME 2025 ISSUE 2 © OECD 2025\n\n### **Figure 1.15. Elevated equity valuations are supported by the higher expected earnings of technology companies**\n\nNote: Panel A shows monthly values of the cyclically adjusted price-to-earnings ratio (CAPE), computed with 10-year average earnings adjusted for subsequent changes in consumer prices. The value for November 2025 is based on observations up to 19 November. Source: LSEG; and OECD calculations.\n\nSpreads on high-yield corporate bonds remain close to historical lows, despite a recent uptick, and investment-grade bond yields have declined in recent months. Low credit risk premia and sustained portfolio inflows in the United States and the euro area have helped to support strong corporate bond and leveraged loan issuance during the first three quarters of 2025 (Figure 1.16, Panel A). Bank credit growth has also been recovering gradually in major advanced economies, albeit from low levels, supported by an ongoing decline in lending rates (Figure 1.16, Panel B).\n\nEmerging-market economies have benefitted from higher global risk appetite. A weaker US dollar and declining exchange rate volatility have made emerging-market assets relatively more attractive. This has been accompanied by strong growth in equity prices, partly driven by a substantial rise of AI-related equities, with the main exception of India, where the stock market has been broadly flat, possibly reflecting the marked rise in the US tariff rate on imports of goods from India. Spreads on corporate bonds issued by emerging-market companies have continued to fall, reaching their lowest levels since 2007. In the sovereign bond markets of most emerging-market economies, long-term yields have broadly eased since June and spreads on US dollar-denominated bonds have declined.\n\nIn contrast, developments in sovereign debt markets in advanced economies have been less favourable. Ten-year yields have edged up in recent months in many jurisdictions and 30-year yields have increased by larger amounts, steepening the long end of the yield curve. Amid ongoing quantitative tightening and large net government debt issuance needs in the near future, that steepening could reflect growing concerns about long-run fiscal sustainability. In the United States, the 10-year sovereign yield has been underpinned by a persistently high estimated real term premium. In France, mounting political challenges have clouded the fiscal outlook, markedly widening French-German sovereign spreads since mid-August.","publish_date":null,"title":"Figure 1.15. Elevated equity valuations are supported by the higher expected earnings of technology companies","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/9f653ca1-en_p29.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/9f653ca1-en_p29.pdf"},"request":{"example_id":"chart/9f653ca1-en_p29","product_type":"parse","source_file_path":"9f653ca1-en_p29.pdf"},"started_at":"2026-11-04T19:46:16.658921"} +{"completed_at":"2026-11-04T19:46:23.298590","latency_in_ms":4335,"output":{"example_id":"chart/9f653ca1-en_p31","grounded_pages":[],"layout_pages":[],"markdown":" **29**\n\nOECD ECONOMIC OUTLOOK, VOLUME 2025 ISSUE 2 © OECD 2025\n\n### **Figure 1.17. Global growth is projected to weaken before recovering gradually**\n\nNote: In Panel A, ‘Advanced economies’ include the OECD member countries except Chile, Colombia, Costa Rica, Mexico, and Türkiye, plus Croatia. ‘Emerging-market economies’ comprises all other economies. In Panel B, Emerging Asia comprises China, Hong Kong (China), India, Indonesia, Malaysia, the Philippines, Singapore, Chinese Taipei, Thailand and Viet Nam. Latin America comprises Argentina, Brazil, Chile, Colombia, Costa Rica, Mexico and Peru. Contributions calculated using moving PPP shares of global GDP. Source: OECD Economic Outlook 118 database; OECD Economic Outlook 117 database; OECD Interim Economic Outlook 118 database; and OECD calculations.\n\nStatLink 2 https://stat.link/j3hlps\n\nThe near-term growth projections are close to those in the September 2025 OECD Interim Economic Outlook but stronger than those in the June 2025 OECD Economic Outlook (Figure 1.17, Panel C). The change since June reflects several factors, including easier global financial conditions, evidence of stronger investment momentum in some OECD economies and more supportive macroeconomic policies than anticipated, with policy interest rates continuing to be lowered in many countries, and little fiscal tightening in the median OECD economy this year and next. Fiscal policies are also providing support for domestic demand in China, India (via the goods and services tax reform) and Indonesia. At the same time, inflation projections for G20 countries have been revised downwards to incorporate weaker-than-expected inflation outturns, most notably for the United States. Inflation in the G20 countries is anticipated to further moderate through 2026, with the inflationary impact from higher trade costs generally offset by softer growth (Figure 1.17, Panel D).","pages":[{"line_numbers":[],"markdown":" **29**\n\nOECD ECONOMIC OUTLOOK, VOLUME 2025 ISSUE 2 © OECD 2025\n\n### **Figure 1.17. Global growth is projected to weaken before recovering gradually**\n\nNote: In Panel A, ‘Advanced economies’ include the OECD member countries except Chile, Colombia, Costa Rica, Mexico, and Türkiye, plus Croatia. ‘Emerging-market economies’ comprises all other economies. In Panel B, Emerging Asia comprises China, Hong Kong (China), India, Indonesia, Malaysia, the Philippines, Singapore, Chinese Taipei, Thailand and Viet Nam. Latin America comprises Argentina, Brazil, Chile, Colombia, Costa Rica, Mexico and Peru. Contributions calculated using moving PPP shares of global GDP. Source: OECD Economic Outlook 118 database; OECD Economic Outlook 117 database; OECD Interim Economic Outlook 118 database; and OECD calculations.\n\nStatLink 2 https://stat.link/j3hlps\n\nThe near-term growth projections are close to those in the September 2025 OECD Interim Economic Outlook but stronger than those in the June 2025 OECD Economic Outlook (Figure 1.17, Panel C). The change since June reflects several factors, including easier global financial conditions, evidence of stronger investment momentum in some OECD economies and more supportive macroeconomic policies than anticipated, with policy interest rates continuing to be lowered in many countries, and little fiscal tightening in the median OECD economy this year and next. Fiscal policies are also providing support for domestic demand in China, India (via the goods and services tax reform) and Indonesia. At the same time, inflation projections for G20 countries have been revised downwards to incorporate weaker-than-expected inflation outturns, most notably for the United States. Inflation in the G20 countries is anticipated to further moderate through 2026, with the inflationary impact from higher trade costs generally offset by softer growth (Figure 1.17, Panel D).","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":" **29**\n\nOECD ECONOMIC OUTLOOK, VOLUME 2025 ISSUE 2 © OECD 2025\n\n### **Figure 1.17. Global growth is projected to weaken before recovering gradually**\n\nNote: In Panel A, ‘Advanced economies’ include the OECD member countries except Chile, Colombia, Costa Rica, Mexico, and Türkiye, plus Croatia. ‘Emerging-market economies’ comprises all other economies. In Panel B, Emerging Asia comprises China, Hong Kong (China), India, Indonesia, Malaysia, the Philippines, Singapore, Chinese Taipei, Thailand and Viet Nam. Latin America comprises Argentina, Brazil, Chile, Colombia, Costa Rica, Mexico and Peru. Contributions calculated using moving PPP shares of global GDP. Source: OECD Economic Outlook 118 database; OECD Economic Outlook 117 database; OECD Interim Economic Outlook 118 database; and OECD calculations.\n\nStatLink 2 https://stat.link/j3hlps\n\nThe near-term growth projections are close to those in the September 2025 OECD Interim Economic Outlook but stronger than those in the June 2025 OECD Economic Outlook (Figure 1.17, Panel C). The change since June reflects several factors, including easier global financial conditions, evidence of stronger investment momentum in some OECD economies and more supportive macroeconomic policies than anticipated, with policy interest rates continuing to be lowered in many countries, and little fiscal tightening in the median OECD economy this year and next. Fiscal policies are also providing support for domestic demand in China, India (via the goods and services tax reform) and Indonesia. At the same time, inflation projections for G20 countries have been revised downwards to incorporate weaker-than-expected inflation outturns, most notably for the United States. Inflation in the G20 countries is anticipated to further moderate through 2026, with the inflationary impact from higher trade costs generally offset by softer growth (Figure 1.17, Panel D).","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/9f653ca1-en_p31.pdf"]},"response":{"errors":[],"extract_id":"extract_631cc0f5a75f98f7c2b48061d7d848ab","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":[" **29**\n\nOECD ECONOMIC OUTLOOK, VOLUME 2025 ISSUE 2 © OECD 2025\n\n### **Figure 1.17. Global growth is projected to weaken before recovering gradually**\n\nNote: In Panel A, ‘Advanced economies’ include the OECD member countries except Chile, Colombia, Costa Rica, Mexico, and Türkiye, plus Croatia. ‘Emerging-market economies’ comprises all other economies. In Panel B, Emerging Asia comprises China, Hong Kong (China), India, Indonesia, Malaysia, the Philippines, Singapore, Chinese Taipei, Thailand and Viet Nam. Latin America comprises Argentina, Brazil, Chile, Colombia, Costa Rica, Mexico and Peru. Contributions calculated using moving PPP shares of global GDP. Source: OECD Economic Outlook 118 database; OECD Economic Outlook 117 database; OECD Interim Economic Outlook 118 database; and OECD calculations.\n\nStatLink 2 https://stat.link/j3hlps\n\nThe near-term growth projections are close to those in the September 2025 OECD Interim Economic Outlook but stronger than those in the June 2025 OECD Economic Outlook (Figure 1.17, Panel C). The change since June reflects several factors, including easier global financial conditions, evidence of stronger investment momentum in some OECD economies and more supportive macroeconomic policies than anticipated, with policy interest rates continuing to be lowered in many countries, and little fiscal tightening in the median OECD economy this year and next. Fiscal policies are also providing support for domestic demand in China, India (via the goods and services tax reform) and Indonesia. At the same time, inflation projections for G20 countries have been revised downwards to incorporate weaker-than-expected inflation outturns, most notably for the United States. Inflation in the G20 countries is anticipated to further moderate through 2026, with the inflationary impact from higher trade costs generally offset by softer growth (Figure 1.17, Panel D)."],"full_content":" **29**\n\nOECD ECONOMIC OUTLOOK, VOLUME 2025 ISSUE 2 © OECD 2025\n\n### **Figure 1.17. Global growth is projected to weaken before recovering gradually**\n\nNote: In Panel A, ‘Advanced economies’ include the OECD member countries except Chile, Colombia, Costa Rica, Mexico, and Türkiye, plus Croatia. ‘Emerging-market economies’ comprises all other economies. In Panel B, Emerging Asia comprises China, Hong Kong (China), India, Indonesia, Malaysia, the Philippines, Singapore, Chinese Taipei, Thailand and Viet Nam. Latin America comprises Argentina, Brazil, Chile, Colombia, Costa Rica, Mexico and Peru. Contributions calculated using moving PPP shares of global GDP. Source: OECD Economic Outlook 118 database; OECD Economic Outlook 117 database; OECD Interim Economic Outlook 118 database; and OECD calculations.\n\nStatLink 2 https://stat.link/j3hlps\n\nThe near-term growth projections are close to those in the September 2025 OECD Interim Economic Outlook but stronger than those in the June 2025 OECD Economic Outlook (Figure 1.17, Panel C). The change since June reflects several factors, including easier global financial conditions, evidence of stronger investment momentum in some OECD economies and more supportive macroeconomic policies than anticipated, with policy interest rates continuing to be lowered in many countries, and little fiscal tightening in the median OECD economy this year and next. Fiscal policies are also providing support for domestic demand in China, India (via the goods and services tax reform) and Indonesia. At the same time, inflation projections for G20 countries have been revised downwards to incorporate weaker-than-expected inflation outturns, most notably for the United States. Inflation in the G20 countries is anticipated to further moderate through 2026, with the inflationary impact from higher trade costs generally offset by softer growth (Figure 1.17, Panel D).","publish_date":null,"title":"Figure 1.17. Global growth is projected to weaken before recovering gradually","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/9f653ca1-en_p31.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/9f653ca1-en_p31.pdf"},"request":{"example_id":"chart/9f653ca1-en_p31","product_type":"parse","source_file_path":"9f653ca1-en_p31.pdf"},"started_at":"2026-11-04T19:46:18.963025"} +{"completed_at":"2026-11-04T19:46:23.544159","latency_in_ms":2673,"output":{"example_id":"chart/9f653ca1-en_p33","grounded_pages":[],"layout_pages":[],"markdown":" **31**\n\nOECD ECONOMIC OUTLOOK, VOLUME 2025 ISSUE 2 © OECD 2025\n\n• The Mexican economy is projected to recover slowly as inflation eases and further cuts to policy rates take effect. However, higher tariffs on exports to the United States and weak public demand are headwinds to growth, with real GDP projected to rise by 0.7% in 2025, 1.2% in 2026 and 1.7% in 2027. In Brazil, growth is anticipated to weaken in the near-term as investment slows in response to higher interest rates, though monetary policy easing through 2026 will support activity. After easing from 2.4% in 2025 to 1.7% in 2026, output growth is projected to recover to 2.2% in 2027.\n\nGlobal trade growth for 2025 is now expected to be 4.2%, reflecting stronger than expected front-loading in the first half of the year. This is being unwound only partially given the momentum of investment- and tech-related trade through 2025. Nonetheless, trade growth is projected to moderate to 2.3% in 2026 as the full effects of higher tariffs are felt, before recovering to 2.8% in 2027 as growth picks up.\n\nThe composition of trade growth is different to the pre-pandemic period (Figure 1.18, Panel A), with the emerging-market economies accounting for a larger share, even as all regions grow more slowly. In particular, the share of trade accounted for by the South-East Asian economies is expected to rise in 2026-27. In contrast, the contribution of China is projected to weaken slightly, as trade growth slows, and the contribution of advanced economies to global trade growth is expected to decline, especially that of Europe and the North American economies.\n\n### **Figure 1.18. Trade patterns are evolving with emerging markets becoming a key driver of growth**\n\nNote: North America includes Canada, Mexico and the United States. Europe includes the 22 OECD countries which are members of the European Union, plus Iceland, Norway, Switzerland, Türkiye and the United Kingdom. South-East Asian economies include Hong Kong (China), Malaysia, the Philippines, Singapore, Chinese Taipei, Thailand, and Viet Nam. Other advanced economies include Australia, Israel, Japan, Korea and New Zealand. Statistical discrepancies mean that current account balances based on national data rarely equal zero at the world level. Source: OECD Economic Outlook 118 database; and OECD calculations.\n\nStatLink 2 https://stat.link/mhq1ks\n\nGlobal current account balances are projected to decrease slightly as a share of global GDP over the next two years (Figure 1.18, Panel B). In the United States, lower imports and reduced remittances linked to international migration, and continued strong growth in tech-related exports and services help to reduce the current account deficit from 2024 levels, despite a decline in export market performance. China’s continued export growth, coupled with weak domestic demand and a continued policy orientation towards import substitution is projected to result in a higher external surplus. Sizeable surpluses are also projected to persist in Germany, Japan, Korea and a number of smaller economies, including Denmark, Ireland, the Netherlands, Norway, Sweden, Switzerland and Thailand. There is little change in the current account surplus in the euro area as a whole.","pages":[{"line_numbers":[],"markdown":" **31**\n\nOECD ECONOMIC OUTLOOK, VOLUME 2025 ISSUE 2 © OECD 2025\n\n• The Mexican economy is projected to recover slowly as inflation eases and further cuts to policy rates take effect. However, higher tariffs on exports to the United States and weak public demand are headwinds to growth, with real GDP projected to rise by 0.7% in 2025, 1.2% in 2026 and 1.7% in 2027. In Brazil, growth is anticipated to weaken in the near-term as investment slows in response to higher interest rates, though monetary policy easing through 2026 will support activity. After easing from 2.4% in 2025 to 1.7% in 2026, output growth is projected to recover to 2.2% in 2027.\n\nGlobal trade growth for 2025 is now expected to be 4.2%, reflecting stronger than expected front-loading in the first half of the year. This is being unwound only partially given the momentum of investment- and tech-related trade through 2025. Nonetheless, trade growth is projected to moderate to 2.3% in 2026 as the full effects of higher tariffs are felt, before recovering to 2.8% in 2027 as growth picks up.\n\nThe composition of trade growth is different to the pre-pandemic period (Figure 1.18, Panel A), with the emerging-market economies accounting for a larger share, even as all regions grow more slowly. In particular, the share of trade accounted for by the South-East Asian economies is expected to rise in 2026-27. In contrast, the contribution of China is projected to weaken slightly, as trade growth slows, and the contribution of advanced economies to global trade growth is expected to decline, especially that of Europe and the North American economies.\n\n### **Figure 1.18. Trade patterns are evolving with emerging markets becoming a key driver of growth**\n\nNote: North America includes Canada, Mexico and the United States. Europe includes the 22 OECD countries which are members of the European Union, plus Iceland, Norway, Switzerland, Türkiye and the United Kingdom. South-East Asian economies include Hong Kong (China), Malaysia, the Philippines, Singapore, Chinese Taipei, Thailand, and Viet Nam. Other advanced economies include Australia, Israel, Japan, Korea and New Zealand. Statistical discrepancies mean that current account balances based on national data rarely equal zero at the world level. Source: OECD Economic Outlook 118 database; and OECD calculations.\n\nStatLink 2 https://stat.link/mhq1ks\n\nGlobal current account balances are projected to decrease slightly as a share of global GDP over the next two years (Figure 1.18, Panel B). In the United States, lower imports and reduced remittances linked to international migration, and continued strong growth in tech-related exports and services help to reduce the current account deficit from 2024 levels, despite a decline in export market performance. China’s continued export growth, coupled with weak domestic demand and a continued policy orientation towards import substitution is projected to result in a higher external surplus. Sizeable surpluses are also projected to persist in Germany, Japan, Korea and a number of smaller economies, including Denmark, Ireland, the Netherlands, Norway, Sweden, Switzerland and Thailand. There is little change in the current account surplus in the euro area as a whole.","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":" **31**\n\nOECD ECONOMIC OUTLOOK, VOLUME 2025 ISSUE 2 © OECD 2025\n\n• The Mexican economy is projected to recover slowly as inflation eases and further cuts to policy rates take effect. However, higher tariffs on exports to the United States and weak public demand are headwinds to growth, with real GDP projected to rise by 0.7% in 2025, 1.2% in 2026 and 1.7% in 2027. In Brazil, growth is anticipated to weaken in the near-term as investment slows in response to higher interest rates, though monetary policy easing through 2026 will support activity. After easing from 2.4% in 2025 to 1.7% in 2026, output growth is projected to recover to 2.2% in 2027.\n\nGlobal trade growth for 2025 is now expected to be 4.2%, reflecting stronger than expected front-loading in the first half of the year. This is being unwound only partially given the momentum of investment- and tech-related trade through 2025. Nonetheless, trade growth is projected to moderate to 2.3% in 2026 as the full effects of higher tariffs are felt, before recovering to 2.8% in 2027 as growth picks up.\n\nThe composition of trade growth is different to the pre-pandemic period (Figure 1.18, Panel A), with the emerging-market economies accounting for a larger share, even as all regions grow more slowly. In particular, the share of trade accounted for by the South-East Asian economies is expected to rise in 2026-27. In contrast, the contribution of China is projected to weaken slightly, as trade growth slows, and the contribution of advanced economies to global trade growth is expected to decline, especially that of Europe and the North American economies.\n\n### **Figure 1.18. Trade patterns are evolving with emerging markets becoming a key driver of growth**\n\nNote: North America includes Canada, Mexico and the United States. Europe includes the 22 OECD countries which are members of the European Union, plus Iceland, Norway, Switzerland, Türkiye and the United Kingdom. South-East Asian economies include Hong Kong (China), Malaysia, the Philippines, Singapore, Chinese Taipei, Thailand, and Viet Nam. Other advanced economies include Australia, Israel, Japan, Korea and New Zealand. Statistical discrepancies mean that current account balances based on national data rarely equal zero at the world level. Source: OECD Economic Outlook 118 database; and OECD calculations.\n\nStatLink 2 https://stat.link/mhq1ks\n\nGlobal current account balances are projected to decrease slightly as a share of global GDP over the next two years (Figure 1.18, Panel B). In the United States, lower imports and reduced remittances linked to international migration, and continued strong growth in tech-related exports and services help to reduce the current account deficit from 2024 levels, despite a decline in export market performance. China’s continued export growth, coupled with weak domestic demand and a continued policy orientation towards import substitution is projected to result in a higher external surplus. Sizeable surpluses are also projected to persist in Germany, Japan, Korea and a number of smaller economies, including Denmark, Ireland, the Netherlands, Norway, Sweden, Switzerland and Thailand. There is little change in the current account surplus in the euro area as a whole.","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/9f653ca1-en_p33.pdf"]},"response":{"errors":[],"extract_id":"extract_a64ef4f0e7a321e6171a37463c4b5352","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":[" **31**\n\nOECD ECONOMIC OUTLOOK, VOLUME 2025 ISSUE 2 © OECD 2025\n\n• The Mexican economy is projected to recover slowly as inflation eases and further cuts to policy rates take effect. However, higher tariffs on exports to the United States and weak public demand are headwinds to growth, with real GDP projected to rise by 0.7% in 2025, 1.2% in 2026 and 1.7% in 2027. In Brazil, growth is anticipated to weaken in the near-term as investment slows in response to higher interest rates, though monetary policy easing through 2026 will support activity. After easing from 2.4% in 2025 to 1.7% in 2026, output growth is projected to recover to 2.2% in 2027.\n\nGlobal trade growth for 2025 is now expected to be 4.2%, reflecting stronger than expected front-loading in the first half of the year. This is being unwound only partially given the momentum of investment- and tech-related trade through 2025. Nonetheless, trade growth is projected to moderate to 2.3% in 2026 as the full effects of higher tariffs are felt, before recovering to 2.8% in 2027 as growth picks up.\n\nThe composition of trade growth is different to the pre-pandemic period (Figure 1.18, Panel A), with the emerging-market economies accounting for a larger share, even as all regions grow more slowly. In particular, the share of trade accounted for by the South-East Asian economies is expected to rise in 2026-27. In contrast, the contribution of China is projected to weaken slightly, as trade growth slows, and the contribution of advanced economies to global trade growth is expected to decline, especially that of Europe and the North American economies.\n\n### **Figure 1.18. Trade patterns are evolving with emerging markets becoming a key driver of growth**\n\nNote: North America includes Canada, Mexico and the United States. Europe includes the 22 OECD countries which are members of the European Union, plus Iceland, Norway, Switzerland, Türkiye and the United Kingdom. South-East Asian economies include Hong Kong (China), Malaysia, the Philippines, Singapore, Chinese Taipei, Thailand, and Viet Nam. Other advanced economies include Australia, Israel, Japan, Korea and New Zealand. Statistical discrepancies mean that current account balances based on national data rarely equal zero at the world level. Source: OECD Economic Outlook 118 database; and OECD calculations.\n\nStatLink 2 https://stat.link/mhq1ks\n\nGlobal current account balances are projected to decrease slightly as a share of global GDP over the next two years (Figure 1.18, Panel B). In the United States, lower imports and reduced remittances linked to international migration, and continued strong growth in tech-related exports and services help to reduce the current account deficit from 2024 levels, despite a decline in export market performance. China’s continued export growth, coupled with weak domestic demand and a continued policy orientation towards import substitution is projected to result in a higher external surplus. Sizeable surpluses are also projected to persist in Germany, Japan, Korea and a number of smaller economies, including Denmark, Ireland, the Netherlands, Norway, Sweden, Switzerland and Thailand. There is little change in the current account surplus in the euro area as a whole."],"full_content":" **31**\n\nOECD ECONOMIC OUTLOOK, VOLUME 2025 ISSUE 2 © OECD 2025\n\n• The Mexican economy is projected to recover slowly as inflation eases and further cuts to policy rates take effect. However, higher tariffs on exports to the United States and weak public demand are headwinds to growth, with real GDP projected to rise by 0.7% in 2025, 1.2% in 2026 and 1.7% in 2027. In Brazil, growth is anticipated to weaken in the near-term as investment slows in response to higher interest rates, though monetary policy easing through 2026 will support activity. After easing from 2.4% in 2025 to 1.7% in 2026, output growth is projected to recover to 2.2% in 2027.\n\nGlobal trade growth for 2025 is now expected to be 4.2%, reflecting stronger than expected front-loading in the first half of the year. This is being unwound only partially given the momentum of investment- and tech-related trade through 2025. Nonetheless, trade growth is projected to moderate to 2.3% in 2026 as the full effects of higher tariffs are felt, before recovering to 2.8% in 2027 as growth picks up.\n\nThe composition of trade growth is different to the pre-pandemic period (Figure 1.18, Panel A), with the emerging-market economies accounting for a larger share, even as all regions grow more slowly. In particular, the share of trade accounted for by the South-East Asian economies is expected to rise in 2026-27. In contrast, the contribution of China is projected to weaken slightly, as trade growth slows, and the contribution of advanced economies to global trade growth is expected to decline, especially that of Europe and the North American economies.\n\n### **Figure 1.18. Trade patterns are evolving with emerging markets becoming a key driver of growth**\n\nNote: North America includes Canada, Mexico and the United States. Europe includes the 22 OECD countries which are members of the European Union, plus Iceland, Norway, Switzerland, Türkiye and the United Kingdom. South-East Asian economies include Hong Kong (China), Malaysia, the Philippines, Singapore, Chinese Taipei, Thailand, and Viet Nam. Other advanced economies include Australia, Israel, Japan, Korea and New Zealand. Statistical discrepancies mean that current account balances based on national data rarely equal zero at the world level. Source: OECD Economic Outlook 118 database; and OECD calculations.\n\nStatLink 2 https://stat.link/mhq1ks\n\nGlobal current account balances are projected to decrease slightly as a share of global GDP over the next two years (Figure 1.18, Panel B). In the United States, lower imports and reduced remittances linked to international migration, and continued strong growth in tech-related exports and services help to reduce the current account deficit from 2024 levels, despite a decline in export market performance. China’s continued export growth, coupled with weak domestic demand and a continued policy orientation towards import substitution is projected to result in a higher external surplus. Sizeable surpluses are also projected to persist in Germany, Japan, Korea and a number of smaller economies, including Denmark, Ireland, the Netherlands, Norway, Sweden, Switzerland and Thailand. There is little change in the current account surplus in the euro area as a whole.","publish_date":null,"title":"Figure 1.18. Trade patterns are evolving with emerging markets becoming a key driver of growth","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/9f653ca1-en_p33.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/9f653ca1-en_p33.pdf"},"request":{"example_id":"chart/9f653ca1-en_p33","product_type":"parse","source_file_path":"9f653ca1-en_p33.pdf"},"started_at":"2026-11-04T19:46:20.870650"} +{"completed_at":"2026-11-04T19:46:26.094976","latency_in_ms":3597,"output":{"example_id":"chart/ADL_Future_of_automotive_mobility_2024_1_p10","grounded_pages":[],"layout_pages":[],"markdown":"Among our sample, expectation of de-\n\nmotorization is not widespread; it is limited\n\nto Western Europe and densely populated\n\nareas in Asia (e.g., Singapore and Hong Kong).\n\nExpected de-motorization is essentially a\n\nwealthy, urban phenomenon, with high-quality\n\npublic transportation a prerequisite. Three-\n\nquarters (76%) of those in European cities with\n\na population of over 5 million are prepared to\n\ngive up their car, compared to 62% in towns with\n\nunder 250,000 people. ADL’s “Future of Mobility”\n\ncity ranking ³ is led by cities such as Singapore,\n\nStockholm, Amsterdam, and Copenhagen,\n\nwhich have highly mature, cost-effective public\n\ntransport and strong support for alternative\n\nmobility types like cycling.\n\nRelatedly, there is an inverse correlation\n\nbetween city size and car usage. In cities\n\nwith over 250,000 inhabitants, around half\n\nof respondents drive under 6,000 km a year.\n\nIn smaller towns and suburban environments\n\n(under 250,000), people who drive between\n\n6,000 and 15,000 km are the majority. Two-thirds\n\n(67%) of those in towns with a population of\n\nunder 10,000 drive over 6,000 km each year.\n\nYoung people expect the importance of having\n\na car to be higher in 10 years, even in mature\n\nmarkets such as Europe and North America. Only\n\nthe older age group (over 45) in Europe and North\n\nAmerica expect to de-motorize as they age\n\n(refer back to Figure 3).\n\n3 Van Audenhove, François-Joseph, et al. ”Future of Mobility 3.0: Reinventing Mobility in the Era of Disruption and Creativity.” Arthur D. Little, March 2018.\n\n### **WHAT DO DRIVERS WANT IN THEIR NEXT CAR?**\n\nWhen deciding on their next vehicle, customers\n\nin mature markets are much more likely\n\nto choose a used car compared to those in\n\nemerging regions, such as China and India\n\n(see Figure 4). This is partly aspirational in terms\n\nof a desire to own a new vehicle but is also due\n\nto the availability of quality used vehicles.\n\nMature markets have a plentiful supply of a\n\nbroad range of used cars, often sourced from a\n\nwell-developed leasing market and backed by a\n\ntrusted used vehicle distribution brand. These\n\nconditions drive higher trust in the quality of\n\nused vehicles and in used car sales channels.\n\n### **CUS TOMERS IN MATURE MARK E T S ARE MUCH MORE LIK ELY TO CHOOSE A USED CAR COMPARED TO THOSE IN EMERGING REGIONS**\n\n**Figure 4. Future car ownership plans**\n\nNote: To facilitate the reading of this Report, where numbers are more than 20% above the global average, we highlight them in black/white bold. Where they are more than 20% below, we highlight them in red bold. Source: Arthur D. Little\n\nNote: To facilitate the reading of this Report, where numbers are more than 10% points above the global average, we highlight them in black/white bold. Where they are more than 10% points below, we highlight them in red bold. Source: Arthur D. Little\n\n### **Figure 4. Future car ownership plans**\n\n78%\n\n**96%** 90% 89% 84%\n\n**52% 61%**\n\n75%\n\n17% **7% 10%**\n\n**36% 32%**\n\n20%\n\n**1%**\n\n**4%**\n\nBuy new car Buy used car Private leasing Company car Private car subscription Shared mobility service Other modes of transport\n\n**What is your plan regarding car ownership and/or use for your next car?**\n\nGlobal China SEA India ME Europe US NEA\n\n1 0\n\n**R E P O R T:** T H E F U T U R E O F AU T O M O T I V E M O B I L I T Y, 2 0 2 4","pages":[{"line_numbers":[],"markdown":"Among our sample, expectation of de-\n\nmotorization is not widespread; it is limited\n\nto Western Europe and densely populated\n\nareas in Asia (e.g., Singapore and Hong Kong).\n\nExpected de-motorization is essentially a\n\nwealthy, urban phenomenon, with high-quality\n\npublic transportation a prerequisite. Three-\n\nquarters (76%) of those in European cities with\n\na population of over 5 million are prepared to\n\ngive up their car, compared to 62% in towns with\n\nunder 250,000 people. ADL’s “Future of Mobility”\n\ncity ranking ³ is led by cities such as Singapore,\n\nStockholm, Amsterdam, and Copenhagen,\n\nwhich have highly mature, cost-effective public\n\ntransport and strong support for alternative\n\nmobility types like cycling.\n\nRelatedly, there is an inverse correlation\n\nbetween city size and car usage. In cities\n\nwith over 250,000 inhabitants, around half\n\nof respondents drive under 6,000 km a year.\n\nIn smaller towns and suburban environments\n\n(under 250,000), people who drive between\n\n6,000 and 15,000 km are the majority. Two-thirds\n\n(67%) of those in towns with a population of\n\nunder 10,000 drive over 6,000 km each year.\n\nYoung people expect the importance of having\n\na car to be higher in 10 years, even in mature\n\nmarkets such as Europe and North America. Only\n\nthe older age group (over 45) in Europe and North\n\nAmerica expect to de-motorize as they age\n\n(refer back to Figure 3).\n\n3 Van Audenhove, François-Joseph, et al. ”Future of Mobility 3.0: Reinventing Mobility in the Era of Disruption and Creativity.” Arthur D. Little, March 2018.\n\n### **WHAT DO DRIVERS WANT IN THEIR NEXT CAR?**\n\nWhen deciding on their next vehicle, customers\n\nin mature markets are much more likely\n\nto choose a used car compared to those in\n\nemerging regions, such as China and India\n\n(see Figure 4). This is partly aspirational in terms\n\nof a desire to own a new vehicle but is also due\n\nto the availability of quality used vehicles.\n\nMature markets have a plentiful supply of a\n\nbroad range of used cars, often sourced from a\n\nwell-developed leasing market and backed by a\n\ntrusted used vehicle distribution brand. These\n\nconditions drive higher trust in the quality of\n\nused vehicles and in used car sales channels.\n\n### **CUS TOMERS IN MATURE MARK E T S ARE MUCH MORE LIK ELY TO CHOOSE A USED CAR COMPARED TO THOSE IN EMERGING REGIONS**\n\n**Figure 4. Future car ownership plans**\n\nNote: To facilitate the reading of this Report, where numbers are more than 20% above the global average, we highlight them in black/white bold. Where they are more than 20% below, we highlight them in red bold. Source: Arthur D. Little\n\nNote: To facilitate the reading of this Report, where numbers are more than 10% points above the global average, we highlight them in black/white bold. Where they are more than 10% points below, we highlight them in red bold. Source: Arthur D. Little\n\n### **Figure 4. Future car ownership plans**\n\n78%\n\n**96%** 90% 89% 84%\n\n**52% 61%**\n\n75%\n\n17% **7% 10%**\n\n**36% 32%**\n\n20%\n\n**1%**\n\n**4%**\n\nBuy new car Buy used car Private leasing Company car Private car subscription Shared mobility service Other modes of transport\n\n**What is your plan regarding car ownership and/or use for your next car?**\n\nGlobal China SEA India ME Europe US NEA\n\n1 0\n\n**R E P O R T:** T H E F U T U R E O F AU T O M O T I V E M O B I L I T Y, 2 0 2 4","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Among our sample, expectation of de-\n\nmotorization is not widespread; it is limited\n\nto Western Europe and densely populated\n\nareas in Asia (e.g., Singapore and Hong Kong).\n\nExpected de-motorization is essentially a\n\nwealthy, urban phenomenon, with high-quality\n\npublic transportation a prerequisite. Three-\n\nquarters (76%) of those in European cities with\n\na population of over 5 million are prepared to\n\ngive up their car, compared to 62% in towns with\n\nunder 250,000 people. ADL’s “Future of Mobility”\n\ncity ranking ³ is led by cities such as Singapore,\n\nStockholm, Amsterdam, and Copenhagen,\n\nwhich have highly mature, cost-effective public\n\ntransport and strong support for alternative\n\nmobility types like cycling.\n\nRelatedly, there is an inverse correlation\n\nbetween city size and car usage. In cities\n\nwith over 250,000 inhabitants, around half\n\nof respondents drive under 6,000 km a year.\n\nIn smaller towns and suburban environments\n\n(under 250,000), people who drive between\n\n6,000 and 15,000 km are the majority. Two-thirds\n\n(67%) of those in towns with a population of\n\nunder 10,000 drive over 6,000 km each year.\n\nYoung people expect the importance of having\n\na car to be higher in 10 years, even in mature\n\nmarkets such as Europe and North America. Only\n\nthe older age group (over 45) in Europe and North\n\nAmerica expect to de-motorize as they age\n\n(refer back to Figure 3).\n\n3 Van Audenhove, François-Joseph, et al. ”Future of Mobility 3.0: Reinventing Mobility in the Era of Disruption and Creativity.” Arthur D. Little, March 2018.\n\n### **WHAT DO DRIVERS WANT IN THEIR NEXT CAR?**\n\nWhen deciding on their next vehicle, customers\n\nin mature markets are much more likely\n\nto choose a used car compared to those in\n\nemerging regions, such as China and India\n\n(see Figure 4). This is partly aspirational in terms\n\nof a desire to own a new vehicle but is also due\n\nto the availability of quality used vehicles.\n\nMature markets have a plentiful supply of a\n\nbroad range of used cars, often sourced from a\n\nwell-developed leasing market and backed by a\n\ntrusted used vehicle distribution brand. These\n\nconditions drive higher trust in the quality of\n\nused vehicles and in used car sales channels.\n\n### **CUS TOMERS IN MATURE MARK E T S ARE MUCH MORE LIK ELY TO CHOOSE A USED CAR COMPARED TO THOSE IN EMERGING REGIONS**\n\n**Figure 4. Future car ownership plans**\n\nNote: To facilitate the reading of this Report, where numbers are more than 20% above the global average, we highlight them in black/white bold. Where they are more than 20% below, we highlight them in red bold. Source: Arthur D. Little\n\nNote: To facilitate the reading of this Report, where numbers are more than 10% points above the global average, we highlight them in black/white bold. Where they are more than 10% points below, we highlight them in red bold. Source: Arthur D. Little\n\n### **Figure 4. Future car ownership plans**\n\n78%\n\n**96%** 90% 89% 84%\n\n**52% 61%**\n\n75%\n\n17% **7% 10%**\n\n**36% 32%**\n\n20%\n\n**1%**\n\n**4%**\n\nBuy new car Buy used car Private leasing Company car Private car subscription Shared mobility service Other modes of transport\n\n**What is your plan regarding car ownership and/or use for your next car?**\n\nGlobal China SEA India ME Europe US NEA\n\n1 0\n\n**R E P O R T:** T H E F U T U R E O F AU T O M O T I V E M O B I L I T Y, 2 0 2 4","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ADL_Future_of_automotive_mobility_2024_1_p10.pdf"]},"response":{"errors":[],"extract_id":"extract_511efb5be9410e17c1596da7f2b1c208","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Among our sample, expectation of de-\n\nmotorization is not widespread; it is limited\n\nto Western Europe and densely populated\n\nareas in Asia (e.g., Singapore and Hong Kong).\n\nExpected de-motorization is essentially a\n\nwealthy, urban phenomenon, with high-quality\n\npublic transportation a prerequisite. Three-\n\nquarters (76%) of those in European cities with\n\na population of over 5 million are prepared to\n\ngive up their car, compared to 62% in towns with\n\nunder 250,000 people. ADL’s “Future of Mobility”\n\ncity ranking ³ is led by cities such as Singapore,\n\nStockholm, Amsterdam, and Copenhagen,\n\nwhich have highly mature, cost-effective public\n\ntransport and strong support for alternative\n\nmobility types like cycling.\n\nRelatedly, there is an inverse correlation\n\nbetween city size and car usage. In cities\n\nwith over 250,000 inhabitants, around half\n\nof respondents drive under 6,000 km a year.\n\nIn smaller towns and suburban environments\n\n(under 250,000), people who drive between\n\n6,000 and 15,000 km are the majority. Two-thirds\n\n(67%) of those in towns with a population of\n\nunder 10,000 drive over 6,000 km each year.\n\nYoung people expect the importance of having\n\na car to be higher in 10 years, even in mature\n\nmarkets such as Europe and North America. Only\n\nthe older age group (over 45) in Europe and North\n\nAmerica expect to de-motorize as they age\n\n(refer back to Figure 3).\n\n3 Van Audenhove, François-Joseph, et al. ”Future of Mobility 3.0: Reinventing Mobility in the Era of Disruption and Creativity.” Arthur D. Little, March 2018.\n\n### **WHAT DO DRIVERS WANT IN THEIR NEXT CAR?**\n\nWhen deciding on their next vehicle, customers\n\nin mature markets are much more likely\n\nto choose a used car compared to those in\n\nemerging regions, such as China and India\n\n(see Figure 4). This is partly aspirational in terms\n\nof a desire to own a new vehicle but is also due\n\nto the availability of quality used vehicles.\n\nMature markets have a plentiful supply of a\n\nbroad range of used cars, often sourced from a\n\nwell-developed leasing market and backed by a\n\ntrusted used vehicle distribution brand. These\n\nconditions drive higher trust in the quality of\n\nused vehicles and in used car sales channels.\n\n### **CUS TOMERS IN MATURE MARK E T S ARE MUCH MORE LIK ELY TO CHOOSE A USED CAR COMPARED TO THOSE IN EMERGING REGIONS**\n\n**Figure 4. Future car ownership plans**\n\nNote: To facilitate the reading of this Report, where numbers are more than 20% above the global average, we highlight them in black/white bold. Where they are more than 20% below, we highlight them in red bold. Source: Arthur D. Little\n\nNote: To facilitate the reading of this Report, where numbers are more than 10% points above the global average, we highlight them in black/white bold. Where they are more than 10% points below, we highlight them in red bold. Source: Arthur D. Little\n\n### **Figure 4. Future car ownership plans**\n\n78%\n\n**96%** 90% 89% 84%\n\n**52% 61%**\n\n75%\n\n17% **7% 10%**\n\n**36% 32%**\n\n20%\n\n**1%**\n\n**4%**\n\nBuy new car Buy used car Private leasing Company car Private car subscription Shared mobility service Other modes of transport\n\n**What is your plan regarding car ownership and/or use for your next car?**\n\nGlobal China SEA India ME Europe US NEA\n\n1 0\n\n**R E P O R T:** T H E F U T U R E O F AU T O M O T I V E M O B I L I T Y, 2 0 2 4"],"full_content":"Among our sample, expectation of de-\n\nmotorization is not widespread; it is limited\n\nto Western Europe and densely populated\n\nareas in Asia (e.g., Singapore and Hong Kong).\n\nExpected de-motorization is essentially a\n\nwealthy, urban phenomenon, with high-quality\n\npublic transportation a prerequisite. Three-\n\nquarters (76%) of those in European cities with\n\na population of over 5 million are prepared to\n\ngive up their car, compared to 62% in towns with\n\nunder 250,000 people. ADL’s “Future of Mobility”\n\ncity ranking ³ is led by cities such as Singapore,\n\nStockholm, Amsterdam, and Copenhagen,\n\nwhich have highly mature, cost-effective public\n\ntransport and strong support for alternative\n\nmobility types like cycling.\n\nRelatedly, there is an inverse correlation\n\nbetween city size and car usage. In cities\n\nwith over 250,000 inhabitants, around half\n\nof respondents drive under 6,000 km a year.\n\nIn smaller towns and suburban environments\n\n(under 250,000), people who drive between\n\n6,000 and 15,000 km are the majority. Two-thirds\n\n(67%) of those in towns with a population of\n\nunder 10,000 drive over 6,000 km each year.\n\nYoung people expect the importance of having\n\na car to be higher in 10 years, even in mature\n\nmarkets such as Europe and North America. Only\n\nthe older age group (over 45) in Europe and North\n\nAmerica expect to de-motorize as they age\n\n(refer back to Figure 3).\n\n3 Van Audenhove, François-Joseph, et al. ”Future of Mobility 3.0: Reinventing Mobility in the Era of Disruption and Creativity.” Arthur D. Little, March 2018.\n\n### **WHAT DO DRIVERS WANT IN THEIR NEXT CAR?**\n\nWhen deciding on their next vehicle, customers\n\nin mature markets are much more likely\n\nto choose a used car compared to those in\n\nemerging regions, such as China and India\n\n(see Figure 4). This is partly aspirational in terms\n\nof a desire to own a new vehicle but is also due\n\nto the availability of quality used vehicles.\n\nMature markets have a plentiful supply of a\n\nbroad range of used cars, often sourced from a\n\nwell-developed leasing market and backed by a\n\ntrusted used vehicle distribution brand. These\n\nconditions drive higher trust in the quality of\n\nused vehicles and in used car sales channels.\n\n### **CUS TOMERS IN MATURE MARK E T S ARE MUCH MORE LIK ELY TO CHOOSE A USED CAR COMPARED TO THOSE IN EMERGING REGIONS**\n\n**Figure 4. Future car ownership plans**\n\nNote: To facilitate the reading of this Report, where numbers are more than 20% above the global average, we highlight them in black/white bold. Where they are more than 20% below, we highlight them in red bold. Source: Arthur D. Little\n\nNote: To facilitate the reading of this Report, where numbers are more than 10% points above the global average, we highlight them in black/white bold. Where they are more than 10% points below, we highlight them in red bold. Source: Arthur D. Little\n\n### **Figure 4. Future car ownership plans**\n\n78%\n\n**96%** 90% 89% 84%\n\n**52% 61%**\n\n75%\n\n17% **7% 10%**\n\n**36% 32%**\n\n20%\n\n**1%**\n\n**4%**\n\nBuy new car Buy used car Private leasing Company car Private car subscription Shared mobility service Other modes of transport\n\n**What is your plan regarding car ownership and/or use for your next car?**\n\nGlobal China SEA India ME Europe US NEA\n\n1 0\n\n**R E P O R T:** T H E F U T U R E O F AU T O M O T I V E M O B I L I T Y, 2 0 2 4","publish_date":null,"title":"Figure 4. Future car ownership plans","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ADL_Future_of_automotive_mobility_2024_1_p10.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ADL_Future_of_automotive_mobility_2024_1_p10.pdf"},"request":{"example_id":"chart/ADL_Future_of_automotive_mobility_2024_1_p10","product_type":"parse","source_file_path":"ADL_Future_of_automotive_mobility_2024_1_p10.pdf"},"started_at":"2026-11-04T19:46:22.496984"} +{"completed_at":"2026-11-04T19:46:27.405416","latency_in_ms":3292,"output":{"example_id":"chart/ADL_Future_of_automotive_mobility_2024_1_p13","grounded_pages":[],"layout_pages":[],"markdown":"**A R T H U R D . L I T T L E**\n\nIn fact, the most often experienced new mobility\n\nservice for urban respondents is ride hailing,\n\ntried by 50%-75% of people in every region or\n\ncountry except Europe and Northeast Asia.\n\nGlobally, those in the 30-44 age range are the\n\nheaviest experimenters, with about 45% trying\n\nride, car, or two-wheeler sharing, compared to\n\nonly about 5% of people over 60 who have tried\n\nmost service types (see Figure 6).\n\n### **NEW MOBILITY SERVICES FOCUS ON RIDE HAILING**\n\nThere are significant regional differences in\n\nthe use of new mobility services, influenced\n\nby car ownership rates and access to reliable,\n\ncomprehensive public transport. Urban dwellers\n\nin India, Southeast Asia, and the Middle East\n\nmake an average of 14 to 17 journeys each\n\nmonth using new mobility services, compared\n\nto residents of Europe, the US, China, and\n\nNortheast Asia, who make dramatically fewer,\n\nbetween 11 and 12 (see Figure 7).\n\n**Figure 6. Types of new mobility services tried by urban dwellers, by age**\n\nSource: Arthur D. Little\n\n**Figure 7. Average monthly usage of new mobility services**\n\nSource: Arthur D. Little\n\nSource: Arthur D. Little\n\n### **Figure 6. Types of new mobility services tried by urban dwellers, by age**\n\n**Sharing of a personal transportation device**\n\n**Public transportation**\n\n**Which types of new mobility services have you used?**\n\nOnly urban respondents (in %)\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n<30 years 45-60 years 30-44 years >60 years\n\nLocal underground/ metro, tram, bus service\n\nLocal train service\n\nRide hailing\n\n2-wheeler sharing\n\nRide sharing\n\nCar sharing\n\nSource: Arthur D. Little\n\n### **Figure 7. Average monthly usage of new mobility services**\n\n**How many times do you use new mobility services in a normal month?**\n\nOnly urban respondents\n\n2\\.9\n\n2\\.4\n\n**3\\.6 3.9**\n\n3\\.3 3.0 3.3\n\n**2\\.2**\n\n3\\.5 3.8 **4\\.4 4.5**\n\n3\\.5 **2\\.9**\n\n3\\.9 **3\\.7 3.3**\n\n2\\.3\n\n2\\.9\n\n2\\.5\n\n3\\.6 3.9 3.6\n\n**4\\.5** 4\\.1\n\n3\\.1\n\n2\\.5\n\n3\\.8\n\nGlobal China SEA India ME Europe US NEA\n\n2\\.7\n\n2\\.3\n\n**3\\.6**\n\n**2\\.3**\n\nCar sharing Ride hailing Ride sharing 2-wheeler sharing\n\n1 3","pages":[{"line_numbers":[],"markdown":"**A R T H U R D . L I T T L E**\n\nIn fact, the most often experienced new mobility\n\nservice for urban respondents is ride hailing,\n\ntried by 50%-75% of people in every region or\n\ncountry except Europe and Northeast Asia.\n\nGlobally, those in the 30-44 age range are the\n\nheaviest experimenters, with about 45% trying\n\nride, car, or two-wheeler sharing, compared to\n\nonly about 5% of people over 60 who have tried\n\nmost service types (see Figure 6).\n\n### **NEW MOBILITY SERVICES FOCUS ON RIDE HAILING**\n\nThere are significant regional differences in\n\nthe use of new mobility services, influenced\n\nby car ownership rates and access to reliable,\n\ncomprehensive public transport. Urban dwellers\n\nin India, Southeast Asia, and the Middle East\n\nmake an average of 14 to 17 journeys each\n\nmonth using new mobility services, compared\n\nto residents of Europe, the US, China, and\n\nNortheast Asia, who make dramatically fewer,\n\nbetween 11 and 12 (see Figure 7).\n\n**Figure 6. Types of new mobility services tried by urban dwellers, by age**\n\nSource: Arthur D. Little\n\n**Figure 7. Average monthly usage of new mobility services**\n\nSource: Arthur D. Little\n\nSource: Arthur D. Little\n\n### **Figure 6. Types of new mobility services tried by urban dwellers, by age**\n\n**Sharing of a personal transportation device**\n\n**Public transportation**\n\n**Which types of new mobility services have you used?**\n\nOnly urban respondents (in %)\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n<30 years 45-60 years 30-44 years >60 years\n\nLocal underground/ metro, tram, bus service\n\nLocal train service\n\nRide hailing\n\n2-wheeler sharing\n\nRide sharing\n\nCar sharing\n\nSource: Arthur D. Little\n\n### **Figure 7. Average monthly usage of new mobility services**\n\n**How many times do you use new mobility services in a normal month?**\n\nOnly urban respondents\n\n2\\.9\n\n2\\.4\n\n**3\\.6 3.9**\n\n3\\.3 3.0 3.3\n\n**2\\.2**\n\n3\\.5 3.8 **4\\.4 4.5**\n\n3\\.5 **2\\.9**\n\n3\\.9 **3\\.7 3.3**\n\n2\\.3\n\n2\\.9\n\n2\\.5\n\n3\\.6 3.9 3.6\n\n**4\\.5** 4\\.1\n\n3\\.1\n\n2\\.5\n\n3\\.8\n\nGlobal China SEA India ME Europe US NEA\n\n2\\.7\n\n2\\.3\n\n**3\\.6**\n\n**2\\.3**\n\nCar sharing Ride hailing Ride sharing 2-wheeler sharing\n\n1 3","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**A R T H U R D . L I T T L E**\n\nIn fact, the most often experienced new mobility\n\nservice for urban respondents is ride hailing,\n\ntried by 50%-75% of people in every region or\n\ncountry except Europe and Northeast Asia.\n\nGlobally, those in the 30-44 age range are the\n\nheaviest experimenters, with about 45% trying\n\nride, car, or two-wheeler sharing, compared to\n\nonly about 5% of people over 60 who have tried\n\nmost service types (see Figure 6).\n\n### **NEW MOBILITY SERVICES FOCUS ON RIDE HAILING**\n\nThere are significant regional differences in\n\nthe use of new mobility services, influenced\n\nby car ownership rates and access to reliable,\n\ncomprehensive public transport. Urban dwellers\n\nin India, Southeast Asia, and the Middle East\n\nmake an average of 14 to 17 journeys each\n\nmonth using new mobility services, compared\n\nto residents of Europe, the US, China, and\n\nNortheast Asia, who make dramatically fewer,\n\nbetween 11 and 12 (see Figure 7).\n\n**Figure 6. Types of new mobility services tried by urban dwellers, by age**\n\nSource: Arthur D. Little\n\n**Figure 7. Average monthly usage of new mobility services**\n\nSource: Arthur D. Little\n\nSource: Arthur D. Little\n\n### **Figure 6. Types of new mobility services tried by urban dwellers, by age**\n\n**Sharing of a personal transportation device**\n\n**Public transportation**\n\n**Which types of new mobility services have you used?**\n\nOnly urban respondents (in %)\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n<30 years 45-60 years 30-44 years >60 years\n\nLocal underground/ metro, tram, bus service\n\nLocal train service\n\nRide hailing\n\n2-wheeler sharing\n\nRide sharing\n\nCar sharing\n\nSource: Arthur D. Little\n\n### **Figure 7. Average monthly usage of new mobility services**\n\n**How many times do you use new mobility services in a normal month?**\n\nOnly urban respondents\n\n2\\.9\n\n2\\.4\n\n**3\\.6 3.9**\n\n3\\.3 3.0 3.3\n\n**2\\.2**\n\n3\\.5 3.8 **4\\.4 4.5**\n\n3\\.5 **2\\.9**\n\n3\\.9 **3\\.7 3.3**\n\n2\\.3\n\n2\\.9\n\n2\\.5\n\n3\\.6 3.9 3.6\n\n**4\\.5** 4\\.1\n\n3\\.1\n\n2\\.5\n\n3\\.8\n\nGlobal China SEA India ME Europe US NEA\n\n2\\.7\n\n2\\.3\n\n**3\\.6**\n\n**2\\.3**\n\nCar sharing Ride hailing Ride sharing 2-wheeler sharing\n\n1 3","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ADL_Future_of_automotive_mobility_2024_1_p13.pdf"]},"response":{"errors":[],"extract_id":"extract_e4638a6b0b68983e005475f037abf4a8","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**A R T H U R D . L I T T L E**\n\nIn fact, the most often experienced new mobility\n\nservice for urban respondents is ride hailing,\n\ntried by 50%-75% of people in every region or\n\ncountry except Europe and Northeast Asia.\n\nGlobally, those in the 30-44 age range are the\n\nheaviest experimenters, with about 45% trying\n\nride, car, or two-wheeler sharing, compared to\n\nonly about 5% of people over 60 who have tried\n\nmost service types (see Figure 6).\n\n### **NEW MOBILITY SERVICES FOCUS ON RIDE HAILING**\n\nThere are significant regional differences in\n\nthe use of new mobility services, influenced\n\nby car ownership rates and access to reliable,\n\ncomprehensive public transport. Urban dwellers\n\nin India, Southeast Asia, and the Middle East\n\nmake an average of 14 to 17 journeys each\n\nmonth using new mobility services, compared\n\nto residents of Europe, the US, China, and\n\nNortheast Asia, who make dramatically fewer,\n\nbetween 11 and 12 (see Figure 7).\n\n**Figure 6. Types of new mobility services tried by urban dwellers, by age**\n\nSource: Arthur D. Little\n\n**Figure 7. Average monthly usage of new mobility services**\n\nSource: Arthur D. Little\n\nSource: Arthur D. Little\n\n### **Figure 6. Types of new mobility services tried by urban dwellers, by age**\n\n**Sharing of a personal transportation device**\n\n**Public transportation**\n\n**Which types of new mobility services have you used?**\n\nOnly urban respondents (in %)\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n<30 years 45-60 years 30-44 years >60 years\n\nLocal underground/ metro, tram, bus service\n\nLocal train service\n\nRide hailing\n\n2-wheeler sharing\n\nRide sharing\n\nCar sharing\n\nSource: Arthur D. Little\n\n### **Figure 7. Average monthly usage of new mobility services**\n\n**How many times do you use new mobility services in a normal month?**\n\nOnly urban respondents\n\n2\\.9\n\n2\\.4\n\n**3\\.6 3.9**\n\n3\\.3 3.0 3.3\n\n**2\\.2**\n\n3\\.5 3.8 **4\\.4 4.5**\n\n3\\.5 **2\\.9**\n\n3\\.9 **3\\.7 3.3**\n\n2\\.3\n\n2\\.9\n\n2\\.5\n\n3\\.6 3.9 3.6\n\n**4\\.5** 4\\.1\n\n3\\.1\n\n2\\.5\n\n3\\.8\n\nGlobal China SEA India ME Europe US NEA\n\n2\\.7\n\n2\\.3\n\n**3\\.6**\n\n**2\\.3**\n\nCar sharing Ride hailing Ride sharing 2-wheeler sharing\n\n1 3"],"full_content":"**A R T H U R D . L I T T L E**\n\nIn fact, the most often experienced new mobility\n\nservice for urban respondents is ride hailing,\n\ntried by 50%-75% of people in every region or\n\ncountry except Europe and Northeast Asia.\n\nGlobally, those in the 30-44 age range are the\n\nheaviest experimenters, with about 45% trying\n\nride, car, or two-wheeler sharing, compared to\n\nonly about 5% of people over 60 who have tried\n\nmost service types (see Figure 6).\n\n### **NEW MOBILITY SERVICES FOCUS ON RIDE HAILING**\n\nThere are significant regional differences in\n\nthe use of new mobility services, influenced\n\nby car ownership rates and access to reliable,\n\ncomprehensive public transport. Urban dwellers\n\nin India, Southeast Asia, and the Middle East\n\nmake an average of 14 to 17 journeys each\n\nmonth using new mobility services, compared\n\nto residents of Europe, the US, China, and\n\nNortheast Asia, who make dramatically fewer,\n\nbetween 11 and 12 (see Figure 7).\n\n**Figure 6. Types of new mobility services tried by urban dwellers, by age**\n\nSource: Arthur D. Little\n\n**Figure 7. Average monthly usage of new mobility services**\n\nSource: Arthur D. Little\n\nSource: Arthur D. Little\n\n### **Figure 6. Types of new mobility services tried by urban dwellers, by age**\n\n**Sharing of a personal transportation device**\n\n**Public transportation**\n\n**Which types of new mobility services have you used?**\n\nOnly urban respondents (in %)\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n<30 years 45-60 years 30-44 years >60 years\n\nLocal underground/ metro, tram, bus service\n\nLocal train service\n\nRide hailing\n\n2-wheeler sharing\n\nRide sharing\n\nCar sharing\n\nSource: Arthur D. Little\n\n### **Figure 7. Average monthly usage of new mobility services**\n\n**How many times do you use new mobility services in a normal month?**\n\nOnly urban respondents\n\n2\\.9\n\n2\\.4\n\n**3\\.6 3.9**\n\n3\\.3 3.0 3.3\n\n**2\\.2**\n\n3\\.5 3.8 **4\\.4 4.5**\n\n3\\.5 **2\\.9**\n\n3\\.9 **3\\.7 3.3**\n\n2\\.3\n\n2\\.9\n\n2\\.5\n\n3\\.6 3.9 3.6\n\n**4\\.5** 4\\.1\n\n3\\.1\n\n2\\.5\n\n3\\.8\n\nGlobal China SEA India ME Europe US NEA\n\n2\\.7\n\n2\\.3\n\n**3\\.6**\n\n**2\\.3**\n\nCar sharing Ride hailing Ride sharing 2-wheeler sharing\n\n1 3","publish_date":null,"title":"Figure 6. Types of new mobility services tried by urban dwellers, by age Figure 7. Average monthly usage of new","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ADL_Future_of_automotive_mobility_2024_1_p13.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ADL_Future_of_automotive_mobility_2024_1_p13.pdf"},"request":{"example_id":"chart/ADL_Future_of_automotive_mobility_2024_1_p13","product_type":"parse","source_file_path":"ADL_Future_of_automotive_mobility_2024_1_p13.pdf"},"started_at":"2026-11-04T19:46:24.113072"} +{"completed_at":"2026-11-04T19:46:27.477987","latency_in_ms":2442,"output":{"example_id":"chart/ADL_Future_of_automotive_mobility_2024_1_p14","grounded_pages":[],"layout_pages":[],"markdown":"Globally, ride hailing usage occurs, on average,\n\n3\\.5 times per month, with usage being higher\n\nin all parts of Asia and lower in the US (2.9\n\njourneys) and Europe (2.3 journeys). Car sharing\n\nis the most popular new mobility service in the\n\nUS, with users taking an average of 3.3 monthly\n\njourneys.\n\n### **WHAT DRIVES ADOPTION OF NEW MOBILITY SERVICES?**\n\nWhen asked why they choose new mobility\n\nservices, respondents listed flexibility (62%),\n\ncost (52%), and the environment (44%) as\n\ntheir top three reasons, as shown in Figure 8.\n\nFlexibility covers a range of areas — customers\n\ndon’t need to own or maintain a vehicle, battle\n\ntraffic in congested urban areas, or find\n\nparking; they can also switch easily between\n\ndifferent transport modes. Flexibility is clearly\n\nthe number one reason for using new mobility\n\nservices everywhere outside Europe (where cost\n\nslightly outweighs it) and the US (where cost is\n\nalmost as important as flexibility).\n\nIncome had a minor impact on the reasons for\n\nchoosing new mobility services among urban\n\nrespondents. However, cost unsurprisingly\n\nincreased in importance for lower-income\n\ngroups, while flexibility and environmental\n\nconsiderations decreased:\n\n\\- **Very-high-income groups** — 66% cited\n\nflexibility as the key reason, 47% said cost,\n\nand 51% cited the environment. \\- **Low-income groups** — 58% chose flexibility\n\nas a key factor, 55% cited cost, and 44% cited\n\nthe environment.\n\nWhen asked what changes to new mobility\n\nservices would encourage people to give up their\n\ncars, the major areas respondents highlighted\n\nglobally are cost (50%) and availability (44%).\n\nTwenty-eight percent would not give up their\n\ncars under any circumstances. There are\n\nsignificant regional differences: 77% of Chinese\n\nrespondents require greater availability to\n\nreplace their cars with mobility services; 75%\n\nof Indians cite improved ease of use; and nearly\n\nhalf (49%) of US respondents and 37% of those\n\nin Europe would not give up their car under\n\nany circumstances. Essentially, to increase\n\nthe adoption of new mobility services, people\n\nrequire lower costs, higher availability, greater\n\nflexibility, improved reliability, and increased\n\nquality. This combination of requirements\n\nfor higher service and lower cost typically\n\nmakes the “shared” element of CASE unviable\n\neconomically.\n\n**Figure 8. Reasons for choosing new mobility services**\n\nSource: Arthur D. Little\n\nSource: Arthur D. Little\n\n### **Figure 8. Reasons for choosing new mobility services**\n\n**What motivates you to use new mobility services?**\n\nOnly urban respondents\n\nChina SEA India ME Europe US NEA\n\n62%\n\n52%\n\n44%\n\n73%\n\n45% 51%\n\n67%\n\n59%\n\nGlobal\n\n71%\n\n50% **56%** 60%\n\n49% 43%\n\n**49%** 51%\n\n41%\n\n56% 54%\n\n**34%**\n\n69%\n\n44%\n\n**34%**\n\n44%\n\nFlexibility Cost Environment\n\n1 4\n\n**R E P O R T:** T H E F U T U R E O F AU T O M O T I V E M O B I L I T Y, 2 0 2 4","pages":[{"line_numbers":[],"markdown":"Globally, ride hailing usage occurs, on average,\n\n3\\.5 times per month, with usage being higher\n\nin all parts of Asia and lower in the US (2.9\n\njourneys) and Europe (2.3 journeys). Car sharing\n\nis the most popular new mobility service in the\n\nUS, with users taking an average of 3.3 monthly\n\njourneys.\n\n### **WHAT DRIVES ADOPTION OF NEW MOBILITY SERVICES?**\n\nWhen asked why they choose new mobility\n\nservices, respondents listed flexibility (62%),\n\ncost (52%), and the environment (44%) as\n\ntheir top three reasons, as shown in Figure 8.\n\nFlexibility covers a range of areas — customers\n\ndon’t need to own or maintain a vehicle, battle\n\ntraffic in congested urban areas, or find\n\nparking; they can also switch easily between\n\ndifferent transport modes. Flexibility is clearly\n\nthe number one reason for using new mobility\n\nservices everywhere outside Europe (where cost\n\nslightly outweighs it) and the US (where cost is\n\nalmost as important as flexibility).\n\nIncome had a minor impact on the reasons for\n\nchoosing new mobility services among urban\n\nrespondents. However, cost unsurprisingly\n\nincreased in importance for lower-income\n\ngroups, while flexibility and environmental\n\nconsiderations decreased:\n\n\\- **Very-high-income groups** — 66% cited\n\nflexibility as the key reason, 47% said cost,\n\nand 51% cited the environment. \\- **Low-income groups** — 58% chose flexibility\n\nas a key factor, 55% cited cost, and 44% cited\n\nthe environment.\n\nWhen asked what changes to new mobility\n\nservices would encourage people to give up their\n\ncars, the major areas respondents highlighted\n\nglobally are cost (50%) and availability (44%).\n\nTwenty-eight percent would not give up their\n\ncars under any circumstances. There are\n\nsignificant regional differences: 77% of Chinese\n\nrespondents require greater availability to\n\nreplace their cars with mobility services; 75%\n\nof Indians cite improved ease of use; and nearly\n\nhalf (49%) of US respondents and 37% of those\n\nin Europe would not give up their car under\n\nany circumstances. Essentially, to increase\n\nthe adoption of new mobility services, people\n\nrequire lower costs, higher availability, greater\n\nflexibility, improved reliability, and increased\n\nquality. This combination of requirements\n\nfor higher service and lower cost typically\n\nmakes the “shared” element of CASE unviable\n\neconomically.\n\n**Figure 8. Reasons for choosing new mobility services**\n\nSource: Arthur D. Little\n\nSource: Arthur D. Little\n\n### **Figure 8. Reasons for choosing new mobility services**\n\n**What motivates you to use new mobility services?**\n\nOnly urban respondents\n\nChina SEA India ME Europe US NEA\n\n62%\n\n52%\n\n44%\n\n73%\n\n45% 51%\n\n67%\n\n59%\n\nGlobal\n\n71%\n\n50% **56%** 60%\n\n49% 43%\n\n**49%** 51%\n\n41%\n\n56% 54%\n\n**34%**\n\n69%\n\n44%\n\n**34%**\n\n44%\n\nFlexibility Cost Environment\n\n1 4\n\n**R E P O R T:** T H E F U T U R E O F AU T O M O T I V E M O B I L I T Y, 2 0 2 4","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Globally, ride hailing usage occurs, on average,\n\n3\\.5 times per month, with usage being higher\n\nin all parts of Asia and lower in the US (2.9\n\njourneys) and Europe (2.3 journeys). Car sharing\n\nis the most popular new mobility service in the\n\nUS, with users taking an average of 3.3 monthly\n\njourneys.\n\n### **WHAT DRIVES ADOPTION OF NEW MOBILITY SERVICES?**\n\nWhen asked why they choose new mobility\n\nservices, respondents listed flexibility (62%),\n\ncost (52%), and the environment (44%) as\n\ntheir top three reasons, as shown in Figure 8.\n\nFlexibility covers a range of areas — customers\n\ndon’t need to own or maintain a vehicle, battle\n\ntraffic in congested urban areas, or find\n\nparking; they can also switch easily between\n\ndifferent transport modes. Flexibility is clearly\n\nthe number one reason for using new mobility\n\nservices everywhere outside Europe (where cost\n\nslightly outweighs it) and the US (where cost is\n\nalmost as important as flexibility).\n\nIncome had a minor impact on the reasons for\n\nchoosing new mobility services among urban\n\nrespondents. However, cost unsurprisingly\n\nincreased in importance for lower-income\n\ngroups, while flexibility and environmental\n\nconsiderations decreased:\n\n\\- **Very-high-income groups** — 66% cited\n\nflexibility as the key reason, 47% said cost,\n\nand 51% cited the environment. \\- **Low-income groups** — 58% chose flexibility\n\nas a key factor, 55% cited cost, and 44% cited\n\nthe environment.\n\nWhen asked what changes to new mobility\n\nservices would encourage people to give up their\n\ncars, the major areas respondents highlighted\n\nglobally are cost (50%) and availability (44%).\n\nTwenty-eight percent would not give up their\n\ncars under any circumstances. There are\n\nsignificant regional differences: 77% of Chinese\n\nrespondents require greater availability to\n\nreplace their cars with mobility services; 75%\n\nof Indians cite improved ease of use; and nearly\n\nhalf (49%) of US respondents and 37% of those\n\nin Europe would not give up their car under\n\nany circumstances. Essentially, to increase\n\nthe adoption of new mobility services, people\n\nrequire lower costs, higher availability, greater\n\nflexibility, improved reliability, and increased\n\nquality. This combination of requirements\n\nfor higher service and lower cost typically\n\nmakes the “shared” element of CASE unviable\n\neconomically.\n\n**Figure 8. Reasons for choosing new mobility services**\n\nSource: Arthur D. Little\n\nSource: Arthur D. Little\n\n### **Figure 8. Reasons for choosing new mobility services**\n\n**What motivates you to use new mobility services?**\n\nOnly urban respondents\n\nChina SEA India ME Europe US NEA\n\n62%\n\n52%\n\n44%\n\n73%\n\n45% 51%\n\n67%\n\n59%\n\nGlobal\n\n71%\n\n50% **56%** 60%\n\n49% 43%\n\n**49%** 51%\n\n41%\n\n56% 54%\n\n**34%**\n\n69%\n\n44%\n\n**34%**\n\n44%\n\nFlexibility Cost Environment\n\n1 4\n\n**R E P O R T:** T H E F U T U R E O F AU T O M O T I V E M O B I L I T Y, 2 0 2 4","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ADL_Future_of_automotive_mobility_2024_1_p14.pdf"]},"response":{"errors":[],"extract_id":"extract_b9099afb06b8ed59e943e7a06e35a696","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Globally, ride hailing usage occurs, on average,\n\n3\\.5 times per month, with usage being higher\n\nin all parts of Asia and lower in the US (2.9\n\njourneys) and Europe (2.3 journeys). Car sharing\n\nis the most popular new mobility service in the\n\nUS, with users taking an average of 3.3 monthly\n\njourneys.\n\n### **WHAT DRIVES ADOPTION OF NEW MOBILITY SERVICES?**\n\nWhen asked why they choose new mobility\n\nservices, respondents listed flexibility (62%),\n\ncost (52%), and the environment (44%) as\n\ntheir top three reasons, as shown in Figure 8.\n\nFlexibility covers a range of areas — customers\n\ndon’t need to own or maintain a vehicle, battle\n\ntraffic in congested urban areas, or find\n\nparking; they can also switch easily between\n\ndifferent transport modes. Flexibility is clearly\n\nthe number one reason for using new mobility\n\nservices everywhere outside Europe (where cost\n\nslightly outweighs it) and the US (where cost is\n\nalmost as important as flexibility).\n\nIncome had a minor impact on the reasons for\n\nchoosing new mobility services among urban\n\nrespondents. However, cost unsurprisingly\n\nincreased in importance for lower-income\n\ngroups, while flexibility and environmental\n\nconsiderations decreased:\n\n\\- **Very-high-income groups** — 66% cited\n\nflexibility as the key reason, 47% said cost,\n\nand 51% cited the environment. \\- **Low-income groups** — 58% chose flexibility\n\nas a key factor, 55% cited cost, and 44% cited\n\nthe environment.\n\nWhen asked what changes to new mobility\n\nservices would encourage people to give up their\n\ncars, the major areas respondents highlighted\n\nglobally are cost (50%) and availability (44%).\n\nTwenty-eight percent would not give up their\n\ncars under any circumstances. There are\n\nsignificant regional differences: 77% of Chinese\n\nrespondents require greater availability to\n\nreplace their cars with mobility services; 75%\n\nof Indians cite improved ease of use; and nearly\n\nhalf (49%) of US respondents and 37% of those\n\nin Europe would not give up their car under\n\nany circumstances. Essentially, to increase\n\nthe adoption of new mobility services, people\n\nrequire lower costs, higher availability, greater\n\nflexibility, improved reliability, and increased\n\nquality. This combination of requirements\n\nfor higher service and lower cost typically\n\nmakes the “shared” element of CASE unviable\n\neconomically.\n\n**Figure 8. Reasons for choosing new mobility services**\n\nSource: Arthur D. Little\n\nSource: Arthur D. Little\n\n### **Figure 8. Reasons for choosing new mobility services**\n\n**What motivates you to use new mobility services?**\n\nOnly urban respondents\n\nChina SEA India ME Europe US NEA\n\n62%\n\n52%\n\n44%\n\n73%\n\n45% 51%\n\n67%\n\n59%\n\nGlobal\n\n71%\n\n50% **56%** 60%\n\n49% 43%\n\n**49%** 51%\n\n41%\n\n56% 54%\n\n**34%**\n\n69%\n\n44%\n\n**34%**\n\n44%\n\nFlexibility Cost Environment\n\n1 4\n\n**R E P O R T:** T H E F U T U R E O F AU T O M O T I V E M O B I L I T Y, 2 0 2 4"],"full_content":"Globally, ride hailing usage occurs, on average,\n\n3\\.5 times per month, with usage being higher\n\nin all parts of Asia and lower in the US (2.9\n\njourneys) and Europe (2.3 journeys). Car sharing\n\nis the most popular new mobility service in the\n\nUS, with users taking an average of 3.3 monthly\n\njourneys.\n\n### **WHAT DRIVES ADOPTION OF NEW MOBILITY SERVICES?**\n\nWhen asked why they choose new mobility\n\nservices, respondents listed flexibility (62%),\n\ncost (52%), and the environment (44%) as\n\ntheir top three reasons, as shown in Figure 8.\n\nFlexibility covers a range of areas — customers\n\ndon’t need to own or maintain a vehicle, battle\n\ntraffic in congested urban areas, or find\n\nparking; they can also switch easily between\n\ndifferent transport modes. Flexibility is clearly\n\nthe number one reason for using new mobility\n\nservices everywhere outside Europe (where cost\n\nslightly outweighs it) and the US (where cost is\n\nalmost as important as flexibility).\n\nIncome had a minor impact on the reasons for\n\nchoosing new mobility services among urban\n\nrespondents. However, cost unsurprisingly\n\nincreased in importance for lower-income\n\ngroups, while flexibility and environmental\n\nconsiderations decreased:\n\n\\- **Very-high-income groups** — 66% cited\n\nflexibility as the key reason, 47% said cost,\n\nand 51% cited the environment. \\- **Low-income groups** — 58% chose flexibility\n\nas a key factor, 55% cited cost, and 44% cited\n\nthe environment.\n\nWhen asked what changes to new mobility\n\nservices would encourage people to give up their\n\ncars, the major areas respondents highlighted\n\nglobally are cost (50%) and availability (44%).\n\nTwenty-eight percent would not give up their\n\ncars under any circumstances. There are\n\nsignificant regional differences: 77% of Chinese\n\nrespondents require greater availability to\n\nreplace their cars with mobility services; 75%\n\nof Indians cite improved ease of use; and nearly\n\nhalf (49%) of US respondents and 37% of those\n\nin Europe would not give up their car under\n\nany circumstances. Essentially, to increase\n\nthe adoption of new mobility services, people\n\nrequire lower costs, higher availability, greater\n\nflexibility, improved reliability, and increased\n\nquality. This combination of requirements\n\nfor higher service and lower cost typically\n\nmakes the “shared” element of CASE unviable\n\neconomically.\n\n**Figure 8. Reasons for choosing new mobility services**\n\nSource: Arthur D. Little\n\nSource: Arthur D. Little\n\n### **Figure 8. Reasons for choosing new mobility services**\n\n**What motivates you to use new mobility services?**\n\nOnly urban respondents\n\nChina SEA India ME Europe US NEA\n\n62%\n\n52%\n\n44%\n\n73%\n\n45% 51%\n\n67%\n\n59%\n\nGlobal\n\n71%\n\n50% **56%** 60%\n\n49% 43%\n\n**49%** 51%\n\n41%\n\n56% 54%\n\n**34%**\n\n69%\n\n44%\n\n**34%**\n\n44%\n\nFlexibility Cost Environment\n\n1 4\n\n**R E P O R T:** T H E F U T U R E O F AU T O M O T I V E M O B I L I T Y, 2 0 2 4","publish_date":null,"title":"- -","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ADL_Future_of_automotive_mobility_2024_1_p14.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ADL_Future_of_automotive_mobility_2024_1_p14.pdf"},"request":{"example_id":"chart/ADL_Future_of_automotive_mobility_2024_1_p14","product_type":"parse","source_file_path":"ADL_Future_of_automotive_mobility_2024_1_p14.pdf"},"started_at":"2026-11-04T19:46:25.035804"} +{"completed_at":"2026-11-04T19:31:16.097609","latency_in_ms":1947,"output":{"example_id":"chart/ADL_Future_of_automotive_mobility_2024_1_p17","grounded_pages":[],"layout_pages":[],"markdown":"**A R T H U R D . L I T T L E**\n\nThis correlation between income and\n\nacceptance of autonomous driving holds in all\n\nregions, though the threshold for acceptance\n\nvaries. In China, we see 34% net approval at\n\naverage income and city size between 1 and 5\n\nmillion people, while in the US we need to move\n\nto cities with 5 million inhabitants and high\n\nincomes to see similar net approval levels.\n\nEuropean, Northeast Asian, and US respondents\n\nare the least willing to adopt autonomous\n\nvehicles, while those in India, China, and the\n\nMiddle East demonstrate a much more favorable\n\nattitude (see Figure 10).\n\n### **THE IMPORTANCE OF SAFETY TO ACCEPTANCE**\n\nMirroring the examples cited above from\n\nCalifornia, the primary concern with\n\nautonomous driving across all incomes and\n\nregions is safety, encompassing both machine\n\nand human errors, as shown in Figure 11.\n\nUser concerns about safety mainly reflect the\n\nstill-immature technology for L4/L5 autonomous\n\ndriving and a resulting lack of positive experiences\n\nreported in the media. People are aware that\n\ncurrent autonomous vehicles depend on backup\n\nhuman drivers who are ready to intervene,\n\nmust go slower than a human driver, and only\n\noperate in dedicated environments with special\n\ninfrastructure. The promised benefits of\n\nautomation around convenience and ease of use\n\nhave not materialized yet. On the contrary, recent\n\nincidents in the US, which led to pausing trials of\n\nthe most advanced robo-taxis, coupled with high-\n\nprofile stories on crashes of private vehicles, have\n\nnot helped to increase customers’ confidence in\n\nthe maturity of the technology.\n\nHowever, if these safety fears could be overcome\n\nby positive user experiences, it would lead to\n\ngreater usage of personal autonomous vehicles.\n\nWhen asked whether a fully autonomous car\n\nwould increase or decrease their car usage,\n\nall respondents globally gave an average\n\nweighted score of +4, rising to +26 for those\n\nwho were positive about autonomous driving.\n\nAgain, this was highest in India, with a score of\n\n\\+32 for all respondents and +42 for those open\n\nto autonomous driving. This increased usage\n\nwould replace: \\- Existing personal non-autonomous vehicle (51%) \\- Public transport (44%) \\- Taxis (32%)\n\nOverall, since 2015, the expectation of the\n\npercentage of public transport journeys\n\npotentially replaced with autonomous cars\n\nhas grown, which would represent a switch\n\nfrom shared to personal mobility options.\n\n**Figure 10. Desire to use autonomous/semiautonomous cars**\n\nSource: Arthur D. Little\n\nSource: Arthur D. Little\n\n### **Figure 10. Desire to use autonomous/semiautonomous cars**\n\n**Unfavorable**\n\n**attitude**\n\n**Favorable attitude**\n\n16\n\n**29**\n\n**23**\n\n**48**\n\n**28**\n\n**\\-17 -17 -2**\n\n**Do you, or would you, use cars that are fully or semiautonomous?**\n\nWeighted net favorability score (scale: -100 to 100)\n\n16\n\nGlobal China SEA India ME Europe US NEA\n\n1 7","pages":[{"line_numbers":[],"markdown":"**A R T H U R D . L I T T L E**\n\nThis correlation between income and\n\nacceptance of autonomous driving holds in all\n\nregions, though the threshold for acceptance\n\nvaries. In China, we see 34% net approval at\n\naverage income and city size between 1 and 5\n\nmillion people, while in the US we need to move\n\nto cities with 5 million inhabitants and high\n\nincomes to see similar net approval levels.\n\nEuropean, Northeast Asian, and US respondents\n\nare the least willing to adopt autonomous\n\nvehicles, while those in India, China, and the\n\nMiddle East demonstrate a much more favorable\n\nattitude (see Figure 10).\n\n### **THE IMPORTANCE OF SAFETY TO ACCEPTANCE**\n\nMirroring the examples cited above from\n\nCalifornia, the primary concern with\n\nautonomous driving across all incomes and\n\nregions is safety, encompassing both machine\n\nand human errors, as shown in Figure 11.\n\nUser concerns about safety mainly reflect the\n\nstill-immature technology for L4/L5 autonomous\n\ndriving and a resulting lack of positive experiences\n\nreported in the media. People are aware that\n\ncurrent autonomous vehicles depend on backup\n\nhuman drivers who are ready to intervene,\n\nmust go slower than a human driver, and only\n\noperate in dedicated environments with special\n\ninfrastructure. The promised benefits of\n\nautomation around convenience and ease of use\n\nhave not materialized yet. On the contrary, recent\n\nincidents in the US, which led to pausing trials of\n\nthe most advanced robo-taxis, coupled with high-\n\nprofile stories on crashes of private vehicles, have\n\nnot helped to increase customers’ confidence in\n\nthe maturity of the technology.\n\nHowever, if these safety fears could be overcome\n\nby positive user experiences, it would lead to\n\ngreater usage of personal autonomous vehicles.\n\nWhen asked whether a fully autonomous car\n\nwould increase or decrease their car usage,\n\nall respondents globally gave an average\n\nweighted score of +4, rising to +26 for those\n\nwho were positive about autonomous driving.\n\nAgain, this was highest in India, with a score of\n\n\\+32 for all respondents and +42 for those open\n\nto autonomous driving. This increased usage\n\nwould replace: \\- Existing personal non-autonomous vehicle (51%) \\- Public transport (44%) \\- Taxis (32%)\n\nOverall, since 2015, the expectation of the\n\npercentage of public transport journeys\n\npotentially replaced with autonomous cars\n\nhas grown, which would represent a switch\n\nfrom shared to personal mobility options.\n\n**Figure 10. Desire to use autonomous/semiautonomous cars**\n\nSource: Arthur D. Little\n\nSource: Arthur D. Little\n\n### **Figure 10. Desire to use autonomous/semiautonomous cars**\n\n**Unfavorable**\n\n**attitude**\n\n**Favorable attitude**\n\n16\n\n**29**\n\n**23**\n\n**48**\n\n**28**\n\n**\\-17 -17 -2**\n\n**Do you, or would you, use cars that are fully or semiautonomous?**\n\nWeighted net favorability score (scale: -100 to 100)\n\n16\n\nGlobal China SEA India ME Europe US NEA\n\n1 7","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**A R T H U R D . L I T T L E**\n\nThis correlation between income and\n\nacceptance of autonomous driving holds in all\n\nregions, though the threshold for acceptance\n\nvaries. In China, we see 34% net approval at\n\naverage income and city size between 1 and 5\n\nmillion people, while in the US we need to move\n\nto cities with 5 million inhabitants and high\n\nincomes to see similar net approval levels.\n\nEuropean, Northeast Asian, and US respondents\n\nare the least willing to adopt autonomous\n\nvehicles, while those in India, China, and the\n\nMiddle East demonstrate a much more favorable\n\nattitude (see Figure 10).\n\n### **THE IMPORTANCE OF SAFETY TO ACCEPTANCE**\n\nMirroring the examples cited above from\n\nCalifornia, the primary concern with\n\nautonomous driving across all incomes and\n\nregions is safety, encompassing both machine\n\nand human errors, as shown in Figure 11.\n\nUser concerns about safety mainly reflect the\n\nstill-immature technology for L4/L5 autonomous\n\ndriving and a resulting lack of positive experiences\n\nreported in the media. People are aware that\n\ncurrent autonomous vehicles depend on backup\n\nhuman drivers who are ready to intervene,\n\nmust go slower than a human driver, and only\n\noperate in dedicated environments with special\n\ninfrastructure. The promised benefits of\n\nautomation around convenience and ease of use\n\nhave not materialized yet. On the contrary, recent\n\nincidents in the US, which led to pausing trials of\n\nthe most advanced robo-taxis, coupled with high-\n\nprofile stories on crashes of private vehicles, have\n\nnot helped to increase customers’ confidence in\n\nthe maturity of the technology.\n\nHowever, if these safety fears could be overcome\n\nby positive user experiences, it would lead to\n\ngreater usage of personal autonomous vehicles.\n\nWhen asked whether a fully autonomous car\n\nwould increase or decrease their car usage,\n\nall respondents globally gave an average\n\nweighted score of +4, rising to +26 for those\n\nwho were positive about autonomous driving.\n\nAgain, this was highest in India, with a score of\n\n\\+32 for all respondents and +42 for those open\n\nto autonomous driving. This increased usage\n\nwould replace: \\- Existing personal non-autonomous vehicle (51%) \\- Public transport (44%) \\- Taxis (32%)\n\nOverall, since 2015, the expectation of the\n\npercentage of public transport journeys\n\npotentially replaced with autonomous cars\n\nhas grown, which would represent a switch\n\nfrom shared to personal mobility options.\n\n**Figure 10. Desire to use autonomous/semiautonomous cars**\n\nSource: Arthur D. Little\n\nSource: Arthur D. Little\n\n### **Figure 10. Desire to use autonomous/semiautonomous cars**\n\n**Unfavorable**\n\n**attitude**\n\n**Favorable attitude**\n\n16\n\n**29**\n\n**23**\n\n**48**\n\n**28**\n\n**\\-17 -17 -2**\n\n**Do you, or would you, use cars that are fully or semiautonomous?**\n\nWeighted net favorability score (scale: -100 to 100)\n\n16\n\nGlobal China SEA India ME Europe US NEA\n\n1 7","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ADL_Future_of_automotive_mobility_2024_1_p17.pdf"]},"response":{"errors":[],"extract_id":"extract_9bfb6b548160d0f073c11f7337efa84d","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**A R T H U R D . L I T T L E**\n\nThis correlation between income and\n\nacceptance of autonomous driving holds in all\n\nregions, though the threshold for acceptance\n\nvaries. In China, we see 34% net approval at\n\naverage income and city size between 1 and 5\n\nmillion people, while in the US we need to move\n\nto cities with 5 million inhabitants and high\n\nincomes to see similar net approval levels.\n\nEuropean, Northeast Asian, and US respondents\n\nare the least willing to adopt autonomous\n\nvehicles, while those in India, China, and the\n\nMiddle East demonstrate a much more favorable\n\nattitude (see Figure 10).\n\n### **THE IMPORTANCE OF SAFETY TO ACCEPTANCE**\n\nMirroring the examples cited above from\n\nCalifornia, the primary concern with\n\nautonomous driving across all incomes and\n\nregions is safety, encompassing both machine\n\nand human errors, as shown in Figure 11.\n\nUser concerns about safety mainly reflect the\n\nstill-immature technology for L4/L5 autonomous\n\ndriving and a resulting lack of positive experiences\n\nreported in the media. People are aware that\n\ncurrent autonomous vehicles depend on backup\n\nhuman drivers who are ready to intervene,\n\nmust go slower than a human driver, and only\n\noperate in dedicated environments with special\n\ninfrastructure. The promised benefits of\n\nautomation around convenience and ease of use\n\nhave not materialized yet. On the contrary, recent\n\nincidents in the US, which led to pausing trials of\n\nthe most advanced robo-taxis, coupled with high-\n\nprofile stories on crashes of private vehicles, have\n\nnot helped to increase customers’ confidence in\n\nthe maturity of the technology.\n\nHowever, if these safety fears could be overcome\n\nby positive user experiences, it would lead to\n\ngreater usage of personal autonomous vehicles.\n\nWhen asked whether a fully autonomous car\n\nwould increase or decrease their car usage,\n\nall respondents globally gave an average\n\nweighted score of +4, rising to +26 for those\n\nwho were positive about autonomous driving.\n\nAgain, this was highest in India, with a score of\n\n\\+32 for all respondents and +42 for those open\n\nto autonomous driving. This increased usage\n\nwould replace: \\- Existing personal non-autonomous vehicle (51%) \\- Public transport (44%) \\- Taxis (32%)\n\nOverall, since 2015, the expectation of the\n\npercentage of public transport journeys\n\npotentially replaced with autonomous cars\n\nhas grown, which would represent a switch\n\nfrom shared to personal mobility options.\n\n**Figure 10. Desire to use autonomous/semiautonomous cars**\n\nSource: Arthur D. Little\n\nSource: Arthur D. Little\n\n### **Figure 10. Desire to use autonomous/semiautonomous cars**\n\n**Unfavorable**\n\n**attitude**\n\n**Favorable attitude**\n\n16\n\n**29**\n\n**23**\n\n**48**\n\n**28**\n\n**\\-17 -17 -2**\n\n**Do you, or would you, use cars that are fully or semiautonomous?**\n\nWeighted net favorability score (scale: -100 to 100)\n\n16\n\nGlobal China SEA India ME Europe US NEA\n\n1 7"],"full_content":"**A R T H U R D . L I T T L E**\n\nThis correlation between income and\n\nacceptance of autonomous driving holds in all\n\nregions, though the threshold for acceptance\n\nvaries. In China, we see 34% net approval at\n\naverage income and city size between 1 and 5\n\nmillion people, while in the US we need to move\n\nto cities with 5 million inhabitants and high\n\nincomes to see similar net approval levels.\n\nEuropean, Northeast Asian, and US respondents\n\nare the least willing to adopt autonomous\n\nvehicles, while those in India, China, and the\n\nMiddle East demonstrate a much more favorable\n\nattitude (see Figure 10).\n\n### **THE IMPORTANCE OF SAFETY TO ACCEPTANCE**\n\nMirroring the examples cited above from\n\nCalifornia, the primary concern with\n\nautonomous driving across all incomes and\n\nregions is safety, encompassing both machine\n\nand human errors, as shown in Figure 11.\n\nUser concerns about safety mainly reflect the\n\nstill-immature technology for L4/L5 autonomous\n\ndriving and a resulting lack of positive experiences\n\nreported in the media. People are aware that\n\ncurrent autonomous vehicles depend on backup\n\nhuman drivers who are ready to intervene,\n\nmust go slower than a human driver, and only\n\noperate in dedicated environments with special\n\ninfrastructure. The promised benefits of\n\nautomation around convenience and ease of use\n\nhave not materialized yet. On the contrary, recent\n\nincidents in the US, which led to pausing trials of\n\nthe most advanced robo-taxis, coupled with high-\n\nprofile stories on crashes of private vehicles, have\n\nnot helped to increase customers’ confidence in\n\nthe maturity of the technology.\n\nHowever, if these safety fears could be overcome\n\nby positive user experiences, it would lead to\n\ngreater usage of personal autonomous vehicles.\n\nWhen asked whether a fully autonomous car\n\nwould increase or decrease their car usage,\n\nall respondents globally gave an average\n\nweighted score of +4, rising to +26 for those\n\nwho were positive about autonomous driving.\n\nAgain, this was highest in India, with a score of\n\n\\+32 for all respondents and +42 for those open\n\nto autonomous driving. This increased usage\n\nwould replace: \\- Existing personal non-autonomous vehicle (51%) \\- Public transport (44%) \\- Taxis (32%)\n\nOverall, since 2015, the expectation of the\n\npercentage of public transport journeys\n\npotentially replaced with autonomous cars\n\nhas grown, which would represent a switch\n\nfrom shared to personal mobility options.\n\n**Figure 10. Desire to use autonomous/semiautonomous cars**\n\nSource: Arthur D. Little\n\nSource: Arthur D. Little\n\n### **Figure 10. Desire to use autonomous/semiautonomous cars**\n\n**Unfavorable**\n\n**attitude**\n\n**Favorable attitude**\n\n16\n\n**29**\n\n**23**\n\n**48**\n\n**28**\n\n**\\-17 -17 -2**\n\n**Do you, or would you, use cars that are fully or semiautonomous?**\n\nWeighted net favorability score (scale: -100 to 100)\n\n16\n\nGlobal China SEA India ME Europe US NEA\n\n1 7","publish_date":null,"title":"- - -","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ADL_Future_of_automotive_mobility_2024_1_p17.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ADL_Future_of_automotive_mobility_2024_1_p17.pdf"},"request":{"example_id":"chart/ADL_Future_of_automotive_mobility_2024_1_p17","product_type":"parse","source_file_path":"ADL_Future_of_automotive_mobility_2024_1_p17.pdf"},"started_at":"2026-11-04T19:31:14.150148"} +{"completed_at":"2026-11-04T19:46:29.545433","latency_in_ms":1872,"output":{"example_id":"chart/ADL_Future_of_automotive_mobility_2024_1_p25","grounded_pages":[],"layout_pages":[],"markdown":"**A R T H U R D . L I T T L E**\n\n### **BEVs & BRAND LOYALTY: THE RISE OF CHINA**\n\nShopping for a BEV may prompt customers to\n\nreconsider the different options and brands\n\navailable to them, potentially reducing brand\n\nloyalty built up over many years in the ICE world.\n\nGlobally, 19% of existing owners of premium\n\nbrand models ⁵ say they would be more likely\n\nto switch, compared to 17% of those who own\n\nvolume brand vehicles. This shift is highest in\n\nIndia (58% premium and 42% volume) and lowest\n\nin Europe (9% premium and 8% volume) and\n\nthe US volume market (3%). However, the US\n\npremium market is being and will continue to be\n\ndisrupted, with over a quarter (28%) more likely\n\nto change.\n\n5 Premium brands include: Volvo, Audi, BMW, Jaguar, Tesla, Lexus, Cadillac, Mercedes, and Nio; volume brands include: Citroën, Fiat, Ford, Hyundai, Jeep, Kia, Nissan, Opel, Peugeot, Renault, SEAT, Skoda, Toyota, Volkswagen, and BYD.\n\nChinese EV manufacturers are rapidly growing\n\nin terms of global presence. Their rising\n\nvisibility (and lower prices) leads to potential\n\ntariffs and regulatory action in Europe and the\n\nUS, including exclusion from some European\n\nnational incentive programs, due to low content\n\nof European components. However, outside\n\nChina more than half (60%) of respondents say\n\nthey were not interested in purchasing a Chinese\n\nEV (see Figure 18), naming perception of quality\n\nand customer sales/service experience as the\n\nmain reasons.\n\n**Figure 17. Factors encouraging EV purchase**\n\nSource: Arthur D. Little\n\n**Environment**\n\n**Trend & technology front-runner**\n\n**Price**\n\n**More fun to drive than ICE cars (acceleration)**\n\nSource: Arthur D. Little\n\n### **Figure 17. Factors encouraging EV purchase**\n\n82\n\n80\n\n**46**\n\n21\n\n79\n\n83\n\n**56**\n\n20\n\n80\n\n79\n\n**36**\n\n16\n\n69\n\n80\n\n**27**\n\n23\n\n70\n\n86\n\n**20**\n\n18\n\n60\n\n75\n\n**15**\n\n15\n\n84\n\n89\n\n**50**\n\n23\n\n81\n\n81\n\n**48**\n\n**34**\n\n71\n\n85\n\n33\n\n20\n\n**What factors would encourage you to purchase an electric car?**\n\n(in %)\n\n68\n\n80\n\n**23**\n\n26\n\n68\n\n81\n\n**21**\n\n26\n\n**52**\n\n72\n\n**12**\n\n17\n\n76\n\n82\n\n37\n\n24\n\n76\n\n85\n\n38\n\n22\n\n71\n\n82\n\n30\n\n18\n\n82\n\n84\n\n**53**\n\n26\n\n79\n\n88\n\n**53**\n\n20\n\n76\n\n85\n\n**37**\n\n18\n\n90\n\n84\n\n40\n\n**29**\n\n89\n\n84\n\n**62**\n\n26\n\n84\n\n90\n\n**41**\n\n25\n\n74\n\n87\n\n33\n\n28\n\n72\n\n87\n\n**28**\n\n18\n\n62\n\n84\n\n26\n\n17\n\nConventional powertrain owners a) EV intenders Conventional powertrain owners b) EV non intenders EV or hybrid owners\n\n**Figure 18. Reasons for not purchasing a Chinese EV**\n\nSource: Arthur D. Little\n\nSource: Arthur D. Little\n\n### **Figure 18. Reasons for not purchasing a Chinese EV**\n\n22%\n\n53%\n\n47%\n\n42%\n\n30%\n\n43%\n\n19%\n\nPoor quality\n\nChallenges associated with service or purchase\n\nBrand reputation\n\nPreference for domestic brands\n\nEnvironmental, social & governance standards\n\nVehicle design\n\n51%\n\n40%\n\n45%\n\n38% 38%\n\n30%\n\n37% 40%\n\n22%\n\n19%\n\n16%\n\n**Why would you not be interested in purchasing an EV from a Chinese manufacturer?**\n\n**GLOBAL OUTSIDE CHINA** – 60% of respondents would not be interested in purchasing an EV from a Chinese manufacturer\n\nConventional powertrain owners a) EV intenders\n\nConventional powertrain owners b) EV non-intenders\n\nEV or hybrid owners\n\n2 5","pages":[{"line_numbers":[],"markdown":"**A R T H U R D . L I T T L E**\n\n### **BEVs & BRAND LOYALTY: THE RISE OF CHINA**\n\nShopping for a BEV may prompt customers to\n\nreconsider the different options and brands\n\navailable to them, potentially reducing brand\n\nloyalty built up over many years in the ICE world.\n\nGlobally, 19% of existing owners of premium\n\nbrand models ⁵ say they would be more likely\n\nto switch, compared to 17% of those who own\n\nvolume brand vehicles. This shift is highest in\n\nIndia (58% premium and 42% volume) and lowest\n\nin Europe (9% premium and 8% volume) and\n\nthe US volume market (3%). However, the US\n\npremium market is being and will continue to be\n\ndisrupted, with over a quarter (28%) more likely\n\nto change.\n\n5 Premium brands include: Volvo, Audi, BMW, Jaguar, Tesla, Lexus, Cadillac, Mercedes, and Nio; volume brands include: Citroën, Fiat, Ford, Hyundai, Jeep, Kia, Nissan, Opel, Peugeot, Renault, SEAT, Skoda, Toyota, Volkswagen, and BYD.\n\nChinese EV manufacturers are rapidly growing\n\nin terms of global presence. Their rising\n\nvisibility (and lower prices) leads to potential\n\ntariffs and regulatory action in Europe and the\n\nUS, including exclusion from some European\n\nnational incentive programs, due to low content\n\nof European components. However, outside\n\nChina more than half (60%) of respondents say\n\nthey were not interested in purchasing a Chinese\n\nEV (see Figure 18), naming perception of quality\n\nand customer sales/service experience as the\n\nmain reasons.\n\n**Figure 17. Factors encouraging EV purchase**\n\nSource: Arthur D. Little\n\n**Environment**\n\n**Trend & technology front-runner**\n\n**Price**\n\n**More fun to drive than ICE cars (acceleration)**\n\nSource: Arthur D. Little\n\n### **Figure 17. Factors encouraging EV purchase**\n\n82\n\n80\n\n**46**\n\n21\n\n79\n\n83\n\n**56**\n\n20\n\n80\n\n79\n\n**36**\n\n16\n\n69\n\n80\n\n**27**\n\n23\n\n70\n\n86\n\n**20**\n\n18\n\n60\n\n75\n\n**15**\n\n15\n\n84\n\n89\n\n**50**\n\n23\n\n81\n\n81\n\n**48**\n\n**34**\n\n71\n\n85\n\n33\n\n20\n\n**What factors would encourage you to purchase an electric car?**\n\n(in %)\n\n68\n\n80\n\n**23**\n\n26\n\n68\n\n81\n\n**21**\n\n26\n\n**52**\n\n72\n\n**12**\n\n17\n\n76\n\n82\n\n37\n\n24\n\n76\n\n85\n\n38\n\n22\n\n71\n\n82\n\n30\n\n18\n\n82\n\n84\n\n**53**\n\n26\n\n79\n\n88\n\n**53**\n\n20\n\n76\n\n85\n\n**37**\n\n18\n\n90\n\n84\n\n40\n\n**29**\n\n89\n\n84\n\n**62**\n\n26\n\n84\n\n90\n\n**41**\n\n25\n\n74\n\n87\n\n33\n\n28\n\n72\n\n87\n\n**28**\n\n18\n\n62\n\n84\n\n26\n\n17\n\nConventional powertrain owners a) EV intenders Conventional powertrain owners b) EV non intenders EV or hybrid owners\n\n**Figure 18. Reasons for not purchasing a Chinese EV**\n\nSource: Arthur D. Little\n\nSource: Arthur D. Little\n\n### **Figure 18. Reasons for not purchasing a Chinese EV**\n\n22%\n\n53%\n\n47%\n\n42%\n\n30%\n\n43%\n\n19%\n\nPoor quality\n\nChallenges associated with service or purchase\n\nBrand reputation\n\nPreference for domestic brands\n\nEnvironmental, social & governance standards\n\nVehicle design\n\n51%\n\n40%\n\n45%\n\n38% 38%\n\n30%\n\n37% 40%\n\n22%\n\n19%\n\n16%\n\n**Why would you not be interested in purchasing an EV from a Chinese manufacturer?**\n\n**GLOBAL OUTSIDE CHINA** – 60% of respondents would not be interested in purchasing an EV from a Chinese manufacturer\n\nConventional powertrain owners a) EV intenders\n\nConventional powertrain owners b) EV non-intenders\n\nEV or hybrid owners\n\n2 5","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**A R T H U R D . L I T T L E**\n\n### **BEVs & BRAND LOYALTY: THE RISE OF CHINA**\n\nShopping for a BEV may prompt customers to\n\nreconsider the different options and brands\n\navailable to them, potentially reducing brand\n\nloyalty built up over many years in the ICE world.\n\nGlobally, 19% of existing owners of premium\n\nbrand models ⁵ say they would be more likely\n\nto switch, compared to 17% of those who own\n\nvolume brand vehicles. This shift is highest in\n\nIndia (58% premium and 42% volume) and lowest\n\nin Europe (9% premium and 8% volume) and\n\nthe US volume market (3%). However, the US\n\npremium market is being and will continue to be\n\ndisrupted, with over a quarter (28%) more likely\n\nto change.\n\n5 Premium brands include: Volvo, Audi, BMW, Jaguar, Tesla, Lexus, Cadillac, Mercedes, and Nio; volume brands include: Citroën, Fiat, Ford, Hyundai, Jeep, Kia, Nissan, Opel, Peugeot, Renault, SEAT, Skoda, Toyota, Volkswagen, and BYD.\n\nChinese EV manufacturers are rapidly growing\n\nin terms of global presence. Their rising\n\nvisibility (and lower prices) leads to potential\n\ntariffs and regulatory action in Europe and the\n\nUS, including exclusion from some European\n\nnational incentive programs, due to low content\n\nof European components. However, outside\n\nChina more than half (60%) of respondents say\n\nthey were not interested in purchasing a Chinese\n\nEV (see Figure 18), naming perception of quality\n\nand customer sales/service experience as the\n\nmain reasons.\n\n**Figure 17. Factors encouraging EV purchase**\n\nSource: Arthur D. Little\n\n**Environment**\n\n**Trend & technology front-runner**\n\n**Price**\n\n**More fun to drive than ICE cars (acceleration)**\n\nSource: Arthur D. Little\n\n### **Figure 17. Factors encouraging EV purchase**\n\n82\n\n80\n\n**46**\n\n21\n\n79\n\n83\n\n**56**\n\n20\n\n80\n\n79\n\n**36**\n\n16\n\n69\n\n80\n\n**27**\n\n23\n\n70\n\n86\n\n**20**\n\n18\n\n60\n\n75\n\n**15**\n\n15\n\n84\n\n89\n\n**50**\n\n23\n\n81\n\n81\n\n**48**\n\n**34**\n\n71\n\n85\n\n33\n\n20\n\n**What factors would encourage you to purchase an electric car?**\n\n(in %)\n\n68\n\n80\n\n**23**\n\n26\n\n68\n\n81\n\n**21**\n\n26\n\n**52**\n\n72\n\n**12**\n\n17\n\n76\n\n82\n\n37\n\n24\n\n76\n\n85\n\n38\n\n22\n\n71\n\n82\n\n30\n\n18\n\n82\n\n84\n\n**53**\n\n26\n\n79\n\n88\n\n**53**\n\n20\n\n76\n\n85\n\n**37**\n\n18\n\n90\n\n84\n\n40\n\n**29**\n\n89\n\n84\n\n**62**\n\n26\n\n84\n\n90\n\n**41**\n\n25\n\n74\n\n87\n\n33\n\n28\n\n72\n\n87\n\n**28**\n\n18\n\n62\n\n84\n\n26\n\n17\n\nConventional powertrain owners a) EV intenders Conventional powertrain owners b) EV non intenders EV or hybrid owners\n\n**Figure 18. Reasons for not purchasing a Chinese EV**\n\nSource: Arthur D. Little\n\nSource: Arthur D. Little\n\n### **Figure 18. Reasons for not purchasing a Chinese EV**\n\n22%\n\n53%\n\n47%\n\n42%\n\n30%\n\n43%\n\n19%\n\nPoor quality\n\nChallenges associated with service or purchase\n\nBrand reputation\n\nPreference for domestic brands\n\nEnvironmental, social & governance standards\n\nVehicle design\n\n51%\n\n40%\n\n45%\n\n38% 38%\n\n30%\n\n37% 40%\n\n22%\n\n19%\n\n16%\n\n**Why would you not be interested in purchasing an EV from a Chinese manufacturer?**\n\n**GLOBAL OUTSIDE CHINA** – 60% of respondents would not be interested in purchasing an EV from a Chinese manufacturer\n\nConventional powertrain owners a) EV intenders\n\nConventional powertrain owners b) EV non-intenders\n\nEV or hybrid owners\n\n2 5","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ADL_Future_of_automotive_mobility_2024_1_p25.pdf"]},"response":{"errors":[],"extract_id":"extract_ad612cf60f10031e172133d39a5ed33f","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**A R T H U R D . L I T T L E**\n\n### **BEVs & BRAND LOYALTY: THE RISE OF CHINA**\n\nShopping for a BEV may prompt customers to\n\nreconsider the different options and brands\n\navailable to them, potentially reducing brand\n\nloyalty built up over many years in the ICE world.\n\nGlobally, 19% of existing owners of premium\n\nbrand models ⁵ say they would be more likely\n\nto switch, compared to 17% of those who own\n\nvolume brand vehicles. This shift is highest in\n\nIndia (58% premium and 42% volume) and lowest\n\nin Europe (9% premium and 8% volume) and\n\nthe US volume market (3%). However, the US\n\npremium market is being and will continue to be\n\ndisrupted, with over a quarter (28%) more likely\n\nto change.\n\n5 Premium brands include: Volvo, Audi, BMW, Jaguar, Tesla, Lexus, Cadillac, Mercedes, and Nio; volume brands include: Citroën, Fiat, Ford, Hyundai, Jeep, Kia, Nissan, Opel, Peugeot, Renault, SEAT, Skoda, Toyota, Volkswagen, and BYD.\n\nChinese EV manufacturers are rapidly growing\n\nin terms of global presence. Their rising\n\nvisibility (and lower prices) leads to potential\n\ntariffs and regulatory action in Europe and the\n\nUS, including exclusion from some European\n\nnational incentive programs, due to low content\n\nof European components. However, outside\n\nChina more than half (60%) of respondents say\n\nthey were not interested in purchasing a Chinese\n\nEV (see Figure 18), naming perception of quality\n\nand customer sales/service experience as the\n\nmain reasons.\n\n**Figure 17. Factors encouraging EV purchase**\n\nSource: Arthur D. Little\n\n**Environment**\n\n**Trend & technology front-runner**\n\n**Price**\n\n**More fun to drive than ICE cars (acceleration)**\n\nSource: Arthur D. Little\n\n### **Figure 17. Factors encouraging EV purchase**\n\n82\n\n80\n\n**46**\n\n21\n\n79\n\n83\n\n**56**\n\n20\n\n80\n\n79\n\n**36**\n\n16\n\n69\n\n80\n\n**27**\n\n23\n\n70\n\n86\n\n**20**\n\n18\n\n60\n\n75\n\n**15**\n\n15\n\n84\n\n89\n\n**50**\n\n23\n\n81\n\n81\n\n**48**\n\n**34**\n\n71\n\n85\n\n33\n\n20\n\n**What factors would encourage you to purchase an electric car?**\n\n(in %)\n\n68\n\n80\n\n**23**\n\n26\n\n68\n\n81\n\n**21**\n\n26\n\n**52**\n\n72\n\n**12**\n\n17\n\n76\n\n82\n\n37\n\n24\n\n76\n\n85\n\n38\n\n22\n\n71\n\n82\n\n30\n\n18\n\n82\n\n84\n\n**53**\n\n26\n\n79\n\n88\n\n**53**\n\n20\n\n76\n\n85\n\n**37**\n\n18\n\n90\n\n84\n\n40\n\n**29**\n\n89\n\n84\n\n**62**\n\n26\n\n84\n\n90\n\n**41**\n\n25\n\n74\n\n87\n\n33\n\n28\n\n72\n\n87\n\n**28**\n\n18\n\n62\n\n84\n\n26\n\n17\n\nConventional powertrain owners a) EV intenders Conventional powertrain owners b) EV non intenders EV or hybrid owners\n\n**Figure 18. Reasons for not purchasing a Chinese EV**\n\nSource: Arthur D. Little\n\nSource: Arthur D. Little\n\n### **Figure 18. Reasons for not purchasing a Chinese EV**\n\n22%\n\n53%\n\n47%\n\n42%\n\n30%\n\n43%\n\n19%\n\nPoor quality\n\nChallenges associated with service or purchase\n\nBrand reputation\n\nPreference for domestic brands\n\nEnvironmental, social & governance standards\n\nVehicle design\n\n51%\n\n40%\n\n45%\n\n38% 38%\n\n30%\n\n37% 40%\n\n22%\n\n19%\n\n16%\n\n**Why would you not be interested in purchasing an EV from a Chinese manufacturer?**\n\n**GLOBAL OUTSIDE CHINA** – 60% of respondents would not be interested in purchasing an EV from a Chinese manufacturer\n\nConventional powertrain owners a) EV intenders\n\nConventional powertrain owners b) EV non-intenders\n\nEV or hybrid owners\n\n2 5"],"full_content":"**A R T H U R D . L I T T L E**\n\n### **BEVs & BRAND LOYALTY: THE RISE OF CHINA**\n\nShopping for a BEV may prompt customers to\n\nreconsider the different options and brands\n\navailable to them, potentially reducing brand\n\nloyalty built up over many years in the ICE world.\n\nGlobally, 19% of existing owners of premium\n\nbrand models ⁵ say they would be more likely\n\nto switch, compared to 17% of those who own\n\nvolume brand vehicles. This shift is highest in\n\nIndia (58% premium and 42% volume) and lowest\n\nin Europe (9% premium and 8% volume) and\n\nthe US volume market (3%). However, the US\n\npremium market is being and will continue to be\n\ndisrupted, with over a quarter (28%) more likely\n\nto change.\n\n5 Premium brands include: Volvo, Audi, BMW, Jaguar, Tesla, Lexus, Cadillac, Mercedes, and Nio; volume brands include: Citroën, Fiat, Ford, Hyundai, Jeep, Kia, Nissan, Opel, Peugeot, Renault, SEAT, Skoda, Toyota, Volkswagen, and BYD.\n\nChinese EV manufacturers are rapidly growing\n\nin terms of global presence. Their rising\n\nvisibility (and lower prices) leads to potential\n\ntariffs and regulatory action in Europe and the\n\nUS, including exclusion from some European\n\nnational incentive programs, due to low content\n\nof European components. However, outside\n\nChina more than half (60%) of respondents say\n\nthey were not interested in purchasing a Chinese\n\nEV (see Figure 18), naming perception of quality\n\nand customer sales/service experience as the\n\nmain reasons.\n\n**Figure 17. Factors encouraging EV purchase**\n\nSource: Arthur D. Little\n\n**Environment**\n\n**Trend & technology front-runner**\n\n**Price**\n\n**More fun to drive than ICE cars (acceleration)**\n\nSource: Arthur D. Little\n\n### **Figure 17. Factors encouraging EV purchase**\n\n82\n\n80\n\n**46**\n\n21\n\n79\n\n83\n\n**56**\n\n20\n\n80\n\n79\n\n**36**\n\n16\n\n69\n\n80\n\n**27**\n\n23\n\n70\n\n86\n\n**20**\n\n18\n\n60\n\n75\n\n**15**\n\n15\n\n84\n\n89\n\n**50**\n\n23\n\n81\n\n81\n\n**48**\n\n**34**\n\n71\n\n85\n\n33\n\n20\n\n**What factors would encourage you to purchase an electric car?**\n\n(in %)\n\n68\n\n80\n\n**23**\n\n26\n\n68\n\n81\n\n**21**\n\n26\n\n**52**\n\n72\n\n**12**\n\n17\n\n76\n\n82\n\n37\n\n24\n\n76\n\n85\n\n38\n\n22\n\n71\n\n82\n\n30\n\n18\n\n82\n\n84\n\n**53**\n\n26\n\n79\n\n88\n\n**53**\n\n20\n\n76\n\n85\n\n**37**\n\n18\n\n90\n\n84\n\n40\n\n**29**\n\n89\n\n84\n\n**62**\n\n26\n\n84\n\n90\n\n**41**\n\n25\n\n74\n\n87\n\n33\n\n28\n\n72\n\n87\n\n**28**\n\n18\n\n62\n\n84\n\n26\n\n17\n\nConventional powertrain owners a) EV intenders Conventional powertrain owners b) EV non intenders EV or hybrid owners\n\n**Figure 18. Reasons for not purchasing a Chinese EV**\n\nSource: Arthur D. Little\n\nSource: Arthur D. Little\n\n### **Figure 18. Reasons for not purchasing a Chinese EV**\n\n22%\n\n53%\n\n47%\n\n42%\n\n30%\n\n43%\n\n19%\n\nPoor quality\n\nChallenges associated with service or purchase\n\nBrand reputation\n\nPreference for domestic brands\n\nEnvironmental, social & governance standards\n\nVehicle design\n\n51%\n\n40%\n\n45%\n\n38% 38%\n\n30%\n\n37% 40%\n\n22%\n\n19%\n\n16%\n\n**Why would you not be interested in purchasing an EV from a Chinese manufacturer?**\n\n**GLOBAL OUTSIDE CHINA** – 60% of respondents would not be interested in purchasing an EV from a Chinese manufacturer\n\nConventional powertrain owners a) EV intenders\n\nConventional powertrain owners b) EV non-intenders\n\nEV or hybrid owners\n\n2 5","publish_date":null,"title":"Figure 17. Factors encouraging EV purchase Figure 18. Reasons for not purchasing a Chinese EV","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ADL_Future_of_automotive_mobility_2024_1_p25.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ADL_Future_of_automotive_mobility_2024_1_p25.pdf"},"request":{"example_id":"chart/ADL_Future_of_automotive_mobility_2024_1_p25","product_type":"parse","source_file_path":"ADL_Future_of_automotive_mobility_2024_1_p25.pdf"},"started_at":"2026-11-04T19:46:27.672795"} +{"completed_at":"2026-11-04T19:46:31.899630","latency_in_ms":2471,"output":{"example_id":"chart/Activate_Consulting_Technology_&_Media_Outlook_2026_(10)_p25","grounded_pages":[],"layout_pages":[],"markdown":"# **Super Users are present across all age cohorts, with higher density in younger generations**\n\n### **24**\n\n## **SUPER USERS BY GENERATION** , U.S., 2025, % ADULTS AGED 18+ / % ADULTS AGED 18+ BY GENERATION\n\n### Sources: Activate analysis, Activate 2025 Consumer Technology & Media Research Study (n = 4,026)\n\n## **GEN Z**\n\n### (aged 18-28)\n\n## **MILLENNIALS**\n\n### (aged 29-44)\n\n## **GEN X**\n\n### (aged 45-60)\n\n## **BABY BOOMERS**\n\n### (aged 61-79)\n\n## **TOTAL ADULT POPULATION**\n\n## ALL OTHER USERS\n\n# **SUPER USERS**\n\n# **72%**\n\n# **28%**\n\n# **8%**\n\n# **66%**\n\n# **34%**\n\n# **53%**\n\n# **47%**\n\n# **73%**\n\n# **27%**\n\n# **92%**\n\n# **8%** **8%**\n\n### **SUPER USERS**","pages":[{"line_numbers":[],"markdown":"# **Super Users are present across all age cohorts, with higher density in younger generations**\n\n### **24**\n\n## **SUPER USERS BY GENERATION** , U.S., 2025, % ADULTS AGED 18+ / % ADULTS AGED 18+ BY GENERATION\n\n### Sources: Activate analysis, Activate 2025 Consumer Technology & Media Research Study (n = 4,026)\n\n## **GEN Z**\n\n### (aged 18-28)\n\n## **MILLENNIALS**\n\n### (aged 29-44)\n\n## **GEN X**\n\n### (aged 45-60)\n\n## **BABY BOOMERS**\n\n### (aged 61-79)\n\n## **TOTAL ADULT POPULATION**\n\n## ALL OTHER USERS\n\n# **SUPER USERS**\n\n# **72%**\n\n# **28%**\n\n# **8%**\n\n# **66%**\n\n# **34%**\n\n# **53%**\n\n# **47%**\n\n# **73%**\n\n# **27%**\n\n# **92%**\n\n# **8%** **8%**\n\n### **SUPER USERS**","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"# **Super Users are present across all age cohorts, with higher density in younger generations**\n\n### **24**\n\n## **SUPER USERS BY GENERATION** , U.S., 2025, % ADULTS AGED 18+ / % ADULTS AGED 18+ BY GENERATION\n\n### Sources: Activate analysis, Activate 2025 Consumer Technology & Media Research Study (n = 4,026)\n\n## **GEN Z**\n\n### (aged 18-28)\n\n## **MILLENNIALS**\n\n### (aged 29-44)\n\n## **GEN X**\n\n### (aged 45-60)\n\n## **BABY BOOMERS**\n\n### (aged 61-79)\n\n## **TOTAL ADULT POPULATION**\n\n## ALL OTHER USERS\n\n# **SUPER USERS**\n\n# **72%**\n\n# **28%**\n\n# **8%**\n\n# **66%**\n\n# **34%**\n\n# **53%**\n\n# **47%**\n\n# **73%**\n\n# **27%**\n\n# **92%**\n\n# **8%** **8%**\n\n### **SUPER USERS**","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Activate_Consulting_Technology_%26_Media_Outlook_2026_%2810%29_p25.pdf"]},"response":{"errors":[],"extract_id":"extract_f4998889a385b1ab0e9ff93853958a21","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["# **Super Users are present across all age cohorts, with higher density in younger generations**\n\n### **24**\n\n## **SUPER USERS BY GENERATION** , U.S., 2025, % ADULTS AGED 18+ / % ADULTS AGED 18+ BY GENERATION\n\n### Sources: Activate analysis, Activate 2025 Consumer Technology & Media Research Study (n = 4,026)\n\n## **GEN Z**\n\n### (aged 18-28)\n\n## **MILLENNIALS**\n\n### (aged 29-44)\n\n## **GEN X**\n\n### (aged 45-60)\n\n## **BABY BOOMERS**\n\n### (aged 61-79)\n\n## **TOTAL ADULT POPULATION**\n\n## ALL OTHER USERS\n\n# **SUPER USERS**\n\n# **72%**\n\n# **28%**\n\n# **8%**\n\n# **66%**\n\n# **34%**\n\n# **53%**\n\n# **47%**\n\n# **73%**\n\n# **27%**\n\n# **92%**\n\n# **8%** **8%**\n\n### **SUPER USERS**"],"full_content":"# **Super Users are present across all age cohorts, with higher density in younger generations**\n\n### **24**\n\n## **SUPER USERS BY GENERATION** , U.S., 2025, % ADULTS AGED 18+ / % ADULTS AGED 18+ BY GENERATION\n\n### Sources: Activate analysis, Activate 2025 Consumer Technology & Media Research Study (n = 4,026)\n\n## **GEN Z**\n\n### (aged 18-28)\n\n## **MILLENNIALS**\n\n### (aged 29-44)\n\n## **GEN X**\n\n### (aged 45-60)\n\n## **BABY BOOMERS**\n\n### (aged 61-79)\n\n## **TOTAL ADULT POPULATION**\n\n## ALL OTHER USERS\n\n# **SUPER USERS**\n\n# **72%**\n\n# **28%**\n\n# **8%**\n\n# **66%**\n\n# **34%**\n\n# **53%**\n\n# **47%**\n\n# **73%**\n\n# **27%**\n\n# **92%**\n\n# **8%** **8%**\n\n### **SUPER USERS**","publish_date":null,"title":"Super Users are present across all age cohorts, with higher density in younger generations","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Activate_Consulting_Technology_%26_Media_Outlook_2026_%2810%29_p25.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Activate_Consulting_Technology_%26_Media_Outlook_2026_%2810%29_p25.pdf"},"request":{"example_id":"chart/Activate_Consulting_Technology_&_Media_Outlook_2026_(10)_p25","product_type":"parse","source_file_path":"Activate_Consulting_Technology_&_Media_Outlook_2026_(10)_p25.pdf"},"started_at":"2026-11-04T19:46:29.427892"} +{"completed_at":"2026-11-04T19:46:32.616473","latency_in_ms":1617,"output":{"example_id":"chart/Activate_Consulting_Technology_&_Media_Outlook_2026_(10)_p60","grounded_pages":[],"layout_pages":[],"markdown":"# **We forecast that global consumer video game revenue will reach $214B by 2029**\n\n### **59**\n\n## **CONSUMER VIDEO GAME REVENUE BY REGION** **1** , GLOBAL, 2019 VS. 2021 VS. 2025E VS. 2029E, BILLIONS USD\n\n### 1\\. Excludes hardware and device sales, augmented/virtual reality content, and advertising. Sources: Activate analysis, Newzoo, Omdia, PricewaterhouseCoopers, Statista\n\n## 2019 2021 2025E 2029E\n\n# **$214B**\n\n# **$180B $169B**\n\n# **$136B $155B**\n\n# **$130B $121B**\n\n# **$101B**\n\n# **$59B $50B $48B $35B**\n\n## **10%**\n\n## **16%**\n\n## **2%**\n\n## **1%**\n\n### 2019-2021\n\n# CAGR:\n\n### 2021-2025E\n\n# CAGR:\n\n## **11%**\n\n# **2%**\n\n### **REST OF WORLD**\n\n### **U.S.**\n\n## **4%**\n\n## **4%**\n\n### 2025E-2029E\n\n# CAGR:\n\n# **4%**\n\n### **ACTIVATE FORECAST**\n\n### **VIDEO GAMING**","pages":[{"line_numbers":[],"markdown":"# **We forecast that global consumer video game revenue will reach $214B by 2029**\n\n### **59**\n\n## **CONSUMER VIDEO GAME REVENUE BY REGION** **1** , GLOBAL, 2019 VS. 2021 VS. 2025E VS. 2029E, BILLIONS USD\n\n### 1\\. Excludes hardware and device sales, augmented/virtual reality content, and advertising. Sources: Activate analysis, Newzoo, Omdia, PricewaterhouseCoopers, Statista\n\n## 2019 2021 2025E 2029E\n\n# **$214B**\n\n# **$180B $169B**\n\n# **$136B $155B**\n\n# **$130B $121B**\n\n# **$101B**\n\n# **$59B $50B $48B $35B**\n\n## **10%**\n\n## **16%**\n\n## **2%**\n\n## **1%**\n\n### 2019-2021\n\n# CAGR:\n\n### 2021-2025E\n\n# CAGR:\n\n## **11%**\n\n# **2%**\n\n### **REST OF WORLD**\n\n### **U.S.**\n\n## **4%**\n\n## **4%**\n\n### 2025E-2029E\n\n# CAGR:\n\n# **4%**\n\n### **ACTIVATE FORECAST**\n\n### **VIDEO GAMING**","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"# **We forecast that global consumer video game revenue will reach $214B by 2029**\n\n### **59**\n\n## **CONSUMER VIDEO GAME REVENUE BY REGION** **1** , GLOBAL, 2019 VS. 2021 VS. 2025E VS. 2029E, BILLIONS USD\n\n### 1\\. Excludes hardware and device sales, augmented/virtual reality content, and advertising. Sources: Activate analysis, Newzoo, Omdia, PricewaterhouseCoopers, Statista\n\n## 2019 2021 2025E 2029E\n\n# **$214B**\n\n# **$180B $169B**\n\n# **$136B $155B**\n\n# **$130B $121B**\n\n# **$101B**\n\n# **$59B $50B $48B $35B**\n\n## **10%**\n\n## **16%**\n\n## **2%**\n\n## **1%**\n\n### 2019-2021\n\n# CAGR:\n\n### 2021-2025E\n\n# CAGR:\n\n## **11%**\n\n# **2%**\n\n### **REST OF WORLD**\n\n### **U.S.**\n\n## **4%**\n\n## **4%**\n\n### 2025E-2029E\n\n# CAGR:\n\n# **4%**\n\n### **ACTIVATE FORECAST**\n\n### **VIDEO GAMING**","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Activate_Consulting_Technology_%26_Media_Outlook_2026_%2810%29_p60.pdf"]},"response":{"errors":[],"extract_id":"extract_e0cea712e0d199ae8a72938bfa729861","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["# **We forecast that global consumer video game revenue will reach $214B by 2029**\n\n### **59**\n\n## **CONSUMER VIDEO GAME REVENUE BY REGION** **1** , GLOBAL, 2019 VS. 2021 VS. 2025E VS. 2029E, BILLIONS USD\n\n### 1\\. Excludes hardware and device sales, augmented/virtual reality content, and advertising. Sources: Activate analysis, Newzoo, Omdia, PricewaterhouseCoopers, Statista\n\n## 2019 2021 2025E 2029E\n\n# **$214B**\n\n# **$180B $169B**\n\n# **$136B $155B**\n\n# **$130B $121B**\n\n# **$101B**\n\n# **$59B $50B $48B $35B**\n\n## **10%**\n\n## **16%**\n\n## **2%**\n\n## **1%**\n\n### 2019-2021\n\n# CAGR:\n\n### 2021-2025E\n\n# CAGR:\n\n## **11%**\n\n# **2%**\n\n### **REST OF WORLD**\n\n### **U.S.**\n\n## **4%**\n\n## **4%**\n\n### 2025E-2029E\n\n# CAGR:\n\n# **4%**\n\n### **ACTIVATE FORECAST**\n\n### **VIDEO GAMING**"],"full_content":"# **We forecast that global consumer video game revenue will reach $214B by 2029**\n\n### **59**\n\n## **CONSUMER VIDEO GAME REVENUE BY REGION** **1** , GLOBAL, 2019 VS. 2021 VS. 2025E VS. 2029E, BILLIONS USD\n\n### 1\\. Excludes hardware and device sales, augmented/virtual reality content, and advertising. Sources: Activate analysis, Newzoo, Omdia, PricewaterhouseCoopers, Statista\n\n## 2019 2021 2025E 2029E\n\n# **$214B**\n\n# **$180B $169B**\n\n# **$136B $155B**\n\n# **$130B $121B**\n\n# **$101B**\n\n# **$59B $50B $48B $35B**\n\n## **10%**\n\n## **16%**\n\n## **2%**\n\n## **1%**\n\n### 2019-2021\n\n# CAGR:\n\n### 2021-2025E\n\n# CAGR:\n\n## **11%**\n\n# **2%**\n\n### **REST OF WORLD**\n\n### **U.S.**\n\n## **4%**\n\n## **4%**\n\n### 2025E-2029E\n\n# CAGR:\n\n# **4%**\n\n### **ACTIVATE FORECAST**\n\n### **VIDEO GAMING**","publish_date":null,"title":"We forecast that global consumer video game revenue will reach $214B by 2029","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Activate_Consulting_Technology_%26_Media_Outlook_2026_%2810%29_p60.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Activate_Consulting_Technology_%26_Media_Outlook_2026_%2810%29_p60.pdf"},"request":{"example_id":"chart/Activate_Consulting_Technology_&_Media_Outlook_2026_(10)_p60","product_type":"parse","source_file_path":"Activate_Consulting_Technology_&_Media_Outlook_2026_(10)_p60.pdf"},"started_at":"2026-11-04T19:46:30.999155"} +{"completed_at":"2026-11-04T19:46:33.947002","latency_in_ms":1367,"output":{"example_id":"chart/Activate_Consulting_Technology_&_Media_Outlook_2026_(10)_p65","grounded_pages":[],"layout_pages":[],"markdown":"### **64**\n\n## **SHARE OF GAMERS** **1** **USING EACH PLATFORM FOR GAMING IN THE LAST 12 MONTHS** , U.S., 2025, % GAMERS ¹ BY SEGMENT\n\n# **24%**\n\n# **70%**\n\n# **89%**\n\n# **34%**\n\n# **56%**\n\n# **82% 87% 89% 93%**\n\n## **MOBILE**\n\n## **CONSOLE**\n\n## **PC**\n\n## **SUPER GAMERS** **2** **AVID GAMERS** **3** **CASUAL GAMERS** **4**\n\n### 1\\. “Gamers” are defined as adults aged 18+ who play video games. 2. “Super Gamers” are defined as gamers who view gaming as their primary source of entertainment and follow gaming content online. 3. “Avid Gamers” are defined as gamers who view gaming as one of a few equivalent options for entertainment, or gamers who view gaming as their primary source of entertainment but do not follow gaming content online. 4. “Casual Gamers” are defined as gamers who only view gaming as an entertainment option when other options are not available, or gamers who view gaming as one of a few equivalent options for entertainment but do not follow gaming content online. Sources: Activate analysis, Activate 2025 Consumer Technology & Media Research Study (n = 4,026)\n\n### **VIDEO GAMING**\n\n# **Super Gamers play video games across all platforms, while Avid and Casual Gamers primarily use mobile**","pages":[{"line_numbers":[],"markdown":"### **64**\n\n## **SHARE OF GAMERS** **1** **USING EACH PLATFORM FOR GAMING IN THE LAST 12 MONTHS** , U.S., 2025, % GAMERS ¹ BY SEGMENT\n\n# **24%**\n\n# **70%**\n\n# **89%**\n\n# **34%**\n\n# **56%**\n\n# **82% 87% 89% 93%**\n\n## **MOBILE**\n\n## **CONSOLE**\n\n## **PC**\n\n## **SUPER GAMERS** **2** **AVID GAMERS** **3** **CASUAL GAMERS** **4**\n\n### 1\\. “Gamers” are defined as adults aged 18+ who play video games. 2. “Super Gamers” are defined as gamers who view gaming as their primary source of entertainment and follow gaming content online. 3. “Avid Gamers” are defined as gamers who view gaming as one of a few equivalent options for entertainment, or gamers who view gaming as their primary source of entertainment but do not follow gaming content online. 4. “Casual Gamers” are defined as gamers who only view gaming as an entertainment option when other options are not available, or gamers who view gaming as one of a few equivalent options for entertainment but do not follow gaming content online. Sources: Activate analysis, Activate 2025 Consumer Technology & Media Research Study (n = 4,026)\n\n### **VIDEO GAMING**\n\n# **Super Gamers play video games across all platforms, while Avid and Casual Gamers primarily use mobile**","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"### **64**\n\n## **SHARE OF GAMERS** **1** **USING EACH PLATFORM FOR GAMING IN THE LAST 12 MONTHS** , U.S., 2025, % GAMERS ¹ BY SEGMENT\n\n# **24%**\n\n# **70%**\n\n# **89%**\n\n# **34%**\n\n# **56%**\n\n# **82% 87% 89% 93%**\n\n## **MOBILE**\n\n## **CONSOLE**\n\n## **PC**\n\n## **SUPER GAMERS** **2** **AVID GAMERS** **3** **CASUAL GAMERS** **4**\n\n### 1\\. “Gamers” are defined as adults aged 18+ who play video games. 2. “Super Gamers” are defined as gamers who view gaming as their primary source of entertainment and follow gaming content online. 3. “Avid Gamers” are defined as gamers who view gaming as one of a few equivalent options for entertainment, or gamers who view gaming as their primary source of entertainment but do not follow gaming content online. 4. “Casual Gamers” are defined as gamers who only view gaming as an entertainment option when other options are not available, or gamers who view gaming as one of a few equivalent options for entertainment but do not follow gaming content online. Sources: Activate analysis, Activate 2025 Consumer Technology & Media Research Study (n = 4,026)\n\n### **VIDEO GAMING**\n\n# **Super Gamers play video games across all platforms, while Avid and Casual Gamers primarily use mobile**","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Activate_Consulting_Technology_%26_Media_Outlook_2026_%2810%29_p65.pdf"]},"response":{"errors":[],"extract_id":"extract_0d34c58d4b1a72db1c4a0150a836b7e0","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["### **64**\n\n## **SHARE OF GAMERS** **1** **USING EACH PLATFORM FOR GAMING IN THE LAST 12 MONTHS** , U.S., 2025, % GAMERS ¹ BY SEGMENT\n\n# **24%**\n\n# **70%**\n\n# **89%**\n\n# **34%**\n\n# **56%**\n\n# **82% 87% 89% 93%**\n\n## **MOBILE**\n\n## **CONSOLE**\n\n## **PC**\n\n## **SUPER GAMERS** **2** **AVID GAMERS** **3** **CASUAL GAMERS** **4**\n\n### 1\\. “Gamers” are defined as adults aged 18+ who play video games. 2. “Super Gamers” are defined as gamers who view gaming as their primary source of entertainment and follow gaming content online. 3. “Avid Gamers” are defined as gamers who view gaming as one of a few equivalent options for entertainment, or gamers who view gaming as their primary source of entertainment but do not follow gaming content online. 4. “Casual Gamers” are defined as gamers who only view gaming as an entertainment option when other options are not available, or gamers who view gaming as one of a few equivalent options for entertainment but do not follow gaming content online. Sources: Activate analysis, Activate 2025 Consumer Technology & Media Research Study (n = 4,026)\n\n### **VIDEO GAMING**\n\n# **Super Gamers play video games across all platforms, while Avid and Casual Gamers primarily use mobile**"],"full_content":"### **64**\n\n## **SHARE OF GAMERS** **1** **USING EACH PLATFORM FOR GAMING IN THE LAST 12 MONTHS** , U.S., 2025, % GAMERS ¹ BY SEGMENT\n\n# **24%**\n\n# **70%**\n\n# **89%**\n\n# **34%**\n\n# **56%**\n\n# **82% 87% 89% 93%**\n\n## **MOBILE**\n\n## **CONSOLE**\n\n## **PC**\n\n## **SUPER GAMERS** **2** **AVID GAMERS** **3** **CASUAL GAMERS** **4**\n\n### 1\\. “Gamers” are defined as adults aged 18+ who play video games. 2. “Super Gamers” are defined as gamers who view gaming as their primary source of entertainment and follow gaming content online. 3. “Avid Gamers” are defined as gamers who view gaming as one of a few equivalent options for entertainment, or gamers who view gaming as their primary source of entertainment but do not follow gaming content online. 4. “Casual Gamers” are defined as gamers who only view gaming as an entertainment option when other options are not available, or gamers who view gaming as one of a few equivalent options for entertainment but do not follow gaming content online. Sources: Activate analysis, Activate 2025 Consumer Technology & Media Research Study (n = 4,026)\n\n### **VIDEO GAMING**\n\n# **Super Gamers play video games across all platforms, while Avid and Casual Gamers primarily use mobile**","publish_date":null,"title":"Super Gamers play video games across all platforms, while Avid and Casual Gamers primarily use mobile","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Activate_Consulting_Technology_%26_Media_Outlook_2026_%2810%29_p65.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Activate_Consulting_Technology_%26_Media_Outlook_2026_%2810%29_p65.pdf"},"request":{"example_id":"chart/Activate_Consulting_Technology_&_Media_Outlook_2026_(10)_p65","product_type":"parse","source_file_path":"Activate_Consulting_Technology_&_Media_Outlook_2026_(10)_p65.pdf"},"started_at":"2026-11-04T19:46:32.579598"} +{"completed_at":"2026-11-04T19:46:35.980027","latency_in_ms":1505,"output":{"example_id":"chart/Activate_Consulting_Technology_&_Media_Outlook_2026_(10)_p8","grounded_pages":[],"layout_pages":[],"markdown":"## **Listening to Music**\n\n## **Listening to / Watching Podcasts**\n\n## **Using Social Media**\n\n## **Watching Videos**\n\n## **Playing Video Games**\n\n## **Reading**\n\n# **Multitasking is the time multiplier across all major technology and media behaviors**\n\n### **7**\n\n## **MULTITASKING** **1** **BEHAVIOR BY ACTIVITY** **2** , U.S., 2024 VS. 2025, % ADULTS AGED 18+ WHO ENGAGE IN EACH ACTIVITY\n\n### 1\\. “Multitasking” is defined as simultaneously doing another activity, such as working, cleaning, cooking, or exercising. 2. Figures do not sum due to rounding. Sources: Activate analysis, Activate 2024 Consumer Technology & Media Research Study (n = 4,004), Activate 2025 Consumer Technology & Media Research Study (n = 4,026)\n\n# **_Consumers who multitask_** **_1_** **_all, almost all, or most of the time when…_**\n\n## **2pp 3pp 0pp 0pp 2pp**\n\n## **32%**\n\n## **45% 49%**\n\n## **55%**\n\n## **62%**\n\n## **72%**\n\n## **30%**\n\n## **44% 49% 52%**\n\n## **61%**\n\n## **72%**\n\n## 2024 2025\n\n### **CONSUMER TIME AND ATTENTION**\n\n## **1pp**","pages":[{"line_numbers":[],"markdown":"## **Listening to Music**\n\n## **Listening to / Watching Podcasts**\n\n## **Using Social Media**\n\n## **Watching Videos**\n\n## **Playing Video Games**\n\n## **Reading**\n\n# **Multitasking is the time multiplier across all major technology and media behaviors**\n\n### **7**\n\n## **MULTITASKING** **1** **BEHAVIOR BY ACTIVITY** **2** , U.S., 2024 VS. 2025, % ADULTS AGED 18+ WHO ENGAGE IN EACH ACTIVITY\n\n### 1\\. “Multitasking” is defined as simultaneously doing another activity, such as working, cleaning, cooking, or exercising. 2. Figures do not sum due to rounding. Sources: Activate analysis, Activate 2024 Consumer Technology & Media Research Study (n = 4,004), Activate 2025 Consumer Technology & Media Research Study (n = 4,026)\n\n# **_Consumers who multitask_** **_1_** **_all, almost all, or most of the time when…_**\n\n## **2pp 3pp 0pp 0pp 2pp**\n\n## **32%**\n\n## **45% 49%**\n\n## **55%**\n\n## **62%**\n\n## **72%**\n\n## **30%**\n\n## **44% 49% 52%**\n\n## **61%**\n\n## **72%**\n\n## 2024 2025\n\n### **CONSUMER TIME AND ATTENTION**\n\n## **1pp**","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"## **Listening to Music**\n\n## **Listening to / Watching Podcasts**\n\n## **Using Social Media**\n\n## **Watching Videos**\n\n## **Playing Video Games**\n\n## **Reading**\n\n# **Multitasking is the time multiplier across all major technology and media behaviors**\n\n### **7**\n\n## **MULTITASKING** **1** **BEHAVIOR BY ACTIVITY** **2** , U.S., 2024 VS. 2025, % ADULTS AGED 18+ WHO ENGAGE IN EACH ACTIVITY\n\n### 1\\. “Multitasking” is defined as simultaneously doing another activity, such as working, cleaning, cooking, or exercising. 2. Figures do not sum due to rounding. Sources: Activate analysis, Activate 2024 Consumer Technology & Media Research Study (n = 4,004), Activate 2025 Consumer Technology & Media Research Study (n = 4,026)\n\n# **_Consumers who multitask_** **_1_** **_all, almost all, or most of the time when…_**\n\n## **2pp 3pp 0pp 0pp 2pp**\n\n## **32%**\n\n## **45% 49%**\n\n## **55%**\n\n## **62%**\n\n## **72%**\n\n## **30%**\n\n## **44% 49% 52%**\n\n## **61%**\n\n## **72%**\n\n## 2024 2025\n\n### **CONSUMER TIME AND ATTENTION**\n\n## **1pp**","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Activate_Consulting_Technology_%26_Media_Outlook_2026_%2810%29_p8.pdf"]},"response":{"errors":[],"extract_id":"extract_aa971bfbd3d570b62f206f9442452b83","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["## **Listening to Music**\n\n## **Listening to / Watching Podcasts**\n\n## **Using Social Media**\n\n## **Watching Videos**\n\n## **Playing Video Games**\n\n## **Reading**\n\n# **Multitasking is the time multiplier across all major technology and media behaviors**\n\n### **7**\n\n## **MULTITASKING** **1** **BEHAVIOR BY ACTIVITY** **2** , U.S., 2024 VS. 2025, % ADULTS AGED 18+ WHO ENGAGE IN EACH ACTIVITY\n\n### 1\\. “Multitasking” is defined as simultaneously doing another activity, such as working, cleaning, cooking, or exercising. 2. Figures do not sum due to rounding. Sources: Activate analysis, Activate 2024 Consumer Technology & Media Research Study (n = 4,004), Activate 2025 Consumer Technology & Media Research Study (n = 4,026)\n\n# **_Consumers who multitask_** **_1_** **_all, almost all, or most of the time when…_**\n\n## **2pp 3pp 0pp 0pp 2pp**\n\n## **32%**\n\n## **45% 49%**\n\n## **55%**\n\n## **62%**\n\n## **72%**\n\n## **30%**\n\n## **44% 49% 52%**\n\n## **61%**\n\n## **72%**\n\n## 2024 2025\n\n### **CONSUMER TIME AND ATTENTION**\n\n## **1pp**"],"full_content":"## **Listening to Music**\n\n## **Listening to / Watching Podcasts**\n\n## **Using Social Media**\n\n## **Watching Videos**\n\n## **Playing Video Games**\n\n## **Reading**\n\n# **Multitasking is the time multiplier across all major technology and media behaviors**\n\n### **7**\n\n## **MULTITASKING** **1** **BEHAVIOR BY ACTIVITY** **2** , U.S., 2024 VS. 2025, % ADULTS AGED 18+ WHO ENGAGE IN EACH ACTIVITY\n\n### 1\\. “Multitasking” is defined as simultaneously doing another activity, such as working, cleaning, cooking, or exercising. 2. Figures do not sum due to rounding. Sources: Activate analysis, Activate 2024 Consumer Technology & Media Research Study (n = 4,004), Activate 2025 Consumer Technology & Media Research Study (n = 4,026)\n\n# **_Consumers who multitask_** **_1_** **_all, almost all, or most of the time when…_**\n\n## **2pp 3pp 0pp 0pp 2pp**\n\n## **32%**\n\n## **45% 49%**\n\n## **55%**\n\n## **62%**\n\n## **72%**\n\n## **30%**\n\n## **44% 49% 52%**\n\n## **61%**\n\n## **72%**\n\n## 2024 2025\n\n### **CONSUMER TIME AND ATTENTION**\n\n## **1pp**","publish_date":null,"title":"Multitasking is the time multiplier across all major technology and media behaviors","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Activate_Consulting_Technology_%26_Media_Outlook_2026_%2810%29_p8.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Activate_Consulting_Technology_%26_Media_Outlook_2026_%2810%29_p8.pdf"},"request":{"example_id":"chart/Activate_Consulting_Technology_&_Media_Outlook_2026_(10)_p8","product_type":"parse","source_file_path":"Activate_Consulting_Technology_&_Media_Outlook_2026_(10)_p8.pdf"},"started_at":"2026-11-04T19:46:34.474060"} +{"completed_at":"2026-11-04T19:46:39.887726","latency_in_ms":3905,"output":{"example_id":"chart/BLK_AR24_p8","grounded_pages":[],"layout_pages":[],"markdown":"$0\n\n$20\n\n$40\n\n$60\n\n$80\n\n$100\n\n$120\n\n$140\n\n$0\n\n$5,000\n\n$10,000\n\n$15,000\n\n$20,000\n\n$25,000\n\n$30,000\n\n$35,000\n\n$40,000\n\n$45,000\n\n1988 1993 1998 2003 2008 2013 2018 2023\n\nGlobal GDP ($T)\n\nS&P 500\n\n S&P 500 (11% annualized total return)\n\n Global GDP Growth (5% annualized growth)\n\nThe performance graph is not necessarily indicative of future investment performance.\n\nS&P 500 and global GDP growth ⁵\n\n5\\. Figure 1.1. Source: FactSet, World Bank. Data as of December 31, 2023. S&P 500 return assumes reinvestment of all dividends. The performance graph is not necessarily indicative of future investment performance\n\n6\\. Our World in Data, Data compiled from multiple sources by World Bank (2025); _Bold and van Zanden — Maddison Project Database 2023; Maddison Database 2010_\n\nto the New York Stock Exchange, most financing has come from banks, corporations, and governments — not the capital markets.\n\nWhy banks, corporations, and governments? Because that’s where people put their money. They parked their savings in bank accounts, drove corporate growth through consumption, and paid taxes that fund public spending.\n\nBut when my partners and I founded BlackRock in 1988, we believed the world was changing. The capital markets wouldn’t just supplement banks, corporations, and governments — they’d stand alongside them as a coequal source of capital.\n\nThe logic was simple: Markets delivered better returns than the other three. Better returns would attract more investors. More investors would deepen markets. And deeper markets meant more capital. Plus, asset managers could accelerate this shift through innovation. For BlackRock, that meant first developing better technology to manage risk, then expanding choice and lowering fees through products like exchange-traded funds (ETFs).\n\nWe’ve been fortunate these past 37 years. Our logic panned out. But what’s striking now is how early\n\nwe still are in the story of market expansion. The real payoff is only just beginning.\n\nAs we enter our century’s second quarter, there’s a growing mismatch between the demand for investment and the capital available from traditional sources.\n\nGovernments can’t fund infrastructure through deficits. The deficits can’t get much higher.\n\nInstead, they’ll turn to private investors.\n\nMeanwhile, companies won’t rely solely on banks for credit. Bank lending is constrained. Instead, businesses will go to the markets.\n\nThe money is already there. In fact, more capital is sitting idle today than at any point in my career. In the U.S. alone, roughly $25 trillion is parked in banks and money market funds. ⁷\n\nBut we’re repeating a mistake from the earliest days of finance: Abundant capital. Deployed too narrowly. As one historian wrote, Amsterdam’s first stock exchange “could have made a much greater contribution to the economy” if investors had more companies to invest in. The same is true today. ⁸\n\nAssets that will define the future — data centers, ports, power\n\ngrids, the world’s fastest-growing private companies — aren’t available to most investors. They’re in private markets, locked behind high walls, with gates that open only for the wealthiest or largest market participants.\n\nThe reason for the exclusivity has always been risk. Illiquidity. Complexity. That’s why only certain investors are allowed in.\n\nBut nothing in finance is immutable. Private markets don’t have to be as risky. Or opaque. Or out of reach. Not if the investment industry is willing to innovate — and that’s exactly what we’ve spent the past year doing at BlackRock.\n\nBlackRock has always had a foot in private markets. But we’ve been — first and foremost — a traditional asset manager. That’s who we were at the start of 2024. But it’s not who we are anymore.\n\nIn the past 14 months, we’ve announced the acquisition of two of the top firms in the fastest- growing areas of private markets: infrastructure and private credit. We bought another firm to get better data and analytics, so we can better measure risk, spot opportunities, and unlock access to private markets.\n\nWe’ve transformed our company. The next section details why we did it, how we did it, and why it matters.\n\nThe $68 trillion investment boom: Who will own it?\n\nThroughout history, infrastructure has driven a surprising amount of economic growth. Between 1860 and 1890, just building railroads boosted U.S. GDP by roughly 25%. ⁹ A century later, highways did something similar:\n\nThese investors have benefited from the greatest period of wealth creation in human history. Over the past 40 years, global gross domestic product (GDP) has grown more than in the previous two thousand combined. ⁶ This extraordinary growth — partly propelled, it must be noted, by historically low interest rates — has driven exceptional long-\n\nterm returns. But of course, not everyone has shared in this wealth.\n\nThis extraordinary era of market expansion has coincided with — and was largely fueled by — globalization. And while a flatter world lifted 1 billion people out of $1-a-day poverty, it also held back millions in wealthier nations striving for a better life.\n\nToday, many countries have twin, inverted economies: one where wealth builds on wealth; another where hardship builds on hardship. The divide has reshaped our politics, our policies, even our sense of what’s possible. Protectionism has returned with force. The unspoken assumption is that capitalism didn’t work and it’s time to try something new.\n\nBut there’s another way to look at it: Capitalism did work — just for too few people.\n\nMarkets, like everything humans build, aren’t perfect. They reflect us — unfinished, sometimes flawed, but always improvable. The solution isn’t to abandon markets; it’s to expand them, to finish the market democratization that began 400 years ago and let more\n\npeople own a meaningful stake in the growth happening around them.\n\nIn the pages (or pixels) ahead, I’ll offer some thoughts on how we can further democratize investing in two broad ways:\n\n1\\. By helping current investors access parts of the market they’ve previously been restricted from.\n\n2\\. By enabling more people to become investors in the first place.\n\nI’ll begin with BlackRock’s work, but both of these efforts extend beyond asset management, touching broader issues like retirement policy, energy, and tokenization.\n\nMore investment. More investors. That’s the answer.\n\nGiven that BlackRock is a fiduciary and the largest asset manager in the world, some readers might contend I’m talking my own book. Fair enough. But it’s also the book we intentionally chose — long before it was a bestseller. From the start, we believed that when people can invest better, they can live better — and that’s exactly why we built BlackRock.\n\n## Unlocking private markets\n\nBlackRock’s past 14 months — and the future\n\nEconomies run on capital. Whether you’re assembling 17th-century trading fleets or 21st-century data centers, the money has to come from somewhere. But historically, it hasn’t been investors. Despite 400 years of financial innovation, from Amsterdam to Change Alley\n\n7\\. Source: ICI and The Federal Reserve. Cash in Money Market Funds and Commercial Bank Deposits, as of March 5, 2025\n\n8\\. Lodewijk Petram, _The World’s First Stock Exchange_ , (Columbia University Press, 2014)\n\n9\\. National Bureau of Economic Research, _Railroads, Reallocation, and the Rise of American Manufacturing_ , (2019), pg. 2. Note: The study estimates that U.S. aggregate productivity would have been 25% lower in 1890 in the absence of railroads, with an associated annual loss of $3 billion or 25% of GDP\n\n## BlackRock has always had a foot in private markets. But we’ve been — first and foremost — a traditional asset manager. That’s who we were at the start of 2024. But it’s not who we are anymore.\n\n## We believed the world was changing. The capital markets wouldn’t just supplement banks, corporations, and governments — they’d stand alongside them as a coequal source of capital.\n\nBlackRock 2024 Annual Report 13 12 BlackRock 2024 Annual Report","pages":[{"line_numbers":[],"markdown":"$0\n\n$20\n\n$40\n\n$60\n\n$80\n\n$100\n\n$120\n\n$140\n\n$0\n\n$5,000\n\n$10,000\n\n$15,000\n\n$20,000\n\n$25,000\n\n$30,000\n\n$35,000\n\n$40,000\n\n$45,000\n\n1988 1993 1998 2003 2008 2013 2018 2023\n\nGlobal GDP ($T)\n\nS&P 500\n\n S&P 500 (11% annualized total return)\n\n Global GDP Growth (5% annualized growth)\n\nThe performance graph is not necessarily indicative of future investment performance.\n\nS&P 500 and global GDP growth ⁵\n\n5\\. Figure 1.1. Source: FactSet, World Bank. Data as of December 31, 2023. S&P 500 return assumes reinvestment of all dividends. The performance graph is not necessarily indicative of future investment performance\n\n6\\. Our World in Data, Data compiled from multiple sources by World Bank (2025); _Bold and van Zanden — Maddison Project Database 2023; Maddison Database 2010_\n\nto the New York Stock Exchange, most financing has come from banks, corporations, and governments — not the capital markets.\n\nWhy banks, corporations, and governments? Because that’s where people put their money. They parked their savings in bank accounts, drove corporate growth through consumption, and paid taxes that fund public spending.\n\nBut when my partners and I founded BlackRock in 1988, we believed the world was changing. The capital markets wouldn’t just supplement banks, corporations, and governments — they’d stand alongside them as a coequal source of capital.\n\nThe logic was simple: Markets delivered better returns than the other three. Better returns would attract more investors. More investors would deepen markets. And deeper markets meant more capital. Plus, asset managers could accelerate this shift through innovation. For BlackRock, that meant first developing better technology to manage risk, then expanding choice and lowering fees through products like exchange-traded funds (ETFs).\n\nWe’ve been fortunate these past 37 years. Our logic panned out. But what’s striking now is how early\n\nwe still are in the story of market expansion. The real payoff is only just beginning.\n\nAs we enter our century’s second quarter, there’s a growing mismatch between the demand for investment and the capital available from traditional sources.\n\nGovernments can’t fund infrastructure through deficits. The deficits can’t get much higher.\n\nInstead, they’ll turn to private investors.\n\nMeanwhile, companies won’t rely solely on banks for credit. Bank lending is constrained. Instead, businesses will go to the markets.\n\nThe money is already there. In fact, more capital is sitting idle today than at any point in my career. In the U.S. alone, roughly $25 trillion is parked in banks and money market funds. ⁷\n\nBut we’re repeating a mistake from the earliest days of finance: Abundant capital. Deployed too narrowly. As one historian wrote, Amsterdam’s first stock exchange “could have made a much greater contribution to the economy” if investors had more companies to invest in. The same is true today. ⁸\n\nAssets that will define the future — data centers, ports, power\n\ngrids, the world’s fastest-growing private companies — aren’t available to most investors. They’re in private markets, locked behind high walls, with gates that open only for the wealthiest or largest market participants.\n\nThe reason for the exclusivity has always been risk. Illiquidity. Complexity. That’s why only certain investors are allowed in.\n\nBut nothing in finance is immutable. Private markets don’t have to be as risky. Or opaque. Or out of reach. Not if the investment industry is willing to innovate — and that’s exactly what we’ve spent the past year doing at BlackRock.\n\nBlackRock has always had a foot in private markets. But we’ve been — first and foremost — a traditional asset manager. That’s who we were at the start of 2024. But it’s not who we are anymore.\n\nIn the past 14 months, we’ve announced the acquisition of two of the top firms in the fastest- growing areas of private markets: infrastructure and private credit. We bought another firm to get better data and analytics, so we can better measure risk, spot opportunities, and unlock access to private markets.\n\nWe’ve transformed our company. The next section details why we did it, how we did it, and why it matters.\n\nThe $68 trillion investment boom: Who will own it?\n\nThroughout history, infrastructure has driven a surprising amount of economic growth. Between 1860 and 1890, just building railroads boosted U.S. GDP by roughly 25%. ⁹ A century later, highways did something similar:\n\nThese investors have benefited from the greatest period of wealth creation in human history. Over the past 40 years, global gross domestic product (GDP) has grown more than in the previous two thousand combined. �� This extraordinary growth — partly propelled, it must be noted, by historically low interest rates — has driven exceptional long-\n\nterm returns. But of course, not everyone has shared in this wealth.\n\nThis extraordinary era of market expansion has coincided with — and was largely fueled by — globalization. And while a flatter world lifted 1 billion people out of $1-a-day poverty, it also held back millions in wealthier nations striving for a better life.\n\nToday, many countries have twin, inverted economies: one where wealth builds on wealth; another where hardship builds on hardship. The divide has reshaped our politics, our policies, even our sense of what’s possible. Protectionism has returned with force. The unspoken assumption is that capitalism didn’t work and it’s time to try something new.\n\nBut there’s another way to look at it: Capitalism did work — just for too few people.\n\nMarkets, like everything humans build, aren’t perfect. They reflect us — unfinished, sometimes flawed, but always improvable. The solution isn’t to abandon markets; it’s to expand them, to finish the market democratization that began 400 years ago and let more\n\npeople own a meaningful stake in the growth happening around them.\n\nIn the pages (or pixels) ahead, I’ll offer some thoughts on how we can further democratize investing in two broad ways:\n\n1\\. By helping current investors access parts of the market they’ve previously been restricted from.\n\n2\\. By enabling more people to become investors in the first place.\n\nI’ll begin with BlackRock’s work, but both of these efforts extend beyond asset management, touching broader issues like retirement policy, energy, and tokenization.\n\nMore investment. More investors. That’s the answer.\n\nGiven that BlackRock is a fiduciary and the largest asset manager in the world, some readers might contend I’m talking my own book. Fair enough. But it’s also the book we intentionally chose — long before it was a bestseller. From the start, we believed that when people can invest better, they can live better — and that’s exactly why we built BlackRock.\n\n## Unlocking private markets\n\nBlackRock’s past 14 months — and the future\n\nEconomies run on capital. Whether you’re assembling 17th-century trading fleets or 21st-century data centers, the money has to come from somewhere. But historically, it hasn’t been investors. Despite 400 years of financial innovation, from Amsterdam to Change Alley\n\n7\\. Source: ICI and The Federal Reserve. Cash in Money Market Funds and Commercial Bank Deposits, as of March 5, 2025\n\n8\\. Lodewijk Petram, _The World’s First Stock Exchange_ , (Columbia University Press, 2014)\n\n9\\. National Bureau of Economic Research, _Railroads, Reallocation, and the Rise of American Manufacturing_ , (2019), pg. 2. Note: The study estimates that U.S. aggregate productivity would have been 25% lower in 1890 in the absence of railroads, with an associated annual loss of $3 billion or 25% of GDP\n\n## BlackRock has always had a foot in private markets. But we’ve been — first and foremost — a traditional asset manager. That’s who we were at the start of 2024. But it’s not who we are anymore.\n\n## We believed the world was changing. The capital markets wouldn’t just supplement banks, corporations, and governments — they’d stand alongside them as a coequal source of capital.\n\nBlackRock 2024 Annual Report 13 12 BlackRock 2024 Annual Report","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"$0\n\n$20\n\n$40\n\n$60\n\n$80\n\n$100\n\n$120\n\n$140\n\n$0\n\n$5,000\n\n$10,000\n\n$15,000\n\n$20,000\n\n$25,000\n\n$30,000\n\n$35,000\n\n$40,000\n\n$45,000\n\n1988 1993 1998 2003 2008 2013 2018 2023\n\nGlobal GDP ($T)\n\nS&P 500\n\n S&P 500 (11% annualized total return)\n\n Global GDP Growth (5% annualized growth)\n\nThe performance graph is not necessarily indicative of future investment performance.\n\nS&P 500 and global GDP growth ⁵\n\n5\\. Figure 1.1. Source: FactSet, World Bank. Data as of December 31, 2023. S&P 500 return assumes reinvestment of all dividends. The performance graph is not necessarily indicative of future investment performance\n\n6\\. Our World in Data, Data compiled from multiple sources by World Bank (2025); _Bold and van Zanden — Maddison Project Database 2023; Maddison Database 2010_\n\nto the New York Stock Exchange, most financing has come from banks, corporations, and governments — not the capital markets.\n\nWhy banks, corporations, and governments? Because that’s where people put their money. They parked their savings in bank accounts, drove corporate growth through consumption, and paid taxes that fund public spending.\n\nBut when my partners and I founded BlackRock in 1988, we believed the world was changing. The capital markets wouldn’t just supplement banks, corporations, and governments — they’d stand alongside them as a coequal source of capital.\n\nThe logic was simple: Markets delivered better returns than the other three. Better returns would attract more investors. More investors would deepen markets. And deeper markets meant more capital. Plus, asset managers could accelerate this shift through innovation. For BlackRock, that meant first developing better technology to manage risk, then expanding choice and lowering fees through products like exchange-traded funds (ETFs).\n\nWe’ve been fortunate these past 37 years. Our logic panned out. But what’s striking now is how early\n\nwe still are in the story of market expansion. The real payoff is only just beginning.\n\nAs we enter our century’s second quarter, there’s a growing mismatch between the demand for investment and the capital available from traditional sources.\n\nGovernments can’t fund infrastructure through deficits. The deficits can’t get much higher.\n\nInstead, they’ll turn to private investors.\n\nMeanwhile, companies won’t rely solely on banks for credit. Bank lending is constrained. Instead, businesses will go to the markets.\n\nThe money is already there. In fact, more capital is sitting idle today than at any point in my career. In the U.S. alone, roughly $25 trillion is parked in banks and money market funds. ⁷\n\nBut we’re repeating a mistake from the earliest days of finance: Abundant capital. Deployed too narrowly. As one historian wrote, Amsterdam’s first stock exchange “could have made a much greater contribution to the economy” if investors had more companies to invest in. The same is true today. ⁸\n\nAssets that will define the future — data centers, ports, power\n\ngrids, the world’s fastest-growing private companies — aren’t available to most investors. They’re in private markets, locked behind high walls, with gates that open only for the wealthiest or largest market participants.\n\nThe reason for the exclusivity has always been risk. Illiquidity. Complexity. That’s why only certain investors are allowed in.\n\nBut nothing in finance is immutable. Private markets don’t have to be as risky. Or opaque. Or out of reach. Not if the investment industry is willing to innovate — and that’s exactly what we’ve spent the past year doing at BlackRock.\n\nBlackRock has always had a foot in private markets. But we’ve been — first and foremost — a traditional asset manager. That’s who we were at the start of 2024. But it’s not who we are anymore.\n\nIn the past 14 months, we’ve announced the acquisition of two of the top firms in the fastest- growing areas of private markets: infrastructure and private credit. We bought another firm to get better data and analytics, so we can better measure risk, spot opportunities, and unlock access to private markets.\n\nWe’ve transformed our company. The next section details why we did it, how we did it, and why it matters.\n\nThe $68 trillion investment boom: Who will own it?\n\nThroughout history, infrastructure has driven a surprising amount of economic growth. Between 1860 and 1890, just building railroads boosted U.S. GDP by roughly 25%. ⁹ A century later, highways did something similar:\n\nThese investors have benefited from the greatest period of wealth creation in human history. Over the past 40 years, global gross domestic product (GDP) has grown more than in the previous two thousand combined. ⁶ This extraordinary growth — partly propelled, it must be noted, by historically low interest rates — has driven exceptional long-\n\nterm returns. But of course, not everyone has shared in this wealth.\n\nThis extraordinary era of market expansion has coincided with — and was largely fueled by — globalization. And while a flatter world lifted 1 billion people out of $1-a-day poverty, it also held back millions in wealthier nations striving for a better life.\n\nToday, many countries have twin, inverted economies: one where wealth builds on wealth; another where hardship builds on hardship. The divide has reshaped our politics, our policies, even our sense of what’s possible. Protectionism has returned with force. The unspoken assumption is that capitalism didn’t work and it’s time to try something new.\n\nBut there’s another way to look at it: Capitalism did work — just for too few people.\n\nMarkets, like everything humans build, aren’t perfect. They reflect us — unfinished, sometimes flawed, but always improvable. The solution isn’t to abandon markets; it’s to expand them, to finish the market democratization that began 400 years ago and let more\n\npeople own a meaningful stake in the growth happening around them.\n\nIn the pages (or pixels) ahead, I’ll offer some thoughts on how we can further democratize investing in two broad ways:\n\n1\\. By helping current investors access parts of the market they’ve previously been restricted from.\n\n2\\. By enabling more people to become investors in the first place.\n\nI’ll begin with BlackRock’s work, but both of these efforts extend beyond asset management, touching broader issues like retirement policy, energy, and tokenization.\n\nMore investment. More investors. That’s the answer.\n\nGiven that BlackRock is a fiduciary and the largest asset manager in the world, some readers might contend I’m talking my own book. Fair enough. But it’s also the book we intentionally chose — long before it was a bestseller. From the start, we believed that when people can invest better, they can live better — and that’s exactly why we built BlackRock.\n\n## Unlocking private markets\n\nBlackRock’s past 14 months — and the future\n\nEconomies run on capital. Whether you’re assembling 17th-century trading fleets or 21st-century data centers, the money has to come from somewhere. But historically, it hasn’t been investors. Despite 400 years of financial innovation, from Amsterdam to Change Alley\n\n7\\. Source: ICI and The Federal Reserve. Cash in Money Market Funds and Commercial Bank Deposits, as of March 5, 2025\n\n8\\. Lodewijk Petram, _The World’s First Stock Exchange_ , (Columbia University Press, 2014)\n\n9\\. National Bureau of Economic Research, _Railroads, Reallocation, and the Rise of American Manufacturing_ , (2019), pg. 2. Note: The study estimates that U.S. aggregate productivity would have been 25% lower in 1890 in the absence of railroads, with an associated annual loss of $3 billion or 25% of GDP\n\n## BlackRock has always had a foot in private markets. But we’ve been — first and foremost — a traditional asset manager. That’s who we were at the start of 2024. But it’s not who we are anymore.\n\n## We believed the world was changing. The capital markets wouldn’t just supplement banks, corporations, and governments — they’d stand alongside them as a coequal source of capital.\n\nBlackRock 2024 Annual Report 13 12 BlackRock 2024 Annual Report","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/BLK_AR24_p8.pdf"]},"response":{"errors":[],"extract_id":"extract_24e10b7793b7f6ee9d97a9d88d150ab6","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["$0\n\n$20\n\n$40\n\n$60\n\n$80\n\n$100\n\n$120\n\n$140\n\n$0\n\n$5,000\n\n$10,000\n\n$15,000\n\n$20,000\n\n$25,000\n\n$30,000\n\n$35,000\n\n$40,000\n\n$45,000\n\n1988 1993 1998 2003 2008 2013 2018 2023\n\nGlobal GDP ($T)\n\nS&P 500\n\n S&P 500 (11% annualized total return)\n\n Global GDP Growth (5% annualized growth)\n\nThe performance graph is not necessarily indicative of future investment performance.\n\nS&P 500 and global GDP growth ⁵\n\n5\\. Figure 1.1. Source: FactSet, World Bank. Data as of December 31, 2023. S&P 500 return assumes reinvestment of all dividends. The performance graph is not necessarily indicative of future investment performance\n\n6\\. Our World in Data, Data compiled from multiple sources by World Bank (2025); _Bold and van Zanden — Maddison Project Database 2023; Maddison Database 2010_\n\nto the New York Stock Exchange, most financing has come from banks, corporations, and governments — not the capital markets.\n\nWhy banks, corporations, and governments? Because that’s where people put their money. They parked their savings in bank accounts, drove corporate growth through consumption, and paid taxes that fund public spending.\n\nBut when my partners and I founded BlackRock in 1988, we believed the world was changing. The capital markets wouldn’t just supplement banks, corporations, and governments — they’d stand alongside them as a coequal source of capital.\n\nThe logic was simple: Markets delivered better returns than the other three. Better returns would attract more investors. More investors would deepen markets. And deeper markets meant more capital. Plus, asset managers could accelerate this shift through innovation. For BlackRock, that meant first developing better technology to manage risk, then expanding choice and lowering fees through products like exchange-traded funds (ETFs).\n\nWe’ve been fortunate these past 37 years. Our logic panned out. But what’s striking now is how early\n\nwe still are in the story of market expansion. The real payoff is only just beginning.\n\nAs we enter our century’s second quarter, there’s a growing mismatch between the demand for investment and the capital available from traditional sources.\n\nGovernments can’t fund infrastructure through deficits. The deficits can’t get much higher.\n\nInstead, they’ll turn to private investors.\n\nMeanwhile, companies won’t rely solely on banks for credit. Bank lending is constrained. Instead, businesses will go to the markets.\n\nThe money is already there. In fact, more capital is sitting idle today than at any point in my career. In the U.S. alone, roughly $25 trillion is parked in banks and money market funds. ⁷\n\nBut we’re repeating a mistake from the earliest days of finance: Abundant capital. Deployed too narrowly. As one historian wrote, Amsterdam’s first stock exchange “could have made a much greater contribution to the economy” if investors had more companies to invest in. The same is true today. ⁸\n\nAssets that will define the future — data centers, ports, power\n\ngrids, the world’s fastest-growing private companies — aren’t available to most investors. They’re in private markets, locked behind high walls, with gates that open only for the wealthiest or largest market participants.\n\nThe reason for the exclusivity has always been risk. Illiquidity. Complexity. That’s why only certain investors are allowed in.\n\nBut nothing in finance is immutable. Private markets don’t have to be as risky. Or opaque. Or out of reach. Not if the investment industry is willing to innovate — and that’s exactly what we’ve spent the past year doing at BlackRock.\n\nBlackRock has always had a foot in private markets. But we’ve been — first and foremost — a traditional asset manager. That’s who we were at the start of 2024. But it’s not who we are anymore.\n\nIn the past 14 months, we’ve announced the acquisition of two of the top firms in the fastest- growing areas of private markets: infrastructure and private credit. We bought another firm to get better data and analytics, so we can better measure risk, spot opportunities, and unlock access to private markets.\n\nWe’ve transformed our company. The next section details why we did it, how we did it, and why it matters.\n\nThe $68 trillion investment boom: Who will own it?\n\nThroughout history, infrastructure has driven a surprising amount of economic growth. Between 1860 and 1890, just building railroads boosted U.S. GDP by roughly 25%. ⁹ A century later, highways did something similar:\n\nThese investors have benefited from the greatest period of wealth creation in human history. Over the past 40 years, global gross domestic product (GDP) has grown more than in the previous two thousand combined. ⁶ This extraordinary growth — partly propelled, it must be noted, by historically low interest rates — has driven exceptional long-\n\nterm returns. But of course, not everyone has shared in this wealth.\n\nThis extraordinary era of market expansion has coincided with — and was largely fueled by — globalization. And while a flatter world lifted 1 billion people out of $1-a-day poverty, it also held back millions in wealthier nations striving for a better life.\n\nToday, many countries have twin, inverted economies: one where wealth builds on wealth; another where hardship builds on hardship. The divide has reshaped our politics, our policies, even our sense of what’s possible. Protectionism has returned with force. The unspoken assumption is that capitalism didn’t work and it’s time to try something new.\n\nBut there’s another way to look at it: Capitalism did work — just for too few people.\n\nMarkets, like everything humans build, aren’t perfect. They reflect us — unfinished, sometimes flawed, but always improvable. The solution isn’t to abandon markets; it’s to expand them, to finish the market democratization that began 400 years ago and let more\n\npeople own a meaningful stake in the growth happening around them.\n\nIn the pages (or pixels) ahead, I’ll offer some thoughts on how we can further democratize investing in two broad ways:\n\n1\\. By helping current investors access parts of the market they’ve previously been restricted from.\n\n2\\. By enabling more people to become investors in the first place.\n\nI’ll begin with BlackRock’s work, but both of these efforts extend beyond asset management, touching broader issues like retirement policy, energy, and tokenization.\n\nMore investment. More investors. That’s the answer.\n\nGiven that BlackRock is a fiduciary and the largest asset manager in the world, some readers might contend I’m talking my own book. Fair enough. But it’s also the book we intentionally chose — long before it was a bestseller. From the start, we believed that when people can invest better, they can live better — and that’s exactly why we built BlackRock.\n\n## Unlocking private markets\n\nBlackRock’s past 14 months — and the future\n\nEconomies run on capital. Whether you’re assembling 17th-century trading fleets or 21st-century data centers, the money has to come from somewhere. But historically, it hasn’t been investors. Despite 400 years of financial innovation, from Amsterdam to Change Alley\n\n7\\. Source: ICI and The Federal Reserve. Cash in Money Market Funds and Commercial Bank Deposits, as of March 5, 2025\n\n8\\. Lodewijk Petram, _The World’s First Stock Exchange_ , (Columbia University Press, 2014)\n\n9\\. National Bureau of Economic Research, _Railroads, Reallocation, and the Rise of American Manufacturing_ , (2019), pg. 2. Note: The study estimates that U.S. aggregate productivity would have been 25% lower in 1890 in the absence of railroads, with an associated annual loss of $3 billion or 25% of GDP\n\n## BlackRock has always had a foot in private markets. But we’ve been — first and foremost — a traditional asset manager. That’s who we were at the start of 2024. But it’s not who we are anymore.\n\n## We believed the world was changing. The capital markets wouldn’t just supplement banks, corporations, and governments — they’d stand alongside them as a coequal source of capital.\n\nBlackRock 2024 Annual Report 13 12 BlackRock 2024 Annual Report"],"full_content":"$0\n\n$20\n\n$40\n\n$60\n\n$80\n\n$100\n\n$120\n\n$140\n\n$0\n\n$5,000\n\n$10,000\n\n$15,000\n\n$20,000\n\n$25,000\n\n$30,000\n\n$35,000\n\n$40,000\n\n$45,000\n\n1988 1993 1998 2003 2008 2013 2018 2023\n\nGlobal GDP ($T)\n\nS&P 500\n\n S&P 500 (11% annualized total return)\n\n Global GDP Growth (5% annualized growth)\n\nThe performance graph is not necessarily indicative of future investment performance.\n\nS&P 500 and global GDP growth ⁵\n\n5\\. Figure 1.1. Source: FactSet, World Bank. Data as of December 31, 2023. S&P 500 return assumes reinvestment of all dividends. The performance graph is not necessarily indicative of future investment performance\n\n6\\. Our World in Data, Data compiled from multiple sources by World Bank (2025); _Bold and van Zanden — Maddison Project Database 2023; Maddison Database 2010_\n\nto the New York Stock Exchange, most financing has come from banks, corporations, and governments — not the capital markets.\n\nWhy banks, corporations, and governments? Because that’s where people put their money. They parked their savings in bank accounts, drove corporate growth through consumption, and paid taxes that fund public spending.\n\nBut when my partners and I founded BlackRock in 1988, we believed the world was changing. The capital markets wouldn’t just supplement banks, corporations, and governments — they’d stand alongside them as a coequal source of capital.\n\nThe logic was simple: Markets delivered better returns than the other three. Better returns would attract more investors. More investors would deepen markets. And deeper markets meant more capital. Plus, asset managers could accelerate this shift through innovation. For BlackRock, that meant first developing better technology to manage risk, then expanding choice and lowering fees through products like exchange-traded funds (ETFs).\n\nWe’ve been fortunate these past 37 years. Our logic panned out. But what’s striking now is how early\n\nwe still are in the story of market expansion. The real payoff is only just beginning.\n\nAs we enter our century’s second quarter, there’s a growing mismatch between the demand for investment and the capital available from traditional sources.\n\nGovernments can’t fund infrastructure through deficits. The deficits can’t get much higher.\n\nInstead, they’ll turn to private investors.\n\nMeanwhile, companies won’t rely solely on banks for credit. Bank lending is constrained. Instead, businesses will go to the markets.\n\nThe money is already there. In fact, more capital is sitting idle today than at any point in my career. In the U.S. alone, roughly $25 trillion is parked in banks and money market funds. ⁷\n\nBut we’re repeating a mistake from the earliest days of finance: Abundant capital. Deployed too narrowly. As one historian wrote, Amsterdam’s first stock exchange “could have made a much greater contribution to the economy” if investors had more companies to invest in. The same is true today. ⁸\n\nAssets that will define the future — data centers, ports, power\n\ngrids, the world’s fastest-growing private companies — aren’t available to most investors. They’re in private markets, locked behind high walls, with gates that open only for the wealthiest or largest market participants.\n\nThe reason for the exclusivity has always been risk. Illiquidity. Complexity. That’s why only certain investors are allowed in.\n\nBut nothing in finance is immutable. Private markets don’t have to be as risky. Or opaque. Or out of reach. Not if the investment industry is willing to innovate — and that’s exactly what we’ve spent the past year doing at BlackRock.\n\nBlackRock has always had a foot in private markets. But we’ve been — first and foremost — a traditional asset manager. That’s who we were at the start of 2024. But it’s not who we are anymore.\n\nIn the past 14 months, we’ve announced the acquisition of two of the top firms in the fastest- growing areas of private markets: infrastructure and private credit. We bought another firm to get better data and analytics, so we can better measure risk, spot opportunities, and unlock access to private markets.\n\nWe’ve transformed our company. The next section details why we did it, how we did it, and why it matters.\n\nThe $68 trillion investment boom: Who will own it?\n\nThroughout history, infrastructure has driven a surprising amount of economic growth. Between 1860 and 1890, just building railroads boosted U.S. GDP by roughly 25%. ⁹ A century later, highways did something similar:\n\nThese investors have benefited from the greatest period of wealth creation in human history. Over the past 40 years, global gross domestic product (GDP) has grown more than in the previous two thousand combined. ⁶ This extraordinary growth — partly propelled, it must be noted, by historically low interest rates — has driven exceptional long-\n\nterm returns. But of course, not everyone has shared in this wealth.\n\nThis extraordinary era of market expansion has coincided with — and was largely fueled by — globalization. And while a flatter world lifted 1 billion people out of $1-a-day poverty, it also held back millions in wealthier nations striving for a better life.\n\nToday, many countries have twin, inverted economies: one where wealth builds on wealth; another where hardship builds on hardship. The divide has reshaped our politics, our policies, even our sense of what’s possible. Protectionism has returned with force. The unspoken assumption is that capitalism didn’t work and it’s time to try something new.\n\nBut there’s another way to look at it: Capitalism did work — just for too few people.\n\nMarkets, like everything humans build, aren’t perfect. They reflect us — unfinished, sometimes flawed, but always improvable. The solution isn’t to abandon markets; it’s to expand them, to finish the market democratization that began 400 years ago and let more\n\npeople own a meaningful stake in the growth happening around them.\n\nIn the pages (or pixels) ahead, I’ll offer some thoughts on how we can further democratize investing in two broad ways:\n\n1\\. By helping current investors access parts of the market they’ve previously been restricted from.\n\n2\\. By enabling more people to become investors in the first place.\n\nI’ll begin with BlackRock’s work, but both of these efforts extend beyond asset management, touching broader issues like retirement policy, energy, and tokenization.\n\nMore investment. More investors. That’s the answer.\n\nGiven that BlackRock is a fiduciary and the largest asset manager in the world, some readers might contend I’m talking my own book. Fair enough. But it’s also the book we intentionally chose — long before it was a bestseller. From the start, we believed that when people can invest better, they can live better — and that’s exactly why we built BlackRock.\n\n## Unlocking private markets\n\nBlackRock’s past 14 months — and the future\n\nEconomies run on capital. Whether you’re assembling 17th-century trading fleets or 21st-century data centers, the money has to come from somewhere. But historically, it hasn’t been investors. Despite 400 years of financial innovation, from Amsterdam to Change Alley\n\n7\\. Source: ICI and The Federal Reserve. Cash in Money Market Funds and Commercial Bank Deposits, as of March 5, 2025\n\n8\\. Lodewijk Petram, _The World’s First Stock Exchange_ , (Columbia University Press, 2014)\n\n9\\. National Bureau of Economic Research, _Railroads, Reallocation, and the Rise of American Manufacturing_ , (2019), pg. 2. Note: The study estimates that U.S. aggregate productivity would have been 25% lower in 1890 in the absence of railroads, with an associated annual loss of $3 billion or 25% of GDP\n\n## BlackRock has always had a foot in private markets. But we’ve been — first and foremost — a traditional asset manager. That’s who we were at the start of 2024. But it’s not who we are anymore.\n\n## We believed the world was changing. The capital markets wouldn’t just supplement banks, corporations, and governments — they’d stand alongside them as a coequal source of capital.\n\nBlackRock 2024 Annual Report 13 12 BlackRock 2024 Annual Report","publish_date":null,"title":"BlackRock has always had a foot in private markets. But we’ve been — first and foremost — a traditional asset","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/BLK_AR24_p8.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/BLK_AR24_p8.pdf"},"request":{"example_id":"chart/BLK_AR24_p8","product_type":"parse","source_file_path":"BLK_AR24_p8.pdf"},"started_at":"2026-11-04T19:46:35.981913"} +{"completed_at":"2026-11-04T19:46:39.086693","latency_in_ms":1471,"output":{"example_id":"chart/COM_2025_26_F1_COMMUNICATION_FROM_COMMISSION_TO_INST_EN_V10_P1_3877708_p19","grounded_pages":[],"layout_pages":[],"markdown":"17\n\n### **Compared to other advanced economies, a much smaller share of EU private savings goes directly into investment in companies and riskier forms of investment.** Productive\n\ninvestment is low and private saving is high ⁹⁴ . The share of EU household savings going\n\n### towards company bonds, listed shares, investment funds and similar is relatively low,\n\ncorresponding to 43% of the levels of savings in traditional bank accounts (KPI 15) ⁹⁵ . The corresponding level in the UK is 55% and in the US 72% (see Figure 4) ⁹⁶ . High participation\n\n### rates in financial markets help companies to diversify their funding. Strong capital markets are\n\nimportant to ease companies access to finance and unlock finance for scale-ups ⁹⁷ .\n\n### **Significant amounts of EU savers’ money are either tied up in bank accounts or invested abroad.** Although the volume of private savings in the EU was nearly 65%\n\ngreater than in the US in 2022, EU\n\n### households have considerably lower total\n\nwealth than their US peers ⁹⁸ , largely because\n\n### of the lower returns they generally receive from financial markets. While the net wealth of US households grew by around 150% over the past 15 years, the corresponding growth\n\nwas only 55% in the Euro area ⁹⁹ . This is\n\n### largely due to a lack of capacity in the EU financial system to stimulate high-yielding investments, which can be explained by a combination of lack of fiscal incentives and cumbersome tax reporting requirements in many Member States, along with a public perception of the overall business climate not being sufficiently promising, lowering the confidence in the prospects of return on investment. Of the EU household savings invested in company bonds, listed shares and investment funds, a significant share, approximately EUR 300 billion annually, is invested not\n\nin the EU, but abroad ¹⁰⁰ , and mainly in the US. EU foreign direct investment into the US simultaneously amounted to EUR 3.27 trillion in 2023 ¹⁰¹ .\n\n### **New bank loans to SMEs have decreased since the COVID-19 pandemic, putting new investments at risk** . Although a higher share of financing to companies channelled through corporate bonds, listed shares, venture capital and investment funds would be desirable, bank financing remains crucial to fostering growth and competitiveness of the bulk of European SMEs who use traditional bank loans to finance investments (counting for 57% of their total finance, see Figure 5). However, SME lending, which spiked following the on-set of the\n\n94 Productive investment is here defined as gross fixed capital formation minus residential investment.\n\n95 Note: Part of these savings will contribute to corporate investment activity via the portfolio allocation made by banks and other financial intermediaries.\n\n96 Note: The Commission’s set of indicators on the Capital Markets Union provide detailed indicators that help track capital markets’ developments. List of indicators to monitor progress towards the CMU objectives.\n\n97 European Investment Bank (EIB), The scale-up gap.\n\n98 Former Prime Minister Draghi, “The future of European competitiveness – A competitiveness strategy for Europe”, part B, page 1.\n\n99 ECB Distributional Wealth Accounts; US Federal Reserve Economic Data.\n\n100 Former Prime Minister Letta, “Much more than a Market”, page 11. 101 Statista, Foreign direct investment from Europe into the United States from 2000 to 2023.\n\n### **Figure 4:** Household savings invested in bonds, shares, investment funds and similar.\n\n**Note:** Volumes of household savings in company bonds; listed shares; and investment, insurance and pension funds, relative to the volumes of households’ cash holdings and bank deposits.\n\n**Source:** European Commission, DG FISMA","pages":[{"line_numbers":[],"markdown":"17\n\n### **Compared to other advanced economies, a much smaller share of EU private savings goes directly into investment in companies and riskier forms of investment.** Productive\n\ninvestment is low and private saving is high ⁹⁴ . The share of EU household savings going\n\n### towards company bonds, listed shares, investment funds and similar is relatively low,\n\ncorresponding to 43% of the levels of savings in traditional bank accounts (KPI 15) ⁹⁵ . The corresponding level in the UK is 55% and in the US 72% (see Figure 4) ⁹⁶ . High participation\n\n### rates in financial markets help companies to diversify their funding. Strong capital markets are\n\nimportant to ease companies access to finance and unlock finance for scale-ups ⁹⁷ .\n\n### **Significant amounts of EU savers’ money are either tied up in bank accounts or invested abroad.** Although the volume of private savings in the EU was nearly 65%\n\ngreater than in the US in 2022, EU\n\n### households have considerably lower total\n\nwealth than their US peers ⁹⁸ , largely because\n\n### of the lower returns they generally receive from financial markets. While the net wealth of US households grew by around 150% over the past 15 years, the corresponding growth\n\nwas only 55% in the Euro area ⁹⁹ . This is\n\n### largely due to a lack of capacity in the EU financial system to stimulate high-yielding investments, which can be explained by a combination of lack of fiscal incentives and cumbersome tax reporting requirements in many Member States, along with a public perception of the overall business climate not being sufficiently promising, lowering the confidence in the prospects of return on investment. Of the EU household savings invested in company bonds, listed shares and investment funds, a significant share, approximately EUR 300 billion annually, is invested not\n\nin the EU, but abroad ¹⁰⁰ , and mainly in the US. EU foreign direct investment into the US simultaneously amounted to EUR 3.27 trillion in 2023 ¹⁰¹ .\n\n### **New bank loans to SMEs have decreased since the COVID-19 pandemic, putting new investments at risk** . Although a higher share of financing to companies channelled through corporate bonds, listed shares, venture capital and investment funds would be desirable, bank financing remains crucial to fostering growth and competitiveness of the bulk of European SMEs who use traditional bank loans to finance investments (counting for 57% of their total finance, see Figure 5). However, SME lending, which spiked following the on-set of the\n\n94 Productive investment is here defined as gross fixed capital formation minus residential investment.\n\n95 Note: Part of these savings will contribute to corporate investment activity via the portfolio allocation made by banks and other financial intermediaries.\n\n96 Note: The Commission’s set of indicators on the Capital Markets Union provide detailed indicators that help track capital markets’ developments. List of indicators to monitor progress towards the CMU objectives.\n\n97 European Investment Bank (EIB), The scale-up gap.\n\n98 Former Prime Minister Draghi, “The future of European competitiveness – A competitiveness strategy for Europe”, part B, page 1.\n\n99 ECB Distributional Wealth Accounts; US Federal Reserve Economic Data.\n\n100 Former Prime Minister Letta, “Much more than a Market”, page 11. 101 Statista, Foreign direct investment from Europe into the United States from 2000 to 2023.\n\n### **Figure 4:** Household savings invested in bonds, shares, investment funds and similar.\n\n**Note:** Volumes of household savings in company bonds; listed shares; and investment, insurance and pension funds, relative to the volumes of households’ cash holdings and bank deposits.\n\n**Source:** European Commission, DG FISMA","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"17\n\n### **Compared to other advanced economies, a much smaller share of EU private savings goes directly into investment in companies and riskier forms of investment.** Productive\n\ninvestment is low and private saving is high ⁹⁴ . The share of EU household savings going\n\n### towards company bonds, listed shares, investment funds and similar is relatively low,\n\ncorresponding to 43% of the levels of savings in traditional bank accounts (KPI 15) ⁹⁵ . The corresponding level in the UK is 55% and in the US 72% (see Figure 4) ⁹⁶ . High participation\n\n### rates in financial markets help companies to diversify their funding. Strong capital markets are\n\nimportant to ease companies access to finance and unlock finance for scale-ups ⁹⁷ .\n\n### **Significant amounts of EU savers’ money are either tied up in bank accounts or invested abroad.** Although the volume of private savings in the EU was nearly 65%\n\ngreater than in the US in 2022, EU\n\n### households have considerably lower total\n\nwealth than their US peers ⁹⁸ , largely because\n\n### of the lower returns they generally receive from financial markets. While the net wealth of US households grew by around 150% over the past 15 years, the corresponding growth\n\nwas only 55% in the Euro area ⁹⁹ . This is\n\n### largely due to a lack of capacity in the EU financial system to stimulate high-yielding investments, which can be explained by a combination of lack of fiscal incentives and cumbersome tax reporting requirements in many Member States, along with a public perception of the overall business climate not being sufficiently promising, lowering the confidence in the prospects of return on investment. Of the EU household savings invested in company bonds, listed shares and investment funds, a significant share, approximately EUR 300 billion annually, is invested not\n\nin the EU, but abroad ¹⁰⁰ , and mainly in the US. EU foreign direct investment into the US simultaneously amounted to EUR 3.27 trillion in 2023 ¹⁰¹ .\n\n### **New bank loans to SMEs have decreased since the COVID-19 pandemic, putting new investments at risk** . Although a higher share of financing to companies channelled through corporate bonds, listed shares, venture capital and investment funds would be desirable, bank financing remains crucial to fostering growth and competitiveness of the bulk of European SMEs who use traditional bank loans to finance investments (counting for 57% of their total finance, see Figure 5). However, SME lending, which spiked following the on-set of the\n\n94 Productive investment is here defined as gross fixed capital formation minus residential investment.\n\n95 Note: Part of these savings will contribute to corporate investment activity via the portfolio allocation made by banks and other financial intermediaries.\n\n96 Note: The Commission’s set of indicators on the Capital Markets Union provide detailed indicators that help track capital markets’ developments. List of indicators to monitor progress towards the CMU objectives.\n\n97 European Investment Bank (EIB), The scale-up gap.\n\n98 Former Prime Minister Draghi, “The future of European competitiveness – A competitiveness strategy for Europe”, part B, page 1.\n\n99 ECB Distributional Wealth Accounts; US Federal Reserve Economic Data.\n\n100 Former Prime Minister Letta, “Much more than a Market”, page 11. 101 Statista, Foreign direct investment from Europe into the United States from 2000 to 2023.\n\n### **Figure 4:** Household savings invested in bonds, shares, investment funds and similar.\n\n**Note:** Volumes of household savings in company bonds; listed shares; and investment, insurance and pension funds, relative to the volumes of households’ cash holdings and bank deposits.\n\n**Source:** European Commission, DG FISMA","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/COM_2025_26_F1_COMMUNICATION_FROM_COMMISSION_TO_INST_EN_V10_P1_3877708_p19.pdf"]},"response":{"errors":[],"extract_id":"extract_00f5b69e69a8b9dd3d9d8c5868eafd1e","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["17\n\n### **Compared to other advanced economies, a much smaller share of EU private savings goes directly into investment in companies and riskier forms of investment.** Productive\n\ninvestment is low and private saving is high ⁹⁴ . The share of EU household savings going\n\n### towards company bonds, listed shares, investment funds and similar is relatively low,\n\ncorresponding to 43% of the levels of savings in traditional bank accounts (KPI 15) ⁹⁵ . The corresponding level in the UK is 55% and in the US 72% (see Figure 4) ⁹⁶ . High participation\n\n### rates in financial markets help companies to diversify their funding. Strong capital markets are\n\nimportant to ease companies access to finance and unlock finance for scale-ups ⁹⁷ .\n\n### **Significant amounts of EU savers’ money are either tied up in bank accounts or invested abroad.** Although the volume of private savings in the EU was nearly 65%\n\ngreater than in the US in 2022, EU\n\n### households have considerably lower total\n\nwealth than their US peers ⁹⁸ , largely because\n\n### of the lower returns they generally receive from financial markets. While the net wealth of US households grew by around 150% over the past 15 years, the corresponding growth\n\nwas only 55% in the Euro area ⁹⁹ . This is\n\n### largely due to a lack of capacity in the EU financial system to stimulate high-yielding investments, which can be explained by a combination of lack of fiscal incentives and cumbersome tax reporting requirements in many Member States, along with a public perception of the overall business climate not being sufficiently promising, lowering the confidence in the prospects of return on investment. Of the EU household savings invested in company bonds, listed shares and investment funds, a significant share, approximately EUR 300 billion annually, is invested not\n\nin the EU, but abroad ¹⁰⁰ , and mainly in the US. EU foreign direct investment into the US simultaneously amounted to EUR 3.27 trillion in 2023 ¹⁰¹ .\n\n### **New bank loans to SMEs have decreased since the COVID-19 pandemic, putting new investments at risk** . Although a higher share of financing to companies channelled through corporate bonds, listed shares, venture capital and investment funds would be desirable, bank financing remains crucial to fostering growth and competitiveness of the bulk of European SMEs who use traditional bank loans to finance investments (counting for 57% of their total finance, see Figure 5). However, SME lending, which spiked following the on-set of the\n\n94 Productive investment is here defined as gross fixed capital formation minus residential investment.\n\n95 Note: Part of these savings will contribute to corporate investment activity via the portfolio allocation made by banks and other financial intermediaries.\n\n96 Note: The Commission’s set of indicators on the Capital Markets Union provide detailed indicators that help track capital markets’ developments. List of indicators to monitor progress towards the CMU objectives.\n\n97 European Investment Bank (EIB), The scale-up gap.\n\n98 Former Prime Minister Draghi, “The future of European competitiveness – A competitiveness strategy for Europe”, part B, page 1.\n\n99 ECB Distributional Wealth Accounts; US Federal Reserve Economic Data.\n\n100 Former Prime Minister Letta, “Much more than a Market”, page 11. 101 Statista, Foreign direct investment from Europe into the United States from 2000 to 2023.\n\n### **Figure 4:** Household savings invested in bonds, shares, investment funds and similar.\n\n**Note:** Volumes of household savings in company bonds; listed shares; and investment, insurance and pension funds, relative to the volumes of households’ cash holdings and bank deposits.\n\n**Source:** European Commission, DG FISMA"],"full_content":"17\n\n### **Compared to other advanced economies, a much smaller share of EU private savings goes directly into investment in companies and riskier forms of investment.** Productive\n\ninvestment is low and private saving is high ⁹⁴ . The share of EU household savings going\n\n### towards company bonds, listed shares, investment funds and similar is relatively low,\n\ncorresponding to 43% of the levels of savings in traditional bank accounts (KPI 15) ⁹⁵ . The corresponding level in the UK is 55% and in the US 72% (see Figure 4) ⁹⁶ . High participation\n\n### rates in financial markets help companies to diversify their funding. Strong capital markets are\n\nimportant to ease companies access to finance and unlock finance for scale-ups ⁹⁷ .\n\n### **Significant amounts of EU savers’ money are either tied up in bank accounts or invested abroad.** Although the volume of private savings in the EU was nearly 65%\n\ngreater than in the US in 2022, EU\n\n### households have considerably lower total\n\nwealth than their US peers ⁹⁸ , largely because\n\n### of the lower returns they generally receive from financial markets. While the net wealth of US households grew by around 150% over the past 15 years, the corresponding growth\n\nwas only 55% in the Euro area ⁹⁹ . This is\n\n### largely due to a lack of capacity in the EU financial system to stimulate high-yielding investments, which can be explained by a combination of lack of fiscal incentives and cumbersome tax reporting requirements in many Member States, along with a public perception of the overall business climate not being sufficiently promising, lowering the confidence in the prospects of return on investment. Of the EU household savings invested in company bonds, listed shares and investment funds, a significant share, approximately EUR 300 billion annually, is invested not\n\nin the EU, but abroad ¹⁰⁰ , and mainly in the US. EU foreign direct investment into the US simultaneously amounted to EUR 3.27 trillion in 2023 ¹⁰¹ .\n\n### **New bank loans to SMEs have decreased since the COVID-19 pandemic, putting new investments at risk** . Although a higher share of financing to companies channelled through corporate bonds, listed shares, venture capital and investment funds would be desirable, bank financing remains crucial to fostering growth and competitiveness of the bulk of European SMEs who use traditional bank loans to finance investments (counting for 57% of their total finance, see Figure 5). However, SME lending, which spiked following the on-set of the\n\n94 Productive investment is here defined as gross fixed capital formation minus residential investment.\n\n95 Note: Part of these savings will contribute to corporate investment activity via the portfolio allocation made by banks and other financial intermediaries.\n\n96 Note: The Commission’s set of indicators on the Capital Markets Union provide detailed indicators that help track capital markets’ developments. List of indicators to monitor progress towards the CMU objectives.\n\n97 European Investment Bank (EIB), The scale-up gap.\n\n98 Former Prime Minister Draghi, “The future of European competitiveness – A competitiveness strategy for Europe”, part B, page 1.\n\n99 ECB Distributional Wealth Accounts; US Federal Reserve Economic Data.\n\n100 Former Prime Minister Letta, “Much more than a Market”, page 11. 101 Statista, Foreign direct investment from Europe into the United States from 2000 to 2023.\n\n### **Figure 4:** Household savings invested in bonds, shares, investment funds and similar.\n\n**Note:** Volumes of household savings in company bonds; listed shares; and investment, insurance and pension funds, relative to the volumes of households’ cash holdings and bank deposits.\n\n**Source:** European Commission, DG FISMA","publish_date":null,"title":"Compared to other advanced economies, a much smaller share of EU private savings goes Productive directly into investment in companies","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/COM_2025_26_F1_COMMUNICATION_FROM_COMMISSION_TO_INST_EN_V10_P1_3877708_p19.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/COM_2025_26_F1_COMMUNICATION_FROM_COMMISSION_TO_INST_EN_V10_P1_3877708_p19.pdf"},"request":{"example_id":"chart/COM_2025_26_F1_COMMUNICATION_FROM_COMMISSION_TO_INST_EN_V10_P1_3877708_p19","product_type":"parse","source_file_path":"COM_2025_26_F1_COMMUNICATION_FROM_COMMISSION_TO_INST_EN_V10_P1_3877708_p19.pdf"},"started_at":"2026-11-04T19:46:37.615212"} +{"completed_at":"2026-11-04T19:46:40.328702","latency_in_ms":1474,"output":{"example_id":"chart/COM_2025_26_F1_COMMUNICATION_FROM_COMMISSION_TO_INST_EN_V10_P1_3877708_p20","grounded_pages":[],"layout_pages":[],"markdown":"18\n\n### COVID pandemic due to the massive public intervention schemes, is contracting significantly, as the public support is withdrawn. Current levels of new bank finance have fallen below pre- pandemic levels (Figure 6), which can partly also be attributed to the increase in interest rates seen up until 2024.\n\n### **Figure 5:** Type of finance used in 2023 by SMEs (share of total)\n\n### **Figure 6:** New bank lending to non-financial corporations\n\n**Source** : The Survey on the Access to Finance of Enterprises (SAFE), December 2023 ¹\n\n**Source** : European Central Bank, MFI Interest Rate Statistics ²\n\n### **3\\.2** **Public investment and infrastructure**\n\n**KPI What it measures Latest EU value**\n\nKPI 16: Public investment Public investment (share of GDP) 3.49% (2023) 3.24% (2022)\n\n### **Public investment has slowly trended up over the past years reaching 3.5% of GDP in 2024,** up from 3.1% of GDP in 2018 (KPI 16) **.** While the EU level equals the US public spending that also stands at 3\\.5% of GDP, the funding landscape in the EU is fragmented and complex, with most funding carried out at national level. For example, in the case of R&D,\n\n93% of public funds invested annually are provided through national programmes ¹⁰² .\n\n### **At the EU level, the cohesion policy funds, the Recovery and Resilience Facility (RRF)**\n\n**and other instruments help finance the green and digital transition** **103** **.** Since 2021, the RRF has paid out EUR 306 billion ¹⁰⁴ , thorough implementation of national plans is needed to ensure\n\n### timely and comprehensive disbursement of remaining amounts. The Cohesion Policy funds have disbursed EUR 249 billion in the period 2021 - 2024. The Innovation Fund will provide an estimated EUR 40 billion in the current decade for development and deployment of low- carbon technologies, notably in energy intensive industries, power generation and energy storage.\n\n### **Despite relevant sources of public financing, such as InvestEU and STEP, a funding gap remains for scaling up manufacturing capacities** , as less than 5% of the EU clean tech\n\nfunding supports net-zero manufacturing at the highest technology readiness levels (8-9) ¹⁰⁵ .\n\n102 Former Prime Minister Draghi, “The future of European competitiveness – A competitiveness strategy for Europe”, Part\n\nB, p.236.\n\n103 Overall, in 2023, the EU budget and NextGenerationEU dedicated 38% of their resources to climate-relevant measures, and 19% of their resources to the EU’s digital priorities.\n\n104 European Commission, Recovery and Resilience Scoreboard. 105 Technology readiness levels are a measurement of the maturity of a technology, with 1 being the least mature (basic research) and 9 being the most mature (ready for launch/operations).","pages":[{"line_numbers":[],"markdown":"18\n\n### COVID pandemic due to the massive public intervention schemes, is contracting significantly, as the public support is withdrawn. Current levels of new bank finance have fallen below pre- pandemic levels (Figure 6), which can partly also be attributed to the increase in interest rates seen up until 2024.\n\n### **Figure 5:** Type of finance used in 2023 by SMEs (share of total)\n\n### **Figure 6:** New bank lending to non-financial corporations\n\n**Source** : The Survey on the Access to Finance of Enterprises (SAFE), December 2023 ¹\n\n**Source** : European Central Bank, MFI Interest Rate Statistics ²\n\n### **3\\.2** **Public investment and infrastructure**\n\n**KPI What it measures Latest EU value**\n\nKPI 16: Public investment Public investment (share of GDP) 3.49% (2023) 3.24% (2022)\n\n### **Public investment has slowly trended up over the past years reaching 3.5% of GDP in 2024,** up from 3.1% of GDP in 2018 (KPI 16) **.** While the EU level equals the US public spending that also stands at 3\\.5% of GDP, the funding landscape in the EU is fragmented and complex, with most funding carried out at national level. For example, in the case of R&D,\n\n93% of public funds invested annually are provided through national programmes ¹⁰² .\n\n### **At the EU level, the cohesion policy funds, the Recovery and Resilience Facility (RRF)**\n\n**and other instruments help finance the green and digital transition** **103** **.** Since 2021, the RRF has paid out EUR 306 billion ¹⁰⁴ , thorough implementation of national plans is needed to ensure\n\n### timely and comprehensive disbursement of remaining amounts. The Cohesion Policy funds have disbursed EUR 249 billion in the period 2021 - 2024. The Innovation Fund will provide an estimated EUR 40 billion in the current decade for development and deployment of low- carbon technologies, notably in energy intensive industries, power generation and energy storage.\n\n### **Despite relevant sources of public financing, such as InvestEU and STEP, a funding gap remains for scaling up manufacturing capacities** , as less than 5% of the EU clean tech\n\nfunding supports net-zero manufacturing at the highest technology readiness levels (8-9) ¹⁰⁵ .\n\n102 Former Prime Minister Draghi, “The future of European competitiveness – A competitiveness strategy for Europe”, Part\n\nB, p.236.\n\n103 Overall, in 2023, the EU budget and NextGenerationEU dedicated 38% of their resources to climate-relevant measures, and 19% of their resources to the EU’s digital priorities.\n\n104 European Commission, Recovery and Resilience Scoreboard. 105 Technology readiness levels are a measurement of the maturity of a technology, with 1 being the least mature (basic research) and 9 being the most mature (ready for launch/operations).","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"18\n\n### COVID pandemic due to the massive public intervention schemes, is contracting significantly, as the public support is withdrawn. Current levels of new bank finance have fallen below pre- pandemic levels (Figure 6), which can partly also be attributed to the increase in interest rates seen up until 2024.\n\n### **Figure 5:** Type of finance used in 2023 by SMEs (share of total)\n\n### **Figure 6:** New bank lending to non-financial corporations\n\n**Source** : The Survey on the Access to Finance of Enterprises (SAFE), December 2023 ¹\n\n**Source** : European Central Bank, MFI Interest Rate Statistics ²\n\n### **3\\.2** **Public investment and infrastructure**\n\n**KPI What it measures Latest EU value**\n\nKPI 16: Public investment Public investment (share of GDP) 3.49% (2023) 3.24% (2022)\n\n### **Public investment has slowly trended up over the past years reaching 3.5% of GDP in 2024,** up from 3.1% of GDP in 2018 (KPI 16) **.** While the EU level equals the US public spending that also stands at 3\\.5% of GDP, the funding landscape in the EU is fragmented and complex, with most funding carried out at national level. For example, in the case of R&D,\n\n93% of public funds invested annually are provided through national programmes ¹⁰² .\n\n### **At the EU level, the cohesion policy funds, the Recovery and Resilience Facility (RRF)**\n\n**and other instruments help finance the green and digital transition** **103** **.** Since 2021, the RRF has paid out EUR 306 billion ¹⁰⁴ , thorough implementation of national plans is needed to ensure\n\n### timely and comprehensive disbursement of remaining amounts. The Cohesion Policy funds have disbursed EUR 249 billion in the period 2021 - 2024. The Innovation Fund will provide an estimated EUR 40 billion in the current decade for development and deployment of low- carbon technologies, notably in energy intensive industries, power generation and energy storage.\n\n### **Despite relevant sources of public financing, such as InvestEU and STEP, a funding gap remains for scaling up manufacturing capacities** , as less than 5% of the EU clean tech\n\nfunding supports net-zero manufacturing at the highest technology readiness levels (8-9) ¹⁰⁵ .\n\n102 Former Prime Minister Draghi, “The future of European competitiveness – A competitiveness strategy for Europe”, Part\n\nB, p.236.\n\n103 Overall, in 2023, the EU budget and NextGenerationEU dedicated 38% of their resources to climate-relevant measures, and 19% of their resources to the EU’s digital priorities.\n\n104 European Commission, Recovery and Resilience Scoreboard. 105 Technology readiness levels are a measurement of the maturity of a technology, with 1 being the least mature (basic research) and 9 being the most mature (ready for launch/operations).","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/COM_2025_26_F1_COMMUNICATION_FROM_COMMISSION_TO_INST_EN_V10_P1_3877708_p20.pdf"]},"response":{"errors":[],"extract_id":"extract_9bf31a36ac260240e61883d89c2a4c3a","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["18\n\n### COVID pandemic due to the massive public intervention schemes, is contracting significantly, as the public support is withdrawn. Current levels of new bank finance have fallen below pre- pandemic levels (Figure 6), which can partly also be attributed to the increase in interest rates seen up until 2024.\n\n### **Figure 5:** Type of finance used in 2023 by SMEs (share of total)\n\n### **Figure 6:** New bank lending to non-financial corporations\n\n**Source** : The Survey on the Access to Finance of Enterprises (SAFE), December 2023 ¹\n\n**Source** : European Central Bank, MFI Interest Rate Statistics ²\n\n### **3\\.2** **Public investment and infrastructure**\n\n**KPI What it measures Latest EU value**\n\nKPI 16: Public investment Public investment (share of GDP) 3.49% (2023) 3.24% (2022)\n\n### **Public investment has slowly trended up over the past years reaching 3.5% of GDP in 2024,** up from 3.1% of GDP in 2018 (KPI 16) **.** While the EU level equals the US public spending that also stands at 3\\.5% of GDP, the funding landscape in the EU is fragmented and complex, with most funding carried out at national level. For example, in the case of R&D,\n\n93% of public funds invested annually are provided through national programmes ¹⁰² .\n\n### **At the EU level, the cohesion policy funds, the Recovery and Resilience Facility (RRF)**\n\n**and other instruments help finance the green and digital transition** **103** **.** Since 2021, the RRF has paid out EUR 306 billion ¹⁰⁴ , thorough implementation of national plans is needed to ensure\n\n### timely and comprehensive disbursement of remaining amounts. The Cohesion Policy funds have disbursed EUR 249 billion in the period 2021 - 2024. The Innovation Fund will provide an estimated EUR 40 billion in the current decade for development and deployment of low- carbon technologies, notably in energy intensive industries, power generation and energy storage.\n\n### **Despite relevant sources of public financing, such as InvestEU and STEP, a funding gap remains for scaling up manufacturing capacities** , as less than 5% of the EU clean tech\n\nfunding supports net-zero manufacturing at the highest technology readiness levels (8-9) ¹⁰⁵ .\n\n102 Former Prime Minister Draghi, “The future of European competitiveness – A competitiveness strategy for Europe”, Part\n\nB, p.236.\n\n103 Overall, in 2023, the EU budget and NextGenerationEU dedicated 38% of their resources to climate-relevant measures, and 19% of their resources to the EU’s digital priorities.\n\n104 European Commission, Recovery and Resilience Scoreboard. 105 Technology readiness levels are a measurement of the maturity of a technology, with 1 being the least mature (basic research) and 9 being the most mature (ready for launch/operations)."],"full_content":"18\n\n### COVID pandemic due to the massive public intervention schemes, is contracting significantly, as the public support is withdrawn. Current levels of new bank finance have fallen below pre- pandemic levels (Figure 6), which can partly also be attributed to the increase in interest rates seen up until 2024.\n\n### **Figure 5:** Type of finance used in 2023 by SMEs (share of total)\n\n### **Figure 6:** New bank lending to non-financial corporations\n\n**Source** : The Survey on the Access to Finance of Enterprises (SAFE), December 2023 ¹\n\n**Source** : European Central Bank, MFI Interest Rate Statistics ²\n\n### **3\\.2** **Public investment and infrastructure**\n\n**KPI What it measures Latest EU value**\n\nKPI 16: Public investment Public investment (share of GDP) 3.49% (2023) 3.24% (2022)\n\n### **Public investment has slowly trended up over the past years reaching 3.5% of GDP in 2024,** up from 3.1% of GDP in 2018 (KPI 16) **.** While the EU level equals the US public spending that also stands at 3\\.5% of GDP, the funding landscape in the EU is fragmented and complex, with most funding carried out at national level. For example, in the case of R&D,\n\n93% of public funds invested annually are provided through national programmes ¹⁰² .\n\n### **At the EU level, the cohesion policy funds, the Recovery and Resilience Facility (RRF)**\n\n**and other instruments help finance the green and digital transition** **103** **.** Since 2021, the RRF has paid out EUR 306 billion ¹⁰⁴ , thorough implementation of national plans is needed to ensure\n\n### timely and comprehensive disbursement of remaining amounts. The Cohesion Policy funds have disbursed EUR 249 billion in the period 2021 - 2024. The Innovation Fund will provide an estimated EUR 40 billion in the current decade for development and deployment of low- carbon technologies, notably in energy intensive industries, power generation and energy storage.\n\n### **Despite relevant sources of public financing, such as InvestEU and STEP, a funding gap remains for scaling up manufacturing capacities** , as less than 5% of the EU clean tech\n\nfunding supports net-zero manufacturing at the highest technology readiness levels (8-9) ¹⁰⁵ .\n\n102 Former Prime Minister Draghi, “The future of European competitiveness – A competitiveness strategy for Europe”, Part\n\nB, p.236.\n\n103 Overall, in 2023, the EU budget and NextGenerationEU dedicated 38% of their resources to climate-relevant measures, and 19% of their resources to the EU’s digital priorities.\n\n104 European Commission, Recovery and Resilience Scoreboard. 105 Technology readiness levels are a measurement of the maturity of a technology, with 1 being the least mature (basic research) and 9 being the most mature (ready for launch/operations).","publish_date":null,"title":"COVID pandemic due to the massive public intervention schemes, is contracting significantly, as the public support is withdrawn. Current levels","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/COM_2025_26_F1_COMMUNICATION_FROM_COMMISSION_TO_INST_EN_V10_P1_3877708_p20.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/COM_2025_26_F1_COMMUNICATION_FROM_COMMISSION_TO_INST_EN_V10_P1_3877708_p20.pdf"},"request":{"example_id":"chart/COM_2025_26_F1_COMMUNICATION_FROM_COMMISSION_TO_INST_EN_V10_P1_3877708_p20","product_type":"parse","source_file_path":"COM_2025_26_F1_COMMUNICATION_FROM_COMMISSION_TO_INST_EN_V10_P1_3877708_p20.pdf"},"started_at":"2026-11-04T19:46:38.854116"} +{"completed_at":"2026-11-04T19:46:42.974900","latency_in_ms":2620,"output":{"example_id":"chart/COM_2025_26_F1_COMMUNICATION_FROM_COMMISSION_TO_INST_EN_V10_P1_3877708_p21","grounded_pages":[],"layout_pages":[],"markdown":"19\n\n### **The EU State aid framework has enabled targeted public investment by Member States while preventing undue distortions of competition and maintaining a level playing field.** In 2022, Member States spent almost EUR 228 billion, corresponding to 1.4% of their GDP, on State aid (including crisis measures). This year’s Single Market and Competitiveness Scoreboard – published alongside this report – includes a new indicator that compares the concentration of State aid with the concentration of GDP in the EU. It finds State aid has\n\nbecome more unevenly allocated across the Member States over the past 10 years ¹⁰⁶ .\n\n### **Important Projects of Common European Interest (IPCEIs) are a key tool for coordinated public and private investments in the EU to support breakthrough innovation and infrastructure projects in critical technologies areas.** IPCEIs are a step towards increased cross-country coordination in industrial policies within the EU, an evolution that needs to go further. To date, ten integrated IPCEIs have been approved, for a value of more than EUR 37 billion in national public support, unlocking EUR 66 billion of private investments (see Figure 7). There is a need to make the IPCEI design and review process simpler and faster to get strategic projects off the ground quickly. The Joint European Forum on IPCEI, launched in October 2023, helps to address these issues by identifying strategic areas for future IPCEIs and improving their design and implementation.\n\n### **Figure 7:** Overview of IPCEIs and the unlocked investment volumes (by autumn 2024)\n\n**Source:** Approved IPCEIs - European Commission\n\n### **Public procurement can serve as a strategic tool to channel public investment towards shaping the future of the European economy in support of objectives, such as green transition and resilience of the EU economy, but its implementation can be challenging.** The public procurement directives ensure common rules across the Single Market and each year over 250 000 public authorities in the EU spend around 14% of GDP (EUR 2 000 billion in 2022) on services, works and supplies. While existing rules allow for social, sustainability and resilience criteria, their uptake has been limited, including due to implementation challenges.\n\n106 European Commission, Single Market Scoreboard.","pages":[{"line_numbers":[],"markdown":"19\n\n### **The EU State aid framework has enabled targeted public investment by Member States while preventing undue distortions of competition and maintaining a level playing field.** In 2022, Member States spent almost EUR 228 billion, corresponding to 1.4% of their GDP, on State aid (including crisis measures). This year’s Single Market and Competitiveness Scoreboard – published alongside this report – includes a new indicator that compares the concentration of State aid with the concentration of GDP in the EU. It finds State aid has\n\nbecome more unevenly allocated across the Member States over the past 10 years ¹⁰⁶ .\n\n### **Important Projects of Common European Interest (IPCEIs) are a key tool for coordinated public and private investments in the EU to support breakthrough innovation and infrastructure projects in critical technologies areas.** IPCEIs are a step towards increased cross-country coordination in industrial policies within the EU, an evolution that needs to go further. To date, ten integrated IPCEIs have been approved, for a value of more than EUR 37 billion in national public support, unlocking EUR 66 billion of private investments (see Figure 7). There is a need to make the IPCEI design and review process simpler and faster to get strategic projects off the ground quickly. The Joint European Forum on IPCEI, launched in October 2023, helps to address these issues by identifying strategic areas for future IPCEIs and improving their design and implementation.\n\n### **Figure 7:** Overview of IPCEIs and the unlocked investment volumes (by autumn 2024)\n\n**Source:** Approved IPCEIs - European Commission\n\n### **Public procurement can serve as a strategic tool to channel public investment towards shaping the future of the European economy in support of objectives, such as green transition and resilience of the EU economy, but its implementation can be challenging.** The public procurement directives ensure common rules across the Single Market and each year over 250 000 public authorities in the EU spend around 14% of GDP (EUR 2 000 billion in 2022) on services, works and supplies. While existing rules allow for social, sustainability and resilience criteria, their uptake has been limited, including due to implementation challenges.\n\n106 European Commission, Single Market Scoreboard.","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"19\n\n### **The EU State aid framework has enabled targeted public investment by Member States while preventing undue distortions of competition and maintaining a level playing field.** In 2022, Member States spent almost EUR 228 billion, corresponding to 1.4% of their GDP, on State aid (including crisis measures). This year’s Single Market and Competitiveness Scoreboard – published alongside this report – includes a new indicator that compares the concentration of State aid with the concentration of GDP in the EU. It finds State aid has\n\nbecome more unevenly allocated across the Member States over the past 10 years ¹⁰⁶ .\n\n### **Important Projects of Common European Interest (IPCEIs) are a key tool for coordinated public and private investments in the EU to support breakthrough innovation and infrastructure projects in critical technologies areas.** IPCEIs are a step towards increased cross-country coordination in industrial policies within the EU, an evolution that needs to go further. To date, ten integrated IPCEIs have been approved, for a value of more than EUR 37 billion in national public support, unlocking EUR 66 billion of private investments (see Figure 7). There is a need to make the IPCEI design and review process simpler and faster to get strategic projects off the ground quickly. The Joint European Forum on IPCEI, launched in October 2023, helps to address these issues by identifying strategic areas for future IPCEIs and improving their design and implementation.\n\n### **Figure 7:** Overview of IPCEIs and the unlocked investment volumes (by autumn 2024)\n\n**Source:** Approved IPCEIs - European Commission\n\n### **Public procurement can serve as a strategic tool to channel public investment towards shaping the future of the European economy in support of objectives, such as green transition and resilience of the EU economy, but its implementation can be challenging.** The public procurement directives ensure common rules across the Single Market and each year over 250 000 public authorities in the EU spend around 14% of GDP (EUR 2 000 billion in 2022) on services, works and supplies. While existing rules allow for social, sustainability and resilience criteria, their uptake has been limited, including due to implementation challenges.\n\n106 European Commission, Single Market Scoreboard.","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/COM_2025_26_F1_COMMUNICATION_FROM_COMMISSION_TO_INST_EN_V10_P1_3877708_p21.pdf"]},"response":{"errors":[],"extract_id":"extract_91c76a9f24dfd941e3efaf21a45c50ba","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["19\n\n### **The EU State aid framework has enabled targeted public investment by Member States while preventing undue distortions of competition and maintaining a level playing field.** In 2022, Member States spent almost EUR 228 billion, corresponding to 1.4% of their GDP, on State aid (including crisis measures). This year’s Single Market and Competitiveness Scoreboard – published alongside this report – includes a new indicator that compares the concentration of State aid with the concentration of GDP in the EU. It finds State aid has\n\nbecome more unevenly allocated across the Member States over the past 10 years ¹⁰⁶ .\n\n### **Important Projects of Common European Interest (IPCEIs) are a key tool for coordinated public and private investments in the EU to support breakthrough innovation and infrastructure projects in critical technologies areas.** IPCEIs are a step towards increased cross-country coordination in industrial policies within the EU, an evolution that needs to go further. To date, ten integrated IPCEIs have been approved, for a value of more than EUR 37 billion in national public support, unlocking EUR 66 billion of private investments (see Figure 7). There is a need to make the IPCEI design and review process simpler and faster to get strategic projects off the ground quickly. The Joint European Forum on IPCEI, launched in October 2023, helps to address these issues by identifying strategic areas for future IPCEIs and improving their design and implementation.\n\n### **Figure 7:** Overview of IPCEIs and the unlocked investment volumes (by autumn 2024)\n\n**Source:** Approved IPCEIs - European Commission\n\n### **Public procurement can serve as a strategic tool to channel public investment towards shaping the future of the European economy in support of objectives, such as green transition and resilience of the EU economy, but its implementation can be challenging.** The public procurement directives ensure common rules across the Single Market and each year over 250 000 public authorities in the EU spend around 14% of GDP (EUR 2 000 billion in 2022) on services, works and supplies. While existing rules allow for social, sustainability and resilience criteria, their uptake has been limited, including due to implementation challenges.\n\n106 European Commission, Single Market Scoreboard."],"full_content":"19\n\n### **The EU State aid framework has enabled targeted public investment by Member States while preventing undue distortions of competition and maintaining a level playing field.** In 2022, Member States spent almost EUR 228 billion, corresponding to 1.4% of their GDP, on State aid (including crisis measures). This year’s Single Market and Competitiveness Scoreboard – published alongside this report – includes a new indicator that compares the concentration of State aid with the concentration of GDP in the EU. It finds State aid has\n\nbecome more unevenly allocated across the Member States over the past 10 years ¹⁰⁶ .\n\n### **Important Projects of Common European Interest (IPCEIs) are a key tool for coordinated public and private investments in the EU to support breakthrough innovation and infrastructure projects in critical technologies areas.** IPCEIs are a step towards increased cross-country coordination in industrial policies within the EU, an evolution that needs to go further. To date, ten integrated IPCEIs have been approved, for a value of more than EUR 37 billion in national public support, unlocking EUR 66 billion of private investments (see Figure 7). There is a need to make the IPCEI design and review process simpler and faster to get strategic projects off the ground quickly. The Joint European Forum on IPCEI, launched in October 2023, helps to address these issues by identifying strategic areas for future IPCEIs and improving their design and implementation.\n\n### **Figure 7:** Overview of IPCEIs and the unlocked investment volumes (by autumn 2024)\n\n**Source:** Approved IPCEIs - European Commission\n\n### **Public procurement can serve as a strategic tool to channel public investment towards shaping the future of the European economy in support of objectives, such as green transition and resilience of the EU economy, but its implementation can be challenging.** The public procurement directives ensure common rules across the Single Market and each year over 250 000 public authorities in the EU spend around 14% of GDP (EUR 2 000 billion in 2022) on services, works and supplies. While existing rules allow for social, sustainability and resilience criteria, their uptake has been limited, including due to implementation challenges.\n\n106 European Commission, Single Market Scoreboard.","publish_date":null,"title":"The EU State aid framework has enabled targeted public investment by Member States while preventing undue distortions of competition and","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/COM_2025_26_F1_COMMUNICATION_FROM_COMMISSION_TO_INST_EN_V10_P1_3877708_p21.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/COM_2025_26_F1_COMMUNICATION_FROM_COMMISSION_TO_INST_EN_V10_P1_3877708_p21.pdf"},"request":{"example_id":"chart/COM_2025_26_F1_COMMUNICATION_FROM_COMMISSION_TO_INST_EN_V10_P1_3877708_p21","product_type":"parse","source_file_path":"COM_2025_26_F1_COMMUNICATION_FROM_COMMISSION_TO_INST_EN_V10_P1_3877708_p21.pdf"},"started_at":"2026-11-04T19:46:40.354465"} +{"completed_at":"2026-11-04T19:46:43.711500","latency_in_ms":2085,"output":{"example_id":"chart/COM_2025_26_F1_COMMUNICATION_FROM_COMMISSION_TO_INST_EN_V10_P1_3877708_p22","grounded_pages":[],"layout_pages":[],"markdown":"20\n\n### **3\\.3** **Energy**\n\n**KPI What it measures Target Latest EU value**\n\nKPI 17: Electricity prices for non-household consumers\n\nElectricity prices for non-household consumers (EU ID price band, large commercial consumers) with recoverable taxes and levies excluded.\n\nEUR 0.16 per kWh (2024) EUR 0.20 per kWh (2023)\n\nKPI 18: Electrification Electricity as a share of the total energy consumption. ¹⁰⁷\n\n21\\.3% (2022) 20.8% (2021)\n\nKPI 19: Share of energy from renewable sources\n\nRenewable energy generation as a share of the overall energy consumption.\n\n45% in 2030\n\n24\\.5% (2023) 23% (2022)\n\n### **Recent years’ energy price spikes have taken a toll on European energy-intensive industries,** such as steel, cement, glass, paper and chemicals manufacturers. Production has\n\ndeclined steeply – in some segments by more than 10% compared to before 2021 ¹⁰⁸ . For\n\n### instance, in aluminium manufacturing, energy costs typically amount to half of the total\n\nproduction costs ¹⁰⁹ . Energy costs significantly affect the ability of EU businesses to compete\n\n### on international markets.\n\n### **EU electricity prices have fallen from their peak but are still near twice as high as the historical levels and significantly higher than in competing regions (see KPI 17).** EU companies face electricity prices that are on average 3 times those in the US and natural gas\n\nprices 4-5 times higher ¹¹⁰ . There are also substantial price differences within the EU ¹¹¹ . The\n\n### price hikes have had direct negative impact on investor confidence and resulted in withdrawn foreign direct investments and halted expansion projects. 33% of businesses say that volatile and too high energy prices are the main factors negatively affecting EU attractiveness as a\n\nbusiness location ¹¹² .\n\n### **Figure 8:** Electricity prices for companies in the EU and other advanced economies.\n\n**Source:** Eurostat, US Energy Information Administration (EIA), the UK Department for Energy Security and Net Zero (DESNZ) and the International Energy Agency (IEA) ¹¹³ .\n\n107 Electricity as a share of gross final energy consumption. 108 Eurostat; European Commission –DG GROW. 109 International Aluminium Institute. 110 Eurostat; US Energy Information Administration (EIA); the UK Department for Energy Security and Net Zero (DESNZ); and the International Energy Agency (IEA).\n\n111 See the national numbers on KPI 17 in Annex 1 of this report. 112 Ernst & Young, EY Europe Attractiveness Survey June 2024. 113 EU IC band = medium-sized industrial consumers with annual consumption of between 500 MWh and 2 000 MWh. EU ID band = large-sized industrial consumers with annual consumption of between 2 000 MWh and 20 000 MWh.","pages":[{"line_numbers":[],"markdown":"20\n\n### **3\\.3** **Energy**\n\n**KPI What it measures Target Latest EU value**\n\nKPI 17: Electricity prices for non-household consumers\n\nElectricity prices for non-household consumers (EU ID price band, large commercial consumers) with recoverable taxes and levies excluded.\n\nEUR 0.16 per kWh (2024) EUR 0.20 per kWh (2023)\n\nKPI 18: Electrification Electricity as a share of the total energy consumption. ¹⁰⁷\n\n21\\.3% (2022) 20.8% (2021)\n\nKPI 19: Share of energy from renewable sources\n\nRenewable energy generation as a share of the overall energy consumption.\n\n45% in 2030\n\n24\\.5% (2023) 23% (2022)\n\n### **Recent years’ energy price spikes have taken a toll on European energy-intensive industries,** such as steel, cement, glass, paper and chemicals manufacturers. Production has\n\ndeclined steeply – in some segments by more than 10% compared to before 2021 ¹⁰⁸ . For\n\n### instance, in aluminium manufacturing, energy costs typically amount to half of the total\n\nproduction costs ¹⁰⁹ . Energy costs significantly affect the ability of EU businesses to compete\n\n### on international markets.\n\n### **EU electricity prices have fallen from their peak but are still near twice as high as the historical levels and significantly higher than in competing regions (see KPI 17).** EU companies face electricity prices that are on average 3 times those in the US and natural gas\n\nprices 4-5 times higher ¹¹⁰ . There are also substantial price differences within the EU ¹¹¹ . The\n\n### price hikes have had direct negative impact on investor confidence and resulted in withdrawn foreign direct investments and halted expansion projects. 33% of businesses say that volatile and too high energy prices are the main factors negatively affecting EU attractiveness as a\n\nbusiness location ¹¹² .\n\n### **Figure 8:** Electricity prices for companies in the EU and other advanced economies.\n\n**Source:** Eurostat, US Energy Information Administration (EIA), the UK Department for Energy Security and Net Zero (DESNZ) and the International Energy Agency (IEA) ¹¹³ .\n\n107 Electricity as a share of gross final energy consumption. 108 Eurostat; European Commission –DG GROW. 109 International Aluminium Institute. 110 Eurostat; US Energy Information Administration (EIA); the UK Department for Energy Security and Net Zero (DESNZ); and the International Energy Agency (IEA).\n\n111 See the national numbers on KPI 17 in Annex 1 of this report. 112 Ernst & Young, EY Europe Attractiveness Survey June 2024. 113 EU IC band = medium-sized industrial consumers with annual consumption of between 500 MWh and 2 000 MWh. EU ID band = large-sized industrial consumers with annual consumption of between 2 000 MWh and 20 000 MWh.","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"20\n\n### **3\\.3** **Energy**\n\n**KPI What it measures Target Latest EU value**\n\nKPI 17: Electricity prices for non-household consumers\n\nElectricity prices for non-household consumers (EU ID price band, large commercial consumers) with recoverable taxes and levies excluded.\n\nEUR 0.16 per kWh (2024) EUR 0.20 per kWh (2023)\n\nKPI 18: Electrification Electricity as a share of the total energy consumption. ¹⁰⁷\n\n21\\.3% (2022) 20.8% (2021)\n\nKPI 19: Share of energy from renewable sources\n\nRenewable energy generation as a share of the overall energy consumption.\n\n45% in 2030\n\n24\\.5% (2023) 23% (2022)\n\n### **Recent years’ energy price spikes have taken a toll on European energy-intensive industries,** such as steel, cement, glass, paper and chemicals manufacturers. Production has\n\ndeclined steeply – in some segments by more than 10% compared to before 2021 ¹⁰⁸ . For\n\n### instance, in aluminium manufacturing, energy costs typically amount to half of the total\n\nproduction costs ¹⁰⁹ . Energy costs significantly affect the ability of EU businesses to compete\n\n### on international markets.\n\n### **EU electricity prices have fallen from their peak but are still near twice as high as the historical levels and significantly higher than in competing regions (see KPI 17).** EU companies face electricity prices that are on average 3 times those in the US and natural gas\n\nprices 4-5 times higher ¹¹⁰ . There are also substantial price differences within the EU ¹¹¹ . The\n\n### price hikes have had direct negative impact on investor confidence and resulted in withdrawn foreign direct investments and halted expansion projects. 33% of businesses say that volatile and too high energy prices are the main factors negatively affecting EU attractiveness as a\n\nbusiness location ¹¹² .\n\n### **Figure 8:** Electricity prices for companies in the EU and other advanced economies.\n\n**Source:** Eurostat, US Energy Information Administration (EIA), the UK Department for Energy Security and Net Zero (DESNZ) and the International Energy Agency (IEA) ¹¹³ .\n\n107 Electricity as a share of gross final energy consumption. 108 Eurostat; European Commission –DG GROW. 109 International Aluminium Institute. 110 Eurostat; US Energy Information Administration (EIA); the UK Department for Energy Security and Net Zero (DESNZ); and the International Energy Agency (IEA).\n\n111 See the national numbers on KPI 17 in Annex 1 of this report. 112 Ernst & Young, EY Europe Attractiveness Survey June 2024. 113 EU IC band = medium-sized industrial consumers with annual consumption of between 500 MWh and 2 000 MWh. EU ID band = large-sized industrial consumers with annual consumption of between 2 000 MWh and 20 000 MWh.","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/COM_2025_26_F1_COMMUNICATION_FROM_COMMISSION_TO_INST_EN_V10_P1_3877708_p22.pdf"]},"response":{"errors":[],"extract_id":"extract_84333d7389b5a74e284053e45c9e84ed","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["20\n\n### **3\\.3** **Energy**\n\n**KPI What it measures Target Latest EU value**\n\nKPI 17: Electricity prices for non-household consumers\n\nElectricity prices for non-household consumers (EU ID price band, large commercial consumers) with recoverable taxes and levies excluded.\n\nEUR 0.16 per kWh (2024) EUR 0.20 per kWh (2023)\n\nKPI 18: Electrification Electricity as a share of the total energy consumption. ¹⁰⁷\n\n21\\.3% (2022) 20.8% (2021)\n\nKPI 19: Share of energy from renewable sources\n\nRenewable energy generation as a share of the overall energy consumption.\n\n45% in 2030\n\n24\\.5% (2023) 23% (2022)\n\n### **Recent years’ energy price spikes have taken a toll on European energy-intensive industries,** such as steel, cement, glass, paper and chemicals manufacturers. Production has\n\ndeclined steeply – in some segments by more than 10% compared to before 2021 ¹⁰⁸ . For\n\n### instance, in aluminium manufacturing, energy costs typically amount to half of the total\n\nproduction costs ¹⁰⁹ . Energy costs significantly affect the ability of EU businesses to compete\n\n### on international markets.\n\n### **EU electricity prices have fallen from their peak but are still near twice as high as the historical levels and significantly higher than in competing regions (see KPI 17).** EU companies face electricity prices that are on average 3 times those in the US and natural gas\n\nprices 4-5 times higher ¹¹⁰ . There are also substantial price differences within the EU ¹¹¹ . The\n\n### price hikes have had direct negative impact on investor confidence and resulted in withdrawn foreign direct investments and halted expansion projects. 33% of businesses say that volatile and too high energy prices are the main factors negatively affecting EU attractiveness as a\n\nbusiness location ¹¹² .\n\n### **Figure 8:** Electricity prices for companies in the EU and other advanced economies.\n\n**Source:** Eurostat, US Energy Information Administration (EIA), the UK Department for Energy Security and Net Zero (DESNZ) and the International Energy Agency (IEA) ¹¹³ .\n\n107 Electricity as a share of gross final energy consumption. 108 Eurostat; European Commission –DG GROW. 109 International Aluminium Institute. 110 Eurostat; US Energy Information Administration (EIA); the UK Department for Energy Security and Net Zero (DESNZ); and the International Energy Agency (IEA).\n\n111 See the national numbers on KPI 17 in Annex 1 of this report. 112 Ernst & Young, EY Europe Attractiveness Survey June 2024. 113 EU IC band = medium-sized industrial consumers with annual consumption of between 500 MWh and 2 000 MWh. EU ID band = large-sized industrial consumers with annual consumption of between 2 000 MWh and 20 000 MWh."],"full_content":"20\n\n### **3\\.3** **Energy**\n\n**KPI What it measures Target Latest EU value**\n\nKPI 17: Electricity prices for non-household consumers\n\nElectricity prices for non-household consumers (EU ID price band, large commercial consumers) with recoverable taxes and levies excluded.\n\nEUR 0.16 per kWh (2024) EUR 0.20 per kWh (2023)\n\nKPI 18: Electrification Electricity as a share of the total energy consumption. ¹⁰⁷\n\n21\\.3% (2022) 20.8% (2021)\n\nKPI 19: Share of energy from renewable sources\n\nRenewable energy generation as a share of the overall energy consumption.\n\n45% in 2030\n\n24\\.5% (2023) 23% (2022)\n\n### **Recent years’ energy price spikes have taken a toll on European energy-intensive industries,** such as steel, cement, glass, paper and chemicals manufacturers. Production has\n\ndeclined steeply – in some segments by more than 10% compared to before 2021 ¹⁰⁸ . For\n\n### instance, in aluminium manufacturing, energy costs typically amount to half of the total\n\nproduction costs ¹⁰⁹ . Energy costs significantly affect the ability of EU businesses to compete\n\n### on international markets.\n\n### **EU electricity prices have fallen from their peak but are still near twice as high as the historical levels and significantly higher than in competing regions (see KPI 17).** EU companies face electricity prices that are on average 3 times those in the US and natural gas\n\nprices 4-5 times higher ¹¹⁰ . There are also substantial price differences within the EU ¹¹¹ . The\n\n### price hikes have had direct negative impact on investor confidence and resulted in withdrawn foreign direct investments and halted expansion projects. 33% of businesses say that volatile and too high energy prices are the main factors negatively affecting EU attractiveness as a\n\nbusiness location ¹¹² .\n\n### **Figure 8:** Electricity prices for companies in the EU and other advanced economies.\n\n**Source:** Eurostat, US Energy Information Administration (EIA), the UK Department for Energy Security and Net Zero (DESNZ) and the International Energy Agency (IEA) ¹¹³ .\n\n107 Electricity as a share of gross final energy consumption. 108 Eurostat; European Commission –DG GROW. 109 International Aluminium Institute. 110 Eurostat; US Energy Information Administration (EIA); the UK Department for Energy Security and Net Zero (DESNZ); and the International Energy Agency (IEA).\n\n111 See the national numbers on KPI 17 in Annex 1 of this report. 112 Ernst & Young, EY Europe Attractiveness Survey June 2024. 113 EU IC band = medium-sized industrial consumers with annual consumption of between 500 MWh and 2 000 MWh. EU ID band = large-sized industrial consumers with annual consumption of between 2 000 MWh and 20 000 MWh.","publish_date":null,"title":"3.3 Energy Recent years’ energy price spikes have taken a toll on European energy-intensive such as steel, cement, glass, paper","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/COM_2025_26_F1_COMMUNICATION_FROM_COMMISSION_TO_INST_EN_V10_P1_3877708_p22.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/COM_2025_26_F1_COMMUNICATION_FROM_COMMISSION_TO_INST_EN_V10_P1_3877708_p22.pdf"},"request":{"example_id":"chart/COM_2025_26_F1_COMMUNICATION_FROM_COMMISSION_TO_INST_EN_V10_P1_3877708_p22","product_type":"parse","source_file_path":"COM_2025_26_F1_COMMUNICATION_FROM_COMMISSION_TO_INST_EN_V10_P1_3877708_p22.pdf"},"started_at":"2026-11-04T19:46:41.625968"} +{"completed_at":"2026-11-04T19:45:39.743271","latency_in_ms":1744,"output":{"example_id":"chart/COM_2025_26_F1_COMMUNICATION_FROM_COMMISSION_TO_INST_EN_V10_P1_3877708_p4","grounded_pages":[],"layout_pages":[],"markdown":"2\n\n### Deal and the Single Market Strategy. It builds on the reports by former Prime Minister Letta\n\non the Single Market ⁵ , former Prime Minister Draghi on competitiveness ⁶ , and former President Niinistö on preparedness ⁷ . It responds to calls from businesses to put competitiveness at the core of the EU’s agenda ⁸ .\n\n### **Figure 1:** Evolution of labour productivity in the EU and other advanced economies\n\n**Source:** AMECO database. GDP per capita in purchasing power parities (PPP) per hour worked. Values indexed with EU at 100 in 2023. **With its 22 Key Performance Indicators (KPIs)** **9** **, the report provides an analysis of the**\n\n### **key drivers of the EU’s long-term competitiveness and productivity, as well as the state of the Single Market.** It provides a basis for discussion with Member States and the European Parliament, allowing to take stock of the Single Market and the EU’s competitiveness as part\n\nof an annual progress review ¹⁰ , following up on a request of the European Council ¹¹ building on the 2023 Communication on Long-term competitiveness of the EU ¹² . It is supported by two\n\n### Staff Working Documents providing additional data coverage and analysis related to the KPIs, an overview of resilience measures by selected global players, and a monitor of industrial ecosystems. These elements are complemented by the Single Market and Competitiveness\n\nScoreboard featuring an additional 150 indicators ¹³ .\n\n### **The report identifies strengths to build on and weaknesses to address.** It starts from an assessment of the functioning of the Single Market and moves on to take stock of the EU’s competitiveness along the three axes of the Competitiveness Compass. It examines the EU’s progress in closing the innovation gap. The report then presents the state of play of drivers advancing the decarbonisation of industry and investment. Finally, the report looks at progress in strengthening economic security and reducing dependencies.\n\n5 Former Prime Minister Letta, “Much more than a Market”.\n\n6 Former Prime Minister Draghi, “The future of European competitiveness – A competitiveness strategy for Europe”.\n\n7 Former President Niinistö, “Enhancing the European Union’s Resilience for Future Uncertainties”.\n\n8 Calls from the business community include, for example, the Antwerp Declaration for a European Industrial Deal.\n\n9 This year’s edition of the report departs from the 17+2 KPIs from the 2024 ASCMR, complementing with alimited number of additions to deepen the analysis in areas of crucial interest.\n\n10 Communication ‘The Single Market at 30’ (COM(2023)162).\n\n11 The European Council (EUCO) Conclusions of December 2022 asked the Commission to present a strategy at EU level to boost competitiveness and productivity.\n\n12 Communication ‘Long-term competitiveness of the EU: looking beyond 2030’(COM(2023) 168).\n\n13 European Commission, Single Market Scoreboard.","pages":[{"line_numbers":[],"markdown":"2\n\n### Deal and the Single Market Strategy. It builds on the reports by former Prime Minister Letta\n\non the Single Market ⁵ , former Prime Minister Draghi on competitiveness ⁶ , and former President Niinistö on preparedness ⁷ . It responds to calls from businesses to put competitiveness at the core of the EU’s agenda ⁸ .\n\n### **Figure 1:** Evolution of labour productivity in the EU and other advanced economies\n\n**Source:** AMECO database. GDP per capita in purchasing power parities (PPP) per hour worked. Values indexed with EU at 100 in 2023. **With its 22 Key Performance Indicators (KPIs)** **9** **, the report provides an analysis of the**\n\n### **key drivers of the EU’s long-term competitiveness and productivity, as well as the state of the Single Market.** It provides a basis for discussion with Member States and the European Parliament, allowing to take stock of the Single Market and the EU’s competitiveness as part\n\nof an annual progress review ¹⁰ , following up on a request of the European Council ¹¹ building on the 2023 Communication on Long-term competitiveness of the EU ¹² . It is supported by two\n\n### Staff Working Documents providing additional data coverage and analysis related to the KPIs, an overview of resilience measures by selected global players, and a monitor of industrial ecosystems. These elements are complemented by the Single Market and Competitiveness\n\nScoreboard featuring an additional 150 indicators ¹³ .\n\n### **The report identifies strengths to build on and weaknesses to address.** It starts from an assessment of the functioning of the Single Market and moves on to take stock of the EU’s competitiveness along the three axes of the Competitiveness Compass. It examines the EU’s progress in closing the innovation gap. The report then presents the state of play of drivers advancing the decarbonisation of industry and investment. Finally, the report looks at progress in strengthening economic security and reducing dependencies.\n\n5 Former Prime Minister Letta, “Much more than a Market”.\n\n6 Former Prime Minister Draghi, “The future of European competitiveness – A competitiveness strategy for Europe”.\n\n7 Former President Niinistö, “Enhancing the European Union’s Resilience for Future Uncertainties”.\n\n8 Calls from the business community include, for example, the Antwerp Declaration for a European Industrial Deal.\n\n9 This year’s edition of the report departs from the 17+2 KPIs from the 2024 ASCMR, complementing with alimited number of additions to deepen the analysis in areas of crucial interest.\n\n10 Communication ‘The Single Market at 30’ (COM(2023)162).\n\n11 The European Council (EUCO) Conclusions of December 2022 asked the Commission to present a strategy at EU level to boost competitiveness and productivity.\n\n12 Communication ‘Long-term competitiveness of the EU: looking beyond 2030’(COM(2023) 168).\n\n13 European Commission, Single Market Scoreboard.","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"2\n\n### Deal and the Single Market Strategy. It builds on the reports by former Prime Minister Letta\n\non the Single Market ⁵ , former Prime Minister Draghi on competitiveness ⁶ , and former President Niinistö on preparedness ⁷ . It responds to calls from businesses to put competitiveness at the core of the EU’s agenda ⁸ .\n\n### **Figure 1:** Evolution of labour productivity in the EU and other advanced economies\n\n**Source:** AMECO database. GDP per capita in purchasing power parities (PPP) per hour worked. Values indexed with EU at 100 in 2023. **With its 22 Key Performance Indicators (KPIs)** **9** **, the report provides an analysis of the**\n\n### **key drivers of the EU’s long-term competitiveness and productivity, as well as the state of the Single Market.** It provides a basis for discussion with Member States and the European Parliament, allowing to take stock of the Single Market and the EU’s competitiveness as part\n\nof an annual progress review ¹⁰ , following up on a request of the European Council ¹¹ building on the 2023 Communication on Long-term competitiveness of the EU ¹² . It is supported by two\n\n### Staff Working Documents providing additional data coverage and analysis related to the KPIs, an overview of resilience measures by selected global players, and a monitor of industrial ecosystems. These elements are complemented by the Single Market and Competitiveness\n\nScoreboard featuring an additional 150 indicators ¹³ .\n\n### **The report identifies strengths to build on and weaknesses to address.** It starts from an assessment of the functioning of the Single Market and moves on to take stock of the EU’s competitiveness along the three axes of the Competitiveness Compass. It examines the EU’s progress in closing the innovation gap. The report then presents the state of play of drivers advancing the decarbonisation of industry and investment. Finally, the report looks at progress in strengthening economic security and reducing dependencies.\n\n5 Former Prime Minister Letta, “Much more than a Market”.\n\n6 Former Prime Minister Draghi, “The future of European competitiveness – A competitiveness strategy for Europe”.\n\n7 Former President Niinistö, “Enhancing the European Union’s Resilience for Future Uncertainties”.\n\n8 Calls from the business community include, for example, the Antwerp Declaration for a European Industrial Deal.\n\n9 This year’s edition of the report departs from the 17+2 KPIs from the 2024 ASCMR, complementing with alimited number of additions to deepen the analysis in areas of crucial interest.\n\n10 Communication ‘The Single Market at 30’ (COM(2023)162).\n\n11 The European Council (EUCO) Conclusions of December 2022 asked the Commission to present a strategy at EU level to boost competitiveness and productivity.\n\n12 Communication ‘Long-term competitiveness of the EU: looking beyond 2030’(COM(2023) 168).\n\n13 European Commission, Single Market Scoreboard.","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/COM_2025_26_F1_COMMUNICATION_FROM_COMMISSION_TO_INST_EN_V10_P1_3877708_p4.pdf"]},"response":{"errors":[],"extract_id":"extract_f7c28276f44b4a4ef940c96843875141","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["2\n\n### Deal and the Single Market Strategy. It builds on the reports by former Prime Minister Letta\n\non the Single Market ⁵ , former Prime Minister Draghi on competitiveness ⁶ , and former President Niinistö on preparedness ⁷ . It responds to calls from businesses to put competitiveness at the core of the EU’s agenda ⁸ .\n\n### **Figure 1:** Evolution of labour productivity in the EU and other advanced economies\n\n**Source:** AMECO database. GDP per capita in purchasing power parities (PPP) per hour worked. Values indexed with EU at 100 in 2023. **With its 22 Key Performance Indicators (KPIs)** **9** **, the report provides an analysis of the**\n\n### **key drivers of the EU’s long-term competitiveness and productivity, as well as the state of the Single Market.** It provides a basis for discussion with Member States and the European Parliament, allowing to take stock of the Single Market and the EU’s competitiveness as part\n\nof an annual progress review ¹⁰ , following up on a request of the European Council ¹¹ building on the 2023 Communication on Long-term competitiveness of the EU ¹² . It is supported by two\n\n### Staff Working Documents providing additional data coverage and analysis related to the KPIs, an overview of resilience measures by selected global players, and a monitor of industrial ecosystems. These elements are complemented by the Single Market and Competitiveness\n\nScoreboard featuring an additional 150 indicators ¹³ .\n\n### **The report identifies strengths to build on and weaknesses to address.** It starts from an assessment of the functioning of the Single Market and moves on to take stock of the EU’s competitiveness along the three axes of the Competitiveness Compass. It examines the EU’s progress in closing the innovation gap. The report then presents the state of play of drivers advancing the decarbonisation of industry and investment. Finally, the report looks at progress in strengthening economic security and reducing dependencies.\n\n5 Former Prime Minister Letta, “Much more than a Market”.\n\n6 Former Prime Minister Draghi, “The future of European competitiveness – A competitiveness strategy for Europe”.\n\n7 Former President Niinistö, “Enhancing the European Union’s Resilience for Future Uncertainties”.\n\n8 Calls from the business community include, for example, the Antwerp Declaration for a European Industrial Deal.\n\n9 This year’s edition of the report departs from the 17+2 KPIs from the 2024 ASCMR, complementing with alimited number of additions to deepen the analysis in areas of crucial interest.\n\n10 Communication ‘The Single Market at 30’ (COM(2023)162).\n\n11 The European Council (EUCO) Conclusions of December 2022 asked the Commission to present a strategy at EU level to boost competitiveness and productivity.\n\n12 Communication ‘Long-term competitiveness of the EU: looking beyond 2030’(COM(2023) 168).\n\n13 European Commission, Single Market Scoreboard."],"full_content":"2\n\n### Deal and the Single Market Strategy. It builds on the reports by former Prime Minister Letta\n\non the Single Market ⁵ , former Prime Minister Draghi on competitiveness ⁶ , and former President Niinistö on preparedness ⁷ . It responds to calls from businesses to put competitiveness at the core of the EU’s agenda ⁸ .\n\n### **Figure 1:** Evolution of labour productivity in the EU and other advanced economies\n\n**Source:** AMECO database. GDP per capita in purchasing power parities (PPP) per hour worked. Values indexed with EU at 100 in 2023. **With its 22 Key Performance Indicators (KPIs)** **9** **, the report provides an analysis of the**\n\n### **key drivers of the EU’s long-term competitiveness and productivity, as well as the state of the Single Market.** It provides a basis for discussion with Member States and the European Parliament, allowing to take stock of the Single Market and the EU’s competitiveness as part\n\nof an annual progress review ¹⁰ , following up on a request of the European Council ¹¹ building on the 2023 Communication on Long-term competitiveness of the EU ¹² . It is supported by two\n\n### Staff Working Documents providing additional data coverage and analysis related to the KPIs, an overview of resilience measures by selected global players, and a monitor of industrial ecosystems. These elements are complemented by the Single Market and Competitiveness\n\nScoreboard featuring an additional 150 indicators ¹³ .\n\n### **The report identifies strengths to build on and weaknesses to address.** It starts from an assessment of the functioning of the Single Market and moves on to take stock of the EU’s competitiveness along the three axes of the Competitiveness Compass. It examines the EU’s progress in closing the innovation gap. The report then presents the state of play of drivers advancing the decarbonisation of industry and investment. Finally, the report looks at progress in strengthening economic security and reducing dependencies.\n\n5 Former Prime Minister Letta, “Much more than a Market”.\n\n6 Former Prime Minister Draghi, “The future of European competitiveness – A competitiveness strategy for Europe”.\n\n7 Former President Niinistö, “Enhancing the European Union’s Resilience for Future Uncertainties”.\n\n8 Calls from the business community include, for example, the Antwerp Declaration for a European Industrial Deal.\n\n9 This year’s edition of the report departs from the 17+2 KPIs from the 2024 ASCMR, complementing with alimited number of additions to deepen the analysis in areas of crucial interest.\n\n10 Communication ‘The Single Market at 30’ (COM(2023)162).\n\n11 The European Council (EUCO) Conclusions of December 2022 asked the Commission to present a strategy at EU level to boost competitiveness and productivity.\n\n12 Communication ‘Long-term competitiveness of the EU: looking beyond 2030’(COM(2023) 168).\n\n13 European Commission, Single Market Scoreboard.","publish_date":null,"title":"Deal and the Single Market Strategy. It builds on the reports by former Prime Minister Letta on the Single Market","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/COM_2025_26_F1_COMMUNICATION_FROM_COMMISSION_TO_INST_EN_V10_P1_3877708_p4.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/COM_2025_26_F1_COMMUNICATION_FROM_COMMISSION_TO_INST_EN_V10_P1_3877708_p4.pdf"},"request":{"example_id":"chart/COM_2025_26_F1_COMMUNICATION_FROM_COMMISSION_TO_INST_EN_V10_P1_3877708_p4","product_type":"parse","source_file_path":"COM_2025_26_F1_COMMUNICATION_FROM_COMMISSION_TO_INST_EN_V10_P1_3877708_p4.pdf"},"started_at":"2026-11-04T19:45:37.998471"} +{"completed_at":"2026-11-04T19:45:40.851571","latency_in_ms":1422,"output":{"example_id":"chart/CleanEnergyMarketMonitorMarch2024_p11","grounded_pages":[],"layout_pages":[],"markdown":"Clean Energy Market Monitor – March 2024\n\nPAGE | 11\n\nOverview\n\nIEA. CC BY 4.0.\n\n## **As clean energy deployment climbed new heights for some key technologies**\n\n### Annual deployment of selected clean energy technologies, 2022 and 2023\n\nIEA 2024. CC BY 4.0.\n\nNote: “Annual deployment” refers to sales or capacity additions. GW = gigawatt; MW = megawatt; Mn = million.\n\n25\n\n50\n\n75\n\n100\n\n125\n\n2022 2023\n\n**Wind** (GW)\n\n**\\+60%**\n\n100\n\n200\n\n300\n\n400\n\n500\n\n2022 2023\n\n**Solar PV**\n\n(GW)\n\n**\\+85%**\n\n2\n\n4\n\n6\n\n8\n\n10\n\n2022 2023\n\n**Nuclear**\n\n(GW)\n\n**\\-30%**\n\n3\n\n6\n\n9\n\n12\n\n15\n\n2022 2023\n\n**Electric cars**\n\n(Million)\n\n**\\+35%**\n\n25\n\n50\n\n75\n\n100\n\n125\n\n2022 2023\n\n**Heat pumps**\n\n(GW)\n\n**\\-3%**\n\n140\n\n280\n\n420\n\n560\n\n700\n\n2022 2023\n\n**Electrolysers**\n\n(MW)\n\n**\\+360%**","pages":[{"line_numbers":[],"markdown":"Clean Energy Market Monitor – March 2024\n\nPAGE | 11\n\nOverview\n\nIEA. CC BY 4.0.\n\n## **As clean energy deployment climbed new heights for some key technologies**\n\n### Annual deployment of selected clean energy technologies, 2022 and 2023\n\nIEA 2024. CC BY 4.0.\n\nNote: “Annual deployment” refers to sales or capacity additions. GW = gigawatt; MW = megawatt; Mn = million.\n\n25\n\n50\n\n75\n\n100\n\n125\n\n2022 2023\n\n**Wind** (GW)\n\n**\\+60%**\n\n100\n\n200\n\n300\n\n400\n\n500\n\n2022 2023\n\n**Solar PV**\n\n(GW)\n\n**\\+85%**\n\n2\n\n4\n\n6\n\n8\n\n10\n\n2022 2023\n\n**Nuclear**\n\n(GW)\n\n**\\-30%**\n\n3\n\n6\n\n9\n\n12\n\n15\n\n2022 2023\n\n**Electric cars**\n\n(Million)\n\n**\\+35%**\n\n25\n\n50\n\n75\n\n100\n\n125\n\n2022 2023\n\n**Heat pumps**\n\n(GW)\n\n**\\-3%**\n\n140\n\n280\n\n420\n\n560\n\n700\n\n2022 2023\n\n**Electrolysers**\n\n(MW)\n\n**\\+360%**","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Clean Energy Market Monitor – March 2024\n\nPAGE | 11\n\nOverview\n\nIEA. CC BY 4.0.\n\n## **As clean energy deployment climbed new heights for some key technologies**\n\n### Annual deployment of selected clean energy technologies, 2022 and 2023\n\nIEA 2024. CC BY 4.0.\n\nNote: “Annual deployment” refers to sales or capacity additions. GW = gigawatt; MW = megawatt; Mn = million.\n\n25\n\n50\n\n75\n\n100\n\n125\n\n2022 2023\n\n**Wind** (GW)\n\n**\\+60%**\n\n100\n\n200\n\n300\n\n400\n\n500\n\n2022 2023\n\n**Solar PV**\n\n(GW)\n\n**\\+85%**\n\n2\n\n4\n\n6\n\n8\n\n10\n\n2022 2023\n\n**Nuclear**\n\n(GW)\n\n**\\-30%**\n\n3\n\n6\n\n9\n\n12\n\n15\n\n2022 2023\n\n**Electric cars**\n\n(Million)\n\n**\\+35%**\n\n25\n\n50\n\n75\n\n100\n\n125\n\n2022 2023\n\n**Heat pumps**\n\n(GW)\n\n**\\-3%**\n\n140\n\n280\n\n420\n\n560\n\n700\n\n2022 2023\n\n**Electrolysers**\n\n(MW)\n\n**\\+360%**","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/CleanEnergyMarketMonitorMarch2024_p11.pdf"]},"response":{"errors":[],"extract_id":"extract_09a901061c76d660f9931779aac4eff5","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Clean Energy Market Monitor – March 2024\n\nPAGE | 11\n\nOverview\n\nIEA. CC BY 4.0.\n\n## **As clean energy deployment climbed new heights for some key technologies**\n\n### Annual deployment of selected clean energy technologies, 2022 and 2023\n\nIEA 2024. CC BY 4.0.\n\nNote: “Annual deployment” refers to sales or capacity additions. GW = gigawatt; MW = megawatt; Mn = million.\n\n25\n\n50\n\n75\n\n100\n\n125\n\n2022 2023\n\n**Wind** (GW)\n\n**\\+60%**\n\n100\n\n200\n\n300\n\n400\n\n500\n\n2022 2023\n\n**Solar PV**\n\n(GW)\n\n**\\+85%**\n\n2\n\n4\n\n6\n\n8\n\n10\n\n2022 2023\n\n**Nuclear**\n\n(GW)\n\n**\\-30%**\n\n3\n\n6\n\n9\n\n12\n\n15\n\n2022 2023\n\n**Electric cars**\n\n(Million)\n\n**\\+35%**\n\n25\n\n50\n\n75\n\n100\n\n125\n\n2022 2023\n\n**Heat pumps**\n\n(GW)\n\n**\\-3%**\n\n140\n\n280\n\n420\n\n560\n\n700\n\n2022 2023\n\n**Electrolysers**\n\n(MW)\n\n**\\+360%**"],"full_content":"Clean Energy Market Monitor – March 2024\n\nPAGE | 11\n\nOverview\n\nIEA. CC BY 4.0.\n\n## **As clean energy deployment climbed new heights for some key technologies**\n\n### Annual deployment of selected clean energy technologies, 2022 and 2023\n\nIEA 2024. CC BY 4.0.\n\nNote: “Annual deployment” refers to sales or capacity additions. GW = gigawatt; MW = megawatt; Mn = million.\n\n25\n\n50\n\n75\n\n100\n\n125\n\n2022 2023\n\n**Wind** (GW)\n\n**\\+60%**\n\n100\n\n200\n\n300\n\n400\n\n500\n\n2022 2023\n\n**Solar PV**\n\n(GW)\n\n**\\+85%**\n\n2\n\n4\n\n6\n\n8\n\n10\n\n2022 2023\n\n**Nuclear**\n\n(GW)\n\n**\\-30%**\n\n3\n\n6\n\n9\n\n12\n\n15\n\n2022 2023\n\n**Electric cars**\n\n(Million)\n\n**\\+35%**\n\n25\n\n50\n\n75\n\n100\n\n125\n\n2022 2023\n\n**Heat pumps**\n\n(GW)\n\n**\\-3%**\n\n140\n\n280\n\n420\n\n560\n\n700\n\n2022 2023\n\n**Electrolysers**\n\n(MW)\n\n**\\+360%**","publish_date":null,"title":"As clean energy deployment climbed new heights for some key technologies","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/CleanEnergyMarketMonitorMarch2024_p11.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/CleanEnergyMarketMonitorMarch2024_p11.pdf"},"request":{"example_id":"chart/CleanEnergyMarketMonitorMarch2024_p11","product_type":"parse","source_file_path":"CleanEnergyMarketMonitorMarch2024_p11.pdf"},"started_at":"2026-11-04T19:45:39.428608"} +{"completed_at":"2026-11-04T19:45:42.649737","latency_in_ms":1445,"output":{"example_id":"chart/CleanEnergyMarketMonitorMarch2024_p13","grounded_pages":[],"layout_pages":[],"markdown":"Clean Energy Market Monitor – March 2024\n\nPAGE | 13\n\nClean energy deployment\n\nIEA. CC BY 4.0.\n\n## **Solar PV capacity additions and avoided emissions**\n\nIEA 2024. CC BY 4.0.\n\nNotes: “Avoided emissions” refers to avoided emissions in one year from cumulative deployment since 2019 to 2023.\n\n**117**\n\n**151 171**\n\n**228**\n\n**420**\n\n100\n\n200\n\n300\n\n400\n\n2019 2020 2021 2022 2023\n\n**World**\n\nGW\n\n**619**\n\n**101**\n\n**85**\n\n**80**\n\n**221**\n\n200 400 600\n\nChina\n\nEuropean Union\n\nUnited States\n\nIndia\n\nRest of world\n\n**Avoided emissions**\n\nMt CO₂\n\n**30**\n\n**60 64**\n\n**100**\n\n**261**\n\n100\n\n200\n\n300\n\n2019 2020 2021 2022 2023\n\n**China**\n\nGW\n\n**13**\n\n**19**\n\n**24 21**\n\n**32**\n\n10\n\n20\n\n30\n\n40\n\n2019 2020 2021 2022 2023\n\n**United States**\n\nGW\n\n**18 19**\n\n**27**\n\n**41**\n\n**53**\n\n20\n\n40\n\n60\n\n2019 2020 2021 2022 2023\n\n**European Union**\n\nGW\n\n**12**\n\n**5**\n\n**13**\n\n**18**\n\n**12**\n\n5\n\n10\n\n15\n\n20\n\n2019 2020 2021 2022 2023\n\n**India**\n\nGW","pages":[{"line_numbers":[],"markdown":"Clean Energy Market Monitor – March 2024\n\nPAGE | 13\n\nClean energy deployment\n\nIEA. CC BY 4.0.\n\n## **Solar PV capacity additions and avoided emissions**\n\nIEA 2024. CC BY 4.0.\n\nNotes: “Avoided emissions” refers to avoided emissions in one year from cumulative deployment since 2019 to 2023.\n\n**117**\n\n**151 171**\n\n**228**\n\n**420**\n\n100\n\n200\n\n300\n\n400\n\n2019 2020 2021 2022 2023\n\n**World**\n\nGW\n\n**619**\n\n**101**\n\n**85**\n\n**80**\n\n**221**\n\n200 400 600\n\nChina\n\nEuropean Union\n\nUnited States\n\nIndia\n\nRest of world\n\n**Avoided emissions**\n\nMt CO₂\n\n**30**\n\n**60 64**\n\n**100**\n\n**261**\n\n100\n\n200\n\n300\n\n2019 2020 2021 2022 2023\n\n**China**\n\nGW\n\n**13**\n\n**19**\n\n**24 21**\n\n**32**\n\n10\n\n20\n\n30\n\n40\n\n2019 2020 2021 2022 2023\n\n**United States**\n\nGW\n\n**18 19**\n\n**27**\n\n**41**\n\n**53**\n\n20\n\n40\n\n60\n\n2019 2020 2021 2022 2023\n\n**European Union**\n\nGW\n\n**12**\n\n**5**\n\n**13**\n\n**18**\n\n**12**\n\n5\n\n10\n\n15\n\n20\n\n2019 2020 2021 2022 2023\n\n**India**\n\nGW","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Clean Energy Market Monitor – March 2024\n\nPAGE | 13\n\nClean energy deployment\n\nIEA. CC BY 4.0.\n\n## **Solar PV capacity additions and avoided emissions**\n\nIEA 2024. CC BY 4.0.\n\nNotes: “Avoided emissions” refers to avoided emissions in one year from cumulative deployment since 2019 to 2023.\n\n**117**\n\n**151 171**\n\n**228**\n\n**420**\n\n100\n\n200\n\n300\n\n400\n\n2019 2020 2021 2022 2023\n\n**World**\n\nGW\n\n**619**\n\n**101**\n\n**85**\n\n**80**\n\n**221**\n\n200 400 600\n\nChina\n\nEuropean Union\n\nUnited States\n\nIndia\n\nRest of world\n\n**Avoided emissions**\n\nMt CO₂\n\n**30**\n\n**60 64**\n\n**100**\n\n**261**\n\n100\n\n200\n\n300\n\n2019 2020 2021 2022 2023\n\n**China**\n\nGW\n\n**13**\n\n**19**\n\n**24 21**\n\n**32**\n\n10\n\n20\n\n30\n\n40\n\n2019 2020 2021 2022 2023\n\n**United States**\n\nGW\n\n**18 19**\n\n**27**\n\n**41**\n\n**53**\n\n20\n\n40\n\n60\n\n2019 2020 2021 2022 2023\n\n**European Union**\n\nGW\n\n**12**\n\n**5**\n\n**13**\n\n**18**\n\n**12**\n\n5\n\n10\n\n15\n\n20\n\n2019 2020 2021 2022 2023\n\n**India**\n\nGW","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/CleanEnergyMarketMonitorMarch2024_p13.pdf"]},"response":{"errors":[],"extract_id":"extract_7ac265833af0a2c53003dc309c5b2afc","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Clean Energy Market Monitor – March 2024\n\nPAGE | 13\n\nClean energy deployment\n\nIEA. CC BY 4.0.\n\n## **Solar PV capacity additions and avoided emissions**\n\nIEA 2024. CC BY 4.0.\n\nNotes: “Avoided emissions” refers to avoided emissions in one year from cumulative deployment since 2019 to 2023.\n\n**117**\n\n**151 171**\n\n**228**\n\n**420**\n\n100\n\n200\n\n300\n\n400\n\n2019 2020 2021 2022 2023\n\n**World**\n\nGW\n\n**619**\n\n**101**\n\n**85**\n\n**80**\n\n**221**\n\n200 400 600\n\nChina\n\nEuropean Union\n\nUnited States\n\nIndia\n\nRest of world\n\n**Avoided emissions**\n\nMt CO₂\n\n**30**\n\n**60 64**\n\n**100**\n\n**261**\n\n100\n\n200\n\n300\n\n2019 2020 2021 2022 2023\n\n**China**\n\nGW\n\n**13**\n\n**19**\n\n**24 21**\n\n**32**\n\n10\n\n20\n\n30\n\n40\n\n2019 2020 2021 2022 2023\n\n**United States**\n\nGW\n\n**18 19**\n\n**27**\n\n**41**\n\n**53**\n\n20\n\n40\n\n60\n\n2019 2020 2021 2022 2023\n\n**European Union**\n\nGW\n\n**12**\n\n**5**\n\n**13**\n\n**18**\n\n**12**\n\n5\n\n10\n\n15\n\n20\n\n2019 2020 2021 2022 2023\n\n**India**\n\nGW"],"full_content":"Clean Energy Market Monitor – March 2024\n\nPAGE | 13\n\nClean energy deployment\n\nIEA. CC BY 4.0.\n\n## **Solar PV capacity additions and avoided emissions**\n\nIEA 2024. CC BY 4.0.\n\nNotes: “Avoided emissions” refers to avoided emissions in one year from cumulative deployment since 2019 to 2023.\n\n**117**\n\n**151 171**\n\n**228**\n\n**420**\n\n100\n\n200\n\n300\n\n400\n\n2019 2020 2021 2022 2023\n\n**World**\n\nGW\n\n**619**\n\n**101**\n\n**85**\n\n**80**\n\n**221**\n\n200 400 600\n\nChina\n\nEuropean Union\n\nUnited States\n\nIndia\n\nRest of world\n\n**Avoided emissions**\n\nMt CO₂\n\n**30**\n\n**60 64**\n\n**100**\n\n**261**\n\n100\n\n200\n\n300\n\n2019 2020 2021 2022 2023\n\n**China**\n\nGW\n\n**13**\n\n**19**\n\n**24 21**\n\n**32**\n\n10\n\n20\n\n30\n\n40\n\n2019 2020 2021 2022 2023\n\n**United States**\n\nGW\n\n**18 19**\n\n**27**\n\n**41**\n\n**53**\n\n20\n\n40\n\n60\n\n2019 2020 2021 2022 2023\n\n**European Union**\n\nGW\n\n**12**\n\n**5**\n\n**13**\n\n**18**\n\n**12**\n\n5\n\n10\n\n15\n\n20\n\n2019 2020 2021 2022 2023\n\n**India**\n\nGW","publish_date":null,"title":"Solar PV capacity additions and avoided emissions","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/CleanEnergyMarketMonitorMarch2024_p13.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/CleanEnergyMarketMonitorMarch2024_p13.pdf"},"request":{"example_id":"chart/CleanEnergyMarketMonitorMarch2024_p13","product_type":"parse","source_file_path":"CleanEnergyMarketMonitorMarch2024_p13.pdf"},"started_at":"2026-11-04T19:45:41.204488"} +{"completed_at":"2026-11-04T19:45:45.465704","latency_in_ms":2814,"output":{"example_id":"chart/CleanEnergyMarketMonitorMarch2024_p15","grounded_pages":[],"layout_pages":[],"markdown":"Clean Energy Market Monitor – March 2024\n\nPAGE | 15\n\nClean energy deployment\n\nIEA. CC BY 4.0.\n\n## **Wind capacity additions and avoided emissions**\n\nIEA 2024. CC BY 4.0.\n\nNotes: “Avoided emissions” refers to avoided emissions in one year from cumulative deployment since 2019 to 2023.\n\n**59**\n\n**108**\n\n**95**\n\n**74**\n\n**117**\n\n30\n\n60\n\n90\n\n120\n\n2019 2020 2021 2022 2023\n\n**World**\n\nGW\n\n**487**\n\n**99**\n\n**95**\n\n**18**\n\n**131**\n\n250 500\n\nChina\n\nUnited States\n\nEuropean Union\n\nIndia\n\nRest of world\n\n**Avoided emissions**\n\nMt CO₂\n\n**26**\n\n**70**\n\n**48**\n\n**37**\n\n**74**\n\n20\n\n40\n\n60\n\n80\n\n2019 2020 2021 2022 2023\n\n**China**\n\nGW\n\n**9**\n\n**15 14**\n\n**9**\n\n**6** 5\n\n10\n\n15\n\n2019 2020 2021 2022 2023\n\n**United States**\n\nGW\n\n**10 10**\n\n**12**\n\n**15 16**\n\n5\n\n10\n\n15\n\n20\n\n2019 2020 2021 2022 2023\n\n**European Union**\n\nGW\n\n**2\\.2**\n\n**1\\.1**\n\n**1\\.6 1.8**\n\n**2\\.7**\n\n1\n\n2\n\n3\n\n2019 2020 2021 2022 2023\n\n**India**\n\nGW","pages":[{"line_numbers":[],"markdown":"Clean Energy Market Monitor – March 2024\n\nPAGE | 15\n\nClean energy deployment\n\nIEA. CC BY 4.0.\n\n## **Wind capacity additions and avoided emissions**\n\nIEA 2024. CC BY 4.0.\n\nNotes: “Avoided emissions” refers to avoided emissions in one year from cumulative deployment since 2019 to 2023.\n\n**59**\n\n**108**\n\n**95**\n\n**74**\n\n**117**\n\n30\n\n60\n\n90\n\n120\n\n2019 2020 2021 2022 2023\n\n**World**\n\nGW\n\n**487**\n\n**99**\n\n**95**\n\n**18**\n\n**131**\n\n250 500\n\nChina\n\nUnited States\n\nEuropean Union\n\nIndia\n\nRest of world\n\n**Avoided emissions**\n\nMt CO₂\n\n**26**\n\n**70**\n\n**48**\n\n**37**\n\n**74**\n\n20\n\n40\n\n60\n\n80\n\n2019 2020 2021 2022 2023\n\n**China**\n\nGW\n\n**9**\n\n**15 14**\n\n**9**\n\n**6** 5\n\n10\n\n15\n\n2019 2020 2021 2022 2023\n\n**United States**\n\nGW\n\n**10 10**\n\n**12**\n\n**15 16**\n\n5\n\n10\n\n15\n\n20\n\n2019 2020 2021 2022 2023\n\n**European Union**\n\nGW\n\n**2\\.2**\n\n**1\\.1**\n\n**1\\.6 1.8**\n\n**2\\.7**\n\n1\n\n2\n\n3\n\n2019 2020 2021 2022 2023\n\n**India**\n\nGW","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Clean Energy Market Monitor – March 2024\n\nPAGE | 15\n\nClean energy deployment\n\nIEA. CC BY 4.0.\n\n## **Wind capacity additions and avoided emissions**\n\nIEA 2024. CC BY 4.0.\n\nNotes: “Avoided emissions” refers to avoided emissions in one year from cumulative deployment since 2019 to 2023.\n\n**59**\n\n**108**\n\n**95**\n\n**74**\n\n**117**\n\n30\n\n60\n\n90\n\n120\n\n2019 2020 2021 2022 2023\n\n**World**\n\nGW\n\n**487**\n\n**99**\n\n**95**\n\n**18**\n\n**131**\n\n250 500\n\nChina\n\nUnited States\n\nEuropean Union\n\nIndia\n\nRest of world\n\n**Avoided emissions**\n\nMt CO₂\n\n**26**\n\n**70**\n\n**48**\n\n**37**\n\n**74**\n\n20\n\n40\n\n60\n\n80\n\n2019 2020 2021 2022 2023\n\n**China**\n\nGW\n\n**9**\n\n**15 14**\n\n**9**\n\n**6** 5\n\n10\n\n15\n\n2019 2020 2021 2022 2023\n\n**United States**\n\nGW\n\n**10 10**\n\n**12**\n\n**15 16**\n\n5\n\n10\n\n15\n\n20\n\n2019 2020 2021 2022 2023\n\n**European Union**\n\nGW\n\n**2\\.2**\n\n**1\\.1**\n\n**1\\.6 1.8**\n\n**2\\.7**\n\n1\n\n2\n\n3\n\n2019 2020 2021 2022 2023\n\n**India**\n\nGW","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/CleanEnergyMarketMonitorMarch2024_p15.pdf"]},"response":{"errors":[],"extract_id":"extract_c523fa70216ba1487274489e3f900a64","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Clean Energy Market Monitor – March 2024\n\nPAGE | 15\n\nClean energy deployment\n\nIEA. CC BY 4.0.\n\n## **Wind capacity additions and avoided emissions**\n\nIEA 2024. CC BY 4.0.\n\nNotes: “Avoided emissions” refers to avoided emissions in one year from cumulative deployment since 2019 to 2023.\n\n**59**\n\n**108**\n\n**95**\n\n**74**\n\n**117**\n\n30\n\n60\n\n90\n\n120\n\n2019 2020 2021 2022 2023\n\n**World**\n\nGW\n\n**487**\n\n**99**\n\n**95**\n\n**18**\n\n**131**\n\n250 500\n\nChina\n\nUnited States\n\nEuropean Union\n\nIndia\n\nRest of world\n\n**Avoided emissions**\n\nMt CO₂\n\n**26**\n\n**70**\n\n**48**\n\n**37**\n\n**74**\n\n20\n\n40\n\n60\n\n80\n\n2019 2020 2021 2022 2023\n\n**China**\n\nGW\n\n**9**\n\n**15 14**\n\n**9**\n\n**6** 5\n\n10\n\n15\n\n2019 2020 2021 2022 2023\n\n**United States**\n\nGW\n\n**10 10**\n\n**12**\n\n**15 16**\n\n5\n\n10\n\n15\n\n20\n\n2019 2020 2021 2022 2023\n\n**European Union**\n\nGW\n\n**2\\.2**\n\n**1\\.1**\n\n**1\\.6 1.8**\n\n**2\\.7**\n\n1\n\n2\n\n3\n\n2019 2020 2021 2022 2023\n\n**India**\n\nGW"],"full_content":"Clean Energy Market Monitor – March 2024\n\nPAGE | 15\n\nClean energy deployment\n\nIEA. CC BY 4.0.\n\n## **Wind capacity additions and avoided emissions**\n\nIEA 2024. CC BY 4.0.\n\nNotes: “Avoided emissions” refers to avoided emissions in one year from cumulative deployment since 2019 to 2023.\n\n**59**\n\n**108**\n\n**95**\n\n**74**\n\n**117**\n\n30\n\n60\n\n90\n\n120\n\n2019 2020 2021 2022 2023\n\n**World**\n\nGW\n\n**487**\n\n**99**\n\n**95**\n\n**18**\n\n**131**\n\n250 500\n\nChina\n\nUnited States\n\nEuropean Union\n\nIndia\n\nRest of world\n\n**Avoided emissions**\n\nMt CO₂\n\n**26**\n\n**70**\n\n**48**\n\n**37**\n\n**74**\n\n20\n\n40\n\n60\n\n80\n\n2019 2020 2021 2022 2023\n\n**China**\n\nGW\n\n**9**\n\n**15 14**\n\n**9**\n\n**6** 5\n\n10\n\n15\n\n2019 2020 2021 2022 2023\n\n**United States**\n\nGW\n\n**10 10**\n\n**12**\n\n**15 16**\n\n5\n\n10\n\n15\n\n20\n\n2019 2020 2021 2022 2023\n\n**European Union**\n\nGW\n\n**2\\.2**\n\n**1\\.1**\n\n**1\\.6 1.8**\n\n**2\\.7**\n\n1\n\n2\n\n3\n\n2019 2020 2021 2022 2023\n\n**India**\n\nGW","publish_date":null,"title":"Wind capacity additions and avoided emissions","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/CleanEnergyMarketMonitorMarch2024_p15.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/CleanEnergyMarketMonitorMarch2024_p15.pdf"},"request":{"example_id":"chart/CleanEnergyMarketMonitorMarch2024_p15","product_type":"parse","source_file_path":"CleanEnergyMarketMonitorMarch2024_p15.pdf"},"started_at":"2026-11-04T19:45:42.650979"} +{"completed_at":"2026-11-04T19:45:45.171316","latency_in_ms":1235,"output":{"example_id":"chart/CleanEnergyMarketMonitorMarch2024_p17","grounded_pages":[],"layout_pages":[],"markdown":"Clean Energy Market Monitor – March 2024\n\nPAGE | 17\n\nClean energy deployment\n\nIEA. CC BY 4.0.\n\n## **Nuclear capacity additions and avoided emissions**\n\nIEA 2024. CC BY 4.0.\n\nNotes: “Avoided emissions” refers to avoided emissions in one year from cumulative deployment since 2019 to 2023.\n\n**5\\.5**\n\n**6\\.0 5.6**\n\n**7\\.9**\n\n**5\\.5**\n\n2\n\n4\n\n6\n\n8\n\n2019 2020 2021 2022 2023\n\n**World**\n\nGW\n\n**74**\n\n**20**\n\n**15**\n\n**9**\n\n**44**\n\n40 80\n\nChina\n\nKorea\n\nUnited Arab\n\nEmirates\n\nEuropean Union\n\nRest of world\n\n**Avoided emissions**\n\nMt CO₂\n\n**2\\.8**\n\n**2\\.3 2.4 2.3**\n\n**1\\.2** 1\n\n2\n\n3\n\n2019 2020 2021 2022 2023\n\n**China**\n\nGW\n\n**0 0 0 0**\n\n**1\\.3**\n\n0\\.5\n\n1\\.0\n\n1\\.5\n\n2019 2020 2021 2022 2023\n\n**United States**\n\nGW\n\n**0\\.0 0.0 0.0**\n\n**1\\.7**\n\n**0\\.5**\n\n0\\.6\n\n1\\.2\n\n1\\.8\n\n2019 2020 2021 2022 2023\n\n**European Union**\n\nGW\n\n**1\\.4**\n\n**0\\.0 0.0**\n\n**1\\.4 1.4**\n\n0\\.5\n\n1\\.0\n\n1\\.5\n\n2019 2020 2021 2022 2023\n\n**Korea**\n\nGW","pages":[{"line_numbers":[],"markdown":"Clean Energy Market Monitor – March 2024\n\nPAGE | 17\n\nClean energy deployment\n\nIEA. CC BY 4.0.\n\n## **Nuclear capacity additions and avoided emissions**\n\nIEA 2024. CC BY 4.0.\n\nNotes: “Avoided emissions” refers to avoided emissions in one year from cumulative deployment since 2019 to 2023.\n\n**5\\.5**\n\n**6\\.0 5.6**\n\n**7\\.9**\n\n**5\\.5**\n\n2\n\n4\n\n6\n\n8\n\n2019 2020 2021 2022 2023\n\n**World**\n\nGW\n\n**74**\n\n**20**\n\n**15**\n\n**9**\n\n**44**\n\n40 80\n\nChina\n\nKorea\n\nUnited Arab\n\nEmirates\n\nEuropean Union\n\nRest of world\n\n**Avoided emissions**\n\nMt CO₂\n\n**2\\.8**\n\n**2\\.3 2.4 2.3**\n\n**1\\.2** 1\n\n2\n\n3\n\n2019 2020 2021 2022 2023\n\n**China**\n\nGW\n\n**0 0 0 0**\n\n**1\\.3**\n\n0\\.5\n\n1\\.0\n\n1\\.5\n\n2019 2020 2021 2022 2023\n\n**United States**\n\nGW\n\n**0\\.0 0.0 0.0**\n\n**1\\.7**\n\n**0\\.5**\n\n0\\.6\n\n1\\.2\n\n1\\.8\n\n2019 2020 2021 2022 2023\n\n**European Union**\n\nGW\n\n**1\\.4**\n\n**0\\.0 0.0**\n\n**1\\.4 1.4**\n\n0\\.5\n\n1\\.0\n\n1\\.5\n\n2019 2020 2021 2022 2023\n\n**Korea**\n\nGW","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Clean Energy Market Monitor – March 2024\n\nPAGE | 17\n\nClean energy deployment\n\nIEA. CC BY 4.0.\n\n## **Nuclear capacity additions and avoided emissions**\n\nIEA 2024. CC BY 4.0.\n\nNotes: “Avoided emissions” refers to avoided emissions in one year from cumulative deployment since 2019 to 2023.\n\n**5\\.5**\n\n**6\\.0 5.6**\n\n**7\\.9**\n\n**5\\.5**\n\n2\n\n4\n\n6\n\n8\n\n2019 2020 2021 2022 2023\n\n**World**\n\nGW\n\n**74**\n\n**20**\n\n**15**\n\n**9**\n\n**44**\n\n40 80\n\nChina\n\nKorea\n\nUnited Arab\n\nEmirates\n\nEuropean Union\n\nRest of world\n\n**Avoided emissions**\n\nMt CO₂\n\n**2\\.8**\n\n**2\\.3 2.4 2.3**\n\n**1\\.2** 1\n\n2\n\n3\n\n2019 2020 2021 2022 2023\n\n**China**\n\nGW\n\n**0 0 0 0**\n\n**1\\.3**\n\n0\\.5\n\n1\\.0\n\n1\\.5\n\n2019 2020 2021 2022 2023\n\n**United States**\n\nGW\n\n**0\\.0 0.0 0.0**\n\n**1\\.7**\n\n**0\\.5**\n\n0\\.6\n\n1\\.2\n\n1\\.8\n\n2019 2020 2021 2022 2023\n\n**European Union**\n\nGW\n\n**1\\.4**\n\n**0\\.0 0.0**\n\n**1\\.4 1.4**\n\n0\\.5\n\n1\\.0\n\n1\\.5\n\n2019 2020 2021 2022 2023\n\n**Korea**\n\nGW","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/CleanEnergyMarketMonitorMarch2024_p17.pdf"]},"response":{"errors":[],"extract_id":"extract_b77dfaf2e36fcc40c0511e8a594c0446","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Clean Energy Market Monitor – March 2024\n\nPAGE | 17\n\nClean energy deployment\n\nIEA. CC BY 4.0.\n\n## **Nuclear capacity additions and avoided emissions**\n\nIEA 2024. CC BY 4.0.\n\nNotes: “Avoided emissions” refers to avoided emissions in one year from cumulative deployment since 2019 to 2023.\n\n**5\\.5**\n\n**6\\.0 5.6**\n\n**7\\.9**\n\n**5\\.5**\n\n2\n\n4\n\n6\n\n8\n\n2019 2020 2021 2022 2023\n\n**World**\n\nGW\n\n**74**\n\n**20**\n\n**15**\n\n**9**\n\n**44**\n\n40 80\n\nChina\n\nKorea\n\nUnited Arab\n\nEmirates\n\nEuropean Union\n\nRest of world\n\n**Avoided emissions**\n\nMt CO₂\n\n**2\\.8**\n\n**2\\.3 2.4 2.3**\n\n**1\\.2** 1\n\n2\n\n3\n\n2019 2020 2021 2022 2023\n\n**China**\n\nGW\n\n**0 0 0 0**\n\n**1\\.3**\n\n0\\.5\n\n1\\.0\n\n1\\.5\n\n2019 2020 2021 2022 2023\n\n**United States**\n\nGW\n\n**0\\.0 0.0 0.0**\n\n**1\\.7**\n\n**0\\.5**\n\n0\\.6\n\n1\\.2\n\n1\\.8\n\n2019 2020 2021 2022 2023\n\n**European Union**\n\nGW\n\n**1\\.4**\n\n**0\\.0 0.0**\n\n**1\\.4 1.4**\n\n0\\.5\n\n1\\.0\n\n1\\.5\n\n2019 2020 2021 2022 2023\n\n**Korea**\n\nGW"],"full_content":"Clean Energy Market Monitor – March 2024\n\nPAGE | 17\n\nClean energy deployment\n\nIEA. CC BY 4.0.\n\n## **Nuclear capacity additions and avoided emissions**\n\nIEA 2024. CC BY 4.0.\n\nNotes: “Avoided emissions” refers to avoided emissions in one year from cumulative deployment since 2019 to 2023.\n\n**5\\.5**\n\n**6\\.0 5.6**\n\n**7\\.9**\n\n**5\\.5**\n\n2\n\n4\n\n6\n\n8\n\n2019 2020 2021 2022 2023\n\n**World**\n\nGW\n\n**74**\n\n**20**\n\n**15**\n\n**9**\n\n**44**\n\n40 80\n\nChina\n\nKorea\n\nUnited Arab\n\nEmirates\n\nEuropean Union\n\nRest of world\n\n**Avoided emissions**\n\nMt CO₂\n\n**2\\.8**\n\n**2\\.3 2.4 2.3**\n\n**1\\.2** 1\n\n2\n\n3\n\n2019 2020 2021 2022 2023\n\n**China**\n\nGW\n\n**0 0 0 0**\n\n**1\\.3**\n\n0\\.5\n\n1\\.0\n\n1\\.5\n\n2019 2020 2021 2022 2023\n\n**United States**\n\nGW\n\n**0\\.0 0.0 0.0**\n\n**1\\.7**\n\n**0\\.5**\n\n0\\.6\n\n1\\.2\n\n1\\.8\n\n2019 2020 2021 2022 2023\n\n**European Union**\n\nGW\n\n**1\\.4**\n\n**0\\.0 0.0**\n\n**1\\.4 1.4**\n\n0\\.5\n\n1\\.0\n\n1\\.5\n\n2019 2020 2021 2022 2023\n\n**Korea**\n\nGW","publish_date":null,"title":"Nuclear capacity additions and avoided emissions","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/CleanEnergyMarketMonitorMarch2024_p17.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/CleanEnergyMarketMonitorMarch2024_p17.pdf"},"request":{"example_id":"chart/CleanEnergyMarketMonitorMarch2024_p17","product_type":"parse","source_file_path":"CleanEnergyMarketMonitorMarch2024_p17.pdf"},"started_at":"2026-11-04T19:45:43.935318"} +{"completed_at":"2026-11-04T19:45:47.816623","latency_in_ms":2111,"output":{"example_id":"chart/CleanEnergyMarketMonitorMarch2024_p19","grounded_pages":[],"layout_pages":[],"markdown":"Clean Energy Market Monitor – March 2024\n\nPAGE | 19\n\nClean energy deployment\n\nIEA. CC BY 4.0.\n\n## **Electric car sales and avoided emissions**\n\nIEA 2024. CC BY 4.0.\n\nNotes: “Avoided emissions” refers to avoided emissions in one year from cumulative deployment since 2019 to 2023.\n\n**2\\.2 3.0**\n\n**6\\.5**\n\n**10\\.2**\n\n**13\\.7**\n\n5\n\n10\n\n15\n\n2019 2020 2021 2022 2023\n\n**World**\n\nMillion units\n\n**22**\n\n**15**\n\n**14**\n\n**3**\n\n**7**\n\n10 20\n\nChina\n\nUnited States\n\nEuropean Union\n\nUnited Kingdom\n\nRest of world\n\n**Avoided emissions**\n\nMt CO₂\n\n**1\\.1 1.1**\n\n**3\\.3**\n\n**6\\.0**\n\n**8\\.1**\n\n3\n\n6\n\n9\n\n2019 2020 2021 2022 2023\n\n**China**\n\nMillion units\n\n**0\\.3 0.3**\n\n**0\\.6**\n\n**1\\.0**\n\n**1\\.4**\n\n0\\.5\n\n1\\.0\n\n1\\.5\n\n2019 2020 2021 2022 2023\n\n**United States**\n\nMillion units\n\n**0\\.4**\n\n**1\\.0**\n\n**1\\.7**\n\n**2\\.0**\n\n**2\\.4**\n\n1\n\n2\n\n3\n\n2019 2020 2021 2022 2023\n\n**European Union**\n\nMillion units\n\n**1 3 12**\n\n**48**\n\n**82**\n\n30\n\n60\n\n90\n\n2019 2020 2021 2022 2023\n\n**India**\n\nThousand units","pages":[{"line_numbers":[],"markdown":"Clean Energy Market Monitor – March 2024\n\nPAGE | 19\n\nClean energy deployment\n\nIEA. CC BY 4.0.\n\n## **Electric car sales and avoided emissions**\n\nIEA 2024. CC BY 4.0.\n\nNotes: “Avoided emissions” refers to avoided emissions in one year from cumulative deployment since 2019 to 2023.\n\n**2\\.2 3.0**\n\n**6\\.5**\n\n**10\\.2**\n\n**13\\.7**\n\n5\n\n10\n\n15\n\n2019 2020 2021 2022 2023\n\n**World**\n\nMillion units\n\n**22**\n\n**15**\n\n**14**\n\n**3**\n\n**7**\n\n10 20\n\nChina\n\nUnited States\n\nEuropean Union\n\nUnited Kingdom\n\nRest of world\n\n**Avoided emissions**\n\nMt CO₂\n\n**1\\.1 1.1**\n\n**3\\.3**\n\n**6\\.0**\n\n**8\\.1**\n\n3\n\n6\n\n9\n\n2019 2020 2021 2022 2023\n\n**China**\n\nMillion units\n\n**0\\.3 0.3**\n\n**0\\.6**\n\n**1\\.0**\n\n**1\\.4**\n\n0\\.5\n\n1\\.0\n\n1\\.5\n\n2019 2020 2021 2022 2023\n\n**United States**\n\nMillion units\n\n**0\\.4**\n\n**1\\.0**\n\n**1\\.7**\n\n**2\\.0**\n\n**2\\.4**\n\n1\n\n2\n\n3\n\n2019 2020 2021 2022 2023\n\n**European Union**\n\nMillion units\n\n**1 3 12**\n\n**48**\n\n**82**\n\n30\n\n60\n\n90\n\n2019 2020 2021 2022 2023\n\n**India**\n\nThousand units","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Clean Energy Market Monitor – March 2024\n\nPAGE | 19\n\nClean energy deployment\n\nIEA. CC BY 4.0.\n\n## **Electric car sales and avoided emissions**\n\nIEA 2024. CC BY 4.0.\n\nNotes: “Avoided emissions” refers to avoided emissions in one year from cumulative deployment since 2019 to 2023.\n\n**2\\.2 3.0**\n\n**6\\.5**\n\n**10\\.2**\n\n**13\\.7**\n\n5\n\n10\n\n15\n\n2019 2020 2021 2022 2023\n\n**World**\n\nMillion units\n\n**22**\n\n**15**\n\n**14**\n\n**3**\n\n**7**\n\n10 20\n\nChina\n\nUnited States\n\nEuropean Union\n\nUnited Kingdom\n\nRest of world\n\n**Avoided emissions**\n\nMt CO₂\n\n**1\\.1 1.1**\n\n**3\\.3**\n\n**6\\.0**\n\n**8\\.1**\n\n3\n\n6\n\n9\n\n2019 2020 2021 2022 2023\n\n**China**\n\nMillion units\n\n**0\\.3 0.3**\n\n**0\\.6**\n\n**1\\.0**\n\n**1\\.4**\n\n0\\.5\n\n1\\.0\n\n1\\.5\n\n2019 2020 2021 2022 2023\n\n**United States**\n\nMillion units\n\n**0\\.4**\n\n**1\\.0**\n\n**1\\.7**\n\n**2\\.0**\n\n**2\\.4**\n\n1\n\n2\n\n3\n\n2019 2020 2021 2022 2023\n\n**European Union**\n\nMillion units\n\n**1 3 12**\n\n**48**\n\n**82**\n\n30\n\n60\n\n90\n\n2019 2020 2021 2022 2023\n\n**India**\n\nThousand units","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/CleanEnergyMarketMonitorMarch2024_p19.pdf"]},"response":{"errors":[],"extract_id":"extract_4afac4041850ca4f7ceccaaaea741ee0","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Clean Energy Market Monitor – March 2024\n\nPAGE | 19\n\nClean energy deployment\n\nIEA. CC BY 4.0.\n\n## **Electric car sales and avoided emissions**\n\nIEA 2024. CC BY 4.0.\n\nNotes: “Avoided emissions” refers to avoided emissions in one year from cumulative deployment since 2019 to 2023.\n\n**2\\.2 3.0**\n\n**6\\.5**\n\n**10\\.2**\n\n**13\\.7**\n\n5\n\n10\n\n15\n\n2019 2020 2021 2022 2023\n\n**World**\n\nMillion units\n\n**22**\n\n**15**\n\n**14**\n\n**3**\n\n**7**\n\n10 20\n\nChina\n\nUnited States\n\nEuropean Union\n\nUnited Kingdom\n\nRest of world\n\n**Avoided emissions**\n\nMt CO₂\n\n**1\\.1 1.1**\n\n**3\\.3**\n\n**6\\.0**\n\n**8\\.1**\n\n3\n\n6\n\n9\n\n2019 2020 2021 2022 2023\n\n**China**\n\nMillion units\n\n**0\\.3 0.3**\n\n**0\\.6**\n\n**1\\.0**\n\n**1\\.4**\n\n0\\.5\n\n1\\.0\n\n1\\.5\n\n2019 2020 2021 2022 2023\n\n**United States**\n\nMillion units\n\n**0\\.4**\n\n**1\\.0**\n\n**1\\.7**\n\n**2\\.0**\n\n**2\\.4**\n\n1\n\n2\n\n3\n\n2019 2020 2021 2022 2023\n\n**European Union**\n\nMillion units\n\n**1 3 12**\n\n**48**\n\n**82**\n\n30\n\n60\n\n90\n\n2019 2020 2021 2022 2023\n\n**India**\n\nThousand units"],"full_content":"Clean Energy Market Monitor – March 2024\n\nPAGE | 19\n\nClean energy deployment\n\nIEA. CC BY 4.0.\n\n## **Electric car sales and avoided emissions**\n\nIEA 2024. CC BY 4.0.\n\nNotes: “Avoided emissions” refers to avoided emissions in one year from cumulative deployment since 2019 to 2023.\n\n**2\\.2 3.0**\n\n**6\\.5**\n\n**10\\.2**\n\n**13\\.7**\n\n5\n\n10\n\n15\n\n2019 2020 2021 2022 2023\n\n**World**\n\nMillion units\n\n**22**\n\n**15**\n\n**14**\n\n**3**\n\n**7**\n\n10 20\n\nChina\n\nUnited States\n\nEuropean Union\n\nUnited Kingdom\n\nRest of world\n\n**Avoided emissions**\n\nMt CO₂\n\n**1\\.1 1.1**\n\n**3\\.3**\n\n**6\\.0**\n\n**8\\.1**\n\n3\n\n6\n\n9\n\n2019 2020 2021 2022 2023\n\n**China**\n\nMillion units\n\n**0\\.3 0.3**\n\n**0\\.6**\n\n**1\\.0**\n\n**1\\.4**\n\n0\\.5\n\n1\\.0\n\n1\\.5\n\n2019 2020 2021 2022 2023\n\n**United States**\n\nMillion units\n\n**0\\.4**\n\n**1\\.0**\n\n**1\\.7**\n\n**2\\.0**\n\n**2\\.4**\n\n1\n\n2\n\n3\n\n2019 2020 2021 2022 2023\n\n**European Union**\n\nMillion units\n\n**1 3 12**\n\n**48**\n\n**82**\n\n30\n\n60\n\n90\n\n2019 2020 2021 2022 2023\n\n**India**\n\nThousand units","publish_date":null,"title":"Electric car sales and avoided emissions","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/CleanEnergyMarketMonitorMarch2024_p19.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/CleanEnergyMarketMonitorMarch2024_p19.pdf"},"request":{"example_id":"chart/CleanEnergyMarketMonitorMarch2024_p19","product_type":"parse","source_file_path":"CleanEnergyMarketMonitorMarch2024_p19.pdf"},"started_at":"2026-11-04T19:45:45.705103"} +{"completed_at":"2026-11-04T19:45:49.627715","latency_in_ms":2532,"output":{"example_id":"chart/CleanEnergyMarketMonitorMarch2024_p9","grounded_pages":[],"layout_pages":[],"markdown":"Clean Energy Market Monitor – March 2024\n\nPAGE | 9\n\nOverview\n\nIEA. CC BY 4.0.\n\n## **Without this clean energy deployment, the development of energy markets would have looked very different**\n\nIEA 2024. CC BY 4.0.\n\n\\-800 -600 -400 -200 0 200 400\n\n**Coal**\n\n(Mtce)\n\nAvoided annual demand due to 2019-23 clean tech. deployment\n\nActual 2019-23\n\nchange in demand\n\n**2023 coal demand in G7 countries**\n\n\\-200 -150 -100 -50 0 50 100\n\n**Natural gas**\n\n(bcm)\n\nAvoided annual demand due to 2019-23 clean tech. deployment\n\nActual 2019-23\n\nchange in demand\n\n**Pre-invasion natural gas exports from Russia to the European Union**\n\n\\-1.00 -0.75 -0.50 -0.25 0.00 0.25 0.50\n\n**Oil**\n\n(mboe/d)\n\nAvoided annual demand due to 2019-23 clean tech. deployment\n\nActual 2019-23\n\nchange in demand\n\n**2023 United Kingdom oil demand**","pages":[{"line_numbers":[],"markdown":"Clean Energy Market Monitor – March 2024\n\nPAGE | 9\n\nOverview\n\nIEA. CC BY 4.0.\n\n## **Without this clean energy deployment, the development of energy markets would have looked very different**\n\nIEA 2024. CC BY 4.0.\n\n\\-800 -600 -400 -200 0 200 400\n\n**Coal**\n\n(Mtce)\n\nAvoided annual demand due to 2019-23 clean tech. deployment\n\nActual 2019-23\n\nchange in demand\n\n**2023 coal demand in G7 countries**\n\n\\-200 -150 -100 -50 0 50 100\n\n**Natural gas**\n\n(bcm)\n\nAvoided annual demand due to 2019-23 clean tech. deployment\n\nActual 2019-23\n\nchange in demand\n\n**Pre-invasion natural gas exports from Russia to the European Union**\n\n\\-1.00 -0.75 -0.50 -0.25 0.00 0.25 0.50\n\n**Oil**\n\n(mboe/d)\n\nAvoided annual demand due to 2019-23 clean tech. deployment\n\nActual 2019-23\n\nchange in demand\n\n**2023 United Kingdom oil demand**","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Clean Energy Market Monitor – March 2024\n\nPAGE | 9\n\nOverview\n\nIEA. CC BY 4.0.\n\n## **Without this clean energy deployment, the development of energy markets would have looked very different**\n\nIEA 2024. CC BY 4.0.\n\n\\-800 -600 -400 -200 0 200 400\n\n**Coal**\n\n(Mtce)\n\nAvoided annual demand due to 2019-23 clean tech. deployment\n\nActual 2019-23\n\nchange in demand\n\n**2023 coal demand in G7 countries**\n\n\\-200 -150 -100 -50 0 50 100\n\n**Natural gas**\n\n(bcm)\n\nAvoided annual demand due to 2019-23 clean tech. deployment\n\nActual 2019-23\n\nchange in demand\n\n**Pre-invasion natural gas exports from Russia to the European Union**\n\n\\-1.00 -0.75 -0.50 -0.25 0.00 0.25 0.50\n\n**Oil**\n\n(mboe/d)\n\nAvoided annual demand due to 2019-23 clean tech. deployment\n\nActual 2019-23\n\nchange in demand\n\n**2023 United Kingdom oil demand**","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/CleanEnergyMarketMonitorMarch2024_p9.pdf"]},"response":{"errors":[],"extract_id":"extract_41f02bbb49988b79660ae6d460ece70e","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Clean Energy Market Monitor – March 2024\n\nPAGE | 9\n\nOverview\n\nIEA. CC BY 4.0.\n\n## **Without this clean energy deployment, the development of energy markets would have looked very different**\n\nIEA 2024. CC BY 4.0.\n\n\\-800 -600 -400 -200 0 200 400\n\n**Coal**\n\n(Mtce)\n\nAvoided annual demand due to 2019-23 clean tech. deployment\n\nActual 2019-23\n\nchange in demand\n\n**2023 coal demand in G7 countries**\n\n\\-200 -150 -100 -50 0 50 100\n\n**Natural gas**\n\n(bcm)\n\nAvoided annual demand due to 2019-23 clean tech. deployment\n\nActual 2019-23\n\nchange in demand\n\n**Pre-invasion natural gas exports from Russia to the European Union**\n\n\\-1.00 -0.75 -0.50 -0.25 0.00 0.25 0.50\n\n**Oil**\n\n(mboe/d)\n\nAvoided annual demand due to 2019-23 clean tech. deployment\n\nActual 2019-23\n\nchange in demand\n\n**2023 United Kingdom oil demand**"],"full_content":"Clean Energy Market Monitor – March 2024\n\nPAGE | 9\n\nOverview\n\nIEA. CC BY 4.0.\n\n## **Without this clean energy deployment, the development of energy markets would have looked very different**\n\nIEA 2024. CC BY 4.0.\n\n\\-800 -600 -400 -200 0 200 400\n\n**Coal**\n\n(Mtce)\n\nAvoided annual demand due to 2019-23 clean tech. deployment\n\nActual 2019-23\n\nchange in demand\n\n**2023 coal demand in G7 countries**\n\n\\-200 -150 -100 -50 0 50 100\n\n**Natural gas**\n\n(bcm)\n\nAvoided annual demand due to 2019-23 clean tech. deployment\n\nActual 2019-23\n\nchange in demand\n\n**Pre-invasion natural gas exports from Russia to the European Union**\n\n\\-1.00 -0.75 -0.50 -0.25 0.00 0.25 0.50\n\n**Oil**\n\n(mboe/d)\n\nAvoided annual demand due to 2019-23 clean tech. deployment\n\nActual 2019-23\n\nchange in demand\n\n**2023 United Kingdom oil demand**","publish_date":null,"title":"Without this clean energy deployment, the development of energy markets would have looked very different","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/CleanEnergyMarketMonitorMarch2024_p9.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/CleanEnergyMarketMonitorMarch2024_p9.pdf"},"request":{"example_id":"chart/CleanEnergyMarketMonitorMarch2024_p9","product_type":"parse","source_file_path":"CleanEnergyMarketMonitorMarch2024_p9.pdf"},"started_at":"2026-11-04T19:45:47.095120"} +{"completed_at":"2026-11-04T19:45:52.851722","latency_in_ms":4731,"output":{"example_id":"chart/Consultancy-in-Spain.-The-industry-in-figures-2024_p10","grounded_pages":[],"layout_pages":[],"markdown":"CONSULTANCY IN SPAIN / The 2024 industry in figures\n\n10\n\n/ /\n\n14,517 13,714 12,944 12,190 11,270\n\n14,538 15,921\n\n19,659\n\n\\+0.1% +5.9% +5.9% +6.2% +9.5% +15.2%\\*\n\n21,575\n\n\\+9.7%\n\n21,982\n\n\\+8.8%\\*\n\n### 2024\n\nThe revenue brought in by consulting firms in 2024 totaled 21,982 million euros, which accounts for 8.8% growth in comparison to the year before. This upward trend observed held steady another yet another year again, reaffirming the relevant contribution of consulting firms to the digital transformation within companies and government agencies, and the industry’s leading role in adapting business models to a continuously evolving environment.\n\n# **Consulting Industry Revenue**\n\n### The consulting industry grew 8.8% in 2024, driven by increased revenue in the national market\n\n### The consulting industry continues to grow (millions of euros)\n\nSource: AEC (Spanish Association of Consulting Firms)\n\nIn the last year, there was a new increase in revenue in the national market while there was a slight decrease in the foreign market. Revenue in Spain totaled 15,381 million euros, which accounts for 13.6% growth over 2023, while it totaled 6,601 million euros in the international market, which is 1.1% less than the year before. This behavior reinforces the weight of the national market in the industry as a whole, all while indicating a slight\n\n\\*For a more accurate reading of the data, the comparative analysis of the evolution of revenue and employment between financial years 2021-2022 and 2023-2024 (variation rate) was completed taking into consideration only companies that have not gone throu- gh significant integration or corporate reorganization processes. These companies represented 82% of all sector revenue in 2021 and 89% in 2023.\n\nadjustment in the foreign market following several years of sustained growth.\n\nNext page Previous page Contents","pages":[{"line_numbers":[],"markdown":"CONSULTANCY IN SPAIN / The 2024 industry in figures\n\n10\n\n/ /\n\n14,517 13,714 12,944 12,190 11,270\n\n14,538 15,921\n\n19,659\n\n\\+0.1% +5.9% +5.9% +6.2% +9.5% +15.2%\\*\n\n21,575\n\n\\+9.7%\n\n21,982\n\n\\+8.8%\\*\n\n### 2024\n\nThe revenue brought in by consulting firms in 2024 totaled 21,982 million euros, which accounts for 8.8% growth in comparison to the year before. This upward trend observed held steady another yet another year again, reaffirming the relevant contribution of consulting firms to the digital transformation within companies and government agencies, and the industry’s leading role in adapting business models to a continuously evolving environment.\n\n# **Consulting Industry Revenue**\n\n### The consulting industry grew 8.8% in 2024, driven by increased revenue in the national market\n\n### The consulting industry continues to grow (millions of euros)\n\nSource: AEC (Spanish Association of Consulting Firms)\n\nIn the last year, there was a new increase in revenue in the national market while there was a slight decrease in the foreign market. Revenue in Spain totaled 15,381 million euros, which accounts for 13.6% growth over 2023, while it totaled 6,601 million euros in the international market, which is 1.1% less than the year before. This behavior reinforces the weight of the national market in the industry as a whole, all while indicating a slight\n\n\\*For a more accurate reading of the data, the comparative analysis of the evolution of revenue and employment between financial years 2021-2022 and 2023-2024 (variation rate) was completed taking into consideration only companies that have not gone throu- gh significant integration or corporate reorganization processes. These companies represented 82% of all sector revenue in 2021 and 89% in 2023.\n\nadjustment in the foreign market following several years of sustained growth.\n\nNext page Previous page Contents","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"CONSULTANCY IN SPAIN / The 2024 industry in figures\n\n10\n\n/ /\n\n14,517 13,714 12,944 12,190 11,270\n\n14,538 15,921\n\n19,659\n\n\\+0.1% +5.9% +5.9% +6.2% +9.5% +15.2%\\*\n\n21,575\n\n\\+9.7%\n\n21,982\n\n\\+8.8%\\*\n\n### 2024\n\nThe revenue brought in by consulting firms in 2024 totaled 21,982 million euros, which accounts for 8.8% growth in comparison to the year before. This upward trend observed held steady another yet another year again, reaffirming the relevant contribution of consulting firms to the digital transformation within companies and government agencies, and the industry’s leading role in adapting business models to a continuously evolving environment.\n\n# **Consulting Industry Revenue**\n\n### The consulting industry grew 8.8% in 2024, driven by increased revenue in the national market\n\n### The consulting industry continues to grow (millions of euros)\n\nSource: AEC (Spanish Association of Consulting Firms)\n\nIn the last year, there was a new increase in revenue in the national market while there was a slight decrease in the foreign market. Revenue in Spain totaled 15,381 million euros, which accounts for 13.6% growth over 2023, while it totaled 6,601 million euros in the international market, which is 1.1% less than the year before. This behavior reinforces the weight of the national market in the industry as a whole, all while indicating a slight\n\n\\*For a more accurate reading of the data, the comparative analysis of the evolution of revenue and employment between financial years 2021-2022 and 2023-2024 (variation rate) was completed taking into consideration only companies that have not gone throu- gh significant integration or corporate reorganization processes. These companies represented 82% of all sector revenue in 2021 and 89% in 2023.\n\nadjustment in the foreign market following several years of sustained growth.\n\nNext page Previous page Contents","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Consultancy-in-Spain.-The-industry-in-figures-2024_p10.pdf"]},"response":{"errors":[],"extract_id":"extract_4cb56c10a3f405b6db555a02925b9afe","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["CONSULTANCY IN SPAIN / The 2024 industry in figures\n\n10\n\n/ /\n\n14,517 13,714 12,944 12,190 11,270\n\n14,538 15,921\n\n19,659\n\n\\+0.1% +5.9% +5.9% +6.2% +9.5% +15.2%\\*\n\n21,575\n\n\\+9.7%\n\n21,982\n\n\\+8.8%\\*\n\n### 2024\n\nThe revenue brought in by consulting firms in 2024 totaled 21,982 million euros, which accounts for 8.8% growth in comparison to the year before. This upward trend observed held steady another yet another year again, reaffirming the relevant contribution of consulting firms to the digital transformation within companies and government agencies, and the industry’s leading role in adapting business models to a continuously evolving environment.\n\n# **Consulting Industry Revenue**\n\n### The consulting industry grew 8.8% in 2024, driven by increased revenue in the national market\n\n### The consulting industry continues to grow (millions of euros)\n\nSource: AEC (Spanish Association of Consulting Firms)\n\nIn the last year, there was a new increase in revenue in the national market while there was a slight decrease in the foreign market. Revenue in Spain totaled 15,381 million euros, which accounts for 13.6% growth over 2023, while it totaled 6,601 million euros in the international market, which is 1.1% less than the year before. This behavior reinforces the weight of the national market in the industry as a whole, all while indicating a slight\n\n\\*For a more accurate reading of the data, the comparative analysis of the evolution of revenue and employment between financial years 2021-2022 and 2023-2024 (variation rate) was completed taking into consideration only companies that have not gone throu- gh significant integration or corporate reorganization processes. These companies represented 82% of all sector revenue in 2021 and 89% in 2023.\n\nadjustment in the foreign market following several years of sustained growth.\n\nNext page Previous page Contents"],"full_content":"CONSULTANCY IN SPAIN / The 2024 industry in figures\n\n10\n\n/ /\n\n14,517 13,714 12,944 12,190 11,270\n\n14,538 15,921\n\n19,659\n\n\\+0.1% +5.9% +5.9% +6.2% +9.5% +15.2%\\*\n\n21,575\n\n\\+9.7%\n\n21,982\n\n\\+8.8%\\*\n\n### 2024\n\nThe revenue brought in by consulting firms in 2024 totaled 21,982 million euros, which accounts for 8.8% growth in comparison to the year before. This upward trend observed held steady another yet another year again, reaffirming the relevant contribution of consulting firms to the digital transformation within companies and government agencies, and the industry’s leading role in adapting business models to a continuously evolving environment.\n\n# **Consulting Industry Revenue**\n\n### The consulting industry grew 8.8% in 2024, driven by increased revenue in the national market\n\n### The consulting industry continues to grow (millions of euros)\n\nSource: AEC (Spanish Association of Consulting Firms)\n\nIn the last year, there was a new increase in revenue in the national market while there was a slight decrease in the foreign market. Revenue in Spain totaled 15,381 million euros, which accounts for 13.6% growth over 2023, while it totaled 6,601 million euros in the international market, which is 1.1% less than the year before. This behavior reinforces the weight of the national market in the industry as a whole, all while indicating a slight\n\n\\*For a more accurate reading of the data, the comparative analysis of the evolution of revenue and employment between financial years 2021-2022 and 2023-2024 (variation rate) was completed taking into consideration only companies that have not gone throu- gh significant integration or corporate reorganization processes. These companies represented 82% of all sector revenue in 2021 and 89% in 2023.\n\nadjustment in the foreign market following several years of sustained growth.\n\nNext page Previous page Contents","publish_date":null,"title":"Consulting Industry Revenue","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Consultancy-in-Spain.-The-industry-in-figures-2024_p10.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Consultancy-in-Spain.-The-industry-in-figures-2024_p10.pdf"},"request":{"example_id":"chart/Consultancy-in-Spain.-The-industry-in-figures-2024_p10","product_type":"parse","source_file_path":"Consultancy-in-Spain.-The-industry-in-figures-2024_p10.pdf"},"started_at":"2026-11-04T19:45:48.120337"} +{"completed_at":"2026-11-04T19:45:56.492739","latency_in_ms":7018,"output":{"example_id":"chart/Consultancy-in-Spain.-The-industry-in-figures-2024_p12","grounded_pages":[],"layout_pages":[],"markdown":"CONSULTANCY IN SPAIN / The 2024 industry in figures\n\n12\n\n/ /\n\n### 2019 2018 2017 2016 2015 2021 2022 2023 2024 2020\n\n### 5\\.1% 6.2% 5.9%\n\n### 3\\.8% 3.6% 4.0% 3.5%\n\n### 5\\.9%\n\n### 3\\.6%\n\n### 0\\.1%\n\n### \\-9.9%\n\n### 9\\.5%\n\n### 15\\.2%\n\n### 7\\.4%\n\n### 10\\.0%\n\n### 9\\.7%\n\n### 8\\.6%\n\n### 8\\.8%\n\n### 6\\.3%\n\n### 8\\.2%\n\nIn 2024, the consultancy industry continued its rise, driven by the growing demand for technological, strategic and operational solutions helping organi- zations in all sectors face the challenges of an ever more complex and changing environment, subject to fast-paced technological development.\n\nSector firms have proven their capacity to adapt, responding with agility to the new challenges and needs facing companies and government agencies. The increase in revenue recorded in 2024 strength- ens the position of consultancy as a key industry in the transformation of the production model and the evolution towards a more resilient economy.\n\nOnce again, the industry dynamics can be seen upon comparing its evolution to that of the entire\n\neconomy. In 2024, revenue generated by consulting firms grew 8.8%, in comparison to the 6.3% increase in nominal GDP. This difference — 2.5 percentage points — confirms a structural trend that has been repeating itself for more than a decade: consultan- cy has consistently grown above the GDP, position- ing itself as one of the main drivers of growth in the Spanish economy.\n\n### The increase in consultancy revenue in 2024 remained above the growth in nominal GDP in Spain\n\n### Consulting industry revenue is growing above the GDP (2015-2024)\n\nSources: : sector revenue: AEC (Spanish Association of Consulting Firms); GDP at current prices: INE (Spanish National Statistics Institute) (https://www.ine.es/dyngs/Prensa/avCNTR4T24.htm)\n\nMethodological change.\n\nConsulting industry growth GDP growth at current prices\n\nNext page Previous page Contents","pages":[{"line_numbers":[],"markdown":"CONSULTANCY IN SPAIN / The 2024 industry in figures\n\n12\n\n/ /\n\n### 2019 2018 2017 2016 2015 2021 2022 2023 2024 2020\n\n### 5\\.1% 6.2% 5.9%\n\n### 3\\.8% 3.6% 4.0% 3.5%\n\n### 5\\.9%\n\n### 3\\.6%\n\n### 0\\.1%\n\n### \\-9.9%\n\n### 9\\.5%\n\n### 15\\.2%\n\n### 7\\.4%\n\n### 10\\.0%\n\n### 9\\.7%\n\n### 8\\.6%\n\n### 8\\.8%\n\n### 6\\.3%\n\n### 8\\.2%\n\nIn 2024, the consultancy industry continued its rise, driven by the growing demand for technological, strategic and operational solutions helping organi- zations in all sectors face the challenges of an ever more complex and changing environment, subject to fast-paced technological development.\n\nSector firms have proven their capacity to adapt, responding with agility to the new challenges and needs facing companies and government agencies. The increase in revenue recorded in 2024 strength- ens the position of consultancy as a key industry in the transformation of the production model and the evolution towards a more resilient economy.\n\nOnce again, the industry dynamics can be seen upon comparing its evolution to that of the entire\n\neconomy. In 2024, revenue generated by consulting firms grew 8.8%, in comparison to the 6.3% increase in nominal GDP. This difference — 2.5 percentage points — confirms a structural trend that has been repeating itself for more than a decade: consultan- cy has consistently grown above the GDP, position- ing itself as one of the main drivers of growth in the Spanish economy.\n\n### The increase in consultancy revenue in 2024 remained above the growth in nominal GDP in Spain\n\n### Consulting industry revenue is growing above the GDP (2015-2024)\n\nSources: : sector revenue: AEC (Spanish Association of Consulting Firms); GDP at current prices: INE (Spanish National Statistics Institute) (https://www.ine.es/dyngs/Prensa/avCNTR4T24.htm)\n\nMethodological change.\n\nConsulting industry growth GDP growth at current prices\n\nNext page Previous page Contents","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"CONSULTANCY IN SPAIN / The 2024 industry in figures\n\n12\n\n/ /\n\n### 2019 2018 2017 2016 2015 2021 2022 2023 2024 2020\n\n### 5\\.1% 6.2% 5.9%\n\n### 3\\.8% 3.6% 4.0% 3.5%\n\n### 5\\.9%\n\n### 3\\.6%\n\n### 0\\.1%\n\n### \\-9.9%\n\n### 9\\.5%\n\n### 15\\.2%\n\n### 7\\.4%\n\n### 10\\.0%\n\n### 9\\.7%\n\n### 8\\.6%\n\n### 8\\.8%\n\n### 6\\.3%\n\n### 8\\.2%\n\nIn 2024, the consultancy industry continued its rise, driven by the growing demand for technological, strategic and operational solutions helping organi- zations in all sectors face the challenges of an ever more complex and changing environment, subject to fast-paced technological development.\n\nSector firms have proven their capacity to adapt, responding with agility to the new challenges and needs facing companies and government agencies. The increase in revenue recorded in 2024 strength- ens the position of consultancy as a key industry in the transformation of the production model and the evolution towards a more resilient economy.\n\nOnce again, the industry dynamics can be seen upon comparing its evolution to that of the entire\n\neconomy. In 2024, revenue generated by consulting firms grew 8.8%, in comparison to the 6.3% increase in nominal GDP. This difference — 2.5 percentage points — confirms a structural trend that has been repeating itself for more than a decade: consultan- cy has consistently grown above the GDP, position- ing itself as one of the main drivers of growth in the Spanish economy.\n\n### The increase in consultancy revenue in 2024 remained above the growth in nominal GDP in Spain\n\n### Consulting industry revenue is growing above the GDP (2015-2024)\n\nSources: : sector revenue: AEC (Spanish Association of Consulting Firms); GDP at current prices: INE (Spanish National Statistics Institute) (https://www.ine.es/dyngs/Prensa/avCNTR4T24.htm)\n\nMethodological change.\n\nConsulting industry growth GDP growth at current prices\n\nNext page Previous page Contents","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Consultancy-in-Spain.-The-industry-in-figures-2024_p12.pdf"]},"response":{"errors":[],"extract_id":"extract_98d82a3704e2cb41f761f2c6d7f4655d","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["CONSULTANCY IN SPAIN / The 2024 industry in figures\n\n12\n\n/ /\n\n### 2019 2018 2017 2016 2015 2021 2022 2023 2024 2020\n\n### 5\\.1% 6.2% 5.9%\n\n### 3\\.8% 3.6% 4.0% 3.5%\n\n### 5\\.9%\n\n### 3\\.6%\n\n### 0\\.1%\n\n### \\-9.9%\n\n### 9\\.5%\n\n### 15\\.2%\n\n### 7\\.4%\n\n### 10\\.0%\n\n### 9\\.7%\n\n### 8\\.6%\n\n### 8\\.8%\n\n### 6\\.3%\n\n### 8\\.2%\n\nIn 2024, the consultancy industry continued its rise, driven by the growing demand for technological, strategic and operational solutions helping organi- zations in all sectors face the challenges of an ever more complex and changing environment, subject to fast-paced technological development.\n\nSector firms have proven their capacity to adapt, responding with agility to the new challenges and needs facing companies and government agencies. The increase in revenue recorded in 2024 strength- ens the position of consultancy as a key industry in the transformation of the production model and the evolution towards a more resilient economy.\n\nOnce again, the industry dynamics can be seen upon comparing its evolution to that of the entire\n\neconomy. In 2024, revenue generated by consulting firms grew 8.8%, in comparison to the 6.3% increase in nominal GDP. This difference — 2.5 percentage points — confirms a structural trend that has been repeating itself for more than a decade: consultan- cy has consistently grown above the GDP, position- ing itself as one of the main drivers of growth in the Spanish economy.\n\n### The increase in consultancy revenue in 2024 remained above the growth in nominal GDP in Spain\n\n### Consulting industry revenue is growing above the GDP (2015-2024)\n\nSources: : sector revenue: AEC (Spanish Association of Consulting Firms); GDP at current prices: INE (Spanish National Statistics Institute) (https://www.ine.es/dyngs/Prensa/avCNTR4T24.htm)\n\nMethodological change.\n\nConsulting industry growth GDP growth at current prices\n\nNext page Previous page Contents"],"full_content":"CONSULTANCY IN SPAIN / The 2024 industry in figures\n\n12\n\n/ /\n\n### 2019 2018 2017 2016 2015 2021 2022 2023 2024 2020\n\n### 5\\.1% 6.2% 5.9%\n\n### 3\\.8% 3.6% 4.0% 3.5%\n\n### 5\\.9%\n\n### 3\\.6%\n\n### 0\\.1%\n\n### \\-9.9%\n\n### 9\\.5%\n\n### 15\\.2%\n\n### 7\\.4%\n\n### 10\\.0%\n\n### 9\\.7%\n\n### 8\\.6%\n\n### 8\\.8%\n\n### 6\\.3%\n\n### 8\\.2%\n\nIn 2024, the consultancy industry continued its rise, driven by the growing demand for technological, strategic and operational solutions helping organi- zations in all sectors face the challenges of an ever more complex and changing environment, subject to fast-paced technological development.\n\nSector firms have proven their capacity to adapt, responding with agility to the new challenges and needs facing companies and government agencies. The increase in revenue recorded in 2024 strength- ens the position of consultancy as a key industry in the transformation of the production model and the evolution towards a more resilient economy.\n\nOnce again, the industry dynamics can be seen upon comparing its evolution to that of the entire\n\neconomy. In 2024, revenue generated by consulting firms grew 8.8%, in comparison to the 6.3% increase in nominal GDP. This difference — 2.5 percentage points — confirms a structural trend that has been repeating itself for more than a decade: consultan- cy has consistently grown above the GDP, position- ing itself as one of the main drivers of growth in the Spanish economy.\n\n### The increase in consultancy revenue in 2024 remained above the growth in nominal GDP in Spain\n\n### Consulting industry revenue is growing above the GDP (2015-2024)\n\nSources: : sector revenue: AEC (Spanish Association of Consulting Firms); GDP at current prices: INE (Spanish National Statistics Institute) (https://www.ine.es/dyngs/Prensa/avCNTR4T24.htm)\n\nMethodological change.\n\nConsulting industry growth GDP growth at current prices\n\nNext page Previous page Contents","publish_date":null,"title":"The increase in consultancy revenue in 2024 remained above the growth in nominal GDP in Spain (2015-2024) 2019 2018 2017","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Consultancy-in-Spain.-The-industry-in-figures-2024_p12.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Consultancy-in-Spain.-The-industry-in-figures-2024_p12.pdf"},"request":{"example_id":"chart/Consultancy-in-Spain.-The-industry-in-figures-2024_p12","product_type":"parse","source_file_path":"Consultancy-in-Spain.-The-industry-in-figures-2024_p12.pdf"},"started_at":"2026-11-04T19:45:49.474163"} +{"completed_at":"2026-11-04T19:45:55.877399","latency_in_ms":5217,"output":{"example_id":"chart/Consultancy-in-Spain.-The-industry-in-figures-2024_p15","grounded_pages":[],"layout_pages":[],"markdown":"/ /\n\n15\n\n### 2018 2019 2020 2021 2022 2023 2017 2016 2015 2024\n\n100 97 99 96 94 93 93 97 100\n\n100\n\n118 129 140 141\n\n166\n\n183 194\n\n129 151\n\n174 191\n\n99\n\n195\n\n198\n\n129 122 115\n\n114 100\n\n108\n\n### Evolution of the workforce, revenue and productivity (measured as revenue per employee) (2015= base of 100)\n\nSource: AEC (Spanish Association of Consulting Firms)\n\n### 2018 2020 2022 2016 2024 2015 2017 2019 2021 2023\n\n3\\.2%\n\n3\\.5% 3.6%\n\n2\\.5% 2.0% 2.7% 2.5%\n\n3\\.1%\n\n2\\.3%\n\n2\\.6%\n\n5\\.3% 4.0%\n\n\\-2.1%\n\n2\\.5% 3.9%\n\n4\\.4%\n\n3\\.9%\n\n3\\.5%\n\n2\\.3%\n\n9\\.4%\n\n3\\.4%\n\n8\\.9%\n\n0\\.7%\n\n11\\.9%\n\n8\\.6%\n\n6\\.3% 8.5%\n\n3\\.4%\n\n14\\.1%\n\n0\\.4%\n\n### Workforce growth rate at Spanish consulting firms, of third-party employees in Spain and of employment in knowledge-intensive jobs in Spain (2015-2024)\n\nSources: Spanish consulting firm workforce: AEC (Spanish Association of Consulting Firms); third- party employees registered with the Spanish Social Security system in Spain: Spanish Ministry of Labor and Social Economy Statistics Annual; knowledge-intensive employment: Eurostat.\n\nMethodological change.\n\nConsulting firm workforce\n\nThird-party employees registered with the Spanish Social Security system\n\nKnowledge-intensive employment\n\nRevenue Productivity (revenue per employee) Workforce\n\nNext page Previous page Contents","pages":[{"line_numbers":[],"markdown":"/ /\n\n15\n\n### 2018 2019 2020 2021 2022 2023 2017 2016 2015 2024\n\n100 97 99 96 94 93 93 97 100\n\n100\n\n118 129 140 141\n\n166\n\n183 194\n\n129 151\n\n174 191\n\n99\n\n195\n\n198\n\n129 122 115\n\n114 100\n\n108\n\n### Evolution of the workforce, revenue and productivity (measured as revenue per employee) (2015= base of 100)\n\nSource: AEC (Spanish Association of Consulting Firms)\n\n### 2018 2020 2022 2016 2024 2015 2017 2019 2021 2023\n\n3\\.2%\n\n3\\.5% 3.6%\n\n2\\.5% 2.0% 2.7% 2.5%\n\n3\\.1%\n\n2\\.3%\n\n2\\.6%\n\n5\\.3% 4.0%\n\n\\-2.1%\n\n2\\.5% 3.9%\n\n4\\.4%\n\n3\\.9%\n\n3\\.5%\n\n2\\.3%\n\n9\\.4%\n\n3\\.4%\n\n8\\.9%\n\n0\\.7%\n\n11\\.9%\n\n8\\.6%\n\n6\\.3% 8.5%\n\n3\\.4%\n\n14\\.1%\n\n0\\.4%\n\n### Workforce growth rate at Spanish consulting firms, of third-party employees in Spain and of employment in knowledge-intensive jobs in Spain (2015-2024)\n\nSources: Spanish consulting firm workforce: AEC (Spanish Association of Consulting Firms); third- party employees registered with the Spanish Social Security system in Spain: Spanish Ministry of Labor and Social Economy Statistics Annual; knowledge-intensive employment: Eurostat.\n\nMethodological change.\n\nConsulting firm workforce\n\nThird-party employees registered with the Spanish Social Security system\n\nKnowledge-intensive employment\n\nRevenue Productivity (revenue per employee) Workforce\n\nNext page Previous page Contents","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"/ /\n\n15\n\n### 2018 2019 2020 2021 2022 2023 2017 2016 2015 2024\n\n100 97 99 96 94 93 93 97 100\n\n100\n\n118 129 140 141\n\n166\n\n183 194\n\n129 151\n\n174 191\n\n99\n\n195\n\n198\n\n129 122 115\n\n114 100\n\n108\n\n### Evolution of the workforce, revenue and productivity (measured as revenue per employee) (2015= base of 100)\n\nSource: AEC (Spanish Association of Consulting Firms)\n\n### 2018 2020 2022 2016 2024 2015 2017 2019 2021 2023\n\n3\\.2%\n\n3\\.5% 3.6%\n\n2\\.5% 2.0% 2.7% 2.5%\n\n3\\.1%\n\n2\\.3%\n\n2\\.6%\n\n5\\.3% 4.0%\n\n\\-2.1%\n\n2\\.5% 3.9%\n\n4\\.4%\n\n3\\.9%\n\n3\\.5%\n\n2\\.3%\n\n9\\.4%\n\n3\\.4%\n\n8\\.9%\n\n0\\.7%\n\n11\\.9%\n\n8\\.6%\n\n6\\.3% 8.5%\n\n3\\.4%\n\n14\\.1%\n\n0\\.4%\n\n### Workforce growth rate at Spanish consulting firms, of third-party employees in Spain and of employment in knowledge-intensive jobs in Spain (2015-2024)\n\nSources: Spanish consulting firm workforce: AEC (Spanish Association of Consulting Firms); third- party employees registered with the Spanish Social Security system in Spain: Spanish Ministry of Labor and Social Economy Statistics Annual; knowledge-intensive employment: Eurostat.\n\nMethodological change.\n\nConsulting firm workforce\n\nThird-party employees registered with the Spanish Social Security system\n\nKnowledge-intensive employment\n\nRevenue Productivity (revenue per employee) Workforce\n\nNext page Previous page Contents","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Consultancy-in-Spain.-The-industry-in-figures-2024_p15.pdf"]},"response":{"errors":[],"extract_id":"extract_e083c63bdb1129f023e38e4b34fc35db","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["/ /\n\n15\n\n### 2018 2019 2020 2021 2022 2023 2017 2016 2015 2024\n\n100 97 99 96 94 93 93 97 100\n\n100\n\n118 129 140 141\n\n166\n\n183 194\n\n129 151\n\n174 191\n\n99\n\n195\n\n198\n\n129 122 115\n\n114 100\n\n108\n\n### Evolution of the workforce, revenue and productivity (measured as revenue per employee) (2015= base of 100)\n\nSource: AEC (Spanish Association of Consulting Firms)\n\n### 2018 2020 2022 2016 2024 2015 2017 2019 2021 2023\n\n3\\.2%\n\n3\\.5% 3.6%\n\n2\\.5% 2.0% 2.7% 2.5%\n\n3\\.1%\n\n2\\.3%\n\n2\\.6%\n\n5\\.3% 4.0%\n\n\\-2.1%\n\n2\\.5% 3.9%\n\n4\\.4%\n\n3\\.9%\n\n3\\.5%\n\n2\\.3%\n\n9\\.4%\n\n3\\.4%\n\n8\\.9%\n\n0\\.7%\n\n11\\.9%\n\n8\\.6%\n\n6\\.3% 8.5%\n\n3\\.4%\n\n14\\.1%\n\n0\\.4%\n\n### Workforce growth rate at Spanish consulting firms, of third-party employees in Spain and of employment in knowledge-intensive jobs in Spain (2015-2024)\n\nSources: Spanish consulting firm workforce: AEC (Spanish Association of Consulting Firms); third- party employees registered with the Spanish Social Security system in Spain: Spanish Ministry of Labor and Social Economy Statistics Annual; knowledge-intensive employment: Eurostat.\n\nMethodological change.\n\nConsulting firm workforce\n\nThird-party employees registered with the Spanish Social Security system\n\nKnowledge-intensive employment\n\nRevenue Productivity (revenue per employee) Workforce\n\nNext page Previous page Contents"],"full_content":"/ /\n\n15\n\n### 2018 2019 2020 2021 2022 2023 2017 2016 2015 2024\n\n100 97 99 96 94 93 93 97 100\n\n100\n\n118 129 140 141\n\n166\n\n183 194\n\n129 151\n\n174 191\n\n99\n\n195\n\n198\n\n129 122 115\n\n114 100\n\n108\n\n### Evolution of the workforce, revenue and productivity (measured as revenue per employee) (2015= base of 100)\n\nSource: AEC (Spanish Association of Consulting Firms)\n\n### 2018 2020 2022 2016 2024 2015 2017 2019 2021 2023\n\n3\\.2%\n\n3\\.5% 3.6%\n\n2\\.5% 2.0% 2.7% 2.5%\n\n3\\.1%\n\n2\\.3%\n\n2\\.6%\n\n5\\.3% 4.0%\n\n\\-2.1%\n\n2\\.5% 3.9%\n\n4\\.4%\n\n3\\.9%\n\n3\\.5%\n\n2\\.3%\n\n9\\.4%\n\n3\\.4%\n\n8\\.9%\n\n0\\.7%\n\n11\\.9%\n\n8\\.6%\n\n6\\.3% 8.5%\n\n3\\.4%\n\n14\\.1%\n\n0\\.4%\n\n### Workforce growth rate at Spanish consulting firms, of third-party employees in Spain and of employment in knowledge-intensive jobs in Spain (2015-2024)\n\nSources: Spanish consulting firm workforce: AEC (Spanish Association of Consulting Firms); third- party employees registered with the Spanish Social Security system in Spain: Spanish Ministry of Labor and Social Economy Statistics Annual; knowledge-intensive employment: Eurostat.\n\nMethodological change.\n\nConsulting firm workforce\n\nThird-party employees registered with the Spanish Social Security system\n\nKnowledge-intensive employment\n\nRevenue Productivity (revenue per employee) Workforce\n\nNext page Previous page Contents","publish_date":null,"title":"Evolution of the workforce, revenue and productivity (measured as revenue per employee) (2015= base of 100) Workforce growth rate at","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Consultancy-in-Spain.-The-industry-in-figures-2024_p15.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Consultancy-in-Spain.-The-industry-in-figures-2024_p15.pdf"},"request":{"example_id":"chart/Consultancy-in-Spain.-The-industry-in-figures-2024_p15","product_type":"parse","source_file_path":"Consultancy-in-Spain.-The-industry-in-figures-2024_p15.pdf"},"started_at":"2026-11-04T19:45:50.659481"} +{"completed_at":"2026-11-04T19:45:57.912995","latency_in_ms":5303,"output":{"example_id":"chart/Consultancy-in-Spain.-The-industry-in-figures-2024_p16","grounded_pages":[],"layout_pages":[],"markdown":"CONSULTANCY IN SPAIN / The 2024 industry in figures\n\n16\n\n/ /\n\nTitulados áreas STEM Titulados otras áreas\n\n### men Women Total\n\n30\\.2%\n\n69\\.8%\n\n44\\.1%\n\n55\\.9%\n\n35\\.1%\n\n64\\.9%\n\nSource: AEC (Spanish Association of Consulting Firms)\n\n### University degree holders from STEM areas (percentage in Spain)\n\nUniversitarios FP Otros\n\n### Men Women Total\n\n### 63\\.8%\n\n### 22\\.9%\n\n### 13\\.3%\n\n### 70\\.8%\n\n### 13\\.9% 15\\.3%\n\n### 65\\.8%\n\n### 20\\.1% 14.1%\n\nSource: AEC (Spanish Association of Consulting Firms)\n\n### Most of the consultancy industry’s professionals hold university degrees (percentage in Spain)\n\nDue to the type of services offered, consulting fir- ms attract the most qualified profiles in the job market. 65.8% of the total number of professio- nals in 2024 were university graduates. Another year again, the percentage of women with a hi- gher education was above that of men (70.8% ver- sus 63.8%).\n\nOn the other hand, the presence of vocational tra- ining graduates dropped with respect to the year before, representing 20.1% of the total in 2.24. In this case, the percentage of women, 13.9% of the total, was less than the men, which represented 22.9%.\n\nDegree holders in STEM fields Degree holders in other fields\n\nUniversity graduates Vocational training Other\n\nNext page Previous page Contents","pages":[{"line_numbers":[],"markdown":"CONSULTANCY IN SPAIN / The 2024 industry in figures\n\n16\n\n/ /\n\nTitulados áreas STEM Titulados otras áreas\n\n### men Women Total\n\n30\\.2%\n\n69\\.8%\n\n44\\.1%\n\n55\\.9%\n\n35\\.1%\n\n64\\.9%\n\nSource: AEC (Spanish Association of Consulting Firms)\n\n### University degree holders from STEM areas (percentage in Spain)\n\nUniversitarios FP Otros\n\n### Men Women Total\n\n### 63\\.8%\n\n### 22\\.9%\n\n### 13\\.3%\n\n### 70\\.8%\n\n### 13\\.9% 15\\.3%\n\n### 65\\.8%\n\n### 20\\.1% 14.1%\n\nSource: AEC (Spanish Association of Consulting Firms)\n\n### Most of the consultancy industry’s professionals hold university degrees (percentage in Spain)\n\nDue to the type of services offered, consulting fir- ms attract the most qualified profiles in the job market. 65.8% of the total number of professio- nals in 2024 were university graduates. Another year again, the percentage of women with a hi- gher education was above that of men (70.8% ver- sus 63.8%).\n\nOn the other hand, the presence of vocational tra- ining graduates dropped with respect to the year before, representing 20.1% of the total in 2.24. In this case, the percentage of women, 13.9% of the total, was less than the men, which represented 22.9%.\n\nDegree holders in STEM fields Degree holders in other fields\n\nUniversity graduates Vocational training Other\n\nNext page Previous page Contents","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"CONSULTANCY IN SPAIN / The 2024 industry in figures\n\n16\n\n/ /\n\nTitulados áreas STEM Titulados otras áreas\n\n### men Women Total\n\n30\\.2%\n\n69\\.8%\n\n44\\.1%\n\n55\\.9%\n\n35\\.1%\n\n64\\.9%\n\nSource: AEC (Spanish Association of Consulting Firms)\n\n### University degree holders from STEM areas (percentage in Spain)\n\nUniversitarios FP Otros\n\n### Men Women Total\n\n### 63\\.8%\n\n### 22\\.9%\n\n### 13\\.3%\n\n### 70\\.8%\n\n### 13\\.9% 15\\.3%\n\n### 65\\.8%\n\n### 20\\.1% 14.1%\n\nSource: AEC (Spanish Association of Consulting Firms)\n\n### Most of the consultancy industry’s professionals hold university degrees (percentage in Spain)\n\nDue to the type of services offered, consulting fir- ms attract the most qualified profiles in the job market. 65.8% of the total number of professio- nals in 2024 were university graduates. Another year again, the percentage of women with a hi- gher education was above that of men (70.8% ver- sus 63.8%).\n\nOn the other hand, the presence of vocational tra- ining graduates dropped with respect to the year before, representing 20.1% of the total in 2.24. In this case, the percentage of women, 13.9% of the total, was less than the men, which represented 22.9%.\n\nDegree holders in STEM fields Degree holders in other fields\n\nUniversity graduates Vocational training Other\n\nNext page Previous page Contents","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Consultancy-in-Spain.-The-industry-in-figures-2024_p16.pdf"]},"response":{"errors":[],"extract_id":"extract_cfbaa14f1a6b7ce46194141341e842b1","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["CONSULTANCY IN SPAIN / The 2024 industry in figures\n\n16\n\n/ /\n\nTitulados áreas STEM Titulados otras áreas\n\n### men Women Total\n\n30\\.2%\n\n69\\.8%\n\n44\\.1%\n\n55\\.9%\n\n35\\.1%\n\n64\\.9%\n\nSource: AEC (Spanish Association of Consulting Firms)\n\n### University degree holders from STEM areas (percentage in Spain)\n\nUniversitarios FP Otros\n\n### Men Women Total\n\n### 63\\.8%\n\n### 22\\.9%\n\n### 13\\.3%\n\n### 70\\.8%\n\n### 13\\.9% 15\\.3%\n\n### 65\\.8%\n\n### 20\\.1% 14.1%\n\nSource: AEC (Spanish Association of Consulting Firms)\n\n### Most of the consultancy industry’s professionals hold university degrees (percentage in Spain)\n\nDue to the type of services offered, consulting fir- ms attract the most qualified profiles in the job market. 65.8% of the total number of professio- nals in 2024 were university graduates. Another year again, the percentage of women with a hi- gher education was above that of men (70.8% ver- sus 63.8%).\n\nOn the other hand, the presence of vocational tra- ining graduates dropped with respect to the year before, representing 20.1% of the total in 2.24. In this case, the percentage of women, 13.9% of the total, was less than the men, which represented 22.9%.\n\nDegree holders in STEM fields Degree holders in other fields\n\nUniversity graduates Vocational training Other\n\nNext page Previous page Contents"],"full_content":"CONSULTANCY IN SPAIN / The 2024 industry in figures\n\n16\n\n/ /\n\nTitulados áreas STEM Titulados otras áreas\n\n### men Women Total\n\n30\\.2%\n\n69\\.8%\n\n44\\.1%\n\n55\\.9%\n\n35\\.1%\n\n64\\.9%\n\nSource: AEC (Spanish Association of Consulting Firms)\n\n### University degree holders from STEM areas (percentage in Spain)\n\nUniversitarios FP Otros\n\n### Men Women Total\n\n### 63\\.8%\n\n### 22\\.9%\n\n### 13\\.3%\n\n### 70\\.8%\n\n### 13\\.9% 15\\.3%\n\n### 65\\.8%\n\n### 20\\.1% 14.1%\n\nSource: AEC (Spanish Association of Consulting Firms)\n\n### Most of the consultancy industry’s professionals hold university degrees (percentage in Spain)\n\nDue to the type of services offered, consulting fir- ms attract the most qualified profiles in the job market. 65.8% of the total number of professio- nals in 2024 were university graduates. Another year again, the percentage of women with a hi- gher education was above that of men (70.8% ver- sus 63.8%).\n\nOn the other hand, the presence of vocational tra- ining graduates dropped with respect to the year before, representing 20.1% of the total in 2.24. In this case, the percentage of women, 13.9% of the total, was less than the men, which represented 22.9%.\n\nDegree holders in STEM fields Degree holders in other fields\n\nUniversity graduates Vocational training Other\n\nNext page Previous page Contents","publish_date":null,"title":"Most of the consultancy industry’s professionals hold university degrees (percentage in Spain) University degree holders from STEM areas","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Consultancy-in-Spain.-The-industry-in-figures-2024_p16.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Consultancy-in-Spain.-The-industry-in-figures-2024_p16.pdf"},"request":{"example_id":"chart/Consultancy-in-Spain.-The-industry-in-figures-2024_p16","product_type":"parse","source_file_path":"Consultancy-in-Spain.-The-industry-in-figures-2024_p16.pdf"},"started_at":"2026-11-04T19:45:52.609761"} +{"completed_at":"2026-11-04T19:45:59.486030","latency_in_ms":5274,"output":{"example_id":"chart/Consultancy-in-Spain.-The-industry-in-figures-2024_p19","grounded_pages":[],"layout_pages":[],"markdown":"/ /\n\n19\n\nInvestment in training per employee at Spanish consulting firms (euros)\n\nInvestment in training per employee in Spain (euros)\n\nTraining hours per employee at Spanish consulting firms (annual hours)\n\nInvestment in training by Spanish consulting firms (millions of euros) 64.7\n\n348\n\n100\n\n30\\.5\n\n2018\n\n74\\.7\n\n370\n\n104\n\n29\\.1\n\n2019\n\n68\\.1\n\n335\n\n106\n\n26\\.5\n\n2020\n\n79\n\n347\n\n109\n\n25\\.7\n\n2021\n\n78\\.5\n\n297\n\n112\n\n26\\.6\n\n2022\n\n92\\.2\n\n328\n\n120\n\n26\\.1\n\n2023\n\n102\\.0\n\n358\n\n125\n\n29\\.7\n\n2024\n\n58\\.2\n\n342\n\n94\n\n32\\.2\n\n2017\n\n53\\.8\n\n329\n\n91\n\n30\\.9\n\n2016\n\n### Indicators of the investment in training among Spanish consulting firms and the economy\n\nSource: AEC (Spanish Association of Consulting Firms)\n\n### 329\n\n### 91\n\n### 342\n\n### 94\n\n### 348\n\n### 100\n\n### 370\n\n### 104\n\n### 2018\n\n### 335\n\n### 106\n\n### 2020 2019\n\n### 347\n\n### 109\n\n### 2021\n\n### 297\n\n### 112\n\n### 2022\n\n### 328\n\n### 120\n\n### 2023\n\n### 358\n\n### 125\n\n### 2024 2017 2016\n\n### Sector de la consultoría Economía española\n\n### Increase in the investment in training per employee (euros)\n\nSource: AEC (Spanish Association of Consulting Firms)\n\n## Investment in training totaled 102 million euros in 2024, which is 10.6% higher than the year before\n\nConsultancy industry Spanish economy\n\nNext page Previous page Contents","pages":[{"line_numbers":[],"markdown":"/ /\n\n19\n\nInvestment in training per employee at Spanish consulting firms (euros)\n\nInvestment in training per employee in Spain (euros)\n\nTraining hours per employee at Spanish consulting firms (annual hours)\n\nInvestment in training by Spanish consulting firms (millions of euros) 64.7\n\n348\n\n100\n\n30\\.5\n\n2018\n\n74\\.7\n\n370\n\n104\n\n29\\.1\n\n2019\n\n68\\.1\n\n335\n\n106\n\n26\\.5\n\n2020\n\n79\n\n347\n\n109\n\n25\\.7\n\n2021\n\n78\\.5\n\n297\n\n112\n\n26\\.6\n\n2022\n\n92\\.2\n\n328\n\n120\n\n26\\.1\n\n2023\n\n102\\.0\n\n358\n\n125\n\n29\\.7\n\n2024\n\n58\\.2\n\n342\n\n94\n\n32\\.2\n\n2017\n\n53\\.8\n\n329\n\n91\n\n30\\.9\n\n2016\n\n### Indicators of the investment in training among Spanish consulting firms and the economy\n\nSource: AEC (Spanish Association of Consulting Firms)\n\n### 329\n\n### 91\n\n### 342\n\n### 94\n\n### 348\n\n### 100\n\n### 370\n\n### 104\n\n### 2018\n\n### 335\n\n### 106\n\n### 2020 2019\n\n### 347\n\n### 109\n\n### 2021\n\n### 297\n\n### 112\n\n### 2022\n\n### 328\n\n### 120\n\n### 2023\n\n### 358\n\n### 125\n\n### 2024 2017 2016\n\n### Sector de la consultoría Economía española\n\n### Increase in the investment in training per employee (euros)\n\nSource: AEC (Spanish Association of Consulting Firms)\n\n## Investment in training totaled 102 million euros in 2024, which is 10.6% higher than the year before\n\nConsultancy industry Spanish economy\n\nNext page Previous page Contents","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"/ /\n\n19\n\nInvestment in training per employee at Spanish consulting firms (euros)\n\nInvestment in training per employee in Spain (euros)\n\nTraining hours per employee at Spanish consulting firms (annual hours)\n\nInvestment in training by Spanish consulting firms (millions of euros) 64.7\n\n348\n\n100\n\n30\\.5\n\n2018\n\n74\\.7\n\n370\n\n104\n\n29\\.1\n\n2019\n\n68\\.1\n\n335\n\n106\n\n26\\.5\n\n2020\n\n79\n\n347\n\n109\n\n25\\.7\n\n2021\n\n78\\.5\n\n297\n\n112\n\n26\\.6\n\n2022\n\n92\\.2\n\n328\n\n120\n\n26\\.1\n\n2023\n\n102\\.0\n\n358\n\n125\n\n29\\.7\n\n2024\n\n58\\.2\n\n342\n\n94\n\n32\\.2\n\n2017\n\n53\\.8\n\n329\n\n91\n\n30\\.9\n\n2016\n\n### Indicators of the investment in training among Spanish consulting firms and the economy\n\nSource: AEC (Spanish Association of Consulting Firms)\n\n### 329\n\n### 91\n\n### 342\n\n### 94\n\n### 348\n\n### 100\n\n### 370\n\n### 104\n\n### 2018\n\n### 335\n\n### 106\n\n### 2020 2019\n\n### 347\n\n### 109\n\n### 2021\n\n### 297\n\n### 112\n\n### 2022\n\n### 328\n\n### 120\n\n### 2023\n\n### 358\n\n### 125\n\n### 2024 2017 2016\n\n### Sector de la consultoría Economía española\n\n### Increase in the investment in training per employee (euros)\n\nSource: AEC (Spanish Association of Consulting Firms)\n\n## Investment in training totaled 102 million euros in 2024, which is 10.6% higher than the year before\n\nConsultancy industry Spanish economy\n\nNext page Previous page Contents","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Consultancy-in-Spain.-The-industry-in-figures-2024_p19.pdf"]},"response":{"errors":[],"extract_id":"extract_7a91b0ce2fb691bc2a48bd9b10a33b81","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["/ /\n\n19\n\nInvestment in training per employee at Spanish consulting firms (euros)\n\nInvestment in training per employee in Spain (euros)\n\nTraining hours per employee at Spanish consulting firms (annual hours)\n\nInvestment in training by Spanish consulting firms (millions of euros) 64.7\n\n348\n\n100\n\n30\\.5\n\n2018\n\n74\\.7\n\n370\n\n104\n\n29\\.1\n\n2019\n\n68\\.1\n\n335\n\n106\n\n26\\.5\n\n2020\n\n79\n\n347\n\n109\n\n25\\.7\n\n2021\n\n78\\.5\n\n297\n\n112\n\n26\\.6\n\n2022\n\n92\\.2\n\n328\n\n120\n\n26\\.1\n\n2023\n\n102\\.0\n\n358\n\n125\n\n29\\.7\n\n2024\n\n58\\.2\n\n342\n\n94\n\n32\\.2\n\n2017\n\n53\\.8\n\n329\n\n91\n\n30\\.9\n\n2016\n\n### Indicators of the investment in training among Spanish consulting firms and the economy\n\nSource: AEC (Spanish Association of Consulting Firms)\n\n### 329\n\n### 91\n\n### 342\n\n### 94\n\n### 348\n\n### 100\n\n### 370\n\n### 104\n\n### 2018\n\n### 335\n\n### 106\n\n### 2020 2019\n\n### 347\n\n### 109\n\n### 2021\n\n### 297\n\n### 112\n\n### 2022\n\n### 328\n\n### 120\n\n### 2023\n\n### 358\n\n### 125\n\n### 2024 2017 2016\n\n### Sector de la consultoría Economía española\n\n### Increase in the investment in training per employee (euros)\n\nSource: AEC (Spanish Association of Consulting Firms)\n\n## Investment in training totaled 102 million euros in 2024, which is 10.6% higher than the year before\n\nConsultancy industry Spanish economy\n\nNext page Previous page Contents"],"full_content":"/ /\n\n19\n\nInvestment in training per employee at Spanish consulting firms (euros)\n\nInvestment in training per employee in Spain (euros)\n\nTraining hours per employee at Spanish consulting firms (annual hours)\n\nInvestment in training by Spanish consulting firms (millions of euros) 64.7\n\n348\n\n100\n\n30\\.5\n\n2018\n\n74\\.7\n\n370\n\n104\n\n29\\.1\n\n2019\n\n68\\.1\n\n335\n\n106\n\n26\\.5\n\n2020\n\n79\n\n347\n\n109\n\n25\\.7\n\n2021\n\n78\\.5\n\n297\n\n112\n\n26\\.6\n\n2022\n\n92\\.2\n\n328\n\n120\n\n26\\.1\n\n2023\n\n102\\.0\n\n358\n\n125\n\n29\\.7\n\n2024\n\n58\\.2\n\n342\n\n94\n\n32\\.2\n\n2017\n\n53\\.8\n\n329\n\n91\n\n30\\.9\n\n2016\n\n### Indicators of the investment in training among Spanish consulting firms and the economy\n\nSource: AEC (Spanish Association of Consulting Firms)\n\n### 329\n\n### 91\n\n### 342\n\n### 94\n\n### 348\n\n### 100\n\n### 370\n\n### 104\n\n### 2018\n\n### 335\n\n### 106\n\n### 2020 2019\n\n### 347\n\n### 109\n\n### 2021\n\n### 297\n\n### 112\n\n### 2022\n\n### 328\n\n### 120\n\n### 2023\n\n### 358\n\n### 125\n\n### 2024 2017 2016\n\n### Sector de la consultoría Economía española\n\n### Increase in the investment in training per employee (euros)\n\nSource: AEC (Spanish Association of Consulting Firms)\n\n## Investment in training totaled 102 million euros in 2024, which is 10.6% higher than the year before\n\nConsultancy industry Spanish economy\n\nNext page Previous page Contents","publish_date":null,"title":"Investment in training totaled 102 million euros in 2024, which is 10.6% higher than the year before","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Consultancy-in-Spain.-The-industry-in-figures-2024_p19.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Consultancy-in-Spain.-The-industry-in-figures-2024_p19.pdf"},"request":{"example_id":"chart/Consultancy-in-Spain.-The-industry-in-figures-2024_p19","product_type":"parse","source_file_path":"Consultancy-in-Spain.-The-industry-in-figures-2024_p19.pdf"},"started_at":"2026-11-04T19:45:54.211968"} +{"completed_at":"2026-11-04T19:45:56.947381","latency_in_ms":1063,"output":{"example_id":"chart/DI_2025-Global-Human-Capital-Trends_p14","grounded_pages":[],"layout_pages":[],"markdown":"**12** _Stagility: Creating stability for workers for organizations to move at speed_\n\ncreate value for workers and the organization. ³ And 85% say that organizations need to create more agile ways of organizing work to swiftly adapt to market changes. ⁴\n\nA disconnect between executives and workers is emerging, with leaders preparing for more agile ways of working and workers favoring stability (figure 1). The challenge is that the ability for organizations to evolve and adopt new models of work depends largely on the ability of the workforce to do the same. How can leaders find the right balance between the stability that workers crave and the agility that organizations need to create _stagility_ ?\n\nFor many organizations there is a gap between knowledge and action. According to our 2025 Global Human Capital Trends survey (see “ **Methodology** ”), a significant majority of organizations (72%) recognize the importance of balancing agility and stability, yet only 39% are doing something meaningful about it (figure 2). Whether organizations can strike this balance will likely depend on their ability to find ways to anchor the work, the organization, and the worker. These new anchors should provide a sense of identity, a clearly defined path of action, and a mental model for the way work _works_ . They should also provide a sense of grounding and direction, binding worker to organization in a shared vision that benefits both parties.\n\nTraditional anchors that have long provided stability for both organizations and workers are being challenged. These anchors include static job descriptions, defined teams, and linear, internal career pathways and employment models. As these anchors become upended, workers may struggle to find the grounding they require to act with confidence and creativity and to have a dynamic capability to respond and evolve.\n\nOrganizations and workers must find new sources of stability to reinforce the connection between worker and organization, putting both in a position to adapt and thrive. These new anchors will come in the form of reimagined organization structures, greater intention- ality in the design and resourcing of work, a better understanding of workers as individuals, and stronger networks within and beyond the four walls of the organization.\n\nSources of instability for work, organization, and worker\n\nFor organizations and workers to establish new anchors, they should first assess the sources of instability they face. These fall under three broad categories: work, organization, and worker.\n\n_Figure 1_\n\n### The changing nature of work no longer reflects workers’ ideal ways of working\n\n_Percentage of workers answering the question, “Which of the following describes your ideal way of working,” and percentage of leaders answering the question, “Which of the following statements best describes the changing nature of work in your organization?”_\n\nSource: Analysis of Deloitte’s 2025 Global Human Capital Trends survey data.\n\n0% 10% 20% 30% 40% 50% 60% 70%\n\n**Frequent changes in where and when work is done depending on the need**\n\n**Frequent changes in how work is done depending on the need**\n\n**Broadly defined jobs or positions, anchored to outcomes or challenges instead of specific tasks**\n\n**Shifting team(s) or workgroups**\n\n**Less stability**\n\n**Leaders Workers**","pages":[{"line_numbers":[],"markdown":"**12** _Stagility: Creating stability for workers for organizations to move at speed_\n\ncreate value for workers and the organization. ³ And 85% say that organizations need to create more agile ways of organizing work to swiftly adapt to market changes. ⁴\n\nA disconnect between executives and workers is emerging, with leaders preparing for more agile ways of working and workers favoring stability (figure 1). The challenge is that the ability for organizations to evolve and adopt new models of work depends largely on the ability of the workforce to do the same. How can leaders find the right balance between the stability that workers crave and the agility that organizations need to create _stagility_ ?\n\nFor many organizations there is a gap between knowledge and action. According to our 2025 Global Human Capital Trends survey (see “ **Methodology** ”), a significant majority of organizations (72%) recognize the importance of balancing agility and stability, yet only 39% are doing something meaningful about it (figure 2). Whether organizations can strike this balance will likely depend on their ability to find ways to anchor the work, the organization, and the worker. These new anchors should provide a sense of identity, a clearly defined path of action, and a mental model for the way work _works_ . They should also provide a sense of grounding and direction, binding worker to organization in a shared vision that benefits both parties.\n\nTraditional anchors that have long provided stability for both organizations and workers are being challenged. These anchors include static job descriptions, defined teams, and linear, internal career pathways and employment models. As these anchors become upended, workers may struggle to find the grounding they require to act with confidence and creativity and to have a dynamic capability to respond and evolve.\n\nOrganizations and workers must find new sources of stability to reinforce the connection between worker and organization, putting both in a position to adapt and thrive. These new anchors will come in the form of reimagined organization structures, greater intention- ality in the design and resourcing of work, a better understanding of workers as individuals, and stronger networks within and beyond the four walls of the organization.\n\nSources of instability for work, organization, and worker\n\nFor organizations and workers to establish new anchors, they should first assess the sources of instability they face. These fall under three broad categories: work, organization, and worker.\n\n_Figure 1_\n\n### The changing nature of work no longer reflects workers’ ideal ways of working\n\n_Percentage of workers answering the question, “Which of the following describes your ideal way of working,” and percentage of leaders answering the question, “Which of the following statements best describes the changing nature of work in your organization?”_\n\nSource: Analysis of Deloitte’s 2025 Global Human Capital Trends survey data.\n\n0% 10% 20% 30% 40% 50% 60% 70%\n\n**Frequent changes in where and when work is done depending on the need**\n\n**Frequent changes in how work is done depending on the need**\n\n**Broadly defined jobs or positions, anchored to outcomes or challenges instead of specific tasks**\n\n**Shifting team(s) or workgroups**\n\n**Less stability**\n\n**Leaders Workers**","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**12** _Stagility: Creating stability for workers for organizations to move at speed_\n\ncreate value for workers and the organization. ³ And 85% say that organizations need to create more agile ways of organizing work to swiftly adapt to market changes. ⁴\n\nA disconnect between executives and workers is emerging, with leaders preparing for more agile ways of working and workers favoring stability (figure 1). The challenge is that the ability for organizations to evolve and adopt new models of work depends largely on the ability of the workforce to do the same. How can leaders find the right balance between the stability that workers crave and the agility that organizations need to create _stagility_ ?\n\nFor many organizations there is a gap between knowledge and action. According to our 2025 Global Human Capital Trends survey (see “ **Methodology** ”), a significant majority of organizations (72%) recognize the importance of balancing agility and stability, yet only 39% are doing something meaningful about it (figure 2). Whether organizations can strike this balance will likely depend on their ability to find ways to anchor the work, the organization, and the worker. These new anchors should provide a sense of identity, a clearly defined path of action, and a mental model for the way work _works_ . They should also provide a sense of grounding and direction, binding worker to organization in a shared vision that benefits both parties.\n\nTraditional anchors that have long provided stability for both organizations and workers are being challenged. These anchors include static job descriptions, defined teams, and linear, internal career pathways and employment models. As these anchors become upended, workers may struggle to find the grounding they require to act with confidence and creativity and to have a dynamic capability to respond and evolve.\n\nOrganizations and workers must find new sources of stability to reinforce the connection between worker and organization, putting both in a position to adapt and thrive. These new anchors will come in the form of reimagined organization structures, greater intention- ality in the design and resourcing of work, a better understanding of workers as individuals, and stronger networks within and beyond the four walls of the organization.\n\nSources of instability for work, organization, and worker\n\nFor organizations and workers to establish new anchors, they should first assess the sources of instability they face. These fall under three broad categories: work, organization, and worker.\n\n_Figure 1_\n\n### The changing nature of work no longer reflects workers’ ideal ways of working\n\n_Percentage of workers answering the question, “Which of the following describes your ideal way of working,” and percentage of leaders answering the question, “Which of the following statements best describes the changing nature of work in your organization?”_\n\nSource: Analysis of Deloitte’s 2025 Global Human Capital Trends survey data.\n\n0% 10% 20% 30% 40% 50% 60% 70%\n\n**Frequent changes in where and when work is done depending on the need**\n\n**Frequent changes in how work is done depending on the need**\n\n**Broadly defined jobs or positions, anchored to outcomes or challenges instead of specific tasks**\n\n**Shifting team(s) or workgroups**\n\n**Less stability**\n\n**Leaders Workers**","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/DI_2025-Global-Human-Capital-Trends_p14.pdf"]},"response":{"errors":[],"extract_id":"extract_79329384d915f0f3d4710442ce3b1b71","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**12** _Stagility: Creating stability for workers for organizations to move at speed_\n\ncreate value for workers and the organization. ³ And 85% say that organizations need to create more agile ways of organizing work to swiftly adapt to market changes. ⁴\n\nA disconnect between executives and workers is emerging, with leaders preparing for more agile ways of working and workers favoring stability (figure 1). The challenge is that the ability for organizations to evolve and adopt new models of work depends largely on the ability of the workforce to do the same. How can leaders find the right balance between the stability that workers crave and the agility that organizations need to create _stagility_ ?\n\nFor many organizations there is a gap between knowledge and action. According to our 2025 Global Human Capital Trends survey (see “ **Methodology** ”), a significant majority of organizations (72%) recognize the importance of balancing agility and stability, yet only 39% are doing something meaningful about it (figure 2). Whether organizations can strike this balance will likely depend on their ability to find ways to anchor the work, the organization, and the worker. These new anchors should provide a sense of identity, a clearly defined path of action, and a mental model for the way work _works_ . They should also provide a sense of grounding and direction, binding worker to organization in a shared vision that benefits both parties.\n\nTraditional anchors that have long provided stability for both organizations and workers are being challenged. These anchors include static job descriptions, defined teams, and linear, internal career pathways and employment models. As these anchors become upended, workers may struggle to find the grounding they require to act with confidence and creativity and to have a dynamic capability to respond and evolve.\n\nOrganizations and workers must find new sources of stability to reinforce the connection between worker and organization, putting both in a position to adapt and thrive. These new anchors will come in the form of reimagined organization structures, greater intention- ality in the design and resourcing of work, a better understanding of workers as individuals, and stronger networks within and beyond the four walls of the organization.\n\nSources of instability for work, organization, and worker\n\nFor organizations and workers to establish new anchors, they should first assess the sources of instability they face. These fall under three broad categories: work, organization, and worker.\n\n_Figure 1_\n\n### The changing nature of work no longer reflects workers’ ideal ways of working\n\n_Percentage of workers answering the question, “Which of the following describes your ideal way of working,” and percentage of leaders answering the question, “Which of the following statements best describes the changing nature of work in your organization?”_\n\nSource: Analysis of Deloitte’s 2025 Global Human Capital Trends survey data.\n\n0% 10% 20% 30% 40% 50% 60% 70%\n\n**Frequent changes in where and when work is done depending on the need**\n\n**Frequent changes in how work is done depending on the need**\n\n**Broadly defined jobs or positions, anchored to outcomes or challenges instead of specific tasks**\n\n**Shifting team(s) or workgroups**\n\n**Less stability**\n\n**Leaders Workers**"],"full_content":"**12** _Stagility: Creating stability for workers for organizations to move at speed_\n\ncreate value for workers and the organization. ³ And 85% say that organizations need to create more agile ways of organizing work to swiftly adapt to market changes. ⁴\n\nA disconnect between executives and workers is emerging, with leaders preparing for more agile ways of working and workers favoring stability (figure 1). The challenge is that the ability for organizations to evolve and adopt new models of work depends largely on the ability of the workforce to do the same. How can leaders find the right balance between the stability that workers crave and the agility that organizations need to create _stagility_ ?\n\nFor many organizations there is a gap between knowledge and action. According to our 2025 Global Human Capital Trends survey (see “ **Methodology** ”), a significant majority of organizations (72%) recognize the importance of balancing agility and stability, yet only 39% are doing something meaningful about it (figure 2). Whether organizations can strike this balance will likely depend on their ability to find ways to anchor the work, the organization, and the worker. These new anchors should provide a sense of identity, a clearly defined path of action, and a mental model for the way work _works_ . They should also provide a sense of grounding and direction, binding worker to organization in a shared vision that benefits both parties.\n\nTraditional anchors that have long provided stability for both organizations and workers are being challenged. These anchors include static job descriptions, defined teams, and linear, internal career pathways and employment models. As these anchors become upended, workers may struggle to find the grounding they require to act with confidence and creativity and to have a dynamic capability to respond and evolve.\n\nOrganizations and workers must find new sources of stability to reinforce the connection between worker and organization, putting both in a position to adapt and thrive. These new anchors will come in the form of reimagined organization structures, greater intention- ality in the design and resourcing of work, a better understanding of workers as individuals, and stronger networks within and beyond the four walls of the organization.\n\nSources of instability for work, organization, and worker\n\nFor organizations and workers to establish new anchors, they should first assess the sources of instability they face. These fall under three broad categories: work, organization, and worker.\n\n_Figure 1_\n\n### The changing nature of work no longer reflects workers’ ideal ways of working\n\n_Percentage of workers answering the question, “Which of the following describes your ideal way of working,” and percentage of leaders answering the question, “Which of the following statements best describes the changing nature of work in your organization?”_\n\nSource: Analysis of Deloitte’s 2025 Global Human Capital Trends survey data.\n\n0% 10% 20% 30% 40% 50% 60% 70%\n\n**Frequent changes in where and when work is done depending on the need**\n\n**Frequent changes in how work is done depending on the need**\n\n**Broadly defined jobs or positions, anchored to outcomes or challenges instead of specific tasks**\n\n**Shifting team(s) or workgroups**\n\n**Less stability**\n\n**Leaders Workers**","publish_date":null,"title":"The changing nature of work no longer reflects workers’ ideal ways of working","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/DI_2025-Global-Human-Capital-Trends_p14.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/DI_2025-Global-Human-Capital-Trends_p14.pdf"},"request":{"example_id":"chart/DI_2025-Global-Human-Capital-Trends_p14","product_type":"parse","source_file_path":"DI_2025-Global-Human-Capital-Trends_p14.pdf"},"started_at":"2026-11-04T19:45:55.884226"} +{"completed_at":"2026-11-04T19:45:59.030183","latency_in_ms":1198,"output":{"example_id":"chart/Digital_News-Report_2022_p11","grounded_pages":[],"layout_pages":[],"markdown":"• But in the face of rapidly rising household bills, we find some\n\nrespondents rethink the number of media subscriptions they can afford this year – which include news, television, music, and books. While most say they expect to retain the same number of media subscriptions, others say they expect to take out _fewer_ , as they look to save money on non-essential items.\n\n• With first-party data collection becoming more important for\n\npublishers with the imminent demise of third-party cookies, we find that most consumers are still reluctant to register their email address with news sites. Across our entire sample, only around a third (32%) say they trust news websites to use their personal data responsibly – comparable to online retailers such as Amazon – and the figure is even lower in the United States (18%) and France (19%).\n\n• Access to news continues to become more distributed. Across\n\nall markets, less than a quarter (23%) prefer to start their news journeys with a website or app, down nine points since 2018. Those aged 18–24 have an even weaker connection with websites and apps, preferring to access news via side-door routes such as social media, search, and mobile aggregators.\n\n• Facebook remains the most-used social network for news but\n\nusers are more likely to say they see _too much news_ in their feed compared with other networks. While older groups remain loyal to the platform, we show how the youngest generation has switched much of its attention to more visual networks over the last three years.\n\n• TikTok has become the fastest growing network in this\n\nyear’s survey, reaching 40% of 18–24s, with 15% using the platform for news. Usage is much higher in parts of Latin America, Asia, and Africa than it is in the United States or Northern Europe. Telegram has also grown significantly in some markets, providing a flexible alternative to Meta-owned WhatsApp.\n\n• While social media have increased the profile of many digital\n\njournalists, we find that the most well-known journalists are still TV anchors and presenters in most countries. When asked to name journalists they pay attention to, few people can name foreign correspondents, while newspaper columnists have higher name recognition in the UK and Finland than in Brazil, the United States, or France.\n\n• The smartphone has become the dominant way in which\n\nmost people _first access_ news in the morning, though we find different patterns across countries. In Norway, Spain, Finland, and the UK, the smartphone is now accessed first ahead of television, while radio retains an important role in Ireland. Morning newspaper reading is still surprisingly popular in the Netherlands; television still dominates in Japan.\n\n• After last year’s slowdown in part caused by restrictions\n\non movement during the COVID-19 pandemic, growth in podcasts seems to have resumed, with 34% consuming one or more podcasts in the last month. Our data show Spotify continuing to gain ground over Apple and Google podcasts in a number of countries and YouTube also benefiting from the popularity of video-led and hybrid podcasts.\n\n**CONSUMPTION PATTERNS REVEAL DISCONNECTION AND DISENGAGEMENT WITH NEWS – AMONGST SOME NEWS CONSUMERS**\n\nWhile a succession of crises including the pandemic and the invasion of Ukraine demonstrates the importance of independent professional journalism, and significant growth for some individual media brands, we find that many people are becoming increasingly disconnected from news – with falling interest in many countries, a rise in selective news avoidance, and low trust further underlining the critical challenge news media face today: connecting with people who have access to an unprecedented amount of content online and convincing them that paying attention to _news_ is worth their while.\n\nWe now have solid data spanning ten years that enable us to see consistent and relentless falls across countries in the reach of traditional channels such as print, radio, and television news. At the same time, we find that online reach is flat or, at best, increasing slightly – but certainly not making up the gap. Digital and social media offer a much wider range of stories, but this environment can often be overwhelming and confusing. While many people remain extremely active and engaged with online news, the abundance of choice in an online context may be leading others to engage far less regularly than they did in the past.\n\nIn the chart below, which shows weekly access in the United States between 2013 and 2022, we have added an extra line for those who say they accessed _none of the listed options_ in the week they completed our survey. This has grown from 3% in 2013 to 15% in 2022.\n\nWhile the United States seems to have the largest group of disconnected news users, we see similarly high figures in Japan (15%) the United Kingdom (9%), France (8%), and Australia (8%). Even Germany, a country with often very traditional media habits, is not immune. Since 2013, weekly print consumption has fallen from 63% to 26% and TV news usage from 82% to 65%. Although online and social media have grown a little, overtaking television for the first time, we also find growing numbers who seem to be disconnecting from news altogether – this proportion reaching 5% in 2022.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 2\n\nPROPORTION THAT USED EACH SOURCE OF NEWS IN THE LAST WEEK (2013–22) – USA\n\n**Q3.** Which, if any, of the following have you used in the last week as a source of news? _Base: Total 2013–22 samples (n ≈ 2000). Note: No data for 2014 due to a scripting error in the survey. The figure for no news would have been 12% without a methodology change where we previously excluded those who said they consumed no news in last month._\n\nTV Print Online (incl. social media)\n\nSocial media None of these Radio news\n\n69%\n\n27% 28%\n\n3%\n\n47%\n\n72%\n\n48%\n\n16%\n\n42%\n\n67%\n\n15% 15%\n\nTrump Bump Biden Slump?\n\n0%\n\n25%\n\n50%\n\n75%\n\n100% R\n\nS\n\nO\n\nP\n\nT\n\n2014 2015 2016 2017 2018 2019 2020 2021 2022 2013\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 11","pages":[{"line_numbers":[],"markdown":"• But in the face of rapidly rising household bills, we find some\n\nrespondents rethink the number of media subscriptions they can afford this year – which include news, television, music, and books. While most say they expect to retain the same number of media subscriptions, others say they expect to take out _fewer_ , as they look to save money on non-essential items.\n\n• With first-party data collection becoming more important for\n\npublishers with the imminent demise of third-party cookies, we find that most consumers are still reluctant to register their email address with news sites. Across our entire sample, only around a third (32%) say they trust news websites to use their personal data responsibly – comparable to online retailers such as Amazon – and the figure is even lower in the United States (18%) and France (19%).\n\n• Access to news continues to become more distributed. Across\n\nall markets, less than a quarter (23%) prefer to start their news journeys with a website or app, down nine points since 2018. Those aged 18–24 have an even weaker connection with websites and apps, preferring to access news via side-door routes such as social media, search, and mobile aggregators.\n\n• Facebook remains the most-used social network for news but\n\nusers are more likely to say they see _too much news_ in their feed compared with other networks. While older groups remain loyal to the platform, we show how the youngest generation has switched much of its attention to more visual networks over the last three years.\n\n• TikTok has become the fastest growing network in this\n\nyear’s survey, reaching 40% of 18–24s, with 15% using the platform for news. Usage is much higher in parts of Latin America, Asia, and Africa than it is in the United States or Northern Europe. Telegram has also grown significantly in some markets, providing a flexible alternative to Meta-owned WhatsApp.\n\n• While social media have increased the profile of many digital\n\njournalists, we find that the most well-known journalists are still TV anchors and presenters in most countries. When asked to name journalists they pay attention to, few people can name foreign correspondents, while newspaper columnists have higher name recognition in the UK and Finland than in Brazil, the United States, or France.\n\n• The smartphone has become the dominant way in which\n\nmost people _first access_ news in the morning, though we find different patterns across countries. In Norway, Spain, Finland, and the UK, the smartphone is now accessed first ahead of television, while radio retains an important role in Ireland. Morning newspaper reading is still surprisingly popular in the Netherlands; television still dominates in Japan.\n\n• After last year’s slowdown in part caused by restrictions\n\non movement during the COVID-19 pandemic, growth in podcasts seems to have resumed, with 34% consuming one or more podcasts in the last month. Our data show Spotify continuing to gain ground over Apple and Google podcasts in a number of countries and YouTube also benefiting from the popularity of video-led and hybrid podcasts.\n\n**CONSUMPTION PATTERNS REVEAL DISCONNECTION AND DISENGAGEMENT WITH NEWS – AMONGST SOME NEWS CONSUMERS**\n\nWhile a succession of crises including the pandemic and the invasion of Ukraine demonstrates the importance of independent professional journalism, and significant growth for some individual media brands, we find that many people are becoming increasingly disconnected from news – with falling interest in many countries, a rise in selective news avoidance, and low trust further underlining the critical challenge news media face today: connecting with people who have access to an unprecedented amount of content online and convincing them that paying attention to _news_ is worth their while.\n\nWe now have solid data spanning ten years that enable us to see consistent and relentless falls across countries in the reach of traditional channels such as print, radio, and television news. At the same time, we find that online reach is flat or, at best, increasing slightly – but certainly not making up the gap. Digital and social media offer a much wider range of stories, but this environment can often be overwhelming and confusing. While many people remain extremely active and engaged with online news, the abundance of choice in an online context may be leading others to engage far less regularly than they did in the past.\n\nIn the chart below, which shows weekly access in the United States between 2013 and 2022, we have added an extra line for those who say they accessed _none of the listed options_ in the week they completed our survey. This has grown from 3% in 2013 to 15% in 2022.\n\nWhile the United States seems to have the largest group of disconnected news users, we see similarly high figures in Japan (15%) the United Kingdom (9%), France (8%), and Australia (8%). Even Germany, a country with often very traditional media habits, is not immune. Since 2013, weekly print consumption has fallen from 63% to 26% and TV news usage from 82% to 65%. Although online and social media have grown a little, overtaking television for the first time, we also find growing numbers who seem to be disconnecting from news altogether – this proportion reaching 5% in 2022.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 2\n\nPROPORTION THAT USED EACH SOURCE OF NEWS IN THE LAST WEEK (2013–22) – USA\n\n**Q3.** Which, if any, of the following have you used in the last week as a source of news? _Base: Total 2013–22 samples (n ≈ 2000). Note: No data for 2014 due to a scripting error in the survey. The figure for no news would have been 12% without a methodology change where we previously excluded those who said they consumed no news in last month._\n\nTV Print Online (incl. social media)\n\nSocial media None of these Radio news\n\n69%\n\n27% 28%\n\n3%\n\n47%\n\n72%\n\n48%\n\n16%\n\n42%\n\n67%\n\n15% 15%\n\nTrump Bump Biden Slump?\n\n0%\n\n25%\n\n50%\n\n75%\n\n100% R\n\nS\n\nO\n\nP\n\nT\n\n2014 2015 2016 2017 2018 2019 2020 2021 2022 2013\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 11","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"• But in the face of rapidly rising household bills, we find some\n\nrespondents rethink the number of media subscriptions they can afford this year – which include news, television, music, and books. While most say they expect to retain the same number of media subscriptions, others say they expect to take out _fewer_ , as they look to save money on non-essential items.\n\n• With first-party data collection becoming more important for\n\npublishers with the imminent demise of third-party cookies, we find that most consumers are still reluctant to register their email address with news sites. Across our entire sample, only around a third (32%) say they trust news websites to use their personal data responsibly – comparable to online retailers such as Amazon – and the figure is even lower in the United States (18%) and France (19%).\n\n• Access to news continues to become more distributed. Across\n\nall markets, less than a quarter (23%) prefer to start their news journeys with a website or app, down nine points since 2018. Those aged 18–24 have an even weaker connection with websites and apps, preferring to access news via side-door routes such as social media, search, and mobile aggregators.\n\n• Facebook remains the most-used social network for news but\n\nusers are more likely to say they see _too much news_ in their feed compared with other networks. While older groups remain loyal to the platform, we show how the youngest generation has switched much of its attention to more visual networks over the last three years.\n\n• TikTok has become the fastest growing network in this\n\nyear’s survey, reaching 40% of 18–24s, with 15% using the platform for news. Usage is much higher in parts of Latin America, Asia, and Africa than it is in the United States or Northern Europe. Telegram has also grown significantly in some markets, providing a flexible alternative to Meta-owned WhatsApp.\n\n• While social media have increased the profile of many digital\n\njournalists, we find that the most well-known journalists are still TV anchors and presenters in most countries. When asked to name journalists they pay attention to, few people can name foreign correspondents, while newspaper columnists have higher name recognition in the UK and Finland than in Brazil, the United States, or France.\n\n• The smartphone has become the dominant way in which\n\nmost people _first access_ news in the morning, though we find different patterns across countries. In Norway, Spain, Finland, and the UK, the smartphone is now accessed first ahead of television, while radio retains an important role in Ireland. Morning newspaper reading is still surprisingly popular in the Netherlands; television still dominates in Japan.\n\n• After last year’s slowdown in part caused by restrictions\n\non movement during the COVID-19 pandemic, growth in podcasts seems to have resumed, with 34% consuming one or more podcasts in the last month. Our data show Spotify continuing to gain ground over Apple and Google podcasts in a number of countries and YouTube also benefiting from the popularity of video-led and hybrid podcasts.\n\n**CONSUMPTION PATTERNS REVEAL DISCONNECTION AND DISENGAGEMENT WITH NEWS – AMONGST SOME NEWS CONSUMERS**\n\nWhile a succession of crises including the pandemic and the invasion of Ukraine demonstrates the importance of independent professional journalism, and significant growth for some individual media brands, we find that many people are becoming increasingly disconnected from news – with falling interest in many countries, a rise in selective news avoidance, and low trust further underlining the critical challenge news media face today: connecting with people who have access to an unprecedented amount of content online and convincing them that paying attention to _news_ is worth their while.\n\nWe now have solid data spanning ten years that enable us to see consistent and relentless falls across countries in the reach of traditional channels such as print, radio, and television news. At the same time, we find that online reach is flat or, at best, increasing slightly – but certainly not making up the gap. Digital and social media offer a much wider range of stories, but this environment can often be overwhelming and confusing. While many people remain extremely active and engaged with online news, the abundance of choice in an online context may be leading others to engage far less regularly than they did in the past.\n\nIn the chart below, which shows weekly access in the United States between 2013 and 2022, we have added an extra line for those who say they accessed _none of the listed options_ in the week they completed our survey. This has grown from 3% in 2013 to 15% in 2022.\n\nWhile the United States seems to have the largest group of disconnected news users, we see similarly high figures in Japan (15%) the United Kingdom (9%), France (8%), and Australia (8%). Even Germany, a country with often very traditional media habits, is not immune. Since 2013, weekly print consumption has fallen from 63% to 26% and TV news usage from 82% to 65%. Although online and social media have grown a little, overtaking television for the first time, we also find growing numbers who seem to be disconnecting from news altogether – this proportion reaching 5% in 2022.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 2\n\nPROPORTION THAT USED EACH SOURCE OF NEWS IN THE LAST WEEK (2013–22) – USA\n\n**Q3.** Which, if any, of the following have you used in the last week as a source of news? _Base: Total 2013–22 samples (n ≈ 2000). Note: No data for 2014 due to a scripting error in the survey. The figure for no news would have been 12% without a methodology change where we previously excluded those who said they consumed no news in last month._\n\nTV Print Online (incl. social media)\n\nSocial media None of these Radio news\n\n69%\n\n27% 28%\n\n3%\n\n47%\n\n72%\n\n48%\n\n16%\n\n42%\n\n67%\n\n15% 15%\n\nTrump Bump Biden Slump?\n\n0%\n\n25%\n\n50%\n\n75%\n\n100% R\n\nS\n\nO\n\nP\n\nT\n\n2014 2015 2016 2017 2018 2019 2020 2021 2022 2013\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 11","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Digital_News-Report_2022_p11.pdf"]},"response":{"errors":[],"extract_id":"extract_62dbce8a3d01f071126d645d851d11c4","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["• But in the face of rapidly rising household bills, we find some\n\nrespondents rethink the number of media subscriptions they can afford this year – which include news, television, music, and books. While most say they expect to retain the same number of media subscriptions, others say they expect to take out _fewer_ , as they look to save money on non-essential items.\n\n• With first-party data collection becoming more important for\n\npublishers with the imminent demise of third-party cookies, we find that most consumers are still reluctant to register their email address with news sites. Across our entire sample, only around a third (32%) say they trust news websites to use their personal data responsibly – comparable to online retailers such as Amazon – and the figure is even lower in the United States (18%) and France (19%).\n\n• Access to news continues to become more distributed. Across\n\nall markets, less than a quarter (23%) prefer to start their news journeys with a website or app, down nine points since 2018. Those aged 18–24 have an even weaker connection with websites and apps, preferring to access news via side-door routes such as social media, search, and mobile aggregators.\n\n• Facebook remains the most-used social network for news but\n\nusers are more likely to say they see _too much news_ in their feed compared with other networks. While older groups remain loyal to the platform, we show how the youngest generation has switched much of its attention to more visual networks over the last three years.\n\n• TikTok has become the fastest growing network in this\n\nyear’s survey, reaching 40% of 18–24s, with 15% using the platform for news. Usage is much higher in parts of Latin America, Asia, and Africa than it is in the United States or Northern Europe. Telegram has also grown significantly in some markets, providing a flexible alternative to Meta-owned WhatsApp.\n\n• While social media have increased the profile of many digital\n\njournalists, we find that the most well-known journalists are still TV anchors and presenters in most countries. When asked to name journalists they pay attention to, few people can name foreign correspondents, while newspaper columnists have higher name recognition in the UK and Finland than in Brazil, the United States, or France.\n\n• The smartphone has become the dominant way in which\n\nmost people _first access_ news in the morning, though we find different patterns across countries. In Norway, Spain, Finland, and the UK, the smartphone is now accessed first ahead of television, while radio retains an important role in Ireland. Morning newspaper reading is still surprisingly popular in the Netherlands; television still dominates in Japan.\n\n• After last year’s slowdown in part caused by restrictions\n\non movement during the COVID-19 pandemic, growth in podcasts seems to have resumed, with 34% consuming one or more podcasts in the last month. Our data show Spotify continuing to gain ground over Apple and Google podcasts in a number of countries and YouTube also benefiting from the popularity of video-led and hybrid podcasts.\n\n**CONSUMPTION PATTERNS REVEAL DISCONNECTION AND DISENGAGEMENT WITH NEWS – AMONGST SOME NEWS CONSUMERS**\n\nWhile a succession of crises including the pandemic and the invasion of Ukraine demonstrates the importance of independent professional journalism, and significant growth for some individual media brands, we find that many people are becoming increasingly disconnected from news – with falling interest in many countries, a rise in selective news avoidance, and low trust further underlining the critical challenge news media face today: connecting with people who have access to an unprecedented amount of content online and convincing them that paying attention to _news_ is worth their while.\n\nWe now have solid data spanning ten years that enable us to see consistent and relentless falls across countries in the reach of traditional channels such as print, radio, and television news. At the same time, we find that online reach is flat or, at best, increasing slightly – but certainly not making up the gap. Digital and social media offer a much wider range of stories, but this environment can often be overwhelming and confusing. While many people remain extremely active and engaged with online news, the abundance of choice in an online context may be leading others to engage far less regularly than they did in the past.\n\nIn the chart below, which shows weekly access in the United States between 2013 and 2022, we have added an extra line for those who say they accessed _none of the listed options_ in the week they completed our survey. This has grown from 3% in 2013 to 15% in 2022.\n\nWhile the United States seems to have the largest group of disconnected news users, we see similarly high figures in Japan (15%) the United Kingdom (9%), France (8%), and Australia (8%). Even Germany, a country with often very traditional media habits, is not immune. Since 2013, weekly print consumption has fallen from 63% to 26% and TV news usage from 82% to 65%. Although online and social media have grown a little, overtaking television for the first time, we also find growing numbers who seem to be disconnecting from news altogether – this proportion reaching 5% in 2022.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 2\n\nPROPORTION THAT USED EACH SOURCE OF NEWS IN THE LAST WEEK (2013–22) – USA\n\n**Q3.** Which, if any, of the following have you used in the last week as a source of news? _Base: Total 2013–22 samples (n ≈ 2000). Note: No data for 2014 due to a scripting error in the survey. The figure for no news would have been 12% without a methodology change where we previously excluded those who said they consumed no news in last month._\n\nTV Print Online (incl. social media)\n\nSocial media None of these Radio news\n\n69%\n\n27% 28%\n\n3%\n\n47%\n\n72%\n\n48%\n\n16%\n\n42%\n\n67%\n\n15% 15%\n\nTrump Bump Biden Slump?\n\n0%\n\n25%\n\n50%\n\n75%\n\n100% R\n\nS\n\nO\n\nP\n\nT\n\n2014 2015 2016 2017 2018 2019 2020 2021 2022 2013\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 11"],"full_content":"• But in the face of rapidly rising household bills, we find some\n\nrespondents rethink the number of media subscriptions they can afford this year – which include news, television, music, and books. While most say they expect to retain the same number of media subscriptions, others say they expect to take out _fewer_ , as they look to save money on non-essential items.\n\n• With first-party data collection becoming more important for\n\npublishers with the imminent demise of third-party cookies, we find that most consumers are still reluctant to register their email address with news sites. Across our entire sample, only around a third (32%) say they trust news websites to use their personal data responsibly – comparable to online retailers such as Amazon – and the figure is even lower in the United States (18%) and France (19%).\n\n• Access to news continues to become more distributed. Across\n\nall markets, less than a quarter (23%) prefer to start their news journeys with a website or app, down nine points since 2018. Those aged 18–24 have an even weaker connection with websites and apps, preferring to access news via side-door routes such as social media, search, and mobile aggregators.\n\n• Facebook remains the most-used social network for news but\n\nusers are more likely to say they see _too much news_ in their feed compared with other networks. While older groups remain loyal to the platform, we show how the youngest generation has switched much of its attention to more visual networks over the last three years.\n\n• TikTok has become the fastest growing network in this\n\nyear’s survey, reaching 40% of 18–24s, with 15% using the platform for news. Usage is much higher in parts of Latin America, Asia, and Africa than it is in the United States or Northern Europe. Telegram has also grown significantly in some markets, providing a flexible alternative to Meta-owned WhatsApp.\n\n• While social media have increased the profile of many digital\n\njournalists, we find that the most well-known journalists are still TV anchors and presenters in most countries. When asked to name journalists they pay attention to, few people can name foreign correspondents, while newspaper columnists have higher name recognition in the UK and Finland than in Brazil, the United States, or France.\n\n• The smartphone has become the dominant way in which\n\nmost people _first access_ news in the morning, though we find different patterns across countries. In Norway, Spain, Finland, and the UK, the smartphone is now accessed first ahead of television, while radio retains an important role in Ireland. Morning newspaper reading is still surprisingly popular in the Netherlands; television still dominates in Japan.\n\n• After last year’s slowdown in part caused by restrictions\n\non movement during the COVID-19 pandemic, growth in podcasts seems to have resumed, with 34% consuming one or more podcasts in the last month. Our data show Spotify continuing to gain ground over Apple and Google podcasts in a number of countries and YouTube also benefiting from the popularity of video-led and hybrid podcasts.\n\n**CONSUMPTION PATTERNS REVEAL DISCONNECTION AND DISENGAGEMENT WITH NEWS – AMONGST SOME NEWS CONSUMERS**\n\nWhile a succession of crises including the pandemic and the invasion of Ukraine demonstrates the importance of independent professional journalism, and significant growth for some individual media brands, we find that many people are becoming increasingly disconnected from news – with falling interest in many countries, a rise in selective news avoidance, and low trust further underlining the critical challenge news media face today: connecting with people who have access to an unprecedented amount of content online and convincing them that paying attention to _news_ is worth their while.\n\nWe now have solid data spanning ten years that enable us to see consistent and relentless falls across countries in the reach of traditional channels such as print, radio, and television news. At the same time, we find that online reach is flat or, at best, increasing slightly – but certainly not making up the gap. Digital and social media offer a much wider range of stories, but this environment can often be overwhelming and confusing. While many people remain extremely active and engaged with online news, the abundance of choice in an online context may be leading others to engage far less regularly than they did in the past.\n\nIn the chart below, which shows weekly access in the United States between 2013 and 2022, we have added an extra line for those who say they accessed _none of the listed options_ in the week they completed our survey. This has grown from 3% in 2013 to 15% in 2022.\n\nWhile the United States seems to have the largest group of disconnected news users, we see similarly high figures in Japan (15%) the United Kingdom (9%), France (8%), and Australia (8%). Even Germany, a country with often very traditional media habits, is not immune. Since 2013, weekly print consumption has fallen from 63% to 26% and TV news usage from 82% to 65%. Although online and social media have grown a little, overtaking television for the first time, we also find growing numbers who seem to be disconnecting from news altogether – this proportion reaching 5% in 2022.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 2\n\nPROPORTION THAT USED EACH SOURCE OF NEWS IN THE LAST WEEK (2013–22) – USA\n\n**Q3.** Which, if any, of the following have you used in the last week as a source of news? _Base: Total 2013–22 samples (n ≈ 2000). Note: No data for 2014 due to a scripting error in the survey. The figure for no news would have been 12% without a methodology change where we previously excluded those who said they consumed no news in last month._\n\nTV Print Online (incl. social media)\n\nSocial media None of these Radio news\n\n69%\n\n27% 28%\n\n3%\n\n47%\n\n72%\n\n48%\n\n16%\n\n42%\n\n67%\n\n15% 15%\n\nTrump Bump Biden Slump?\n\n0%\n\n25%\n\n50%\n\n75%\n\n100% R\n\nS\n\nO\n\nP\n\nT\n\n2014 2015 2016 2017 2018 2019 2020 2021 2022 2013\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 11","publish_date":null,"title":"Graph 2","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Digital_News-Report_2022_p11.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Digital_News-Report_2022_p11.pdf"},"request":{"example_id":"chart/Digital_News-Report_2022_p11","product_type":"parse","source_file_path":"Digital_News-Report_2022_p11.pdf"},"started_at":"2026-11-04T19:45:57.832150"} +{"completed_at":"2026-11-04T19:46:03.873105","latency_in_ms":5388,"output":{"example_id":"chart/Digital_News-Report_2022_p12","grounded_pages":[],"layout_pages":[],"markdown":"It is important to note that high levels of disconnection are not evident everywhere. The proportion consuming none of our listed news sources weekly is limited in Portugal (2%) and Finland (2%) and extremely rare in South Africa (1%), Nigeria (<1%), and Kenya (<1%), even though we also see growing pressure on traditional sources of news such as print and television news in these countries.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 3\n\nPROPORTION THAT USED EACH SOURCE OF NEWS IN THE LAST WEEK (2013–22) – GERMANY\n\n**Q3.** Which, if any, of the following have you used in the last week as a source of news? _Base: Total 2013–22 samples (most n ≈ 2000). Note: No data for 2014 due to a scripting error in the survey. The figure for no news would have been 4% without a methodology change where we previously excluded those who said they consumed no news in last month._\n\nTV Print Online (incl. social media)\n\nSocial media Radio news\n\n60%\n\n18%\n\n51%\n\n1%\n\n63%\n\n82%\n\n65%\n\n39%\n\n5%\n\n32%\n\n68%\n\n26%\n\n0%\n\n25%\n\n50%\n\n75%\n\n100% Radio news\n\nNone of these\n\nSocial media\n\nOnline (incl. social media)\n\nPrint\n\nTV\n\n2014 2015 2016 2017 2018 2019 2020 2021 2022 2013\n\nNone of these\n\n**Falling interest in the news**\n\nDisconnection is just one sign of the difficulties of engaging some audiences in a more digital environment. At the same time we find that the proportion that says they are very or extremely interested in news has fallen sharply over time across markets – a trend that has accelerated despite the continuing COVID-19 pandemic. This year we find news interest lower in the vast majority of countries in our survey. In some countries, such as Argentina, Brazil, Spain, and the United Kingdom, these falls have been going on for some time, while in the United States we see a slightly different pattern. Interest remained high during the Trump years but seems to have declined significantly since Joe Biden became president. Today, less than half of our sample (47%) say they are very or extremely interested in news compared with 67% in 2015.\n\nAgain, it is important to note that there are some countries that buck these trends or where declines are happening more slowly. These include richer Central/Western European nations where there has been less political or economic turmoil over the past few years.\n\nThese data suggest two different but related problems. First, the emergence of a minority of people who are active online, many of them younger or less well educated, but who have become largely disconnected from the news, perhaps because they don’t feel that it is relevant to their lives. And then, separately, we find a more generalised decline in news interest and consumption affecting a much bigger group, which may relate to structural changes in the way the news is distributed, such as the shift to online, the nature of the news cycle itself, or both.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 4\n\nPROPORTION WHO ARE VERY OR EXTREMELY INTERESTED IN NEWS (2015–22)\n\n**Q1c.** How interested, if at all, would you say you are in news? _Base: Total 2015–22 samples (n≈2000)._\n\n**Selected countries with largest falls**\n\n**Selected countries with more stable levels**\n\n0%\n\n25%\n\n50%\n\n75%\n\n100%\n\n2022 2017 2022 2015 2022 2015 2022 2015 2022 2015\n\n0%\n\n25%\n\n50%\n\n75%\n\n100%\n\n2022 2015 2022 2016 2022 2016 2022 2015 2022 2015\n\n77%\n\n48%\n\n82%\n\n57%\n\n56%\n\n51%\n\n59%\n\n50%\n\n74%\n\n57%\n\n85%\n\n55%\n\n70%\n\n43%\n\n57%\n\n55%\n\n67%\n\n47%\n\n64%\n\n67%\n\n0%\n\n25%\n\n50%\n\n75%\n\n100%\n\n2022 2017 2022 2015 2022 2015 2022 2015 2022 2015\n\n0%\n\n25%\n\n50%\n\n75%\n\n100%\n\n2022 2015 2022 2016 2022 2016 2022 2015 2022 2015\n\n77%\n\n48%\n\n82%\n\n57%\n\n56%\n\n51%\n\n59%\n\n50%\n\n74%\n\n57%\n\n85%\n\n55%\n\n70%\n\n43%\n\n57%\n\n55%\n\n67%\n\n47%\n\n64%\n\n67%\n\n**Argentina**\n\n**Germany**\n\n**Brazil**\n\n**Sweden**\n\n**Spain**\n\n**Switzerland**\n\n**UK**\n\n**Netherlands**\n\n**USA**\n\n**Finland**\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 12","pages":[{"line_numbers":[],"markdown":"It is important to note that high levels of disconnection are not evident everywhere. The proportion consuming none of our listed news sources weekly is limited in Portugal (2%) and Finland (2%) and extremely rare in South Africa (1%), Nigeria (<1%), and Kenya (<1%), even though we also see growing pressure on traditional sources of news such as print and television news in these countries.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 3\n\nPROPORTION THAT USED EACH SOURCE OF NEWS IN THE LAST WEEK (2013–22) – GERMANY\n\n**Q3.** Which, if any, of the following have you used in the last week as a source of news? _Base: Total 2013–22 samples (most n ≈ 2000). Note: No data for 2014 due to a scripting error in the survey. The figure for no news would have been 4% without a methodology change where we previously excluded those who said they consumed no news in last month._\n\nTV Print Online (incl. social media)\n\nSocial media Radio news\n\n60%\n\n18%\n\n51%\n\n1%\n\n63%\n\n82%\n\n65%\n\n39%\n\n5%\n\n32%\n\n68%\n\n26%\n\n0%\n\n25%\n\n50%\n\n75%\n\n100% Radio news\n\nNone of these\n\nSocial media\n\nOnline (incl. social media)\n\nPrint\n\nTV\n\n2014 2015 2016 2017 2018 2019 2020 2021 2022 2013\n\nNone of these\n\n**Falling interest in the news**\n\nDisconnection is just one sign of the difficulties of engaging some audiences in a more digital environment. At the same time we find that the proportion that says they are very or extremely interested in news has fallen sharply over time across markets – a trend that has accelerated despite the continuing COVID-19 pandemic. This year we find news interest lower in the vast majority of countries in our survey. In some countries, such as Argentina, Brazil, Spain, and the United Kingdom, these falls have been going on for some time, while in the United States we see a slightly different pattern. Interest remained high during the Trump years but seems to have declined significantly since Joe Biden became president. Today, less than half of our sample (47%) say they are very or extremely interested in news compared with 67% in 2015.\n\nAgain, it is important to note that there are some countries that buck these trends or where declines are happening more slowly. These include richer Central/Western European nations where there has been less political or economic turmoil over the past few years.\n\nThese data suggest two different but related problems. First, the emergence of a minority of people who are active online, many of them younger or less well educated, but who have become largely disconnected from the news, perhaps because they don’t feel that it is relevant to their lives. And then, separately, we find a more generalised decline in news interest and consumption affecting a much bigger group, which may relate to structural changes in the way the news is distributed, such as the shift to online, the nature of the news cycle itself, or both.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 4\n\nPROPORTION WHO ARE VERY OR EXTREMELY INTERESTED IN NEWS (2015–22)\n\n**Q1c.** How interested, if at all, would you say you are in news? _Base: Total 2015–22 samples (n≈2000)._\n\n**Selected countries with largest falls**\n\n**Selected countries with more stable levels**\n\n0%\n\n25%\n\n50%\n\n75%\n\n100%\n\n2022 2017 2022 2015 2022 2015 2022 2015 2022 2015\n\n0%\n\n25%\n\n50%\n\n75%\n\n100%\n\n2022 2015 2022 2016 2022 2016 2022 2015 2022 2015\n\n77%\n\n48%\n\n82%\n\n57%\n\n56%\n\n51%\n\n59%\n\n50%\n\n74%\n\n57%\n\n85%\n\n55%\n\n70%\n\n43%\n\n57%\n\n55%\n\n67%\n\n47%\n\n64%\n\n67%\n\n0%\n\n25%\n\n50%\n\n75%\n\n100%\n\n2022 2017 2022 2015 2022 2015 2022 2015 2022 2015\n\n0%\n\n25%\n\n50%\n\n75%\n\n100%\n\n2022 2015 2022 2016 2022 2016 2022 2015 2022 2015\n\n77%\n\n48%\n\n82%\n\n57%\n\n56%\n\n51%\n\n59%\n\n50%\n\n74%\n\n57%\n\n85%\n\n55%\n\n70%\n\n43%\n\n57%\n\n55%\n\n67%\n\n47%\n\n64%\n\n67%\n\n**Argentina**\n\n**Germany**\n\n**Brazil**\n\n**Sweden**\n\n**Spain**\n\n**Switzerland**\n\n**UK**\n\n**Netherlands**\n\n**USA**\n\n**Finland**\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 12","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"It is important to note that high levels of disconnection are not evident everywhere. The proportion consuming none of our listed news sources weekly is limited in Portugal (2%) and Finland (2%) and extremely rare in South Africa (1%), Nigeria (<1%), and Kenya (<1%), even though we also see growing pressure on traditional sources of news such as print and television news in these countries.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 3\n\nPROPORTION THAT USED EACH SOURCE OF NEWS IN THE LAST WEEK (2013–22) – GERMANY\n\n**Q3.** Which, if any, of the following have you used in the last week as a source of news? _Base: Total 2013–22 samples (most n ≈ 2000). Note: No data for 2014 due to a scripting error in the survey. The figure for no news would have been 4% without a methodology change where we previously excluded those who said they consumed no news in last month._\n\nTV Print Online (incl. social media)\n\nSocial media Radio news\n\n60%\n\n18%\n\n51%\n\n1%\n\n63%\n\n82%\n\n65%\n\n39%\n\n5%\n\n32%\n\n68%\n\n26%\n\n0%\n\n25%\n\n50%\n\n75%\n\n100% Radio news\n\nNone of these\n\nSocial media\n\nOnline (incl. social media)\n\nPrint\n\nTV\n\n2014 2015 2016 2017 2018 2019 2020 2021 2022 2013\n\nNone of these\n\n**Falling interest in the news**\n\nDisconnection is just one sign of the difficulties of engaging some audiences in a more digital environment. At the same time we find that the proportion that says they are very or extremely interested in news has fallen sharply over time across markets – a trend that has accelerated despite the continuing COVID-19 pandemic. This year we find news interest lower in the vast majority of countries in our survey. In some countries, such as Argentina, Brazil, Spain, and the United Kingdom, these falls have been going on for some time, while in the United States we see a slightly different pattern. Interest remained high during the Trump years but seems to have declined significantly since Joe Biden became president. Today, less than half of our sample (47%) say they are very or extremely interested in news compared with 67% in 2015.\n\nAgain, it is important to note that there are some countries that buck these trends or where declines are happening more slowly. These include richer Central/Western European nations where there has been less political or economic turmoil over the past few years.\n\nThese data suggest two different but related problems. First, the emergence of a minority of people who are active online, many of them younger or less well educated, but who have become largely disconnected from the news, perhaps because they don’t feel that it is relevant to their lives. And then, separately, we find a more generalised decline in news interest and consumption affecting a much bigger group, which may relate to structural changes in the way the news is distributed, such as the shift to online, the nature of the news cycle itself, or both.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 4\n\nPROPORTION WHO ARE VERY OR EXTREMELY INTERESTED IN NEWS (2015–22)\n\n**Q1c.** How interested, if at all, would you say you are in news? _Base: Total 2015–22 samples (n≈2000)._\n\n**Selected countries with largest falls**\n\n**Selected countries with more stable levels**\n\n0%\n\n25%\n\n50%\n\n75%\n\n100%\n\n2022 2017 2022 2015 2022 2015 2022 2015 2022 2015\n\n0%\n\n25%\n\n50%\n\n75%\n\n100%\n\n2022 2015 2022 2016 2022 2016 2022 2015 2022 2015\n\n77%\n\n48%\n\n82%\n\n57%\n\n56%\n\n51%\n\n59%\n\n50%\n\n74%\n\n57%\n\n85%\n\n55%\n\n70%\n\n43%\n\n57%\n\n55%\n\n67%\n\n47%\n\n64%\n\n67%\n\n0%\n\n25%\n\n50%\n\n75%\n\n100%\n\n2022 2017 2022 2015 2022 2015 2022 2015 2022 2015\n\n0%\n\n25%\n\n50%\n\n75%\n\n100%\n\n2022 2015 2022 2016 2022 2016 2022 2015 2022 2015\n\n77%\n\n48%\n\n82%\n\n57%\n\n56%\n\n51%\n\n59%\n\n50%\n\n74%\n\n57%\n\n85%\n\n55%\n\n70%\n\n43%\n\n57%\n\n55%\n\n67%\n\n47%\n\n64%\n\n67%\n\n**Argentina**\n\n**Germany**\n\n**Brazil**\n\n**Sweden**\n\n**Spain**\n\n**Switzerland**\n\n**UK**\n\n**Netherlands**\n\n**USA**\n\n**Finland**\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 12","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Digital_News-Report_2022_p12.pdf"]},"response":{"errors":[],"extract_id":"extract_3c4a9bcf2c853b07427ad11936c0c936","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["It is important to note that high levels of disconnection are not evident everywhere. The proportion consuming none of our listed news sources weekly is limited in Portugal (2%) and Finland (2%) and extremely rare in South Africa (1%), Nigeria (<1%), and Kenya (<1%), even though we also see growing pressure on traditional sources of news such as print and television news in these countries.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 3\n\nPROPORTION THAT USED EACH SOURCE OF NEWS IN THE LAST WEEK (2013–22) – GERMANY\n\n**Q3.** Which, if any, of the following have you used in the last week as a source of news? _Base: Total 2013–22 samples (most n ≈ 2000). Note: No data for 2014 due to a scripting error in the survey. The figure for no news would have been 4% without a methodology change where we previously excluded those who said they consumed no news in last month._\n\nTV Print Online (incl. social media)\n\nSocial media Radio news\n\n60%\n\n18%\n\n51%\n\n1%\n\n63%\n\n82%\n\n65%\n\n39%\n\n5%\n\n32%\n\n68%\n\n26%\n\n0%\n\n25%\n\n50%\n\n75%\n\n100% Radio news\n\nNone of these\n\nSocial media\n\nOnline (incl. social media)\n\nPrint\n\nTV\n\n2014 2015 2016 2017 2018 2019 2020 2021 2022 2013\n\nNone of these\n\n**Falling interest in the news**\n\nDisconnection is just one sign of the difficulties of engaging some audiences in a more digital environment. At the same time we find that the proportion that says they are very or extremely interested in news has fallen sharply over time across markets – a trend that has accelerated despite the continuing COVID-19 pandemic. This year we find news interest lower in the vast majority of countries in our survey. In some countries, such as Argentina, Brazil, Spain, and the United Kingdom, these falls have been going on for some time, while in the United States we see a slightly different pattern. Interest remained high during the Trump years but seems to have declined significantly since Joe Biden became president. Today, less than half of our sample (47%) say they are very or extremely interested in news compared with 67% in 2015.\n\nAgain, it is important to note that there are some countries that buck these trends or where declines are happening more slowly. These include richer Central/Western European nations where there has been less political or economic turmoil over the past few years.\n\nThese data suggest two different but related problems. First, the emergence of a minority of people who are active online, many of them younger or less well educated, but who have become largely disconnected from the news, perhaps because they don’t feel that it is relevant to their lives. And then, separately, we find a more generalised decline in news interest and consumption affecting a much bigger group, which may relate to structural changes in the way the news is distributed, such as the shift to online, the nature of the news cycle itself, or both.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 4\n\nPROPORTION WHO ARE VERY OR EXTREMELY INTERESTED IN NEWS (2015–22)\n\n**Q1c.** How interested, if at all, would you say you are in news? _Base: Total 2015–22 samples (n≈2000)._\n\n**Selected countries with largest falls**\n\n**Selected countries with more stable levels**\n\n0%\n\n25%\n\n50%\n\n75%\n\n100%\n\n2022 2017 2022 2015 2022 2015 2022 2015 2022 2015\n\n0%\n\n25%\n\n50%\n\n75%\n\n100%\n\n2022 2015 2022 2016 2022 2016 2022 2015 2022 2015\n\n77%\n\n48%\n\n82%\n\n57%\n\n56%\n\n51%\n\n59%\n\n50%\n\n74%\n\n57%\n\n85%\n\n55%\n\n70%\n\n43%\n\n57%\n\n55%\n\n67%\n\n47%\n\n64%\n\n67%\n\n0%\n\n25%\n\n50%\n\n75%\n\n100%\n\n2022 2017 2022 2015 2022 2015 2022 2015 2022 2015\n\n0%\n\n25%\n\n50%\n\n75%\n\n100%\n\n2022 2015 2022 2016 2022 2016 2022 2015 2022 2015\n\n77%\n\n48%\n\n82%\n\n57%\n\n56%\n\n51%\n\n59%\n\n50%\n\n74%\n\n57%\n\n85%\n\n55%\n\n70%\n\n43%\n\n57%\n\n55%\n\n67%\n\n47%\n\n64%\n\n67%\n\n**Argentina**\n\n**Germany**\n\n**Brazil**\n\n**Sweden**\n\n**Spain**\n\n**Switzerland**\n\n**UK**\n\n**Netherlands**\n\n**USA**\n\n**Finland**\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 12"],"full_content":"It is important to note that high levels of disconnection are not evident everywhere. The proportion consuming none of our listed news sources weekly is limited in Portugal (2%) and Finland (2%) and extremely rare in South Africa (1%), Nigeria (<1%), and Kenya (<1%), even though we also see growing pressure on traditional sources of news such as print and television news in these countries.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 3\n\nPROPORTION THAT USED EACH SOURCE OF NEWS IN THE LAST WEEK (2013–22) – GERMANY\n\n**Q3.** Which, if any, of the following have you used in the last week as a source of news? _Base: Total 2013–22 samples (most n ≈ 2000). Note: No data for 2014 due to a scripting error in the survey. The figure for no news would have been 4% without a methodology change where we previously excluded those who said they consumed no news in last month._\n\nTV Print Online (incl. social media)\n\nSocial media Radio news\n\n60%\n\n18%\n\n51%\n\n1%\n\n63%\n\n82%\n\n65%\n\n39%\n\n5%\n\n32%\n\n68%\n\n26%\n\n0%\n\n25%\n\n50%\n\n75%\n\n100% Radio news\n\nNone of these\n\nSocial media\n\nOnline (incl. social media)\n\nPrint\n\nTV\n\n2014 2015 2016 2017 2018 2019 2020 2021 2022 2013\n\nNone of these\n\n**Falling interest in the news**\n\nDisconnection is just one sign of the difficulties of engaging some audiences in a more digital environment. At the same time we find that the proportion that says they are very or extremely interested in news has fallen sharply over time across markets – a trend that has accelerated despite the continuing COVID-19 pandemic. This year we find news interest lower in the vast majority of countries in our survey. In some countries, such as Argentina, Brazil, Spain, and the United Kingdom, these falls have been going on for some time, while in the United States we see a slightly different pattern. Interest remained high during the Trump years but seems to have declined significantly since Joe Biden became president. Today, less than half of our sample (47%) say they are very or extremely interested in news compared with 67% in 2015.\n\nAgain, it is important to note that there are some countries that buck these trends or where declines are happening more slowly. These include richer Central/Western European nations where there has been less political or economic turmoil over the past few years.\n\nThese data suggest two different but related problems. First, the emergence of a minority of people who are active online, many of them younger or less well educated, but who have become largely disconnected from the news, perhaps because they don’t feel that it is relevant to their lives. And then, separately, we find a more generalised decline in news interest and consumption affecting a much bigger group, which may relate to structural changes in the way the news is distributed, such as the shift to online, the nature of the news cycle itself, or both.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 4\n\nPROPORTION WHO ARE VERY OR EXTREMELY INTERESTED IN NEWS (2015–22)\n\n**Q1c.** How interested, if at all, would you say you are in news? _Base: Total 2015–22 samples (n≈2000)._\n\n**Selected countries with largest falls**\n\n**Selected countries with more stable levels**\n\n0%\n\n25%\n\n50%\n\n75%\n\n100%\n\n2022 2017 2022 2015 2022 2015 2022 2015 2022 2015\n\n0%\n\n25%\n\n50%\n\n75%\n\n100%\n\n2022 2015 2022 2016 2022 2016 2022 2015 2022 2015\n\n77%\n\n48%\n\n82%\n\n57%\n\n56%\n\n51%\n\n59%\n\n50%\n\n74%\n\n57%\n\n85%\n\n55%\n\n70%\n\n43%\n\n57%\n\n55%\n\n67%\n\n47%\n\n64%\n\n67%\n\n0%\n\n25%\n\n50%\n\n75%\n\n100%\n\n2022 2017 2022 2015 2022 2015 2022 2015 2022 2015\n\n0%\n\n25%\n\n50%\n\n75%\n\n100%\n\n2022 2015 2022 2016 2022 2016 2022 2015 2022 2015\n\n77%\n\n48%\n\n82%\n\n57%\n\n56%\n\n51%\n\n59%\n\n50%\n\n74%\n\n57%\n\n85%\n\n55%\n\n70%\n\n43%\n\n57%\n\n55%\n\n67%\n\n47%\n\n64%\n\n67%\n\n**Argentina**\n\n**Germany**\n\n**Brazil**\n\n**Sweden**\n\n**Spain**\n\n**Switzerland**\n\n**UK**\n\n**Netherlands**\n\n**USA**\n\n**Finland**\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 12","publish_date":null,"title":"Graph 3 Graph 4","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Digital_News-Report_2022_p12.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Digital_News-Report_2022_p12.pdf"},"request":{"example_id":"chart/Digital_News-Report_2022_p12","product_type":"parse","source_file_path":"Digital_News-Report_2022_p12.pdf"},"started_at":"2026-11-04T19:45:58.484313"} +{"completed_at":"2026-11-04T19:46:02.284333","latency_in_ms":2791,"output":{"example_id":"chart/Digital_News-Report_2022_p13","grounded_pages":[],"layout_pages":[],"markdown":"**Selective news avoidance and its implications**\n\nWhile the majority of people across countries remain engaged and use the news regularly, we find that many also increasingly choose to ration or limit their exposure to it – or at least to certain types of news. We call this behaviour _selective news avoidance_ and the growth of this activity may help to explain why consumption levels have mostly not increased, despite the uncertain times in which we live. The proportion that says they avoid the news, sometimes or often, has doubled in Brazil (54%) and the UK (46%) since 2017 – and also increased in all other markets (see next chart). This type of selective avoidance seems to be less widespread in Northern European countries such as Germany (29%), Denmark, and Finland (20%), as well as in some Asian countries such as Japan (14%).\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 5\n\n0% 25% 50% 75%\n\n54\n\n46\n\n42\n\n41\n\n41\n\n36\n\n35\n\n34\n\n29\n\n20\n\n20\n\n14\n\n34\n\n35\n\n41\n\n32\n\n29\n\n33\n\n33\n\n36\n\n25\n\n15\n\n17\n\n11\n\n27\n\n24\n\n38\n\n30\n\n30\n\n29\n\n26\n\n28\n\n24\n\n14\n\n18\n\n6\n\n2022\n\n2019\n\n2017\n\nJapan\n\nFinland\n\nDenmark\n\nGermany\n\nItaly\n\nSpain\n\nFrance\n\nAustralia\n\nIreland\n\nUSA\n\nUK\n\nBrazil\n\nPROPORTION WHO SOMETIMES OR OFTEN ACTIVELY AVOID THE NEWS (2017–22) – SELECTED MARKETS\n\n**Q1di\\_2017.** Do you find yourself actively trying to avoid news these days? _Base: Total 2017–22 samples (n≈2000)._\n\n2017\n\n2019\n\n2022\n\n# **38%**\n\nAll country average (was 29% in 2017)\n\nSelective news avoiders give a variety of reasons for their behaviour. Across markets, many respondents say they are put off by the repetitiveness of the news agenda – especially around politics and COVID-19 (43%), or that they often feel worn out by the news (29%). A significant proportion say they avoid news because they think it can’t be trusted (29%). Around a third (36%), particularly those who are under 35, say that the news brings down their mood. Others say the news leads to arguments they would rather avoid (17%), or leads to feelings of powerlessness (16%). A small proportion say they don’t have enough time for news (14%) or that it is too hard to understand (8%).\n\nConcerns about the news having a negative effect on their mood are higher amongst avoiders in the United Kingdom (55%) and United States (49%) than they are elsewhere.\n\n**_I actively avoid things that trigger my anxiety and things that can have a negative impact on my day. I will try to avoid reading news about things like deaths and disasters._** _Male, 27, UK_\n\n**_Truthfully, I don’t like to dwell too much on the mainstream news. I find sometimes it can be repetitive and negative._** _Female, 19, UK_\n\nPolitical allegiances can also make a striking difference to why people choose to avoid news. In the United States, those who self-identify on the right are far more likely to avoid news because they think it is untrustworthy or biased, but those on the left are more likely to feel overwhelmed, carry feelings of powerlessness, or worry that the news might create arguments.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 6\n\nMOST COMMON REASONS FOR NEWS AVOIDANCE – ALL MARKETS\n\n**Q1di\\_2017ii.** Why do you find yourself actively trying to avoid the news? _Base: All who avoid the news often, sometimes, or occasionally. All markets = 64,120._\n\n## **43%**\n\nsay there is too much politics and COVID-19\n\n## **36%**\n\nsay news has negative effect on mood\n\n## **29%**\n\nsay they are worn out by amount of news\n\n## **29%**\n\nsay the news is untrustworthy or biased\n\n## **17%**\n\nsay it leads to arguments I’d rather avoid\n\n## **16%**\n\nsay there is nothing I can do with the information\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 13","pages":[{"line_numbers":[],"markdown":"**Selective news avoidance and its implications**\n\nWhile the majority of people across countries remain engaged and use the news regularly, we find that many also increasingly choose to ration or limit their exposure to it – or at least to certain types of news. We call this behaviour _selective news avoidance_ and the growth of this activity may help to explain why consumption levels have mostly not increased, despite the uncertain times in which we live. The proportion that says they avoid the news, sometimes or often, has doubled in Brazil (54%) and the UK (46%) since 2017 – and also increased in all other markets (see next chart). This type of selective avoidance seems to be less widespread in Northern European countries such as Germany (29%), Denmark, and Finland (20%), as well as in some Asian countries such as Japan (14%).\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 5\n\n0% 25% 50% 75%\n\n54\n\n46\n\n42\n\n41\n\n41\n\n36\n\n35\n\n34\n\n29\n\n20\n\n20\n\n14\n\n34\n\n35\n\n41\n\n32\n\n29\n\n33\n\n33\n\n36\n\n25\n\n15\n\n17\n\n11\n\n27\n\n24\n\n38\n\n30\n\n30\n\n29\n\n26\n\n28\n\n24\n\n14\n\n18\n\n6\n\n2022\n\n2019\n\n2017\n\nJapan\n\nFinland\n\nDenmark\n\nGermany\n\nItaly\n\nSpain\n\nFrance\n\nAustralia\n\nIreland\n\nUSA\n\nUK\n\nBrazil\n\nPROPORTION WHO SOMETIMES OR OFTEN ACTIVELY AVOID THE NEWS (2017–22) – SELECTED MARKETS\n\n**Q1di\\_2017.** Do you find yourself actively trying to avoid news these days? _Base: Total 2017–22 samples (n≈2000)._\n\n2017\n\n2019\n\n2022\n\n# **38%**\n\nAll country average (was 29% in 2017)\n\nSelective news avoiders give a variety of reasons for their behaviour. Across markets, many respondents say they are put off by the repetitiveness of the news agenda – especially around politics and COVID-19 (43%), or that they often feel worn out by the news (29%). A significant proportion say they avoid news because they think it can’t be trusted (29%). Around a third (36%), particularly those who are under 35, say that the news brings down their mood. Others say the news leads to arguments they would rather avoid (17%), or leads to feelings of powerlessness (16%). A small proportion say they don’t have enough time for news (14%) or that it is too hard to understand (8%).\n\nConcerns about the news having a negative effect on their mood are higher amongst avoiders in the United Kingdom (55%) and United States (49%) than they are elsewhere.\n\n**_I actively avoid things that trigger my anxiety and things that can have a negative impact on my day. I will try to avoid reading news about things like deaths and disasters._** _Male, 27, UK_\n\n**_Truthfully, I don’t like to dwell too much on the mainstream news. I find sometimes it can be repetitive and negative._** _Female, 19, UK_\n\nPolitical allegiances can also make a striking difference to why people choose to avoid news. In the United States, those who self-identify on the right are far more likely to avoid news because they think it is untrustworthy or biased, but those on the left are more likely to feel overwhelmed, carry feelings of powerlessness, or worry that the news might create arguments.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 6\n\nMOST COMMON REASONS FOR NEWS AVOIDANCE – ALL MARKETS\n\n**Q1di\\_2017ii.** Why do you find yourself actively trying to avoid the news? _Base: All who avoid the news often, sometimes, or occasionally. All markets = 64,120._\n\n## **43%**\n\nsay there is too much politics and COVID-19\n\n## **36%**\n\nsay news has negative effect on mood\n\n## **29%**\n\nsay they are worn out by amount of news\n\n## **29%**\n\nsay the news is untrustworthy or biased\n\n## **17%**\n\nsay it leads to arguments I’d rather avoid\n\n## **16%**\n\nsay there is nothing I can do with the information\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 13","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**Selective news avoidance and its implications**\n\nWhile the majority of people across countries remain engaged and use the news regularly, we find that many also increasingly choose to ration or limit their exposure to it – or at least to certain types of news. We call this behaviour _selective news avoidance_ and the growth of this activity may help to explain why consumption levels have mostly not increased, despite the uncertain times in which we live. The proportion that says they avoid the news, sometimes or often, has doubled in Brazil (54%) and the UK (46%) since 2017 – and also increased in all other markets (see next chart). This type of selective avoidance seems to be less widespread in Northern European countries such as Germany (29%), Denmark, and Finland (20%), as well as in some Asian countries such as Japan (14%).\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 5\n\n0% 25% 50% 75%\n\n54\n\n46\n\n42\n\n41\n\n41\n\n36\n\n35\n\n34\n\n29\n\n20\n\n20\n\n14\n\n34\n\n35\n\n41\n\n32\n\n29\n\n33\n\n33\n\n36\n\n25\n\n15\n\n17\n\n11\n\n27\n\n24\n\n38\n\n30\n\n30\n\n29\n\n26\n\n28\n\n24\n\n14\n\n18\n\n6\n\n2022\n\n2019\n\n2017\n\nJapan\n\nFinland\n\nDenmark\n\nGermany\n\nItaly\n\nSpain\n\nFrance\n\nAustralia\n\nIreland\n\nUSA\n\nUK\n\nBrazil\n\nPROPORTION WHO SOMETIMES OR OFTEN ACTIVELY AVOID THE NEWS (2017–22) – SELECTED MARKETS\n\n**Q1di\\_2017.** Do you find yourself actively trying to avoid news these days? _Base: Total 2017–22 samples (n≈2000)._\n\n2017\n\n2019\n\n2022\n\n# **38%**\n\nAll country average (was 29% in 2017)\n\nSelective news avoiders give a variety of reasons for their behaviour. Across markets, many respondents say they are put off by the repetitiveness of the news agenda – especially around politics and COVID-19 (43%), or that they often feel worn out by the news (29%). A significant proportion say they avoid news because they think it can’t be trusted (29%). Around a third (36%), particularly those who are under 35, say that the news brings down their mood. Others say the news leads to arguments they would rather avoid (17%), or leads to feelings of powerlessness (16%). A small proportion say they don’t have enough time for news (14%) or that it is too hard to understand (8%).\n\nConcerns about the news having a negative effect on their mood are higher amongst avoiders in the United Kingdom (55%) and United States (49%) than they are elsewhere.\n\n**_I actively avoid things that trigger my anxiety and things that can have a negative impact on my day. I will try to avoid reading news about things like deaths and disasters._** _Male, 27, UK_\n\n**_Truthfully, I don’t like to dwell too much on the mainstream news. I find sometimes it can be repetitive and negative._** _Female, 19, UK_\n\nPolitical allegiances can also make a striking difference to why people choose to avoid news. In the United States, those who self-identify on the right are far more likely to avoid news because they think it is untrustworthy or biased, but those on the left are more likely to feel overwhelmed, carry feelings of powerlessness, or worry that the news might create arguments.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 6\n\nMOST COMMON REASONS FOR NEWS AVOIDANCE – ALL MARKETS\n\n**Q1di\\_2017ii.** Why do you find yourself actively trying to avoid the news? _Base: All who avoid the news often, sometimes, or occasionally. All markets = 64,120._\n\n## **43%**\n\nsay there is too much politics and COVID-19\n\n## **36%**\n\nsay news has negative effect on mood\n\n## **29%**\n\nsay they are worn out by amount of news\n\n## **29%**\n\nsay the news is untrustworthy or biased\n\n## **17%**\n\nsay it leads to arguments I’d rather avoid\n\n## **16%**\n\nsay there is nothing I can do with the information\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 13","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Digital_News-Report_2022_p13.pdf"]},"response":{"errors":[],"extract_id":"extract_d8d9c19d552a0d914bee5cdbe2f92cce","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**Selective news avoidance and its implications**\n\nWhile the majority of people across countries remain engaged and use the news regularly, we find that many also increasingly choose to ration or limit their exposure to it – or at least to certain types of news. We call this behaviour _selective news avoidance_ and the growth of this activity may help to explain why consumption levels have mostly not increased, despite the uncertain times in which we live. The proportion that says they avoid the news, sometimes or often, has doubled in Brazil (54%) and the UK (46%) since 2017 – and also increased in all other markets (see next chart). This type of selective avoidance seems to be less widespread in Northern European countries such as Germany (29%), Denmark, and Finland (20%), as well as in some Asian countries such as Japan (14%).\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 5\n\n0% 25% 50% 75%\n\n54\n\n46\n\n42\n\n41\n\n41\n\n36\n\n35\n\n34\n\n29\n\n20\n\n20\n\n14\n\n34\n\n35\n\n41\n\n32\n\n29\n\n33\n\n33\n\n36\n\n25\n\n15\n\n17\n\n11\n\n27\n\n24\n\n38\n\n30\n\n30\n\n29\n\n26\n\n28\n\n24\n\n14\n\n18\n\n6\n\n2022\n\n2019\n\n2017\n\nJapan\n\nFinland\n\nDenmark\n\nGermany\n\nItaly\n\nSpain\n\nFrance\n\nAustralia\n\nIreland\n\nUSA\n\nUK\n\nBrazil\n\nPROPORTION WHO SOMETIMES OR OFTEN ACTIVELY AVOID THE NEWS (2017–22) – SELECTED MARKETS\n\n**Q1di\\_2017.** Do you find yourself actively trying to avoid news these days? _Base: Total 2017–22 samples (n≈2000)._\n\n2017\n\n2019\n\n2022\n\n# **38%**\n\nAll country average (was 29% in 2017)\n\nSelective news avoiders give a variety of reasons for their behaviour. Across markets, many respondents say they are put off by the repetitiveness of the news agenda – especially around politics and COVID-19 (43%), or that they often feel worn out by the news (29%). A significant proportion say they avoid news because they think it can’t be trusted (29%). Around a third (36%), particularly those who are under 35, say that the news brings down their mood. Others say the news leads to arguments they would rather avoid (17%), or leads to feelings of powerlessness (16%). A small proportion say they don’t have enough time for news (14%) or that it is too hard to understand (8%).\n\nConcerns about the news having a negative effect on their mood are higher amongst avoiders in the United Kingdom (55%) and United States (49%) than they are elsewhere.\n\n**_I actively avoid things that trigger my anxiety and things that can have a negative impact on my day. I will try to avoid reading news about things like deaths and disasters._** _Male, 27, UK_\n\n**_Truthfully, I don’t like to dwell too much on the mainstream news. I find sometimes it can be repetitive and negative._** _Female, 19, UK_\n\nPolitical allegiances can also make a striking difference to why people choose to avoid news. In the United States, those who self-identify on the right are far more likely to avoid news because they think it is untrustworthy or biased, but those on the left are more likely to feel overwhelmed, carry feelings of powerlessness, or worry that the news might create arguments.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 6\n\nMOST COMMON REASONS FOR NEWS AVOIDANCE – ALL MARKETS\n\n**Q1di\\_2017ii.** Why do you find yourself actively trying to avoid the news? _Base: All who avoid the news often, sometimes, or occasionally. All markets = 64,120._\n\n## **43%**\n\nsay there is too much politics and COVID-19\n\n## **36%**\n\nsay news has negative effect on mood\n\n## **29%**\n\nsay they are worn out by amount of news\n\n## **29%**\n\nsay the news is untrustworthy or biased\n\n## **17%**\n\nsay it leads to arguments I’d rather avoid\n\n## **16%**\n\nsay there is nothing I can do with the information\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 13"],"full_content":"**Selective news avoidance and its implications**\n\nWhile the majority of people across countries remain engaged and use the news regularly, we find that many also increasingly choose to ration or limit their exposure to it – or at least to certain types of news. We call this behaviour _selective news avoidance_ and the growth of this activity may help to explain why consumption levels have mostly not increased, despite the uncertain times in which we live. The proportion that says they avoid the news, sometimes or often, has doubled in Brazil (54%) and the UK (46%) since 2017 – and also increased in all other markets (see next chart). This type of selective avoidance seems to be less widespread in Northern European countries such as Germany (29%), Denmark, and Finland (20%), as well as in some Asian countries such as Japan (14%).\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 5\n\n0% 25% 50% 75%\n\n54\n\n46\n\n42\n\n41\n\n41\n\n36\n\n35\n\n34\n\n29\n\n20\n\n20\n\n14\n\n34\n\n35\n\n41\n\n32\n\n29\n\n33\n\n33\n\n36\n\n25\n\n15\n\n17\n\n11\n\n27\n\n24\n\n38\n\n30\n\n30\n\n29\n\n26\n\n28\n\n24\n\n14\n\n18\n\n6\n\n2022\n\n2019\n\n2017\n\nJapan\n\nFinland\n\nDenmark\n\nGermany\n\nItaly\n\nSpain\n\nFrance\n\nAustralia\n\nIreland\n\nUSA\n\nUK\n\nBrazil\n\nPROPORTION WHO SOMETIMES OR OFTEN ACTIVELY AVOID THE NEWS (2017–22) – SELECTED MARKETS\n\n**Q1di\\_2017.** Do you find yourself actively trying to avoid news these days? _Base: Total 2017–22 samples (n≈2000)._\n\n2017\n\n2019\n\n2022\n\n# **38%**\n\nAll country average (was 29% in 2017)\n\nSelective news avoiders give a variety of reasons for their behaviour. Across markets, many respondents say they are put off by the repetitiveness of the news agenda – especially around politics and COVID-19 (43%), or that they often feel worn out by the news (29%). A significant proportion say they avoid news because they think it can’t be trusted (29%). Around a third (36%), particularly those who are under 35, say that the news brings down their mood. Others say the news leads to arguments they would rather avoid (17%), or leads to feelings of powerlessness (16%). A small proportion say they don’t have enough time for news (14%) or that it is too hard to understand (8%).\n\nConcerns about the news having a negative effect on their mood are higher amongst avoiders in the United Kingdom (55%) and United States (49%) than they are elsewhere.\n\n**_I actively avoid things that trigger my anxiety and things that can have a negative impact on my day. I will try to avoid reading news about things like deaths and disasters._** _Male, 27, UK_\n\n**_Truthfully, I don’t like to dwell too much on the mainstream news. I find sometimes it can be repetitive and negative._** _Female, 19, UK_\n\nPolitical allegiances can also make a striking difference to why people choose to avoid news. In the United States, those who self-identify on the right are far more likely to avoid news because they think it is untrustworthy or biased, but those on the left are more likely to feel overwhelmed, carry feelings of powerlessness, or worry that the news might create arguments.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 6\n\nMOST COMMON REASONS FOR NEWS AVOIDANCE – ALL MARKETS\n\n**Q1di\\_2017ii.** Why do you find yourself actively trying to avoid the news? _Base: All who avoid the news often, sometimes, or occasionally. All markets = 64,120._\n\n## **43%**\n\nsay there is too much politics and COVID-19\n\n## **36%**\n\nsay news has negative effect on mood\n\n## **29%**\n\nsay they are worn out by amount of news\n\n## **29%**\n\nsay the news is untrustworthy or biased\n\n## **17%**\n\nsay it leads to arguments I’d rather avoid\n\n## **16%**\n\nsay there is nothing I can do with the information\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 13","publish_date":null,"title":"38%","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Digital_News-Report_2022_p13.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Digital_News-Report_2022_p13.pdf"},"request":{"example_id":"chart/Digital_News-Report_2022_p13","product_type":"parse","source_file_path":"Digital_News-Report_2022_p13.pdf"},"started_at":"2026-11-04T19:45:59.493087"} +{"completed_at":"2026-11-04T19:46:03.771882","latency_in_ms":2390,"output":{"example_id":"chart/Digital_News-Report_2022_p14","grounded_pages":[],"layout_pages":[],"markdown":"REUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 7\n\nPROPORTION WHO AVOID NEWS FOR EACH REASON – BY POLITICAL LEANING – USA\n\n**Q1di\\_2017ii.** Why do you find yourself actively trying to avoid the news? _Base: All who avoid news often, sometimes, or occasionally in USA. Left =327, Right= 336._\n\n0% 25% 50% 75%\n\n54\n\n29\n\n49\n\n65\n\n14\n\n15\n\n3\n\n57\n\n41\n\n23\n\n20\n\n22\n\n18\n\n9\n\nRight\n\nLeft\n\nHard to follow/ understand\n\nLeads to arguments\n\nNothing I can do about it\n\nNews is untrustworthy or biased\n\nToo much politics or COVID-19 news\n\nWorn out by the amount of news\n\nBrings down my mood\n\nLeft Right\n\nThese findings will be particularly challenging for the news industry. Subjects that journalists consider most important, such as political crises, international conflicts, global pandemics, and climate catastrophes, seem to be precisely the ones that are turning some people away from news – especially amongst those who are younger or harder to reach.\n\nMany news organisations are embracing approaches such as solutions journalism around subjects like climate change, that aim to give people a sense of hope or personal agency. Others are looking to find ways to widen the agenda to softer subjects or make news more relevant at a personal level, but there will be a limit to how far journalists can go – or should go – to make the news more palatable.\n\n**News is** **_too hard_** **for many people to work out**\n\nOne other important data point relates to the difficulty that many younger audiences, and less educated groups, have in understanding journalism as currently practised. In countries such as Australia, the United States, and Brazil, around 15% of younger avoiders say they find news hard to follow – a much higher proportion than older news consumers.\n\n**_I do sometimes finish reading articles maybe more stressed about a situation, or sometimes just confused._** _Female, 24, UK_\n\nThis could relate to the complexity of the language or assumed knowledge often contained in news reports. But the increase in news consumption via social media or word of mouth through friends and family, may also be playing a part. News is often accessed by young people in more fragmented ways, meaning that people sometimes miss key context that was previously carefully packaged into linear narratives by the mainstream media.\n\nDuring the COVID-19 crisis (and now Ukraine) we saw many news organisations using explainer and Question & Answer formats to try to address these issues on websites and via social media to engage younger and less educated audiences. Our data suggest this process needs to go much further.\n\n_For further analysis see Section 2.3: The Changing News Habits and Attitudes of Younger Audiences_\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 8\n\nPROPORTION OF NEWS AVOIDERS WHO STRUGGLE TO UNDERSTAND THE NEWS – BY AGE – SELECTED MARKETS\n\n**Q1di\\_2017ii.** Why do you find yourself actively trying to avoid the news? _Base: Under 35s/35s and over who avoid news: Australia = 428/963, Brazil = 559/1053, Denmark = 283/648, Finland = 260/705, France = 319/1075, Germany = 333/950, Ireland = 367/999, Italy = 279/1080, Japan = 235/825, Spain = 336/1035, UK = 386/1310, USA = 450/987._\n\n0% 25%\n\n4\n\n6\n\n5\n\n7\n\n6\n\n3\n\n7\n\n4\n\n4\n\n2\n\n2\n\n6\n\n16\n\n15\n\n15\n\n11\n\n11\n\n11\n\n11\n\n10\n\n10\n\n8\n\n6\n\n6\n\nOv\n\nU\n\nItaly\n\nUK\n\nFinland\n\nIreland\n\nGermany\n\nJapan\n\nDenmark\n\nSpain\n\nFrance\n\nUSA\n\nBrazil\n\nAustralia\n\nU35\n\n35+\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 14","pages":[{"line_numbers":[],"markdown":"REUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 7\n\nPROPORTION WHO AVOID NEWS FOR EACH REASON – BY POLITICAL LEANING – USA\n\n**Q1di\\_2017ii.** Why do you find yourself actively trying to avoid the news? _Base: All who avoid news often, sometimes, or occasionally in USA. Left =327, Right= 336._\n\n0% 25% 50% 75%\n\n54\n\n29\n\n49\n\n65\n\n14\n\n15\n\n3\n\n57\n\n41\n\n23\n\n20\n\n22\n\n18\n\n9\n\nRight\n\nLeft\n\nHard to follow/ understand\n\nLeads to arguments\n\nNothing I can do about it\n\nNews is untrustworthy or biased\n\nToo much politics or COVID-19 news\n\nWorn out by the amount of news\n\nBrings down my mood\n\nLeft Right\n\nThese findings will be particularly challenging for the news industry. Subjects that journalists consider most important, such as political crises, international conflicts, global pandemics, and climate catastrophes, seem to be precisely the ones that are turning some people away from news – especially amongst those who are younger or harder to reach.\n\nMany news organisations are embracing approaches such as solutions journalism around subjects like climate change, that aim to give people a sense of hope or personal agency. Others are looking to find ways to widen the agenda to softer subjects or make news more relevant at a personal level, but there will be a limit to how far journalists can go – or should go – to make the news more palatable.\n\n**News is** **_too hard_** **for many people to work out**\n\nOne other important data point relates to the difficulty that many younger audiences, and less educated groups, have in understanding journalism as currently practised. In countries such as Australia, the United States, and Brazil, around 15% of younger avoiders say they find news hard to follow – a much higher proportion than older news consumers.\n\n**_I do sometimes finish reading articles maybe more stressed about a situation, or sometimes just confused._** _Female, 24, UK_\n\nThis could relate to the complexity of the language or assumed knowledge often contained in news reports. But the increase in news consumption via social media or word of mouth through friends and family, may also be playing a part. News is often accessed by young people in more fragmented ways, meaning that people sometimes miss key context that was previously carefully packaged into linear narratives by the mainstream media.\n\nDuring the COVID-19 crisis (and now Ukraine) we saw many news organisations using explainer and Question & Answer formats to try to address these issues on websites and via social media to engage younger and less educated audiences. Our data suggest this process needs to go much further.\n\n_For further analysis see Section 2.3: The Changing News Habits and Attitudes of Younger Audiences_\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 8\n\nPROPORTION OF NEWS AVOIDERS WHO STRUGGLE TO UNDERSTAND THE NEWS – BY AGE – SELECTED MARKETS\n\n**Q1di\\_2017ii.** Why do you find yourself actively trying to avoid the news? _Base: Under 35s/35s and over who avoid news: Australia = 428/963, Brazil = 559/1053, Denmark = 283/648, Finland = 260/705, France = 319/1075, Germany = 333/950, Ireland = 367/999, Italy = 279/1080, Japan = 235/825, Spain = 336/1035, UK = 386/1310, USA = 450/987._\n\n0% 25%\n\n4\n\n6\n\n5\n\n7\n\n6\n\n3\n\n7\n\n4\n\n4\n\n2\n\n2\n\n6\n\n16\n\n15\n\n15\n\n11\n\n11\n\n11\n\n11\n\n10\n\n10\n\n8\n\n6\n\n6\n\nOv\n\nU\n\nItaly\n\nUK\n\nFinland\n\nIreland\n\nGermany\n\nJapan\n\nDenmark\n\nSpain\n\nFrance\n\nUSA\n\nBrazil\n\nAustralia\n\nU35\n\n35+\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 14","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"REUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 7\n\nPROPORTION WHO AVOID NEWS FOR EACH REASON – BY POLITICAL LEANING – USA\n\n**Q1di\\_2017ii.** Why do you find yourself actively trying to avoid the news? _Base: All who avoid news often, sometimes, or occasionally in USA. Left =327, Right= 336._\n\n0% 25% 50% 75%\n\n54\n\n29\n\n49\n\n65\n\n14\n\n15\n\n3\n\n57\n\n41\n\n23\n\n20\n\n22\n\n18\n\n9\n\nRight\n\nLeft\n\nHard to follow/ understand\n\nLeads to arguments\n\nNothing I can do about it\n\nNews is untrustworthy or biased\n\nToo much politics or COVID-19 news\n\nWorn out by the amount of news\n\nBrings down my mood\n\nLeft Right\n\nThese findings will be particularly challenging for the news industry. Subjects that journalists consider most important, such as political crises, international conflicts, global pandemics, and climate catastrophes, seem to be precisely the ones that are turning some people away from news – especially amongst those who are younger or harder to reach.\n\nMany news organisations are embracing approaches such as solutions journalism around subjects like climate change, that aim to give people a sense of hope or personal agency. Others are looking to find ways to widen the agenda to softer subjects or make news more relevant at a personal level, but there will be a limit to how far journalists can go – or should go – to make the news more palatable.\n\n**News is** **_too hard_** **for many people to work out**\n\nOne other important data point relates to the difficulty that many younger audiences, and less educated groups, have in understanding journalism as currently practised. In countries such as Australia, the United States, and Brazil, around 15% of younger avoiders say they find news hard to follow – a much higher proportion than older news consumers.\n\n**_I do sometimes finish reading articles maybe more stressed about a situation, or sometimes just confused._** _Female, 24, UK_\n\nThis could relate to the complexity of the language or assumed knowledge often contained in news reports. But the increase in news consumption via social media or word of mouth through friends and family, may also be playing a part. News is often accessed by young people in more fragmented ways, meaning that people sometimes miss key context that was previously carefully packaged into linear narratives by the mainstream media.\n\nDuring the COVID-19 crisis (and now Ukraine) we saw many news organisations using explainer and Question & Answer formats to try to address these issues on websites and via social media to engage younger and less educated audiences. Our data suggest this process needs to go much further.\n\n_For further analysis see Section 2.3: The Changing News Habits and Attitudes of Younger Audiences_\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 8\n\nPROPORTION OF NEWS AVOIDERS WHO STRUGGLE TO UNDERSTAND THE NEWS – BY AGE – SELECTED MARKETS\n\n**Q1di\\_2017ii.** Why do you find yourself actively trying to avoid the news? _Base: Under 35s/35s and over who avoid news: Australia = 428/963, Brazil = 559/1053, Denmark = 283/648, Finland = 260/705, France = 319/1075, Germany = 333/950, Ireland = 367/999, Italy = 279/1080, Japan = 235/825, Spain = 336/1035, UK = 386/1310, USA = 450/987._\n\n0% 25%\n\n4\n\n6\n\n5\n\n7\n\n6\n\n3\n\n7\n\n4\n\n4\n\n2\n\n2\n\n6\n\n16\n\n15\n\n15\n\n11\n\n11\n\n11\n\n11\n\n10\n\n10\n\n8\n\n6\n\n6\n\nOv\n\nU\n\nItaly\n\nUK\n\nFinland\n\nIreland\n\nGermany\n\nJapan\n\nDenmark\n\nSpain\n\nFrance\n\nUSA\n\nBrazil\n\nAustralia\n\nU35\n\n35+\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 14","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Digital_News-Report_2022_p14.pdf"]},"response":{"errors":[],"extract_id":"extract_9ac882274b39378a682e2c6bc75b36db","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["REUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 7\n\nPROPORTION WHO AVOID NEWS FOR EACH REASON – BY POLITICAL LEANING – USA\n\n**Q1di\\_2017ii.** Why do you find yourself actively trying to avoid the news? _Base: All who avoid news often, sometimes, or occasionally in USA. Left =327, Right= 336._\n\n0% 25% 50% 75%\n\n54\n\n29\n\n49\n\n65\n\n14\n\n15\n\n3\n\n57\n\n41\n\n23\n\n20\n\n22\n\n18\n\n9\n\nRight\n\nLeft\n\nHard to follow/ understand\n\nLeads to arguments\n\nNothing I can do about it\n\nNews is untrustworthy or biased\n\nToo much politics or COVID-19 news\n\nWorn out by the amount of news\n\nBrings down my mood\n\nLeft Right\n\nThese findings will be particularly challenging for the news industry. Subjects that journalists consider most important, such as political crises, international conflicts, global pandemics, and climate catastrophes, seem to be precisely the ones that are turning some people away from news – especially amongst those who are younger or harder to reach.\n\nMany news organisations are embracing approaches such as solutions journalism around subjects like climate change, that aim to give people a sense of hope or personal agency. Others are looking to find ways to widen the agenda to softer subjects or make news more relevant at a personal level, but there will be a limit to how far journalists can go – or should go – to make the news more palatable.\n\n**News is** **_too hard_** **for many people to work out**\n\nOne other important data point relates to the difficulty that many younger audiences, and less educated groups, have in understanding journalism as currently practised. In countries such as Australia, the United States, and Brazil, around 15% of younger avoiders say they find news hard to follow – a much higher proportion than older news consumers.\n\n**_I do sometimes finish reading articles maybe more stressed about a situation, or sometimes just confused._** _Female, 24, UK_\n\nThis could relate to the complexity of the language or assumed knowledge often contained in news reports. But the increase in news consumption via social media or word of mouth through friends and family, may also be playing a part. News is often accessed by young people in more fragmented ways, meaning that people sometimes miss key context that was previously carefully packaged into linear narratives by the mainstream media.\n\nDuring the COVID-19 crisis (and now Ukraine) we saw many news organisations using explainer and Question & Answer formats to try to address these issues on websites and via social media to engage younger and less educated audiences. Our data suggest this process needs to go much further.\n\n_For further analysis see Section 2.3: The Changing News Habits and Attitudes of Younger Audiences_\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 8\n\nPROPORTION OF NEWS AVOIDERS WHO STRUGGLE TO UNDERSTAND THE NEWS – BY AGE – SELECTED MARKETS\n\n**Q1di\\_2017ii.** Why do you find yourself actively trying to avoid the news? _Base: Under 35s/35s and over who avoid news: Australia = 428/963, Brazil = 559/1053, Denmark = 283/648, Finland = 260/705, France = 319/1075, Germany = 333/950, Ireland = 367/999, Italy = 279/1080, Japan = 235/825, Spain = 336/1035, UK = 386/1310, USA = 450/987._\n\n0% 25%\n\n4\n\n6\n\n5\n\n7\n\n6\n\n3\n\n7\n\n4\n\n4\n\n2\n\n2\n\n6\n\n16\n\n15\n\n15\n\n11\n\n11\n\n11\n\n11\n\n10\n\n10\n\n8\n\n6\n\n6\n\nOv\n\nU\n\nItaly\n\nUK\n\nFinland\n\nIreland\n\nGermany\n\nJapan\n\nDenmark\n\nSpain\n\nFrance\n\nUSA\n\nBrazil\n\nAustralia\n\nU35\n\n35+\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 14"],"full_content":"REUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 7\n\nPROPORTION WHO AVOID NEWS FOR EACH REASON – BY POLITICAL LEANING – USA\n\n**Q1di\\_2017ii.** Why do you find yourself actively trying to avoid the news? _Base: All who avoid news often, sometimes, or occasionally in USA. Left =327, Right= 336._\n\n0% 25% 50% 75%\n\n54\n\n29\n\n49\n\n65\n\n14\n\n15\n\n3\n\n57\n\n41\n\n23\n\n20\n\n22\n\n18\n\n9\n\nRight\n\nLeft\n\nHard to follow/ understand\n\nLeads to arguments\n\nNothing I can do about it\n\nNews is untrustworthy or biased\n\nToo much politics or COVID-19 news\n\nWorn out by the amount of news\n\nBrings down my mood\n\nLeft Right\n\nThese findings will be particularly challenging for the news industry. Subjects that journalists consider most important, such as political crises, international conflicts, global pandemics, and climate catastrophes, seem to be precisely the ones that are turning some people away from news – especially amongst those who are younger or harder to reach.\n\nMany news organisations are embracing approaches such as solutions journalism around subjects like climate change, that aim to give people a sense of hope or personal agency. Others are looking to find ways to widen the agenda to softer subjects or make news more relevant at a personal level, but there will be a limit to how far journalists can go – or should go – to make the news more palatable.\n\n**News is** **_too hard_** **for many people to work out**\n\nOne other important data point relates to the difficulty that many younger audiences, and less educated groups, have in understanding journalism as currently practised. In countries such as Australia, the United States, and Brazil, around 15% of younger avoiders say they find news hard to follow – a much higher proportion than older news consumers.\n\n**_I do sometimes finish reading articles maybe more stressed about a situation, or sometimes just confused._** _Female, 24, UK_\n\nThis could relate to the complexity of the language or assumed knowledge often contained in news reports. But the increase in news consumption via social media or word of mouth through friends and family, may also be playing a part. News is often accessed by young people in more fragmented ways, meaning that people sometimes miss key context that was previously carefully packaged into linear narratives by the mainstream media.\n\nDuring the COVID-19 crisis (and now Ukraine) we saw many news organisations using explainer and Question & Answer formats to try to address these issues on websites and via social media to engage younger and less educated audiences. Our data suggest this process needs to go much further.\n\n_For further analysis see Section 2.3: The Changing News Habits and Attitudes of Younger Audiences_\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 8\n\nPROPORTION OF NEWS AVOIDERS WHO STRUGGLE TO UNDERSTAND THE NEWS – BY AGE – SELECTED MARKETS\n\n**Q1di\\_2017ii.** Why do you find yourself actively trying to avoid the news? _Base: Under 35s/35s and over who avoid news: Australia = 428/963, Brazil = 559/1053, Denmark = 283/648, Finland = 260/705, France = 319/1075, Germany = 333/950, Ireland = 367/999, Italy = 279/1080, Japan = 235/825, Spain = 336/1035, UK = 386/1310, USA = 450/987._\n\n0% 25%\n\n4\n\n6\n\n5\n\n7\n\n6\n\n3\n\n7\n\n4\n\n4\n\n2\n\n2\n\n6\n\n16\n\n15\n\n15\n\n11\n\n11\n\n11\n\n11\n\n10\n\n10\n\n8\n\n6\n\n6\n\nOv\n\nU\n\nItaly\n\nUK\n\nFinland\n\nIreland\n\nGermany\n\nJapan\n\nDenmark\n\nSpain\n\nFrance\n\nUSA\n\nBrazil\n\nAustralia\n\nU35\n\n35+\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 14","publish_date":null,"title":"Graph 7 Graph 8","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Digital_News-Report_2022_p14.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Digital_News-Report_2022_p14.pdf"},"request":{"example_id":"chart/Digital_News-Report_2022_p14","product_type":"parse","source_file_path":"Digital_News-Report_2022_p14.pdf"},"started_at":"2026-11-04T19:46:01.381779"} +{"completed_at":"2026-11-04T19:31:16.221012","latency_in_ms":1391,"output":{"example_id":"chart/Digital_News-Report_2022_p15","grounded_pages":[],"layout_pages":[],"markdown":"**TRUST FALLS BACK AFTER COVID-19 BUMPS**\n\nThis year we find lower levels of overall trust in 21 of our 46 markets, partly reversing the gains made at the height of the Coronavirus pandemic. Eighteen markets are at a similar level, with just seven on an upward path. The average level of trust (42%) is also a little lower than last year and trust in individual news brands is on a downward trend in most countries.\n\nFinland remains the country with the highest levels of overall trust (69%) – up four points on last year and 13 points on 2020. At the other end of the scale, news trust in the USA (26%) has fallen by a further three percentage points and remains the lowest level in our survey along with Slovakia (26%). Notable changes in Europe include falls in Romania (-9), Croatia (-7), Poland (-6), Switzerland (-5), Austria (-5), Greece (-5), Italy (-5), and Spain (-4). In Latin America, trust is down in Brazil (-6) and Colombia (-3), but level or slightly up elsewhere. It is a mixed picture in Africa, with a fall in Kenya (-4) but strong growth in South Africa (+9) and Nigeria (+4). Finally, in Asia, trust has risen in the Philippines (+5) and Japan (+2) but is down in Malaysia (-5) and Taiwan (-4).\n\nIn some regions we find a widening gap between markets with the highest levels of trust and those with the lowest. We also find that those markets with the most trust – such as Finland – also tend to have higher interest in news and lower levels of active news avoidance. By contrast, low trust countries, such as the United States, France, the United Kingdom, and Slovakia, see some of the highest levels of selective avoidance, as well as news disconnection, with declining interest in news.\n\nIt is important to put this year’s trust changes in context. In most cases the headline trust level is still higher than it was before the Coronavirus pandemic reinforced the importance of reliable media for many people. On the other hand, the chart below also reminds us that, in many countries, trust had been on a downward trajectory for some time – partly driven by a series of polarising events like Brexit in the UK, the turmoil of the Trump years in the United States, and the Gilets Jaunes protests in France. In this context, last year’s COVID-19 optimism looks more like a short- term rally rather than any kind of longer-term renaissance. In all the examples below, apart from Finland, trust is still considerably lower than in 2015.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 9\n\nPROPORTION THAT TRUSTS MOST NEWS MOST OF THE TIME – ALL MARKETS\n\n**Q6\\_2016\\_1.** Thinking about news in general, do you agree or disagree with the following statements? – I think you can trust most news most of the time. _Base: Total sample in each market (n ≈ 2000)._\n\n0% 25% 50% 75% 100%\n\nCanada USA **North America**\n\n42\n\n26\n\nBrazil Peru Chile Colombia Mexico Argentina\n\n**Latin America**\n\n48\n\n41\n\n38 37 37 35\n\nThailand Japan Singapore Australia Hong Kong India Indonesia Philippines Malaysia South Korea Taiwan\n\n**Asia-Pacific** 53\n\n44 43 41 41 41\n\n39 37 36\n\n30\n\n27\n\nSouth Africa Nigeria Kenya\n\n**Africa** 61\n\n58 57\n\nPoland Bulgaria Czech Republic Romania Hungary Slovakia\n\n**Eastern Europe**\n\n42\n\n35 34 33\n\n27 26\n\nPortugal Croatia Turkey Italy Spain Greece\n\n**Southern Europe**\n\n61\n\n38 36 35\n\n32\n\n27\n\nNetherlands Belgium Germany Switzerland Austria France\n\n**Western Europe**\n\n56\n\n51 50\n\n46\n\n41\n\n29\n\nFinland **Northern Europe** Denmark Norway Ireland Sweden UK\n\n69\n\n58 56\n\n52\n\n50\n\n34\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 10\n\nPROPORTION THAT TRUSTS MOST NEWS MOST OF THE TIME (2015–22) – SELECTED MARKETS\n\n**Q6\\_2016\\_1.** Thinking about news in general, do you agree or disagree with the following statements? – I think you can trust most news most of the time. _Base: Total 2015–22 samples (n≈2000)._\n\nFinland Germany UK France USA\n\nTrump elected\n\nPost Brexit fall\n\nGeneral election\n\nGilet Jaunes protests\n\nCorona bump\n\n60%\n\n38%\n\n32%\n\n51%\n\n68%\n\n50%\n\n29% 26%\n\n34%\n\n69%\n\n0%\n\n25%\n\n50%\n\n75%\n\n100%\n\n2022 2021 2020 2019 2018 2017 2015 2016\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 15","pages":[{"line_numbers":[],"markdown":"**TRUST FALLS BACK AFTER COVID-19 BUMPS**\n\nThis year we find lower levels of overall trust in 21 of our 46 markets, partly reversing the gains made at the height of the Coronavirus pandemic. Eighteen markets are at a similar level, with just seven on an upward path. The average level of trust (42%) is also a little lower than last year and trust in individual news brands is on a downward trend in most countries.\n\nFinland remains the country with the highest levels of overall trust (69%) – up four points on last year and 13 points on 2020. At the other end of the scale, news trust in the USA (26%) has fallen by a further three percentage points and remains the lowest level in our survey along with Slovakia (26%). Notable changes in Europe include falls in Romania (-9), Croatia (-7), Poland (-6), Switzerland (-5), Austria (-5), Greece (-5), Italy (-5), and Spain (-4). In Latin America, trust is down in Brazil (-6) and Colombia (-3), but level or slightly up elsewhere. It is a mixed picture in Africa, with a fall in Kenya (-4) but strong growth in South Africa (+9) and Nigeria (+4). Finally, in Asia, trust has risen in the Philippines (+5) and Japan (+2) but is down in Malaysia (-5) and Taiwan (-4).\n\nIn some regions we find a widening gap between markets with the highest levels of trust and those with the lowest. We also find that those markets with the most trust – such as Finland – also tend to have higher interest in news and lower levels of active news avoidance. By contrast, low trust countries, such as the United States, France, the United Kingdom, and Slovakia, see some of the highest levels of selective avoidance, as well as news disconnection, with declining interest in news.\n\nIt is important to put this year’s trust changes in context. In most cases the headline trust level is still higher than it was before the Coronavirus pandemic reinforced the importance of reliable media for many people. On the other hand, the chart below also reminds us that, in many countries, trust had been on a downward trajectory for some time – partly driven by a series of polarising events like Brexit in the UK, the turmoil of the Trump years in the United States, and the Gilets Jaunes protests in France. In this context, last year’s COVID-19 optimism looks more like a short- term rally rather than any kind of longer-term renaissance. In all the examples below, apart from Finland, trust is still considerably lower than in 2015.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 9\n\nPROPORTION THAT TRUSTS MOST NEWS MOST OF THE TIME – ALL MARKETS\n\n**Q6\\_2016\\_1.** Thinking about news in general, do you agree or disagree with the following statements? – I think you can trust most news most of the time. _Base: Total sample in each market (n ≈ 2000)._\n\n0% 25% 50% 75% 100%\n\nCanada USA **North America**\n\n42\n\n26\n\nBrazil Peru Chile Colombia Mexico Argentina\n\n**Latin America**\n\n48\n\n41\n\n38 37 37 35\n\nThailand Japan Singapore Australia Hong Kong India Indonesia Philippines Malaysia South Korea Taiwan\n\n**Asia-Pacific** 53\n\n44 43 41 41 41\n\n39 37 36\n\n30\n\n27\n\nSouth Africa Nigeria Kenya\n\n**Africa** 61\n\n58 57\n\nPoland Bulgaria Czech Republic Romania Hungary Slovakia\n\n**Eastern Europe**\n\n42\n\n35 34 33\n\n27 26\n\nPortugal Croatia Turkey Italy Spain Greece\n\n**Southern Europe**\n\n61\n\n38 36 35\n\n32\n\n27\n\nNetherlands Belgium Germany Switzerland Austria France\n\n**Western Europe**\n\n56\n\n51 50\n\n46\n\n41\n\n29\n\nFinland **Northern Europe** Denmark Norway Ireland Sweden UK\n\n69\n\n58 56\n\n52\n\n50\n\n34\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 10\n\nPROPORTION THAT TRUSTS MOST NEWS MOST OF THE TIME (2015–22) – SELECTED MARKETS\n\n**Q6\\_2016\\_1.** Thinking about news in general, do you agree or disagree with the following statements? – I think you can trust most news most of the time. _Base: Total 2015–22 samples (n≈2000)._\n\nFinland Germany UK France USA\n\nTrump elected\n\nPost Brexit fall\n\nGeneral election\n\nGilet Jaunes protests\n\nCorona bump\n\n60%\n\n38%\n\n32%\n\n51%\n\n68%\n\n50%\n\n29% 26%\n\n34%\n\n69%\n\n0%\n\n25%\n\n50%\n\n75%\n\n100%\n\n2022 2021 2020 2019 2018 2017 2015 2016\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 15","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**TRUST FALLS BACK AFTER COVID-19 BUMPS**\n\nThis year we find lower levels of overall trust in 21 of our 46 markets, partly reversing the gains made at the height of the Coronavirus pandemic. Eighteen markets are at a similar level, with just seven on an upward path. The average level of trust (42%) is also a little lower than last year and trust in individual news brands is on a downward trend in most countries.\n\nFinland remains the country with the highest levels of overall trust (69%) – up four points on last year and 13 points on 2020. At the other end of the scale, news trust in the USA (26%) has fallen by a further three percentage points and remains the lowest level in our survey along with Slovakia (26%). Notable changes in Europe include falls in Romania (-9), Croatia (-7), Poland (-6), Switzerland (-5), Austria (-5), Greece (-5), Italy (-5), and Spain (-4). In Latin America, trust is down in Brazil (-6) and Colombia (-3), but level or slightly up elsewhere. It is a mixed picture in Africa, with a fall in Kenya (-4) but strong growth in South Africa (+9) and Nigeria (+4). Finally, in Asia, trust has risen in the Philippines (+5) and Japan (+2) but is down in Malaysia (-5) and Taiwan (-4).\n\nIn some regions we find a widening gap between markets with the highest levels of trust and those with the lowest. We also find that those markets with the most trust – such as Finland – also tend to have higher interest in news and lower levels of active news avoidance. By contrast, low trust countries, such as the United States, France, the United Kingdom, and Slovakia, see some of the highest levels of selective avoidance, as well as news disconnection, with declining interest in news.\n\nIt is important to put this year’s trust changes in context. In most cases the headline trust level is still higher than it was before the Coronavirus pandemic reinforced the importance of reliable media for many people. On the other hand, the chart below also reminds us that, in many countries, trust had been on a downward trajectory for some time – partly driven by a series of polarising events like Brexit in the UK, the turmoil of the Trump years in the United States, and the Gilets Jaunes protests in France. In this context, last year’s COVID-19 optimism looks more like a short- term rally rather than any kind of longer-term renaissance. In all the examples below, apart from Finland, trust is still considerably lower than in 2015.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 9\n\nPROPORTION THAT TRUSTS MOST NEWS MOST OF THE TIME – ALL MARKETS\n\n**Q6\\_2016\\_1.** Thinking about news in general, do you agree or disagree with the following statements? – I think you can trust most news most of the time. _Base: Total sample in each market (n ≈ 2000)._\n\n0% 25% 50% 75% 100%\n\nCanada USA **North America**\n\n42\n\n26\n\nBrazil Peru Chile Colombia Mexico Argentina\n\n**Latin America**\n\n48\n\n41\n\n38 37 37 35\n\nThailand Japan Singapore Australia Hong Kong India Indonesia Philippines Malaysia South Korea Taiwan\n\n**Asia-Pacific** 53\n\n44 43 41 41 41\n\n39 37 36\n\n30\n\n27\n\nSouth Africa Nigeria Kenya\n\n**Africa** 61\n\n58 57\n\nPoland Bulgaria Czech Republic Romania Hungary Slovakia\n\n**Eastern Europe**\n\n42\n\n35 34 33\n\n27 26\n\nPortugal Croatia Turkey Italy Spain Greece\n\n**Southern Europe**\n\n61\n\n38 36 35\n\n32\n\n27\n\nNetherlands Belgium Germany Switzerland Austria France\n\n**Western Europe**\n\n56\n\n51 50\n\n46\n\n41\n\n29\n\nFinland **Northern Europe** Denmark Norway Ireland Sweden UK\n\n69\n\n58 56\n\n52\n\n50\n\n34\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 10\n\nPROPORTION THAT TRUSTS MOST NEWS MOST OF THE TIME (2015–22) – SELECTED MARKETS\n\n**Q6\\_2016\\_1.** Thinking about news in general, do you agree or disagree with the following statements? – I think you can trust most news most of the time. _Base: Total 2015–22 samples (n≈2000)._\n\nFinland Germany UK France USA\n\nTrump elected\n\nPost Brexit fall\n\nGeneral election\n\nGilet Jaunes protests\n\nCorona bump\n\n60%\n\n38%\n\n32%\n\n51%\n\n68%\n\n50%\n\n29% 26%\n\n34%\n\n69%\n\n0%\n\n25%\n\n50%\n\n75%\n\n100%\n\n2022 2021 2020 2019 2018 2017 2015 2016\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 15","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Digital_News-Report_2022_p15.pdf"]},"response":{"errors":[],"extract_id":"extract_29d96c29b0f573f784ac6226f8fde095","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**TRUST FALLS BACK AFTER COVID-19 BUMPS**\n\nThis year we find lower levels of overall trust in 21 of our 46 markets, partly reversing the gains made at the height of the Coronavirus pandemic. Eighteen markets are at a similar level, with just seven on an upward path. The average level of trust (42%) is also a little lower than last year and trust in individual news brands is on a downward trend in most countries.\n\nFinland remains the country with the highest levels of overall trust (69%) – up four points on last year and 13 points on 2020. At the other end of the scale, news trust in the USA (26%) has fallen by a further three percentage points and remains the lowest level in our survey along with Slovakia (26%). Notable changes in Europe include falls in Romania (-9), Croatia (-7), Poland (-6), Switzerland (-5), Austria (-5), Greece (-5), Italy (-5), and Spain (-4). In Latin America, trust is down in Brazil (-6) and Colombia (-3), but level or slightly up elsewhere. It is a mixed picture in Africa, with a fall in Kenya (-4) but strong growth in South Africa (+9) and Nigeria (+4). Finally, in Asia, trust has risen in the Philippines (+5) and Japan (+2) but is down in Malaysia (-5) and Taiwan (-4).\n\nIn some regions we find a widening gap between markets with the highest levels of trust and those with the lowest. We also find that those markets with the most trust – such as Finland – also tend to have higher interest in news and lower levels of active news avoidance. By contrast, low trust countries, such as the United States, France, the United Kingdom, and Slovakia, see some of the highest levels of selective avoidance, as well as news disconnection, with declining interest in news.\n\nIt is important to put this year’s trust changes in context. In most cases the headline trust level is still higher than it was before the Coronavirus pandemic reinforced the importance of reliable media for many people. On the other hand, the chart below also reminds us that, in many countries, trust had been on a downward trajectory for some time – partly driven by a series of polarising events like Brexit in the UK, the turmoil of the Trump years in the United States, and the Gilets Jaunes protests in France. In this context, last year’s COVID-19 optimism looks more like a short- term rally rather than any kind of longer-term renaissance. In all the examples below, apart from Finland, trust is still considerably lower than in 2015.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 9\n\nPROPORTION THAT TRUSTS MOST NEWS MOST OF THE TIME – ALL MARKETS\n\n**Q6\\_2016\\_1.** Thinking about news in general, do you agree or disagree with the following statements? – I think you can trust most news most of the time. _Base: Total sample in each market (n ≈ 2000)._\n\n0% 25% 50% 75% 100%\n\nCanada USA **North America**\n\n42\n\n26\n\nBrazil Peru Chile Colombia Mexico Argentina\n\n**Latin America**\n\n48\n\n41\n\n38 37 37 35\n\nThailand Japan Singapore Australia Hong Kong India Indonesia Philippines Malaysia South Korea Taiwan\n\n**Asia-Pacific** 53\n\n44 43 41 41 41\n\n39 37 36\n\n30\n\n27\n\nSouth Africa Nigeria Kenya\n\n**Africa** 61\n\n58 57\n\nPoland Bulgaria Czech Republic Romania Hungary Slovakia\n\n**Eastern Europe**\n\n42\n\n35 34 33\n\n27 26\n\nPortugal Croatia Turkey Italy Spain Greece\n\n**Southern Europe**\n\n61\n\n38 36 35\n\n32\n\n27\n\nNetherlands Belgium Germany Switzerland Austria France\n\n**Western Europe**\n\n56\n\n51 50\n\n46\n\n41\n\n29\n\nFinland **Northern Europe** Denmark Norway Ireland Sweden UK\n\n69\n\n58 56\n\n52\n\n50\n\n34\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 10\n\nPROPORTION THAT TRUSTS MOST NEWS MOST OF THE TIME (2015–22) – SELECTED MARKETS\n\n**Q6\\_2016\\_1.** Thinking about news in general, do you agree or disagree with the following statements? – I think you can trust most news most of the time. _Base: Total 2015–22 samples (n≈2000)._\n\nFinland Germany UK France USA\n\nTrump elected\n\nPost Brexit fall\n\nGeneral election\n\nGilet Jaunes protests\n\nCorona bump\n\n60%\n\n38%\n\n32%\n\n51%\n\n68%\n\n50%\n\n29% 26%\n\n34%\n\n69%\n\n0%\n\n25%\n\n50%\n\n75%\n\n100%\n\n2022 2021 2020 2019 2018 2017 2015 2016\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 15"],"full_content":"**TRUST FALLS BACK AFTER COVID-19 BUMPS**\n\nThis year we find lower levels of overall trust in 21 of our 46 markets, partly reversing the gains made at the height of the Coronavirus pandemic. Eighteen markets are at a similar level, with just seven on an upward path. The average level of trust (42%) is also a little lower than last year and trust in individual news brands is on a downward trend in most countries.\n\nFinland remains the country with the highest levels of overall trust (69%) – up four points on last year and 13 points on 2020. At the other end of the scale, news trust in the USA (26%) has fallen by a further three percentage points and remains the lowest level in our survey along with Slovakia (26%). Notable changes in Europe include falls in Romania (-9), Croatia (-7), Poland (-6), Switzerland (-5), Austria (-5), Greece (-5), Italy (-5), and Spain (-4). In Latin America, trust is down in Brazil (-6) and Colombia (-3), but level or slightly up elsewhere. It is a mixed picture in Africa, with a fall in Kenya (-4) but strong growth in South Africa (+9) and Nigeria (+4). Finally, in Asia, trust has risen in the Philippines (+5) and Japan (+2) but is down in Malaysia (-5) and Taiwan (-4).\n\nIn some regions we find a widening gap between markets with the highest levels of trust and those with the lowest. We also find that those markets with the most trust – such as Finland – also tend to have higher interest in news and lower levels of active news avoidance. By contrast, low trust countries, such as the United States, France, the United Kingdom, and Slovakia, see some of the highest levels of selective avoidance, as well as news disconnection, with declining interest in news.\n\nIt is important to put this year’s trust changes in context. In most cases the headline trust level is still higher than it was before the Coronavirus pandemic reinforced the importance of reliable media for many people. On the other hand, the chart below also reminds us that, in many countries, trust had been on a downward trajectory for some time – partly driven by a series of polarising events like Brexit in the UK, the turmoil of the Trump years in the United States, and the Gilets Jaunes protests in France. In this context, last year’s COVID-19 optimism looks more like a short- term rally rather than any kind of longer-term renaissance. In all the examples below, apart from Finland, trust is still considerably lower than in 2015.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 9\n\nPROPORTION THAT TRUSTS MOST NEWS MOST OF THE TIME – ALL MARKETS\n\n**Q6\\_2016\\_1.** Thinking about news in general, do you agree or disagree with the following statements? – I think you can trust most news most of the time. _Base: Total sample in each market (n ≈ 2000)._\n\n0% 25% 50% 75% 100%\n\nCanada USA **North America**\n\n42\n\n26\n\nBrazil Peru Chile Colombia Mexico Argentina\n\n**Latin America**\n\n48\n\n41\n\n38 37 37 35\n\nThailand Japan Singapore Australia Hong Kong India Indonesia Philippines Malaysia South Korea Taiwan\n\n**Asia-Pacific** 53\n\n44 43 41 41 41\n\n39 37 36\n\n30\n\n27\n\nSouth Africa Nigeria Kenya\n\n**Africa** 61\n\n58 57\n\nPoland Bulgaria Czech Republic Romania Hungary Slovakia\n\n**Eastern Europe**\n\n42\n\n35 34 33\n\n27 26\n\nPortugal Croatia Turkey Italy Spain Greece\n\n**Southern Europe**\n\n61\n\n38 36 35\n\n32\n\n27\n\nNetherlands Belgium Germany Switzerland Austria France\n\n**Western Europe**\n\n56\n\n51 50\n\n46\n\n41\n\n29\n\nFinland **Northern Europe** Denmark Norway Ireland Sweden UK\n\n69\n\n58 56\n\n52\n\n50\n\n34\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 10\n\nPROPORTION THAT TRUSTS MOST NEWS MOST OF THE TIME (2015–22) – SELECTED MARKETS\n\n**Q6\\_2016\\_1.** Thinking about news in general, do you agree or disagree with the following statements? – I think you can trust most news most of the time. _Base: Total 2015–22 samples (n≈2000)._\n\nFinland Germany UK France USA\n\nTrump elected\n\nPost Brexit fall\n\nGeneral election\n\nGilet Jaunes protests\n\nCorona bump\n\n60%\n\n38%\n\n32%\n\n51%\n\n68%\n\n50%\n\n29% 26%\n\n34%\n\n69%\n\n0%\n\n25%\n\n50%\n\n75%\n\n100%\n\n2022 2021 2020 2019 2018 2017 2015 2016\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 15","publish_date":null,"title":"Graph 9 Graph 10","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Digital_News-Report_2022_p15.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Digital_News-Report_2022_p15.pdf"},"request":{"example_id":"chart/Digital_News-Report_2022_p15","product_type":"parse","source_file_path":"Digital_News-Report_2022_p15.pdf"},"started_at":"2026-11-04T19:31:14.829186"} +{"completed_at":"2026-11-04T19:46:06.159041","latency_in_ms":1777,"output":{"example_id":"chart/Digital_News-Report_2022_p17","grounded_pages":[],"layout_pages":[],"markdown":"**TRUST IN NEWS BRANDS: THE PARTICULAR CASE OF PUBLIC SERVICE MEDIA**\n\nUnderpinning news trust in many Northern and Western European countries, as well as Canada and Australia, we find a number of public service broadcasting organisations with a strong track record of independence. These generally appear at the top of our brand trust lists in each country and are often the first port of call for audiences when looking for reliable news around important stories such as COVID-19 and the Ukraine conflict. These institutions have a very different reputation to public and state broadcasters in Southern and Eastern Europe that often take a more partisan editorial line.\n\nBut independent public media are under increasing pressure in a number of countries, with attacks on funding, questions about impartiality, and challenges in reaching younger audiences who are increasingly turning to digital and social media. The BBC faces another round of cuts after a difficult licence settlement while DR, the Danish public broadcaster, recently cut three linear channels as part of a wider restructuring after a narrow parliamentary majority imposed a major cut in funding. Against this background, it is interesting to see how well trust is holding up, with some very notable exceptions. Nordic PSBs as well as those of other smaller countries like Ireland and Portugal have retained or even increased trust since 2018, but it has been a very different story in the UK, Australia, and Canada, with significant falls for the main public broadcasters.\n\nThe BBC, which remains the most widely used news source in the UK, reported record audiences in the first weeks of the Ukraine crisis, ³ but it has also come under intense criticism for its reporting of a range of more divisive issues – notably Brexit but also immigration, race, gender identity, and attitudes to COVID-19 vaccinations. In many cases, these criticisms of the BBC have been amplified on social media, with senior correspondents exposed to physical attacks along with personal abuse on social media. ⁴\n\nTrust in BBC News has fallen 20 percentage points in the last five years, from 75% to 55%. Equally telling is the proportion who say they distrust the BBC, which has grown from 11% to 26% (see next chart). The majority of these are from the political right, echoing criticism from Boris Johnson’s government about an alleged anti- Brexit and liberal bias, but we also find that low trust in the BBC also comes from those who are less interested in news altogether. It is important to note that other big brands in the UK (e.g. the _Guardian_ and the _Mail_ ) have been affected by growing levels of distrust, though not as severely. Declining trust is a particular challenge for public media organisations, as they try to fulfil their mission to appeal to all audiences.\n\nThe continued audience success of some European public media makes them a particular target for those who want to influence the debates on politics and wider issues around culture. Our survey shows that journalists from these organisations are often first to be recognised by the public. The outgoing BBC political editor Laura Kuenssberg is by far the most well-known journalist in the UK, according to respondents to our survey, even when it comes to digital news. Broadcasters with a commitment to impartiality, including the BBC, make up 62% of all mentions. Journalists from public media also feature strongly in data from Finland, France, and Germany. But it is a very different story in the United States, where partisan cable TV hosts Tucker Carlson, Rachel Maddow, and Sean Hannity attract the most attention.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 13\n\n0% 25% 50% 75% 100%\n\n84\n\n81\n\n80\n\n74\n\n66\n\n59\n\n55\n\n76\n\n75\n\n67\n\n66\n\n62\n\n57\n\n53\n\n86\n\n83\n\n82\n\n80\n\n78\n\n76\n\n75\n\n71\n\n71\n\n69\n\n68\n\n66\n\n57\n\n55\n\n2022\n\n2018\n\nFrance Télévisions\n\nNHK (Japan)\n\nORF (Austria)\n\nZDF (Germany)\n\nARD (Germany)\n\nSR (Sweden)\n\nSVT (Sweden)\n\nBBC (UK)\n\nCBC (Canada)\n\nABC (Australia)\n\nRTÉ (Ireland)\n\nNRK (Norway)\n\nDR (Denmark)\n\nYle (Finland)\n\nPROPORTION THAT TRUSTS SELECTED PUBLIC SERVICE MEDIA BRANDS 2018 AND 2022 – SELECTED MARKETS\n\n**Q6\\_2018\\_trust.** How trustworthy would you say news from the following brands is? Please use the scale below, where 0 is ‘not at all trustworthy’ and 10 is ‘completely trustworthy’. _Base: All in 2018 and 2022 that had heard of each brand (n ≈ 2000). Note: Showing scores 6–10 on 10-point scale._\n\n2018\n\n2022\n\nBiggest drops Biggest drops\n\n3 During the first week of the conflict, the BBC estimated 280 million people used its news output in the UK and around the world. https://www.independent.co.uk/news/uk/ bbc-ukraine-tim-davie-nadine-dorries-bbc-news-b2047209.html\n\n4 https://www.theguardian.com/uk-news/2022/feb/17/bbcs-sarah-smith-relieved-to-escape-scottish-bile-and-hatred\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 14\n\nPROPORTION THAT TRUSTS THE BBC (2018–22) – UK\n\n**Q6\\_2018\\_trust.** How trustworthy would you say news from the following brands is? Please use the scale below, where 0 is ‘not at all trustworthy’ and 10 is ‘completely trustworthy’. _Base: All in 2018–22 that had heard of the BBC (n ≈ 2000). Note: Trust = % scored 6–10 on 10-point scale, Don’t trust = 0–4, Neither = 5._\n\n0%\n\n25%\n\n50%\n\n75%\n\n100% 11 16 17 21 26\n\n75 73\n\n64 62 55\n\n2022 2021 2020 2019 2018\n\nConservative Lib Dem Other\n\nSNP Labour\n\nTrust Don’t trust Neither\n\nThose that **don’t trust** the BBC are **more likely to be men** . Tend to be less interested in news and have lower levels of education.\n\nMore likely to **only use social media** as a source of news.\n\nOver half of this group **voted Conservative** at the last election (2019), **two thirds voted for Brexit** .\n\n53%\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 17","pages":[{"line_numbers":[],"markdown":"**TRUST IN NEWS BRANDS: THE PARTICULAR CASE OF PUBLIC SERVICE MEDIA**\n\nUnderpinning news trust in many Northern and Western European countries, as well as Canada and Australia, we find a number of public service broadcasting organisations with a strong track record of independence. These generally appear at the top of our brand trust lists in each country and are often the first port of call for audiences when looking for reliable news around important stories such as COVID-19 and the Ukraine conflict. These institutions have a very different reputation to public and state broadcasters in Southern and Eastern Europe that often take a more partisan editorial line.\n\nBut independent public media are under increasing pressure in a number of countries, with attacks on funding, questions about impartiality, and challenges in reaching younger audiences who are increasingly turning to digital and social media. The BBC faces another round of cuts after a difficult licence settlement while DR, the Danish public broadcaster, recently cut three linear channels as part of a wider restructuring after a narrow parliamentary majority imposed a major cut in funding. Against this background, it is interesting to see how well trust is holding up, with some very notable exceptions. Nordic PSBs as well as those of other smaller countries like Ireland and Portugal have retained or even increased trust since 2018, but it has been a very different story in the UK, Australia, and Canada, with significant falls for the main public broadcasters.\n\nThe BBC, which remains the most widely used news source in the UK, reported record audiences in the first weeks of the Ukraine crisis, ³ but it has also come under intense criticism for its reporting of a range of more divisive issues – notably Brexit but also immigration, race, gender identity, and attitudes to COVID-19 vaccinations. In many cases, these criticisms of the BBC have been amplified on social media, with senior correspondents exposed to physical attacks along with personal abuse on social media. ⁴\n\nTrust in BBC News has fallen 20 percentage points in the last five years, from 75% to 55%. Equally telling is the proportion who say they distrust the BBC, which has grown from 11% to 26% (see next chart). The majority of these are from the political right, echoing criticism from Boris Johnson’s government about an alleged anti- Brexit and liberal bias, but we also find that low trust in the BBC also comes from those who are less interested in news altogether. It is important to note that other big brands in the UK (e.g. the _Guardian_ and the _Mail_ ) have been affected by growing levels of distrust, though not as severely. Declining trust is a particular challenge for public media organisations, as they try to fulfil their mission to appeal to all audiences.\n\nThe continued audience success of some European public media makes them a particular target for those who want to influence the debates on politics and wider issues around culture. Our survey shows that journalists from these organisations are often first to be recognised by the public. The outgoing BBC political editor Laura Kuenssberg is by far the most well-known journalist in the UK, according to respondents to our survey, even when it comes to digital news. Broadcasters with a commitment to impartiality, including the BBC, make up 62% of all mentions. Journalists from public media also feature strongly in data from Finland, France, and Germany. But it is a very different story in the United States, where partisan cable TV hosts Tucker Carlson, Rachel Maddow, and Sean Hannity attract the most attention.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 13\n\n0% 25% 50% 75% 100%\n\n84\n\n81\n\n80\n\n74\n\n66\n\n59\n\n55\n\n76\n\n75\n\n67\n\n66\n\n62\n\n57\n\n53\n\n86\n\n83\n\n82\n\n80\n\n78\n\n76\n\n75\n\n71\n\n71\n\n69\n\n68\n\n66\n\n57\n\n55\n\n2022\n\n2018\n\nFrance Télévisions\n\nNHK (Japan)\n\nORF (Austria)\n\nZDF (Germany)\n\nARD (Germany)\n\nSR (Sweden)\n\nSVT (Sweden)\n\nBBC (UK)\n\nCBC (Canada)\n\nABC (Australia)\n\nRTÉ (Ireland)\n\nNRK (Norway)\n\nDR (Denmark)\n\nYle (Finland)\n\nPROPORTION THAT TRUSTS SELECTED PUBLIC SERVICE MEDIA BRANDS 2018 AND 2022 – SELECTED MARKETS\n\n**Q6\\_2018\\_trust.** How trustworthy would you say news from the following brands is? Please use the scale below, where 0 is ‘not at all trustworthy’ and 10 is ‘completely trustworthy’. _Base: All in 2018 and 2022 that had heard of each brand (n ≈ 2000). Note: Showing scores 6–10 on 10-point scale._\n\n2018\n\n2022\n\nBiggest drops Biggest drops\n\n3 During the first week of the conflict, the BBC estimated 280 million people used its news output in the UK and around the world. https://www.independent.co.uk/news/uk/ bbc-ukraine-tim-davie-nadine-dorries-bbc-news-b2047209.html\n\n4 https://www.theguardian.com/uk-news/2022/feb/17/bbcs-sarah-smith-relieved-to-escape-scottish-bile-and-hatred\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 14\n\nPROPORTION THAT TRUSTS THE BBC (2018–22) – UK\n\n**Q6\\_2018\\_trust.** How trustworthy would you say news from the following brands is? Please use the scale below, where 0 is ‘not at all trustworthy’ and 10 is ‘completely trustworthy’. _Base: All in 2018–22 that had heard of the BBC (n ≈ 2000). Note: Trust = % scored 6–10 on 10-point scale, Don’t trust = 0–4, Neither = 5._\n\n0%\n\n25%\n\n50%\n\n75%\n\n100% 11 16 17 21 26\n\n75 73\n\n64 62 55\n\n2022 2021 2020 2019 2018\n\nConservative Lib Dem Other\n\nSNP Labour\n\nTrust Don’t trust Neither\n\nThose that **don’t trust** the BBC are **more likely to be men** . Tend to be less interested in news and have lower levels of education.\n\nMore likely to **only use social media** as a source of news.\n\nOver half of this group **voted Conservative** at the last election (2019), **two thirds voted for Brexit** .\n\n53%\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 17","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**TRUST IN NEWS BRANDS: THE PARTICULAR CASE OF PUBLIC SERVICE MEDIA**\n\nUnderpinning news trust in many Northern and Western European countries, as well as Canada and Australia, we find a number of public service broadcasting organisations with a strong track record of independence. These generally appear at the top of our brand trust lists in each country and are often the first port of call for audiences when looking for reliable news around important stories such as COVID-19 and the Ukraine conflict. These institutions have a very different reputation to public and state broadcasters in Southern and Eastern Europe that often take a more partisan editorial line.\n\nBut independent public media are under increasing pressure in a number of countries, with attacks on funding, questions about impartiality, and challenges in reaching younger audiences who are increasingly turning to digital and social media. The BBC faces another round of cuts after a difficult licence settlement while DR, the Danish public broadcaster, recently cut three linear channels as part of a wider restructuring after a narrow parliamentary majority imposed a major cut in funding. Against this background, it is interesting to see how well trust is holding up, with some very notable exceptions. Nordic PSBs as well as those of other smaller countries like Ireland and Portugal have retained or even increased trust since 2018, but it has been a very different story in the UK, Australia, and Canada, with significant falls for the main public broadcasters.\n\nThe BBC, which remains the most widely used news source in the UK, reported record audiences in the first weeks of the Ukraine crisis, ³ but it has also come under intense criticism for its reporting of a range of more divisive issues – notably Brexit but also immigration, race, gender identity, and attitudes to COVID-19 vaccinations. In many cases, these criticisms of the BBC have been amplified on social media, with senior correspondents exposed to physical attacks along with personal abuse on social media. ⁴\n\nTrust in BBC News has fallen 20 percentage points in the last five years, from 75% to 55%. Equally telling is the proportion who say they distrust the BBC, which has grown from 11% to 26% (see next chart). The majority of these are from the political right, echoing criticism from Boris Johnson’s government about an alleged anti- Brexit and liberal bias, but we also find that low trust in the BBC also comes from those who are less interested in news altogether. It is important to note that other big brands in the UK (e.g. the _Guardian_ and the _Mail_ ) have been affected by growing levels of distrust, though not as severely. Declining trust is a particular challenge for public media organisations, as they try to fulfil their mission to appeal to all audiences.\n\nThe continued audience success of some European public media makes them a particular target for those who want to influence the debates on politics and wider issues around culture. Our survey shows that journalists from these organisations are often first to be recognised by the public. The outgoing BBC political editor Laura Kuenssberg is by far the most well-known journalist in the UK, according to respondents to our survey, even when it comes to digital news. Broadcasters with a commitment to impartiality, including the BBC, make up 62% of all mentions. Journalists from public media also feature strongly in data from Finland, France, and Germany. But it is a very different story in the United States, where partisan cable TV hosts Tucker Carlson, Rachel Maddow, and Sean Hannity attract the most attention.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 13\n\n0% 25% 50% 75% 100%\n\n84\n\n81\n\n80\n\n74\n\n66\n\n59\n\n55\n\n76\n\n75\n\n67\n\n66\n\n62\n\n57\n\n53\n\n86\n\n83\n\n82\n\n80\n\n78\n\n76\n\n75\n\n71\n\n71\n\n69\n\n68\n\n66\n\n57\n\n55\n\n2022\n\n2018\n\nFrance Télévisions\n\nNHK (Japan)\n\nORF (Austria)\n\nZDF (Germany)\n\nARD (Germany)\n\nSR (Sweden)\n\nSVT (Sweden)\n\nBBC (UK)\n\nCBC (Canada)\n\nABC (Australia)\n\nRTÉ (Ireland)\n\nNRK (Norway)\n\nDR (Denmark)\n\nYle (Finland)\n\nPROPORTION THAT TRUSTS SELECTED PUBLIC SERVICE MEDIA BRANDS 2018 AND 2022 – SELECTED MARKETS\n\n**Q6\\_2018\\_trust.** How trustworthy would you say news from the following brands is? Please use the scale below, where 0 is ‘not at all trustworthy’ and 10 is ‘completely trustworthy’. _Base: All in 2018 and 2022 that had heard of each brand (n ≈ 2000). Note: Showing scores 6–10 on 10-point scale._\n\n2018\n\n2022\n\nBiggest drops Biggest drops\n\n3 During the first week of the conflict, the BBC estimated 280 million people used its news output in the UK and around the world. https://www.independent.co.uk/news/uk/ bbc-ukraine-tim-davie-nadine-dorries-bbc-news-b2047209.html\n\n4 https://www.theguardian.com/uk-news/2022/feb/17/bbcs-sarah-smith-relieved-to-escape-scottish-bile-and-hatred\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 14\n\nPROPORTION THAT TRUSTS THE BBC (2018–22) – UK\n\n**Q6\\_2018\\_trust.** How trustworthy would you say news from the following brands is? Please use the scale below, where 0 is ‘not at all trustworthy’ and 10 is ‘completely trustworthy’. _Base: All in 2018–22 that had heard of the BBC (n ≈ 2000). Note: Trust = % scored 6–10 on 10-point scale, Don’t trust = 0–4, Neither = 5._\n\n0%\n\n25%\n\n50%\n\n75%\n\n100% 11 16 17 21 26\n\n75 73\n\n64 62 55\n\n2022 2021 2020 2019 2018\n\nConservative Lib Dem Other\n\nSNP Labour\n\nTrust Don’t trust Neither\n\nThose that **don’t trust** the BBC are **more likely to be men** . Tend to be less interested in news and have lower levels of education.\n\nMore likely to **only use social media** as a source of news.\n\nOver half of this group **voted Conservative** at the last election (2019), **two thirds voted for Brexit** .\n\n53%\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 17","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Digital_News-Report_2022_p17.pdf"]},"response":{"errors":[],"extract_id":"extract_edb57cd3cac7d9230b2276f3fdcaa0a3","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**TRUST IN NEWS BRANDS: THE PARTICULAR CASE OF PUBLIC SERVICE MEDIA**\n\nUnderpinning news trust in many Northern and Western European countries, as well as Canada and Australia, we find a number of public service broadcasting organisations with a strong track record of independence. These generally appear at the top of our brand trust lists in each country and are often the first port of call for audiences when looking for reliable news around important stories such as COVID-19 and the Ukraine conflict. These institutions have a very different reputation to public and state broadcasters in Southern and Eastern Europe that often take a more partisan editorial line.\n\nBut independent public media are under increasing pressure in a number of countries, with attacks on funding, questions about impartiality, and challenges in reaching younger audiences who are increasingly turning to digital and social media. The BBC faces another round of cuts after a difficult licence settlement while DR, the Danish public broadcaster, recently cut three linear channels as part of a wider restructuring after a narrow parliamentary majority imposed a major cut in funding. Against this background, it is interesting to see how well trust is holding up, with some very notable exceptions. Nordic PSBs as well as those of other smaller countries like Ireland and Portugal have retained or even increased trust since 2018, but it has been a very different story in the UK, Australia, and Canada, with significant falls for the main public broadcasters.\n\nThe BBC, which remains the most widely used news source in the UK, reported record audiences in the first weeks of the Ukraine crisis, ³ but it has also come under intense criticism for its reporting of a range of more divisive issues – notably Brexit but also immigration, race, gender identity, and attitudes to COVID-19 vaccinations. In many cases, these criticisms of the BBC have been amplified on social media, with senior correspondents exposed to physical attacks along with personal abuse on social media. ⁴\n\nTrust in BBC News has fallen 20 percentage points in the last five years, from 75% to 55%. Equally telling is the proportion who say they distrust the BBC, which has grown from 11% to 26% (see next chart). The majority of these are from the political right, echoing criticism from Boris Johnson’s government about an alleged anti- Brexit and liberal bias, but we also find that low trust in the BBC also comes from those who are less interested in news altogether. It is important to note that other big brands in the UK (e.g. the _Guardian_ and the _Mail_ ) have been affected by growing levels of distrust, though not as severely. Declining trust is a particular challenge for public media organisations, as they try to fulfil their mission to appeal to all audiences.\n\nThe continued audience success of some European public media makes them a particular target for those who want to influence the debates on politics and wider issues around culture. Our survey shows that journalists from these organisations are often first to be recognised by the public. The outgoing BBC political editor Laura Kuenssberg is by far the most well-known journalist in the UK, according to respondents to our survey, even when it comes to digital news. Broadcasters with a commitment to impartiality, including the BBC, make up 62% of all mentions. Journalists from public media also feature strongly in data from Finland, France, and Germany. But it is a very different story in the United States, where partisan cable TV hosts Tucker Carlson, Rachel Maddow, and Sean Hannity attract the most attention.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 13\n\n0% 25% 50% 75% 100%\n\n84\n\n81\n\n80\n\n74\n\n66\n\n59\n\n55\n\n76\n\n75\n\n67\n\n66\n\n62\n\n57\n\n53\n\n86\n\n83\n\n82\n\n80\n\n78\n\n76\n\n75\n\n71\n\n71\n\n69\n\n68\n\n66\n\n57\n\n55\n\n2022\n\n2018\n\nFrance Télévisions\n\nNHK (Japan)\n\nORF (Austria)\n\nZDF (Germany)\n\nARD (Germany)\n\nSR (Sweden)\n\nSVT (Sweden)\n\nBBC (UK)\n\nCBC (Canada)\n\nABC (Australia)\n\nRTÉ (Ireland)\n\nNRK (Norway)\n\nDR (Denmark)\n\nYle (Finland)\n\nPROPORTION THAT TRUSTS SELECTED PUBLIC SERVICE MEDIA BRANDS 2018 AND 2022 – SELECTED MARKETS\n\n**Q6\\_2018\\_trust.** How trustworthy would you say news from the following brands is? Please use the scale below, where 0 is ‘not at all trustworthy’ and 10 is ‘completely trustworthy’. _Base: All in 2018 and 2022 that had heard of each brand (n ≈ 2000). Note: Showing scores 6–10 on 10-point scale._\n\n2018\n\n2022\n\nBiggest drops Biggest drops\n\n3 During the first week of the conflict, the BBC estimated 280 million people used its news output in the UK and around the world. https://www.independent.co.uk/news/uk/ bbc-ukraine-tim-davie-nadine-dorries-bbc-news-b2047209.html\n\n4 https://www.theguardian.com/uk-news/2022/feb/17/bbcs-sarah-smith-relieved-to-escape-scottish-bile-and-hatred\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 14\n\nPROPORTION THAT TRUSTS THE BBC (2018–22) – UK\n\n**Q6\\_2018\\_trust.** How trustworthy would you say news from the following brands is? Please use the scale below, where 0 is ‘not at all trustworthy’ and 10 is ‘completely trustworthy’. _Base: All in 2018–22 that had heard of the BBC (n ≈ 2000). Note: Trust = % scored 6–10 on 10-point scale, Don’t trust = 0–4, Neither = 5._\n\n0%\n\n25%\n\n50%\n\n75%\n\n100% 11 16 17 21 26\n\n75 73\n\n64 62 55\n\n2022 2021 2020 2019 2018\n\nConservative Lib Dem Other\n\nSNP Labour\n\nTrust Don’t trust Neither\n\nThose that **don’t trust** the BBC are **more likely to be men** . Tend to be less interested in news and have lower levels of education.\n\nMore likely to **only use social media** as a source of news.\n\nOver half of this group **voted Conservative** at the last election (2019), **two thirds voted for Brexit** .\n\n53%\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 17"],"full_content":"**TRUST IN NEWS BRANDS: THE PARTICULAR CASE OF PUBLIC SERVICE MEDIA**\n\nUnderpinning news trust in many Northern and Western European countries, as well as Canada and Australia, we find a number of public service broadcasting organisations with a strong track record of independence. These generally appear at the top of our brand trust lists in each country and are often the first port of call for audiences when looking for reliable news around important stories such as COVID-19 and the Ukraine conflict. These institutions have a very different reputation to public and state broadcasters in Southern and Eastern Europe that often take a more partisan editorial line.\n\nBut independent public media are under increasing pressure in a number of countries, with attacks on funding, questions about impartiality, and challenges in reaching younger audiences who are increasingly turning to digital and social media. The BBC faces another round of cuts after a difficult licence settlement while DR, the Danish public broadcaster, recently cut three linear channels as part of a wider restructuring after a narrow parliamentary majority imposed a major cut in funding. Against this background, it is interesting to see how well trust is holding up, with some very notable exceptions. Nordic PSBs as well as those of other smaller countries like Ireland and Portugal have retained or even increased trust since 2018, but it has been a very different story in the UK, Australia, and Canada, with significant falls for the main public broadcasters.\n\nThe BBC, which remains the most widely used news source in the UK, reported record audiences in the first weeks of the Ukraine crisis, ³ but it has also come under intense criticism for its reporting of a range of more divisive issues – notably Brexit but also immigration, race, gender identity, and attitudes to COVID-19 vaccinations. In many cases, these criticisms of the BBC have been amplified on social media, with senior correspondents exposed to physical attacks along with personal abuse on social media. ⁴\n\nTrust in BBC News has fallen 20 percentage points in the last five years, from 75% to 55%. Equally telling is the proportion who say they distrust the BBC, which has grown from 11% to 26% (see next chart). The majority of these are from the political right, echoing criticism from Boris Johnson’s government about an alleged anti- Brexit and liberal bias, but we also find that low trust in the BBC also comes from those who are less interested in news altogether. It is important to note that other big brands in the UK (e.g. the _Guardian_ and the _Mail_ ) have been affected by growing levels of distrust, though not as severely. Declining trust is a particular challenge for public media organisations, as they try to fulfil their mission to appeal to all audiences.\n\nThe continued audience success of some European public media makes them a particular target for those who want to influence the debates on politics and wider issues around culture. Our survey shows that journalists from these organisations are often first to be recognised by the public. The outgoing BBC political editor Laura Kuenssberg is by far the most well-known journalist in the UK, according to respondents to our survey, even when it comes to digital news. Broadcasters with a commitment to impartiality, including the BBC, make up 62% of all mentions. Journalists from public media also feature strongly in data from Finland, France, and Germany. But it is a very different story in the United States, where partisan cable TV hosts Tucker Carlson, Rachel Maddow, and Sean Hannity attract the most attention.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 13\n\n0% 25% 50% 75% 100%\n\n84\n\n81\n\n80\n\n74\n\n66\n\n59\n\n55\n\n76\n\n75\n\n67\n\n66\n\n62\n\n57\n\n53\n\n86\n\n83\n\n82\n\n80\n\n78\n\n76\n\n75\n\n71\n\n71\n\n69\n\n68\n\n66\n\n57\n\n55\n\n2022\n\n2018\n\nFrance Télévisions\n\nNHK (Japan)\n\nORF (Austria)\n\nZDF (Germany)\n\nARD (Germany)\n\nSR (Sweden)\n\nSVT (Sweden)\n\nBBC (UK)\n\nCBC (Canada)\n\nABC (Australia)\n\nRTÉ (Ireland)\n\nNRK (Norway)\n\nDR (Denmark)\n\nYle (Finland)\n\nPROPORTION THAT TRUSTS SELECTED PUBLIC SERVICE MEDIA BRANDS 2018 AND 2022 – SELECTED MARKETS\n\n**Q6\\_2018\\_trust.** How trustworthy would you say news from the following brands is? Please use the scale below, where 0 is ‘not at all trustworthy’ and 10 is ‘completely trustworthy’. _Base: All in 2018 and 2022 that had heard of each brand (n ≈ 2000). Note: Showing scores 6–10 on 10-point scale._\n\n2018\n\n2022\n\nBiggest drops Biggest drops\n\n3 During the first week of the conflict, the BBC estimated 280 million people used its news output in the UK and around the world. https://www.independent.co.uk/news/uk/ bbc-ukraine-tim-davie-nadine-dorries-bbc-news-b2047209.html\n\n4 https://www.theguardian.com/uk-news/2022/feb/17/bbcs-sarah-smith-relieved-to-escape-scottish-bile-and-hatred\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 14\n\nPROPORTION THAT TRUSTS THE BBC (2018–22) – UK\n\n**Q6\\_2018\\_trust.** How trustworthy would you say news from the following brands is? Please use the scale below, where 0 is ‘not at all trustworthy’ and 10 is ‘completely trustworthy’. _Base: All in 2018–22 that had heard of the BBC (n ≈ 2000). Note: Trust = % scored 6–10 on 10-point scale, Don’t trust = 0–4, Neither = 5._\n\n0%\n\n25%\n\n50%\n\n75%\n\n100% 11 16 17 21 26\n\n75 73\n\n64 62 55\n\n2022 2021 2020 2019 2018\n\nConservative Lib Dem Other\n\nSNP Labour\n\nTrust Don’t trust Neither\n\nThose that **don’t trust** the BBC are **more likely to be men** . Tend to be less interested in news and have lower levels of education.\n\nMore likely to **only use social media** as a source of news.\n\nOver half of this group **voted Conservative** at the last election (2019), **two thirds voted for Brexit** .\n\n53%\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 17","publish_date":null,"title":"Graph 14 Graph 13","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Digital_News-Report_2022_p17.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Digital_News-Report_2022_p17.pdf"},"request":{"example_id":"chart/Digital_News-Report_2022_p17","product_type":"parse","source_file_path":"Digital_News-Report_2022_p17.pdf"},"started_at":"2026-11-04T19:46:04.382033"} +{"completed_at":"2026-11-04T19:46:07.999333","latency_in_ms":1562,"output":{"example_id":"chart/Digital_News-Report_2022_p18","grounded_pages":[],"layout_pages":[],"markdown":"_For further analysis see Section 2.5: Which Journalists do People Pay Most Attention to and Why? A Study of Six Countries_\n\n**Should journalists remain impartial when on social media?**\n\nWith more journalists building channels and direct relationships on social media, there is a growing debate about how they should interact when on platforms such as Facebook and Twitter. These are informal spaces where social media users can let their hair down and show more personality. But how compatible is this with the impartial or objective approaches still practised by many media brands (and that last year we found many members of the public value)?\n\nIn our survey, around half of respondents or more in most countries feel that journalists should stick to reporting the news, but a sizeable minority believed they should be allowed to express their personal opinions on social media at the same time. Brazil and Japan seem to be outliers here, showing opposite views.\n\nBut as we also found in last year’s _Digital News Report_ study on impartiality (Vir 2021), there are clear generational differences, with younger groups holding a stronger preference for journalists being able to express their personal opinions freely on social media. As some media organisations tighten social media guidelines, they are facing resistance from younger journalists who take a different view, and are trying to push the boundaries. This is just another way in which journalistic norms are being challenged by social and digital media.\n\n**PAYING FOR NEWS, SUBSCRIPTION FATIGUE, AND THE BATTLE FOR REGISTRATION**\n\nIn the last few years, many publishers have been increasing efforts to get audiences to pay for content online via subscription, membership, or donations – to reduce their reliance on advertising revenue, which online tends to flow towards the big platforms such as Google and Meta (formerly Facebook). These are now worldwide trends, and our country pages (Section 3) document how leading publishers in Argentina, Colombia, Japan, Nigeria, and Kenya, for example, have recently launched or consolidated paywalls.\n\nThis year’s data show mixed progress, with significant increases in a small number of wealthier countries, though there are signs elsewhere that growth may be levelling off. Across a basket of 20 countries where payment is relatively widespread, 17% paid for any online news – the same figure as last year. Norway continues to lead the way (41%), followed by Sweden (33%), Finland (19%), and the United States (19%). Australia (18%) and Germany (14%) showed the biggest increases this year at five percentage points in each case. By contrast, our survey data in Norway and the United States show a slight decline this year, though in both cases industry figures suggest that there has still been some progress. Either way, longer term trends do suggest a slowing down in some of these early-mover markets – raising questions about whether they might be reaching a more mature phase.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 15\n\nBest known journalists in Northern/Western Europe have reputation for **impartiality**\n\n**LAURA KUENSSBERG** BBC Political Editor (outgoing), UK\n\n**ALSO**\n\n**Marietta Slomka** ZDF TV Anchor, Germany\n\n**Matti Rönkä** Yle TV Anchor, Finland\n\n**ALSO**\n\n**Sean Hannity** Fox News\n\n**Rachel Maddow** MSNBC\n\n**TUCKER CARLSON** TV Anchor, Fox News\n\nBest known journalists in USA have reputation for **strong opinion**\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 16\n\nPROPORTION WHO THINK JOURNALISTS SHOULD STICK TO REPORTING THE NEWS ON SOCIAL MEDIA – SELECTED MARKETS\n\n**Q\\_Journalists\\_4.** Which of the following comes closest to how you think journalists should use social media like Twitter and Facebook? _Base: Total sample in each market._\n\n0%\n\n25%\n\n50%\n\n75%\n\n100%\n\n35 33 31 34\n\n29\n\n60\n\n44\n\n54 53 52 52 51\n\n28 27\n\nThey should be able to express personal opinions as well as reporting news\n\nStick to reporting the news\n\nJapan Brazil USA Germany Australia UK South Korea\n\nStick to reporting the news\n\nThey should be able to express personal opinions as well as reporting news\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 17\n\nPROPORTION WHO THINK JOURNALISTS SHOULD STICK TO REPORTING THE NEWS ON SOCIAL MEDIA – BY AGE – SELECTED MARKETS\n\n**Q\\_Journalists\\_4.** Which of the following comes closest to how you think journalists should use social media like Twitter and Facebook? _Base: 18–24/25–34/35–44/45–54/55+ in Australia, Brazil, Germany, Ireland, Japan, Norway, South Korea, UK, USA ≈ 1884/2974/3306/2884/7527. Note: This question was only asked in the countries listed._\n\n0%\n\n25%\n\n50%\n\n75%\n\n100%\n\n57\n\n29\n\n47\n\n37 42 41 38\n\n44\n\n34\n\n46\n\nThey should be able to express their personal opinions alongside reporting the news\n\nThey should stick to reporting the news\n\n18-24 25-34 35-44 45-54 55+\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 18","pages":[{"line_numbers":[],"markdown":"_For further analysis see Section 2.5: Which Journalists do People Pay Most Attention to and Why? A Study of Six Countries_\n\n**Should journalists remain impartial when on social media?**\n\nWith more journalists building channels and direct relationships on social media, there is a growing debate about how they should interact when on platforms such as Facebook and Twitter. These are informal spaces where social media users can let their hair down and show more personality. But how compatible is this with the impartial or objective approaches still practised by many media brands (and that last year we found many members of the public value)?\n\nIn our survey, around half of respondents or more in most countries feel that journalists should stick to reporting the news, but a sizeable minority believed they should be allowed to express their personal opinions on social media at the same time. Brazil and Japan seem to be outliers here, showing opposite views.\n\nBut as we also found in last year’s _Digital News Report_ study on impartiality (Vir 2021), there are clear generational differences, with younger groups holding a stronger preference for journalists being able to express their personal opinions freely on social media. As some media organisations tighten social media guidelines, they are facing resistance from younger journalists who take a different view, and are trying to push the boundaries. This is just another way in which journalistic norms are being challenged by social and digital media.\n\n**PAYING FOR NEWS, SUBSCRIPTION FATIGUE, AND THE BATTLE FOR REGISTRATION**\n\nIn the last few years, many publishers have been increasing efforts to get audiences to pay for content online via subscription, membership, or donations – to reduce their reliance on advertising revenue, which online tends to flow towards the big platforms such as Google and Meta (formerly Facebook). These are now worldwide trends, and our country pages (Section 3) document how leading publishers in Argentina, Colombia, Japan, Nigeria, and Kenya, for example, have recently launched or consolidated paywalls.\n\nThis year’s data show mixed progress, with significant increases in a small number of wealthier countries, though there are signs elsewhere that growth may be levelling off. Across a basket of 20 countries where payment is relatively widespread, 17% paid for any online news – the same figure as last year. Norway continues to lead the way (41%), followed by Sweden (33%), Finland (19%), and the United States (19%). Australia (18%) and Germany (14%) showed the biggest increases this year at five percentage points in each case. By contrast, our survey data in Norway and the United States show a slight decline this year, though in both cases industry figures suggest that there has still been some progress. Either way, longer term trends do suggest a slowing down in some of these early-mover markets – raising questions about whether they might be reaching a more mature phase.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 15\n\nBest known journalists in Northern/Western Europe have reputation for **impartiality**\n\n**LAURA KUENSSBERG** BBC Political Editor (outgoing), UK\n\n**ALSO**\n\n**Marietta Slomka** ZDF TV Anchor, Germany\n\n**Matti Rönkä** Yle TV Anchor, Finland\n\n**ALSO**\n\n**Sean Hannity** Fox News\n\n**Rachel Maddow** MSNBC\n\n**TUCKER CARLSON** TV Anchor, Fox News\n\nBest known journalists in USA have reputation for **strong opinion**\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 16\n\nPROPORTION WHO THINK JOURNALISTS SHOULD STICK TO REPORTING THE NEWS ON SOCIAL MEDIA – SELECTED MARKETS\n\n**Q\\_Journalists\\_4.** Which of the following comes closest to how you think journalists should use social media like Twitter and Facebook? _Base: Total sample in each market._\n\n0%\n\n25%\n\n50%\n\n75%\n\n100%\n\n35 33 31 34\n\n29\n\n60\n\n44\n\n54 53 52 52 51\n\n28 27\n\nThey should be able to express personal opinions as well as reporting news\n\nStick to reporting the news\n\nJapan Brazil USA Germany Australia UK South Korea\n\nStick to reporting the news\n\nThey should be able to express personal opinions as well as reporting news\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 17\n\nPROPORTION WHO THINK JOURNALISTS SHOULD STICK TO REPORTING THE NEWS ON SOCIAL MEDIA – BY AGE – SELECTED MARKETS\n\n**Q\\_Journalists\\_4.** Which of the following comes closest to how you think journalists should use social media like Twitter and Facebook? _Base: 18–24/25–34/35–44/45–54/55+ in Australia, Brazil, Germany, Ireland, Japan, Norway, South Korea, UK, USA ≈ 1884/2974/3306/2884/7527. Note: This question was only asked in the countries listed._\n\n0%\n\n25%\n\n50%\n\n75%\n\n100%\n\n57\n\n29\n\n47\n\n37 42 41 38\n\n44\n\n34\n\n46\n\nThey should be able to express their personal opinions alongside reporting the news\n\nThey should stick to reporting the news\n\n18-24 25-34 35-44 45-54 55+\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 18","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"_For further analysis see Section 2.5: Which Journalists do People Pay Most Attention to and Why? A Study of Six Countries_\n\n**Should journalists remain impartial when on social media?**\n\nWith more journalists building channels and direct relationships on social media, there is a growing debate about how they should interact when on platforms such as Facebook and Twitter. These are informal spaces where social media users can let their hair down and show more personality. But how compatible is this with the impartial or objective approaches still practised by many media brands (and that last year we found many members of the public value)?\n\nIn our survey, around half of respondents or more in most countries feel that journalists should stick to reporting the news, but a sizeable minority believed they should be allowed to express their personal opinions on social media at the same time. Brazil and Japan seem to be outliers here, showing opposite views.\n\nBut as we also found in last year’s _Digital News Report_ study on impartiality (Vir 2021), there are clear generational differences, with younger groups holding a stronger preference for journalists being able to express their personal opinions freely on social media. As some media organisations tighten social media guidelines, they are facing resistance from younger journalists who take a different view, and are trying to push the boundaries. This is just another way in which journalistic norms are being challenged by social and digital media.\n\n**PAYING FOR NEWS, SUBSCRIPTION FATIGUE, AND THE BATTLE FOR REGISTRATION**\n\nIn the last few years, many publishers have been increasing efforts to get audiences to pay for content online via subscription, membership, or donations – to reduce their reliance on advertising revenue, which online tends to flow towards the big platforms such as Google and Meta (formerly Facebook). These are now worldwide trends, and our country pages (Section 3) document how leading publishers in Argentina, Colombia, Japan, Nigeria, and Kenya, for example, have recently launched or consolidated paywalls.\n\nThis year’s data show mixed progress, with significant increases in a small number of wealthier countries, though there are signs elsewhere that growth may be levelling off. Across a basket of 20 countries where payment is relatively widespread, 17% paid for any online news – the same figure as last year. Norway continues to lead the way (41%), followed by Sweden (33%), Finland (19%), and the United States (19%). Australia (18%) and Germany (14%) showed the biggest increases this year at five percentage points in each case. By contrast, our survey data in Norway and the United States show a slight decline this year, though in both cases industry figures suggest that there has still been some progress. Either way, longer term trends do suggest a slowing down in some of these early-mover markets – raising questions about whether they might be reaching a more mature phase.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 15\n\nBest known journalists in Northern/Western Europe have reputation for **impartiality**\n\n**LAURA KUENSSBERG** BBC Political Editor (outgoing), UK\n\n**ALSO**\n\n**Marietta Slomka** ZDF TV Anchor, Germany\n\n**Matti Rönkä** Yle TV Anchor, Finland\n\n**ALSO**\n\n**Sean Hannity** Fox News\n\n**Rachel Maddow** MSNBC\n\n**TUCKER CARLSON** TV Anchor, Fox News\n\nBest known journalists in USA have reputation for **strong opinion**\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 16\n\nPROPORTION WHO THINK JOURNALISTS SHOULD STICK TO REPORTING THE NEWS ON SOCIAL MEDIA – SELECTED MARKETS\n\n**Q\\_Journalists\\_4.** Which of the following comes closest to how you think journalists should use social media like Twitter and Facebook? _Base: Total sample in each market._\n\n0%\n\n25%\n\n50%\n\n75%\n\n100%\n\n35 33 31 34\n\n29\n\n60\n\n44\n\n54 53 52 52 51\n\n28 27\n\nThey should be able to express personal opinions as well as reporting news\n\nStick to reporting the news\n\nJapan Brazil USA Germany Australia UK South Korea\n\nStick to reporting the news\n\nThey should be able to express personal opinions as well as reporting news\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 17\n\nPROPORTION WHO THINK JOURNALISTS SHOULD STICK TO REPORTING THE NEWS ON SOCIAL MEDIA – BY AGE – SELECTED MARKETS\n\n**Q\\_Journalists\\_4.** Which of the following comes closest to how you think journalists should use social media like Twitter and Facebook? _Base: 18–24/25–34/35–44/45–54/55+ in Australia, Brazil, Germany, Ireland, Japan, Norway, South Korea, UK, USA ≈ 1884/2974/3306/2884/7527. Note: This question was only asked in the countries listed._\n\n0%\n\n25%\n\n50%\n\n75%\n\n100%\n\n57\n\n29\n\n47\n\n37 42 41 38\n\n44\n\n34\n\n46\n\nThey should be able to express their personal opinions alongside reporting the news\n\nThey should stick to reporting the news\n\n18-24 25-34 35-44 45-54 55+\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 18","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Digital_News-Report_2022_p18.pdf"]},"response":{"errors":[],"extract_id":"extract_602778e845542a66dfd22bcf3854e783","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["_For further analysis see Section 2.5: Which Journalists do People Pay Most Attention to and Why? A Study of Six Countries_\n\n**Should journalists remain impartial when on social media?**\n\nWith more journalists building channels and direct relationships on social media, there is a growing debate about how they should interact when on platforms such as Facebook and Twitter. These are informal spaces where social media users can let their hair down and show more personality. But how compatible is this with the impartial or objective approaches still practised by many media brands (and that last year we found many members of the public value)?\n\nIn our survey, around half of respondents or more in most countries feel that journalists should stick to reporting the news, but a sizeable minority believed they should be allowed to express their personal opinions on social media at the same time. Brazil and Japan seem to be outliers here, showing opposite views.\n\nBut as we also found in last year’s _Digital News Report_ study on impartiality (Vir 2021), there are clear generational differences, with younger groups holding a stronger preference for journalists being able to express their personal opinions freely on social media. As some media organisations tighten social media guidelines, they are facing resistance from younger journalists who take a different view, and are trying to push the boundaries. This is just another way in which journalistic norms are being challenged by social and digital media.\n\n**PAYING FOR NEWS, SUBSCRIPTION FATIGUE, AND THE BATTLE FOR REGISTRATION**\n\nIn the last few years, many publishers have been increasing efforts to get audiences to pay for content online via subscription, membership, or donations – to reduce their reliance on advertising revenue, which online tends to flow towards the big platforms such as Google and Meta (formerly Facebook). These are now worldwide trends, and our country pages (Section 3) document how leading publishers in Argentina, Colombia, Japan, Nigeria, and Kenya, for example, have recently launched or consolidated paywalls.\n\nThis year’s data show mixed progress, with significant increases in a small number of wealthier countries, though there are signs elsewhere that growth may be levelling off. Across a basket of 20 countries where payment is relatively widespread, 17% paid for any online news – the same figure as last year. Norway continues to lead the way (41%), followed by Sweden (33%), Finland (19%), and the United States (19%). Australia (18%) and Germany (14%) showed the biggest increases this year at five percentage points in each case. By contrast, our survey data in Norway and the United States show a slight decline this year, though in both cases industry figures suggest that there has still been some progress. Either way, longer term trends do suggest a slowing down in some of these early-mover markets – raising questions about whether they might be reaching a more mature phase.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 15\n\nBest known journalists in Northern/Western Europe have reputation for **impartiality**\n\n**LAURA KUENSSBERG** BBC Political Editor (outgoing), UK\n\n**ALSO**\n\n**Marietta Slomka** ZDF TV Anchor, Germany\n\n**Matti Rönkä** Yle TV Anchor, Finland\n\n**ALSO**\n\n**Sean Hannity** Fox News\n\n**Rachel Maddow** MSNBC\n\n**TUCKER CARLSON** TV Anchor, Fox News\n\nBest known journalists in USA have reputation for **strong opinion**\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 16\n\nPROPORTION WHO THINK JOURNALISTS SHOULD STICK TO REPORTING THE NEWS ON SOCIAL MEDIA – SELECTED MARKETS\n\n**Q\\_Journalists\\_4.** Which of the following comes closest to how you think journalists should use social media like Twitter and Facebook? _Base: Total sample in each market._\n\n0%\n\n25%\n\n50%\n\n75%\n\n100%\n\n35 33 31 34\n\n29\n\n60\n\n44\n\n54 53 52 52 51\n\n28 27\n\nThey should be able to express personal opinions as well as reporting news\n\nStick to reporting the news\n\nJapan Brazil USA Germany Australia UK South Korea\n\nStick to reporting the news\n\nThey should be able to express personal opinions as well as reporting news\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 17\n\nPROPORTION WHO THINK JOURNALISTS SHOULD STICK TO REPORTING THE NEWS ON SOCIAL MEDIA – BY AGE – SELECTED MARKETS\n\n**Q\\_Journalists\\_4.** Which of the following comes closest to how you think journalists should use social media like Twitter and Facebook? _Base: 18–24/25–34/35–44/45–54/55+ in Australia, Brazil, Germany, Ireland, Japan, Norway, South Korea, UK, USA ≈ 1884/2974/3306/2884/7527. Note: This question was only asked in the countries listed._\n\n0%\n\n25%\n\n50%\n\n75%\n\n100%\n\n57\n\n29\n\n47\n\n37 42 41 38\n\n44\n\n34\n\n46\n\nThey should be able to express their personal opinions alongside reporting the news\n\nThey should stick to reporting the news\n\n18-24 25-34 35-44 45-54 55+\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 18"],"full_content":"_For further analysis see Section 2.5: Which Journalists do People Pay Most Attention to and Why? A Study of Six Countries_\n\n**Should journalists remain impartial when on social media?**\n\nWith more journalists building channels and direct relationships on social media, there is a growing debate about how they should interact when on platforms such as Facebook and Twitter. These are informal spaces where social media users can let their hair down and show more personality. But how compatible is this with the impartial or objective approaches still practised by many media brands (and that last year we found many members of the public value)?\n\nIn our survey, around half of respondents or more in most countries feel that journalists should stick to reporting the news, but a sizeable minority believed they should be allowed to express their personal opinions on social media at the same time. Brazil and Japan seem to be outliers here, showing opposite views.\n\nBut as we also found in last year’s _Digital News Report_ study on impartiality (Vir 2021), there are clear generational differences, with younger groups holding a stronger preference for journalists being able to express their personal opinions freely on social media. As some media organisations tighten social media guidelines, they are facing resistance from younger journalists who take a different view, and are trying to push the boundaries. This is just another way in which journalistic norms are being challenged by social and digital media.\n\n**PAYING FOR NEWS, SUBSCRIPTION FATIGUE, AND THE BATTLE FOR REGISTRATION**\n\nIn the last few years, many publishers have been increasing efforts to get audiences to pay for content online via subscription, membership, or donations – to reduce their reliance on advertising revenue, which online tends to flow towards the big platforms such as Google and Meta (formerly Facebook). These are now worldwide trends, and our country pages (Section 3) document how leading publishers in Argentina, Colombia, Japan, Nigeria, and Kenya, for example, have recently launched or consolidated paywalls.\n\nThis year’s data show mixed progress, with significant increases in a small number of wealthier countries, though there are signs elsewhere that growth may be levelling off. Across a basket of 20 countries where payment is relatively widespread, 17% paid for any online news – the same figure as last year. Norway continues to lead the way (41%), followed by Sweden (33%), Finland (19%), and the United States (19%). Australia (18%) and Germany (14%) showed the biggest increases this year at five percentage points in each case. By contrast, our survey data in Norway and the United States show a slight decline this year, though in both cases industry figures suggest that there has still been some progress. Either way, longer term trends do suggest a slowing down in some of these early-mover markets – raising questions about whether they might be reaching a more mature phase.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 15\n\nBest known journalists in Northern/Western Europe have reputation for **impartiality**\n\n**LAURA KUENSSBERG** BBC Political Editor (outgoing), UK\n\n**ALSO**\n\n**Marietta Slomka** ZDF TV Anchor, Germany\n\n**Matti Rönkä** Yle TV Anchor, Finland\n\n**ALSO**\n\n**Sean Hannity** Fox News\n\n**Rachel Maddow** MSNBC\n\n**TUCKER CARLSON** TV Anchor, Fox News\n\nBest known journalists in USA have reputation for **strong opinion**\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 16\n\nPROPORTION WHO THINK JOURNALISTS SHOULD STICK TO REPORTING THE NEWS ON SOCIAL MEDIA – SELECTED MARKETS\n\n**Q\\_Journalists\\_4.** Which of the following comes closest to how you think journalists should use social media like Twitter and Facebook? _Base: Total sample in each market._\n\n0%\n\n25%\n\n50%\n\n75%\n\n100%\n\n35 33 31 34\n\n29\n\n60\n\n44\n\n54 53 52 52 51\n\n28 27\n\nThey should be able to express personal opinions as well as reporting news\n\nStick to reporting the news\n\nJapan Brazil USA Germany Australia UK South Korea\n\nStick to reporting the news\n\nThey should be able to express personal opinions as well as reporting news\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 17\n\nPROPORTION WHO THINK JOURNALISTS SHOULD STICK TO REPORTING THE NEWS ON SOCIAL MEDIA – BY AGE – SELECTED MARKETS\n\n**Q\\_Journalists\\_4.** Which of the following comes closest to how you think journalists should use social media like Twitter and Facebook? _Base: 18–24/25–34/35–44/45–54/55+ in Australia, Brazil, Germany, Ireland, Japan, Norway, South Korea, UK, USA ≈ 1884/2974/3306/2884/7527. Note: This question was only asked in the countries listed._\n\n0%\n\n25%\n\n50%\n\n75%\n\n100%\n\n57\n\n29\n\n47\n\n37 42 41 38\n\n44\n\n34\n\n46\n\nThey should be able to express their personal opinions alongside reporting the news\n\nThey should stick to reporting the news\n\n18-24 25-34 35-44 45-54 55+\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 18","publish_date":null,"title":"Graph 15 Graph 17 Graph 16","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Digital_News-Report_2022_p18.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Digital_News-Report_2022_p18.pdf"},"request":{"example_id":"chart/Digital_News-Report_2022_p18","product_type":"parse","source_file_path":"Digital_News-Report_2022_p18.pdf"},"started_at":"2026-11-04T19:46:06.436366"} +{"completed_at":"2026-11-04T19:46:11.395672","latency_in_ms":3542,"output":{"example_id":"chart/Digital_News-Report_2022_p20","grounded_pages":[],"layout_pages":[],"markdown":"In almost all of the listed countries, the majority of subscribers pay for one publication. But in the United States and Australia, around half (56% and 51% respectively) now pay for _two or more_ – often a national and local paper combination. Second subscriptions in the United States include political and cultural magazines such as _The Atlantic_ and _The New Yorker_ , partisan digital outlets such as Blaze Media and Epoch Times, or passion-based titles such as the Athletic (sports). We also see growing levels of payment for platform-based news subscription products Apple+ and Twitter Blue. In Australia, second subscriptions include a relatively high proportion (9%) from US or UK publications such as the _New York Times_ and _The Times_ of London.\n\n**The influence of individual journalist brands may be overstated**\n\nSome news brands continue to have the reputation, scale, and technical capability to charge for their services, but in parallel the emergence of low-cost subscription and membership platforms, like Substack and Patreon, has enabled some individual journalists, podcasters, or other creators to also convince people to pay for their work. Substack says that there are now around a million paying subscribers to its premium content emails. ⁵\n\nDespite this, our data show the number of these individual subscriptions is still relatively modest, certainly when compared with established news brands. Even in the United States, we find that just 7% of news subscribers in the United States pay for one or more journalist emails – around 1% of our overall sample. The proportions are even lower elsewhere, though we do see some emerging payment for individual email writers or podcasters in parts of Europe and Australia.\n\nOver time, it is possible that these solo or small businesses, mostly operating in closely defined niches, could make up a far bigger part of the overall pie, expanding the market but also creating more competition for big news brands. If nothing else, they have already made the value of some individual journalists more explicit, pushing up wages of top talent.\n\nThese data also highlight the growing importance of formats like email and podcasts, both of which have become much more important in driving regular engagement for all publishers – from general news to specialist information and entertainment.\n\n_For further analysis see Section 2.4: Email News: Its Contribution to Engagement and Monetisation_\n\n5 https://www.forbes.com/sites/marisadellatto/2021/11/15/substack-announces-one-million-paid-subscribers/\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 20\n\nPROPORTION OF SUBSCRIBERS WHO PAY FOR INDIVIDUAL JOURNALISTS VIA EMAIL OR OTHER SERVICES – SELECTED COUNTRIES\n\n**Q7\\_SUBS\\_name\\_2022.** You said you have paid a subscription/membership to a digital news service in the last year … Which of the following did you subscribe to? _Base: All who made a regular payment to at least one digital news service in last year. USA = 304, Spain = 153, Belgium = 211, Netherlands = 221, Sweden = 487, France = 135, Germany = 183, Australia = 270, UK = 160._\n\n0%\n\n25%\n\n50%\n\n75%\n\n100%\n\n57\n\n29\n\n47 37 42 41 38 44 34\n\n46\n\n55+\n\n45-54\n\n35-44\n\n25-34\n\n18-24\n\nThey should be able to express their personal opinions alongside reporting the news\n\nThey should stick to reporting the news\n\n0%\n\n2%\n\n4%\n\n6%\n\n8%\n\n10%\n\n4\n\n5\n\n3\n\n2\n\n3\n\n1\n\n3\n\n2\n\n1\n\n7\n\n5\n\n4\n\n3\n\n2 2\n\n1 1 0\n\nJournalist via YouTube/podcast\n\nJournalist via Newsletter/website\n\nUK Germany Australia France Belgium Netherlands Sweden Spain USA\n\nJournalist via newsletter/website\n\nJournalist via YouTube/podcast\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 21\n\nPROPORTION WHO SUBSCRIBE TO DIFFERENT MEDIA SERVICES – UK\n\n**Q\\_Sub\\_fatigue1.** There are many different online media subscriptions you can pay for these days. How many of the following online media subscriptions do you currently have, if any, that you pay for with your own money? **Q7\\_SUBS.** You said you have paid a subscription or made an ongoing donation to a digital news service in the last year. How many different news providers do you regularly pay money to? _Base: UK = 2410._\n\nKey: No Sub 1 2 3 4 5 or more\n\n35 25 20 13 4\n\n**TV/Movies** (e.g., Netflix, Amazon Prime Video, Disney+) **65% pay**\n\n63 30 3 2\n\n**Music** (e.g., Spotify, Apple Music, Pandora, Deezer, etc.) **37% pay**\n\n78 14 5 2\n\n**Sports** (e.g., Sky Sports, BT Sport, etc.) **22% pay**\n\n\\* Note: 7% pay for a news subscription or other ongoing subscription, 9% for any payment incl. one-off payments and donations.\n\n93 6\n\n**News** (e.g., Telegraph, Times, Guardian, etc.) **7% pay\\***\n\n0% 25% 50% 75% 100%\n\nIn the vast majority of markets in our survey, people say they still identify most with traditional news brands rather than individual journalists. This is especially true in Nordic countries but less true in the United States, Southern Europe, Latin America, and parts of Asia. Perhaps the greater affinity with journalists in these countries offers a different opportunity to monetise content than brand-led subscription.\n\n**The impact of wider media subscriptions and prospects for subscription fatigue**\n\nWith rising energy prices and inflation, and with many different types of online media (TV, music, books) now competing for a share of household budgets, there are genuine fears in the industry about whether recent growth in news subscriptions can continue. Against this backdrop, we explored the characteristics of different media subscriptions and asked respondents whether they were thinking of taking out more – or cutting back.\n\nTaking the UK as an example of an advanced subscription market (chart below), we find almost two-thirds of our sample (65%) have taken out at least one subscription to a TV or film service like Netflix, 37% have a music subscription to Spotify or similar, 22% pay for a premium sports service, and 7% have a news subscription or other ongoing payment such as a donation. Because premium content is spread across TV platforms, many people feel the need to take out more than one subscription, with 19% taking out three or more. By contrast, one music or news service is normally enough, as rival services often offer broadly similar content. But as we noted earlier, in some countries we are seeing more differentiated news services and a consequent rise in the number of subscriptions per person. The other big difference is the age profile, with TV, music, and audio books skewing younger, while sport has an equal age split and news skews much older. People who subscribe to news are more likely to subscribe to other services and vice versa.\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 20","pages":[{"line_numbers":[],"markdown":"In almost all of the listed countries, the majority of subscribers pay for one publication. But in the United States and Australia, around half (56% and 51% respectively) now pay for _two or more_ – often a national and local paper combination. Second subscriptions in the United States include political and cultural magazines such as _The Atlantic_ and _The New Yorker_ , partisan digital outlets such as Blaze Media and Epoch Times, or passion-based titles such as the Athletic (sports). We also see growing levels of payment for platform-based news subscription products Apple+ and Twitter Blue. In Australia, second subscriptions include a relatively high proportion (9%) from US or UK publications such as the _New York Times_ and _The Times_ of London.\n\n**The influence of individual journalist brands may be overstated**\n\nSome news brands continue to have the reputation, scale, and technical capability to charge for their services, but in parallel the emergence of low-cost subscription and membership platforms, like Substack and Patreon, has enabled some individual journalists, podcasters, or other creators to also convince people to pay for their work. Substack says that there are now around a million paying subscribers to its premium content emails. ⁵\n\nDespite this, our data show the number of these individual subscriptions is still relatively modest, certainly when compared with established news brands. Even in the United States, we find that just 7% of news subscribers in the United States pay for one or more journalist emails – around 1% of our overall sample. The proportions are even lower elsewhere, though we do see some emerging payment for individual email writers or podcasters in parts of Europe and Australia.\n\nOver time, it is possible that these solo or small businesses, mostly operating in closely defined niches, could make up a far bigger part of the overall pie, expanding the market but also creating more competition for big news brands. If nothing else, they have already made the value of some individual journalists more explicit, pushing up wages of top talent.\n\nThese data also highlight the growing importance of formats like email and podcasts, both of which have become much more important in driving regular engagement for all publishers – from general news to specialist information and entertainment.\n\n_For further analysis see Section 2.4: Email News: Its Contribution to Engagement and Monetisation_\n\n5 https://www.forbes.com/sites/marisadellatto/2021/11/15/substack-announces-one-million-paid-subscribers/\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 20\n\nPROPORTION OF SUBSCRIBERS WHO PAY FOR INDIVIDUAL JOURNALISTS VIA EMAIL OR OTHER SERVICES – SELECTED COUNTRIES\n\n**Q7\\_SUBS\\_name\\_2022.** You said you have paid a subscription/membership to a digital news service in the last year … Which of the following did you subscribe to? _Base: All who made a regular payment to at least one digital news service in last year. USA = 304, Spain = 153, Belgium = 211, Netherlands = 221, Sweden = 487, France = 135, Germany = 183, Australia = 270, UK = 160._\n\n0%\n\n25%\n\n50%\n\n75%\n\n100%\n\n57\n\n29\n\n47 37 42 41 38 44 34\n\n46\n\n55+\n\n45-54\n\n35-44\n\n25-34\n\n18-24\n\nThey should be able to express their personal opinions alongside reporting the news\n\nThey should stick to reporting the news\n\n0%\n\n2%\n\n4%\n\n6%\n\n8%\n\n10%\n\n4\n\n5\n\n3\n\n2\n\n3\n\n1\n\n3\n\n2\n\n1\n\n7\n\n5\n\n4\n\n3\n\n2 2\n\n1 1 0\n\nJournalist via YouTube/podcast\n\nJournalist via Newsletter/website\n\nUK Germany Australia France Belgium Netherlands Sweden Spain USA\n\nJournalist via newsletter/website\n\nJournalist via YouTube/podcast\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 21\n\nPROPORTION WHO SUBSCRIBE TO DIFFERENT MEDIA SERVICES – UK\n\n**Q\\_Sub\\_fatigue1.** There are many different online media subscriptions you can pay for these days. How many of the following online media subscriptions do you currently have, if any, that you pay for with your own money? **Q7\\_SUBS.** You said you have paid a subscription or made an ongoing donation to a digital news service in the last year. How many different news providers do you regularly pay money to? _Base: UK = 2410._\n\nKey: No Sub 1 2 3 4 5 or more\n\n35 25 20 13 4\n\n**TV/Movies** (e.g., Netflix, Amazon Prime Video, Disney+) **65% pay**\n\n63 30 3 2\n\n**Music** (e.g., Spotify, Apple Music, Pandora, Deezer, etc.) **37% pay**\n\n78 14 5 2\n\n**Sports** (e.g., Sky Sports, BT Sport, etc.) **22% pay**\n\n\\* Note: 7% pay for a news subscription or other ongoing subscription, 9% for any payment incl. one-off payments and donations.\n\n93 6\n\n**News** (e.g., Telegraph, Times, Guardian, etc.) **7% pay\\***\n\n0% 25% 50% 75% 100%\n\nIn the vast majority of markets in our survey, people say they still identify most with traditional news brands rather than individual journalists. This is especially true in Nordic countries but less true in the United States, Southern Europe, Latin America, and parts of Asia. Perhaps the greater affinity with journalists in these countries offers a different opportunity to monetise content than brand-led subscription.\n\n**The impact of wider media subscriptions and prospects for subscription fatigue**\n\nWith rising energy prices and inflation, and with many different types of online media (TV, music, books) now competing for a share of household budgets, there are genuine fears in the industry about whether recent growth in news subscriptions can continue. Against this backdrop, we explored the characteristics of different media subscriptions and asked respondents whether they were thinking of taking out more – or cutting back.\n\nTaking the UK as an example of an advanced subscription market (chart below), we find almost two-thirds of our sample (65%) have taken out at least one subscription to a TV or film service like Netflix, 37% have a music subscription to Spotify or similar, 22% pay for a premium sports service, and 7% have a news subscription or other ongoing payment such as a donation. Because premium content is spread across TV platforms, many people feel the need to take out more than one subscription, with 19% taking out three or more. By contrast, one music or news service is normally enough, as rival services often offer broadly similar content. But as we noted earlier, in some countries we are seeing more differentiated news services and a consequent rise in the number of subscriptions per person. The other big difference is the age profile, with TV, music, and audio books skewing younger, while sport has an equal age split and news skews much older. People who subscribe to news are more likely to subscribe to other services and vice versa.\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 20","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"In almost all of the listed countries, the majority of subscribers pay for one publication. But in the United States and Australia, around half (56% and 51% respectively) now pay for _two or more_ – often a national and local paper combination. Second subscriptions in the United States include political and cultural magazines such as _The Atlantic_ and _The New Yorker_ , partisan digital outlets such as Blaze Media and Epoch Times, or passion-based titles such as the Athletic (sports). We also see growing levels of payment for platform-based news subscription products Apple+ and Twitter Blue. In Australia, second subscriptions include a relatively high proportion (9%) from US or UK publications such as the _New York Times_ and _The Times_ of London.\n\n**The influence of individual journalist brands may be overstated**\n\nSome news brands continue to have the reputation, scale, and technical capability to charge for their services, but in parallel the emergence of low-cost subscription and membership platforms, like Substack and Patreon, has enabled some individual journalists, podcasters, or other creators to also convince people to pay for their work. Substack says that there are now around a million paying subscribers to its premium content emails. ⁵\n\nDespite this, our data show the number of these individual subscriptions is still relatively modest, certainly when compared with established news brands. Even in the United States, we find that just 7% of news subscribers in the United States pay for one or more journalist emails – around 1% of our overall sample. The proportions are even lower elsewhere, though we do see some emerging payment for individual email writers or podcasters in parts of Europe and Australia.\n\nOver time, it is possible that these solo or small businesses, mostly operating in closely defined niches, could make up a far bigger part of the overall pie, expanding the market but also creating more competition for big news brands. If nothing else, they have already made the value of some individual journalists more explicit, pushing up wages of top talent.\n\nThese data also highlight the growing importance of formats like email and podcasts, both of which have become much more important in driving regular engagement for all publishers – from general news to specialist information and entertainment.\n\n_For further analysis see Section 2.4: Email News: Its Contribution to Engagement and Monetisation_\n\n5 https://www.forbes.com/sites/marisadellatto/2021/11/15/substack-announces-one-million-paid-subscribers/\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 20\n\nPROPORTION OF SUBSCRIBERS WHO PAY FOR INDIVIDUAL JOURNALISTS VIA EMAIL OR OTHER SERVICES – SELECTED COUNTRIES\n\n**Q7\\_SUBS\\_name\\_2022.** You said you have paid a subscription/membership to a digital news service in the last year … Which of the following did you subscribe to? _Base: All who made a regular payment to at least one digital news service in last year. USA = 304, Spain = 153, Belgium = 211, Netherlands = 221, Sweden = 487, France = 135, Germany = 183, Australia = 270, UK = 160._\n\n0%\n\n25%\n\n50%\n\n75%\n\n100%\n\n57\n\n29\n\n47 37 42 41 38 44 34\n\n46\n\n55+\n\n45-54\n\n35-44\n\n25-34\n\n18-24\n\nThey should be able to express their personal opinions alongside reporting the news\n\nThey should stick to reporting the news\n\n0%\n\n2%\n\n4%\n\n6%\n\n8%\n\n10%\n\n4\n\n5\n\n3\n\n2\n\n3\n\n1\n\n3\n\n2\n\n1\n\n7\n\n5\n\n4\n\n3\n\n2 2\n\n1 1 0\n\nJournalist via YouTube/podcast\n\nJournalist via Newsletter/website\n\nUK Germany Australia France Belgium Netherlands Sweden Spain USA\n\nJournalist via newsletter/website\n\nJournalist via YouTube/podcast\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 21\n\nPROPORTION WHO SUBSCRIBE TO DIFFERENT MEDIA SERVICES – UK\n\n**Q\\_Sub\\_fatigue1.** There are many different online media subscriptions you can pay for these days. How many of the following online media subscriptions do you currently have, if any, that you pay for with your own money? **Q7\\_SUBS.** You said you have paid a subscription or made an ongoing donation to a digital news service in the last year. How many different news providers do you regularly pay money to? _Base: UK = 2410._\n\nKey: No Sub 1 2 3 4 5 or more\n\n35 25 20 13 4\n\n**TV/Movies** (e.g., Netflix, Amazon Prime Video, Disney+) **65% pay**\n\n63 30 3 2\n\n**Music** (e.g., Spotify, Apple Music, Pandora, Deezer, etc.) **37% pay**\n\n78 14 5 2\n\n**Sports** (e.g., Sky Sports, BT Sport, etc.) **22% pay**\n\n\\* Note: 7% pay for a news subscription or other ongoing subscription, 9% for any payment incl. one-off payments and donations.\n\n93 6\n\n**News** (e.g., Telegraph, Times, Guardian, etc.) **7% pay\\***\n\n0% 25% 50% 75% 100%\n\nIn the vast majority of markets in our survey, people say they still identify most with traditional news brands rather than individual journalists. This is especially true in Nordic countries but less true in the United States, Southern Europe, Latin America, and parts of Asia. Perhaps the greater affinity with journalists in these countries offers a different opportunity to monetise content than brand-led subscription.\n\n**The impact of wider media subscriptions and prospects for subscription fatigue**\n\nWith rising energy prices and inflation, and with many different types of online media (TV, music, books) now competing for a share of household budgets, there are genuine fears in the industry about whether recent growth in news subscriptions can continue. Against this backdrop, we explored the characteristics of different media subscriptions and asked respondents whether they were thinking of taking out more – or cutting back.\n\nTaking the UK as an example of an advanced subscription market (chart below), we find almost two-thirds of our sample (65%) have taken out at least one subscription to a TV or film service like Netflix, 37% have a music subscription to Spotify or similar, 22% pay for a premium sports service, and 7% have a news subscription or other ongoing payment such as a donation. Because premium content is spread across TV platforms, many people feel the need to take out more than one subscription, with 19% taking out three or more. By contrast, one music or news service is normally enough, as rival services often offer broadly similar content. But as we noted earlier, in some countries we are seeing more differentiated news services and a consequent rise in the number of subscriptions per person. The other big difference is the age profile, with TV, music, and audio books skewing younger, while sport has an equal age split and news skews much older. People who subscribe to news are more likely to subscribe to other services and vice versa.\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 20","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Digital_News-Report_2022_p20.pdf"]},"response":{"errors":[],"extract_id":"extract_ab565326ab51d2d79b307f8b5b6ff2c6","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["In almost all of the listed countries, the majority of subscribers pay for one publication. But in the United States and Australia, around half (56% and 51% respectively) now pay for _two or more_ – often a national and local paper combination. Second subscriptions in the United States include political and cultural magazines such as _The Atlantic_ and _The New Yorker_ , partisan digital outlets such as Blaze Media and Epoch Times, or passion-based titles such as the Athletic (sports). We also see growing levels of payment for platform-based news subscription products Apple+ and Twitter Blue. In Australia, second subscriptions include a relatively high proportion (9%) from US or UK publications such as the _New York Times_ and _The Times_ of London.\n\n**The influence of individual journalist brands may be overstated**\n\nSome news brands continue to have the reputation, scale, and technical capability to charge for their services, but in parallel the emergence of low-cost subscription and membership platforms, like Substack and Patreon, has enabled some individual journalists, podcasters, or other creators to also convince people to pay for their work. Substack says that there are now around a million paying subscribers to its premium content emails. ⁵\n\nDespite this, our data show the number of these individual subscriptions is still relatively modest, certainly when compared with established news brands. Even in the United States, we find that just 7% of news subscribers in the United States pay for one or more journalist emails – around 1% of our overall sample. The proportions are even lower elsewhere, though we do see some emerging payment for individual email writers or podcasters in parts of Europe and Australia.\n\nOver time, it is possible that these solo or small businesses, mostly operating in closely defined niches, could make up a far bigger part of the overall pie, expanding the market but also creating more competition for big news brands. If nothing else, they have already made the value of some individual journalists more explicit, pushing up wages of top talent.\n\nThese data also highlight the growing importance of formats like email and podcasts, both of which have become much more important in driving regular engagement for all publishers – from general news to specialist information and entertainment.\n\n_For further analysis see Section 2.4: Email News: Its Contribution to Engagement and Monetisation_\n\n5 https://www.forbes.com/sites/marisadellatto/2021/11/15/substack-announces-one-million-paid-subscribers/\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 20\n\nPROPORTION OF SUBSCRIBERS WHO PAY FOR INDIVIDUAL JOURNALISTS VIA EMAIL OR OTHER SERVICES – SELECTED COUNTRIES\n\n**Q7\\_SUBS\\_name\\_2022.** You said you have paid a subscription/membership to a digital news service in the last year … Which of the following did you subscribe to? _Base: All who made a regular payment to at least one digital news service in last year. USA = 304, Spain = 153, Belgium = 211, Netherlands = 221, Sweden = 487, France = 135, Germany = 183, Australia = 270, UK = 160._\n\n0%\n\n25%\n\n50%\n\n75%\n\n100%\n\n57\n\n29\n\n47 37 42 41 38 44 34\n\n46\n\n55+\n\n45-54\n\n35-44\n\n25-34\n\n18-24\n\nThey should be able to express their personal opinions alongside reporting the news\n\nThey should stick to reporting the news\n\n0%\n\n2%\n\n4%\n\n6%\n\n8%\n\n10%\n\n4\n\n5\n\n3\n\n2\n\n3\n\n1\n\n3\n\n2\n\n1\n\n7\n\n5\n\n4\n\n3\n\n2 2\n\n1 1 0\n\nJournalist via YouTube/podcast\n\nJournalist via Newsletter/website\n\nUK Germany Australia France Belgium Netherlands Sweden Spain USA\n\nJournalist via newsletter/website\n\nJournalist via YouTube/podcast\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 21\n\nPROPORTION WHO SUBSCRIBE TO DIFFERENT MEDIA SERVICES – UK\n\n**Q\\_Sub\\_fatigue1.** There are many different online media subscriptions you can pay for these days. How many of the following online media subscriptions do you currently have, if any, that you pay for with your own money? **Q7\\_SUBS.** You said you have paid a subscription or made an ongoing donation to a digital news service in the last year. How many different news providers do you regularly pay money to? _Base: UK = 2410._\n\nKey: No Sub 1 2 3 4 5 or more\n\n35 25 20 13 4\n\n**TV/Movies** (e.g., Netflix, Amazon Prime Video, Disney+) **65% pay**\n\n63 30 3 2\n\n**Music** (e.g., Spotify, Apple Music, Pandora, Deezer, etc.) **37% pay**\n\n78 14 5 2\n\n**Sports** (e.g., Sky Sports, BT Sport, etc.) **22% pay**\n\n\\* Note: 7% pay for a news subscription or other ongoing subscription, 9% for any payment incl. one-off payments and donations.\n\n93 6\n\n**News** (e.g., Telegraph, Times, Guardian, etc.) **7% pay\\***\n\n0% 25% 50% 75% 100%\n\nIn the vast majority of markets in our survey, people say they still identify most with traditional news brands rather than individual journalists. This is especially true in Nordic countries but less true in the United States, Southern Europe, Latin America, and parts of Asia. Perhaps the greater affinity with journalists in these countries offers a different opportunity to monetise content than brand-led subscription.\n\n**The impact of wider media subscriptions and prospects for subscription fatigue**\n\nWith rising energy prices and inflation, and with many different types of online media (TV, music, books) now competing for a share of household budgets, there are genuine fears in the industry about whether recent growth in news subscriptions can continue. Against this backdrop, we explored the characteristics of different media subscriptions and asked respondents whether they were thinking of taking out more – or cutting back.\n\nTaking the UK as an example of an advanced subscription market (chart below), we find almost two-thirds of our sample (65%) have taken out at least one subscription to a TV or film service like Netflix, 37% have a music subscription to Spotify or similar, 22% pay for a premium sports service, and 7% have a news subscription or other ongoing payment such as a donation. Because premium content is spread across TV platforms, many people feel the need to take out more than one subscription, with 19% taking out three or more. By contrast, one music or news service is normally enough, as rival services often offer broadly similar content. But as we noted earlier, in some countries we are seeing more differentiated news services and a consequent rise in the number of subscriptions per person. The other big difference is the age profile, with TV, music, and audio books skewing younger, while sport has an equal age split and news skews much older. People who subscribe to news are more likely to subscribe to other services and vice versa.\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 20"],"full_content":"In almost all of the listed countries, the majority of subscribers pay for one publication. But in the United States and Australia, around half (56% and 51% respectively) now pay for _two or more_ – often a national and local paper combination. Second subscriptions in the United States include political and cultural magazines such as _The Atlantic_ and _The New Yorker_ , partisan digital outlets such as Blaze Media and Epoch Times, or passion-based titles such as the Athletic (sports). We also see growing levels of payment for platform-based news subscription products Apple+ and Twitter Blue. In Australia, second subscriptions include a relatively high proportion (9%) from US or UK publications such as the _New York Times_ and _The Times_ of London.\n\n**The influence of individual journalist brands may be overstated**\n\nSome news brands continue to have the reputation, scale, and technical capability to charge for their services, but in parallel the emergence of low-cost subscription and membership platforms, like Substack and Patreon, has enabled some individual journalists, podcasters, or other creators to also convince people to pay for their work. Substack says that there are now around a million paying subscribers to its premium content emails. ⁵\n\nDespite this, our data show the number of these individual subscriptions is still relatively modest, certainly when compared with established news brands. Even in the United States, we find that just 7% of news subscribers in the United States pay for one or more journalist emails – around 1% of our overall sample. The proportions are even lower elsewhere, though we do see some emerging payment for individual email writers or podcasters in parts of Europe and Australia.\n\nOver time, it is possible that these solo or small businesses, mostly operating in closely defined niches, could make up a far bigger part of the overall pie, expanding the market but also creating more competition for big news brands. If nothing else, they have already made the value of some individual journalists more explicit, pushing up wages of top talent.\n\nThese data also highlight the growing importance of formats like email and podcasts, both of which have become much more important in driving regular engagement for all publishers – from general news to specialist information and entertainment.\n\n_For further analysis see Section 2.4: Email News: Its Contribution to Engagement and Monetisation_\n\n5 https://www.forbes.com/sites/marisadellatto/2021/11/15/substack-announces-one-million-paid-subscribers/\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 20\n\nPROPORTION OF SUBSCRIBERS WHO PAY FOR INDIVIDUAL JOURNALISTS VIA EMAIL OR OTHER SERVICES – SELECTED COUNTRIES\n\n**Q7\\_SUBS\\_name\\_2022.** You said you have paid a subscription/membership to a digital news service in the last year … Which of the following did you subscribe to? _Base: All who made a regular payment to at least one digital news service in last year. USA = 304, Spain = 153, Belgium = 211, Netherlands = 221, Sweden = 487, France = 135, Germany = 183, Australia = 270, UK = 160._\n\n0%\n\n25%\n\n50%\n\n75%\n\n100%\n\n57\n\n29\n\n47 37 42 41 38 44 34\n\n46\n\n55+\n\n45-54\n\n35-44\n\n25-34\n\n18-24\n\nThey should be able to express their personal opinions alongside reporting the news\n\nThey should stick to reporting the news\n\n0%\n\n2%\n\n4%\n\n6%\n\n8%\n\n10%\n\n4\n\n5\n\n3\n\n2\n\n3\n\n1\n\n3\n\n2\n\n1\n\n7\n\n5\n\n4\n\n3\n\n2 2\n\n1 1 0\n\nJournalist via YouTube/podcast\n\nJournalist via Newsletter/website\n\nUK Germany Australia France Belgium Netherlands Sweden Spain USA\n\nJournalist via newsletter/website\n\nJournalist via YouTube/podcast\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 21\n\nPROPORTION WHO SUBSCRIBE TO DIFFERENT MEDIA SERVICES – UK\n\n**Q\\_Sub\\_fatigue1.** There are many different online media subscriptions you can pay for these days. How many of the following online media subscriptions do you currently have, if any, that you pay for with your own money? **Q7\\_SUBS.** You said you have paid a subscription or made an ongoing donation to a digital news service in the last year. How many different news providers do you regularly pay money to? _Base: UK = 2410._\n\nKey: No Sub 1 2 3 4 5 or more\n\n35 25 20 13 4\n\n**TV/Movies** (e.g., Netflix, Amazon Prime Video, Disney+) **65% pay**\n\n63 30 3 2\n\n**Music** (e.g., Spotify, Apple Music, Pandora, Deezer, etc.) **37% pay**\n\n78 14 5 2\n\n**Sports** (e.g., Sky Sports, BT Sport, etc.) **22% pay**\n\n\\* Note: 7% pay for a news subscription or other ongoing subscription, 9% for any payment incl. one-off payments and donations.\n\n93 6\n\n**News** (e.g., Telegraph, Times, Guardian, etc.) **7% pay\\***\n\n0% 25% 50% 75% 100%\n\nIn the vast majority of markets in our survey, people say they still identify most with traditional news brands rather than individual journalists. This is especially true in Nordic countries but less true in the United States, Southern Europe, Latin America, and parts of Asia. Perhaps the greater affinity with journalists in these countries offers a different opportunity to monetise content than brand-led subscription.\n\n**The impact of wider media subscriptions and prospects for subscription fatigue**\n\nWith rising energy prices and inflation, and with many different types of online media (TV, music, books) now competing for a share of household budgets, there are genuine fears in the industry about whether recent growth in news subscriptions can continue. Against this backdrop, we explored the characteristics of different media subscriptions and asked respondents whether they were thinking of taking out more – or cutting back.\n\nTaking the UK as an example of an advanced subscription market (chart below), we find almost two-thirds of our sample (65%) have taken out at least one subscription to a TV or film service like Netflix, 37% have a music subscription to Spotify or similar, 22% pay for a premium sports service, and 7% have a news subscription or other ongoing payment such as a donation. Because premium content is spread across TV platforms, many people feel the need to take out more than one subscription, with 19% taking out three or more. By contrast, one music or news service is normally enough, as rival services often offer broadly similar content. But as we noted earlier, in some countries we are seeing more differentiated news services and a consequent rise in the number of subscriptions per person. The other big difference is the age profile, with TV, music, and audio books skewing younger, while sport has an equal age split and news skews much older. People who subscribe to news are more likely to subscribe to other services and vice versa.\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 20","publish_date":null,"title":"Graph 21 Graph 20","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Digital_News-Report_2022_p20.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Digital_News-Report_2022_p20.pdf"},"request":{"example_id":"chart/Digital_News-Report_2022_p20","product_type":"parse","source_file_path":"Digital_News-Report_2022_p20.pdf"},"started_at":"2026-11-04T19:46:07.853500"} +{"completed_at":"2026-11-04T19:46:14.096376","latency_in_ms":4720,"output":{"example_id":"chart/Digital_News-Report_2022_p21","grounded_pages":[],"layout_pages":[],"markdown":"6 https://edition.cnn.com/2022/04/21/media/cnn-shutting-down/index.html\n\nThinking about the year ahead, most respondents say the number of media subscriptions they have (across TV, music, sports, books, and news) will stay the same, but there is an appetite to take out more in Germany, Austria, and Switzerland. By contrast, the Portuguese are most likely to be looking to _reduce_ their media subscriptions this year, while respondents in the UK and US are evenly split.\n\nThis overall picture is perhaps more optimistic than might be expected, with relatively low subscription churn, providing publishers with a more stable and secure income than the frequent ups and downs of the advertising market. On the other hand, the cost-of-living crisis has been building since our survey in January and will clearly be a significant factor for those on lower incomes.\n\n**_Energy costs, along with petrol and food prices are all rising. Luxuries will need to be cut to free up more money._** _Female, 57, UK_\n\n**_I’m on a tight budget and with soaring inflation and gas prices going up and groceries prices are outrageous … I have no extra money for media subscriptions._** _Female, 62, USA_\n\nIt’s not entirely clear whether news will be treated in the same bucket as entertainment subscriptions that could be seen as more of a luxury. The news demographic tends to be older and richer, perhaps making the group less affected by rising prices. On the other hand, the sector is unlikely to be immune from these trends and a number of respondents said they would be cancelling subscriptions to news sites because they were too expensive. Others say they are planning to increase the number of subscriptions this year, with some seeing the current national and international instability as a good reason to invest more in high-quality information.\n\n**_I feel that this time is the most important time to know as much news and information about our surroundings as possible._** _Male, 33, USA_\n\nInvestment in new products may encourage further take-up and much of the news industry remains confident about longer-term trends. But some publishers, notably CNN, recently reversed its much-hyped online paywall strategy (CNN+) amid doubts from its new owners about the viability of yet another stand-alone subscription. ⁶ Uncertainty remains about how far and fast the market can grow, especially in more troubled economic times.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 22\n\nPROPORTION WHO SUBSCRIBE TO EACH TYPE OF MEDIA SERVICE – BY AGE – UK\n\n**Q\\_Sub\\_fatigue2.** There are many different online media subscriptions you can pay for these days. How many of the following online media subscriptions do you currently have, if any, that you pay for with your own money? **Q7ai.** Which, if any, of the following ways have you used to pay for online news content in the last year? _Base: UK = 2410._\n\n18-24 25-34 35-44 45-54 55+\n\n0% 25% 50% 75% 100%\n\n49\n\n15\n\n5\n\n21\n\n9\n\n73\n\n38\n\n12\n\n21\n\n8\n\n78\n\n44\n\n13\n\n26\n\n4\n\n81\n\n61\n\n18\n\n19\n\n6\n\n75\n\n75\n\n21\n\n21\n\n4\n\n**TV/movies streaming**\n\n**Music**\n\n**Sports**\n\n**News**\n\n**Audiobooks/ podcasts**\n\nSkews older\n\nSkews younger\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 23\n\nPROPORTION WHO THINK THEY WILL HAVE MORE/FEWER MEDIA SUBSCRIPTIONS NEXT YEAR – SELECTED MARKETS\n\n31 60 9 Austria\n\n24 67 9 Switzerland\n\n21 71 8 Germany\n\n14 72 14 USA\n\n13 70 17 Portugal\n\n10 82 9 UK\n\n**Q\\_Sub\\_fatigue2a.** In the next year, do you expect the number of online media subscriptions you pay for with your own money to increase, decrease, or stay the same? _Base: Total sample in each market. Note: Don’t knows were excluded._\n\n0% 25% 50% 75% 100%\n\nMore About the same Fewer\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 21","pages":[{"line_numbers":[],"markdown":"6 https://edition.cnn.com/2022/04/21/media/cnn-shutting-down/index.html\n\nThinking about the year ahead, most respondents say the number of media subscriptions they have (across TV, music, sports, books, and news) will stay the same, but there is an appetite to take out more in Germany, Austria, and Switzerland. By contrast, the Portuguese are most likely to be looking to _reduce_ their media subscriptions this year, while respondents in the UK and US are evenly split.\n\nThis overall picture is perhaps more optimistic than might be expected, with relatively low subscription churn, providing publishers with a more stable and secure income than the frequent ups and downs of the advertising market. On the other hand, the cost-of-living crisis has been building since our survey in January and will clearly be a significant factor for those on lower incomes.\n\n**_Energy costs, along with petrol and food prices are all rising. Luxuries will need to be cut to free up more money._** _Female, 57, UK_\n\n**_I’m on a tight budget and with soaring inflation and gas prices going up and groceries prices are outrageous … I have no extra money for media subscriptions._** _Female, 62, USA_\n\nIt’s not entirely clear whether news will be treated in the same bucket as entertainment subscriptions that could be seen as more of a luxury. The news demographic tends to be older and richer, perhaps making the group less affected by rising prices. On the other hand, the sector is unlikely to be immune from these trends and a number of respondents said they would be cancelling subscriptions to news sites because they were too expensive. Others say they are planning to increase the number of subscriptions this year, with some seeing the current national and international instability as a good reason to invest more in high-quality information.\n\n**_I feel that this time is the most important time to know as much news and information about our surroundings as possible._** _Male, 33, USA_\n\nInvestment in new products may encourage further take-up and much of the news industry remains confident about longer-term trends. But some publishers, notably CNN, recently reversed its much-hyped online paywall strategy (CNN+) amid doubts from its new owners about the viability of yet another stand-alone subscription. ⁶ Uncertainty remains about how far and fast the market can grow, especially in more troubled economic times.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 22\n\nPROPORTION WHO SUBSCRIBE TO EACH TYPE OF MEDIA SERVICE – BY AGE – UK\n\n**Q\\_Sub\\_fatigue2.** There are many different online media subscriptions you can pay for these days. How many of the following online media subscriptions do you currently have, if any, that you pay for with your own money? **Q7ai.** Which, if any, of the following ways have you used to pay for online news content in the last year? _Base: UK = 2410._\n\n18-24 25-34 35-44 45-54 55+\n\n0% 25% 50% 75% 100%\n\n49\n\n15\n\n5\n\n21\n\n9\n\n73\n\n38\n\n12\n\n21\n\n8\n\n78\n\n44\n\n13\n\n26\n\n4\n\n81\n\n61\n\n18\n\n19\n\n6\n\n75\n\n75\n\n21\n\n21\n\n4\n\n**TV/movies streaming**\n\n**Music**\n\n**Sports**\n\n**News**\n\n**Audiobooks/ podcasts**\n\nSkews older\n\nSkews younger\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 23\n\nPROPORTION WHO THINK THEY WILL HAVE MORE/FEWER MEDIA SUBSCRIPTIONS NEXT YEAR – SELECTED MARKETS\n\n31 60 9 Austria\n\n24 67 9 Switzerland\n\n21 71 8 Germany\n\n14 72 14 USA\n\n13 70 17 Portugal\n\n10 82 9 UK\n\n**Q\\_Sub\\_fatigue2a.** In the next year, do you expect the number of online media subscriptions you pay for with your own money to increase, decrease, or stay the same? _Base: Total sample in each market. Note: Don’t knows were excluded._\n\n0% 25% 50% 75% 100%\n\nMore About the same Fewer\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 21","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"6 https://edition.cnn.com/2022/04/21/media/cnn-shutting-down/index.html\n\nThinking about the year ahead, most respondents say the number of media subscriptions they have (across TV, music, sports, books, and news) will stay the same, but there is an appetite to take out more in Germany, Austria, and Switzerland. By contrast, the Portuguese are most likely to be looking to _reduce_ their media subscriptions this year, while respondents in the UK and US are evenly split.\n\nThis overall picture is perhaps more optimistic than might be expected, with relatively low subscription churn, providing publishers with a more stable and secure income than the frequent ups and downs of the advertising market. On the other hand, the cost-of-living crisis has been building since our survey in January and will clearly be a significant factor for those on lower incomes.\n\n**_Energy costs, along with petrol and food prices are all rising. Luxuries will need to be cut to free up more money._** _Female, 57, UK_\n\n**_I’m on a tight budget and with soaring inflation and gas prices going up and groceries prices are outrageous … I have no extra money for media subscriptions._** _Female, 62, USA_\n\nIt’s not entirely clear whether news will be treated in the same bucket as entertainment subscriptions that could be seen as more of a luxury. The news demographic tends to be older and richer, perhaps making the group less affected by rising prices. On the other hand, the sector is unlikely to be immune from these trends and a number of respondents said they would be cancelling subscriptions to news sites because they were too expensive. Others say they are planning to increase the number of subscriptions this year, with some seeing the current national and international instability as a good reason to invest more in high-quality information.\n\n**_I feel that this time is the most important time to know as much news and information about our surroundings as possible._** _Male, 33, USA_\n\nInvestment in new products may encourage further take-up and much of the news industry remains confident about longer-term trends. But some publishers, notably CNN, recently reversed its much-hyped online paywall strategy (CNN+) amid doubts from its new owners about the viability of yet another stand-alone subscription. ⁶ Uncertainty remains about how far and fast the market can grow, especially in more troubled economic times.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 22\n\nPROPORTION WHO SUBSCRIBE TO EACH TYPE OF MEDIA SERVICE – BY AGE – UK\n\n**Q\\_Sub\\_fatigue2.** There are many different online media subscriptions you can pay for these days. How many of the following online media subscriptions do you currently have, if any, that you pay for with your own money? **Q7ai.** Which, if any, of the following ways have you used to pay for online news content in the last year? _Base: UK = 2410._\n\n18-24 25-34 35-44 45-54 55+\n\n0% 25% 50% 75% 100%\n\n49\n\n15\n\n5\n\n21\n\n9\n\n73\n\n38\n\n12\n\n21\n\n8\n\n78\n\n44\n\n13\n\n26\n\n4\n\n81\n\n61\n\n18\n\n19\n\n6\n\n75\n\n75\n\n21\n\n21\n\n4\n\n**TV/movies streaming**\n\n**Music**\n\n**Sports**\n\n**News**\n\n**Audiobooks/ podcasts**\n\nSkews older\n\nSkews younger\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 23\n\nPROPORTION WHO THINK THEY WILL HAVE MORE/FEWER MEDIA SUBSCRIPTIONS NEXT YEAR – SELECTED MARKETS\n\n31 60 9 Austria\n\n24 67 9 Switzerland\n\n21 71 8 Germany\n\n14 72 14 USA\n\n13 70 17 Portugal\n\n10 82 9 UK\n\n**Q\\_Sub\\_fatigue2a.** In the next year, do you expect the number of online media subscriptions you pay for with your own money to increase, decrease, or stay the same? _Base: Total sample in each market. Note: Don’t knows were excluded._\n\n0% 25% 50% 75% 100%\n\nMore About the same Fewer\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 21","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Digital_News-Report_2022_p21.pdf"]},"response":{"errors":[],"extract_id":"extract_f9737813a954f417953e75bf14f0adbf","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["6 https://edition.cnn.com/2022/04/21/media/cnn-shutting-down/index.html\n\nThinking about the year ahead, most respondents say the number of media subscriptions they have (across TV, music, sports, books, and news) will stay the same, but there is an appetite to take out more in Germany, Austria, and Switzerland. By contrast, the Portuguese are most likely to be looking to _reduce_ their media subscriptions this year, while respondents in the UK and US are evenly split.\n\nThis overall picture is perhaps more optimistic than might be expected, with relatively low subscription churn, providing publishers with a more stable and secure income than the frequent ups and downs of the advertising market. On the other hand, the cost-of-living crisis has been building since our survey in January and will clearly be a significant factor for those on lower incomes.\n\n**_Energy costs, along with petrol and food prices are all rising. Luxuries will need to be cut to free up more money._** _Female, 57, UK_\n\n**_I’m on a tight budget and with soaring inflation and gas prices going up and groceries prices are outrageous … I have no extra money for media subscriptions._** _Female, 62, USA_\n\nIt’s not entirely clear whether news will be treated in the same bucket as entertainment subscriptions that could be seen as more of a luxury. The news demographic tends to be older and richer, perhaps making the group less affected by rising prices. On the other hand, the sector is unlikely to be immune from these trends and a number of respondents said they would be cancelling subscriptions to news sites because they were too expensive. Others say they are planning to increase the number of subscriptions this year, with some seeing the current national and international instability as a good reason to invest more in high-quality information.\n\n**_I feel that this time is the most important time to know as much news and information about our surroundings as possible._** _Male, 33, USA_\n\nInvestment in new products may encourage further take-up and much of the news industry remains confident about longer-term trends. But some publishers, notably CNN, recently reversed its much-hyped online paywall strategy (CNN+) amid doubts from its new owners about the viability of yet another stand-alone subscription. ⁶ Uncertainty remains about how far and fast the market can grow, especially in more troubled economic times.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 22\n\nPROPORTION WHO SUBSCRIBE TO EACH TYPE OF MEDIA SERVICE – BY AGE – UK\n\n**Q\\_Sub\\_fatigue2.** There are many different online media subscriptions you can pay for these days. How many of the following online media subscriptions do you currently have, if any, that you pay for with your own money? **Q7ai.** Which, if any, of the following ways have you used to pay for online news content in the last year? _Base: UK = 2410._\n\n18-24 25-34 35-44 45-54 55+\n\n0% 25% 50% 75% 100%\n\n49\n\n15\n\n5\n\n21\n\n9\n\n73\n\n38\n\n12\n\n21\n\n8\n\n78\n\n44\n\n13\n\n26\n\n4\n\n81\n\n61\n\n18\n\n19\n\n6\n\n75\n\n75\n\n21\n\n21\n\n4\n\n**TV/movies streaming**\n\n**Music**\n\n**Sports**\n\n**News**\n\n**Audiobooks/ podcasts**\n\nSkews older\n\nSkews younger\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 23\n\nPROPORTION WHO THINK THEY WILL HAVE MORE/FEWER MEDIA SUBSCRIPTIONS NEXT YEAR – SELECTED MARKETS\n\n31 60 9 Austria\n\n24 67 9 Switzerland\n\n21 71 8 Germany\n\n14 72 14 USA\n\n13 70 17 Portugal\n\n10 82 9 UK\n\n**Q\\_Sub\\_fatigue2a.** In the next year, do you expect the number of online media subscriptions you pay for with your own money to increase, decrease, or stay the same? _Base: Total sample in each market. Note: Don’t knows were excluded._\n\n0% 25% 50% 75% 100%\n\nMore About the same Fewer\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 21"],"full_content":"6 https://edition.cnn.com/2022/04/21/media/cnn-shutting-down/index.html\n\nThinking about the year ahead, most respondents say the number of media subscriptions they have (across TV, music, sports, books, and news) will stay the same, but there is an appetite to take out more in Germany, Austria, and Switzerland. By contrast, the Portuguese are most likely to be looking to _reduce_ their media subscriptions this year, while respondents in the UK and US are evenly split.\n\nThis overall picture is perhaps more optimistic than might be expected, with relatively low subscription churn, providing publishers with a more stable and secure income than the frequent ups and downs of the advertising market. On the other hand, the cost-of-living crisis has been building since our survey in January and will clearly be a significant factor for those on lower incomes.\n\n**_Energy costs, along with petrol and food prices are all rising. Luxuries will need to be cut to free up more money._** _Female, 57, UK_\n\n**_I’m on a tight budget and with soaring inflation and gas prices going up and groceries prices are outrageous … I have no extra money for media subscriptions._** _Female, 62, USA_\n\nIt’s not entirely clear whether news will be treated in the same bucket as entertainment subscriptions that could be seen as more of a luxury. The news demographic tends to be older and richer, perhaps making the group less affected by rising prices. On the other hand, the sector is unlikely to be immune from these trends and a number of respondents said they would be cancelling subscriptions to news sites because they were too expensive. Others say they are planning to increase the number of subscriptions this year, with some seeing the current national and international instability as a good reason to invest more in high-quality information.\n\n**_I feel that this time is the most important time to know as much news and information about our surroundings as possible._** _Male, 33, USA_\n\nInvestment in new products may encourage further take-up and much of the news industry remains confident about longer-term trends. But some publishers, notably CNN, recently reversed its much-hyped online paywall strategy (CNN+) amid doubts from its new owners about the viability of yet another stand-alone subscription. ⁶ Uncertainty remains about how far and fast the market can grow, especially in more troubled economic times.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 22\n\nPROPORTION WHO SUBSCRIBE TO EACH TYPE OF MEDIA SERVICE – BY AGE – UK\n\n**Q\\_Sub\\_fatigue2.** There are many different online media subscriptions you can pay for these days. How many of the following online media subscriptions do you currently have, if any, that you pay for with your own money? **Q7ai.** Which, if any, of the following ways have you used to pay for online news content in the last year? _Base: UK = 2410._\n\n18-24 25-34 35-44 45-54 55+\n\n0% 25% 50% 75% 100%\n\n49\n\n15\n\n5\n\n21\n\n9\n\n73\n\n38\n\n12\n\n21\n\n8\n\n78\n\n44\n\n13\n\n26\n\n4\n\n81\n\n61\n\n18\n\n19\n\n6\n\n75\n\n75\n\n21\n\n21\n\n4\n\n**TV/movies streaming**\n\n**Music**\n\n**Sports**\n\n**News**\n\n**Audiobooks/ podcasts**\n\nSkews older\n\nSkews younger\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 23\n\nPROPORTION WHO THINK THEY WILL HAVE MORE/FEWER MEDIA SUBSCRIPTIONS NEXT YEAR – SELECTED MARKETS\n\n31 60 9 Austria\n\n24 67 9 Switzerland\n\n21 71 8 Germany\n\n14 72 14 USA\n\n13 70 17 Portugal\n\n10 82 9 UK\n\n**Q\\_Sub\\_fatigue2a.** In the next year, do you expect the number of online media subscriptions you pay for with your own money to increase, decrease, or stay the same? _Base: Total sample in each market. Note: Don’t knows were excluded._\n\n0% 25% 50% 75% 100%\n\nMore About the same Fewer\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 21","publish_date":null,"title":"Graph 23 Graph 22","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Digital_News-Report_2022_p21.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Digital_News-Report_2022_p21.pdf"},"request":{"example_id":"chart/Digital_News-Report_2022_p21","product_type":"parse","source_file_path":"Digital_News-Report_2022_p21.pdf"},"started_at":"2026-11-04T19:46:09.375690"} +{"completed_at":"2026-11-04T19:46:14.266388","latency_in_ms":2594,"output":{"example_id":"chart/Digital_News-Report_2022_p23","grounded_pages":[],"layout_pages":[],"markdown":"REUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 28\n\nPROPORTION WHO SAY EACH ACCESS POINT IS THEIR MAIN GATEWAY TO ONLINE NEWS (2018–22) – ALL MARKETS\n\n**Q10a.** Thinking about how you got news online (via computer, mobile, or any device) in the last week, which of these was the \\*\\*MAIN\\*\\* way in which you came across news? _Base: All markets 2018–22 and total sample in each market (n ≈ 2000). Note: Number of markets grew from 37 in 2018 to 46 in 2021–22._\n\n32%\n\n23% 23%\n\n28%\n\n0%\n\n25%\n\n50%\n\n2022 2021 2020 2019 2018\n\nDirect access to websites/apps Social media access\n\n**MAINLY DIRECT**\n\nFinland 65%\n\nNorway 59%\n\nUK 46%\n\n**MUCH LESS DIRECT**\n\nSouth Korea 5%\n\nBrazil 11%\n\nMexico 11%\n\n**MAINLY SOCIAL**\n\nThailand 59%\n\nPhilippines 51%\n\nKenya 51%\n\n**MUCH LESS SOCIAL**\n\nJapan 9%\n\nSweden 10%\n\nFinland 11%\n\nThis year, we revived a question about how different groups access news _first thing in the morning_ to give a sense of how news habits have changed since 2019. The next chart shows striking differences between countries. In Norway, Spain, Finland, and the UK, the smartphone is now the go-to route to news in the morning, with dependence on it growing substantially in the last three years. Radio remains an important part of morning routines in Ireland – even though it has been overtaken by the smartphone for the first time. Morning newspaper reading is still surprisingly popular in the Netherlands and Finland. Television remains a key influence in the US, France, and Italy and is the dominant medium in Japan.\n\nAcross countries, almost half of under 35s (47%) say they accessed news first using a smartphone, compared with just a quarter (28%) of those aged 35 and older and just 15% of those aged 65 and older. Japan’s older population may explain why it is bucking wider trends.\n\nBut what channels do people use when accessing news on a smartphone at the start of each day? Since 2019, we find that, across countries, social media (39%) have overtaken websites and apps (31%), followed by the smartphone home screen (12%) and aggregator apps like Apple News (9%).\n\n**Main gateways to digital news**\n\nWe continue to monitor the main access points to online news, not just via a smartphone and not just first thing in the morning. Across all devices, our data show direct access to apps and websites becoming less important over time and social media becoming more important, partly due to their ubiquity and convenience. At an aggregate level, we have reached something of a tipping point this year, with social media preference (28%) surging ahead of direct access (23%). But these are cross-market averages and there are major differences between markets. Audiences in Nordic countries and the UK still have strong direct connections, while people in Japan and South Korea tend to access news via powerful aggregators and search engines – relying less on direct access. These differences, which we can perhaps treat as a proxy for the strength of news brands in a digital world, help explain why it may be easier to charge for online news in some countries and not others.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 27\n\n0% 25% 50%\n\n43\n\n39\n\n38\n\n36\n\n35\n\n35\n\n26\n\n29\n\n25\n\n24\n\n13\n\n24\n\n32\n\n21\n\n15\n\n21\n\n24\n\n29\n\n27\n\n45\n\n16\n\n15\n\n11\n\n12\n\n31\n\n17\n\n14\n\n7\n\n23\n\n3\n\n5\n\n3\n\n3\n\n10\n\n3\n\n4\n\n13\n\n3\n\n3\n\n8\n\nJapan\n\nFrance\n\nUSA\n\nNetherlands\n\nUK\n\nIreland\n\nFinland\n\nItaly\n\nSpain\n\nNorway\n\nPROPORTION WHO SAY EACH DEVICE IS THEIR FIRST WAY OF ACCESSING NEWS IN THE MORNING – SELECTED MARKETS\n\n**Q9c\\_new2016.** What is the FIRST way you typically come across news in the morning? _Base: Total sample in each market (n ≈ 2000)._\n\nPrint Radio TV Smartphone\n\nWas 44% in 2019\n\nWas 27% in 2019\n\nWas 28% in 2019\n\nWas 31% in 2019\n\nWas 27% in 2019\n\nWas 26% in 2019\n\nWas 32% in 2019\n\n**TV first**\n\n**Mobile**\n\n**first**\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 23","pages":[{"line_numbers":[],"markdown":"REUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 28\n\nPROPORTION WHO SAY EACH ACCESS POINT IS THEIR MAIN GATEWAY TO ONLINE NEWS (2018–22) – ALL MARKETS\n\n**Q10a.** Thinking about how you got news online (via computer, mobile, or any device) in the last week, which of these was the \\*\\*MAIN\\*\\* way in which you came across news? _Base: All markets 2018–22 and total sample in each market (n ≈ 2000). Note: Number of markets grew from 37 in 2018 to 46 in 2021–22._\n\n32%\n\n23% 23%\n\n28%\n\n0%\n\n25%\n\n50%\n\n2022 2021 2020 2019 2018\n\nDirect access to websites/apps Social media access\n\n**MAINLY DIRECT**\n\nFinland 65%\n\nNorway 59%\n\nUK 46%\n\n**MUCH LESS DIRECT**\n\nSouth Korea 5%\n\nBrazil 11%\n\nMexico 11%\n\n**MAINLY SOCIAL**\n\nThailand 59%\n\nPhilippines 51%\n\nKenya 51%\n\n**MUCH LESS SOCIAL**\n\nJapan 9%\n\nSweden 10%\n\nFinland 11%\n\nThis year, we revived a question about how different groups access news _first thing in the morning_ to give a sense of how news habits have changed since 2019. The next chart shows striking differences between countries. In Norway, Spain, Finland, and the UK, the smartphone is now the go-to route to news in the morning, with dependence on it growing substantially in the last three years. Radio remains an important part of morning routines in Ireland – even though it has been overtaken by the smartphone for the first time. Morning newspaper reading is still surprisingly popular in the Netherlands and Finland. Television remains a key influence in the US, France, and Italy and is the dominant medium in Japan.\n\nAcross countries, almost half of under 35s (47%) say they accessed news first using a smartphone, compared with just a quarter (28%) of those aged 35 and older and just 15% of those aged 65 and older. Japan’s older population may explain why it is bucking wider trends.\n\nBut what channels do people use when accessing news on a smartphone at the start of each day? Since 2019, we find that, across countries, social media (39%) have overtaken websites and apps (31%), followed by the smartphone home screen (12%) and aggregator apps like Apple News (9%).\n\n**Main gateways to digital news**\n\nWe continue to monitor the main access points to online news, not just via a smartphone and not just first thing in the morning. Across all devices, our data show direct access to apps and websites becoming less important over time and social media becoming more important, partly due to their ubiquity and convenience. At an aggregate level, we have reached something of a tipping point this year, with social media preference (28%) surging ahead of direct access (23%). But these are cross-market averages and there are major differences between markets. Audiences in Nordic countries and the UK still have strong direct connections, while people in Japan and South Korea tend to access news via powerful aggregators and search engines – relying less on direct access. These differences, which we can perhaps treat as a proxy for the strength of news brands in a digital world, help explain why it may be easier to charge for online news in some countries and not others.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 27\n\n0% 25% 50%\n\n43\n\n39\n\n38\n\n36\n\n35\n\n35\n\n26\n\n29\n\n25\n\n24\n\n13\n\n24\n\n32\n\n21\n\n15\n\n21\n\n24\n\n29\n\n27\n\n45\n\n16\n\n15\n\n11\n\n12\n\n31\n\n17\n\n14\n\n7\n\n23\n\n3\n\n5\n\n3\n\n3\n\n10\n\n3\n\n4\n\n13\n\n3\n\n3\n\n8\n\nJapan\n\nFrance\n\nUSA\n\nNetherlands\n\nUK\n\nIreland\n\nFinland\n\nItaly\n\nSpain\n\nNorway\n\nPROPORTION WHO SAY EACH DEVICE IS THEIR FIRST WAY OF ACCESSING NEWS IN THE MORNING – SELECTED MARKETS\n\n**Q9c\\_new2016.** What is the FIRST way you typically come across news in the morning? _Base: Total sample in each market (n ≈ 2000)._\n\nPrint Radio TV Smartphone\n\nWas 44% in 2019\n\nWas 27% in 2019\n\nWas 28% in 2019\n\nWas 31% in 2019\n\nWas 27% in 2019\n\nWas 26% in 2019\n\nWas 32% in 2019\n\n**TV first**\n\n**Mobile**\n\n**first**\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 23","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"REUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 28\n\nPROPORTION WHO SAY EACH ACCESS POINT IS THEIR MAIN GATEWAY TO ONLINE NEWS (2018–22) – ALL MARKETS\n\n**Q10a.** Thinking about how you got news online (via computer, mobile, or any device) in the last week, which of these was the \\*\\*MAIN\\*\\* way in which you came across news? _Base: All markets 2018–22 and total sample in each market (n ≈ 2000). Note: Number of markets grew from 37 in 2018 to 46 in 2021–22._\n\n32%\n\n23% 23%\n\n28%\n\n0%\n\n25%\n\n50%\n\n2022 2021 2020 2019 2018\n\nDirect access to websites/apps Social media access\n\n**MAINLY DIRECT**\n\nFinland 65%\n\nNorway 59%\n\nUK 46%\n\n**MUCH LESS DIRECT**\n\nSouth Korea 5%\n\nBrazil 11%\n\nMexico 11%\n\n**MAINLY SOCIAL**\n\nThailand 59%\n\nPhilippines 51%\n\nKenya 51%\n\n**MUCH LESS SOCIAL**\n\nJapan 9%\n\nSweden 10%\n\nFinland 11%\n\nThis year, we revived a question about how different groups access news _first thing in the morning_ to give a sense of how news habits have changed since 2019. The next chart shows striking differences between countries. In Norway, Spain, Finland, and the UK, the smartphone is now the go-to route to news in the morning, with dependence on it growing substantially in the last three years. Radio remains an important part of morning routines in Ireland – even though it has been overtaken by the smartphone for the first time. Morning newspaper reading is still surprisingly popular in the Netherlands and Finland. Television remains a key influence in the US, France, and Italy and is the dominant medium in Japan.\n\nAcross countries, almost half of under 35s (47%) say they accessed news first using a smartphone, compared with just a quarter (28%) of those aged 35 and older and just 15% of those aged 65 and older. Japan’s older population may explain why it is bucking wider trends.\n\nBut what channels do people use when accessing news on a smartphone at the start of each day? Since 2019, we find that, across countries, social media (39%) have overtaken websites and apps (31%), followed by the smartphone home screen (12%) and aggregator apps like Apple News (9%).\n\n**Main gateways to digital news**\n\nWe continue to monitor the main access points to online news, not just via a smartphone and not just first thing in the morning. Across all devices, our data show direct access to apps and websites becoming less important over time and social media becoming more important, partly due to their ubiquity and convenience. At an aggregate level, we have reached something of a tipping point this year, with social media preference (28%) surging ahead of direct access (23%). But these are cross-market averages and there are major differences between markets. Audiences in Nordic countries and the UK still have strong direct connections, while people in Japan and South Korea tend to access news via powerful aggregators and search engines – relying less on direct access. These differences, which we can perhaps treat as a proxy for the strength of news brands in a digital world, help explain why it may be easier to charge for online news in some countries and not others.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 27\n\n0% 25% 50%\n\n43\n\n39\n\n38\n\n36\n\n35\n\n35\n\n26\n\n29\n\n25\n\n24\n\n13\n\n24\n\n32\n\n21\n\n15\n\n21\n\n24\n\n29\n\n27\n\n45\n\n16\n\n15\n\n11\n\n12\n\n31\n\n17\n\n14\n\n7\n\n23\n\n3\n\n5\n\n3\n\n3\n\n10\n\n3\n\n4\n\n13\n\n3\n\n3\n\n8\n\nJapan\n\nFrance\n\nUSA\n\nNetherlands\n\nUK\n\nIreland\n\nFinland\n\nItaly\n\nSpain\n\nNorway\n\nPROPORTION WHO SAY EACH DEVICE IS THEIR FIRST WAY OF ACCESSING NEWS IN THE MORNING – SELECTED MARKETS\n\n**Q9c\\_new2016.** What is the FIRST way you typically come across news in the morning? _Base: Total sample in each market (n ≈ 2000)._\n\nPrint Radio TV Smartphone\n\nWas 44% in 2019\n\nWas 27% in 2019\n\nWas 28% in 2019\n\nWas 31% in 2019\n\nWas 27% in 2019\n\nWas 26% in 2019\n\nWas 32% in 2019\n\n**TV first**\n\n**Mobile**\n\n**first**\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 23","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Digital_News-Report_2022_p23.pdf"]},"response":{"errors":[],"extract_id":"extract_22a1bef610789581a815857769073c92","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["REUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 28\n\nPROPORTION WHO SAY EACH ACCESS POINT IS THEIR MAIN GATEWAY TO ONLINE NEWS (2018–22) – ALL MARKETS\n\n**Q10a.** Thinking about how you got news online (via computer, mobile, or any device) in the last week, which of these was the \\*\\*MAIN\\*\\* way in which you came across news? _Base: All markets 2018–22 and total sample in each market (n ≈ 2000). Note: Number of markets grew from 37 in 2018 to 46 in 2021–22._\n\n32%\n\n23% 23%\n\n28%\n\n0%\n\n25%\n\n50%\n\n2022 2021 2020 2019 2018\n\nDirect access to websites/apps Social media access\n\n**MAINLY DIRECT**\n\nFinland 65%\n\nNorway 59%\n\nUK 46%\n\n**MUCH LESS DIRECT**\n\nSouth Korea 5%\n\nBrazil 11%\n\nMexico 11%\n\n**MAINLY SOCIAL**\n\nThailand 59%\n\nPhilippines 51%\n\nKenya 51%\n\n**MUCH LESS SOCIAL**\n\nJapan 9%\n\nSweden 10%\n\nFinland 11%\n\nThis year, we revived a question about how different groups access news _first thing in the morning_ to give a sense of how news habits have changed since 2019. The next chart shows striking differences between countries. In Norway, Spain, Finland, and the UK, the smartphone is now the go-to route to news in the morning, with dependence on it growing substantially in the last three years. Radio remains an important part of morning routines in Ireland – even though it has been overtaken by the smartphone for the first time. Morning newspaper reading is still surprisingly popular in the Netherlands and Finland. Television remains a key influence in the US, France, and Italy and is the dominant medium in Japan.\n\nAcross countries, almost half of under 35s (47%) say they accessed news first using a smartphone, compared with just a quarter (28%) of those aged 35 and older and just 15% of those aged 65 and older. Japan’s older population may explain why it is bucking wider trends.\n\nBut what channels do people use when accessing news on a smartphone at the start of each day? Since 2019, we find that, across countries, social media (39%) have overtaken websites and apps (31%), followed by the smartphone home screen (12%) and aggregator apps like Apple News (9%).\n\n**Main gateways to digital news**\n\nWe continue to monitor the main access points to online news, not just via a smartphone and not just first thing in the morning. Across all devices, our data show direct access to apps and websites becoming less important over time and social media becoming more important, partly due to their ubiquity and convenience. At an aggregate level, we have reached something of a tipping point this year, with social media preference (28%) surging ahead of direct access (23%). But these are cross-market averages and there are major differences between markets. Audiences in Nordic countries and the UK still have strong direct connections, while people in Japan and South Korea tend to access news via powerful aggregators and search engines – relying less on direct access. These differences, which we can perhaps treat as a proxy for the strength of news brands in a digital world, help explain why it may be easier to charge for online news in some countries and not others.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 27\n\n0% 25% 50%\n\n43\n\n39\n\n38\n\n36\n\n35\n\n35\n\n26\n\n29\n\n25\n\n24\n\n13\n\n24\n\n32\n\n21\n\n15\n\n21\n\n24\n\n29\n\n27\n\n45\n\n16\n\n15\n\n11\n\n12\n\n31\n\n17\n\n14\n\n7\n\n23\n\n3\n\n5\n\n3\n\n3\n\n10\n\n3\n\n4\n\n13\n\n3\n\n3\n\n8\n\nJapan\n\nFrance\n\nUSA\n\nNetherlands\n\nUK\n\nIreland\n\nFinland\n\nItaly\n\nSpain\n\nNorway\n\nPROPORTION WHO SAY EACH DEVICE IS THEIR FIRST WAY OF ACCESSING NEWS IN THE MORNING – SELECTED MARKETS\n\n**Q9c\\_new2016.** What is the FIRST way you typically come across news in the morning? _Base: Total sample in each market (n ≈ 2000)._\n\nPrint Radio TV Smartphone\n\nWas 44% in 2019\n\nWas 27% in 2019\n\nWas 28% in 2019\n\nWas 31% in 2019\n\nWas 27% in 2019\n\nWas 26% in 2019\n\nWas 32% in 2019\n\n**TV first**\n\n**Mobile**\n\n**first**\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 23"],"full_content":"REUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 28\n\nPROPORTION WHO SAY EACH ACCESS POINT IS THEIR MAIN GATEWAY TO ONLINE NEWS (2018–22) – ALL MARKETS\n\n**Q10a.** Thinking about how you got news online (via computer, mobile, or any device) in the last week, which of these was the \\*\\*MAIN\\*\\* way in which you came across news? _Base: All markets 2018–22 and total sample in each market (n ≈ 2000). Note: Number of markets grew from 37 in 2018 to 46 in 2021–22._\n\n32%\n\n23% 23%\n\n28%\n\n0%\n\n25%\n\n50%\n\n2022 2021 2020 2019 2018\n\nDirect access to websites/apps Social media access\n\n**MAINLY DIRECT**\n\nFinland 65%\n\nNorway 59%\n\nUK 46%\n\n**MUCH LESS DIRECT**\n\nSouth Korea 5%\n\nBrazil 11%\n\nMexico 11%\n\n**MAINLY SOCIAL**\n\nThailand 59%\n\nPhilippines 51%\n\nKenya 51%\n\n**MUCH LESS SOCIAL**\n\nJapan 9%\n\nSweden 10%\n\nFinland 11%\n\nThis year, we revived a question about how different groups access news _first thing in the morning_ to give a sense of how news habits have changed since 2019. The next chart shows striking differences between countries. In Norway, Spain, Finland, and the UK, the smartphone is now the go-to route to news in the morning, with dependence on it growing substantially in the last three years. Radio remains an important part of morning routines in Ireland – even though it has been overtaken by the smartphone for the first time. Morning newspaper reading is still surprisingly popular in the Netherlands and Finland. Television remains a key influence in the US, France, and Italy and is the dominant medium in Japan.\n\nAcross countries, almost half of under 35s (47%) say they accessed news first using a smartphone, compared with just a quarter (28%) of those aged 35 and older and just 15% of those aged 65 and older. Japan’s older population may explain why it is bucking wider trends.\n\nBut what channels do people use when accessing news on a smartphone at the start of each day? Since 2019, we find that, across countries, social media (39%) have overtaken websites and apps (31%), followed by the smartphone home screen (12%) and aggregator apps like Apple News (9%).\n\n**Main gateways to digital news**\n\nWe continue to monitor the main access points to online news, not just via a smartphone and not just first thing in the morning. Across all devices, our data show direct access to apps and websites becoming less important over time and social media becoming more important, partly due to their ubiquity and convenience. At an aggregate level, we have reached something of a tipping point this year, with social media preference (28%) surging ahead of direct access (23%). But these are cross-market averages and there are major differences between markets. Audiences in Nordic countries and the UK still have strong direct connections, while people in Japan and South Korea tend to access news via powerful aggregators and search engines – relying less on direct access. These differences, which we can perhaps treat as a proxy for the strength of news brands in a digital world, help explain why it may be easier to charge for online news in some countries and not others.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 27\n\n0% 25% 50%\n\n43\n\n39\n\n38\n\n36\n\n35\n\n35\n\n26\n\n29\n\n25\n\n24\n\n13\n\n24\n\n32\n\n21\n\n15\n\n21\n\n24\n\n29\n\n27\n\n45\n\n16\n\n15\n\n11\n\n12\n\n31\n\n17\n\n14\n\n7\n\n23\n\n3\n\n5\n\n3\n\n3\n\n10\n\n3\n\n4\n\n13\n\n3\n\n3\n\n8\n\nJapan\n\nFrance\n\nUSA\n\nNetherlands\n\nUK\n\nIreland\n\nFinland\n\nItaly\n\nSpain\n\nNorway\n\nPROPORTION WHO SAY EACH DEVICE IS THEIR FIRST WAY OF ACCESSING NEWS IN THE MORNING – SELECTED MARKETS\n\n**Q9c\\_new2016.** What is the FIRST way you typically come across news in the morning? _Base: Total sample in each market (n ≈ 2000)._\n\nPrint Radio TV Smartphone\n\nWas 44% in 2019\n\nWas 27% in 2019\n\nWas 28% in 2019\n\nWas 31% in 2019\n\nWas 27% in 2019\n\nWas 26% in 2019\n\nWas 32% in 2019\n\n**TV first**\n\n**Mobile**\n\n**first**\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 23","publish_date":null,"title":"Graph 27 Graph 28","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Digital_News-Report_2022_p23.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Digital_News-Report_2022_p23.pdf"},"request":{"example_id":"chart/Digital_News-Report_2022_p23","product_type":"parse","source_file_path":"Digital_News-Report_2022_p23.pdf"},"started_at":"2026-11-04T19:46:11.672109"} +{"completed_at":"2026-11-04T19:46:16.093328","latency_in_ms":1995,"output":{"example_id":"chart/Digital_News-Report_2022_p24","grounded_pages":[],"layout_pages":[],"markdown":"These changes are in large part driven by the emerging habits of a new generation of _social natives_ as they come into adulthood. In the next chart, which shows main access by age in the UK, we find that millennials within our 25–34 group (orange line) and those older than 35 (magenta line) have only slightly changed preferences over time, but the 18–24 group (turquoise line) has become significantly less likely to use a news website or app. This is another illustration that this youngest generation, which has grown up with social media, is not just different but is more different than the one that came before.\n\n_For further analysis see Section 2.3: The Changing News Habits and Attitudes of Younger Audiences_\n\n**IS FACEBOOK LOSING ITS SHEEN?**\n\nEach year since 2014, we have tracked the different social networks used for any purpose and for news across 12 countries. Facebook usage for any purpose (60%) is down five percentage points since its peak in 2017 and is now at a similar level to YouTube. Instagram (40%), TikTok (16%), and Telegram (11%) are the only networks to have grown in the last year.\n\nOnce again, most of these changes – such as the decline in Facebook use over the last few years – result from shifts in behaviour of the youngest cohort of social natives, not older respondents, who tend to have more entrenched habits. In the following chart from the UK, we illustrate how the attention of this group has gradually shifted – from Facebook, to Instagram, and now to TikTok.\n\nWhen it comes to _news usage_ specifically, Facebook remains the most important network across our basket of 12 countries, but has dropped by 12 percentage points since 2016. Twitter has largely stagnated over the last decade in terms of its user base, though it remains hugely influential with journalists and politicians. Confusion over its future direction is likely to persist following the attempt to take over the company by Elon Musk ⁸ and the subsequent loss of senior executives. Meta-owned Instagram is now more widely used for news, while TikTok has overtaken Snapchat from a low base.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 29\n\nPROPORTION WHO START THEIR NEWS JOURNEY WITH A NEWS WEBSITE OR APP – BY AGE – UK\n\n**Q10.** Thinking about how you got news online (via computer, mobile, or any device) in the last week, which were the ways in which you came across news stories? _Base: 2018–22; 18–24 ≈ 200, 25–34 ≈ 300, 35+ ≈ 1500._\n\nCoronavirus\n\n53% 52% 52%\n\n28%\n\n45% 49%\n\n0%\n\n25%\n\n50%\n\n75%\n\n100%\n\n35+\n\n25-34\n\n18-24\n\n2016 2017 2018 2019 2020 2021 2022 2015\n\n18-24 25-34 35+\n\n**Gap opening up between 18-24s and the rest**\n\nDecline mirrors rise in social media use by 18-24s\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 30\n\nPROPORTION THAT USED EACH SOCIAL NETWORK FOR ANY PURPOSE IN THE LAST WEEK (2014-22) – AVERAGE OF 12 MARKETS\n\n**Q12a.** Which, if any, of the following have you used in the last week for any purpose? _Base: Total 2014–22 sample in selected markets (most n ≈ 2000). Note: From 2015–21 the 12 countries included are: UK, USA, Germany, France, Spain, Italy, Ireland, Denmark, Finland, Japan, Australia, and Brazil. In 2014 we did not poll in Australia or Ireland._\n\n27% 24%\n\n0%\n\n25%\n\n50%\n\n75% Telegram\n\nTikTok\n\nYouTube\n\nWhatsApp\n\nSnapchat\n\nInstagram\n\nTwitter\n\nFacebook Messenger\n\nFacebook\n\n2022 2021 2020 2019 2018 2017 2016 2015 2014\n\n51%\n\n17%\n\n60% 60%\n\n31%\n\n8% 5% 2% 5%\n\n21% 19%\n\n16% 12% 11%\n\n34%\n\n61%\n\n53%\n\nFacebook Snapchat WhatsApp YouTube Telegram TikTok\n\nFacebook Messenger Twitter Instagram\n\n40%\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 32\n\n0%\n\n25%\n\n50% Ti\n\nYo\n\nW\n\nSn\n\nIn\n\nFa\n\n2022 2021 2020 2019 2018 2017 2016 2015 2014\n\n7% 4% 2%\n\nPROPORTION THAT USED EACH SOCIAL NETWORK FOR NEWS IN THE LAST WEEK (2014-22) – AVERAGE OF 12 MARKETS\n\n**Q12b.** Which, if any, of the following have you used in the last week for news? _Base: Total 2014–22 sample in selected markets (most n ≈ 2000). Note: From 2015–21 the 12 countries included are: UK, USA, Germany, France, Spain, Italy, Ireland, Denmark, Finland, Japan, Australia, and Brazil. In 2014 we did not poll in Australia or Ireland._\n\n15%\n\n7%\n\n30%\n\n36%\n\n6%\n\n2% 1% 1%\n\n11% 9%\n\n19% 16%\n\nFacebook Snapchat WhatsApp YouTube TikTok\n\nFacebook Messenger Twitter Instagram\n\n12%\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 31\n\nPROPORTION OF 18–24s (SOCIAL NATIVES) WHO USED EACH SOCIAL NETWORK FOR ANY PURPOSE IN THE LAST WEEK (2014–22) – UK\n\n**Q12a.** Which, if any, of the following have you used in the last week for any purpose? _Base: 2014–22: 18–24s ≈ 200._\n\n0%\n\n25%\n\n50%\n\n75%\n\n100%\n\n2022 2021 2020 2019 2018 2017 2016 2015 2014\n\n17%\n\n62%\n\n51%\n\n78%\n\n20%\n\n4%\n\nFacebook WhatsApp TikTok Instagram\n\n68%\n\n32%\n\n18-24 cliff edge started in 2018 which is when Instagram started to take over\n\n8 https://www.cnbc.com/2022/04/26/elon-musks-twitter-takeover-sets-up-a-potential-clash-with-europe.html\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 24","pages":[{"line_numbers":[],"markdown":"These changes are in large part driven by the emerging habits of a new generation of _social natives_ as they come into adulthood. In the next chart, which shows main access by age in the UK, we find that millennials within our 25–34 group (orange line) and those older than 35 (magenta line) have only slightly changed preferences over time, but the 18–24 group (turquoise line) has become significantly less likely to use a news website or app. This is another illustration that this youngest generation, which has grown up with social media, is not just different but is more different than the one that came before.\n\n_For further analysis see Section 2.3: The Changing News Habits and Attitudes of Younger Audiences_\n\n**IS FACEBOOK LOSING ITS SHEEN?**\n\nEach year since 2014, we have tracked the different social networks used for any purpose and for news across 12 countries. Facebook usage for any purpose (60%) is down five percentage points since its peak in 2017 and is now at a similar level to YouTube. Instagram (40%), TikTok (16%), and Telegram (11%) are the only networks to have grown in the last year.\n\nOnce again, most of these changes – such as the decline in Facebook use over the last few years – result from shifts in behaviour of the youngest cohort of social natives, not older respondents, who tend to have more entrenched habits. In the following chart from the UK, we illustrate how the attention of this group has gradually shifted – from Facebook, to Instagram, and now to TikTok.\n\nWhen it comes to _news usage_ specifically, Facebook remains the most important network across our basket of 12 countries, but has dropped by 12 percentage points since 2016. Twitter has largely stagnated over the last decade in terms of its user base, though it remains hugely influential with journalists and politicians. Confusion over its future direction is likely to persist following the attempt to take over the company by Elon Musk ⁸ and the subsequent loss of senior executives. Meta-owned Instagram is now more widely used for news, while TikTok has overtaken Snapchat from a low base.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 29\n\nPROPORTION WHO START THEIR NEWS JOURNEY WITH A NEWS WEBSITE OR APP – BY AGE – UK\n\n**Q10.** Thinking about how you got news online (via computer, mobile, or any device) in the last week, which were the ways in which you came across news stories? _Base: 2018–22; 18–24 ≈ 200, 25–34 ≈ 300, 35+ ≈ 1500._\n\nCoronavirus\n\n53% 52% 52%\n\n28%\n\n45% 49%\n\n0%\n\n25%\n\n50%\n\n75%\n\n100%\n\n35+\n\n25-34\n\n18-24\n\n2016 2017 2018 2019 2020 2021 2022 2015\n\n18-24 25-34 35+\n\n**Gap opening up between 18-24s and the rest**\n\nDecline mirrors rise in social media use by 18-24s\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 30\n\nPROPORTION THAT USED EACH SOCIAL NETWORK FOR ANY PURPOSE IN THE LAST WEEK (2014-22) – AVERAGE OF 12 MARKETS\n\n**Q12a.** Which, if any, of the following have you used in the last week for any purpose? _Base: Total 2014–22 sample in selected markets (most n ≈ 2000). Note: From 2015–21 the 12 countries included are: UK, USA, Germany, France, Spain, Italy, Ireland, Denmark, Finland, Japan, Australia, and Brazil. In 2014 we did not poll in Australia or Ireland._\n\n27% 24%\n\n0%\n\n25%\n\n50%\n\n75% Telegram\n\nTikTok\n\nYouTube\n\nWhatsApp\n\nSnapchat\n\nInstagram\n\nTwitter\n\nFacebook Messenger\n\nFacebook\n\n2022 2021 2020 2019 2018 2017 2016 2015 2014\n\n51%\n\n17%\n\n60% 60%\n\n31%\n\n8% 5% 2% 5%\n\n21% 19%\n\n16% 12% 11%\n\n34%\n\n61%\n\n53%\n\nFacebook Snapchat WhatsApp YouTube Telegram TikTok\n\nFacebook Messenger Twitter Instagram\n\n40%\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 32\n\n0%\n\n25%\n\n50% Ti\n\nYo\n\nW\n\nSn\n\nIn\n\nFa\n\n2022 2021 2020 2019 2018 2017 2016 2015 2014\n\n7% 4% 2%\n\nPROPORTION THAT USED EACH SOCIAL NETWORK FOR NEWS IN THE LAST WEEK (2014-22) – AVERAGE OF 12 MARKETS\n\n**Q12b.** Which, if any, of the following have you used in the last week for news? _Base: Total 2014–22 sample in selected markets (most n ≈ 2000). Note: From 2015–21 the 12 countries included are: UK, USA, Germany, France, Spain, Italy, Ireland, Denmark, Finland, Japan, Australia, and Brazil. In 2014 we did not poll in Australia or Ireland._\n\n15%\n\n7%\n\n30%\n\n36%\n\n6%\n\n2% 1% 1%\n\n11% 9%\n\n19% 16%\n\nFacebook Snapchat WhatsApp YouTube TikTok\n\nFacebook Messenger Twitter Instagram\n\n12%\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 31\n\nPROPORTION OF 18–24s (SOCIAL NATIVES) WHO USED EACH SOCIAL NETWORK FOR ANY PURPOSE IN THE LAST WEEK (2014–22) – UK\n\n**Q12a.** Which, if any, of the following have you used in the last week for any purpose? _Base: 2014–22: 18–24s ≈ 200._\n\n0%\n\n25%\n\n50%\n\n75%\n\n100%\n\n2022 2021 2020 2019 2018 2017 2016 2015 2014\n\n17%\n\n62%\n\n51%\n\n78%\n\n20%\n\n4%\n\nFacebook WhatsApp TikTok Instagram\n\n68%\n\n32%\n\n18-24 cliff edge started in 2018 which is when Instagram started to take over\n\n8 https://www.cnbc.com/2022/04/26/elon-musks-twitter-takeover-sets-up-a-potential-clash-with-europe.html\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 24","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"These changes are in large part driven by the emerging habits of a new generation of _social natives_ as they come into adulthood. In the next chart, which shows main access by age in the UK, we find that millennials within our 25–34 group (orange line) and those older than 35 (magenta line) have only slightly changed preferences over time, but the 18–24 group (turquoise line) has become significantly less likely to use a news website or app. This is another illustration that this youngest generation, which has grown up with social media, is not just different but is more different than the one that came before.\n\n_For further analysis see Section 2.3: The Changing News Habits and Attitudes of Younger Audiences_\n\n**IS FACEBOOK LOSING ITS SHEEN?**\n\nEach year since 2014, we have tracked the different social networks used for any purpose and for news across 12 countries. Facebook usage for any purpose (60%) is down five percentage points since its peak in 2017 and is now at a similar level to YouTube. Instagram (40%), TikTok (16%), and Telegram (11%) are the only networks to have grown in the last year.\n\nOnce again, most of these changes – such as the decline in Facebook use over the last few years – result from shifts in behaviour of the youngest cohort of social natives, not older respondents, who tend to have more entrenched habits. In the following chart from the UK, we illustrate how the attention of this group has gradually shifted – from Facebook, to Instagram, and now to TikTok.\n\nWhen it comes to _news usage_ specifically, Facebook remains the most important network across our basket of 12 countries, but has dropped by 12 percentage points since 2016. Twitter has largely stagnated over the last decade in terms of its user base, though it remains hugely influential with journalists and politicians. Confusion over its future direction is likely to persist following the attempt to take over the company by Elon Musk ⁸ and the subsequent loss of senior executives. Meta-owned Instagram is now more widely used for news, while TikTok has overtaken Snapchat from a low base.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 29\n\nPROPORTION WHO START THEIR NEWS JOURNEY WITH A NEWS WEBSITE OR APP – BY AGE – UK\n\n**Q10.** Thinking about how you got news online (via computer, mobile, or any device) in the last week, which were the ways in which you came across news stories? _Base: 2018–22; 18–24 ≈ 200, 25–34 ≈ 300, 35+ ≈ 1500._\n\nCoronavirus\n\n53% 52% 52%\n\n28%\n\n45% 49%\n\n0%\n\n25%\n\n50%\n\n75%\n\n100%\n\n35+\n\n25-34\n\n18-24\n\n2016 2017 2018 2019 2020 2021 2022 2015\n\n18-24 25-34 35+\n\n**Gap opening up between 18-24s and the rest**\n\nDecline mirrors rise in social media use by 18-24s\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 30\n\nPROPORTION THAT USED EACH SOCIAL NETWORK FOR ANY PURPOSE IN THE LAST WEEK (2014-22) – AVERAGE OF 12 MARKETS\n\n**Q12a.** Which, if any, of the following have you used in the last week for any purpose? _Base: Total 2014–22 sample in selected markets (most n ≈ 2000). Note: From 2015–21 the 12 countries included are: UK, USA, Germany, France, Spain, Italy, Ireland, Denmark, Finland, Japan, Australia, and Brazil. In 2014 we did not poll in Australia or Ireland._\n\n27% 24%\n\n0%\n\n25%\n\n50%\n\n75% Telegram\n\nTikTok\n\nYouTube\n\nWhatsApp\n\nSnapchat\n\nInstagram\n\nTwitter\n\nFacebook Messenger\n\nFacebook\n\n2022 2021 2020 2019 2018 2017 2016 2015 2014\n\n51%\n\n17%\n\n60% 60%\n\n31%\n\n8% 5% 2% 5%\n\n21% 19%\n\n16% 12% 11%\n\n34%\n\n61%\n\n53%\n\nFacebook Snapchat WhatsApp YouTube Telegram TikTok\n\nFacebook Messenger Twitter Instagram\n\n40%\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 32\n\n0%\n\n25%\n\n50% Ti\n\nYo\n\nW\n\nSn\n\nIn\n\nFa\n\n2022 2021 2020 2019 2018 2017 2016 2015 2014\n\n7% 4% 2%\n\nPROPORTION THAT USED EACH SOCIAL NETWORK FOR NEWS IN THE LAST WEEK (2014-22) – AVERAGE OF 12 MARKETS\n\n**Q12b.** Which, if any, of the following have you used in the last week for news? _Base: Total 2014–22 sample in selected markets (most n ≈ 2000). Note: From 2015–21 the 12 countries included are: UK, USA, Germany, France, Spain, Italy, Ireland, Denmark, Finland, Japan, Australia, and Brazil. In 2014 we did not poll in Australia or Ireland._\n\n15%\n\n7%\n\n30%\n\n36%\n\n6%\n\n2% 1% 1%\n\n11% 9%\n\n19% 16%\n\nFacebook Snapchat WhatsApp YouTube TikTok\n\nFacebook Messenger Twitter Instagram\n\n12%\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 31\n\nPROPORTION OF 18–24s (SOCIAL NATIVES) WHO USED EACH SOCIAL NETWORK FOR ANY PURPOSE IN THE LAST WEEK (2014–22) – UK\n\n**Q12a.** Which, if any, of the following have you used in the last week for any purpose? _Base: 2014–22: 18–24s ≈ 200._\n\n0%\n\n25%\n\n50%\n\n75%\n\n100%\n\n2022 2021 2020 2019 2018 2017 2016 2015 2014\n\n17%\n\n62%\n\n51%\n\n78%\n\n20%\n\n4%\n\nFacebook WhatsApp TikTok Instagram\n\n68%\n\n32%\n\n18-24 cliff edge started in 2018 which is when Instagram started to take over\n\n8 https://www.cnbc.com/2022/04/26/elon-musks-twitter-takeover-sets-up-a-potential-clash-with-europe.html\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 24","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Digital_News-Report_2022_p24.pdf"]},"response":{"errors":[],"extract_id":"extract_9032bed380bdec3db2e6d849a6e10b47","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["These changes are in large part driven by the emerging habits of a new generation of _social natives_ as they come into adulthood. In the next chart, which shows main access by age in the UK, we find that millennials within our 25–34 group (orange line) and those older than 35 (magenta line) have only slightly changed preferences over time, but the 18–24 group (turquoise line) has become significantly less likely to use a news website or app. This is another illustration that this youngest generation, which has grown up with social media, is not just different but is more different than the one that came before.\n\n_For further analysis see Section 2.3: The Changing News Habits and Attitudes of Younger Audiences_\n\n**IS FACEBOOK LOSING ITS SHEEN?**\n\nEach year since 2014, we have tracked the different social networks used for any purpose and for news across 12 countries. Facebook usage for any purpose (60%) is down five percentage points since its peak in 2017 and is now at a similar level to YouTube. Instagram (40%), TikTok (16%), and Telegram (11%) are the only networks to have grown in the last year.\n\nOnce again, most of these changes – such as the decline in Facebook use over the last few years – result from shifts in behaviour of the youngest cohort of social natives, not older respondents, who tend to have more entrenched habits. In the following chart from the UK, we illustrate how the attention of this group has gradually shifted – from Facebook, to Instagram, and now to TikTok.\n\nWhen it comes to _news usage_ specifically, Facebook remains the most important network across our basket of 12 countries, but has dropped by 12 percentage points since 2016. Twitter has largely stagnated over the last decade in terms of its user base, though it remains hugely influential with journalists and politicians. Confusion over its future direction is likely to persist following the attempt to take over the company by Elon Musk ⁸ and the subsequent loss of senior executives. Meta-owned Instagram is now more widely used for news, while TikTok has overtaken Snapchat from a low base.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 29\n\nPROPORTION WHO START THEIR NEWS JOURNEY WITH A NEWS WEBSITE OR APP – BY AGE – UK\n\n**Q10.** Thinking about how you got news online (via computer, mobile, or any device) in the last week, which were the ways in which you came across news stories? _Base: 2018–22; 18–24 ≈ 200, 25–34 ≈ 300, 35+ ≈ 1500._\n\nCoronavirus\n\n53% 52% 52%\n\n28%\n\n45% 49%\n\n0%\n\n25%\n\n50%\n\n75%\n\n100%\n\n35+\n\n25-34\n\n18-24\n\n2016 2017 2018 2019 2020 2021 2022 2015\n\n18-24 25-34 35+\n\n**Gap opening up between 18-24s and the rest**\n\nDecline mirrors rise in social media use by 18-24s\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 30\n\nPROPORTION THAT USED EACH SOCIAL NETWORK FOR ANY PURPOSE IN THE LAST WEEK (2014-22) – AVERAGE OF 12 MARKETS\n\n**Q12a.** Which, if any, of the following have you used in the last week for any purpose? _Base: Total 2014–22 sample in selected markets (most n ≈ 2000). Note: From 2015–21 the 12 countries included are: UK, USA, Germany, France, Spain, Italy, Ireland, Denmark, Finland, Japan, Australia, and Brazil. In 2014 we did not poll in Australia or Ireland._\n\n27% 24%\n\n0%\n\n25%\n\n50%\n\n75% Telegram\n\nTikTok\n\nYouTube\n\nWhatsApp\n\nSnapchat\n\nInstagram\n\nTwitter\n\nFacebook Messenger\n\nFacebook\n\n2022 2021 2020 2019 2018 2017 2016 2015 2014\n\n51%\n\n17%\n\n60% 60%\n\n31%\n\n8% 5% 2% 5%\n\n21% 19%\n\n16% 12% 11%\n\n34%\n\n61%\n\n53%\n\nFacebook Snapchat WhatsApp YouTube Telegram TikTok\n\nFacebook Messenger Twitter Instagram\n\n40%\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 32\n\n0%\n\n25%\n\n50% Ti\n\nYo\n\nW\n\nSn\n\nIn\n\nFa\n\n2022 2021 2020 2019 2018 2017 2016 2015 2014\n\n7% 4% 2%\n\nPROPORTION THAT USED EACH SOCIAL NETWORK FOR NEWS IN THE LAST WEEK (2014-22) – AVERAGE OF 12 MARKETS\n\n**Q12b.** Which, if any, of the following have you used in the last week for news? _Base: Total 2014–22 sample in selected markets (most n ≈ 2000). Note: From 2015–21 the 12 countries included are: UK, USA, Germany, France, Spain, Italy, Ireland, Denmark, Finland, Japan, Australia, and Brazil. In 2014 we did not poll in Australia or Ireland._\n\n15%\n\n7%\n\n30%\n\n36%\n\n6%\n\n2% 1% 1%\n\n11% 9%\n\n19% 16%\n\nFacebook Snapchat WhatsApp YouTube TikTok\n\nFacebook Messenger Twitter Instagram\n\n12%\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 31\n\nPROPORTION OF 18–24s (SOCIAL NATIVES) WHO USED EACH SOCIAL NETWORK FOR ANY PURPOSE IN THE LAST WEEK (2014–22) – UK\n\n**Q12a.** Which, if any, of the following have you used in the last week for any purpose? _Base: 2014–22: 18–24s ≈ 200._\n\n0%\n\n25%\n\n50%\n\n75%\n\n100%\n\n2022 2021 2020 2019 2018 2017 2016 2015 2014\n\n17%\n\n62%\n\n51%\n\n78%\n\n20%\n\n4%\n\nFacebook WhatsApp TikTok Instagram\n\n68%\n\n32%\n\n18-24 cliff edge started in 2018 which is when Instagram started to take over\n\n8 https://www.cnbc.com/2022/04/26/elon-musks-twitter-takeover-sets-up-a-potential-clash-with-europe.html\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 24"],"full_content":"These changes are in large part driven by the emerging habits of a new generation of _social natives_ as they come into adulthood. In the next chart, which shows main access by age in the UK, we find that millennials within our 25–34 group (orange line) and those older than 35 (magenta line) have only slightly changed preferences over time, but the 18–24 group (turquoise line) has become significantly less likely to use a news website or app. This is another illustration that this youngest generation, which has grown up with social media, is not just different but is more different than the one that came before.\n\n_For further analysis see Section 2.3: The Changing News Habits and Attitudes of Younger Audiences_\n\n**IS FACEBOOK LOSING ITS SHEEN?**\n\nEach year since 2014, we have tracked the different social networks used for any purpose and for news across 12 countries. Facebook usage for any purpose (60%) is down five percentage points since its peak in 2017 and is now at a similar level to YouTube. Instagram (40%), TikTok (16%), and Telegram (11%) are the only networks to have grown in the last year.\n\nOnce again, most of these changes – such as the decline in Facebook use over the last few years – result from shifts in behaviour of the youngest cohort of social natives, not older respondents, who tend to have more entrenched habits. In the following chart from the UK, we illustrate how the attention of this group has gradually shifted – from Facebook, to Instagram, and now to TikTok.\n\nWhen it comes to _news usage_ specifically, Facebook remains the most important network across our basket of 12 countries, but has dropped by 12 percentage points since 2016. Twitter has largely stagnated over the last decade in terms of its user base, though it remains hugely influential with journalists and politicians. Confusion over its future direction is likely to persist following the attempt to take over the company by Elon Musk ⁸ and the subsequent loss of senior executives. Meta-owned Instagram is now more widely used for news, while TikTok has overtaken Snapchat from a low base.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 29\n\nPROPORTION WHO START THEIR NEWS JOURNEY WITH A NEWS WEBSITE OR APP – BY AGE – UK\n\n**Q10.** Thinking about how you got news online (via computer, mobile, or any device) in the last week, which were the ways in which you came across news stories? _Base: 2018–22; 18–24 ≈ 200, 25–34 ≈ 300, 35+ ≈ 1500._\n\nCoronavirus\n\n53% 52% 52%\n\n28%\n\n45% 49%\n\n0%\n\n25%\n\n50%\n\n75%\n\n100%\n\n35+\n\n25-34\n\n18-24\n\n2016 2017 2018 2019 2020 2021 2022 2015\n\n18-24 25-34 35+\n\n**Gap opening up between 18-24s and the rest**\n\nDecline mirrors rise in social media use by 18-24s\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 30\n\nPROPORTION THAT USED EACH SOCIAL NETWORK FOR ANY PURPOSE IN THE LAST WEEK (2014-22) – AVERAGE OF 12 MARKETS\n\n**Q12a.** Which, if any, of the following have you used in the last week for any purpose? _Base: Total 2014–22 sample in selected markets (most n ≈ 2000). Note: From 2015–21 the 12 countries included are: UK, USA, Germany, France, Spain, Italy, Ireland, Denmark, Finland, Japan, Australia, and Brazil. In 2014 we did not poll in Australia or Ireland._\n\n27% 24%\n\n0%\n\n25%\n\n50%\n\n75% Telegram\n\nTikTok\n\nYouTube\n\nWhatsApp\n\nSnapchat\n\nInstagram\n\nTwitter\n\nFacebook Messenger\n\nFacebook\n\n2022 2021 2020 2019 2018 2017 2016 2015 2014\n\n51%\n\n17%\n\n60% 60%\n\n31%\n\n8% 5% 2% 5%\n\n21% 19%\n\n16% 12% 11%\n\n34%\n\n61%\n\n53%\n\nFacebook Snapchat WhatsApp YouTube Telegram TikTok\n\nFacebook Messenger Twitter Instagram\n\n40%\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 32\n\n0%\n\n25%\n\n50% Ti\n\nYo\n\nW\n\nSn\n\nIn\n\nFa\n\n2022 2021 2020 2019 2018 2017 2016 2015 2014\n\n7% 4% 2%\n\nPROPORTION THAT USED EACH SOCIAL NETWORK FOR NEWS IN THE LAST WEEK (2014-22) – AVERAGE OF 12 MARKETS\n\n**Q12b.** Which, if any, of the following have you used in the last week for news? _Base: Total 2014–22 sample in selected markets (most n ≈ 2000). Note: From 2015–21 the 12 countries included are: UK, USA, Germany, France, Spain, Italy, Ireland, Denmark, Finland, Japan, Australia, and Brazil. In 2014 we did not poll in Australia or Ireland._\n\n15%\n\n7%\n\n30%\n\n36%\n\n6%\n\n2% 1% 1%\n\n11% 9%\n\n19% 16%\n\nFacebook Snapchat WhatsApp YouTube TikTok\n\nFacebook Messenger Twitter Instagram\n\n12%\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 31\n\nPROPORTION OF 18–24s (SOCIAL NATIVES) WHO USED EACH SOCIAL NETWORK FOR ANY PURPOSE IN THE LAST WEEK (2014–22) – UK\n\n**Q12a.** Which, if any, of the following have you used in the last week for any purpose? _Base: 2014–22: 18–24s ≈ 200._\n\n0%\n\n25%\n\n50%\n\n75%\n\n100%\n\n2022 2021 2020 2019 2018 2017 2016 2015 2014\n\n17%\n\n62%\n\n51%\n\n78%\n\n20%\n\n4%\n\nFacebook WhatsApp TikTok Instagram\n\n68%\n\n32%\n\n18-24 cliff edge started in 2018 which is when Instagram started to take over\n\n8 https://www.cnbc.com/2022/04/26/elon-musks-twitter-takeover-sets-up-a-potential-clash-with-europe.html\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 24","publish_date":null,"title":"Graph 31 Graph 29 Graph 32 Graph 30","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Digital_News-Report_2022_p24.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Digital_News-Report_2022_p24.pdf"},"request":{"example_id":"chart/Digital_News-Report_2022_p24","product_type":"parse","source_file_path":"Digital_News-Report_2022_p24.pdf"},"started_at":"2026-11-04T19:46:14.098049"} +{"completed_at":"2026-11-04T19:46:24.144944","latency_in_ms":8528,"output":{"example_id":"chart/Digital_News-Report_2022_p25","grounded_pages":[],"layout_pages":[],"markdown":"Outside Western countries – especially in Latin America and Africa – we find a much higher proportion of our sample using social media for news, but we also find different networks in play. In African countries such as Nigeria, South Africa, and Kenya, WhatsApp (55%) and Telegram (18%) combined are more important for finding, sharing, and discussing news than Facebook (59%). Latin Americans tend to use a combination of all the main networks, while it is a mixed picture in Asia, with almost three-quarters (73%) using Facebook for news in the Philippines but just 5% in Japan, where Twitter (18%) and Line (16%) are more popular. YouTube (44%) is the main social network for news in South Korea, along with home-grown app KakaoTalk (24%) and KakaoStory (5%).\n\n**TikTok emerges as a significant new player in the news ecosystem**\n\nIn our data, we find significant and rapidly growing usage of TikTok, especially in Africa, Asia, and Latin America as well as across Eastern Europe. Usage is mostly still with under 25s, but is extending into all age brackets in countries with higher levels of usage, such as Kenya, South Africa, Thailand, Indonesia, Brazil, and Peru. This quote from our Brazil qualitative research sums up how perceptions of content on TikTok have changed in the last year or so.\n\n**_When TikTok was launched, it was just about dancing, choreography; not today, even though the videos are fast, they bring immediate information._** _Male, 24, Brazil_\n\nThe Russia-Ukraine conflict has increased the profile of the network globally. Ukrainians have been documenting their experience of the war, including leaving their homes as refugees, clocking up millions of views in the process (see picture above right). As we noted last year, many publishers have been increasing their investment, but some worry that a platform where entertainment content is so sought-after may not be the most effective place for news. BBC News originally decided to avoid TikTok, but has now set up channels in Russian and English after seeing misleading information being shared on the platform around the war in Ukraine.\n\nIn contrast to traditional networks such as Facebook or Twitter, the content of the main TikTok feed is driven less by who you follow and more by an algorithm that takes into account what you like and what others are watching.\n\n**_It’s so addictive … and where it lacks in trustworthiness, it excels in presentation. It’s a news source I end up consuming because I’m also often scrolling TikTok for other reasons, but the algorithm ends up providing news anyways._** _Female, 22, USA_\n\nBut others remain sceptical about the value of very short videos designed for popularity.\n\n**_I greatly dislike everything about the presentation of TikTok [and Instagram]. Everything about it screams a lack of professionalism and research and I cannot trust it, along with it being obnoxiously loud._** _Male, 18, USA_\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 33\n\nPROPORTION THAT USED EACH SOCIAL NETWORK FOR NEWS IN THE LAST WEEK – SELECTED REGIONS\n\n**Europe**\n\nFacebook\n\nInstagram\n\nFB Messenger\n\nTwitter\n\nTelegram\n\nTikTok\n\nYouTube\n\nWhatsApp\n\n**Asia Latin America Africa North America**\n\n**16%**\n\n**23%**\n\n**6%**\n\n**25%**\n\n**24%**\n\n**34%**\n\n**34%**\n\n**10%**\n\n**43%**\n\n**18%**\n\n**16%**\n\n**57%**\n\n**28% 29%**\n\n**15%**\n\n**59%**\n\n**55%**\n\n**42% 35% 44%**\n\n**35%**\n\n**40%**\n\n**14%**\n\n**13% 11%**\n\n**14% 11%**\n\n**11% 11%**\n\n**9%**\n\n**4%**\n\n**5%**\n\n**2%**\n\n**6%**\n\n**9%**\n\n**9%**\n\n**6%**\n\n**11%**\n\n**18%**\n\n**14%**\n\n**Q12B.** Which, if any, of the following have you used for news in the last week? _Base: Total sample in each region. Europe = 48,836, North America = 4048, Asia = 20,349, Latin America = 12,104, Africa = 6057. Note: Africa average is Kenya, South Africa, and Nigeria only (English speakers in South Africa and Nigeria)._\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 34\n\n## **40%**\n\nof 18-24s use TikTok for any purpose\n\n## **15%**\n\nof 18-24s use TikTok for news\n\n(Across all markets)\n\nKenya **43%** / **15%**\n\nThailand **42%** / **22%**\n\nSouth Africa **41%** / **15%**\n\nPeru **37%** / **16%**\n\nMexico **35%** / **10%**\n\nMarkets with highest usage (all ages) **% for any purpose** / **for news**\n\n**Q12a/b.** Which, if any, of the following have you used in the last week for any purpose/news? _Base: Total sample in each market (n ≈ 2000)._\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 25","pages":[{"line_numbers":[],"markdown":"Outside Western countries – especially in Latin America and Africa – we find a much higher proportion of our sample using social media for news, but we also find different networks in play. In African countries such as Nigeria, South Africa, and Kenya, WhatsApp (55%) and Telegram (18%) combined are more important for finding, sharing, and discussing news than Facebook (59%). Latin Americans tend to use a combination of all the main networks, while it is a mixed picture in Asia, with almost three-quarters (73%) using Facebook for news in the Philippines but just 5% in Japan, where Twitter (18%) and Line (16%) are more popular. YouTube (44%) is the main social network for news in South Korea, along with home-grown app KakaoTalk (24%) and KakaoStory (5%).\n\n**TikTok emerges as a significant new player in the news ecosystem**\n\nIn our data, we find significant and rapidly growing usage of TikTok, especially in Africa, Asia, and Latin America as well as across Eastern Europe. Usage is mostly still with under 25s, but is extending into all age brackets in countries with higher levels of usage, such as Kenya, South Africa, Thailand, Indonesia, Brazil, and Peru. This quote from our Brazil qualitative research sums up how perceptions of content on TikTok have changed in the last year or so.\n\n**_When TikTok was launched, it was just about dancing, choreography; not today, even though the videos are fast, they bring immediate information._** _Male, 24, Brazil_\n\nThe Russia-Ukraine conflict has increased the profile of the network globally. Ukrainians have been documenting their experience of the war, including leaving their homes as refugees, clocking up millions of views in the process (see picture above right). As we noted last year, many publishers have been increasing their investment, but some worry that a platform where entertainment content is so sought-after may not be the most effective place for news. BBC News originally decided to avoid TikTok, but has now set up channels in Russian and English after seeing misleading information being shared on the platform around the war in Ukraine.\n\nIn contrast to traditional networks such as Facebook or Twitter, the content of the main TikTok feed is driven less by who you follow and more by an algorithm that takes into account what you like and what others are watching.\n\n**_It’s so addictive … and where it lacks in trustworthiness, it excels in presentation. It’s a news source I end up consuming because I’m also often scrolling TikTok for other reasons, but the algorithm ends up providing news anyways._** _Female, 22, USA_\n\nBut others remain sceptical about the value of very short videos designed for popularity.\n\n**_I greatly dislike everything about the presentation of TikTok [and Instagram]. Everything about it screams a lack of professionalism and research and I cannot trust it, along with it being obnoxiously loud._** _Male, 18, USA_\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 33\n\nPROPORTION THAT USED EACH SOCIAL NETWORK FOR NEWS IN THE LAST WEEK – SELECTED REGIONS\n\n**Europe**\n\nFacebook\n\nInstagram\n\nFB Messenger\n\nTwitter\n\nTelegram\n\nTikTok\n\nYouTube\n\nWhatsApp\n\n**Asia Latin America Africa North America**\n\n**16%**\n\n**23%**\n\n**6%**\n\n**25%**\n\n**24%**\n\n**34%**\n\n**34%**\n\n**10%**\n\n**43%**\n\n**18%**\n\n**16%**\n\n**57%**\n\n**28% 29%**\n\n**15%**\n\n**59%**\n\n**55%**\n\n**42% 35% 44%**\n\n**35%**\n\n**40%**\n\n**14%**\n\n**13% 11%**\n\n**14% 11%**\n\n**11% 11%**\n\n**9%**\n\n**4%**\n\n**5%**\n\n**2%**\n\n**6%**\n\n**9%**\n\n**9%**\n\n**6%**\n\n**11%**\n\n**18%**\n\n**14%**\n\n**Q12B.** Which, if any, of the following have you used for news in the last week? _Base: Total sample in each region. Europe = 48,836, North America = 4048, Asia = 20,349, Latin America = 12,104, Africa = 6057. Note: Africa average is Kenya, South Africa, and Nigeria only (English speakers in South Africa and Nigeria)._\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 34\n\n## **40%**\n\nof 18-24s use TikTok for any purpose\n\n## **15%**\n\nof 18-24s use TikTok for news\n\n(Across all markets)\n\nKenya **43%** / **15%**\n\nThailand **42%** / **22%**\n\nSouth Africa **41%** / **15%**\n\nPeru **37%** / **16%**\n\nMexico **35%** / **10%**\n\nMarkets with highest usage (all ages) **% for any purpose** / **for news**\n\n**Q12a/b.** Which, if any, of the following have you used in the last week for any purpose/news? _Base: Total sample in each market (n ≈ 2000)._\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 25","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Outside Western countries – especially in Latin America and Africa – we find a much higher proportion of our sample using social media for news, but we also find different networks in play. In African countries such as Nigeria, South Africa, and Kenya, WhatsApp (55%) and Telegram (18%) combined are more important for finding, sharing, and discussing news than Facebook (59%). Latin Americans tend to use a combination of all the main networks, while it is a mixed picture in Asia, with almost three-quarters (73%) using Facebook for news in the Philippines but just 5% in Japan, where Twitter (18%) and Line (16%) are more popular. YouTube (44%) is the main social network for news in South Korea, along with home-grown app KakaoTalk (24%) and KakaoStory (5%).\n\n**TikTok emerges as a significant new player in the news ecosystem**\n\nIn our data, we find significant and rapidly growing usage of TikTok, especially in Africa, Asia, and Latin America as well as across Eastern Europe. Usage is mostly still with under 25s, but is extending into all age brackets in countries with higher levels of usage, such as Kenya, South Africa, Thailand, Indonesia, Brazil, and Peru. This quote from our Brazil qualitative research sums up how perceptions of content on TikTok have changed in the last year or so.\n\n**_When TikTok was launched, it was just about dancing, choreography; not today, even though the videos are fast, they bring immediate information._** _Male, 24, Brazil_\n\nThe Russia-Ukraine conflict has increased the profile of the network globally. Ukrainians have been documenting their experience of the war, including leaving their homes as refugees, clocking up millions of views in the process (see picture above right). As we noted last year, many publishers have been increasing their investment, but some worry that a platform where entertainment content is so sought-after may not be the most effective place for news. BBC News originally decided to avoid TikTok, but has now set up channels in Russian and English after seeing misleading information being shared on the platform around the war in Ukraine.\n\nIn contrast to traditional networks such as Facebook or Twitter, the content of the main TikTok feed is driven less by who you follow and more by an algorithm that takes into account what you like and what others are watching.\n\n**_It’s so addictive … and where it lacks in trustworthiness, it excels in presentation. It’s a news source I end up consuming because I’m also often scrolling TikTok for other reasons, but the algorithm ends up providing news anyways._** _Female, 22, USA_\n\nBut others remain sceptical about the value of very short videos designed for popularity.\n\n**_I greatly dislike everything about the presentation of TikTok [and Instagram]. Everything about it screams a lack of professionalism and research and I cannot trust it, along with it being obnoxiously loud._** _Male, 18, USA_\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 33\n\nPROPORTION THAT USED EACH SOCIAL NETWORK FOR NEWS IN THE LAST WEEK – SELECTED REGIONS\n\n**Europe**\n\nFacebook\n\nInstagram\n\nFB Messenger\n\nTwitter\n\nTelegram\n\nTikTok\n\nYouTube\n\nWhatsApp\n\n**Asia Latin America Africa North America**\n\n**16%**\n\n**23%**\n\n**6%**\n\n**25%**\n\n**24%**\n\n**34%**\n\n**34%**\n\n**10%**\n\n**43%**\n\n**18%**\n\n**16%**\n\n**57%**\n\n**28% 29%**\n\n**15%**\n\n**59%**\n\n**55%**\n\n**42% 35% 44%**\n\n**35%**\n\n**40%**\n\n**14%**\n\n**13% 11%**\n\n**14% 11%**\n\n**11% 11%**\n\n**9%**\n\n**4%**\n\n**5%**\n\n**2%**\n\n**6%**\n\n**9%**\n\n**9%**\n\n**6%**\n\n**11%**\n\n**18%**\n\n**14%**\n\n**Q12B.** Which, if any, of the following have you used for news in the last week? _Base: Total sample in each region. Europe = 48,836, North America = 4048, Asia = 20,349, Latin America = 12,104, Africa = 6057. Note: Africa average is Kenya, South Africa, and Nigeria only (English speakers in South Africa and Nigeria)._\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 34\n\n## **40%**\n\nof 18-24s use TikTok for any purpose\n\n## **15%**\n\nof 18-24s use TikTok for news\n\n(Across all markets)\n\nKenya **43%** / **15%**\n\nThailand **42%** / **22%**\n\nSouth Africa **41%** / **15%**\n\nPeru **37%** / **16%**\n\nMexico **35%** / **10%**\n\nMarkets with highest usage (all ages) **% for any purpose** / **for news**\n\n**Q12a/b.** Which, if any, of the following have you used in the last week for any purpose/news? _Base: Total sample in each market (n ≈ 2000)._\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 25","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Digital_News-Report_2022_p25.pdf"]},"response":{"errors":[],"extract_id":"extract_ec0f07b585614572b8e299ca38fa06e9","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Outside Western countries – especially in Latin America and Africa – we find a much higher proportion of our sample using social media for news, but we also find different networks in play. In African countries such as Nigeria, South Africa, and Kenya, WhatsApp (55%) and Telegram (18%) combined are more important for finding, sharing, and discussing news than Facebook (59%). Latin Americans tend to use a combination of all the main networks, while it is a mixed picture in Asia, with almost three-quarters (73%) using Facebook for news in the Philippines but just 5% in Japan, where Twitter (18%) and Line (16%) are more popular. YouTube (44%) is the main social network for news in South Korea, along with home-grown app KakaoTalk (24%) and KakaoStory (5%).\n\n**TikTok emerges as a significant new player in the news ecosystem**\n\nIn our data, we find significant and rapidly growing usage of TikTok, especially in Africa, Asia, and Latin America as well as across Eastern Europe. Usage is mostly still with under 25s, but is extending into all age brackets in countries with higher levels of usage, such as Kenya, South Africa, Thailand, Indonesia, Brazil, and Peru. This quote from our Brazil qualitative research sums up how perceptions of content on TikTok have changed in the last year or so.\n\n**_When TikTok was launched, it was just about dancing, choreography; not today, even though the videos are fast, they bring immediate information._** _Male, 24, Brazil_\n\nThe Russia-Ukraine conflict has increased the profile of the network globally. Ukrainians have been documenting their experience of the war, including leaving their homes as refugees, clocking up millions of views in the process (see picture above right). As we noted last year, many publishers have been increasing their investment, but some worry that a platform where entertainment content is so sought-after may not be the most effective place for news. BBC News originally decided to avoid TikTok, but has now set up channels in Russian and English after seeing misleading information being shared on the platform around the war in Ukraine.\n\nIn contrast to traditional networks such as Facebook or Twitter, the content of the main TikTok feed is driven less by who you follow and more by an algorithm that takes into account what you like and what others are watching.\n\n**_It’s so addictive … and where it lacks in trustworthiness, it excels in presentation. It’s a news source I end up consuming because I’m also often scrolling TikTok for other reasons, but the algorithm ends up providing news anyways._** _Female, 22, USA_\n\nBut others remain sceptical about the value of very short videos designed for popularity.\n\n**_I greatly dislike everything about the presentation of TikTok [and Instagram]. Everything about it screams a lack of professionalism and research and I cannot trust it, along with it being obnoxiously loud._** _Male, 18, USA_\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 33\n\nPROPORTION THAT USED EACH SOCIAL NETWORK FOR NEWS IN THE LAST WEEK – SELECTED REGIONS\n\n**Europe**\n\nFacebook\n\nInstagram\n\nFB Messenger\n\nTwitter\n\nTelegram\n\nTikTok\n\nYouTube\n\nWhatsApp\n\n**Asia Latin America Africa North America**\n\n**16%**\n\n**23%**\n\n**6%**\n\n**25%**\n\n**24%**\n\n**34%**\n\n**34%**\n\n**10%**\n\n**43%**\n\n**18%**\n\n**16%**\n\n**57%**\n\n**28% 29%**\n\n**15%**\n\n**59%**\n\n**55%**\n\n**42% 35% 44%**\n\n**35%**\n\n**40%**\n\n**14%**\n\n**13% 11%**\n\n**14% 11%**\n\n**11% 11%**\n\n**9%**\n\n**4%**\n\n**5%**\n\n**2%**\n\n**6%**\n\n**9%**\n\n**9%**\n\n**6%**\n\n**11%**\n\n**18%**\n\n**14%**\n\n**Q12B.** Which, if any, of the following have you used for news in the last week? _Base: Total sample in each region. Europe = 48,836, North America = 4048, Asia = 20,349, Latin America = 12,104, Africa = 6057. Note: Africa average is Kenya, South Africa, and Nigeria only (English speakers in South Africa and Nigeria)._\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 34\n\n## **40%**\n\nof 18-24s use TikTok for any purpose\n\n## **15%**\n\nof 18-24s use TikTok for news\n\n(Across all markets)\n\nKenya **43%** / **15%**\n\nThailand **42%** / **22%**\n\nSouth Africa **41%** / **15%**\n\nPeru **37%** / **16%**\n\nMexico **35%** / **10%**\n\nMarkets with highest usage (all ages) **% for any purpose** / **for news**\n\n**Q12a/b.** Which, if any, of the following have you used in the last week for any purpose/news? _Base: Total sample in each market (n ≈ 2000)._\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 25"],"full_content":"Outside Western countries – especially in Latin America and Africa – we find a much higher proportion of our sample using social media for news, but we also find different networks in play. In African countries such as Nigeria, South Africa, and Kenya, WhatsApp (55%) and Telegram (18%) combined are more important for finding, sharing, and discussing news than Facebook (59%). Latin Americans tend to use a combination of all the main networks, while it is a mixed picture in Asia, with almost three-quarters (73%) using Facebook for news in the Philippines but just 5% in Japan, where Twitter (18%) and Line (16%) are more popular. YouTube (44%) is the main social network for news in South Korea, along with home-grown app KakaoTalk (24%) and KakaoStory (5%).\n\n**TikTok emerges as a significant new player in the news ecosystem**\n\nIn our data, we find significant and rapidly growing usage of TikTok, especially in Africa, Asia, and Latin America as well as across Eastern Europe. Usage is mostly still with under 25s, but is extending into all age brackets in countries with higher levels of usage, such as Kenya, South Africa, Thailand, Indonesia, Brazil, and Peru. This quote from our Brazil qualitative research sums up how perceptions of content on TikTok have changed in the last year or so.\n\n**_When TikTok was launched, it was just about dancing, choreography; not today, even though the videos are fast, they bring immediate information._** _Male, 24, Brazil_\n\nThe Russia-Ukraine conflict has increased the profile of the network globally. Ukrainians have been documenting their experience of the war, including leaving their homes as refugees, clocking up millions of views in the process (see picture above right). As we noted last year, many publishers have been increasing their investment, but some worry that a platform where entertainment content is so sought-after may not be the most effective place for news. BBC News originally decided to avoid TikTok, but has now set up channels in Russian and English after seeing misleading information being shared on the platform around the war in Ukraine.\n\nIn contrast to traditional networks such as Facebook or Twitter, the content of the main TikTok feed is driven less by who you follow and more by an algorithm that takes into account what you like and what others are watching.\n\n**_It’s so addictive … and where it lacks in trustworthiness, it excels in presentation. It’s a news source I end up consuming because I’m also often scrolling TikTok for other reasons, but the algorithm ends up providing news anyways._** _Female, 22, USA_\n\nBut others remain sceptical about the value of very short videos designed for popularity.\n\n**_I greatly dislike everything about the presentation of TikTok [and Instagram]. Everything about it screams a lack of professionalism and research and I cannot trust it, along with it being obnoxiously loud._** _Male, 18, USA_\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 33\n\nPROPORTION THAT USED EACH SOCIAL NETWORK FOR NEWS IN THE LAST WEEK – SELECTED REGIONS\n\n**Europe**\n\nFacebook\n\nInstagram\n\nFB Messenger\n\nTwitter\n\nTelegram\n\nTikTok\n\nYouTube\n\nWhatsApp\n\n**Asia Latin America Africa North America**\n\n**16%**\n\n**23%**\n\n**6%**\n\n**25%**\n\n**24%**\n\n**34%**\n\n**34%**\n\n**10%**\n\n**43%**\n\n**18%**\n\n**16%**\n\n**57%**\n\n**28% 29%**\n\n**15%**\n\n**59%**\n\n**55%**\n\n**42% 35% 44%**\n\n**35%**\n\n**40%**\n\n**14%**\n\n**13% 11%**\n\n**14% 11%**\n\n**11% 11%**\n\n**9%**\n\n**4%**\n\n**5%**\n\n**2%**\n\n**6%**\n\n**9%**\n\n**9%**\n\n**6%**\n\n**11%**\n\n**18%**\n\n**14%**\n\n**Q12B.** Which, if any, of the following have you used for news in the last week? _Base: Total sample in each region. Europe = 48,836, North America = 4048, Asia = 20,349, Latin America = 12,104, Africa = 6057. Note: Africa average is Kenya, South Africa, and Nigeria only (English speakers in South Africa and Nigeria)._\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 34\n\n## **40%**\n\nof 18-24s use TikTok for any purpose\n\n## **15%**\n\nof 18-24s use TikTok for news\n\n(Across all markets)\n\nKenya **43%** / **15%**\n\nThailand **42%** / **22%**\n\nSouth Africa **41%** / **15%**\n\nPeru **37%** / **16%**\n\nMexico **35%** / **10%**\n\nMarkets with highest usage (all ages) **% for any purpose** / **for news**\n\n**Q12a/b.** Which, if any, of the following have you used in the last week for any purpose/news? _Base: Total sample in each market (n ≈ 2000)._\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 25","publish_date":null,"title":"Graph 33 Graph 34 40% 15%","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Digital_News-Report_2022_p25.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Digital_News-Report_2022_p25.pdf"},"request":{"example_id":"chart/Digital_News-Report_2022_p25","product_type":"parse","source_file_path":"Digital_News-Report_2022_p25.pdf"},"started_at":"2026-11-04T19:46:15.615973"} +{"completed_at":"2026-11-04T19:46:22.070614","latency_in_ms":4795,"output":{"example_id":"chart/Digital_News-Report_2022_p26","grounded_pages":[],"layout_pages":[],"markdown":"**How much news do people see on each platform?**\n\nFacebook tells us that about 4% of the content of the average person’s feed comes from news outlets, though for some that proportion will be much higher. ⁹ Twitter does not provide equivalent data, but our survey shows that across markets it has a high proportion (56%) who use it for news weekly. By contrast, only around four in ten (39%) Instagram users access news via the platform weekly and just a third of TikTok users do the same (33%). But do people think they are seeing too much or too little news in these networks? ¹⁰\n\nMore than a fifth of UK respondents (21%) feel that they see too much news on Facebook and only 3% would like to see more. Facebook has been trying to reduce the amount of news on the platform partly for reputational reasons but also because, as previous research shows, people are mainly there to connect with friends and family rather than consume news. On the other hand, Twitter is seen as having more of a reputation for news so it is somewhat surprising to find one in ten (11%) saying they see too much. There is a clear link with news avoidance: those who _often_ avoid the news are twice as likely to say they see too much news on both Facebook and Twitter when compared with the average user.\n\nThese data suggest the ambivalent relationship many people have with news on social media. It drives much of the conversation and engagement but can also interrupt other activities and create arguments. This dilemma may be one of the factors behind Facebook’s recent decision to create a separate tab in five countries (Facebook News) with content from a selection of partner publishers, in parallel with an ongoing string of decisions in recent years to reduce the amount of news in the main ‘Feed’ (recently renamed from ‘News Feed’). Platforms like Twitter might also need to think about why even some people who love the news want to see less of it, given the network’s reputation for often abusive debate. TikTok and Instagram will also be aware, as they grow, of the careful balance they need to strike if they are not to put people off.\n\n9 https://www.cnbc.com/2018/01/19/facebooks-news-feed-survey-users-to-find-high-quality-news.html 10 We did ask respondents directly to estimate how much news they saw in each of their social media feeds, partly to prime them for this subsequent question about too much/too little.\n\nWe have preferred to use data on the proportion who say they use each network for any purpose/for news as they show similar trends and are data we collect every year.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 35\n\nPROPORTION WHO THINK THEY SEE TOO MUCH NEWS ON EACH NETWORK – UK\n\n**Q12\\_Social\\_rightsize ...** Do you think that the amount of content you see from news outlets on . is too much, about right, or not enough? _Base: All that use each social network: Facebook = 1501, Twitter = 685, Instagram = 859, TikTok = 270._\n\n8%\n\n20%\n\n66%\n\n6%\n\n9%\n\n20%\n\n65%\n\n6%\n\nFacebook Twitter Instagram TikTok\n\n21%\n\n20%\n\n55%\n\n3%\n\n11%\n\n13%\n\n73%\n\n3%\n\nToo much About right Not enough Don’t know\n\n**Twitter**\n\nIndia 23%\n\nAustralia 23%\n\nSouth Africa 19%\n\nSouth Korea 15%\n\nBelgium 14%\n\n**Facebook**\n\nPortugal 26%\n\nUSA 22%\n\nHungary 21%\n\nUK 21%\n\nAustralia 20%\n\nCanada 20%\n\n**Top markets saying too much**\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 36\n\nPROPORTION WHO SAW FALSE OR MISLEADING INFORMATION ABOUT EACH TOPIC IN THE LAST WEEK – SELECTED REGIONS\n\n**Q. FAKE\\_NEWS 2021a.** Have you seen false or misleading information about any of the following topics, in the last week? _Base: Total sample in each region: Europe = 48,836, North America = 4048, Asia = 20,349, Latin America = 12,104, Africa = 6057. Note: Africa average is Kenya, South Africa, and Nigeria only (English speakers in South Africa and Nigeria)._\n\nCOVID-19 **47%**\n\n**53%**\n\n**45%**\n\n**54%**\n\n**47%**\n\nPolitics **34%**\n\n**43% 42%**\n\n**51%**\n\n**49%**\n\nCelebrities **20%**\n\n**22%**\n\n**29%**\n\n**33%**\n\n**48%**\n\nClimate change\n\n**19%**\n\n**32%**\n\n**15%**\n\n**21%**\n\n**19%**\n\nImmigration **20%**\n\n**27%**\n\n**9%**\n\n**18%**\n\n**16%**\n\nProducts (e.g. scams)\n\n**16%**\n\n**19%**\n\n**22%**\n\n**26%**\n\n**29%**\n\nEurope North America Asia Latin America Africa\n\n**MISINFORMATION**\n\nIn this year’s survey we find a link between online misinformation fears and the widespread use of social media. Across markets, just over half (54%) say they worry about identifying the difference between what is real and fake on the internet when it comes to news, but people who say they _mainly_ use social media as a source of news are more worried (61%) than people who don’t use it at all (48%). Additionally, regions with the highest levels of concern – Africa and Latin America – correspond closely with high levels of social media news use. This is not to say that social media use causes misinformation, but that usage may generate awareness of and potential exposure to false information, including giving voice to extreme perspectives that previously would not have been widely heard.\n\nWhen looking at the types of misinformation that people claim to see, we find that dubious health claims around COVID-19, including from anti-vaccination groups, are, as in 2021, still most widespread across most regions, ahead of politics. Notable exceptions are in Kenya – where political misinformation is more widely seen – as well as Colombia and the Philippines, both countries where elections were held this year.\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 26","pages":[{"line_numbers":[],"markdown":"**How much news do people see on each platform?**\n\nFacebook tells us that about 4% of the content of the average person’s feed comes from news outlets, though for some that proportion will be much higher. ⁹ Twitter does not provide equivalent data, but our survey shows that across markets it has a high proportion (56%) who use it for news weekly. By contrast, only around four in ten (39%) Instagram users access news via the platform weekly and just a third of TikTok users do the same (33%). But do people think they are seeing too much or too little news in these networks? ¹⁰\n\nMore than a fifth of UK respondents (21%) feel that they see too much news on Facebook and only 3% would like to see more. Facebook has been trying to reduce the amount of news on the platform partly for reputational reasons but also because, as previous research shows, people are mainly there to connect with friends and family rather than consume news. On the other hand, Twitter is seen as having more of a reputation for news so it is somewhat surprising to find one in ten (11%) saying they see too much. There is a clear link with news avoidance: those who _often_ avoid the news are twice as likely to say they see too much news on both Facebook and Twitter when compared with the average user.\n\nThese data suggest the ambivalent relationship many people have with news on social media. It drives much of the conversation and engagement but can also interrupt other activities and create arguments. This dilemma may be one of the factors behind Facebook’s recent decision to create a separate tab in five countries (Facebook News) with content from a selection of partner publishers, in parallel with an ongoing string of decisions in recent years to reduce the amount of news in the main ‘Feed’ (recently renamed from ‘News Feed’). Platforms like Twitter might also need to think about why even some people who love the news want to see less of it, given the network’s reputation for often abusive debate. TikTok and Instagram will also be aware, as they grow, of the careful balance they need to strike if they are not to put people off.\n\n9 https://www.cnbc.com/2018/01/19/facebooks-news-feed-survey-users-to-find-high-quality-news.html 10 We did ask respondents directly to estimate how much news they saw in each of their social media feeds, partly to prime them for this subsequent question about too much/too little.\n\nWe have preferred to use data on the proportion who say they use each network for any purpose/for news as they show similar trends and are data we collect every year.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 35\n\nPROPORTION WHO THINK THEY SEE TOO MUCH NEWS ON EACH NETWORK – UK\n\n**Q12\\_Social\\_rightsize ...** Do you think that the amount of content you see from news outlets on . is too much, about right, or not enough? _Base: All that use each social network: Facebook = 1501, Twitter = 685, Instagram = 859, TikTok = 270._\n\n8%\n\n20%\n\n66%\n\n6%\n\n9%\n\n20%\n\n65%\n\n6%\n\nFacebook Twitter Instagram TikTok\n\n21%\n\n20%\n\n55%\n\n3%\n\n11%\n\n13%\n\n73%\n\n3%\n\nToo much About right Not enough Don’t know\n\n**Twitter**\n\nIndia 23%\n\nAustralia 23%\n\nSouth Africa 19%\n\nSouth Korea 15%\n\nBelgium 14%\n\n**Facebook**\n\nPortugal 26%\n\nUSA 22%\n\nHungary 21%\n\nUK 21%\n\nAustralia 20%\n\nCanada 20%\n\n**Top markets saying too much**\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 36\n\nPROPORTION WHO SAW FALSE OR MISLEADING INFORMATION ABOUT EACH TOPIC IN THE LAST WEEK – SELECTED REGIONS\n\n**Q. FAKE\\_NEWS 2021a.** Have you seen false or misleading information about any of the following topics, in the last week? _Base: Total sample in each region: Europe = 48,836, North America = 4048, Asia = 20,349, Latin America = 12,104, Africa = 6057. Note: Africa average is Kenya, South Africa, and Nigeria only (English speakers in South Africa and Nigeria)._\n\nCOVID-19 **47%**\n\n**53%**\n\n**45%**\n\n**54%**\n\n**47%**\n\nPolitics **34%**\n\n**43% 42%**\n\n**51%**\n\n**49%**\n\nCelebrities **20%**\n\n**22%**\n\n**29%**\n\n**33%**\n\n**48%**\n\nClimate change\n\n**19%**\n\n**32%**\n\n**15%**\n\n**21%**\n\n**19%**\n\nImmigration **20%**\n\n**27%**\n\n**9%**\n\n**18%**\n\n**16%**\n\nProducts (e.g. scams)\n\n**16%**\n\n**19%**\n\n**22%**\n\n**26%**\n\n**29%**\n\nEurope North America Asia Latin America Africa\n\n**MISINFORMATION**\n\nIn this year’s survey we find a link between online misinformation fears and the widespread use of social media. Across markets, just over half (54%) say they worry about identifying the difference between what is real and fake on the internet when it comes to news, but people who say they _mainly_ use social media as a source of news are more worried (61%) than people who don’t use it at all (48%). Additionally, regions with the highest levels of concern – Africa and Latin America – correspond closely with high levels of social media news use. This is not to say that social media use causes misinformation, but that usage may generate awareness of and potential exposure to false information, including giving voice to extreme perspectives that previously would not have been widely heard.\n\nWhen looking at the types of misinformation that people claim to see, we find that dubious health claims around COVID-19, including from anti-vaccination groups, are, as in 2021, still most widespread across most regions, ahead of politics. Notable exceptions are in Kenya – where political misinformation is more widely seen – as well as Colombia and the Philippines, both countries where elections were held this year.\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 26","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**How much news do people see on each platform?**\n\nFacebook tells us that about 4% of the content of the average person’s feed comes from news outlets, though for some that proportion will be much higher. ⁹ Twitter does not provide equivalent data, but our survey shows that across markets it has a high proportion (56%) who use it for news weekly. By contrast, only around four in ten (39%) Instagram users access news via the platform weekly and just a third of TikTok users do the same (33%). But do people think they are seeing too much or too little news in these networks? ¹⁰\n\nMore than a fifth of UK respondents (21%) feel that they see too much news on Facebook and only 3% would like to see more. Facebook has been trying to reduce the amount of news on the platform partly for reputational reasons but also because, as previous research shows, people are mainly there to connect with friends and family rather than consume news. On the other hand, Twitter is seen as having more of a reputation for news so it is somewhat surprising to find one in ten (11%) saying they see too much. There is a clear link with news avoidance: those who _often_ avoid the news are twice as likely to say they see too much news on both Facebook and Twitter when compared with the average user.\n\nThese data suggest the ambivalent relationship many people have with news on social media. It drives much of the conversation and engagement but can also interrupt other activities and create arguments. This dilemma may be one of the factors behind Facebook’s recent decision to create a separate tab in five countries (Facebook News) with content from a selection of partner publishers, in parallel with an ongoing string of decisions in recent years to reduce the amount of news in the main ‘Feed’ (recently renamed from ‘News Feed’). Platforms like Twitter might also need to think about why even some people who love the news want to see less of it, given the network’s reputation for often abusive debate. TikTok and Instagram will also be aware, as they grow, of the careful balance they need to strike if they are not to put people off.\n\n9 https://www.cnbc.com/2018/01/19/facebooks-news-feed-survey-users-to-find-high-quality-news.html 10 We did ask respondents directly to estimate how much news they saw in each of their social media feeds, partly to prime them for this subsequent question about too much/too little.\n\nWe have preferred to use data on the proportion who say they use each network for any purpose/for news as they show similar trends and are data we collect every year.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 35\n\nPROPORTION WHO THINK THEY SEE TOO MUCH NEWS ON EACH NETWORK – UK\n\n**Q12\\_Social\\_rightsize ...** Do you think that the amount of content you see from news outlets on . is too much, about right, or not enough? _Base: All that use each social network: Facebook = 1501, Twitter = 685, Instagram = 859, TikTok = 270._\n\n8%\n\n20%\n\n66%\n\n6%\n\n9%\n\n20%\n\n65%\n\n6%\n\nFacebook Twitter Instagram TikTok\n\n21%\n\n20%\n\n55%\n\n3%\n\n11%\n\n13%\n\n73%\n\n3%\n\nToo much About right Not enough Don’t know\n\n**Twitter**\n\nIndia 23%\n\nAustralia 23%\n\nSouth Africa 19%\n\nSouth Korea 15%\n\nBelgium 14%\n\n**Facebook**\n\nPortugal 26%\n\nUSA 22%\n\nHungary 21%\n\nUK 21%\n\nAustralia 20%\n\nCanada 20%\n\n**Top markets saying too much**\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 36\n\nPROPORTION WHO SAW FALSE OR MISLEADING INFORMATION ABOUT EACH TOPIC IN THE LAST WEEK – SELECTED REGIONS\n\n**Q. FAKE\\_NEWS 2021a.** Have you seen false or misleading information about any of the following topics, in the last week? _Base: Total sample in each region: Europe = 48,836, North America = 4048, Asia = 20,349, Latin America = 12,104, Africa = 6057. Note: Africa average is Kenya, South Africa, and Nigeria only (English speakers in South Africa and Nigeria)._\n\nCOVID-19 **47%**\n\n**53%**\n\n**45%**\n\n**54%**\n\n**47%**\n\nPolitics **34%**\n\n**43% 42%**\n\n**51%**\n\n**49%**\n\nCelebrities **20%**\n\n**22%**\n\n**29%**\n\n**33%**\n\n**48%**\n\nClimate change\n\n**19%**\n\n**32%**\n\n**15%**\n\n**21%**\n\n**19%**\n\nImmigration **20%**\n\n**27%**\n\n**9%**\n\n**18%**\n\n**16%**\n\nProducts (e.g. scams)\n\n**16%**\n\n**19%**\n\n**22%**\n\n**26%**\n\n**29%**\n\nEurope North America Asia Latin America Africa\n\n**MISINFORMATION**\n\nIn this year’s survey we find a link between online misinformation fears and the widespread use of social media. Across markets, just over half (54%) say they worry about identifying the difference between what is real and fake on the internet when it comes to news, but people who say they _mainly_ use social media as a source of news are more worried (61%) than people who don’t use it at all (48%). Additionally, regions with the highest levels of concern – Africa and Latin America – correspond closely with high levels of social media news use. This is not to say that social media use causes misinformation, but that usage may generate awareness of and potential exposure to false information, including giving voice to extreme perspectives that previously would not have been widely heard.\n\nWhen looking at the types of misinformation that people claim to see, we find that dubious health claims around COVID-19, including from anti-vaccination groups, are, as in 2021, still most widespread across most regions, ahead of politics. Notable exceptions are in Kenya – where political misinformation is more widely seen – as well as Colombia and the Philippines, both countries where elections were held this year.\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 26","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Digital_News-Report_2022_p26.pdf"]},"response":{"errors":[],"extract_id":"extract_fe33008c3c872593a994bdb6fc453d35","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**How much news do people see on each platform?**\n\nFacebook tells us that about 4% of the content of the average person’s feed comes from news outlets, though for some that proportion will be much higher. ⁹ Twitter does not provide equivalent data, but our survey shows that across markets it has a high proportion (56%) who use it for news weekly. By contrast, only around four in ten (39%) Instagram users access news via the platform weekly and just a third of TikTok users do the same (33%). But do people think they are seeing too much or too little news in these networks? ¹⁰\n\nMore than a fifth of UK respondents (21%) feel that they see too much news on Facebook and only 3% would like to see more. Facebook has been trying to reduce the amount of news on the platform partly for reputational reasons but also because, as previous research shows, people are mainly there to connect with friends and family rather than consume news. On the other hand, Twitter is seen as having more of a reputation for news so it is somewhat surprising to find one in ten (11%) saying they see too much. There is a clear link with news avoidance: those who _often_ avoid the news are twice as likely to say they see too much news on both Facebook and Twitter when compared with the average user.\n\nThese data suggest the ambivalent relationship many people have with news on social media. It drives much of the conversation and engagement but can also interrupt other activities and create arguments. This dilemma may be one of the factors behind Facebook’s recent decision to create a separate tab in five countries (Facebook News) with content from a selection of partner publishers, in parallel with an ongoing string of decisions in recent years to reduce the amount of news in the main ‘Feed’ (recently renamed from ‘News Feed’). Platforms like Twitter might also need to think about why even some people who love the news want to see less of it, given the network’s reputation for often abusive debate. TikTok and Instagram will also be aware, as they grow, of the careful balance they need to strike if they are not to put people off.\n\n9 https://www.cnbc.com/2018/01/19/facebooks-news-feed-survey-users-to-find-high-quality-news.html 10 We did ask respondents directly to estimate how much news they saw in each of their social media feeds, partly to prime them for this subsequent question about too much/too little.\n\nWe have preferred to use data on the proportion who say they use each network for any purpose/for news as they show similar trends and are data we collect every year.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 35\n\nPROPORTION WHO THINK THEY SEE TOO MUCH NEWS ON EACH NETWORK – UK\n\n**Q12\\_Social\\_rightsize ...** Do you think that the amount of content you see from news outlets on . is too much, about right, or not enough? _Base: All that use each social network: Facebook = 1501, Twitter = 685, Instagram = 859, TikTok = 270._\n\n8%\n\n20%\n\n66%\n\n6%\n\n9%\n\n20%\n\n65%\n\n6%\n\nFacebook Twitter Instagram TikTok\n\n21%\n\n20%\n\n55%\n\n3%\n\n11%\n\n13%\n\n73%\n\n3%\n\nToo much About right Not enough Don’t know\n\n**Twitter**\n\nIndia 23%\n\nAustralia 23%\n\nSouth Africa 19%\n\nSouth Korea 15%\n\nBelgium 14%\n\n**Facebook**\n\nPortugal 26%\n\nUSA 22%\n\nHungary 21%\n\nUK 21%\n\nAustralia 20%\n\nCanada 20%\n\n**Top markets saying too much**\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 36\n\nPROPORTION WHO SAW FALSE OR MISLEADING INFORMATION ABOUT EACH TOPIC IN THE LAST WEEK – SELECTED REGIONS\n\n**Q. FAKE\\_NEWS 2021a.** Have you seen false or misleading information about any of the following topics, in the last week? _Base: Total sample in each region: Europe = 48,836, North America = 4048, Asia = 20,349, Latin America = 12,104, Africa = 6057. Note: Africa average is Kenya, South Africa, and Nigeria only (English speakers in South Africa and Nigeria)._\n\nCOVID-19 **47%**\n\n**53%**\n\n**45%**\n\n**54%**\n\n**47%**\n\nPolitics **34%**\n\n**43% 42%**\n\n**51%**\n\n**49%**\n\nCelebrities **20%**\n\n**22%**\n\n**29%**\n\n**33%**\n\n**48%**\n\nClimate change\n\n**19%**\n\n**32%**\n\n**15%**\n\n**21%**\n\n**19%**\n\nImmigration **20%**\n\n**27%**\n\n**9%**\n\n**18%**\n\n**16%**\n\nProducts (e.g. scams)\n\n**16%**\n\n**19%**\n\n**22%**\n\n**26%**\n\n**29%**\n\nEurope North America Asia Latin America Africa\n\n**MISINFORMATION**\n\nIn this year’s survey we find a link between online misinformation fears and the widespread use of social media. Across markets, just over half (54%) say they worry about identifying the difference between what is real and fake on the internet when it comes to news, but people who say they _mainly_ use social media as a source of news are more worried (61%) than people who don’t use it at all (48%). Additionally, regions with the highest levels of concern – Africa and Latin America – correspond closely with high levels of social media news use. This is not to say that social media use causes misinformation, but that usage may generate awareness of and potential exposure to false information, including giving voice to extreme perspectives that previously would not have been widely heard.\n\nWhen looking at the types of misinformation that people claim to see, we find that dubious health claims around COVID-19, including from anti-vaccination groups, are, as in 2021, still most widespread across most regions, ahead of politics. Notable exceptions are in Kenya – where political misinformation is more widely seen – as well as Colombia and the Philippines, both countries where elections were held this year.\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 26"],"full_content":"**How much news do people see on each platform?**\n\nFacebook tells us that about 4% of the content of the average person’s feed comes from news outlets, though for some that proportion will be much higher. ⁹ Twitter does not provide equivalent data, but our survey shows that across markets it has a high proportion (56%) who use it for news weekly. By contrast, only around four in ten (39%) Instagram users access news via the platform weekly and just a third of TikTok users do the same (33%). But do people think they are seeing too much or too little news in these networks? ¹⁰\n\nMore than a fifth of UK respondents (21%) feel that they see too much news on Facebook and only 3% would like to see more. Facebook has been trying to reduce the amount of news on the platform partly for reputational reasons but also because, as previous research shows, people are mainly there to connect with friends and family rather than consume news. On the other hand, Twitter is seen as having more of a reputation for news so it is somewhat surprising to find one in ten (11%) saying they see too much. There is a clear link with news avoidance: those who _often_ avoid the news are twice as likely to say they see too much news on both Facebook and Twitter when compared with the average user.\n\nThese data suggest the ambivalent relationship many people have with news on social media. It drives much of the conversation and engagement but can also interrupt other activities and create arguments. This dilemma may be one of the factors behind Facebook’s recent decision to create a separate tab in five countries (Facebook News) with content from a selection of partner publishers, in parallel with an ongoing string of decisions in recent years to reduce the amount of news in the main ‘Feed’ (recently renamed from ‘News Feed’). Platforms like Twitter might also need to think about why even some people who love the news want to see less of it, given the network’s reputation for often abusive debate. TikTok and Instagram will also be aware, as they grow, of the careful balance they need to strike if they are not to put people off.\n\n9 https://www.cnbc.com/2018/01/19/facebooks-news-feed-survey-users-to-find-high-quality-news.html 10 We did ask respondents directly to estimate how much news they saw in each of their social media feeds, partly to prime them for this subsequent question about too much/too little.\n\nWe have preferred to use data on the proportion who say they use each network for any purpose/for news as they show similar trends and are data we collect every year.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 35\n\nPROPORTION WHO THINK THEY SEE TOO MUCH NEWS ON EACH NETWORK – UK\n\n**Q12\\_Social\\_rightsize ...** Do you think that the amount of content you see from news outlets on . is too much, about right, or not enough? _Base: All that use each social network: Facebook = 1501, Twitter = 685, Instagram = 859, TikTok = 270._\n\n8%\n\n20%\n\n66%\n\n6%\n\n9%\n\n20%\n\n65%\n\n6%\n\nFacebook Twitter Instagram TikTok\n\n21%\n\n20%\n\n55%\n\n3%\n\n11%\n\n13%\n\n73%\n\n3%\n\nToo much About right Not enough Don’t know\n\n**Twitter**\n\nIndia 23%\n\nAustralia 23%\n\nSouth Africa 19%\n\nSouth Korea 15%\n\nBelgium 14%\n\n**Facebook**\n\nPortugal 26%\n\nUSA 22%\n\nHungary 21%\n\nUK 21%\n\nAustralia 20%\n\nCanada 20%\n\n**Top markets saying too much**\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 36\n\nPROPORTION WHO SAW FALSE OR MISLEADING INFORMATION ABOUT EACH TOPIC IN THE LAST WEEK – SELECTED REGIONS\n\n**Q. FAKE\\_NEWS 2021a.** Have you seen false or misleading information about any of the following topics, in the last week? _Base: Total sample in each region: Europe = 48,836, North America = 4048, Asia = 20,349, Latin America = 12,104, Africa = 6057. Note: Africa average is Kenya, South Africa, and Nigeria only (English speakers in South Africa and Nigeria)._\n\nCOVID-19 **47%**\n\n**53%**\n\n**45%**\n\n**54%**\n\n**47%**\n\nPolitics **34%**\n\n**43% 42%**\n\n**51%**\n\n**49%**\n\nCelebrities **20%**\n\n**22%**\n\n**29%**\n\n**33%**\n\n**48%**\n\nClimate change\n\n**19%**\n\n**32%**\n\n**15%**\n\n**21%**\n\n**19%**\n\nImmigration **20%**\n\n**27%**\n\n**9%**\n\n**18%**\n\n**16%**\n\nProducts (e.g. scams)\n\n**16%**\n\n**19%**\n\n**22%**\n\n**26%**\n\n**29%**\n\nEurope North America Asia Latin America Africa\n\n**MISINFORMATION**\n\nIn this year’s survey we find a link between online misinformation fears and the widespread use of social media. Across markets, just over half (54%) say they worry about identifying the difference between what is real and fake on the internet when it comes to news, but people who say they _mainly_ use social media as a source of news are more worried (61%) than people who don’t use it at all (48%). Additionally, regions with the highest levels of concern – Africa and Latin America – correspond closely with high levels of social media news use. This is not to say that social media use causes misinformation, but that usage may generate awareness of and potential exposure to false information, including giving voice to extreme perspectives that previously would not have been widely heard.\n\nWhen looking at the types of misinformation that people claim to see, we find that dubious health claims around COVID-19, including from anti-vaccination groups, are, as in 2021, still most widespread across most regions, ahead of politics. Notable exceptions are in Kenya – where political misinformation is more widely seen – as well as Colombia and the Philippines, both countries where elections were held this year.\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 26","publish_date":null,"title":"Graph 35 Graph 36","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Digital_News-Report_2022_p26.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Digital_News-Report_2022_p26.pdf"},"request":{"example_id":"chart/Digital_News-Report_2022_p26","product_type":"parse","source_file_path":"Digital_News-Report_2022_p26.pdf"},"started_at":"2026-11-04T19:46:17.274854"} +{"completed_at":"2026-11-04T19:46:21.222716","latency_in_ms":2252,"output":{"example_id":"chart/Digital_News-Report_2022_p27","grounded_pages":[],"layout_pages":[],"markdown":"11 https://eu.usatoday.com/story/tech/2022/01/21/climate-change-misinformation-facebook-youtube-twitter/6594691001/\n\nLevels of perceived misinformation around climate change and the environment are around three times higher in the United States (34%) than they are in Taiwan (10%) or Denmark (13%). Despite pledges to crack down, social media posts and videos denying climate change or disputing its causes remain widespread on Twitter, Facebook, YouTube, and TikTok, according to recent research. ¹¹\n\nIn this year’s report we find increased polarisation around climate change in a number of countries, including the United States, and identify the most important sources for climate change news with different groups.\n\n_For further analysis see Section 2.6: How People Access and Think about Climate Change News_\n\n**TEXT IS STILL KING BUT WHAT ROLE FOR ONLINE VIDEO?**\n\nSince the emergence of the internet, consumption on news websites (and apps) has primarily been about reading text articles. But that has begun to change with the growing supply of video news formats on social media.\n\nAgainst this background, it is surprising to find that all age groups, on average, say they still prefer to read news online rather than watch it – and we have seen little change in underlying preferences since we last asked the question in 2019. Younger audiences, however, are significantly more likely to say they watch the news, perhaps because they are more exposed to networks like Facebook, Instagram, and TikTok.\n\nWe do find significant market-level differences in preferences around reading or watching, partly conditioned by consumption patterns offline. We find markets with historic patterns of high newspaper consumption such as Finland and Japan near the top of the list in terms of reading preference, with low-newspaper- circulation countries like Thailand and Brazil near the bottom. But other factors are also likely to be at play. Higher use of social media in general in Latin America (Mexico, Peru, Colombia, and Chile) as well as in parts of Asia-Pacific (Philippines and Taiwan) may be encouraging more video news use in these markets.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 37\n\nPROPORTION WHO MOSTLY READ NEWS IN TEXT – BY AGE – ALL MARKETS\n\n**OPTQ11D.** In thinking about your online news habits, which of the following statements applies best to you? _Base: Total sample; all markets = 93,432._\n\n67\n\n65\n\n60\n\n55\n\n14\n\n16\n\n18\n\n19\n\n11\n\n12\n\n14\n\n17\n\n8\n\n8\n\n8\n\n9\n\n55+\n\n35+\n\n25-34\n\n18-24\n\nMostly read Same Mostly watch Don’t know\n\n0% 25% 50% 75% 100%\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 38\n\nPROPORTION WHO MOSTLY READ NEWS IN TEXT – ALL MARKETS\n\n**OPTQ11D.** In thinking about your online news habits, which of the following statements applies best to you? Please select one. Showing net of different option statements. _Base: Total sample in each market (n ≈ 2000)._\n\nMostly read Same Mostly watch Don’t know\n\n0% 25% 50% 75% 100%\n\n6\n\n6\n\n12\n\n6\n\n7\n\n11\n\n17\n\n10\n\n11\n\n8\n\n12\n\n9\n\n8\n\n9\n\n6\n\n8\n\n10\n\n16\n\n7\n\n4\n\n8\n\n9\n\n15\n\n6\n\n11\n\n8\n\n21\n\n6\n\n9\n\n18\n\n14\n\n10\n\n8\n\n6\n\n6\n\n5\n\n6\n\n6\n\n5\n\n8\n\n8\n\n6\n\n6\n\n11\n\n10\n\n12\n\n8\n\n9\n\n11\n\n10\n\n8\n\n15\n\n10\n\n10\n\n11\n\n13\n\n12\n\n9\n\n12\n\n11\n\n13\n\n13\n\n18\n\n14\n\n10\n\n17\n\n10\n\n17\n\n10\n\n16\n\n22\n\n19\n\n16\n\n15\n\n18\n\n20\n\n22\n\n24\n\n26\n\n24\n\n7\n\n10\n\n12\n\n6\n\n15\n\n11\n\n15\n\n12\n\n6\n\n15\n\n19\n\n16\n\n20\n\n13\n\n11\n\n15\n\n13\n\n18\n\n16\n\n17\n\n16\n\n17\n\n17\n\n18\n\n11\n\n24\n\n20\n\n22\n\n13\n\n18\n\n16\n\n19\n\n21\n\n18\n\n14\n\n24\n\n22\n\n20\n\n22\n\n15\n\n17\n\n21\n\n24\n\n23\n\n23\n\n25\n\n85\n\n79\n\n78\n\n76\n\n74\n\n74\n\n74\n\n72\n\n71\n\n70\n\n69\n\n69\n\n68\n\n68\n\n67\n\n67\n\n67\n\n66\n\n66\n\n65\n\n64\n\n63\n\n63\n\n63\n\n61\n\n61\n\n61\n\n60\n\n60\n\n60\n\n58\n\n58\n\n58\n\n58\n\n56\n\n56\n\n55\n\n54\n\n53\n\n53\n\n51\n\n50\n\n47\n\n47\n\n47\n\n45\n\nDon’t\n\nMostly\n\nSame\n\nMostly\n\nMexico\n\nPhilippines\n\nPeru\n\nColombia\n\nBrazil\n\nCanada\n\nUSA\n\nChile\n\nTaiwan\n\nKenya\n\nThailand\n\nFrance\n\nSouth Korea\n\nAustria\n\nIndia\n\nGermany\n\nArgentina\n\nItaly\n\nHong Kong\n\nTurkey\n\nRomania\n\nAustralia\n\nSwitzerland\n\nSlovakia\n\nMalaysia\n\nSpain\n\nSingapore\n\nPortugal\n\nSouth Africa\n\nBelgium\n\nIreland\n\nJapan\n\nPoland\n\nCroatia\n\nIndonesia\n\nNigeria\n\nBulgaria\n\nUK\n\nNetherlands\n\nGreece\n\nSweden\n\nHungary\n\nDenmark\n\nCzech Republic\n\nNorway\n\nFinland 3\n\n4\n\n3\n\n4\n\n5\n\n1\n\n4\n\n4\n\n1\n\n4\n\n1\n\n5\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 27","pages":[{"line_numbers":[],"markdown":"11 https://eu.usatoday.com/story/tech/2022/01/21/climate-change-misinformation-facebook-youtube-twitter/6594691001/\n\nLevels of perceived misinformation around climate change and the environment are around three times higher in the United States (34%) than they are in Taiwan (10%) or Denmark (13%). Despite pledges to crack down, social media posts and videos denying climate change or disputing its causes remain widespread on Twitter, Facebook, YouTube, and TikTok, according to recent research. ¹¹\n\nIn this year’s report we find increased polarisation around climate change in a number of countries, including the United States, and identify the most important sources for climate change news with different groups.\n\n_For further analysis see Section 2.6: How People Access and Think about Climate Change News_\n\n**TEXT IS STILL KING BUT WHAT ROLE FOR ONLINE VIDEO?**\n\nSince the emergence of the internet, consumption on news websites (and apps) has primarily been about reading text articles. But that has begun to change with the growing supply of video news formats on social media.\n\nAgainst this background, it is surprising to find that all age groups, on average, say they still prefer to read news online rather than watch it – and we have seen little change in underlying preferences since we last asked the question in 2019. Younger audiences, however, are significantly more likely to say they watch the news, perhaps because they are more exposed to networks like Facebook, Instagram, and TikTok.\n\nWe do find significant market-level differences in preferences around reading or watching, partly conditioned by consumption patterns offline. We find markets with historic patterns of high newspaper consumption such as Finland and Japan near the top of the list in terms of reading preference, with low-newspaper- circulation countries like Thailand and Brazil near the bottom. But other factors are also likely to be at play. Higher use of social media in general in Latin America (Mexico, Peru, Colombia, and Chile) as well as in parts of Asia-Pacific (Philippines and Taiwan) may be encouraging more video news use in these markets.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 37\n\nPROPORTION WHO MOSTLY READ NEWS IN TEXT – BY AGE – ALL MARKETS\n\n**OPTQ11D.** In thinking about your online news habits, which of the following statements applies best to you? _Base: Total sample; all markets = 93,432._\n\n67\n\n65\n\n60\n\n55\n\n14\n\n16\n\n18\n\n19\n\n11\n\n12\n\n14\n\n17\n\n8\n\n8\n\n8\n\n9\n\n55+\n\n35+\n\n25-34\n\n18-24\n\nMostly read Same Mostly watch Don’t know\n\n0% 25% 50% 75% 100%\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 38\n\nPROPORTION WHO MOSTLY READ NEWS IN TEXT – ALL MARKETS\n\n**OPTQ11D.** In thinking about your online news habits, which of the following statements applies best to you? Please select one. Showing net of different option statements. _Base: Total sample in each market (n ≈ 2000)._\n\nMostly read Same Mostly watch Don’t know\n\n0% 25% 50% 75% 100%\n\n6\n\n6\n\n12\n\n6\n\n7\n\n11\n\n17\n\n10\n\n11\n\n8\n\n12\n\n9\n\n8\n\n9\n\n6\n\n8\n\n10\n\n16\n\n7\n\n4\n\n8\n\n9\n\n15\n\n6\n\n11\n\n8\n\n21\n\n6\n\n9\n\n18\n\n14\n\n10\n\n8\n\n6\n\n6\n\n5\n\n6\n\n6\n\n5\n\n8\n\n8\n\n6\n\n6\n\n11\n\n10\n\n12\n\n8\n\n9\n\n11\n\n10\n\n8\n\n15\n\n10\n\n10\n\n11\n\n13\n\n12\n\n9\n\n12\n\n11\n\n13\n\n13\n\n18\n\n14\n\n10\n\n17\n\n10\n\n17\n\n10\n\n16\n\n22\n\n19\n\n16\n\n15\n\n18\n\n20\n\n22\n\n24\n\n26\n\n24\n\n7\n\n10\n\n12\n\n6\n\n15\n\n11\n\n15\n\n12\n\n6\n\n15\n\n19\n\n16\n\n20\n\n13\n\n11\n\n15\n\n13\n\n18\n\n16\n\n17\n\n16\n\n17\n\n17\n\n18\n\n11\n\n24\n\n20\n\n22\n\n13\n\n18\n\n16\n\n19\n\n21\n\n18\n\n14\n\n24\n\n22\n\n20\n\n22\n\n15\n\n17\n\n21\n\n24\n\n23\n\n23\n\n25\n\n85\n\n79\n\n78\n\n76\n\n74\n\n74\n\n74\n\n72\n\n71\n\n70\n\n69\n\n69\n\n68\n\n68\n\n67\n\n67\n\n67\n\n66\n\n66\n\n65\n\n64\n\n63\n\n63\n\n63\n\n61\n\n61\n\n61\n\n60\n\n60\n\n60\n\n58\n\n58\n\n58\n\n58\n\n56\n\n56\n\n55\n\n54\n\n53\n\n53\n\n51\n\n50\n\n47\n\n47\n\n47\n\n45\n\nDon’t\n\nMostly\n\nSame\n\nMostly\n\nMexico\n\nPhilippines\n\nPeru\n\nColombia\n\nBrazil\n\nCanada\n\nUSA\n\nChile\n\nTaiwan\n\nKenya\n\nThailand\n\nFrance\n\nSouth Korea\n\nAustria\n\nIndia\n\nGermany\n\nArgentina\n\nItaly\n\nHong Kong\n\nTurkey\n\nRomania\n\nAustralia\n\nSwitzerland\n\nSlovakia\n\nMalaysia\n\nSpain\n\nSingapore\n\nPortugal\n\nSouth Africa\n\nBelgium\n\nIreland\n\nJapan\n\nPoland\n\nCroatia\n\nIndonesia\n\nNigeria\n\nBulgaria\n\nUK\n\nNetherlands\n\nGreece\n\nSweden\n\nHungary\n\nDenmark\n\nCzech Republic\n\nNorway\n\nFinland 3\n\n4\n\n3\n\n4\n\n5\n\n1\n\n4\n\n4\n\n1\n\n4\n\n1\n\n5\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 27","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"11 https://eu.usatoday.com/story/tech/2022/01/21/climate-change-misinformation-facebook-youtube-twitter/6594691001/\n\nLevels of perceived misinformation around climate change and the environment are around three times higher in the United States (34%) than they are in Taiwan (10%) or Denmark (13%). Despite pledges to crack down, social media posts and videos denying climate change or disputing its causes remain widespread on Twitter, Facebook, YouTube, and TikTok, according to recent research. ¹¹\n\nIn this year’s report we find increased polarisation around climate change in a number of countries, including the United States, and identify the most important sources for climate change news with different groups.\n\n_For further analysis see Section 2.6: How People Access and Think about Climate Change News_\n\n**TEXT IS STILL KING BUT WHAT ROLE FOR ONLINE VIDEO?**\n\nSince the emergence of the internet, consumption on news websites (and apps) has primarily been about reading text articles. But that has begun to change with the growing supply of video news formats on social media.\n\nAgainst this background, it is surprising to find that all age groups, on average, say they still prefer to read news online rather than watch it – and we have seen little change in underlying preferences since we last asked the question in 2019. Younger audiences, however, are significantly more likely to say they watch the news, perhaps because they are more exposed to networks like Facebook, Instagram, and TikTok.\n\nWe do find significant market-level differences in preferences around reading or watching, partly conditioned by consumption patterns offline. We find markets with historic patterns of high newspaper consumption such as Finland and Japan near the top of the list in terms of reading preference, with low-newspaper- circulation countries like Thailand and Brazil near the bottom. But other factors are also likely to be at play. Higher use of social media in general in Latin America (Mexico, Peru, Colombia, and Chile) as well as in parts of Asia-Pacific (Philippines and Taiwan) may be encouraging more video news use in these markets.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 37\n\nPROPORTION WHO MOSTLY READ NEWS IN TEXT – BY AGE – ALL MARKETS\n\n**OPTQ11D.** In thinking about your online news habits, which of the following statements applies best to you? _Base: Total sample; all markets = 93,432._\n\n67\n\n65\n\n60\n\n55\n\n14\n\n16\n\n18\n\n19\n\n11\n\n12\n\n14\n\n17\n\n8\n\n8\n\n8\n\n9\n\n55+\n\n35+\n\n25-34\n\n18-24\n\nMostly read Same Mostly watch Don’t know\n\n0% 25% 50% 75% 100%\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 38\n\nPROPORTION WHO MOSTLY READ NEWS IN TEXT – ALL MARKETS\n\n**OPTQ11D.** In thinking about your online news habits, which of the following statements applies best to you? Please select one. Showing net of different option statements. _Base: Total sample in each market (n ≈ 2000)._\n\nMostly read Same Mostly watch Don’t know\n\n0% 25% 50% 75% 100%\n\n6\n\n6\n\n12\n\n6\n\n7\n\n11\n\n17\n\n10\n\n11\n\n8\n\n12\n\n9\n\n8\n\n9\n\n6\n\n8\n\n10\n\n16\n\n7\n\n4\n\n8\n\n9\n\n15\n\n6\n\n11\n\n8\n\n21\n\n6\n\n9\n\n18\n\n14\n\n10\n\n8\n\n6\n\n6\n\n5\n\n6\n\n6\n\n5\n\n8\n\n8\n\n6\n\n6\n\n11\n\n10\n\n12\n\n8\n\n9\n\n11\n\n10\n\n8\n\n15\n\n10\n\n10\n\n11\n\n13\n\n12\n\n9\n\n12\n\n11\n\n13\n\n13\n\n18\n\n14\n\n10\n\n17\n\n10\n\n17\n\n10\n\n16\n\n22\n\n19\n\n16\n\n15\n\n18\n\n20\n\n22\n\n24\n\n26\n\n24\n\n7\n\n10\n\n12\n\n6\n\n15\n\n11\n\n15\n\n12\n\n6\n\n15\n\n19\n\n16\n\n20\n\n13\n\n11\n\n15\n\n13\n\n18\n\n16\n\n17\n\n16\n\n17\n\n17\n\n18\n\n11\n\n24\n\n20\n\n22\n\n13\n\n18\n\n16\n\n19\n\n21\n\n18\n\n14\n\n24\n\n22\n\n20\n\n22\n\n15\n\n17\n\n21\n\n24\n\n23\n\n23\n\n25\n\n85\n\n79\n\n78\n\n76\n\n74\n\n74\n\n74\n\n72\n\n71\n\n70\n\n69\n\n69\n\n68\n\n68\n\n67\n\n67\n\n67\n\n66\n\n66\n\n65\n\n64\n\n63\n\n63\n\n63\n\n61\n\n61\n\n61\n\n60\n\n60\n\n60\n\n58\n\n58\n\n58\n\n58\n\n56\n\n56\n\n55\n\n54\n\n53\n\n53\n\n51\n\n50\n\n47\n\n47\n\n47\n\n45\n\nDon’t\n\nMostly\n\nSame\n\nMostly\n\nMexico\n\nPhilippines\n\nPeru\n\nColombia\n\nBrazil\n\nCanada\n\nUSA\n\nChile\n\nTaiwan\n\nKenya\n\nThailand\n\nFrance\n\nSouth Korea\n\nAustria\n\nIndia\n\nGermany\n\nArgentina\n\nItaly\n\nHong Kong\n\nTurkey\n\nRomania\n\nAustralia\n\nSwitzerland\n\nSlovakia\n\nMalaysia\n\nSpain\n\nSingapore\n\nPortugal\n\nSouth Africa\n\nBelgium\n\nIreland\n\nJapan\n\nPoland\n\nCroatia\n\nIndonesia\n\nNigeria\n\nBulgaria\n\nUK\n\nNetherlands\n\nGreece\n\nSweden\n\nHungary\n\nDenmark\n\nCzech Republic\n\nNorway\n\nFinland 3\n\n4\n\n3\n\n4\n\n5\n\n1\n\n4\n\n4\n\n1\n\n4\n\n1\n\n5\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 27","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Digital_News-Report_2022_p27.pdf"]},"response":{"errors":[],"extract_id":"extract_ad5b8133731f778b8e5e571ea6591b40","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["11 https://eu.usatoday.com/story/tech/2022/01/21/climate-change-misinformation-facebook-youtube-twitter/6594691001/\n\nLevels of perceived misinformation around climate change and the environment are around three times higher in the United States (34%) than they are in Taiwan (10%) or Denmark (13%). Despite pledges to crack down, social media posts and videos denying climate change or disputing its causes remain widespread on Twitter, Facebook, YouTube, and TikTok, according to recent research. ¹¹\n\nIn this year’s report we find increased polarisation around climate change in a number of countries, including the United States, and identify the most important sources for climate change news with different groups.\n\n_For further analysis see Section 2.6: How People Access and Think about Climate Change News_\n\n**TEXT IS STILL KING BUT WHAT ROLE FOR ONLINE VIDEO?**\n\nSince the emergence of the internet, consumption on news websites (and apps) has primarily been about reading text articles. But that has begun to change with the growing supply of video news formats on social media.\n\nAgainst this background, it is surprising to find that all age groups, on average, say they still prefer to read news online rather than watch it – and we have seen little change in underlying preferences since we last asked the question in 2019. Younger audiences, however, are significantly more likely to say they watch the news, perhaps because they are more exposed to networks like Facebook, Instagram, and TikTok.\n\nWe do find significant market-level differences in preferences around reading or watching, partly conditioned by consumption patterns offline. We find markets with historic patterns of high newspaper consumption such as Finland and Japan near the top of the list in terms of reading preference, with low-newspaper- circulation countries like Thailand and Brazil near the bottom. But other factors are also likely to be at play. Higher use of social media in general in Latin America (Mexico, Peru, Colombia, and Chile) as well as in parts of Asia-Pacific (Philippines and Taiwan) may be encouraging more video news use in these markets.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 37\n\nPROPORTION WHO MOSTLY READ NEWS IN TEXT – BY AGE – ALL MARKETS\n\n**OPTQ11D.** In thinking about your online news habits, which of the following statements applies best to you? _Base: Total sample; all markets = 93,432._\n\n67\n\n65\n\n60\n\n55\n\n14\n\n16\n\n18\n\n19\n\n11\n\n12\n\n14\n\n17\n\n8\n\n8\n\n8\n\n9\n\n55+\n\n35+\n\n25-34\n\n18-24\n\nMostly read Same Mostly watch Don’t know\n\n0% 25% 50% 75% 100%\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 38\n\nPROPORTION WHO MOSTLY READ NEWS IN TEXT – ALL MARKETS\n\n**OPTQ11D.** In thinking about your online news habits, which of the following statements applies best to you? Please select one. Showing net of different option statements. _Base: Total sample in each market (n ≈ 2000)._\n\nMostly read Same Mostly watch Don’t know\n\n0% 25% 50% 75% 100%\n\n6\n\n6\n\n12\n\n6\n\n7\n\n11\n\n17\n\n10\n\n11\n\n8\n\n12\n\n9\n\n8\n\n9\n\n6\n\n8\n\n10\n\n16\n\n7\n\n4\n\n8\n\n9\n\n15\n\n6\n\n11\n\n8\n\n21\n\n6\n\n9\n\n18\n\n14\n\n10\n\n8\n\n6\n\n6\n\n5\n\n6\n\n6\n\n5\n\n8\n\n8\n\n6\n\n6\n\n11\n\n10\n\n12\n\n8\n\n9\n\n11\n\n10\n\n8\n\n15\n\n10\n\n10\n\n11\n\n13\n\n12\n\n9\n\n12\n\n11\n\n13\n\n13\n\n18\n\n14\n\n10\n\n17\n\n10\n\n17\n\n10\n\n16\n\n22\n\n19\n\n16\n\n15\n\n18\n\n20\n\n22\n\n24\n\n26\n\n24\n\n7\n\n10\n\n12\n\n6\n\n15\n\n11\n\n15\n\n12\n\n6\n\n15\n\n19\n\n16\n\n20\n\n13\n\n11\n\n15\n\n13\n\n18\n\n16\n\n17\n\n16\n\n17\n\n17\n\n18\n\n11\n\n24\n\n20\n\n22\n\n13\n\n18\n\n16\n\n19\n\n21\n\n18\n\n14\n\n24\n\n22\n\n20\n\n22\n\n15\n\n17\n\n21\n\n24\n\n23\n\n23\n\n25\n\n85\n\n79\n\n78\n\n76\n\n74\n\n74\n\n74\n\n72\n\n71\n\n70\n\n69\n\n69\n\n68\n\n68\n\n67\n\n67\n\n67\n\n66\n\n66\n\n65\n\n64\n\n63\n\n63\n\n63\n\n61\n\n61\n\n61\n\n60\n\n60\n\n60\n\n58\n\n58\n\n58\n\n58\n\n56\n\n56\n\n55\n\n54\n\n53\n\n53\n\n51\n\n50\n\n47\n\n47\n\n47\n\n45\n\nDon’t\n\nMostly\n\nSame\n\nMostly\n\nMexico\n\nPhilippines\n\nPeru\n\nColombia\n\nBrazil\n\nCanada\n\nUSA\n\nChile\n\nTaiwan\n\nKenya\n\nThailand\n\nFrance\n\nSouth Korea\n\nAustria\n\nIndia\n\nGermany\n\nArgentina\n\nItaly\n\nHong Kong\n\nTurkey\n\nRomania\n\nAustralia\n\nSwitzerland\n\nSlovakia\n\nMalaysia\n\nSpain\n\nSingapore\n\nPortugal\n\nSouth Africa\n\nBelgium\n\nIreland\n\nJapan\n\nPoland\n\nCroatia\n\nIndonesia\n\nNigeria\n\nBulgaria\n\nUK\n\nNetherlands\n\nGreece\n\nSweden\n\nHungary\n\nDenmark\n\nCzech Republic\n\nNorway\n\nFinland 3\n\n4\n\n3\n\n4\n\n5\n\n1\n\n4\n\n4\n\n1\n\n4\n\n1\n\n5\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 27"],"full_content":"11 https://eu.usatoday.com/story/tech/2022/01/21/climate-change-misinformation-facebook-youtube-twitter/6594691001/\n\nLevels of perceived misinformation around climate change and the environment are around three times higher in the United States (34%) than they are in Taiwan (10%) or Denmark (13%). Despite pledges to crack down, social media posts and videos denying climate change or disputing its causes remain widespread on Twitter, Facebook, YouTube, and TikTok, according to recent research. ¹¹\n\nIn this year’s report we find increased polarisation around climate change in a number of countries, including the United States, and identify the most important sources for climate change news with different groups.\n\n_For further analysis see Section 2.6: How People Access and Think about Climate Change News_\n\n**TEXT IS STILL KING BUT WHAT ROLE FOR ONLINE VIDEO?**\n\nSince the emergence of the internet, consumption on news websites (and apps) has primarily been about reading text articles. But that has begun to change with the growing supply of video news formats on social media.\n\nAgainst this background, it is surprising to find that all age groups, on average, say they still prefer to read news online rather than watch it – and we have seen little change in underlying preferences since we last asked the question in 2019. Younger audiences, however, are significantly more likely to say they watch the news, perhaps because they are more exposed to networks like Facebook, Instagram, and TikTok.\n\nWe do find significant market-level differences in preferences around reading or watching, partly conditioned by consumption patterns offline. We find markets with historic patterns of high newspaper consumption such as Finland and Japan near the top of the list in terms of reading preference, with low-newspaper- circulation countries like Thailand and Brazil near the bottom. But other factors are also likely to be at play. Higher use of social media in general in Latin America (Mexico, Peru, Colombia, and Chile) as well as in parts of Asia-Pacific (Philippines and Taiwan) may be encouraging more video news use in these markets.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 37\n\nPROPORTION WHO MOSTLY READ NEWS IN TEXT – BY AGE – ALL MARKETS\n\n**OPTQ11D.** In thinking about your online news habits, which of the following statements applies best to you? _Base: Total sample; all markets = 93,432._\n\n67\n\n65\n\n60\n\n55\n\n14\n\n16\n\n18\n\n19\n\n11\n\n12\n\n14\n\n17\n\n8\n\n8\n\n8\n\n9\n\n55+\n\n35+\n\n25-34\n\n18-24\n\nMostly read Same Mostly watch Don’t know\n\n0% 25% 50% 75% 100%\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / EXECUTIVE SUMMARY GRAPHS\n\n## Graph 38\n\nPROPORTION WHO MOSTLY READ NEWS IN TEXT – ALL MARKETS\n\n**OPTQ11D.** In thinking about your online news habits, which of the following statements applies best to you? Please select one. Showing net of different option statements. _Base: Total sample in each market (n ≈ 2000)._\n\nMostly read Same Mostly watch Don’t know\n\n0% 25% 50% 75% 100%\n\n6\n\n6\n\n12\n\n6\n\n7\n\n11\n\n17\n\n10\n\n11\n\n8\n\n12\n\n9\n\n8\n\n9\n\n6\n\n8\n\n10\n\n16\n\n7\n\n4\n\n8\n\n9\n\n15\n\n6\n\n11\n\n8\n\n21\n\n6\n\n9\n\n18\n\n14\n\n10\n\n8\n\n6\n\n6\n\n5\n\n6\n\n6\n\n5\n\n8\n\n8\n\n6\n\n6\n\n11\n\n10\n\n12\n\n8\n\n9\n\n11\n\n10\n\n8\n\n15\n\n10\n\n10\n\n11\n\n13\n\n12\n\n9\n\n12\n\n11\n\n13\n\n13\n\n18\n\n14\n\n10\n\n17\n\n10\n\n17\n\n10\n\n16\n\n22\n\n19\n\n16\n\n15\n\n18\n\n20\n\n22\n\n24\n\n26\n\n24\n\n7\n\n10\n\n12\n\n6\n\n15\n\n11\n\n15\n\n12\n\n6\n\n15\n\n19\n\n16\n\n20\n\n13\n\n11\n\n15\n\n13\n\n18\n\n16\n\n17\n\n16\n\n17\n\n17\n\n18\n\n11\n\n24\n\n20\n\n22\n\n13\n\n18\n\n16\n\n19\n\n21\n\n18\n\n14\n\n24\n\n22\n\n20\n\n22\n\n15\n\n17\n\n21\n\n24\n\n23\n\n23\n\n25\n\n85\n\n79\n\n78\n\n76\n\n74\n\n74\n\n74\n\n72\n\n71\n\n70\n\n69\n\n69\n\n68\n\n68\n\n67\n\n67\n\n67\n\n66\n\n66\n\n65\n\n64\n\n63\n\n63\n\n63\n\n61\n\n61\n\n61\n\n60\n\n60\n\n60\n\n58\n\n58\n\n58\n\n58\n\n56\n\n56\n\n55\n\n54\n\n53\n\n53\n\n51\n\n50\n\n47\n\n47\n\n47\n\n45\n\nDon’t\n\nMostly\n\nSame\n\nMostly\n\nMexico\n\nPhilippines\n\nPeru\n\nColombia\n\nBrazil\n\nCanada\n\nUSA\n\nChile\n\nTaiwan\n\nKenya\n\nThailand\n\nFrance\n\nSouth Korea\n\nAustria\n\nIndia\n\nGermany\n\nArgentina\n\nItaly\n\nHong Kong\n\nTurkey\n\nRomania\n\nAustralia\n\nSwitzerland\n\nSlovakia\n\nMalaysia\n\nSpain\n\nSingapore\n\nPortugal\n\nSouth Africa\n\nBelgium\n\nIreland\n\nJapan\n\nPoland\n\nCroatia\n\nIndonesia\n\nNigeria\n\nBulgaria\n\nUK\n\nNetherlands\n\nGreece\n\nSweden\n\nHungary\n\nDenmark\n\nCzech Republic\n\nNorway\n\nFinland 3\n\n4\n\n3\n\n4\n\n5\n\n1\n\n4\n\n4\n\n1\n\n4\n\n1\n\n5\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 27","publish_date":null,"title":"Graph 38 Graph 37","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Digital_News-Report_2022_p27.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Digital_News-Report_2022_p27.pdf"},"request":{"example_id":"chart/Digital_News-Report_2022_p27","product_type":"parse","source_file_path":"Digital_News-Report_2022_p27.pdf"},"started_at":"2026-11-04T19:46:18.970171"} +{"completed_at":"2026-11-04T19:46:23.652255","latency_in_ms":2614,"output":{"example_id":"chart/Digital_News-Report_2022_p34","grounded_pages":[],"layout_pages":[],"markdown":"# 2\\.1 Perceptions of Media Coverage of the War\n\n# in Ukraine\n\n### **Kirsten Eddy** and **Richard Fletcher**\n\nOn 24 February, Russia launched a full-scale invasion of Ukraine, marking the start of the biggest war in Europe since the Second World War. Major news organisations around the world have embedded journalists in Ukraine to cover bombings and violence in hard-hit towns and cities across broadcast, digital, and print media. Journalists, civilians, and politicians – most notably Ukrainian President Volodymyr Zelensky – have also taken to social networks like TikTok, Telegram, and Twitter to document the horrors of the war for a global audience in real time. The humanitarian crisis unfolding in Ukraine, along with the scale of the Western response to Russia’s invasion, has far-reaching political and economic effects.\n\nTo better understand how people have been accessing news about the Russia–Ukraine conflict, and to explore the impact this might have on overall trends in attitudes and behaviours, we commissioned YouGov to carry out a separate follow-up survey in five countries: Brazil, Germany, Poland, the UK, and the US. These countries were selected because they represent different levels of proximity to the conflict, ranging from Poland, which borders Ukraine, to Brazil and the US, which are on different continents. Fieldwork took place from 29 March to 7 April 2022 – roughly one month after the invasion began. We polled around 1,000 respondents in each country and samples were assembled using the same quotas as the main survey. The questionnaire contained several new questions specifically about the Russia–Ukraine conflict, as well a handful of key questions from the main Digital News Report survey on news interest, use, avoidance, and trust. This allows us to directly compare our ‘pre-conflict’ results from the main 2022 survey and the ‘during conflict’ follow-up survey. However, because the Russia-Ukraine conflict overlapped with other domestic news stories, not all changes we see will be a direct result of coverage of the conflict.\n\n**MOST PEOPLE ARE FOLLOWING THE CONFLICT CLOSELY**\n\nFirst, how much attention are people paying to news about the conflict? In all five countries, a majority are following the Russia– Ukraine conflict at least somewhat closely. Although the conflict has global consequences, attention is highest in Germany, which is both geographically close to the conflict and where the effects are already filtering down to the lives of ordinary people, for example, in terms of energy prices. In Brazil, which is politically and geographically farther from the conflict, around 40% are not following it closely.\n\nSome of these findings are in line with earlier data on those who are not consuming any news sources at all in markets like the US and UK. But in countries like Brazil and Germany, the proportion of 18–24s not following the conflict at all is particularly high compared with other age groups. And in the US and UK, women are slightly more likely than men to not be paying close attention to the war.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 1\n\nPROPORTION WHO ARE FOLLOWING THE RUSSIA–UKRAINE CONFLICT CLOSELY – SELECTED MARKETS\n\nGermany\n\nPoland\n\nUK\n\nUSA\n\nBrazil\n\n**Q1\\_ukraine.** How closely, if at all, are you following the current Russia–Ukraine conflict? _Base: Total sample: Brazil = 1002, Germany = 1015, Poland = 1003, UK = 1191, USA = 1356._\n\n0% 25% 50% 75% 100%\n\n18\n\n7\n\n7\n\n7\n\n31\n\n18\n\n19\n\n15\n\n20\n\n32\n\n47\n\n44\n\n37\n\n34\n\n11\n\n25\n\n20\n\n29\n\n20\n\n5\n\n6\n\n10\n\n7\n\n2\n\n15\n\nExtremely closely\n\nNot very closely\n\nVery closely\n\nNot at all\n\nSomewhat closely\n\nDon’t know\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 34","pages":[{"line_numbers":[],"markdown":"# 2\\.1 Perceptions of Media Coverage of the War\n\n# in Ukraine\n\n### **Kirsten Eddy** and **Richard Fletcher**\n\nOn 24 February, Russia launched a full-scale invasion of Ukraine, marking the start of the biggest war in Europe since the Second World War. Major news organisations around the world have embedded journalists in Ukraine to cover bombings and violence in hard-hit towns and cities across broadcast, digital, and print media. Journalists, civilians, and politicians – most notably Ukrainian President Volodymyr Zelensky – have also taken to social networks like TikTok, Telegram, and Twitter to document the horrors of the war for a global audience in real time. The humanitarian crisis unfolding in Ukraine, along with the scale of the Western response to Russia’s invasion, has far-reaching political and economic effects.\n\nTo better understand how people have been accessing news about the Russia–Ukraine conflict, and to explore the impact this might have on overall trends in attitudes and behaviours, we commissioned YouGov to carry out a separate follow-up survey in five countries: Brazil, Germany, Poland, the UK, and the US. These countries were selected because they represent different levels of proximity to the conflict, ranging from Poland, which borders Ukraine, to Brazil and the US, which are on different continents. Fieldwork took place from 29 March to 7 April 2022 – roughly one month after the invasion began. We polled around 1,000 respondents in each country and samples were assembled using the same quotas as the main survey. The questionnaire contained several new questions specifically about the Russia–Ukraine conflict, as well a handful of key questions from the main Digital News Report survey on news interest, use, avoidance, and trust. This allows us to directly compare our ‘pre-conflict’ results from the main 2022 survey and the ‘during conflict’ follow-up survey. However, because the Russia-Ukraine conflict overlapped with other domestic news stories, not all changes we see will be a direct result of coverage of the conflict.\n\n**MOST PEOPLE ARE FOLLOWING THE CONFLICT CLOSELY**\n\nFirst, how much attention are people paying to news about the conflict? In all five countries, a majority are following the Russia– Ukraine conflict at least somewhat closely. Although the conflict has global consequences, attention is highest in Germany, which is both geographically close to the conflict and where the effects are already filtering down to the lives of ordinary people, for example, in terms of energy prices. In Brazil, which is politically and geographically farther from the conflict, around 40% are not following it closely.\n\nSome of these findings are in line with earlier data on those who are not consuming any news sources at all in markets like the US and UK. But in countries like Brazil and Germany, the proportion of 18–24s not following the conflict at all is particularly high compared with other age groups. And in the US and UK, women are slightly more likely than men to not be paying close attention to the war.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 1\n\nPROPORTION WHO ARE FOLLOWING THE RUSSIA–UKRAINE CONFLICT CLOSELY – SELECTED MARKETS\n\nGermany\n\nPoland\n\nUK\n\nUSA\n\nBrazil\n\n**Q1\\_ukraine.** How closely, if at all, are you following the current Russia–Ukraine conflict? _Base: Total sample: Brazil = 1002, Germany = 1015, Poland = 1003, UK = 1191, USA = 1356._\n\n0% 25% 50% 75% 100%\n\n18\n\n7\n\n7\n\n7\n\n31\n\n18\n\n19\n\n15\n\n20\n\n32\n\n47\n\n44\n\n37\n\n34\n\n11\n\n25\n\n20\n\n29\n\n20\n\n5\n\n6\n\n10\n\n7\n\n2\n\n15\n\nExtremely closely\n\nNot very closely\n\nVery closely\n\nNot at all\n\nSomewhat closely\n\nDon’t know\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 34","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"# 2\\.1 Perceptions of Media Coverage of the War\n\n# in Ukraine\n\n### **Kirsten Eddy** and **Richard Fletcher**\n\nOn 24 February, Russia launched a full-scale invasion of Ukraine, marking the start of the biggest war in Europe since the Second World War. Major news organisations around the world have embedded journalists in Ukraine to cover bombings and violence in hard-hit towns and cities across broadcast, digital, and print media. Journalists, civilians, and politicians – most notably Ukrainian President Volodymyr Zelensky – have also taken to social networks like TikTok, Telegram, and Twitter to document the horrors of the war for a global audience in real time. The humanitarian crisis unfolding in Ukraine, along with the scale of the Western response to Russia’s invasion, has far-reaching political and economic effects.\n\nTo better understand how people have been accessing news about the Russia–Ukraine conflict, and to explore the impact this might have on overall trends in attitudes and behaviours, we commissioned YouGov to carry out a separate follow-up survey in five countries: Brazil, Germany, Poland, the UK, and the US. These countries were selected because they represent different levels of proximity to the conflict, ranging from Poland, which borders Ukraine, to Brazil and the US, which are on different continents. Fieldwork took place from 29 March to 7 April 2022 – roughly one month after the invasion began. We polled around 1,000 respondents in each country and samples were assembled using the same quotas as the main survey. The questionnaire contained several new questions specifically about the Russia–Ukraine conflict, as well a handful of key questions from the main Digital News Report survey on news interest, use, avoidance, and trust. This allows us to directly compare our ‘pre-conflict’ results from the main 2022 survey and the ‘during conflict’ follow-up survey. However, because the Russia-Ukraine conflict overlapped with other domestic news stories, not all changes we see will be a direct result of coverage of the conflict.\n\n**MOST PEOPLE ARE FOLLOWING THE CONFLICT CLOSELY**\n\nFirst, how much attention are people paying to news about the conflict? In all five countries, a majority are following the Russia– Ukraine conflict at least somewhat closely. Although the conflict has global consequences, attention is highest in Germany, which is both geographically close to the conflict and where the effects are already filtering down to the lives of ordinary people, for example, in terms of energy prices. In Brazil, which is politically and geographically farther from the conflict, around 40% are not following it closely.\n\nSome of these findings are in line with earlier data on those who are not consuming any news sources at all in markets like the US and UK. But in countries like Brazil and Germany, the proportion of 18–24s not following the conflict at all is particularly high compared with other age groups. And in the US and UK, women are slightly more likely than men to not be paying close attention to the war.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 1\n\nPROPORTION WHO ARE FOLLOWING THE RUSSIA–UKRAINE CONFLICT CLOSELY – SELECTED MARKETS\n\nGermany\n\nPoland\n\nUK\n\nUSA\n\nBrazil\n\n**Q1\\_ukraine.** How closely, if at all, are you following the current Russia–Ukraine conflict? _Base: Total sample: Brazil = 1002, Germany = 1015, Poland = 1003, UK = 1191, USA = 1356._\n\n0% 25% 50% 75% 100%\n\n18\n\n7\n\n7\n\n7\n\n31\n\n18\n\n19\n\n15\n\n20\n\n32\n\n47\n\n44\n\n37\n\n34\n\n11\n\n25\n\n20\n\n29\n\n20\n\n5\n\n6\n\n10\n\n7\n\n2\n\n15\n\nExtremely closely\n\nNot very closely\n\nVery closely\n\nNot at all\n\nSomewhat closely\n\nDon’t know\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 34","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Digital_News-Report_2022_p34.pdf"]},"response":{"errors":[],"extract_id":"extract_e5357c96f4f1f586e84da15393166450","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["# 2\\.1 Perceptions of Media Coverage of the War\n\n# in Ukraine\n\n### **Kirsten Eddy** and **Richard Fletcher**\n\nOn 24 February, Russia launched a full-scale invasion of Ukraine, marking the start of the biggest war in Europe since the Second World War. Major news organisations around the world have embedded journalists in Ukraine to cover bombings and violence in hard-hit towns and cities across broadcast, digital, and print media. Journalists, civilians, and politicians – most notably Ukrainian President Volodymyr Zelensky – have also taken to social networks like TikTok, Telegram, and Twitter to document the horrors of the war for a global audience in real time. The humanitarian crisis unfolding in Ukraine, along with the scale of the Western response to Russia’s invasion, has far-reaching political and economic effects.\n\nTo better understand how people have been accessing news about the Russia–Ukraine conflict, and to explore the impact this might have on overall trends in attitudes and behaviours, we commissioned YouGov to carry out a separate follow-up survey in five countries: Brazil, Germany, Poland, the UK, and the US. These countries were selected because they represent different levels of proximity to the conflict, ranging from Poland, which borders Ukraine, to Brazil and the US, which are on different continents. Fieldwork took place from 29 March to 7 April 2022 – roughly one month after the invasion began. We polled around 1,000 respondents in each country and samples were assembled using the same quotas as the main survey. The questionnaire contained several new questions specifically about the Russia–Ukraine conflict, as well a handful of key questions from the main Digital News Report survey on news interest, use, avoidance, and trust. This allows us to directly compare our ‘pre-conflict’ results from the main 2022 survey and the ‘during conflict’ follow-up survey. However, because the Russia-Ukraine conflict overlapped with other domestic news stories, not all changes we see will be a direct result of coverage of the conflict.\n\n**MOST PEOPLE ARE FOLLOWING THE CONFLICT CLOSELY**\n\nFirst, how much attention are people paying to news about the conflict? In all five countries, a majority are following the Russia– Ukraine conflict at least somewhat closely. Although the conflict has global consequences, attention is highest in Germany, which is both geographically close to the conflict and where the effects are already filtering down to the lives of ordinary people, for example, in terms of energy prices. In Brazil, which is politically and geographically farther from the conflict, around 40% are not following it closely.\n\nSome of these findings are in line with earlier data on those who are not consuming any news sources at all in markets like the US and UK. But in countries like Brazil and Germany, the proportion of 18–24s not following the conflict at all is particularly high compared with other age groups. And in the US and UK, women are slightly more likely than men to not be paying close attention to the war.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 1\n\nPROPORTION WHO ARE FOLLOWING THE RUSSIA–UKRAINE CONFLICT CLOSELY – SELECTED MARKETS\n\nGermany\n\nPoland\n\nUK\n\nUSA\n\nBrazil\n\n**Q1\\_ukraine.** How closely, if at all, are you following the current Russia–Ukraine conflict? _Base: Total sample: Brazil = 1002, Germany = 1015, Poland = 1003, UK = 1191, USA = 1356._\n\n0% 25% 50% 75% 100%\n\n18\n\n7\n\n7\n\n7\n\n31\n\n18\n\n19\n\n15\n\n20\n\n32\n\n47\n\n44\n\n37\n\n34\n\n11\n\n25\n\n20\n\n29\n\n20\n\n5\n\n6\n\n10\n\n7\n\n2\n\n15\n\nExtremely closely\n\nNot very closely\n\nVery closely\n\nNot at all\n\nSomewhat closely\n\nDon’t know\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 34"],"full_content":"# 2\\.1 Perceptions of Media Coverage of the War\n\n# in Ukraine\n\n### **Kirsten Eddy** and **Richard Fletcher**\n\nOn 24 February, Russia launched a full-scale invasion of Ukraine, marking the start of the biggest war in Europe since the Second World War. Major news organisations around the world have embedded journalists in Ukraine to cover bombings and violence in hard-hit towns and cities across broadcast, digital, and print media. Journalists, civilians, and politicians – most notably Ukrainian President Volodymyr Zelensky – have also taken to social networks like TikTok, Telegram, and Twitter to document the horrors of the war for a global audience in real time. The humanitarian crisis unfolding in Ukraine, along with the scale of the Western response to Russia’s invasion, has far-reaching political and economic effects.\n\nTo better understand how people have been accessing news about the Russia–Ukraine conflict, and to explore the impact this might have on overall trends in attitudes and behaviours, we commissioned YouGov to carry out a separate follow-up survey in five countries: Brazil, Germany, Poland, the UK, and the US. These countries were selected because they represent different levels of proximity to the conflict, ranging from Poland, which borders Ukraine, to Brazil and the US, which are on different continents. Fieldwork took place from 29 March to 7 April 2022 – roughly one month after the invasion began. We polled around 1,000 respondents in each country and samples were assembled using the same quotas as the main survey. The questionnaire contained several new questions specifically about the Russia–Ukraine conflict, as well a handful of key questions from the main Digital News Report survey on news interest, use, avoidance, and trust. This allows us to directly compare our ‘pre-conflict’ results from the main 2022 survey and the ‘during conflict’ follow-up survey. However, because the Russia-Ukraine conflict overlapped with other domestic news stories, not all changes we see will be a direct result of coverage of the conflict.\n\n**MOST PEOPLE ARE FOLLOWING THE CONFLICT CLOSELY**\n\nFirst, how much attention are people paying to news about the conflict? In all five countries, a majority are following the Russia– Ukraine conflict at least somewhat closely. Although the conflict has global consequences, attention is highest in Germany, which is both geographically close to the conflict and where the effects are already filtering down to the lives of ordinary people, for example, in terms of energy prices. In Brazil, which is politically and geographically farther from the conflict, around 40% are not following it closely.\n\nSome of these findings are in line with earlier data on those who are not consuming any news sources at all in markets like the US and UK. But in countries like Brazil and Germany, the proportion of 18–24s not following the conflict at all is particularly high compared with other age groups. And in the US and UK, women are slightly more likely than men to not be paying close attention to the war.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 1\n\nPROPORTION WHO ARE FOLLOWING THE RUSSIA–UKRAINE CONFLICT CLOSELY – SELECTED MARKETS\n\nGermany\n\nPoland\n\nUK\n\nUSA\n\nBrazil\n\n**Q1\\_ukraine.** How closely, if at all, are you following the current Russia–Ukraine conflict? _Base: Total sample: Brazil = 1002, Germany = 1015, Poland = 1003, UK = 1191, USA = 1356._\n\n0% 25% 50% 75% 100%\n\n18\n\n7\n\n7\n\n7\n\n31\n\n18\n\n19\n\n15\n\n20\n\n32\n\n47\n\n44\n\n37\n\n34\n\n11\n\n25\n\n20\n\n29\n\n20\n\n5\n\n6\n\n10\n\n7\n\n2\n\n15\n\nExtremely closely\n\nNot very closely\n\nVery closely\n\nNot at all\n\nSomewhat closely\n\nDon’t know\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 34","publish_date":null,"title":"2.1 Perceptions of Media Coverage of the War in Ukraine","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Digital_News-Report_2022_p34.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Digital_News-Report_2022_p34.pdf"},"request":{"example_id":"chart/Digital_News-Report_2022_p34","product_type":"parse","source_file_path":"Digital_News-Report_2022_p34.pdf"},"started_at":"2026-11-04T19:46:21.038225"} +{"completed_at":"2026-11-04T19:46:24.460706","latency_in_ms":1665,"output":{"example_id":"chart/Digital_News-Report_2022_p35","grounded_pages":[],"layout_pages":[],"markdown":"REUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 3\n\nPROPORTION WHO SOMETIMES OR OFTEN ACTIVELY AVOID THE NEWS – SELECTED MARKETS\n\n**Q1di 2017.** Do you find yourself actively trying to avoid news these days? _Base: Total pre/during-conflict samples: Brazil = 2022/1002, Germany = 2002/1015, Poland = 2009/1003, UK = 2410/1191, USA = 2036/1356. Note: Pre-conflict sample from the main 2022 DNR survey. During-conflict sample from 29 March to 7 April._\n\n0% 25% 50% 75%\n\nUK 46 | 46\n\nGermany 29\n\nPre-conflict During conflict\n\n36\n\nPoland 41 47\n\nUSA 42 46\n\nBrazil 54 55\n\nAs we have seen with other major world events, people turn more to TV news at times of crisis. When asking respondents which news source they are paying the most attention to when it comes to the Russia–Ukraine conflict, television tops the list for three of the five countries – with the _most_ attention to TV news on the conflict spread out geographically, from Germany (46%) and Brazil (44%). In the US and Poland, online news sites, non-mainstream sites, and social media combined account for a larger share – but TV is still the most widely used individual source for news about the conflict.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 2\n\nPROPORTION WHO PAY MOST ATTENTION TO EACH SOURCE FOR NEWS ABOUT THE RUSSIA–UKRAINE CONFLICT – SELECTED MARKETS\n\n**Q2\\_ukraine.** When it comes to the current Russia–Ukraine conflict, which ONE of the following sources of news, if any, are you paying most attention to? _Base: Total sample: Brazil = 1002, Germany = 1015, Poland = 1003, UK = 1191, USA = 1356._\n\n0% 25% 50%\n\nBrazil Television\n\nUSA UK Germany Poland\n\n44\n\n31\n\n39\n\n46\n\n38\n\nBrazil Online news sites from major news organisations USA UK Germany Poland\n\n13\n\n18\n\n23\n\n14\n\n20\n\nBrazil **Online** (news sites + non-mainstream sites + social)\n\nUSA UK Germany Poland\n\n41\n\n40\n\n33\n\n24\n\n40\n\nBrazil Social media sites USA UK Germany Poland\n\n23\n\n13\n\n7 6\n\n11\n\nBrazil Radio USA UK Germany Poland\n\n5 6\n\n9 9\n\n2\n\nBrazil Non-mainstream sites (e.g. specialist websites, or alternative sites)\n\nUSA UK Germany Poland\n\n5\n\n9\n\n4\n\n9\n\n3\n\nBrazil Conversations with colleagues, friends, or family USA UK Germany Poland\n\n5 5\n\n5 5\n\n3\n\nBrazil Printed newspapers USA UK Germany Poland\n\n4 4\n\n1\n\n1\n\n3\n\nWe see some differences in preferred sources by both age and market. For instance, the same splits we see in TV versus social media use between younger and older cohorts also applies to the most important platform for following the Russia–Ukraine conflict. In the UK, for instance, more than half (55%) of those 55 or older, but only 13% of 18–24s, are paying the most attention to TV news on the Russia–Ukraine conflict. And nearly eight times as many 18–24s (15%) are paying the most attention to social media news on the war as those 55 and older (2%). Further, while TV is especially dominant for information about the conflict in countries like Germany, social media are a particularly important source in Brazil, with 23% of people paying the most attention to these platforms for related information.\n\n**HOW HAS THE CONFLICT INFLUENCED PEOPLE’S NEWS BEHAVIOURS AND ATTITUDES?**\n\nWe also used this follow-up survey to explore whether the trends we described in the Executive Summary – namely, decreases in news use and interest and increases in news avoidance – were reversed by the extensive coverage of the Russia–Ukraine conflict. We did this by repeating the same key questions (albeit to a different sample of respondents) and looking for differences between the ‘pre-conflict’ sample from the main 2022 survey and the ‘during conflict’ follow-up survey. Of course, we need to be cautious about attributing all changes to coverage of the conflict. Major domestic stories – such as the ‘Partygate’ scandal in the UK – were unfolding in parallel, and these may also have influenced people’s attitudes and behaviours around news.\n\nIn general, we find little evidence that these long-term trends have been reversed – even temporarily – and in some cases there’s possible evidence that they have been accelerated. The clearest example of this is news avoidance. In Germany, Poland, and the US, the proportion who say they sometimes or often actively avoid the news has increased. The biggest increase of all was in Germany (+7pp), but significant increases can also be seen in Poland (+6pp) and the US (+4pp). To put some of these changes in context, the increase of 7pp in Germany in just two months is larger than the 5pp increase we saw in the _five years_ from 2017 to 2022. We know that one of the main reasons people avoid the news is because of the negative effect it has on their mood, so it would be unsurprising if the deeply depressing and concerning nature of conflict has caused more people to turn away from it. In the UK and Brazil, where news avoidance was already high, we do not see evidence of a further increase – but it is equally important to note that it has not decreased either. Furthermore, in these two countries, news avoidance has already increased markedly in recent years – by 11pp from 2019 to 2022 in the case of the UK, and by 20pp in Brazil.\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 35","pages":[{"line_numbers":[],"markdown":"REUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 3\n\nPROPORTION WHO SOMETIMES OR OFTEN ACTIVELY AVOID THE NEWS – SELECTED MARKETS\n\n**Q1di 2017.** Do you find yourself actively trying to avoid news these days? _Base: Total pre/during-conflict samples: Brazil = 2022/1002, Germany = 2002/1015, Poland = 2009/1003, UK = 2410/1191, USA = 2036/1356. Note: Pre-conflict sample from the main 2022 DNR survey. During-conflict sample from 29 March to 7 April._\n\n0% 25% 50% 75%\n\nUK 46 | 46\n\nGermany 29\n\nPre-conflict During conflict\n\n36\n\nPoland 41 47\n\nUSA 42 46\n\nBrazil 54 55\n\nAs we have seen with other major world events, people turn more to TV news at times of crisis. When asking respondents which news source they are paying the most attention to when it comes to the Russia–Ukraine conflict, television tops the list for three of the five countries – with the _most_ attention to TV news on the conflict spread out geographically, from Germany (46%) and Brazil (44%). In the US and Poland, online news sites, non-mainstream sites, and social media combined account for a larger share – but TV is still the most widely used individual source for news about the conflict.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 2\n\nPROPORTION WHO PAY MOST ATTENTION TO EACH SOURCE FOR NEWS ABOUT THE RUSSIA–UKRAINE CONFLICT – SELECTED MARKETS\n\n**Q2\\_ukraine.** When it comes to the current Russia–Ukraine conflict, which ONE of the following sources of news, if any, are you paying most attention to? _Base: Total sample: Brazil = 1002, Germany = 1015, Poland = 1003, UK = 1191, USA = 1356._\n\n0% 25% 50%\n\nBrazil Television\n\nUSA UK Germany Poland\n\n44\n\n31\n\n39\n\n46\n\n38\n\nBrazil Online news sites from major news organisations USA UK Germany Poland\n\n13\n\n18\n\n23\n\n14\n\n20\n\nBrazil **Online** (news sites + non-mainstream sites + social)\n\nUSA UK Germany Poland\n\n41\n\n40\n\n33\n\n24\n\n40\n\nBrazil Social media sites USA UK Germany Poland\n\n23\n\n13\n\n7 6\n\n11\n\nBrazil Radio USA UK Germany Poland\n\n5 6\n\n9 9\n\n2\n\nBrazil Non-mainstream sites (e.g. specialist websites, or alternative sites)\n\nUSA UK Germany Poland\n\n5\n\n9\n\n4\n\n9\n\n3\n\nBrazil Conversations with colleagues, friends, or family USA UK Germany Poland\n\n5 5\n\n5 5\n\n3\n\nBrazil Printed newspapers USA UK Germany Poland\n\n4 4\n\n1\n\n1\n\n3\n\nWe see some differences in preferred sources by both age and market. For instance, the same splits we see in TV versus social media use between younger and older cohorts also applies to the most important platform for following the Russia–Ukraine conflict. In the UK, for instance, more than half (55%) of those 55 or older, but only 13% of 18–24s, are paying the most attention to TV news on the Russia–Ukraine conflict. And nearly eight times as many 18–24s (15%) are paying the most attention to social media news on the war as those 55 and older (2%). Further, while TV is especially dominant for information about the conflict in countries like Germany, social media are a particularly important source in Brazil, with 23% of people paying the most attention to these platforms for related information.\n\n**HOW HAS THE CONFLICT INFLUENCED PEOPLE’S NEWS BEHAVIOURS AND ATTITUDES?**\n\nWe also used this follow-up survey to explore whether the trends we described in the Executive Summary – namely, decreases in news use and interest and increases in news avoidance – were reversed by the extensive coverage of the Russia–Ukraine conflict. We did this by repeating the same key questions (albeit to a different sample of respondents) and looking for differences between the ‘pre-conflict’ sample from the main 2022 survey and the ‘during conflict’ follow-up survey. Of course, we need to be cautious about attributing all changes to coverage of the conflict. Major domestic stories – such as the ‘Partygate’ scandal in the UK – were unfolding in parallel, and these may also have influenced people’s attitudes and behaviours around news.\n\nIn general, we find little evidence that these long-term trends have been reversed – even temporarily – and in some cases there’s possible evidence that they have been accelerated. The clearest example of this is news avoidance. In Germany, Poland, and the US, the proportion who say they sometimes or often actively avoid the news has increased. The biggest increase of all was in Germany (+7pp), but significant increases can also be seen in Poland (+6pp) and the US (+4pp). To put some of these changes in context, the increase of 7pp in Germany in just two months is larger than the 5pp increase we saw in the _five years_ from 2017 to 2022. We know that one of the main reasons people avoid the news is because of the negative effect it has on their mood, so it would be unsurprising if the deeply depressing and concerning nature of conflict has caused more people to turn away from it. In the UK and Brazil, where news avoidance was already high, we do not see evidence of a further increase – but it is equally important to note that it has not decreased either. Furthermore, in these two countries, news avoidance has already increased markedly in recent years – by 11pp from 2019 to 2022 in the case of the UK, and by 20pp in Brazil.\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 35","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"REUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 3\n\nPROPORTION WHO SOMETIMES OR OFTEN ACTIVELY AVOID THE NEWS – SELECTED MARKETS\n\n**Q1di 2017.** Do you find yourself actively trying to avoid news these days? _Base: Total pre/during-conflict samples: Brazil = 2022/1002, Germany = 2002/1015, Poland = 2009/1003, UK = 2410/1191, USA = 2036/1356. Note: Pre-conflict sample from the main 2022 DNR survey. During-conflict sample from 29 March to 7 April._\n\n0% 25% 50% 75%\n\nUK 46 | 46\n\nGermany 29\n\nPre-conflict During conflict\n\n36\n\nPoland 41 47\n\nUSA 42 46\n\nBrazil 54 55\n\nAs we have seen with other major world events, people turn more to TV news at times of crisis. When asking respondents which news source they are paying the most attention to when it comes to the Russia–Ukraine conflict, television tops the list for three of the five countries – with the _most_ attention to TV news on the conflict spread out geographically, from Germany (46%) and Brazil (44%). In the US and Poland, online news sites, non-mainstream sites, and social media combined account for a larger share – but TV is still the most widely used individual source for news about the conflict.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 2\n\nPROPORTION WHO PAY MOST ATTENTION TO EACH SOURCE FOR NEWS ABOUT THE RUSSIA–UKRAINE CONFLICT – SELECTED MARKETS\n\n**Q2\\_ukraine.** When it comes to the current Russia–Ukraine conflict, which ONE of the following sources of news, if any, are you paying most attention to? _Base: Total sample: Brazil = 1002, Germany = 1015, Poland = 1003, UK = 1191, USA = 1356._\n\n0% 25% 50%\n\nBrazil Television\n\nUSA UK Germany Poland\n\n44\n\n31\n\n39\n\n46\n\n38\n\nBrazil Online news sites from major news organisations USA UK Germany Poland\n\n13\n\n18\n\n23\n\n14\n\n20\n\nBrazil **Online** (news sites + non-mainstream sites + social)\n\nUSA UK Germany Poland\n\n41\n\n40\n\n33\n\n24\n\n40\n\nBrazil Social media sites USA UK Germany Poland\n\n23\n\n13\n\n7 6\n\n11\n\nBrazil Radio USA UK Germany Poland\n\n5 6\n\n9 9\n\n2\n\nBrazil Non-mainstream sites (e.g. specialist websites, or alternative sites)\n\nUSA UK Germany Poland\n\n5\n\n9\n\n4\n\n9\n\n3\n\nBrazil Conversations with colleagues, friends, or family USA UK Germany Poland\n\n5 5\n\n5 5\n\n3\n\nBrazil Printed newspapers USA UK Germany Poland\n\n4 4\n\n1\n\n1\n\n3\n\nWe see some differences in preferred sources by both age and market. For instance, the same splits we see in TV versus social media use between younger and older cohorts also applies to the most important platform for following the Russia–Ukraine conflict. In the UK, for instance, more than half (55%) of those 55 or older, but only 13% of 18–24s, are paying the most attention to TV news on the Russia–Ukraine conflict. And nearly eight times as many 18–24s (15%) are paying the most attention to social media news on the war as those 55 and older (2%). Further, while TV is especially dominant for information about the conflict in countries like Germany, social media are a particularly important source in Brazil, with 23% of people paying the most attention to these platforms for related information.\n\n**HOW HAS THE CONFLICT INFLUENCED PEOPLE’S NEWS BEHAVIOURS AND ATTITUDES?**\n\nWe also used this follow-up survey to explore whether the trends we described in the Executive Summary – namely, decreases in news use and interest and increases in news avoidance – were reversed by the extensive coverage of the Russia–Ukraine conflict. We did this by repeating the same key questions (albeit to a different sample of respondents) and looking for differences between the ‘pre-conflict’ sample from the main 2022 survey and the ‘during conflict’ follow-up survey. Of course, we need to be cautious about attributing all changes to coverage of the conflict. Major domestic stories – such as the ‘Partygate’ scandal in the UK – were unfolding in parallel, and these may also have influenced people’s attitudes and behaviours around news.\n\nIn general, we find little evidence that these long-term trends have been reversed – even temporarily – and in some cases there’s possible evidence that they have been accelerated. The clearest example of this is news avoidance. In Germany, Poland, and the US, the proportion who say they sometimes or often actively avoid the news has increased. The biggest increase of all was in Germany (+7pp), but significant increases can also be seen in Poland (+6pp) and the US (+4pp). To put some of these changes in context, the increase of 7pp in Germany in just two months is larger than the 5pp increase we saw in the _five years_ from 2017 to 2022. We know that one of the main reasons people avoid the news is because of the negative effect it has on their mood, so it would be unsurprising if the deeply depressing and concerning nature of conflict has caused more people to turn away from it. In the UK and Brazil, where news avoidance was already high, we do not see evidence of a further increase – but it is equally important to note that it has not decreased either. Furthermore, in these two countries, news avoidance has already increased markedly in recent years – by 11pp from 2019 to 2022 in the case of the UK, and by 20pp in Brazil.\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 35","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Digital_News-Report_2022_p35.pdf"]},"response":{"errors":[],"extract_id":"extract_42e4abf298617d59ef5c5043b7576711","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["REUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 3\n\nPROPORTION WHO SOMETIMES OR OFTEN ACTIVELY AVOID THE NEWS – SELECTED MARKETS\n\n**Q1di 2017.** Do you find yourself actively trying to avoid news these days? _Base: Total pre/during-conflict samples: Brazil = 2022/1002, Germany = 2002/1015, Poland = 2009/1003, UK = 2410/1191, USA = 2036/1356. Note: Pre-conflict sample from the main 2022 DNR survey. During-conflict sample from 29 March to 7 April._\n\n0% 25% 50% 75%\n\nUK 46 | 46\n\nGermany 29\n\nPre-conflict During conflict\n\n36\n\nPoland 41 47\n\nUSA 42 46\n\nBrazil 54 55\n\nAs we have seen with other major world events, people turn more to TV news at times of crisis. When asking respondents which news source they are paying the most attention to when it comes to the Russia–Ukraine conflict, television tops the list for three of the five countries – with the _most_ attention to TV news on the conflict spread out geographically, from Germany (46%) and Brazil (44%). In the US and Poland, online news sites, non-mainstream sites, and social media combined account for a larger share – but TV is still the most widely used individual source for news about the conflict.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 2\n\nPROPORTION WHO PAY MOST ATTENTION TO EACH SOURCE FOR NEWS ABOUT THE RUSSIA–UKRAINE CONFLICT – SELECTED MARKETS\n\n**Q2\\_ukraine.** When it comes to the current Russia–Ukraine conflict, which ONE of the following sources of news, if any, are you paying most attention to? _Base: Total sample: Brazil = 1002, Germany = 1015, Poland = 1003, UK = 1191, USA = 1356._\n\n0% 25% 50%\n\nBrazil Television\n\nUSA UK Germany Poland\n\n44\n\n31\n\n39\n\n46\n\n38\n\nBrazil Online news sites from major news organisations USA UK Germany Poland\n\n13\n\n18\n\n23\n\n14\n\n20\n\nBrazil **Online** (news sites + non-mainstream sites + social)\n\nUSA UK Germany Poland\n\n41\n\n40\n\n33\n\n24\n\n40\n\nBrazil Social media sites USA UK Germany Poland\n\n23\n\n13\n\n7 6\n\n11\n\nBrazil Radio USA UK Germany Poland\n\n5 6\n\n9 9\n\n2\n\nBrazil Non-mainstream sites (e.g. specialist websites, or alternative sites)\n\nUSA UK Germany Poland\n\n5\n\n9\n\n4\n\n9\n\n3\n\nBrazil Conversations with colleagues, friends, or family USA UK Germany Poland\n\n5 5\n\n5 5\n\n3\n\nBrazil Printed newspapers USA UK Germany Poland\n\n4 4\n\n1\n\n1\n\n3\n\nWe see some differences in preferred sources by both age and market. For instance, the same splits we see in TV versus social media use between younger and older cohorts also applies to the most important platform for following the Russia–Ukraine conflict. In the UK, for instance, more than half (55%) of those 55 or older, but only 13% of 18–24s, are paying the most attention to TV news on the Russia–Ukraine conflict. And nearly eight times as many 18–24s (15%) are paying the most attention to social media news on the war as those 55 and older (2%). Further, while TV is especially dominant for information about the conflict in countries like Germany, social media are a particularly important source in Brazil, with 23% of people paying the most attention to these platforms for related information.\n\n**HOW HAS THE CONFLICT INFLUENCED PEOPLE’S NEWS BEHAVIOURS AND ATTITUDES?**\n\nWe also used this follow-up survey to explore whether the trends we described in the Executive Summary – namely, decreases in news use and interest and increases in news avoidance – were reversed by the extensive coverage of the Russia–Ukraine conflict. We did this by repeating the same key questions (albeit to a different sample of respondents) and looking for differences between the ‘pre-conflict’ sample from the main 2022 survey and the ‘during conflict’ follow-up survey. Of course, we need to be cautious about attributing all changes to coverage of the conflict. Major domestic stories – such as the ‘Partygate’ scandal in the UK – were unfolding in parallel, and these may also have influenced people’s attitudes and behaviours around news.\n\nIn general, we find little evidence that these long-term trends have been reversed – even temporarily – and in some cases there’s possible evidence that they have been accelerated. The clearest example of this is news avoidance. In Germany, Poland, and the US, the proportion who say they sometimes or often actively avoid the news has increased. The biggest increase of all was in Germany (+7pp), but significant increases can also be seen in Poland (+6pp) and the US (+4pp). To put some of these changes in context, the increase of 7pp in Germany in just two months is larger than the 5pp increase we saw in the _five years_ from 2017 to 2022. We know that one of the main reasons people avoid the news is because of the negative effect it has on their mood, so it would be unsurprising if the deeply depressing and concerning nature of conflict has caused more people to turn away from it. In the UK and Brazil, where news avoidance was already high, we do not see evidence of a further increase – but it is equally important to note that it has not decreased either. Furthermore, in these two countries, news avoidance has already increased markedly in recent years – by 11pp from 2019 to 2022 in the case of the UK, and by 20pp in Brazil.\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 35"],"full_content":"REUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 3\n\nPROPORTION WHO SOMETIMES OR OFTEN ACTIVELY AVOID THE NEWS – SELECTED MARKETS\n\n**Q1di 2017.** Do you find yourself actively trying to avoid news these days? _Base: Total pre/during-conflict samples: Brazil = 2022/1002, Germany = 2002/1015, Poland = 2009/1003, UK = 2410/1191, USA = 2036/1356. Note: Pre-conflict sample from the main 2022 DNR survey. During-conflict sample from 29 March to 7 April._\n\n0% 25% 50% 75%\n\nUK 46 | 46\n\nGermany 29\n\nPre-conflict During conflict\n\n36\n\nPoland 41 47\n\nUSA 42 46\n\nBrazil 54 55\n\nAs we have seen with other major world events, people turn more to TV news at times of crisis. When asking respondents which news source they are paying the most attention to when it comes to the Russia–Ukraine conflict, television tops the list for three of the five countries – with the _most_ attention to TV news on the conflict spread out geographically, from Germany (46%) and Brazil (44%). In the US and Poland, online news sites, non-mainstream sites, and social media combined account for a larger share – but TV is still the most widely used individual source for news about the conflict.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 2\n\nPROPORTION WHO PAY MOST ATTENTION TO EACH SOURCE FOR NEWS ABOUT THE RUSSIA–UKRAINE CONFLICT – SELECTED MARKETS\n\n**Q2\\_ukraine.** When it comes to the current Russia–Ukraine conflict, which ONE of the following sources of news, if any, are you paying most attention to? _Base: Total sample: Brazil = 1002, Germany = 1015, Poland = 1003, UK = 1191, USA = 1356._\n\n0% 25% 50%\n\nBrazil Television\n\nUSA UK Germany Poland\n\n44\n\n31\n\n39\n\n46\n\n38\n\nBrazil Online news sites from major news organisations USA UK Germany Poland\n\n13\n\n18\n\n23\n\n14\n\n20\n\nBrazil **Online** (news sites + non-mainstream sites + social)\n\nUSA UK Germany Poland\n\n41\n\n40\n\n33\n\n24\n\n40\n\nBrazil Social media sites USA UK Germany Poland\n\n23\n\n13\n\n7 6\n\n11\n\nBrazil Radio USA UK Germany Poland\n\n5 6\n\n9 9\n\n2\n\nBrazil Non-mainstream sites (e.g. specialist websites, or alternative sites)\n\nUSA UK Germany Poland\n\n5\n\n9\n\n4\n\n9\n\n3\n\nBrazil Conversations with colleagues, friends, or family USA UK Germany Poland\n\n5 5\n\n5 5\n\n3\n\nBrazil Printed newspapers USA UK Germany Poland\n\n4 4\n\n1\n\n1\n\n3\n\nWe see some differences in preferred sources by both age and market. For instance, the same splits we see in TV versus social media use between younger and older cohorts also applies to the most important platform for following the Russia–Ukraine conflict. In the UK, for instance, more than half (55%) of those 55 or older, but only 13% of 18–24s, are paying the most attention to TV news on the Russia–Ukraine conflict. And nearly eight times as many 18–24s (15%) are paying the most attention to social media news on the war as those 55 and older (2%). Further, while TV is especially dominant for information about the conflict in countries like Germany, social media are a particularly important source in Brazil, with 23% of people paying the most attention to these platforms for related information.\n\n**HOW HAS THE CONFLICT INFLUENCED PEOPLE’S NEWS BEHAVIOURS AND ATTITUDES?**\n\nWe also used this follow-up survey to explore whether the trends we described in the Executive Summary – namely, decreases in news use and interest and increases in news avoidance – were reversed by the extensive coverage of the Russia–Ukraine conflict. We did this by repeating the same key questions (albeit to a different sample of respondents) and looking for differences between the ‘pre-conflict’ sample from the main 2022 survey and the ‘during conflict’ follow-up survey. Of course, we need to be cautious about attributing all changes to coverage of the conflict. Major domestic stories – such as the ‘Partygate’ scandal in the UK – were unfolding in parallel, and these may also have influenced people’s attitudes and behaviours around news.\n\nIn general, we find little evidence that these long-term trends have been reversed – even temporarily – and in some cases there’s possible evidence that they have been accelerated. The clearest example of this is news avoidance. In Germany, Poland, and the US, the proportion who say they sometimes or often actively avoid the news has increased. The biggest increase of all was in Germany (+7pp), but significant increases can also be seen in Poland (+6pp) and the US (+4pp). To put some of these changes in context, the increase of 7pp in Germany in just two months is larger than the 5pp increase we saw in the _five years_ from 2017 to 2022. We know that one of the main reasons people avoid the news is because of the negative effect it has on their mood, so it would be unsurprising if the deeply depressing and concerning nature of conflict has caused more people to turn away from it. In the UK and Brazil, where news avoidance was already high, we do not see evidence of a further increase – but it is equally important to note that it has not decreased either. Furthermore, in these two countries, news avoidance has already increased markedly in recent years – by 11pp from 2019 to 2022 in the case of the UK, and by 20pp in Brazil.\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 35","publish_date":null,"title":"2.1 Perceptions of Media Coverage of the War in Ukraine","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Digital_News-Report_2022_p35.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Digital_News-Report_2022_p35.pdf"},"request":{"example_id":"chart/Digital_News-Report_2022_p35","product_type":"parse","source_file_path":"Digital_News-Report_2022_p35.pdf"},"started_at":"2026-11-04T19:46:22.795688"} +{"completed_at":"2026-11-04T19:46:25.454029","latency_in_ms":1307,"output":{"example_id":"chart/Digital_News-Report_2022_p36","grounded_pages":[],"layout_pages":[],"markdown":"Despite the increase in news avoidance, the proportion who say they access news several times a day increased in Poland by 6pp. This highlights that news avoidance and news use are not mutually exclusive, and that people can make a conscious decision to moderate their news use – or perhaps coverage on specific topics – while still regularly checking in. In Poland, around 40% of people who accessed news several times a day during the conflict also say they sometimes or often actively avoid the news. The proportion who accessed the news several times a day fell by 6pp in Brazil and 4pp in Germany, and remained stable elsewhere.\n\nIn Poland and Germany, closest to the conflict, TV use for news _in general_ is also up from before the conflict (3pp for Germany and 9pp for Poland). Use of social media for news in general remains even or is down across all countries – in Germany, for instance, by as much as 7pp for social media – perhaps because people are seeking out professional coverage or want to step back from continuous updates. But in general, it is difficult to draw any firm conclusions about the effect of the Russia– Ukraine crisis on frequency of news access, especially as the polling took place several weeks after the invasion began.\n\nLooking instead at interest levels in news, we again see the biggest change in Poland – a 7pp increase from pre-conflict levels. However, because declines in news interest have been so steep since 2020, this increase only represents a return to 2021 levels. Elsewhere, interest has either been unaffected, or – as in Brazil (-6pp) – continues to fall. Despite its severity, the Russia–Ukraine conflict has done little to reverse declining levels of interest in most countries, even in the short term. We see a similar picture when we look at levels of trust in news. Despite the bravery of journalists on the ground in Ukraine and the remarkable reporting they have produced, trust in news has not been affected – aside from a 4pp increase in the UK.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 4\n\nPROPORTION VERY OR EXTREMELY INTERESTED IN NEWS (2015–2022) – SELECTED MARKETS\n\n**Q1c.** How interested, if at all, would you say you are in news? _Base: Total 2015–22 samples (n ≈ 2000). Total during-conflict sample (n ≈ 1000)._\n\nUSA UK Germany Poland Brazil\n\n0%\n\n25%\n\n50%\n\n75%\n\n100%\n\nBrazil\n\nPoland\n\nGermany\n\nUK\n\nUS\n\n2022 2021 2020 2019 2018 2017 2016 2015\n\nRussia-Ukraine confict\n\n**HOW DO PEOPLE FEEL ABOUT THE NEWS MEDIA’S COVERAGE OF THE CONFLICT?**\n\nFinally, we wanted to gauge how people feel about the media’s performance covering the Russia–Ukraine conflict. To do this, we asked people to rate how well the news media have done keeping people up to date, explaining the wider implications, and providing a range of different perspectives on the conflict. The news media are broadly seen to be doing a good job with coverage, especially on keeping people up to date on the latest news – with nearly half or more of respondents in all five countries agreeing the media have done a good job with this. But in general, they feel the media have not performed quite as well for explaining the wider implications of the conflict or providing a different range of perspectives on it – clearly illustrating, across all five countries, where news organisations could be doing a better job as the crisis continues to unfold. In general, people in the older age groups, and those with more formal education, tend to rate all aspects of the coverage more favourably.\n\n**CONCLUSIONS**\n\nIn the wake of thousands of deaths and the displacement of more than six million Ukrainian refugees ¹⁴ – and with Russia now being accused by the international community of widespread human rights violations – the war in Ukraine continues to attract global attention. Across the five countries we study here, a majority are following the conflict at least somewhat closely, especially those nearest to it.\n\nAs we’ve seen with other major world events, many people are turning to television news for the latest information on the conflict – illustrating the continuing resonance of broadcast media in times of crisis. This may not be surprising, particularly given concerns about false or misleading information circulating on social media platforms. But in some countries, such as the US and Poland, attention to news from online sources (including mainstream and alternative news sites as well as social media) is also high.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 5\n\nPROPORTION WHO THINK THE NEWS MEDIA HAVE DONE A GOOD JOB OF EACH – SELECTED MARKETS\n\n**Q3\\_ukraine.** Do you think the news media in your country have done a good job or a bad job … _Base: Total sample: Brazil = 1002, Germany = 1015, Poland = 1003, UK = 1191, USA = 1356._\n\n0% 25% 75% 50%\n\nBrazil ...keeping you up to date with the current Russia- Ukraine conflict?\n\nUSA UK Germany Poland\n\n58\n\n47\n\n65\n\n55\n\n64\n\nBrazil ...explaining the wider implications of the current Russia-Ukraine conflict?\n\nUSA UK Germany Poland\n\n52\n\n38\n\n50\n\n46\n\n49\n\nBrazil ...providing a range of different perspectives on the current Russia-Ukraine conflict?\n\nUSA UK Germany Poland\n\n49\n\n36\n\n43\n\n40\n\n51\n\n14 https://data2.unhcr.org/en/situations/ukraine/location?secret=unhcrrestricted\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 36","pages":[{"line_numbers":[],"markdown":"Despite the increase in news avoidance, the proportion who say they access news several times a day increased in Poland by 6pp. This highlights that news avoidance and news use are not mutually exclusive, and that people can make a conscious decision to moderate their news use – or perhaps coverage on specific topics – while still regularly checking in. In Poland, around 40% of people who accessed news several times a day during the conflict also say they sometimes or often actively avoid the news. The proportion who accessed the news several times a day fell by 6pp in Brazil and 4pp in Germany, and remained stable elsewhere.\n\nIn Poland and Germany, closest to the conflict, TV use for news _in general_ is also up from before the conflict (3pp for Germany and 9pp for Poland). Use of social media for news in general remains even or is down across all countries – in Germany, for instance, by as much as 7pp for social media – perhaps because people are seeking out professional coverage or want to step back from continuous updates. But in general, it is difficult to draw any firm conclusions about the effect of the Russia– Ukraine crisis on frequency of news access, especially as the polling took place several weeks after the invasion began.\n\nLooking instead at interest levels in news, we again see the biggest change in Poland – a 7pp increase from pre-conflict levels. However, because declines in news interest have been so steep since 2020, this increase only represents a return to 2021 levels. Elsewhere, interest has either been unaffected, or – as in Brazil (-6pp) – continues to fall. Despite its severity, the Russia–Ukraine conflict has done little to reverse declining levels of interest in most countries, even in the short term. We see a similar picture when we look at levels of trust in news. Despite the bravery of journalists on the ground in Ukraine and the remarkable reporting they have produced, trust in news has not been affected – aside from a 4pp increase in the UK.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 4\n\nPROPORTION VERY OR EXTREMELY INTERESTED IN NEWS (2015–2022) – SELECTED MARKETS\n\n**Q1c.** How interested, if at all, would you say you are in news? _Base: Total 2015–22 samples (n ≈ 2000). Total during-conflict sample (n ≈ 1000)._\n\nUSA UK Germany Poland Brazil\n\n0%\n\n25%\n\n50%\n\n75%\n\n100%\n\nBrazil\n\nPoland\n\nGermany\n\nUK\n\nUS\n\n2022 2021 2020 2019 2018 2017 2016 2015\n\nRussia-Ukraine confict\n\n**HOW DO PEOPLE FEEL ABOUT THE NEWS MEDIA’S COVERAGE OF THE CONFLICT?**\n\nFinally, we wanted to gauge how people feel about the media’s performance covering the Russia–Ukraine conflict. To do this, we asked people to rate how well the news media have done keeping people up to date, explaining the wider implications, and providing a range of different perspectives on the conflict. The news media are broadly seen to be doing a good job with coverage, especially on keeping people up to date on the latest news – with nearly half or more of respondents in all five countries agreeing the media have done a good job with this. But in general, they feel the media have not performed quite as well for explaining the wider implications of the conflict or providing a different range of perspectives on it – clearly illustrating, across all five countries, where news organisations could be doing a better job as the crisis continues to unfold. In general, people in the older age groups, and those with more formal education, tend to rate all aspects of the coverage more favourably.\n\n**CONCLUSIONS**\n\nIn the wake of thousands of deaths and the displacement of more than six million Ukrainian refugees ¹⁴ – and with Russia now being accused by the international community of widespread human rights violations – the war in Ukraine continues to attract global attention. Across the five countries we study here, a majority are following the conflict at least somewhat closely, especially those nearest to it.\n\nAs we’ve seen with other major world events, many people are turning to television news for the latest information on the conflict – illustrating the continuing resonance of broadcast media in times of crisis. This may not be surprising, particularly given concerns about false or misleading information circulating on social media platforms. But in some countries, such as the US and Poland, attention to news from online sources (including mainstream and alternative news sites as well as social media) is also high.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 5\n\nPROPORTION WHO THINK THE NEWS MEDIA HAVE DONE A GOOD JOB OF EACH – SELECTED MARKETS\n\n**Q3\\_ukraine.** Do you think the news media in your country have done a good job or a bad job … _Base: Total sample: Brazil = 1002, Germany = 1015, Poland = 1003, UK = 1191, USA = 1356._\n\n0% 25% 75% 50%\n\nBrazil ...keeping you up to date with the current Russia- Ukraine conflict?\n\nUSA UK Germany Poland\n\n58\n\n47\n\n65\n\n55\n\n64\n\nBrazil ...explaining the wider implications of the current Russia-Ukraine conflict?\n\nUSA UK Germany Poland\n\n52\n\n38\n\n50\n\n46\n\n49\n\nBrazil ...providing a range of different perspectives on the current Russia-Ukraine conflict?\n\nUSA UK Germany Poland\n\n49\n\n36\n\n43\n\n40\n\n51\n\n14 https://data2.unhcr.org/en/situations/ukraine/location?secret=unhcrrestricted\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 36","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Despite the increase in news avoidance, the proportion who say they access news several times a day increased in Poland by 6pp. This highlights that news avoidance and news use are not mutually exclusive, and that people can make a conscious decision to moderate their news use – or perhaps coverage on specific topics – while still regularly checking in. In Poland, around 40% of people who accessed news several times a day during the conflict also say they sometimes or often actively avoid the news. The proportion who accessed the news several times a day fell by 6pp in Brazil and 4pp in Germany, and remained stable elsewhere.\n\nIn Poland and Germany, closest to the conflict, TV use for news _in general_ is also up from before the conflict (3pp for Germany and 9pp for Poland). Use of social media for news in general remains even or is down across all countries – in Germany, for instance, by as much as 7pp for social media – perhaps because people are seeking out professional coverage or want to step back from continuous updates. But in general, it is difficult to draw any firm conclusions about the effect of the Russia– Ukraine crisis on frequency of news access, especially as the polling took place several weeks after the invasion began.\n\nLooking instead at interest levels in news, we again see the biggest change in Poland – a 7pp increase from pre-conflict levels. However, because declines in news interest have been so steep since 2020, this increase only represents a return to 2021 levels. Elsewhere, interest has either been unaffected, or – as in Brazil (-6pp) – continues to fall. Despite its severity, the Russia–Ukraine conflict has done little to reverse declining levels of interest in most countries, even in the short term. We see a similar picture when we look at levels of trust in news. Despite the bravery of journalists on the ground in Ukraine and the remarkable reporting they have produced, trust in news has not been affected – aside from a 4pp increase in the UK.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 4\n\nPROPORTION VERY OR EXTREMELY INTERESTED IN NEWS (2015–2022) – SELECTED MARKETS\n\n**Q1c.** How interested, if at all, would you say you are in news? _Base: Total 2015–22 samples (n ≈ 2000). Total during-conflict sample (n ≈ 1000)._\n\nUSA UK Germany Poland Brazil\n\n0%\n\n25%\n\n50%\n\n75%\n\n100%\n\nBrazil\n\nPoland\n\nGermany\n\nUK\n\nUS\n\n2022 2021 2020 2019 2018 2017 2016 2015\n\nRussia-Ukraine confict\n\n**HOW DO PEOPLE FEEL ABOUT THE NEWS MEDIA’S COVERAGE OF THE CONFLICT?**\n\nFinally, we wanted to gauge how people feel about the media’s performance covering the Russia–Ukraine conflict. To do this, we asked people to rate how well the news media have done keeping people up to date, explaining the wider implications, and providing a range of different perspectives on the conflict. The news media are broadly seen to be doing a good job with coverage, especially on keeping people up to date on the latest news – with nearly half or more of respondents in all five countries agreeing the media have done a good job with this. But in general, they feel the media have not performed quite as well for explaining the wider implications of the conflict or providing a different range of perspectives on it – clearly illustrating, across all five countries, where news organisations could be doing a better job as the crisis continues to unfold. In general, people in the older age groups, and those with more formal education, tend to rate all aspects of the coverage more favourably.\n\n**CONCLUSIONS**\n\nIn the wake of thousands of deaths and the displacement of more than six million Ukrainian refugees ¹⁴ – and with Russia now being accused by the international community of widespread human rights violations – the war in Ukraine continues to attract global attention. Across the five countries we study here, a majority are following the conflict at least somewhat closely, especially those nearest to it.\n\nAs we’ve seen with other major world events, many people are turning to television news for the latest information on the conflict – illustrating the continuing resonance of broadcast media in times of crisis. This may not be surprising, particularly given concerns about false or misleading information circulating on social media platforms. But in some countries, such as the US and Poland, attention to news from online sources (including mainstream and alternative news sites as well as social media) is also high.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 5\n\nPROPORTION WHO THINK THE NEWS MEDIA HAVE DONE A GOOD JOB OF EACH – SELECTED MARKETS\n\n**Q3\\_ukraine.** Do you think the news media in your country have done a good job or a bad job … _Base: Total sample: Brazil = 1002, Germany = 1015, Poland = 1003, UK = 1191, USA = 1356._\n\n0% 25% 75% 50%\n\nBrazil ...keeping you up to date with the current Russia- Ukraine conflict?\n\nUSA UK Germany Poland\n\n58\n\n47\n\n65\n\n55\n\n64\n\nBrazil ...explaining the wider implications of the current Russia-Ukraine conflict?\n\nUSA UK Germany Poland\n\n52\n\n38\n\n50\n\n46\n\n49\n\nBrazil ...providing a range of different perspectives on the current Russia-Ukraine conflict?\n\nUSA UK Germany Poland\n\n49\n\n36\n\n43\n\n40\n\n51\n\n14 https://data2.unhcr.org/en/situations/ukraine/location?secret=unhcrrestricted\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 36","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Digital_News-Report_2022_p36.pdf"]},"response":{"errors":[],"extract_id":"extract_33d6ef4d29ad9b7c76ab6f929a1fad62","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Despite the increase in news avoidance, the proportion who say they access news several times a day increased in Poland by 6pp. This highlights that news avoidance and news use are not mutually exclusive, and that people can make a conscious decision to moderate their news use – or perhaps coverage on specific topics – while still regularly checking in. In Poland, around 40% of people who accessed news several times a day during the conflict also say they sometimes or often actively avoid the news. The proportion who accessed the news several times a day fell by 6pp in Brazil and 4pp in Germany, and remained stable elsewhere.\n\nIn Poland and Germany, closest to the conflict, TV use for news _in general_ is also up from before the conflict (3pp for Germany and 9pp for Poland). Use of social media for news in general remains even or is down across all countries – in Germany, for instance, by as much as 7pp for social media – perhaps because people are seeking out professional coverage or want to step back from continuous updates. But in general, it is difficult to draw any firm conclusions about the effect of the Russia– Ukraine crisis on frequency of news access, especially as the polling took place several weeks after the invasion began.\n\nLooking instead at interest levels in news, we again see the biggest change in Poland – a 7pp increase from pre-conflict levels. However, because declines in news interest have been so steep since 2020, this increase only represents a return to 2021 levels. Elsewhere, interest has either been unaffected, or – as in Brazil (-6pp) – continues to fall. Despite its severity, the Russia–Ukraine conflict has done little to reverse declining levels of interest in most countries, even in the short term. We see a similar picture when we look at levels of trust in news. Despite the bravery of journalists on the ground in Ukraine and the remarkable reporting they have produced, trust in news has not been affected – aside from a 4pp increase in the UK.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 4\n\nPROPORTION VERY OR EXTREMELY INTERESTED IN NEWS (2015–2022) – SELECTED MARKETS\n\n**Q1c.** How interested, if at all, would you say you are in news? _Base: Total 2015–22 samples (n ≈ 2000). Total during-conflict sample (n ≈ 1000)._\n\nUSA UK Germany Poland Brazil\n\n0%\n\n25%\n\n50%\n\n75%\n\n100%\n\nBrazil\n\nPoland\n\nGermany\n\nUK\n\nUS\n\n2022 2021 2020 2019 2018 2017 2016 2015\n\nRussia-Ukraine confict\n\n**HOW DO PEOPLE FEEL ABOUT THE NEWS MEDIA’S COVERAGE OF THE CONFLICT?**\n\nFinally, we wanted to gauge how people feel about the media’s performance covering the Russia–Ukraine conflict. To do this, we asked people to rate how well the news media have done keeping people up to date, explaining the wider implications, and providing a range of different perspectives on the conflict. The news media are broadly seen to be doing a good job with coverage, especially on keeping people up to date on the latest news – with nearly half or more of respondents in all five countries agreeing the media have done a good job with this. But in general, they feel the media have not performed quite as well for explaining the wider implications of the conflict or providing a different range of perspectives on it – clearly illustrating, across all five countries, where news organisations could be doing a better job as the crisis continues to unfold. In general, people in the older age groups, and those with more formal education, tend to rate all aspects of the coverage more favourably.\n\n**CONCLUSIONS**\n\nIn the wake of thousands of deaths and the displacement of more than six million Ukrainian refugees ¹⁴ – and with Russia now being accused by the international community of widespread human rights violations – the war in Ukraine continues to attract global attention. Across the five countries we study here, a majority are following the conflict at least somewhat closely, especially those nearest to it.\n\nAs we’ve seen with other major world events, many people are turning to television news for the latest information on the conflict – illustrating the continuing resonance of broadcast media in times of crisis. This may not be surprising, particularly given concerns about false or misleading information circulating on social media platforms. But in some countries, such as the US and Poland, attention to news from online sources (including mainstream and alternative news sites as well as social media) is also high.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 5\n\nPROPORTION WHO THINK THE NEWS MEDIA HAVE DONE A GOOD JOB OF EACH – SELECTED MARKETS\n\n**Q3\\_ukraine.** Do you think the news media in your country have done a good job or a bad job … _Base: Total sample: Brazil = 1002, Germany = 1015, Poland = 1003, UK = 1191, USA = 1356._\n\n0% 25% 75% 50%\n\nBrazil ...keeping you up to date with the current Russia- Ukraine conflict?\n\nUSA UK Germany Poland\n\n58\n\n47\n\n65\n\n55\n\n64\n\nBrazil ...explaining the wider implications of the current Russia-Ukraine conflict?\n\nUSA UK Germany Poland\n\n52\n\n38\n\n50\n\n46\n\n49\n\nBrazil ...providing a range of different perspectives on the current Russia-Ukraine conflict?\n\nUSA UK Germany Poland\n\n49\n\n36\n\n43\n\n40\n\n51\n\n14 https://data2.unhcr.org/en/situations/ukraine/location?secret=unhcrrestricted\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 36"],"full_content":"Despite the increase in news avoidance, the proportion who say they access news several times a day increased in Poland by 6pp. This highlights that news avoidance and news use are not mutually exclusive, and that people can make a conscious decision to moderate their news use – or perhaps coverage on specific topics – while still regularly checking in. In Poland, around 40% of people who accessed news several times a day during the conflict also say they sometimes or often actively avoid the news. The proportion who accessed the news several times a day fell by 6pp in Brazil and 4pp in Germany, and remained stable elsewhere.\n\nIn Poland and Germany, closest to the conflict, TV use for news _in general_ is also up from before the conflict (3pp for Germany and 9pp for Poland). Use of social media for news in general remains even or is down across all countries – in Germany, for instance, by as much as 7pp for social media – perhaps because people are seeking out professional coverage or want to step back from continuous updates. But in general, it is difficult to draw any firm conclusions about the effect of the Russia– Ukraine crisis on frequency of news access, especially as the polling took place several weeks after the invasion began.\n\nLooking instead at interest levels in news, we again see the biggest change in Poland – a 7pp increase from pre-conflict levels. However, because declines in news interest have been so steep since 2020, this increase only represents a return to 2021 levels. Elsewhere, interest has either been unaffected, or – as in Brazil (-6pp) – continues to fall. Despite its severity, the Russia–Ukraine conflict has done little to reverse declining levels of interest in most countries, even in the short term. We see a similar picture when we look at levels of trust in news. Despite the bravery of journalists on the ground in Ukraine and the remarkable reporting they have produced, trust in news has not been affected – aside from a 4pp increase in the UK.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 4\n\nPROPORTION VERY OR EXTREMELY INTERESTED IN NEWS (2015–2022) – SELECTED MARKETS\n\n**Q1c.** How interested, if at all, would you say you are in news? _Base: Total 2015–22 samples (n ≈ 2000). Total during-conflict sample (n ≈ 1000)._\n\nUSA UK Germany Poland Brazil\n\n0%\n\n25%\n\n50%\n\n75%\n\n100%\n\nBrazil\n\nPoland\n\nGermany\n\nUK\n\nUS\n\n2022 2021 2020 2019 2018 2017 2016 2015\n\nRussia-Ukraine confict\n\n**HOW DO PEOPLE FEEL ABOUT THE NEWS MEDIA’S COVERAGE OF THE CONFLICT?**\n\nFinally, we wanted to gauge how people feel about the media’s performance covering the Russia–Ukraine conflict. To do this, we asked people to rate how well the news media have done keeping people up to date, explaining the wider implications, and providing a range of different perspectives on the conflict. The news media are broadly seen to be doing a good job with coverage, especially on keeping people up to date on the latest news – with nearly half or more of respondents in all five countries agreeing the media have done a good job with this. But in general, they feel the media have not performed quite as well for explaining the wider implications of the conflict or providing a different range of perspectives on it – clearly illustrating, across all five countries, where news organisations could be doing a better job as the crisis continues to unfold. In general, people in the older age groups, and those with more formal education, tend to rate all aspects of the coverage more favourably.\n\n**CONCLUSIONS**\n\nIn the wake of thousands of deaths and the displacement of more than six million Ukrainian refugees ¹⁴ – and with Russia now being accused by the international community of widespread human rights violations – the war in Ukraine continues to attract global attention. Across the five countries we study here, a majority are following the conflict at least somewhat closely, especially those nearest to it.\n\nAs we’ve seen with other major world events, many people are turning to television news for the latest information on the conflict – illustrating the continuing resonance of broadcast media in times of crisis. This may not be surprising, particularly given concerns about false or misleading information circulating on social media platforms. But in some countries, such as the US and Poland, attention to news from online sources (including mainstream and alternative news sites as well as social media) is also high.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 5\n\nPROPORTION WHO THINK THE NEWS MEDIA HAVE DONE A GOOD JOB OF EACH – SELECTED MARKETS\n\n**Q3\\_ukraine.** Do you think the news media in your country have done a good job or a bad job … _Base: Total sample: Brazil = 1002, Germany = 1015, Poland = 1003, UK = 1191, USA = 1356._\n\n0% 25% 75% 50%\n\nBrazil ...keeping you up to date with the current Russia- Ukraine conflict?\n\nUSA UK Germany Poland\n\n58\n\n47\n\n65\n\n55\n\n64\n\nBrazil ...explaining the wider implications of the current Russia-Ukraine conflict?\n\nUSA UK Germany Poland\n\n52\n\n38\n\n50\n\n46\n\n49\n\nBrazil ...providing a range of different perspectives on the current Russia-Ukraine conflict?\n\nUSA UK Germany Poland\n\n49\n\n36\n\n43\n\n40\n\n51\n\n14 https://data2.unhcr.org/en/situations/ukraine/location?secret=unhcrrestricted\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 36","publish_date":null,"title":"Graph 5 Graph 4","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Digital_News-Report_2022_p36.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Digital_News-Report_2022_p36.pdf"},"request":{"example_id":"chart/Digital_News-Report_2022_p36","product_type":"parse","source_file_path":"Digital_News-Report_2022_p36.pdf"},"started_at":"2026-11-04T19:46:24.146430"} +{"completed_at":"2026-11-04T19:46:27.494191","latency_in_ms":2038,"output":{"example_id":"chart/Digital_News-Report_2022_p39","grounded_pages":[],"layout_pages":[],"markdown":"**NEWS AUDIENCE POLARISATION VARIES BY COUNTRY**\n\nThe UK has a relatively high level of news audience polarisation. The spread of the bubbles in the chart above shows that news audience polarisation is higher than in Norway and Germany, but lower than in the USA where there is no large outlet in the middle of the map, highlighting the importance of public service media as a central anchor in much of Europe (though local news media, which we do not plot here, may partly fulfil this role in the USA). News audiences are less polarised in Norway, but while most traditional newspaper and broadcast brands have mixed and/or centrist audiences, they have recently been joined by several partisan/alternative brands – like Document.no and HRS – that have audiences with a stronger political skew – occupying space further from the centre of the map. In Germany, although some new digital-born brands have emerged in recent years, audiences for most outlets are predominantly centrist. This is partly because fewer people self-identify with either the left or the right in Germany, precluding very high levels of news audience polarisation along this dimension (though, of course, both the public and news audiences may be polarised along different dimensions).\n\nWe can compute a news audience polarisation score for each country – based on the standard deviation of outlet audiences, weighted by their size – that reflects the degree to which polarisation approaches a theoretical maximum, where only outlets with an entirely left-leaning and entirely right-leaning audiences exist, with nothing in between. This allows us to more easily compare across countries, but it also highlights that in most countries news audience polarisation is low. In Germany, for example, polarisation is just 10% of what it could be, and even in the USA (34%), polarisation is far from the theoretical limit.\n\n**IS NEWS AUDIENCE POLARISATION ON THE RISE?**\n\nTo explore whether news audience polarisation is growing, we can compare the most recent data with those from 2016 – the first point at which we started asking respondents whether they self-identify politically with the left or the right outside of the UK and the USA.\n\nAt least in the four countries we focus on here, news audience polarisation has not increased substantially – if at all – in the last six years. The most striking aspect of the maps is how little they have changed over time. Across all four countries, news audience polarisation has changed by 3pp or less since 2016, indicating only minimal shifts in audience behaviour.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 8\n\n0%\n\n25%\n\n50%\n\n75%\n\n100%\n\n34\n\n25\n\n15 10\n\nGermany Norway UK USA\n\nEXTENT OF CROSS-PLATFORM NEWS AUDIENCE POLARISATION – SELECTED COUNTRIES\n\n**Q1F.** Some people talk about ‘left’, ‘right’, and ‘centre’ to describe parties and politicians. With this in mind, where would you place yourself on the following scale? **Q5A/B.** Which of the following brands have you used to access news offline/online in the last week? _Base: Germany = 2002, Norway = 2010, UK = 2410, USA = 2036._\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 7\n\nCROSS-PLATFORM NEWS AUDIENCE POLARISATION – SELECTED COUNTRIES\n\nVery right-leaning audience Average political leaning of population Very lef-leaning audience\n\nARD News Bild ZEIT\n\nVery right-leaning audience Average political leaning of population Very lef-leaning audience\n\nGuardian Daily Mail BBC News\n\nVery right-leaning audience Average political leaning of population Very lef-leaning audience\n\nDocument.no NRK Dagsavisen\n\nVery right-leaning audience Average political leaning of population Very lef-leaning audience\n\nFox News New York Times CNN\n\n**Q1F.** Some people talk about ‘left’, ‘right’, and ‘centre’ to describe parties and politicians. With this in mind, where would you place yourself on the following scale? **Q5A/B.** Which of the following brands have you used to access news offline/online in the last week? _Base: Germany = 2002, Norway = 2010, UK = 2410, USA = 2036._\n\n**UK**\n\n**Germany**\n\n**USA**\n\n**Norway**\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 39","pages":[{"line_numbers":[],"markdown":"**NEWS AUDIENCE POLARISATION VARIES BY COUNTRY**\n\nThe UK has a relatively high level of news audience polarisation. The spread of the bubbles in the chart above shows that news audience polarisation is higher than in Norway and Germany, but lower than in the USA where there is no large outlet in the middle of the map, highlighting the importance of public service media as a central anchor in much of Europe (though local news media, which we do not plot here, may partly fulfil this role in the USA). News audiences are less polarised in Norway, but while most traditional newspaper and broadcast brands have mixed and/or centrist audiences, they have recently been joined by several partisan/alternative brands – like Document.no and HRS – that have audiences with a stronger political skew – occupying space further from the centre of the map. In Germany, although some new digital-born brands have emerged in recent years, audiences for most outlets are predominantly centrist. This is partly because fewer people self-identify with either the left or the right in Germany, precluding very high levels of news audience polarisation along this dimension (though, of course, both the public and news audiences may be polarised along different dimensions).\n\nWe can compute a news audience polarisation score for each country – based on the standard deviation of outlet audiences, weighted by their size – that reflects the degree to which polarisation approaches a theoretical maximum, where only outlets with an entirely left-leaning and entirely right-leaning audiences exist, with nothing in between. This allows us to more easily compare across countries, but it also highlights that in most countries news audience polarisation is low. In Germany, for example, polarisation is just 10% of what it could be, and even in the USA (34%), polarisation is far from the theoretical limit.\n\n**IS NEWS AUDIENCE POLARISATION ON THE RISE?**\n\nTo explore whether news audience polarisation is growing, we can compare the most recent data with those from 2016 – the first point at which we started asking respondents whether they self-identify politically with the left or the right outside of the UK and the USA.\n\nAt least in the four countries we focus on here, news audience polarisation has not increased substantially – if at all – in the last six years. The most striking aspect of the maps is how little they have changed over time. Across all four countries, news audience polarisation has changed by 3pp or less since 2016, indicating only minimal shifts in audience behaviour.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 8\n\n0%\n\n25%\n\n50%\n\n75%\n\n100%\n\n34\n\n25\n\n15 10\n\nGermany Norway UK USA\n\nEXTENT OF CROSS-PLATFORM NEWS AUDIENCE POLARISATION – SELECTED COUNTRIES\n\n**Q1F.** Some people talk about ‘left’, ‘right’, and ‘centre’ to describe parties and politicians. With this in mind, where would you place yourself on the following scale? **Q5A/B.** Which of the following brands have you used to access news offline/online in the last week? _Base: Germany = 2002, Norway = 2010, UK = 2410, USA = 2036._\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 7\n\nCROSS-PLATFORM NEWS AUDIENCE POLARISATION – SELECTED COUNTRIES\n\nVery right-leaning audience Average political leaning of population Very lef-leaning audience\n\nARD News Bild ZEIT\n\nVery right-leaning audience Average political leaning of population Very lef-leaning audience\n\nGuardian Daily Mail BBC News\n\nVery right-leaning audience Average political leaning of population Very lef-leaning audience\n\nDocument.no NRK Dagsavisen\n\nVery right-leaning audience Average political leaning of population Very lef-leaning audience\n\nFox News New York Times CNN\n\n**Q1F.** Some people talk about ‘left’, ‘right’, and ‘centre’ to describe parties and politicians. With this in mind, where would you place yourself on the following scale? **Q5A/B.** Which of the following brands have you used to access news offline/online in the last week? _Base: Germany = 2002, Norway = 2010, UK = 2410, USA = 2036._\n\n**UK**\n\n**Germany**\n\n**USA**\n\n**Norway**\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 39","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**NEWS AUDIENCE POLARISATION VARIES BY COUNTRY**\n\nThe UK has a relatively high level of news audience polarisation. The spread of the bubbles in the chart above shows that news audience polarisation is higher than in Norway and Germany, but lower than in the USA where there is no large outlet in the middle of the map, highlighting the importance of public service media as a central anchor in much of Europe (though local news media, which we do not plot here, may partly fulfil this role in the USA). News audiences are less polarised in Norway, but while most traditional newspaper and broadcast brands have mixed and/or centrist audiences, they have recently been joined by several partisan/alternative brands – like Document.no and HRS – that have audiences with a stronger political skew – occupying space further from the centre of the map. In Germany, although some new digital-born brands have emerged in recent years, audiences for most outlets are predominantly centrist. This is partly because fewer people self-identify with either the left or the right in Germany, precluding very high levels of news audience polarisation along this dimension (though, of course, both the public and news audiences may be polarised along different dimensions).\n\nWe can compute a news audience polarisation score for each country – based on the standard deviation of outlet audiences, weighted by their size – that reflects the degree to which polarisation approaches a theoretical maximum, where only outlets with an entirely left-leaning and entirely right-leaning audiences exist, with nothing in between. This allows us to more easily compare across countries, but it also highlights that in most countries news audience polarisation is low. In Germany, for example, polarisation is just 10% of what it could be, and even in the USA (34%), polarisation is far from the theoretical limit.\n\n**IS NEWS AUDIENCE POLARISATION ON THE RISE?**\n\nTo explore whether news audience polarisation is growing, we can compare the most recent data with those from 2016 – the first point at which we started asking respondents whether they self-identify politically with the left or the right outside of the UK and the USA.\n\nAt least in the four countries we focus on here, news audience polarisation has not increased substantially – if at all – in the last six years. The most striking aspect of the maps is how little they have changed over time. Across all four countries, news audience polarisation has changed by 3pp or less since 2016, indicating only minimal shifts in audience behaviour.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 8\n\n0%\n\n25%\n\n50%\n\n75%\n\n100%\n\n34\n\n25\n\n15 10\n\nGermany Norway UK USA\n\nEXTENT OF CROSS-PLATFORM NEWS AUDIENCE POLARISATION – SELECTED COUNTRIES\n\n**Q1F.** Some people talk about ‘left’, ‘right’, and ‘centre’ to describe parties and politicians. With this in mind, where would you place yourself on the following scale? **Q5A/B.** Which of the following brands have you used to access news offline/online in the last week? _Base: Germany = 2002, Norway = 2010, UK = 2410, USA = 2036._\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 7\n\nCROSS-PLATFORM NEWS AUDIENCE POLARISATION – SELECTED COUNTRIES\n\nVery right-leaning audience Average political leaning of population Very lef-leaning audience\n\nARD News Bild ZEIT\n\nVery right-leaning audience Average political leaning of population Very lef-leaning audience\n\nGuardian Daily Mail BBC News\n\nVery right-leaning audience Average political leaning of population Very lef-leaning audience\n\nDocument.no NRK Dagsavisen\n\nVery right-leaning audience Average political leaning of population Very lef-leaning audience\n\nFox News New York Times CNN\n\n**Q1F.** Some people talk about ‘left’, ‘right’, and ‘centre’ to describe parties and politicians. With this in mind, where would you place yourself on the following scale? **Q5A/B.** Which of the following brands have you used to access news offline/online in the last week? _Base: Germany = 2002, Norway = 2010, UK = 2410, USA = 2036._\n\n**UK**\n\n**Germany**\n\n**USA**\n\n**Norway**\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 39","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Digital_News-Report_2022_p39.pdf"]},"response":{"errors":[],"extract_id":"extract_7b01f8a64f5e251df8848a8c69061260","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**NEWS AUDIENCE POLARISATION VARIES BY COUNTRY**\n\nThe UK has a relatively high level of news audience polarisation. The spread of the bubbles in the chart above shows that news audience polarisation is higher than in Norway and Germany, but lower than in the USA where there is no large outlet in the middle of the map, highlighting the importance of public service media as a central anchor in much of Europe (though local news media, which we do not plot here, may partly fulfil this role in the USA). News audiences are less polarised in Norway, but while most traditional newspaper and broadcast brands have mixed and/or centrist audiences, they have recently been joined by several partisan/alternative brands – like Document.no and HRS – that have audiences with a stronger political skew – occupying space further from the centre of the map. In Germany, although some new digital-born brands have emerged in recent years, audiences for most outlets are predominantly centrist. This is partly because fewer people self-identify with either the left or the right in Germany, precluding very high levels of news audience polarisation along this dimension (though, of course, both the public and news audiences may be polarised along different dimensions).\n\nWe can compute a news audience polarisation score for each country – based on the standard deviation of outlet audiences, weighted by their size – that reflects the degree to which polarisation approaches a theoretical maximum, where only outlets with an entirely left-leaning and entirely right-leaning audiences exist, with nothing in between. This allows us to more easily compare across countries, but it also highlights that in most countries news audience polarisation is low. In Germany, for example, polarisation is just 10% of what it could be, and even in the USA (34%), polarisation is far from the theoretical limit.\n\n**IS NEWS AUDIENCE POLARISATION ON THE RISE?**\n\nTo explore whether news audience polarisation is growing, we can compare the most recent data with those from 2016 – the first point at which we started asking respondents whether they self-identify politically with the left or the right outside of the UK and the USA.\n\nAt least in the four countries we focus on here, news audience polarisation has not increased substantially – if at all – in the last six years. The most striking aspect of the maps is how little they have changed over time. Across all four countries, news audience polarisation has changed by 3pp or less since 2016, indicating only minimal shifts in audience behaviour.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 8\n\n0%\n\n25%\n\n50%\n\n75%\n\n100%\n\n34\n\n25\n\n15 10\n\nGermany Norway UK USA\n\nEXTENT OF CROSS-PLATFORM NEWS AUDIENCE POLARISATION – SELECTED COUNTRIES\n\n**Q1F.** Some people talk about ‘left’, ‘right’, and ‘centre’ to describe parties and politicians. With this in mind, where would you place yourself on the following scale? **Q5A/B.** Which of the following brands have you used to access news offline/online in the last week? _Base: Germany = 2002, Norway = 2010, UK = 2410, USA = 2036._\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 7\n\nCROSS-PLATFORM NEWS AUDIENCE POLARISATION – SELECTED COUNTRIES\n\nVery right-leaning audience Average political leaning of population Very lef-leaning audience\n\nARD News Bild ZEIT\n\nVery right-leaning audience Average political leaning of population Very lef-leaning audience\n\nGuardian Daily Mail BBC News\n\nVery right-leaning audience Average political leaning of population Very lef-leaning audience\n\nDocument.no NRK Dagsavisen\n\nVery right-leaning audience Average political leaning of population Very lef-leaning audience\n\nFox News New York Times CNN\n\n**Q1F.** Some people talk about ‘left’, ‘right’, and ‘centre’ to describe parties and politicians. With this in mind, where would you place yourself on the following scale? **Q5A/B.** Which of the following brands have you used to access news offline/online in the last week? _Base: Germany = 2002, Norway = 2010, UK = 2410, USA = 2036._\n\n**UK**\n\n**Germany**\n\n**USA**\n\n**Norway**\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 39"],"full_content":"**NEWS AUDIENCE POLARISATION VARIES BY COUNTRY**\n\nThe UK has a relatively high level of news audience polarisation. The spread of the bubbles in the chart above shows that news audience polarisation is higher than in Norway and Germany, but lower than in the USA where there is no large outlet in the middle of the map, highlighting the importance of public service media as a central anchor in much of Europe (though local news media, which we do not plot here, may partly fulfil this role in the USA). News audiences are less polarised in Norway, but while most traditional newspaper and broadcast brands have mixed and/or centrist audiences, they have recently been joined by several partisan/alternative brands – like Document.no and HRS – that have audiences with a stronger political skew – occupying space further from the centre of the map. In Germany, although some new digital-born brands have emerged in recent years, audiences for most outlets are predominantly centrist. This is partly because fewer people self-identify with either the left or the right in Germany, precluding very high levels of news audience polarisation along this dimension (though, of course, both the public and news audiences may be polarised along different dimensions).\n\nWe can compute a news audience polarisation score for each country – based on the standard deviation of outlet audiences, weighted by their size – that reflects the degree to which polarisation approaches a theoretical maximum, where only outlets with an entirely left-leaning and entirely right-leaning audiences exist, with nothing in between. This allows us to more easily compare across countries, but it also highlights that in most countries news audience polarisation is low. In Germany, for example, polarisation is just 10% of what it could be, and even in the USA (34%), polarisation is far from the theoretical limit.\n\n**IS NEWS AUDIENCE POLARISATION ON THE RISE?**\n\nTo explore whether news audience polarisation is growing, we can compare the most recent data with those from 2016 – the first point at which we started asking respondents whether they self-identify politically with the left or the right outside of the UK and the USA.\n\nAt least in the four countries we focus on here, news audience polarisation has not increased substantially – if at all – in the last six years. The most striking aspect of the maps is how little they have changed over time. Across all four countries, news audience polarisation has changed by 3pp or less since 2016, indicating only minimal shifts in audience behaviour.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 8\n\n0%\n\n25%\n\n50%\n\n75%\n\n100%\n\n34\n\n25\n\n15 10\n\nGermany Norway UK USA\n\nEXTENT OF CROSS-PLATFORM NEWS AUDIENCE POLARISATION – SELECTED COUNTRIES\n\n**Q1F.** Some people talk about ‘left’, ‘right’, and ‘centre’ to describe parties and politicians. With this in mind, where would you place yourself on the following scale? **Q5A/B.** Which of the following brands have you used to access news offline/online in the last week? _Base: Germany = 2002, Norway = 2010, UK = 2410, USA = 2036._\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 7\n\nCROSS-PLATFORM NEWS AUDIENCE POLARISATION – SELECTED COUNTRIES\n\nVery right-leaning audience Average political leaning of population Very lef-leaning audience\n\nARD News Bild ZEIT\n\nVery right-leaning audience Average political leaning of population Very lef-leaning audience\n\nGuardian Daily Mail BBC News\n\nVery right-leaning audience Average political leaning of population Very lef-leaning audience\n\nDocument.no NRK Dagsavisen\n\nVery right-leaning audience Average political leaning of population Very lef-leaning audience\n\nFox News New York Times CNN\n\n**Q1F.** Some people talk about ‘left’, ‘right’, and ‘centre’ to describe parties and politicians. With this in mind, where would you place yourself on the following scale? **Q5A/B.** Which of the following brands have you used to access news offline/online in the last week? _Base: Germany = 2002, Norway = 2010, UK = 2410, USA = 2036._\n\n**UK**\n\n**Germany**\n\n**USA**\n\n**Norway**\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 39","publish_date":null,"title":"Graph 7 Graph 8","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Digital_News-Report_2022_p39.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Digital_News-Report_2022_p39.pdf"},"request":{"example_id":"chart/Digital_News-Report_2022_p39","product_type":"parse","source_file_path":"Digital_News-Report_2022_p39.pdf"},"started_at":"2026-11-04T19:46:25.455736"} +{"completed_at":"2026-11-04T19:46:29.426460","latency_in_ms":2284,"output":{"example_id":"chart/Digital_News-Report_2022_p41","grounded_pages":[],"layout_pages":[],"markdown":"**PERCEPTIONS OF NEWS OUTLET POLARISATION**\n\nTo complement our data on news audience behaviour, we also asked a new question this year on the extent to which people perceive the news landscape in their country to be polarised. Perceived polarisation is an established concept in political science, referring to the extent to which people believe political groups or parties to be different from one another – regardless of the actual similarities or differences in their policies. As such, we asked respondents: ‘In your view, how politically close together or far apart are the main news organisations in your country?’ with response options ranging from ‘Very far apart’ to ‘Very close together’. It is important to keep in mind that polarisation (which is normally seen as bad) shares some qualities (e.g., absence of homogeneity) with concepts like diversity and plurality (which are normally thought of as good). However, given that these typically refer to the extent to which all views are represented equally rather than the distance between groups, the question wording gets us closer to polarisation.\n\nPerceived news outlet polarisation is highest in Poland (54%), Spain (49%), Thailand (48%), and Argentina (47%) where around half think that the main news organisations are politically quite far apart or very far apart. The lowest figures are in South Korea (15%), Portugal (16%), and Singapore (16%). At the regional level, perceived polarisation is, on average, lower in Northern Europe but higher in Southern Europe and Latin America.\n\nTo some extent, these patterns map onto what we might expect based on rules about impartiality, levels of press freedom, and political parallelism – the degree to which the media outlets map onto political parties. And at least in the four countries we focus on in this chapter, levels of news audience polarisation match levels of perceived polarisation, with people in the UK and the US perceiving more polarisation than those in Germany and Norway. But there are also many reasons to expect people’s perceptions to diverge from our expectations. For example, we might infer from the hostile media phenomenon that people who believe the media to be biased against people like them might see news organisations as ‘all the same’ and thus close together – even if the media landscape itself is polarised. At the same time, people may believe there is more polarisation than there is because this is a narrative which has been reinforced by politicians and the media. In this sense, perceived polarisation is similar to trust in the news, whereby perceptions of trust and actual trustworthiness may be quite different, but people’s perceptions are nonetheless important to understand because they can have real consequences.\n\n**CONCLUSIONS**\n\nIn this chapter we have described how, perhaps contrary to expectations, news audience polarisation is often relatively low (especially outside of the USA), is broadly stable over time, and varies significantly by country. In many countries, new partisan/ alternative brands have entered the market, but because their audiences are typically small and because most people still gravitate towards established outlets, their influence on aggregate patterns of audience behaviour is minimal – and in some cases is offset by increasingly mixed/centrist audiences for larger brands.\n\nAt the same time, we have seen how audiences for some individual outlets can change, even in a relatively short space of time. News organisations, many of which have become more audience-focused in recent years, may find it useful to monitor the changing left– right profile of their audience, but this is just one of many possible dimensions along which they can map their users – both political and non-political. The challenge is identifying the most useful, and thinking through how audience profile relates to business and editorial strategy and the changing ways people arrive at and access news – and where this leaves them positioned in the national news landscape.\n\nStepping back, the overall picture we describe may offer some reassurance to those worried about the effects of digital media on society. But it is important to also keep in mind that, although patterns of audience behaviour may appear stable in the short term, the pattern over several decades may be very different. Furthermore, stability at the level of the mass public can also mask increasing polarisation among partisans. All this aside, even if audience behaviour changes little, news content from the same outlets may simultaneously be more partisan. And although smaller, more partisan outlets may have minimal direct reach, they may still exert indirect influence on political elites and media coverage. Audience behaviour certainly matters, but it is only one piece of a much larger puzzle.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 11\n\nPROPORTION WHO THINK NEWS ORGANISATIONS IN THEIR MARKET ARE POLITICALLY FAR APART – SELECTED MARKETS\n\n**Q\\_Polarisation.** In your view, how politically close together or far apart are the main news organisations in your country? _Base: Total sample in each market (n ≈ 2000)._\n\n0% 25% 50% 75% 100%\n\nThailand Taiwan Hong Kong India Philippines Australia Japan Malaysia Indonesia Singapore South Korea\n\n**Asia- Pacific**\n\n48\n\n38 36\n\n33 32\n\n29\n\n26\n\n23\n\n18 16 15\n\nSpain Italy Greece Turkey Croatia Portugal\n\n**Southern Europe**\n\n49\n\n42\n\n39 38\n\n22\n\n16\n\nAustria Netherlands France Switzerland Belgium Germany\n\n**Western Europe**\n\n36\n\n31 31 30\n\n27 26\n\nPoland 54\n\nHungary Czech Republic Slovakia Romania Bulgaria\n\n**Eastern Europe** 44\n\n28 27\n\n24\n\n20\n\nUK Sweden Norway Denmark Finland Ireland\n\n**Northern Europe**\n\n37\n\n30\n\n24 24 22 21\n\nUSA Canada **North America**\n\n41\n\n24\n\nArgentina Peru Mexico Chile Colombia Brazil\n\n**Latin America**\n\n47\n\n43\n\n37\n\n32 32\n\n23\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 41","pages":[{"line_numbers":[],"markdown":"**PERCEPTIONS OF NEWS OUTLET POLARISATION**\n\nTo complement our data on news audience behaviour, we also asked a new question this year on the extent to which people perceive the news landscape in their country to be polarised. Perceived polarisation is an established concept in political science, referring to the extent to which people believe political groups or parties to be different from one another – regardless of the actual similarities or differences in their policies. As such, we asked respondents: ‘In your view, how politically close together or far apart are the main news organisations in your country?’ with response options ranging from ‘Very far apart’ to ‘Very close together’. It is important to keep in mind that polarisation (which is normally seen as bad) shares some qualities (e.g., absence of homogeneity) with concepts like diversity and plurality (which are normally thought of as good). However, given that these typically refer to the extent to which all views are represented equally rather than the distance between groups, the question wording gets us closer to polarisation.\n\nPerceived news outlet polarisation is highest in Poland (54%), Spain (49%), Thailand (48%), and Argentina (47%) where around half think that the main news organisations are politically quite far apart or very far apart. The lowest figures are in South Korea (15%), Portugal (16%), and Singapore (16%). At the regional level, perceived polarisation is, on average, lower in Northern Europe but higher in Southern Europe and Latin America.\n\nTo some extent, these patterns map onto what we might expect based on rules about impartiality, levels of press freedom, and political parallelism – the degree to which the media outlets map onto political parties. And at least in the four countries we focus on in this chapter, levels of news audience polarisation match levels of perceived polarisation, with people in the UK and the US perceiving more polarisation than those in Germany and Norway. But there are also many reasons to expect people’s perceptions to diverge from our expectations. For example, we might infer from the hostile media phenomenon that people who believe the media to be biased against people like them might see news organisations as ‘all the same’ and thus close together – even if the media landscape itself is polarised. At the same time, people may believe there is more polarisation than there is because this is a narrative which has been reinforced by politicians and the media. In this sense, perceived polarisation is similar to trust in the news, whereby perceptions of trust and actual trustworthiness may be quite different, but people’s perceptions are nonetheless important to understand because they can have real consequences.\n\n**CONCLUSIONS**\n\nIn this chapter we have described how, perhaps contrary to expectations, news audience polarisation is often relatively low (especially outside of the USA), is broadly stable over time, and varies significantly by country. In many countries, new partisan/ alternative brands have entered the market, but because their audiences are typically small and because most people still gravitate towards established outlets, their influence on aggregate patterns of audience behaviour is minimal – and in some cases is offset by increasingly mixed/centrist audiences for larger brands.\n\nAt the same time, we have seen how audiences for some individual outlets can change, even in a relatively short space of time. News organisations, many of which have become more audience-focused in recent years, may find it useful to monitor the changing left– right profile of their audience, but this is just one of many possible dimensions along which they can map their users – both political and non-political. The challenge is identifying the most useful, and thinking through how audience profile relates to business and editorial strategy and the changing ways people arrive at and access news – and where this leaves them positioned in the national news landscape.\n\nStepping back, the overall picture we describe may offer some reassurance to those worried about the effects of digital media on society. But it is important to also keep in mind that, although patterns of audience behaviour may appear stable in the short term, the pattern over several decades may be very different. Furthermore, stability at the level of the mass public can also mask increasing polarisation among partisans. All this aside, even if audience behaviour changes little, news content from the same outlets may simultaneously be more partisan. And although smaller, more partisan outlets may have minimal direct reach, they may still exert indirect influence on political elites and media coverage. Audience behaviour certainly matters, but it is only one piece of a much larger puzzle.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 11\n\nPROPORTION WHO THINK NEWS ORGANISATIONS IN THEIR MARKET ARE POLITICALLY FAR APART – SELECTED MARKETS\n\n**Q\\_Polarisation.** In your view, how politically close together or far apart are the main news organisations in your country? _Base: Total sample in each market (n ≈ 2000)._\n\n0% 25% 50% 75% 100%\n\nThailand Taiwan Hong Kong India Philippines Australia Japan Malaysia Indonesia Singapore South Korea\n\n**Asia- Pacific**\n\n48\n\n38 36\n\n33 32\n\n29\n\n26\n\n23\n\n18 16 15\n\nSpain Italy Greece Turkey Croatia Portugal\n\n**Southern Europe**\n\n49\n\n42\n\n39 38\n\n22\n\n16\n\nAustria Netherlands France Switzerland Belgium Germany\n\n**Western Europe**\n\n36\n\n31 31 30\n\n27 26\n\nPoland 54\n\nHungary Czech Republic Slovakia Romania Bulgaria\n\n**Eastern Europe** 44\n\n28 27\n\n24\n\n20\n\nUK Sweden Norway Denmark Finland Ireland\n\n**Northern Europe**\n\n37\n\n30\n\n24 24 22 21\n\nUSA Canada **North America**\n\n41\n\n24\n\nArgentina Peru Mexico Chile Colombia Brazil\n\n**Latin America**\n\n47\n\n43\n\n37\n\n32 32\n\n23\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 41","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**PERCEPTIONS OF NEWS OUTLET POLARISATION**\n\nTo complement our data on news audience behaviour, we also asked a new question this year on the extent to which people perceive the news landscape in their country to be polarised. Perceived polarisation is an established concept in political science, referring to the extent to which people believe political groups or parties to be different from one another – regardless of the actual similarities or differences in their policies. As such, we asked respondents: ‘In your view, how politically close together or far apart are the main news organisations in your country?’ with response options ranging from ‘Very far apart’ to ‘Very close together’. It is important to keep in mind that polarisation (which is normally seen as bad) shares some qualities (e.g., absence of homogeneity) with concepts like diversity and plurality (which are normally thought of as good). However, given that these typically refer to the extent to which all views are represented equally rather than the distance between groups, the question wording gets us closer to polarisation.\n\nPerceived news outlet polarisation is highest in Poland (54%), Spain (49%), Thailand (48%), and Argentina (47%) where around half think that the main news organisations are politically quite far apart or very far apart. The lowest figures are in South Korea (15%), Portugal (16%), and Singapore (16%). At the regional level, perceived polarisation is, on average, lower in Northern Europe but higher in Southern Europe and Latin America.\n\nTo some extent, these patterns map onto what we might expect based on rules about impartiality, levels of press freedom, and political parallelism – the degree to which the media outlets map onto political parties. And at least in the four countries we focus on in this chapter, levels of news audience polarisation match levels of perceived polarisation, with people in the UK and the US perceiving more polarisation than those in Germany and Norway. But there are also many reasons to expect people’s perceptions to diverge from our expectations. For example, we might infer from the hostile media phenomenon that people who believe the media to be biased against people like them might see news organisations as ‘all the same’ and thus close together – even if the media landscape itself is polarised. At the same time, people may believe there is more polarisation than there is because this is a narrative which has been reinforced by politicians and the media. In this sense, perceived polarisation is similar to trust in the news, whereby perceptions of trust and actual trustworthiness may be quite different, but people’s perceptions are nonetheless important to understand because they can have real consequences.\n\n**CONCLUSIONS**\n\nIn this chapter we have described how, perhaps contrary to expectations, news audience polarisation is often relatively low (especially outside of the USA), is broadly stable over time, and varies significantly by country. In many countries, new partisan/ alternative brands have entered the market, but because their audiences are typically small and because most people still gravitate towards established outlets, their influence on aggregate patterns of audience behaviour is minimal – and in some cases is offset by increasingly mixed/centrist audiences for larger brands.\n\nAt the same time, we have seen how audiences for some individual outlets can change, even in a relatively short space of time. News organisations, many of which have become more audience-focused in recent years, may find it useful to monitor the changing left– right profile of their audience, but this is just one of many possible dimensions along which they can map their users – both political and non-political. The challenge is identifying the most useful, and thinking through how audience profile relates to business and editorial strategy and the changing ways people arrive at and access news – and where this leaves them positioned in the national news landscape.\n\nStepping back, the overall picture we describe may offer some reassurance to those worried about the effects of digital media on society. But it is important to also keep in mind that, although patterns of audience behaviour may appear stable in the short term, the pattern over several decades may be very different. Furthermore, stability at the level of the mass public can also mask increasing polarisation among partisans. All this aside, even if audience behaviour changes little, news content from the same outlets may simultaneously be more partisan. And although smaller, more partisan outlets may have minimal direct reach, they may still exert indirect influence on political elites and media coverage. Audience behaviour certainly matters, but it is only one piece of a much larger puzzle.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 11\n\nPROPORTION WHO THINK NEWS ORGANISATIONS IN THEIR MARKET ARE POLITICALLY FAR APART – SELECTED MARKETS\n\n**Q\\_Polarisation.** In your view, how politically close together or far apart are the main news organisations in your country? _Base: Total sample in each market (n ≈ 2000)._\n\n0% 25% 50% 75% 100%\n\nThailand Taiwan Hong Kong India Philippines Australia Japan Malaysia Indonesia Singapore South Korea\n\n**Asia- Pacific**\n\n48\n\n38 36\n\n33 32\n\n29\n\n26\n\n23\n\n18 16 15\n\nSpain Italy Greece Turkey Croatia Portugal\n\n**Southern Europe**\n\n49\n\n42\n\n39 38\n\n22\n\n16\n\nAustria Netherlands France Switzerland Belgium Germany\n\n**Western Europe**\n\n36\n\n31 31 30\n\n27 26\n\nPoland 54\n\nHungary Czech Republic Slovakia Romania Bulgaria\n\n**Eastern Europe** 44\n\n28 27\n\n24\n\n20\n\nUK Sweden Norway Denmark Finland Ireland\n\n**Northern Europe**\n\n37\n\n30\n\n24 24 22 21\n\nUSA Canada **North America**\n\n41\n\n24\n\nArgentina Peru Mexico Chile Colombia Brazil\n\n**Latin America**\n\n47\n\n43\n\n37\n\n32 32\n\n23\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 41","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Digital_News-Report_2022_p41.pdf"]},"response":{"errors":[],"extract_id":"extract_77722788fe19f955d036d2878e2680a6","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**PERCEPTIONS OF NEWS OUTLET POLARISATION**\n\nTo complement our data on news audience behaviour, we also asked a new question this year on the extent to which people perceive the news landscape in their country to be polarised. Perceived polarisation is an established concept in political science, referring to the extent to which people believe political groups or parties to be different from one another – regardless of the actual similarities or differences in their policies. As such, we asked respondents: ‘In your view, how politically close together or far apart are the main news organisations in your country?’ with response options ranging from ‘Very far apart’ to ‘Very close together’. It is important to keep in mind that polarisation (which is normally seen as bad) shares some qualities (e.g., absence of homogeneity) with concepts like diversity and plurality (which are normally thought of as good). However, given that these typically refer to the extent to which all views are represented equally rather than the distance between groups, the question wording gets us closer to polarisation.\n\nPerceived news outlet polarisation is highest in Poland (54%), Spain (49%), Thailand (48%), and Argentina (47%) where around half think that the main news organisations are politically quite far apart or very far apart. The lowest figures are in South Korea (15%), Portugal (16%), and Singapore (16%). At the regional level, perceived polarisation is, on average, lower in Northern Europe but higher in Southern Europe and Latin America.\n\nTo some extent, these patterns map onto what we might expect based on rules about impartiality, levels of press freedom, and political parallelism – the degree to which the media outlets map onto political parties. And at least in the four countries we focus on in this chapter, levels of news audience polarisation match levels of perceived polarisation, with people in the UK and the US perceiving more polarisation than those in Germany and Norway. But there are also many reasons to expect people’s perceptions to diverge from our expectations. For example, we might infer from the hostile media phenomenon that people who believe the media to be biased against people like them might see news organisations as ‘all the same’ and thus close together – even if the media landscape itself is polarised. At the same time, people may believe there is more polarisation than there is because this is a narrative which has been reinforced by politicians and the media. In this sense, perceived polarisation is similar to trust in the news, whereby perceptions of trust and actual trustworthiness may be quite different, but people’s perceptions are nonetheless important to understand because they can have real consequences.\n\n**CONCLUSIONS**\n\nIn this chapter we have described how, perhaps contrary to expectations, news audience polarisation is often relatively low (especially outside of the USA), is broadly stable over time, and varies significantly by country. In many countries, new partisan/ alternative brands have entered the market, but because their audiences are typically small and because most people still gravitate towards established outlets, their influence on aggregate patterns of audience behaviour is minimal – and in some cases is offset by increasingly mixed/centrist audiences for larger brands.\n\nAt the same time, we have seen how audiences for some individual outlets can change, even in a relatively short space of time. News organisations, many of which have become more audience-focused in recent years, may find it useful to monitor the changing left– right profile of their audience, but this is just one of many possible dimensions along which they can map their users – both political and non-political. The challenge is identifying the most useful, and thinking through how audience profile relates to business and editorial strategy and the changing ways people arrive at and access news – and where this leaves them positioned in the national news landscape.\n\nStepping back, the overall picture we describe may offer some reassurance to those worried about the effects of digital media on society. But it is important to also keep in mind that, although patterns of audience behaviour may appear stable in the short term, the pattern over several decades may be very different. Furthermore, stability at the level of the mass public can also mask increasing polarisation among partisans. All this aside, even if audience behaviour changes little, news content from the same outlets may simultaneously be more partisan. And although smaller, more partisan outlets may have minimal direct reach, they may still exert indirect influence on political elites and media coverage. Audience behaviour certainly matters, but it is only one piece of a much larger puzzle.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 11\n\nPROPORTION WHO THINK NEWS ORGANISATIONS IN THEIR MARKET ARE POLITICALLY FAR APART – SELECTED MARKETS\n\n**Q\\_Polarisation.** In your view, how politically close together or far apart are the main news organisations in your country? _Base: Total sample in each market (n ≈ 2000)._\n\n0% 25% 50% 75% 100%\n\nThailand Taiwan Hong Kong India Philippines Australia Japan Malaysia Indonesia Singapore South Korea\n\n**Asia- Pacific**\n\n48\n\n38 36\n\n33 32\n\n29\n\n26\n\n23\n\n18 16 15\n\nSpain Italy Greece Turkey Croatia Portugal\n\n**Southern Europe**\n\n49\n\n42\n\n39 38\n\n22\n\n16\n\nAustria Netherlands France Switzerland Belgium Germany\n\n**Western Europe**\n\n36\n\n31 31 30\n\n27 26\n\nPoland 54\n\nHungary Czech Republic Slovakia Romania Bulgaria\n\n**Eastern Europe** 44\n\n28 27\n\n24\n\n20\n\nUK Sweden Norway Denmark Finland Ireland\n\n**Northern Europe**\n\n37\n\n30\n\n24 24 22 21\n\nUSA Canada **North America**\n\n41\n\n24\n\nArgentina Peru Mexico Chile Colombia Brazil\n\n**Latin America**\n\n47\n\n43\n\n37\n\n32 32\n\n23\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 41"],"full_content":"**PERCEPTIONS OF NEWS OUTLET POLARISATION**\n\nTo complement our data on news audience behaviour, we also asked a new question this year on the extent to which people perceive the news landscape in their country to be polarised. Perceived polarisation is an established concept in political science, referring to the extent to which people believe political groups or parties to be different from one another – regardless of the actual similarities or differences in their policies. As such, we asked respondents: ‘In your view, how politically close together or far apart are the main news organisations in your country?’ with response options ranging from ‘Very far apart’ to ‘Very close together’. It is important to keep in mind that polarisation (which is normally seen as bad) shares some qualities (e.g., absence of homogeneity) with concepts like diversity and plurality (which are normally thought of as good). However, given that these typically refer to the extent to which all views are represented equally rather than the distance between groups, the question wording gets us closer to polarisation.\n\nPerceived news outlet polarisation is highest in Poland (54%), Spain (49%), Thailand (48%), and Argentina (47%) where around half think that the main news organisations are politically quite far apart or very far apart. The lowest figures are in South Korea (15%), Portugal (16%), and Singapore (16%). At the regional level, perceived polarisation is, on average, lower in Northern Europe but higher in Southern Europe and Latin America.\n\nTo some extent, these patterns map onto what we might expect based on rules about impartiality, levels of press freedom, and political parallelism – the degree to which the media outlets map onto political parties. And at least in the four countries we focus on in this chapter, levels of news audience polarisation match levels of perceived polarisation, with people in the UK and the US perceiving more polarisation than those in Germany and Norway. But there are also many reasons to expect people’s perceptions to diverge from our expectations. For example, we might infer from the hostile media phenomenon that people who believe the media to be biased against people like them might see news organisations as ‘all the same’ and thus close together – even if the media landscape itself is polarised. At the same time, people may believe there is more polarisation than there is because this is a narrative which has been reinforced by politicians and the media. In this sense, perceived polarisation is similar to trust in the news, whereby perceptions of trust and actual trustworthiness may be quite different, but people’s perceptions are nonetheless important to understand because they can have real consequences.\n\n**CONCLUSIONS**\n\nIn this chapter we have described how, perhaps contrary to expectations, news audience polarisation is often relatively low (especially outside of the USA), is broadly stable over time, and varies significantly by country. In many countries, new partisan/ alternative brands have entered the market, but because their audiences are typically small and because most people still gravitate towards established outlets, their influence on aggregate patterns of audience behaviour is minimal – and in some cases is offset by increasingly mixed/centrist audiences for larger brands.\n\nAt the same time, we have seen how audiences for some individual outlets can change, even in a relatively short space of time. News organisations, many of which have become more audience-focused in recent years, may find it useful to monitor the changing left– right profile of their audience, but this is just one of many possible dimensions along which they can map their users – both political and non-political. The challenge is identifying the most useful, and thinking through how audience profile relates to business and editorial strategy and the changing ways people arrive at and access news – and where this leaves them positioned in the national news landscape.\n\nStepping back, the overall picture we describe may offer some reassurance to those worried about the effects of digital media on society. But it is important to also keep in mind that, although patterns of audience behaviour may appear stable in the short term, the pattern over several decades may be very different. Furthermore, stability at the level of the mass public can also mask increasing polarisation among partisans. All this aside, even if audience behaviour changes little, news content from the same outlets may simultaneously be more partisan. And although smaller, more partisan outlets may have minimal direct reach, they may still exert indirect influence on political elites and media coverage. Audience behaviour certainly matters, but it is only one piece of a much larger puzzle.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 11\n\nPROPORTION WHO THINK NEWS ORGANISATIONS IN THEIR MARKET ARE POLITICALLY FAR APART – SELECTED MARKETS\n\n**Q\\_Polarisation.** In your view, how politically close together or far apart are the main news organisations in your country? _Base: Total sample in each market (n ≈ 2000)._\n\n0% 25% 50% 75% 100%\n\nThailand Taiwan Hong Kong India Philippines Australia Japan Malaysia Indonesia Singapore South Korea\n\n**Asia- Pacific**\n\n48\n\n38 36\n\n33 32\n\n29\n\n26\n\n23\n\n18 16 15\n\nSpain Italy Greece Turkey Croatia Portugal\n\n**Southern Europe**\n\n49\n\n42\n\n39 38\n\n22\n\n16\n\nAustria Netherlands France Switzerland Belgium Germany\n\n**Western Europe**\n\n36\n\n31 31 30\n\n27 26\n\nPoland 54\n\nHungary Czech Republic Slovakia Romania Bulgaria\n\n**Eastern Europe** 44\n\n28 27\n\n24\n\n20\n\nUK Sweden Norway Denmark Finland Ireland\n\n**Northern Europe**\n\n37\n\n30\n\n24 24 22 21\n\nUSA Canada **North America**\n\n41\n\n24\n\nArgentina Peru Mexico Chile Colombia Brazil\n\n**Latin America**\n\n47\n\n43\n\n37\n\n32 32\n\n23\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 41","publish_date":null,"title":"Graph 11","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Digital_News-Report_2022_p41.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Digital_News-Report_2022_p41.pdf"},"request":{"example_id":"chart/Digital_News-Report_2022_p41","product_type":"parse","source_file_path":"Digital_News-Report_2022_p41.pdf"},"started_at":"2026-11-04T19:46:27.141824"} +{"completed_at":"2026-11-04T19:46:32.118574","latency_in_ms":3536,"output":{"example_id":"chart/Digital_News-Report_2022_p42","grounded_pages":[],"layout_pages":[],"markdown":"# 2\\.3 The Changing News Habits and Attitudes of Younger\n\n# Audiences\n\n### **Kirsten Eddy**\n\nIn 2019, the _Digital News Report_ looked at how young people get their news, ¹⁶ finding stark differences in news consumption and behaviours among younger people, including a greater reliance on digital and social media and a weaker identification with and loyalty to news brands compared with older groups. Three years later, we now turn our attention to how young people’s news habits and attitudes have changed amid rising concerns about news distrust and avoidance, increasing public attention to social issues such as climate change and social justice, and the growth of newer platforms such as TikTok and Telegram.\n\nHere, we aim to unpack these new behaviours as well as to dismantle some broad narratives of ‘young people’. Instead, we consider how _social natives_ (18–24s) – who largely grew up in the world of the social, participatory web – differ meaningfully from _digital natives_ (25–34s) – who largely grew up in the information age but before the rise of social networks – when it comes to news access, formats, and attitudes. ¹⁷ These groups are critical audiences for publishers and journalists around the world, and for the sustainability of the news, but are increasingly hard to reach and may require different strategies to engage them.\n\nIn this chapter, we supplement our survey data with quotes drawn from qualitative research conducted by the market research agency Craft with 72 young people (aged 18–30) in Brazil, the UK, and the US. ¹⁸\n\n**THE ROLE OF SOCIAL MEDIA IN YOUNG PEOPLE’S NEWS BEHAVIOURS**\n\nSince the _Digital News Report_ began tracking respondents’ main source of news, social networks have steadily replaced news websites as a primary source for younger audiences overall, with 39% of social natives (18–24s) across 12 markets now using social media as their main source of news, compared with 34% who prefer to go direct to a news website or app. We also find that social natives are far more likely to access news using ‘side-door’ sources such as social media, aggregator sites, and search engines than older groups.\n\nThe social media landscape continues to evolve dramatically, with new social networks like TikTok entering the field as well as existing platforms like Instagram and Telegram gaining markedly in popularity among young audiences. As social natives shift their attention away from Facebook (or in many cases never really start using it), more visually focused platforms such as Instagram, TikTok, and YouTube have become increasingly popular for news among this group. Use of TikTok for news has increased _fivefold_ among 18–24s across all markets over just three years, from 3% in 2020 to 15% in 2022, while YouTube is increasingly popular among young people in Eastern Europe, Asia-Pacific, and Latin America.\n\n16 https://www.digitalnewsreport.org/survey/2019/how-younger-generations-consume-news-differently/ 17 Because these age ranges are not entirely representative of the generational cohorts of ‘Gen Y’ and ‘Gen Z’, there is no clear consensus on what year of birth separates the two generations,\n\nand to avoid broad generational claims, we instead distinguish digital natives (25–34s) from social natives (18–24s).\n\n18 This work included digital tasks, blogging, and vlogging, and in-depth interviews with 24 participants per country in February–March 2022. Participants represented a range of demographic\n\ntraits, life stages, and news habits.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 12\n\nPROPORTION OF 18–24s WHO SAY EACH IS THEIR MAIN SOURCE OF NEWS – AVERAGE OF 12 MARKETS\n\n**Q4.** You say you’ve used these sources of news in the last week, which would you say is your MAIN source of news? _Base: 18–24s from 2015–22 in selected markets (n = 200). Note: The 12 markets included are UK, USA, Germany, France, Spain, Italy, Ireland, Denmark, Finland, Australia, Brazil, and Japan._\n\nSocial media News websites/apps\n\n0%\n\n25%\n\n50%\n\n2022 2021 2020 2019 2018 2017 2016 2015\n\n0%\n\n25%\n\n50%\n\n2022 2021 2020 2019 2018 2017 2016 2015\n\n39%\n\n34%\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 42","pages":[{"line_numbers":[],"markdown":"# 2\\.3 The Changing News Habits and Attitudes of Younger\n\n# Audiences\n\n### **Kirsten Eddy**\n\nIn 2019, the _Digital News Report_ looked at how young people get their news, ¹⁶ finding stark differences in news consumption and behaviours among younger people, including a greater reliance on digital and social media and a weaker identification with and loyalty to news brands compared with older groups. Three years later, we now turn our attention to how young people’s news habits and attitudes have changed amid rising concerns about news distrust and avoidance, increasing public attention to social issues such as climate change and social justice, and the growth of newer platforms such as TikTok and Telegram.\n\nHere, we aim to unpack these new behaviours as well as to dismantle some broad narratives of ‘young people’. Instead, we consider how _social natives_ (18–24s) – who largely grew up in the world of the social, participatory web – differ meaningfully from _digital natives_ (25–34s) – who largely grew up in the information age but before the rise of social networks – when it comes to news access, formats, and attitudes. ¹⁷ These groups are critical audiences for publishers and journalists around the world, and for the sustainability of the news, but are increasingly hard to reach and may require different strategies to engage them.\n\nIn this chapter, we supplement our survey data with quotes drawn from qualitative research conducted by the market research agency Craft with 72 young people (aged 18–30) in Brazil, the UK, and the US. ¹⁸\n\n**THE ROLE OF SOCIAL MEDIA IN YOUNG PEOPLE’S NEWS BEHAVIOURS**\n\nSince the _Digital News Report_ began tracking respondents’ main source of news, social networks have steadily replaced news websites as a primary source for younger audiences overall, with 39% of social natives (18–24s) across 12 markets now using social media as their main source of news, compared with 34% who prefer to go direct to a news website or app. We also find that social natives are far more likely to access news using ‘side-door’ sources such as social media, aggregator sites, and search engines than older groups.\n\nThe social media landscape continues to evolve dramatically, with new social networks like TikTok entering the field as well as existing platforms like Instagram and Telegram gaining markedly in popularity among young audiences. As social natives shift their attention away from Facebook (or in many cases never really start using it), more visually focused platforms such as Instagram, TikTok, and YouTube have become increasingly popular for news among this group. Use of TikTok for news has increased _fivefold_ among 18–24s across all markets over just three years, from 3% in 2020 to 15% in 2022, while YouTube is increasingly popular among young people in Eastern Europe, Asia-Pacific, and Latin America.\n\n16 https://www.digitalnewsreport.org/survey/2019/how-younger-generations-consume-news-differently/ 17 Because these age ranges are not entirely representative of the generational cohorts of ‘Gen Y’ and ‘Gen Z’, there is no clear consensus on what year of birth separates the two generations,\n\nand to avoid broad generational claims, we instead distinguish digital natives (25–34s) from social natives (18–24s).\n\n18 This work included digital tasks, blogging, and vlogging, and in-depth interviews with 24 participants per country in February–March 2022. Participants represented a range of demographic\n\ntraits, life stages, and news habits.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 12\n\nPROPORTION OF 18–24s WHO SAY EACH IS THEIR MAIN SOURCE OF NEWS – AVERAGE OF 12 MARKETS\n\n**Q4.** You say you’ve used these sources of news in the last week, which would you say is your MAIN source of news? _Base: 18–24s from 2015–22 in selected markets (n = 200). Note: The 12 markets included are UK, USA, Germany, France, Spain, Italy, Ireland, Denmark, Finland, Australia, Brazil, and Japan._\n\nSocial media News websites/apps\n\n0%\n\n25%\n\n50%\n\n2022 2021 2020 2019 2018 2017 2016 2015\n\n0%\n\n25%\n\n50%\n\n2022 2021 2020 2019 2018 2017 2016 2015\n\n39%\n\n34%\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 42","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"# 2\\.3 The Changing News Habits and Attitudes of Younger\n\n# Audiences\n\n### **Kirsten Eddy**\n\nIn 2019, the _Digital News Report_ looked at how young people get their news, ¹⁶ finding stark differences in news consumption and behaviours among younger people, including a greater reliance on digital and social media and a weaker identification with and loyalty to news brands compared with older groups. Three years later, we now turn our attention to how young people’s news habits and attitudes have changed amid rising concerns about news distrust and avoidance, increasing public attention to social issues such as climate change and social justice, and the growth of newer platforms such as TikTok and Telegram.\n\nHere, we aim to unpack these new behaviours as well as to dismantle some broad narratives of ‘young people’. Instead, we consider how _social natives_ (18–24s) – who largely grew up in the world of the social, participatory web – differ meaningfully from _digital natives_ (25–34s) – who largely grew up in the information age but before the rise of social networks – when it comes to news access, formats, and attitudes. ¹⁷ These groups are critical audiences for publishers and journalists around the world, and for the sustainability of the news, but are increasingly hard to reach and may require different strategies to engage them.\n\nIn this chapter, we supplement our survey data with quotes drawn from qualitative research conducted by the market research agency Craft with 72 young people (aged 18–30) in Brazil, the UK, and the US. ¹⁸\n\n**THE ROLE OF SOCIAL MEDIA IN YOUNG PEOPLE’S NEWS BEHAVIOURS**\n\nSince the _Digital News Report_ began tracking respondents’ main source of news, social networks have steadily replaced news websites as a primary source for younger audiences overall, with 39% of social natives (18–24s) across 12 markets now using social media as their main source of news, compared with 34% who prefer to go direct to a news website or app. We also find that social natives are far more likely to access news using ‘side-door’ sources such as social media, aggregator sites, and search engines than older groups.\n\nThe social media landscape continues to evolve dramatically, with new social networks like TikTok entering the field as well as existing platforms like Instagram and Telegram gaining markedly in popularity among young audiences. As social natives shift their attention away from Facebook (or in many cases never really start using it), more visually focused platforms such as Instagram, TikTok, and YouTube have become increasingly popular for news among this group. Use of TikTok for news has increased _fivefold_ among 18–24s across all markets over just three years, from 3% in 2020 to 15% in 2022, while YouTube is increasingly popular among young people in Eastern Europe, Asia-Pacific, and Latin America.\n\n16 https://www.digitalnewsreport.org/survey/2019/how-younger-generations-consume-news-differently/ 17 Because these age ranges are not entirely representative of the generational cohorts of ‘Gen Y’ and ‘Gen Z’, there is no clear consensus on what year of birth separates the two generations,\n\nand to avoid broad generational claims, we instead distinguish digital natives (25–34s) from social natives (18–24s).\n\n18 This work included digital tasks, blogging, and vlogging, and in-depth interviews with 24 participants per country in February–March 2022. Participants represented a range of demographic\n\ntraits, life stages, and news habits.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 12\n\nPROPORTION OF 18–24s WHO SAY EACH IS THEIR MAIN SOURCE OF NEWS – AVERAGE OF 12 MARKETS\n\n**Q4.** You say you’ve used these sources of news in the last week, which would you say is your MAIN source of news? _Base: 18–24s from 2015–22 in selected markets (n = 200). Note: The 12 markets included are UK, USA, Germany, France, Spain, Italy, Ireland, Denmark, Finland, Australia, Brazil, and Japan._\n\nSocial media News websites/apps\n\n0%\n\n25%\n\n50%\n\n2022 2021 2020 2019 2018 2017 2016 2015\n\n0%\n\n25%\n\n50%\n\n2022 2021 2020 2019 2018 2017 2016 2015\n\n39%\n\n34%\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 42","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Digital_News-Report_2022_p42.pdf"]},"response":{"errors":[],"extract_id":"extract_d9df74f384b12daeec7ae3a9bf58ea3d","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["# 2\\.3 The Changing News Habits and Attitudes of Younger\n\n# Audiences\n\n### **Kirsten Eddy**\n\nIn 2019, the _Digital News Report_ looked at how young people get their news, ¹⁶ finding stark differences in news consumption and behaviours among younger people, including a greater reliance on digital and social media and a weaker identification with and loyalty to news brands compared with older groups. Three years later, we now turn our attention to how young people’s news habits and attitudes have changed amid rising concerns about news distrust and avoidance, increasing public attention to social issues such as climate change and social justice, and the growth of newer platforms such as TikTok and Telegram.\n\nHere, we aim to unpack these new behaviours as well as to dismantle some broad narratives of ‘young people’. Instead, we consider how _social natives_ (18–24s) – who largely grew up in the world of the social, participatory web – differ meaningfully from _digital natives_ (25–34s) – who largely grew up in the information age but before the rise of social networks – when it comes to news access, formats, and attitudes. ¹⁷ These groups are critical audiences for publishers and journalists around the world, and for the sustainability of the news, but are increasingly hard to reach and may require different strategies to engage them.\n\nIn this chapter, we supplement our survey data with quotes drawn from qualitative research conducted by the market research agency Craft with 72 young people (aged 18–30) in Brazil, the UK, and the US. ¹⁸\n\n**THE ROLE OF SOCIAL MEDIA IN YOUNG PEOPLE’S NEWS BEHAVIOURS**\n\nSince the _Digital News Report_ began tracking respondents’ main source of news, social networks have steadily replaced news websites as a primary source for younger audiences overall, with 39% of social natives (18–24s) across 12 markets now using social media as their main source of news, compared with 34% who prefer to go direct to a news website or app. We also find that social natives are far more likely to access news using ‘side-door’ sources such as social media, aggregator sites, and search engines than older groups.\n\nThe social media landscape continues to evolve dramatically, with new social networks like TikTok entering the field as well as existing platforms like Instagram and Telegram gaining markedly in popularity among young audiences. As social natives shift their attention away from Facebook (or in many cases never really start using it), more visually focused platforms such as Instagram, TikTok, and YouTube have become increasingly popular for news among this group. Use of TikTok for news has increased _fivefold_ among 18–24s across all markets over just three years, from 3% in 2020 to 15% in 2022, while YouTube is increasingly popular among young people in Eastern Europe, Asia-Pacific, and Latin America.\n\n16 https://www.digitalnewsreport.org/survey/2019/how-younger-generations-consume-news-differently/ 17 Because these age ranges are not entirely representative of the generational cohorts of ‘Gen Y’ and ‘Gen Z’, there is no clear consensus on what year of birth separates the two generations,\n\nand to avoid broad generational claims, we instead distinguish digital natives (25–34s) from social natives (18–24s).\n\n18 This work included digital tasks, blogging, and vlogging, and in-depth interviews with 24 participants per country in February–March 2022. Participants represented a range of demographic\n\ntraits, life stages, and news habits.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 12\n\nPROPORTION OF 18–24s WHO SAY EACH IS THEIR MAIN SOURCE OF NEWS – AVERAGE OF 12 MARKETS\n\n**Q4.** You say you’ve used these sources of news in the last week, which would you say is your MAIN source of news? _Base: 18–24s from 2015–22 in selected markets (n = 200). Note: The 12 markets included are UK, USA, Germany, France, Spain, Italy, Ireland, Denmark, Finland, Australia, Brazil, and Japan._\n\nSocial media News websites/apps\n\n0%\n\n25%\n\n50%\n\n2022 2021 2020 2019 2018 2017 2016 2015\n\n0%\n\n25%\n\n50%\n\n2022 2021 2020 2019 2018 2017 2016 2015\n\n39%\n\n34%\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 42"],"full_content":"# 2\\.3 The Changing News Habits and Attitudes of Younger\n\n# Audiences\n\n### **Kirsten Eddy**\n\nIn 2019, the _Digital News Report_ looked at how young people get their news, ¹⁶ finding stark differences in news consumption and behaviours among younger people, including a greater reliance on digital and social media and a weaker identification with and loyalty to news brands compared with older groups. Three years later, we now turn our attention to how young people’s news habits and attitudes have changed amid rising concerns about news distrust and avoidance, increasing public attention to social issues such as climate change and social justice, and the growth of newer platforms such as TikTok and Telegram.\n\nHere, we aim to unpack these new behaviours as well as to dismantle some broad narratives of ‘young people’. Instead, we consider how _social natives_ (18–24s) – who largely grew up in the world of the social, participatory web – differ meaningfully from _digital natives_ (25–34s) – who largely grew up in the information age but before the rise of social networks – when it comes to news access, formats, and attitudes. ¹⁷ These groups are critical audiences for publishers and journalists around the world, and for the sustainability of the news, but are increasingly hard to reach and may require different strategies to engage them.\n\nIn this chapter, we supplement our survey data with quotes drawn from qualitative research conducted by the market research agency Craft with 72 young people (aged 18–30) in Brazil, the UK, and the US. ¹⁸\n\n**THE ROLE OF SOCIAL MEDIA IN YOUNG PEOPLE’S NEWS BEHAVIOURS**\n\nSince the _Digital News Report_ began tracking respondents’ main source of news, social networks have steadily replaced news websites as a primary source for younger audiences overall, with 39% of social natives (18–24s) across 12 markets now using social media as their main source of news, compared with 34% who prefer to go direct to a news website or app. We also find that social natives are far more likely to access news using ‘side-door’ sources such as social media, aggregator sites, and search engines than older groups.\n\nThe social media landscape continues to evolve dramatically, with new social networks like TikTok entering the field as well as existing platforms like Instagram and Telegram gaining markedly in popularity among young audiences. As social natives shift their attention away from Facebook (or in many cases never really start using it), more visually focused platforms such as Instagram, TikTok, and YouTube have become increasingly popular for news among this group. Use of TikTok for news has increased _fivefold_ among 18–24s across all markets over just three years, from 3% in 2020 to 15% in 2022, while YouTube is increasingly popular among young people in Eastern Europe, Asia-Pacific, and Latin America.\n\n16 https://www.digitalnewsreport.org/survey/2019/how-younger-generations-consume-news-differently/ 17 Because these age ranges are not entirely representative of the generational cohorts of ‘Gen Y’ and ‘Gen Z’, there is no clear consensus on what year of birth separates the two generations,\n\nand to avoid broad generational claims, we instead distinguish digital natives (25–34s) from social natives (18–24s).\n\n18 This work included digital tasks, blogging, and vlogging, and in-depth interviews with 24 participants per country in February–March 2022. Participants represented a range of demographic\n\ntraits, life stages, and news habits.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 12\n\nPROPORTION OF 18–24s WHO SAY EACH IS THEIR MAIN SOURCE OF NEWS – AVERAGE OF 12 MARKETS\n\n**Q4.** You say you’ve used these sources of news in the last week, which would you say is your MAIN source of news? _Base: 18–24s from 2015–22 in selected markets (n = 200). Note: The 12 markets included are UK, USA, Germany, France, Spain, Italy, Ireland, Denmark, Finland, Australia, Brazil, and Japan._\n\nSocial media News websites/apps\n\n0%\n\n25%\n\n50%\n\n2022 2021 2020 2019 2018 2017 2016 2015\n\n0%\n\n25%\n\n50%\n\n2022 2021 2020 2019 2018 2017 2016 2015\n\n39%\n\n34%\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 42","publish_date":null,"title":"In 2019, the Digital News Report looked at how","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Digital_News-Report_2022_p42.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Digital_News-Report_2022_p42.pdf"},"request":{"example_id":"chart/Digital_News-Report_2022_p42","product_type":"parse","source_file_path":"Digital_News-Report_2022_p42.pdf"},"started_at":"2026-11-04T19:46:28.581785"} +{"completed_at":"2026-11-04T19:46:30.911181","latency_in_ms":1366,"output":{"example_id":"chart/Digital_News-Report_2022_p43","grounded_pages":[],"layout_pages":[],"markdown":"And while 25–34s have largely embraced many of the same networks as social natives in their daily lives and news habits, they have remained much more loyal to Facebook (9pp higher for news than social natives) – the network this cohort largely grew up with – and have been slower to move to new networks like TikTok (5pp lower for news than social natives).\n\nWhat makes these networks so appealing to some younger audiences? Qualitative interviews reveal that they are drawn to the informal, entertaining style of visual media (and particularly online video) platforms – describing it as more personalised and diverse than TV, as a resource for rapidly changing events such as the Russia–Ukraine conflict, and as a venue for niche interests, from pop culture to travel to health and well-being.\n\n**_On TV we always see the same things, but on YouTube, Spotify, TikTok, we have a range of diversity. … We can get all this and see that there is diversity, society far beyond just what we live._** _Male, 18, Brazil_\n\n**_A TV reporter who also has a TikTok page gives us regular updates … feels comforting and more intimate than watching on TV news._** _Female, 22, UK_\n\n**_I find [YouTube] more digestible than mainstream media. I also find YouTube to be more interesting because it is more specifically geared towards me._** _Male, 23, USA_\n\nHowever, the popularity of online video does not mean text- and audio-based formats don’t still have a big role to play in young people’s news habits. Under 35s still largely say they prefer to mostly read (58%) rather than mostly watch (15%) news – particularly, as our qualitative research finds, when looking for live updates and summaries or when keeping up with what is happening on a ‘need to know’ basis. Some say they seek out a mix of text and video content to better understand information. Others, particularly in Asia-Pacific and Latin American markets, are drawn to audio-based formats like podcasts that allow users to multitask while they listen. There is not a one-size-fits-all approach or medium through which newsrooms can attract younger audiences.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 13\n\n27% 24%\n\n0%\n\n25%\n\n50% Snapchat\n\nTikTok\n\nWhatsApp\n\nInstagram\n\nYouTube\n\nTwitter\n\nFacebook\n\n2022 2021 2020 2019 2018 2017 2016 2015\n\nPROPORTION OF 18–24s WHO USED EACH SOCIAL NETWORK FOR NEWS IN THE PAST WEEK – UK\n\n**Q12B.** Which, if any, of the following have you used for news in the last week? _Base: 18–24s from 2015–22 (n ≈ 200)._\n\nFacebook\n\nSnapchat WhatsApp\n\nYouTube\n\nTikTok\n\nTwitter Instagram\n\n**WHY YOUNGER AUDIENCES AVOID NEWS**\n\nAnd yet, as more news outlets and formats compete for audiences’ time and attention, we continue to see longer-term falls in interest and trust in news across age groups and markets – particularly among younger audiences. Under 35s are the lowest-trusting age groups, with only a third (37%) of both 18–24s and 25–34s across all markets saying they trust most news most of the time, compared with nearly half of those 55 and older (47%). Young people also increasingly choose to avoid the news, with substantial rises in avoidance among social natives since we last asked this question in 2019. Across all markets, around four in ten under 35s often or sometimes avoid the news now, compared with a third (36%) of those 35 and older.\n\nWhy is this happening? Most often, younger audiences (under 35) say the news has a negative effect on their mood (34%) and, most recently, that there is too much news coverage of topics like politics or Coronavirus (39%). In particular, the longstanding criticism of the depressing or overwhelming nature of news persists among young people. For instance, in the UK, _two-thirds_ (64%) of news avoiders under 35 say the news brings down their mood. Our qualitative research participants described forming habits of avoiding this negativity.\n\n**_I tend to try and limit the amount of negative news I consume, especially first thing in the morning and last thing at night. I don’t want it to affect my day or make me worry._** _Female, 29, UK_\n\n**_Depending on my mood, if I see news that I know is bad, is going to upset me, sometimes I leave it and read it later._** _Male, 24, Brazil_\n\nYoung people, particularly digital natives (27%), also at times avoid the news because they perceive it as biased or untrustworthy. As under 35s grew up in the digital age and have been socialised by older generations to be critical of the information they consume, our qualitative research suggests they take a particularly sceptical approach to all information and often question the ‘agenda’ of purveyors of news. In this sense, mainstream news brands are not inherently more valued for impartiality by some young people, and their wariness of bias at times pushes them away from consuming news altogether.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 14\n\nPROPORTION WHO SOMETIMES OR OFTEN ACTIVELY AVOID NEWS – BY AGE (2017–22) – ALL MARKETS\n\n**Q1di\\_2017.** Do you find yourself actively trying to avoid news these days? _Base: 18–24/25–34/35+ in all markets from 2017–22: 2017 = 7991/12,711/51,103, 2019 = 8272/13,361/54,116, 2022 = 11,304/16,862/65,266. Note: 36 markets in 2017, 38 markets in 2019, 46 markets in 2022._\n\n0%\n\n25%\n\n50%\n\n2022 2019 2017\n\n42% 40%\n\n36%\n\n25%\n\n50%\n\n2022 2019 2017\n\n18-24 25-34 35+\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 43","pages":[{"line_numbers":[],"markdown":"And while 25–34s have largely embraced many of the same networks as social natives in their daily lives and news habits, they have remained much more loyal to Facebook (9pp higher for news than social natives) – the network this cohort largely grew up with – and have been slower to move to new networks like TikTok (5pp lower for news than social natives).\n\nWhat makes these networks so appealing to some younger audiences? Qualitative interviews reveal that they are drawn to the informal, entertaining style of visual media (and particularly online video) platforms – describing it as more personalised and diverse than TV, as a resource for rapidly changing events such as the Russia–Ukraine conflict, and as a venue for niche interests, from pop culture to travel to health and well-being.\n\n**_On TV we always see the same things, but on YouTube, Spotify, TikTok, we have a range of diversity. … We can get all this and see that there is diversity, society far beyond just what we live._** _Male, 18, Brazil_\n\n**_A TV reporter who also has a TikTok page gives us regular updates … feels comforting and more intimate than watching on TV news._** _Female, 22, UK_\n\n**_I find [YouTube] more digestible than mainstream media. I also find YouTube to be more interesting because it is more specifically geared towards me._** _Male, 23, USA_\n\nHowever, the popularity of online video does not mean text- and audio-based formats don’t still have a big role to play in young people’s news habits. Under 35s still largely say they prefer to mostly read (58%) rather than mostly watch (15%) news – particularly, as our qualitative research finds, when looking for live updates and summaries or when keeping up with what is happening on a ‘need to know’ basis. Some say they seek out a mix of text and video content to better understand information. Others, particularly in Asia-Pacific and Latin American markets, are drawn to audio-based formats like podcasts that allow users to multitask while they listen. There is not a one-size-fits-all approach or medium through which newsrooms can attract younger audiences.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 13\n\n27% 24%\n\n0%\n\n25%\n\n50% Snapchat\n\nTikTok\n\nWhatsApp\n\nInstagram\n\nYouTube\n\nTwitter\n\nFacebook\n\n2022 2021 2020 2019 2018 2017 2016 2015\n\nPROPORTION OF 18–24s WHO USED EACH SOCIAL NETWORK FOR NEWS IN THE PAST WEEK – UK\n\n**Q12B.** Which, if any, of the following have you used for news in the last week? _Base: 18–24s from 2015–22 (n ≈ 200)._\n\nFacebook\n\nSnapchat WhatsApp\n\nYouTube\n\nTikTok\n\nTwitter Instagram\n\n**WHY YOUNGER AUDIENCES AVOID NEWS**\n\nAnd yet, as more news outlets and formats compete for audiences’ time and attention, we continue to see longer-term falls in interest and trust in news across age groups and markets – particularly among younger audiences. Under 35s are the lowest-trusting age groups, with only a third (37%) of both 18–24s and 25–34s across all markets saying they trust most news most of the time, compared with nearly half of those 55 and older (47%). Young people also increasingly choose to avoid the news, with substantial rises in avoidance among social natives since we last asked this question in 2019. Across all markets, around four in ten under 35s often or sometimes avoid the news now, compared with a third (36%) of those 35 and older.\n\nWhy is this happening? Most often, younger audiences (under 35) say the news has a negative effect on their mood (34%) and, most recently, that there is too much news coverage of topics like politics or Coronavirus (39%). In particular, the longstanding criticism of the depressing or overwhelming nature of news persists among young people. For instance, in the UK, _two-thirds_ (64%) of news avoiders under 35 say the news brings down their mood. Our qualitative research participants described forming habits of avoiding this negativity.\n\n**_I tend to try and limit the amount of negative news I consume, especially first thing in the morning and last thing at night. I don’t want it to affect my day or make me worry._** _Female, 29, UK_\n\n**_Depending on my mood, if I see news that I know is bad, is going to upset me, sometimes I leave it and read it later._** _Male, 24, Brazil_\n\nYoung people, particularly digital natives (27%), also at times avoid the news because they perceive it as biased or untrustworthy. As under 35s grew up in the digital age and have been socialised by older generations to be critical of the information they consume, our qualitative research suggests they take a particularly sceptical approach to all information and often question the ‘agenda’ of purveyors of news. In this sense, mainstream news brands are not inherently more valued for impartiality by some young people, and their wariness of bias at times pushes them away from consuming news altogether.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 14\n\nPROPORTION WHO SOMETIMES OR OFTEN ACTIVELY AVOID NEWS – BY AGE (2017–22) – ALL MARKETS\n\n**Q1di\\_2017.** Do you find yourself actively trying to avoid news these days? _Base: 18–24/25–34/35+ in all markets from 2017–22: 2017 = 7991/12,711/51,103, 2019 = 8272/13,361/54,116, 2022 = 11,304/16,862/65,266. Note: 36 markets in 2017, 38 markets in 2019, 46 markets in 2022._\n\n0%\n\n25%\n\n50%\n\n2022 2019 2017\n\n42% 40%\n\n36%\n\n25%\n\n50%\n\n2022 2019 2017\n\n18-24 25-34 35+\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 43","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"And while 25–34s have largely embraced many of the same networks as social natives in their daily lives and news habits, they have remained much more loyal to Facebook (9pp higher for news than social natives) – the network this cohort largely grew up with – and have been slower to move to new networks like TikTok (5pp lower for news than social natives).\n\nWhat makes these networks so appealing to some younger audiences? Qualitative interviews reveal that they are drawn to the informal, entertaining style of visual media (and particularly online video) platforms – describing it as more personalised and diverse than TV, as a resource for rapidly changing events such as the Russia–Ukraine conflict, and as a venue for niche interests, from pop culture to travel to health and well-being.\n\n**_On TV we always see the same things, but on YouTube, Spotify, TikTok, we have a range of diversity. … We can get all this and see that there is diversity, society far beyond just what we live._** _Male, 18, Brazil_\n\n**_A TV reporter who also has a TikTok page gives us regular updates … feels comforting and more intimate than watching on TV news._** _Female, 22, UK_\n\n**_I find [YouTube] more digestible than mainstream media. I also find YouTube to be more interesting because it is more specifically geared towards me._** _Male, 23, USA_\n\nHowever, the popularity of online video does not mean text- and audio-based formats don’t still have a big role to play in young people’s news habits. Under 35s still largely say they prefer to mostly read (58%) rather than mostly watch (15%) news – particularly, as our qualitative research finds, when looking for live updates and summaries or when keeping up with what is happening on a ‘need to know’ basis. Some say they seek out a mix of text and video content to better understand information. Others, particularly in Asia-Pacific and Latin American markets, are drawn to audio-based formats like podcasts that allow users to multitask while they listen. There is not a one-size-fits-all approach or medium through which newsrooms can attract younger audiences.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 13\n\n27% 24%\n\n0%\n\n25%\n\n50% Snapchat\n\nTikTok\n\nWhatsApp\n\nInstagram\n\nYouTube\n\nTwitter\n\nFacebook\n\n2022 2021 2020 2019 2018 2017 2016 2015\n\nPROPORTION OF 18–24s WHO USED EACH SOCIAL NETWORK FOR NEWS IN THE PAST WEEK – UK\n\n**Q12B.** Which, if any, of the following have you used for news in the last week? _Base: 18–24s from 2015–22 (n ≈ 200)._\n\nFacebook\n\nSnapchat WhatsApp\n\nYouTube\n\nTikTok\n\nTwitter Instagram\n\n**WHY YOUNGER AUDIENCES AVOID NEWS**\n\nAnd yet, as more news outlets and formats compete for audiences’ time and attention, we continue to see longer-term falls in interest and trust in news across age groups and markets – particularly among younger audiences. Under 35s are the lowest-trusting age groups, with only a third (37%) of both 18–24s and 25–34s across all markets saying they trust most news most of the time, compared with nearly half of those 55 and older (47%). Young people also increasingly choose to avoid the news, with substantial rises in avoidance among social natives since we last asked this question in 2019. Across all markets, around four in ten under 35s often or sometimes avoid the news now, compared with a third (36%) of those 35 and older.\n\nWhy is this happening? Most often, younger audiences (under 35) say the news has a negative effect on their mood (34%) and, most recently, that there is too much news coverage of topics like politics or Coronavirus (39%). In particular, the longstanding criticism of the depressing or overwhelming nature of news persists among young people. For instance, in the UK, _two-thirds_ (64%) of news avoiders under 35 say the news brings down their mood. Our qualitative research participants described forming habits of avoiding this negativity.\n\n**_I tend to try and limit the amount of negative news I consume, especially first thing in the morning and last thing at night. I don’t want it to affect my day or make me worry._** _Female, 29, UK_\n\n**_Depending on my mood, if I see news that I know is bad, is going to upset me, sometimes I leave it and read it later._** _Male, 24, Brazil_\n\nYoung people, particularly digital natives (27%), also at times avoid the news because they perceive it as biased or untrustworthy. As under 35s grew up in the digital age and have been socialised by older generations to be critical of the information they consume, our qualitative research suggests they take a particularly sceptical approach to all information and often question the ‘agenda’ of purveyors of news. In this sense, mainstream news brands are not inherently more valued for impartiality by some young people, and their wariness of bias at times pushes them away from consuming news altogether.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 14\n\nPROPORTION WHO SOMETIMES OR OFTEN ACTIVELY AVOID NEWS – BY AGE (2017–22) – ALL MARKETS\n\n**Q1di\\_2017.** Do you find yourself actively trying to avoid news these days? _Base: 18–24/25–34/35+ in all markets from 2017–22: 2017 = 7991/12,711/51,103, 2019 = 8272/13,361/54,116, 2022 = 11,304/16,862/65,266. Note: 36 markets in 2017, 38 markets in 2019, 46 markets in 2022._\n\n0%\n\n25%\n\n50%\n\n2022 2019 2017\n\n42% 40%\n\n36%\n\n25%\n\n50%\n\n2022 2019 2017\n\n18-24 25-34 35+\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 43","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Digital_News-Report_2022_p43.pdf"]},"response":{"errors":[],"extract_id":"extract_2da7ea9e5f4afbb1025f47c2d5f453eb","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["And while 25–34s have largely embraced many of the same networks as social natives in their daily lives and news habits, they have remained much more loyal to Facebook (9pp higher for news than social natives) – the network this cohort largely grew up with – and have been slower to move to new networks like TikTok (5pp lower for news than social natives).\n\nWhat makes these networks so appealing to some younger audiences? Qualitative interviews reveal that they are drawn to the informal, entertaining style of visual media (and particularly online video) platforms – describing it as more personalised and diverse than TV, as a resource for rapidly changing events such as the Russia–Ukraine conflict, and as a venue for niche interests, from pop culture to travel to health and well-being.\n\n**_On TV we always see the same things, but on YouTube, Spotify, TikTok, we have a range of diversity. … We can get all this and see that there is diversity, society far beyond just what we live._** _Male, 18, Brazil_\n\n**_A TV reporter who also has a TikTok page gives us regular updates … feels comforting and more intimate than watching on TV news._** _Female, 22, UK_\n\n**_I find [YouTube] more digestible than mainstream media. I also find YouTube to be more interesting because it is more specifically geared towards me._** _Male, 23, USA_\n\nHowever, the popularity of online video does not mean text- and audio-based formats don’t still have a big role to play in young people’s news habits. Under 35s still largely say they prefer to mostly read (58%) rather than mostly watch (15%) news – particularly, as our qualitative research finds, when looking for live updates and summaries or when keeping up with what is happening on a ‘need to know’ basis. Some say they seek out a mix of text and video content to better understand information. Others, particularly in Asia-Pacific and Latin American markets, are drawn to audio-based formats like podcasts that allow users to multitask while they listen. There is not a one-size-fits-all approach or medium through which newsrooms can attract younger audiences.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 13\n\n27% 24%\n\n0%\n\n25%\n\n50% Snapchat\n\nTikTok\n\nWhatsApp\n\nInstagram\n\nYouTube\n\nTwitter\n\nFacebook\n\n2022 2021 2020 2019 2018 2017 2016 2015\n\nPROPORTION OF 18–24s WHO USED EACH SOCIAL NETWORK FOR NEWS IN THE PAST WEEK – UK\n\n**Q12B.** Which, if any, of the following have you used for news in the last week? _Base: 18–24s from 2015–22 (n ≈ 200)._\n\nFacebook\n\nSnapchat WhatsApp\n\nYouTube\n\nTikTok\n\nTwitter Instagram\n\n**WHY YOUNGER AUDIENCES AVOID NEWS**\n\nAnd yet, as more news outlets and formats compete for audiences’ time and attention, we continue to see longer-term falls in interest and trust in news across age groups and markets – particularly among younger audiences. Under 35s are the lowest-trusting age groups, with only a third (37%) of both 18–24s and 25–34s across all markets saying they trust most news most of the time, compared with nearly half of those 55 and older (47%). Young people also increasingly choose to avoid the news, with substantial rises in avoidance among social natives since we last asked this question in 2019. Across all markets, around four in ten under 35s often or sometimes avoid the news now, compared with a third (36%) of those 35 and older.\n\nWhy is this happening? Most often, younger audiences (under 35) say the news has a negative effect on their mood (34%) and, most recently, that there is too much news coverage of topics like politics or Coronavirus (39%). In particular, the longstanding criticism of the depressing or overwhelming nature of news persists among young people. For instance, in the UK, _two-thirds_ (64%) of news avoiders under 35 say the news brings down their mood. Our qualitative research participants described forming habits of avoiding this negativity.\n\n**_I tend to try and limit the amount of negative news I consume, especially first thing in the morning and last thing at night. I don’t want it to affect my day or make me worry._** _Female, 29, UK_\n\n**_Depending on my mood, if I see news that I know is bad, is going to upset me, sometimes I leave it and read it later._** _Male, 24, Brazil_\n\nYoung people, particularly digital natives (27%), also at times avoid the news because they perceive it as biased or untrustworthy. As under 35s grew up in the digital age and have been socialised by older generations to be critical of the information they consume, our qualitative research suggests they take a particularly sceptical approach to all information and often question the ‘agenda’ of purveyors of news. In this sense, mainstream news brands are not inherently more valued for impartiality by some young people, and their wariness of bias at times pushes them away from consuming news altogether.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 14\n\nPROPORTION WHO SOMETIMES OR OFTEN ACTIVELY AVOID NEWS – BY AGE (2017–22) – ALL MARKETS\n\n**Q1di\\_2017.** Do you find yourself actively trying to avoid news these days? _Base: 18–24/25–34/35+ in all markets from 2017–22: 2017 = 7991/12,711/51,103, 2019 = 8272/13,361/54,116, 2022 = 11,304/16,862/65,266. Note: 36 markets in 2017, 38 markets in 2019, 46 markets in 2022._\n\n0%\n\n25%\n\n50%\n\n2022 2019 2017\n\n42% 40%\n\n36%\n\n25%\n\n50%\n\n2022 2019 2017\n\n18-24 25-34 35+\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 43"],"full_content":"And while 25–34s have largely embraced many of the same networks as social natives in their daily lives and news habits, they have remained much more loyal to Facebook (9pp higher for news than social natives) – the network this cohort largely grew up with – and have been slower to move to new networks like TikTok (5pp lower for news than social natives).\n\nWhat makes these networks so appealing to some younger audiences? Qualitative interviews reveal that they are drawn to the informal, entertaining style of visual media (and particularly online video) platforms – describing it as more personalised and diverse than TV, as a resource for rapidly changing events such as the Russia–Ukraine conflict, and as a venue for niche interests, from pop culture to travel to health and well-being.\n\n**_On TV we always see the same things, but on YouTube, Spotify, TikTok, we have a range of diversity. … We can get all this and see that there is diversity, society far beyond just what we live._** _Male, 18, Brazil_\n\n**_A TV reporter who also has a TikTok page gives us regular updates … feels comforting and more intimate than watching on TV news._** _Female, 22, UK_\n\n**_I find [YouTube] more digestible than mainstream media. I also find YouTube to be more interesting because it is more specifically geared towards me._** _Male, 23, USA_\n\nHowever, the popularity of online video does not mean text- and audio-based formats don’t still have a big role to play in young people’s news habits. Under 35s still largely say they prefer to mostly read (58%) rather than mostly watch (15%) news – particularly, as our qualitative research finds, when looking for live updates and summaries or when keeping up with what is happening on a ‘need to know’ basis. Some say they seek out a mix of text and video content to better understand information. Others, particularly in Asia-Pacific and Latin American markets, are drawn to audio-based formats like podcasts that allow users to multitask while they listen. There is not a one-size-fits-all approach or medium through which newsrooms can attract younger audiences.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 13\n\n27% 24%\n\n0%\n\n25%\n\n50% Snapchat\n\nTikTok\n\nWhatsApp\n\nInstagram\n\nYouTube\n\nTwitter\n\nFacebook\n\n2022 2021 2020 2019 2018 2017 2016 2015\n\nPROPORTION OF 18–24s WHO USED EACH SOCIAL NETWORK FOR NEWS IN THE PAST WEEK – UK\n\n**Q12B.** Which, if any, of the following have you used for news in the last week? _Base: 18–24s from 2015–22 (n ≈ 200)._\n\nFacebook\n\nSnapchat WhatsApp\n\nYouTube\n\nTikTok\n\nTwitter Instagram\n\n**WHY YOUNGER AUDIENCES AVOID NEWS**\n\nAnd yet, as more news outlets and formats compete for audiences’ time and attention, we continue to see longer-term falls in interest and trust in news across age groups and markets – particularly among younger audiences. Under 35s are the lowest-trusting age groups, with only a third (37%) of both 18–24s and 25–34s across all markets saying they trust most news most of the time, compared with nearly half of those 55 and older (47%). Young people also increasingly choose to avoid the news, with substantial rises in avoidance among social natives since we last asked this question in 2019. Across all markets, around four in ten under 35s often or sometimes avoid the news now, compared with a third (36%) of those 35 and older.\n\nWhy is this happening? Most often, younger audiences (under 35) say the news has a negative effect on their mood (34%) and, most recently, that there is too much news coverage of topics like politics or Coronavirus (39%). In particular, the longstanding criticism of the depressing or overwhelming nature of news persists among young people. For instance, in the UK, _two-thirds_ (64%) of news avoiders under 35 say the news brings down their mood. Our qualitative research participants described forming habits of avoiding this negativity.\n\n**_I tend to try and limit the amount of negative news I consume, especially first thing in the morning and last thing at night. I don’t want it to affect my day or make me worry._** _Female, 29, UK_\n\n**_Depending on my mood, if I see news that I know is bad, is going to upset me, sometimes I leave it and read it later._** _Male, 24, Brazil_\n\nYoung people, particularly digital natives (27%), also at times avoid the news because they perceive it as biased or untrustworthy. As under 35s grew up in the digital age and have been socialised by older generations to be critical of the information they consume, our qualitative research suggests they take a particularly sceptical approach to all information and often question the ‘agenda’ of purveyors of news. In this sense, mainstream news brands are not inherently more valued for impartiality by some young people, and their wariness of bias at times pushes them away from consuming news altogether.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 14\n\nPROPORTION WHO SOMETIMES OR OFTEN ACTIVELY AVOID NEWS – BY AGE (2017–22) – ALL MARKETS\n\n**Q1di\\_2017.** Do you find yourself actively trying to avoid news these days? _Base: 18–24/25–34/35+ in all markets from 2017–22: 2017 = 7991/12,711/51,103, 2019 = 8272/13,361/54,116, 2022 = 11,304/16,862/65,266. Note: 36 markets in 2017, 38 markets in 2019, 46 markets in 2022._\n\n0%\n\n25%\n\n50%\n\n2022 2019 2017\n\n42% 40%\n\n36%\n\n25%\n\n50%\n\n2022 2019 2017\n\n18-24 25-34 35+\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 43","publish_date":null,"title":"Graph 13 Graph 14","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Digital_News-Report_2022_p43.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Digital_News-Report_2022_p43.pdf"},"request":{"example_id":"chart/Digital_News-Report_2022_p43","product_type":"parse","source_file_path":"Digital_News-Report_2022_p43.pdf"},"started_at":"2026-11-04T19:46:29.544432"} +{"completed_at":"2026-11-04T19:46:34.275330","latency_in_ms":2912,"output":{"example_id":"chart/Digital_News-Report_2022_p45","grounded_pages":[],"layout_pages":[],"markdown":"**MOTIVATIONS TO ACCESS NEWS**\n\nThis year, we also asked about why people personally choose to keep up with the news. All age groups see the news as equally important for learning new things, but news users under 35 are slightly more motivated than older groups by how entertaining the news is and how sharable it is, and they are slightly less motivated than older groups by a sense of duty to stay informed of news or by its personal usefulness to them.\n\nHowever, the extent to which young people feel a sense of duty to consume news looks very different across countries – for instance, with huge gaps between those under versus over 35 in Brazil and the US but very small gaps by age in France and Japan. And in the UK, a sense of duty to be informed and feeling the news helps them learn new things are tied for the top motivation for consuming news among under 35s. Together with insights from our qualitative research, this suggests that young audiences engage in a sort of mix-and-match of motivations depending on their interests as well as the types of content they are thinking of or seeking out.\n\n**_I access the news partially for fun and partially for entertainment but also to be informed. There are also science magazines, where it’s not fun, but it’s not ‘need to know’ either._** _Female, 23, USA_\n\n**_I am looking for what I need to know to stay relevant with current events … That news is important and relevant to my life. Also, I look for news that is going to entertain my personal interests and fun news that is exciting to discover such as current celebrity news._** _Male, 27, USA_\n\n**CONCLUSIONS**\n\nAs the digital media environment rapidly evolves and more young adults who grew up with social media enter our sample, key differences among younger news audiences continue to crystallise. The point here is not simply that some of young people’s behaviours and preferences are different from those of older people. They long have been. It is that the differences seem to be growing, even between social natives and digital natives. Many of these shifts in behaviours are so fundamental that they appear unlikely to be reversed with time. The youngest cohort represents a more casual, less loyal news user. Social natives’ reliance on social media and weak connection with brands make it harder for media organisations to attract and engage them.\n\nAt the same time, younger audiences are also particularly suspicious and less trusting of _all_ information. This, along with the often-depressing nature of news and the overwhelming amount of information they encounter in their daily lives, makes young people sceptical of news organisations’ agendas and increasingly likely to avoid the news – or at least certain types of news. While young people do not all have the same needs, many are looking for more diverse voices and perspectives and for stories that don’t depress and upset them.\n\nSocial natives in particular have increasingly moved towards new visual social networks – but they are not simply all TikTokers, nor do they all have limited attention spans when it comes to serious information. Young people like a range of formats and media, from text to video to audio, and are drawn to information that is curated for them. There will continue to be a place for text, video, audio, and still imagery – sometimes all in one piece of content. And there will be a place for both the serious, impartial tones of traditional media and more casual, entertaining, or advocacy-centred approaches to covering news.\n\nYounger audiences’ definitions of _what news is_ are also wider. Recognising the variety of preferences and tastes that exist within an incredibly diverse cohort presents a new set of challenges for media organisations. But one route to increased relevance for news brands may lie in broadening their appeal – connecting with the topics young people care about, developing multimedia and platform-specific content, and aligning content and tone with format – rather than entirely replacing what they already do or expecting young people to eventually come around to what has always been done. At times, this includes continuing with what news brands currently offer, some of which is highly valued by younger audiences. In moments when they feel they ‘need to know’ what is happening – as with COVID-19 and the Russia–Ukraine conflict – young people still want news brands to be there.\n\n_A full report on the findings of the qualitative study in the UK, US, and Brazil will be published in September 2022 (in conjunction with Craft)._\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 16\n\nPROPORTION WHO SAY EACH IS WHY THEY KEEP UP WITH THE NEWS – BY AGE – ALL MARKETS\n\n**QMotivations\\_2022a.** Which, if any, are the most important reasons for you, personally, to keep up with the news? Please choose up to three options. _Base: Under 35/35 and older: All markets = 28,166/65,266._\n\n0% 25% 50%\n\n45\n\n48\n\n40\n\n26\n\n17\n\n44\n\n38\n\n34\n\n29\n\n22\n\n35+\n\nU35\n\nI fnd it is ofen entertaining and fun\n\nIt gives me something to talk about and share with others\n\nI have a duty to keep myself informed about the news\n\nIt is important and useful for me personally\n\nIt is a good way for me to learn about new things\n\nU35 35+\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 45","pages":[{"line_numbers":[],"markdown":"**MOTIVATIONS TO ACCESS NEWS**\n\nThis year, we also asked about why people personally choose to keep up with the news. All age groups see the news as equally important for learning new things, but news users under 35 are slightly more motivated than older groups by how entertaining the news is and how sharable it is, and they are slightly less motivated than older groups by a sense of duty to stay informed of news or by its personal usefulness to them.\n\nHowever, the extent to which young people feel a sense of duty to consume news looks very different across countries – for instance, with huge gaps between those under versus over 35 in Brazil and the US but very small gaps by age in France and Japan. And in the UK, a sense of duty to be informed and feeling the news helps them learn new things are tied for the top motivation for consuming news among under 35s. Together with insights from our qualitative research, this suggests that young audiences engage in a sort of mix-and-match of motivations depending on their interests as well as the types of content they are thinking of or seeking out.\n\n**_I access the news partially for fun and partially for entertainment but also to be informed. There are also science magazines, where it’s not fun, but it’s not ‘need to know’ either._** _Female, 23, USA_\n\n**_I am looking for what I need to know to stay relevant with current events … That news is important and relevant to my life. Also, I look for news that is going to entertain my personal interests and fun news that is exciting to discover such as current celebrity news._** _Male, 27, USA_\n\n**CONCLUSIONS**\n\nAs the digital media environment rapidly evolves and more young adults who grew up with social media enter our sample, key differences among younger news audiences continue to crystallise. The point here is not simply that some of young people’s behaviours and preferences are different from those of older people. They long have been. It is that the differences seem to be growing, even between social natives and digital natives. Many of these shifts in behaviours are so fundamental that they appear unlikely to be reversed with time. The youngest cohort represents a more casual, less loyal news user. Social natives’ reliance on social media and weak connection with brands make it harder for media organisations to attract and engage them.\n\nAt the same time, younger audiences are also particularly suspicious and less trusting of _all_ information. This, along with the often-depressing nature of news and the overwhelming amount of information they encounter in their daily lives, makes young people sceptical of news organisations’ agendas and increasingly likely to avoid the news – or at least certain types of news. While young people do not all have the same needs, many are looking for more diverse voices and perspectives and for stories that don’t depress and upset them.\n\nSocial natives in particular have increasingly moved towards new visual social networks – but they are not simply all TikTokers, nor do they all have limited attention spans when it comes to serious information. Young people like a range of formats and media, from text to video to audio, and are drawn to information that is curated for them. There will continue to be a place for text, video, audio, and still imagery – sometimes all in one piece of content. And there will be a place for both the serious, impartial tones of traditional media and more casual, entertaining, or advocacy-centred approaches to covering news.\n\nYounger audiences’ definitions of _what news is_ are also wider. Recognising the variety of preferences and tastes that exist within an incredibly diverse cohort presents a new set of challenges for media organisations. But one route to increased relevance for news brands may lie in broadening their appeal – connecting with the topics young people care about, developing multimedia and platform-specific content, and aligning content and tone with format – rather than entirely replacing what they already do or expecting young people to eventually come around to what has always been done. At times, this includes continuing with what news brands currently offer, some of which is highly valued by younger audiences. In moments when they feel they ‘need to know’ what is happening – as with COVID-19 and the Russia–Ukraine conflict – young people still want news brands to be there.\n\n_A full report on the findings of the qualitative study in the UK, US, and Brazil will be published in September 2022 (in conjunction with Craft)._\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 16\n\nPROPORTION WHO SAY EACH IS WHY THEY KEEP UP WITH THE NEWS – BY AGE – ALL MARKETS\n\n**QMotivations\\_2022a.** Which, if any, are the most important reasons for you, personally, to keep up with the news? Please choose up to three options. _Base: Under 35/35 and older: All markets = 28,166/65,266._\n\n0% 25% 50%\n\n45\n\n48\n\n40\n\n26\n\n17\n\n44\n\n38\n\n34\n\n29\n\n22\n\n35+\n\nU35\n\nI fnd it is ofen entertaining and fun\n\nIt gives me something to talk about and share with others\n\nI have a duty to keep myself informed about the news\n\nIt is important and useful for me personally\n\nIt is a good way for me to learn about new things\n\nU35 35+\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 45","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**MOTIVATIONS TO ACCESS NEWS**\n\nThis year, we also asked about why people personally choose to keep up with the news. All age groups see the news as equally important for learning new things, but news users under 35 are slightly more motivated than older groups by how entertaining the news is and how sharable it is, and they are slightly less motivated than older groups by a sense of duty to stay informed of news or by its personal usefulness to them.\n\nHowever, the extent to which young people feel a sense of duty to consume news looks very different across countries – for instance, with huge gaps between those under versus over 35 in Brazil and the US but very small gaps by age in France and Japan. And in the UK, a sense of duty to be informed and feeling the news helps them learn new things are tied for the top motivation for consuming news among under 35s. Together with insights from our qualitative research, this suggests that young audiences engage in a sort of mix-and-match of motivations depending on their interests as well as the types of content they are thinking of or seeking out.\n\n**_I access the news partially for fun and partially for entertainment but also to be informed. There are also science magazines, where it’s not fun, but it’s not ‘need to know’ either._** _Female, 23, USA_\n\n**_I am looking for what I need to know to stay relevant with current events … That news is important and relevant to my life. Also, I look for news that is going to entertain my personal interests and fun news that is exciting to discover such as current celebrity news._** _Male, 27, USA_\n\n**CONCLUSIONS**\n\nAs the digital media environment rapidly evolves and more young adults who grew up with social media enter our sample, key differences among younger news audiences continue to crystallise. The point here is not simply that some of young people’s behaviours and preferences are different from those of older people. They long have been. It is that the differences seem to be growing, even between social natives and digital natives. Many of these shifts in behaviours are so fundamental that they appear unlikely to be reversed with time. The youngest cohort represents a more casual, less loyal news user. Social natives’ reliance on social media and weak connection with brands make it harder for media organisations to attract and engage them.\n\nAt the same time, younger audiences are also particularly suspicious and less trusting of _all_ information. This, along with the often-depressing nature of news and the overwhelming amount of information they encounter in their daily lives, makes young people sceptical of news organisations’ agendas and increasingly likely to avoid the news – or at least certain types of news. While young people do not all have the same needs, many are looking for more diverse voices and perspectives and for stories that don’t depress and upset them.\n\nSocial natives in particular have increasingly moved towards new visual social networks – but they are not simply all TikTokers, nor do they all have limited attention spans when it comes to serious information. Young people like a range of formats and media, from text to video to audio, and are drawn to information that is curated for them. There will continue to be a place for text, video, audio, and still imagery – sometimes all in one piece of content. And there will be a place for both the serious, impartial tones of traditional media and more casual, entertaining, or advocacy-centred approaches to covering news.\n\nYounger audiences’ definitions of _what news is_ are also wider. Recognising the variety of preferences and tastes that exist within an incredibly diverse cohort presents a new set of challenges for media organisations. But one route to increased relevance for news brands may lie in broadening their appeal – connecting with the topics young people care about, developing multimedia and platform-specific content, and aligning content and tone with format – rather than entirely replacing what they already do or expecting young people to eventually come around to what has always been done. At times, this includes continuing with what news brands currently offer, some of which is highly valued by younger audiences. In moments when they feel they ‘need to know’ what is happening – as with COVID-19 and the Russia–Ukraine conflict – young people still want news brands to be there.\n\n_A full report on the findings of the qualitative study in the UK, US, and Brazil will be published in September 2022 (in conjunction with Craft)._\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 16\n\nPROPORTION WHO SAY EACH IS WHY THEY KEEP UP WITH THE NEWS – BY AGE – ALL MARKETS\n\n**QMotivations\\_2022a.** Which, if any, are the most important reasons for you, personally, to keep up with the news? Please choose up to three options. _Base: Under 35/35 and older: All markets = 28,166/65,266._\n\n0% 25% 50%\n\n45\n\n48\n\n40\n\n26\n\n17\n\n44\n\n38\n\n34\n\n29\n\n22\n\n35+\n\nU35\n\nI fnd it is ofen entertaining and fun\n\nIt gives me something to talk about and share with others\n\nI have a duty to keep myself informed about the news\n\nIt is important and useful for me personally\n\nIt is a good way for me to learn about new things\n\nU35 35+\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 45","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Digital_News-Report_2022_p45.pdf"]},"response":{"errors":[],"extract_id":"extract_4d1cb75c0349ea89e73f21396901e219","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**MOTIVATIONS TO ACCESS NEWS**\n\nThis year, we also asked about why people personally choose to keep up with the news. All age groups see the news as equally important for learning new things, but news users under 35 are slightly more motivated than older groups by how entertaining the news is and how sharable it is, and they are slightly less motivated than older groups by a sense of duty to stay informed of news or by its personal usefulness to them.\n\nHowever, the extent to which young people feel a sense of duty to consume news looks very different across countries – for instance, with huge gaps between those under versus over 35 in Brazil and the US but very small gaps by age in France and Japan. And in the UK, a sense of duty to be informed and feeling the news helps them learn new things are tied for the top motivation for consuming news among under 35s. Together with insights from our qualitative research, this suggests that young audiences engage in a sort of mix-and-match of motivations depending on their interests as well as the types of content they are thinking of or seeking out.\n\n**_I access the news partially for fun and partially for entertainment but also to be informed. There are also science magazines, where it’s not fun, but it’s not ‘need to know’ either._** _Female, 23, USA_\n\n**_I am looking for what I need to know to stay relevant with current events … That news is important and relevant to my life. Also, I look for news that is going to entertain my personal interests and fun news that is exciting to discover such as current celebrity news._** _Male, 27, USA_\n\n**CONCLUSIONS**\n\nAs the digital media environment rapidly evolves and more young adults who grew up with social media enter our sample, key differences among younger news audiences continue to crystallise. The point here is not simply that some of young people’s behaviours and preferences are different from those of older people. They long have been. It is that the differences seem to be growing, even between social natives and digital natives. Many of these shifts in behaviours are so fundamental that they appear unlikely to be reversed with time. The youngest cohort represents a more casual, less loyal news user. Social natives’ reliance on social media and weak connection with brands make it harder for media organisations to attract and engage them.\n\nAt the same time, younger audiences are also particularly suspicious and less trusting of _all_ information. This, along with the often-depressing nature of news and the overwhelming amount of information they encounter in their daily lives, makes young people sceptical of news organisations’ agendas and increasingly likely to avoid the news – or at least certain types of news. While young people do not all have the same needs, many are looking for more diverse voices and perspectives and for stories that don’t depress and upset them.\n\nSocial natives in particular have increasingly moved towards new visual social networks – but they are not simply all TikTokers, nor do they all have limited attention spans when it comes to serious information. Young people like a range of formats and media, from text to video to audio, and are drawn to information that is curated for them. There will continue to be a place for text, video, audio, and still imagery – sometimes all in one piece of content. And there will be a place for both the serious, impartial tones of traditional media and more casual, entertaining, or advocacy-centred approaches to covering news.\n\nYounger audiences’ definitions of _what news is_ are also wider. Recognising the variety of preferences and tastes that exist within an incredibly diverse cohort presents a new set of challenges for media organisations. But one route to increased relevance for news brands may lie in broadening their appeal – connecting with the topics young people care about, developing multimedia and platform-specific content, and aligning content and tone with format – rather than entirely replacing what they already do or expecting young people to eventually come around to what has always been done. At times, this includes continuing with what news brands currently offer, some of which is highly valued by younger audiences. In moments when they feel they ‘need to know’ what is happening – as with COVID-19 and the Russia–Ukraine conflict – young people still want news brands to be there.\n\n_A full report on the findings of the qualitative study in the UK, US, and Brazil will be published in September 2022 (in conjunction with Craft)._\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 16\n\nPROPORTION WHO SAY EACH IS WHY THEY KEEP UP WITH THE NEWS – BY AGE – ALL MARKETS\n\n**QMotivations\\_2022a.** Which, if any, are the most important reasons for you, personally, to keep up with the news? Please choose up to three options. _Base: Under 35/35 and older: All markets = 28,166/65,266._\n\n0% 25% 50%\n\n45\n\n48\n\n40\n\n26\n\n17\n\n44\n\n38\n\n34\n\n29\n\n22\n\n35+\n\nU35\n\nI fnd it is ofen entertaining and fun\n\nIt gives me something to talk about and share with others\n\nI have a duty to keep myself informed about the news\n\nIt is important and useful for me personally\n\nIt is a good way for me to learn about new things\n\nU35 35+\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 45"],"full_content":"**MOTIVATIONS TO ACCESS NEWS**\n\nThis year, we also asked about why people personally choose to keep up with the news. All age groups see the news as equally important for learning new things, but news users under 35 are slightly more motivated than older groups by how entertaining the news is and how sharable it is, and they are slightly less motivated than older groups by a sense of duty to stay informed of news or by its personal usefulness to them.\n\nHowever, the extent to which young people feel a sense of duty to consume news looks very different across countries – for instance, with huge gaps between those under versus over 35 in Brazil and the US but very small gaps by age in France and Japan. And in the UK, a sense of duty to be informed and feeling the news helps them learn new things are tied for the top motivation for consuming news among under 35s. Together with insights from our qualitative research, this suggests that young audiences engage in a sort of mix-and-match of motivations depending on their interests as well as the types of content they are thinking of or seeking out.\n\n**_I access the news partially for fun and partially for entertainment but also to be informed. There are also science magazines, where it’s not fun, but it’s not ‘need to know’ either._** _Female, 23, USA_\n\n**_I am looking for what I need to know to stay relevant with current events … That news is important and relevant to my life. Also, I look for news that is going to entertain my personal interests and fun news that is exciting to discover such as current celebrity news._** _Male, 27, USA_\n\n**CONCLUSIONS**\n\nAs the digital media environment rapidly evolves and more young adults who grew up with social media enter our sample, key differences among younger news audiences continue to crystallise. The point here is not simply that some of young people’s behaviours and preferences are different from those of older people. They long have been. It is that the differences seem to be growing, even between social natives and digital natives. Many of these shifts in behaviours are so fundamental that they appear unlikely to be reversed with time. The youngest cohort represents a more casual, less loyal news user. Social natives’ reliance on social media and weak connection with brands make it harder for media organisations to attract and engage them.\n\nAt the same time, younger audiences are also particularly suspicious and less trusting of _all_ information. This, along with the often-depressing nature of news and the overwhelming amount of information they encounter in their daily lives, makes young people sceptical of news organisations’ agendas and increasingly likely to avoid the news – or at least certain types of news. While young people do not all have the same needs, many are looking for more diverse voices and perspectives and for stories that don’t depress and upset them.\n\nSocial natives in particular have increasingly moved towards new visual social networks – but they are not simply all TikTokers, nor do they all have limited attention spans when it comes to serious information. Young people like a range of formats and media, from text to video to audio, and are drawn to information that is curated for them. There will continue to be a place for text, video, audio, and still imagery – sometimes all in one piece of content. And there will be a place for both the serious, impartial tones of traditional media and more casual, entertaining, or advocacy-centred approaches to covering news.\n\nYounger audiences’ definitions of _what news is_ are also wider. Recognising the variety of preferences and tastes that exist within an incredibly diverse cohort presents a new set of challenges for media organisations. But one route to increased relevance for news brands may lie in broadening their appeal – connecting with the topics young people care about, developing multimedia and platform-specific content, and aligning content and tone with format – rather than entirely replacing what they already do or expecting young people to eventually come around to what has always been done. At times, this includes continuing with what news brands currently offer, some of which is highly valued by younger audiences. In moments when they feel they ‘need to know’ what is happening – as with COVID-19 and the Russia–Ukraine conflict – young people still want news brands to be there.\n\n_A full report on the findings of the qualitative study in the UK, US, and Brazil will be published in September 2022 (in conjunction with Craft)._\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 16\n\nPROPORTION WHO SAY EACH IS WHY THEY KEEP UP WITH THE NEWS – BY AGE – ALL MARKETS\n\n**QMotivations\\_2022a.** Which, if any, are the most important reasons for you, personally, to keep up with the news? Please choose up to three options. _Base: Under 35/35 and older: All markets = 28,166/65,266._\n\n0% 25% 50%\n\n45\n\n48\n\n40\n\n26\n\n17\n\n44\n\n38\n\n34\n\n29\n\n22\n\n35+\n\nU35\n\nI fnd it is ofen entertaining and fun\n\nIt gives me something to talk about and share with others\n\nI have a duty to keep myself informed about the news\n\nIt is important and useful for me personally\n\nIt is a good way for me to learn about new things\n\nU35 35+\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 45","publish_date":null,"title":"Graph 16","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Digital_News-Report_2022_p45.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Digital_News-Report_2022_p45.pdf"},"request":{"example_id":"chart/Digital_News-Report_2022_p45","product_type":"parse","source_file_path":"Digital_News-Report_2022_p45.pdf"},"started_at":"2026-11-04T19:46:31.363076"} +{"completed_at":"2026-11-04T19:46:34.469138","latency_in_ms":1851,"output":{"example_id":"chart/Digital_News-Report_2022_p46","grounded_pages":[],"layout_pages":[],"markdown":"# 2\\.4 Email News: Its Contribution to Engagement and\n\n# Monetisation\n\n### **Nic Newman**\n\nIn the last few years, the emergence of paid newsletter platforms, such as Substack, Revue, and Bulletin, have opened up new opportunities for individuals and small publishers to distribute and monetise content. A number of high-profile columnists have left big news organisations to run newsletter-based businesses, with a few of the most successful earning salaries in the high six figures. ¹⁹\n\nAt the same time, digital-born brands such as Politico and Axios have found that smart, journalist-curated emails have been a key driver of growth for businesses which cover a range of niches from politics to health, technology, sports, and local. ²⁰ Meanwhile, mainstream news organisations have been shifting resources into email production as they try to attract new subscribers, build loyalty with existing users, and introduce more personalisation into their digital products (Jack 2016; Newman et al. 2020). The _New York Times_ , for example, now produces 50 different emails read by 15 million people a week. ²¹\n\nMuch of this recent free and paid newsletter activity has been focused on the United States, but we were keen to know if this had extended elsewhere. More widely, we wanted to understand more about the appeal of newsletters in general. In what ways can this low-tech and often unfashionable medium help build or support sustainable journalism?\n\n**WEEKLY EMAIL CONSUMPTION**\n\nOur _Digital News Report_ data show that email newsletters remain an important channel across countries, with an average of 17% using them weekly. In the United States, 22% use newsletters or email alerts, with almost half of them (10%) saying it is their _main way_ of accessing digital news. Austria (24%), Belgium (23%) and Portugal (22%) also have surprisingly high email usage, but Norway (11%) and the UK (9%) have some of the lowest levels. This may be because their stronger brand connections – with users more likely to go directly to website and apps – mean that publishers feel there is less need to push content to audiences.\n\n19 Substack now has 1m paid subscribers with the top 20 newsletters collectively making more than $20 million a year. More at https://www.vanityfair.com/news/2021/11/substack-paid-\n\nsubscriber-milestone-new-york-times-atlantic\n\n20 https://www.axios.com/newsletters/local 21 https://www.niemanlab.org/2021/08/the-new-york-times-is-making-about-a-third-of-its-newsletters-subscriber-only/\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 17\n\n0% 25%\n\n24\n\n23\n\n22 22\n\n20 20 20 20\n\n19 19\n\n18 18 18\n\n17 17\n\n16 16 16 16 16 16\n\n15 15 15 15 15 15 15 15\n\n14 14 14\n\n13 13 13\n\n12 12\n\n11 11 11\n\n10\n\n9\n\nGo\n\nUK South Korea Norway Finland Croatia Argentina Taiwan Bulgaria Indonesia Japan Hong Kong Sweden Turkey Czech Republic Poland Ireland Spain Italy Denmark Slovakia Malaysia Thailand Singapore Netherlands Peru France Chile Canada Australia Hungary Philippines Mexico Germany Colombia Romania Greece Brazil Switzerland Portugal USA Belgium Austria\n\nPROPORTION WHO ACCESSED NEWS VIA EMAIL IN THE LAST WEEK – SELECTED MARKETS\n\n**Q10.** Thinking about how you got news online (via computer, mobile, or any device) in the last week, which were the ways in which you came across news stories? _Base: Total sample in each market (n ≈ 2000)._\n\n**17%** Average shown across all 42 markets\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 46","pages":[{"line_numbers":[],"markdown":"# 2\\.4 Email News: Its Contribution to Engagement and\n\n# Monetisation\n\n### **Nic Newman**\n\nIn the last few years, the emergence of paid newsletter platforms, such as Substack, Revue, and Bulletin, have opened up new opportunities for individuals and small publishers to distribute and monetise content. A number of high-profile columnists have left big news organisations to run newsletter-based businesses, with a few of the most successful earning salaries in the high six figures. ¹⁹\n\nAt the same time, digital-born brands such as Politico and Axios have found that smart, journalist-curated emails have been a key driver of growth for businesses which cover a range of niches from politics to health, technology, sports, and local. ²⁰ Meanwhile, mainstream news organisations have been shifting resources into email production as they try to attract new subscribers, build loyalty with existing users, and introduce more personalisation into their digital products (Jack 2016; Newman et al. 2020). The _New York Times_ , for example, now produces 50 different emails read by 15 million people a week. ²¹\n\nMuch of this recent free and paid newsletter activity has been focused on the United States, but we were keen to know if this had extended elsewhere. More widely, we wanted to understand more about the appeal of newsletters in general. In what ways can this low-tech and often unfashionable medium help build or support sustainable journalism?\n\n**WEEKLY EMAIL CONSUMPTION**\n\nOur _Digital News Report_ data show that email newsletters remain an important channel across countries, with an average of 17% using them weekly. In the United States, 22% use newsletters or email alerts, with almost half of them (10%) saying it is their _main way_ of accessing digital news. Austria (24%), Belgium (23%) and Portugal (22%) also have surprisingly high email usage, but Norway (11%) and the UK (9%) have some of the lowest levels. This may be because their stronger brand connections – with users more likely to go directly to website and apps – mean that publishers feel there is less need to push content to audiences.\n\n19 Substack now has 1m paid subscribers with the top 20 newsletters collectively making more than $20 million a year. More at https://www.vanityfair.com/news/2021/11/substack-paid-\n\nsubscriber-milestone-new-york-times-atlantic\n\n20 https://www.axios.com/newsletters/local 21 https://www.niemanlab.org/2021/08/the-new-york-times-is-making-about-a-third-of-its-newsletters-subscriber-only/\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 17\n\n0% 25%\n\n24\n\n23\n\n22 22\n\n20 20 20 20\n\n19 19\n\n18 18 18\n\n17 17\n\n16 16 16 16 16 16\n\n15 15 15 15 15 15 15 15\n\n14 14 14\n\n13 13 13\n\n12 12\n\n11 11 11\n\n10\n\n9\n\nGo\n\nUK South Korea Norway Finland Croatia Argentina Taiwan Bulgaria Indonesia Japan Hong Kong Sweden Turkey Czech Republic Poland Ireland Spain Italy Denmark Slovakia Malaysia Thailand Singapore Netherlands Peru France Chile Canada Australia Hungary Philippines Mexico Germany Colombia Romania Greece Brazil Switzerland Portugal USA Belgium Austria\n\nPROPORTION WHO ACCESSED NEWS VIA EMAIL IN THE LAST WEEK – SELECTED MARKETS\n\n**Q10.** Thinking about how you got news online (via computer, mobile, or any device) in the last week, which were the ways in which you came across news stories? _Base: Total sample in each market (n ≈ 2000)._\n\n**17%** Average shown across all 42 markets\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 46","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"# 2\\.4 Email News: Its Contribution to Engagement and\n\n# Monetisation\n\n### **Nic Newman**\n\nIn the last few years, the emergence of paid newsletter platforms, such as Substack, Revue, and Bulletin, have opened up new opportunities for individuals and small publishers to distribute and monetise content. A number of high-profile columnists have left big news organisations to run newsletter-based businesses, with a few of the most successful earning salaries in the high six figures. ¹⁹\n\nAt the same time, digital-born brands such as Politico and Axios have found that smart, journalist-curated emails have been a key driver of growth for businesses which cover a range of niches from politics to health, technology, sports, and local. ²⁰ Meanwhile, mainstream news organisations have been shifting resources into email production as they try to attract new subscribers, build loyalty with existing users, and introduce more personalisation into their digital products (Jack 2016; Newman et al. 2020). The _New York Times_ , for example, now produces 50 different emails read by 15 million people a week. ²¹\n\nMuch of this recent free and paid newsletter activity has been focused on the United States, but we were keen to know if this had extended elsewhere. More widely, we wanted to understand more about the appeal of newsletters in general. In what ways can this low-tech and often unfashionable medium help build or support sustainable journalism?\n\n**WEEKLY EMAIL CONSUMPTION**\n\nOur _Digital News Report_ data show that email newsletters remain an important channel across countries, with an average of 17% using them weekly. In the United States, 22% use newsletters or email alerts, with almost half of them (10%) saying it is their _main way_ of accessing digital news. Austria (24%), Belgium (23%) and Portugal (22%) also have surprisingly high email usage, but Norway (11%) and the UK (9%) have some of the lowest levels. This may be because their stronger brand connections – with users more likely to go directly to website and apps – mean that publishers feel there is less need to push content to audiences.\n\n19 Substack now has 1m paid subscribers with the top 20 newsletters collectively making more than $20 million a year. More at https://www.vanityfair.com/news/2021/11/substack-paid-\n\nsubscriber-milestone-new-york-times-atlantic\n\n20 https://www.axios.com/newsletters/local 21 https://www.niemanlab.org/2021/08/the-new-york-times-is-making-about-a-third-of-its-newsletters-subscriber-only/\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 17\n\n0% 25%\n\n24\n\n23\n\n22 22\n\n20 20 20 20\n\n19 19\n\n18 18 18\n\n17 17\n\n16 16 16 16 16 16\n\n15 15 15 15 15 15 15 15\n\n14 14 14\n\n13 13 13\n\n12 12\n\n11 11 11\n\n10\n\n9\n\nGo\n\nUK South Korea Norway Finland Croatia Argentina Taiwan Bulgaria Indonesia Japan Hong Kong Sweden Turkey Czech Republic Poland Ireland Spain Italy Denmark Slovakia Malaysia Thailand Singapore Netherlands Peru France Chile Canada Australia Hungary Philippines Mexico Germany Colombia Romania Greece Brazil Switzerland Portugal USA Belgium Austria\n\nPROPORTION WHO ACCESSED NEWS VIA EMAIL IN THE LAST WEEK – SELECTED MARKETS\n\n**Q10.** Thinking about how you got news online (via computer, mobile, or any device) in the last week, which were the ways in which you came across news stories? _Base: Total sample in each market (n ≈ 2000)._\n\n**17%** Average shown across all 42 markets\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 46","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Digital_News-Report_2022_p46.pdf"]},"response":{"errors":[],"extract_id":"extract_60147113eb70066f98e2b868214eb97a","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["# 2\\.4 Email News: Its Contribution to Engagement and\n\n# Monetisation\n\n### **Nic Newman**\n\nIn the last few years, the emergence of paid newsletter platforms, such as Substack, Revue, and Bulletin, have opened up new opportunities for individuals and small publishers to distribute and monetise content. A number of high-profile columnists have left big news organisations to run newsletter-based businesses, with a few of the most successful earning salaries in the high six figures. ¹⁹\n\nAt the same time, digital-born brands such as Politico and Axios have found that smart, journalist-curated emails have been a key driver of growth for businesses which cover a range of niches from politics to health, technology, sports, and local. ²⁰ Meanwhile, mainstream news organisations have been shifting resources into email production as they try to attract new subscribers, build loyalty with existing users, and introduce more personalisation into their digital products (Jack 2016; Newman et al. 2020). The _New York Times_ , for example, now produces 50 different emails read by 15 million people a week. ²¹\n\nMuch of this recent free and paid newsletter activity has been focused on the United States, but we were keen to know if this had extended elsewhere. More widely, we wanted to understand more about the appeal of newsletters in general. In what ways can this low-tech and often unfashionable medium help build or support sustainable journalism?\n\n**WEEKLY EMAIL CONSUMPTION**\n\nOur _Digital News Report_ data show that email newsletters remain an important channel across countries, with an average of 17% using them weekly. In the United States, 22% use newsletters or email alerts, with almost half of them (10%) saying it is their _main way_ of accessing digital news. Austria (24%), Belgium (23%) and Portugal (22%) also have surprisingly high email usage, but Norway (11%) and the UK (9%) have some of the lowest levels. This may be because their stronger brand connections – with users more likely to go directly to website and apps – mean that publishers feel there is less need to push content to audiences.\n\n19 Substack now has 1m paid subscribers with the top 20 newsletters collectively making more than $20 million a year. More at https://www.vanityfair.com/news/2021/11/substack-paid-\n\nsubscriber-milestone-new-york-times-atlantic\n\n20 https://www.axios.com/newsletters/local 21 https://www.niemanlab.org/2021/08/the-new-york-times-is-making-about-a-third-of-its-newsletters-subscriber-only/\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 17\n\n0% 25%\n\n24\n\n23\n\n22 22\n\n20 20 20 20\n\n19 19\n\n18 18 18\n\n17 17\n\n16 16 16 16 16 16\n\n15 15 15 15 15 15 15 15\n\n14 14 14\n\n13 13 13\n\n12 12\n\n11 11 11\n\n10\n\n9\n\nGo\n\nUK South Korea Norway Finland Croatia Argentina Taiwan Bulgaria Indonesia Japan Hong Kong Sweden Turkey Czech Republic Poland Ireland Spain Italy Denmark Slovakia Malaysia Thailand Singapore Netherlands Peru France Chile Canada Australia Hungary Philippines Mexico Germany Colombia Romania Greece Brazil Switzerland Portugal USA Belgium Austria\n\nPROPORTION WHO ACCESSED NEWS VIA EMAIL IN THE LAST WEEK – SELECTED MARKETS\n\n**Q10.** Thinking about how you got news online (via computer, mobile, or any device) in the last week, which were the ways in which you came across news stories? _Base: Total sample in each market (n ≈ 2000)._\n\n**17%** Average shown across all 42 markets\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 46"],"full_content":"# 2\\.4 Email News: Its Contribution to Engagement and\n\n# Monetisation\n\n### **Nic Newman**\n\nIn the last few years, the emergence of paid newsletter platforms, such as Substack, Revue, and Bulletin, have opened up new opportunities for individuals and small publishers to distribute and monetise content. A number of high-profile columnists have left big news organisations to run newsletter-based businesses, with a few of the most successful earning salaries in the high six figures. ¹⁹\n\nAt the same time, digital-born brands such as Politico and Axios have found that smart, journalist-curated emails have been a key driver of growth for businesses which cover a range of niches from politics to health, technology, sports, and local. ²⁰ Meanwhile, mainstream news organisations have been shifting resources into email production as they try to attract new subscribers, build loyalty with existing users, and introduce more personalisation into their digital products (Jack 2016; Newman et al. 2020). The _New York Times_ , for example, now produces 50 different emails read by 15 million people a week. ²¹\n\nMuch of this recent free and paid newsletter activity has been focused on the United States, but we were keen to know if this had extended elsewhere. More widely, we wanted to understand more about the appeal of newsletters in general. In what ways can this low-tech and often unfashionable medium help build or support sustainable journalism?\n\n**WEEKLY EMAIL CONSUMPTION**\n\nOur _Digital News Report_ data show that email newsletters remain an important channel across countries, with an average of 17% using them weekly. In the United States, 22% use newsletters or email alerts, with almost half of them (10%) saying it is their _main way_ of accessing digital news. Austria (24%), Belgium (23%) and Portugal (22%) also have surprisingly high email usage, but Norway (11%) and the UK (9%) have some of the lowest levels. This may be because their stronger brand connections – with users more likely to go directly to website and apps – mean that publishers feel there is less need to push content to audiences.\n\n19 Substack now has 1m paid subscribers with the top 20 newsletters collectively making more than $20 million a year. More at https://www.vanityfair.com/news/2021/11/substack-paid-\n\nsubscriber-milestone-new-york-times-atlantic\n\n20 https://www.axios.com/newsletters/local 21 https://www.niemanlab.org/2021/08/the-new-york-times-is-making-about-a-third-of-its-newsletters-subscriber-only/\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 17\n\n0% 25%\n\n24\n\n23\n\n22 22\n\n20 20 20 20\n\n19 19\n\n18 18 18\n\n17 17\n\n16 16 16 16 16 16\n\n15 15 15 15 15 15 15 15\n\n14 14 14\n\n13 13 13\n\n12 12\n\n11 11 11\n\n10\n\n9\n\nGo\n\nUK South Korea Norway Finland Croatia Argentina Taiwan Bulgaria Indonesia Japan Hong Kong Sweden Turkey Czech Republic Poland Ireland Spain Italy Denmark Slovakia Malaysia Thailand Singapore Netherlands Peru France Chile Canada Australia Hungary Philippines Mexico Germany Colombia Romania Greece Brazil Switzerland Portugal USA Belgium Austria\n\nPROPORTION WHO ACCESSED NEWS VIA EMAIL IN THE LAST WEEK – SELECTED MARKETS\n\n**Q10.** Thinking about how you got news online (via computer, mobile, or any device) in the last week, which were the ways in which you came across news stories? _Base: Total sample in each market (n ≈ 2000)._\n\n**17%** Average shown across all 42 markets\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 46","publish_date":null,"title":"2.4 Email News: Its Contribution to Engagement and Monetisation","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Digital_News-Report_2022_p46.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Digital_News-Report_2022_p46.pdf"},"request":{"example_id":"chart/Digital_News-Report_2022_p46","product_type":"parse","source_file_path":"Digital_News-Report_2022_p46.pdf"},"started_at":"2026-11-04T19:46:32.617967"} +{"completed_at":"2026-11-04T19:46:37.051895","latency_in_ms":2577,"output":{"example_id":"chart/Digital_News-Report_2022_p50","grounded_pages":[],"layout_pages":[],"markdown":"# 2\\.5 Which Journalists do People Pay Most Attention to and\n\n# Why? A Study of Six Countries\n\n### **Craig T. Robertson** and **Nic Newman**\n\nIn the pre-internet age, journalists’ careers were inextricably linked with the outlets they worked for. But the rise of social media and other tech platforms has allowed many individual journalists – along with many others, whether activists, creators, influencers, or political figures – the opportunity to break away and build their own profiles independently of any particular news brand. As we heard in the previous chapter, some journalists have used email or subscription platforms to monetise their personal connection with audiences – enabling the most successful few to make significant amounts from subscriptions. Elsewhere, within mainstream media itself, the price of top journalistic talent has risen substantially in an increasingly competitive media landscape – especially for those news anchors, columnists, commentators, and podcasters who are able to attract significant audiences.\n\nBut to what extent do people identify with individual journalists rather than news brands? Do the journalists people say they pay attention to represent individual entrepreneurs, newer digital- born brands, or are they mainly drawn from traditional brands in radio, television, or print? And are the journalists whom people are able to name mainly trading in facts or opinion? To address these questions, for the purposes of this chapter, which combines qualitative and quantitative survey responses, we focus on six countries – the US, UK, Brazil, Germany, France, and Finland – that represent different media landscapes and journalistic traditions across our wider sample.\n\n**NEWS BRANDS OR JOURNALISTS?**\n\nWe asked respondents in our survey whether they paid most attention to specific journalists or to news brands when looking for news online. Looking across our six countries, we see significant differences in terms of the attention paid, with stronger affinity toward news brands in Northern/Western European markets such as Finland, Germany, and the UK, but greater identification with individual journalists or commentators in Brazil and France. The United States is somewhere in the middle, with around a third paying most attention to brands and a fifth to journalists.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 25\n\nPROPORTION WHO PAY MOST ATTENTION TO NEWS BRANDS VERSUS JOURNALISTS – SELECTED MARKETS\n\n**Q\\_Journalists\\_1.** When looking for news online, which of the following do you tend to pay most attention to? _Base: Finland = 2064, Germany = 2002, UK = 2410, USA = 2036, Brazil = 2022, France = 2059._\n\n0%\n\n25%\n\n50%\n\n75%\n\n100% 5 15 19 20 7 24\n\n20\n\n28 26 23\n\n16\n\n26 7\n\n12 11 21\n\n46\n\n29\n\n69\n\n45 43\n\n37\n\n32\n\n21\n\nFrance Brazil USA UK Germany Finland\n\n5\n\nSpecific news brands\n\nDon’t know\n\nSpecific journalists or commentators\n\nI don’t use online news\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 50","pages":[{"line_numbers":[],"markdown":"# 2\\.5 Which Journalists do People Pay Most Attention to and\n\n# Why? A Study of Six Countries\n\n### **Craig T. Robertson** and **Nic Newman**\n\nIn the pre-internet age, journalists’ careers were inextricably linked with the outlets they worked for. But the rise of social media and other tech platforms has allowed many individual journalists – along with many others, whether activists, creators, influencers, or political figures – the opportunity to break away and build their own profiles independently of any particular news brand. As we heard in the previous chapter, some journalists have used email or subscription platforms to monetise their personal connection with audiences – enabling the most successful few to make significant amounts from subscriptions. Elsewhere, within mainstream media itself, the price of top journalistic talent has risen substantially in an increasingly competitive media landscape – especially for those news anchors, columnists, commentators, and podcasters who are able to attract significant audiences.\n\nBut to what extent do people identify with individual journalists rather than news brands? Do the journalists people say they pay attention to represent individual entrepreneurs, newer digital- born brands, or are they mainly drawn from traditional brands in radio, television, or print? And are the journalists whom people are able to name mainly trading in facts or opinion? To address these questions, for the purposes of this chapter, which combines qualitative and quantitative survey responses, we focus on six countries – the US, UK, Brazil, Germany, France, and Finland – that represent different media landscapes and journalistic traditions across our wider sample.\n\n**NEWS BRANDS OR JOURNALISTS?**\n\nWe asked respondents in our survey whether they paid most attention to specific journalists or to news brands when looking for news online. Looking across our six countries, we see significant differences in terms of the attention paid, with stronger affinity toward news brands in Northern/Western European markets such as Finland, Germany, and the UK, but greater identification with individual journalists or commentators in Brazil and France. The United States is somewhere in the middle, with around a third paying most attention to brands and a fifth to journalists.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 25\n\nPROPORTION WHO PAY MOST ATTENTION TO NEWS BRANDS VERSUS JOURNALISTS – SELECTED MARKETS\n\n**Q\\_Journalists\\_1.** When looking for news online, which of the following do you tend to pay most attention to? _Base: Finland = 2064, Germany = 2002, UK = 2410, USA = 2036, Brazil = 2022, France = 2059._\n\n0%\n\n25%\n\n50%\n\n75%\n\n100% 5 15 19 20 7 24\n\n20\n\n28 26 23\n\n16\n\n26 7\n\n12 11 21\n\n46\n\n29\n\n69\n\n45 43\n\n37\n\n32\n\n21\n\nFrance Brazil USA UK Germany Finland\n\n5\n\nSpecific news brands\n\nDon’t know\n\nSpecific journalists or commentators\n\nI don’t use online news\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 50","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"# 2\\.5 Which Journalists do People Pay Most Attention to and\n\n# Why? A Study of Six Countries\n\n### **Craig T. Robertson** and **Nic Newman**\n\nIn the pre-internet age, journalists’ careers were inextricably linked with the outlets they worked for. But the rise of social media and other tech platforms has allowed many individual journalists – along with many others, whether activists, creators, influencers, or political figures – the opportunity to break away and build their own profiles independently of any particular news brand. As we heard in the previous chapter, some journalists have used email or subscription platforms to monetise their personal connection with audiences – enabling the most successful few to make significant amounts from subscriptions. Elsewhere, within mainstream media itself, the price of top journalistic talent has risen substantially in an increasingly competitive media landscape – especially for those news anchors, columnists, commentators, and podcasters who are able to attract significant audiences.\n\nBut to what extent do people identify with individual journalists rather than news brands? Do the journalists people say they pay attention to represent individual entrepreneurs, newer digital- born brands, or are they mainly drawn from traditional brands in radio, television, or print? And are the journalists whom people are able to name mainly trading in facts or opinion? To address these questions, for the purposes of this chapter, which combines qualitative and quantitative survey responses, we focus on six countries – the US, UK, Brazil, Germany, France, and Finland – that represent different media landscapes and journalistic traditions across our wider sample.\n\n**NEWS BRANDS OR JOURNALISTS?**\n\nWe asked respondents in our survey whether they paid most attention to specific journalists or to news brands when looking for news online. Looking across our six countries, we see significant differences in terms of the attention paid, with stronger affinity toward news brands in Northern/Western European markets such as Finland, Germany, and the UK, but greater identification with individual journalists or commentators in Brazil and France. The United States is somewhere in the middle, with around a third paying most attention to brands and a fifth to journalists.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 25\n\nPROPORTION WHO PAY MOST ATTENTION TO NEWS BRANDS VERSUS JOURNALISTS – SELECTED MARKETS\n\n**Q\\_Journalists\\_1.** When looking for news online, which of the following do you tend to pay most attention to? _Base: Finland = 2064, Germany = 2002, UK = 2410, USA = 2036, Brazil = 2022, France = 2059._\n\n0%\n\n25%\n\n50%\n\n75%\n\n100% 5 15 19 20 7 24\n\n20\n\n28 26 23\n\n16\n\n26 7\n\n12 11 21\n\n46\n\n29\n\n69\n\n45 43\n\n37\n\n32\n\n21\n\nFrance Brazil USA UK Germany Finland\n\n5\n\nSpecific news brands\n\nDon’t know\n\nSpecific journalists or commentators\n\nI don’t use online news\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 50","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Digital_News-Report_2022_p50.pdf"]},"response":{"errors":[],"extract_id":"extract_c57df5e76200dff7fe8ac314c357651e","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["# 2\\.5 Which Journalists do People Pay Most Attention to and\n\n# Why? A Study of Six Countries\n\n### **Craig T. Robertson** and **Nic Newman**\n\nIn the pre-internet age, journalists’ careers were inextricably linked with the outlets they worked for. But the rise of social media and other tech platforms has allowed many individual journalists – along with many others, whether activists, creators, influencers, or political figures – the opportunity to break away and build their own profiles independently of any particular news brand. As we heard in the previous chapter, some journalists have used email or subscription platforms to monetise their personal connection with audiences – enabling the most successful few to make significant amounts from subscriptions. Elsewhere, within mainstream media itself, the price of top journalistic talent has risen substantially in an increasingly competitive media landscape – especially for those news anchors, columnists, commentators, and podcasters who are able to attract significant audiences.\n\nBut to what extent do people identify with individual journalists rather than news brands? Do the journalists people say they pay attention to represent individual entrepreneurs, newer digital- born brands, or are they mainly drawn from traditional brands in radio, television, or print? And are the journalists whom people are able to name mainly trading in facts or opinion? To address these questions, for the purposes of this chapter, which combines qualitative and quantitative survey responses, we focus on six countries – the US, UK, Brazil, Germany, France, and Finland – that represent different media landscapes and journalistic traditions across our wider sample.\n\n**NEWS BRANDS OR JOURNALISTS?**\n\nWe asked respondents in our survey whether they paid most attention to specific journalists or to news brands when looking for news online. Looking across our six countries, we see significant differences in terms of the attention paid, with stronger affinity toward news brands in Northern/Western European markets such as Finland, Germany, and the UK, but greater identification with individual journalists or commentators in Brazil and France. The United States is somewhere in the middle, with around a third paying most attention to brands and a fifth to journalists.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 25\n\nPROPORTION WHO PAY MOST ATTENTION TO NEWS BRANDS VERSUS JOURNALISTS – SELECTED MARKETS\n\n**Q\\_Journalists\\_1.** When looking for news online, which of the following do you tend to pay most attention to? _Base: Finland = 2064, Germany = 2002, UK = 2410, USA = 2036, Brazil = 2022, France = 2059._\n\n0%\n\n25%\n\n50%\n\n75%\n\n100% 5 15 19 20 7 24\n\n20\n\n28 26 23\n\n16\n\n26 7\n\n12 11 21\n\n46\n\n29\n\n69\n\n45 43\n\n37\n\n32\n\n21\n\nFrance Brazil USA UK Germany Finland\n\n5\n\nSpecific news brands\n\nDon’t know\n\nSpecific journalists or commentators\n\nI don’t use online news\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 50"],"full_content":"# 2\\.5 Which Journalists do People Pay Most Attention to and\n\n# Why? A Study of Six Countries\n\n### **Craig T. Robertson** and **Nic Newman**\n\nIn the pre-internet age, journalists’ careers were inextricably linked with the outlets they worked for. But the rise of social media and other tech platforms has allowed many individual journalists – along with many others, whether activists, creators, influencers, or political figures – the opportunity to break away and build their own profiles independently of any particular news brand. As we heard in the previous chapter, some journalists have used email or subscription platforms to monetise their personal connection with audiences – enabling the most successful few to make significant amounts from subscriptions. Elsewhere, within mainstream media itself, the price of top journalistic talent has risen substantially in an increasingly competitive media landscape – especially for those news anchors, columnists, commentators, and podcasters who are able to attract significant audiences.\n\nBut to what extent do people identify with individual journalists rather than news brands? Do the journalists people say they pay attention to represent individual entrepreneurs, newer digital- born brands, or are they mainly drawn from traditional brands in radio, television, or print? And are the journalists whom people are able to name mainly trading in facts or opinion? To address these questions, for the purposes of this chapter, which combines qualitative and quantitative survey responses, we focus on six countries – the US, UK, Brazil, Germany, France, and Finland – that represent different media landscapes and journalistic traditions across our wider sample.\n\n**NEWS BRANDS OR JOURNALISTS?**\n\nWe asked respondents in our survey whether they paid most attention to specific journalists or to news brands when looking for news online. Looking across our six countries, we see significant differences in terms of the attention paid, with stronger affinity toward news brands in Northern/Western European markets such as Finland, Germany, and the UK, but greater identification with individual journalists or commentators in Brazil and France. The United States is somewhere in the middle, with around a third paying most attention to brands and a fifth to journalists.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 25\n\nPROPORTION WHO PAY MOST ATTENTION TO NEWS BRANDS VERSUS JOURNALISTS – SELECTED MARKETS\n\n**Q\\_Journalists\\_1.** When looking for news online, which of the following do you tend to pay most attention to? _Base: Finland = 2064, Germany = 2002, UK = 2410, USA = 2036, Brazil = 2022, France = 2059._\n\n0%\n\n25%\n\n50%\n\n75%\n\n100% 5 15 19 20 7 24\n\n20\n\n28 26 23\n\n16\n\n26 7\n\n12 11 21\n\n46\n\n29\n\n69\n\n45 43\n\n37\n\n32\n\n21\n\nFrance Brazil USA UK Germany Finland\n\n5\n\nSpecific news brands\n\nDon’t know\n\nSpecific journalists or commentators\n\nI don’t use online news\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 50","publish_date":null,"title":"2.5 Which Journalists do People Pay Most Attention to and Why? A Study of Six Countries","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Digital_News-Report_2022_p50.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Digital_News-Report_2022_p50.pdf"},"request":{"example_id":"chart/Digital_News-Report_2022_p50","product_type":"parse","source_file_path":"Digital_News-Report_2022_p50.pdf"},"started_at":"2026-11-04T19:46:34.474752"} +{"completed_at":"2026-11-04T19:46:38.003567","latency_in_ms":1799,"output":{"example_id":"chart/Digital_News-Report_2022_p52","grounded_pages":[],"layout_pages":[],"markdown":"**TV DOMINATES IN BRAZIL AND FRANCE BUT NOT SO MUCH IN FINLAND**\n\nSimilar to the US, television journalists are prominent in the most-named lists in Brazil and France. In Brazil, William Bonner, anchor for the country’s popular _Jornal Nacional_ , heads the list, followed by crime TV presenter José Luiz Datena and another TV Globo presenter, Maria Júlia (‘Maju’) Coutinho. The vast majority (86%) of named journalists in Brazil are presenters or anchors from broadcast media, and more than four in ten (41%) work for TV Globo, just one part of Brazil’s biggest news and entertainment company, reflecting the powerful position of this brand. A further fifth (17%) work for other top broadcasters Record TV or Band TV. Just 4% of the journalists mentioned work for any print outlet.\n\nThe story in France is very similar, with, if anything, an even greater focus on TV news presenters. Nine in ten journalists named in France work in broadcast media, primarily at television outlets such as TF1 and France Télévisions – which received 23% and 21% of mentions respectively. Topping the French list are TV presenters Pascal Praud and Anne-Claire Coudray. As in Brazil, journalists from national newspapers like _Le Monde_ and _Le Figaro_ are rarely mentioned, with only 6% of names coming from print media. Neither country sees any significant number of journalists from digital media outlets named (4% in Brazil, 3% in France).\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 27\n\nTHREE MOST-MENTIONED JOURNALISTS AND THEIR ROLES – SELECTED MARKETS\n\nJournalists Brand Role\n\n1 **William Bonner** TV Globo Presenter/Anchor\n\n2 **José Luiz Datena** Band TV Presenter/Anchor\n\n3 **Maria Júlia Coutinho** TV Globo Presenter/Anchor\n\n**BRAZIL**\n\nJournalists Brand Role\n\n1 **Pascal Praud** CNews Presenter/Anchor\n\n2 **Anne-Claire Coudray** TF1 Presenter/Anchor\n\n3 **Jean-Jacques Bourdin** RMC Presenter/Anchor\n\n**FRANCE**\n\nJournalists Brand Role\n\n1 **Marietta Slomka** ZDF Presenter/Anchor\n\n2 **Claus Kleber** ZDF Presenter/Anchor\n\n3 **Peter Kloeppel** RTL Presenter/Anchor\n\n**GERMANY**\n\nJournalists Brand Role\n\n1 **Sanna Ukkola** AlfaTV/ Iltalehti TV Presenter (AlfaTV) and Columnist (Iltalehti)\n\n2 **Matti Rönkä** Yle Presenter/Anchor\n\n3 **Timo Haapala** Ilta-Sanomat Political Correspondent/ Columnist\n\n**FINLAND**\n\n**Q \\_Journalists\\_2.** Please type in the names of up to five journalists that you regularly pay attention to (e.g. news presenters, columnists, or reporters). _Base: Total sample in each market (n ≈ 2000)._\n\nThe picture in Finland, meanwhile, is a little different. Like the UK, there is more focus on journalists from print media, with over a third (37%) of names being columnists or reporters for outlets such as _Iltalehti_ , _Ilta-Sanomat_ , and _Helsingin Sanomat_ . Together, these papers received a third (32%) of mentions, reflecting the continued prominence of print news brands in Finland. Public broadcaster Yle received 35% of mentions, with presenter Matti Rönkä ranking first equal alongside former Yle presenter (and current AlfaTV presenter/ _Iltalehti_ columnist) Sanna Ukkola. Among the most-mentioned brands is also MTV3, with 22% of named journalists working for this outlet. In all, six in ten (60%) journalists named in Finland work for broadcast media – a lower figure than other countries, but still high. Again, journalists working for digital-born media do not feature prominently.\n\nFinally, an in-between country is Germany, where there are somewhat more print journalists (19%) named than in Brazil or France, but fewer than the UK and Finland. Those named are spread across outlets including _Die Welt_ , _Die Zeit_ , and _Der Spiegel_ . The proportion of journalists named from broadcast media is also in the middle of the pack (70%), with this seven-in-ten figure being largely driven by the prominence of ARD, ZDF, and RTL, which make up two-thirds (64%) of mentions. Reflecting the wide popularity of these broadcasters, the most-named journalists are _ZDF heute-journal_ anchors Marietta Slomka and (formerly) Claus Kleber, alongside _RTL aktuell_ host Peter Kloeppel.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 28\n\nPROPORTION OF JOURNALISTS MENTIONED FROM EACH MEDIA TYPE – SELECTED MARKETS\n\n**Q \\_Journalists\\_2.** Please type in the names of up to five journalists that you regularly pay attention to (e.g. news presenters, columnists, or reporters). Base: All who mentioned the name of a journalist: France = 723, Brazil = 1008, USA = 635, Germany = 327, UK = 576, Finland = 431.\n\n0%\n\n25%\n\n50%\n\n75%\n\n100% 10 17 11 8 6\n\n4\n\n9\n\n19 30\n\n37\n\n90\n\n86\n\n74\n\n70\n\n62 60\n\nFinland UK Germany USA Brazil France\n\nBroadcast (TV & Radio) Online/other Print\n\nMost occurring brands by journalists named\n\nFrance Brazil USA Germany UK Finland\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 52","pages":[{"line_numbers":[],"markdown":"**TV DOMINATES IN BRAZIL AND FRANCE BUT NOT SO MUCH IN FINLAND**\n\nSimilar to the US, television journalists are prominent in the most-named lists in Brazil and France. In Brazil, William Bonner, anchor for the country’s popular _Jornal Nacional_ , heads the list, followed by crime TV presenter José Luiz Datena and another TV Globo presenter, Maria Júlia (‘Maju’) Coutinho. The vast majority (86%) of named journalists in Brazil are presenters or anchors from broadcast media, and more than four in ten (41%) work for TV Globo, just one part of Brazil’s biggest news and entertainment company, reflecting the powerful position of this brand. A further fifth (17%) work for other top broadcasters Record TV or Band TV. Just 4% of the journalists mentioned work for any print outlet.\n\nThe story in France is very similar, with, if anything, an even greater focus on TV news presenters. Nine in ten journalists named in France work in broadcast media, primarily at television outlets such as TF1 and France Télévisions – which received 23% and 21% of mentions respectively. Topping the French list are TV presenters Pascal Praud and Anne-Claire Coudray. As in Brazil, journalists from national newspapers like _Le Monde_ and _Le Figaro_ are rarely mentioned, with only 6% of names coming from print media. Neither country sees any significant number of journalists from digital media outlets named (4% in Brazil, 3% in France).\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 27\n\nTHREE MOST-MENTIONED JOURNALISTS AND THEIR ROLES – SELECTED MARKETS\n\nJournalists Brand Role\n\n1 **William Bonner** TV Globo Presenter/Anchor\n\n2 **José Luiz Datena** Band TV Presenter/Anchor\n\n3 **Maria Júlia Coutinho** TV Globo Presenter/Anchor\n\n**BRAZIL**\n\nJournalists Brand Role\n\n1 **Pascal Praud** CNews Presenter/Anchor\n\n2 **Anne-Claire Coudray** TF1 Presenter/Anchor\n\n3 **Jean-Jacques Bourdin** RMC Presenter/Anchor\n\n**FRANCE**\n\nJournalists Brand Role\n\n1 **Marietta Slomka** ZDF Presenter/Anchor\n\n2 **Claus Kleber** ZDF Presenter/Anchor\n\n3 **Peter Kloeppel** RTL Presenter/Anchor\n\n**GERMANY**\n\nJournalists Brand Role\n\n1 **Sanna Ukkola** AlfaTV/ Iltalehti TV Presenter (AlfaTV) and Columnist (Iltalehti)\n\n2 **Matti Rönkä** Yle Presenter/Anchor\n\n3 **Timo Haapala** Ilta-Sanomat Political Correspondent/ Columnist\n\n**FINLAND**\n\n**Q \\_Journalists\\_2.** Please type in the names of up to five journalists that you regularly pay attention to (e.g. news presenters, columnists, or reporters). _Base: Total sample in each market (n ≈ 2000)._\n\nThe picture in Finland, meanwhile, is a little different. Like the UK, there is more focus on journalists from print media, with over a third (37%) of names being columnists or reporters for outlets such as _Iltalehti_ , _Ilta-Sanomat_ , and _Helsingin Sanomat_ . Together, these papers received a third (32%) of mentions, reflecting the continued prominence of print news brands in Finland. Public broadcaster Yle received 35% of mentions, with presenter Matti Rönkä ranking first equal alongside former Yle presenter (and current AlfaTV presenter/ _Iltalehti_ columnist) Sanna Ukkola. Among the most-mentioned brands is also MTV3, with 22% of named journalists working for this outlet. In all, six in ten (60%) journalists named in Finland work for broadcast media – a lower figure than other countries, but still high. Again, journalists working for digital-born media do not feature prominently.\n\nFinally, an in-between country is Germany, where there are somewhat more print journalists (19%) named than in Brazil or France, but fewer than the UK and Finland. Those named are spread across outlets including _Die Welt_ , _Die Zeit_ , and _Der Spiegel_ . The proportion of journalists named from broadcast media is also in the middle of the pack (70%), with this seven-in-ten figure being largely driven by the prominence of ARD, ZDF, and RTL, which make up two-thirds (64%) of mentions. Reflecting the wide popularity of these broadcasters, the most-named journalists are _ZDF heute-journal_ anchors Marietta Slomka and (formerly) Claus Kleber, alongside _RTL aktuell_ host Peter Kloeppel.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 28\n\nPROPORTION OF JOURNALISTS MENTIONED FROM EACH MEDIA TYPE – SELECTED MARKETS\n\n**Q \\_Journalists\\_2.** Please type in the names of up to five journalists that you regularly pay attention to (e.g. news presenters, columnists, or reporters). Base: All who mentioned the name of a journalist: France = 723, Brazil = 1008, USA = 635, Germany = 327, UK = 576, Finland = 431.\n\n0%\n\n25%\n\n50%\n\n75%\n\n100% 10 17 11 8 6\n\n4\n\n9\n\n19 30\n\n37\n\n90\n\n86\n\n74\n\n70\n\n62 60\n\nFinland UK Germany USA Brazil France\n\nBroadcast (TV & Radio) Online/other Print\n\nMost occurring brands by journalists named\n\nFrance Brazil USA Germany UK Finland\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 52","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**TV DOMINATES IN BRAZIL AND FRANCE BUT NOT SO MUCH IN FINLAND**\n\nSimilar to the US, television journalists are prominent in the most-named lists in Brazil and France. In Brazil, William Bonner, anchor for the country’s popular _Jornal Nacional_ , heads the list, followed by crime TV presenter José Luiz Datena and another TV Globo presenter, Maria Júlia (‘Maju’) Coutinho. The vast majority (86%) of named journalists in Brazil are presenters or anchors from broadcast media, and more than four in ten (41%) work for TV Globo, just one part of Brazil’s biggest news and entertainment company, reflecting the powerful position of this brand. A further fifth (17%) work for other top broadcasters Record TV or Band TV. Just 4% of the journalists mentioned work for any print outlet.\n\nThe story in France is very similar, with, if anything, an even greater focus on TV news presenters. Nine in ten journalists named in France work in broadcast media, primarily at television outlets such as TF1 and France Télévisions – which received 23% and 21% of mentions respectively. Topping the French list are TV presenters Pascal Praud and Anne-Claire Coudray. As in Brazil, journalists from national newspapers like _Le Monde_ and _Le Figaro_ are rarely mentioned, with only 6% of names coming from print media. Neither country sees any significant number of journalists from digital media outlets named (4% in Brazil, 3% in France).\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 27\n\nTHREE MOST-MENTIONED JOURNALISTS AND THEIR ROLES – SELECTED MARKETS\n\nJournalists Brand Role\n\n1 **William Bonner** TV Globo Presenter/Anchor\n\n2 **José Luiz Datena** Band TV Presenter/Anchor\n\n3 **Maria Júlia Coutinho** TV Globo Presenter/Anchor\n\n**BRAZIL**\n\nJournalists Brand Role\n\n1 **Pascal Praud** CNews Presenter/Anchor\n\n2 **Anne-Claire Coudray** TF1 Presenter/Anchor\n\n3 **Jean-Jacques Bourdin** RMC Presenter/Anchor\n\n**FRANCE**\n\nJournalists Brand Role\n\n1 **Marietta Slomka** ZDF Presenter/Anchor\n\n2 **Claus Kleber** ZDF Presenter/Anchor\n\n3 **Peter Kloeppel** RTL Presenter/Anchor\n\n**GERMANY**\n\nJournalists Brand Role\n\n1 **Sanna Ukkola** AlfaTV/ Iltalehti TV Presenter (AlfaTV) and Columnist (Iltalehti)\n\n2 **Matti Rönkä** Yle Presenter/Anchor\n\n3 **Timo Haapala** Ilta-Sanomat Political Correspondent/ Columnist\n\n**FINLAND**\n\n**Q \\_Journalists\\_2.** Please type in the names of up to five journalists that you regularly pay attention to (e.g. news presenters, columnists, or reporters). _Base: Total sample in each market (n ≈ 2000)._\n\nThe picture in Finland, meanwhile, is a little different. Like the UK, there is more focus on journalists from print media, with over a third (37%) of names being columnists or reporters for outlets such as _Iltalehti_ , _Ilta-Sanomat_ , and _Helsingin Sanomat_ . Together, these papers received a third (32%) of mentions, reflecting the continued prominence of print news brands in Finland. Public broadcaster Yle received 35% of mentions, with presenter Matti Rönkä ranking first equal alongside former Yle presenter (and current AlfaTV presenter/ _Iltalehti_ columnist) Sanna Ukkola. Among the most-mentioned brands is also MTV3, with 22% of named journalists working for this outlet. In all, six in ten (60%) journalists named in Finland work for broadcast media – a lower figure than other countries, but still high. Again, journalists working for digital-born media do not feature prominently.\n\nFinally, an in-between country is Germany, where there are somewhat more print journalists (19%) named than in Brazil or France, but fewer than the UK and Finland. Those named are spread across outlets including _Die Welt_ , _Die Zeit_ , and _Der Spiegel_ . The proportion of journalists named from broadcast media is also in the middle of the pack (70%), with this seven-in-ten figure being largely driven by the prominence of ARD, ZDF, and RTL, which make up two-thirds (64%) of mentions. Reflecting the wide popularity of these broadcasters, the most-named journalists are _ZDF heute-journal_ anchors Marietta Slomka and (formerly) Claus Kleber, alongside _RTL aktuell_ host Peter Kloeppel.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 28\n\nPROPORTION OF JOURNALISTS MENTIONED FROM EACH MEDIA TYPE – SELECTED MARKETS\n\n**Q \\_Journalists\\_2.** Please type in the names of up to five journalists that you regularly pay attention to (e.g. news presenters, columnists, or reporters). Base: All who mentioned the name of a journalist: France = 723, Brazil = 1008, USA = 635, Germany = 327, UK = 576, Finland = 431.\n\n0%\n\n25%\n\n50%\n\n75%\n\n100% 10 17 11 8 6\n\n4\n\n9\n\n19 30\n\n37\n\n90\n\n86\n\n74\n\n70\n\n62 60\n\nFinland UK Germany USA Brazil France\n\nBroadcast (TV & Radio) Online/other Print\n\nMost occurring brands by journalists named\n\nFrance Brazil USA Germany UK Finland\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 52","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Digital_News-Report_2022_p52.pdf"]},"response":{"errors":[],"extract_id":"extract_33181b94cf644dce74f4fe3a370dcb5d","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**TV DOMINATES IN BRAZIL AND FRANCE BUT NOT SO MUCH IN FINLAND**\n\nSimilar to the US, television journalists are prominent in the most-named lists in Brazil and France. In Brazil, William Bonner, anchor for the country’s popular _Jornal Nacional_ , heads the list, followed by crime TV presenter José Luiz Datena and another TV Globo presenter, Maria Júlia (‘Maju’) Coutinho. The vast majority (86%) of named journalists in Brazil are presenters or anchors from broadcast media, and more than four in ten (41%) work for TV Globo, just one part of Brazil’s biggest news and entertainment company, reflecting the powerful position of this brand. A further fifth (17%) work for other top broadcasters Record TV or Band TV. Just 4% of the journalists mentioned work for any print outlet.\n\nThe story in France is very similar, with, if anything, an even greater focus on TV news presenters. Nine in ten journalists named in France work in broadcast media, primarily at television outlets such as TF1 and France Télévisions – which received 23% and 21% of mentions respectively. Topping the French list are TV presenters Pascal Praud and Anne-Claire Coudray. As in Brazil, journalists from national newspapers like _Le Monde_ and _Le Figaro_ are rarely mentioned, with only 6% of names coming from print media. Neither country sees any significant number of journalists from digital media outlets named (4% in Brazil, 3% in France).\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 27\n\nTHREE MOST-MENTIONED JOURNALISTS AND THEIR ROLES – SELECTED MARKETS\n\nJournalists Brand Role\n\n1 **William Bonner** TV Globo Presenter/Anchor\n\n2 **José Luiz Datena** Band TV Presenter/Anchor\n\n3 **Maria Júlia Coutinho** TV Globo Presenter/Anchor\n\n**BRAZIL**\n\nJournalists Brand Role\n\n1 **Pascal Praud** CNews Presenter/Anchor\n\n2 **Anne-Claire Coudray** TF1 Presenter/Anchor\n\n3 **Jean-Jacques Bourdin** RMC Presenter/Anchor\n\n**FRANCE**\n\nJournalists Brand Role\n\n1 **Marietta Slomka** ZDF Presenter/Anchor\n\n2 **Claus Kleber** ZDF Presenter/Anchor\n\n3 **Peter Kloeppel** RTL Presenter/Anchor\n\n**GERMANY**\n\nJournalists Brand Role\n\n1 **Sanna Ukkola** AlfaTV/ Iltalehti TV Presenter (AlfaTV) and Columnist (Iltalehti)\n\n2 **Matti Rönkä** Yle Presenter/Anchor\n\n3 **Timo Haapala** Ilta-Sanomat Political Correspondent/ Columnist\n\n**FINLAND**\n\n**Q \\_Journalists\\_2.** Please type in the names of up to five journalists that you regularly pay attention to (e.g. news presenters, columnists, or reporters). _Base: Total sample in each market (n ≈ 2000)._\n\nThe picture in Finland, meanwhile, is a little different. Like the UK, there is more focus on journalists from print media, with over a third (37%) of names being columnists or reporters for outlets such as _Iltalehti_ , _Ilta-Sanomat_ , and _Helsingin Sanomat_ . Together, these papers received a third (32%) of mentions, reflecting the continued prominence of print news brands in Finland. Public broadcaster Yle received 35% of mentions, with presenter Matti Rönkä ranking first equal alongside former Yle presenter (and current AlfaTV presenter/ _Iltalehti_ columnist) Sanna Ukkola. Among the most-mentioned brands is also MTV3, with 22% of named journalists working for this outlet. In all, six in ten (60%) journalists named in Finland work for broadcast media – a lower figure than other countries, but still high. Again, journalists working for digital-born media do not feature prominently.\n\nFinally, an in-between country is Germany, where there are somewhat more print journalists (19%) named than in Brazil or France, but fewer than the UK and Finland. Those named are spread across outlets including _Die Welt_ , _Die Zeit_ , and _Der Spiegel_ . The proportion of journalists named from broadcast media is also in the middle of the pack (70%), with this seven-in-ten figure being largely driven by the prominence of ARD, ZDF, and RTL, which make up two-thirds (64%) of mentions. Reflecting the wide popularity of these broadcasters, the most-named journalists are _ZDF heute-journal_ anchors Marietta Slomka and (formerly) Claus Kleber, alongside _RTL aktuell_ host Peter Kloeppel.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 28\n\nPROPORTION OF JOURNALISTS MENTIONED FROM EACH MEDIA TYPE – SELECTED MARKETS\n\n**Q \\_Journalists\\_2.** Please type in the names of up to five journalists that you regularly pay attention to (e.g. news presenters, columnists, or reporters). Base: All who mentioned the name of a journalist: France = 723, Brazil = 1008, USA = 635, Germany = 327, UK = 576, Finland = 431.\n\n0%\n\n25%\n\n50%\n\n75%\n\n100% 10 17 11 8 6\n\n4\n\n9\n\n19 30\n\n37\n\n90\n\n86\n\n74\n\n70\n\n62 60\n\nFinland UK Germany USA Brazil France\n\nBroadcast (TV & Radio) Online/other Print\n\nMost occurring brands by journalists named\n\nFrance Brazil USA Germany UK Finland\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 52"],"full_content":"**TV DOMINATES IN BRAZIL AND FRANCE BUT NOT SO MUCH IN FINLAND**\n\nSimilar to the US, television journalists are prominent in the most-named lists in Brazil and France. In Brazil, William Bonner, anchor for the country’s popular _Jornal Nacional_ , heads the list, followed by crime TV presenter José Luiz Datena and another TV Globo presenter, Maria Júlia (‘Maju’) Coutinho. The vast majority (86%) of named journalists in Brazil are presenters or anchors from broadcast media, and more than four in ten (41%) work for TV Globo, just one part of Brazil’s biggest news and entertainment company, reflecting the powerful position of this brand. A further fifth (17%) work for other top broadcasters Record TV or Band TV. Just 4% of the journalists mentioned work for any print outlet.\n\nThe story in France is very similar, with, if anything, an even greater focus on TV news presenters. Nine in ten journalists named in France work in broadcast media, primarily at television outlets such as TF1 and France Télévisions – which received 23% and 21% of mentions respectively. Topping the French list are TV presenters Pascal Praud and Anne-Claire Coudray. As in Brazil, journalists from national newspapers like _Le Monde_ and _Le Figaro_ are rarely mentioned, with only 6% of names coming from print media. Neither country sees any significant number of journalists from digital media outlets named (4% in Brazil, 3% in France).\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 27\n\nTHREE MOST-MENTIONED JOURNALISTS AND THEIR ROLES – SELECTED MARKETS\n\nJournalists Brand Role\n\n1 **William Bonner** TV Globo Presenter/Anchor\n\n2 **José Luiz Datena** Band TV Presenter/Anchor\n\n3 **Maria Júlia Coutinho** TV Globo Presenter/Anchor\n\n**BRAZIL**\n\nJournalists Brand Role\n\n1 **Pascal Praud** CNews Presenter/Anchor\n\n2 **Anne-Claire Coudray** TF1 Presenter/Anchor\n\n3 **Jean-Jacques Bourdin** RMC Presenter/Anchor\n\n**FRANCE**\n\nJournalists Brand Role\n\n1 **Marietta Slomka** ZDF Presenter/Anchor\n\n2 **Claus Kleber** ZDF Presenter/Anchor\n\n3 **Peter Kloeppel** RTL Presenter/Anchor\n\n**GERMANY**\n\nJournalists Brand Role\n\n1 **Sanna Ukkola** AlfaTV/ Iltalehti TV Presenter (AlfaTV) and Columnist (Iltalehti)\n\n2 **Matti Rönkä** Yle Presenter/Anchor\n\n3 **Timo Haapala** Ilta-Sanomat Political Correspondent/ Columnist\n\n**FINLAND**\n\n**Q \\_Journalists\\_2.** Please type in the names of up to five journalists that you regularly pay attention to (e.g. news presenters, columnists, or reporters). _Base: Total sample in each market (n ≈ 2000)._\n\nThe picture in Finland, meanwhile, is a little different. Like the UK, there is more focus on journalists from print media, with over a third (37%) of names being columnists or reporters for outlets such as _Iltalehti_ , _Ilta-Sanomat_ , and _Helsingin Sanomat_ . Together, these papers received a third (32%) of mentions, reflecting the continued prominence of print news brands in Finland. Public broadcaster Yle received 35% of mentions, with presenter Matti Rönkä ranking first equal alongside former Yle presenter (and current AlfaTV presenter/ _Iltalehti_ columnist) Sanna Ukkola. Among the most-mentioned brands is also MTV3, with 22% of named journalists working for this outlet. In all, six in ten (60%) journalists named in Finland work for broadcast media – a lower figure than other countries, but still high. Again, journalists working for digital-born media do not feature prominently.\n\nFinally, an in-between country is Germany, where there are somewhat more print journalists (19%) named than in Brazil or France, but fewer than the UK and Finland. Those named are spread across outlets including _Die Welt_ , _Die Zeit_ , and _Der Spiegel_ . The proportion of journalists named from broadcast media is also in the middle of the pack (70%), with this seven-in-ten figure being largely driven by the prominence of ARD, ZDF, and RTL, which make up two-thirds (64%) of mentions. Reflecting the wide popularity of these broadcasters, the most-named journalists are _ZDF heute-journal_ anchors Marietta Slomka and (formerly) Claus Kleber, alongside _RTL aktuell_ host Peter Kloeppel.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 28\n\nPROPORTION OF JOURNALISTS MENTIONED FROM EACH MEDIA TYPE – SELECTED MARKETS\n\n**Q \\_Journalists\\_2.** Please type in the names of up to five journalists that you regularly pay attention to (e.g. news presenters, columnists, or reporters). Base: All who mentioned the name of a journalist: France = 723, Brazil = 1008, USA = 635, Germany = 327, UK = 576, Finland = 431.\n\n0%\n\n25%\n\n50%\n\n75%\n\n100% 10 17 11 8 6\n\n4\n\n9\n\n19 30\n\n37\n\n90\n\n86\n\n74\n\n70\n\n62 60\n\nFinland UK Germany USA Brazil France\n\nBroadcast (TV & Radio) Online/other Print\n\nMost occurring brands by journalists named\n\nFrance Brazil USA Germany UK Finland\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 52","publish_date":null,"title":"Graph 27 Graph 28","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Digital_News-Report_2022_p52.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Digital_News-Report_2022_p52.pdf"},"request":{"example_id":"chart/Digital_News-Report_2022_p52","product_type":"parse","source_file_path":"Digital_News-Report_2022_p52.pdf"},"started_at":"2026-11-04T19:46:36.203683"} +{"completed_at":"2026-11-04T19:31:17.914202","latency_in_ms":1811,"output":{"example_id":"chart/Digital_News-Report_2022_p55","grounded_pages":[],"layout_pages":[],"markdown":"27 https://www.reuters.com/world/americas/chiles-record-breaking-drought-makes-climate-change-very-easy-see-2021-08-10/ 28 https://about.netflix.com/en/news/net-zero-nature-our-climate-commitment 29 https://www.theguardian.com/environment/2021/mar/31/seaspiracy-netflix-documentary-accused-of-misrepresentation-by-participants and https://www.independent.co.uk/voices/\n\nbbc-david-attenborough-nature-documentaries-fake-a8291961.html\n\nOn the other hand, in markets with the highest levels of interest, there is less left–right polarisation. We can see this dynamic play out in Greece and Portugal. In Greece, the difference in the proportion of people on the political left versus right who are interested in climate change news is only 16pp, while in Portugal it is just 10pp.\n\nAnother influence on our figures may be the impacts of climate change itself. Many countries have witnessed extreme weather events linked to climate change; Greece and Portugal, for instance, have suffered from devastating wildfires in recent years, and Chile continues to suffer from severe drought, which has made climate change ‘easy to see’. ²⁷ It may be that audiences are more interested in news about the topic when they are not so polarised and can clearly see the negative effects of extreme weather where they live.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 31\n\nA cow is seen on land that used to be filled with water, at the Aculeo Lagoon in Paine, Chile January 9, 2019. REUTERS Rodrigo Garrido.\n\nA cow is seen on land that used to be filled with water, at the Aculeo Lagoon in Paine, Chile January 9, 2019. REUTERS Rodrigo Garrido.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 32\n\n0% 25% 50% 75%\n\n43\n\n51\n\n40\n\n39\n\n39\n\n20\n\n23\n\n14\n\n56\n\n55\n\n58\n\n44\n\n48\n\n37\n\n35\n\n33\n\n59\n\n61\n\n62\n\n45\n\n41\n\n56\n\n49\n\n55\n\nRight\n\nCentre\n\nLef\n\nUSA\n\nNorway\n\nAustralia\n\nJapan\n\nS. Korea\n\nChile\n\nPortugal\n\nGreece\n\nPROPORTION INTERESTED IN CLIMATE CHANGE NEWS – BY POLITICAL LEANING – SELECTED MARKETS\n\n**Q1d\\_2022.** Which of the following types of news, if any, are you interested in? **Q1F.** Some people talk about ‘left’, ‘right’, and ‘centre’ to describe parties and politicians. With this in mind, where would you place yourself on the following scale? _Base: Left/Centre/Right: Greece = 272/1137/189, Portugal = 513/832/257, Chile = 261/977/192, South Korea = 373/1277/260, Japan = 65/1249/140, Australia = 390/1035/298, Norway = 385/1082/296, USA = 459/778/498._\n\nLeft Centre Right\n\n**16pp** gap\n\n**10pp** gap\n\n**36pp** gap\n\n**41pp** gap\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 33\n\nPROPORTION WHO PAY ATTENTION TO EACH SOURCE FOR CLIMATE CHANGE NEWS – BY AGE – ALL MARKETS\n\n**C1\\_2022a.** When it comes to climate change news, which of the following sources, if any, do you pay most attention to? _Please select all that apply. Base: U35/35+ in all markets = 28,001/65,431. Total sample in all markets = 93,432. Note: Attention to ‘scientists/experts/academics’ as a source was asked but data excluded as they are mostly accessed via other sources._\n\nU35 35+\n\n0% 25% 50%\n\n42\n\n33\n\n13\n\n9\n\n13\n\n10\n\n11\n\n34\n\n32\n\n17\n\n16\n\n14\n\n12\n\n8\n\n35+\n\nU35\n\nDon't pay attention to climate change\n\nPoliticians and political parties\n\nOrdinary people\n\nCelebrities/social media personalities/activists\n\nSmaller or alternative news sources\n\nMajor news organisations\n\nDocumentaries (TV or flm)\n\n0% 25% 50%\n\n42\n\n33\n\n13\n\n9\n\n13\n\n10\n\n11\n\n34\n\n32\n\n17\n\n16\n\n14\n\n12\n\n8 Don't pay attention to climate change\n\nPoliticians and political parties\n\nOrdinary people\n\nCelebrities/social media personalities/activists\n\nSmaller or alternative news sources\n\nMajor news organisations\n\nDocumentaries (TV or flm)\n\n**39%** total\n\n**33%** total\n\n**14%** total\n\n**11%** total\n\n**13%** total\n\n**10%** total\n\n**10%** total\n\n**SOURCES OF CLIMATE CHANGE NEWS**\n\nBeyond interest, we can examine the sources people say they pay most attention to for climate change news. Overall, across the markets we cover, more people say they pay attention to documentaries (39%) than to major news organisations (33%) for information about this topic. This is the case across all markets in the aggregate, as well as across age groups.\n\nThis disparity between the use of documentaries and major news organisations for climate change news is also the case across many individual markets. In Greece, for instance, there is a large gap in the proportion of people paying attention to documentaries (55%) versus major news organisations (35%). Exceptions to this trend are markets like Japan, where larger proportions of people say they pay attention to major news organisations (41%) than documentaries (27%). But, more often than not, respondents across most markets say that documentaries draw their attention.\n\nIt may be that films and television shows, from Al Gore’s 2006 _An Inconvenient Truth_ – which drew global attention to climate change – to the numerous nature documentaries produced since then by the BBC (e.g. _Blue Planet II_ ), Netflix (e.g. _Seaspiracy_ ), and Disney+ (e.g. _Elephant_ ), pack a bigger punch and remain more memorable to people than daily news reports. These are films and television series which, while not always about climate change, combine stunning visuals with compelling storytelling and environmental messages, reaching millions of people. Netflix says its series _Our Planet_ , narrated by Sir David Attenborough, pictured right, has reached 100m households since its release in 2019. ²⁸ While some documentaries have been criticised for their accuracy, ²⁹ the emotive power of these films certainly seems to resonate with audiences. Their popularity here is also further evidence of the importance of TV in driving attention to climate change information – something we noted in 2020.\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 55","pages":[{"line_numbers":[],"markdown":"27 https://www.reuters.com/world/americas/chiles-record-breaking-drought-makes-climate-change-very-easy-see-2021-08-10/ 28 https://about.netflix.com/en/news/net-zero-nature-our-climate-commitment 29 https://www.theguardian.com/environment/2021/mar/31/seaspiracy-netflix-documentary-accused-of-misrepresentation-by-participants and https://www.independent.co.uk/voices/\n\nbbc-david-attenborough-nature-documentaries-fake-a8291961.html\n\nOn the other hand, in markets with the highest levels of interest, there is less left–right polarisation. We can see this dynamic play out in Greece and Portugal. In Greece, the difference in the proportion of people on the political left versus right who are interested in climate change news is only 16pp, while in Portugal it is just 10pp.\n\nAnother influence on our figures may be the impacts of climate change itself. Many countries have witnessed extreme weather events linked to climate change; Greece and Portugal, for instance, have suffered from devastating wildfires in recent years, and Chile continues to suffer from severe drought, which has made climate change ‘easy to see’. ²⁷ It may be that audiences are more interested in news about the topic when they are not so polarised and can clearly see the negative effects of extreme weather where they live.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 31\n\nA cow is seen on land that used to be filled with water, at the Aculeo Lagoon in Paine, Chile January 9, 2019. REUTERS Rodrigo Garrido.\n\nA cow is seen on land that used to be filled with water, at the Aculeo Lagoon in Paine, Chile January 9, 2019. REUTERS Rodrigo Garrido.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 32\n\n0% 25% 50% 75%\n\n43\n\n51\n\n40\n\n39\n\n39\n\n20\n\n23\n\n14\n\n56\n\n55\n\n58\n\n44\n\n48\n\n37\n\n35\n\n33\n\n59\n\n61\n\n62\n\n45\n\n41\n\n56\n\n49\n\n55\n\nRight\n\nCentre\n\nLef\n\nUSA\n\nNorway\n\nAustralia\n\nJapan\n\nS. Korea\n\nChile\n\nPortugal\n\nGreece\n\nPROPORTION INTERESTED IN CLIMATE CHANGE NEWS – BY POLITICAL LEANING – SELECTED MARKETS\n\n**Q1d\\_2022.** Which of the following types of news, if any, are you interested in? **Q1F.** Some people talk about ‘left’, ‘right’, and ‘centre’ to describe parties and politicians. With this in mind, where would you place yourself on the following scale? _Base: Left/Centre/Right: Greece = 272/1137/189, Portugal = 513/832/257, Chile = 261/977/192, South Korea = 373/1277/260, Japan = 65/1249/140, Australia = 390/1035/298, Norway = 385/1082/296, USA = 459/778/498._\n\nLeft Centre Right\n\n**16pp** gap\n\n**10pp** gap\n\n**36pp** gap\n\n**41pp** gap\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 33\n\nPROPORTION WHO PAY ATTENTION TO EACH SOURCE FOR CLIMATE CHANGE NEWS – BY AGE – ALL MARKETS\n\n**C1\\_2022a.** When it comes to climate change news, which of the following sources, if any, do you pay most attention to? _Please select all that apply. Base: U35/35+ in all markets = 28,001/65,431. Total sample in all markets = 93,432. Note: Attention to ‘scientists/experts/academics’ as a source was asked but data excluded as they are mostly accessed via other sources._\n\nU35 35+\n\n0% 25% 50%\n\n42\n\n33\n\n13\n\n9\n\n13\n\n10\n\n11\n\n34\n\n32\n\n17\n\n16\n\n14\n\n12\n\n8\n\n35+\n\nU35\n\nDon't pay attention to climate change\n\nPoliticians and political parties\n\nOrdinary people\n\nCelebrities/social media personalities/activists\n\nSmaller or alternative news sources\n\nMajor news organisations\n\nDocumentaries (TV or flm)\n\n0% 25% 50%\n\n42\n\n33\n\n13\n\n9\n\n13\n\n10\n\n11\n\n34\n\n32\n\n17\n\n16\n\n14\n\n12\n\n8 Don't pay attention to climate change\n\nPoliticians and political parties\n\nOrdinary people\n\nCelebrities/social media personalities/activists\n\nSmaller or alternative news sources\n\nMajor news organisations\n\nDocumentaries (TV or flm)\n\n**39%** total\n\n**33%** total\n\n**14%** total\n\n**11%** total\n\n**13%** total\n\n**10%** total\n\n**10%** total\n\n**SOURCES OF CLIMATE CHANGE NEWS**\n\nBeyond interest, we can examine the sources people say they pay most attention to for climate change news. Overall, across the markets we cover, more people say they pay attention to documentaries (39%) than to major news organisations (33%) for information about this topic. This is the case across all markets in the aggregate, as well as across age groups.\n\nThis disparity between the use of documentaries and major news organisations for climate change news is also the case across many individual markets. In Greece, for instance, there is a large gap in the proportion of people paying attention to documentaries (55%) versus major news organisations (35%). Exceptions to this trend are markets like Japan, where larger proportions of people say they pay attention to major news organisations (41%) than documentaries (27%). But, more often than not, respondents across most markets say that documentaries draw their attention.\n\nIt may be that films and television shows, from Al Gore’s 2006 _An Inconvenient Truth_ – which drew global attention to climate change – to the numerous nature documentaries produced since then by the BBC (e.g. _Blue Planet II_ ), Netflix (e.g. _Seaspiracy_ ), and Disney+ (e.g. _Elephant_ ), pack a bigger punch and remain more memorable to people than daily news reports. These are films and television series which, while not always about climate change, combine stunning visuals with compelling storytelling and environmental messages, reaching millions of people. Netflix says its series _Our Planet_ , narrated by Sir David Attenborough, pictured right, has reached 100m households since its release in 2019. ²⁸ While some documentaries have been criticised for their accuracy, ²⁹ the emotive power of these films certainly seems to resonate with audiences. Their popularity here is also further evidence of the importance of TV in driving attention to climate change information – something we noted in 2020.\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 55","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"27 https://www.reuters.com/world/americas/chiles-record-breaking-drought-makes-climate-change-very-easy-see-2021-08-10/ 28 https://about.netflix.com/en/news/net-zero-nature-our-climate-commitment 29 https://www.theguardian.com/environment/2021/mar/31/seaspiracy-netflix-documentary-accused-of-misrepresentation-by-participants and https://www.independent.co.uk/voices/\n\nbbc-david-attenborough-nature-documentaries-fake-a8291961.html\n\nOn the other hand, in markets with the highest levels of interest, there is less left–right polarisation. We can see this dynamic play out in Greece and Portugal. In Greece, the difference in the proportion of people on the political left versus right who are interested in climate change news is only 16pp, while in Portugal it is just 10pp.\n\nAnother influence on our figures may be the impacts of climate change itself. Many countries have witnessed extreme weather events linked to climate change; Greece and Portugal, for instance, have suffered from devastating wildfires in recent years, and Chile continues to suffer from severe drought, which has made climate change ‘easy to see’. ²⁷ It may be that audiences are more interested in news about the topic when they are not so polarised and can clearly see the negative effects of extreme weather where they live.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 31\n\nA cow is seen on land that used to be filled with water, at the Aculeo Lagoon in Paine, Chile January 9, 2019. REUTERS Rodrigo Garrido.\n\nA cow is seen on land that used to be filled with water, at the Aculeo Lagoon in Paine, Chile January 9, 2019. REUTERS Rodrigo Garrido.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 32\n\n0% 25% 50% 75%\n\n43\n\n51\n\n40\n\n39\n\n39\n\n20\n\n23\n\n14\n\n56\n\n55\n\n58\n\n44\n\n48\n\n37\n\n35\n\n33\n\n59\n\n61\n\n62\n\n45\n\n41\n\n56\n\n49\n\n55\n\nRight\n\nCentre\n\nLef\n\nUSA\n\nNorway\n\nAustralia\n\nJapan\n\nS. Korea\n\nChile\n\nPortugal\n\nGreece\n\nPROPORTION INTERESTED IN CLIMATE CHANGE NEWS – BY POLITICAL LEANING – SELECTED MARKETS\n\n**Q1d\\_2022.** Which of the following types of news, if any, are you interested in? **Q1F.** Some people talk about ‘left’, ‘right’, and ‘centre’ to describe parties and politicians. With this in mind, where would you place yourself on the following scale? _Base: Left/Centre/Right: Greece = 272/1137/189, Portugal = 513/832/257, Chile = 261/977/192, South Korea = 373/1277/260, Japan = 65/1249/140, Australia = 390/1035/298, Norway = 385/1082/296, USA = 459/778/498._\n\nLeft Centre Right\n\n**16pp** gap\n\n**10pp** gap\n\n**36pp** gap\n\n**41pp** gap\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 33\n\nPROPORTION WHO PAY ATTENTION TO EACH SOURCE FOR CLIMATE CHANGE NEWS – BY AGE – ALL MARKETS\n\n**C1\\_2022a.** When it comes to climate change news, which of the following sources, if any, do you pay most attention to? _Please select all that apply. Base: U35/35+ in all markets = 28,001/65,431. Total sample in all markets = 93,432. Note: Attention to ‘scientists/experts/academics’ as a source was asked but data excluded as they are mostly accessed via other sources._\n\nU35 35+\n\n0% 25% 50%\n\n42\n\n33\n\n13\n\n9\n\n13\n\n10\n\n11\n\n34\n\n32\n\n17\n\n16\n\n14\n\n12\n\n8\n\n35+\n\nU35\n\nDon't pay attention to climate change\n\nPoliticians and political parties\n\nOrdinary people\n\nCelebrities/social media personalities/activists\n\nSmaller or alternative news sources\n\nMajor news organisations\n\nDocumentaries (TV or flm)\n\n0% 25% 50%\n\n42\n\n33\n\n13\n\n9\n\n13\n\n10\n\n11\n\n34\n\n32\n\n17\n\n16\n\n14\n\n12\n\n8 Don't pay attention to climate change\n\nPoliticians and political parties\n\nOrdinary people\n\nCelebrities/social media personalities/activists\n\nSmaller or alternative news sources\n\nMajor news organisations\n\nDocumentaries (TV or flm)\n\n**39%** total\n\n**33%** total\n\n**14%** total\n\n**11%** total\n\n**13%** total\n\n**10%** total\n\n**10%** total\n\n**SOURCES OF CLIMATE CHANGE NEWS**\n\nBeyond interest, we can examine the sources people say they pay most attention to for climate change news. Overall, across the markets we cover, more people say they pay attention to documentaries (39%) than to major news organisations (33%) for information about this topic. This is the case across all markets in the aggregate, as well as across age groups.\n\nThis disparity between the use of documentaries and major news organisations for climate change news is also the case across many individual markets. In Greece, for instance, there is a large gap in the proportion of people paying attention to documentaries (55%) versus major news organisations (35%). Exceptions to this trend are markets like Japan, where larger proportions of people say they pay attention to major news organisations (41%) than documentaries (27%). But, more often than not, respondents across most markets say that documentaries draw their attention.\n\nIt may be that films and television shows, from Al Gore’s 2006 _An Inconvenient Truth_ – which drew global attention to climate change – to the numerous nature documentaries produced since then by the BBC (e.g. _Blue Planet II_ ), Netflix (e.g. _Seaspiracy_ ), and Disney+ (e.g. _Elephant_ ), pack a bigger punch and remain more memorable to people than daily news reports. These are films and television series which, while not always about climate change, combine stunning visuals with compelling storytelling and environmental messages, reaching millions of people. Netflix says its series _Our Planet_ , narrated by Sir David Attenborough, pictured right, has reached 100m households since its release in 2019. ²⁸ While some documentaries have been criticised for their accuracy, ²⁹ the emotive power of these films certainly seems to resonate with audiences. Their popularity here is also further evidence of the importance of TV in driving attention to climate change information – something we noted in 2020.\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 55","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Digital_News-Report_2022_p55.pdf"]},"response":{"errors":[],"extract_id":"extract_972d6943eeff38df4ea546ea0dbc0abf","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["27 https://www.reuters.com/world/americas/chiles-record-breaking-drought-makes-climate-change-very-easy-see-2021-08-10/ 28 https://about.netflix.com/en/news/net-zero-nature-our-climate-commitment 29 https://www.theguardian.com/environment/2021/mar/31/seaspiracy-netflix-documentary-accused-of-misrepresentation-by-participants and https://www.independent.co.uk/voices/\n\nbbc-david-attenborough-nature-documentaries-fake-a8291961.html\n\nOn the other hand, in markets with the highest levels of interest, there is less left–right polarisation. We can see this dynamic play out in Greece and Portugal. In Greece, the difference in the proportion of people on the political left versus right who are interested in climate change news is only 16pp, while in Portugal it is just 10pp.\n\nAnother influence on our figures may be the impacts of climate change itself. Many countries have witnessed extreme weather events linked to climate change; Greece and Portugal, for instance, have suffered from devastating wildfires in recent years, and Chile continues to suffer from severe drought, which has made climate change ‘easy to see’. ²⁷ It may be that audiences are more interested in news about the topic when they are not so polarised and can clearly see the negative effects of extreme weather where they live.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 31\n\nA cow is seen on land that used to be filled with water, at the Aculeo Lagoon in Paine, Chile January 9, 2019. REUTERS Rodrigo Garrido.\n\nA cow is seen on land that used to be filled with water, at the Aculeo Lagoon in Paine, Chile January 9, 2019. REUTERS Rodrigo Garrido.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 32\n\n0% 25% 50% 75%\n\n43\n\n51\n\n40\n\n39\n\n39\n\n20\n\n23\n\n14\n\n56\n\n55\n\n58\n\n44\n\n48\n\n37\n\n35\n\n33\n\n59\n\n61\n\n62\n\n45\n\n41\n\n56\n\n49\n\n55\n\nRight\n\nCentre\n\nLef\n\nUSA\n\nNorway\n\nAustralia\n\nJapan\n\nS. Korea\n\nChile\n\nPortugal\n\nGreece\n\nPROPORTION INTERESTED IN CLIMATE CHANGE NEWS – BY POLITICAL LEANING – SELECTED MARKETS\n\n**Q1d\\_2022.** Which of the following types of news, if any, are you interested in? **Q1F.** Some people talk about ‘left’, ‘right’, and ‘centre’ to describe parties and politicians. With this in mind, where would you place yourself on the following scale? _Base: Left/Centre/Right: Greece = 272/1137/189, Portugal = 513/832/257, Chile = 261/977/192, South Korea = 373/1277/260, Japan = 65/1249/140, Australia = 390/1035/298, Norway = 385/1082/296, USA = 459/778/498._\n\nLeft Centre Right\n\n**16pp** gap\n\n**10pp** gap\n\n**36pp** gap\n\n**41pp** gap\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 33\n\nPROPORTION WHO PAY ATTENTION TO EACH SOURCE FOR CLIMATE CHANGE NEWS – BY AGE – ALL MARKETS\n\n**C1\\_2022a.** When it comes to climate change news, which of the following sources, if any, do you pay most attention to? _Please select all that apply. Base: U35/35+ in all markets = 28,001/65,431. Total sample in all markets = 93,432. Note: Attention to ‘scientists/experts/academics’ as a source was asked but data excluded as they are mostly accessed via other sources._\n\nU35 35+\n\n0% 25% 50%\n\n42\n\n33\n\n13\n\n9\n\n13\n\n10\n\n11\n\n34\n\n32\n\n17\n\n16\n\n14\n\n12\n\n8\n\n35+\n\nU35\n\nDon't pay attention to climate change\n\nPoliticians and political parties\n\nOrdinary people\n\nCelebrities/social media personalities/activists\n\nSmaller or alternative news sources\n\nMajor news organisations\n\nDocumentaries (TV or flm)\n\n0% 25% 50%\n\n42\n\n33\n\n13\n\n9\n\n13\n\n10\n\n11\n\n34\n\n32\n\n17\n\n16\n\n14\n\n12\n\n8 Don't pay attention to climate change\n\nPoliticians and political parties\n\nOrdinary people\n\nCelebrities/social media personalities/activists\n\nSmaller or alternative news sources\n\nMajor news organisations\n\nDocumentaries (TV or flm)\n\n**39%** total\n\n**33%** total\n\n**14%** total\n\n**11%** total\n\n**13%** total\n\n**10%** total\n\n**10%** total\n\n**SOURCES OF CLIMATE CHANGE NEWS**\n\nBeyond interest, we can examine the sources people say they pay most attention to for climate change news. Overall, across the markets we cover, more people say they pay attention to documentaries (39%) than to major news organisations (33%) for information about this topic. This is the case across all markets in the aggregate, as well as across age groups.\n\nThis disparity between the use of documentaries and major news organisations for climate change news is also the case across many individual markets. In Greece, for instance, there is a large gap in the proportion of people paying attention to documentaries (55%) versus major news organisations (35%). Exceptions to this trend are markets like Japan, where larger proportions of people say they pay attention to major news organisations (41%) than documentaries (27%). But, more often than not, respondents across most markets say that documentaries draw their attention.\n\nIt may be that films and television shows, from Al Gore’s 2006 _An Inconvenient Truth_ – which drew global attention to climate change – to the numerous nature documentaries produced since then by the BBC (e.g. _Blue Planet II_ ), Netflix (e.g. _Seaspiracy_ ), and Disney+ (e.g. _Elephant_ ), pack a bigger punch and remain more memorable to people than daily news reports. These are films and television series which, while not always about climate change, combine stunning visuals with compelling storytelling and environmental messages, reaching millions of people. Netflix says its series _Our Planet_ , narrated by Sir David Attenborough, pictured right, has reached 100m households since its release in 2019. ²⁸ While some documentaries have been criticised for their accuracy, ²⁹ the emotive power of these films certainly seems to resonate with audiences. Their popularity here is also further evidence of the importance of TV in driving attention to climate change information – something we noted in 2020.\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 55"],"full_content":"27 https://www.reuters.com/world/americas/chiles-record-breaking-drought-makes-climate-change-very-easy-see-2021-08-10/ 28 https://about.netflix.com/en/news/net-zero-nature-our-climate-commitment 29 https://www.theguardian.com/environment/2021/mar/31/seaspiracy-netflix-documentary-accused-of-misrepresentation-by-participants and https://www.independent.co.uk/voices/\n\nbbc-david-attenborough-nature-documentaries-fake-a8291961.html\n\nOn the other hand, in markets with the highest levels of interest, there is less left–right polarisation. We can see this dynamic play out in Greece and Portugal. In Greece, the difference in the proportion of people on the political left versus right who are interested in climate change news is only 16pp, while in Portugal it is just 10pp.\n\nAnother influence on our figures may be the impacts of climate change itself. Many countries have witnessed extreme weather events linked to climate change; Greece and Portugal, for instance, have suffered from devastating wildfires in recent years, and Chile continues to suffer from severe drought, which has made climate change ‘easy to see’. ²⁷ It may be that audiences are more interested in news about the topic when they are not so polarised and can clearly see the negative effects of extreme weather where they live.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 31\n\nA cow is seen on land that used to be filled with water, at the Aculeo Lagoon in Paine, Chile January 9, 2019. REUTERS Rodrigo Garrido.\n\nA cow is seen on land that used to be filled with water, at the Aculeo Lagoon in Paine, Chile January 9, 2019. REUTERS Rodrigo Garrido.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 32\n\n0% 25% 50% 75%\n\n43\n\n51\n\n40\n\n39\n\n39\n\n20\n\n23\n\n14\n\n56\n\n55\n\n58\n\n44\n\n48\n\n37\n\n35\n\n33\n\n59\n\n61\n\n62\n\n45\n\n41\n\n56\n\n49\n\n55\n\nRight\n\nCentre\n\nLef\n\nUSA\n\nNorway\n\nAustralia\n\nJapan\n\nS. Korea\n\nChile\n\nPortugal\n\nGreece\n\nPROPORTION INTERESTED IN CLIMATE CHANGE NEWS – BY POLITICAL LEANING – SELECTED MARKETS\n\n**Q1d\\_2022.** Which of the following types of news, if any, are you interested in? **Q1F.** Some people talk about ‘left’, ‘right’, and ‘centre’ to describe parties and politicians. With this in mind, where would you place yourself on the following scale? _Base: Left/Centre/Right: Greece = 272/1137/189, Portugal = 513/832/257, Chile = 261/977/192, South Korea = 373/1277/260, Japan = 65/1249/140, Australia = 390/1035/298, Norway = 385/1082/296, USA = 459/778/498._\n\nLeft Centre Right\n\n**16pp** gap\n\n**10pp** gap\n\n**36pp** gap\n\n**41pp** gap\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 33\n\nPROPORTION WHO PAY ATTENTION TO EACH SOURCE FOR CLIMATE CHANGE NEWS – BY AGE – ALL MARKETS\n\n**C1\\_2022a.** When it comes to climate change news, which of the following sources, if any, do you pay most attention to? _Please select all that apply. Base: U35/35+ in all markets = 28,001/65,431. Total sample in all markets = 93,432. Note: Attention to ‘scientists/experts/academics’ as a source was asked but data excluded as they are mostly accessed via other sources._\n\nU35 35+\n\n0% 25% 50%\n\n42\n\n33\n\n13\n\n9\n\n13\n\n10\n\n11\n\n34\n\n32\n\n17\n\n16\n\n14\n\n12\n\n8\n\n35+\n\nU35\n\nDon't pay attention to climate change\n\nPoliticians and political parties\n\nOrdinary people\n\nCelebrities/social media personalities/activists\n\nSmaller or alternative news sources\n\nMajor news organisations\n\nDocumentaries (TV or flm)\n\n0% 25% 50%\n\n42\n\n33\n\n13\n\n9\n\n13\n\n10\n\n11\n\n34\n\n32\n\n17\n\n16\n\n14\n\n12\n\n8 Don't pay attention to climate change\n\nPoliticians and political parties\n\nOrdinary people\n\nCelebrities/social media personalities/activists\n\nSmaller or alternative news sources\n\nMajor news organisations\n\nDocumentaries (TV or flm)\n\n**39%** total\n\n**33%** total\n\n**14%** total\n\n**11%** total\n\n**13%** total\n\n**10%** total\n\n**10%** total\n\n**SOURCES OF CLIMATE CHANGE NEWS**\n\nBeyond interest, we can examine the sources people say they pay most attention to for climate change news. Overall, across the markets we cover, more people say they pay attention to documentaries (39%) than to major news organisations (33%) for information about this topic. This is the case across all markets in the aggregate, as well as across age groups.\n\nThis disparity between the use of documentaries and major news organisations for climate change news is also the case across many individual markets. In Greece, for instance, there is a large gap in the proportion of people paying attention to documentaries (55%) versus major news organisations (35%). Exceptions to this trend are markets like Japan, where larger proportions of people say they pay attention to major news organisations (41%) than documentaries (27%). But, more often than not, respondents across most markets say that documentaries draw their attention.\n\nIt may be that films and television shows, from Al Gore’s 2006 _An Inconvenient Truth_ – which drew global attention to climate change – to the numerous nature documentaries produced since then by the BBC (e.g. _Blue Planet II_ ), Netflix (e.g. _Seaspiracy_ ), and Disney+ (e.g. _Elephant_ ), pack a bigger punch and remain more memorable to people than daily news reports. These are films and television series which, while not always about climate change, combine stunning visuals with compelling storytelling and environmental messages, reaching millions of people. Netflix says its series _Our Planet_ , narrated by Sir David Attenborough, pictured right, has reached 100m households since its release in 2019. ²⁸ While some documentaries have been criticised for their accuracy, ²⁹ the emotive power of these films certainly seems to resonate with audiences. Their popularity here is also further evidence of the importance of TV in driving attention to climate change information – something we noted in 2020.\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 55","publish_date":null,"title":"Graph 32 Graph 33 Graph 31","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Digital_News-Report_2022_p55.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Digital_News-Report_2022_p55.pdf"},"request":{"example_id":"chart/Digital_News-Report_2022_p55","product_type":"parse","source_file_path":"Digital_News-Report_2022_p55.pdf"},"started_at":"2026-11-04T19:31:16.102402"} +{"completed_at":"2026-11-04T19:46:41.241070","latency_in_ms":2152,"output":{"example_id":"chart/Digital_News-Report_2022_p56","grounded_pages":[],"layout_pages":[],"markdown":"REUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 35\n\nPROPORTION WHO PAY ATTENTION TO EACH SOURCE FOR CLIMATE CHANGE NEWS – SELECTED MARKETS\n\n**C1\\_2022a.** When it comes to climate change news, which of the following sources, if any, do you pay most attention to? _Base: Total sample in each market (n ≈ 2,000)._\n\nMajor news organisations Documentaries\n\nDon’t pay attention to climate change Smaller/alternative news sources\n\n0% 25% 50% 75%\n\n4\n\n4\n\n10\n\n6\n\n20\n\n18\n\n23\n\n15\n\n14\n\n5\n\n22\n\n10\n\n15\n\n15\n\n42\n\n50\n\n27\n\n55\n\n31\n\n33\n\n24\n\n45\n\n43\n\n41\n\n35\n\n28\n\n27\n\n24\n\nDon't pay attention to climate change\n\nSmaller/alternative news sources\n\nDocumentaries\n\nMajor news organisations\n\nUSA\n\nAustralia\n\nNorway\n\nGreece\n\nJapan\n\nPortugal\n\nChile\n\nWhen we drill down into the attention paid to major news organisations, we find that usage for climate change news is lower in markets where interest and attention are polarised, such as the US and Australia. In these markets, along with Norway, we see the largest proportions of people saying they don’t pay attention to climate change at all, with these high figures largely being driven by those on the political right. Half (49%) of those on the political right in the US say they don’t pay attention to climate change at all, and it is a third in Australia (34%) and Norway (31%). This compares to just 5% of those on the political right in Portugal and Chile who say they don’t pay attention.\n\nIndeed, the comparisons between the US, Australia, and other markets are stark, showing the role that polarised politics – and media coverage – can play in driving down interest in and attention to climate change as an issue. Attention to sources across the board is low in both the US and Australia despite the visible effects of climate change witnessed in both countries in recent years. In contrast, countries like Greece and Chile do not appear to suffer from the same levels of audience polarisation and see higher rates of attention.\n\nFinally, there are also interesting differences in attention to sources by age. Respondents under 35 in many markets are often two or three times more likely to say they pay attention to celebrities, social media personalities, or activists for climate change news than people over 35. So-called ‘green influencers’ have amassed large followings online, drawing attention to the linkages between climate change and social justice issues. They include Vanessa Nakate, a climate activist from Uganda with hundreds of thousands of followers across Twitter and Instagram; Jack Harries, an environmentalist YouTuber whose channel has 3.7m subscribers; and Jerome Foster, a young climate activist who is a White House adviser with over 41,000 Instagram followers.\n\nGroups like the EcoTok collective on TikTok also seek to connect with the younger audience on the platform. Younger respondents are also somewhat more likely to say they get climate change news from smaller or alternative news sources, some of which have been started by young climate activists.\n\n**COVERAGE OF CLIMATE CHANGE**\n\nAudiences appear to be more interested in and pay more attention to climate change news in places where the negative effects are felt more acutely. But it is arguably too late if audiences only pay attention once disasters have already struck. A central challenge is how to cover the issue in a way that draws attention to the causes and decisions that lead to disasters, not ‘just’ the disasters themselves.\n\nOne approach to this is taking a clearer stance on the issue, as exemplified by the _Guardian_ calling for more action on the ‘climate crisis’ and many documentaries that take a clear activist stance on environmental issues. But how do audiences feel about news outlets adopting a position?\n\nWe find that, in general, there are strong regional differences in views. These reflect similar patterns to those already identified in this chapter: audiences in Latin America, Southern Europe, and several markets in Asia-Pacific are more open to news outlets taking a stand in favour of climate change action. Large proportions of respondents in Chile (58%), Portugal (48%), and the Philippines (42%) say that journalists should advocate for change. On the other hand, pluralities in Northern European, Western European, and North American markets favour impartiality. Respondents in Germany (45%), Norway (44%), and the US (42%) say they would prefer news outlets to reflect a range of views and leave it up to them to decide what to think (see chart next page).\n\nExamining these trends, it may be that audiences in markets such as Norway and Germany are predisposed to expect an impartial approach to reporting – particularly from public broadcasters. On the other hand, political polarisation may again be playing a role.\n\nIn many markets, large majorities of those on the political right want journalists to remain impartial on climate change and their views drive some of the overall trends. The proportion of those on the right favouring impartiality is 60% in Norway and 69% in the US. Conversely, those on the political left tend to favour journalists advocating for climate change action. Views in Chile and Portugal are not as polarised on this question, with both those on the political left and right favouring climate action.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 36\n\nPROPORTION WHO PAY ATTENTION TO CELEBRITIES, SOCIAL MEDIA PERSONALITIES, OR ACTIVISTS FOR CLIMATE CHANGE NEWS – BY AGE – SELECTED MARKETS\n\n**C1\\_2022a.** When it comes to climate change news, which of the following sources, if any, do you pay most attention to? Please select all that apply. _Base: U35/35+: Greece = 468/1536, Portugal = 462/1549, USA = 668/1368, Norway = 496/1514, Australia = 561/1477, Japan = 406/1609._\n\n0%\n\n4\n\n8\n\n4 3 5\n\n8\n\n18 16 14 12 12 12\n\nJapan Australia Norway USA Portugal Greece\n\n0%\n\n25%\n\n4\n\n8\n\n4 3\n\n5\n\n8\n\n18\n\n16\n\n14\n\n12 12 12\n\nJapan Australia Norway USA Portugal Greece\n\nU35 35+\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 56","pages":[{"line_numbers":[],"markdown":"REUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 35\n\nPROPORTION WHO PAY ATTENTION TO EACH SOURCE FOR CLIMATE CHANGE NEWS – SELECTED MARKETS\n\n**C1\\_2022a.** When it comes to climate change news, which of the following sources, if any, do you pay most attention to? _Base: Total sample in each market (n ≈ 2,000)._\n\nMajor news organisations Documentaries\n\nDon’t pay attention to climate change Smaller/alternative news sources\n\n0% 25% 50% 75%\n\n4\n\n4\n\n10\n\n6\n\n20\n\n18\n\n23\n\n15\n\n14\n\n5\n\n22\n\n10\n\n15\n\n15\n\n42\n\n50\n\n27\n\n55\n\n31\n\n33\n\n24\n\n45\n\n43\n\n41\n\n35\n\n28\n\n27\n\n24\n\nDon't pay attention to climate change\n\nSmaller/alternative news sources\n\nDocumentaries\n\nMajor news organisations\n\nUSA\n\nAustralia\n\nNorway\n\nGreece\n\nJapan\n\nPortugal\n\nChile\n\nWhen we drill down into the attention paid to major news organisations, we find that usage for climate change news is lower in markets where interest and attention are polarised, such as the US and Australia. In these markets, along with Norway, we see the largest proportions of people saying they don’t pay attention to climate change at all, with these high figures largely being driven by those on the political right. Half (49%) of those on the political right in the US say they don’t pay attention to climate change at all, and it is a third in Australia (34%) and Norway (31%). This compares to just 5% of those on the political right in Portugal and Chile who say they don’t pay attention.\n\nIndeed, the comparisons between the US, Australia, and other markets are stark, showing the role that polarised politics – and media coverage – can play in driving down interest in and attention to climate change as an issue. Attention to sources across the board is low in both the US and Australia despite the visible effects of climate change witnessed in both countries in recent years. In contrast, countries like Greece and Chile do not appear to suffer from the same levels of audience polarisation and see higher rates of attention.\n\nFinally, there are also interesting differences in attention to sources by age. Respondents under 35 in many markets are often two or three times more likely to say they pay attention to celebrities, social media personalities, or activists for climate change news than people over 35. So-called ‘green influencers’ have amassed large followings online, drawing attention to the linkages between climate change and social justice issues. They include Vanessa Nakate, a climate activist from Uganda with hundreds of thousands of followers across Twitter and Instagram; Jack Harries, an environmentalist YouTuber whose channel has 3.7m subscribers; and Jerome Foster, a young climate activist who is a White House adviser with over 41,000 Instagram followers.\n\nGroups like the EcoTok collective on TikTok also seek to connect with the younger audience on the platform. Younger respondents are also somewhat more likely to say they get climate change news from smaller or alternative news sources, some of which have been started by young climate activists.\n\n**COVERAGE OF CLIMATE CHANGE**\n\nAudiences appear to be more interested in and pay more attention to climate change news in places where the negative effects are felt more acutely. But it is arguably too late if audiences only pay attention once disasters have already struck. A central challenge is how to cover the issue in a way that draws attention to the causes and decisions that lead to disasters, not ‘just’ the disasters themselves.\n\nOne approach to this is taking a clearer stance on the issue, as exemplified by the _Guardian_ calling for more action on the ‘climate crisis’ and many documentaries that take a clear activist stance on environmental issues. But how do audiences feel about news outlets adopting a position?\n\nWe find that, in general, there are strong regional differences in views. These reflect similar patterns to those already identified in this chapter: audiences in Latin America, Southern Europe, and several markets in Asia-Pacific are more open to news outlets taking a stand in favour of climate change action. Large proportions of respondents in Chile (58%), Portugal (48%), and the Philippines (42%) say that journalists should advocate for change. On the other hand, pluralities in Northern European, Western European, and North American markets favour impartiality. Respondents in Germany (45%), Norway (44%), and the US (42%) say they would prefer news outlets to reflect a range of views and leave it up to them to decide what to think (see chart next page).\n\nExamining these trends, it may be that audiences in markets such as Norway and Germany are predisposed to expect an impartial approach to reporting – particularly from public broadcasters. On the other hand, political polarisation may again be playing a role.\n\nIn many markets, large majorities of those on the political right want journalists to remain impartial on climate change and their views drive some of the overall trends. The proportion of those on the right favouring impartiality is 60% in Norway and 69% in the US. Conversely, those on the political left tend to favour journalists advocating for climate change action. Views in Chile and Portugal are not as polarised on this question, with both those on the political left and right favouring climate action.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 36\n\nPROPORTION WHO PAY ATTENTION TO CELEBRITIES, SOCIAL MEDIA PERSONALITIES, OR ACTIVISTS FOR CLIMATE CHANGE NEWS – BY AGE – SELECTED MARKETS\n\n**C1\\_2022a.** When it comes to climate change news, which of the following sources, if any, do you pay most attention to? Please select all that apply. _Base: U35/35+: Greece = 468/1536, Portugal = 462/1549, USA = 668/1368, Norway = 496/1514, Australia = 561/1477, Japan = 406/1609._\n\n0%\n\n4\n\n8\n\n4 3 5\n\n8\n\n18 16 14 12 12 12\n\nJapan Australia Norway USA Portugal Greece\n\n0%\n\n25%\n\n4\n\n8\n\n4 3\n\n5\n\n8\n\n18\n\n16\n\n14\n\n12 12 12\n\nJapan Australia Norway USA Portugal Greece\n\nU35 35+\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 56","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"REUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 35\n\nPROPORTION WHO PAY ATTENTION TO EACH SOURCE FOR CLIMATE CHANGE NEWS – SELECTED MARKETS\n\n**C1\\_2022a.** When it comes to climate change news, which of the following sources, if any, do you pay most attention to? _Base: Total sample in each market (n ≈ 2,000)._\n\nMajor news organisations Documentaries\n\nDon’t pay attention to climate change Smaller/alternative news sources\n\n0% 25% 50% 75%\n\n4\n\n4\n\n10\n\n6\n\n20\n\n18\n\n23\n\n15\n\n14\n\n5\n\n22\n\n10\n\n15\n\n15\n\n42\n\n50\n\n27\n\n55\n\n31\n\n33\n\n24\n\n45\n\n43\n\n41\n\n35\n\n28\n\n27\n\n24\n\nDon't pay attention to climate change\n\nSmaller/alternative news sources\n\nDocumentaries\n\nMajor news organisations\n\nUSA\n\nAustralia\n\nNorway\n\nGreece\n\nJapan\n\nPortugal\n\nChile\n\nWhen we drill down into the attention paid to major news organisations, we find that usage for climate change news is lower in markets where interest and attention are polarised, such as the US and Australia. In these markets, along with Norway, we see the largest proportions of people saying they don’t pay attention to climate change at all, with these high figures largely being driven by those on the political right. Half (49%) of those on the political right in the US say they don’t pay attention to climate change at all, and it is a third in Australia (34%) and Norway (31%). This compares to just 5% of those on the political right in Portugal and Chile who say they don’t pay attention.\n\nIndeed, the comparisons between the US, Australia, and other markets are stark, showing the role that polarised politics – and media coverage – can play in driving down interest in and attention to climate change as an issue. Attention to sources across the board is low in both the US and Australia despite the visible effects of climate change witnessed in both countries in recent years. In contrast, countries like Greece and Chile do not appear to suffer from the same levels of audience polarisation and see higher rates of attention.\n\nFinally, there are also interesting differences in attention to sources by age. Respondents under 35 in many markets are often two or three times more likely to say they pay attention to celebrities, social media personalities, or activists for climate change news than people over 35. So-called ‘green influencers’ have amassed large followings online, drawing attention to the linkages between climate change and social justice issues. They include Vanessa Nakate, a climate activist from Uganda with hundreds of thousands of followers across Twitter and Instagram; Jack Harries, an environmentalist YouTuber whose channel has 3.7m subscribers; and Jerome Foster, a young climate activist who is a White House adviser with over 41,000 Instagram followers.\n\nGroups like the EcoTok collective on TikTok also seek to connect with the younger audience on the platform. Younger respondents are also somewhat more likely to say they get climate change news from smaller or alternative news sources, some of which have been started by young climate activists.\n\n**COVERAGE OF CLIMATE CHANGE**\n\nAudiences appear to be more interested in and pay more attention to climate change news in places where the negative effects are felt more acutely. But it is arguably too late if audiences only pay attention once disasters have already struck. A central challenge is how to cover the issue in a way that draws attention to the causes and decisions that lead to disasters, not ‘just’ the disasters themselves.\n\nOne approach to this is taking a clearer stance on the issue, as exemplified by the _Guardian_ calling for more action on the ‘climate crisis’ and many documentaries that take a clear activist stance on environmental issues. But how do audiences feel about news outlets adopting a position?\n\nWe find that, in general, there are strong regional differences in views. These reflect similar patterns to those already identified in this chapter: audiences in Latin America, Southern Europe, and several markets in Asia-Pacific are more open to news outlets taking a stand in favour of climate change action. Large proportions of respondents in Chile (58%), Portugal (48%), and the Philippines (42%) say that journalists should advocate for change. On the other hand, pluralities in Northern European, Western European, and North American markets favour impartiality. Respondents in Germany (45%), Norway (44%), and the US (42%) say they would prefer news outlets to reflect a range of views and leave it up to them to decide what to think (see chart next page).\n\nExamining these trends, it may be that audiences in markets such as Norway and Germany are predisposed to expect an impartial approach to reporting – particularly from public broadcasters. On the other hand, political polarisation may again be playing a role.\n\nIn many markets, large majorities of those on the political right want journalists to remain impartial on climate change and their views drive some of the overall trends. The proportion of those on the right favouring impartiality is 60% in Norway and 69% in the US. Conversely, those on the political left tend to favour journalists advocating for climate change action. Views in Chile and Portugal are not as polarised on this question, with both those on the political left and right favouring climate action.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 36\n\nPROPORTION WHO PAY ATTENTION TO CELEBRITIES, SOCIAL MEDIA PERSONALITIES, OR ACTIVISTS FOR CLIMATE CHANGE NEWS – BY AGE – SELECTED MARKETS\n\n**C1\\_2022a.** When it comes to climate change news, which of the following sources, if any, do you pay most attention to? Please select all that apply. _Base: U35/35+: Greece = 468/1536, Portugal = 462/1549, USA = 668/1368, Norway = 496/1514, Australia = 561/1477, Japan = 406/1609._\n\n0%\n\n4\n\n8\n\n4 3 5\n\n8\n\n18 16 14 12 12 12\n\nJapan Australia Norway USA Portugal Greece\n\n0%\n\n25%\n\n4\n\n8\n\n4 3\n\n5\n\n8\n\n18\n\n16\n\n14\n\n12 12 12\n\nJapan Australia Norway USA Portugal Greece\n\nU35 35+\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 56","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Digital_News-Report_2022_p56.pdf"]},"response":{"errors":[],"extract_id":"extract_40111e206147de7ff395c036476c4ab9","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["REUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 35\n\nPROPORTION WHO PAY ATTENTION TO EACH SOURCE FOR CLIMATE CHANGE NEWS – SELECTED MARKETS\n\n**C1\\_2022a.** When it comes to climate change news, which of the following sources, if any, do you pay most attention to? _Base: Total sample in each market (n ≈ 2,000)._\n\nMajor news organisations Documentaries\n\nDon’t pay attention to climate change Smaller/alternative news sources\n\n0% 25% 50% 75%\n\n4\n\n4\n\n10\n\n6\n\n20\n\n18\n\n23\n\n15\n\n14\n\n5\n\n22\n\n10\n\n15\n\n15\n\n42\n\n50\n\n27\n\n55\n\n31\n\n33\n\n24\n\n45\n\n43\n\n41\n\n35\n\n28\n\n27\n\n24\n\nDon't pay attention to climate change\n\nSmaller/alternative news sources\n\nDocumentaries\n\nMajor news organisations\n\nUSA\n\nAustralia\n\nNorway\n\nGreece\n\nJapan\n\nPortugal\n\nChile\n\nWhen we drill down into the attention paid to major news organisations, we find that usage for climate change news is lower in markets where interest and attention are polarised, such as the US and Australia. In these markets, along with Norway, we see the largest proportions of people saying they don’t pay attention to climate change at all, with these high figures largely being driven by those on the political right. Half (49%) of those on the political right in the US say they don’t pay attention to climate change at all, and it is a third in Australia (34%) and Norway (31%). This compares to just 5% of those on the political right in Portugal and Chile who say they don’t pay attention.\n\nIndeed, the comparisons between the US, Australia, and other markets are stark, showing the role that polarised politics – and media coverage – can play in driving down interest in and attention to climate change as an issue. Attention to sources across the board is low in both the US and Australia despite the visible effects of climate change witnessed in both countries in recent years. In contrast, countries like Greece and Chile do not appear to suffer from the same levels of audience polarisation and see higher rates of attention.\n\nFinally, there are also interesting differences in attention to sources by age. Respondents under 35 in many markets are often two or three times more likely to say they pay attention to celebrities, social media personalities, or activists for climate change news than people over 35. So-called ‘green influencers’ have amassed large followings online, drawing attention to the linkages between climate change and social justice issues. They include Vanessa Nakate, a climate activist from Uganda with hundreds of thousands of followers across Twitter and Instagram; Jack Harries, an environmentalist YouTuber whose channel has 3.7m subscribers; and Jerome Foster, a young climate activist who is a White House adviser with over 41,000 Instagram followers.\n\nGroups like the EcoTok collective on TikTok also seek to connect with the younger audience on the platform. Younger respondents are also somewhat more likely to say they get climate change news from smaller or alternative news sources, some of which have been started by young climate activists.\n\n**COVERAGE OF CLIMATE CHANGE**\n\nAudiences appear to be more interested in and pay more attention to climate change news in places where the negative effects are felt more acutely. But it is arguably too late if audiences only pay attention once disasters have already struck. A central challenge is how to cover the issue in a way that draws attention to the causes and decisions that lead to disasters, not ‘just’ the disasters themselves.\n\nOne approach to this is taking a clearer stance on the issue, as exemplified by the _Guardian_ calling for more action on the ‘climate crisis’ and many documentaries that take a clear activist stance on environmental issues. But how do audiences feel about news outlets adopting a position?\n\nWe find that, in general, there are strong regional differences in views. These reflect similar patterns to those already identified in this chapter: audiences in Latin America, Southern Europe, and several markets in Asia-Pacific are more open to news outlets taking a stand in favour of climate change action. Large proportions of respondents in Chile (58%), Portugal (48%), and the Philippines (42%) say that journalists should advocate for change. On the other hand, pluralities in Northern European, Western European, and North American markets favour impartiality. Respondents in Germany (45%), Norway (44%), and the US (42%) say they would prefer news outlets to reflect a range of views and leave it up to them to decide what to think (see chart next page).\n\nExamining these trends, it may be that audiences in markets such as Norway and Germany are predisposed to expect an impartial approach to reporting – particularly from public broadcasters. On the other hand, political polarisation may again be playing a role.\n\nIn many markets, large majorities of those on the political right want journalists to remain impartial on climate change and their views drive some of the overall trends. The proportion of those on the right favouring impartiality is 60% in Norway and 69% in the US. Conversely, those on the political left tend to favour journalists advocating for climate change action. Views in Chile and Portugal are not as polarised on this question, with both those on the political left and right favouring climate action.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 36\n\nPROPORTION WHO PAY ATTENTION TO CELEBRITIES, SOCIAL MEDIA PERSONALITIES, OR ACTIVISTS FOR CLIMATE CHANGE NEWS – BY AGE – SELECTED MARKETS\n\n**C1\\_2022a.** When it comes to climate change news, which of the following sources, if any, do you pay most attention to? Please select all that apply. _Base: U35/35+: Greece = 468/1536, Portugal = 462/1549, USA = 668/1368, Norway = 496/1514, Australia = 561/1477, Japan = 406/1609._\n\n0%\n\n4\n\n8\n\n4 3 5\n\n8\n\n18 16 14 12 12 12\n\nJapan Australia Norway USA Portugal Greece\n\n0%\n\n25%\n\n4\n\n8\n\n4 3\n\n5\n\n8\n\n18\n\n16\n\n14\n\n12 12 12\n\nJapan Australia Norway USA Portugal Greece\n\nU35 35+\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 56"],"full_content":"REUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 35\n\nPROPORTION WHO PAY ATTENTION TO EACH SOURCE FOR CLIMATE CHANGE NEWS – SELECTED MARKETS\n\n**C1\\_2022a.** When it comes to climate change news, which of the following sources, if any, do you pay most attention to? _Base: Total sample in each market (n ≈ 2,000)._\n\nMajor news organisations Documentaries\n\nDon’t pay attention to climate change Smaller/alternative news sources\n\n0% 25% 50% 75%\n\n4\n\n4\n\n10\n\n6\n\n20\n\n18\n\n23\n\n15\n\n14\n\n5\n\n22\n\n10\n\n15\n\n15\n\n42\n\n50\n\n27\n\n55\n\n31\n\n33\n\n24\n\n45\n\n43\n\n41\n\n35\n\n28\n\n27\n\n24\n\nDon't pay attention to climate change\n\nSmaller/alternative news sources\n\nDocumentaries\n\nMajor news organisations\n\nUSA\n\nAustralia\n\nNorway\n\nGreece\n\nJapan\n\nPortugal\n\nChile\n\nWhen we drill down into the attention paid to major news organisations, we find that usage for climate change news is lower in markets where interest and attention are polarised, such as the US and Australia. In these markets, along with Norway, we see the largest proportions of people saying they don’t pay attention to climate change at all, with these high figures largely being driven by those on the political right. Half (49%) of those on the political right in the US say they don’t pay attention to climate change at all, and it is a third in Australia (34%) and Norway (31%). This compares to just 5% of those on the political right in Portugal and Chile who say they don’t pay attention.\n\nIndeed, the comparisons between the US, Australia, and other markets are stark, showing the role that polarised politics – and media coverage – can play in driving down interest in and attention to climate change as an issue. Attention to sources across the board is low in both the US and Australia despite the visible effects of climate change witnessed in both countries in recent years. In contrast, countries like Greece and Chile do not appear to suffer from the same levels of audience polarisation and see higher rates of attention.\n\nFinally, there are also interesting differences in attention to sources by age. Respondents under 35 in many markets are often two or three times more likely to say they pay attention to celebrities, social media personalities, or activists for climate change news than people over 35. So-called ‘green influencers’ have amassed large followings online, drawing attention to the linkages between climate change and social justice issues. They include Vanessa Nakate, a climate activist from Uganda with hundreds of thousands of followers across Twitter and Instagram; Jack Harries, an environmentalist YouTuber whose channel has 3.7m subscribers; and Jerome Foster, a young climate activist who is a White House adviser with over 41,000 Instagram followers.\n\nGroups like the EcoTok collective on TikTok also seek to connect with the younger audience on the platform. Younger respondents are also somewhat more likely to say they get climate change news from smaller or alternative news sources, some of which have been started by young climate activists.\n\n**COVERAGE OF CLIMATE CHANGE**\n\nAudiences appear to be more interested in and pay more attention to climate change news in places where the negative effects are felt more acutely. But it is arguably too late if audiences only pay attention once disasters have already struck. A central challenge is how to cover the issue in a way that draws attention to the causes and decisions that lead to disasters, not ‘just’ the disasters themselves.\n\nOne approach to this is taking a clearer stance on the issue, as exemplified by the _Guardian_ calling for more action on the ‘climate crisis’ and many documentaries that take a clear activist stance on environmental issues. But how do audiences feel about news outlets adopting a position?\n\nWe find that, in general, there are strong regional differences in views. These reflect similar patterns to those already identified in this chapter: audiences in Latin America, Southern Europe, and several markets in Asia-Pacific are more open to news outlets taking a stand in favour of climate change action. Large proportions of respondents in Chile (58%), Portugal (48%), and the Philippines (42%) say that journalists should advocate for change. On the other hand, pluralities in Northern European, Western European, and North American markets favour impartiality. Respondents in Germany (45%), Norway (44%), and the US (42%) say they would prefer news outlets to reflect a range of views and leave it up to them to decide what to think (see chart next page).\n\nExamining these trends, it may be that audiences in markets such as Norway and Germany are predisposed to expect an impartial approach to reporting – particularly from public broadcasters. On the other hand, political polarisation may again be playing a role.\n\nIn many markets, large majorities of those on the political right want journalists to remain impartial on climate change and their views drive some of the overall trends. The proportion of those on the right favouring impartiality is 60% in Norway and 69% in the US. Conversely, those on the political left tend to favour journalists advocating for climate change action. Views in Chile and Portugal are not as polarised on this question, with both those on the political left and right favouring climate action.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 36\n\nPROPORTION WHO PAY ATTENTION TO CELEBRITIES, SOCIAL MEDIA PERSONALITIES, OR ACTIVISTS FOR CLIMATE CHANGE NEWS – BY AGE – SELECTED MARKETS\n\n**C1\\_2022a.** When it comes to climate change news, which of the following sources, if any, do you pay most attention to? Please select all that apply. _Base: U35/35+: Greece = 468/1536, Portugal = 462/1549, USA = 668/1368, Norway = 496/1514, Australia = 561/1477, Japan = 406/1609._\n\n0%\n\n4\n\n8\n\n4 3 5\n\n8\n\n18 16 14 12 12 12\n\nJapan Australia Norway USA Portugal Greece\n\n0%\n\n25%\n\n4\n\n8\n\n4 3\n\n5\n\n8\n\n18\n\n16\n\n14\n\n12 12 12\n\nJapan Australia Norway USA Portugal Greece\n\nU35 35+\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 56","publish_date":null,"title":"Graph 35 Graph 36","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Digital_News-Report_2022_p56.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Digital_News-Report_2022_p56.pdf"},"request":{"example_id":"chart/Digital_News-Report_2022_p56","product_type":"parse","source_file_path":"Digital_News-Report_2022_p56.pdf"},"started_at":"2026-11-04T19:46:39.088303"} +{"completed_at":"2026-11-04T19:46:43.392129","latency_in_ms":2821,"output":{"example_id":"chart/Digital_News-Report_2022_p57","grounded_pages":[],"layout_pages":[],"markdown":"Notably, these polarised views on impartiality are reflective of trends we identified in the _Digital News Report 2021_ where, on the question of impartiality, audiences split by their political views. They also split by age, and the same is true this year: younger people are more likely to think news outlets should take a clear position in favour of climate change action. Across all markets, 43% of those aged 18–24 think this, compared to just 34% of over 55s. Those over 55 instead favour news outlets taking an impartial position.\n\n**CONCLUSIONS**\n\nOur findings show that trends around climate change news are driven by factors such as politics, age, and the impacts of climate change itself. Interest in and attention to climate change, as a news topic, is much more politically polarised in some countries than others, with many of those on the political right tuning out. Conservative politicians and alternative media have played key roles in this. For newsrooms, the difficult question is how to engage this critical segment of the audience, particularly if conservatives view climate change as a politicised issue of the ‘left’ and disfavour newsrooms taking a stand.\n\nAre there other steps journalists can take to make climate change stories feel more relevant? Perhaps lessons could be learned from the wide appeal of many environmental documentaries, which provide clear stories and engaging visuals. These narratives help audiences connect with what is a very large and sometimes abstract story – while not necessarily being ‘political’. But at the same time, some documentaries have been accused of selectively using facts, and a ‘slick-but-lacking-substance’ approach may backfire among those not already convinced. Yet there is always room for better storytelling.\n\nBut what about taking a more active stance? Among those on the political left, while they are more interested and engaged, there is still a large proportion in most markets who do not express interest in climate change news and who are not paying attention to major news organisations. For this audience, taking a clearer stance may be one way to increase interest or attention, but doing so risks running up against journalistic norms of impartiality adopted in many countries.\n\nIn the UK, the _Guardian_ has taken a stand and this may win the favour of younger (and left-leaning) audiences who are more inclined to want this. Engaging younger audiences on this issue means reaching them with accurate and compelling stories in the digital spaces they occupy – where influencers and activists make calls to action. At the same time, taking a clear stand may risk further alienating already disengaged audiences, further entrenching divisions around the issue.\n\nWhat seems apparent, however, is that audiences express more interest, pay more attention to climate change news, and feel more inclined to support journalists taking a stand in places where the direct impacts of climate change are acutely felt. But if the very visible manifestations of climate change – fires, floods, and droughts – are what draw attention, this raises a problem in itself. Action – and, therefore, credible information to inform it – is needed before disaster. Yet long-term trends, future unseen risks, and the scientific complexities of climate change are harder to engage audiences with, especially when other immediate crises, from inflation to war in Ukraine, are on the doorstep. Climate change is a difficult news topic to cover and it is not clear that there is a one-size-fits-all approach to it.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 37\n\nPROPORTION WHO THINK NEWS OUTLETS SHOULD TAKE A STAND WHEN COVERING CLIMATE CHANGE – SELECTED MARKETS\n\n**QAdvocacy\\_2022.** Thinking about the way news outlets cover climate change, which of the following statements best describes your view? News outlets should… _Base: Total sample in each market (n ≈ 2,000)._\n\n0% 25% 50% 75% 100%\n\nChile 58 22 9 11\n\nPortugal 48 21 15 16\n\nPhilippines 42 29 21 8\n\nGermany 30 45 8 16\n\nNorway 27 44 6 22\n\nUSA 26 42 8 23\n\nTake a clear position in favour of climate change action\n\nReflect a range of views and leave it up to people to decide\n\nTake a clear position against climate change action\n\nDon’t know\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 38\n\nPROPORTION WHO THINK NEWS OUTLETS SHOULD TAKE A STAND WHEN COVERING CLIMATE CHANGE – BY AGE – ALL MARKETS\n\n**QAdvocacy\\_2022.** Thinking about the way news outlets cover climate change, which of the following statements best describes your view? News outlets should… _Base: 18-24/25-34/35-44/ 45-54/55+: 11,304/16,862/17,205/15,074/32,987._\n\n0% 25% 50%\n\nTake a clear position in favour of climate change action\n\nReflect a range of views on climate change and leave it up to people to decide\n\nTake a clear position against climate change action\n\n43\n\n27\n\n14\n\n40\n\n29\n\n14\n\n37\n\n33\n\n13\n\n37\n\n35\n\n11\n\n34\n\n43\n\n9\n\n18-24 25-34 35-44 45-54 55+\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 57","pages":[{"line_numbers":[],"markdown":"Notably, these polarised views on impartiality are reflective of trends we identified in the _Digital News Report 2021_ where, on the question of impartiality, audiences split by their political views. They also split by age, and the same is true this year: younger people are more likely to think news outlets should take a clear position in favour of climate change action. Across all markets, 43% of those aged 18–24 think this, compared to just 34% of over 55s. Those over 55 instead favour news outlets taking an impartial position.\n\n**CONCLUSIONS**\n\nOur findings show that trends around climate change news are driven by factors such as politics, age, and the impacts of climate change itself. Interest in and attention to climate change, as a news topic, is much more politically polarised in some countries than others, with many of those on the political right tuning out. Conservative politicians and alternative media have played key roles in this. For newsrooms, the difficult question is how to engage this critical segment of the audience, particularly if conservatives view climate change as a politicised issue of the ‘left’ and disfavour newsrooms taking a stand.\n\nAre there other steps journalists can take to make climate change stories feel more relevant? Perhaps lessons could be learned from the wide appeal of many environmental documentaries, which provide clear stories and engaging visuals. These narratives help audiences connect with what is a very large and sometimes abstract story – while not necessarily being ‘political’. But at the same time, some documentaries have been accused of selectively using facts, and a ‘slick-but-lacking-substance’ approach may backfire among those not already convinced. Yet there is always room for better storytelling.\n\nBut what about taking a more active stance? Among those on the political left, while they are more interested and engaged, there is still a large proportion in most markets who do not express interest in climate change news and who are not paying attention to major news organisations. For this audience, taking a clearer stance may be one way to increase interest or attention, but doing so risks running up against journalistic norms of impartiality adopted in many countries.\n\nIn the UK, the _Guardian_ has taken a stand and this may win the favour of younger (and left-leaning) audiences who are more inclined to want this. Engaging younger audiences on this issue means reaching them with accurate and compelling stories in the digital spaces they occupy – where influencers and activists make calls to action. At the same time, taking a clear stand may risk further alienating already disengaged audiences, further entrenching divisions around the issue.\n\nWhat seems apparent, however, is that audiences express more interest, pay more attention to climate change news, and feel more inclined to support journalists taking a stand in places where the direct impacts of climate change are acutely felt. But if the very visible manifestations of climate change – fires, floods, and droughts – are what draw attention, this raises a problem in itself. Action – and, therefore, credible information to inform it – is needed before disaster. Yet long-term trends, future unseen risks, and the scientific complexities of climate change are harder to engage audiences with, especially when other immediate crises, from inflation to war in Ukraine, are on the doorstep. Climate change is a difficult news topic to cover and it is not clear that there is a one-size-fits-all approach to it.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 37\n\nPROPORTION WHO THINK NEWS OUTLETS SHOULD TAKE A STAND WHEN COVERING CLIMATE CHANGE – SELECTED MARKETS\n\n**QAdvocacy\\_2022.** Thinking about the way news outlets cover climate change, which of the following statements best describes your view? News outlets should… _Base: Total sample in each market (n ≈ 2,000)._\n\n0% 25% 50% 75% 100%\n\nChile 58 22 9 11\n\nPortugal 48 21 15 16\n\nPhilippines 42 29 21 8\n\nGermany 30 45 8 16\n\nNorway 27 44 6 22\n\nUSA 26 42 8 23\n\nTake a clear position in favour of climate change action\n\nReflect a range of views and leave it up to people to decide\n\nTake a clear position against climate change action\n\nDon’t know\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 38\n\nPROPORTION WHO THINK NEWS OUTLETS SHOULD TAKE A STAND WHEN COVERING CLIMATE CHANGE – BY AGE – ALL MARKETS\n\n**QAdvocacy\\_2022.** Thinking about the way news outlets cover climate change, which of the following statements best describes your view? News outlets should… _Base: 18-24/25-34/35-44/ 45-54/55+: 11,304/16,862/17,205/15,074/32,987._\n\n0% 25% 50%\n\nTake a clear position in favour of climate change action\n\nReflect a range of views on climate change and leave it up to people to decide\n\nTake a clear position against climate change action\n\n43\n\n27\n\n14\n\n40\n\n29\n\n14\n\n37\n\n33\n\n13\n\n37\n\n35\n\n11\n\n34\n\n43\n\n9\n\n18-24 25-34 35-44 45-54 55+\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 57","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Notably, these polarised views on impartiality are reflective of trends we identified in the _Digital News Report 2021_ where, on the question of impartiality, audiences split by their political views. They also split by age, and the same is true this year: younger people are more likely to think news outlets should take a clear position in favour of climate change action. Across all markets, 43% of those aged 18–24 think this, compared to just 34% of over 55s. Those over 55 instead favour news outlets taking an impartial position.\n\n**CONCLUSIONS**\n\nOur findings show that trends around climate change news are driven by factors such as politics, age, and the impacts of climate change itself. Interest in and attention to climate change, as a news topic, is much more politically polarised in some countries than others, with many of those on the political right tuning out. Conservative politicians and alternative media have played key roles in this. For newsrooms, the difficult question is how to engage this critical segment of the audience, particularly if conservatives view climate change as a politicised issue of the ‘left’ and disfavour newsrooms taking a stand.\n\nAre there other steps journalists can take to make climate change stories feel more relevant? Perhaps lessons could be learned from the wide appeal of many environmental documentaries, which provide clear stories and engaging visuals. These narratives help audiences connect with what is a very large and sometimes abstract story – while not necessarily being ‘political’. But at the same time, some documentaries have been accused of selectively using facts, and a ‘slick-but-lacking-substance’ approach may backfire among those not already convinced. Yet there is always room for better storytelling.\n\nBut what about taking a more active stance? Among those on the political left, while they are more interested and engaged, there is still a large proportion in most markets who do not express interest in climate change news and who are not paying attention to major news organisations. For this audience, taking a clearer stance may be one way to increase interest or attention, but doing so risks running up against journalistic norms of impartiality adopted in many countries.\n\nIn the UK, the _Guardian_ has taken a stand and this may win the favour of younger (and left-leaning) audiences who are more inclined to want this. Engaging younger audiences on this issue means reaching them with accurate and compelling stories in the digital spaces they occupy – where influencers and activists make calls to action. At the same time, taking a clear stand may risk further alienating already disengaged audiences, further entrenching divisions around the issue.\n\nWhat seems apparent, however, is that audiences express more interest, pay more attention to climate change news, and feel more inclined to support journalists taking a stand in places where the direct impacts of climate change are acutely felt. But if the very visible manifestations of climate change – fires, floods, and droughts – are what draw attention, this raises a problem in itself. Action – and, therefore, credible information to inform it – is needed before disaster. Yet long-term trends, future unseen risks, and the scientific complexities of climate change are harder to engage audiences with, especially when other immediate crises, from inflation to war in Ukraine, are on the doorstep. Climate change is a difficult news topic to cover and it is not clear that there is a one-size-fits-all approach to it.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 37\n\nPROPORTION WHO THINK NEWS OUTLETS SHOULD TAKE A STAND WHEN COVERING CLIMATE CHANGE – SELECTED MARKETS\n\n**QAdvocacy\\_2022.** Thinking about the way news outlets cover climate change, which of the following statements best describes your view? News outlets should… _Base: Total sample in each market (n ≈ 2,000)._\n\n0% 25% 50% 75% 100%\n\nChile 58 22 9 11\n\nPortugal 48 21 15 16\n\nPhilippines 42 29 21 8\n\nGermany 30 45 8 16\n\nNorway 27 44 6 22\n\nUSA 26 42 8 23\n\nTake a clear position in favour of climate change action\n\nReflect a range of views and leave it up to people to decide\n\nTake a clear position against climate change action\n\nDon’t know\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 38\n\nPROPORTION WHO THINK NEWS OUTLETS SHOULD TAKE A STAND WHEN COVERING CLIMATE CHANGE – BY AGE – ALL MARKETS\n\n**QAdvocacy\\_2022.** Thinking about the way news outlets cover climate change, which of the following statements best describes your view? News outlets should… _Base: 18-24/25-34/35-44/ 45-54/55+: 11,304/16,862/17,205/15,074/32,987._\n\n0% 25% 50%\n\nTake a clear position in favour of climate change action\n\nReflect a range of views on climate change and leave it up to people to decide\n\nTake a clear position against climate change action\n\n43\n\n27\n\n14\n\n40\n\n29\n\n14\n\n37\n\n33\n\n13\n\n37\n\n35\n\n11\n\n34\n\n43\n\n9\n\n18-24 25-34 35-44 45-54 55+\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 57","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Digital_News-Report_2022_p57.pdf"]},"response":{"errors":[],"extract_id":"extract_60205b70e2100ff07644a975458bc93e","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Notably, these polarised views on impartiality are reflective of trends we identified in the _Digital News Report 2021_ where, on the question of impartiality, audiences split by their political views. They also split by age, and the same is true this year: younger people are more likely to think news outlets should take a clear position in favour of climate change action. Across all markets, 43% of those aged 18–24 think this, compared to just 34% of over 55s. Those over 55 instead favour news outlets taking an impartial position.\n\n**CONCLUSIONS**\n\nOur findings show that trends around climate change news are driven by factors such as politics, age, and the impacts of climate change itself. Interest in and attention to climate change, as a news topic, is much more politically polarised in some countries than others, with many of those on the political right tuning out. Conservative politicians and alternative media have played key roles in this. For newsrooms, the difficult question is how to engage this critical segment of the audience, particularly if conservatives view climate change as a politicised issue of the ‘left’ and disfavour newsrooms taking a stand.\n\nAre there other steps journalists can take to make climate change stories feel more relevant? Perhaps lessons could be learned from the wide appeal of many environmental documentaries, which provide clear stories and engaging visuals. These narratives help audiences connect with what is a very large and sometimes abstract story – while not necessarily being ‘political’. But at the same time, some documentaries have been accused of selectively using facts, and a ‘slick-but-lacking-substance’ approach may backfire among those not already convinced. Yet there is always room for better storytelling.\n\nBut what about taking a more active stance? Among those on the political left, while they are more interested and engaged, there is still a large proportion in most markets who do not express interest in climate change news and who are not paying attention to major news organisations. For this audience, taking a clearer stance may be one way to increase interest or attention, but doing so risks running up against journalistic norms of impartiality adopted in many countries.\n\nIn the UK, the _Guardian_ has taken a stand and this may win the favour of younger (and left-leaning) audiences who are more inclined to want this. Engaging younger audiences on this issue means reaching them with accurate and compelling stories in the digital spaces they occupy – where influencers and activists make calls to action. At the same time, taking a clear stand may risk further alienating already disengaged audiences, further entrenching divisions around the issue.\n\nWhat seems apparent, however, is that audiences express more interest, pay more attention to climate change news, and feel more inclined to support journalists taking a stand in places where the direct impacts of climate change are acutely felt. But if the very visible manifestations of climate change – fires, floods, and droughts – are what draw attention, this raises a problem in itself. Action – and, therefore, credible information to inform it – is needed before disaster. Yet long-term trends, future unseen risks, and the scientific complexities of climate change are harder to engage audiences with, especially when other immediate crises, from inflation to war in Ukraine, are on the doorstep. Climate change is a difficult news topic to cover and it is not clear that there is a one-size-fits-all approach to it.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 37\n\nPROPORTION WHO THINK NEWS OUTLETS SHOULD TAKE A STAND WHEN COVERING CLIMATE CHANGE – SELECTED MARKETS\n\n**QAdvocacy\\_2022.** Thinking about the way news outlets cover climate change, which of the following statements best describes your view? News outlets should… _Base: Total sample in each market (n ≈ 2,000)._\n\n0% 25% 50% 75% 100%\n\nChile 58 22 9 11\n\nPortugal 48 21 15 16\n\nPhilippines 42 29 21 8\n\nGermany 30 45 8 16\n\nNorway 27 44 6 22\n\nUSA 26 42 8 23\n\nTake a clear position in favour of climate change action\n\nReflect a range of views and leave it up to people to decide\n\nTake a clear position against climate change action\n\nDon’t know\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 38\n\nPROPORTION WHO THINK NEWS OUTLETS SHOULD TAKE A STAND WHEN COVERING CLIMATE CHANGE – BY AGE – ALL MARKETS\n\n**QAdvocacy\\_2022.** Thinking about the way news outlets cover climate change, which of the following statements best describes your view? News outlets should… _Base: 18-24/25-34/35-44/ 45-54/55+: 11,304/16,862/17,205/15,074/32,987._\n\n0% 25% 50%\n\nTake a clear position in favour of climate change action\n\nReflect a range of views on climate change and leave it up to people to decide\n\nTake a clear position against climate change action\n\n43\n\n27\n\n14\n\n40\n\n29\n\n14\n\n37\n\n33\n\n13\n\n37\n\n35\n\n11\n\n34\n\n43\n\n9\n\n18-24 25-34 35-44 45-54 55+\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 57"],"full_content":"Notably, these polarised views on impartiality are reflective of trends we identified in the _Digital News Report 2021_ where, on the question of impartiality, audiences split by their political views. They also split by age, and the same is true this year: younger people are more likely to think news outlets should take a clear position in favour of climate change action. Across all markets, 43% of those aged 18–24 think this, compared to just 34% of over 55s. Those over 55 instead favour news outlets taking an impartial position.\n\n**CONCLUSIONS**\n\nOur findings show that trends around climate change news are driven by factors such as politics, age, and the impacts of climate change itself. Interest in and attention to climate change, as a news topic, is much more politically polarised in some countries than others, with many of those on the political right tuning out. Conservative politicians and alternative media have played key roles in this. For newsrooms, the difficult question is how to engage this critical segment of the audience, particularly if conservatives view climate change as a politicised issue of the ‘left’ and disfavour newsrooms taking a stand.\n\nAre there other steps journalists can take to make climate change stories feel more relevant? Perhaps lessons could be learned from the wide appeal of many environmental documentaries, which provide clear stories and engaging visuals. These narratives help audiences connect with what is a very large and sometimes abstract story – while not necessarily being ‘political’. But at the same time, some documentaries have been accused of selectively using facts, and a ‘slick-but-lacking-substance’ approach may backfire among those not already convinced. Yet there is always room for better storytelling.\n\nBut what about taking a more active stance? Among those on the political left, while they are more interested and engaged, there is still a large proportion in most markets who do not express interest in climate change news and who are not paying attention to major news organisations. For this audience, taking a clearer stance may be one way to increase interest or attention, but doing so risks running up against journalistic norms of impartiality adopted in many countries.\n\nIn the UK, the _Guardian_ has taken a stand and this may win the favour of younger (and left-leaning) audiences who are more inclined to want this. Engaging younger audiences on this issue means reaching them with accurate and compelling stories in the digital spaces they occupy – where influencers and activists make calls to action. At the same time, taking a clear stand may risk further alienating already disengaged audiences, further entrenching divisions around the issue.\n\nWhat seems apparent, however, is that audiences express more interest, pay more attention to climate change news, and feel more inclined to support journalists taking a stand in places where the direct impacts of climate change are acutely felt. But if the very visible manifestations of climate change – fires, floods, and droughts – are what draw attention, this raises a problem in itself. Action – and, therefore, credible information to inform it – is needed before disaster. Yet long-term trends, future unseen risks, and the scientific complexities of climate change are harder to engage audiences with, especially when other immediate crises, from inflation to war in Ukraine, are on the doorstep. Climate change is a difficult news topic to cover and it is not clear that there is a one-size-fits-all approach to it.\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 37\n\nPROPORTION WHO THINK NEWS OUTLETS SHOULD TAKE A STAND WHEN COVERING CLIMATE CHANGE – SELECTED MARKETS\n\n**QAdvocacy\\_2022.** Thinking about the way news outlets cover climate change, which of the following statements best describes your view? News outlets should… _Base: Total sample in each market (n ≈ 2,000)._\n\n0% 25% 50% 75% 100%\n\nChile 58 22 9 11\n\nPortugal 48 21 15 16\n\nPhilippines 42 29 21 8\n\nGermany 30 45 8 16\n\nNorway 27 44 6 22\n\nUSA 26 42 8 23\n\nTake a clear position in favour of climate change action\n\nReflect a range of views and leave it up to people to decide\n\nTake a clear position against climate change action\n\nDon’t know\n\nREUTERS INSTITUTE FOR THE STUDY OF JOURNALISM / FURTHER ANALYSIS GRAPHS\n\n## Graph 38\n\nPROPORTION WHO THINK NEWS OUTLETS SHOULD TAKE A STAND WHEN COVERING CLIMATE CHANGE – BY AGE – ALL MARKETS\n\n**QAdvocacy\\_2022.** Thinking about the way news outlets cover climate change, which of the following statements best describes your view? News outlets should… _Base: 18-24/25-34/35-44/ 45-54/55+: 11,304/16,862/17,205/15,074/32,987._\n\n0% 25% 50%\n\nTake a clear position in favour of climate change action\n\nReflect a range of views on climate change and leave it up to people to decide\n\nTake a clear position against climate change action\n\n43\n\n27\n\n14\n\n40\n\n29\n\n14\n\n37\n\n33\n\n13\n\n37\n\n35\n\n11\n\n34\n\n43\n\n9\n\n18-24 25-34 35-44 45-54 55+\n\nReuters Institute for the Study of Journalism | Digital News Report 2022 57","publish_date":null,"title":"Graph 37 Graph 38","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Digital_News-Report_2022_p57.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Digital_News-Report_2022_p57.pdf"},"request":{"example_id":"chart/Digital_News-Report_2022_p57","product_type":"parse","source_file_path":"Digital_News-Report_2022_p57.pdf"},"started_at":"2026-11-04T19:46:40.570203"} +{"completed_at":"2026-11-04T19:46:43.990222","latency_in_ms":1849,"output":{"example_id":"chart/Digital_News-Report_2022_p63","grounded_pages":[],"layout_pages":[],"markdown":"63\n\n59%\n\n74%\n\n20%\n\n79%\n\n53%\n\n73%\n\n17%\n\n38%\n\n0%\n\n50%\n\n100%\n\n2022 2019 2020 2021 2018 2017 2016 2015 2013 2014\n\n67%\n\n29%\n\n16%\n\n40%\n\n64%\n\n22%\n\n0%\n\n50%\n\n100%\n\n2022 2019 2020 2021 2018 2017 2016 2015 2014 2013\n\n10\n\n11\n\n6\n\n5\n\n5\n\n2\n\n5\n\n3\n\n2\n\n3\n\n3\n\nThe ‘i’ newspaper\n\nDaily Mirror (& Sunday)\n\nDaily Telegraph (& Sunday)\n\nChannel 5 News\n\nGuardian/Observer\n\nSun/Sun on Sunday\n\nThe Times/Sunday Times\n\nGB News\n\nMetro\n\nRegional or local newspaper\n\nCommercial radio news\n\nDaily Mail/Mail on Sunday\n\nChannel 4 News\n\nSky News\n\nITV News\n\nBBC News (TV & radio) 50\n\n27\n\n15\n\n9\n\n9\n\n7\n\n7\n\n5\n\n5\n\n5\n\n5\n\n4\n\n3\n\n3\n\n3\n\n2\n\n10\n\n7\n\n5\n\n4\n\n4\n\n3\n\n3\n\n3\n\n2\n\n4\n\n3\n\n4\n\nBuzzFeed News\n\nYahoo! News\n\nThe Times online\n\nMSN News\n\nMetro online\n\nITV News online\n\nHufPost\n\nSun online\n\nMirror online\n\nIndependent/ i100 online\n\nTelegraph online\n\nRegional or local newspaper website\n\nMailOnline\n\nSky News online\n\nGuardian online\n\nBBC News online 43\n\n18\n\n13\n\n12\n\n9\n\n6\n\n5\n\n5\n\n5\n\n5\n\n5\n\n5\n\n4\n\n4\n\n4\n\n4\n\nDigital News Report 2022 | United Kingdom 63\n\nTV, RADIO AND PRINT ONLINE\n\n**TOP BRANDS** % Weekly usage\n\n**Weekly use** TV, radio & print\n\n**More than 3 days per week** TV, radio & print\n\n**Weekly use** online brands\n\n**More than 3 days per week** online brands\n\n**WEEKLY REACH OFFLINE AND ONLINE**\n\nSOURCES OF NEWS 2013–22\n\nDEVICES FOR NEWS 2013–22\n\nOnline (incl. social media) TV Social media Print\n\n**CHANGING MEDIA**\n\nAudiences for traditional news sources like TV and print have fallen significantly over the last decade, with younger groups preferring to get their news online or via social media. After last year’s COVID-19 bump, overall news consumption is down across all devices.\n\n# 9%\n\npay for **ONLINE NEWS**\n\n# 25%\n\nlisten to **PODCASTS** in the last month\n\n**Rank Brand For News For All**\n\n**1** Facebook 19% (-4) 62%\n\n**2** Twitter 13% (-3) 28%\n\n**3** WhatsApp 10% (-4) 66%\n\n**Rank Brand For News For All**\n\n**4** YouTube 9% (-1) 54%\n\n**5** Instagram 6% (+1) 36%\n\n**6** Facebook Messenger 4% (-3) 43%\n\nTOP SOCIAL MEDIA AND MESSAGING\n\n# 21%\n\n**SHARE NEWS** via social, messaging or email\n\nBBC TV News: 44%\n\nBBC Radio News: 18%\n\n# **20%**\n\n34% in 2017\n\n# **20%**\n\n29% in 2017\n\nBRAND TRUST SCORES\n\nTrust Neither Don’t trust\n\nOVERALL TRUST SCORE\n\nMEDIA IS FREE FROM ...\n\n**TRUST**\n\nPublic broadcasters such as the BBC, ITV, and Channel 4 remain the most trusted news brands. Popular outlets such as the _Sun_ , the _Mail_ and the _Mirror_ attract big audiences online but their news coverage is often strongly distrusted. New TV channel GB News is struggling to build trust with general audiences even as it builds loyalty with some on the right.\n\nTrust = % scored 6–10 on 10-point scale, Don’t trust = 0–4, Neither = 5. Those that\n\nhaven’t heard of each brand were excluded. Only the above brands were included in\n\nthe survey so should not be treated as a list of the most trusted brands.\n\n0% 100%\n\n26\n\n18\n\n16\n\n14\n\n16\n\n22\n\n24\n\n22\n\n20\n\n29\n\n39\n\n29\n\n51\n\n49\n\n67\n\n55\n\n55\n\n54\n\n52\n\n52\n\n48\n\n45\n\n43\n\n42\n\n36\n\n27\n\n24\n\n23\n\n22\n\n12 Sun\n\nDaily Mirror\n\nDaily Mail/MailOnline\n\nHufPost\n\nGB News\n\nDaily Telegraph\n\nIndependent\n\nThe Times\n\nSky News\n\nGuardian\n\nLocal or regional newspaper\n\nFinancial Times\n\nChannel 4 News\n\nITV News\n\nBBC News _Change over time 2015-2022_\n\n_Undue political_\n\n_influence_\n\n_Undue business_\n\n_influence_\n\n51%\n\n0%\n\n25%\n\n50%\n\n75%\n\n2022 2015\n\n# **34%**\n\n=36/46 markets","pages":[{"line_numbers":[],"markdown":"63\n\n59%\n\n74%\n\n20%\n\n79%\n\n53%\n\n73%\n\n17%\n\n38%\n\n0%\n\n50%\n\n100%\n\n2022 2019 2020 2021 2018 2017 2016 2015 2013 2014\n\n67%\n\n29%\n\n16%\n\n40%\n\n64%\n\n22%\n\n0%\n\n50%\n\n100%\n\n2022 2019 2020 2021 2018 2017 2016 2015 2014 2013\n\n10\n\n11\n\n6\n\n5\n\n5\n\n2\n\n5\n\n3\n\n2\n\n3\n\n3\n\nThe ‘i’ newspaper\n\nDaily Mirror (& Sunday)\n\nDaily Telegraph (& Sunday)\n\nChannel 5 News\n\nGuardian/Observer\n\nSun/Sun on Sunday\n\nThe Times/Sunday Times\n\nGB News\n\nMetro\n\nRegional or local newspaper\n\nCommercial radio news\n\nDaily Mail/Mail on Sunday\n\nChannel 4 News\n\nSky News\n\nITV News\n\nBBC News (TV & radio) 50\n\n27\n\n15\n\n9\n\n9\n\n7\n\n7\n\n5\n\n5\n\n5\n\n5\n\n4\n\n3\n\n3\n\n3\n\n2\n\n10\n\n7\n\n5\n\n4\n\n4\n\n3\n\n3\n\n3\n\n2\n\n4\n\n3\n\n4\n\nBuzzFeed News\n\nYahoo! News\n\nThe Times online\n\nMSN News\n\nMetro online\n\nITV News online\n\nHufPost\n\nSun online\n\nMirror online\n\nIndependent/ i100 online\n\nTelegraph online\n\nRegional or local newspaper website\n\nMailOnline\n\nSky News online\n\nGuardian online\n\nBBC News online 43\n\n18\n\n13\n\n12\n\n9\n\n6\n\n5\n\n5\n\n5\n\n5\n\n5\n\n5\n\n4\n\n4\n\n4\n\n4\n\nDigital News Report 2022 | United Kingdom 63\n\nTV, RADIO AND PRINT ONLINE\n\n**TOP BRANDS** % Weekly usage\n\n**Weekly use** TV, radio & print\n\n**More than 3 days per week** TV, radio & print\n\n**Weekly use** online brands\n\n**More than 3 days per week** online brands\n\n**WEEKLY REACH OFFLINE AND ONLINE**\n\nSOURCES OF NEWS 2013–22\n\nDEVICES FOR NEWS 2013–22\n\nOnline (incl. social media) TV Social media Print\n\n**CHANGING MEDIA**\n\nAudiences for traditional news sources like TV and print have fallen significantly over the last decade, with younger groups preferring to get their news online or via social media. After last year’s COVID-19 bump, overall news consumption is down across all devices.\n\n# 9%\n\npay for **ONLINE NEWS**\n\n# 25%\n\nlisten to **PODCASTS** in the last month\n\n**Rank Brand For News For All**\n\n**1** Facebook 19% (-4) 62%\n\n**2** Twitter 13% (-3) 28%\n\n**3** WhatsApp 10% (-4) 66%\n\n**Rank Brand For News For All**\n\n**4** YouTube 9% (-1) 54%\n\n**5** Instagram 6% (+1) 36%\n\n**6** Facebook Messenger 4% (-3) 43%\n\nTOP SOCIAL MEDIA AND MESSAGING\n\n# 21%\n\n**SHARE NEWS** via social, messaging or email\n\nBBC TV News: 44%\n\nBBC Radio News: 18%\n\n# **20%**\n\n34% in 2017\n\n# **20%**\n\n29% in 2017\n\nBRAND TRUST SCORES\n\nTrust Neither Don’t trust\n\nOVERALL TRUST SCORE\n\nMEDIA IS FREE FROM ...\n\n**TRUST**\n\nPublic broadcasters such as the BBC, ITV, and Channel 4 remain the most trusted news brands. Popular outlets such as the _Sun_ , the _Mail_ and the _Mirror_ attract big audiences online but their news coverage is often strongly distrusted. New TV channel GB News is struggling to build trust with general audiences even as it builds loyalty with some on the right.\n\nTrust = % scored 6–10 on 10-point scale, Don’t trust = 0–4, Neither = 5. Those that\n\nhaven’t heard of each brand were excluded. Only the above brands were included in\n\nthe survey so should not be treated as a list of the most trusted brands.\n\n0% 100%\n\n26\n\n18\n\n16\n\n14\n\n16\n\n22\n\n24\n\n22\n\n20\n\n29\n\n39\n\n29\n\n51\n\n49\n\n67\n\n55\n\n55\n\n54\n\n52\n\n52\n\n48\n\n45\n\n43\n\n42\n\n36\n\n27\n\n24\n\n23\n\n22\n\n12 Sun\n\nDaily Mirror\n\nDaily Mail/MailOnline\n\nHufPost\n\nGB News\n\nDaily Telegraph\n\nIndependent\n\nThe Times\n\nSky News\n\nGuardian\n\nLocal or regional newspaper\n\nFinancial Times\n\nChannel 4 News\n\nITV News\n\nBBC News _Change over time 2015-2022_\n\n_Undue political_\n\n_influence_\n\n_Undue business_\n\n_influence_\n\n51%\n\n0%\n\n25%\n\n50%\n\n75%\n\n2022 2015\n\n# **34%**\n\n=36/46 markets","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"63\n\n59%\n\n74%\n\n20%\n\n79%\n\n53%\n\n73%\n\n17%\n\n38%\n\n0%\n\n50%\n\n100%\n\n2022 2019 2020 2021 2018 2017 2016 2015 2013 2014\n\n67%\n\n29%\n\n16%\n\n40%\n\n64%\n\n22%\n\n0%\n\n50%\n\n100%\n\n2022 2019 2020 2021 2018 2017 2016 2015 2014 2013\n\n10\n\n11\n\n6\n\n5\n\n5\n\n2\n\n5\n\n3\n\n2\n\n3\n\n3\n\nThe ‘i’ newspaper\n\nDaily Mirror (& Sunday)\n\nDaily Telegraph (& Sunday)\n\nChannel 5 News\n\nGuardian/Observer\n\nSun/Sun on Sunday\n\nThe Times/Sunday Times\n\nGB News\n\nMetro\n\nRegional or local newspaper\n\nCommercial radio news\n\nDaily Mail/Mail on Sunday\n\nChannel 4 News\n\nSky News\n\nITV News\n\nBBC News (TV & radio) 50\n\n27\n\n15\n\n9\n\n9\n\n7\n\n7\n\n5\n\n5\n\n5\n\n5\n\n4\n\n3\n\n3\n\n3\n\n2\n\n10\n\n7\n\n5\n\n4\n\n4\n\n3\n\n3\n\n3\n\n2\n\n4\n\n3\n\n4\n\nBuzzFeed News\n\nYahoo! News\n\nThe Times online\n\nMSN News\n\nMetro online\n\nITV News online\n\nHufPost\n\nSun online\n\nMirror online\n\nIndependent/ i100 online\n\nTelegraph online\n\nRegional or local newspaper website\n\nMailOnline\n\nSky News online\n\nGuardian online\n\nBBC News online 43\n\n18\n\n13\n\n12\n\n9\n\n6\n\n5\n\n5\n\n5\n\n5\n\n5\n\n5\n\n4\n\n4\n\n4\n\n4\n\nDigital News Report 2022 | United Kingdom 63\n\nTV, RADIO AND PRINT ONLINE\n\n**TOP BRANDS** % Weekly usage\n\n**Weekly use** TV, radio & print\n\n**More than 3 days per week** TV, radio & print\n\n**Weekly use** online brands\n\n**More than 3 days per week** online brands\n\n**WEEKLY REACH OFFLINE AND ONLINE**\n\nSOURCES OF NEWS 2013–22\n\nDEVICES FOR NEWS 2013–22\n\nOnline (incl. social media) TV Social media Print\n\n**CHANGING MEDIA**\n\nAudiences for traditional news sources like TV and print have fallen significantly over the last decade, with younger groups preferring to get their news online or via social media. After last year’s COVID-19 bump, overall news consumption is down across all devices.\n\n# 9%\n\npay for **ONLINE NEWS**\n\n# 25%\n\nlisten to **PODCASTS** in the last month\n\n**Rank Brand For News For All**\n\n**1** Facebook 19% (-4) 62%\n\n**2** Twitter 13% (-3) 28%\n\n**3** WhatsApp 10% (-4) 66%\n\n**Rank Brand For News For All**\n\n**4** YouTube 9% (-1) 54%\n\n**5** Instagram 6% (+1) 36%\n\n**6** Facebook Messenger 4% (-3) 43%\n\nTOP SOCIAL MEDIA AND MESSAGING\n\n# 21%\n\n**SHARE NEWS** via social, messaging or email\n\nBBC TV News: 44%\n\nBBC Radio News: 18%\n\n# **20%**\n\n34% in 2017\n\n# **20%**\n\n29% in 2017\n\nBRAND TRUST SCORES\n\nTrust Neither Don’t trust\n\nOVERALL TRUST SCORE\n\nMEDIA IS FREE FROM ...\n\n**TRUST**\n\nPublic broadcasters such as the BBC, ITV, and Channel 4 remain the most trusted news brands. Popular outlets such as the _Sun_ , the _Mail_ and the _Mirror_ attract big audiences online but their news coverage is often strongly distrusted. New TV channel GB News is struggling to build trust with general audiences even as it builds loyalty with some on the right.\n\nTrust = % scored 6–10 on 10-point scale, Don’t trust = 0–4, Neither = 5. Those that\n\nhaven’t heard of each brand were excluded. Only the above brands were included in\n\nthe survey so should not be treated as a list of the most trusted brands.\n\n0% 100%\n\n26\n\n18\n\n16\n\n14\n\n16\n\n22\n\n24\n\n22\n\n20\n\n29\n\n39\n\n29\n\n51\n\n49\n\n67\n\n55\n\n55\n\n54\n\n52\n\n52\n\n48\n\n45\n\n43\n\n42\n\n36\n\n27\n\n24\n\n23\n\n22\n\n12 Sun\n\nDaily Mirror\n\nDaily Mail/MailOnline\n\nHufPost\n\nGB News\n\nDaily Telegraph\n\nIndependent\n\nThe Times\n\nSky News\n\nGuardian\n\nLocal or regional newspaper\n\nFinancial Times\n\nChannel 4 News\n\nITV News\n\nBBC News _Change over time 2015-2022_\n\n_Undue political_\n\n_influence_\n\n_Undue business_\n\n_influence_\n\n51%\n\n0%\n\n25%\n\n50%\n\n75%\n\n2022 2015\n\n# **34%**\n\n=36/46 markets","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Digital_News-Report_2022_p63.pdf"]},"response":{"errors":[],"extract_id":"extract_afffd42a8b4bffeb6801362b8c5a68da","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["63\n\n59%\n\n74%\n\n20%\n\n79%\n\n53%\n\n73%\n\n17%\n\n38%\n\n0%\n\n50%\n\n100%\n\n2022 2019 2020 2021 2018 2017 2016 2015 2013 2014\n\n67%\n\n29%\n\n16%\n\n40%\n\n64%\n\n22%\n\n0%\n\n50%\n\n100%\n\n2022 2019 2020 2021 2018 2017 2016 2015 2014 2013\n\n10\n\n11\n\n6\n\n5\n\n5\n\n2\n\n5\n\n3\n\n2\n\n3\n\n3\n\nThe ‘i’ newspaper\n\nDaily Mirror (& Sunday)\n\nDaily Telegraph (& Sunday)\n\nChannel 5 News\n\nGuardian/Observer\n\nSun/Sun on Sunday\n\nThe Times/Sunday Times\n\nGB News\n\nMetro\n\nRegional or local newspaper\n\nCommercial radio news\n\nDaily Mail/Mail on Sunday\n\nChannel 4 News\n\nSky News\n\nITV News\n\nBBC News (TV & radio) 50\n\n27\n\n15\n\n9\n\n9\n\n7\n\n7\n\n5\n\n5\n\n5\n\n5\n\n4\n\n3\n\n3\n\n3\n\n2\n\n10\n\n7\n\n5\n\n4\n\n4\n\n3\n\n3\n\n3\n\n2\n\n4\n\n3\n\n4\n\nBuzzFeed News\n\nYahoo! News\n\nThe Times online\n\nMSN News\n\nMetro online\n\nITV News online\n\nHufPost\n\nSun online\n\nMirror online\n\nIndependent/ i100 online\n\nTelegraph online\n\nRegional or local newspaper website\n\nMailOnline\n\nSky News online\n\nGuardian online\n\nBBC News online 43\n\n18\n\n13\n\n12\n\n9\n\n6\n\n5\n\n5\n\n5\n\n5\n\n5\n\n5\n\n4\n\n4\n\n4\n\n4\n\nDigital News Report 2022 | United Kingdom 63\n\nTV, RADIO AND PRINT ONLINE\n\n**TOP BRANDS** % Weekly usage\n\n**Weekly use** TV, radio & print\n\n**More than 3 days per week** TV, radio & print\n\n**Weekly use** online brands\n\n**More than 3 days per week** online brands\n\n**WEEKLY REACH OFFLINE AND ONLINE**\n\nSOURCES OF NEWS 2013–22\n\nDEVICES FOR NEWS 2013–22\n\nOnline (incl. social media) TV Social media Print\n\n**CHANGING MEDIA**\n\nAudiences for traditional news sources like TV and print have fallen significantly over the last decade, with younger groups preferring to get their news online or via social media. After last year’s COVID-19 bump, overall news consumption is down across all devices.\n\n# 9%\n\npay for **ONLINE NEWS**\n\n# 25%\n\nlisten to **PODCASTS** in the last month\n\n**Rank Brand For News For All**\n\n**1** Facebook 19% (-4) 62%\n\n**2** Twitter 13% (-3) 28%\n\n**3** WhatsApp 10% (-4) 66%\n\n**Rank Brand For News For All**\n\n**4** YouTube 9% (-1) 54%\n\n**5** Instagram 6% (+1) 36%\n\n**6** Facebook Messenger 4% (-3) 43%\n\nTOP SOCIAL MEDIA AND MESSAGING\n\n# 21%\n\n**SHARE NEWS** via social, messaging or email\n\nBBC TV News: 44%\n\nBBC Radio News: 18%\n\n# **20%**\n\n34% in 2017\n\n# **20%**\n\n29% in 2017\n\nBRAND TRUST SCORES\n\nTrust Neither Don’t trust\n\nOVERALL TRUST SCORE\n\nMEDIA IS FREE FROM ...\n\n**TRUST**\n\nPublic broadcasters such as the BBC, ITV, and Channel 4 remain the most trusted news brands. Popular outlets such as the _Sun_ , the _Mail_ and the _Mirror_ attract big audiences online but their news coverage is often strongly distrusted. New TV channel GB News is struggling to build trust with general audiences even as it builds loyalty with some on the right.\n\nTrust = % scored 6–10 on 10-point scale, Don’t trust = 0–4, Neither = 5. Those that\n\nhaven’t heard of each brand were excluded. Only the above brands were included in\n\nthe survey so should not be treated as a list of the most trusted brands.\n\n0% 100%\n\n26\n\n18\n\n16\n\n14\n\n16\n\n22\n\n24\n\n22\n\n20\n\n29\n\n39\n\n29\n\n51\n\n49\n\n67\n\n55\n\n55\n\n54\n\n52\n\n52\n\n48\n\n45\n\n43\n\n42\n\n36\n\n27\n\n24\n\n23\n\n22\n\n12 Sun\n\nDaily Mirror\n\nDaily Mail/MailOnline\n\nHufPost\n\nGB News\n\nDaily Telegraph\n\nIndependent\n\nThe Times\n\nSky News\n\nGuardian\n\nLocal or regional newspaper\n\nFinancial Times\n\nChannel 4 News\n\nITV News\n\nBBC News _Change over time 2015-2022_\n\n_Undue political_\n\n_influence_\n\n_Undue business_\n\n_influence_\n\n51%\n\n0%\n\n25%\n\n50%\n\n75%\n\n2022 2015\n\n# **34%**\n\n=36/46 markets"],"full_content":"63\n\n59%\n\n74%\n\n20%\n\n79%\n\n53%\n\n73%\n\n17%\n\n38%\n\n0%\n\n50%\n\n100%\n\n2022 2019 2020 2021 2018 2017 2016 2015 2013 2014\n\n67%\n\n29%\n\n16%\n\n40%\n\n64%\n\n22%\n\n0%\n\n50%\n\n100%\n\n2022 2019 2020 2021 2018 2017 2016 2015 2014 2013\n\n10\n\n11\n\n6\n\n5\n\n5\n\n2\n\n5\n\n3\n\n2\n\n3\n\n3\n\nThe ‘i’ newspaper\n\nDaily Mirror (& Sunday)\n\nDaily Telegraph (& Sunday)\n\nChannel 5 News\n\nGuardian/Observer\n\nSun/Sun on Sunday\n\nThe Times/Sunday Times\n\nGB News\n\nMetro\n\nRegional or local newspaper\n\nCommercial radio news\n\nDaily Mail/Mail on Sunday\n\nChannel 4 News\n\nSky News\n\nITV News\n\nBBC News (TV & radio) 50\n\n27\n\n15\n\n9\n\n9\n\n7\n\n7\n\n5\n\n5\n\n5\n\n5\n\n4\n\n3\n\n3\n\n3\n\n2\n\n10\n\n7\n\n5\n\n4\n\n4\n\n3\n\n3\n\n3\n\n2\n\n4\n\n3\n\n4\n\nBuzzFeed News\n\nYahoo! News\n\nThe Times online\n\nMSN News\n\nMetro online\n\nITV News online\n\nHufPost\n\nSun online\n\nMirror online\n\nIndependent/ i100 online\n\nTelegraph online\n\nRegional or local newspaper website\n\nMailOnline\n\nSky News online\n\nGuardian online\n\nBBC News online 43\n\n18\n\n13\n\n12\n\n9\n\n6\n\n5\n\n5\n\n5\n\n5\n\n5\n\n5\n\n4\n\n4\n\n4\n\n4\n\nDigital News Report 2022 | United Kingdom 63\n\nTV, RADIO AND PRINT ONLINE\n\n**TOP BRANDS** % Weekly usage\n\n**Weekly use** TV, radio & print\n\n**More than 3 days per week** TV, radio & print\n\n**Weekly use** online brands\n\n**More than 3 days per week** online brands\n\n**WEEKLY REACH OFFLINE AND ONLINE**\n\nSOURCES OF NEWS 2013–22\n\nDEVICES FOR NEWS 2013–22\n\nOnline (incl. social media) TV Social media Print\n\n**CHANGING MEDIA**\n\nAudiences for traditional news sources like TV and print have fallen significantly over the last decade, with younger groups preferring to get their news online or via social media. After last year’s COVID-19 bump, overall news consumption is down across all devices.\n\n# 9%\n\npay for **ONLINE NEWS**\n\n# 25%\n\nlisten to **PODCASTS** in the last month\n\n**Rank Brand For News For All**\n\n**1** Facebook 19% (-4) 62%\n\n**2** Twitter 13% (-3) 28%\n\n**3** WhatsApp 10% (-4) 66%\n\n**Rank Brand For News For All**\n\n**4** YouTube 9% (-1) 54%\n\n**5** Instagram 6% (+1) 36%\n\n**6** Facebook Messenger 4% (-3) 43%\n\nTOP SOCIAL MEDIA AND MESSAGING\n\n# 21%\n\n**SHARE NEWS** via social, messaging or email\n\nBBC TV News: 44%\n\nBBC Radio News: 18%\n\n# **20%**\n\n34% in 2017\n\n# **20%**\n\n29% in 2017\n\nBRAND TRUST SCORES\n\nTrust Neither Don’t trust\n\nOVERALL TRUST SCORE\n\nMEDIA IS FREE FROM ...\n\n**TRUST**\n\nPublic broadcasters such as the BBC, ITV, and Channel 4 remain the most trusted news brands. Popular outlets such as the _Sun_ , the _Mail_ and the _Mirror_ attract big audiences online but their news coverage is often strongly distrusted. New TV channel GB News is struggling to build trust with general audiences even as it builds loyalty with some on the right.\n\nTrust = % scored 6–10 on 10-point scale, Don’t trust = 0–4, Neither = 5. Those that\n\nhaven’t heard of each brand were excluded. Only the above brands were included in\n\nthe survey so should not be treated as a list of the most trusted brands.\n\n0% 100%\n\n26\n\n18\n\n16\n\n14\n\n16\n\n22\n\n24\n\n22\n\n20\n\n29\n\n39\n\n29\n\n51\n\n49\n\n67\n\n55\n\n55\n\n54\n\n52\n\n52\n\n48\n\n45\n\n43\n\n42\n\n36\n\n27\n\n24\n\n23\n\n22\n\n12 Sun\n\nDaily Mirror\n\nDaily Mail/MailOnline\n\nHufPost\n\nGB News\n\nDaily Telegraph\n\nIndependent\n\nThe Times\n\nSky News\n\nGuardian\n\nLocal or regional newspaper\n\nFinancial Times\n\nChannel 4 News\n\nITV News\n\nBBC News _Change over time 2015-2022_\n\n_Undue political_\n\n_influence_\n\n_Undue business_\n\n_influence_\n\n51%\n\n0%\n\n25%\n\n50%\n\n75%\n\n2022 2015\n\n# **34%**\n\n=36/46 markets","publish_date":null,"title":"9% 25% 21%","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Digital_News-Report_2022_p63.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Digital_News-Report_2022_p63.pdf"},"request":{"example_id":"chart/Digital_News-Report_2022_p63","product_type":"parse","source_file_path":"Digital_News-Report_2022_p63.pdf"},"started_at":"2026-11-04T19:46:42.140844"} +{"completed_at":"2026-11-04T19:44:48.000078","latency_in_ms":1853,"output":{"example_id":"chart/Earnings-Presentation-FY24-Q4_1_p9","grounded_pages":[],"layout_pages":[],"markdown":"## Through dividends and share repurchases\n\nAmounts in billions, except as noted. Dollar changes may not recalculate due to rounding.\n\n### • Share repurchases during the\n\n### quarter totaled $1,497 million representing 9.5 million shares, at an average price of $156.61 per share\n\n### • Remaining share repurchase\n\n### authorization is $16.5 billion\n\nReturns to shareholders $2.7 $2.2 $2.0 $1.6 $3.0\n\n# **Returns to shareholders**\n\n$1.5 $1.5 $1.5 $1.5 $1.5\n\n$1.2\n\n$0.7\n\n$0.5 $0.1 $1.5\n\nDividends Share repurchases\n\nQ4 FY23 Q1 FY24 Q2 FY24 Q3 FY24 Q4 FY24\n\n9","pages":[{"line_numbers":[],"markdown":"## Through dividends and share repurchases\n\nAmounts in billions, except as noted. Dollar changes may not recalculate due to rounding.\n\n### • Share repurchases during the\n\n### quarter totaled $1,497 million representing 9.5 million shares, at an average price of $156.61 per share\n\n### • Remaining share repurchase\n\n### authorization is $16.5 billion\n\nReturns to shareholders $2.7 $2.2 $2.0 $1.6 $3.0\n\n# **Returns to shareholders**\n\n$1.5 $1.5 $1.5 $1.5 $1.5\n\n$1.2\n\n$0.7\n\n$0.5 $0.1 $1.5\n\nDividends Share repurchases\n\nQ4 FY23 Q1 FY24 Q2 FY24 Q3 FY24 Q4 FY24\n\n9","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"## Through dividends and share repurchases\n\nAmounts in billions, except as noted. Dollar changes may not recalculate due to rounding.\n\n### • Share repurchases during the\n\n### quarter totaled $1,497 million representing 9.5 million shares, at an average price of $156.61 per share\n\n### • Remaining share repurchase\n\n### authorization is $16.5 billion\n\nReturns to shareholders $2.7 $2.2 $2.0 $1.6 $3.0\n\n# **Returns to shareholders**\n\n$1.5 $1.5 $1.5 $1.5 $1.5\n\n$1.2\n\n$0.7\n\n$0.5 $0.1 $1.5\n\nDividends Share repurchases\n\nQ4 FY23 Q1 FY24 Q2 FY24 Q3 FY24 Q4 FY24\n\n9","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Earnings-Presentation-FY24-Q4_1_p9.pdf"]},"response":{"errors":[],"extract_id":"extract_674e9c733534240131d7e069dbaf775e","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["## Through dividends and share repurchases\n\nAmounts in billions, except as noted. Dollar changes may not recalculate due to rounding.\n\n### • Share repurchases during the\n\n### quarter totaled $1,497 million representing 9.5 million shares, at an average price of $156.61 per share\n\n### • Remaining share repurchase\n\n### authorization is $16.5 billion\n\nReturns to shareholders $2.7 $2.2 $2.0 $1.6 $3.0\n\n# **Returns to shareholders**\n\n$1.5 $1.5 $1.5 $1.5 $1.5\n\n$1.2\n\n$0.7\n\n$0.5 $0.1 $1.5\n\nDividends Share repurchases\n\nQ4 FY23 Q1 FY24 Q2 FY24 Q3 FY24 Q4 FY24\n\n9"],"full_content":"## Through dividends and share repurchases\n\nAmounts in billions, except as noted. Dollar changes may not recalculate due to rounding.\n\n### • Share repurchases during the\n\n### quarter totaled $1,497 million representing 9.5 million shares, at an average price of $156.61 per share\n\n### • Remaining share repurchase\n\n### authorization is $16.5 billion\n\nReturns to shareholders $2.7 $2.2 $2.0 $1.6 $3.0\n\n# **Returns to shareholders**\n\n$1.5 $1.5 $1.5 $1.5 $1.5\n\n$1.2\n\n$0.7\n\n$0.5 $0.1 $1.5\n\nDividends Share repurchases\n\nQ4 FY23 Q1 FY24 Q2 FY24 Q3 FY24 Q4 FY24\n\n9","publish_date":null,"title":"Returns to shareholders","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Earnings-Presentation-FY24-Q4_1_p9.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Earnings-Presentation-FY24-Q4_1_p9.pdf"},"request":{"example_id":"chart/Earnings-Presentation-FY24-Q4_1_p9","product_type":"parse","source_file_path":"Earnings-Presentation-FY24-Q4_1_p9.pdf"},"started_at":"2026-11-04T19:44:46.146238"} +{"completed_at":"2026-11-04T19:44:48.301721","latency_in_ms":2152,"output":{"example_id":"chart/Earnings_Presentation_(FY25_Q4)_p19","grounded_pages":[],"layout_pages":[],"markdown":"## Net sales $23.1 billion, +5.7%, Net sales without fuel +7.1%, eCommerce +24%\n\n### • Comp sales strength driven by increases in transactions and unit volumes\n\n### ◦ Transactions ex fuel: +5.4% ◦ Average ticket ex fuel: +1.3%\n\n### • Strength in food and health & wellness\n\n### • Share gains in grocery and general merchandise categories, including apparel and consumer electronics (per Circana)\n\n### • Growth in eCommerce sales of +24%; driven by momentum in delivery from club\n\n### • Digital penetration (members using Scan and Go or shopping online) hit all-time high\n\n### • Member's Mark grew high single-digits, outpacing segment comp\n\neComm Cont. without fuel +190 bps +180 bps +230 bps +290 bps +280 bps\n\n1 Comp sales for the 14-week period ended January 31, 2025 compared to the 14-week period ended February 2, 2024.\n\n# Sam’s Club U.S. revenues\n\n1\\.9%\n\n3\\.5%\n\n4\\.6%\n\n3\\.7%\n\n5\\.3%\n\n3\\.1%\n\n4\\.4%\n\n5\\.2%\n\n7\\.0% 6.8%\n\nWith fuel Without fuel\n\nQ4 FY24 Q1 FY25 Q2 FY25 Q3 FY25 Q4 FY25\n\n19\n\nSam's Club U.S. comp sales ¹","pages":[{"line_numbers":[],"markdown":"## Net sales $23.1 billion, +5.7%, Net sales without fuel +7.1%, eCommerce +24%\n\n### • Comp sales strength driven by increases in transactions and unit volumes\n\n### ◦ Transactions ex fuel: +5.4% ◦ Average ticket ex fuel: +1.3%\n\n### • Strength in food and health & wellness\n\n### • Share gains in grocery and general merchandise categories, including apparel and consumer electronics (per Circana)\n\n### • Growth in eCommerce sales of +24%; driven by momentum in delivery from club\n\n### • Digital penetration (members using Scan and Go or shopping online) hit all-time high\n\n### • Member's Mark grew high single-digits, outpacing segment comp\n\neComm Cont. without fuel +190 bps +180 bps +230 bps +290 bps +280 bps\n\n1 Comp sales for the 14-week period ended January 31, 2025 compared to the 14-week period ended February 2, 2024.\n\n# Sam’s Club U.S. revenues\n\n1\\.9%\n\n3\\.5%\n\n4\\.6%\n\n3\\.7%\n\n5\\.3%\n\n3\\.1%\n\n4\\.4%\n\n5\\.2%\n\n7\\.0% 6.8%\n\nWith fuel Without fuel\n\nQ4 FY24 Q1 FY25 Q2 FY25 Q3 FY25 Q4 FY25\n\n19\n\nSam's Club U.S. comp sales ¹","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"## Net sales $23.1 billion, +5.7%, Net sales without fuel +7.1%, eCommerce +24%\n\n### • Comp sales strength driven by increases in transactions and unit volumes\n\n### ◦ Transactions ex fuel: +5.4% ◦ Average ticket ex fuel: +1.3%\n\n### • Strength in food and health & wellness\n\n### • Share gains in grocery and general merchandise categories, including apparel and consumer electronics (per Circana)\n\n### • Growth in eCommerce sales of +24%; driven by momentum in delivery from club\n\n### • Digital penetration (members using Scan and Go or shopping online) hit all-time high\n\n### • Member's Mark grew high single-digits, outpacing segment comp\n\neComm Cont. without fuel +190 bps +180 bps +230 bps +290 bps +280 bps\n\n1 Comp sales for the 14-week period ended January 31, 2025 compared to the 14-week period ended February 2, 2024.\n\n# Sam’s Club U.S. revenues\n\n1\\.9%\n\n3\\.5%\n\n4\\.6%\n\n3\\.7%\n\n5\\.3%\n\n3\\.1%\n\n4\\.4%\n\n5\\.2%\n\n7\\.0% 6.8%\n\nWith fuel Without fuel\n\nQ4 FY24 Q1 FY25 Q2 FY25 Q3 FY25 Q4 FY25\n\n19\n\nSam's Club U.S. comp sales ¹","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Earnings_Presentation_%28FY25_Q4%29_p19.pdf"]},"response":{"errors":[],"extract_id":"extract_5f642feeb53d7c2bd1b96c936fdae261","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["## Net sales $23.1 billion, +5.7%, Net sales without fuel +7.1%, eCommerce +24%\n\n### • Comp sales strength driven by increases in transactions and unit volumes\n\n### ◦ Transactions ex fuel: +5.4% ◦ Average ticket ex fuel: +1.3%\n\n### • Strength in food and health & wellness\n\n### • Share gains in grocery and general merchandise categories, including apparel and consumer electronics (per Circana)\n\n### • Growth in eCommerce sales of +24%; driven by momentum in delivery from club\n\n### • Digital penetration (members using Scan and Go or shopping online) hit all-time high\n\n### • Member's Mark grew high single-digits, outpacing segment comp\n\neComm Cont. without fuel +190 bps +180 bps +230 bps +290 bps +280 bps\n\n1 Comp sales for the 14-week period ended January 31, 2025 compared to the 14-week period ended February 2, 2024.\n\n# Sam’s Club U.S. revenues\n\n1\\.9%\n\n3\\.5%\n\n4\\.6%\n\n3\\.7%\n\n5\\.3%\n\n3\\.1%\n\n4\\.4%\n\n5\\.2%\n\n7\\.0% 6.8%\n\nWith fuel Without fuel\n\nQ4 FY24 Q1 FY25 Q2 FY25 Q3 FY25 Q4 FY25\n\n19\n\nSam's Club U.S. comp sales ¹"],"full_content":"## Net sales $23.1 billion, +5.7%, Net sales without fuel +7.1%, eCommerce +24%\n\n### • Comp sales strength driven by increases in transactions and unit volumes\n\n### ◦ Transactions ex fuel: +5.4% ◦ Average ticket ex fuel: +1.3%\n\n### • Strength in food and health & wellness\n\n### • Share gains in grocery and general merchandise categories, including apparel and consumer electronics (per Circana)\n\n### • Growth in eCommerce sales of +24%; driven by momentum in delivery from club\n\n### • Digital penetration (members using Scan and Go or shopping online) hit all-time high\n\n### • Member's Mark grew high single-digits, outpacing segment comp\n\neComm Cont. without fuel +190 bps +180 bps +230 bps +290 bps +280 bps\n\n1 Comp sales for the 14-week period ended January 31, 2025 compared to the 14-week period ended February 2, 2024.\n\n# Sam’s Club U.S. revenues\n\n1\\.9%\n\n3\\.5%\n\n4\\.6%\n\n3\\.7%\n\n5\\.3%\n\n3\\.1%\n\n4\\.4%\n\n5\\.2%\n\n7\\.0% 6.8%\n\nWith fuel Without fuel\n\nQ4 FY24 Q1 FY25 Q2 FY25 Q3 FY25 Q4 FY25\n\n19\n\nSam's Club U.S. comp sales ¹","publish_date":null,"title":"Sam’s Club U.S. revenues","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Earnings_Presentation_%28FY25_Q4%29_p19.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Earnings_Presentation_%28FY25_Q4%29_p19.pdf"},"request":{"example_id":"chart/Earnings_Presentation_(FY25_Q4)_p19","product_type":"parse","source_file_path":"Earnings_Presentation_(FY25_Q4)_p19.pdf"},"started_at":"2026-11-04T19:44:46.149313"} +{"completed_at":"2026-11-04T19:44:51.192690","latency_in_ms":2584,"output":{"example_id":"chart/Earnings_Presentation_(FY26_Q1)_1_p19","grounded_pages":[],"layout_pages":[],"markdown":"## Net sales $22.1 billion, +2.9%, Net sales without fuel +5.5%, eCommerce +27%\n\n### • Continued strength in grocery and health & wellness; fourth consecutive quarter of positive general merchandise sales\n\n### • Comp sales driven by increases in transactions and unit volumes\n\n### ◦ Transactions ex fuel: +4.8% ◦ Average ticket ex fuel: +1.7%\n\n### • eCommerce delivery growth of ~160%\n\n### • Share gains in grocery and general merchandise categories (per Circana)\n\n### • Member's Mark grew low double-digits, outpacing segment comp\n\neComm Cont. without fuel +180 bps +230 bps +290 bps +280 bps +350 bps\n\n1 Comp sales for the 13-week period ended May 2, 2025 compared to the 13-week period ended May 3, 2024.\n\n# Sam’s Club U.S. revenues\n\n3\\.5%\n\n4\\.6%\n\n3\\.7%\n\n5\\.3%\n\n4\\.0%\n\n4\\.4%\n\n5\\.2%\n\n7\\.0% 6.8% 6.7%\n\nWith fuel Without fuel\n\nQ1 FY25 Q2 FY25 Q3 FY25 Q4 FY25 Q1 FY26\n\n19\n\nSam's Club U.S. comp sales ¹","pages":[{"line_numbers":[],"markdown":"## Net sales $22.1 billion, +2.9%, Net sales without fuel +5.5%, eCommerce +27%\n\n### • Continued strength in grocery and health & wellness; fourth consecutive quarter of positive general merchandise sales\n\n### • Comp sales driven by increases in transactions and unit volumes\n\n### ◦ Transactions ex fuel: +4.8% ◦ Average ticket ex fuel: +1.7%\n\n### • eCommerce delivery growth of ~160%\n\n### • Share gains in grocery and general merchandise categories (per Circana)\n\n### • Member's Mark grew low double-digits, outpacing segment comp\n\neComm Cont. without fuel +180 bps +230 bps +290 bps +280 bps +350 bps\n\n1 Comp sales for the 13-week period ended May 2, 2025 compared to the 13-week period ended May 3, 2024.\n\n# Sam’s Club U.S. revenues\n\n3\\.5%\n\n4\\.6%\n\n3\\.7%\n\n5\\.3%\n\n4\\.0%\n\n4\\.4%\n\n5\\.2%\n\n7\\.0% 6.8% 6.7%\n\nWith fuel Without fuel\n\nQ1 FY25 Q2 FY25 Q3 FY25 Q4 FY25 Q1 FY26\n\n19\n\nSam's Club U.S. comp sales ¹","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"## Net sales $22.1 billion, +2.9%, Net sales without fuel +5.5%, eCommerce +27%\n\n### • Continued strength in grocery and health & wellness; fourth consecutive quarter of positive general merchandise sales\n\n### • Comp sales driven by increases in transactions and unit volumes\n\n### ◦ Transactions ex fuel: +4.8% ◦ Average ticket ex fuel: +1.7%\n\n### • eCommerce delivery growth of ~160%\n\n### • Share gains in grocery and general merchandise categories (per Circana)\n\n### • Member's Mark grew low double-digits, outpacing segment comp\n\neComm Cont. without fuel +180 bps +230 bps +290 bps +280 bps +350 bps\n\n1 Comp sales for the 13-week period ended May 2, 2025 compared to the 13-week period ended May 3, 2024.\n\n# Sam’s Club U.S. revenues\n\n3\\.5%\n\n4\\.6%\n\n3\\.7%\n\n5\\.3%\n\n4\\.0%\n\n4\\.4%\n\n5\\.2%\n\n7\\.0% 6.8% 6.7%\n\nWith fuel Without fuel\n\nQ1 FY25 Q2 FY25 Q3 FY25 Q4 FY25 Q1 FY26\n\n19\n\nSam's Club U.S. comp sales ¹","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Earnings_Presentation_%28FY26_Q1%29_1_p19.pdf"]},"response":{"errors":[],"extract_id":"extract_e9482c6cbb47f20c7ee83baa485bc667","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["## Net sales $22.1 billion, +2.9%, Net sales without fuel +5.5%, eCommerce +27%\n\n### • Continued strength in grocery and health & wellness; fourth consecutive quarter of positive general merchandise sales\n\n### • Comp sales driven by increases in transactions and unit volumes\n\n### ◦ Transactions ex fuel: +4.8% ◦ Average ticket ex fuel: +1.7%\n\n### • eCommerce delivery growth of ~160%\n\n### • Share gains in grocery and general merchandise categories (per Circana)\n\n### • Member's Mark grew low double-digits, outpacing segment comp\n\neComm Cont. without fuel +180 bps +230 bps +290 bps +280 bps +350 bps\n\n1 Comp sales for the 13-week period ended May 2, 2025 compared to the 13-week period ended May 3, 2024.\n\n# Sam’s Club U.S. revenues\n\n3\\.5%\n\n4\\.6%\n\n3\\.7%\n\n5\\.3%\n\n4\\.0%\n\n4\\.4%\n\n5\\.2%\n\n7\\.0% 6.8% 6.7%\n\nWith fuel Without fuel\n\nQ1 FY25 Q2 FY25 Q3 FY25 Q4 FY25 Q1 FY26\n\n19\n\nSam's Club U.S. comp sales ¹"],"full_content":"## Net sales $22.1 billion, +2.9%, Net sales without fuel +5.5%, eCommerce +27%\n\n### • Continued strength in grocery and health & wellness; fourth consecutive quarter of positive general merchandise sales\n\n### • Comp sales driven by increases in transactions and unit volumes\n\n### ◦ Transactions ex fuel: +4.8% ◦ Average ticket ex fuel: +1.7%\n\n### • eCommerce delivery growth of ~160%\n\n### • Share gains in grocery and general merchandise categories (per Circana)\n\n### • Member's Mark grew low double-digits, outpacing segment comp\n\neComm Cont. without fuel +180 bps +230 bps +290 bps +280 bps +350 bps\n\n1 Comp sales for the 13-week period ended May 2, 2025 compared to the 13-week period ended May 3, 2024.\n\n# Sam’s Club U.S. revenues\n\n3\\.5%\n\n4\\.6%\n\n3\\.7%\n\n5\\.3%\n\n4\\.0%\n\n4\\.4%\n\n5\\.2%\n\n7\\.0% 6.8% 6.7%\n\nWith fuel Without fuel\n\nQ1 FY25 Q2 FY25 Q3 FY25 Q4 FY25 Q1 FY26\n\n19\n\nSam's Club U.S. comp sales ¹","publish_date":null,"title":"Sam’s Club U.S. revenues","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Earnings_Presentation_%28FY26_Q1%29_1_p19.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Earnings_Presentation_%28FY26_Q1%29_1_p19.pdf"},"request":{"example_id":"chart/Earnings_Presentation_(FY26_Q1)_1_p19","product_type":"parse","source_file_path":"Earnings_Presentation_(FY26_Q1)_1_p19.pdf"},"started_at":"2026-11-04T19:44:48.608508"} +{"completed_at":"2026-11-04T19:44:50.361192","latency_in_ms":1079,"output":{"example_id":"chart/Earnings_Presentation_(FY26_Q1)_1_p9","grounded_pages":[],"layout_pages":[],"markdown":"## Dividends and share repurchases\n\nAmounts in billions, except as noted. Dollar amounts may not recalculate due to rounding.\n\n### • Share repurchases during the quarter totaled $4.6 billion representing 50.4 million shares, at an average price of $90.35 per share\n\n### • Remaining share repurchase authorization is $7.5 billion\n\nReturns to shareholders $2.7 $2.7 $2.6 $3.1 $6.4\n\n# Returns to shareholders\n\n$1.7 $1.7 $1.7 $1.7 $1.9\n\n$1.1 $1.0 $1.0 $1.4 $4.6\n\nDividends Share repurchases\n\nQ1 FY25 Q2 FY25 Q3 FY25 Q4 FY25 Q1 FY26\n\n9","pages":[{"line_numbers":[],"markdown":"## Dividends and share repurchases\n\nAmounts in billions, except as noted. Dollar amounts may not recalculate due to rounding.\n\n### • Share repurchases during the quarter totaled $4.6 billion representing 50.4 million shares, at an average price of $90.35 per share\n\n### • Remaining share repurchase authorization is $7.5 billion\n\nReturns to shareholders $2.7 $2.7 $2.6 $3.1 $6.4\n\n# Returns to shareholders\n\n$1.7 $1.7 $1.7 $1.7 $1.9\n\n$1.1 $1.0 $1.0 $1.4 $4.6\n\nDividends Share repurchases\n\nQ1 FY25 Q2 FY25 Q3 FY25 Q4 FY25 Q1 FY26\n\n9","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"## Dividends and share repurchases\n\nAmounts in billions, except as noted. Dollar amounts may not recalculate due to rounding.\n\n### • Share repurchases during the quarter totaled $4.6 billion representing 50.4 million shares, at an average price of $90.35 per share\n\n### • Remaining share repurchase authorization is $7.5 billion\n\nReturns to shareholders $2.7 $2.7 $2.6 $3.1 $6.4\n\n# Returns to shareholders\n\n$1.7 $1.7 $1.7 $1.7 $1.9\n\n$1.1 $1.0 $1.0 $1.4 $4.6\n\nDividends Share repurchases\n\nQ1 FY25 Q2 FY25 Q3 FY25 Q4 FY25 Q1 FY26\n\n9","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Earnings_Presentation_%28FY26_Q1%29_1_p9.pdf"]},"response":{"errors":[],"extract_id":"extract_a6c43433487cdd7b4e24b7efd015c219","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["## Dividends and share repurchases\n\nAmounts in billions, except as noted. Dollar amounts may not recalculate due to rounding.\n\n### • Share repurchases during the quarter totaled $4.6 billion representing 50.4 million shares, at an average price of $90.35 per share\n\n### • Remaining share repurchase authorization is $7.5 billion\n\nReturns to shareholders $2.7 $2.7 $2.6 $3.1 $6.4\n\n# Returns to shareholders\n\n$1.7 $1.7 $1.7 $1.7 $1.9\n\n$1.1 $1.0 $1.0 $1.4 $4.6\n\nDividends Share repurchases\n\nQ1 FY25 Q2 FY25 Q3 FY25 Q4 FY25 Q1 FY26\n\n9"],"full_content":"## Dividends and share repurchases\n\nAmounts in billions, except as noted. Dollar amounts may not recalculate due to rounding.\n\n### • Share repurchases during the quarter totaled $4.6 billion representing 50.4 million shares, at an average price of $90.35 per share\n\n### • Remaining share repurchase authorization is $7.5 billion\n\nReturns to shareholders $2.7 $2.7 $2.6 $3.1 $6.4\n\n# Returns to shareholders\n\n$1.7 $1.7 $1.7 $1.7 $1.9\n\n$1.1 $1.0 $1.0 $1.4 $4.6\n\nDividends Share repurchases\n\nQ1 FY25 Q2 FY25 Q3 FY25 Q4 FY25 Q1 FY26\n\n9","publish_date":null,"title":"Returns to shareholders","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Earnings_Presentation_%28FY26_Q1%29_1_p9.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Earnings_Presentation_%28FY26_Q1%29_1_p9.pdf"},"request":{"example_id":"chart/Earnings_Presentation_(FY26_Q1)_1_p9","product_type":"parse","source_file_path":"Earnings_Presentation_(FY26_Q1)_1_p9.pdf"},"started_at":"2026-11-04T19:44:49.281241"} +{"completed_at":"2026-11-04T19:44:52.810817","latency_in_ms":1660,"output":{"example_id":"chart/Economic_and_Market_Report-Full_year-2024_p11","grounded_pages":[],"layout_pages":[],"markdown":"www.acea.auto 11\n\n### Figure 3. Top 10 global car producers\n\nSOURCE: **S&P GLOBAL MOBILITY**","pages":[{"line_numbers":[],"markdown":"www.acea.auto 11\n\n### Figure 3. Top 10 global car producers\n\nSOURCE: **S&P GLOBAL MOBILITY**","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"www.acea.auto 11\n\n### Figure 3. Top 10 global car producers\n\nSOURCE: **S&P GLOBAL MOBILITY**","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Economic_and_Market_Report-Full_year-2024_p11.pdf"]},"response":{"errors":[],"extract_id":"extract_8d4a2c25ddee6f96cb6222ee9f74da23","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["www.acea.auto 11\n\n### Figure 3. Top 10 global car producers\n\nSOURCE: **S&P GLOBAL MOBILITY**"],"full_content":"www.acea.auto 11\n\n### Figure 3. Top 10 global car producers\n\nSOURCE: **S&P GLOBAL MOBILITY**","publish_date":null,"title":"Figure 3. Top 10 global car producers","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Economic_and_Market_Report-Full_year-2024_p11.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Economic_and_Market_Report-Full_year-2024_p11.pdf"},"request":{"example_id":"chart/Economic_and_Market_Report-Full_year-2024_p11","product_type":"parse","source_file_path":"Economic_and_Market_Report-Full_year-2024_p11.pdf"},"started_at":"2026-11-04T19:44:51.150318"} +{"completed_at":"2026-11-04T19:44:53.238665","latency_in_ms":1142,"output":{"example_id":"chart/Economic_and_Market_Report-Full_year-2024_p16","grounded_pages":[],"layout_pages":[],"markdown":"www.acea.auto 16\n\n# COMMERCIAL VEHICLES\n\n## REGISTRATIONS\n\n### Global\n\nIn 2024, North America led with a 1.7% increase in sales compared to 2023, reaching 4 million units and representing a 30% market share. In contrast, Europe saw a growth of 5.7%, totalling 2.5 million units and capturing an 18% market share. Greater China ⁹ experienced a decline of 5.9%, with sales dropping to nearly 3 million units, yet maintaining a substantial 22% market share.\n\nThe truck segment followed, with Greater China holding a stable market at 1 million units, a slight increase of 0.9%, commanding a 34% market share. North America and Europe saw declines in truck sales, decreasing by 1.9% and 7%, respectively. South Asia experienced a notable drop of 7.8%, while South America saw an 8.1% increase.\n\nIn the bus segment, Greater China led with a 29% increase in sales to 120,997 units, capturing a 31% market share. South Asia saw a 14.5% rise to 96,949 units, while Europe experienced a 7.3% growth to 60,625. North America faced a 4% decline to 55,322 units.\n\nFigure 4. Global new commercial vehicles and bus sales **10**\n\nSOURCE: **S&P GLOBAL MOBILITY**\n\n9 Includes Mainland China, and Taiwan 10 Based on production type for North America region. Europe includes the EU, EFTA, the United Kingdom, and other countries, notably Belarus, Bosnia-Herzegovina, Kazakhstan, North Macedonia, Russia, Serbia, Türkiye, Ukraine, and Uzbekistan","pages":[{"line_numbers":[],"markdown":"www.acea.auto 16\n\n# COMMERCIAL VEHICLES\n\n## REGISTRATIONS\n\n### Global\n\nIn 2024, North America led with a 1.7% increase in sales compared to 2023, reaching 4 million units and representing a 30% market share. In contrast, Europe saw a growth of 5.7%, totalling 2.5 million units and capturing an 18% market share. Greater China ⁹ experienced a decline of 5.9%, with sales dropping to nearly 3 million units, yet maintaining a substantial 22% market share.\n\nThe truck segment followed, with Greater China holding a stable market at 1 million units, a slight increase of 0.9%, commanding a 34% market share. North America and Europe saw declines in truck sales, decreasing by 1.9% and 7%, respectively. South Asia experienced a notable drop of 7.8%, while South America saw an 8.1% increase.\n\nIn the bus segment, Greater China led with a 29% increase in sales to 120,997 units, capturing a 31% market share. South Asia saw a 14.5% rise to 96,949 units, while Europe experienced a 7.3% growth to 60,625. North America faced a 4% decline to 55,322 units.\n\nFigure 4. Global new commercial vehicles and bus sales **10**\n\nSOURCE: **S&P GLOBAL MOBILITY**\n\n9 Includes Mainland China, and Taiwan 10 Based on production type for North America region. Europe includes the EU, EFTA, the United Kingdom, and other countries, notably Belarus, Bosnia-Herzegovina, Kazakhstan, North Macedonia, Russia, Serbia, Türkiye, Ukraine, and Uzbekistan","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"www.acea.auto 16\n\n# COMMERCIAL VEHICLES\n\n## REGISTRATIONS\n\n### Global\n\nIn 2024, North America led with a 1.7% increase in sales compared to 2023, reaching 4 million units and representing a 30% market share. In contrast, Europe saw a growth of 5.7%, totalling 2.5 million units and capturing an 18% market share. Greater China ⁹ experienced a decline of 5.9%, with sales dropping to nearly 3 million units, yet maintaining a substantial 22% market share.\n\nThe truck segment followed, with Greater China holding a stable market at 1 million units, a slight increase of 0.9%, commanding a 34% market share. North America and Europe saw declines in truck sales, decreasing by 1.9% and 7%, respectively. South Asia experienced a notable drop of 7.8%, while South America saw an 8.1% increase.\n\nIn the bus segment, Greater China led with a 29% increase in sales to 120,997 units, capturing a 31% market share. South Asia saw a 14.5% rise to 96,949 units, while Europe experienced a 7.3% growth to 60,625. North America faced a 4% decline to 55,322 units.\n\nFigure 4. Global new commercial vehicles and bus sales **10**\n\nSOURCE: **S&P GLOBAL MOBILITY**\n\n9 Includes Mainland China, and Taiwan 10 Based on production type for North America region. Europe includes the EU, EFTA, the United Kingdom, and other countries, notably Belarus, Bosnia-Herzegovina, Kazakhstan, North Macedonia, Russia, Serbia, Türkiye, Ukraine, and Uzbekistan","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Economic_and_Market_Report-Full_year-2024_p16.pdf"]},"response":{"errors":[],"extract_id":"extract_b8549f8a9aba903f4a4782bc0f3426fc","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["www.acea.auto 16\n\n# COMMERCIAL VEHICLES\n\n## REGISTRATIONS\n\n### Global\n\nIn 2024, North America led with a 1.7% increase in sales compared to 2023, reaching 4 million units and representing a 30% market share. In contrast, Europe saw a growth of 5.7%, totalling 2.5 million units and capturing an 18% market share. Greater China ⁹ experienced a decline of 5.9%, with sales dropping to nearly 3 million units, yet maintaining a substantial 22% market share.\n\nThe truck segment followed, with Greater China holding a stable market at 1 million units, a slight increase of 0.9%, commanding a 34% market share. North America and Europe saw declines in truck sales, decreasing by 1.9% and 7%, respectively. South Asia experienced a notable drop of 7.8%, while South America saw an 8.1% increase.\n\nIn the bus segment, Greater China led with a 29% increase in sales to 120,997 units, capturing a 31% market share. South Asia saw a 14.5% rise to 96,949 units, while Europe experienced a 7.3% growth to 60,625. North America faced a 4% decline to 55,322 units.\n\nFigure 4. Global new commercial vehicles and bus sales **10**\n\nSOURCE: **S&P GLOBAL MOBILITY**\n\n9 Includes Mainland China, and Taiwan 10 Based on production type for North America region. Europe includes the EU, EFTA, the United Kingdom, and other countries, notably Belarus, Bosnia-Herzegovina, Kazakhstan, North Macedonia, Russia, Serbia, Türkiye, Ukraine, and Uzbekistan"],"full_content":"www.acea.auto 16\n\n# COMMERCIAL VEHICLES\n\n## REGISTRATIONS\n\n### Global\n\nIn 2024, North America led with a 1.7% increase in sales compared to 2023, reaching 4 million units and representing a 30% market share. In contrast, Europe saw a growth of 5.7%, totalling 2.5 million units and capturing an 18% market share. Greater China ⁹ experienced a decline of 5.9%, with sales dropping to nearly 3 million units, yet maintaining a substantial 22% market share.\n\nThe truck segment followed, with Greater China holding a stable market at 1 million units, a slight increase of 0.9%, commanding a 34% market share. North America and Europe saw declines in truck sales, decreasing by 1.9% and 7%, respectively. South Asia experienced a notable drop of 7.8%, while South America saw an 8.1% increase.\n\nIn the bus segment, Greater China led with a 29% increase in sales to 120,997 units, capturing a 31% market share. South Asia saw a 14.5% rise to 96,949 units, while Europe experienced a 7.3% growth to 60,625. North America faced a 4% decline to 55,322 units.\n\nFigure 4. Global new commercial vehicles and bus sales **10**\n\nSOURCE: **S&P GLOBAL MOBILITY**\n\n9 Includes Mainland China, and Taiwan 10 Based on production type for North America region. Europe includes the EU, EFTA, the United Kingdom, and other countries, notably Belarus, Bosnia-Herzegovina, Kazakhstan, North Macedonia, Russia, Serbia, Türkiye, Ukraine, and Uzbekistan","publish_date":null,"title":"COMMERCIAL VEHICLES","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Economic_and_Market_Report-Full_year-2024_p16.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Economic_and_Market_Report-Full_year-2024_p16.pdf"},"request":{"example_id":"chart/Economic_and_Market_Report-Full_year-2024_p16","product_type":"parse","source_file_path":"Economic_and_Market_Report-Full_year-2024_p16.pdf"},"started_at":"2026-11-04T19:44:52.095791"} +{"completed_at":"2026-11-04T19:44:54.166560","latency_in_ms":1168,"output":{"example_id":"chart/Economic_and_Market_Report-Full_year-2024_p18","grounded_pages":[],"layout_pages":[],"markdown":"www.acea.auto 18\n\ndecline of 9.1%, reducing their market share to 6.1% from 7.2% the previous year. Hybrid- electric van sales also fell by 4.8%, accounting for just 2% of the market.\n\nDiesel trucks continued to dominate in 2024, accounting for 95.1% of new EU registrations, despite a 6.2% decline. Electrically chargeable truck registrations fell by 4.6%, with their market share remaining stable at 2.3% compared to last year.\n\nNew EU electrically chargeable bus registrations rose by 26.8% in 2024, with the market share increasing from 15.9% to 18.5%. Hybrid-electric bus sales fell by 16.1%, making up a 9.8% share. Diesel bus registrations grew by 11.1%, increasing their market share to 63.1%, up 1 percentage point compared to 2023.\n\n### Figure 5. New commercial vehicles registrations by power source","pages":[{"line_numbers":[],"markdown":"www.acea.auto 18\n\ndecline of 9.1%, reducing their market share to 6.1% from 7.2% the previous year. Hybrid- electric van sales also fell by 4.8%, accounting for just 2% of the market.\n\nDiesel trucks continued to dominate in 2024, accounting for 95.1% of new EU registrations, despite a 6.2% decline. Electrically chargeable truck registrations fell by 4.6%, with their market share remaining stable at 2.3% compared to last year.\n\nNew EU electrically chargeable bus registrations rose by 26.8% in 2024, with the market share increasing from 15.9% to 18.5%. Hybrid-electric bus sales fell by 16.1%, making up a 9.8% share. Diesel bus registrations grew by 11.1%, increasing their market share to 63.1%, up 1 percentage point compared to 2023.\n\n### Figure 5. New commercial vehicles registrations by power source","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"www.acea.auto 18\n\ndecline of 9.1%, reducing their market share to 6.1% from 7.2% the previous year. Hybrid- electric van sales also fell by 4.8%, accounting for just 2% of the market.\n\nDiesel trucks continued to dominate in 2024, accounting for 95.1% of new EU registrations, despite a 6.2% decline. Electrically chargeable truck registrations fell by 4.6%, with their market share remaining stable at 2.3% compared to last year.\n\nNew EU electrically chargeable bus registrations rose by 26.8% in 2024, with the market share increasing from 15.9% to 18.5%. Hybrid-electric bus sales fell by 16.1%, making up a 9.8% share. Diesel bus registrations grew by 11.1%, increasing their market share to 63.1%, up 1 percentage point compared to 2023.\n\n### Figure 5. New commercial vehicles registrations by power source","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Economic_and_Market_Report-Full_year-2024_p18.pdf"]},"response":{"errors":[],"extract_id":"extract_183e494ad75af190dc2c4cbf9c35fe13","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["www.acea.auto 18\n\ndecline of 9.1%, reducing their market share to 6.1% from 7.2% the previous year. Hybrid- electric van sales also fell by 4.8%, accounting for just 2% of the market.\n\nDiesel trucks continued to dominate in 2024, accounting for 95.1% of new EU registrations, despite a 6.2% decline. Electrically chargeable truck registrations fell by 4.6%, with their market share remaining stable at 2.3% compared to last year.\n\nNew EU electrically chargeable bus registrations rose by 26.8% in 2024, with the market share increasing from 15.9% to 18.5%. Hybrid-electric bus sales fell by 16.1%, making up a 9.8% share. Diesel bus registrations grew by 11.1%, increasing their market share to 63.1%, up 1 percentage point compared to 2023.\n\n### Figure 5. New commercial vehicles registrations by power source"],"full_content":"www.acea.auto 18\n\ndecline of 9.1%, reducing their market share to 6.1% from 7.2% the previous year. Hybrid- electric van sales also fell by 4.8%, accounting for just 2% of the market.\n\nDiesel trucks continued to dominate in 2024, accounting for 95.1% of new EU registrations, despite a 6.2% decline. Electrically chargeable truck registrations fell by 4.6%, with their market share remaining stable at 2.3% compared to last year.\n\nNew EU electrically chargeable bus registrations rose by 26.8% in 2024, with the market share increasing from 15.9% to 18.5%. Hybrid-electric bus sales fell by 16.1%, making up a 9.8% share. Diesel bus registrations grew by 11.1%, increasing their market share to 63.1%, up 1 percentage point compared to 2023.\n\n### Figure 5. New commercial vehicles registrations by power source","publish_date":null,"title":"Figure 5. New commercial vehicles registrations by power source","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Economic_and_Market_Report-Full_year-2024_p18.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Economic_and_Market_Report-Full_year-2024_p18.pdf"},"request":{"example_id":"chart/Economic_and_Market_Report-Full_year-2024_p18","product_type":"parse","source_file_path":"Economic_and_Market_Report-Full_year-2024_p18.pdf"},"started_at":"2026-11-04T19:44:52.997991"} +{"completed_at":"2026-11-04T19:44:55.303658","latency_in_ms":1116,"output":{"example_id":"chart/Economic_and_Market_Report-Full_year_2023_p11","grounded_pages":[],"layout_pages":[],"markdown":"www.acea.auto 11\n\n### Figure 3: Top 10 global car producers\n\nSOURCE: **S&P GLOBAL MOBILITY**","pages":[{"line_numbers":[],"markdown":"www.acea.auto 11\n\n### Figure 3: Top 10 global car producers\n\nSOURCE: **S&P GLOBAL MOBILITY**","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"www.acea.auto 11\n\n### Figure 3: Top 10 global car producers\n\nSOURCE: **S&P GLOBAL MOBILITY**","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Economic_and_Market_Report-Full_year_2023_p11.pdf"]},"response":{"errors":[],"extract_id":"extract_1f33cdc3922824aea60351d808a96ba9","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["www.acea.auto 11\n\n### Figure 3: Top 10 global car producers\n\nSOURCE: **S&P GLOBAL MOBILITY**"],"full_content":"www.acea.auto 11\n\n### Figure 3: Top 10 global car producers\n\nSOURCE: **S&P GLOBAL MOBILITY**","publish_date":null,"title":"Figure 3: Top 10 global car producers","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Economic_and_Market_Report-Full_year_2023_p11.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Economic_and_Market_Report-Full_year_2023_p11.pdf"},"request":{"example_id":"chart/Economic_and_Market_Report-Full_year_2023_p11","product_type":"parse","source_file_path":"Economic_and_Market_Report-Full_year_2023_p11.pdf"},"started_at":"2026-11-04T19:44:54.187497"} +{"completed_at":"2026-11-04T19:44:57.127108","latency_in_ms":1213,"output":{"example_id":"chart/EnergyEfficiency2024_p83","grounded_pages":[],"layout_pages":[],"markdown":"Energy Efficiency 2024 Chapter 3. System-wide themes\n\nPAGE | 83\n\nI EA. CC BY 4.0.\n\n# **3\\.3 Finance**\n\n## **Efficiency investment is resilient and projected to reach about USD 660 billion in 2024**\n\nCombined public and private investment in end-use sectors on efficiency, including electrification, is set to increase slightly by nearly 4% in 2024, matching the all-time high in annual investment of about USD 660 billion set in 2022. This is around 10% more than all upstream oil and gas investment in 2024. Investment has risen by a strong 45% since 2019, as the energy crisis and Covid-19 pandemic prompted massive government spending to stimulate consumer investment in energy efficiency. Since 2019, end-use investment in transport risen by an estimated 77%, followed by 34% for buildings and 13% in industry.\n\nMore recently, investment trends have been more mixed. Between 2022 and 2024, investment is estimated to have declined by 7% in the buildings sector and risen by 14% in transport, while it stayed steady in the industrial sector. Spending on efficient electrification is rising rapidly, primarily through EVs in China, Europe and North America. However, most other areas of efficiency spending are declining from recent highs as the effects of high energy prices ease and Covid-19 stimulus spending winds down. As a result, overall global end-use investment has plateaued over the last two years, with higher inflation and interest rates making it more expensive to implement and finance energy efficiency upgrades.\n\n**Global investment in energy efficiency, electrification and renewables for end uses by sector and region, 2019-2024e**\n\nIEA. CC BY 4.0.\n\nNotes: An energy efficiency investment is defined as the incremental spending on new energy-efficient equipment or the full cost of refurbishments that reduce energy use. The intention is to capture spending that leads to reduced energy consumption. 2024e are estimated values. Source: IEA (2024), World Energy Investment.\n\n0\n\n250\n\n500\n\n750\n\n2019 2020 2021 2022 2023 2024e\n\nBillion USD (2022)\n\nIndustry Buildings Transport\n\n0\n\n250\n\n500\n\n750\n\n2019 2020 2021 2022 2023 2024e\n\nOther Asia Pacific (excl. China)\n\nChina Europe\n\nNorth America","pages":[{"line_numbers":[],"markdown":"Energy Efficiency 2024 Chapter 3. System-wide themes\n\nPAGE | 83\n\nI EA. CC BY 4.0.\n\n# **3\\.3 Finance**\n\n## **Efficiency investment is resilient and projected to reach about USD 660 billion in 2024**\n\nCombined public and private investment in end-use sectors on efficiency, including electrification, is set to increase slightly by nearly 4% in 2024, matching the all-time high in annual investment of about USD 660 billion set in 2022. This is around 10% more than all upstream oil and gas investment in 2024. Investment has risen by a strong 45% since 2019, as the energy crisis and Covid-19 pandemic prompted massive government spending to stimulate consumer investment in energy efficiency. Since 2019, end-use investment in transport risen by an estimated 77%, followed by 34% for buildings and 13% in industry.\n\nMore recently, investment trends have been more mixed. Between 2022 and 2024, investment is estimated to have declined by 7% in the buildings sector and risen by 14% in transport, while it stayed steady in the industrial sector. Spending on efficient electrification is rising rapidly, primarily through EVs in China, Europe and North America. However, most other areas of efficiency spending are declining from recent highs as the effects of high energy prices ease and Covid-19 stimulus spending winds down. As a result, overall global end-use investment has plateaued over the last two years, with higher inflation and interest rates making it more expensive to implement and finance energy efficiency upgrades.\n\n**Global investment in energy efficiency, electrification and renewables for end uses by sector and region, 2019-2024e**\n\nIEA. CC BY 4.0.\n\nNotes: An energy efficiency investment is defined as the incremental spending on new energy-efficient equipment or the full cost of refurbishments that reduce energy use. The intention is to capture spending that leads to reduced energy consumption. 2024e are estimated values. Source: IEA (2024), World Energy Investment.\n\n0\n\n250\n\n500\n\n750\n\n2019 2020 2021 2022 2023 2024e\n\nBillion USD (2022)\n\nIndustry Buildings Transport\n\n0\n\n250\n\n500\n\n750\n\n2019 2020 2021 2022 2023 2024e\n\nOther Asia Pacific (excl. China)\n\nChina Europe\n\nNorth America","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Energy Efficiency 2024 Chapter 3. System-wide themes\n\nPAGE | 83\n\nI EA. CC BY 4.0.\n\n# **3\\.3 Finance**\n\n## **Efficiency investment is resilient and projected to reach about USD 660 billion in 2024**\n\nCombined public and private investment in end-use sectors on efficiency, including electrification, is set to increase slightly by nearly 4% in 2024, matching the all-time high in annual investment of about USD 660 billion set in 2022. This is around 10% more than all upstream oil and gas investment in 2024. Investment has risen by a strong 45% since 2019, as the energy crisis and Covid-19 pandemic prompted massive government spending to stimulate consumer investment in energy efficiency. Since 2019, end-use investment in transport risen by an estimated 77%, followed by 34% for buildings and 13% in industry.\n\nMore recently, investment trends have been more mixed. Between 2022 and 2024, investment is estimated to have declined by 7% in the buildings sector and risen by 14% in transport, while it stayed steady in the industrial sector. Spending on efficient electrification is rising rapidly, primarily through EVs in China, Europe and North America. However, most other areas of efficiency spending are declining from recent highs as the effects of high energy prices ease and Covid-19 stimulus spending winds down. As a result, overall global end-use investment has plateaued over the last two years, with higher inflation and interest rates making it more expensive to implement and finance energy efficiency upgrades.\n\n**Global investment in energy efficiency, electrification and renewables for end uses by sector and region, 2019-2024e**\n\nIEA. CC BY 4.0.\n\nNotes: An energy efficiency investment is defined as the incremental spending on new energy-efficient equipment or the full cost of refurbishments that reduce energy use. The intention is to capture spending that leads to reduced energy consumption. 2024e are estimated values. Source: IEA (2024), World Energy Investment.\n\n0\n\n250\n\n500\n\n750\n\n2019 2020 2021 2022 2023 2024e\n\nBillion USD (2022)\n\nIndustry Buildings Transport\n\n0\n\n250\n\n500\n\n750\n\n2019 2020 2021 2022 2023 2024e\n\nOther Asia Pacific (excl. China)\n\nChina Europe\n\nNorth America","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/EnergyEfficiency2024_p83.pdf"]},"response":{"errors":[],"extract_id":"extract_8861cd50fb568c4ec20138b29aa1bb34","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Energy Efficiency 2024 Chapter 3. System-wide themes\n\nPAGE | 83\n\nI EA. CC BY 4.0.\n\n# **3\\.3 Finance**\n\n## **Efficiency investment is resilient and projected to reach about USD 660 billion in 2024**\n\nCombined public and private investment in end-use sectors on efficiency, including electrification, is set to increase slightly by nearly 4% in 2024, matching the all-time high in annual investment of about USD 660 billion set in 2022. This is around 10% more than all upstream oil and gas investment in 2024. Investment has risen by a strong 45% since 2019, as the energy crisis and Covid-19 pandemic prompted massive government spending to stimulate consumer investment in energy efficiency. Since 2019, end-use investment in transport risen by an estimated 77%, followed by 34% for buildings and 13% in industry.\n\nMore recently, investment trends have been more mixed. Between 2022 and 2024, investment is estimated to have declined by 7% in the buildings sector and risen by 14% in transport, while it stayed steady in the industrial sector. Spending on efficient electrification is rising rapidly, primarily through EVs in China, Europe and North America. However, most other areas of efficiency spending are declining from recent highs as the effects of high energy prices ease and Covid-19 stimulus spending winds down. As a result, overall global end-use investment has plateaued over the last two years, with higher inflation and interest rates making it more expensive to implement and finance energy efficiency upgrades.\n\n**Global investment in energy efficiency, electrification and renewables for end uses by sector and region, 2019-2024e**\n\nIEA. CC BY 4.0.\n\nNotes: An energy efficiency investment is defined as the incremental spending on new energy-efficient equipment or the full cost of refurbishments that reduce energy use. The intention is to capture spending that leads to reduced energy consumption. 2024e are estimated values. Source: IEA (2024), World Energy Investment.\n\n0\n\n250\n\n500\n\n750\n\n2019 2020 2021 2022 2023 2024e\n\nBillion USD (2022)\n\nIndustry Buildings Transport\n\n0\n\n250\n\n500\n\n750\n\n2019 2020 2021 2022 2023 2024e\n\nOther Asia Pacific (excl. China)\n\nChina Europe\n\nNorth America"],"full_content":"Energy Efficiency 2024 Chapter 3. System-wide themes\n\nPAGE | 83\n\nI EA. CC BY 4.0.\n\n# **3\\.3 Finance**\n\n## **Efficiency investment is resilient and projected to reach about USD 660 billion in 2024**\n\nCombined public and private investment in end-use sectors on efficiency, including electrification, is set to increase slightly by nearly 4% in 2024, matching the all-time high in annual investment of about USD 660 billion set in 2022. This is around 10% more than all upstream oil and gas investment in 2024. Investment has risen by a strong 45% since 2019, as the energy crisis and Covid-19 pandemic prompted massive government spending to stimulate consumer investment in energy efficiency. Since 2019, end-use investment in transport risen by an estimated 77%, followed by 34% for buildings and 13% in industry.\n\nMore recently, investment trends have been more mixed. Between 2022 and 2024, investment is estimated to have declined by 7% in the buildings sector and risen by 14% in transport, while it stayed steady in the industrial sector. Spending on efficient electrification is rising rapidly, primarily through EVs in China, Europe and North America. However, most other areas of efficiency spending are declining from recent highs as the effects of high energy prices ease and Covid-19 stimulus spending winds down. As a result, overall global end-use investment has plateaued over the last two years, with higher inflation and interest rates making it more expensive to implement and finance energy efficiency upgrades.\n\n**Global investment in energy efficiency, electrification and renewables for end uses by sector and region, 2019-2024e**\n\nIEA. CC BY 4.0.\n\nNotes: An energy efficiency investment is defined as the incremental spending on new energy-efficient equipment or the full cost of refurbishments that reduce energy use. The intention is to capture spending that leads to reduced energy consumption. 2024e are estimated values. Source: IEA (2024), World Energy Investment.\n\n0\n\n250\n\n500\n\n750\n\n2019 2020 2021 2022 2023 2024e\n\nBillion USD (2022)\n\nIndustry Buildings Transport\n\n0\n\n250\n\n500\n\n750\n\n2019 2020 2021 2022 2023 2024e\n\nOther Asia Pacific (excl. China)\n\nChina Europe\n\nNorth America","publish_date":null,"title":"3.3 Finance","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/EnergyEfficiency2024_p83.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/EnergyEfficiency2024_p83.pdf"},"request":{"example_id":"chart/EnergyEfficiency2024_p83","product_type":"parse","source_file_path":"EnergyEfficiency2024_p83.pdf"},"started_at":"2026-11-04T19:44:55.913662"} +{"completed_at":"2026-11-04T19:44:57.647247","latency_in_ms":1236,"output":{"example_id":"chart/EnergyEfficiency2024_p90","grounded_pages":[],"layout_pages":[],"markdown":"Energy Efficiency 2024 Chapter 3. System-wide themes\n\nPAGE | 90\n\nI EA. CC BY 4.0.\n\nand industry. Retrofits of energy-intensive buildings – such as hospitals, shopping malls, office buildings, schools, and universities – as well as district heating and cooling also offer quick wins to accelerate energy efficiency progress. Behavioural change also plays an important role and can be supported through investments in public transport systems and digitally-enabled devices such as smart thermostats.\n\n## **Which sectors and regions are key for increasing investment?**\n\nWhile investment in energy efficiency has risen by 50% compared with 2019 levels, spending has not been evenly distributed around the world. Efficiency investment is highly concentrated, with almost 90% of spending occurring in China and advanced economies.\n\nIn the NZE Scenario, energy efficiency investments in advanced economies almost double by 2030, while in China and other EMDEs, investments grow four to seven times, compared to today. Due to the rapid rate of urbanisation in most developing economies and the need to construct highly efficient zero-carbon- ready buildings, the investment seen in the buildings sector stands out in the NZE Scenario. A sixfold increase in spending is seen in China and even higher increases in other EMDEs. While industry is one of the most difficult sectors to decarbonise, its funding often yields the best results. Key actions include upgrading facilities to handle more recycled materials; electrification, especially in light industry; and switching to efficient, electric motor-driven systems.\n\n**Energy investment in end-use sectors, 2024e, and Net Zero Emissions by 2050 Scenario, 2030**\n\nIEA. CC BY 4.0\n\nNotes: NZE = Net Zero Emissions by 2050 Scenario. An energy efficiency investment is defined as the incremental spending on new energy-efficient equipment or the full cost of refurbishments that reduce energy use. The intention is to capture spending that leads to reduced energy consumption. 2024e = estimated values. Sources: IEA (2024), World Energy Investment 2024, and World Energy Outlook 2024.\n\n0 200 400 600 800 1 000 1 200 1 400 1 600 1 800 2 000\n\nRegions\n\nSectors\n\nRegions\n\nSectors\n\nRegions\n\nSectors\n\n2030 NZE Change 2024e\n\nAnnual investment (Billion USD, 2023)\n\nIndustry\n\nBuildings\n\nTransport\n\nAdvanced economies\n\nEMDE (excl. China)\n\nChina","pages":[{"line_numbers":[],"markdown":"Energy Efficiency 2024 Chapter 3. System-wide themes\n\nPAGE | 90\n\nI EA. CC BY 4.0.\n\nand industry. Retrofits of energy-intensive buildings – such as hospitals, shopping malls, office buildings, schools, and universities – as well as district heating and cooling also offer quick wins to accelerate energy efficiency progress. Behavioural change also plays an important role and can be supported through investments in public transport systems and digitally-enabled devices such as smart thermostats.\n\n## **Which sectors and regions are key for increasing investment?**\n\nWhile investment in energy efficiency has risen by 50% compared with 2019 levels, spending has not been evenly distributed around the world. Efficiency investment is highly concentrated, with almost 90% of spending occurring in China and advanced economies.\n\nIn the NZE Scenario, energy efficiency investments in advanced economies almost double by 2030, while in China and other EMDEs, investments grow four to seven times, compared to today. Due to the rapid rate of urbanisation in most developing economies and the need to construct highly efficient zero-carbon- ready buildings, the investment seen in the buildings sector stands out in the NZE Scenario. A sixfold increase in spending is seen in China and even higher increases in other EMDEs. While industry is one of the most difficult sectors to decarbonise, its funding often yields the best results. Key actions include upgrading facilities to handle more recycled materials; electrification, especially in light industry; and switching to efficient, electric motor-driven systems.\n\n**Energy investment in end-use sectors, 2024e, and Net Zero Emissions by 2050 Scenario, 2030**\n\nIEA. CC BY 4.0\n\nNotes: NZE = Net Zero Emissions by 2050 Scenario. An energy efficiency investment is defined as the incremental spending on new energy-efficient equipment or the full cost of refurbishments that reduce energy use. The intention is to capture spending that leads to reduced energy consumption. 2024e = estimated values. Sources: IEA (2024), World Energy Investment 2024, and World Energy Outlook 2024.\n\n0 200 400 600 800 1 000 1 200 1 400 1 600 1 800 2 000\n\nRegions\n\nSectors\n\nRegions\n\nSectors\n\nRegions\n\nSectors\n\n2030 NZE Change 2024e\n\nAnnual investment (Billion USD, 2023)\n\nIndustry\n\nBuildings\n\nTransport\n\nAdvanced economies\n\nEMDE (excl. China)\n\nChina","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Energy Efficiency 2024 Chapter 3. System-wide themes\n\nPAGE | 90\n\nI EA. CC BY 4.0.\n\nand industry. Retrofits of energy-intensive buildings – such as hospitals, shopping malls, office buildings, schools, and universities – as well as district heating and cooling also offer quick wins to accelerate energy efficiency progress. Behavioural change also plays an important role and can be supported through investments in public transport systems and digitally-enabled devices such as smart thermostats.\n\n## **Which sectors and regions are key for increasing investment?**\n\nWhile investment in energy efficiency has risen by 50% compared with 2019 levels, spending has not been evenly distributed around the world. Efficiency investment is highly concentrated, with almost 90% of spending occurring in China and advanced economies.\n\nIn the NZE Scenario, energy efficiency investments in advanced economies almost double by 2030, while in China and other EMDEs, investments grow four to seven times, compared to today. Due to the rapid rate of urbanisation in most developing economies and the need to construct highly efficient zero-carbon- ready buildings, the investment seen in the buildings sector stands out in the NZE Scenario. A sixfold increase in spending is seen in China and even higher increases in other EMDEs. While industry is one of the most difficult sectors to decarbonise, its funding often yields the best results. Key actions include upgrading facilities to handle more recycled materials; electrification, especially in light industry; and switching to efficient, electric motor-driven systems.\n\n**Energy investment in end-use sectors, 2024e, and Net Zero Emissions by 2050 Scenario, 2030**\n\nIEA. CC BY 4.0\n\nNotes: NZE = Net Zero Emissions by 2050 Scenario. An energy efficiency investment is defined as the incremental spending on new energy-efficient equipment or the full cost of refurbishments that reduce energy use. The intention is to capture spending that leads to reduced energy consumption. 2024e = estimated values. Sources: IEA (2024), World Energy Investment 2024, and World Energy Outlook 2024.\n\n0 200 400 600 800 1 000 1 200 1 400 1 600 1 800 2 000\n\nRegions\n\nSectors\n\nRegions\n\nSectors\n\nRegions\n\nSectors\n\n2030 NZE Change 2024e\n\nAnnual investment (Billion USD, 2023)\n\nIndustry\n\nBuildings\n\nTransport\n\nAdvanced economies\n\nEMDE (excl. China)\n\nChina","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/EnergyEfficiency2024_p90.pdf"]},"response":{"errors":[],"extract_id":"extract_d4719ee7b46f550ff46348d14a65512a","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Energy Efficiency 2024 Chapter 3. System-wide themes\n\nPAGE | 90\n\nI EA. CC BY 4.0.\n\nand industry. Retrofits of energy-intensive buildings – such as hospitals, shopping malls, office buildings, schools, and universities – as well as district heating and cooling also offer quick wins to accelerate energy efficiency progress. Behavioural change also plays an important role and can be supported through investments in public transport systems and digitally-enabled devices such as smart thermostats.\n\n## **Which sectors and regions are key for increasing investment?**\n\nWhile investment in energy efficiency has risen by 50% compared with 2019 levels, spending has not been evenly distributed around the world. Efficiency investment is highly concentrated, with almost 90% of spending occurring in China and advanced economies.\n\nIn the NZE Scenario, energy efficiency investments in advanced economies almost double by 2030, while in China and other EMDEs, investments grow four to seven times, compared to today. Due to the rapid rate of urbanisation in most developing economies and the need to construct highly efficient zero-carbon- ready buildings, the investment seen in the buildings sector stands out in the NZE Scenario. A sixfold increase in spending is seen in China and even higher increases in other EMDEs. While industry is one of the most difficult sectors to decarbonise, its funding often yields the best results. Key actions include upgrading facilities to handle more recycled materials; electrification, especially in light industry; and switching to efficient, electric motor-driven systems.\n\n**Energy investment in end-use sectors, 2024e, and Net Zero Emissions by 2050 Scenario, 2030**\n\nIEA. CC BY 4.0\n\nNotes: NZE = Net Zero Emissions by 2050 Scenario. An energy efficiency investment is defined as the incremental spending on new energy-efficient equipment or the full cost of refurbishments that reduce energy use. The intention is to capture spending that leads to reduced energy consumption. 2024e = estimated values. Sources: IEA (2024), World Energy Investment 2024, and World Energy Outlook 2024.\n\n0 200 400 600 800 1 000 1 200 1 400 1 600 1 800 2 000\n\nRegions\n\nSectors\n\nRegions\n\nSectors\n\nRegions\n\nSectors\n\n2030 NZE Change 2024e\n\nAnnual investment (Billion USD, 2023)\n\nIndustry\n\nBuildings\n\nTransport\n\nAdvanced economies\n\nEMDE (excl. China)\n\nChina"],"full_content":"Energy Efficiency 2024 Chapter 3. System-wide themes\n\nPAGE | 90\n\nI EA. CC BY 4.0.\n\nand industry. Retrofits of energy-intensive buildings – such as hospitals, shopping malls, office buildings, schools, and universities – as well as district heating and cooling also offer quick wins to accelerate energy efficiency progress. Behavioural change also plays an important role and can be supported through investments in public transport systems and digitally-enabled devices such as smart thermostats.\n\n## **Which sectors and regions are key for increasing investment?**\n\nWhile investment in energy efficiency has risen by 50% compared with 2019 levels, spending has not been evenly distributed around the world. Efficiency investment is highly concentrated, with almost 90% of spending occurring in China and advanced economies.\n\nIn the NZE Scenario, energy efficiency investments in advanced economies almost double by 2030, while in China and other EMDEs, investments grow four to seven times, compared to today. Due to the rapid rate of urbanisation in most developing economies and the need to construct highly efficient zero-carbon- ready buildings, the investment seen in the buildings sector stands out in the NZE Scenario. A sixfold increase in spending is seen in China and even higher increases in other EMDEs. While industry is one of the most difficult sectors to decarbonise, its funding often yields the best results. Key actions include upgrading facilities to handle more recycled materials; electrification, especially in light industry; and switching to efficient, electric motor-driven systems.\n\n**Energy investment in end-use sectors, 2024e, and Net Zero Emissions by 2050 Scenario, 2030**\n\nIEA. CC BY 4.0\n\nNotes: NZE = Net Zero Emissions by 2050 Scenario. An energy efficiency investment is defined as the incremental spending on new energy-efficient equipment or the full cost of refurbishments that reduce energy use. The intention is to capture spending that leads to reduced energy consumption. 2024e = estimated values. Sources: IEA (2024), World Energy Investment 2024, and World Energy Outlook 2024.\n\n0 200 400 600 800 1 000 1 200 1 400 1 600 1 800 2 000\n\nRegions\n\nSectors\n\nRegions\n\nSectors\n\nRegions\n\nSectors\n\n2030 NZE Change 2024e\n\nAnnual investment (Billion USD, 2023)\n\nIndustry\n\nBuildings\n\nTransport\n\nAdvanced economies\n\nEMDE (excl. China)\n\nChina","publish_date":null,"title":"Which sectors and regions are key for increasing investment?","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/EnergyEfficiency2024_p90.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/EnergyEfficiency2024_p90.pdf"},"request":{"example_id":"chart/EnergyEfficiency2024_p90","product_type":"parse","source_file_path":"EnergyEfficiency2024_p90.pdf"},"started_at":"2026-11-04T19:44:56.410547"} +{"completed_at":"2026-11-04T19:44:59.579947","latency_in_ms":1878,"output":{"example_id":"chart/EnergyTechnologyPerspectives2024_p35","grounded_pages":[],"layout_pages":[],"markdown":"Energy Technology Perspectives 2024 Chapter 1: The state of manufacturing and trade\n\nPAGE | 33\n\nI EA. CC BY 4.0.\n\n**Manufacturing sector investment by country/region, 2005-2023**\n\nIEA. CC BY 4.0.\n\nSources: IEA analysis based on Oxford Economics Limited (2024a).\n\n**Manufacturing investment has risen most rapidly in China over the past two decades, while the share of investment in GDP has stagnated in major advanced economies.**\n\n### Investment in clean energy technology supply chains\n\nGlobal investment in manufacturing in the five key clean technology supply chains this report focuses on – solar PV, wind, EVs (including batteries), electrolysers and heat pumps – jumped 50% to USD 235 billion in 2023, up from USD 160 billion in 2022 (Figure 1.3). Investments were led by solar PV and batteries, which together accounted for 80% of the total in 2023. China accounted for nearly three-quarters of total investment in 2023, with the United States and the European Union together accounting for around one-fifth. India, Japan, Korea and Southeast Asia accounted for most of the rest, with virtually no investment taking place in either Africa or Central and South America.\n\nOverall investments in the steel, aluminium and ammonia industries increased more slowly than in clean energy technologies in 2023, rising from around USD 50 billion to just under USD 60 billion globally. Investment in aluminium capacity halved, while that in ammonia facilities decreased slightly, whereas investment in steel jumped by nearly 65%. Annual investment figures for these industries depend on the spending associated with the scheduling of a handful of projects and so tends to be cyclical. The majority of investment in these three industries took place in emerging markets and developing economies (EMDEs).\n\n0\n\n2\n\n4\n\n6\n\n8\n\n2005 2023\n\nTrillion USD (2023, MER)\n\nChina European Union India Japan United States Rest of World\n\n0%\n\n5%\n\n10%\n\n15%\n\n20%\n\n2005 2023\n\n**Investment Share of GDP**","pages":[{"line_numbers":[],"markdown":"Energy Technology Perspectives 2024 Chapter 1: The state of manufacturing and trade\n\nPAGE | 33\n\nI EA. CC BY 4.0.\n\n**Manufacturing sector investment by country/region, 2005-2023**\n\nIEA. CC BY 4.0.\n\nSources: IEA analysis based on Oxford Economics Limited (2024a).\n\n**Manufacturing investment has risen most rapidly in China over the past two decades, while the share of investment in GDP has stagnated in major advanced economies.**\n\n### Investment in clean energy technology supply chains\n\nGlobal investment in manufacturing in the five key clean technology supply chains this report focuses on – solar PV, wind, EVs (including batteries), electrolysers and heat pumps – jumped 50% to USD 235 billion in 2023, up from USD 160 billion in 2022 (Figure 1.3). Investments were led by solar PV and batteries, which together accounted for 80% of the total in 2023. China accounted for nearly three-quarters of total investment in 2023, with the United States and the European Union together accounting for around one-fifth. India, Japan, Korea and Southeast Asia accounted for most of the rest, with virtually no investment taking place in either Africa or Central and South America.\n\nOverall investments in the steel, aluminium and ammonia industries increased more slowly than in clean energy technologies in 2023, rising from around USD 50 billion to just under USD 60 billion globally. Investment in aluminium capacity halved, while that in ammonia facilities decreased slightly, whereas investment in steel jumped by nearly 65%. Annual investment figures for these industries depend on the spending associated with the scheduling of a handful of projects and so tends to be cyclical. The majority of investment in these three industries took place in emerging markets and developing economies (EMDEs).\n\n0\n\n2\n\n4\n\n6\n\n8\n\n2005 2023\n\nTrillion USD (2023, MER)\n\nChina European Union India Japan United States Rest of World\n\n0%\n\n5%\n\n10%\n\n15%\n\n20%\n\n2005 2023\n\n**Investment Share of GDP**","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Energy Technology Perspectives 2024 Chapter 1: The state of manufacturing and trade\n\nPAGE | 33\n\nI EA. CC BY 4.0.\n\n**Manufacturing sector investment by country/region, 2005-2023**\n\nIEA. CC BY 4.0.\n\nSources: IEA analysis based on Oxford Economics Limited (2024a).\n\n**Manufacturing investment has risen most rapidly in China over the past two decades, while the share of investment in GDP has stagnated in major advanced economies.**\n\n### Investment in clean energy technology supply chains\n\nGlobal investment in manufacturing in the five key clean technology supply chains this report focuses on – solar PV, wind, EVs (including batteries), electrolysers and heat pumps – jumped 50% to USD 235 billion in 2023, up from USD 160 billion in 2022 (Figure 1.3). Investments were led by solar PV and batteries, which together accounted for 80% of the total in 2023. China accounted for nearly three-quarters of total investment in 2023, with the United States and the European Union together accounting for around one-fifth. India, Japan, Korea and Southeast Asia accounted for most of the rest, with virtually no investment taking place in either Africa or Central and South America.\n\nOverall investments in the steel, aluminium and ammonia industries increased more slowly than in clean energy technologies in 2023, rising from around USD 50 billion to just under USD 60 billion globally. Investment in aluminium capacity halved, while that in ammonia facilities decreased slightly, whereas investment in steel jumped by nearly 65%. Annual investment figures for these industries depend on the spending associated with the scheduling of a handful of projects and so tends to be cyclical. The majority of investment in these three industries took place in emerging markets and developing economies (EMDEs).\n\n0\n\n2\n\n4\n\n6\n\n8\n\n2005 2023\n\nTrillion USD (2023, MER)\n\nChina European Union India Japan United States Rest of World\n\n0%\n\n5%\n\n10%\n\n15%\n\n20%\n\n2005 2023\n\n**Investment Share of GDP**","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/EnergyTechnologyPerspectives2024_p35.pdf"]},"response":{"errors":[],"extract_id":"extract_0c7b9988635b4bb94c4601b9d72eaac5","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Energy Technology Perspectives 2024 Chapter 1: The state of manufacturing and trade\n\nPAGE | 33\n\nI EA. CC BY 4.0.\n\n**Manufacturing sector investment by country/region, 2005-2023**\n\nIEA. CC BY 4.0.\n\nSources: IEA analysis based on Oxford Economics Limited (2024a).\n\n**Manufacturing investment has risen most rapidly in China over the past two decades, while the share of investment in GDP has stagnated in major advanced economies.**\n\n### Investment in clean energy technology supply chains\n\nGlobal investment in manufacturing in the five key clean technology supply chains this report focuses on – solar PV, wind, EVs (including batteries), electrolysers and heat pumps – jumped 50% to USD 235 billion in 2023, up from USD 160 billion in 2022 (Figure 1.3). Investments were led by solar PV and batteries, which together accounted for 80% of the total in 2023. China accounted for nearly three-quarters of total investment in 2023, with the United States and the European Union together accounting for around one-fifth. India, Japan, Korea and Southeast Asia accounted for most of the rest, with virtually no investment taking place in either Africa or Central and South America.\n\nOverall investments in the steel, aluminium and ammonia industries increased more slowly than in clean energy technologies in 2023, rising from around USD 50 billion to just under USD 60 billion globally. Investment in aluminium capacity halved, while that in ammonia facilities decreased slightly, whereas investment in steel jumped by nearly 65%. Annual investment figures for these industries depend on the spending associated with the scheduling of a handful of projects and so tends to be cyclical. The majority of investment in these three industries took place in emerging markets and developing economies (EMDEs).\n\n0\n\n2\n\n4\n\n6\n\n8\n\n2005 2023\n\nTrillion USD (2023, MER)\n\nChina European Union India Japan United States Rest of World\n\n0%\n\n5%\n\n10%\n\n15%\n\n20%\n\n2005 2023\n\n**Investment Share of GDP**"],"full_content":"Energy Technology Perspectives 2024 Chapter 1: The state of manufacturing and trade\n\nPAGE | 33\n\nI EA. CC BY 4.0.\n\n**Manufacturing sector investment by country/region, 2005-2023**\n\nIEA. CC BY 4.0.\n\nSources: IEA analysis based on Oxford Economics Limited (2024a).\n\n**Manufacturing investment has risen most rapidly in China over the past two decades, while the share of investment in GDP has stagnated in major advanced economies.**\n\n### Investment in clean energy technology supply chains\n\nGlobal investment in manufacturing in the five key clean technology supply chains this report focuses on – solar PV, wind, EVs (including batteries), electrolysers and heat pumps – jumped 50% to USD 235 billion in 2023, up from USD 160 billion in 2022 (Figure 1.3). Investments were led by solar PV and batteries, which together accounted for 80% of the total in 2023. China accounted for nearly three-quarters of total investment in 2023, with the United States and the European Union together accounting for around one-fifth. India, Japan, Korea and Southeast Asia accounted for most of the rest, with virtually no investment taking place in either Africa or Central and South America.\n\nOverall investments in the steel, aluminium and ammonia industries increased more slowly than in clean energy technologies in 2023, rising from around USD 50 billion to just under USD 60 billion globally. Investment in aluminium capacity halved, while that in ammonia facilities decreased slightly, whereas investment in steel jumped by nearly 65%. Annual investment figures for these industries depend on the spending associated with the scheduling of a handful of projects and so tends to be cyclical. The majority of investment in these three industries took place in emerging markets and developing economies (EMDEs).\n\n0\n\n2\n\n4\n\n6\n\n8\n\n2005 2023\n\nTrillion USD (2023, MER)\n\nChina European Union India Japan United States Rest of World\n\n0%\n\n5%\n\n10%\n\n15%\n\n20%\n\n2005 2023\n\n**Investment Share of GDP**","publish_date":null,"title":"Investment in clean energy technology supply chains","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/EnergyTechnologyPerspectives2024_p35.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/EnergyTechnologyPerspectives2024_p35.pdf"},"request":{"example_id":"chart/EnergyTechnologyPerspectives2024_p35","product_type":"parse","source_file_path":"EnergyTechnologyPerspectives2024_p35.pdf"},"started_at":"2026-11-04T19:44:57.701378"} +{"completed_at":"2026-11-04T19:45:00.313359","latency_in_ms":1387,"output":{"example_id":"chart/EnergyTechnologyPerspectives2024_p36","grounded_pages":[],"layout_pages":[],"markdown":"Energy Technology Perspectives 2024 Chapter 1: The state of manufacturing and trade\n\nPAGE | 34\n\nI EA. CC BY 4.0.\n\n**Global investment in clean energy technology and materials manufacturing, 2022-2023**\n\nIEA. CC BY 4.0.\n\nNotes: FID = final investment decision. Materials includes investment associated with global capacity additions for crude steel and iron for steel, and alumina and primary production for aluminium. Only investments in new capacity are included. Completed projects include all projects in operation at end-2023. FID or under construction is as of end-June 2024. Sources: IEA analysis based on S&P Global (2024); WindEurope (2023); BNEF (2024a); GWEC (2023); (Wood Mackenzie (2024); InfoLink (2024); SPV Market Research (2024); BMI (2024); EV Volumes (2024); BNEF (2024b); BNEF (2024c); IEA (2024a); Atlas EV hub (2024); IFA (2024); OECD (2024b); CRU (2024); Oxford Economics Limited (2024b); CEPII (2024) as well as announcements by manufacturers and personal communications, gathered by the IEA.\n\n**Investments in clean technology manufacturing grew by 50% year-on-year in 2023; much faster than investments in materials manufacturing.**\n\nCumulative investment in near-zero emissions technologies for materials production reached just under USD 6 billion in 2023, with iron and steel accounting for around 90% and ammonia for the rest. A further USD 15 billion of investment is expected to take place in projects that have either reached final investment decision or are already under construction. The majority of this is for ammonia facilities using low-emissions hydrogen for new energy applications like shipping and power generation. Only around USD 300 million of capital investment has been committed to near-zero emission aluminium production, and this is limited to demonstration projects with capacity typically around 1% of that of commercial-scale facilities. Commercial investment targeting emissions reductions in the aluminium industry is mostly limited to the utilisation of greater shares of low-emissions electricity (which reduces indirect emissions), energy efficiency and fuel switching in alumina production.\n\nProjecting forward investments for the project pipeline for clean technologies is highly uncertain, as projects may well be cancelled, postponed or brought forward according to changes in policy incentives and market conditions. Assuming the whole project pipeline is completed and there is no change in the inflation-adjusted cost of building manufacturing facilities, total investment in clean technology manufacturing projects is expected to stand at roughly USD 200 billion (in 2023 dollars) in 2024 (Figure 1.4). Current projects alone point to nearly USD 180 billion per year of investment over the period 2025-30, around 35% of which is\n\n0\n\n5\n\n10\n\n15\n\n20\n\nCompleted\n\nprojects, 2023\n\nFID or under construction\n\n0\n\n50\n\n100\n\n150\n\n200\n\n250\n\n2022 2023\n\nBillion USD (2023, MER)\n\nBatteries Electric vehicles Solar PV Wind Other\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\n2022 2023\n\nAluminium Ammonia Steel\n\n**Clean technologies Materials (total) Materials**\n\n**(near-zero emissions)**","pages":[{"line_numbers":[],"markdown":"Energy Technology Perspectives 2024 Chapter 1: The state of manufacturing and trade\n\nPAGE | 34\n\nI EA. CC BY 4.0.\n\n**Global investment in clean energy technology and materials manufacturing, 2022-2023**\n\nIEA. CC BY 4.0.\n\nNotes: FID = final investment decision. Materials includes investment associated with global capacity additions for crude steel and iron for steel, and alumina and primary production for aluminium. Only investments in new capacity are included. Completed projects include all projects in operation at end-2023. FID or under construction is as of end-June 2024. Sources: IEA analysis based on S&P Global (2024); WindEurope (2023); BNEF (2024a); GWEC (2023); (Wood Mackenzie (2024); InfoLink (2024); SPV Market Research (2024); BMI (2024); EV Volumes (2024); BNEF (2024b); BNEF (2024c); IEA (2024a); Atlas EV hub (2024); IFA (2024); OECD (2024b); CRU (2024); Oxford Economics Limited (2024b); CEPII (2024) as well as announcements by manufacturers and personal communications, gathered by the IEA.\n\n**Investments in clean technology manufacturing grew by 50% year-on-year in 2023; much faster than investments in materials manufacturing.**\n\nCumulative investment in near-zero emissions technologies for materials production reached just under USD 6 billion in 2023, with iron and steel accounting for around 90% and ammonia for the rest. A further USD 15 billion of investment is expected to take place in projects that have either reached final investment decision or are already under construction. The majority of this is for ammonia facilities using low-emissions hydrogen for new energy applications like shipping and power generation. Only around USD 300 million of capital investment has been committed to near-zero emission aluminium production, and this is limited to demonstration projects with capacity typically around 1% of that of commercial-scale facilities. Commercial investment targeting emissions reductions in the aluminium industry is mostly limited to the utilisation of greater shares of low-emissions electricity (which reduces indirect emissions), energy efficiency and fuel switching in alumina production.\n\nProjecting forward investments for the project pipeline for clean technologies is highly uncertain, as projects may well be cancelled, postponed or brought forward according to changes in policy incentives and market conditions. Assuming the whole project pipeline is completed and there is no change in the inflation-adjusted cost of building manufacturing facilities, total investment in clean technology manufacturing projects is expected to stand at roughly USD 200 billion (in 2023 dollars) in 2024 (Figure 1.4). Current projects alone point to nearly USD 180 billion per year of investment over the period 2025-30, around 35% of which is\n\n0\n\n5\n\n10\n\n15\n\n20\n\nCompleted\n\nprojects, 2023\n\nFID or under construction\n\n0\n\n50\n\n100\n\n150\n\n200\n\n250\n\n2022 2023\n\nBillion USD (2023, MER)\n\nBatteries Electric vehicles Solar PV Wind Other\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\n2022 2023\n\nAluminium Ammonia Steel\n\n**Clean technologies Materials (total) Materials**\n\n**(near-zero emissions)**","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Energy Technology Perspectives 2024 Chapter 1: The state of manufacturing and trade\n\nPAGE | 34\n\nI EA. CC BY 4.0.\n\n**Global investment in clean energy technology and materials manufacturing, 2022-2023**\n\nIEA. CC BY 4.0.\n\nNotes: FID = final investment decision. Materials includes investment associated with global capacity additions for crude steel and iron for steel, and alumina and primary production for aluminium. Only investments in new capacity are included. Completed projects include all projects in operation at end-2023. FID or under construction is as of end-June 2024. Sources: IEA analysis based on S&P Global (2024); WindEurope (2023); BNEF (2024a); GWEC (2023); (Wood Mackenzie (2024); InfoLink (2024); SPV Market Research (2024); BMI (2024); EV Volumes (2024); BNEF (2024b); BNEF (2024c); IEA (2024a); Atlas EV hub (2024); IFA (2024); OECD (2024b); CRU (2024); Oxford Economics Limited (2024b); CEPII (2024) as well as announcements by manufacturers and personal communications, gathered by the IEA.\n\n**Investments in clean technology manufacturing grew by 50% year-on-year in 2023; much faster than investments in materials manufacturing.**\n\nCumulative investment in near-zero emissions technologies for materials production reached just under USD 6 billion in 2023, with iron and steel accounting for around 90% and ammonia for the rest. A further USD 15 billion of investment is expected to take place in projects that have either reached final investment decision or are already under construction. The majority of this is for ammonia facilities using low-emissions hydrogen for new energy applications like shipping and power generation. Only around USD 300 million of capital investment has been committed to near-zero emission aluminium production, and this is limited to demonstration projects with capacity typically around 1% of that of commercial-scale facilities. Commercial investment targeting emissions reductions in the aluminium industry is mostly limited to the utilisation of greater shares of low-emissions electricity (which reduces indirect emissions), energy efficiency and fuel switching in alumina production.\n\nProjecting forward investments for the project pipeline for clean technologies is highly uncertain, as projects may well be cancelled, postponed or brought forward according to changes in policy incentives and market conditions. Assuming the whole project pipeline is completed and there is no change in the inflation-adjusted cost of building manufacturing facilities, total investment in clean technology manufacturing projects is expected to stand at roughly USD 200 billion (in 2023 dollars) in 2024 (Figure 1.4). Current projects alone point to nearly USD 180 billion per year of investment over the period 2025-30, around 35% of which is\n\n0\n\n5\n\n10\n\n15\n\n20\n\nCompleted\n\nprojects, 2023\n\nFID or under construction\n\n0\n\n50\n\n100\n\n150\n\n200\n\n250\n\n2022 2023\n\nBillion USD (2023, MER)\n\nBatteries Electric vehicles Solar PV Wind Other\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\n2022 2023\n\nAluminium Ammonia Steel\n\n**Clean technologies Materials (total) Materials**\n\n**(near-zero emissions)**","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/EnergyTechnologyPerspectives2024_p36.pdf"]},"response":{"errors":[],"extract_id":"extract_10fa4680e52d8dc054b2e4210e8b9bed","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Energy Technology Perspectives 2024 Chapter 1: The state of manufacturing and trade\n\nPAGE | 34\n\nI EA. CC BY 4.0.\n\n**Global investment in clean energy technology and materials manufacturing, 2022-2023**\n\nIEA. CC BY 4.0.\n\nNotes: FID = final investment decision. Materials includes investment associated with global capacity additions for crude steel and iron for steel, and alumina and primary production for aluminium. Only investments in new capacity are included. Completed projects include all projects in operation at end-2023. FID or under construction is as of end-June 2024. Sources: IEA analysis based on S&P Global (2024); WindEurope (2023); BNEF (2024a); GWEC (2023); (Wood Mackenzie (2024); InfoLink (2024); SPV Market Research (2024); BMI (2024); EV Volumes (2024); BNEF (2024b); BNEF (2024c); IEA (2024a); Atlas EV hub (2024); IFA (2024); OECD (2024b); CRU (2024); Oxford Economics Limited (2024b); CEPII (2024) as well as announcements by manufacturers and personal communications, gathered by the IEA.\n\n**Investments in clean technology manufacturing grew by 50% year-on-year in 2023; much faster than investments in materials manufacturing.**\n\nCumulative investment in near-zero emissions technologies for materials production reached just under USD 6 billion in 2023, with iron and steel accounting for around 90% and ammonia for the rest. A further USD 15 billion of investment is expected to take place in projects that have either reached final investment decision or are already under construction. The majority of this is for ammonia facilities using low-emissions hydrogen for new energy applications like shipping and power generation. Only around USD 300 million of capital investment has been committed to near-zero emission aluminium production, and this is limited to demonstration projects with capacity typically around 1% of that of commercial-scale facilities. Commercial investment targeting emissions reductions in the aluminium industry is mostly limited to the utilisation of greater shares of low-emissions electricity (which reduces indirect emissions), energy efficiency and fuel switching in alumina production.\n\nProjecting forward investments for the project pipeline for clean technologies is highly uncertain, as projects may well be cancelled, postponed or brought forward according to changes in policy incentives and market conditions. Assuming the whole project pipeline is completed and there is no change in the inflation-adjusted cost of building manufacturing facilities, total investment in clean technology manufacturing projects is expected to stand at roughly USD 200 billion (in 2023 dollars) in 2024 (Figure 1.4). Current projects alone point to nearly USD 180 billion per year of investment over the period 2025-30, around 35% of which is\n\n0\n\n5\n\n10\n\n15\n\n20\n\nCompleted\n\nprojects, 2023\n\nFID or under construction\n\n0\n\n50\n\n100\n\n150\n\n200\n\n250\n\n2022 2023\n\nBillion USD (2023, MER)\n\nBatteries Electric vehicles Solar PV Wind Other\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\n2022 2023\n\nAluminium Ammonia Steel\n\n**Clean technologies Materials (total) Materials**\n\n**(near-zero emissions)**"],"full_content":"Energy Technology Perspectives 2024 Chapter 1: The state of manufacturing and trade\n\nPAGE | 34\n\nI EA. CC BY 4.0.\n\n**Global investment in clean energy technology and materials manufacturing, 2022-2023**\n\nIEA. CC BY 4.0.\n\nNotes: FID = final investment decision. Materials includes investment associated with global capacity additions for crude steel and iron for steel, and alumina and primary production for aluminium. Only investments in new capacity are included. Completed projects include all projects in operation at end-2023. FID or under construction is as of end-June 2024. Sources: IEA analysis based on S&P Global (2024); WindEurope (2023); BNEF (2024a); GWEC (2023); (Wood Mackenzie (2024); InfoLink (2024); SPV Market Research (2024); BMI (2024); EV Volumes (2024); BNEF (2024b); BNEF (2024c); IEA (2024a); Atlas EV hub (2024); IFA (2024); OECD (2024b); CRU (2024); Oxford Economics Limited (2024b); CEPII (2024) as well as announcements by manufacturers and personal communications, gathered by the IEA.\n\n**Investments in clean technology manufacturing grew by 50% year-on-year in 2023; much faster than investments in materials manufacturing.**\n\nCumulative investment in near-zero emissions technologies for materials production reached just under USD 6 billion in 2023, with iron and steel accounting for around 90% and ammonia for the rest. A further USD 15 billion of investment is expected to take place in projects that have either reached final investment decision or are already under construction. The majority of this is for ammonia facilities using low-emissions hydrogen for new energy applications like shipping and power generation. Only around USD 300 million of capital investment has been committed to near-zero emission aluminium production, and this is limited to demonstration projects with capacity typically around 1% of that of commercial-scale facilities. Commercial investment targeting emissions reductions in the aluminium industry is mostly limited to the utilisation of greater shares of low-emissions electricity (which reduces indirect emissions), energy efficiency and fuel switching in alumina production.\n\nProjecting forward investments for the project pipeline for clean technologies is highly uncertain, as projects may well be cancelled, postponed or brought forward according to changes in policy incentives and market conditions. Assuming the whole project pipeline is completed and there is no change in the inflation-adjusted cost of building manufacturing facilities, total investment in clean technology manufacturing projects is expected to stand at roughly USD 200 billion (in 2023 dollars) in 2024 (Figure 1.4). Current projects alone point to nearly USD 180 billion per year of investment over the period 2025-30, around 35% of which is\n\n0\n\n5\n\n10\n\n15\n\n20\n\nCompleted\n\nprojects, 2023\n\nFID or under construction\n\n0\n\n50\n\n100\n\n150\n\n200\n\n250\n\n2022 2023\n\nBillion USD (2023, MER)\n\nBatteries Electric vehicles Solar PV Wind Other\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\n2022 2023\n\nAluminium Ammonia Steel\n\n**Clean technologies Materials (total) Materials**\n\n**(near-zero emissions)**","publish_date":null,"title":"Cumulative investment in near-zero emissions technologies for materials production reached just under USD 6 billion in 2023, with iron and","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/EnergyTechnologyPerspectives2024_p36.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/EnergyTechnologyPerspectives2024_p36.pdf"},"request":{"example_id":"chart/EnergyTechnologyPerspectives2024_p36","product_type":"parse","source_file_path":"EnergyTechnologyPerspectives2024_p36.pdf"},"started_at":"2026-11-04T19:44:58.925604"} +{"completed_at":"2026-11-04T19:45:02.216627","latency_in_ms":1609,"output":{"example_id":"chart/EnergyTechnologyPerspectives2024_p38","grounded_pages":[],"layout_pages":[],"markdown":"Energy Technology Perspectives 2024 Chapter 1: The state of manufacturing and trade\n\nPAGE | 36\n\nI EA. CC BY 4.0.\n\napplications. This raises the share of near-zero emissions technology in total ammonia investment to around 85%, up from under 15% in 2023. Of the total investment associated with the project pipelines for materials, less than 25% can be considered committed _._\n\n## **Manufacturing capacity**\n\n### Clean technology manufacturing\n\nClean technology manufacturing has been growing fast since _ETP-2023_ , as have announcements for further capacity additions (IEA, 2023b). Even solar PV manufacturing, a comparatively mature industry, has expanded rapidly over the past 4 years. For example, at the end of 2021 – the base year for _ETP-2023_ – capacity for making solar PV modules stood at just over 450 GW; it is more than doubled to 1.2 TW by the end of 2023 – the base year for _ETP-2024_ (Figure 1.5). At the end of November 2022, announcements of manufacturing capacity expansions for solar PV modules implied an increase in capacity to just 790 GW by 2030. In other words, global manufacturing capacity exceeded 2030 expectations by 50% in the space of only about 12 months.\n\nIn 2020, around 75 GW of PV module manufacturing capacity was added (Figure 1.6). In 2023, this number jumped to 500 GW, well above the record levels of solar PV installations for electricity generation that year, which stood around 425 GW. Manufacturing capacity additions of 430 GW in China alone in 2023 were larger than the levels of global installations for electricity generation, and larger than the global manufacturing capacity additions between 2020 and 2022.\n\n**Manufacturing capacity and additions associated with announced projects for selected clean technologies**\n\nIEA. CC BY 4.0.\n\nNotes: _ETP-23_ assessed the capacity additions related to announced projects as of the end of November 2022. The cutoff point for _ETP-24_ is the end of June 2024. For batteries, values for 2021 and announcements for _ETP-24_ may differ slightly from _ETP-23_ due to the inclusion of Tier 3 (See the Annex) battery makers for comparability with _ETP-24._ Source: IEA analysis based on IEA (2023b); and IEA (2024a).\n\n**Since the last edition of** **_ETP_** **, manufacturing capacity and announced capacity additions for most technologies have expanded substantially.**\n\n0\n\n5\n\n10\n\nETP-23 ETP-24\n\nBattery cells\n\nTWh\n\nCapacity in 2021 Capacity in 2023 Announcements to 2030\n\n0\n\n1\n\n2\n\nETP-23 ETP-24\n\nSolar PV modules\n\nTW\n\n0\n\n100\n\n200\n\n300\n\nETP-23 ETP-24 ETP-23 ETP-24 ETP-23 ETP-24\n\nWind nacelles Electrolysers Heat pumps\n\nGW","pages":[{"line_numbers":[],"markdown":"Energy Technology Perspectives 2024 Chapter 1: The state of manufacturing and trade\n\nPAGE | 36\n\nI EA. CC BY 4.0.\n\napplications. This raises the share of near-zero emissions technology in total ammonia investment to around 85%, up from under 15% in 2023. Of the total investment associated with the project pipelines for materials, less than 25% can be considered committed _._\n\n## **Manufacturing capacity**\n\n### Clean technology manufacturing\n\nClean technology manufacturing has been growing fast since _ETP-2023_ , as have announcements for further capacity additions (IEA, 2023b). Even solar PV manufacturing, a comparatively mature industry, has expanded rapidly over the past 4 years. For example, at the end of 2021 – the base year for _ETP-2023_ – capacity for making solar PV modules stood at just over 450 GW; it is more than doubled to 1.2 TW by the end of 2023 – the base year for _ETP-2024_ (Figure 1.5). At the end of November 2022, announcements of manufacturing capacity expansions for solar PV modules implied an increase in capacity to just 790 GW by 2030. In other words, global manufacturing capacity exceeded 2030 expectations by 50% in the space of only about 12 months.\n\nIn 2020, around 75 GW of PV module manufacturing capacity was added (Figure 1.6). In 2023, this number jumped to 500 GW, well above the record levels of solar PV installations for electricity generation that year, which stood around 425 GW. Manufacturing capacity additions of 430 GW in China alone in 2023 were larger than the levels of global installations for electricity generation, and larger than the global manufacturing capacity additions between 2020 and 2022.\n\n**Manufacturing capacity and additions associated with announced projects for selected clean technologies**\n\nIEA. CC BY 4.0.\n\nNotes: _ETP-23_ assessed the capacity additions related to announced projects as of the end of November 2022. The cutoff point for _ETP-24_ is the end of June 2024. For batteries, values for 2021 and announcements for _ETP-24_ may differ slightly from _ETP-23_ due to the inclusion of Tier 3 (See the Annex) battery makers for comparability with _ETP-24._ Source: IEA analysis based on IEA (2023b); and IEA (2024a).\n\n**Since the last edition of** **_ETP_** **, manufacturing capacity and announced capacity additions for most technologies have expanded substantially.**\n\n0\n\n5\n\n10\n\nETP-23 ETP-24\n\nBattery cells\n\nTWh\n\nCapacity in 2021 Capacity in 2023 Announcements to 2030\n\n0\n\n1\n\n2\n\nETP-23 ETP-24\n\nSolar PV modules\n\nTW\n\n0\n\n100\n\n200\n\n300\n\nETP-23 ETP-24 ETP-23 ETP-24 ETP-23 ETP-24\n\nWind nacelles Electrolysers Heat pumps\n\nGW","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Energy Technology Perspectives 2024 Chapter 1: The state of manufacturing and trade\n\nPAGE | 36\n\nI EA. CC BY 4.0.\n\napplications. This raises the share of near-zero emissions technology in total ammonia investment to around 85%, up from under 15% in 2023. Of the total investment associated with the project pipelines for materials, less than 25% can be considered committed _._\n\n## **Manufacturing capacity**\n\n### Clean technology manufacturing\n\nClean technology manufacturing has been growing fast since _ETP-2023_ , as have announcements for further capacity additions (IEA, 2023b). Even solar PV manufacturing, a comparatively mature industry, has expanded rapidly over the past 4 years. For example, at the end of 2021 – the base year for _ETP-2023_ – capacity for making solar PV modules stood at just over 450 GW; it is more than doubled to 1.2 TW by the end of 2023 – the base year for _ETP-2024_ (Figure 1.5). At the end of November 2022, announcements of manufacturing capacity expansions for solar PV modules implied an increase in capacity to just 790 GW by 2030. In other words, global manufacturing capacity exceeded 2030 expectations by 50% in the space of only about 12 months.\n\nIn 2020, around 75 GW of PV module manufacturing capacity was added (Figure 1.6). In 2023, this number jumped to 500 GW, well above the record levels of solar PV installations for electricity generation that year, which stood around 425 GW. Manufacturing capacity additions of 430 GW in China alone in 2023 were larger than the levels of global installations for electricity generation, and larger than the global manufacturing capacity additions between 2020 and 2022.\n\n**Manufacturing capacity and additions associated with announced projects for selected clean technologies**\n\nIEA. CC BY 4.0.\n\nNotes: _ETP-23_ assessed the capacity additions related to announced projects as of the end of November 2022. The cutoff point for _ETP-24_ is the end of June 2024. For batteries, values for 2021 and announcements for _ETP-24_ may differ slightly from _ETP-23_ due to the inclusion of Tier 3 (See the Annex) battery makers for comparability with _ETP-24._ Source: IEA analysis based on IEA (2023b); and IEA (2024a).\n\n**Since the last edition of** **_ETP_** **, manufacturing capacity and announced capacity additions for most technologies have expanded substantially.**\n\n0\n\n5\n\n10\n\nETP-23 ETP-24\n\nBattery cells\n\nTWh\n\nCapacity in 2021 Capacity in 2023 Announcements to 2030\n\n0\n\n1\n\n2\n\nETP-23 ETP-24\n\nSolar PV modules\n\nTW\n\n0\n\n100\n\n200\n\n300\n\nETP-23 ETP-24 ETP-23 ETP-24 ETP-23 ETP-24\n\nWind nacelles Electrolysers Heat pumps\n\nGW","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/EnergyTechnologyPerspectives2024_p38.pdf"]},"response":{"errors":[],"extract_id":"extract_11788dc837d6ed7cd849b1213b957cdc","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Energy Technology Perspectives 2024 Chapter 1: The state of manufacturing and trade\n\nPAGE | 36\n\nI EA. CC BY 4.0.\n\napplications. This raises the share of near-zero emissions technology in total ammonia investment to around 85%, up from under 15% in 2023. Of the total investment associated with the project pipelines for materials, less than 25% can be considered committed _._\n\n## **Manufacturing capacity**\n\n### Clean technology manufacturing\n\nClean technology manufacturing has been growing fast since _ETP-2023_ , as have announcements for further capacity additions (IEA, 2023b). Even solar PV manufacturing, a comparatively mature industry, has expanded rapidly over the past 4 years. For example, at the end of 2021 – the base year for _ETP-2023_ – capacity for making solar PV modules stood at just over 450 GW; it is more than doubled to 1.2 TW by the end of 2023 – the base year for _ETP-2024_ (Figure 1.5). At the end of November 2022, announcements of manufacturing capacity expansions for solar PV modules implied an increase in capacity to just 790 GW by 2030. In other words, global manufacturing capacity exceeded 2030 expectations by 50% in the space of only about 12 months.\n\nIn 2020, around 75 GW of PV module manufacturing capacity was added (Figure 1.6). In 2023, this number jumped to 500 GW, well above the record levels of solar PV installations for electricity generation that year, which stood around 425 GW. Manufacturing capacity additions of 430 GW in China alone in 2023 were larger than the levels of global installations for electricity generation, and larger than the global manufacturing capacity additions between 2020 and 2022.\n\n**Manufacturing capacity and additions associated with announced projects for selected clean technologies**\n\nIEA. CC BY 4.0.\n\nNotes: _ETP-23_ assessed the capacity additions related to announced projects as of the end of November 2022. The cutoff point for _ETP-24_ is the end of June 2024. For batteries, values for 2021 and announcements for _ETP-24_ may differ slightly from _ETP-23_ due to the inclusion of Tier 3 (See the Annex) battery makers for comparability with _ETP-24._ Source: IEA analysis based on IEA (2023b); and IEA (2024a).\n\n**Since the last edition of** **_ETP_** **, manufacturing capacity and announced capacity additions for most technologies have expanded substantially.**\n\n0\n\n5\n\n10\n\nETP-23 ETP-24\n\nBattery cells\n\nTWh\n\nCapacity in 2021 Capacity in 2023 Announcements to 2030\n\n0\n\n1\n\n2\n\nETP-23 ETP-24\n\nSolar PV modules\n\nTW\n\n0\n\n100\n\n200\n\n300\n\nETP-23 ETP-24 ETP-23 ETP-24 ETP-23 ETP-24\n\nWind nacelles Electrolysers Heat pumps\n\nGW"],"full_content":"Energy Technology Perspectives 2024 Chapter 1: The state of manufacturing and trade\n\nPAGE | 36\n\nI EA. CC BY 4.0.\n\napplications. This raises the share of near-zero emissions technology in total ammonia investment to around 85%, up from under 15% in 2023. Of the total investment associated with the project pipelines for materials, less than 25% can be considered committed _._\n\n## **Manufacturing capacity**\n\n### Clean technology manufacturing\n\nClean technology manufacturing has been growing fast since _ETP-2023_ , as have announcements for further capacity additions (IEA, 2023b). Even solar PV manufacturing, a comparatively mature industry, has expanded rapidly over the past 4 years. For example, at the end of 2021 – the base year for _ETP-2023_ – capacity for making solar PV modules stood at just over 450 GW; it is more than doubled to 1.2 TW by the end of 2023 – the base year for _ETP-2024_ (Figure 1.5). At the end of November 2022, announcements of manufacturing capacity expansions for solar PV modules implied an increase in capacity to just 790 GW by 2030. In other words, global manufacturing capacity exceeded 2030 expectations by 50% in the space of only about 12 months.\n\nIn 2020, around 75 GW of PV module manufacturing capacity was added (Figure 1.6). In 2023, this number jumped to 500 GW, well above the record levels of solar PV installations for electricity generation that year, which stood around 425 GW. Manufacturing capacity additions of 430 GW in China alone in 2023 were larger than the levels of global installations for electricity generation, and larger than the global manufacturing capacity additions between 2020 and 2022.\n\n**Manufacturing capacity and additions associated with announced projects for selected clean technologies**\n\nIEA. CC BY 4.0.\n\nNotes: _ETP-23_ assessed the capacity additions related to announced projects as of the end of November 2022. The cutoff point for _ETP-24_ is the end of June 2024. For batteries, values for 2021 and announcements for _ETP-24_ may differ slightly from _ETP-23_ due to the inclusion of Tier 3 (See the Annex) battery makers for comparability with _ETP-24._ Source: IEA analysis based on IEA (2023b); and IEA (2024a).\n\n**Since the last edition of** **_ETP_** **, manufacturing capacity and announced capacity additions for most technologies have expanded substantially.**\n\n0\n\n5\n\n10\n\nETP-23 ETP-24\n\nBattery cells\n\nTWh\n\nCapacity in 2021 Capacity in 2023 Announcements to 2030\n\n0\n\n1\n\n2\n\nETP-23 ETP-24\n\nSolar PV modules\n\nTW\n\n0\n\n100\n\n200\n\n300\n\nETP-23 ETP-24 ETP-23 ETP-24 ETP-23 ETP-24\n\nWind nacelles Electrolysers Heat pumps\n\nGW","publish_date":null,"title":"Manufacturing capacity","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/EnergyTechnologyPerspectives2024_p38.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/EnergyTechnologyPerspectives2024_p38.pdf"},"request":{"example_id":"chart/EnergyTechnologyPerspectives2024_p38","product_type":"parse","source_file_path":"EnergyTechnologyPerspectives2024_p38.pdf"},"started_at":"2026-11-04T19:45:00.607427"} +{"completed_at":"2026-11-04T19:45:03.239523","latency_in_ms":1277,"output":{"example_id":"chart/EnergyTechnologyPerspectives2024_p39","grounded_pages":[],"layout_pages":[],"markdown":"Energy Technology Perspectives 2024 Chapter 1: The state of manufacturing and trade\n\nPAGE | 37\n\nI EA. CC BY 4.0.\n\n**Net manufacturing capacity additions for selected clean energy technologies by country/region, 2020-2023**\n\nIEA. CC BY 4.0.\n\nNote: RoW = Rest of World. Sources: IEA analysis based on S&P Global (2024); WindEurope (2023); BNEF (2024a); GWEC (2023); Wood Mackenzie (2024); InfoLink (2024); SPV Market Research (2024); BMI (2024); EV Volumes (2024), BNEF (2024b); BNEF (2024c); Oxford Economics Limited (2024b); CEPII (2024); as well as announcements by manufacturers and personal communications, gathered by the IEA.\n\n**Clean technology manufacturing capacity surged in 2023, with only heat pumps and electric vehicles seeing smaller capacity additions than in 2022.**\n\nIn the case of solar PV, manufacturing capacity additions for assembling modules have generally exceeded those of the key components – polysilicon, wafers and cells – with the notable exception of China over the last 2 years (Figure 1.7). However, the utilisation rates of component manufacturing facilities generally remain low, averting bottlenecks. Average utilisation rates across PV module manufacturing facilities worldwide remained steady in 2023, at around 55%. In parallel, facilities for newer technologies like tunnel oxide passivated contact (TOPCon), heterojunction (HJT) and back contact (BC) cells are gaining market share over the older ones like passivated emitter rear cells (PERC). This has driven down prices and led to some downscaling of expansion plans, especially in China (Box 1.2).\n\n**Solar PV (modules)**\n\n0\n\n200\n\n400\n\n600\n\n2020 2021 2022 2023\n\nGW\n\nChina European Union India Japan United States RoW\n\n0\n\n12\n\n24\n\n36\n\n2020 2021 2022 2023\n\n**Wind (nacelles)**\n\n0\n\n250\n\n500\n\n750\n\n1 000\n\n2020 2021 2022 2023\n\nGWh\n\n**Batteries (cells)**\n\n0\n\n5\n\n10\n\n15\n\n2020 2021 2022 2023\n\n**Electrolysers**\n\n0\n\n5\n\n10\n\n15\n\n20\n\n2020 2021 2022 2023\n\nGW\n\n**Heat pumps**\n\n0\n\n2\n\n4\n\n6\n\n8\n\n2020 2021 2022 2023\n\nMillion units\n\n**Electric vehicles**","pages":[{"line_numbers":[],"markdown":"Energy Technology Perspectives 2024 Chapter 1: The state of manufacturing and trade\n\nPAGE | 37\n\nI EA. CC BY 4.0.\n\n**Net manufacturing capacity additions for selected clean energy technologies by country/region, 2020-2023**\n\nIEA. CC BY 4.0.\n\nNote: RoW = Rest of World. Sources: IEA analysis based on S&P Global (2024); WindEurope (2023); BNEF (2024a); GWEC (2023); Wood Mackenzie (2024); InfoLink (2024); SPV Market Research (2024); BMI (2024); EV Volumes (2024), BNEF (2024b); BNEF (2024c); Oxford Economics Limited (2024b); CEPII (2024); as well as announcements by manufacturers and personal communications, gathered by the IEA.\n\n**Clean technology manufacturing capacity surged in 2023, with only heat pumps and electric vehicles seeing smaller capacity additions than in 2022.**\n\nIn the case of solar PV, manufacturing capacity additions for assembling modules have generally exceeded those of the key components – polysilicon, wafers and cells – with the notable exception of China over the last 2 years (Figure 1.7). However, the utilisation rates of component manufacturing facilities generally remain low, averting bottlenecks. Average utilisation rates across PV module manufacturing facilities worldwide remained steady in 2023, at around 55%. In parallel, facilities for newer technologies like tunnel oxide passivated contact (TOPCon), heterojunction (HJT) and back contact (BC) cells are gaining market share over the older ones like passivated emitter rear cells (PERC). This has driven down prices and led to some downscaling of expansion plans, especially in China (Box 1.2).\n\n**Solar PV (modules)**\n\n0\n\n200\n\n400\n\n600\n\n2020 2021 2022 2023\n\nGW\n\nChina European Union India Japan United States RoW\n\n0\n\n12\n\n24\n\n36\n\n2020 2021 2022 2023\n\n**Wind (nacelles)**\n\n0\n\n250\n\n500\n\n750\n\n1 000\n\n2020 2021 2022 2023\n\nGWh\n\n**Batteries (cells)**\n\n0\n\n5\n\n10\n\n15\n\n2020 2021 2022 2023\n\n**Electrolysers**\n\n0\n\n5\n\n10\n\n15\n\n20\n\n2020 2021 2022 2023\n\nGW\n\n**Heat pumps**\n\n0\n\n2\n\n4\n\n6\n\n8\n\n2020 2021 2022 2023\n\nMillion units\n\n**Electric vehicles**","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Energy Technology Perspectives 2024 Chapter 1: The state of manufacturing and trade\n\nPAGE | 37\n\nI EA. CC BY 4.0.\n\n**Net manufacturing capacity additions for selected clean energy technologies by country/region, 2020-2023**\n\nIEA. CC BY 4.0.\n\nNote: RoW = Rest of World. Sources: IEA analysis based on S&P Global (2024); WindEurope (2023); BNEF (2024a); GWEC (2023); Wood Mackenzie (2024); InfoLink (2024); SPV Market Research (2024); BMI (2024); EV Volumes (2024), BNEF (2024b); BNEF (2024c); Oxford Economics Limited (2024b); CEPII (2024); as well as announcements by manufacturers and personal communications, gathered by the IEA.\n\n**Clean technology manufacturing capacity surged in 2023, with only heat pumps and electric vehicles seeing smaller capacity additions than in 2022.**\n\nIn the case of solar PV, manufacturing capacity additions for assembling modules have generally exceeded those of the key components – polysilicon, wafers and cells – with the notable exception of China over the last 2 years (Figure 1.7). However, the utilisation rates of component manufacturing facilities generally remain low, averting bottlenecks. Average utilisation rates across PV module manufacturing facilities worldwide remained steady in 2023, at around 55%. In parallel, facilities for newer technologies like tunnel oxide passivated contact (TOPCon), heterojunction (HJT) and back contact (BC) cells are gaining market share over the older ones like passivated emitter rear cells (PERC). This has driven down prices and led to some downscaling of expansion plans, especially in China (Box 1.2).\n\n**Solar PV (modules)**\n\n0\n\n200\n\n400\n\n600\n\n2020 2021 2022 2023\n\nGW\n\nChina European Union India Japan United States RoW\n\n0\n\n12\n\n24\n\n36\n\n2020 2021 2022 2023\n\n**Wind (nacelles)**\n\n0\n\n250\n\n500\n\n750\n\n1 000\n\n2020 2021 2022 2023\n\nGWh\n\n**Batteries (cells)**\n\n0\n\n5\n\n10\n\n15\n\n2020 2021 2022 2023\n\n**Electrolysers**\n\n0\n\n5\n\n10\n\n15\n\n20\n\n2020 2021 2022 2023\n\nGW\n\n**Heat pumps**\n\n0\n\n2\n\n4\n\n6\n\n8\n\n2020 2021 2022 2023\n\nMillion units\n\n**Electric vehicles**","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/EnergyTechnologyPerspectives2024_p39.pdf"]},"response":{"errors":[],"extract_id":"extract_e71ad22f5200f535759957ef3e53e690","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Energy Technology Perspectives 2024 Chapter 1: The state of manufacturing and trade\n\nPAGE | 37\n\nI EA. CC BY 4.0.\n\n**Net manufacturing capacity additions for selected clean energy technologies by country/region, 2020-2023**\n\nIEA. CC BY 4.0.\n\nNote: RoW = Rest of World. Sources: IEA analysis based on S&P Global (2024); WindEurope (2023); BNEF (2024a); GWEC (2023); Wood Mackenzie (2024); InfoLink (2024); SPV Market Research (2024); BMI (2024); EV Volumes (2024), BNEF (2024b); BNEF (2024c); Oxford Economics Limited (2024b); CEPII (2024); as well as announcements by manufacturers and personal communications, gathered by the IEA.\n\n**Clean technology manufacturing capacity surged in 2023, with only heat pumps and electric vehicles seeing smaller capacity additions than in 2022.**\n\nIn the case of solar PV, manufacturing capacity additions for assembling modules have generally exceeded those of the key components – polysilicon, wafers and cells – with the notable exception of China over the last 2 years (Figure 1.7). However, the utilisation rates of component manufacturing facilities generally remain low, averting bottlenecks. Average utilisation rates across PV module manufacturing facilities worldwide remained steady in 2023, at around 55%. In parallel, facilities for newer technologies like tunnel oxide passivated contact (TOPCon), heterojunction (HJT) and back contact (BC) cells are gaining market share over the older ones like passivated emitter rear cells (PERC). This has driven down prices and led to some downscaling of expansion plans, especially in China (Box 1.2).\n\n**Solar PV (modules)**\n\n0\n\n200\n\n400\n\n600\n\n2020 2021 2022 2023\n\nGW\n\nChina European Union India Japan United States RoW\n\n0\n\n12\n\n24\n\n36\n\n2020 2021 2022 2023\n\n**Wind (nacelles)**\n\n0\n\n250\n\n500\n\n750\n\n1 000\n\n2020 2021 2022 2023\n\nGWh\n\n**Batteries (cells)**\n\n0\n\n5\n\n10\n\n15\n\n2020 2021 2022 2023\n\n**Electrolysers**\n\n0\n\n5\n\n10\n\n15\n\n20\n\n2020 2021 2022 2023\n\nGW\n\n**Heat pumps**\n\n0\n\n2\n\n4\n\n6\n\n8\n\n2020 2021 2022 2023\n\nMillion units\n\n**Electric vehicles**"],"full_content":"Energy Technology Perspectives 2024 Chapter 1: The state of manufacturing and trade\n\nPAGE | 37\n\nI EA. CC BY 4.0.\n\n**Net manufacturing capacity additions for selected clean energy technologies by country/region, 2020-2023**\n\nIEA. CC BY 4.0.\n\nNote: RoW = Rest of World. Sources: IEA analysis based on S&P Global (2024); WindEurope (2023); BNEF (2024a); GWEC (2023); Wood Mackenzie (2024); InfoLink (2024); SPV Market Research (2024); BMI (2024); EV Volumes (2024), BNEF (2024b); BNEF (2024c); Oxford Economics Limited (2024b); CEPII (2024); as well as announcements by manufacturers and personal communications, gathered by the IEA.\n\n**Clean technology manufacturing capacity surged in 2023, with only heat pumps and electric vehicles seeing smaller capacity additions than in 2022.**\n\nIn the case of solar PV, manufacturing capacity additions for assembling modules have generally exceeded those of the key components – polysilicon, wafers and cells – with the notable exception of China over the last 2 years (Figure 1.7). However, the utilisation rates of component manufacturing facilities generally remain low, averting bottlenecks. Average utilisation rates across PV module manufacturing facilities worldwide remained steady in 2023, at around 55%. In parallel, facilities for newer technologies like tunnel oxide passivated contact (TOPCon), heterojunction (HJT) and back contact (BC) cells are gaining market share over the older ones like passivated emitter rear cells (PERC). This has driven down prices and led to some downscaling of expansion plans, especially in China (Box 1.2).\n\n**Solar PV (modules)**\n\n0\n\n200\n\n400\n\n600\n\n2020 2021 2022 2023\n\nGW\n\nChina European Union India Japan United States RoW\n\n0\n\n12\n\n24\n\n36\n\n2020 2021 2022 2023\n\n**Wind (nacelles)**\n\n0\n\n250\n\n500\n\n750\n\n1 000\n\n2020 2021 2022 2023\n\nGWh\n\n**Batteries (cells)**\n\n0\n\n5\n\n10\n\n15\n\n2020 2021 2022 2023\n\n**Electrolysers**\n\n0\n\n5\n\n10\n\n15\n\n20\n\n2020 2021 2022 2023\n\nGW\n\n**Heat pumps**\n\n0\n\n2\n\n4\n\n6\n\n8\n\n2020 2021 2022 2023\n\nMillion units\n\n**Electric vehicles**","publish_date":null,"title":"In the case of solar PV, manufacturing capacity additions for assembling modules have generally exceeded those of the key components","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/EnergyTechnologyPerspectives2024_p39.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/EnergyTechnologyPerspectives2024_p39.pdf"},"request":{"example_id":"chart/EnergyTechnologyPerspectives2024_p39","product_type":"parse","source_file_path":"EnergyTechnologyPerspectives2024_p39.pdf"},"started_at":"2026-11-04T19:45:01.962039"} +{"completed_at":"2026-11-04T19:45:04.605656","latency_in_ms":1364,"output":{"example_id":"chart/EnergyTechnologyPerspectives2024_p46","grounded_pages":[],"layout_pages":[],"markdown":"Energy Technology Perspectives 2024 Chapter 1: The state of manufacturing and trade\n\nPAGE | 44\n\nI EA. CC BY 4.0.\n\n### Materials production\n\nThe rate of expansion of global capacity for producing key materials is generally much slower than for clean technology manufacturing, as they also serve the needs of other manufacturing sectors and construction, where materials demand is growing less rapidly. In the case of crude steel, nearly 60 Mt per year of new capacity was added on a net basis worldwide in 2023, boosting total capacity by around 2.5% (Figure 1.9). Around 40% of the steelmaking capacity added in 2023, and 50% in 2022, was associated with iron production, whereas the remainder consisted of standalone steelmaking capacity, much of which is used to recycle scrap steel in electric furnaces. This is reflected in the greater absolute quantities of steelmaking capacity added over the past 4 years (105 Mt cumulatively) relative to ironmaking capacity (70 Mt cumulatively), despite the fact that it takes around 1.1 tonnes of iron to make a tonne of steel, when not using any scrap.\n\n**Global net manufacturing capacity additions for selected materials, 2020- 2023**\n\nIEA. CC BY 4.0.\n\nNotes: RoW = Rest of World. Capacity additions exclude regions where capacity has fallen. Aluminium refers to primary production capacity only. Sources: IEA analysis based on IFA (2024); IEA (2024c); OECD (2024b); GEM (2024); and CRU (2024).\n\n**China, India and the Middle East have together accounted for more than half of global net capacity additions for key materials and their intermediates since 2020.**\n\nInstalled primary aluminium production capacity grew by just 1% in 2023, reflecting a slowdown in China as production levels approached annual limits imposed by the government (S&P Global, 2022). Capacity for alumina production grew by around 5% in both 2022 and 2023.\n\nAmmonia capacity expanded by 7 Mt, or 3%, in 2022 – equivalent to around eight modern plants being completed – but was virtually unchanged in 2023. While the pipeline of electrolysis projects for ammonia for energy applications is growing\n\n0\n\n25\n\n50\n\n75\n\n2020\n\n2021\n\n2022\n\n2023\n\n2020\n\n2021\n\n2022\n\n2023\n\nMt\n\nChina European Union India Middle East United States RoW\n\n**Crude steel Iron**\n\n0\n\n3\n\n6\n\n9\n\n2020\n\n2021\n\n2022\n\n2023\n\n**Ammonia**\n\n0\n\n3\n\n6\n\n9\n\n2020\n\n2021\n\n2022\n\n2023\n\n2020\n\n2021\n\n2022\n\n2023\n\n**Aluminium Alumina**","pages":[{"line_numbers":[],"markdown":"Energy Technology Perspectives 2024 Chapter 1: The state of manufacturing and trade\n\nPAGE | 44\n\nI EA. CC BY 4.0.\n\n### Materials production\n\nThe rate of expansion of global capacity for producing key materials is generally much slower than for clean technology manufacturing, as they also serve the needs of other manufacturing sectors and construction, where materials demand is growing less rapidly. In the case of crude steel, nearly 60 Mt per year of new capacity was added on a net basis worldwide in 2023, boosting total capacity by around 2.5% (Figure 1.9). Around 40% of the steelmaking capacity added in 2023, and 50% in 2022, was associated with iron production, whereas the remainder consisted of standalone steelmaking capacity, much of which is used to recycle scrap steel in electric furnaces. This is reflected in the greater absolute quantities of steelmaking capacity added over the past 4 years (105 Mt cumulatively) relative to ironmaking capacity (70 Mt cumulatively), despite the fact that it takes around 1.1 tonnes of iron to make a tonne of steel, when not using any scrap.\n\n**Global net manufacturing capacity additions for selected materials, 2020- 2023**\n\nIEA. CC BY 4.0.\n\nNotes: RoW = Rest of World. Capacity additions exclude regions where capacity has fallen. Aluminium refers to primary production capacity only. Sources: IEA analysis based on IFA (2024); IEA (2024c); OECD (2024b); GEM (2024); and CRU (2024).\n\n**China, India and the Middle East have together accounted for more than half of global net capacity additions for key materials and their intermediates since 2020.**\n\nInstalled primary aluminium production capacity grew by just 1% in 2023, reflecting a slowdown in China as production levels approached annual limits imposed by the government (S&P Global, 2022). Capacity for alumina production grew by around 5% in both 2022 and 2023.\n\nAmmonia capacity expanded by 7 Mt, or 3%, in 2022 – equivalent to around eight modern plants being completed – but was virtually unchanged in 2023. While the pipeline of electrolysis projects for ammonia for energy applications is growing\n\n0\n\n25\n\n50\n\n75\n\n2020\n\n2021\n\n2022\n\n2023\n\n2020\n\n2021\n\n2022\n\n2023\n\nMt\n\nChina European Union India Middle East United States RoW\n\n**Crude steel Iron**\n\n0\n\n3\n\n6\n\n9\n\n2020\n\n2021\n\n2022\n\n2023\n\n**Ammonia**\n\n0\n\n3\n\n6\n\n9\n\n2020\n\n2021\n\n2022\n\n2023\n\n2020\n\n2021\n\n2022\n\n2023\n\n**Aluminium Alumina**","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Energy Technology Perspectives 2024 Chapter 1: The state of manufacturing and trade\n\nPAGE | 44\n\nI EA. CC BY 4.0.\n\n### Materials production\n\nThe rate of expansion of global capacity for producing key materials is generally much slower than for clean technology manufacturing, as they also serve the needs of other manufacturing sectors and construction, where materials demand is growing less rapidly. In the case of crude steel, nearly 60 Mt per year of new capacity was added on a net basis worldwide in 2023, boosting total capacity by around 2.5% (Figure 1.9). Around 40% of the steelmaking capacity added in 2023, and 50% in 2022, was associated with iron production, whereas the remainder consisted of standalone steelmaking capacity, much of which is used to recycle scrap steel in electric furnaces. This is reflected in the greater absolute quantities of steelmaking capacity added over the past 4 years (105 Mt cumulatively) relative to ironmaking capacity (70 Mt cumulatively), despite the fact that it takes around 1.1 tonnes of iron to make a tonne of steel, when not using any scrap.\n\n**Global net manufacturing capacity additions for selected materials, 2020- 2023**\n\nIEA. CC BY 4.0.\n\nNotes: RoW = Rest of World. Capacity additions exclude regions where capacity has fallen. Aluminium refers to primary production capacity only. Sources: IEA analysis based on IFA (2024); IEA (2024c); OECD (2024b); GEM (2024); and CRU (2024).\n\n**China, India and the Middle East have together accounted for more than half of global net capacity additions for key materials and their intermediates since 2020.**\n\nInstalled primary aluminium production capacity grew by just 1% in 2023, reflecting a slowdown in China as production levels approached annual limits imposed by the government (S&P Global, 2022). Capacity for alumina production grew by around 5% in both 2022 and 2023.\n\nAmmonia capacity expanded by 7 Mt, or 3%, in 2022 – equivalent to around eight modern plants being completed – but was virtually unchanged in 2023. While the pipeline of electrolysis projects for ammonia for energy applications is growing\n\n0\n\n25\n\n50\n\n75\n\n2020\n\n2021\n\n2022\n\n2023\n\n2020\n\n2021\n\n2022\n\n2023\n\nMt\n\nChina European Union India Middle East United States RoW\n\n**Crude steel Iron**\n\n0\n\n3\n\n6\n\n9\n\n2020\n\n2021\n\n2022\n\n2023\n\n**Ammonia**\n\n0\n\n3\n\n6\n\n9\n\n2020\n\n2021\n\n2022\n\n2023\n\n2020\n\n2021\n\n2022\n\n2023\n\n**Aluminium Alumina**","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/EnergyTechnologyPerspectives2024_p46.pdf"]},"response":{"errors":[],"extract_id":"extract_db71fd3627da315701619dadcc299697","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Energy Technology Perspectives 2024 Chapter 1: The state of manufacturing and trade\n\nPAGE | 44\n\nI EA. CC BY 4.0.\n\n### Materials production\n\nThe rate of expansion of global capacity for producing key materials is generally much slower than for clean technology manufacturing, as they also serve the needs of other manufacturing sectors and construction, where materials demand is growing less rapidly. In the case of crude steel, nearly 60 Mt per year of new capacity was added on a net basis worldwide in 2023, boosting total capacity by around 2.5% (Figure 1.9). Around 40% of the steelmaking capacity added in 2023, and 50% in 2022, was associated with iron production, whereas the remainder consisted of standalone steelmaking capacity, much of which is used to recycle scrap steel in electric furnaces. This is reflected in the greater absolute quantities of steelmaking capacity added over the past 4 years (105 Mt cumulatively) relative to ironmaking capacity (70 Mt cumulatively), despite the fact that it takes around 1.1 tonnes of iron to make a tonne of steel, when not using any scrap.\n\n**Global net manufacturing capacity additions for selected materials, 2020- 2023**\n\nIEA. CC BY 4.0.\n\nNotes: RoW = Rest of World. Capacity additions exclude regions where capacity has fallen. Aluminium refers to primary production capacity only. Sources: IEA analysis based on IFA (2024); IEA (2024c); OECD (2024b); GEM (2024); and CRU (2024).\n\n**China, India and the Middle East have together accounted for more than half of global net capacity additions for key materials and their intermediates since 2020.**\n\nInstalled primary aluminium production capacity grew by just 1% in 2023, reflecting a slowdown in China as production levels approached annual limits imposed by the government (S&P Global, 2022). Capacity for alumina production grew by around 5% in both 2022 and 2023.\n\nAmmonia capacity expanded by 7 Mt, or 3%, in 2022 – equivalent to around eight modern plants being completed – but was virtually unchanged in 2023. While the pipeline of electrolysis projects for ammonia for energy applications is growing\n\n0\n\n25\n\n50\n\n75\n\n2020\n\n2021\n\n2022\n\n2023\n\n2020\n\n2021\n\n2022\n\n2023\n\nMt\n\nChina European Union India Middle East United States RoW\n\n**Crude steel Iron**\n\n0\n\n3\n\n6\n\n9\n\n2020\n\n2021\n\n2022\n\n2023\n\n**Ammonia**\n\n0\n\n3\n\n6\n\n9\n\n2020\n\n2021\n\n2022\n\n2023\n\n2020\n\n2021\n\n2022\n\n2023\n\n**Aluminium Alumina**"],"full_content":"Energy Technology Perspectives 2024 Chapter 1: The state of manufacturing and trade\n\nPAGE | 44\n\nI EA. CC BY 4.0.\n\n### Materials production\n\nThe rate of expansion of global capacity for producing key materials is generally much slower than for clean technology manufacturing, as they also serve the needs of other manufacturing sectors and construction, where materials demand is growing less rapidly. In the case of crude steel, nearly 60 Mt per year of new capacity was added on a net basis worldwide in 2023, boosting total capacity by around 2.5% (Figure 1.9). Around 40% of the steelmaking capacity added in 2023, and 50% in 2022, was associated with iron production, whereas the remainder consisted of standalone steelmaking capacity, much of which is used to recycle scrap steel in electric furnaces. This is reflected in the greater absolute quantities of steelmaking capacity added over the past 4 years (105 Mt cumulatively) relative to ironmaking capacity (70 Mt cumulatively), despite the fact that it takes around 1.1 tonnes of iron to make a tonne of steel, when not using any scrap.\n\n**Global net manufacturing capacity additions for selected materials, 2020- 2023**\n\nIEA. CC BY 4.0.\n\nNotes: RoW = Rest of World. Capacity additions exclude regions where capacity has fallen. Aluminium refers to primary production capacity only. Sources: IEA analysis based on IFA (2024); IEA (2024c); OECD (2024b); GEM (2024); and CRU (2024).\n\n**China, India and the Middle East have together accounted for more than half of global net capacity additions for key materials and their intermediates since 2020.**\n\nInstalled primary aluminium production capacity grew by just 1% in 2023, reflecting a slowdown in China as production levels approached annual limits imposed by the government (S&P Global, 2022). Capacity for alumina production grew by around 5% in both 2022 and 2023.\n\nAmmonia capacity expanded by 7 Mt, or 3%, in 2022 – equivalent to around eight modern plants being completed – but was virtually unchanged in 2023. While the pipeline of electrolysis projects for ammonia for energy applications is growing\n\n0\n\n25\n\n50\n\n75\n\n2020\n\n2021\n\n2022\n\n2023\n\n2020\n\n2021\n\n2022\n\n2023\n\nMt\n\nChina European Union India Middle East United States RoW\n\n**Crude steel Iron**\n\n0\n\n3\n\n6\n\n9\n\n2020\n\n2021\n\n2022\n\n2023\n\n**Ammonia**\n\n0\n\n3\n\n6\n\n9\n\n2020\n\n2021\n\n2022\n\n2023\n\n2020\n\n2021\n\n2022\n\n2023\n\n**Aluminium Alumina**","publish_date":null,"title":"Materials production","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/EnergyTechnologyPerspectives2024_p46.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/EnergyTechnologyPerspectives2024_p46.pdf"},"request":{"example_id":"chart/EnergyTechnologyPerspectives2024_p46","product_type":"parse","source_file_path":"EnergyTechnologyPerspectives2024_p46.pdf"},"started_at":"2026-11-04T19:45:03.241306"} +{"completed_at":"2026-11-04T19:45:06.799195","latency_in_ms":2330,"output":{"example_id":"chart/EnergyTechnologyPerspectives2024_p47","grounded_pages":[],"layout_pages":[],"markdown":"Energy Technology Perspectives 2024 Chapter 1: The state of manufacturing and trade\n\nPAGE | 45\n\nI EA. CC BY 4.0.\n\nrapidly, the average size of these projects today is very small compared with plants serving existing agricultural and industrial demand (IEA, 2024c).\n\nThe project pipelines for the production of steel, aluminium and ammonia are also relatively limited compared with planned capacity additions for clean technology manufacturing, with capacity set to grow only modestly (Figure 1.10). ⁴ Only a small share of the announced materials manufacturing projects to 2030 involves near-zero emissions technologies for iron, steel and aluminium, though there is potential for their deployment to grow quickly once these technologies have been demonstrated at commercial scale and there is a clear demand signal. If all the projects using near-zero emissions technologies came to fruition, their share of global capacity would reach less than 1% for iron and steel, less than 0.1% for aluminium, and just 10% for ammonia – far below the levels required to get on track with global climate goals (see Chapter 2). Achieving net zero emissions by mid-century calls for the rapid and widespread deployment of these technologies, both for new capacity and retrofitting of existing plants.\n\n**Global installed manufacturing capacity and announced capacity**\n\n**additions for selected materials, 2023-2030**\n\nIEA. CC BY 4.0.\n\nNotes: At year-end. Committed = existing projects, those under construction and other projects having achieved a final investment decision. Aluminium refers to primary production only. Near-zero emissions capable projects are not included in the share of near-zero emissions. Sources: IEA analysis based on IFA (2024); OECD (2024b); GEM (2024); CRU (2024); and IEA (2024c).\n\n**The share of near-zero emissions technologies in planned capacity additions is highest in the ammonia industry.**\n\n4 This excludes the capacity at existing plants that needs to be refurbished or replaced at the end of its economic life, which is assumed to be around 25 years (i.e. roughly 4% of that capacity needs to be replaced or refurbished in any given year to maintain the same overall level of capacity).\n\n**Aluminium Alumina Iron Ammonia**\n\n0%\n\n4%\n\n8%\n\n12%\n\n0\n\n120\n\n240\n\n360\n\n2023 2030\n\nInstalled Announced (committed) Announced (preliminary) Share near-zero emissions (right axis)\n\n0%\n\n4%\n\n8%\n\n12%\n\n0\n\n1 000\n\n2 000\n\n3 000\n\n2023 2030 2023 2030\n\nMt\n\n0%\n\n4%\n\n8%\n\n12%\n\n0\n\n70\n\n140\n\n210\n\n2023 2030 2023 2030\n\n**Crude steel**","pages":[{"line_numbers":[],"markdown":"Energy Technology Perspectives 2024 Chapter 1: The state of manufacturing and trade\n\nPAGE | 45\n\nI EA. CC BY 4.0.\n\nrapidly, the average size of these projects today is very small compared with plants serving existing agricultural and industrial demand (IEA, 2024c).\n\nThe project pipelines for the production of steel, aluminium and ammonia are also relatively limited compared with planned capacity additions for clean technology manufacturing, with capacity set to grow only modestly (Figure 1.10). ⁴ Only a small share of the announced materials manufacturing projects to 2030 involves near-zero emissions technologies for iron, steel and aluminium, though there is potential for their deployment to grow quickly once these technologies have been demonstrated at commercial scale and there is a clear demand signal. If all the projects using near-zero emissions technologies came to fruition, their share of global capacity would reach less than 1% for iron and steel, less than 0.1% for aluminium, and just 10% for ammonia – far below the levels required to get on track with global climate goals (see Chapter 2). Achieving net zero emissions by mid-century calls for the rapid and widespread deployment of these technologies, both for new capacity and retrofitting of existing plants.\n\n**Global installed manufacturing capacity and announced capacity**\n\n**additions for selected materials, 2023-2030**\n\nIEA. CC BY 4.0.\n\nNotes: At year-end. Committed = existing projects, those under construction and other projects having achieved a final investment decision. Aluminium refers to primary production only. Near-zero emissions capable projects are not included in the share of near-zero emissions. Sources: IEA analysis based on IFA (2024); OECD (2024b); GEM (2024); CRU (2024); and IEA (2024c).\n\n**The share of near-zero emissions technologies in planned capacity additions is highest in the ammonia industry.**\n\n4 This excludes the capacity at existing plants that needs to be refurbished or replaced at the end of its economic life, which is assumed to be around 25 years (i.e. roughly 4% of that capacity needs to be replaced or refurbished in any given year to maintain the same overall level of capacity).\n\n**Aluminium Alumina Iron Ammonia**\n\n0%\n\n4%\n\n8%\n\n12%\n\n0\n\n120\n\n240\n\n360\n\n2023 2030\n\nInstalled Announced (committed) Announced (preliminary) Share near-zero emissions (right axis)\n\n0%\n\n4%\n\n8%\n\n12%\n\n0\n\n1 000\n\n2 000\n\n3 000\n\n2023 2030 2023 2030\n\nMt\n\n0%\n\n4%\n\n8%\n\n12%\n\n0\n\n70\n\n140\n\n210\n\n2023 2030 2023 2030\n\n**Crude steel**","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Energy Technology Perspectives 2024 Chapter 1: The state of manufacturing and trade\n\nPAGE | 45\n\nI EA. CC BY 4.0.\n\nrapidly, the average size of these projects today is very small compared with plants serving existing agricultural and industrial demand (IEA, 2024c).\n\nThe project pipelines for the production of steel, aluminium and ammonia are also relatively limited compared with planned capacity additions for clean technology manufacturing, with capacity set to grow only modestly (Figure 1.10). ⁴ Only a small share of the announced materials manufacturing projects to 2030 involves near-zero emissions technologies for iron, steel and aluminium, though there is potential for their deployment to grow quickly once these technologies have been demonstrated at commercial scale and there is a clear demand signal. If all the projects using near-zero emissions technologies came to fruition, their share of global capacity would reach less than 1% for iron and steel, less than 0.1% for aluminium, and just 10% for ammonia – far below the levels required to get on track with global climate goals (see Chapter 2). Achieving net zero emissions by mid-century calls for the rapid and widespread deployment of these technologies, both for new capacity and retrofitting of existing plants.\n\n**Global installed manufacturing capacity and announced capacity**\n\n**additions for selected materials, 2023-2030**\n\nIEA. CC BY 4.0.\n\nNotes: At year-end. Committed = existing projects, those under construction and other projects having achieved a final investment decision. Aluminium refers to primary production only. Near-zero emissions capable projects are not included in the share of near-zero emissions. Sources: IEA analysis based on IFA (2024); OECD (2024b); GEM (2024); CRU (2024); and IEA (2024c).\n\n**The share of near-zero emissions technologies in planned capacity additions is highest in the ammonia industry.**\n\n4 This excludes the capacity at existing plants that needs to be refurbished or replaced at the end of its economic life, which is assumed to be around 25 years (i.e. roughly 4% of that capacity needs to be replaced or refurbished in any given year to maintain the same overall level of capacity).\n\n**Aluminium Alumina Iron Ammonia**\n\n0%\n\n4%\n\n8%\n\n12%\n\n0\n\n120\n\n240\n\n360\n\n2023 2030\n\nInstalled Announced (committed) Announced (preliminary) Share near-zero emissions (right axis)\n\n0%\n\n4%\n\n8%\n\n12%\n\n0\n\n1 000\n\n2 000\n\n3 000\n\n2023 2030 2023 2030\n\nMt\n\n0%\n\n4%\n\n8%\n\n12%\n\n0\n\n70\n\n140\n\n210\n\n2023 2030 2023 2030\n\n**Crude steel**","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/EnergyTechnologyPerspectives2024_p47.pdf"]},"response":{"errors":[],"extract_id":"extract_3ebd8859b21c0cb63014ad0284c03509","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Energy Technology Perspectives 2024 Chapter 1: The state of manufacturing and trade\n\nPAGE | 45\n\nI EA. CC BY 4.0.\n\nrapidly, the average size of these projects today is very small compared with plants serving existing agricultural and industrial demand (IEA, 2024c).\n\nThe project pipelines for the production of steel, aluminium and ammonia are also relatively limited compared with planned capacity additions for clean technology manufacturing, with capacity set to grow only modestly (Figure 1.10). ⁴ Only a small share of the announced materials manufacturing projects to 2030 involves near-zero emissions technologies for iron, steel and aluminium, though there is potential for their deployment to grow quickly once these technologies have been demonstrated at commercial scale and there is a clear demand signal. If all the projects using near-zero emissions technologies came to fruition, their share of global capacity would reach less than 1% for iron and steel, less than 0.1% for aluminium, and just 10% for ammonia – far below the levels required to get on track with global climate goals (see Chapter 2). Achieving net zero emissions by mid-century calls for the rapid and widespread deployment of these technologies, both for new capacity and retrofitting of existing plants.\n\n**Global installed manufacturing capacity and announced capacity**\n\n**additions for selected materials, 2023-2030**\n\nIEA. CC BY 4.0.\n\nNotes: At year-end. Committed = existing projects, those under construction and other projects having achieved a final investment decision. Aluminium refers to primary production only. Near-zero emissions capable projects are not included in the share of near-zero emissions. Sources: IEA analysis based on IFA (2024); OECD (2024b); GEM (2024); CRU (2024); and IEA (2024c).\n\n**The share of near-zero emissions technologies in planned capacity additions is highest in the ammonia industry.**\n\n4 This excludes the capacity at existing plants that needs to be refurbished or replaced at the end of its economic life, which is assumed to be around 25 years (i.e. roughly 4% of that capacity needs to be replaced or refurbished in any given year to maintain the same overall level of capacity).\n\n**Aluminium Alumina Iron Ammonia**\n\n0%\n\n4%\n\n8%\n\n12%\n\n0\n\n120\n\n240\n\n360\n\n2023 2030\n\nInstalled Announced (committed) Announced (preliminary) Share near-zero emissions (right axis)\n\n0%\n\n4%\n\n8%\n\n12%\n\n0\n\n1 000\n\n2 000\n\n3 000\n\n2023 2030 2023 2030\n\nMt\n\n0%\n\n4%\n\n8%\n\n12%\n\n0\n\n70\n\n140\n\n210\n\n2023 2030 2023 2030\n\n**Crude steel**"],"full_content":"Energy Technology Perspectives 2024 Chapter 1: The state of manufacturing and trade\n\nPAGE | 45\n\nI EA. CC BY 4.0.\n\nrapidly, the average size of these projects today is very small compared with plants serving existing agricultural and industrial demand (IEA, 2024c).\n\nThe project pipelines for the production of steel, aluminium and ammonia are also relatively limited compared with planned capacity additions for clean technology manufacturing, with capacity set to grow only modestly (Figure 1.10). ⁴ Only a small share of the announced materials manufacturing projects to 2030 involves near-zero emissions technologies for iron, steel and aluminium, though there is potential for their deployment to grow quickly once these technologies have been demonstrated at commercial scale and there is a clear demand signal. If all the projects using near-zero emissions technologies came to fruition, their share of global capacity would reach less than 1% for iron and steel, less than 0.1% for aluminium, and just 10% for ammonia – far below the levels required to get on track with global climate goals (see Chapter 2). Achieving net zero emissions by mid-century calls for the rapid and widespread deployment of these technologies, both for new capacity and retrofitting of existing plants.\n\n**Global installed manufacturing capacity and announced capacity**\n\n**additions for selected materials, 2023-2030**\n\nIEA. CC BY 4.0.\n\nNotes: At year-end. Committed = existing projects, those under construction and other projects having achieved a final investment decision. Aluminium refers to primary production only. Near-zero emissions capable projects are not included in the share of near-zero emissions. Sources: IEA analysis based on IFA (2024); OECD (2024b); GEM (2024); CRU (2024); and IEA (2024c).\n\n**The share of near-zero emissions technologies in planned capacity additions is highest in the ammonia industry.**\n\n4 This excludes the capacity at existing plants that needs to be refurbished or replaced at the end of its economic life, which is assumed to be around 25 years (i.e. roughly 4% of that capacity needs to be replaced or refurbished in any given year to maintain the same overall level of capacity).\n\n**Aluminium Alumina Iron Ammonia**\n\n0%\n\n4%\n\n8%\n\n12%\n\n0\n\n120\n\n240\n\n360\n\n2023 2030\n\nInstalled Announced (committed) Announced (preliminary) Share near-zero emissions (right axis)\n\n0%\n\n4%\n\n8%\n\n12%\n\n0\n\n1 000\n\n2 000\n\n3 000\n\n2023 2030 2023 2030\n\nMt\n\n0%\n\n4%\n\n8%\n\n12%\n\n0\n\n70\n\n140\n\n210\n\n2023 2030 2023 2030\n\n**Crude steel**","publish_date":null,"title":"rapidly, the average size of these projects today is very small compared with plants serving existing agricultural and industrial demand","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/EnergyTechnologyPerspectives2024_p47.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/EnergyTechnologyPerspectives2024_p47.pdf"},"request":{"example_id":"chart/EnergyTechnologyPerspectives2024_p47","product_type":"parse","source_file_path":"EnergyTechnologyPerspectives2024_p47.pdf"},"started_at":"2026-11-04T19:45:04.468640"} +{"completed_at":"2026-11-04T19:45:07.184189","latency_in_ms":1237,"output":{"example_id":"chart/EnergyTechnologyPerspectives2024_p48","grounded_pages":[],"layout_pages":[],"markdown":"Energy Technology Perspectives 2024 Chapter 1: The state of manufacturing and trade\n\nPAGE | 46\n\nI EA. CC BY 4.0.\n\n## **Geographical concentration**\n\nEach of the different steps of most clean energy technology supply chains are highly geographically concentrated, from the extraction and refining of raw materials to the manufacturing of those technologies, their components and the materials required as inputs (IEA, 2023b). This concentration can create supply chain risks or vulnerabilities. The global landscape has not changed much between 2021 and 2023, with manufacturing remaining far more concentrated than fossil fuel supply, and China continuing to play a leading role in the world’s clean energy technology supply chains (Figure 1.11).\n\nGeographical concentration is expected to persist despite the growing number of new project announcements (IEA, 2024a). If all announced capacity additions come to fruition, the three major producing countries and regions are expected to continue to account for around 80% or more of global capacity in all cases, with only minor shifts in relative shares. In the case of battery manufacturing, China’s share could fall as that of both the European Union and the United States increases. In contrast, China’s share of wind manufacturing is set to grow. For heat pumps, Europe’s share is set to increase its share the most.\n\n**Installed manufacturing capacity by country/region, 2023**\n\nIEA. CC BY 4.0.\n\nNote: RoW = Rest of World. “Electric cars” values are calculated based on 2023 production numbers, adjusted according to the utilisation rates of car assembly plants in the region. Source: IEA analysis based on IEA (2024a); and IEA (2023b).\n\n**Manufacturing capacity for clean technologies and materials today is highly concentrated geographically, with China the largest single producer in all cases.**\n\n0% 20% 40% 60% 80% 100%\n\nAnodes\n\nCathodes\n\nCells\n\nElectric cars\n\nPolysilicon\n\nWafers\n\nCells\n\nModules\n\nBlades\n\nNacelles\n\nTowers\n\nElectrolysers\n\nHeat pumps\n\nAlumina\n\nAluminium\n\nAmmonia\n\nIron\n\nSteel\n\nBatteries EV Solar PV Wind Other Materials\n\nChina European Union India Japan Korea United States RoW China in 2021","pages":[{"line_numbers":[],"markdown":"Energy Technology Perspectives 2024 Chapter 1: The state of manufacturing and trade\n\nPAGE | 46\n\nI EA. CC BY 4.0.\n\n## **Geographical concentration**\n\nEach of the different steps of most clean energy technology supply chains are highly geographically concentrated, from the extraction and refining of raw materials to the manufacturing of those technologies, their components and the materials required as inputs (IEA, 2023b). This concentration can create supply chain risks or vulnerabilities. The global landscape has not changed much between 2021 and 2023, with manufacturing remaining far more concentrated than fossil fuel supply, and China continuing to play a leading role in the world’s clean energy technology supply chains (Figure 1.11).\n\nGeographical concentration is expected to persist despite the growing number of new project announcements (IEA, 2024a). If all announced capacity additions come to fruition, the three major producing countries and regions are expected to continue to account for around 80% or more of global capacity in all cases, with only minor shifts in relative shares. In the case of battery manufacturing, China’s share could fall as that of both the European Union and the United States increases. In contrast, China’s share of wind manufacturing is set to grow. For heat pumps, Europe’s share is set to increase its share the most.\n\n**Installed manufacturing capacity by country/region, 2023**\n\nIEA. CC BY 4.0.\n\nNote: RoW = Rest of World. “Electric cars” values are calculated based on 2023 production numbers, adjusted according to the utilisation rates of car assembly plants in the region. Source: IEA analysis based on IEA (2024a); and IEA (2023b).\n\n**Manufacturing capacity for clean technologies and materials today is highly concentrated geographically, with China the largest single producer in all cases.**\n\n0% 20% 40% 60% 80% 100%\n\nAnodes\n\nCathodes\n\nCells\n\nElectric cars\n\nPolysilicon\n\nWafers\n\nCells\n\nModules\n\nBlades\n\nNacelles\n\nTowers\n\nElectrolysers\n\nHeat pumps\n\nAlumina\n\nAluminium\n\nAmmonia\n\nIron\n\nSteel\n\nBatteries EV Solar PV Wind Other Materials\n\nChina European Union India Japan Korea United States RoW China in 2021","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Energy Technology Perspectives 2024 Chapter 1: The state of manufacturing and trade\n\nPAGE | 46\n\nI EA. CC BY 4.0.\n\n## **Geographical concentration**\n\nEach of the different steps of most clean energy technology supply chains are highly geographically concentrated, from the extraction and refining of raw materials to the manufacturing of those technologies, their components and the materials required as inputs (IEA, 2023b). This concentration can create supply chain risks or vulnerabilities. The global landscape has not changed much between 2021 and 2023, with manufacturing remaining far more concentrated than fossil fuel supply, and China continuing to play a leading role in the world’s clean energy technology supply chains (Figure 1.11).\n\nGeographical concentration is expected to persist despite the growing number of new project announcements (IEA, 2024a). If all announced capacity additions come to fruition, the three major producing countries and regions are expected to continue to account for around 80% or more of global capacity in all cases, with only minor shifts in relative shares. In the case of battery manufacturing, China’s share could fall as that of both the European Union and the United States increases. In contrast, China’s share of wind manufacturing is set to grow. For heat pumps, Europe’s share is set to increase its share the most.\n\n**Installed manufacturing capacity by country/region, 2023**\n\nIEA. CC BY 4.0.\n\nNote: RoW = Rest of World. “Electric cars” values are calculated based on 2023 production numbers, adjusted according to the utilisation rates of car assembly plants in the region. Source: IEA analysis based on IEA (2024a); and IEA (2023b).\n\n**Manufacturing capacity for clean technologies and materials today is highly concentrated geographically, with China the largest single producer in all cases.**\n\n0% 20% 40% 60% 80% 100%\n\nAnodes\n\nCathodes\n\nCells\n\nElectric cars\n\nPolysilicon\n\nWafers\n\nCells\n\nModules\n\nBlades\n\nNacelles\n\nTowers\n\nElectrolysers\n\nHeat pumps\n\nAlumina\n\nAluminium\n\nAmmonia\n\nIron\n\nSteel\n\nBatteries EV Solar PV Wind Other Materials\n\nChina European Union India Japan Korea United States RoW China in 2021","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/EnergyTechnologyPerspectives2024_p48.pdf"]},"response":{"errors":[],"extract_id":"extract_3475fc22107630c788eb2b3420ec0652","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Energy Technology Perspectives 2024 Chapter 1: The state of manufacturing and trade\n\nPAGE | 46\n\nI EA. CC BY 4.0.\n\n## **Geographical concentration**\n\nEach of the different steps of most clean energy technology supply chains are highly geographically concentrated, from the extraction and refining of raw materials to the manufacturing of those technologies, their components and the materials required as inputs (IEA, 2023b). This concentration can create supply chain risks or vulnerabilities. The global landscape has not changed much between 2021 and 2023, with manufacturing remaining far more concentrated than fossil fuel supply, and China continuing to play a leading role in the world’s clean energy technology supply chains (Figure 1.11).\n\nGeographical concentration is expected to persist despite the growing number of new project announcements (IEA, 2024a). If all announced capacity additions come to fruition, the three major producing countries and regions are expected to continue to account for around 80% or more of global capacity in all cases, with only minor shifts in relative shares. In the case of battery manufacturing, China’s share could fall as that of both the European Union and the United States increases. In contrast, China’s share of wind manufacturing is set to grow. For heat pumps, Europe’s share is set to increase its share the most.\n\n**Installed manufacturing capacity by country/region, 2023**\n\nIEA. CC BY 4.0.\n\nNote: RoW = Rest of World. “Electric cars” values are calculated based on 2023 production numbers, adjusted according to the utilisation rates of car assembly plants in the region. Source: IEA analysis based on IEA (2024a); and IEA (2023b).\n\n**Manufacturing capacity for clean technologies and materials today is highly concentrated geographically, with China the largest single producer in all cases.**\n\n0% 20% 40% 60% 80% 100%\n\nAnodes\n\nCathodes\n\nCells\n\nElectric cars\n\nPolysilicon\n\nWafers\n\nCells\n\nModules\n\nBlades\n\nNacelles\n\nTowers\n\nElectrolysers\n\nHeat pumps\n\nAlumina\n\nAluminium\n\nAmmonia\n\nIron\n\nSteel\n\nBatteries EV Solar PV Wind Other Materials\n\nChina European Union India Japan Korea United States RoW China in 2021"],"full_content":"Energy Technology Perspectives 2024 Chapter 1: The state of manufacturing and trade\n\nPAGE | 46\n\nI EA. CC BY 4.0.\n\n## **Geographical concentration**\n\nEach of the different steps of most clean energy technology supply chains are highly geographically concentrated, from the extraction and refining of raw materials to the manufacturing of those technologies, their components and the materials required as inputs (IEA, 2023b). This concentration can create supply chain risks or vulnerabilities. The global landscape has not changed much between 2021 and 2023, with manufacturing remaining far more concentrated than fossil fuel supply, and China continuing to play a leading role in the world’s clean energy technology supply chains (Figure 1.11).\n\nGeographical concentration is expected to persist despite the growing number of new project announcements (IEA, 2024a). If all announced capacity additions come to fruition, the three major producing countries and regions are expected to continue to account for around 80% or more of global capacity in all cases, with only minor shifts in relative shares. In the case of battery manufacturing, China’s share could fall as that of both the European Union and the United States increases. In contrast, China’s share of wind manufacturing is set to grow. For heat pumps, Europe’s share is set to increase its share the most.\n\n**Installed manufacturing capacity by country/region, 2023**\n\nIEA. CC BY 4.0.\n\nNote: RoW = Rest of World. “Electric cars” values are calculated based on 2023 production numbers, adjusted according to the utilisation rates of car assembly plants in the region. Source: IEA analysis based on IEA (2024a); and IEA (2023b).\n\n**Manufacturing capacity for clean technologies and materials today is highly concentrated geographically, with China the largest single producer in all cases.**\n\n0% 20% 40% 60% 80% 100%\n\nAnodes\n\nCathodes\n\nCells\n\nElectric cars\n\nPolysilicon\n\nWafers\n\nCells\n\nModules\n\nBlades\n\nNacelles\n\nTowers\n\nElectrolysers\n\nHeat pumps\n\nAlumina\n\nAluminium\n\nAmmonia\n\nIron\n\nSteel\n\nBatteries EV Solar PV Wind Other Materials\n\nChina European Union India Japan Korea United States RoW China in 2021","publish_date":null,"title":"Geographical concentration","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/EnergyTechnologyPerspectives2024_p48.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/EnergyTechnologyPerspectives2024_p48.pdf"},"request":{"example_id":"chart/EnergyTechnologyPerspectives2024_p48","product_type":"parse","source_file_path":"EnergyTechnologyPerspectives2024_p48.pdf"},"started_at":"2026-11-04T19:45:05.947163"} +{"completed_at":"2026-11-04T19:45:08.296325","latency_in_ms":1110,"output":{"example_id":"chart/EnergyTechnologyPerspectives2024_p50","grounded_pages":[],"layout_pages":[],"markdown":"Energy Technology Perspectives 2024 Chapter 1: The state of manufacturing and trade\n\nPAGE | 48\n\nI EA. CC BY 4.0.\n\n**Production of selected clean energy technologies and materials relative to**\n\n**domestic demand by country/region, 2023**\n\nIEA. CC BY 4.0.\n\nNotes: Battery cells are for both electric vehicles and stationary storage. Steel includes crude steel as well as semi-finished and finished products.\n\n**China is a net exporter of key clean energy technologies, while the European Union and the United States rely heavily on imports to meet domestic demand.**\n\nThe resilience of clean energy technology supply chains can be jeopardised by strong reliance on a very limited number of component suppliers. For example, an explosion at a polysilicon facility in China in 2020 put 8% of global polysilicon production capacity out of operation (IEA, 2022b). This was the largest of four polysilicon plant closures in 2020, the others resulting from flooding or technical problems. While each incident occurred at a different time, together they led to an estimated 4% decline in annual production in an already-tight polysilicon market, contributing to the near tripling of prices between 2020 and 2021. Vertical integration of supply chains can, in principle, enhance the resilience of supply chains – China’s wind turbine manufacturers were more resilient to commodity price fluctuations in 2022 in part due to vertical integration – though it may lead to greater dependence on a small number of suppliers (IEA, 2023c).\n\n0%\n\n50%\n\n100%\n\n150%\n\n200%\n\nSolar PV modules\n\nWind nacelles Battery cells Electric cars ICE cars Heat pumps Electrolysers\n\n0%\n\n50%\n\n100%\n\n150%\n\nAlumina Aluminium Iron Steel Ammonia Fertilisers\n\nChina European Union United States Local demand","pages":[{"line_numbers":[],"markdown":"Energy Technology Perspectives 2024 Chapter 1: The state of manufacturing and trade\n\nPAGE | 48\n\nI EA. CC BY 4.0.\n\n**Production of selected clean energy technologies and materials relative to**\n\n**domestic demand by country/region, 2023**\n\nIEA. CC BY 4.0.\n\nNotes: Battery cells are for both electric vehicles and stationary storage. Steel includes crude steel as well as semi-finished and finished products.\n\n**China is a net exporter of key clean energy technologies, while the European Union and the United States rely heavily on imports to meet domestic demand.**\n\nThe resilience of clean energy technology supply chains can be jeopardised by strong reliance on a very limited number of component suppliers. For example, an explosion at a polysilicon facility in China in 2020 put 8% of global polysilicon production capacity out of operation (IEA, 2022b). This was the largest of four polysilicon plant closures in 2020, the others resulting from flooding or technical problems. While each incident occurred at a different time, together they led to an estimated 4% decline in annual production in an already-tight polysilicon market, contributing to the near tripling of prices between 2020 and 2021. Vertical integration of supply chains can, in principle, enhance the resilience of supply chains – China’s wind turbine manufacturers were more resilient to commodity price fluctuations in 2022 in part due to vertical integration – though it may lead to greater dependence on a small number of suppliers (IEA, 2023c).\n\n0%\n\n50%\n\n100%\n\n150%\n\n200%\n\nSolar PV modules\n\nWind nacelles Battery cells Electric cars ICE cars Heat pumps Electrolysers\n\n0%\n\n50%\n\n100%\n\n150%\n\nAlumina Aluminium Iron Steel Ammonia Fertilisers\n\nChina European Union United States Local demand","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Energy Technology Perspectives 2024 Chapter 1: The state of manufacturing and trade\n\nPAGE | 48\n\nI EA. CC BY 4.0.\n\n**Production of selected clean energy technologies and materials relative to**\n\n**domestic demand by country/region, 2023**\n\nIEA. CC BY 4.0.\n\nNotes: Battery cells are for both electric vehicles and stationary storage. Steel includes crude steel as well as semi-finished and finished products.\n\n**China is a net exporter of key clean energy technologies, while the European Union and the United States rely heavily on imports to meet domestic demand.**\n\nThe resilience of clean energy technology supply chains can be jeopardised by strong reliance on a very limited number of component suppliers. For example, an explosion at a polysilicon facility in China in 2020 put 8% of global polysilicon production capacity out of operation (IEA, 2022b). This was the largest of four polysilicon plant closures in 2020, the others resulting from flooding or technical problems. While each incident occurred at a different time, together they led to an estimated 4% decline in annual production in an already-tight polysilicon market, contributing to the near tripling of prices between 2020 and 2021. Vertical integration of supply chains can, in principle, enhance the resilience of supply chains – China’s wind turbine manufacturers were more resilient to commodity price fluctuations in 2022 in part due to vertical integration – though it may lead to greater dependence on a small number of suppliers (IEA, 2023c).\n\n0%\n\n50%\n\n100%\n\n150%\n\n200%\n\nSolar PV modules\n\nWind nacelles Battery cells Electric cars ICE cars Heat pumps Electrolysers\n\n0%\n\n50%\n\n100%\n\n150%\n\nAlumina Aluminium Iron Steel Ammonia Fertilisers\n\nChina European Union United States Local demand","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/EnergyTechnologyPerspectives2024_p50.pdf"]},"response":{"errors":[],"extract_id":"extract_f770b26f4df4cae0c6a31e4503be0ca9","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Energy Technology Perspectives 2024 Chapter 1: The state of manufacturing and trade\n\nPAGE | 48\n\nI EA. CC BY 4.0.\n\n**Production of selected clean energy technologies and materials relative to**\n\n**domestic demand by country/region, 2023**\n\nIEA. CC BY 4.0.\n\nNotes: Battery cells are for both electric vehicles and stationary storage. Steel includes crude steel as well as semi-finished and finished products.\n\n**China is a net exporter of key clean energy technologies, while the European Union and the United States rely heavily on imports to meet domestic demand.**\n\nThe resilience of clean energy technology supply chains can be jeopardised by strong reliance on a very limited number of component suppliers. For example, an explosion at a polysilicon facility in China in 2020 put 8% of global polysilicon production capacity out of operation (IEA, 2022b). This was the largest of four polysilicon plant closures in 2020, the others resulting from flooding or technical problems. While each incident occurred at a different time, together they led to an estimated 4% decline in annual production in an already-tight polysilicon market, contributing to the near tripling of prices between 2020 and 2021. Vertical integration of supply chains can, in principle, enhance the resilience of supply chains – China’s wind turbine manufacturers were more resilient to commodity price fluctuations in 2022 in part due to vertical integration – though it may lead to greater dependence on a small number of suppliers (IEA, 2023c).\n\n0%\n\n50%\n\n100%\n\n150%\n\n200%\n\nSolar PV modules\n\nWind nacelles Battery cells Electric cars ICE cars Heat pumps Electrolysers\n\n0%\n\n50%\n\n100%\n\n150%\n\nAlumina Aluminium Iron Steel Ammonia Fertilisers\n\nChina European Union United States Local demand"],"full_content":"Energy Technology Perspectives 2024 Chapter 1: The state of manufacturing and trade\n\nPAGE | 48\n\nI EA. CC BY 4.0.\n\n**Production of selected clean energy technologies and materials relative to**\n\n**domestic demand by country/region, 2023**\n\nIEA. CC BY 4.0.\n\nNotes: Battery cells are for both electric vehicles and stationary storage. Steel includes crude steel as well as semi-finished and finished products.\n\n**China is a net exporter of key clean energy technologies, while the European Union and the United States rely heavily on imports to meet domestic demand.**\n\nThe resilience of clean energy technology supply chains can be jeopardised by strong reliance on a very limited number of component suppliers. For example, an explosion at a polysilicon facility in China in 2020 put 8% of global polysilicon production capacity out of operation (IEA, 2022b). This was the largest of four polysilicon plant closures in 2020, the others resulting from flooding or technical problems. While each incident occurred at a different time, together they led to an estimated 4% decline in annual production in an already-tight polysilicon market, contributing to the near tripling of prices between 2020 and 2021. Vertical integration of supply chains can, in principle, enhance the resilience of supply chains – China’s wind turbine manufacturers were more resilient to commodity price fluctuations in 2022 in part due to vertical integration – though it may lead to greater dependence on a small number of suppliers (IEA, 2023c).\n\n0%\n\n50%\n\n100%\n\n150%\n\n200%\n\nSolar PV modules\n\nWind nacelles Battery cells Electric cars ICE cars Heat pumps Electrolysers\n\n0%\n\n50%\n\n100%\n\n150%\n\nAlumina Aluminium Iron Steel Ammonia Fertilisers\n\nChina European Union United States Local demand","publish_date":null,"title":"The resilience of clean energy technology supply chains can be jeopardised by strong reliance on a very limited number of","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/EnergyTechnologyPerspectives2024_p50.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/EnergyTechnologyPerspectives2024_p50.pdf"},"request":{"example_id":"chart/EnergyTechnologyPerspectives2024_p50","product_type":"parse","source_file_path":"EnergyTechnologyPerspectives2024_p50.pdf"},"started_at":"2026-11-04T19:45:07.185766"} +{"completed_at":"2026-11-04T19:45:10.826729","latency_in_ms":2271,"output":{"example_id":"chart/EnergyTechnologyPerspectives2024_p52","grounded_pages":[],"layout_pages":[],"markdown":"Energy Technology Perspectives 2024 Chapter 1: The state of manufacturing and trade\n\nPAGE | 50\n\nI EA. CC BY 4.0.\n\n**Energy employment in selected technology areas, 2022**\n\nIEA. CC BY 4.0.\n\nNotes: EV = electric vehicle. “EV and battery” excludes stationary battery storage. “Bioenergy” excludes waste and traditional use of biomass. “Power grids” refers to transmission and distribution, and stationary battery storage. Non- manufacturing work includes raw materials extraction, construction/installation, utilities, professional services, wholesale and transport. Manufacturing jobs include the final product and direct components (e.g. solar PV includes wafers and cells, while batteries includes cathodes and anodes). Non-manufacturing jobs in the EV and battery category include jobs installing batteries in EVs, but not EV-related jobs such as mechanics or sales, which are not estimated by the IEA. Fossil fuel supply includes upstream (raw materials extraction), midstream (manufacturing of equipment and oil refining) and some downstream activities (wholesale, transport and professional services), but excludes fossil fuel combustion for power generation or distribution. Source: Adapted from IEA (2023d).\n\n**Manufacturing of clean energy technologies is a major source of employment, with around 3 million jobs globally in EVs and batteries, solar PV, wind and heat pumps.**\n\nClean energy manufacturing jobs are typically very concentrated geographically. In general, manufacturing sectors in EMDEs tend to be more labour-intensive per unit produced than in advanced economies, where mechanisation, automation and other productivity-enhancing measures have been adopted to a greater extent. Most of these jobs are in China: 75% in EVs and battery manufacturing, 85% for solar PV, 55% for wind and 50% for heat pumps (Figure 1.14). Even in the nuclear industry, over 40% of manufacturing jobs are now located in China. By contrast, jobs in the oil and gas industry are concentrated in the Middle East and the United States, with only 10% of jobs in China. Similarly, only about 25% of the manufacturing jobs in ICE vehicles are in China, compared to more than 15% in the European Union, 12% in India and 7% in both Japan and the United States.\n\nThe nature of energy jobs is also changing as the clean energy transition advances, with newly created jobs in the clean energy sector requiring different skills and qualifications. As jobs in fossil fuel supply and in manufacturing the equipment and vehicles that use them are displaced, retraining and reskilling will be important to ensure a fair transition. Many fossil fuel workers already possess\n\n0 1 2 3 4 5 6 7 8\n\nEV and battery\n\nSolar PV\n\nBioenergy\n\nHydro Wind\n\nNuclear\n\nGeothermal Power grids\n\nBuilding retrofit\n\nHeat pump\n\nCoal\n\nNatural gas\n\nOil\n\nMillion workers\n\nTechnology manufacturing Non-manufacturing work Fossil fuel supply","pages":[{"line_numbers":[],"markdown":"Energy Technology Perspectives 2024 Chapter 1: The state of manufacturing and trade\n\nPAGE | 50\n\nI EA. CC BY 4.0.\n\n**Energy employment in selected technology areas, 2022**\n\nIEA. CC BY 4.0.\n\nNotes: EV = electric vehicle. “EV and battery” excludes stationary battery storage. “Bioenergy” excludes waste and traditional use of biomass. “Power grids” refers to transmission and distribution, and stationary battery storage. Non- manufacturing work includes raw materials extraction, construction/installation, utilities, professional services, wholesale and transport. Manufacturing jobs include the final product and direct components (e.g. solar PV includes wafers and cells, while batteries includes cathodes and anodes). Non-manufacturing jobs in the EV and battery category include jobs installing batteries in EVs, but not EV-related jobs such as mechanics or sales, which are not estimated by the IEA. Fossil fuel supply includes upstream (raw materials extraction), midstream (manufacturing of equipment and oil refining) and some downstream activities (wholesale, transport and professional services), but excludes fossil fuel combustion for power generation or distribution. Source: Adapted from IEA (2023d).\n\n**Manufacturing of clean energy technologies is a major source of employment, with around 3 million jobs globally in EVs and batteries, solar PV, wind and heat pumps.**\n\nClean energy manufacturing jobs are typically very concentrated geographically. In general, manufacturing sectors in EMDEs tend to be more labour-intensive per unit produced than in advanced economies, where mechanisation, automation and other productivity-enhancing measures have been adopted to a greater extent. Most of these jobs are in China: 75% in EVs and battery manufacturing, 85% for solar PV, 55% for wind and 50% for heat pumps (Figure 1.14). Even in the nuclear industry, over 40% of manufacturing jobs are now located in China. By contrast, jobs in the oil and gas industry are concentrated in the Middle East and the United States, with only 10% of jobs in China. Similarly, only about 25% of the manufacturing jobs in ICE vehicles are in China, compared to more than 15% in the European Union, 12% in India and 7% in both Japan and the United States.\n\nThe nature of energy jobs is also changing as the clean energy transition advances, with newly created jobs in the clean energy sector requiring different skills and qualifications. As jobs in fossil fuel supply and in manufacturing the equipment and vehicles that use them are displaced, retraining and reskilling will be important to ensure a fair transition. Many fossil fuel workers already possess\n\n0 1 2 3 4 5 6 7 8\n\nEV and battery\n\nSolar PV\n\nBioenergy\n\nHydro Wind\n\nNuclear\n\nGeothermal Power grids\n\nBuilding retrofit\n\nHeat pump\n\nCoal\n\nNatural gas\n\nOil\n\nMillion workers\n\nTechnology manufacturing Non-manufacturing work Fossil fuel supply","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Energy Technology Perspectives 2024 Chapter 1: The state of manufacturing and trade\n\nPAGE | 50\n\nI EA. CC BY 4.0.\n\n**Energy employment in selected technology areas, 2022**\n\nIEA. CC BY 4.0.\n\nNotes: EV = electric vehicle. “EV and battery” excludes stationary battery storage. “Bioenergy” excludes waste and traditional use of biomass. “Power grids” refers to transmission and distribution, and stationary battery storage. Non- manufacturing work includes raw materials extraction, construction/installation, utilities, professional services, wholesale and transport. Manufacturing jobs include the final product and direct components (e.g. solar PV includes wafers and cells, while batteries includes cathodes and anodes). Non-manufacturing jobs in the EV and battery category include jobs installing batteries in EVs, but not EV-related jobs such as mechanics or sales, which are not estimated by the IEA. Fossil fuel supply includes upstream (raw materials extraction), midstream (manufacturing of equipment and oil refining) and some downstream activities (wholesale, transport and professional services), but excludes fossil fuel combustion for power generation or distribution. Source: Adapted from IEA (2023d).\n\n**Manufacturing of clean energy technologies is a major source of employment, with around 3 million jobs globally in EVs and batteries, solar PV, wind and heat pumps.**\n\nClean energy manufacturing jobs are typically very concentrated geographically. In general, manufacturing sectors in EMDEs tend to be more labour-intensive per unit produced than in advanced economies, where mechanisation, automation and other productivity-enhancing measures have been adopted to a greater extent. Most of these jobs are in China: 75% in EVs and battery manufacturing, 85% for solar PV, 55% for wind and 50% for heat pumps (Figure 1.14). Even in the nuclear industry, over 40% of manufacturing jobs are now located in China. By contrast, jobs in the oil and gas industry are concentrated in the Middle East and the United States, with only 10% of jobs in China. Similarly, only about 25% of the manufacturing jobs in ICE vehicles are in China, compared to more than 15% in the European Union, 12% in India and 7% in both Japan and the United States.\n\nThe nature of energy jobs is also changing as the clean energy transition advances, with newly created jobs in the clean energy sector requiring different skills and qualifications. As jobs in fossil fuel supply and in manufacturing the equipment and vehicles that use them are displaced, retraining and reskilling will be important to ensure a fair transition. Many fossil fuel workers already possess\n\n0 1 2 3 4 5 6 7 8\n\nEV and battery\n\nSolar PV\n\nBioenergy\n\nHydro Wind\n\nNuclear\n\nGeothermal Power grids\n\nBuilding retrofit\n\nHeat pump\n\nCoal\n\nNatural gas\n\nOil\n\nMillion workers\n\nTechnology manufacturing Non-manufacturing work Fossil fuel supply","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/EnergyTechnologyPerspectives2024_p52.pdf"]},"response":{"errors":[],"extract_id":"extract_5606fd76147a6cdea94da7b2b3cee297","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Energy Technology Perspectives 2024 Chapter 1: The state of manufacturing and trade\n\nPAGE | 50\n\nI EA. CC BY 4.0.\n\n**Energy employment in selected technology areas, 2022**\n\nIEA. CC BY 4.0.\n\nNotes: EV = electric vehicle. “EV and battery” excludes stationary battery storage. “Bioenergy” excludes waste and traditional use of biomass. “Power grids” refers to transmission and distribution, and stationary battery storage. Non- manufacturing work includes raw materials extraction, construction/installation, utilities, professional services, wholesale and transport. Manufacturing jobs include the final product and direct components (e.g. solar PV includes wafers and cells, while batteries includes cathodes and anodes). Non-manufacturing jobs in the EV and battery category include jobs installing batteries in EVs, but not EV-related jobs such as mechanics or sales, which are not estimated by the IEA. Fossil fuel supply includes upstream (raw materials extraction), midstream (manufacturing of equipment and oil refining) and some downstream activities (wholesale, transport and professional services), but excludes fossil fuel combustion for power generation or distribution. Source: Adapted from IEA (2023d).\n\n**Manufacturing of clean energy technologies is a major source of employment, with around 3 million jobs globally in EVs and batteries, solar PV, wind and heat pumps.**\n\nClean energy manufacturing jobs are typically very concentrated geographically. In general, manufacturing sectors in EMDEs tend to be more labour-intensive per unit produced than in advanced economies, where mechanisation, automation and other productivity-enhancing measures have been adopted to a greater extent. Most of these jobs are in China: 75% in EVs and battery manufacturing, 85% for solar PV, 55% for wind and 50% for heat pumps (Figure 1.14). Even in the nuclear industry, over 40% of manufacturing jobs are now located in China. By contrast, jobs in the oil and gas industry are concentrated in the Middle East and the United States, with only 10% of jobs in China. Similarly, only about 25% of the manufacturing jobs in ICE vehicles are in China, compared to more than 15% in the European Union, 12% in India and 7% in both Japan and the United States.\n\nThe nature of energy jobs is also changing as the clean energy transition advances, with newly created jobs in the clean energy sector requiring different skills and qualifications. As jobs in fossil fuel supply and in manufacturing the equipment and vehicles that use them are displaced, retraining and reskilling will be important to ensure a fair transition. Many fossil fuel workers already possess\n\n0 1 2 3 4 5 6 7 8\n\nEV and battery\n\nSolar PV\n\nBioenergy\n\nHydro Wind\n\nNuclear\n\nGeothermal Power grids\n\nBuilding retrofit\n\nHeat pump\n\nCoal\n\nNatural gas\n\nOil\n\nMillion workers\n\nTechnology manufacturing Non-manufacturing work Fossil fuel supply"],"full_content":"Energy Technology Perspectives 2024 Chapter 1: The state of manufacturing and trade\n\nPAGE | 50\n\nI EA. CC BY 4.0.\n\n**Energy employment in selected technology areas, 2022**\n\nIEA. CC BY 4.0.\n\nNotes: EV = electric vehicle. “EV and battery” excludes stationary battery storage. “Bioenergy” excludes waste and traditional use of biomass. “Power grids” refers to transmission and distribution, and stationary battery storage. Non- manufacturing work includes raw materials extraction, construction/installation, utilities, professional services, wholesale and transport. Manufacturing jobs include the final product and direct components (e.g. solar PV includes wafers and cells, while batteries includes cathodes and anodes). Non-manufacturing jobs in the EV and battery category include jobs installing batteries in EVs, but not EV-related jobs such as mechanics or sales, which are not estimated by the IEA. Fossil fuel supply includes upstream (raw materials extraction), midstream (manufacturing of equipment and oil refining) and some downstream activities (wholesale, transport and professional services), but excludes fossil fuel combustion for power generation or distribution. Source: Adapted from IEA (2023d).\n\n**Manufacturing of clean energy technologies is a major source of employment, with around 3 million jobs globally in EVs and batteries, solar PV, wind and heat pumps.**\n\nClean energy manufacturing jobs are typically very concentrated geographically. In general, manufacturing sectors in EMDEs tend to be more labour-intensive per unit produced than in advanced economies, where mechanisation, automation and other productivity-enhancing measures have been adopted to a greater extent. Most of these jobs are in China: 75% in EVs and battery manufacturing, 85% for solar PV, 55% for wind and 50% for heat pumps (Figure 1.14). Even in the nuclear industry, over 40% of manufacturing jobs are now located in China. By contrast, jobs in the oil and gas industry are concentrated in the Middle East and the United States, with only 10% of jobs in China. Similarly, only about 25% of the manufacturing jobs in ICE vehicles are in China, compared to more than 15% in the European Union, 12% in India and 7% in both Japan and the United States.\n\nThe nature of energy jobs is also changing as the clean energy transition advances, with newly created jobs in the clean energy sector requiring different skills and qualifications. As jobs in fossil fuel supply and in manufacturing the equipment and vehicles that use them are displaced, retraining and reskilling will be important to ensure a fair transition. Many fossil fuel workers already possess\n\n0 1 2 3 4 5 6 7 8\n\nEV and battery\n\nSolar PV\n\nBioenergy\n\nHydro Wind\n\nNuclear\n\nGeothermal Power grids\n\nBuilding retrofit\n\nHeat pump\n\nCoal\n\nNatural gas\n\nOil\n\nMillion workers\n\nTechnology manufacturing Non-manufacturing work Fossil fuel supply","publish_date":null,"title":"Clean energy manufacturing jobs are typically very concentrated geographically. In general, manufacturing sectors in EMDEs tend to be more labour-intensive","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/EnergyTechnologyPerspectives2024_p52.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/EnergyTechnologyPerspectives2024_p52.pdf"},"request":{"example_id":"chart/EnergyTechnologyPerspectives2024_p52","product_type":"parse","source_file_path":"EnergyTechnologyPerspectives2024_p52.pdf"},"started_at":"2026-11-04T19:45:08.555074"} +{"completed_at":"2026-11-04T19:45:10.928712","latency_in_ms":1237,"output":{"example_id":"chart/EnergyTechnologyPerspectives2024_p53","grounded_pages":[],"layout_pages":[],"markdown":"Energy Technology Perspectives 2024 Chapter 1: The state of manufacturing and trade\n\nPAGE | 51\n\nI EA. CC BY 4.0.\n\nsimilar skillsets to those needed in clean energy sectors. Oil and gas workers, for example, are already some of the most highly sought-after employees due to their extensive skills and mobility. Much of the oil and gas workforce possesses skills relevant to bioenergy processing, CCUS, hydrogen production, offshore wind installations and geothermal production. Similarly, many workers engaged in ICE manufacturing are expected to switch to making EVs, especially as a growing number of established Original Equipment Manufacturers offer their own electric models. However, fewer transfers are expected from ICE vehicle supply chains, and fuel storage technology in particular, to EV battery production, since this requires a distinct skillset and is generally carried out by different companies in different locations.\n\n**Energy employment in selected energy technologies by region, 2022**\n\nIEA. CC BY 4.0.\n\nNote: EV = electric vehicles. Source: Adapted from IEA (2023d).\n\n**Most manufacturing jobs for electric vehicles and batteries, renewables and heat pumps are located in China, in contrast to those in the traditional fossil fuel sectors.**\n\n# **1\\.2 Trade**\n\n## **Trade volumes and values**\n\nTrade has always played a key role in the global economy and the energy sector, with energy sector trade traditionally being synonymous with trade in fossil fuels, though this landscape is now increasingly being shaped by trade in clean energy technologies. Fossil fuels account for nearly 40% of international trade by mass, but in 2023, fossil fuels accounted for only around 10% in value terms. Notably, manufactured goods have a much higher value-to-mass ratio than fossil fuels –\n\n0%\n\n20%\n\n40%\n\n60%\n\n80%\n\n100%\n\nEV and battery\n\nBattery storage\n\nSolar PV Wind Nuclear Bioenergy Hydro Heat pumps\n\nCoal Natural\n\ngas\n\nOil\n\nTechnology manufacturing Fossil fuel supply\n\nChina European Union Middle East North America Other Asia Pacific Rest of World","pages":[{"line_numbers":[],"markdown":"Energy Technology Perspectives 2024 Chapter 1: The state of manufacturing and trade\n\nPAGE | 51\n\nI EA. CC BY 4.0.\n\nsimilar skillsets to those needed in clean energy sectors. Oil and gas workers, for example, are already some of the most highly sought-after employees due to their extensive skills and mobility. Much of the oil and gas workforce possesses skills relevant to bioenergy processing, CCUS, hydrogen production, offshore wind installations and geothermal production. Similarly, many workers engaged in ICE manufacturing are expected to switch to making EVs, especially as a growing number of established Original Equipment Manufacturers offer their own electric models. However, fewer transfers are expected from ICE vehicle supply chains, and fuel storage technology in particular, to EV battery production, since this requires a distinct skillset and is generally carried out by different companies in different locations.\n\n**Energy employment in selected energy technologies by region, 2022**\n\nIEA. CC BY 4.0.\n\nNote: EV = electric vehicles. Source: Adapted from IEA (2023d).\n\n**Most manufacturing jobs for electric vehicles and batteries, renewables and heat pumps are located in China, in contrast to those in the traditional fossil fuel sectors.**\n\n# **1\\.2 Trade**\n\n## **Trade volumes and values**\n\nTrade has always played a key role in the global economy and the energy sector, with energy sector trade traditionally being synonymous with trade in fossil fuels, though this landscape is now increasingly being shaped by trade in clean energy technologies. Fossil fuels account for nearly 40% of international trade by mass, but in 2023, fossil fuels accounted for only around 10% in value terms. Notably, manufactured goods have a much higher value-to-mass ratio than fossil fuels –\n\n0%\n\n20%\n\n40%\n\n60%\n\n80%\n\n100%\n\nEV and battery\n\nBattery storage\n\nSolar PV Wind Nuclear Bioenergy Hydro Heat pumps\n\nCoal Natural\n\ngas\n\nOil\n\nTechnology manufacturing Fossil fuel supply\n\nChina European Union Middle East North America Other Asia Pacific Rest of World","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Energy Technology Perspectives 2024 Chapter 1: The state of manufacturing and trade\n\nPAGE | 51\n\nI EA. CC BY 4.0.\n\nsimilar skillsets to those needed in clean energy sectors. Oil and gas workers, for example, are already some of the most highly sought-after employees due to their extensive skills and mobility. Much of the oil and gas workforce possesses skills relevant to bioenergy processing, CCUS, hydrogen production, offshore wind installations and geothermal production. Similarly, many workers engaged in ICE manufacturing are expected to switch to making EVs, especially as a growing number of established Original Equipment Manufacturers offer their own electric models. However, fewer transfers are expected from ICE vehicle supply chains, and fuel storage technology in particular, to EV battery production, since this requires a distinct skillset and is generally carried out by different companies in different locations.\n\n**Energy employment in selected energy technologies by region, 2022**\n\nIEA. CC BY 4.0.\n\nNote: EV = electric vehicles. Source: Adapted from IEA (2023d).\n\n**Most manufacturing jobs for electric vehicles and batteries, renewables and heat pumps are located in China, in contrast to those in the traditional fossil fuel sectors.**\n\n# **1\\.2 Trade**\n\n## **Trade volumes and values**\n\nTrade has always played a key role in the global economy and the energy sector, with energy sector trade traditionally being synonymous with trade in fossil fuels, though this landscape is now increasingly being shaped by trade in clean energy technologies. Fossil fuels account for nearly 40% of international trade by mass, but in 2023, fossil fuels accounted for only around 10% in value terms. Notably, manufactured goods have a much higher value-to-mass ratio than fossil fuels –\n\n0%\n\n20%\n\n40%\n\n60%\n\n80%\n\n100%\n\nEV and battery\n\nBattery storage\n\nSolar PV Wind Nuclear Bioenergy Hydro Heat pumps\n\nCoal Natural\n\ngas\n\nOil\n\nTechnology manufacturing Fossil fuel supply\n\nChina European Union Middle East North America Other Asia Pacific Rest of World","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/EnergyTechnologyPerspectives2024_p53.pdf"]},"response":{"errors":[],"extract_id":"extract_c54d70383b0ca7e75b075a841deda6e4","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Energy Technology Perspectives 2024 Chapter 1: The state of manufacturing and trade\n\nPAGE | 51\n\nI EA. CC BY 4.0.\n\nsimilar skillsets to those needed in clean energy sectors. Oil and gas workers, for example, are already some of the most highly sought-after employees due to their extensive skills and mobility. Much of the oil and gas workforce possesses skills relevant to bioenergy processing, CCUS, hydrogen production, offshore wind installations and geothermal production. Similarly, many workers engaged in ICE manufacturing are expected to switch to making EVs, especially as a growing number of established Original Equipment Manufacturers offer their own electric models. However, fewer transfers are expected from ICE vehicle supply chains, and fuel storage technology in particular, to EV battery production, since this requires a distinct skillset and is generally carried out by different companies in different locations.\n\n**Energy employment in selected energy technologies by region, 2022**\n\nIEA. CC BY 4.0.\n\nNote: EV = electric vehicles. Source: Adapted from IEA (2023d).\n\n**Most manufacturing jobs for electric vehicles and batteries, renewables and heat pumps are located in China, in contrast to those in the traditional fossil fuel sectors.**\n\n# **1\\.2 Trade**\n\n## **Trade volumes and values**\n\nTrade has always played a key role in the global economy and the energy sector, with energy sector trade traditionally being synonymous with trade in fossil fuels, though this landscape is now increasingly being shaped by trade in clean energy technologies. Fossil fuels account for nearly 40% of international trade by mass, but in 2023, fossil fuels accounted for only around 10% in value terms. Notably, manufactured goods have a much higher value-to-mass ratio than fossil fuels –\n\n0%\n\n20%\n\n40%\n\n60%\n\n80%\n\n100%\n\nEV and battery\n\nBattery storage\n\nSolar PV Wind Nuclear Bioenergy Hydro Heat pumps\n\nCoal Natural\n\ngas\n\nOil\n\nTechnology manufacturing Fossil fuel supply\n\nChina European Union Middle East North America Other Asia Pacific Rest of World"],"full_content":"Energy Technology Perspectives 2024 Chapter 1: The state of manufacturing and trade\n\nPAGE | 51\n\nI EA. CC BY 4.0.\n\nsimilar skillsets to those needed in clean energy sectors. Oil and gas workers, for example, are already some of the most highly sought-after employees due to their extensive skills and mobility. Much of the oil and gas workforce possesses skills relevant to bioenergy processing, CCUS, hydrogen production, offshore wind installations and geothermal production. Similarly, many workers engaged in ICE manufacturing are expected to switch to making EVs, especially as a growing number of established Original Equipment Manufacturers offer their own electric models. However, fewer transfers are expected from ICE vehicle supply chains, and fuel storage technology in particular, to EV battery production, since this requires a distinct skillset and is generally carried out by different companies in different locations.\n\n**Energy employment in selected energy technologies by region, 2022**\n\nIEA. CC BY 4.0.\n\nNote: EV = electric vehicles. Source: Adapted from IEA (2023d).\n\n**Most manufacturing jobs for electric vehicles and batteries, renewables and heat pumps are located in China, in contrast to those in the traditional fossil fuel sectors.**\n\n# **1\\.2 Trade**\n\n## **Trade volumes and values**\n\nTrade has always played a key role in the global economy and the energy sector, with energy sector trade traditionally being synonymous with trade in fossil fuels, though this landscape is now increasingly being shaped by trade in clean energy technologies. Fossil fuels account for nearly 40% of international trade by mass, but in 2023, fossil fuels accounted for only around 10% in value terms. Notably, manufactured goods have a much higher value-to-mass ratio than fossil fuels –\n\n0%\n\n20%\n\n40%\n\n60%\n\n80%\n\n100%\n\nEV and battery\n\nBattery storage\n\nSolar PV Wind Nuclear Bioenergy Hydro Heat pumps\n\nCoal Natural\n\ngas\n\nOil\n\nTechnology manufacturing Fossil fuel supply\n\nChina European Union Middle East North America Other Asia Pacific Rest of World","publish_date":null,"title":"1.2 Trade","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/EnergyTechnologyPerspectives2024_p53.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/EnergyTechnologyPerspectives2024_p53.pdf"},"request":{"example_id":"chart/EnergyTechnologyPerspectives2024_p53","product_type":"parse","source_file_path":"EnergyTechnologyPerspectives2024_p53.pdf"},"started_at":"2026-11-04T19:45:09.691219"} +{"completed_at":"2026-11-04T19:45:13.258434","latency_in_ms":1308,"output":{"example_id":"chart/EnergyTechnologyPerspectives2024_p54","grounded_pages":[],"layout_pages":[],"markdown":"Energy Technology Perspectives 2024 Chapter 1: The state of manufacturing and trade\n\nPAGE | 52\n\nI EA. CC BY 4.0.\n\nClean energy technologies made up just 0.2% of total mass traded in 2023, but approximately 1% of total traded value. Other manufactured goods made up just 7% of traded mass, but nearly half of traded value (Figure 1.15).\n\n**Shares of physical goods in global trade, 2023**\n\nIEA. CC BY 4.0.\n\nNotes: Clean energy technology includes batteries, electric vehicles, electrolysers, heat pumps, solar PV and wind. Aluminium includes mineral inputs, intermediates (such as alumina and scrap), ingots and semi- and finished products. Iron and steel include mineral inputs, intermediates (such as scrap, pig iron and direct reduced iron) and semi- and finished products. Other energy includes bioenergy, hydrogen, and nuclear fuel. Other manufactured goods include arms and ammunitions, clothing, electronics, machinery, and pharmaceuticals, among others. Other materials and chemicals include cement, chemicals, ethanol, metals, methanol, paper, plastics and rubber and wood. Other transport includes non-electric vehicles including hybrids, aircrafts, ships and other transport equipment. Source: IEA analysis based on CEPII, (2024); Oxford Economics Limited (2024b); and IEA (2024e).\n\n**Fossil fuels account for 40% of international trade by mass, but only 10% in value.**\n\nRising energy demand, in particular, has historically led to increasing trade of fossil fuels. Between 1973 and 2023, global exports of oil and oil products increased by 60%, reaching more than 24 billion barrels in 2023. For coal, exports increased more than sixfold, reaching 1.2 Gt of coal equivalent (Gtce), while natural gas exports expanded more than fifteen-fold to 1.3 trillion cubic metres (tcm). In aggregate, fossil fuel exports in mass terms increased by 160% over 1973-23, reaching 5.8 Gt. Today, exports make up around 20% of global coal supply, 30% of that of natural gas and 75% of that of oil products (Figure 1.16).\n\n0% 20% 40% 60% 80% 100%\n\nValue\n\nMass\n\nShare of global exports in 2023\n\nFossil fuels Iron and steel Aluminium\n\nFood Other materials and chemicals Other manufactured goods\n\nAmmonia and fertilisers Critical minerals Clean energy technologies\n\nOther transport Other energy","pages":[{"line_numbers":[],"markdown":"Energy Technology Perspectives 2024 Chapter 1: The state of manufacturing and trade\n\nPAGE | 52\n\nI EA. CC BY 4.0.\n\nClean energy technologies made up just 0.2% of total mass traded in 2023, but approximately 1% of total traded value. Other manufactured goods made up just 7% of traded mass, but nearly half of traded value (Figure 1.15).\n\n**Shares of physical goods in global trade, 2023**\n\nIEA. CC BY 4.0.\n\nNotes: Clean energy technology includes batteries, electric vehicles, electrolysers, heat pumps, solar PV and wind. Aluminium includes mineral inputs, intermediates (such as alumina and scrap), ingots and semi- and finished products. Iron and steel include mineral inputs, intermediates (such as scrap, pig iron and direct reduced iron) and semi- and finished products. Other energy includes bioenergy, hydrogen, and nuclear fuel. Other manufactured goods include arms and ammunitions, clothing, electronics, machinery, and pharmaceuticals, among others. Other materials and chemicals include cement, chemicals, ethanol, metals, methanol, paper, plastics and rubber and wood. Other transport includes non-electric vehicles including hybrids, aircrafts, ships and other transport equipment. Source: IEA analysis based on CEPII, (2024); Oxford Economics Limited (2024b); and IEA (2024e).\n\n**Fossil fuels account for 40% of international trade by mass, but only 10% in value.**\n\nRising energy demand, in particular, has historically led to increasing trade of fossil fuels. Between 1973 and 2023, global exports of oil and oil products increased by 60%, reaching more than 24 billion barrels in 2023. For coal, exports increased more than sixfold, reaching 1.2 Gt of coal equivalent (Gtce), while natural gas exports expanded more than fifteen-fold to 1.3 trillion cubic metres (tcm). In aggregate, fossil fuel exports in mass terms increased by 160% over 1973-23, reaching 5.8 Gt. Today, exports make up around 20% of global coal supply, 30% of that of natural gas and 75% of that of oil products (Figure 1.16).\n\n0% 20% 40% 60% 80% 100%\n\nValue\n\nMass\n\nShare of global exports in 2023\n\nFossil fuels Iron and steel Aluminium\n\nFood Other materials and chemicals Other manufactured goods\n\nAmmonia and fertilisers Critical minerals Clean energy technologies\n\nOther transport Other energy","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Energy Technology Perspectives 2024 Chapter 1: The state of manufacturing and trade\n\nPAGE | 52\n\nI EA. CC BY 4.0.\n\nClean energy technologies made up just 0.2% of total mass traded in 2023, but approximately 1% of total traded value. Other manufactured goods made up just 7% of traded mass, but nearly half of traded value (Figure 1.15).\n\n**Shares of physical goods in global trade, 2023**\n\nIEA. CC BY 4.0.\n\nNotes: Clean energy technology includes batteries, electric vehicles, electrolysers, heat pumps, solar PV and wind. Aluminium includes mineral inputs, intermediates (such as alumina and scrap), ingots and semi- and finished products. Iron and steel include mineral inputs, intermediates (such as scrap, pig iron and direct reduced iron) and semi- and finished products. Other energy includes bioenergy, hydrogen, and nuclear fuel. Other manufactured goods include arms and ammunitions, clothing, electronics, machinery, and pharmaceuticals, among others. Other materials and chemicals include cement, chemicals, ethanol, metals, methanol, paper, plastics and rubber and wood. Other transport includes non-electric vehicles including hybrids, aircrafts, ships and other transport equipment. Source: IEA analysis based on CEPII, (2024); Oxford Economics Limited (2024b); and IEA (2024e).\n\n**Fossil fuels account for 40% of international trade by mass, but only 10% in value.**\n\nRising energy demand, in particular, has historically led to increasing trade of fossil fuels. Between 1973 and 2023, global exports of oil and oil products increased by 60%, reaching more than 24 billion barrels in 2023. For coal, exports increased more than sixfold, reaching 1.2 Gt of coal equivalent (Gtce), while natural gas exports expanded more than fifteen-fold to 1.3 trillion cubic metres (tcm). In aggregate, fossil fuel exports in mass terms increased by 160% over 1973-23, reaching 5.8 Gt. Today, exports make up around 20% of global coal supply, 30% of that of natural gas and 75% of that of oil products (Figure 1.16).\n\n0% 20% 40% 60% 80% 100%\n\nValue\n\nMass\n\nShare of global exports in 2023\n\nFossil fuels Iron and steel Aluminium\n\nFood Other materials and chemicals Other manufactured goods\n\nAmmonia and fertilisers Critical minerals Clean energy technologies\n\nOther transport Other energy","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/EnergyTechnologyPerspectives2024_p54.pdf"]},"response":{"errors":[],"extract_id":"extract_af34b9217c82976690c6eedda1801e89","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Energy Technology Perspectives 2024 Chapter 1: The state of manufacturing and trade\n\nPAGE | 52\n\nI EA. CC BY 4.0.\n\nClean energy technologies made up just 0.2% of total mass traded in 2023, but approximately 1% of total traded value. Other manufactured goods made up just 7% of traded mass, but nearly half of traded value (Figure 1.15).\n\n**Shares of physical goods in global trade, 2023**\n\nIEA. CC BY 4.0.\n\nNotes: Clean energy technology includes batteries, electric vehicles, electrolysers, heat pumps, solar PV and wind. Aluminium includes mineral inputs, intermediates (such as alumina and scrap), ingots and semi- and finished products. Iron and steel include mineral inputs, intermediates (such as scrap, pig iron and direct reduced iron) and semi- and finished products. Other energy includes bioenergy, hydrogen, and nuclear fuel. Other manufactured goods include arms and ammunitions, clothing, electronics, machinery, and pharmaceuticals, among others. Other materials and chemicals include cement, chemicals, ethanol, metals, methanol, paper, plastics and rubber and wood. Other transport includes non-electric vehicles including hybrids, aircrafts, ships and other transport equipment. Source: IEA analysis based on CEPII, (2024); Oxford Economics Limited (2024b); and IEA (2024e).\n\n**Fossil fuels account for 40% of international trade by mass, but only 10% in value.**\n\nRising energy demand, in particular, has historically led to increasing trade of fossil fuels. Between 1973 and 2023, global exports of oil and oil products increased by 60%, reaching more than 24 billion barrels in 2023. For coal, exports increased more than sixfold, reaching 1.2 Gt of coal equivalent (Gtce), while natural gas exports expanded more than fifteen-fold to 1.3 trillion cubic metres (tcm). In aggregate, fossil fuel exports in mass terms increased by 160% over 1973-23, reaching 5.8 Gt. Today, exports make up around 20% of global coal supply, 30% of that of natural gas and 75% of that of oil products (Figure 1.16).\n\n0% 20% 40% 60% 80% 100%\n\nValue\n\nMass\n\nShare of global exports in 2023\n\nFossil fuels Iron and steel Aluminium\n\nFood Other materials and chemicals Other manufactured goods\n\nAmmonia and fertilisers Critical minerals Clean energy technologies\n\nOther transport Other energy"],"full_content":"Energy Technology Perspectives 2024 Chapter 1: The state of manufacturing and trade\n\nPAGE | 52\n\nI EA. CC BY 4.0.\n\nClean energy technologies made up just 0.2% of total mass traded in 2023, but approximately 1% of total traded value. Other manufactured goods made up just 7% of traded mass, but nearly half of traded value (Figure 1.15).\n\n**Shares of physical goods in global trade, 2023**\n\nIEA. CC BY 4.0.\n\nNotes: Clean energy technology includes batteries, electric vehicles, electrolysers, heat pumps, solar PV and wind. Aluminium includes mineral inputs, intermediates (such as alumina and scrap), ingots and semi- and finished products. Iron and steel include mineral inputs, intermediates (such as scrap, pig iron and direct reduced iron) and semi- and finished products. Other energy includes bioenergy, hydrogen, and nuclear fuel. Other manufactured goods include arms and ammunitions, clothing, electronics, machinery, and pharmaceuticals, among others. Other materials and chemicals include cement, chemicals, ethanol, metals, methanol, paper, plastics and rubber and wood. Other transport includes non-electric vehicles including hybrids, aircrafts, ships and other transport equipment. Source: IEA analysis based on CEPII, (2024); Oxford Economics Limited (2024b); and IEA (2024e).\n\n**Fossil fuels account for 40% of international trade by mass, but only 10% in value.**\n\nRising energy demand, in particular, has historically led to increasing trade of fossil fuels. Between 1973 and 2023, global exports of oil and oil products increased by 60%, reaching more than 24 billion barrels in 2023. For coal, exports increased more than sixfold, reaching 1.2 Gt of coal equivalent (Gtce), while natural gas exports expanded more than fifteen-fold to 1.3 trillion cubic metres (tcm). In aggregate, fossil fuel exports in mass terms increased by 160% over 1973-23, reaching 5.8 Gt. Today, exports make up around 20% of global coal supply, 30% of that of natural gas and 75% of that of oil products (Figure 1.16).\n\n0% 20% 40% 60% 80% 100%\n\nValue\n\nMass\n\nShare of global exports in 2023\n\nFossil fuels Iron and steel Aluminium\n\nFood Other materials and chemicals Other manufactured goods\n\nAmmonia and fertilisers Critical minerals Clean energy technologies\n\nOther transport Other energy","publish_date":null,"title":"Clean energy technologies made up just 0.2% of total mass traded in 2023, but approximately 1% of total traded value.","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/EnergyTechnologyPerspectives2024_p54.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/EnergyTechnologyPerspectives2024_p54.pdf"},"request":{"example_id":"chart/EnergyTechnologyPerspectives2024_p54","product_type":"parse","source_file_path":"EnergyTechnologyPerspectives2024_p54.pdf"},"started_at":"2026-11-04T19:45:11.950324"} +{"completed_at":"2026-11-04T19:45:15.492258","latency_in_ms":2232,"output":{"example_id":"chart/EnergyTechnologyPerspectives2024_p57","grounded_pages":[],"layout_pages":[],"markdown":"Energy Technology Perspectives 2024 Chapter 1: The state of manufacturing and trade\n\nPAGE | 55\n\nI EA. CC BY 4.0.\n\n**Global exports of selected clean technologies and materials, 2010-2023**\n\nIEA. CC BY 4.0.\n\nNotes: For electric cars, the index is based on units (not mass) and scaled in 2015 = 100. Aluminium includes mineral inputs, intermediates (such as alumina and scrap), ingots, and semi- and finished products. Iron and steel include mineral inputs, intermediates (such as scrap, pig iron and direct reduced iron) and semi- and finished products. Trade in materials refers to total trade, not just the materials used for manufacturing clean energy technologies. The trade value of the selected technologies reflects only the final exported unit and does not separately account for the value of the materials and components embedded in them. Source: IEA analysis based on EV Volumes (2024); CEPII (2024); Oxford Economics Limited (2024b); pv-magazine (2024b); BNEF (2024a); SPV Market Research (2024); InfoLink (2024); IEA-PVPS (2024); and RTS Corporation (2024).\n\n**Exports of clean energy technologies are increasing rapidly in value and mass terms.**\n\nGlobal exports for several clean technologies have increased more quickly than those of other more established sectors of the economy like food, pharmaceuticals, and textiles and clothing over 2010 to 2023. For these three sectors, trade increased by between 25% to 50% in mass terms over the period – an increase similar to that of the three selected materials and all of their precursors. The volume of trade in clean technologies nonetheless remains small: in mass terms, the combined exports of the three technologies mentioned above, electric cars, solar PV and heat pumps, were around 60% of that of textile and clothing and just above 5% of that of food in 2023.\n\n0\n\n500\n\n1 000\n\n1 500\n\n2 000\n\n2010 2015 2020\n\nIndex (2010 = 100)\n\n**Solar PV modules**\n\n**​**\n\n**​**\n\n**​**\n\n**​**\n\n0\n\n500\n\n1 000\n\n1 500\n\n2 000\n\n2010 2015 2020\n\n**Electric cars\\***\n\n0\n\n50\n\n100\n\n150\n\n200\n\n2010 2015 2020\n\n**Heat pumps**\n\n0\n\n50\n\n100\n\n150\n\n200\n\n2010 2015 2020\n\nIndex (2010 = 100)\n\n**Aluminum**\n\nValue Mass\n\n**​**\n\n**​**\n\n**​**\n\n**​**\n\n0\n\n50\n\n100\n\n150\n\n200\n\n2010 2015 2020\n\n**Iron and steel**\n\n0\n\n50\n\n100\n\n150\n\n200\n\n2010 2015 2020\n\n**Fertiliser**","pages":[{"line_numbers":[],"markdown":"Energy Technology Perspectives 2024 Chapter 1: The state of manufacturing and trade\n\nPAGE | 55\n\nI EA. CC BY 4.0.\n\n**Global exports of selected clean technologies and materials, 2010-2023**\n\nIEA. CC BY 4.0.\n\nNotes: For electric cars, the index is based on units (not mass) and scaled in 2015 = 100. Aluminium includes mineral inputs, intermediates (such as alumina and scrap), ingots, and semi- and finished products. Iron and steel include mineral inputs, intermediates (such as scrap, pig iron and direct reduced iron) and semi- and finished products. Trade in materials refers to total trade, not just the materials used for manufacturing clean energy technologies. The trade value of the selected technologies reflects only the final exported unit and does not separately account for the value of the materials and components embedded in them. Source: IEA analysis based on EV Volumes (2024); CEPII (2024); Oxford Economics Limited (2024b); pv-magazine (2024b); BNEF (2024a); SPV Market Research (2024); InfoLink (2024); IEA-PVPS (2024); and RTS Corporation (2024).\n\n**Exports of clean energy technologies are increasing rapidly in value and mass terms.**\n\nGlobal exports for several clean technologies have increased more quickly than those of other more established sectors of the economy like food, pharmaceuticals, and textiles and clothing over 2010 to 2023. For these three sectors, trade increased by between 25% to 50% in mass terms over the period – an increase similar to that of the three selected materials and all of their precursors. The volume of trade in clean technologies nonetheless remains small: in mass terms, the combined exports of the three technologies mentioned above, electric cars, solar PV and heat pumps, were around 60% of that of textile and clothing and just above 5% of that of food in 2023.\n\n0\n\n500\n\n1 000\n\n1 500\n\n2 000\n\n2010 2015 2020\n\nIndex (2010 = 100)\n\n**Solar PV modules**\n\n**​**\n\n**​**\n\n**​**\n\n**​**\n\n0\n\n500\n\n1 000\n\n1 500\n\n2 000\n\n2010 2015 2020\n\n**Electric cars\\***\n\n0\n\n50\n\n100\n\n150\n\n200\n\n2010 2015 2020\n\n**Heat pumps**\n\n0\n\n50\n\n100\n\n150\n\n200\n\n2010 2015 2020\n\nIndex (2010 = 100)\n\n**Aluminum**\n\nValue Mass\n\n**​**\n\n**​**\n\n**​**\n\n**​**\n\n0\n\n50\n\n100\n\n150\n\n200\n\n2010 2015 2020\n\n**Iron and steel**\n\n0\n\n50\n\n100\n\n150\n\n200\n\n2010 2015 2020\n\n**Fertiliser**","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Energy Technology Perspectives 2024 Chapter 1: The state of manufacturing and trade\n\nPAGE | 55\n\nI EA. CC BY 4.0.\n\n**Global exports of selected clean technologies and materials, 2010-2023**\n\nIEA. CC BY 4.0.\n\nNotes: For electric cars, the index is based on units (not mass) and scaled in 2015 = 100. Aluminium includes mineral inputs, intermediates (such as alumina and scrap), ingots, and semi- and finished products. Iron and steel include mineral inputs, intermediates (such as scrap, pig iron and direct reduced iron) and semi- and finished products. Trade in materials refers to total trade, not just the materials used for manufacturing clean energy technologies. The trade value of the selected technologies reflects only the final exported unit and does not separately account for the value of the materials and components embedded in them. Source: IEA analysis based on EV Volumes (2024); CEPII (2024); Oxford Economics Limited (2024b); pv-magazine (2024b); BNEF (2024a); SPV Market Research (2024); InfoLink (2024); IEA-PVPS (2024); and RTS Corporation (2024).\n\n**Exports of clean energy technologies are increasing rapidly in value and mass terms.**\n\nGlobal exports for several clean technologies have increased more quickly than those of other more established sectors of the economy like food, pharmaceuticals, and textiles and clothing over 2010 to 2023. For these three sectors, trade increased by between 25% to 50% in mass terms over the period – an increase similar to that of the three selected materials and all of their precursors. The volume of trade in clean technologies nonetheless remains small: in mass terms, the combined exports of the three technologies mentioned above, electric cars, solar PV and heat pumps, were around 60% of that of textile and clothing and just above 5% of that of food in 2023.\n\n0\n\n500\n\n1 000\n\n1 500\n\n2 000\n\n2010 2015 2020\n\nIndex (2010 = 100)\n\n**Solar PV modules**\n\n**​**\n\n**​**\n\n**​**\n\n**​**\n\n0\n\n500\n\n1 000\n\n1 500\n\n2 000\n\n2010 2015 2020\n\n**Electric cars\\***\n\n0\n\n50\n\n100\n\n150\n\n200\n\n2010 2015 2020\n\n**Heat pumps**\n\n0\n\n50\n\n100\n\n150\n\n200\n\n2010 2015 2020\n\nIndex (2010 = 100)\n\n**Aluminum**\n\nValue Mass\n\n**​**\n\n**​**\n\n**​**\n\n**​**\n\n0\n\n50\n\n100\n\n150\n\n200\n\n2010 2015 2020\n\n**Iron and steel**\n\n0\n\n50\n\n100\n\n150\n\n200\n\n2010 2015 2020\n\n**Fertiliser**","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/EnergyTechnologyPerspectives2024_p57.pdf"]},"response":{"errors":[],"extract_id":"extract_6901992e96a6d23f33ac5b1922411c66","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Energy Technology Perspectives 2024 Chapter 1: The state of manufacturing and trade\n\nPAGE | 55\n\nI EA. CC BY 4.0.\n\n**Global exports of selected clean technologies and materials, 2010-2023**\n\nIEA. CC BY 4.0.\n\nNotes: For electric cars, the index is based on units (not mass) and scaled in 2015 = 100. Aluminium includes mineral inputs, intermediates (such as alumina and scrap), ingots, and semi- and finished products. Iron and steel include mineral inputs, intermediates (such as scrap, pig iron and direct reduced iron) and semi- and finished products. Trade in materials refers to total trade, not just the materials used for manufacturing clean energy technologies. The trade value of the selected technologies reflects only the final exported unit and does not separately account for the value of the materials and components embedded in them. Source: IEA analysis based on EV Volumes (2024); CEPII (2024); Oxford Economics Limited (2024b); pv-magazine (2024b); BNEF (2024a); SPV Market Research (2024); InfoLink (2024); IEA-PVPS (2024); and RTS Corporation (2024).\n\n**Exports of clean energy technologies are increasing rapidly in value and mass terms.**\n\nGlobal exports for several clean technologies have increased more quickly than those of other more established sectors of the economy like food, pharmaceuticals, and textiles and clothing over 2010 to 2023. For these three sectors, trade increased by between 25% to 50% in mass terms over the period – an increase similar to that of the three selected materials and all of their precursors. The volume of trade in clean technologies nonetheless remains small: in mass terms, the combined exports of the three technologies mentioned above, electric cars, solar PV and heat pumps, were around 60% of that of textile and clothing and just above 5% of that of food in 2023.\n\n0\n\n500\n\n1 000\n\n1 500\n\n2 000\n\n2010 2015 2020\n\nIndex (2010 = 100)\n\n**Solar PV modules**\n\n**​**\n\n**​**\n\n**​**\n\n**​**\n\n0\n\n500\n\n1 000\n\n1 500\n\n2 000\n\n2010 2015 2020\n\n**Electric cars\\***\n\n0\n\n50\n\n100\n\n150\n\n200\n\n2010 2015 2020\n\n**Heat pumps**\n\n0\n\n50\n\n100\n\n150\n\n200\n\n2010 2015 2020\n\nIndex (2010 = 100)\n\n**Aluminum**\n\nValue Mass\n\n**​**\n\n**​**\n\n**​**\n\n**​**\n\n0\n\n50\n\n100\n\n150\n\n200\n\n2010 2015 2020\n\n**Iron and steel**\n\n0\n\n50\n\n100\n\n150\n\n200\n\n2010 2015 2020\n\n**Fertiliser**"],"full_content":"Energy Technology Perspectives 2024 Chapter 1: The state of manufacturing and trade\n\nPAGE | 55\n\nI EA. CC BY 4.0.\n\n**Global exports of selected clean technologies and materials, 2010-2023**\n\nIEA. CC BY 4.0.\n\nNotes: For electric cars, the index is based on units (not mass) and scaled in 2015 = 100. Aluminium includes mineral inputs, intermediates (such as alumina and scrap), ingots, and semi- and finished products. Iron and steel include mineral inputs, intermediates (such as scrap, pig iron and direct reduced iron) and semi- and finished products. Trade in materials refers to total trade, not just the materials used for manufacturing clean energy technologies. The trade value of the selected technologies reflects only the final exported unit and does not separately account for the value of the materials and components embedded in them. Source: IEA analysis based on EV Volumes (2024); CEPII (2024); Oxford Economics Limited (2024b); pv-magazine (2024b); BNEF (2024a); SPV Market Research (2024); InfoLink (2024); IEA-PVPS (2024); and RTS Corporation (2024).\n\n**Exports of clean energy technologies are increasing rapidly in value and mass terms.**\n\nGlobal exports for several clean technologies have increased more quickly than those of other more established sectors of the economy like food, pharmaceuticals, and textiles and clothing over 2010 to 2023. For these three sectors, trade increased by between 25% to 50% in mass terms over the period – an increase similar to that of the three selected materials and all of their precursors. The volume of trade in clean technologies nonetheless remains small: in mass terms, the combined exports of the three technologies mentioned above, electric cars, solar PV and heat pumps, were around 60% of that of textile and clothing and just above 5% of that of food in 2023.\n\n0\n\n500\n\n1 000\n\n1 500\n\n2 000\n\n2010 2015 2020\n\nIndex (2010 = 100)\n\n**Solar PV modules**\n\n**​**\n\n**​**\n\n**​**\n\n**​**\n\n0\n\n500\n\n1 000\n\n1 500\n\n2 000\n\n2010 2015 2020\n\n**Electric cars\\***\n\n0\n\n50\n\n100\n\n150\n\n200\n\n2010 2015 2020\n\n**Heat pumps**\n\n0\n\n50\n\n100\n\n150\n\n200\n\n2010 2015 2020\n\nIndex (2010 = 100)\n\n**Aluminum**\n\nValue Mass\n\n**​**\n\n**​**\n\n**​**\n\n**​**\n\n0\n\n50\n\n100\n\n150\n\n200\n\n2010 2015 2020\n\n**Iron and steel**\n\n0\n\n50\n\n100\n\n150\n\n200\n\n2010 2015 2020\n\n**Fertiliser**","publish_date":null,"title":"Global exports for several clean technologies have increased more quickly than those of other more established sectors of the economy","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/EnergyTechnologyPerspectives2024_p57.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/EnergyTechnologyPerspectives2024_p57.pdf"},"request":{"example_id":"chart/EnergyTechnologyPerspectives2024_p57","product_type":"parse","source_file_path":"EnergyTechnologyPerspectives2024_p57.pdf"},"started_at":"2026-11-04T19:45:13.260206"} +{"completed_at":"2026-11-04T19:45:16.438184","latency_in_ms":2071,"output":{"example_id":"chart/FPA-guide-to-data-visualization_p20","grounded_pages":[],"layout_pages":[],"markdown":"AFP GUIDE: **Data Visualization: How to Tell Number Stories with Pictures**\n\n### **Pie charts**\n\nThere’s a big debate about the utility of pie charts. Most experts advise against them, noting that they are harder to read and of- ten the small differences between slices are easy to miss. Below is an example of a chart first created as a pie chart and then, when rules of best practices in visualization are applied, recreated as a bar.\n\n“Our current belief is that getting rid of\n\neither completely is not productive,” said Laird. “Our built-in construct library does contain pie charts for certain contexts; we believe they are especially useful when comparing the impact of two different sce- narios on an annual basis when limited variables (3 or 4) are involved (usually a business unit or region). In this case, it might be any measurement: OPEX, bookings, rev- enue, contribution margin, etc. Many would argue that a column or bar format would be preferable even in these cases. We would say that, especially in the example provided here, simplicity is actually preferable and easier for cognitive processing.”\n\nAccording to Mosely, pie charts can act\n\nas a useful visualization\n\n• When you want to affect your audience emotionally. • When you want to quickly communi-\n\ncate a part-to-whole relationship.\n\n• When approximate values are enough\n\nto have a productive discussion.\n\nBut only if … • The parts make up a meaningful\n\nwhole; if not, use a different chart.\n\n• You can define the entire set in a way\n\nthat makes sense to the viewer.\n\n• The parts are mutually exclusive; if\n\nthere is overlap between the parts, use a different chart.\n\n• You want to compare the parts to the\n\nwhole, rather than to each other.\n\n• You have fewer than eight parts; pie\n\ncharts with lots of slices (or slices of very different sizes) are hard to read.\n\nFY 14 Q2 Forecast Total: $750,000\n\nSalaries\n\nand\n\nBenefits\n\n61%\n\nContracts and\n\nProfessional Fees\n\n9%\n\nDepreciation, Interest\n\nand Maintenance\n\n10%\n\nOffice Expenses and Other Costs\\*\n\n7%\n\nConference, Meetings\n\nand Exhibits\n\n7%\n\nTravel and Business Entertainment\n\n2%\n\nComputer Services\n\n2% Editing/Publishing\n\n2%\n\nFY 14 Jan Actual Total: $387,000k\n\n0% 10% 20% 30% 40% 50% 60% 70%\n\nSalaries and Benefits\n\nContracts and Professional Fees\n\nDepreciation, Interest\n\nand Maintenance Office Expenses and Other Costs\\*\n\nConference, Meetings\n\nand Exhibits\n\nTravel and Business\n\nEntertainment\n\nComputer Services\n\nEditing/Publishing\n\nData graphed after applying data visualization rules:\n\nSalaries\n\nand\n\nBenefits 60.3%\n\nContracts and\n\nProfessional Fees\n\n10\\.1%\n\nDepreciation, Interest\n\nand Maintenance\n\n9%\n\nOffice Expenses and Other Costs\\*\n\n8%\n\nConference, Meetings\n\nand Exhibits\n\n5%\n\nTravel and Business Entertainment\n\n3\\.5%\n\nComputer Services\n\n2\\.7% Editing/Publishing\n\n1\\.4%\n\n\\*Other costs category includes marketing and promotions costs.\n\n\\*Other costs category includes marketing and promotions costs.\n\nFY14 Jan Actual FY14 Q2 Forecast\n\nMultiple benefits were achieved by applying data visualization rules: a. Two graphs have been combined into one allowing for\n\na. Easy comparison between data sets (FY14 Q2 Forecast\n\nto Actual)\n\nb. Additional real estate on documents with limited space\n\nb. Different type of display media was used to more clearly\n\nshow the relationship between the expense types\n\nc. Color was used sparingly to mark forecasted expense percentages drawing attention to the most important\n\naspects of the graph\n\nd. By using a different display media there is more data flexibility. For example, another dimension could be added to the bar graph such as FY14 budget data etc.","pages":[{"line_numbers":[],"markdown":"AFP GUIDE: **Data Visualization: How to Tell Number Stories with Pictures**\n\n### **Pie charts**\n\nThere’s a big debate about the utility of pie charts. Most experts advise against them, noting that they are harder to read and of- ten the small differences between slices are easy to miss. Below is an example of a chart first created as a pie chart and then, when rules of best practices in visualization are applied, recreated as a bar.\n\n“Our current belief is that getting rid of\n\neither completely is not productive,” said Laird. “Our built-in construct library does contain pie charts for certain contexts; we believe they are especially useful when comparing the impact of two different sce- narios on an annual basis when limited variables (3 or 4) are involved (usually a business unit or region). In this case, it might be any measurement: OPEX, bookings, rev- enue, contribution margin, etc. Many would argue that a column or bar format would be preferable even in these cases. We would say that, especially in the example provided here, simplicity is actually preferable and easier for cognitive processing.”\n\nAccording to Mosely, pie charts can act\n\nas a useful visualization\n\n• When you want to affect your audience emotionally. • When you want to quickly communi-\n\ncate a part-to-whole relationship.\n\n• When approximate values are enough\n\nto have a productive discussion.\n\nBut only if … • The parts make up a meaningful\n\nwhole; if not, use a different chart.\n\n• You can define the entire set in a way\n\nthat makes sense to the viewer.\n\n• The parts are mutually exclusive; if\n\nthere is overlap between the parts, use a different chart.\n\n• You want to compare the parts to the\n\nwhole, rather than to each other.\n\n• You have fewer than eight parts; pie\n\ncharts with lots of slices (or slices of very different sizes) are hard to read.\n\nFY 14 Q2 Forecast Total: $750,000\n\nSalaries\n\nand\n\nBenefits\n\n61%\n\nContracts and\n\nProfessional Fees\n\n9%\n\nDepreciation, Interest\n\nand Maintenance\n\n10%\n\nOffice Expenses and Other Costs\\*\n\n7%\n\nConference, Meetings\n\nand Exhibits\n\n7%\n\nTravel and Business Entertainment\n\n2%\n\nComputer Services\n\n2% Editing/Publishing\n\n2%\n\nFY 14 Jan Actual Total: $387,000k\n\n0% 10% 20% 30% 40% 50% 60% 70%\n\nSalaries and Benefits\n\nContracts and Professional Fees\n\nDepreciation, Interest\n\nand Maintenance Office Expenses and Other Costs\\*\n\nConference, Meetings\n\nand Exhibits\n\nTravel and Business\n\nEntertainment\n\nComputer Services\n\nEditing/Publishing\n\nData graphed after applying data visualization rules:\n\nSalaries\n\nand\n\nBenefits 60.3%\n\nContracts and\n\nProfessional Fees\n\n10\\.1%\n\nDepreciation, Interest\n\nand Maintenance\n\n9%\n\nOffice Expenses and Other Costs\\*\n\n8%\n\nConference, Meetings\n\nand Exhibits\n\n5%\n\nTravel and Business Entertainment\n\n3\\.5%\n\nComputer Services\n\n2\\.7% Editing/Publishing\n\n1\\.4%\n\n\\*Other costs category includes marketing and promotions costs.\n\n\\*Other costs category includes marketing and promotions costs.\n\nFY14 Jan Actual FY14 Q2 Forecast\n\nMultiple benefits were achieved by applying data visualization rules: a. Two graphs have been combined into one allowing for\n\na. Easy comparison between data sets (FY14 Q2 Forecast\n\nto Actual)\n\nb. Additional real estate on documents with limited space\n\nb. Different type of display media was used to more clearly\n\nshow the relationship between the expense types\n\nc. Color was used sparingly to mark forecasted expense percentages drawing attention to the most important\n\naspects of the graph\n\nd. By using a different display media there is more data flexibility. For example, another dimension could be added to the bar graph such as FY14 budget data etc.","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"AFP GUIDE: **Data Visualization: How to Tell Number Stories with Pictures**\n\n### **Pie charts**\n\nThere’s a big debate about the utility of pie charts. Most experts advise against them, noting that they are harder to read and of- ten the small differences between slices are easy to miss. Below is an example of a chart first created as a pie chart and then, when rules of best practices in visualization are applied, recreated as a bar.\n\n“Our current belief is that getting rid of\n\neither completely is not productive,” said Laird. “Our built-in construct library does contain pie charts for certain contexts; we believe they are especially useful when comparing the impact of two different sce- narios on an annual basis when limited variables (3 or 4) are involved (usually a business unit or region). In this case, it might be any measurement: OPEX, bookings, rev- enue, contribution margin, etc. Many would argue that a column or bar format would be preferable even in these cases. We would say that, especially in the example provided here, simplicity is actually preferable and easier for cognitive processing.”\n\nAccording to Mosely, pie charts can act\n\nas a useful visualization\n\n• When you want to affect your audience emotionally. • When you want to quickly communi-\n\ncate a part-to-whole relationship.\n\n• When approximate values are enough\n\nto have a productive discussion.\n\nBut only if … • The parts make up a meaningful\n\nwhole; if not, use a different chart.\n\n• You can define the entire set in a way\n\nthat makes sense to the viewer.\n\n• The parts are mutually exclusive; if\n\nthere is overlap between the parts, use a different chart.\n\n• You want to compare the parts to the\n\nwhole, rather than to each other.\n\n• You have fewer than eight parts; pie\n\ncharts with lots of slices (or slices of very different sizes) are hard to read.\n\nFY 14 Q2 Forecast Total: $750,000\n\nSalaries\n\nand\n\nBenefits\n\n61%\n\nContracts and\n\nProfessional Fees\n\n9%\n\nDepreciation, Interest\n\nand Maintenance\n\n10%\n\nOffice Expenses and Other Costs\\*\n\n7%\n\nConference, Meetings\n\nand Exhibits\n\n7%\n\nTravel and Business Entertainment\n\n2%\n\nComputer Services\n\n2% Editing/Publishing\n\n2%\n\nFY 14 Jan Actual Total: $387,000k\n\n0% 10% 20% 30% 40% 50% 60% 70%\n\nSalaries and Benefits\n\nContracts and Professional Fees\n\nDepreciation, Interest\n\nand Maintenance Office Expenses and Other Costs\\*\n\nConference, Meetings\n\nand Exhibits\n\nTravel and Business\n\nEntertainment\n\nComputer Services\n\nEditing/Publishing\n\nData graphed after applying data visualization rules:\n\nSalaries\n\nand\n\nBenefits 60.3%\n\nContracts and\n\nProfessional Fees\n\n10\\.1%\n\nDepreciation, Interest\n\nand Maintenance\n\n9%\n\nOffice Expenses and Other Costs\\*\n\n8%\n\nConference, Meetings\n\nand Exhibits\n\n5%\n\nTravel and Business Entertainment\n\n3\\.5%\n\nComputer Services\n\n2\\.7% Editing/Publishing\n\n1\\.4%\n\n\\*Other costs category includes marketing and promotions costs.\n\n\\*Other costs category includes marketing and promotions costs.\n\nFY14 Jan Actual FY14 Q2 Forecast\n\nMultiple benefits were achieved by applying data visualization rules: a. Two graphs have been combined into one allowing for\n\na. Easy comparison between data sets (FY14 Q2 Forecast\n\nto Actual)\n\nb. Additional real estate on documents with limited space\n\nb. Different type of display media was used to more clearly\n\nshow the relationship between the expense types\n\nc. Color was used sparingly to mark forecasted expense percentages drawing attention to the most important\n\naspects of the graph\n\nd. By using a different display media there is more data flexibility. For example, another dimension could be added to the bar graph such as FY14 budget data etc.","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/FPA-guide-to-data-visualization_p20.pdf"]},"response":{"errors":[],"extract_id":"extract_3cf37d310557ceaa9da37af3377ecc7f","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["AFP GUIDE: **Data Visualization: How to Tell Number Stories with Pictures**\n\n### **Pie charts**\n\nThere’s a big debate about the utility of pie charts. Most experts advise against them, noting that they are harder to read and of- ten the small differences between slices are easy to miss. Below is an example of a chart first created as a pie chart and then, when rules of best practices in visualization are applied, recreated as a bar.\n\n“Our current belief is that getting rid of\n\neither completely is not productive,” said Laird. “Our built-in construct library does contain pie charts for certain contexts; we believe they are especially useful when comparing the impact of two different sce- narios on an annual basis when limited variables (3 or 4) are involved (usually a business unit or region). In this case, it might be any measurement: OPEX, bookings, rev- enue, contribution margin, etc. Many would argue that a column or bar format would be preferable even in these cases. We would say that, especially in the example provided here, simplicity is actually preferable and easier for cognitive processing.”\n\nAccording to Mosely, pie charts can act\n\nas a useful visualization\n\n• When you want to affect your audience emotionally. • When you want to quickly communi-\n\ncate a part-to-whole relationship.\n\n• When approximate values are enough\n\nto have a productive discussion.\n\nBut only if … • The parts make up a meaningful\n\nwhole; if not, use a different chart.\n\n• You can define the entire set in a way\n\nthat makes sense to the viewer.\n\n• The parts are mutually exclusive; if\n\nthere is overlap between the parts, use a different chart.\n\n• You want to compare the parts to the\n\nwhole, rather than to each other.\n\n• You have fewer than eight parts; pie\n\ncharts with lots of slices (or slices of very different sizes) are hard to read.\n\nFY 14 Q2 Forecast Total: $750,000\n\nSalaries\n\nand\n\nBenefits\n\n61%\n\nContracts and\n\nProfessional Fees\n\n9%\n\nDepreciation, Interest\n\nand Maintenance\n\n10%\n\nOffice Expenses and Other Costs\\*\n\n7%\n\nConference, Meetings\n\nand Exhibits\n\n7%\n\nTravel and Business Entertainment\n\n2%\n\nComputer Services\n\n2% Editing/Publishing\n\n2%\n\nFY 14 Jan Actual Total: $387,000k\n\n0% 10% 20% 30% 40% 50% 60% 70%\n\nSalaries and Benefits\n\nContracts and Professional Fees\n\nDepreciation, Interest\n\nand Maintenance Office Expenses and Other Costs\\*\n\nConference, Meetings\n\nand Exhibits\n\nTravel and Business\n\nEntertainment\n\nComputer Services\n\nEditing/Publishing\n\nData graphed after applying data visualization rules:\n\nSalaries\n\nand\n\nBenefits 60.3%\n\nContracts and\n\nProfessional Fees\n\n10\\.1%\n\nDepreciation, Interest\n\nand Maintenance\n\n9%\n\nOffice Expenses and Other Costs\\*\n\n8%\n\nConference, Meetings\n\nand Exhibits\n\n5%\n\nTravel and Business Entertainment\n\n3\\.5%\n\nComputer Services\n\n2\\.7% Editing/Publishing\n\n1\\.4%\n\n\\*Other costs category includes marketing and promotions costs.\n\n\\*Other costs category includes marketing and promotions costs.\n\nFY14 Jan Actual FY14 Q2 Forecast\n\nMultiple benefits were achieved by applying data visualization rules: a. Two graphs have been combined into one allowing for\n\na. Easy comparison between data sets (FY14 Q2 Forecast\n\nto Actual)\n\nb. Additional real estate on documents with limited space\n\nb. Different type of display media was used to more clearly\n\nshow the relationship between the expense types\n\nc. Color was used sparingly to mark forecasted expense percentages drawing attention to the most important\n\naspects of the graph\n\nd. By using a different display media there is more data flexibility. For example, another dimension could be added to the bar graph such as FY14 budget data etc."],"full_content":"AFP GUIDE: **Data Visualization: How to Tell Number Stories with Pictures**\n\n### **Pie charts**\n\nThere’s a big debate about the utility of pie charts. Most experts advise against them, noting that they are harder to read and of- ten the small differences between slices are easy to miss. Below is an example of a chart first created as a pie chart and then, when rules of best practices in visualization are applied, recreated as a bar.\n\n“Our current belief is that getting rid of\n\neither completely is not productive,” said Laird. “Our built-in construct library does contain pie charts for certain contexts; we believe they are especially useful when comparing the impact of two different sce- narios on an annual basis when limited variables (3 or 4) are involved (usually a business unit or region). In this case, it might be any measurement: OPEX, bookings, rev- enue, contribution margin, etc. Many would argue that a column or bar format would be preferable even in these cases. We would say that, especially in the example provided here, simplicity is actually preferable and easier for cognitive processing.”\n\nAccording to Mosely, pie charts can act\n\nas a useful visualization\n\n• When you want to affect your audience emotionally. • When you want to quickly communi-\n\ncate a part-to-whole relationship.\n\n• When approximate values are enough\n\nto have a productive discussion.\n\nBut only if … • The parts make up a meaningful\n\nwhole; if not, use a different chart.\n\n• You can define the entire set in a way\n\nthat makes sense to the viewer.\n\n• The parts are mutually exclusive; if\n\nthere is overlap between the parts, use a different chart.\n\n• You want to compare the parts to the\n\nwhole, rather than to each other.\n\n• You have fewer than eight parts; pie\n\ncharts with lots of slices (or slices of very different sizes) are hard to read.\n\nFY 14 Q2 Forecast Total: $750,000\n\nSalaries\n\nand\n\nBenefits\n\n61%\n\nContracts and\n\nProfessional Fees\n\n9%\n\nDepreciation, Interest\n\nand Maintenance\n\n10%\n\nOffice Expenses and Other Costs\\*\n\n7%\n\nConference, Meetings\n\nand Exhibits\n\n7%\n\nTravel and Business Entertainment\n\n2%\n\nComputer Services\n\n2% Editing/Publishing\n\n2%\n\nFY 14 Jan Actual Total: $387,000k\n\n0% 10% 20% 30% 40% 50% 60% 70%\n\nSalaries and Benefits\n\nContracts and Professional Fees\n\nDepreciation, Interest\n\nand Maintenance Office Expenses and Other Costs\\*\n\nConference, Meetings\n\nand Exhibits\n\nTravel and Business\n\nEntertainment\n\nComputer Services\n\nEditing/Publishing\n\nData graphed after applying data visualization rules:\n\nSalaries\n\nand\n\nBenefits 60.3%\n\nContracts and\n\nProfessional Fees\n\n10\\.1%\n\nDepreciation, Interest\n\nand Maintenance\n\n9%\n\nOffice Expenses and Other Costs\\*\n\n8%\n\nConference, Meetings\n\nand Exhibits\n\n5%\n\nTravel and Business Entertainment\n\n3\\.5%\n\nComputer Services\n\n2\\.7% Editing/Publishing\n\n1\\.4%\n\n\\*Other costs category includes marketing and promotions costs.\n\n\\*Other costs category includes marketing and promotions costs.\n\nFY14 Jan Actual FY14 Q2 Forecast\n\nMultiple benefits were achieved by applying data visualization rules: a. Two graphs have been combined into one allowing for\n\na. Easy comparison between data sets (FY14 Q2 Forecast\n\nto Actual)\n\nb. Additional real estate on documents with limited space\n\nb. Different type of display media was used to more clearly\n\nshow the relationship between the expense types\n\nc. Color was used sparingly to mark forecasted expense percentages drawing attention to the most important\n\naspects of the graph\n\nd. By using a different display media there is more data flexibility. For example, another dimension could be added to the bar graph such as FY14 budget data etc.","publish_date":null,"title":"Pie charts","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/FPA-guide-to-data-visualization_p20.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/FPA-guide-to-data-visualization_p20.pdf"},"request":{"example_id":"chart/FPA-guide-to-data-visualization_p20","product_type":"parse","source_file_path":"FPA-guide-to-data-visualization_p20.pdf"},"started_at":"2026-11-04T19:45:14.366476"} +{"completed_at":"2026-11-04T19:45:18.304783","latency_in_ms":2410,"output":{"example_id":"chart/FPA-guide-to-data-visualization_p5","grounded_pages":[],"layout_pages":[],"markdown":"AFP GUIDE: **Data Visualization: How to Tell Number Stories with Pictures**\n\n## Introduction: The Trend\n\n_Survey shows more finance professionals are looking to data visualization to help solve problems, identify patterns and anomalies, and understand what’s driving their organizational performance._\n\nThe trend toward greater use of visualization among\n\nFP&A professionals is still growing. “A lot of them are still focused on numbers,” said Craig Schiff, president and CEO of BPM Partners, a vendor-neutral advi- sory services firm specializing in business performance management (BPM) and business intelligence solutions. However, the data from the 2014 BPM Pulse Survey of finance professionals shows the desire for change is strong. Asked if they require data visualization beyond typical charts or graphs offered in Excel, 40 percent of the finance professionals surveyed responded that they do. “That’s a large and growing number,” Schiff noted.\n\nMore interactive features scored highest (71 percent)\n\nwhen asked what kind of data visualization capability respondents would like to have. “They really want to understand what’s going on behind the scenes. Graphics are great. The important thing is that not all the visu- alization tools provide the ability to drill back into the detail to broaden your own understanding of the graph that brought the issue to your attention.” Fifty percent of respondents also wanted more chart types and tools in order to present the data with specialized graphics.\n\nThe last question the survey asked about data visu-\n\nalization is from what source respondents would prefer to acquire these tools. The choices were: _third-party providers, included in their performance management solution, or doesn’t matter._ According to the results, 71 percent preferred that the data visualization capabilities be integrated into the core product. “One of the chal- lenges of the third-party tools is how to connect them to the source data,” Schiff said. That integration is already built into the budgeting tools. Those systems also speak the financial language and are designed for business end-user self-sufficiency. Many of the standalone tools require more IT knowledge to connect them to the data sources. “We believe that data visualization is a grow- ing trend, and there is still a lot of opportunity for the vendors to add value.”\n\n_Source: BPM Partners 2014 BPM Pulse Survey_\n\nDo you require data visualization capabilities beyond standard charts and graphs?\n\nYes\n\nNo\n\nNot sure\n\n70\\.6%\n\n56\\.5%\n\n43\\.5%\n\n21\\.2%\n\n7\\.1%\n\nMore interactive\n\nMore chart types\n\n(heat chart, fractals,\n\ntree map, bubble\n\nchart, etc.)\n\nAccess to more\n\ndata sources\n\n(external)\n\nNot sure\n\nAccess to more\n\ndata sources\n\n(internal)\n\n0% 20% 40% 60% 80%\n\nWhat data visualization capabilities do you need that you are not getting today? (select all that apply)\n\nWhere would you prefer to get these data visualization capabilities from?\n\n15\\.3%\n\n72\\.9%\n\n11\\.8%\n\nIncluded in performance management solution\n\nDoesn’t matter\n\nThird-party solution that may access my performance management data\n\n24\\.0%\n\n35\\.1%\n\n40\\.9%","pages":[{"line_numbers":[],"markdown":"AFP GUIDE: **Data Visualization: How to Tell Number Stories with Pictures**\n\n## Introduction: The Trend\n\n_Survey shows more finance professionals are looking to data visualization to help solve problems, identify patterns and anomalies, and understand what’s driving their organizational performance._\n\nThe trend toward greater use of visualization among\n\nFP&A professionals is still growing. “A lot of them are still focused on numbers,” said Craig Schiff, president and CEO of BPM Partners, a vendor-neutral advi- sory services firm specializing in business performance management (BPM) and business intelligence solutions. However, the data from the 2014 BPM Pulse Survey of finance professionals shows the desire for change is strong. Asked if they require data visualization beyond typical charts or graphs offered in Excel, 40 percent of the finance professionals surveyed responded that they do. “That’s a large and growing number,” Schiff noted.\n\nMore interactive features scored highest (71 percent)\n\nwhen asked what kind of data visualization capability respondents would like to have. “They really want to understand what’s going on behind the scenes. Graphics are great. The important thing is that not all the visu- alization tools provide the ability to drill back into the detail to broaden your own understanding of the graph that brought the issue to your attention.” Fifty percent of respondents also wanted more chart types and tools in order to present the data with specialized graphics.\n\nThe last question the survey asked about data visu-\n\nalization is from what source respondents would prefer to acquire these tools. The choices were: _third-party providers, included in their performance management solution, or doesn’t matter._ According to the results, 71 percent preferred that the data visualization capabilities be integrated into the core product. “One of the chal- lenges of the third-party tools is how to connect them to the source data,” Schiff said. That integration is already built into the budgeting tools. Those systems also speak the financial language and are designed for business end-user self-sufficiency. Many of the standalone tools require more IT knowledge to connect them to the data sources. “We believe that data visualization is a grow- ing trend, and there is still a lot of opportunity for the vendors to add value.”\n\n_Source: BPM Partners 2014 BPM Pulse Survey_\n\nDo you require data visualization capabilities beyond standard charts and graphs?\n\nYes\n\nNo\n\nNot sure\n\n70\\.6%\n\n56\\.5%\n\n43\\.5%\n\n21\\.2%\n\n7\\.1%\n\nMore interactive\n\nMore chart types\n\n(heat chart, fractals,\n\ntree map, bubble\n\nchart, etc.)\n\nAccess to more\n\ndata sources\n\n(external)\n\nNot sure\n\nAccess to more\n\ndata sources\n\n(internal)\n\n0% 20% 40% 60% 80%\n\nWhat data visualization capabilities do you need that you are not getting today? (select all that apply)\n\nWhere would you prefer to get these data visualization capabilities from?\n\n15\\.3%\n\n72\\.9%\n\n11\\.8%\n\nIncluded in performance management solution\n\nDoesn’t matter\n\nThird-party solution that may access my performance management data\n\n24\\.0%\n\n35\\.1%\n\n40\\.9%","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"AFP GUIDE: **Data Visualization: How to Tell Number Stories with Pictures**\n\n## Introduction: The Trend\n\n_Survey shows more finance professionals are looking to data visualization to help solve problems, identify patterns and anomalies, and understand what’s driving their organizational performance._\n\nThe trend toward greater use of visualization among\n\nFP&A professionals is still growing. “A lot of them are still focused on numbers,” said Craig Schiff, president and CEO of BPM Partners, a vendor-neutral advi- sory services firm specializing in business performance management (BPM) and business intelligence solutions. However, the data from the 2014 BPM Pulse Survey of finance professionals shows the desire for change is strong. Asked if they require data visualization beyond typical charts or graphs offered in Excel, 40 percent of the finance professionals surveyed responded that they do. “That’s a large and growing number,” Schiff noted.\n\nMore interactive features scored highest (71 percent)\n\nwhen asked what kind of data visualization capability respondents would like to have. “They really want to understand what’s going on behind the scenes. Graphics are great. The important thing is that not all the visu- alization tools provide the ability to drill back into the detail to broaden your own understanding of the graph that brought the issue to your attention.” Fifty percent of respondents also wanted more chart types and tools in order to present the data with specialized graphics.\n\nThe last question the survey asked about data visu-\n\nalization is from what source respondents would prefer to acquire these tools. The choices were: _third-party providers, included in their performance management solution, or doesn’t matter._ According to the results, 71 percent preferred that the data visualization capabilities be integrated into the core product. “One of the chal- lenges of the third-party tools is how to connect them to the source data,” Schiff said. That integration is already built into the budgeting tools. Those systems also speak the financial language and are designed for business end-user self-sufficiency. Many of the standalone tools require more IT knowledge to connect them to the data sources. “We believe that data visualization is a grow- ing trend, and there is still a lot of opportunity for the vendors to add value.”\n\n_Source: BPM Partners 2014 BPM Pulse Survey_\n\nDo you require data visualization capabilities beyond standard charts and graphs?\n\nYes\n\nNo\n\nNot sure\n\n70\\.6%\n\n56\\.5%\n\n43\\.5%\n\n21\\.2%\n\n7\\.1%\n\nMore interactive\n\nMore chart types\n\n(heat chart, fractals,\n\ntree map, bubble\n\nchart, etc.)\n\nAccess to more\n\ndata sources\n\n(external)\n\nNot sure\n\nAccess to more\n\ndata sources\n\n(internal)\n\n0% 20% 40% 60% 80%\n\nWhat data visualization capabilities do you need that you are not getting today? (select all that apply)\n\nWhere would you prefer to get these data visualization capabilities from?\n\n15\\.3%\n\n72\\.9%\n\n11\\.8%\n\nIncluded in performance management solution\n\nDoesn’t matter\n\nThird-party solution that may access my performance management data\n\n24\\.0%\n\n35\\.1%\n\n40\\.9%","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/FPA-guide-to-data-visualization_p5.pdf"]},"response":{"errors":[],"extract_id":"extract_43e4995fbde5f488dc952da40bda971b","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["AFP GUIDE: **Data Visualization: How to Tell Number Stories with Pictures**\n\n## Introduction: The Trend\n\n_Survey shows more finance professionals are looking to data visualization to help solve problems, identify patterns and anomalies, and understand what’s driving their organizational performance._\n\nThe trend toward greater use of visualization among\n\nFP&A professionals is still growing. “A lot of them are still focused on numbers,” said Craig Schiff, president and CEO of BPM Partners, a vendor-neutral advi- sory services firm specializing in business performance management (BPM) and business intelligence solutions. However, the data from the 2014 BPM Pulse Survey of finance professionals shows the desire for change is strong. Asked if they require data visualization beyond typical charts or graphs offered in Excel, 40 percent of the finance professionals surveyed responded that they do. “That’s a large and growing number,” Schiff noted.\n\nMore interactive features scored highest (71 percent)\n\nwhen asked what kind of data visualization capability respondents would like to have. “They really want to understand what’s going on behind the scenes. Graphics are great. The important thing is that not all the visu- alization tools provide the ability to drill back into the detail to broaden your own understanding of the graph that brought the issue to your attention.” Fifty percent of respondents also wanted more chart types and tools in order to present the data with specialized graphics.\n\nThe last question the survey asked about data visu-\n\nalization is from what source respondents would prefer to acquire these tools. The choices were: _third-party providers, included in their performance management solution, or doesn’t matter._ According to the results, 71 percent preferred that the data visualization capabilities be integrated into the core product. “One of the chal- lenges of the third-party tools is how to connect them to the source data,” Schiff said. That integration is already built into the budgeting tools. Those systems also speak the financial language and are designed for business end-user self-sufficiency. Many of the standalone tools require more IT knowledge to connect them to the data sources. “We believe that data visualization is a grow- ing trend, and there is still a lot of opportunity for the vendors to add value.”\n\n_Source: BPM Partners 2014 BPM Pulse Survey_\n\nDo you require data visualization capabilities beyond standard charts and graphs?\n\nYes\n\nNo\n\nNot sure\n\n70\\.6%\n\n56\\.5%\n\n43\\.5%\n\n21\\.2%\n\n7\\.1%\n\nMore interactive\n\nMore chart types\n\n(heat chart, fractals,\n\ntree map, bubble\n\nchart, etc.)\n\nAccess to more\n\ndata sources\n\n(external)\n\nNot sure\n\nAccess to more\n\ndata sources\n\n(internal)\n\n0% 20% 40% 60% 80%\n\nWhat data visualization capabilities do you need that you are not getting today? (select all that apply)\n\nWhere would you prefer to get these data visualization capabilities from?\n\n15\\.3%\n\n72\\.9%\n\n11\\.8%\n\nIncluded in performance management solution\n\nDoesn’t matter\n\nThird-party solution that may access my performance management data\n\n24\\.0%\n\n35\\.1%\n\n40\\.9%"],"full_content":"AFP GUIDE: **Data Visualization: How to Tell Number Stories with Pictures**\n\n## Introduction: The Trend\n\n_Survey shows more finance professionals are looking to data visualization to help solve problems, identify patterns and anomalies, and understand what’s driving their organizational performance._\n\nThe trend toward greater use of visualization among\n\nFP&A professionals is still growing. “A lot of them are still focused on numbers,” said Craig Schiff, president and CEO of BPM Partners, a vendor-neutral advi- sory services firm specializing in business performance management (BPM) and business intelligence solutions. However, the data from the 2014 BPM Pulse Survey of finance professionals shows the desire for change is strong. Asked if they require data visualization beyond typical charts or graphs offered in Excel, 40 percent of the finance professionals surveyed responded that they do. “That’s a large and growing number,” Schiff noted.\n\nMore interactive features scored highest (71 percent)\n\nwhen asked what kind of data visualization capability respondents would like to have. “They really want to understand what’s going on behind the scenes. Graphics are great. The important thing is that not all the visu- alization tools provide the ability to drill back into the detail to broaden your own understanding of the graph that brought the issue to your attention.” Fifty percent of respondents also wanted more chart types and tools in order to present the data with specialized graphics.\n\nThe last question the survey asked about data visu-\n\nalization is from what source respondents would prefer to acquire these tools. The choices were: _third-party providers, included in their performance management solution, or doesn’t matter._ According to the results, 71 percent preferred that the data visualization capabilities be integrated into the core product. “One of the chal- lenges of the third-party tools is how to connect them to the source data,” Schiff said. That integration is already built into the budgeting tools. Those systems also speak the financial language and are designed for business end-user self-sufficiency. Many of the standalone tools require more IT knowledge to connect them to the data sources. “We believe that data visualization is a grow- ing trend, and there is still a lot of opportunity for the vendors to add value.”\n\n_Source: BPM Partners 2014 BPM Pulse Survey_\n\nDo you require data visualization capabilities beyond standard charts and graphs?\n\nYes\n\nNo\n\nNot sure\n\n70\\.6%\n\n56\\.5%\n\n43\\.5%\n\n21\\.2%\n\n7\\.1%\n\nMore interactive\n\nMore chart types\n\n(heat chart, fractals,\n\ntree map, bubble\n\nchart, etc.)\n\nAccess to more\n\ndata sources\n\n(external)\n\nNot sure\n\nAccess to more\n\ndata sources\n\n(internal)\n\n0% 20% 40% 60% 80%\n\nWhat data visualization capabilities do you need that you are not getting today? (select all that apply)\n\nWhere would you prefer to get these data visualization capabilities from?\n\n15\\.3%\n\n72\\.9%\n\n11\\.8%\n\nIncluded in performance management solution\n\nDoesn’t matter\n\nThird-party solution that may access my performance management data\n\n24\\.0%\n\n35\\.1%\n\n40\\.9%","publish_date":null,"title":"Introduction: The Trend","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/FPA-guide-to-data-visualization_p5.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/FPA-guide-to-data-visualization_p5.pdf"},"request":{"example_id":"chart/FPA-guide-to-data-visualization_p5","product_type":"parse","source_file_path":"FPA-guide-to-data-visualization_p5.pdf"},"started_at":"2026-11-04T19:45:15.893974"} +{"completed_at":"2026-11-04T19:45:19.107728","latency_in_ms":2474,"output":{"example_id":"chart/FPA-guide-to-data-visualization_p8","grounded_pages":[],"layout_pages":[],"markdown":"AFP GUIDE: **Data Visualization: How to Tell Number Stories with Pictures**\n\nthem to be handled, and what is their geographical loca- tion? “You have to be able to answer the broader ques- tions so people can discover more about the problem,” he said. “It’s a sort of partnership with the user.”\n\n“It’s also important to try to tie in predictive analyt-\n\nics,” continued Wills. “If something happens, what is the total amount of money you’re predicted to lose, and should you be concerned about the average amount or the big amounts that are the outliers? That’s the data you choose to highlight and the context for the analytics. The measure of value or success of a visualization project is whether it helped answer the questions.”\n\nIn order to be able to tailor the image to the audience,\n\n“You have to be asking the right questions,” said Amanda Pype, user researcher at Tableau Software, a leading visualiza- tion software provider. “What do they want to know? The most important thing is to understand the audience and what they’re trying to get out of the data.” Also, use charts that they can easily understand, she advised. “You don’t want to spend 10 minutes explaining the chart.”\n\nOne key concern is the reliability of the forecast,\n\naccording to Andrew Mosely, director at Metapraxis, a London-based visualization vendor. “Across all indus-\n\ntries, the most common business question is: how cred- ible are the forecasts of our business units?’’ said Mosely. “And when a forecast appears either too optimistic or overly pessimistic, how can our FP&A team present the data visually to support their dialogue with the business? Every business needs to be on top of its numbers. The challenge is finding out as soon as possible if you’re not going to make that forecast rather than being presented with an impossible last-minute challenge. Having time to take action puts you back in control.”\n\nMetapraxis’ software contains a calculation engine that\n\ncreates statistical scenarios, which are then presented vi- sually to show trends and variances. Alternative scenarios are automatically created, against which you can test the management forecast, according to Mosely. “We call this visualization the ‘Windsock’ chart” (see below).\n\nThe “Windsock” chart tracks the evolution of the\n\nsubsidiary’s updated year-end forecasts throughout the fi- nancial year, comparing them against the annual budget, prior year performance and a statistical risk range. The chart shows that the current management forecast for Net Sales is at the very top of the current risk range and may be too optimistic.\n\nWestern Europe Net Sales Windsock","pages":[{"line_numbers":[],"markdown":"AFP GUIDE: **Data Visualization: How to Tell Number Stories with Pictures**\n\nthem to be handled, and what is their geographical loca- tion? “You have to be able to answer the broader ques- tions so people can discover more about the problem,” he said. “It’s a sort of partnership with the user.”\n\n“It’s also important to try to tie in predictive analyt-\n\nics,” continued Wills. “If something happens, what is the total amount of money you’re predicted to lose, and should you be concerned about the average amount or the big amounts that are the outliers? That’s the data you choose to highlight and the context for the analytics. The measure of value or success of a visualization project is whether it helped answer the questions.”\n\nIn order to be able to tailor the image to the audience,\n\n“You have to be asking the right questions,” said Amanda Pype, user researcher at Tableau Software, a leading visualiza- tion software provider. “What do they want to know? The most important thing is to understand the audience and what they’re trying to get out of the data.” Also, use charts that they can easily understand, she advised. “You don’t want to spend 10 minutes explaining the chart.”\n\nOne key concern is the reliability of the forecast,\n\naccording to Andrew Mosely, director at Metapraxis, a London-based visualization vendor. “Across all indus-\n\ntries, the most common business question is: how cred- ible are the forecasts of our business units?’’ said Mosely. “And when a forecast appears either too optimistic or overly pessimistic, how can our FP&A team present the data visually to support their dialogue with the business? Every business needs to be on top of its numbers. The challenge is finding out as soon as possible if you’re not going to make that forecast rather than being presented with an impossible last-minute challenge. Having time to take action puts you back in control.”\n\nMetapraxis’ software contains a calculation engine that\n\ncreates statistical scenarios, which are then presented vi- sually to show trends and variances. Alternative scenarios are automatically created, against which you can test the management forecast, according to Mosely. “We call this visualization the ‘Windsock’ chart” (see below).\n\nThe “Windsock” chart tracks the evolution of the\n\nsubsidiary’s updated year-end forecasts throughout the fi- nancial year, comparing them against the annual budget, prior year performance and a statistical risk range. The chart shows that the current management forecast for Net Sales is at the very top of the current risk range and may be too optimistic.\n\nWestern Europe Net Sales Windsock","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"AFP GUIDE: **Data Visualization: How to Tell Number Stories with Pictures**\n\nthem to be handled, and what is their geographical loca- tion? “You have to be able to answer the broader ques- tions so people can discover more about the problem,” he said. “It’s a sort of partnership with the user.”\n\n“It’s also important to try to tie in predictive analyt-\n\nics,” continued Wills. “If something happens, what is the total amount of money you’re predicted to lose, and should you be concerned about the average amount or the big amounts that are the outliers? That’s the data you choose to highlight and the context for the analytics. The measure of value or success of a visualization project is whether it helped answer the questions.”\n\nIn order to be able to tailor the image to the audience,\n\n“You have to be asking the right questions,” said Amanda Pype, user researcher at Tableau Software, a leading visualiza- tion software provider. “What do they want to know? The most important thing is to understand the audience and what they’re trying to get out of the data.” Also, use charts that they can easily understand, she advised. “You don’t want to spend 10 minutes explaining the chart.”\n\nOne key concern is the reliability of the forecast,\n\naccording to Andrew Mosely, director at Metapraxis, a London-based visualization vendor. “Across all indus-\n\ntries, the most common business question is: how cred- ible are the forecasts of our business units?’’ said Mosely. “And when a forecast appears either too optimistic or overly pessimistic, how can our FP&A team present the data visually to support their dialogue with the business? Every business needs to be on top of its numbers. The challenge is finding out as soon as possible if you’re not going to make that forecast rather than being presented with an impossible last-minute challenge. Having time to take action puts you back in control.”\n\nMetapraxis’ software contains a calculation engine that\n\ncreates statistical scenarios, which are then presented vi- sually to show trends and variances. Alternative scenarios are automatically created, against which you can test the management forecast, according to Mosely. “We call this visualization the ‘Windsock’ chart” (see below).\n\nThe “Windsock” chart tracks the evolution of the\n\nsubsidiary’s updated year-end forecasts throughout the fi- nancial year, comparing them against the annual budget, prior year performance and a statistical risk range. The chart shows that the current management forecast for Net Sales is at the very top of the current risk range and may be too optimistic.\n\nWestern Europe Net Sales Windsock","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/FPA-guide-to-data-visualization_p8.pdf"]},"response":{"errors":[],"extract_id":"extract_6c4e4a9e79117394ca4e0c95da4c4f7e","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["AFP GUIDE: **Data Visualization: How to Tell Number Stories with Pictures**\n\nthem to be handled, and what is their geographical loca- tion? “You have to be able to answer the broader ques- tions so people can discover more about the problem,” he said. “It’s a sort of partnership with the user.”\n\n“It’s also important to try to tie in predictive analyt-\n\nics,” continued Wills. “If something happens, what is the total amount of money you’re predicted to lose, and should you be concerned about the average amount or the big amounts that are the outliers? That’s the data you choose to highlight and the context for the analytics. The measure of value or success of a visualization project is whether it helped answer the questions.”\n\nIn order to be able to tailor the image to the audience,\n\n“You have to be asking the right questions,” said Amanda Pype, user researcher at Tableau Software, a leading visualiza- tion software provider. “What do they want to know? The most important thing is to understand the audience and what they’re trying to get out of the data.” Also, use charts that they can easily understand, she advised. “You don’t want to spend 10 minutes explaining the chart.”\n\nOne key concern is the reliability of the forecast,\n\naccording to Andrew Mosely, director at Metapraxis, a London-based visualization vendor. “Across all indus-\n\ntries, the most common business question is: how cred- ible are the forecasts of our business units?’’ said Mosely. “And when a forecast appears either too optimistic or overly pessimistic, how can our FP&A team present the data visually to support their dialogue with the business? Every business needs to be on top of its numbers. The challenge is finding out as soon as possible if you’re not going to make that forecast rather than being presented with an impossible last-minute challenge. Having time to take action puts you back in control.”\n\nMetapraxis’ software contains a calculation engine that\n\ncreates statistical scenarios, which are then presented vi- sually to show trends and variances. Alternative scenarios are automatically created, against which you can test the management forecast, according to Mosely. “We call this visualization the ‘Windsock’ chart” (see below).\n\nThe “Windsock” chart tracks the evolution of the\n\nsubsidiary’s updated year-end forecasts throughout the fi- nancial year, comparing them against the annual budget, prior year performance and a statistical risk range. The chart shows that the current management forecast for Net Sales is at the very top of the current risk range and may be too optimistic.\n\nWestern Europe Net Sales Windsock"],"full_content":"AFP GUIDE: **Data Visualization: How to Tell Number Stories with Pictures**\n\nthem to be handled, and what is their geographical loca- tion? “You have to be able to answer the broader ques- tions so people can discover more about the problem,” he said. “It’s a sort of partnership with the user.”\n\n“It’s also important to try to tie in predictive analyt-\n\nics,” continued Wills. “If something happens, what is the total amount of money you’re predicted to lose, and should you be concerned about the average amount or the big amounts that are the outliers? That’s the data you choose to highlight and the context for the analytics. The measure of value or success of a visualization project is whether it helped answer the questions.”\n\nIn order to be able to tailor the image to the audience,\n\n“You have to be asking the right questions,” said Amanda Pype, user researcher at Tableau Software, a leading visualiza- tion software provider. “What do they want to know? The most important thing is to understand the audience and what they’re trying to get out of the data.” Also, use charts that they can easily understand, she advised. “You don’t want to spend 10 minutes explaining the chart.”\n\nOne key concern is the reliability of the forecast,\n\naccording to Andrew Mosely, director at Metapraxis, a London-based visualization vendor. “Across all indus-\n\ntries, the most common business question is: how cred- ible are the forecasts of our business units?’’ said Mosely. “And when a forecast appears either too optimistic or overly pessimistic, how can our FP&A team present the data visually to support their dialogue with the business? Every business needs to be on top of its numbers. The challenge is finding out as soon as possible if you’re not going to make that forecast rather than being presented with an impossible last-minute challenge. Having time to take action puts you back in control.”\n\nMetapraxis’ software contains a calculation engine that\n\ncreates statistical scenarios, which are then presented vi- sually to show trends and variances. Alternative scenarios are automatically created, against which you can test the management forecast, according to Mosely. “We call this visualization the ‘Windsock’ chart” (see below).\n\nThe “Windsock” chart tracks the evolution of the\n\nsubsidiary’s updated year-end forecasts throughout the fi- nancial year, comparing them against the annual budget, prior year performance and a statistical risk range. The chart shows that the current management forecast for Net Sales is at the very top of the current risk range and may be too optimistic.\n\nWestern Europe Net Sales Windsock","publish_date":null,"title":"them to be handled, and what is their geographical loca- tries, the most common business question is: how cred- tion?","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/FPA-guide-to-data-visualization_p8.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/FPA-guide-to-data-visualization_p8.pdf"},"request":{"example_id":"chart/FPA-guide-to-data-visualization_p8","product_type":"parse","source_file_path":"FPA-guide-to-data-visualization_p8.pdf"},"started_at":"2026-11-04T19:45:16.632845"} +{"completed_at":"2026-11-04T19:45:20.100797","latency_in_ms":1325,"output":{"example_id":"chart/Feaco-Survey-2023-2024_p7","grounded_pages":[],"layout_pages":[],"markdown":"### 7\n\n_Source: Our elaborations on MC turnover split by service lines – feaco survey 2021-2023._ \\* The European panel includes only 9 European countries of the feaco survey as detailed data for Spain and Slovenia are missing. These 9 countries represent 69% of the European GDP. Data from Ukraine have not been included in the European panel.\n\n## MC turnover market share by Service Lines (2021-23)\n\nIn 2023, in the European panel, Technology was the service line with the highest turnover (23.5% market share), followed by Strategy (20.9%) and Operations (20.3%). In the past three years, the market share of Technology has remained rather stable, while Strategy has increased to a relevant extent. People & Change and, to a minor extent, Sales & Marketing have experienced significant ups and downs in their market shares.\n\nSERVICE LINE\\*\\* European panel market share in 2023\\* European panel market share in 2022\\* European panel market share in 2021\\*\n\nTechnology 23.5% 23.1% 23.7%\n\nStrategy 20.9% 20.5% 19.9%\n\nOperations 20.3% 20.2% 20.9%\n\nPeople & Change 12.8% 16.1% 13.8%\n\nFinance & Risk 10.0% 9.7% 10.4%\n\nSales & Marketing 4.0% 3.7% 4.5%\n\nOther 8.5% 6.7% 6.8%\n\nTechnology 23.5%\n\nStrategy 20.9%\n\nOperations 20.3%\n\nPeople & Change 12.8%\n\nFinance & Risk 10%\n\nSales & Marketing 4%\n\nOther 8.5%\n\n### MC turnover market share by Service Lines (2021-23)","pages":[{"line_numbers":[],"markdown":"### 7\n\n_Source: Our elaborations on MC turnover split by service lines – feaco survey 2021-2023._ \\* The European panel includes only 9 European countries of the feaco survey as detailed data for Spain and Slovenia are missing. These 9 countries represent 69% of the European GDP. Data from Ukraine have not been included in the European panel.\n\n## MC turnover market share by Service Lines (2021-23)\n\nIn 2023, in the European panel, Technology was the service line with the highest turnover (23.5% market share), followed by Strategy (20.9%) and Operations (20.3%). In the past three years, the market share of Technology has remained rather stable, while Strategy has increased to a relevant extent. People & Change and, to a minor extent, Sales & Marketing have experienced significant ups and downs in their market shares.\n\nSERVICE LINE\\*\\* European panel market share in 2023\\* European panel market share in 2022\\* European panel market share in 2021\\*\n\nTechnology 23.5% 23.1% 23.7%\n\nStrategy 20.9% 20.5% 19.9%\n\nOperations 20.3% 20.2% 20.9%\n\nPeople & Change 12.8% 16.1% 13.8%\n\nFinance & Risk 10.0% 9.7% 10.4%\n\nSales & Marketing 4.0% 3.7% 4.5%\n\nOther 8.5% 6.7% 6.8%\n\nTechnology 23.5%\n\nStrategy 20.9%\n\nOperations 20.3%\n\nPeople & Change 12.8%\n\nFinance & Risk 10%\n\nSales & Marketing 4%\n\nOther 8.5%\n\n### MC turnover market share by Service Lines (2021-23)","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"### 7\n\n_Source: Our elaborations on MC turnover split by service lines – feaco survey 2021-2023._ \\* The European panel includes only 9 European countries of the feaco survey as detailed data for Spain and Slovenia are missing. These 9 countries represent 69% of the European GDP. Data from Ukraine have not been included in the European panel.\n\n## MC turnover market share by Service Lines (2021-23)\n\nIn 2023, in the European panel, Technology was the service line with the highest turnover (23.5% market share), followed by Strategy (20.9%) and Operations (20.3%). In the past three years, the market share of Technology has remained rather stable, while Strategy has increased to a relevant extent. People & Change and, to a minor extent, Sales & Marketing have experienced significant ups and downs in their market shares.\n\nSERVICE LINE\\*\\* European panel market share in 2023\\* European panel market share in 2022\\* European panel market share in 2021\\*\n\nTechnology 23.5% 23.1% 23.7%\n\nStrategy 20.9% 20.5% 19.9%\n\nOperations 20.3% 20.2% 20.9%\n\nPeople & Change 12.8% 16.1% 13.8%\n\nFinance & Risk 10.0% 9.7% 10.4%\n\nSales & Marketing 4.0% 3.7% 4.5%\n\nOther 8.5% 6.7% 6.8%\n\nTechnology 23.5%\n\nStrategy 20.9%\n\nOperations 20.3%\n\nPeople & Change 12.8%\n\nFinance & Risk 10%\n\nSales & Marketing 4%\n\nOther 8.5%\n\n### MC turnover market share by Service Lines (2021-23)","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Feaco-Survey-2023-2024_p7.pdf"]},"response":{"errors":[],"extract_id":"extract_3a087392c0cf4c2f6d1671ef43bef926","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["### 7\n\n_Source: Our elaborations on MC turnover split by service lines – feaco survey 2021-2023._ \\* The European panel includes only 9 European countries of the feaco survey as detailed data for Spain and Slovenia are missing. These 9 countries represent 69% of the European GDP. Data from Ukraine have not been included in the European panel.\n\n## MC turnover market share by Service Lines (2021-23)\n\nIn 2023, in the European panel, Technology was the service line with the highest turnover (23.5% market share), followed by Strategy (20.9%) and Operations (20.3%). In the past three years, the market share of Technology has remained rather stable, while Strategy has increased to a relevant extent. People & Change and, to a minor extent, Sales & Marketing have experienced significant ups and downs in their market shares.\n\nSERVICE LINE\\*\\* European panel market share in 2023\\* European panel market share in 2022\\* European panel market share in 2021\\*\n\nTechnology 23.5% 23.1% 23.7%\n\nStrategy 20.9% 20.5% 19.9%\n\nOperations 20.3% 20.2% 20.9%\n\nPeople & Change 12.8% 16.1% 13.8%\n\nFinance & Risk 10.0% 9.7% 10.4%\n\nSales & Marketing 4.0% 3.7% 4.5%\n\nOther 8.5% 6.7% 6.8%\n\nTechnology 23.5%\n\nStrategy 20.9%\n\nOperations 20.3%\n\nPeople & Change 12.8%\n\nFinance & Risk 10%\n\nSales & Marketing 4%\n\nOther 8.5%\n\n### MC turnover market share by Service Lines (2021-23)"],"full_content":"### 7\n\n_Source: Our elaborations on MC turnover split by service lines – feaco survey 2021-2023._ \\* The European panel includes only 9 European countries of the feaco survey as detailed data for Spain and Slovenia are missing. These 9 countries represent 69% of the European GDP. Data from Ukraine have not been included in the European panel.\n\n## MC turnover market share by Service Lines (2021-23)\n\nIn 2023, in the European panel, Technology was the service line with the highest turnover (23.5% market share), followed by Strategy (20.9%) and Operations (20.3%). In the past three years, the market share of Technology has remained rather stable, while Strategy has increased to a relevant extent. People & Change and, to a minor extent, Sales & Marketing have experienced significant ups and downs in their market shares.\n\nSERVICE LINE\\*\\* European panel market share in 2023\\* European panel market share in 2022\\* European panel market share in 2021\\*\n\nTechnology 23.5% 23.1% 23.7%\n\nStrategy 20.9% 20.5% 19.9%\n\nOperations 20.3% 20.2% 20.9%\n\nPeople & Change 12.8% 16.1% 13.8%\n\nFinance & Risk 10.0% 9.7% 10.4%\n\nSales & Marketing 4.0% 3.7% 4.5%\n\nOther 8.5% 6.7% 6.8%\n\nTechnology 23.5%\n\nStrategy 20.9%\n\nOperations 20.3%\n\nPeople & Change 12.8%\n\nFinance & Risk 10%\n\nSales & Marketing 4%\n\nOther 8.5%\n\n### MC turnover market share by Service Lines (2021-23)","publish_date":null,"title":"MC turnover market share by Service Lines (2021-23)","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Feaco-Survey-2023-2024_p7.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Feaco-Survey-2023-2024_p7.pdf"},"request":{"example_id":"chart/Feaco-Survey-2023-2024_p7","product_type":"parse","source_file_path":"Feaco-Survey-2023-2024_p7.pdf"},"started_at":"2026-11-04T19:45:18.775341"} +{"completed_at":"2026-11-04T19:45:22.569714","latency_in_ms":3051,"output":{"example_id":"chart/Feaco-Survey-2023-2024_p8","grounded_pages":[],"layout_pages":[],"markdown":"### 8\n\n## MC turnover trend by Service Lines (2021-23)\n\nIn 2023, in the European panel, Sales & Marketing (+20.7%) and Technology (+15.6%) have been the two fastest growing service lines. Sales & Marketing, after a top growth in 2021 and a relevant slow down in 2022, is again the fastest growing service line. People & Change, after a top growth in 2022, is experiencing a major decline (-11.5%) in spite of an overall MC market that is growing double digits. Technology, Strategy and Operations are experiencing a more stable growth path, with highlighting that many clients are leveraging on technology – namely digital technologies – to redefine their post-pandemic strategies and operations. Finance & Risk, after a record growth in 2020 and a decline in 2021, is also back to a more stable growth trend in 2022 and 2023.\n\nSales & Marketing\n\nTechnology\n\nFinance & Risk\n\nStrategy\n\nOperations\n\nPeople & Change\n\n\\-5.0% 0.0% 5.0% 10.0% 15.0% 20.0% 25.0% 30.0% 35.0%\n\n### MC turnover trend by Service Lines (European panel 2021-23)\n\n11\\.2% 9.3%\n\n11\\.0%\n\n11\\.1% 10.6%\n\n15\\.6%\n\n22\\.0%\n\n10\\.9%\n\n13\\.5%\n\n6\\.1% 20.7%\n\n30\\.6%\n\n15\\.1%\n\n13\\.1%\n\n\\-1.2%\n\n30\\.3%\n\n10\\.8%\n\n\\-11.5%\n\nSERVICE LINE\\*\\* European panel trend in 2023\\* European panel trend in 2022\\* European panel trend in 2021\\*\n\nSales & Marketing 20.7% 6.1% 30.6%\n\nTechnology 15.6% 10.6% 11.1%\n\nFinance & Risk 15.1% 13.1% -1.2%\n\nStrategy 13.5% 22.0% 10.9%\n\nOperations 11.0% 9.3% 11.2%\n\nPeople & Change -11.5% 30.3% 10.8%\n\n2021 2022 2023\n\n_Source: Our elaborations on MC turnover split by service lines – feaco survey 2021-2023._ \\* The European panel includes only 9 European countries of the feaco survey as detailed data for Spain and Slovenia are missing. These 9 countries represent 69% of the European GDP. Data from Ukraine have not been included in the European panel. \\*\\* The “Other” service line is not reported in the above table as it is not a homogenous category, and its trend would not convey a clear information on the actual market dynamics.\n\n\\-10.0%","pages":[{"line_numbers":[],"markdown":"### 8\n\n## MC turnover trend by Service Lines (2021-23)\n\nIn 2023, in the European panel, Sales & Marketing (+20.7%) and Technology (+15.6%) have been the two fastest growing service lines. Sales & Marketing, after a top growth in 2021 and a relevant slow down in 2022, is again the fastest growing service line. People & Change, after a top growth in 2022, is experiencing a major decline (-11.5%) in spite of an overall MC market that is growing double digits. Technology, Strategy and Operations are experiencing a more stable growth path, with highlighting that many clients are leveraging on technology – namely digital technologies – to redefine their post-pandemic strategies and operations. Finance & Risk, after a record growth in 2020 and a decline in 2021, is also back to a more stable growth trend in 2022 and 2023.\n\nSales & Marketing\n\nTechnology\n\nFinance & Risk\n\nStrategy\n\nOperations\n\nPeople & Change\n\n\\-5.0% 0.0% 5.0% 10.0% 15.0% 20.0% 25.0% 30.0% 35.0%\n\n### MC turnover trend by Service Lines (European panel 2021-23)\n\n11\\.2% 9.3%\n\n11\\.0%\n\n11\\.1% 10.6%\n\n15\\.6%\n\n22\\.0%\n\n10\\.9%\n\n13\\.5%\n\n6\\.1% 20.7%\n\n30\\.6%\n\n15\\.1%\n\n13\\.1%\n\n\\-1.2%\n\n30\\.3%\n\n10\\.8%\n\n\\-11.5%\n\nSERVICE LINE\\*\\* European panel trend in 2023\\* European panel trend in 2022\\* European panel trend in 2021\\*\n\nSales & Marketing 20.7% 6.1% 30.6%\n\nTechnology 15.6% 10.6% 11.1%\n\nFinance & Risk 15.1% 13.1% -1.2%\n\nStrategy 13.5% 22.0% 10.9%\n\nOperations 11.0% 9.3% 11.2%\n\nPeople & Change -11.5% 30.3% 10.8%\n\n2021 2022 2023\n\n_Source: Our elaborations on MC turnover split by service lines – feaco survey 2021-2023._ \\* The European panel includes only 9 European countries of the feaco survey as detailed data for Spain and Slovenia are missing. These 9 countries represent 69% of the European GDP. Data from Ukraine have not been included in the European panel. \\*\\* The “Other” service line is not reported in the above table as it is not a homogenous category, and its trend would not convey a clear information on the actual market dynamics.\n\n\\-10.0%","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"### 8\n\n## MC turnover trend by Service Lines (2021-23)\n\nIn 2023, in the European panel, Sales & Marketing (+20.7%) and Technology (+15.6%) have been the two fastest growing service lines. Sales & Marketing, after a top growth in 2021 and a relevant slow down in 2022, is again the fastest growing service line. People & Change, after a top growth in 2022, is experiencing a major decline (-11.5%) in spite of an overall MC market that is growing double digits. Technology, Strategy and Operations are experiencing a more stable growth path, with highlighting that many clients are leveraging on technology – namely digital technologies – to redefine their post-pandemic strategies and operations. Finance & Risk, after a record growth in 2020 and a decline in 2021, is also back to a more stable growth trend in 2022 and 2023.\n\nSales & Marketing\n\nTechnology\n\nFinance & Risk\n\nStrategy\n\nOperations\n\nPeople & Change\n\n\\-5.0% 0.0% 5.0% 10.0% 15.0% 20.0% 25.0% 30.0% 35.0%\n\n### MC turnover trend by Service Lines (European panel 2021-23)\n\n11\\.2% 9.3%\n\n11\\.0%\n\n11\\.1% 10.6%\n\n15\\.6%\n\n22\\.0%\n\n10\\.9%\n\n13\\.5%\n\n6\\.1% 20.7%\n\n30\\.6%\n\n15\\.1%\n\n13\\.1%\n\n\\-1.2%\n\n30\\.3%\n\n10\\.8%\n\n\\-11.5%\n\nSERVICE LINE\\*\\* European panel trend in 2023\\* European panel trend in 2022\\* European panel trend in 2021\\*\n\nSales & Marketing 20.7% 6.1% 30.6%\n\nTechnology 15.6% 10.6% 11.1%\n\nFinance & Risk 15.1% 13.1% -1.2%\n\nStrategy 13.5% 22.0% 10.9%\n\nOperations 11.0% 9.3% 11.2%\n\nPeople & Change -11.5% 30.3% 10.8%\n\n2021 2022 2023\n\n_Source: Our elaborations on MC turnover split by service lines – feaco survey 2021-2023._ \\* The European panel includes only 9 European countries of the feaco survey as detailed data for Spain and Slovenia are missing. These 9 countries represent 69% of the European GDP. Data from Ukraine have not been included in the European panel. \\*\\* The “Other” service line is not reported in the above table as it is not a homogenous category, and its trend would not convey a clear information on the actual market dynamics.\n\n\\-10.0%","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Feaco-Survey-2023-2024_p8.pdf"]},"response":{"errors":[],"extract_id":"extract_6317b777a070de546d3d5d3f3211a8c4","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["### 8\n\n## MC turnover trend by Service Lines (2021-23)\n\nIn 2023, in the European panel, Sales & Marketing (+20.7%) and Technology (+15.6%) have been the two fastest growing service lines. Sales & Marketing, after a top growth in 2021 and a relevant slow down in 2022, is again the fastest growing service line. People & Change, after a top growth in 2022, is experiencing a major decline (-11.5%) in spite of an overall MC market that is growing double digits. Technology, Strategy and Operations are experiencing a more stable growth path, with highlighting that many clients are leveraging on technology – namely digital technologies – to redefine their post-pandemic strategies and operations. Finance & Risk, after a record growth in 2020 and a decline in 2021, is also back to a more stable growth trend in 2022 and 2023.\n\nSales & Marketing\n\nTechnology\n\nFinance & Risk\n\nStrategy\n\nOperations\n\nPeople & Change\n\n\\-5.0% 0.0% 5.0% 10.0% 15.0% 20.0% 25.0% 30.0% 35.0%\n\n### MC turnover trend by Service Lines (European panel 2021-23)\n\n11\\.2% 9.3%\n\n11\\.0%\n\n11\\.1% 10.6%\n\n15\\.6%\n\n22\\.0%\n\n10\\.9%\n\n13\\.5%\n\n6\\.1% 20.7%\n\n30\\.6%\n\n15\\.1%\n\n13\\.1%\n\n\\-1.2%\n\n30\\.3%\n\n10\\.8%\n\n\\-11.5%\n\nSERVICE LINE\\*\\* European panel trend in 2023\\* European panel trend in 2022\\* European panel trend in 2021\\*\n\nSales & Marketing 20.7% 6.1% 30.6%\n\nTechnology 15.6% 10.6% 11.1%\n\nFinance & Risk 15.1% 13.1% -1.2%\n\nStrategy 13.5% 22.0% 10.9%\n\nOperations 11.0% 9.3% 11.2%\n\nPeople & Change -11.5% 30.3% 10.8%\n\n2021 2022 2023\n\n_Source: Our elaborations on MC turnover split by service lines – feaco survey 2021-2023._ \\* The European panel includes only 9 European countries of the feaco survey as detailed data for Spain and Slovenia are missing. These 9 countries represent 69% of the European GDP. Data from Ukraine have not been included in the European panel. \\*\\* The “Other” service line is not reported in the above table as it is not a homogenous category, and its trend would not convey a clear information on the actual market dynamics.\n\n\\-10.0%"],"full_content":"### 8\n\n## MC turnover trend by Service Lines (2021-23)\n\nIn 2023, in the European panel, Sales & Marketing (+20.7%) and Technology (+15.6%) have been the two fastest growing service lines. Sales & Marketing, after a top growth in 2021 and a relevant slow down in 2022, is again the fastest growing service line. People & Change, after a top growth in 2022, is experiencing a major decline (-11.5%) in spite of an overall MC market that is growing double digits. Technology, Strategy and Operations are experiencing a more stable growth path, with highlighting that many clients are leveraging on technology – namely digital technologies – to redefine their post-pandemic strategies and operations. Finance & Risk, after a record growth in 2020 and a decline in 2021, is also back to a more stable growth trend in 2022 and 2023.\n\nSales & Marketing\n\nTechnology\n\nFinance & Risk\n\nStrategy\n\nOperations\n\nPeople & Change\n\n\\-5.0% 0.0% 5.0% 10.0% 15.0% 20.0% 25.0% 30.0% 35.0%\n\n### MC turnover trend by Service Lines (European panel 2021-23)\n\n11\\.2% 9.3%\n\n11\\.0%\n\n11\\.1% 10.6%\n\n15\\.6%\n\n22\\.0%\n\n10\\.9%\n\n13\\.5%\n\n6\\.1% 20.7%\n\n30\\.6%\n\n15\\.1%\n\n13\\.1%\n\n\\-1.2%\n\n30\\.3%\n\n10\\.8%\n\n\\-11.5%\n\nSERVICE LINE\\*\\* European panel trend in 2023\\* European panel trend in 2022\\* European panel trend in 2021\\*\n\nSales & Marketing 20.7% 6.1% 30.6%\n\nTechnology 15.6% 10.6% 11.1%\n\nFinance & Risk 15.1% 13.1% -1.2%\n\nStrategy 13.5% 22.0% 10.9%\n\nOperations 11.0% 9.3% 11.2%\n\nPeople & Change -11.5% 30.3% 10.8%\n\n2021 2022 2023\n\n_Source: Our elaborations on MC turnover split by service lines – feaco survey 2021-2023._ \\* The European panel includes only 9 European countries of the feaco survey as detailed data for Spain and Slovenia are missing. These 9 countries represent 69% of the European GDP. Data from Ukraine have not been included in the European panel. \\*\\* The “Other” service line is not reported in the above table as it is not a homogenous category, and its trend would not convey a clear information on the actual market dynamics.\n\n\\-10.0%","publish_date":null,"title":"MC turnover trend by Service Lines (2021-23)","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Feaco-Survey-2023-2024_p8.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Feaco-Survey-2023-2024_p8.pdf"},"request":{"example_id":"chart/Feaco-Survey-2023-2024_p8","product_type":"parse","source_file_path":"Feaco-Survey-2023-2024_p8.pdf"},"started_at":"2026-11-04T19:45:19.518181"} +{"completed_at":"2026-11-04T19:45:22.618712","latency_in_ms":2493,"output":{"example_id":"chart/GWR-2024_Layout_E_RGB_Web_p27","grounded_pages":[],"layout_pages":[],"markdown":"X **Figure 2.1. Annual average economic growth measured as GDP in constant 2015 prices,**\n\n**2006–24** (percentage)\n\n**Source:** IMF 2024b.\n\n\\-0.4\n\n\\-6\n\n\\-4\n\n\\-2\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\nReal GDP growth (%)\n\n2006\n\n2007\n\n2008\n\n2009\n\n2010\n\n2011\n\n2012\n\n2013\n\n2014\n\n2015\n\n2016\n\n2017\n\n2018\n\n2019\n\n2020\n\n2021\n\n2022\n\n2023\n\n2024\n\nWorld Advanced economies Emerging and developing economies\n\n5\\.3 5.3\n\n2\\.9\n\n5\\.2\n\n4\\.0 3.3 3.4 3.5 3.4 3.3 3.8 3.6 2.9\n\n6\\.6\n\n3\\.6 3.3 3.2\n\n7\\.8 8.1\n\n5\\.6\n\n2\\.5\n\n7\\.2\n\n6\\.1\n\n5\\.2 4.9 4.7 4.3 4.4 4.8 4.7\n\n3\\.7\n\n\\-1.8\n\n7\\.0\n\n4\\.0 4\\.4 4\\.2\n\n3\\.1 2.8\n\n0\\.3\n\n\\-3.4\n\n3\\.1\n\n1\\.8 1.2 1.5 2.1 2.4 1.8 2.6 2.3 1.9\n\n\\-4.0\n\n6\\.0\n\n2\\.9\n\n1\\.7 1.8\n\n\\-2.7\n\nRising geopolitical tensions, particularly the war in Ukraine and the Israel–Hamas con- flict, could continue to restrict international trade and deteriorate international relation- ships, thus weighing on growth prospects. In the longer term, climate change remains a threat to sustainable growth for all countries. Developing countries are particularly affected by climate change, in the form of increasingly frequent extreme weather events, with dev- astating effects on private and public infra- structure, disruptions to output and business continuity, and reduced productivity.\n\n## ▶ **2\\.2.**\n\n# **Evolution of** **public debt**\n\nSteep increases in public debt were seen in advanced economies in 2020 due to unprec- edented spending to support firms and households during the COVID-19 pandemic and the subsequent cost-of-living crisis. Following this rapid increase, public debt sub- stantially declined up to 2022 and then stabi- lized. Even so, the debt-to-GDP ratio (that is, government gross debt) in advanced econo- mies is projected to remain higher than pre-pandemic levels at 109 per cent in 2024 (figure 2.2). While the surge in public spend- ing in emerging and developing economies during the COVID-19 pandemic was more modest than what was seen in advanced\n\n# 6\n\n**Global Wage Report 2024–25** 2\\. The global economic and labour market context","pages":[{"line_numbers":[],"markdown":"X **Figure 2.1. Annual average economic growth measured as GDP in constant 2015 prices,**\n\n**2006–24** (percentage)\n\n**Source:** IMF 2024b.\n\n\\-0.4\n\n\\-6\n\n\\-4\n\n\\-2\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\nReal GDP growth (%)\n\n2006\n\n2007\n\n2008\n\n2009\n\n2010\n\n2011\n\n2012\n\n2013\n\n2014\n\n2015\n\n2016\n\n2017\n\n2018\n\n2019\n\n2020\n\n2021\n\n2022\n\n2023\n\n2024\n\nWorld Advanced economies Emerging and developing economies\n\n5\\.3 5.3\n\n2\\.9\n\n5\\.2\n\n4\\.0 3.3 3.4 3.5 3.4 3.3 3.8 3.6 2.9\n\n6\\.6\n\n3\\.6 3.3 3.2\n\n7\\.8 8.1\n\n5\\.6\n\n2\\.5\n\n7\\.2\n\n6\\.1\n\n5\\.2 4.9 4.7 4.3 4.4 4.8 4.7\n\n3\\.7\n\n\\-1.8\n\n7\\.0\n\n4\\.0 4\\.4 4\\.2\n\n3\\.1 2.8\n\n0\\.3\n\n\\-3.4\n\n3\\.1\n\n1\\.8 1.2 1.5 2.1 2.4 1.8 2.6 2.3 1.9\n\n\\-4.0\n\n6\\.0\n\n2\\.9\n\n1\\.7 1.8\n\n\\-2.7\n\nRising geopolitical tensions, particularly the war in Ukraine and the Israel–Hamas con- flict, could continue to restrict international trade and deteriorate international relation- ships, thus weighing on growth prospects. In the longer term, climate change remains a threat to sustainable growth for all countries. Developing countries are particularly affected by climate change, in the form of increasingly frequent extreme weather events, with dev- astating effects on private and public infra- structure, disruptions to output and business continuity, and reduced productivity.\n\n## ▶ **2\\.2.**\n\n# **Evolution of** **public debt**\n\nSteep increases in public debt were seen in advanced economies in 2020 due to unprec- edented spending to support firms and households during the COVID-19 pandemic and the subsequent cost-of-living crisis. Following this rapid increase, public debt sub- stantially declined up to 2022 and then stabi- lized. Even so, the debt-to-GDP ratio (that is, government gross debt) in advanced econo- mies is projected to remain higher than pre-pandemic levels at 109 per cent in 2024 (figure 2.2). While the surge in public spend- ing in emerging and developing economies during the COVID-19 pandemic was more modest than what was seen in advanced\n\n# 6\n\n**Global Wage Report 2024–25** 2\\. The global economic and labour market context","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"X **Figure 2.1. Annual average economic growth measured as GDP in constant 2015 prices,**\n\n**2006–24** (percentage)\n\n**Source:** IMF 2024b.\n\n\\-0.4\n\n\\-6\n\n\\-4\n\n\\-2\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\nReal GDP growth (%)\n\n2006\n\n2007\n\n2008\n\n2009\n\n2010\n\n2011\n\n2012\n\n2013\n\n2014\n\n2015\n\n2016\n\n2017\n\n2018\n\n2019\n\n2020\n\n2021\n\n2022\n\n2023\n\n2024\n\nWorld Advanced economies Emerging and developing economies\n\n5\\.3 5.3\n\n2\\.9\n\n5\\.2\n\n4\\.0 3.3 3.4 3.5 3.4 3.3 3.8 3.6 2.9\n\n6\\.6\n\n3\\.6 3.3 3.2\n\n7\\.8 8.1\n\n5\\.6\n\n2\\.5\n\n7\\.2\n\n6\\.1\n\n5\\.2 4.9 4.7 4.3 4.4 4.8 4.7\n\n3\\.7\n\n\\-1.8\n\n7\\.0\n\n4\\.0 4\\.4 4\\.2\n\n3\\.1 2.8\n\n0\\.3\n\n\\-3.4\n\n3\\.1\n\n1\\.8 1.2 1.5 2.1 2.4 1.8 2.6 2.3 1.9\n\n\\-4.0\n\n6\\.0\n\n2\\.9\n\n1\\.7 1.8\n\n\\-2.7\n\nRising geopolitical tensions, particularly the war in Ukraine and the Israel–Hamas con- flict, could continue to restrict international trade and deteriorate international relation- ships, thus weighing on growth prospects. In the longer term, climate change remains a threat to sustainable growth for all countries. Developing countries are particularly affected by climate change, in the form of increasingly frequent extreme weather events, with dev- astating effects on private and public infra- structure, disruptions to output and business continuity, and reduced productivity.\n\n## ▶ **2\\.2.**\n\n# **Evolution of** **public debt**\n\nSteep increases in public debt were seen in advanced economies in 2020 due to unprec- edented spending to support firms and households during the COVID-19 pandemic and the subsequent cost-of-living crisis. Following this rapid increase, public debt sub- stantially declined up to 2022 and then stabi- lized. Even so, the debt-to-GDP ratio (that is, government gross debt) in advanced econo- mies is projected to remain higher than pre-pandemic levels at 109 per cent in 2024 (figure 2.2). While the surge in public spend- ing in emerging and developing economies during the COVID-19 pandemic was more modest than what was seen in advanced\n\n# 6\n\n**Global Wage Report 2024–25** 2\\. The global economic and labour market context","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/GWR-2024_Layout_E_RGB_Web_p27.pdf"]},"response":{"errors":[],"extract_id":"extract_1cdb1b14f2c011aa4fdccbaee1f667bb","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["X **Figure 2.1. Annual average economic growth measured as GDP in constant 2015 prices,**\n\n**2006–24** (percentage)\n\n**Source:** IMF 2024b.\n\n\\-0.4\n\n\\-6\n\n\\-4\n\n\\-2\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\nReal GDP growth (%)\n\n2006\n\n2007\n\n2008\n\n2009\n\n2010\n\n2011\n\n2012\n\n2013\n\n2014\n\n2015\n\n2016\n\n2017\n\n2018\n\n2019\n\n2020\n\n2021\n\n2022\n\n2023\n\n2024\n\nWorld Advanced economies Emerging and developing economies\n\n5\\.3 5.3\n\n2\\.9\n\n5\\.2\n\n4\\.0 3.3 3.4 3.5 3.4 3.3 3.8 3.6 2.9\n\n6\\.6\n\n3\\.6 3.3 3.2\n\n7\\.8 8.1\n\n5\\.6\n\n2\\.5\n\n7\\.2\n\n6\\.1\n\n5\\.2 4.9 4.7 4.3 4.4 4.8 4.7\n\n3\\.7\n\n\\-1.8\n\n7\\.0\n\n4\\.0 4\\.4 4\\.2\n\n3\\.1 2.8\n\n0\\.3\n\n\\-3.4\n\n3\\.1\n\n1\\.8 1.2 1.5 2.1 2.4 1.8 2.6 2.3 1.9\n\n\\-4.0\n\n6\\.0\n\n2\\.9\n\n1\\.7 1.8\n\n\\-2.7\n\nRising geopolitical tensions, particularly the war in Ukraine and the Israel–Hamas con- flict, could continue to restrict international trade and deteriorate international relation- ships, thus weighing on growth prospects. In the longer term, climate change remains a threat to sustainable growth for all countries. Developing countries are particularly affected by climate change, in the form of increasingly frequent extreme weather events, with dev- astating effects on private and public infra- structure, disruptions to output and business continuity, and reduced productivity.\n\n## ▶ **2\\.2.**\n\n# **Evolution of** **public debt**\n\nSteep increases in public debt were seen in advanced economies in 2020 due to unprec- edented spending to support firms and households during the COVID-19 pandemic and the subsequent cost-of-living crisis. Following this rapid increase, public debt sub- stantially declined up to 2022 and then stabi- lized. Even so, the debt-to-GDP ratio (that is, government gross debt) in advanced econo- mies is projected to remain higher than pre-pandemic levels at 109 per cent in 2024 (figure 2.2). While the surge in public spend- ing in emerging and developing economies during the COVID-19 pandemic was more modest than what was seen in advanced\n\n# 6\n\n**Global Wage Report 2024–25** 2\\. The global economic and labour market context"],"full_content":"X **Figure 2.1. Annual average economic growth measured as GDP in constant 2015 prices,**\n\n**2006–24** (percentage)\n\n**Source:** IMF 2024b.\n\n\\-0.4\n\n\\-6\n\n\\-4\n\n\\-2\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\nReal GDP growth (%)\n\n2006\n\n2007\n\n2008\n\n2009\n\n2010\n\n2011\n\n2012\n\n2013\n\n2014\n\n2015\n\n2016\n\n2017\n\n2018\n\n2019\n\n2020\n\n2021\n\n2022\n\n2023\n\n2024\n\nWorld Advanced economies Emerging and developing economies\n\n5\\.3 5.3\n\n2\\.9\n\n5\\.2\n\n4\\.0 3.3 3.4 3.5 3.4 3.3 3.8 3.6 2.9\n\n6\\.6\n\n3\\.6 3.3 3.2\n\n7\\.8 8.1\n\n5\\.6\n\n2\\.5\n\n7\\.2\n\n6\\.1\n\n5\\.2 4.9 4.7 4.3 4.4 4.8 4.7\n\n3\\.7\n\n\\-1.8\n\n7\\.0\n\n4\\.0 4\\.4 4\\.2\n\n3\\.1 2.8\n\n0\\.3\n\n\\-3.4\n\n3\\.1\n\n1\\.8 1.2 1.5 2.1 2.4 1.8 2.6 2.3 1.9\n\n\\-4.0\n\n6\\.0\n\n2\\.9\n\n1\\.7 1.8\n\n\\-2.7\n\nRising geopolitical tensions, particularly the war in Ukraine and the Israel–Hamas con- flict, could continue to restrict international trade and deteriorate international relation- ships, thus weighing on growth prospects. In the longer term, climate change remains a threat to sustainable growth for all countries. Developing countries are particularly affected by climate change, in the form of increasingly frequent extreme weather events, with dev- astating effects on private and public infra- structure, disruptions to output and business continuity, and reduced productivity.\n\n## ▶ **2\\.2.**\n\n# **Evolution of** **public debt**\n\nSteep increases in public debt were seen in advanced economies in 2020 due to unprec- edented spending to support firms and households during the COVID-19 pandemic and the subsequent cost-of-living crisis. Following this rapid increase, public debt sub- stantially declined up to 2022 and then stabi- lized. Even so, the debt-to-GDP ratio (that is, government gross debt) in advanced econo- mies is projected to remain higher than pre-pandemic levels at 109 per cent in 2024 (figure 2.2). While the surge in public spend- ing in emerging and developing economies during the COVID-19 pandemic was more modest than what was seen in advanced\n\n# 6\n\n**Global Wage Report 2024–25** 2\\. The global economic and labour market context","publish_date":null,"title":"6","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/GWR-2024_Layout_E_RGB_Web_p27.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/GWR-2024_Layout_E_RGB_Web_p27.pdf"},"request":{"example_id":"chart/GWR-2024_Layout_E_RGB_Web_p27","product_type":"parse","source_file_path":"GWR-2024_Layout_E_RGB_Web_p27.pdf"},"started_at":"2026-11-04T19:45:20.124752"} +{"completed_at":"2026-11-04T19:45:24.309570","latency_in_ms":2330,"output":{"example_id":"chart/GWR-2024_Layout_E_RGB_Web_p28","grounded_pages":[],"layout_pages":[],"markdown":"economies, government gross debt in emerg- ing and developing economies has continued to increase in recent years, going from 65 per cent in 2020 to 70 per cent in 2024.\n\nThe difficulty seen among emerging and developing economies to revert to pre- pandemic levels of public debt reflects their limited fiscal leeway, and the further increase in gross government debt in recent years can be partially attributed to the rise in debt- service costs due to historically high interest rates and a strong US dollar (IMF 2023). For many developing countries, higher interest rates in advanced economies also translated into capital outflows, which led to a depletion of foreign exchange reserves and higher bor- rowing costs.\n\n## ▶ **2\\.3.**\n\n# **Inflation rates**\n\nParticularly relevant to the real value of wages has been the acceleration of inflation rates, which peaked in 2022 when they reached 8.6 per cent globally, 7.3 per cent in advanced economies and 9.6 per cent in emerging and developing economies.\n\nSince then, global inflation has decreased from its peak, going from 8.6 per cent in 2022 to 6.7 in 2023, and is predicted to reach 5.8 per cent in 2024 (figure 2.3). However, the moderation of inflation is set to happen at different speeds across the globe. Projections indicate that in 2024 a decline is set to hap- pen in advanced economies, where inflation is estimated at 2.6 per cent, while emerging economies will have to wait at least until 2025 to see a marked slowdown in prices.\n\nX **Figure 2.2. Government gross debt as a share of GDP, 2006–24** (percentage)\n\n**Source:** IMF 2024b.\n\n73 71\n\n78\n\n91\n\n97\n\n101\n\n105 104 103 103 105 103 102 103\n\n122\n\n115\n\n110 109 109\n\n37 36 33\n\n38 37 37 37 38 40\n\n44\n\n49 51 52 55\n\n65 64 64\n\n69 70\n\n30\n\n40\n\n50\n\n60\n\n70\n\n80\n\n90\n\n100\n\n110\n\n120\n\n130\n\nShare of GDP (%)\n\n2006\n\n2007\n\n2008\n\n2009\n\n2010\n\n2011\n\n2012\n\n2013\n\n2014\n\n2015\n\n2016\n\n2017\n\n2018\n\n2019\n\n2020\n\n2021\n\n2022\n\n2023\n\n2024\n\nWorld Advanced economies Emerging and developing economies\n\n# 7\n\n**Global Wage Report 2024–25**\n\n2\\. The global economic and labour market context","pages":[{"line_numbers":[],"markdown":"economies, government gross debt in emerg- ing and developing economies has continued to increase in recent years, going from 65 per cent in 2020 to 70 per cent in 2024.\n\nThe difficulty seen among emerging and developing economies to revert to pre- pandemic levels of public debt reflects their limited fiscal leeway, and the further increase in gross government debt in recent years can be partially attributed to the rise in debt- service costs due to historically high interest rates and a strong US dollar (IMF 2023). For many developing countries, higher interest rates in advanced economies also translated into capital outflows, which led to a depletion of foreign exchange reserves and higher bor- rowing costs.\n\n## ▶ **2\\.3.**\n\n# **Inflation rates**\n\nParticularly relevant to the real value of wages has been the acceleration of inflation rates, which peaked in 2022 when they reached 8.6 per cent globally, 7.3 per cent in advanced economies and 9.6 per cent in emerging and developing economies.\n\nSince then, global inflation has decreased from its peak, going from 8.6 per cent in 2022 to 6.7 in 2023, and is predicted to reach 5.8 per cent in 2024 (figure 2.3). However, the moderation of inflation is set to happen at different speeds across the globe. Projections indicate that in 2024 a decline is set to hap- pen in advanced economies, where inflation is estimated at 2.6 per cent, while emerging economies will have to wait at least until 2025 to see a marked slowdown in prices.\n\nX **Figure 2.2. Government gross debt as a share of GDP, 2006–24** (percentage)\n\n**Source:** IMF 2024b.\n\n73 71\n\n78\n\n91\n\n97\n\n101\n\n105 104 103 103 105 103 102 103\n\n122\n\n115\n\n110 109 109\n\n37 36 33\n\n38 37 37 37 38 40\n\n44\n\n49 51 52 55\n\n65 64 64\n\n69 70\n\n30\n\n40\n\n50\n\n60\n\n70\n\n80\n\n90\n\n100\n\n110\n\n120\n\n130\n\nShare of GDP (%)\n\n2006\n\n2007\n\n2008\n\n2009\n\n2010\n\n2011\n\n2012\n\n2013\n\n2014\n\n2015\n\n2016\n\n2017\n\n2018\n\n2019\n\n2020\n\n2021\n\n2022\n\n2023\n\n2024\n\nWorld Advanced economies Emerging and developing economies\n\n# 7\n\n**Global Wage Report 2024–25**\n\n2\\. The global economic and labour market context","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"economies, government gross debt in emerg- ing and developing economies has continued to increase in recent years, going from 65 per cent in 2020 to 70 per cent in 2024.\n\nThe difficulty seen among emerging and developing economies to revert to pre- pandemic levels of public debt reflects their limited fiscal leeway, and the further increase in gross government debt in recent years can be partially attributed to the rise in debt- service costs due to historically high interest rates and a strong US dollar (IMF 2023). For many developing countries, higher interest rates in advanced economies also translated into capital outflows, which led to a depletion of foreign exchange reserves and higher bor- rowing costs.\n\n## ▶ **2\\.3.**\n\n# **Inflation rates**\n\nParticularly relevant to the real value of wages has been the acceleration of inflation rates, which peaked in 2022 when they reached 8.6 per cent globally, 7.3 per cent in advanced economies and 9.6 per cent in emerging and developing economies.\n\nSince then, global inflation has decreased from its peak, going from 8.6 per cent in 2022 to 6.7 in 2023, and is predicted to reach 5.8 per cent in 2024 (figure 2.3). However, the moderation of inflation is set to happen at different speeds across the globe. Projections indicate that in 2024 a decline is set to hap- pen in advanced economies, where inflation is estimated at 2.6 per cent, while emerging economies will have to wait at least until 2025 to see a marked slowdown in prices.\n\nX **Figure 2.2. Government gross debt as a share of GDP, 2006–24** (percentage)\n\n**Source:** IMF 2024b.\n\n73 71\n\n78\n\n91\n\n97\n\n101\n\n105 104 103 103 105 103 102 103\n\n122\n\n115\n\n110 109 109\n\n37 36 33\n\n38 37 37 37 38 40\n\n44\n\n49 51 52 55\n\n65 64 64\n\n69 70\n\n30\n\n40\n\n50\n\n60\n\n70\n\n80\n\n90\n\n100\n\n110\n\n120\n\n130\n\nShare of GDP (%)\n\n2006\n\n2007\n\n2008\n\n2009\n\n2010\n\n2011\n\n2012\n\n2013\n\n2014\n\n2015\n\n2016\n\n2017\n\n2018\n\n2019\n\n2020\n\n2021\n\n2022\n\n2023\n\n2024\n\nWorld Advanced economies Emerging and developing economies\n\n# 7\n\n**Global Wage Report 2024–25**\n\n2\\. The global economic and labour market context","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/GWR-2024_Layout_E_RGB_Web_p28.pdf"]},"response":{"errors":[],"extract_id":"extract_b1e2c0795002726c245458536c53e638","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["economies, government gross debt in emerg- ing and developing economies has continued to increase in recent years, going from 65 per cent in 2020 to 70 per cent in 2024.\n\nThe difficulty seen among emerging and developing economies to revert to pre- pandemic levels of public debt reflects their limited fiscal leeway, and the further increase in gross government debt in recent years can be partially attributed to the rise in debt- service costs due to historically high interest rates and a strong US dollar (IMF 2023). For many developing countries, higher interest rates in advanced economies also translated into capital outflows, which led to a depletion of foreign exchange reserves and higher bor- rowing costs.\n\n## ▶ **2\\.3.**\n\n# **Inflation rates**\n\nParticularly relevant to the real value of wages has been the acceleration of inflation rates, which peaked in 2022 when they reached 8.6 per cent globally, 7.3 per cent in advanced economies and 9.6 per cent in emerging and developing economies.\n\nSince then, global inflation has decreased from its peak, going from 8.6 per cent in 2022 to 6.7 in 2023, and is predicted to reach 5.8 per cent in 2024 (figure 2.3). However, the moderation of inflation is set to happen at different speeds across the globe. Projections indicate that in 2024 a decline is set to hap- pen in advanced economies, where inflation is estimated at 2.6 per cent, while emerging economies will have to wait at least until 2025 to see a marked slowdown in prices.\n\nX **Figure 2.2. Government gross debt as a share of GDP, 2006–24** (percentage)\n\n**Source:** IMF 2024b.\n\n73 71\n\n78\n\n91\n\n97\n\n101\n\n105 104 103 103 105 103 102 103\n\n122\n\n115\n\n110 109 109\n\n37 36 33\n\n38 37 37 37 38 40\n\n44\n\n49 51 52 55\n\n65 64 64\n\n69 70\n\n30\n\n40\n\n50\n\n60\n\n70\n\n80\n\n90\n\n100\n\n110\n\n120\n\n130\n\nShare of GDP (%)\n\n2006\n\n2007\n\n2008\n\n2009\n\n2010\n\n2011\n\n2012\n\n2013\n\n2014\n\n2015\n\n2016\n\n2017\n\n2018\n\n2019\n\n2020\n\n2021\n\n2022\n\n2023\n\n2024\n\nWorld Advanced economies Emerging and developing economies\n\n# 7\n\n**Global Wage Report 2024–25**\n\n2\\. The global economic and labour market context"],"full_content":"economies, government gross debt in emerg- ing and developing economies has continued to increase in recent years, going from 65 per cent in 2020 to 70 per cent in 2024.\n\nThe difficulty seen among emerging and developing economies to revert to pre- pandemic levels of public debt reflects their limited fiscal leeway, and the further increase in gross government debt in recent years can be partially attributed to the rise in debt- service costs due to historically high interest rates and a strong US dollar (IMF 2023). For many developing countries, higher interest rates in advanced economies also translated into capital outflows, which led to a depletion of foreign exchange reserves and higher bor- rowing costs.\n\n## ▶ **2\\.3.**\n\n# **Inflation rates**\n\nParticularly relevant to the real value of wages has been the acceleration of inflation rates, which peaked in 2022 when they reached 8.6 per cent globally, 7.3 per cent in advanced economies and 9.6 per cent in emerging and developing economies.\n\nSince then, global inflation has decreased from its peak, going from 8.6 per cent in 2022 to 6.7 in 2023, and is predicted to reach 5.8 per cent in 2024 (figure 2.3). However, the moderation of inflation is set to happen at different speeds across the globe. Projections indicate that in 2024 a decline is set to hap- pen in advanced economies, where inflation is estimated at 2.6 per cent, while emerging economies will have to wait at least until 2025 to see a marked slowdown in prices.\n\nX **Figure 2.2. Government gross debt as a share of GDP, 2006–24** (percentage)\n\n**Source:** IMF 2024b.\n\n73 71\n\n78\n\n91\n\n97\n\n101\n\n105 104 103 103 105 103 102 103\n\n122\n\n115\n\n110 109 109\n\n37 36 33\n\n38 37 37 37 38 40\n\n44\n\n49 51 52 55\n\n65 64 64\n\n69 70\n\n30\n\n40\n\n50\n\n60\n\n70\n\n80\n\n90\n\n100\n\n110\n\n120\n\n130\n\nShare of GDP (%)\n\n2006\n\n2007\n\n2008\n\n2009\n\n2010\n\n2011\n\n2012\n\n2013\n\n2014\n\n2015\n\n2016\n\n2017\n\n2018\n\n2019\n\n2020\n\n2021\n\n2022\n\n2023\n\n2024\n\nWorld Advanced economies Emerging and developing economies\n\n# 7\n\n**Global Wage Report 2024–25**\n\n2\\. The global economic and labour market context","publish_date":null,"title":"7","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/GWR-2024_Layout_E_RGB_Web_p28.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/GWR-2024_Layout_E_RGB_Web_p28.pdf"},"request":{"example_id":"chart/GWR-2024_Layout_E_RGB_Web_p28","product_type":"parse","source_file_path":"GWR-2024_Layout_E_RGB_Web_p28.pdf"},"started_at":"2026-11-04T19:45:21.979391"} +{"completed_at":"2026-11-04T19:45:25.517231","latency_in_ms":2350,"output":{"example_id":"chart/GWR-2024_Layout_E_RGB_Web_p32","grounded_pages":[],"layout_pages":[],"markdown":"reporting very fast real wage growth. ⁹ When both China and Türkiye are excluded, global real wage growth is measured at 1.7 per cent in 2024, which still remains the largest in- crease recorded since 2010.\n\nFigure 3.2 presents estimates similar to those of figure 3.1 but focusing on G20 countries, which account for roughly 85 per cent of global GDP and about around 60 per cent of world’s wage employees. Real wage growth across all G20 countries is represented in the figure, along with separate real wage growth estimates for the group of advanced G20 economies and the group of emerging G20 economies.\n\n9\\. Türkiye reported real wage growth of 38.8 per cent in 2023 and 27.2 per cent in 2024. The most recent figures for\n\ninflation in Türkiye are 72.3 per cent in 2022, 53.9 per cent in 2023 and 60.9 per cent in 2024.\n\n10\\. The figure is 0.1 percentage points higher than what estimated in the _Global Wage Report 2022–23_ using data for the\n\nfirst half of 2022. While modest at the aggregate level, this revision is the result of important changes in the estimates for advanced G20 economies and emerging G20 economies, which ended up compensating for each other.\n\n11\\. Average real wage growth in advanced G20 economies has been revised downward by 0.6 percentage points,\n\nfrom –2.2 per cent to –2.8 per cent, while average real wage growth in emerging G20 economies has been revised upward by 1.0 per cent, going from 0.8 per cent to 1.8 per cent. The lowered estimate for advanced G20 economies is driven by lower yearly real wage growth in the Republic of Korea and Australia compared to estimates that used data for just the first half of 2022, while the higher estimate for emerging G20 economies is justified by high real wage growth in India, the Russian Federation and Türkiye.\n\nThe 2022 contraction of real wage growth among G20 countries is in line with what was observed at the global level, reaching –0.8 per cent. ¹⁰ But while average real wage growth in 2022 was negative (–2.8 per cent) in advanced G20 economies, emerging G20 economies experienced a slowdown in real wage growth that nonetheless remained positive at 1.8 per cent. ¹¹ The growth rate in 2023 was even more polarized between advanced and emerging G20 economies: Real wage growth in advanced G20 economies continued to be negative at –0.5 per cent, while real wage growth in emerging G20 economies is esti-\n\nX **Figure 3.1. Annual average global real monthly wage growth, 2006–24** (percentage)\n\n**Source:** ILO estimates based on official national sources as recorded in ILOSTAT and the ILO Global Wage Database.\n\n2\\.1\n\n2\\.4\n\n0\\.6 0.8\n\n1\\.8\n\n0\\.9\n\n1\\.5\n\n1\\.7\n\n1\\.4 1.4\n\n1\\.6\n\n0\\.9\n\n1\\.2 1.3\n\n1\\.1 0.8\n\n\\-1.5\n\n1\\.3\n\n2\\.3\n\n2\\.7\n\n3\\.1\n\n1\\.2\n\n1\\.6\n\n2\\.4\n\n1\\.6\n\n2\\.2 2.5\n\n1\\.9 1.9 2.1\n\n1\\.5\n\n1\\.9 1.9\n\n1\\.5\n\n1\\.8\n\n\\-0.9\n\n1\\.8\n\n2\\.7\n\n\\-2.0\n\n\\-1.0\n\n0\\.0\n\n1\\.0\n\n2\\.0\n\n3\\.0\n\n4\\.0\n\nChange (%)\n\n2006\n\n2007\n\n2008\n\n2009\n\n2010\n\n2011\n\n2012\n\n2013\n\n2014\n\n2015\n\n2016\n\n2017\n\n2018\n\n2019\n\n2020\n\n2021\n\n2022\n\n2023\n\n2024\n\nGlobal Global (without China)\n\n# 12\n\n**Global Wage Report 2024–25** 3\\. Wage trends","pages":[{"line_numbers":[],"markdown":"reporting very fast real wage growth. ⁹ When both China and Türkiye are excluded, global real wage growth is measured at 1.7 per cent in 2024, which still remains the largest in- crease recorded since 2010.\n\nFigure 3.2 presents estimates similar to those of figure 3.1 but focusing on G20 countries, which account for roughly 85 per cent of global GDP and about around 60 per cent of world’s wage employees. Real wage growth across all G20 countries is represented in the figure, along with separate real wage growth estimates for the group of advanced G20 economies and the group of emerging G20 economies.\n\n9\\. Türkiye reported real wage growth of 38.8 per cent in 2023 and 27.2 per cent in 2024. The most recent figures for\n\ninflation in Türkiye are 72.3 per cent in 2022, 53.9 per cent in 2023 and 60.9 per cent in 2024.\n\n10\\. The figure is 0.1 percentage points higher than what estimated in the _Global Wage Report 2022–23_ using data for the\n\nfirst half of 2022. While modest at the aggregate level, this revision is the result of important changes in the estimates for advanced G20 economies and emerging G20 economies, which ended up compensating for each other.\n\n11\\. Average real wage growth in advanced G20 economies has been revised downward by 0.6 percentage points,\n\nfrom –2.2 per cent to –2.8 per cent, while average real wage growth in emerging G20 economies has been revised upward by 1.0 per cent, going from 0.8 per cent to 1.8 per cent. The lowered estimate for advanced G20 economies is driven by lower yearly real wage growth in the Republic of Korea and Australia compared to estimates that used data for just the first half of 2022, while the higher estimate for emerging G20 economies is justified by high real wage growth in India, the Russian Federation and Türkiye.\n\nThe 2022 contraction of real wage growth among G20 countries is in line with what was observed at the global level, reaching –0.8 per cent. ¹⁰ But while average real wage growth in 2022 was negative (–2.8 per cent) in advanced G20 economies, emerging G20 economies experienced a slowdown in real wage growth that nonetheless remained positive at 1.8 per cent. ¹¹ The growth rate in 2023 was even more polarized between advanced and emerging G20 economies: Real wage growth in advanced G20 economies continued to be negative at –0.5 per cent, while real wage growth in emerging G20 economies is esti-\n\nX **Figure 3.1. Annual average global real monthly wage growth, 2006–24** (percentage)\n\n**Source:** ILO estimates based on official national sources as recorded in ILOSTAT and the ILO Global Wage Database.\n\n2\\.1\n\n2\\.4\n\n0\\.6 0.8\n\n1\\.8\n\n0\\.9\n\n1\\.5\n\n1\\.7\n\n1\\.4 1.4\n\n1\\.6\n\n0\\.9\n\n1\\.2 1.3\n\n1\\.1 0.8\n\n\\-1.5\n\n1\\.3\n\n2\\.3\n\n2\\.7\n\n3\\.1\n\n1\\.2\n\n1\\.6\n\n2\\.4\n\n1\\.6\n\n2\\.2 2.5\n\n1\\.9 1.9 2.1\n\n1\\.5\n\n1\\.9 1.9\n\n1\\.5\n\n1\\.8\n\n\\-0.9\n\n1\\.8\n\n2\\.7\n\n\\-2.0\n\n\\-1.0\n\n0\\.0\n\n1\\.0\n\n2\\.0\n\n3\\.0\n\n4\\.0\n\nChange (%)\n\n2006\n\n2007\n\n2008\n\n2009\n\n2010\n\n2011\n\n2012\n\n2013\n\n2014\n\n2015\n\n2016\n\n2017\n\n2018\n\n2019\n\n2020\n\n2021\n\n2022\n\n2023\n\n2024\n\nGlobal Global (without China)\n\n# 12\n\n**Global Wage Report 2024–25** 3\\. Wage trends","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"reporting very fast real wage growth. ⁹ When both China and Türkiye are excluded, global real wage growth is measured at 1.7 per cent in 2024, which still remains the largest in- crease recorded since 2010.\n\nFigure 3.2 presents estimates similar to those of figure 3.1 but focusing on G20 countries, which account for roughly 85 per cent of global GDP and about around 60 per cent of world’s wage employees. Real wage growth across all G20 countries is represented in the figure, along with separate real wage growth estimates for the group of advanced G20 economies and the group of emerging G20 economies.\n\n9\\. Türkiye reported real wage growth of 38.8 per cent in 2023 and 27.2 per cent in 2024. The most recent figures for\n\ninflation in Türkiye are 72.3 per cent in 2022, 53.9 per cent in 2023 and 60.9 per cent in 2024.\n\n10\\. The figure is 0.1 percentage points higher than what estimated in the _Global Wage Report 2022–23_ using data for the\n\nfirst half of 2022. While modest at the aggregate level, this revision is the result of important changes in the estimates for advanced G20 economies and emerging G20 economies, which ended up compensating for each other.\n\n11\\. Average real wage growth in advanced G20 economies has been revised downward by 0.6 percentage points,\n\nfrom –2.2 per cent to –2.8 per cent, while average real wage growth in emerging G20 economies has been revised upward by 1.0 per cent, going from 0.8 per cent to 1.8 per cent. The lowered estimate for advanced G20 economies is driven by lower yearly real wage growth in the Republic of Korea and Australia compared to estimates that used data for just the first half of 2022, while the higher estimate for emerging G20 economies is justified by high real wage growth in India, the Russian Federation and Türkiye.\n\nThe 2022 contraction of real wage growth among G20 countries is in line with what was observed at the global level, reaching –0.8 per cent. ¹⁰ But while average real wage growth in 2022 was negative (–2.8 per cent) in advanced G20 economies, emerging G20 economies experienced a slowdown in real wage growth that nonetheless remained positive at 1.8 per cent. ¹¹ The growth rate in 2023 was even more polarized between advanced and emerging G20 economies: Real wage growth in advanced G20 economies continued to be negative at –0.5 per cent, while real wage growth in emerging G20 economies is esti-\n\nX **Figure 3.1. Annual average global real monthly wage growth, 2006–24** (percentage)\n\n**Source:** ILO estimates based on official national sources as recorded in ILOSTAT and the ILO Global Wage Database.\n\n2\\.1\n\n2\\.4\n\n0\\.6 0.8\n\n1\\.8\n\n0\\.9\n\n1\\.5\n\n1\\.7\n\n1\\.4 1.4\n\n1\\.6\n\n0\\.9\n\n1\\.2 1.3\n\n1\\.1 0.8\n\n\\-1.5\n\n1\\.3\n\n2\\.3\n\n2\\.7\n\n3\\.1\n\n1\\.2\n\n1\\.6\n\n2\\.4\n\n1\\.6\n\n2\\.2 2.5\n\n1\\.9 1.9 2.1\n\n1\\.5\n\n1\\.9 1.9\n\n1\\.5\n\n1\\.8\n\n\\-0.9\n\n1\\.8\n\n2\\.7\n\n\\-2.0\n\n\\-1.0\n\n0\\.0\n\n1\\.0\n\n2\\.0\n\n3\\.0\n\n4\\.0\n\nChange (%)\n\n2006\n\n2007\n\n2008\n\n2009\n\n2010\n\n2011\n\n2012\n\n2013\n\n2014\n\n2015\n\n2016\n\n2017\n\n2018\n\n2019\n\n2020\n\n2021\n\n2022\n\n2023\n\n2024\n\nGlobal Global (without China)\n\n# 12\n\n**Global Wage Report 2024–25** 3\\. Wage trends","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/GWR-2024_Layout_E_RGB_Web_p32.pdf"]},"response":{"errors":[],"extract_id":"extract_77cdafe22aff93d408eaa00e6337e98e","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["reporting very fast real wage growth. ⁹ When both China and Türkiye are excluded, global real wage growth is measured at 1.7 per cent in 2024, which still remains the largest in- crease recorded since 2010.\n\nFigure 3.2 presents estimates similar to those of figure 3.1 but focusing on G20 countries, which account for roughly 85 per cent of global GDP and about around 60 per cent of world’s wage employees. Real wage growth across all G20 countries is represented in the figure, along with separate real wage growth estimates for the group of advanced G20 economies and the group of emerging G20 economies.\n\n9\\. Türkiye reported real wage growth of 38.8 per cent in 2023 and 27.2 per cent in 2024. The most recent figures for\n\ninflation in Türkiye are 72.3 per cent in 2022, 53.9 per cent in 2023 and 60.9 per cent in 2024.\n\n10\\. The figure is 0.1 percentage points higher than what estimated in the _Global Wage Report 2022–23_ using data for the\n\nfirst half of 2022. While modest at the aggregate level, this revision is the result of important changes in the estimates for advanced G20 economies and emerging G20 economies, which ended up compensating for each other.\n\n11\\. Average real wage growth in advanced G20 economies has been revised downward by 0.6 percentage points,\n\nfrom –2.2 per cent to –2.8 per cent, while average real wage growth in emerging G20 economies has been revised upward by 1.0 per cent, going from 0.8 per cent to 1.8 per cent. The lowered estimate for advanced G20 economies is driven by lower yearly real wage growth in the Republic of Korea and Australia compared to estimates that used data for just the first half of 2022, while the higher estimate for emerging G20 economies is justified by high real wage growth in India, the Russian Federation and Türkiye.\n\nThe 2022 contraction of real wage growth among G20 countries is in line with what was observed at the global level, reaching –0.8 per cent. ¹⁰ But while average real wage growth in 2022 was negative (–2.8 per cent) in advanced G20 economies, emerging G20 economies experienced a slowdown in real wage growth that nonetheless remained positive at 1.8 per cent. ¹¹ The growth rate in 2023 was even more polarized between advanced and emerging G20 economies: Real wage growth in advanced G20 economies continued to be negative at –0.5 per cent, while real wage growth in emerging G20 economies is esti-\n\nX **Figure 3.1. Annual average global real monthly wage growth, 2006–24** (percentage)\n\n**Source:** ILO estimates based on official national sources as recorded in ILOSTAT and the ILO Global Wage Database.\n\n2\\.1\n\n2\\.4\n\n0\\.6 0.8\n\n1\\.8\n\n0\\.9\n\n1\\.5\n\n1\\.7\n\n1\\.4 1.4\n\n1\\.6\n\n0\\.9\n\n1\\.2 1.3\n\n1\\.1 0.8\n\n\\-1.5\n\n1\\.3\n\n2\\.3\n\n2\\.7\n\n3\\.1\n\n1\\.2\n\n1\\.6\n\n2\\.4\n\n1\\.6\n\n2\\.2 2.5\n\n1\\.9 1.9 2.1\n\n1\\.5\n\n1\\.9 1.9\n\n1\\.5\n\n1\\.8\n\n\\-0.9\n\n1\\.8\n\n2\\.7\n\n\\-2.0\n\n\\-1.0\n\n0\\.0\n\n1\\.0\n\n2\\.0\n\n3\\.0\n\n4\\.0\n\nChange (%)\n\n2006\n\n2007\n\n2008\n\n2009\n\n2010\n\n2011\n\n2012\n\n2013\n\n2014\n\n2015\n\n2016\n\n2017\n\n2018\n\n2019\n\n2020\n\n2021\n\n2022\n\n2023\n\n2024\n\nGlobal Global (without China)\n\n# 12\n\n**Global Wage Report 2024–25** 3\\. Wage trends"],"full_content":"reporting very fast real wage growth. ⁹ When both China and Türkiye are excluded, global real wage growth is measured at 1.7 per cent in 2024, which still remains the largest in- crease recorded since 2010.\n\nFigure 3.2 presents estimates similar to those of figure 3.1 but focusing on G20 countries, which account for roughly 85 per cent of global GDP and about around 60 per cent of world’s wage employees. Real wage growth across all G20 countries is represented in the figure, along with separate real wage growth estimates for the group of advanced G20 economies and the group of emerging G20 economies.\n\n9\\. Türkiye reported real wage growth of 38.8 per cent in 2023 and 27.2 per cent in 2024. The most recent figures for\n\ninflation in Türkiye are 72.3 per cent in 2022, 53.9 per cent in 2023 and 60.9 per cent in 2024.\n\n10\\. The figure is 0.1 percentage points higher than what estimated in the _Global Wage Report 2022–23_ using data for the\n\nfirst half of 2022. While modest at the aggregate level, this revision is the result of important changes in the estimates for advanced G20 economies and emerging G20 economies, which ended up compensating for each other.\n\n11\\. Average real wage growth in advanced G20 economies has been revised downward by 0.6 percentage points,\n\nfrom –2.2 per cent to –2.8 per cent, while average real wage growth in emerging G20 economies has been revised upward by 1.0 per cent, going from 0.8 per cent to 1.8 per cent. The lowered estimate for advanced G20 economies is driven by lower yearly real wage growth in the Republic of Korea and Australia compared to estimates that used data for just the first half of 2022, while the higher estimate for emerging G20 economies is justified by high real wage growth in India, the Russian Federation and Türkiye.\n\nThe 2022 contraction of real wage growth among G20 countries is in line with what was observed at the global level, reaching –0.8 per cent. ¹⁰ But while average real wage growth in 2022 was negative (–2.8 per cent) in advanced G20 economies, emerging G20 economies experienced a slowdown in real wage growth that nonetheless remained positive at 1.8 per cent. ¹¹ The growth rate in 2023 was even more polarized between advanced and emerging G20 economies: Real wage growth in advanced G20 economies continued to be negative at –0.5 per cent, while real wage growth in emerging G20 economies is esti-\n\nX **Figure 3.1. Annual average global real monthly wage growth, 2006–24** (percentage)\n\n**Source:** ILO estimates based on official national sources as recorded in ILOSTAT and the ILO Global Wage Database.\n\n2\\.1\n\n2\\.4\n\n0\\.6 0.8\n\n1\\.8\n\n0\\.9\n\n1\\.5\n\n1\\.7\n\n1\\.4 1.4\n\n1\\.6\n\n0\\.9\n\n1\\.2 1.3\n\n1\\.1 0.8\n\n\\-1.5\n\n1\\.3\n\n2\\.3\n\n2\\.7\n\n3\\.1\n\n1\\.2\n\n1\\.6\n\n2\\.4\n\n1\\.6\n\n2\\.2 2.5\n\n1\\.9 1.9 2.1\n\n1\\.5\n\n1\\.9 1.9\n\n1\\.5\n\n1\\.8\n\n\\-0.9\n\n1\\.8\n\n2\\.7\n\n\\-2.0\n\n\\-1.0\n\n0\\.0\n\n1\\.0\n\n2\\.0\n\n3\\.0\n\n4\\.0\n\nChange (%)\n\n2006\n\n2007\n\n2008\n\n2009\n\n2010\n\n2011\n\n2012\n\n2013\n\n2014\n\n2015\n\n2016\n\n2017\n\n2018\n\n2019\n\n2020\n\n2021\n\n2022\n\n2023\n\n2024\n\nGlobal Global (without China)\n\n# 12\n\n**Global Wage Report 2024–25** 3\\. Wage trends","publish_date":null,"title":"12","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/GWR-2024_Layout_E_RGB_Web_p32.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/GWR-2024_Layout_E_RGB_Web_p32.pdf"},"request":{"example_id":"chart/GWR-2024_Layout_E_RGB_Web_p32","product_type":"parse","source_file_path":"GWR-2024_Layout_E_RGB_Web_p32.pdf"},"started_at":"2026-11-04T19:45:23.166480"} +{"completed_at":"2026-11-04T19:45:26.735741","latency_in_ms":1641,"output":{"example_id":"chart/GWR-2024_Layout_E_RGB_Web_p33","grounded_pages":[],"layout_pages":[],"markdown":"mated at 6.0 per cent, the highest value re- corded in the last ten years.\n\nDue to the large share of global wage employ- ees that they represent, the real wage growth rate of G20 countries in 2024 is similar to the growth rate estimated at the global level, standing at 3.1 per cent. After two years of negative growth, real wages in advanced G20 economies are projected to return to growth in 2024, with a 0.9 per cent increase, while wage growth in emerging G20 economies is projected to remain almost unchanged at 5.9 per cent.\n\n## ▶ **3\\.2.**\n\n# **Regional wage** **trends**\n\nFigure 3.3 presents regional real wage growth trends to complement the analysis at the global level. When looking at regional-level data, the evolution of real wage growth re- veals substantial heterogeneity. While real wage growth declines driven by the cost- of-living crisis are evident in all regions in 2022, the ways in which real wages evolved in the years preceding and following 2022 vary considerably from region to region.\n\nIn Northern America (Canada and the United States), from 2006 until the COVID-19 pan- demic in 2020, real wage growth fluctuated between 0 and 1 per cent in most years. As\n\nX **Figure 3.2. Annual average real monthly wage growth in the G20 countries, 2006–24**\n\n(percentage)\n\n**Notes:** Advanced G20 economies comprise Australia, Canada, France, Germany, Italy, Japan, the Republic of Korea, the United Kingdom of Great Britain and Northern Ireland, and the United States. Emerging G20 economies comprise Argentina, Brazil, China, India, Indonesia, Mexico, the Russian Federation, Saudi Arabia, South Africa and Türkiye.\n\n**Source:** ILO estimates based on official national sources as recorded in ILOSTAT and the ILO Global Wage Database\n\n2\\.9 0.6 8.0\n\n3\\.2 0.7 8.6\n\n1\\.4 -1.1 5.9\n\n1\\.7 0.7 3.2\n\n2\\.7 1.0 5.6\n\n1\\.8 -0.3 5.1\n\n2\\.6 0.2 6.4\n\n2\\.7 0.4 5.9\n\n1\\.6 0.5 3.2\n\n2\\.0 1.7 2.2\n\n2\\.2 1.0 4.0\n\n1\\.5 0.4 3.2\n\n2\\.2 0.7 4.2\n\n1\\.8 1.0 3.0\n\n2\\.2 1.8 2.7\n\n2\\.1 0.5 4.3\n\n\\-0.8 -2.8 1.8\n\n2\\.3 -0.5 6.0\n\n3\\.1 0.9 5.9\n\n\\-4\n\n\\-2\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n2006\n\n2007\n\n2008\n\n2009\n\n2010\n\n2011\n\n2012\n\n2013\n\n2014\n\n2015\n\n2016\n\n2017\n\n2018\n\n2019\n\n2020\n\n2021\n\n2022\n\n2023\n\n2024\n\nG20 G20 advanced economies G20 emerging countries\n\nChange (%)\n\n# 13\n\n**Global Wage Report 2024–25**\n\n3\\. Wage trends","pages":[{"line_numbers":[],"markdown":"mated at 6.0 per cent, the highest value re- corded in the last ten years.\n\nDue to the large share of global wage employ- ees that they represent, the real wage growth rate of G20 countries in 2024 is similar to the growth rate estimated at the global level, standing at 3.1 per cent. After two years of negative growth, real wages in advanced G20 economies are projected to return to growth in 2024, with a 0.9 per cent increase, while wage growth in emerging G20 economies is projected to remain almost unchanged at 5.9 per cent.\n\n## ▶ **3\\.2.**\n\n# **Regional wage** **trends**\n\nFigure 3.3 presents regional real wage growth trends to complement the analysis at the global level. When looking at regional-level data, the evolution of real wage growth re- veals substantial heterogeneity. While real wage growth declines driven by the cost- of-living crisis are evident in all regions in 2022, the ways in which real wages evolved in the years preceding and following 2022 vary considerably from region to region.\n\nIn Northern America (Canada and the United States), from 2006 until the COVID-19 pan- demic in 2020, real wage growth fluctuated between 0 and 1 per cent in most years. As\n\nX **Figure 3.2. Annual average real monthly wage growth in the G20 countries, 2006–24**\n\n(percentage)\n\n**Notes:** Advanced G20 economies comprise Australia, Canada, France, Germany, Italy, Japan, the Republic of Korea, the United Kingdom of Great Britain and Northern Ireland, and the United States. Emerging G20 economies comprise Argentina, Brazil, China, India, Indonesia, Mexico, the Russian Federation, Saudi Arabia, South Africa and Türkiye.\n\n**Source:** ILO estimates based on official national sources as recorded in ILOSTAT and the ILO Global Wage Database\n\n2\\.9 0.6 8.0\n\n3\\.2 0.7 8.6\n\n1\\.4 -1.1 5.9\n\n1\\.7 0.7 3.2\n\n2\\.7 1.0 5.6\n\n1\\.8 -0.3 5.1\n\n2\\.6 0.2 6.4\n\n2\\.7 0.4 5.9\n\n1\\.6 0.5 3.2\n\n2\\.0 1.7 2.2\n\n2\\.2 1.0 4.0\n\n1\\.5 0.4 3.2\n\n2\\.2 0.7 4.2\n\n1\\.8 1.0 3.0\n\n2\\.2 1.8 2.7\n\n2\\.1 0.5 4.3\n\n\\-0.8 -2.8 1.8\n\n2\\.3 -0.5 6.0\n\n3\\.1 0.9 5.9\n\n\\-4\n\n\\-2\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n2006\n\n2007\n\n2008\n\n2009\n\n2010\n\n2011\n\n2012\n\n2013\n\n2014\n\n2015\n\n2016\n\n2017\n\n2018\n\n2019\n\n2020\n\n2021\n\n2022\n\n2023\n\n2024\n\nG20 G20 advanced economies G20 emerging countries\n\nChange (%)\n\n# 13\n\n**Global Wage Report 2024–25**\n\n3\\. Wage trends","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"mated at 6.0 per cent, the highest value re- corded in the last ten years.\n\nDue to the large share of global wage employ- ees that they represent, the real wage growth rate of G20 countries in 2024 is similar to the growth rate estimated at the global level, standing at 3.1 per cent. After two years of negative growth, real wages in advanced G20 economies are projected to return to growth in 2024, with a 0.9 per cent increase, while wage growth in emerging G20 economies is projected to remain almost unchanged at 5.9 per cent.\n\n## ▶ **3\\.2.**\n\n# **Regional wage** **trends**\n\nFigure 3.3 presents regional real wage growth trends to complement the analysis at the global level. When looking at regional-level data, the evolution of real wage growth re- veals substantial heterogeneity. While real wage growth declines driven by the cost- of-living crisis are evident in all regions in 2022, the ways in which real wages evolved in the years preceding and following 2022 vary considerably from region to region.\n\nIn Northern America (Canada and the United States), from 2006 until the COVID-19 pan- demic in 2020, real wage growth fluctuated between 0 and 1 per cent in most years. As\n\nX **Figure 3.2. Annual average real monthly wage growth in the G20 countries, 2006–24**\n\n(percentage)\n\n**Notes:** Advanced G20 economies comprise Australia, Canada, France, Germany, Italy, Japan, the Republic of Korea, the United Kingdom of Great Britain and Northern Ireland, and the United States. Emerging G20 economies comprise Argentina, Brazil, China, India, Indonesia, Mexico, the Russian Federation, Saudi Arabia, South Africa and Türkiye.\n\n**Source:** ILO estimates based on official national sources as recorded in ILOSTAT and the ILO Global Wage Database\n\n2\\.9 0.6 8.0\n\n3\\.2 0.7 8.6\n\n1\\.4 -1.1 5.9\n\n1\\.7 0.7 3.2\n\n2\\.7 1.0 5.6\n\n1\\.8 -0.3 5.1\n\n2\\.6 0.2 6.4\n\n2\\.7 0.4 5.9\n\n1\\.6 0.5 3.2\n\n2\\.0 1.7 2.2\n\n2\\.2 1.0 4.0\n\n1\\.5 0.4 3.2\n\n2\\.2 0.7 4.2\n\n1\\.8 1.0 3.0\n\n2\\.2 1.8 2.7\n\n2\\.1 0.5 4.3\n\n\\-0.8 -2.8 1.8\n\n2\\.3 -0.5 6.0\n\n3\\.1 0.9 5.9\n\n\\-4\n\n\\-2\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n2006\n\n2007\n\n2008\n\n2009\n\n2010\n\n2011\n\n2012\n\n2013\n\n2014\n\n2015\n\n2016\n\n2017\n\n2018\n\n2019\n\n2020\n\n2021\n\n2022\n\n2023\n\n2024\n\nG20 G20 advanced economies G20 emerging countries\n\nChange (%)\n\n# 13\n\n**Global Wage Report 2024–25**\n\n3\\. Wage trends","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/GWR-2024_Layout_E_RGB_Web_p33.pdf"]},"response":{"errors":[],"extract_id":"extract_0321de7b05fae84365c2829373937505","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["mated at 6.0 per cent, the highest value re- corded in the last ten years.\n\nDue to the large share of global wage employ- ees that they represent, the real wage growth rate of G20 countries in 2024 is similar to the growth rate estimated at the global level, standing at 3.1 per cent. After two years of negative growth, real wages in advanced G20 economies are projected to return to growth in 2024, with a 0.9 per cent increase, while wage growth in emerging G20 economies is projected to remain almost unchanged at 5.9 per cent.\n\n## ▶ **3\\.2.**\n\n# **Regional wage** **trends**\n\nFigure 3.3 presents regional real wage growth trends to complement the analysis at the global level. When looking at regional-level data, the evolution of real wage growth re- veals substantial heterogeneity. While real wage growth declines driven by the cost- of-living crisis are evident in all regions in 2022, the ways in which real wages evolved in the years preceding and following 2022 vary considerably from region to region.\n\nIn Northern America (Canada and the United States), from 2006 until the COVID-19 pan- demic in 2020, real wage growth fluctuated between 0 and 1 per cent in most years. As\n\nX **Figure 3.2. Annual average real monthly wage growth in the G20 countries, 2006–24**\n\n(percentage)\n\n**Notes:** Advanced G20 economies comprise Australia, Canada, France, Germany, Italy, Japan, the Republic of Korea, the United Kingdom of Great Britain and Northern Ireland, and the United States. Emerging G20 economies comprise Argentina, Brazil, China, India, Indonesia, Mexico, the Russian Federation, Saudi Arabia, South Africa and Türkiye.\n\n**Source:** ILO estimates based on official national sources as recorded in ILOSTAT and the ILO Global Wage Database\n\n2\\.9 0.6 8.0\n\n3\\.2 0.7 8.6\n\n1\\.4 -1.1 5.9\n\n1\\.7 0.7 3.2\n\n2\\.7 1.0 5.6\n\n1\\.8 -0.3 5.1\n\n2\\.6 0.2 6.4\n\n2\\.7 0.4 5.9\n\n1\\.6 0.5 3.2\n\n2\\.0 1.7 2.2\n\n2\\.2 1.0 4.0\n\n1\\.5 0.4 3.2\n\n2\\.2 0.7 4.2\n\n1\\.8 1.0 3.0\n\n2\\.2 1.8 2.7\n\n2\\.1 0.5 4.3\n\n\\-0.8 -2.8 1.8\n\n2\\.3 -0.5 6.0\n\n3\\.1 0.9 5.9\n\n\\-4\n\n\\-2\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n2006\n\n2007\n\n2008\n\n2009\n\n2010\n\n2011\n\n2012\n\n2013\n\n2014\n\n2015\n\n2016\n\n2017\n\n2018\n\n2019\n\n2020\n\n2021\n\n2022\n\n2023\n\n2024\n\nG20 G20 advanced economies G20 emerging countries\n\nChange (%)\n\n# 13\n\n**Global Wage Report 2024–25**\n\n3\\. Wage trends"],"full_content":"mated at 6.0 per cent, the highest value re- corded in the last ten years.\n\nDue to the large share of global wage employ- ees that they represent, the real wage growth rate of G20 countries in 2024 is similar to the growth rate estimated at the global level, standing at 3.1 per cent. After two years of negative growth, real wages in advanced G20 economies are projected to return to growth in 2024, with a 0.9 per cent increase, while wage growth in emerging G20 economies is projected to remain almost unchanged at 5.9 per cent.\n\n## ▶ **3\\.2.**\n\n# **Regional wage** **trends**\n\nFigure 3.3 presents regional real wage growth trends to complement the analysis at the global level. When looking at regional-level data, the evolution of real wage growth re- veals substantial heterogeneity. While real wage growth declines driven by the cost- of-living crisis are evident in all regions in 2022, the ways in which real wages evolved in the years preceding and following 2022 vary considerably from region to region.\n\nIn Northern America (Canada and the United States), from 2006 until the COVID-19 pan- demic in 2020, real wage growth fluctuated between 0 and 1 per cent in most years. As\n\nX **Figure 3.2. Annual average real monthly wage growth in the G20 countries, 2006–24**\n\n(percentage)\n\n**Notes:** Advanced G20 economies comprise Australia, Canada, France, Germany, Italy, Japan, the Republic of Korea, the United Kingdom of Great Britain and Northern Ireland, and the United States. Emerging G20 economies comprise Argentina, Brazil, China, India, Indonesia, Mexico, the Russian Federation, Saudi Arabia, South Africa and Türkiye.\n\n**Source:** ILO estimates based on official national sources as recorded in ILOSTAT and the ILO Global Wage Database\n\n2\\.9 0.6 8.0\n\n3\\.2 0.7 8.6\n\n1\\.4 -1.1 5.9\n\n1\\.7 0.7 3.2\n\n2\\.7 1.0 5.6\n\n1\\.8 -0.3 5.1\n\n2\\.6 0.2 6.4\n\n2\\.7 0.4 5.9\n\n1\\.6 0.5 3.2\n\n2\\.0 1.7 2.2\n\n2\\.2 1.0 4.0\n\n1\\.5 0.4 3.2\n\n2\\.2 0.7 4.2\n\n1\\.8 1.0 3.0\n\n2\\.2 1.8 2.7\n\n2\\.1 0.5 4.3\n\n\\-0.8 -2.8 1.8\n\n2\\.3 -0.5 6.0\n\n3\\.1 0.9 5.9\n\n\\-4\n\n\\-2\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n2006\n\n2007\n\n2008\n\n2009\n\n2010\n\n2011\n\n2012\n\n2013\n\n2014\n\n2015\n\n2016\n\n2017\n\n2018\n\n2019\n\n2020\n\n2021\n\n2022\n\n2023\n\n2024\n\nG20 G20 advanced economies G20 emerging countries\n\nChange (%)\n\n# 13\n\n**Global Wage Report 2024–25**\n\n3\\. Wage trends","publish_date":null,"title":"13","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/GWR-2024_Layout_E_RGB_Web_p33.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/GWR-2024_Layout_E_RGB_Web_p33.pdf"},"request":{"example_id":"chart/GWR-2024_Layout_E_RGB_Web_p33","product_type":"parse","source_file_path":"GWR-2024_Layout_E_RGB_Web_p33.pdf"},"started_at":"2026-11-04T19:45:25.094553"} +{"completed_at":"2026-11-04T19:45:28.144267","latency_in_ms":2560,"output":{"example_id":"chart/GWR-2024_Layout_E_RGB_Web_p58","grounded_pages":[],"layout_pages":[],"markdown":"high at the bottom half of the distribution, which is again probably where informal em- ployment among women dominates. When comparing country income groups on a de- cile-by-decile basis, the highest gender wage gaps tend to be found in the low-income countries – with the exception of the bottom decile, where the gender wage gap is higher among lower-middle-income countries.\n\nTable 5.1 shows the gender wage gaps esti- mated among all wage workers in the popu- lation for each country income group using the classic measures for estimating pay difference – namely, the raw mean and median gender wage gaps. The two measure are presented together with estimates of the mean and me- dian factor-weighted gender wage gap, com- plementary measures that can help better\n\nX **Figure 5.3. Share of women among wage employees in each decile of the hourly wage**\n\n**distribution and the decile-specific gender wage gap by country income group, latest available year** (percentage)\n\n**Notes:** Each of the four charts shows two vertical axes: the left-hand axis displays values for the share of women among wage workers in each decile, while the right-hand axis displays values for the decile-specific mean and median gender wage gaps. All three estimates in each chart by decile are based on the weighted value between all of the countries in the sample from the corresponding income group, with weights determined by the size of wage employment in each country and decile. The figures displayed in the charts were obtained using data from the most recent available survey (T1 in Appendix I).\n\n**Source:** ILO estimates based on national survey data from 82 countries. See Appendix I for the sources of data and year of collection.\n\n\\-4 -2 0 2 4 6 8\n\n10 12 14 16 18 20 22 24 26 28 30\n\nGender wage gap (%)\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\nFemale wage workers (%)\n\nD1 D2 D3 D4 D5 D6 D7 D8 D9 D10\n\n\\-4 -2 0 2 4 6 8\n\n10 12 14 16 18 20 22 24 26 28 30\n\nGender wage gap (%)\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\nFemale wage workers (%)\n\nD1 D2 D3 D4 D5 D6 D7 D8 D9 D10\n\nDeciles of the hourly wage distribution\n\nDeciles of the hourly wage distribution\n\n\\-4 -2 0 2 4 6 8\n\n10 12 14 16 18 20 22 24 26 28 30\n\nGender wage gap (%)\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\nFemale wage workers (%)\n\nD1 D2 D3 D4 D5 D6 D7 D8 D9 D10\n\nDeciles of the hourly wage distribution\n\n**Low** \\- **income countries Lower** \\- **middle** \\- **income countries**\n\n**Upper** \\- **middle** \\- **income countries**\n\n\\-4 -2 0 2 4 6 8\n\n10 12 14 16 18 20 22 24 26 28 30\n\nGender wage gap (%)\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\nFemale wage workers (%)\n\nD1 D2 D3 D4 D5 D6 D7 D8 D9 D10\n\nDeciles of the hourly wage distribution\n\n**High** \\- **income countries**\n\nMean gender wage gap by decile Median gender wage gap by decile Share of women by decile\n\n# 38\n\n**Global Wage Report 2024–25** 5\\. Wage inequality, where do we stand?","pages":[{"line_numbers":[],"markdown":"high at the bottom half of the distribution, which is again probably where informal em- ployment among women dominates. When comparing country income groups on a de- cile-by-decile basis, the highest gender wage gaps tend to be found in the low-income countries – with the exception of the bottom decile, where the gender wage gap is higher among lower-middle-income countries.\n\nTable 5.1 shows the gender wage gaps esti- mated among all wage workers in the popu- lation for each country income group using the classic measures for estimating pay difference – namely, the raw mean and median gender wage gaps. The two measure are presented together with estimates of the mean and me- dian factor-weighted gender wage gap, com- plementary measures that can help better\n\nX **Figure 5.3. Share of women among wage employees in each decile of the hourly wage**\n\n**distribution and the decile-specific gender wage gap by country income group, latest available year** (percentage)\n\n**Notes:** Each of the four charts shows two vertical axes: the left-hand axis displays values for the share of women among wage workers in each decile, while the right-hand axis displays values for the decile-specific mean and median gender wage gaps. All three estimates in each chart by decile are based on the weighted value between all of the countries in the sample from the corresponding income group, with weights determined by the size of wage employment in each country and decile. The figures displayed in the charts were obtained using data from the most recent available survey (T1 in Appendix I).\n\n**Source:** ILO estimates based on national survey data from 82 countries. See Appendix I for the sources of data and year of collection.\n\n\\-4 -2 0 2 4 6 8\n\n10 12 14 16 18 20 22 24 26 28 30\n\nGender wage gap (%)\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\nFemale wage workers (%)\n\nD1 D2 D3 D4 D5 D6 D7 D8 D9 D10\n\n\\-4 -2 0 2 4 6 8\n\n10 12 14 16 18 20 22 24 26 28 30\n\nGender wage gap (%)\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\nFemale wage workers (%)\n\nD1 D2 D3 D4 D5 D6 D7 D8 D9 D10\n\nDeciles of the hourly wage distribution\n\nDeciles of the hourly wage distribution\n\n\\-4 -2 0 2 4 6 8\n\n10 12 14 16 18 20 22 24 26 28 30\n\nGender wage gap (%)\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\nFemale wage workers (%)\n\nD1 D2 D3 D4 D5 D6 D7 D8 D9 D10\n\nDeciles of the hourly wage distribution\n\n**Low** \\- **income countries Lower** \\- **middle** \\- **income countries**\n\n**Upper** \\- **middle** \\- **income countries**\n\n\\-4 -2 0 2 4 6 8\n\n10 12 14 16 18 20 22 24 26 28 30\n\nGender wage gap (%)\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\nFemale wage workers (%)\n\nD1 D2 D3 D4 D5 D6 D7 D8 D9 D10\n\nDeciles of the hourly wage distribution\n\n**High** \\- **income countries**\n\nMean gender wage gap by decile Median gender wage gap by decile Share of women by decile\n\n# 38\n\n**Global Wage Report 2024–25** 5\\. Wage inequality, where do we stand?","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"high at the bottom half of the distribution, which is again probably where informal em- ployment among women dominates. When comparing country income groups on a de- cile-by-decile basis, the highest gender wage gaps tend to be found in the low-income countries – with the exception of the bottom decile, where the gender wage gap is higher among lower-middle-income countries.\n\nTable 5.1 shows the gender wage gaps esti- mated among all wage workers in the popu- lation for each country income group using the classic measures for estimating pay difference – namely, the raw mean and median gender wage gaps. The two measure are presented together with estimates of the mean and me- dian factor-weighted gender wage gap, com- plementary measures that can help better\n\nX **Figure 5.3. Share of women among wage employees in each decile of the hourly wage**\n\n**distribution and the decile-specific gender wage gap by country income group, latest available year** (percentage)\n\n**Notes:** Each of the four charts shows two vertical axes: the left-hand axis displays values for the share of women among wage workers in each decile, while the right-hand axis displays values for the decile-specific mean and median gender wage gaps. All three estimates in each chart by decile are based on the weighted value between all of the countries in the sample from the corresponding income group, with weights determined by the size of wage employment in each country and decile. The figures displayed in the charts were obtained using data from the most recent available survey (T1 in Appendix I).\n\n**Source:** ILO estimates based on national survey data from 82 countries. See Appendix I for the sources of data and year of collection.\n\n\\-4 -2 0 2 4 6 8\n\n10 12 14 16 18 20 22 24 26 28 30\n\nGender wage gap (%)\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\nFemale wage workers (%)\n\nD1 D2 D3 D4 D5 D6 D7 D8 D9 D10\n\n\\-4 -2 0 2 4 6 8\n\n10 12 14 16 18 20 22 24 26 28 30\n\nGender wage gap (%)\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\nFemale wage workers (%)\n\nD1 D2 D3 D4 D5 D6 D7 D8 D9 D10\n\nDeciles of the hourly wage distribution\n\nDeciles of the hourly wage distribution\n\n\\-4 -2 0 2 4 6 8\n\n10 12 14 16 18 20 22 24 26 28 30\n\nGender wage gap (%)\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\nFemale wage workers (%)\n\nD1 D2 D3 D4 D5 D6 D7 D8 D9 D10\n\nDeciles of the hourly wage distribution\n\n**Low** \\- **income countries Lower** \\- **middle** \\- **income countries**\n\n**Upper** \\- **middle** \\- **income countries**\n\n\\-4 -2 0 2 4 6 8\n\n10 12 14 16 18 20 22 24 26 28 30\n\nGender wage gap (%)\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\nFemale wage workers (%)\n\nD1 D2 D3 D4 D5 D6 D7 D8 D9 D10\n\nDeciles of the hourly wage distribution\n\n**High** \\- **income countries**\n\nMean gender wage gap by decile Median gender wage gap by decile Share of women by decile\n\n# 38\n\n**Global Wage Report 2024–25** 5\\. Wage inequality, where do we stand?","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/GWR-2024_Layout_E_RGB_Web_p58.pdf"]},"response":{"errors":[],"extract_id":"extract_b8433169b4c09a04a0491af5e5bb6fa0","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["high at the bottom half of the distribution, which is again probably where informal em- ployment among women dominates. When comparing country income groups on a de- cile-by-decile basis, the highest gender wage gaps tend to be found in the low-income countries – with the exception of the bottom decile, where the gender wage gap is higher among lower-middle-income countries.\n\nTable 5.1 shows the gender wage gaps esti- mated among all wage workers in the popu- lation for each country income group using the classic measures for estimating pay difference – namely, the raw mean and median gender wage gaps. The two measure are presented together with estimates of the mean and me- dian factor-weighted gender wage gap, com- plementary measures that can help better\n\nX **Figure 5.3. Share of women among wage employees in each decile of the hourly wage**\n\n**distribution and the decile-specific gender wage gap by country income group, latest available year** (percentage)\n\n**Notes:** Each of the four charts shows two vertical axes: the left-hand axis displays values for the share of women among wage workers in each decile, while the right-hand axis displays values for the decile-specific mean and median gender wage gaps. All three estimates in each chart by decile are based on the weighted value between all of the countries in the sample from the corresponding income group, with weights determined by the size of wage employment in each country and decile. The figures displayed in the charts were obtained using data from the most recent available survey (T1 in Appendix I).\n\n**Source:** ILO estimates based on national survey data from 82 countries. See Appendix I for the sources of data and year of collection.\n\n\\-4 -2 0 2 4 6 8\n\n10 12 14 16 18 20 22 24 26 28 30\n\nGender wage gap (%)\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\nFemale wage workers (%)\n\nD1 D2 D3 D4 D5 D6 D7 D8 D9 D10\n\n\\-4 -2 0 2 4 6 8\n\n10 12 14 16 18 20 22 24 26 28 30\n\nGender wage gap (%)\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\nFemale wage workers (%)\n\nD1 D2 D3 D4 D5 D6 D7 D8 D9 D10\n\nDeciles of the hourly wage distribution\n\nDeciles of the hourly wage distribution\n\n\\-4 -2 0 2 4 6 8\n\n10 12 14 16 18 20 22 24 26 28 30\n\nGender wage gap (%)\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\nFemale wage workers (%)\n\nD1 D2 D3 D4 D5 D6 D7 D8 D9 D10\n\nDeciles of the hourly wage distribution\n\n**Low** \\- **income countries Lower** \\- **middle** \\- **income countries**\n\n**Upper** \\- **middle** \\- **income countries**\n\n\\-4 -2 0 2 4 6 8\n\n10 12 14 16 18 20 22 24 26 28 30\n\nGender wage gap (%)\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\nFemale wage workers (%)\n\nD1 D2 D3 D4 D5 D6 D7 D8 D9 D10\n\nDeciles of the hourly wage distribution\n\n**High** \\- **income countries**\n\nMean gender wage gap by decile Median gender wage gap by decile Share of women by decile\n\n# 38\n\n**Global Wage Report 2024–25** 5\\. Wage inequality, where do we stand?"],"full_content":"high at the bottom half of the distribution, which is again probably where informal em- ployment among women dominates. When comparing country income groups on a de- cile-by-decile basis, the highest gender wage gaps tend to be found in the low-income countries – with the exception of the bottom decile, where the gender wage gap is higher among lower-middle-income countries.\n\nTable 5.1 shows the gender wage gaps esti- mated among all wage workers in the popu- lation for each country income group using the classic measures for estimating pay difference – namely, the raw mean and median gender wage gaps. The two measure are presented together with estimates of the mean and me- dian factor-weighted gender wage gap, com- plementary measures that can help better\n\nX **Figure 5.3. Share of women among wage employees in each decile of the hourly wage**\n\n**distribution and the decile-specific gender wage gap by country income group, latest available year** (percentage)\n\n**Notes:** Each of the four charts shows two vertical axes: the left-hand axis displays values for the share of women among wage workers in each decile, while the right-hand axis displays values for the decile-specific mean and median gender wage gaps. All three estimates in each chart by decile are based on the weighted value between all of the countries in the sample from the corresponding income group, with weights determined by the size of wage employment in each country and decile. The figures displayed in the charts were obtained using data from the most recent available survey (T1 in Appendix I).\n\n**Source:** ILO estimates based on national survey data from 82 countries. See Appendix I for the sources of data and year of collection.\n\n\\-4 -2 0 2 4 6 8\n\n10 12 14 16 18 20 22 24 26 28 30\n\nGender wage gap (%)\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\nFemale wage workers (%)\n\nD1 D2 D3 D4 D5 D6 D7 D8 D9 D10\n\n\\-4 -2 0 2 4 6 8\n\n10 12 14 16 18 20 22 24 26 28 30\n\nGender wage gap (%)\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\nFemale wage workers (%)\n\nD1 D2 D3 D4 D5 D6 D7 D8 D9 D10\n\nDeciles of the hourly wage distribution\n\nDeciles of the hourly wage distribution\n\n\\-4 -2 0 2 4 6 8\n\n10 12 14 16 18 20 22 24 26 28 30\n\nGender wage gap (%)\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\nFemale wage workers (%)\n\nD1 D2 D3 D4 D5 D6 D7 D8 D9 D10\n\nDeciles of the hourly wage distribution\n\n**Low** \\- **income countries Lower** \\- **middle** \\- **income countries**\n\n**Upper** \\- **middle** \\- **income countries**\n\n\\-4 -2 0 2 4 6 8\n\n10 12 14 16 18 20 22 24 26 28 30\n\nGender wage gap (%)\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\nFemale wage workers (%)\n\nD1 D2 D3 D4 D5 D6 D7 D8 D9 D10\n\nDeciles of the hourly wage distribution\n\n**High** \\- **income countries**\n\nMean gender wage gap by decile Median gender wage gap by decile Share of women by decile\n\n# 38\n\n**Global Wage Report 2024–25** 5\\. Wage inequality, where do we stand?","publish_date":null,"title":"38","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/GWR-2024_Layout_E_RGB_Web_p58.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/GWR-2024_Layout_E_RGB_Web_p58.pdf"},"request":{"example_id":"chart/GWR-2024_Layout_E_RGB_Web_p58","product_type":"parse","source_file_path":"GWR-2024_Layout_E_RGB_Web_p58.pdf"},"started_at":"2026-11-04T19:45:25.584178"} +{"completed_at":"2026-11-04T19:45:29.674003","latency_in_ms":2912,"output":{"example_id":"chart/GWR-2024_Layout_E_RGB_Web_p75","grounded_pages":[],"layout_pages":[],"markdown":"X **Figure 6.6. Average annualized real wage growth by percentile of the wage distribution**\n\n**between 2006 and 2021 (or closest available year), by country income group** (percentage)\n\n**Note:** See box 6.1 for explanation of how these charts were created.\n\n**Source:** ILO estimates based on national survey data from 72 countries. See Appendix I for the sources of data and years of collection. The figures displayed in the charts were obtained using data from the earlier survey (T0 in Appendix I) and the most recent available survey (T1 in Appendix I).\n\n\\-15\n\n\\-10\n\n\\-5\n\n0\n\n5\n\nAverage annualized real wage growth (%)\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n35\n\n40\n\n45\n\n50\n\n55\n\n60\n\n65\n\n70\n\n75\n\n80\n\n85\n\n90\n\n95\n\n96-99\n\n100\n\nPercentile of the wage distribution\n\n**Low-income countries**\n\n\\-2\n\n\\-1\n\n0\n\n1\n\n2\n\n3\n\nAverage annualized real wage growth (%)\n\n**Lower-middle-income countries**\n\n0\n\n1\n\n2\n\n3\n\n4\n\nAverage annualized real wage growth (%)\n\n**Upper-middle-income countries**\n\n0\\.5\n\n1\n\n1\\.5\n\n2\n\nAverage annualized real wage growth (%)\n\n**High-income countries**\n\n5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 96-99 100\n\nPercentile of the wage distribution\n\n5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 96-99 100\n\nPercentile of the wage distribution\n\n5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 96-99 100\n\nPercentile of the wage distribution\n\n# 55\n\n**Global Wage Report 2024–25**\n\n6\\. The evolution of wage inequality over time","pages":[{"line_numbers":[],"markdown":"X **Figure 6.6. Average annualized real wage growth by percentile of the wage distribution**\n\n**between 2006 and 2021 (or closest available year), by country income group** (percentage)\n\n**Note:** See box 6.1 for explanation of how these charts were created.\n\n**Source:** ILO estimates based on national survey data from 72 countries. See Appendix I for the sources of data and years of collection. The figures displayed in the charts were obtained using data from the earlier survey (T0 in Appendix I) and the most recent available survey (T1 in Appendix I).\n\n\\-15\n\n\\-10\n\n\\-5\n\n0\n\n5\n\nAverage annualized real wage growth (%)\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n35\n\n40\n\n45\n\n50\n\n55\n\n60\n\n65\n\n70\n\n75\n\n80\n\n85\n\n90\n\n95\n\n96-99\n\n100\n\nPercentile of the wage distribution\n\n**Low-income countries**\n\n\\-2\n\n\\-1\n\n0\n\n1\n\n2\n\n3\n\nAverage annualized real wage growth (%)\n\n**Lower-middle-income countries**\n\n0\n\n1\n\n2\n\n3\n\n4\n\nAverage annualized real wage growth (%)\n\n**Upper-middle-income countries**\n\n0\\.5\n\n1\n\n1\\.5\n\n2\n\nAverage annualized real wage growth (%)\n\n**High-income countries**\n\n5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 96-99 100\n\nPercentile of the wage distribution\n\n5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 96-99 100\n\nPercentile of the wage distribution\n\n5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 96-99 100\n\nPercentile of the wage distribution\n\n# 55\n\n**Global Wage Report 2024–25**\n\n6\\. The evolution of wage inequality over time","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"X **Figure 6.6. Average annualized real wage growth by percentile of the wage distribution**\n\n**between 2006 and 2021 (or closest available year), by country income group** (percentage)\n\n**Note:** See box 6.1 for explanation of how these charts were created.\n\n**Source:** ILO estimates based on national survey data from 72 countries. See Appendix I for the sources of data and years of collection. The figures displayed in the charts were obtained using data from the earlier survey (T0 in Appendix I) and the most recent available survey (T1 in Appendix I).\n\n\\-15\n\n\\-10\n\n\\-5\n\n0\n\n5\n\nAverage annualized real wage growth (%)\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n35\n\n40\n\n45\n\n50\n\n55\n\n60\n\n65\n\n70\n\n75\n\n80\n\n85\n\n90\n\n95\n\n96-99\n\n100\n\nPercentile of the wage distribution\n\n**Low-income countries**\n\n\\-2\n\n\\-1\n\n0\n\n1\n\n2\n\n3\n\nAverage annualized real wage growth (%)\n\n**Lower-middle-income countries**\n\n0\n\n1\n\n2\n\n3\n\n4\n\nAverage annualized real wage growth (%)\n\n**Upper-middle-income countries**\n\n0\\.5\n\n1\n\n1\\.5\n\n2\n\nAverage annualized real wage growth (%)\n\n**High-income countries**\n\n5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 96-99 100\n\nPercentile of the wage distribution\n\n5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 96-99 100\n\nPercentile of the wage distribution\n\n5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 96-99 100\n\nPercentile of the wage distribution\n\n# 55\n\n**Global Wage Report 2024–25**\n\n6\\. The evolution of wage inequality over time","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/GWR-2024_Layout_E_RGB_Web_p75.pdf"]},"response":{"errors":[],"extract_id":"extract_434002d271c64c0a0b92ddedcde08417","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["X **Figure 6.6. Average annualized real wage growth by percentile of the wage distribution**\n\n**between 2006 and 2021 (or closest available year), by country income group** (percentage)\n\n**Note:** See box 6.1 for explanation of how these charts were created.\n\n**Source:** ILO estimates based on national survey data from 72 countries. See Appendix I for the sources of data and years of collection. The figures displayed in the charts were obtained using data from the earlier survey (T0 in Appendix I) and the most recent available survey (T1 in Appendix I).\n\n\\-15\n\n\\-10\n\n\\-5\n\n0\n\n5\n\nAverage annualized real wage growth (%)\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n35\n\n40\n\n45\n\n50\n\n55\n\n60\n\n65\n\n70\n\n75\n\n80\n\n85\n\n90\n\n95\n\n96-99\n\n100\n\nPercentile of the wage distribution\n\n**Low-income countries**\n\n\\-2\n\n\\-1\n\n0\n\n1\n\n2\n\n3\n\nAverage annualized real wage growth (%)\n\n**Lower-middle-income countries**\n\n0\n\n1\n\n2\n\n3\n\n4\n\nAverage annualized real wage growth (%)\n\n**Upper-middle-income countries**\n\n0\\.5\n\n1\n\n1\\.5\n\n2\n\nAverage annualized real wage growth (%)\n\n**High-income countries**\n\n5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 96-99 100\n\nPercentile of the wage distribution\n\n5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 96-99 100\n\nPercentile of the wage distribution\n\n5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 96-99 100\n\nPercentile of the wage distribution\n\n# 55\n\n**Global Wage Report 2024–25**\n\n6\\. The evolution of wage inequality over time"],"full_content":"X **Figure 6.6. Average annualized real wage growth by percentile of the wage distribution**\n\n**between 2006 and 2021 (or closest available year), by country income group** (percentage)\n\n**Note:** See box 6.1 for explanation of how these charts were created.\n\n**Source:** ILO estimates based on national survey data from 72 countries. See Appendix I for the sources of data and years of collection. The figures displayed in the charts were obtained using data from the earlier survey (T0 in Appendix I) and the most recent available survey (T1 in Appendix I).\n\n\\-15\n\n\\-10\n\n\\-5\n\n0\n\n5\n\nAverage annualized real wage growth (%)\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n35\n\n40\n\n45\n\n50\n\n55\n\n60\n\n65\n\n70\n\n75\n\n80\n\n85\n\n90\n\n95\n\n96-99\n\n100\n\nPercentile of the wage distribution\n\n**Low-income countries**\n\n\\-2\n\n\\-1\n\n0\n\n1\n\n2\n\n3\n\nAverage annualized real wage growth (%)\n\n**Lower-middle-income countries**\n\n0\n\n1\n\n2\n\n3\n\n4\n\nAverage annualized real wage growth (%)\n\n**Upper-middle-income countries**\n\n0\\.5\n\n1\n\n1\\.5\n\n2\n\nAverage annualized real wage growth (%)\n\n**High-income countries**\n\n5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 96-99 100\n\nPercentile of the wage distribution\n\n5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 96-99 100\n\nPercentile of the wage distribution\n\n5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 96-99 100\n\nPercentile of the wage distribution\n\n# 55\n\n**Global Wage Report 2024–25**\n\n6\\. The evolution of wage inequality over time","publish_date":null,"title":"Global Wage Report 2024–25 55","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/GWR-2024_Layout_E_RGB_Web_p75.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/GWR-2024_Layout_E_RGB_Web_p75.pdf"},"request":{"example_id":"chart/GWR-2024_Layout_E_RGB_Web_p75","product_type":"parse","source_file_path":"GWR-2024_Layout_E_RGB_Web_p75.pdf"},"started_at":"2026-11-04T19:45:26.761985"} +{"completed_at":"2026-11-04T19:45:29.597543","latency_in_ms":1347,"output":{"example_id":"chart/Goldman_Sachs_BUY_report_p10","grounded_pages":[],"layout_pages":[],"markdown":"KENNETH WOODS PORTFOLIO MANAGEMENT PROGRAM\n\n10\n\n### Strategic direction\n\n**The Goldman Sachs Group** GS derives its revenues through a public / private market focused business model which for the most part capitalizes on fees and commissions for services and structured products. As the environment for global investment banks deters, management at GS is focused on targeting the market share of peers who scale back in many markets across the globe. Credit Suisse and Deutsche Bank in particular have begun scaling back their US and European operations in an effort to refocus their businesses and GS is a prime hunter for their share of the market. The Company has publicly stated that in the US market they will look to strengthen their share of the Institutional Client Services market.\n\nBeyond this, the Company has deployed (and is continuing to deploy) a franchise mix strategy that minimizes their exposure to the riskiness of trading revenues in FICC, focusing more on investment banking and investment management (Exhibit 3). investment banking revenues now\n\nmatch that of the FICC business, incomparable to how things were in past years.\n\nOn a geographic level, GS revenues are derived primarily from North America, however on a profit level, the Company has strong earnings power in the\n\na strong positive for the\n\npositioning as their most operationally effective markets also provide\n\nthe greatest future growth avenues (Exhibit 4).\n\n### Exhibit 3\n\nHistorical revenue mix\n\n7% 8% 9% 12% 12% 11%\n\n5% 7% 6% 6% 7% 10%\n\n21%\n\n29% 24% 21% 20% 23%\n\n35%\n\n31%\n\n29% 25% 25% 22%\n\n19% 7% 17% 21% 20% 16%\n\n13% 17% 15% 16% 17% 18%\n\nFY 2010A FY 2011A FY 2012A FY 2013A FY 2014A FY 2015A\n\nUnderwriting Financial advisory Equities\n\nFICC Investing and lending Investment management","pages":[{"line_numbers":[],"markdown":"KENNETH WOODS PORTFOLIO MANAGEMENT PROGRAM\n\n10\n\n### Strategic direction\n\n**The Goldman Sachs Group** GS derives its revenues through a public / private market focused business model which for the most part capitalizes on fees and commissions for services and structured products. As the environment for global investment banks deters, management at GS is focused on targeting the market share of peers who scale back in many markets across the globe. Credit Suisse and Deutsche Bank in particular have begun scaling back their US and European operations in an effort to refocus their businesses and GS is a prime hunter for their share of the market. The Company has publicly stated that in the US market they will look to strengthen their share of the Institutional Client Services market.\n\nBeyond this, the Company has deployed (and is continuing to deploy) a franchise mix strategy that minimizes their exposure to the riskiness of trading revenues in FICC, focusing more on investment banking and investment management (Exhibit 3). investment banking revenues now\n\nmatch that of the FICC business, incomparable to how things were in past years.\n\nOn a geographic level, GS revenues are derived primarily from North America, however on a profit level, the Company has strong earnings power in the\n\na strong positive for the\n\npositioning as their most operationally effective markets also provide\n\nthe greatest future growth avenues (Exhibit 4).\n\n### Exhibit 3\n\nHistorical revenue mix\n\n7% 8% 9% 12% 12% 11%\n\n5% 7% 6% 6% 7% 10%\n\n21%\n\n29% 24% 21% 20% 23%\n\n35%\n\n31%\n\n29% 25% 25% 22%\n\n19% 7% 17% 21% 20% 16%\n\n13% 17% 15% 16% 17% 18%\n\nFY 2010A FY 2011A FY 2012A FY 2013A FY 2014A FY 2015A\n\nUnderwriting Financial advisory Equities\n\nFICC Investing and lending Investment management","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"KENNETH WOODS PORTFOLIO MANAGEMENT PROGRAM\n\n10\n\n### Strategic direction\n\n**The Goldman Sachs Group** GS derives its revenues through a public / private market focused business model which for the most part capitalizes on fees and commissions for services and structured products. As the environment for global investment banks deters, management at GS is focused on targeting the market share of peers who scale back in many markets across the globe. Credit Suisse and Deutsche Bank in particular have begun scaling back their US and European operations in an effort to refocus their businesses and GS is a prime hunter for their share of the market. The Company has publicly stated that in the US market they will look to strengthen their share of the Institutional Client Services market.\n\nBeyond this, the Company has deployed (and is continuing to deploy) a franchise mix strategy that minimizes their exposure to the riskiness of trading revenues in FICC, focusing more on investment banking and investment management (Exhibit 3). investment banking revenues now\n\nmatch that of the FICC business, incomparable to how things were in past years.\n\nOn a geographic level, GS revenues are derived primarily from North America, however on a profit level, the Company has strong earnings power in the\n\na strong positive for the\n\npositioning as their most operationally effective markets also provide\n\nthe greatest future growth avenues (Exhibit 4).\n\n### Exhibit 3\n\nHistorical revenue mix\n\n7% 8% 9% 12% 12% 11%\n\n5% 7% 6% 6% 7% 10%\n\n21%\n\n29% 24% 21% 20% 23%\n\n35%\n\n31%\n\n29% 25% 25% 22%\n\n19% 7% 17% 21% 20% 16%\n\n13% 17% 15% 16% 17% 18%\n\nFY 2010A FY 2011A FY 2012A FY 2013A FY 2014A FY 2015A\n\nUnderwriting Financial advisory Equities\n\nFICC Investing and lending Investment management","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Goldman_Sachs_BUY_report_p10.pdf"]},"response":{"errors":[],"extract_id":"extract_0975098c6a63f2b21d25bc0f077576d5","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["KENNETH WOODS PORTFOLIO MANAGEMENT PROGRAM\n\n10\n\n### Strategic direction\n\n**The Goldman Sachs Group** GS derives its revenues through a public / private market focused business model which for the most part capitalizes on fees and commissions for services and structured products. As the environment for global investment banks deters, management at GS is focused on targeting the market share of peers who scale back in many markets across the globe. Credit Suisse and Deutsche Bank in particular have begun scaling back their US and European operations in an effort to refocus their businesses and GS is a prime hunter for their share of the market. The Company has publicly stated that in the US market they will look to strengthen their share of the Institutional Client Services market.\n\nBeyond this, the Company has deployed (and is continuing to deploy) a franchise mix strategy that minimizes their exposure to the riskiness of trading revenues in FICC, focusing more on investment banking and investment management (Exhibit 3). investment banking revenues now\n\nmatch that of the FICC business, incomparable to how things were in past years.\n\nOn a geographic level, GS revenues are derived primarily from North America, however on a profit level, the Company has strong earnings power in the\n\na strong positive for the\n\npositioning as their most operationally effective markets also provide\n\nthe greatest future growth avenues (Exhibit 4).\n\n### Exhibit 3\n\nHistorical revenue mix\n\n7% 8% 9% 12% 12% 11%\n\n5% 7% 6% 6% 7% 10%\n\n21%\n\n29% 24% 21% 20% 23%\n\n35%\n\n31%\n\n29% 25% 25% 22%\n\n19% 7% 17% 21% 20% 16%\n\n13% 17% 15% 16% 17% 18%\n\nFY 2010A FY 2011A FY 2012A FY 2013A FY 2014A FY 2015A\n\nUnderwriting Financial advisory Equities\n\nFICC Investing and lending Investment management"],"full_content":"KENNETH WOODS PORTFOLIO MANAGEMENT PROGRAM\n\n10\n\n### Strategic direction\n\n**The Goldman Sachs Group** GS derives its revenues through a public / private market focused business model which for the most part capitalizes on fees and commissions for services and structured products. As the environment for global investment banks deters, management at GS is focused on targeting the market share of peers who scale back in many markets across the globe. Credit Suisse and Deutsche Bank in particular have begun scaling back their US and European operations in an effort to refocus their businesses and GS is a prime hunter for their share of the market. The Company has publicly stated that in the US market they will look to strengthen their share of the Institutional Client Services market.\n\nBeyond this, the Company has deployed (and is continuing to deploy) a franchise mix strategy that minimizes their exposure to the riskiness of trading revenues in FICC, focusing more on investment banking and investment management (Exhibit 3). investment banking revenues now\n\nmatch that of the FICC business, incomparable to how things were in past years.\n\nOn a geographic level, GS revenues are derived primarily from North America, however on a profit level, the Company has strong earnings power in the\n\na strong positive for the\n\npositioning as their most operationally effective markets also provide\n\nthe greatest future growth avenues (Exhibit 4).\n\n### Exhibit 3\n\nHistorical revenue mix\n\n7% 8% 9% 12% 12% 11%\n\n5% 7% 6% 6% 7% 10%\n\n21%\n\n29% 24% 21% 20% 23%\n\n35%\n\n31%\n\n29% 25% 25% 22%\n\n19% 7% 17% 21% 20% 16%\n\n13% 17% 15% 16% 17% 18%\n\nFY 2010A FY 2011A FY 2012A FY 2013A FY 2014A FY 2015A\n\nUnderwriting Financial advisory Equities\n\nFICC Investing and lending Investment management","publish_date":null,"title":"Strategic direction Exhibit 3","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Goldman_Sachs_BUY_report_p10.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Goldman_Sachs_BUY_report_p10.pdf"},"request":{"example_id":"chart/Goldman_Sachs_BUY_report_p10","product_type":"parse","source_file_path":"Goldman_Sachs_BUY_report_p10.pdf"},"started_at":"2026-11-04T19:45:28.249840"} +{"completed_at":"2026-11-04T19:45:30.738826","latency_in_ms":1298,"output":{"example_id":"chart/Goldman_Sachs_BUY_report_p6","grounded_pages":[],"layout_pages":[],"markdown":"KENNETH WOODS PORTFOLIO MANAGEMENT PROGRAM\n\n6\n\n### Business overview and drivers\n\n### Exhibit 1 GS revenue driver dashboard\n\nF1: IB: DCM revenues F2: IB: ECM revenues\n\nF3: IB: Financial advisory revenues F4: IB: Overall investment banking revenues\n\nF5: ICS: FICC revenues F6: ICS: Equities revenues\n\n25\n\n75\n\n125\n\n175\n\n225\n\nPerformance indexed to 100\n\nGlobal M&A deal count Global M&A deal volume\n\nGS Financial advisory revenue\n\n\\--\n\n100\n\n200\n\n300\n\n400\n\nPerformance indexed to 100\n\nGlobal IPO deal count Global IPO value\n\nGS equity underwriting revenue\n\n50\n\n70\n\n90\n\n110\n\n130\n\n150\n\n170\n\nPerformance indexed to 100\n\nUS corporate profits GS investment banking revenues\n\n40\n\n90\n\n140\n\n190\n\n240\n\nFY2010 FY2011 FY2012 FY2013 FY2014 FY2015\n\nPerformance indexed to 100\n\nS&P 500 volume Vix index GS ICS equity revenues\n\n\\--\n\n50\n\n100\n\n150\n\n200\n\nFY2010 FY2011 FY2012 FY2013 FY2014 FY2015\n\nPerformance indexed to 100\n\nInterest rate (US 10yr) High yield spread (inverted)\n\nGS FICC revenues\n\n\\--\n\n50\n\n100\n\n150\n\n200\n\nFY2012 FY2013 FY2014 FY2015\n\nPerformance indexed to 100\n\nDCM deal value Deal count GS debt underwriting revenues","pages":[{"line_numbers":[],"markdown":"KENNETH WOODS PORTFOLIO MANAGEMENT PROGRAM\n\n6\n\n### Business overview and drivers\n\n### Exhibit 1 GS revenue driver dashboard\n\nF1: IB: DCM revenues F2: IB: ECM revenues\n\nF3: IB: Financial advisory revenues F4: IB: Overall investment banking revenues\n\nF5: ICS: FICC revenues F6: ICS: Equities revenues\n\n25\n\n75\n\n125\n\n175\n\n225\n\nPerformance indexed to 100\n\nGlobal M&A deal count Global M&A deal volume\n\nGS Financial advisory revenue\n\n\\--\n\n100\n\n200\n\n300\n\n400\n\nPerformance indexed to 100\n\nGlobal IPO deal count Global IPO value\n\nGS equity underwriting revenue\n\n50\n\n70\n\n90\n\n110\n\n130\n\n150\n\n170\n\nPerformance indexed to 100\n\nUS corporate profits GS investment banking revenues\n\n40\n\n90\n\n140\n\n190\n\n240\n\nFY2010 FY2011 FY2012 FY2013 FY2014 FY2015\n\nPerformance indexed to 100\n\nS&P 500 volume Vix index GS ICS equity revenues\n\n\\--\n\n50\n\n100\n\n150\n\n200\n\nFY2010 FY2011 FY2012 FY2013 FY2014 FY2015\n\nPerformance indexed to 100\n\nInterest rate (US 10yr) High yield spread (inverted)\n\nGS FICC revenues\n\n\\--\n\n50\n\n100\n\n150\n\n200\n\nFY2012 FY2013 FY2014 FY2015\n\nPerformance indexed to 100\n\nDCM deal value Deal count GS debt underwriting revenues","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"KENNETH WOODS PORTFOLIO MANAGEMENT PROGRAM\n\n6\n\n### Business overview and drivers\n\n### Exhibit 1 GS revenue driver dashboard\n\nF1: IB: DCM revenues F2: IB: ECM revenues\n\nF3: IB: Financial advisory revenues F4: IB: Overall investment banking revenues\n\nF5: ICS: FICC revenues F6: ICS: Equities revenues\n\n25\n\n75\n\n125\n\n175\n\n225\n\nPerformance indexed to 100\n\nGlobal M&A deal count Global M&A deal volume\n\nGS Financial advisory revenue\n\n\\--\n\n100\n\n200\n\n300\n\n400\n\nPerformance indexed to 100\n\nGlobal IPO deal count Global IPO value\n\nGS equity underwriting revenue\n\n50\n\n70\n\n90\n\n110\n\n130\n\n150\n\n170\n\nPerformance indexed to 100\n\nUS corporate profits GS investment banking revenues\n\n40\n\n90\n\n140\n\n190\n\n240\n\nFY2010 FY2011 FY2012 FY2013 FY2014 FY2015\n\nPerformance indexed to 100\n\nS&P 500 volume Vix index GS ICS equity revenues\n\n\\--\n\n50\n\n100\n\n150\n\n200\n\nFY2010 FY2011 FY2012 FY2013 FY2014 FY2015\n\nPerformance indexed to 100\n\nInterest rate (US 10yr) High yield spread (inverted)\n\nGS FICC revenues\n\n\\--\n\n50\n\n100\n\n150\n\n200\n\nFY2012 FY2013 FY2014 FY2015\n\nPerformance indexed to 100\n\nDCM deal value Deal count GS debt underwriting revenues","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Goldman_Sachs_BUY_report_p6.pdf"]},"response":{"errors":[],"extract_id":"extract_35bff2696ded31914179f585b6bee48c","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["KENNETH WOODS PORTFOLIO MANAGEMENT PROGRAM\n\n6\n\n### Business overview and drivers\n\n### Exhibit 1 GS revenue driver dashboard\n\nF1: IB: DCM revenues F2: IB: ECM revenues\n\nF3: IB: Financial advisory revenues F4: IB: Overall investment banking revenues\n\nF5: ICS: FICC revenues F6: ICS: Equities revenues\n\n25\n\n75\n\n125\n\n175\n\n225\n\nPerformance indexed to 100\n\nGlobal M&A deal count Global M&A deal volume\n\nGS Financial advisory revenue\n\n\\--\n\n100\n\n200\n\n300\n\n400\n\nPerformance indexed to 100\n\nGlobal IPO deal count Global IPO value\n\nGS equity underwriting revenue\n\n50\n\n70\n\n90\n\n110\n\n130\n\n150\n\n170\n\nPerformance indexed to 100\n\nUS corporate profits GS investment banking revenues\n\n40\n\n90\n\n140\n\n190\n\n240\n\nFY2010 FY2011 FY2012 FY2013 FY2014 FY2015\n\nPerformance indexed to 100\n\nS&P 500 volume Vix index GS ICS equity revenues\n\n\\--\n\n50\n\n100\n\n150\n\n200\n\nFY2010 FY2011 FY2012 FY2013 FY2014 FY2015\n\nPerformance indexed to 100\n\nInterest rate (US 10yr) High yield spread (inverted)\n\nGS FICC revenues\n\n\\--\n\n50\n\n100\n\n150\n\n200\n\nFY2012 FY2013 FY2014 FY2015\n\nPerformance indexed to 100\n\nDCM deal value Deal count GS debt underwriting revenues"],"full_content":"KENNETH WOODS PORTFOLIO MANAGEMENT PROGRAM\n\n6\n\n### Business overview and drivers\n\n### Exhibit 1 GS revenue driver dashboard\n\nF1: IB: DCM revenues F2: IB: ECM revenues\n\nF3: IB: Financial advisory revenues F4: IB: Overall investment banking revenues\n\nF5: ICS: FICC revenues F6: ICS: Equities revenues\n\n25\n\n75\n\n125\n\n175\n\n225\n\nPerformance indexed to 100\n\nGlobal M&A deal count Global M&A deal volume\n\nGS Financial advisory revenue\n\n\\--\n\n100\n\n200\n\n300\n\n400\n\nPerformance indexed to 100\n\nGlobal IPO deal count Global IPO value\n\nGS equity underwriting revenue\n\n50\n\n70\n\n90\n\n110\n\n130\n\n150\n\n170\n\nPerformance indexed to 100\n\nUS corporate profits GS investment banking revenues\n\n40\n\n90\n\n140\n\n190\n\n240\n\nFY2010 FY2011 FY2012 FY2013 FY2014 FY2015\n\nPerformance indexed to 100\n\nS&P 500 volume Vix index GS ICS equity revenues\n\n\\--\n\n50\n\n100\n\n150\n\n200\n\nFY2010 FY2011 FY2012 FY2013 FY2014 FY2015\n\nPerformance indexed to 100\n\nInterest rate (US 10yr) High yield spread (inverted)\n\nGS FICC revenues\n\n\\--\n\n50\n\n100\n\n150\n\n200\n\nFY2012 FY2013 FY2014 FY2015\n\nPerformance indexed to 100\n\nDCM deal value Deal count GS debt underwriting revenues","publish_date":null,"title":"Business overview and drivers Exhibit 1 GS revenue driver dashboard","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Goldman_Sachs_BUY_report_p6.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Goldman_Sachs_BUY_report_p6.pdf"},"request":{"example_id":"chart/Goldman_Sachs_BUY_report_p6","product_type":"parse","source_file_path":"Goldman_Sachs_BUY_report_p6.pdf"},"started_at":"2026-11-04T19:45:29.440482"} +{"completed_at":"2026-11-04T19:45:31.537228","latency_in_ms":1259,"output":{"example_id":"chart/MonthlyTreasuryStatement_202411_p8","grounded_pages":[],"layout_pages":[],"markdown":"8 _Source data: Table 9_\n\n$125B\n\n$129B\n\n$63B $80B\n\n$76B\n\n$79B\n\n$117B\n\n**Month of November 2024: $669B**\n\n$250B\n\n$206B\n\n$108B\n\n$165B\n\n$179B\n\n$160B\n\n$185B\n\n**Fiscal Year 2025 to Date: $1,253B**\n\n**2024 2025**\n\nOct\n\nNov\n\nDec\n\nJan\n\nFeb\n\nMar\n\nApr\n\nMay\n\nJun\n\nJul\n\nAug\n\nSep\n\nOct\n\nNov\n\n$0B $100B\n\n$200B\n\n$300B\n\n$400B\n\n$500B\n\n$600B\n\n$700B\n\n$800B\n\n$900B\n\n$1,000B\n\nOutlay Amount\n\n**Figure 7. Outlays of the U.S. Government, by Source, Fiscal Year 2025**\n\nSocial Security Income Security National Defense\n\nNet Interest\n\nOther\n\nMedicare Health\n\n\\*Disclaimer: Other category is subject to change due to negative balances held.\n\n**Figure 8. Monthly Outlays of the U.S. Government, by Source, Fiscal Years 2024 and 2025**\n\nSocial Security\n\nMedicare\n\nIncome Security\n\nHealth\n\nNational Defense\n\nNet Interest\n\nOther\n\n\"Other\" Includes: Administration of Justice, Agriculture, Community and Regional Development, Education, Training, Employment, and Social Services, Energy, General Government, General Science, Space, and Technology, International Affairs, Natural Resources and Environment, Transportation, Undistributed Offsetting Receipts, Veterans' Benefits and Services. Note: Details may not add to totals due to rounding","pages":[{"line_numbers":[],"markdown":"8 _Source data: Table 9_\n\n$125B\n\n$129B\n\n$63B $80B\n\n$76B\n\n$79B\n\n$117B\n\n**Month of November 2024: $669B**\n\n$250B\n\n$206B\n\n$108B\n\n$165B\n\n$179B\n\n$160B\n\n$185B\n\n**Fiscal Year 2025 to Date: $1,253B**\n\n**2024 2025**\n\nOct\n\nNov\n\nDec\n\nJan\n\nFeb\n\nMar\n\nApr\n\nMay\n\nJun\n\nJul\n\nAug\n\nSep\n\nOct\n\nNov\n\n$0B $100B\n\n$200B\n\n$300B\n\n$400B\n\n$500B\n\n$600B\n\n$700B\n\n$800B\n\n$900B\n\n$1,000B\n\nOutlay Amount\n\n**Figure 7. Outlays of the U.S. Government, by Source, Fiscal Year 2025**\n\nSocial Security Income Security National Defense\n\nNet Interest\n\nOther\n\nMedicare Health\n\n\\*Disclaimer: Other category is subject to change due to negative balances held.\n\n**Figure 8. Monthly Outlays of the U.S. Government, by Source, Fiscal Years 2024 and 2025**\n\nSocial Security\n\nMedicare\n\nIncome Security\n\nHealth\n\nNational Defense\n\nNet Interest\n\nOther\n\n\"Other\" Includes: Administration of Justice, Agriculture, Community and Regional Development, Education, Training, Employment, and Social Services, Energy, General Government, General Science, Space, and Technology, International Affairs, Natural Resources and Environment, Transportation, Undistributed Offsetting Receipts, Veterans' Benefits and Services. Note: Details may not add to totals due to rounding","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"8 _Source data: Table 9_\n\n$125B\n\n$129B\n\n$63B $80B\n\n$76B\n\n$79B\n\n$117B\n\n**Month of November 2024: $669B**\n\n$250B\n\n$206B\n\n$108B\n\n$165B\n\n$179B\n\n$160B\n\n$185B\n\n**Fiscal Year 2025 to Date: $1,253B**\n\n**2024 2025**\n\nOct\n\nNov\n\nDec\n\nJan\n\nFeb\n\nMar\n\nApr\n\nMay\n\nJun\n\nJul\n\nAug\n\nSep\n\nOct\n\nNov\n\n$0B $100B\n\n$200B\n\n$300B\n\n$400B\n\n$500B\n\n$600B\n\n$700B\n\n$800B\n\n$900B\n\n$1,000B\n\nOutlay Amount\n\n**Figure 7. Outlays of the U.S. Government, by Source, Fiscal Year 2025**\n\nSocial Security Income Security National Defense\n\nNet Interest\n\nOther\n\nMedicare Health\n\n\\*Disclaimer: Other category is subject to change due to negative balances held.\n\n**Figure 8. Monthly Outlays of the U.S. Government, by Source, Fiscal Years 2024 and 2025**\n\nSocial Security\n\nMedicare\n\nIncome Security\n\nHealth\n\nNational Defense\n\nNet Interest\n\nOther\n\n\"Other\" Includes: Administration of Justice, Agriculture, Community and Regional Development, Education, Training, Employment, and Social Services, Energy, General Government, General Science, Space, and Technology, International Affairs, Natural Resources and Environment, Transportation, Undistributed Offsetting Receipts, Veterans' Benefits and Services. Note: Details may not add to totals due to rounding","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/MonthlyTreasuryStatement_202411_p8.pdf"]},"response":{"errors":[],"extract_id":"extract_56f4d1c81a6a369019fcdeb9b0c945e2","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["8 _Source data: Table 9_\n\n$125B\n\n$129B\n\n$63B $80B\n\n$76B\n\n$79B\n\n$117B\n\n**Month of November 2024: $669B**\n\n$250B\n\n$206B\n\n$108B\n\n$165B\n\n$179B\n\n$160B\n\n$185B\n\n**Fiscal Year 2025 to Date: $1,253B**\n\n**2024 2025**\n\nOct\n\nNov\n\nDec\n\nJan\n\nFeb\n\nMar\n\nApr\n\nMay\n\nJun\n\nJul\n\nAug\n\nSep\n\nOct\n\nNov\n\n$0B $100B\n\n$200B\n\n$300B\n\n$400B\n\n$500B\n\n$600B\n\n$700B\n\n$800B\n\n$900B\n\n$1,000B\n\nOutlay Amount\n\n**Figure 7. Outlays of the U.S. Government, by Source, Fiscal Year 2025**\n\nSocial Security Income Security National Defense\n\nNet Interest\n\nOther\n\nMedicare Health\n\n\\*Disclaimer: Other category is subject to change due to negative balances held.\n\n**Figure 8. Monthly Outlays of the U.S. Government, by Source, Fiscal Years 2024 and 2025**\n\nSocial Security\n\nMedicare\n\nIncome Security\n\nHealth\n\nNational Defense\n\nNet Interest\n\nOther\n\n\"Other\" Includes: Administration of Justice, Agriculture, Community and Regional Development, Education, Training, Employment, and Social Services, Energy, General Government, General Science, Space, and Technology, International Affairs, Natural Resources and Environment, Transportation, Undistributed Offsetting Receipts, Veterans' Benefits and Services. Note: Details may not add to totals due to rounding"],"full_content":"8 _Source data: Table 9_\n\n$125B\n\n$129B\n\n$63B $80B\n\n$76B\n\n$79B\n\n$117B\n\n**Month of November 2024: $669B**\n\n$250B\n\n$206B\n\n$108B\n\n$165B\n\n$179B\n\n$160B\n\n$185B\n\n**Fiscal Year 2025 to Date: $1,253B**\n\n**2024 2025**\n\nOct\n\nNov\n\nDec\n\nJan\n\nFeb\n\nMar\n\nApr\n\nMay\n\nJun\n\nJul\n\nAug\n\nSep\n\nOct\n\nNov\n\n$0B $100B\n\n$200B\n\n$300B\n\n$400B\n\n$500B\n\n$600B\n\n$700B\n\n$800B\n\n$900B\n\n$1,000B\n\nOutlay Amount\n\n**Figure 7. Outlays of the U.S. Government, by Source, Fiscal Year 2025**\n\nSocial Security Income Security National Defense\n\nNet Interest\n\nOther\n\nMedicare Health\n\n\\*Disclaimer: Other category is subject to change due to negative balances held.\n\n**Figure 8. Monthly Outlays of the U.S. Government, by Source, Fiscal Years 2024 and 2025**\n\nSocial Security\n\nMedicare\n\nIncome Security\n\nHealth\n\nNational Defense\n\nNet Interest\n\nOther\n\n\"Other\" Includes: Administration of Justice, Agriculture, Community and Regional Development, Education, Training, Employment, and Social Services, Energy, General Government, General Science, Space, and Technology, International Affairs, Natural Resources and Environment, Transportation, Undistributed Offsetting Receipts, Veterans' Benefits and Services. Note: Details may not add to totals due to rounding","publish_date":null,"title":"Figure 7. Outlays of the U.S. Government, by Source, Fiscal Year 2025 Month of November 2024: $669B Fiscal Year 2025","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/MonthlyTreasuryStatement_202411_p8.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/MonthlyTreasuryStatement_202411_p8.pdf"},"request":{"example_id":"chart/MonthlyTreasuryStatement_202411_p8","product_type":"parse","source_file_path":"MonthlyTreasuryStatement_202411_p8.pdf"},"started_at":"2026-11-04T19:45:30.278059"} +{"completed_at":"2026-11-04T19:31:19.337174","latency_in_ms":1729,"output":{"example_id":"chart/MonthlyTreasuryStatement_202509_p6","grounded_pages":[],"layout_pages":[],"markdown":"6 _Source_ _d_ _ata: Table 1_\n\nReceipts Outlays Deficit(-)/Surplus\n\n**2024 2025**\n\n**Oct**\n\n**Nov**\n\n**Dec**\n\n**Jan**\n\n**Feb**\n\n**Mar**\n\n**Apr**\n\n**May**\n\n**Jun**\n\n**Jul**\n\n**Aug**\n\n**Sep**\n\n**Oct**\n\n**Nov**\n\n**Dec**\n\n**Jan**\n\n**Feb**\n\n**Mar**\n\n**Apr**\n\n**May**\n\n**Jun**\n\n**Jul**\n\n**Aug**\n\n**Sep**\n\n($1,000B)\n\n($600B)\n\n($200B)\n\n$200B\n\n$600B\n\n$1,000B\n\n($600B)\n\n($400B)\n\n($200B)\n\n$0B\n\n$200B\n\n**2024 2025**\n\n**Oct**\n\n**Nov**\n\n**Dec**\n\n**Jan**\n\n**Feb**\n\n**Mar**\n\n**Apr**\n\n**May**\n\n**Jun**\n\n**Jul**\n\n**Aug**\n\n**Sep**\n\n**Oct**\n\n**Nov**\n\n**Dec**\n\n**Jan**\n\n**Feb**\n\n**Mar**\n\n**Apr**\n\n**May**\n\n**Jun**\n\n**Jul**\n\n**Aug**\n\n**Sep**\n\n$1,000B $2,000B $3,000B $4,000B $5,000B\n\n($7,000B) ($5,000B)\n\n($3,000B)\n\n($1,000B)\n\n($2,000B)\n\n($1,000B)\n\n$0B\n\n**Figure 4. Monthly Receipts, Outlays, and Budget Deficit/Surplus of the U.S. Government, Cumulative, Fiscal Years 2024 and 2025**\n\nReceipts Outlays Deficit(-)/Surplus\n\n**Figure 3. Monthly Receipts, Outlays, and Budget Deficit/Surplus of the U.S. Government, Fiscal Years 2024 and 2025**\n\n$0B","pages":[{"line_numbers":[],"markdown":"6 _Source_ _d_ _ata: Table 1_\n\nReceipts Outlays Deficit(-)/Surplus\n\n**2024 2025**\n\n**Oct**\n\n**Nov**\n\n**Dec**\n\n**Jan**\n\n**Feb**\n\n**Mar**\n\n**Apr**\n\n**May**\n\n**Jun**\n\n**Jul**\n\n**Aug**\n\n**Sep**\n\n**Oct**\n\n**Nov**\n\n**Dec**\n\n**Jan**\n\n**Feb**\n\n**Mar**\n\n**Apr**\n\n**May**\n\n**Jun**\n\n**Jul**\n\n**Aug**\n\n**Sep**\n\n($1,000B)\n\n($600B)\n\n($200B)\n\n$200B\n\n$600B\n\n$1,000B\n\n($600B)\n\n($400B)\n\n($200B)\n\n$0B\n\n$200B\n\n**2024 2025**\n\n**Oct**\n\n**Nov**\n\n**Dec**\n\n**Jan**\n\n**Feb**\n\n**Mar**\n\n**Apr**\n\n**May**\n\n**Jun**\n\n**Jul**\n\n**Aug**\n\n**Sep**\n\n**Oct**\n\n**Nov**\n\n**Dec**\n\n**Jan**\n\n**Feb**\n\n**Mar**\n\n**Apr**\n\n**May**\n\n**Jun**\n\n**Jul**\n\n**Aug**\n\n**Sep**\n\n$1,000B $2,000B $3,000B $4,000B $5,000B\n\n($7,000B) ($5,000B)\n\n($3,000B)\n\n($1,000B)\n\n($2,000B)\n\n($1,000B)\n\n$0B\n\n**Figure 4. Monthly Receipts, Outlays, and Budget Deficit/Surplus of the U.S. Government, Cumulative, Fiscal Years 2024 and 2025**\n\nReceipts Outlays Deficit(-)/Surplus\n\n**Figure 3. Monthly Receipts, Outlays, and Budget Deficit/Surplus of the U.S. Government, Fiscal Years 2024 and 2025**\n\n$0B","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"6 _Source_ _d_ _ata: Table 1_\n\nReceipts Outlays Deficit(-)/Surplus\n\n**2024 2025**\n\n**Oct**\n\n**Nov**\n\n**Dec**\n\n**Jan**\n\n**Feb**\n\n**Mar**\n\n**Apr**\n\n**May**\n\n**Jun**\n\n**Jul**\n\n**Aug**\n\n**Sep**\n\n**Oct**\n\n**Nov**\n\n**Dec**\n\n**Jan**\n\n**Feb**\n\n**Mar**\n\n**Apr**\n\n**May**\n\n**Jun**\n\n**Jul**\n\n**Aug**\n\n**Sep**\n\n($1,000B)\n\n($600B)\n\n($200B)\n\n$200B\n\n$600B\n\n$1,000B\n\n($600B)\n\n($400B)\n\n($200B)\n\n$0B\n\n$200B\n\n**2024 2025**\n\n**Oct**\n\n**Nov**\n\n**Dec**\n\n**Jan**\n\n**Feb**\n\n**Mar**\n\n**Apr**\n\n**May**\n\n**Jun**\n\n**Jul**\n\n**Aug**\n\n**Sep**\n\n**Oct**\n\n**Nov**\n\n**Dec**\n\n**Jan**\n\n**Feb**\n\n**Mar**\n\n**Apr**\n\n**May**\n\n**Jun**\n\n**Jul**\n\n**Aug**\n\n**Sep**\n\n$1,000B $2,000B $3,000B $4,000B $5,000B\n\n($7,000B) ($5,000B)\n\n($3,000B)\n\n($1,000B)\n\n($2,000B)\n\n($1,000B)\n\n$0B\n\n**Figure 4. Monthly Receipts, Outlays, and Budget Deficit/Surplus of the U.S. Government, Cumulative, Fiscal Years 2024 and 2025**\n\nReceipts Outlays Deficit(-)/Surplus\n\n**Figure 3. Monthly Receipts, Outlays, and Budget Deficit/Surplus of the U.S. Government, Fiscal Years 2024 and 2025**\n\n$0B","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/MonthlyTreasuryStatement_202509_p6.pdf"]},"response":{"errors":[],"extract_id":"extract_9241f418675d8cfe6589766804a9b3cd","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["6 _Source_ _d_ _ata: Table 1_\n\nReceipts Outlays Deficit(-)/Surplus\n\n**2024 2025**\n\n**Oct**\n\n**Nov**\n\n**Dec**\n\n**Jan**\n\n**Feb**\n\n**Mar**\n\n**Apr**\n\n**May**\n\n**Jun**\n\n**Jul**\n\n**Aug**\n\n**Sep**\n\n**Oct**\n\n**Nov**\n\n**Dec**\n\n**Jan**\n\n**Feb**\n\n**Mar**\n\n**Apr**\n\n**May**\n\n**Jun**\n\n**Jul**\n\n**Aug**\n\n**Sep**\n\n($1,000B)\n\n($600B)\n\n($200B)\n\n$200B\n\n$600B\n\n$1,000B\n\n($600B)\n\n($400B)\n\n($200B)\n\n$0B\n\n$200B\n\n**2024 2025**\n\n**Oct**\n\n**Nov**\n\n**Dec**\n\n**Jan**\n\n**Feb**\n\n**Mar**\n\n**Apr**\n\n**May**\n\n**Jun**\n\n**Jul**\n\n**Aug**\n\n**Sep**\n\n**Oct**\n\n**Nov**\n\n**Dec**\n\n**Jan**\n\n**Feb**\n\n**Mar**\n\n**Apr**\n\n**May**\n\n**Jun**\n\n**Jul**\n\n**Aug**\n\n**Sep**\n\n$1,000B $2,000B $3,000B $4,000B $5,000B\n\n($7,000B) ($5,000B)\n\n($3,000B)\n\n($1,000B)\n\n($2,000B)\n\n($1,000B)\n\n$0B\n\n**Figure 4. Monthly Receipts, Outlays, and Budget Deficit/Surplus of the U.S. Government, Cumulative, Fiscal Years 2024 and 2025**\n\nReceipts Outlays Deficit(-)/Surplus\n\n**Figure 3. Monthly Receipts, Outlays, and Budget Deficit/Surplus of the U.S. Government, Fiscal Years 2024 and 2025**\n\n$0B"],"full_content":"6 _Source_ _d_ _ata: Table 1_\n\nReceipts Outlays Deficit(-)/Surplus\n\n**2024 2025**\n\n**Oct**\n\n**Nov**\n\n**Dec**\n\n**Jan**\n\n**Feb**\n\n**Mar**\n\n**Apr**\n\n**May**\n\n**Jun**\n\n**Jul**\n\n**Aug**\n\n**Sep**\n\n**Oct**\n\n**Nov**\n\n**Dec**\n\n**Jan**\n\n**Feb**\n\n**Mar**\n\n**Apr**\n\n**May**\n\n**Jun**\n\n**Jul**\n\n**Aug**\n\n**Sep**\n\n($1,000B)\n\n($600B)\n\n($200B)\n\n$200B\n\n$600B\n\n$1,000B\n\n($600B)\n\n($400B)\n\n($200B)\n\n$0B\n\n$200B\n\n**2024 2025**\n\n**Oct**\n\n**Nov**\n\n**Dec**\n\n**Jan**\n\n**Feb**\n\n**Mar**\n\n**Apr**\n\n**May**\n\n**Jun**\n\n**Jul**\n\n**Aug**\n\n**Sep**\n\n**Oct**\n\n**Nov**\n\n**Dec**\n\n**Jan**\n\n**Feb**\n\n**Mar**\n\n**Apr**\n\n**May**\n\n**Jun**\n\n**Jul**\n\n**Aug**\n\n**Sep**\n\n$1,000B $2,000B $3,000B $4,000B $5,000B\n\n($7,000B) ($5,000B)\n\n($3,000B)\n\n($1,000B)\n\n($2,000B)\n\n($1,000B)\n\n$0B\n\n**Figure 4. Monthly Receipts, Outlays, and Budget Deficit/Surplus of the U.S. Government, Cumulative, Fiscal Years 2024 and 2025**\n\nReceipts Outlays Deficit(-)/Surplus\n\n**Figure 3. Monthly Receipts, Outlays, and Budget Deficit/Surplus of the U.S. Government, Fiscal Years 2024 and 2025**\n\n$0B","publish_date":null,"title":"Figure 3. Monthly Receipts, Outlays, and Budget Deficit/Surplus of the U.S. Government, Fiscal Years 2024 and 2025 2024 2025 $1,000B","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/MonthlyTreasuryStatement_202509_p6.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/MonthlyTreasuryStatement_202509_p6.pdf"},"request":{"example_id":"chart/MonthlyTreasuryStatement_202509_p6","product_type":"parse","source_file_path":"MonthlyTreasuryStatement_202509_p6.pdf"},"started_at":"2026-11-04T19:31:17.607785"} +{"completed_at":"2026-11-04T19:45:34.939792","latency_in_ms":2673,"output":{"example_id":"chart/MonthlyTreasuryStatement_202509_p7","grounded_pages":[],"layout_pages":[],"markdown":"7 _Source data: Table 9_\n\n$48B $224B $106B\n\n$452B\n\n$2,656B\n\n$1,748B\n\n**Fiscal Year 2025 to Date: $5,235B**\n\n**2024 2025**\n\nOct\n\nNov\n\nDec\n\nJan\n\nFeb\n\nMar\n\nApr\n\nMay\n\nJun\n\nJul\n\nAug\n\nSep\n\nOct\n\nNov\n\nDec\n\nJan\n\nFeb\n\nMar\n\nApr\n\nMay\n\nJun\n\nJul\n\nAug\n\nSep\n\n$0B\n\n$100B\n\n$200B\n\n$300B\n\n$400B\n\n$500B\n\n$600B\n\n$700B\n\n$800B\n\nReceipt Amount\n\n$3B $33B\n\n$13B\n\n$62B\n\n$298B\n\n$134B\n\n**Month of September 2025: $544B**\n\n**Figure 5. Receipts of the U.S. Government, by Source, Fiscal Year 2025**\n\n**Figure 6. Monthly Receipts of the U.S. Government, by Source, Fiscal Years 2024 and 2025**\n\nMiscellaneous Receipts\n\nOther\n\nExcise Taxes\n\nCorporation Income Taxes\n\nIndividual Income Taxes\n\nSocial Insurance and Retirement Receipts\n\nMiscellaneous Receipts\n\nOther\n\nExcise Taxes\n\nCorporation Income Taxes\n\nIndividual Income Taxes\n\nSocial Insurance and Retirement Receipts\n\n\"Other\" Includes: Customs Duties and Estate and Gift Taxes \\*Disclaimer: Other category is subject to change due to negative balances held. Note: Details may not add to totals due to rounding","pages":[{"line_numbers":[],"markdown":"7 _Source data: Table 9_\n\n$48B $224B $106B\n\n$452B\n\n$2,656B\n\n$1,748B\n\n**Fiscal Year 2025 to Date: $5,235B**\n\n**2024 2025**\n\nOct\n\nNov\n\nDec\n\nJan\n\nFeb\n\nMar\n\nApr\n\nMay\n\nJun\n\nJul\n\nAug\n\nSep\n\nOct\n\nNov\n\nDec\n\nJan\n\nFeb\n\nMar\n\nApr\n\nMay\n\nJun\n\nJul\n\nAug\n\nSep\n\n$0B\n\n$100B\n\n$200B\n\n$300B\n\n$400B\n\n$500B\n\n$600B\n\n$700B\n\n$800B\n\nReceipt Amount\n\n$3B $33B\n\n$13B\n\n$62B\n\n$298B\n\n$134B\n\n**Month of September 2025: $544B**\n\n**Figure 5. Receipts of the U.S. Government, by Source, Fiscal Year 2025**\n\n**Figure 6. Monthly Receipts of the U.S. Government, by Source, Fiscal Years 2024 and 2025**\n\nMiscellaneous Receipts\n\nOther\n\nExcise Taxes\n\nCorporation Income Taxes\n\nIndividual Income Taxes\n\nSocial Insurance and Retirement Receipts\n\nMiscellaneous Receipts\n\nOther\n\nExcise Taxes\n\nCorporation Income Taxes\n\nIndividual Income Taxes\n\nSocial Insurance and Retirement Receipts\n\n\"Other\" Includes: Customs Duties and Estate and Gift Taxes \\*Disclaimer: Other category is subject to change due to negative balances held. Note: Details may not add to totals due to rounding","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"7 _Source data: Table 9_\n\n$48B $224B $106B\n\n$452B\n\n$2,656B\n\n$1,748B\n\n**Fiscal Year 2025 to Date: $5,235B**\n\n**2024 2025**\n\nOct\n\nNov\n\nDec\n\nJan\n\nFeb\n\nMar\n\nApr\n\nMay\n\nJun\n\nJul\n\nAug\n\nSep\n\nOct\n\nNov\n\nDec\n\nJan\n\nFeb\n\nMar\n\nApr\n\nMay\n\nJun\n\nJul\n\nAug\n\nSep\n\n$0B\n\n$100B\n\n$200B\n\n$300B\n\n$400B\n\n$500B\n\n$600B\n\n$700B\n\n$800B\n\nReceipt Amount\n\n$3B $33B\n\n$13B\n\n$62B\n\n$298B\n\n$134B\n\n**Month of September 2025: $544B**\n\n**Figure 5. Receipts of the U.S. Government, by Source, Fiscal Year 2025**\n\n**Figure 6. Monthly Receipts of the U.S. Government, by Source, Fiscal Years 2024 and 2025**\n\nMiscellaneous Receipts\n\nOther\n\nExcise Taxes\n\nCorporation Income Taxes\n\nIndividual Income Taxes\n\nSocial Insurance and Retirement Receipts\n\nMiscellaneous Receipts\n\nOther\n\nExcise Taxes\n\nCorporation Income Taxes\n\nIndividual Income Taxes\n\nSocial Insurance and Retirement Receipts\n\n\"Other\" Includes: Customs Duties and Estate and Gift Taxes \\*Disclaimer: Other category is subject to change due to negative balances held. Note: Details may not add to totals due to rounding","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/MonthlyTreasuryStatement_202509_p7.pdf"]},"response":{"errors":[],"extract_id":"extract_4112c0a8d8ccf0a91fb5ce6c214cfa14","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["7 _Source data: Table 9_\n\n$48B $224B $106B\n\n$452B\n\n$2,656B\n\n$1,748B\n\n**Fiscal Year 2025 to Date: $5,235B**\n\n**2024 2025**\n\nOct\n\nNov\n\nDec\n\nJan\n\nFeb\n\nMar\n\nApr\n\nMay\n\nJun\n\nJul\n\nAug\n\nSep\n\nOct\n\nNov\n\nDec\n\nJan\n\nFeb\n\nMar\n\nApr\n\nMay\n\nJun\n\nJul\n\nAug\n\nSep\n\n$0B\n\n$100B\n\n$200B\n\n$300B\n\n$400B\n\n$500B\n\n$600B\n\n$700B\n\n$800B\n\nReceipt Amount\n\n$3B $33B\n\n$13B\n\n$62B\n\n$298B\n\n$134B\n\n**Month of September 2025: $544B**\n\n**Figure 5. Receipts of the U.S. Government, by Source, Fiscal Year 2025**\n\n**Figure 6. Monthly Receipts of the U.S. Government, by Source, Fiscal Years 2024 and 2025**\n\nMiscellaneous Receipts\n\nOther\n\nExcise Taxes\n\nCorporation Income Taxes\n\nIndividual Income Taxes\n\nSocial Insurance and Retirement Receipts\n\nMiscellaneous Receipts\n\nOther\n\nExcise Taxes\n\nCorporation Income Taxes\n\nIndividual Income Taxes\n\nSocial Insurance and Retirement Receipts\n\n\"Other\" Includes: Customs Duties and Estate and Gift Taxes \\*Disclaimer: Other category is subject to change due to negative balances held. Note: Details may not add to totals due to rounding"],"full_content":"7 _Source data: Table 9_\n\n$48B $224B $106B\n\n$452B\n\n$2,656B\n\n$1,748B\n\n**Fiscal Year 2025 to Date: $5,235B**\n\n**2024 2025**\n\nOct\n\nNov\n\nDec\n\nJan\n\nFeb\n\nMar\n\nApr\n\nMay\n\nJun\n\nJul\n\nAug\n\nSep\n\nOct\n\nNov\n\nDec\n\nJan\n\nFeb\n\nMar\n\nApr\n\nMay\n\nJun\n\nJul\n\nAug\n\nSep\n\n$0B\n\n$100B\n\n$200B\n\n$300B\n\n$400B\n\n$500B\n\n$600B\n\n$700B\n\n$800B\n\nReceipt Amount\n\n$3B $33B\n\n$13B\n\n$62B\n\n$298B\n\n$134B\n\n**Month of September 2025: $544B**\n\n**Figure 5. Receipts of the U.S. Government, by Source, Fiscal Year 2025**\n\n**Figure 6. Monthly Receipts of the U.S. Government, by Source, Fiscal Years 2024 and 2025**\n\nMiscellaneous Receipts\n\nOther\n\nExcise Taxes\n\nCorporation Income Taxes\n\nIndividual Income Taxes\n\nSocial Insurance and Retirement Receipts\n\nMiscellaneous Receipts\n\nOther\n\nExcise Taxes\n\nCorporation Income Taxes\n\nIndividual Income Taxes\n\nSocial Insurance and Retirement Receipts\n\n\"Other\" Includes: Customs Duties and Estate and Gift Taxes \\*Disclaimer: Other category is subject to change due to negative balances held. Note: Details may not add to totals due to rounding","publish_date":null,"title":"Figure 5. Receipts of the U.S. Government, by Source, Fiscal Year 2025 Fiscal Year 2025 to Date: $5,235B Month of","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/MonthlyTreasuryStatement_202509_p7.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/MonthlyTreasuryStatement_202509_p7.pdf"},"request":{"example_id":"chart/MonthlyTreasuryStatement_202509_p7","product_type":"parse","source_file_path":"MonthlyTreasuryStatement_202509_p7.pdf"},"started_at":"2026-11-04T19:45:32.266722"} +{"completed_at":"2026-11-04T19:45:34.427849","latency_in_ms":1348,"output":{"example_id":"chart/MonthlyTreasuryStatement_202509_p8","grounded_pages":[],"layout_pages":[],"markdown":"8 _Source data: Table 9_\n\n$133B\n\n$32B\n\n$34B\n\n$94B\n\n$76B\n\n$37B\n\n($60B)\n\n**Month of September 2025: $346B**\n\n$1,581B\n\n$997B\n\n$702B $979B\n\n$917B\n\n$970B\n\n$865B\n\n**Fiscal Year 2025 to Date: $7,010B**\n\n**2024 2025**\n\nOct\n\nNov\n\nDec\n\nJan\n\nFeb\n\nMar\n\nApr\n\nMay\n\nJun\n\nJul\n\nAug\n\nSep\n\nOct\n\nNov\n\nDec\n\nJan\n\nFeb\n\nMar\n\nApr\n\nMay\n\nJun\n\nJul\n\nAug\n\nSep\n\n($100B) $0B\n\n$100B\n\n$200B\n\n$300B\n\n$400B\n\n$500B\n\n$600B\n\n$700B\n\n$800B\n\n$900B\n\n$1,000B\n\nOutlay Amount\n\n**Figure 7. Outlays of the U.S. Government, by Source, Fiscal Year 2025**\n\nSocial Security Income Security National Defense\n\nNet Interest\n\nOther\n\nMedicare Health\n\n\\*Disclaimer: Other category is subject to change due to negative balances held.\n\n**Figure 8. Monthly Outlays of the U.S. Government, by Source, Fiscal Years 2024 and 2025**\n\nSocial Security\n\nMedicare\n\nIncome Security\n\nHealth\n\nNational Defense\n\nNet Interest\n\nOther\n\n\"Other\" Includes: Administration of Justice, Agriculture, Community and Regional Development, Education, Training, Employment, and Social Services, Energy, General Government, General Science, Space, and Technology, International Affairs, Natural Resources and Environment, Transportation, Undistributed Offsetting Receipts, Veterans' Benefits and Services. Note: Details may not add to totals due to rounding","pages":[{"line_numbers":[],"markdown":"8 _Source data: Table 9_\n\n$133B\n\n$32B\n\n$34B\n\n$94B\n\n$76B\n\n$37B\n\n($60B)\n\n**Month of September 2025: $346B**\n\n$1,581B\n\n$997B\n\n$702B $979B\n\n$917B\n\n$970B\n\n$865B\n\n**Fiscal Year 2025 to Date: $7,010B**\n\n**2024 2025**\n\nOct\n\nNov\n\nDec\n\nJan\n\nFeb\n\nMar\n\nApr\n\nMay\n\nJun\n\nJul\n\nAug\n\nSep\n\nOct\n\nNov\n\nDec\n\nJan\n\nFeb\n\nMar\n\nApr\n\nMay\n\nJun\n\nJul\n\nAug\n\nSep\n\n($100B) $0B\n\n$100B\n\n$200B\n\n$300B\n\n$400B\n\n$500B\n\n$600B\n\n$700B\n\n$800B\n\n$900B\n\n$1,000B\n\nOutlay Amount\n\n**Figure 7. Outlays of the U.S. Government, by Source, Fiscal Year 2025**\n\nSocial Security Income Security National Defense\n\nNet Interest\n\nOther\n\nMedicare Health\n\n\\*Disclaimer: Other category is subject to change due to negative balances held.\n\n**Figure 8. Monthly Outlays of the U.S. Government, by Source, Fiscal Years 2024 and 2025**\n\nSocial Security\n\nMedicare\n\nIncome Security\n\nHealth\n\nNational Defense\n\nNet Interest\n\nOther\n\n\"Other\" Includes: Administration of Justice, Agriculture, Community and Regional Development, Education, Training, Employment, and Social Services, Energy, General Government, General Science, Space, and Technology, International Affairs, Natural Resources and Environment, Transportation, Undistributed Offsetting Receipts, Veterans' Benefits and Services. Note: Details may not add to totals due to rounding","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"8 _Source data: Table 9_\n\n$133B\n\n$32B\n\n$34B\n\n$94B\n\n$76B\n\n$37B\n\n($60B)\n\n**Month of September 2025: $346B**\n\n$1,581B\n\n$997B\n\n$702B $979B\n\n$917B\n\n$970B\n\n$865B\n\n**Fiscal Year 2025 to Date: $7,010B**\n\n**2024 2025**\n\nOct\n\nNov\n\nDec\n\nJan\n\nFeb\n\nMar\n\nApr\n\nMay\n\nJun\n\nJul\n\nAug\n\nSep\n\nOct\n\nNov\n\nDec\n\nJan\n\nFeb\n\nMar\n\nApr\n\nMay\n\nJun\n\nJul\n\nAug\n\nSep\n\n($100B) $0B\n\n$100B\n\n$200B\n\n$300B\n\n$400B\n\n$500B\n\n$600B\n\n$700B\n\n$800B\n\n$900B\n\n$1,000B\n\nOutlay Amount\n\n**Figure 7. Outlays of the U.S. Government, by Source, Fiscal Year 2025**\n\nSocial Security Income Security National Defense\n\nNet Interest\n\nOther\n\nMedicare Health\n\n\\*Disclaimer: Other category is subject to change due to negative balances held.\n\n**Figure 8. Monthly Outlays of the U.S. Government, by Source, Fiscal Years 2024 and 2025**\n\nSocial Security\n\nMedicare\n\nIncome Security\n\nHealth\n\nNational Defense\n\nNet Interest\n\nOther\n\n\"Other\" Includes: Administration of Justice, Agriculture, Community and Regional Development, Education, Training, Employment, and Social Services, Energy, General Government, General Science, Space, and Technology, International Affairs, Natural Resources and Environment, Transportation, Undistributed Offsetting Receipts, Veterans' Benefits and Services. Note: Details may not add to totals due to rounding","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/MonthlyTreasuryStatement_202509_p8.pdf"]},"response":{"errors":[],"extract_id":"extract_f6e0fa0882da19b2746571dba488a443","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["8 _Source data: Table 9_\n\n$133B\n\n$32B\n\n$34B\n\n$94B\n\n$76B\n\n$37B\n\n($60B)\n\n**Month of September 2025: $346B**\n\n$1,581B\n\n$997B\n\n$702B $979B\n\n$917B\n\n$970B\n\n$865B\n\n**Fiscal Year 2025 to Date: $7,010B**\n\n**2024 2025**\n\nOct\n\nNov\n\nDec\n\nJan\n\nFeb\n\nMar\n\nApr\n\nMay\n\nJun\n\nJul\n\nAug\n\nSep\n\nOct\n\nNov\n\nDec\n\nJan\n\nFeb\n\nMar\n\nApr\n\nMay\n\nJun\n\nJul\n\nAug\n\nSep\n\n($100B) $0B\n\n$100B\n\n$200B\n\n$300B\n\n$400B\n\n$500B\n\n$600B\n\n$700B\n\n$800B\n\n$900B\n\n$1,000B\n\nOutlay Amount\n\n**Figure 7. Outlays of the U.S. Government, by Source, Fiscal Year 2025**\n\nSocial Security Income Security National Defense\n\nNet Interest\n\nOther\n\nMedicare Health\n\n\\*Disclaimer: Other category is subject to change due to negative balances held.\n\n**Figure 8. Monthly Outlays of the U.S. Government, by Source, Fiscal Years 2024 and 2025**\n\nSocial Security\n\nMedicare\n\nIncome Security\n\nHealth\n\nNational Defense\n\nNet Interest\n\nOther\n\n\"Other\" Includes: Administration of Justice, Agriculture, Community and Regional Development, Education, Training, Employment, and Social Services, Energy, General Government, General Science, Space, and Technology, International Affairs, Natural Resources and Environment, Transportation, Undistributed Offsetting Receipts, Veterans' Benefits and Services. Note: Details may not add to totals due to rounding"],"full_content":"8 _Source data: Table 9_\n\n$133B\n\n$32B\n\n$34B\n\n$94B\n\n$76B\n\n$37B\n\n($60B)\n\n**Month of September 2025: $346B**\n\n$1,581B\n\n$997B\n\n$702B $979B\n\n$917B\n\n$970B\n\n$865B\n\n**Fiscal Year 2025 to Date: $7,010B**\n\n**2024 2025**\n\nOct\n\nNov\n\nDec\n\nJan\n\nFeb\n\nMar\n\nApr\n\nMay\n\nJun\n\nJul\n\nAug\n\nSep\n\nOct\n\nNov\n\nDec\n\nJan\n\nFeb\n\nMar\n\nApr\n\nMay\n\nJun\n\nJul\n\nAug\n\nSep\n\n($100B) $0B\n\n$100B\n\n$200B\n\n$300B\n\n$400B\n\n$500B\n\n$600B\n\n$700B\n\n$800B\n\n$900B\n\n$1,000B\n\nOutlay Amount\n\n**Figure 7. Outlays of the U.S. Government, by Source, Fiscal Year 2025**\n\nSocial Security Income Security National Defense\n\nNet Interest\n\nOther\n\nMedicare Health\n\n\\*Disclaimer: Other category is subject to change due to negative balances held.\n\n**Figure 8. Monthly Outlays of the U.S. Government, by Source, Fiscal Years 2024 and 2025**\n\nSocial Security\n\nMedicare\n\nIncome Security\n\nHealth\n\nNational Defense\n\nNet Interest\n\nOther\n\n\"Other\" Includes: Administration of Justice, Agriculture, Community and Regional Development, Education, Training, Employment, and Social Services, Energy, General Government, General Science, Space, and Technology, International Affairs, Natural Resources and Environment, Transportation, Undistributed Offsetting Receipts, Veterans' Benefits and Services. Note: Details may not add to totals due to rounding","publish_date":null,"title":"Figure 7. Outlays of the U.S. Government, by Source, Fiscal Year 2025 Month of September 2025: $346B Fiscal Year 2025","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/MonthlyTreasuryStatement_202509_p8.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/MonthlyTreasuryStatement_202509_p8.pdf"},"request":{"example_id":"chart/MonthlyTreasuryStatement_202509_p8","product_type":"parse","source_file_path":"MonthlyTreasuryStatement_202509_p8.pdf"},"started_at":"2026-11-04T19:45:33.079055"} +{"completed_at":"2026-11-04T19:45:37.142668","latency_in_ms":2444,"output":{"example_id":"chart/P505350-59c98ca8-0803-4f23-b470-17f3dab010ab_p106","grounded_pages":[],"layout_pages":[],"markdown":"**84** D I G I T A L P R O G R E S S A N D T R E N D S R E P O R T 2 0 2 5\n\n**FIGURE 5 .6 Share of workers with AI and GenAI skills, by educational attainment, 2024**\n\n0\n\nHICs UMICs LMICs LICs HICs UMICs LMICs LICs\n\n1 2 3 4 5 6 7 8 9 **AI skills (%)** a. Supply of AI skills, excluding GenAI b. Supply of GenAI skills\n\n**GenAI skills (%)**\n\n**Education level Education level**\n\nBelow bachelor’s degree Bachelor’s degree Master’s degree Doctorate\n\n0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6\n\n_Source:_ Original figures for this publication using calculations based on data from Lightcast (https://lightcast.io/). _Note:_ AI = artificial intelligence; GenAI = generative artificial intelligence; HICs = high-income countries; LICs = low-income countries; LMICs = lower-middle-income countries; UMICs = upper-middle-income countries.\n\nHowever, workers are significantly less likely to claim GenAI skills overall. Notably, workers in LMICs report GenAI skills more frequently than those in UMICs and HICs; this is largely attributable to sample bias, because a more elite segment of the workforce in poorer countries typically uses LinkedIn.\n\nThe supply of GenAI skills largely mirrors demand, being concentrated among ICT professionals, followed by administrative and commercial managers and business and administrative professionals across all country income groups. As of June 2025, an analysis of nearly 400 million individual career profiles globally revealed that one-quarter of all users claiming GenAI skills in HICs are ICT professionals. This share rises to one-third in LMICs (refer to figure 5.7, panel a).\n\nThe top five broad occupations in which GenAI skills are concentrated include administrative and commercial managers, business and administrative professionals, chief executives, and production and specialized services managers, dovetailing with demand patterns. Among non- ICT occupations, GenAI skills are most likely claimed by R&D managers, policy and planning managers, management and organizational analysts, survey and market research interviewers, human resources professionals, engineers, and advertising and marketing professionals (refer to figure 5.7, panel b). However, GenAI skill penetration in these roles remains low, typically ranging from 0.5 percent to 1.5 percent of all profiles. This limited uptake, even among highly exposed occupations, corroborates earlier findings that GenAI adoption is still in its nascent stages and has had minimal impact on skills demand and supply to date.","pages":[{"line_numbers":[],"markdown":"**84** D I G I T A L P R O G R E S S A N D T R E N D S R E P O R T 2 0 2 5\n\n**FIGURE 5 .6 Share of workers with AI and GenAI skills, by educational attainment, 2024**\n\n0\n\nHICs UMICs LMICs LICs HICs UMICs LMICs LICs\n\n1 2 3 4 5 6 7 8 9 **AI skills (%)** a. Supply of AI skills, excluding GenAI b. Supply of GenAI skills\n\n**GenAI skills (%)**\n\n**Education level Education level**\n\nBelow bachelor’s degree Bachelor’s degree Master’s degree Doctorate\n\n0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6\n\n_Source:_ Original figures for this publication using calculations based on data from Lightcast (https://lightcast.io/). _Note:_ AI = artificial intelligence; GenAI = generative artificial intelligence; HICs = high-income countries; LICs = low-income countries; LMICs = lower-middle-income countries; UMICs = upper-middle-income countries.\n\nHowever, workers are significantly less likely to claim GenAI skills overall. Notably, workers in LMICs report GenAI skills more frequently than those in UMICs and HICs; this is largely attributable to sample bias, because a more elite segment of the workforce in poorer countries typically uses LinkedIn.\n\nThe supply of GenAI skills largely mirrors demand, being concentrated among ICT professionals, followed by administrative and commercial managers and business and administrative professionals across all country income groups. As of June 2025, an analysis of nearly 400 million individual career profiles globally revealed that one-quarter of all users claiming GenAI skills in HICs are ICT professionals. This share rises to one-third in LMICs (refer to figure 5.7, panel a).\n\nThe top five broad occupations in which GenAI skills are concentrated include administrative and commercial managers, business and administrative professionals, chief executives, and production and specialized services managers, dovetailing with demand patterns. Among non- ICT occupations, GenAI skills are most likely claimed by R&D managers, policy and planning managers, management and organizational analysts, survey and market research interviewers, human resources professionals, engineers, and advertising and marketing professionals (refer to figure 5.7, panel b). However, GenAI skill penetration in these roles remains low, typically ranging from 0.5 percent to 1.5 percent of all profiles. This limited uptake, even among highly exposed occupations, corroborates earlier findings that GenAI adoption is still in its nascent stages and has had minimal impact on skills demand and supply to date.","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**84** D I G I T A L P R O G R E S S A N D T R E N D S R E P O R T 2 0 2 5\n\n**FIGURE 5 .6 Share of workers with AI and GenAI skills, by educational attainment, 2024**\n\n0\n\nHICs UMICs LMICs LICs HICs UMICs LMICs LICs\n\n1 2 3 4 5 6 7 8 9 **AI skills (%)** a. Supply of AI skills, excluding GenAI b. Supply of GenAI skills\n\n**GenAI skills (%)**\n\n**Education level Education level**\n\nBelow bachelor’s degree Bachelor’s degree Master’s degree Doctorate\n\n0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6\n\n_Source:_ Original figures for this publication using calculations based on data from Lightcast (https://lightcast.io/). _Note:_ AI = artificial intelligence; GenAI = generative artificial intelligence; HICs = high-income countries; LICs = low-income countries; LMICs = lower-middle-income countries; UMICs = upper-middle-income countries.\n\nHowever, workers are significantly less likely to claim GenAI skills overall. Notably, workers in LMICs report GenAI skills more frequently than those in UMICs and HICs; this is largely attributable to sample bias, because a more elite segment of the workforce in poorer countries typically uses LinkedIn.\n\nThe supply of GenAI skills largely mirrors demand, being concentrated among ICT professionals, followed by administrative and commercial managers and business and administrative professionals across all country income groups. As of June 2025, an analysis of nearly 400 million individual career profiles globally revealed that one-quarter of all users claiming GenAI skills in HICs are ICT professionals. This share rises to one-third in LMICs (refer to figure 5.7, panel a).\n\nThe top five broad occupations in which GenAI skills are concentrated include administrative and commercial managers, business and administrative professionals, chief executives, and production and specialized services managers, dovetailing with demand patterns. Among non- ICT occupations, GenAI skills are most likely claimed by R&D managers, policy and planning managers, management and organizational analysts, survey and market research interviewers, human resources professionals, engineers, and advertising and marketing professionals (refer to figure 5.7, panel b). However, GenAI skill penetration in these roles remains low, typically ranging from 0.5 percent to 1.5 percent of all profiles. This limited uptake, even among highly exposed occupations, corroborates earlier findings that GenAI adoption is still in its nascent stages and has had minimal impact on skills demand and supply to date.","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/P505350-59c98ca8-0803-4f23-b470-17f3dab010ab_p106.pdf"]},"response":{"errors":[],"extract_id":"extract_4f5462b6652804cea6ba1d8aa963fece","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**84** D I G I T A L P R O G R E S S A N D T R E N D S R E P O R T 2 0 2 5\n\n**FIGURE 5 .6 Share of workers with AI and GenAI skills, by educational attainment, 2024**\n\n0\n\nHICs UMICs LMICs LICs HICs UMICs LMICs LICs\n\n1 2 3 4 5 6 7 8 9 **AI skills (%)** a. Supply of AI skills, excluding GenAI b. Supply of GenAI skills\n\n**GenAI skills (%)**\n\n**Education level Education level**\n\nBelow bachelor’s degree Bachelor’s degree Master’s degree Doctorate\n\n0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6\n\n_Source:_ Original figures for this publication using calculations based on data from Lightcast (https://lightcast.io/). _Note:_ AI = artificial intelligence; GenAI = generative artificial intelligence; HICs = high-income countries; LICs = low-income countries; LMICs = lower-middle-income countries; UMICs = upper-middle-income countries.\n\nHowever, workers are significantly less likely to claim GenAI skills overall. Notably, workers in LMICs report GenAI skills more frequently than those in UMICs and HICs; this is largely attributable to sample bias, because a more elite segment of the workforce in poorer countries typically uses LinkedIn.\n\nThe supply of GenAI skills largely mirrors demand, being concentrated among ICT professionals, followed by administrative and commercial managers and business and administrative professionals across all country income groups. As of June 2025, an analysis of nearly 400 million individual career profiles globally revealed that one-quarter of all users claiming GenAI skills in HICs are ICT professionals. This share rises to one-third in LMICs (refer to figure 5.7, panel a).\n\nThe top five broad occupations in which GenAI skills are concentrated include administrative and commercial managers, business and administrative professionals, chief executives, and production and specialized services managers, dovetailing with demand patterns. Among non- ICT occupations, GenAI skills are most likely claimed by R&D managers, policy and planning managers, management and organizational analysts, survey and market research interviewers, human resources professionals, engineers, and advertising and marketing professionals (refer to figure 5.7, panel b). However, GenAI skill penetration in these roles remains low, typically ranging from 0.5 percent to 1.5 percent of all profiles. This limited uptake, even among highly exposed occupations, corroborates earlier findings that GenAI adoption is still in its nascent stages and has had minimal impact on skills demand and supply to date."],"full_content":"**84** D I G I T A L P R O G R E S S A N D T R E N D S R E P O R T 2 0 2 5\n\n**FIGURE 5 .6 Share of workers with AI and GenAI skills, by educational attainment, 2024**\n\n0\n\nHICs UMICs LMICs LICs HICs UMICs LMICs LICs\n\n1 2 3 4 5 6 7 8 9 **AI skills (%)** a. Supply of AI skills, excluding GenAI b. Supply of GenAI skills\n\n**GenAI skills (%)**\n\n**Education level Education level**\n\nBelow bachelor’s degree Bachelor’s degree Master’s degree Doctorate\n\n0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6\n\n_Source:_ Original figures for this publication using calculations based on data from Lightcast (https://lightcast.io/). _Note:_ AI = artificial intelligence; GenAI = generative artificial intelligence; HICs = high-income countries; LICs = low-income countries; LMICs = lower-middle-income countries; UMICs = upper-middle-income countries.\n\nHowever, workers are significantly less likely to claim GenAI skills overall. Notably, workers in LMICs report GenAI skills more frequently than those in UMICs and HICs; this is largely attributable to sample bias, because a more elite segment of the workforce in poorer countries typically uses LinkedIn.\n\nThe supply of GenAI skills largely mirrors demand, being concentrated among ICT professionals, followed by administrative and commercial managers and business and administrative professionals across all country income groups. As of June 2025, an analysis of nearly 400 million individual career profiles globally revealed that one-quarter of all users claiming GenAI skills in HICs are ICT professionals. This share rises to one-third in LMICs (refer to figure 5.7, panel a).\n\nThe top five broad occupations in which GenAI skills are concentrated include administrative and commercial managers, business and administrative professionals, chief executives, and production and specialized services managers, dovetailing with demand patterns. Among non- ICT occupations, GenAI skills are most likely claimed by R&D managers, policy and planning managers, management and organizational analysts, survey and market research interviewers, human resources professionals, engineers, and advertising and marketing professionals (refer to figure 5.7, panel b). However, GenAI skill penetration in these roles remains low, typically ranging from 0.5 percent to 1.5 percent of all profiles. This limited uptake, even among highly exposed occupations, corroborates earlier findings that GenAI adoption is still in its nascent stages and has had minimal impact on skills demand and supply to date.","publish_date":null,"title":"FIGURE 5 .6 Share of workers with AI and GenAI skills, by educational attainment, 2024 However, workers are significantly less likely","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/P505350-59c98ca8-0803-4f23-b470-17f3dab010ab_p106.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/P505350-59c98ca8-0803-4f23-b470-17f3dab010ab_p106.pdf"},"request":{"example_id":"chart/P505350-59c98ca8-0803-4f23-b470-17f3dab010ab_p106","product_type":"parse","source_file_path":"P505350-59c98ca8-0803-4f23-b470-17f3dab010ab_p106.pdf"},"started_at":"2026-11-04T19:45:34.697878"} +{"completed_at":"2026-11-04T19:45:37.711410","latency_in_ms":2452,"output":{"example_id":"chart/P505350-59c98ca8-0803-4f23-b470-17f3dab010ab_p34","grounded_pages":[],"layout_pages":[],"markdown":"**12** D I G I T A L P R O G R E S S A N D T R E N D S R E P O R T 2 0 2 5\n\nAI tools on top of general-purpose models (for example, services built using the ChatGPT application programming interface), fine-tuning open-source AI models for local languages or government service delivery, or integrating AI components into specialized products.\n\n• _AI adoption: The use of AI technologies without in-house modification by public organizations,_\n\n_private enterprises, or individuals._ This can include deploying AI-powered software, using off- the-shelf AI tools such as Gemini, or integrating ready-made AI features into workflows.\n\nAlthough analytically distinct, these categories are interconnected. Adaptation often builds on innovation, and adoption may involve minor adjustments that edge into adaptation. In practice, many AI activities sit along a continuum rather than in discrete boxes. Investment is a measure for AI innovation and adaptation.\n\n### **AI innovation and adaptation**\n\nAdvancements in AI research have been driven primarily by high-income countries (HICs) notably the United States, and increasingly by industry over academia. Researchers in all LICs and MICs combined produced only 46 percent of the global AI scientific publications during 2000–24 and received only 31 percent of global citations (refer to figure 1.2). Articles from upper-middle-income countries (UMICs) and lower-middle-income countries (LMICs) are driven by China and India, respectively.\n\nAlthough the United States accounts for only 26 percent of HICs’ academic articles and 35 percent of the overall citations, China and India account for 70 percent and 68 percent of UMIC and LMIC publications, respectively. Among the more than 900 notable AI models ¹ published since the 1950s, 62 percent of lead contributors come from the United States; 25 percent from other HICs; 13 percent from China; 0.2 percent, or only 2 models, from India and 0.1 percent, or only 1 model, from Argentina.\n\n**FIGURE 1 .2 AI innovation and adaptation activities, by country income group**\n\n0 20 40 60 80 100\n\n**Percent**\n\n**Global share**\n\nAI scientifc publications (2000–24)\n\nTotal funding raised by AI start-ups (July 2025)\n\nNumber of AI start-ups (July 2025)\n\nGenAI patents (2014–23)\n\nNotable AI models (1950 June 2025)\n\nAI citations (2000–24)\n\nPopulation (2024)\n\nGDP (2024)\n\nUS Other HICs China Other UMICs India Other LMICs LICs\n\n_Sources:_ Original figure for this publication based on analysis of various indicators from Epoch (https://epoch.ai/), OECD.AI (https://oecd .ai/en/), and WIPO (https://www.wipo.int/en/web/ip-statistics). _Note:_ For models with multiple contributors, only the nationality of the first or leading contributor is counted. AI = artificial intelligence; GDP = gross domestic product; GenAI = generative artificial intelligence; HICs = high-income countries; LICs = low-income countries; LMICs = lower-middle-income countries; UMICs = upper-middle-income countries; US = United States.","pages":[{"line_numbers":[],"markdown":"**12** D I G I T A L P R O G R E S S A N D T R E N D S R E P O R T 2 0 2 5\n\nAI tools on top of general-purpose models (for example, services built using the ChatGPT application programming interface), fine-tuning open-source AI models for local languages or government service delivery, or integrating AI components into specialized products.\n\n• _AI adoption: The use of AI technologies without in-house modification by public organizations,_\n\n_private enterprises, or individuals._ This can include deploying AI-powered software, using off- the-shelf AI tools such as Gemini, or integrating ready-made AI features into workflows.\n\nAlthough analytically distinct, these categories are interconnected. Adaptation often builds on innovation, and adoption may involve minor adjustments that edge into adaptation. In practice, many AI activities sit along a continuum rather than in discrete boxes. Investment is a measure for AI innovation and adaptation.\n\n### **AI innovation and adaptation**\n\nAdvancements in AI research have been driven primarily by high-income countries (HICs) notably the United States, and increasingly by industry over academia. Researchers in all LICs and MICs combined produced only 46 percent of the global AI scientific publications during 2000–24 and received only 31 percent of global citations (refer to figure 1.2). Articles from upper-middle-income countries (UMICs) and lower-middle-income countries (LMICs) are driven by China and India, respectively.\n\nAlthough the United States accounts for only 26 percent of HICs’ academic articles and 35 percent of the overall citations, China and India account for 70 percent and 68 percent of UMIC and LMIC publications, respectively. Among the more than 900 notable AI models ¹ published since the 1950s, 62 percent of lead contributors come from the United States; 25 percent from other HICs; 13 percent from China; 0.2 percent, or only 2 models, from India and 0.1 percent, or only 1 model, from Argentina.\n\n**FIGURE 1 .2 AI innovation and adaptation activities, by country income group**\n\n0 20 40 60 80 100\n\n**Percent**\n\n**Global share**\n\nAI scientifc publications (2000–24)\n\nTotal funding raised by AI start-ups (July 2025)\n\nNumber of AI start-ups (July 2025)\n\nGenAI patents (2014–23)\n\nNotable AI models (1950 June 2025)\n\nAI citations (2000–24)\n\nPopulation (2024)\n\nGDP (2024)\n\nUS Other HICs China Other UMICs India Other LMICs LICs\n\n_Sources:_ Original figure for this publication based on analysis of various indicators from Epoch (https://epoch.ai/), OECD.AI (https://oecd .ai/en/), and WIPO (https://www.wipo.int/en/web/ip-statistics). _Note:_ For models with multiple contributors, only the nationality of the first or leading contributor is counted. AI = artificial intelligence; GDP = gross domestic product; GenAI = generative artificial intelligence; HICs = high-income countries; LICs = low-income countries; LMICs = lower-middle-income countries; UMICs = upper-middle-income countries; US = United States.","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**12** D I G I T A L P R O G R E S S A N D T R E N D S R E P O R T 2 0 2 5\n\nAI tools on top of general-purpose models (for example, services built using the ChatGPT application programming interface), fine-tuning open-source AI models for local languages or government service delivery, or integrating AI components into specialized products.\n\n• _AI adoption: The use of AI technologies without in-house modification by public organizations,_\n\n_private enterprises, or individuals._ This can include deploying AI-powered software, using off- the-shelf AI tools such as Gemini, or integrating ready-made AI features into workflows.\n\nAlthough analytically distinct, these categories are interconnected. Adaptation often builds on innovation, and adoption may involve minor adjustments that edge into adaptation. In practice, many AI activities sit along a continuum rather than in discrete boxes. Investment is a measure for AI innovation and adaptation.\n\n### **AI innovation and adaptation**\n\nAdvancements in AI research have been driven primarily by high-income countries (HICs) notably the United States, and increasingly by industry over academia. Researchers in all LICs and MICs combined produced only 46 percent of the global AI scientific publications during 2000–24 and received only 31 percent of global citations (refer to figure 1.2). Articles from upper-middle-income countries (UMICs) and lower-middle-income countries (LMICs) are driven by China and India, respectively.\n\nAlthough the United States accounts for only 26 percent of HICs’ academic articles and 35 percent of the overall citations, China and India account for 70 percent and 68 percent of UMIC and LMIC publications, respectively. Among the more than 900 notable AI models ¹ published since the 1950s, 62 percent of lead contributors come from the United States; 25 percent from other HICs; 13 percent from China; 0.2 percent, or only 2 models, from India and 0.1 percent, or only 1 model, from Argentina.\n\n**FIGURE 1 .2 AI innovation and adaptation activities, by country income group**\n\n0 20 40 60 80 100\n\n**Percent**\n\n**Global share**\n\nAI scientifc publications (2000–24)\n\nTotal funding raised by AI start-ups (July 2025)\n\nNumber of AI start-ups (July 2025)\n\nGenAI patents (2014–23)\n\nNotable AI models (1950 June 2025)\n\nAI citations (2000–24)\n\nPopulation (2024)\n\nGDP (2024)\n\nUS Other HICs China Other UMICs India Other LMICs LICs\n\n_Sources:_ Original figure for this publication based on analysis of various indicators from Epoch (https://epoch.ai/), OECD.AI (https://oecd .ai/en/), and WIPO (https://www.wipo.int/en/web/ip-statistics). _Note:_ For models with multiple contributors, only the nationality of the first or leading contributor is counted. AI = artificial intelligence; GDP = gross domestic product; GenAI = generative artificial intelligence; HICs = high-income countries; LICs = low-income countries; LMICs = lower-middle-income countries; UMICs = upper-middle-income countries; US = United States.","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/P505350-59c98ca8-0803-4f23-b470-17f3dab010ab_p34.pdf"]},"response":{"errors":[],"extract_id":"extract_bc243a27d0e532d1b5e9857a4ea83f1a","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**12** D I G I T A L P R O G R E S S A N D T R E N D S R E P O R T 2 0 2 5\n\nAI tools on top of general-purpose models (for example, services built using the ChatGPT application programming interface), fine-tuning open-source AI models for local languages or government service delivery, or integrating AI components into specialized products.\n\n• _AI adoption: The use of AI technologies without in-house modification by public organizations,_\n\n_private enterprises, or individuals._ This can include deploying AI-powered software, using off- the-shelf AI tools such as Gemini, or integrating ready-made AI features into workflows.\n\nAlthough analytically distinct, these categories are interconnected. Adaptation often builds on innovation, and adoption may involve minor adjustments that edge into adaptation. In practice, many AI activities sit along a continuum rather than in discrete boxes. Investment is a measure for AI innovation and adaptation.\n\n### **AI innovation and adaptation**\n\nAdvancements in AI research have been driven primarily by high-income countries (HICs) notably the United States, and increasingly by industry over academia. Researchers in all LICs and MICs combined produced only 46 percent of the global AI scientific publications during 2000–24 and received only 31 percent of global citations (refer to figure 1.2). Articles from upper-middle-income countries (UMICs) and lower-middle-income countries (LMICs) are driven by China and India, respectively.\n\nAlthough the United States accounts for only 26 percent of HICs’ academic articles and 35 percent of the overall citations, China and India account for 70 percent and 68 percent of UMIC and LMIC publications, respectively. Among the more than 900 notable AI models ¹ published since the 1950s, 62 percent of lead contributors come from the United States; 25 percent from other HICs; 13 percent from China; 0.2 percent, or only 2 models, from India and 0.1 percent, or only 1 model, from Argentina.\n\n**FIGURE 1 .2 AI innovation and adaptation activities, by country income group**\n\n0 20 40 60 80 100\n\n**Percent**\n\n**Global share**\n\nAI scientifc publications (2000–24)\n\nTotal funding raised by AI start-ups (July 2025)\n\nNumber of AI start-ups (July 2025)\n\nGenAI patents (2014–23)\n\nNotable AI models (1950 June 2025)\n\nAI citations (2000–24)\n\nPopulation (2024)\n\nGDP (2024)\n\nUS Other HICs China Other UMICs India Other LMICs LICs\n\n_Sources:_ Original figure for this publication based on analysis of various indicators from Epoch (https://epoch.ai/), OECD.AI (https://oecd .ai/en/), and WIPO (https://www.wipo.int/en/web/ip-statistics). _Note:_ For models with multiple contributors, only the nationality of the first or leading contributor is counted. AI = artificial intelligence; GDP = gross domestic product; GenAI = generative artificial intelligence; HICs = high-income countries; LICs = low-income countries; LMICs = lower-middle-income countries; UMICs = upper-middle-income countries; US = United States."],"full_content":"**12** D I G I T A L P R O G R E S S A N D T R E N D S R E P O R T 2 0 2 5\n\nAI tools on top of general-purpose models (for example, services built using the ChatGPT application programming interface), fine-tuning open-source AI models for local languages or government service delivery, or integrating AI components into specialized products.\n\n• _AI adoption: The use of AI technologies without in-house modification by public organizations,_\n\n_private enterprises, or individuals._ This can include deploying AI-powered software, using off- the-shelf AI tools such as Gemini, or integrating ready-made AI features into workflows.\n\nAlthough analytically distinct, these categories are interconnected. Adaptation often builds on innovation, and adoption may involve minor adjustments that edge into adaptation. In practice, many AI activities sit along a continuum rather than in discrete boxes. Investment is a measure for AI innovation and adaptation.\n\n### **AI innovation and adaptation**\n\nAdvancements in AI research have been driven primarily by high-income countries (HICs) notably the United States, and increasingly by industry over academia. Researchers in all LICs and MICs combined produced only 46 percent of the global AI scientific publications during 2000–24 and received only 31 percent of global citations (refer to figure 1.2). Articles from upper-middle-income countries (UMICs) and lower-middle-income countries (LMICs) are driven by China and India, respectively.\n\nAlthough the United States accounts for only 26 percent of HICs’ academic articles and 35 percent of the overall citations, China and India account for 70 percent and 68 percent of UMIC and LMIC publications, respectively. Among the more than 900 notable AI models ¹ published since the 1950s, 62 percent of lead contributors come from the United States; 25 percent from other HICs; 13 percent from China; 0.2 percent, or only 2 models, from India and 0.1 percent, or only 1 model, from Argentina.\n\n**FIGURE 1 .2 AI innovation and adaptation activities, by country income group**\n\n0 20 40 60 80 100\n\n**Percent**\n\n**Global share**\n\nAI scientifc publications (2000–24)\n\nTotal funding raised by AI start-ups (July 2025)\n\nNumber of AI start-ups (July 2025)\n\nGenAI patents (2014–23)\n\nNotable AI models (1950 June 2025)\n\nAI citations (2000–24)\n\nPopulation (2024)\n\nGDP (2024)\n\nUS Other HICs China Other UMICs India Other LMICs LICs\n\n_Sources:_ Original figure for this publication based on analysis of various indicators from Epoch (https://epoch.ai/), OECD.AI (https://oecd .ai/en/), and WIPO (https://www.wipo.int/en/web/ip-statistics). _Note:_ For models with multiple contributors, only the nationality of the first or leading contributor is counted. AI = artificial intelligence; GDP = gross domestic product; GenAI = generative artificial intelligence; HICs = high-income countries; LICs = low-income countries; LMICs = lower-middle-income countries; UMICs = upper-middle-income countries; US = United States.","publish_date":null,"title":"AI innovation and adaptation","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/P505350-59c98ca8-0803-4f23-b470-17f3dab010ab_p34.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/P505350-59c98ca8-0803-4f23-b470-17f3dab010ab_p34.pdf"},"request":{"example_id":"chart/P505350-59c98ca8-0803-4f23-b470-17f3dab010ab_p34","product_type":"parse","source_file_path":"P505350-59c98ca8-0803-4f23-b470-17f3dab010ab_p34.pdf"},"started_at":"2026-11-04T19:45:35.259263"} +{"completed_at":"2026-11-04T19:45:38.272504","latency_in_ms":1127,"output":{"example_id":"chart/P505350-59c98ca8-0803-4f23-b470-17f3dab010ab_p38","grounded_pages":[],"layout_pages":[],"markdown":"**16** D I G I T A L P R O G R E S S A N D T R E N D S R E P O R T 2 0 2 5\n\nStill, MICs show disproportionately high GenAI adoption relative to their economic size. The share of US traffic to ChatGPT plummeted from 70 percent to 25 percent within 1 month after its debut, but middle-income economies contributed more than 50 percent of global traffic only 6 months later (refer to figure 1.3). MICs have a disproportionately high adoption rate of GenAI relative to their GDP, electricity consumption, and search engine traffic, although low-income economies remain underrepresented, accounting for less than 1 percent of global ChatGPT traffic.\n\nRobust digital infrastructure, specialization in digital services, higher human capital, and a large youth population are key factors driving higher GenAI adoption at the country level. Specifically, higher fixed broadband penetration, faster internet speed, a large share of white-collar workers, comparative advantage in digitally deliverable services, widespread English fluency and digital skills, and a higher share of youth population predict higher GenAI adoption when GDP per capita, population, and internet penetration are controlled for.\n\nThe demographics of GenAI users as captured by Semrush ³ reveal a clear skew toward young, highly educated men. Women account for only one-third of ChatGPT users, a stark contrast to the more balanced gender splits seen on platforms such as Google (48 percent female) and Wikipedia (52 percent female). The most-active GenAI user group is those ages 18–24, particularly for video- based GenAI tools. Nearly half of chatbot users hold a college degree, exceeding the educational attainment of Google’s user base.\n\nThis gender gap in GenAI usage is significantly more pronounced than the global gender divide in internet use overall. According to International Telecommunication Union data, 70 percent of men are internet users worldwide, compared with 65 percent of women. In HICs and UMICs, this gap has nearly closed, with 94 percent of men and 93 percent of women online in HICs and 81 percent of men and 80 percent of women online in UMICs. Around 32 percent of US men use\n\n**FIGURE 1 .3 ChatGPT and GenAI traffic over time, by income level and country**\n\n**Country**\n\n0 10 20 30 40 50 60\n\nFrance United Kingdom Mexico Germany Canada Philippines Indonesia Brazil India\n\n**As a share of US value (percent)**\n\nb. GenAI trafc and other indicators\n\nGenAI trafc Search engine trafc Electricity consumption GDP\n\na. ChatGPT trafc, by income group\n\n0\n\n20\n\n40\n\n60\n\n80\n\n100\n\nNovember 1,\n\n2022 June 1, 2023 January 1, 2024 August 1, 2024 March 1, 2025\n\n**Percent**\n\n**Date**\n\nLICs LMICs UMICs HICs\n\n_Sources:_ Liu, Huang, and Wang 2025; Liu and Wang 2024. _Note:_ GDP = gross domestic product; GenAI = generative artificial intelligence; GPT = general-purpose technology; HIC = high-income countries; LICs = low-income countries; LMICs = lower-middle-income countries; US = United States.","pages":[{"line_numbers":[],"markdown":"**16** D I G I T A L P R O G R E S S A N D T R E N D S R E P O R T 2 0 2 5\n\nStill, MICs show disproportionately high GenAI adoption relative to their economic size. The share of US traffic to ChatGPT plummeted from 70 percent to 25 percent within 1 month after its debut, but middle-income economies contributed more than 50 percent of global traffic only 6 months later (refer to figure 1.3). MICs have a disproportionately high adoption rate of GenAI relative to their GDP, electricity consumption, and search engine traffic, although low-income economies remain underrepresented, accounting for less than 1 percent of global ChatGPT traffic.\n\nRobust digital infrastructure, specialization in digital services, higher human capital, and a large youth population are key factors driving higher GenAI adoption at the country level. Specifically, higher fixed broadband penetration, faster internet speed, a large share of white-collar workers, comparative advantage in digitally deliverable services, widespread English fluency and digital skills, and a higher share of youth population predict higher GenAI adoption when GDP per capita, population, and internet penetration are controlled for.\n\nThe demographics of GenAI users as captured by Semrush ³ reveal a clear skew toward young, highly educated men. Women account for only one-third of ChatGPT users, a stark contrast to the more balanced gender splits seen on platforms such as Google (48 percent female) and Wikipedia (52 percent female). The most-active GenAI user group is those ages 18–24, particularly for video- based GenAI tools. Nearly half of chatbot users hold a college degree, exceeding the educational attainment of Google’s user base.\n\nThis gender gap in GenAI usage is significantly more pronounced than the global gender divide in internet use overall. According to International Telecommunication Union data, 70 percent of men are internet users worldwide, compared with 65 percent of women. In HICs and UMICs, this gap has nearly closed, with 94 percent of men and 93 percent of women online in HICs and 81 percent of men and 80 percent of women online in UMICs. Around 32 percent of US men use\n\n**FIGURE 1 .3 ChatGPT and GenAI traffic over time, by income level and country**\n\n**Country**\n\n0 10 20 30 40 50 60\n\nFrance United Kingdom Mexico Germany Canada Philippines Indonesia Brazil India\n\n**As a share of US value (percent)**\n\nb. GenAI trafc and other indicators\n\nGenAI trafc Search engine trafc Electricity consumption GDP\n\na. ChatGPT trafc, by income group\n\n0\n\n20\n\n40\n\n60\n\n80\n\n100\n\nNovember 1,\n\n2022 June 1, 2023 January 1, 2024 August 1, 2024 March 1, 2025\n\n**Percent**\n\n**Date**\n\nLICs LMICs UMICs HICs\n\n_Sources:_ Liu, Huang, and Wang 2025; Liu and Wang 2024. _Note:_ GDP = gross domestic product; GenAI = generative artificial intelligence; GPT = general-purpose technology; HIC = high-income countries; LICs = low-income countries; LMICs = lower-middle-income countries; US = United States.","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**16** D I G I T A L P R O G R E S S A N D T R E N D S R E P O R T 2 0 2 5\n\nStill, MICs show disproportionately high GenAI adoption relative to their economic size. The share of US traffic to ChatGPT plummeted from 70 percent to 25 percent within 1 month after its debut, but middle-income economies contributed more than 50 percent of global traffic only 6 months later (refer to figure 1.3). MICs have a disproportionately high adoption rate of GenAI relative to their GDP, electricity consumption, and search engine traffic, although low-income economies remain underrepresented, accounting for less than 1 percent of global ChatGPT traffic.\n\nRobust digital infrastructure, specialization in digital services, higher human capital, and a large youth population are key factors driving higher GenAI adoption at the country level. Specifically, higher fixed broadband penetration, faster internet speed, a large share of white-collar workers, comparative advantage in digitally deliverable services, widespread English fluency and digital skills, and a higher share of youth population predict higher GenAI adoption when GDP per capita, population, and internet penetration are controlled for.\n\nThe demographics of GenAI users as captured by Semrush ³ reveal a clear skew toward young, highly educated men. Women account for only one-third of ChatGPT users, a stark contrast to the more balanced gender splits seen on platforms such as Google (48 percent female) and Wikipedia (52 percent female). The most-active GenAI user group is those ages 18–24, particularly for video- based GenAI tools. Nearly half of chatbot users hold a college degree, exceeding the educational attainment of Google’s user base.\n\nThis gender gap in GenAI usage is significantly more pronounced than the global gender divide in internet use overall. According to International Telecommunication Union data, 70 percent of men are internet users worldwide, compared with 65 percent of women. In HICs and UMICs, this gap has nearly closed, with 94 percent of men and 93 percent of women online in HICs and 81 percent of men and 80 percent of women online in UMICs. Around 32 percent of US men use\n\n**FIGURE 1 .3 ChatGPT and GenAI traffic over time, by income level and country**\n\n**Country**\n\n0 10 20 30 40 50 60\n\nFrance United Kingdom Mexico Germany Canada Philippines Indonesia Brazil India\n\n**As a share of US value (percent)**\n\nb. GenAI trafc and other indicators\n\nGenAI trafc Search engine trafc Electricity consumption GDP\n\na. ChatGPT trafc, by income group\n\n0\n\n20\n\n40\n\n60\n\n80\n\n100\n\nNovember 1,\n\n2022 June 1, 2023 January 1, 2024 August 1, 2024 March 1, 2025\n\n**Percent**\n\n**Date**\n\nLICs LMICs UMICs HICs\n\n_Sources:_ Liu, Huang, and Wang 2025; Liu and Wang 2024. _Note:_ GDP = gross domestic product; GenAI = generative artificial intelligence; GPT = general-purpose technology; HIC = high-income countries; LICs = low-income countries; LMICs = lower-middle-income countries; US = United States.","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/P505350-59c98ca8-0803-4f23-b470-17f3dab010ab_p38.pdf"]},"response":{"errors":[],"extract_id":"extract_4edaac98814e178d4250476a042f148d","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**16** D I G I T A L P R O G R E S S A N D T R E N D S R E P O R T 2 0 2 5\n\nStill, MICs show disproportionately high GenAI adoption relative to their economic size. The share of US traffic to ChatGPT plummeted from 70 percent to 25 percent within 1 month after its debut, but middle-income economies contributed more than 50 percent of global traffic only 6 months later (refer to figure 1.3). MICs have a disproportionately high adoption rate of GenAI relative to their GDP, electricity consumption, and search engine traffic, although low-income economies remain underrepresented, accounting for less than 1 percent of global ChatGPT traffic.\n\nRobust digital infrastructure, specialization in digital services, higher human capital, and a large youth population are key factors driving higher GenAI adoption at the country level. Specifically, higher fixed broadband penetration, faster internet speed, a large share of white-collar workers, comparative advantage in digitally deliverable services, widespread English fluency and digital skills, and a higher share of youth population predict higher GenAI adoption when GDP per capita, population, and internet penetration are controlled for.\n\nThe demographics of GenAI users as captured by Semrush ³ reveal a clear skew toward young, highly educated men. Women account for only one-third of ChatGPT users, a stark contrast to the more balanced gender splits seen on platforms such as Google (48 percent female) and Wikipedia (52 percent female). The most-active GenAI user group is those ages 18–24, particularly for video- based GenAI tools. Nearly half of chatbot users hold a college degree, exceeding the educational attainment of Google’s user base.\n\nThis gender gap in GenAI usage is significantly more pronounced than the global gender divide in internet use overall. According to International Telecommunication Union data, 70 percent of men are internet users worldwide, compared with 65 percent of women. In HICs and UMICs, this gap has nearly closed, with 94 percent of men and 93 percent of women online in HICs and 81 percent of men and 80 percent of women online in UMICs. Around 32 percent of US men use\n\n**FIGURE 1 .3 ChatGPT and GenAI traffic over time, by income level and country**\n\n**Country**\n\n0 10 20 30 40 50 60\n\nFrance United Kingdom Mexico Germany Canada Philippines Indonesia Brazil India\n\n**As a share of US value (percent)**\n\nb. GenAI trafc and other indicators\n\nGenAI trafc Search engine trafc Electricity consumption GDP\n\na. ChatGPT trafc, by income group\n\n0\n\n20\n\n40\n\n60\n\n80\n\n100\n\nNovember 1,\n\n2022 June 1, 2023 January 1, 2024 August 1, 2024 March 1, 2025\n\n**Percent**\n\n**Date**\n\nLICs LMICs UMICs HICs\n\n_Sources:_ Liu, Huang, and Wang 2025; Liu and Wang 2024. _Note:_ GDP = gross domestic product; GenAI = generative artificial intelligence; GPT = general-purpose technology; HIC = high-income countries; LICs = low-income countries; LMICs = lower-middle-income countries; US = United States."],"full_content":"**16** D I G I T A L P R O G R E S S A N D T R E N D S R E P O R T 2 0 2 5\n\nStill, MICs show disproportionately high GenAI adoption relative to their economic size. The share of US traffic to ChatGPT plummeted from 70 percent to 25 percent within 1 month after its debut, but middle-income economies contributed more than 50 percent of global traffic only 6 months later (refer to figure 1.3). MICs have a disproportionately high adoption rate of GenAI relative to their GDP, electricity consumption, and search engine traffic, although low-income economies remain underrepresented, accounting for less than 1 percent of global ChatGPT traffic.\n\nRobust digital infrastructure, specialization in digital services, higher human capital, and a large youth population are key factors driving higher GenAI adoption at the country level. Specifically, higher fixed broadband penetration, faster internet speed, a large share of white-collar workers, comparative advantage in digitally deliverable services, widespread English fluency and digital skills, and a higher share of youth population predict higher GenAI adoption when GDP per capita, population, and internet penetration are controlled for.\n\nThe demographics of GenAI users as captured by Semrush ³ reveal a clear skew toward young, highly educated men. Women account for only one-third of ChatGPT users, a stark contrast to the more balanced gender splits seen on platforms such as Google (48 percent female) and Wikipedia (52 percent female). The most-active GenAI user group is those ages 18–24, particularly for video- based GenAI tools. Nearly half of chatbot users hold a college degree, exceeding the educational attainment of Google’s user base.\n\nThis gender gap in GenAI usage is significantly more pronounced than the global gender divide in internet use overall. According to International Telecommunication Union data, 70 percent of men are internet users worldwide, compared with 65 percent of women. In HICs and UMICs, this gap has nearly closed, with 94 percent of men and 93 percent of women online in HICs and 81 percent of men and 80 percent of women online in UMICs. Around 32 percent of US men use\n\n**FIGURE 1 .3 ChatGPT and GenAI traffic over time, by income level and country**\n\n**Country**\n\n0 10 20 30 40 50 60\n\nFrance United Kingdom Mexico Germany Canada Philippines Indonesia Brazil India\n\n**As a share of US value (percent)**\n\nb. GenAI trafc and other indicators\n\nGenAI trafc Search engine trafc Electricity consumption GDP\n\na. ChatGPT trafc, by income group\n\n0\n\n20\n\n40\n\n60\n\n80\n\n100\n\nNovember 1,\n\n2022 June 1, 2023 January 1, 2024 August 1, 2024 March 1, 2025\n\n**Percent**\n\n**Date**\n\nLICs LMICs UMICs HICs\n\n_Sources:_ Liu, Huang, and Wang 2025; Liu and Wang 2024. _Note:_ GDP = gross domestic product; GenAI = generative artificial intelligence; GPT = general-purpose technology; HIC = high-income countries; LICs = low-income countries; LMICs = lower-middle-income countries; US = United States.","publish_date":null,"title":"Still, MICs show disproportionately high GenAI adoption relative to their economic size. The share of US traffic to ChatGPT plummeted","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/P505350-59c98ca8-0803-4f23-b470-17f3dab010ab_p38.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/P505350-59c98ca8-0803-4f23-b470-17f3dab010ab_p38.pdf"},"request":{"example_id":"chart/P505350-59c98ca8-0803-4f23-b470-17f3dab010ab_p38","product_type":"parse","source_file_path":"P505350-59c98ca8-0803-4f23-b470-17f3dab010ab_p38.pdf"},"started_at":"2026-11-04T19:45:37.144716"} +{"completed_at":"2026-11-04T19:45:39.307937","latency_in_ms":1033,"output":{"example_id":"chart/P505350-59c98ca8-0803-4f23-b470-17f3dab010ab_p49","grounded_pages":[],"layout_pages":[],"markdown":"C O N N E C T I v I T y **27**\n\n**FIGURE 2 .1 Mobile network coverage, by country income group, 2022 and 2024**\n\n0\n\n20\n\n40\n\n60\n\n80\n\n100\n\n**Region**\n\n2G 3G 4G 5G\n\n3 5\n\n57\n\n32\n\n2 4\n\n41\n\n51\n\n0 0 1 24\n\n75\n\n1 15\n\n84\n\n1 1\n\n49\n\n48\n\n2 1\n\n32\n\n65\n\n3 4\n\n85\n\n6\n\n4 2\n\n58\n\n35\n\n15\n\n35\n\n39\n\n1\n\n11\n\n28\n\n49\n\n4\n\n2022 2024 2022 2024 2022 2024 2022 2024 2022 2024 Global HICs UMICs LMICs LICs\n\n**Population covered by mobile network (%)**\n\n_Source:_ Original figure for this publication using calculations from the International Telecommunication Union (https://datahub.itu.int/). _Note:_ HICs = high-income countries; LICs = low-income countries; LMICs = lower-middle-income countries; UMICs = upper-middle -income countries.\n\nThe recent rise in the availability of satellite internet worldwide has been driven by significant growth in satellite constellations. As of 2023, there were slightly more than 7,560 satellites in orbit, of which approximately 5,300 are commercial communications satellites providing connectivity services in addition to traditional fixed and mobile networks (Sebastian 2005). The presence of large constellations of nongeosynchronous-orbit (NGSO) satellites covering the earth’s surface remains the main driver of increased availability of satellite connectivity. Since 2015, the number of communications satellites in orbit deployed for commercial purposes has increased more than 14 times, with LEO satellites constituting the bulk of launches (refer to figure 2.2). LEO satellites now make up 90 percent of all satellites in orbit providing communications and connectivity services.\n\nNew business models and price points are increasing the adoption of satellite communications services globally. Although GSO satellites stationed approximately 36,000 kilometers above the Earth’s surface have historically provided effective communications services, primarily focused on business and government clients, and played a vital role in enhancing global navigation, backhauling for telecommunications providers, and emergency response capabilities, recent breakthroughs in technology (leading to a significant reduction in launch costs) and business models are changing the landscape of satellite connectivity. NGSO satellite constellations such as Starlink and OneWeb offer solutions with improved speeds, better prices, and reduced latency, making satellite connectivity attractive to end users seeking improved experiences in developing countries. A 2023 study by Ookla found that in Nigeria and Rwanda, SpaceX’s Starlink recorded median download speeds almost two times faster than those of other fixed broadband providers. ¹ NGSO satellites also provide a viable connectivity solution for rural and remote areas that are hard to reach with terrestrial fixed and mobile networks because of financial, demographic, or geographic constraints, democratizing broadband access to unserved populations. Countries also have their own regulatory processes to decide which providers are best for their own contexts (Stein and Natanson 2025).","pages":[{"line_numbers":[],"markdown":"C O N N E C T I v I T y **27**\n\n**FIGURE 2 .1 Mobile network coverage, by country income group, 2022 and 2024**\n\n0\n\n20\n\n40\n\n60\n\n80\n\n100\n\n**Region**\n\n2G 3G 4G 5G\n\n3 5\n\n57\n\n32\n\n2 4\n\n41\n\n51\n\n0 0 1 24\n\n75\n\n1 15\n\n84\n\n1 1\n\n49\n\n48\n\n2 1\n\n32\n\n65\n\n3 4\n\n85\n\n6\n\n4 2\n\n58\n\n35\n\n15\n\n35\n\n39\n\n1\n\n11\n\n28\n\n49\n\n4\n\n2022 2024 2022 2024 2022 2024 2022 2024 2022 2024 Global HICs UMICs LMICs LICs\n\n**Population covered by mobile network (%)**\n\n_Source:_ Original figure for this publication using calculations from the International Telecommunication Union (https://datahub.itu.int/). _Note:_ HICs = high-income countries; LICs = low-income countries; LMICs = lower-middle-income countries; UMICs = upper-middle -income countries.\n\nThe recent rise in the availability of satellite internet worldwide has been driven by significant growth in satellite constellations. As of 2023, there were slightly more than 7,560 satellites in orbit, of which approximately 5,300 are commercial communications satellites providing connectivity services in addition to traditional fixed and mobile networks (Sebastian 2005). The presence of large constellations of nongeosynchronous-orbit (NGSO) satellites covering the earth’s surface remains the main driver of increased availability of satellite connectivity. Since 2015, the number of communications satellites in orbit deployed for commercial purposes has increased more than 14 times, with LEO satellites constituting the bulk of launches (refer to figure 2.2). LEO satellites now make up 90 percent of all satellites in orbit providing communications and connectivity services.\n\nNew business models and price points are increasing the adoption of satellite communications services globally. Although GSO satellites stationed approximately 36,000 kilometers above the Earth’s surface have historically provided effective communications services, primarily focused on business and government clients, and played a vital role in enhancing global navigation, backhauling for telecommunications providers, and emergency response capabilities, recent breakthroughs in technology (leading to a significant reduction in launch costs) and business models are changing the landscape of satellite connectivity. NGSO satellite constellations such as Starlink and OneWeb offer solutions with improved speeds, better prices, and reduced latency, making satellite connectivity attractive to end users seeking improved experiences in developing countries. A 2023 study by Ookla found that in Nigeria and Rwanda, SpaceX’s Starlink recorded median download speeds almost two times faster than those of other fixed broadband providers. ¹ NGSO satellites also provide a viable connectivity solution for rural and remote areas that are hard to reach with terrestrial fixed and mobile networks because of financial, demographic, or geographic constraints, democratizing broadband access to unserved populations. Countries also have their own regulatory processes to decide which providers are best for their own contexts (Stein and Natanson 2025).","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"C O N N E C T I v I T y **27**\n\n**FIGURE 2 .1 Mobile network coverage, by country income group, 2022 and 2024**\n\n0\n\n20\n\n40\n\n60\n\n80\n\n100\n\n**Region**\n\n2G 3G 4G 5G\n\n3 5\n\n57\n\n32\n\n2 4\n\n41\n\n51\n\n0 0 1 24\n\n75\n\n1 15\n\n84\n\n1 1\n\n49\n\n48\n\n2 1\n\n32\n\n65\n\n3 4\n\n85\n\n6\n\n4 2\n\n58\n\n35\n\n15\n\n35\n\n39\n\n1\n\n11\n\n28\n\n49\n\n4\n\n2022 2024 2022 2024 2022 2024 2022 2024 2022 2024 Global HICs UMICs LMICs LICs\n\n**Population covered by mobile network (%)**\n\n_Source:_ Original figure for this publication using calculations from the International Telecommunication Union (https://datahub.itu.int/). _Note:_ HICs = high-income countries; LICs = low-income countries; LMICs = lower-middle-income countries; UMICs = upper-middle -income countries.\n\nThe recent rise in the availability of satellite internet worldwide has been driven by significant growth in satellite constellations. As of 2023, there were slightly more than 7,560 satellites in orbit, of which approximately 5,300 are commercial communications satellites providing connectivity services in addition to traditional fixed and mobile networks (Sebastian 2005). The presence of large constellations of nongeosynchronous-orbit (NGSO) satellites covering the earth’s surface remains the main driver of increased availability of satellite connectivity. Since 2015, the number of communications satellites in orbit deployed for commercial purposes has increased more than 14 times, with LEO satellites constituting the bulk of launches (refer to figure 2.2). LEO satellites now make up 90 percent of all satellites in orbit providing communications and connectivity services.\n\nNew business models and price points are increasing the adoption of satellite communications services globally. Although GSO satellites stationed approximately 36,000 kilometers above the Earth’s surface have historically provided effective communications services, primarily focused on business and government clients, and played a vital role in enhancing global navigation, backhauling for telecommunications providers, and emergency response capabilities, recent breakthroughs in technology (leading to a significant reduction in launch costs) and business models are changing the landscape of satellite connectivity. NGSO satellite constellations such as Starlink and OneWeb offer solutions with improved speeds, better prices, and reduced latency, making satellite connectivity attractive to end users seeking improved experiences in developing countries. A 2023 study by Ookla found that in Nigeria and Rwanda, SpaceX’s Starlink recorded median download speeds almost two times faster than those of other fixed broadband providers. ¹ NGSO satellites also provide a viable connectivity solution for rural and remote areas that are hard to reach with terrestrial fixed and mobile networks because of financial, demographic, or geographic constraints, democratizing broadband access to unserved populations. Countries also have their own regulatory processes to decide which providers are best for their own contexts (Stein and Natanson 2025).","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/P505350-59c98ca8-0803-4f23-b470-17f3dab010ab_p49.pdf"]},"response":{"errors":[],"extract_id":"extract_bbf7b2fd662dfcc927d3095b847751a3","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["C O N N E C T I v I T y **27**\n\n**FIGURE 2 .1 Mobile network coverage, by country income group, 2022 and 2024**\n\n0\n\n20\n\n40\n\n60\n\n80\n\n100\n\n**Region**\n\n2G 3G 4G 5G\n\n3 5\n\n57\n\n32\n\n2 4\n\n41\n\n51\n\n0 0 1 24\n\n75\n\n1 15\n\n84\n\n1 1\n\n49\n\n48\n\n2 1\n\n32\n\n65\n\n3 4\n\n85\n\n6\n\n4 2\n\n58\n\n35\n\n15\n\n35\n\n39\n\n1\n\n11\n\n28\n\n49\n\n4\n\n2022 2024 2022 2024 2022 2024 2022 2024 2022 2024 Global HICs UMICs LMICs LICs\n\n**Population covered by mobile network (%)**\n\n_Source:_ Original figure for this publication using calculations from the International Telecommunication Union (https://datahub.itu.int/). _Note:_ HICs = high-income countries; LICs = low-income countries; LMICs = lower-middle-income countries; UMICs = upper-middle -income countries.\n\nThe recent rise in the availability of satellite internet worldwide has been driven by significant growth in satellite constellations. As of 2023, there were slightly more than 7,560 satellites in orbit, of which approximately 5,300 are commercial communications satellites providing connectivity services in addition to traditional fixed and mobile networks (Sebastian 2005). The presence of large constellations of nongeosynchronous-orbit (NGSO) satellites covering the earth’s surface remains the main driver of increased availability of satellite connectivity. Since 2015, the number of communications satellites in orbit deployed for commercial purposes has increased more than 14 times, with LEO satellites constituting the bulk of launches (refer to figure 2.2). LEO satellites now make up 90 percent of all satellites in orbit providing communications and connectivity services.\n\nNew business models and price points are increasing the adoption of satellite communications services globally. Although GSO satellites stationed approximately 36,000 kilometers above the Earth’s surface have historically provided effective communications services, primarily focused on business and government clients, and played a vital role in enhancing global navigation, backhauling for telecommunications providers, and emergency response capabilities, recent breakthroughs in technology (leading to a significant reduction in launch costs) and business models are changing the landscape of satellite connectivity. NGSO satellite constellations such as Starlink and OneWeb offer solutions with improved speeds, better prices, and reduced latency, making satellite connectivity attractive to end users seeking improved experiences in developing countries. A 2023 study by Ookla found that in Nigeria and Rwanda, SpaceX’s Starlink recorded median download speeds almost two times faster than those of other fixed broadband providers. ¹ NGSO satellites also provide a viable connectivity solution for rural and remote areas that are hard to reach with terrestrial fixed and mobile networks because of financial, demographic, or geographic constraints, democratizing broadband access to unserved populations. Countries also have their own regulatory processes to decide which providers are best for their own contexts (Stein and Natanson 2025)."],"full_content":"C O N N E C T I v I T y **27**\n\n**FIGURE 2 .1 Mobile network coverage, by country income group, 2022 and 2024**\n\n0\n\n20\n\n40\n\n60\n\n80\n\n100\n\n**Region**\n\n2G 3G 4G 5G\n\n3 5\n\n57\n\n32\n\n2 4\n\n41\n\n51\n\n0 0 1 24\n\n75\n\n1 15\n\n84\n\n1 1\n\n49\n\n48\n\n2 1\n\n32\n\n65\n\n3 4\n\n85\n\n6\n\n4 2\n\n58\n\n35\n\n15\n\n35\n\n39\n\n1\n\n11\n\n28\n\n49\n\n4\n\n2022 2024 2022 2024 2022 2024 2022 2024 2022 2024 Global HICs UMICs LMICs LICs\n\n**Population covered by mobile network (%)**\n\n_Source:_ Original figure for this publication using calculations from the International Telecommunication Union (https://datahub.itu.int/). _Note:_ HICs = high-income countries; LICs = low-income countries; LMICs = lower-middle-income countries; UMICs = upper-middle -income countries.\n\nThe recent rise in the availability of satellite internet worldwide has been driven by significant growth in satellite constellations. As of 2023, there were slightly more than 7,560 satellites in orbit, of which approximately 5,300 are commercial communications satellites providing connectivity services in addition to traditional fixed and mobile networks (Sebastian 2005). The presence of large constellations of nongeosynchronous-orbit (NGSO) satellites covering the earth’s surface remains the main driver of increased availability of satellite connectivity. Since 2015, the number of communications satellites in orbit deployed for commercial purposes has increased more than 14 times, with LEO satellites constituting the bulk of launches (refer to figure 2.2). LEO satellites now make up 90 percent of all satellites in orbit providing communications and connectivity services.\n\nNew business models and price points are increasing the adoption of satellite communications services globally. Although GSO satellites stationed approximately 36,000 kilometers above the Earth’s surface have historically provided effective communications services, primarily focused on business and government clients, and played a vital role in enhancing global navigation, backhauling for telecommunications providers, and emergency response capabilities, recent breakthroughs in technology (leading to a significant reduction in launch costs) and business models are changing the landscape of satellite connectivity. NGSO satellite constellations such as Starlink and OneWeb offer solutions with improved speeds, better prices, and reduced latency, making satellite connectivity attractive to end users seeking improved experiences in developing countries. A 2023 study by Ookla found that in Nigeria and Rwanda, SpaceX’s Starlink recorded median download speeds almost two times faster than those of other fixed broadband providers. ¹ NGSO satellites also provide a viable connectivity solution for rural and remote areas that are hard to reach with terrestrial fixed and mobile networks because of financial, demographic, or geographic constraints, democratizing broadband access to unserved populations. Countries also have their own regulatory processes to decide which providers are best for their own contexts (Stein and Natanson 2025).","publish_date":null,"title":"FIGURE 2 .1 Mobile network coverage, by country income group, 2022 and 2024 The recent rise in the availability of satellite","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/P505350-59c98ca8-0803-4f23-b470-17f3dab010ab_p49.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/P505350-59c98ca8-0803-4f23-b470-17f3dab010ab_p49.pdf"},"request":{"example_id":"chart/P505350-59c98ca8-0803-4f23-b470-17f3dab010ab_p49","product_type":"parse","source_file_path":"P505350-59c98ca8-0803-4f23-b470-17f3dab010ab_p49.pdf"},"started_at":"2026-11-04T19:45:38.274405"} +{"completed_at":"2026-11-04T19:45:40.858095","latency_in_ms":1253,"output":{"example_id":"chart/P505350-59c98ca8-0803-4f23-b470-17f3dab010ab_p56","grounded_pages":[],"layout_pages":[],"markdown":"**34** D I G I T A L P R O G R E S S A N D T R E N D S R E P O R T 2 0 2 5\n\nInternet speed increased significantly in HICs and UMICs between 2023 and 2024, whereas speed in LMICs and LICs stagnated. Median fixed internet speeds in HICs and UMICs increased by more than 50 percent. In LMICs and LICs, marginal improvements kept median download speeds below 25 megabits per second for both mobile and fixed broadband in 2023 and 2024 (refer to figure 2.7). The disparity in internet speed translates into substantial differences in download times.\n\nWithin-country disparities in broadband speeds generally narrowed from 2019 to 2024 but remain pronounced, especially in SSA. The authors created a coefficient of variation (CV) of the logarithm of download speeds as a measure of within-country internet speed inequality. ³ Within-country speed inequality tends to be higher in LICs (refer to figure 2.8). In 2024, fixed broadband inequality was most pronounced in MENA, SAR, and SSA, with median CV values exceeding 0.15. Mobile speed inequality was highest in LAC and SSA. Although median inequality has generally declined, mobile broadband speed inequality rose in LAC and MENA. This reflects ongoing infrastructure projects that initially benefit certain areas, as seen in Cuba, where the introduction of the Arimao cable in early 2023 coincides with an inequality exceeding 0.5.\n\nGaps in data consumption across country income groups continue to widen, with HICs increasingly leaving others behind. In HICs and UMICs, annual fixed broadband traffic per capita is nearly 10 times the mobile traffic per capita (refer to figure 2.9). Total data traffic per capita in HICs reached nearly 1,400 gigabytes (GB) in 2023, around 400 GB in UMICs, 100 GB in LMICs, and only 5 GB in LICs.\n\n**FIGURE 2 .7 Median fixed and mobile download speed, by country income group, 2023 and 2024**\n\n0\n\n50\n\n100\n\n150\n\n0\n\n50\n\n100\n\n150\n\nHICs UMICs LMICs **Region** LICs\n\n**Fixed broadband speed (mbps) Mobile broadband speed (mbps)**\n\nFixed 2023 Fixed 2024 Mobile 2023 Mobile 2024\n\n_Sources:_ Original figure for this publication calculated using data from Ookla (https://www.speedtest.net/global-index). _Note:_ HICs = high-income countries; LICs = low-income countries; LMICs = lower-middle-income countries; mbps = megabits per second; UMICs = upper-middle -income countries.","pages":[{"line_numbers":[],"markdown":"**34** D I G I T A L P R O G R E S S A N D T R E N D S R E P O R T 2 0 2 5\n\nInternet speed increased significantly in HICs and UMICs between 2023 and 2024, whereas speed in LMICs and LICs stagnated. Median fixed internet speeds in HICs and UMICs increased by more than 50 percent. In LMICs and LICs, marginal improvements kept median download speeds below 25 megabits per second for both mobile and fixed broadband in 2023 and 2024 (refer to figure 2.7). The disparity in internet speed translates into substantial differences in download times.\n\nWithin-country disparities in broadband speeds generally narrowed from 2019 to 2024 but remain pronounced, especially in SSA. The authors created a coefficient of variation (CV) of the logarithm of download speeds as a measure of within-country internet speed inequality. ³ Within-country speed inequality tends to be higher in LICs (refer to figure 2.8). In 2024, fixed broadband inequality was most pronounced in MENA, SAR, and SSA, with median CV values exceeding 0.15. Mobile speed inequality was highest in LAC and SSA. Although median inequality has generally declined, mobile broadband speed inequality rose in LAC and MENA. This reflects ongoing infrastructure projects that initially benefit certain areas, as seen in Cuba, where the introduction of the Arimao cable in early 2023 coincides with an inequality exceeding 0.5.\n\nGaps in data consumption across country income groups continue to widen, with HICs increasingly leaving others behind. In HICs and UMICs, annual fixed broadband traffic per capita is nearly 10 times the mobile traffic per capita (refer to figure 2.9). Total data traffic per capita in HICs reached nearly 1,400 gigabytes (GB) in 2023, around 400 GB in UMICs, 100 GB in LMICs, and only 5 GB in LICs.\n\n**FIGURE 2 .7 Median fixed and mobile download speed, by country income group, 2023 and 2024**\n\n0\n\n50\n\n100\n\n150\n\n0\n\n50\n\n100\n\n150\n\nHICs UMICs LMICs **Region** LICs\n\n**Fixed broadband speed (mbps) Mobile broadband speed (mbps)**\n\nFixed 2023 Fixed 2024 Mobile 2023 Mobile 2024\n\n_Sources:_ Original figure for this publication calculated using data from Ookla (https://www.speedtest.net/global-index). _Note:_ HICs = high-income countries; LICs = low-income countries; LMICs = lower-middle-income countries; mbps = megabits per second; UMICs = upper-middle -income countries.","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**34** D I G I T A L P R O G R E S S A N D T R E N D S R E P O R T 2 0 2 5\n\nInternet speed increased significantly in HICs and UMICs between 2023 and 2024, whereas speed in LMICs and LICs stagnated. Median fixed internet speeds in HICs and UMICs increased by more than 50 percent. In LMICs and LICs, marginal improvements kept median download speeds below 25 megabits per second for both mobile and fixed broadband in 2023 and 2024 (refer to figure 2.7). The disparity in internet speed translates into substantial differences in download times.\n\nWithin-country disparities in broadband speeds generally narrowed from 2019 to 2024 but remain pronounced, especially in SSA. The authors created a coefficient of variation (CV) of the logarithm of download speeds as a measure of within-country internet speed inequality. ³ Within-country speed inequality tends to be higher in LICs (refer to figure 2.8). In 2024, fixed broadband inequality was most pronounced in MENA, SAR, and SSA, with median CV values exceeding 0.15. Mobile speed inequality was highest in LAC and SSA. Although median inequality has generally declined, mobile broadband speed inequality rose in LAC and MENA. This reflects ongoing infrastructure projects that initially benefit certain areas, as seen in Cuba, where the introduction of the Arimao cable in early 2023 coincides with an inequality exceeding 0.5.\n\nGaps in data consumption across country income groups continue to widen, with HICs increasingly leaving others behind. In HICs and UMICs, annual fixed broadband traffic per capita is nearly 10 times the mobile traffic per capita (refer to figure 2.9). Total data traffic per capita in HICs reached nearly 1,400 gigabytes (GB) in 2023, around 400 GB in UMICs, 100 GB in LMICs, and only 5 GB in LICs.\n\n**FIGURE 2 .7 Median fixed and mobile download speed, by country income group, 2023 and 2024**\n\n0\n\n50\n\n100\n\n150\n\n0\n\n50\n\n100\n\n150\n\nHICs UMICs LMICs **Region** LICs\n\n**Fixed broadband speed (mbps) Mobile broadband speed (mbps)**\n\nFixed 2023 Fixed 2024 Mobile 2023 Mobile 2024\n\n_Sources:_ Original figure for this publication calculated using data from Ookla (https://www.speedtest.net/global-index). _Note:_ HICs = high-income countries; LICs = low-income countries; LMICs = lower-middle-income countries; mbps = megabits per second; UMICs = upper-middle -income countries.","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/P505350-59c98ca8-0803-4f23-b470-17f3dab010ab_p56.pdf"]},"response":{"errors":[],"extract_id":"extract_9aff499f5607f91695d230efd0aa7e91","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**34** D I G I T A L P R O G R E S S A N D T R E N D S R E P O R T 2 0 2 5\n\nInternet speed increased significantly in HICs and UMICs between 2023 and 2024, whereas speed in LMICs and LICs stagnated. Median fixed internet speeds in HICs and UMICs increased by more than 50 percent. In LMICs and LICs, marginal improvements kept median download speeds below 25 megabits per second for both mobile and fixed broadband in 2023 and 2024 (refer to figure 2.7). The disparity in internet speed translates into substantial differences in download times.\n\nWithin-country disparities in broadband speeds generally narrowed from 2019 to 2024 but remain pronounced, especially in SSA. The authors created a coefficient of variation (CV) of the logarithm of download speeds as a measure of within-country internet speed inequality. ³ Within-country speed inequality tends to be higher in LICs (refer to figure 2.8). In 2024, fixed broadband inequality was most pronounced in MENA, SAR, and SSA, with median CV values exceeding 0.15. Mobile speed inequality was highest in LAC and SSA. Although median inequality has generally declined, mobile broadband speed inequality rose in LAC and MENA. This reflects ongoing infrastructure projects that initially benefit certain areas, as seen in Cuba, where the introduction of the Arimao cable in early 2023 coincides with an inequality exceeding 0.5.\n\nGaps in data consumption across country income groups continue to widen, with HICs increasingly leaving others behind. In HICs and UMICs, annual fixed broadband traffic per capita is nearly 10 times the mobile traffic per capita (refer to figure 2.9). Total data traffic per capita in HICs reached nearly 1,400 gigabytes (GB) in 2023, around 400 GB in UMICs, 100 GB in LMICs, and only 5 GB in LICs.\n\n**FIGURE 2 .7 Median fixed and mobile download speed, by country income group, 2023 and 2024**\n\n0\n\n50\n\n100\n\n150\n\n0\n\n50\n\n100\n\n150\n\nHICs UMICs LMICs **Region** LICs\n\n**Fixed broadband speed (mbps) Mobile broadband speed (mbps)**\n\nFixed 2023 Fixed 2024 Mobile 2023 Mobile 2024\n\n_Sources:_ Original figure for this publication calculated using data from Ookla (https://www.speedtest.net/global-index). _Note:_ HICs = high-income countries; LICs = low-income countries; LMICs = lower-middle-income countries; mbps = megabits per second; UMICs = upper-middle -income countries."],"full_content":"**34** D I G I T A L P R O G R E S S A N D T R E N D S R E P O R T 2 0 2 5\n\nInternet speed increased significantly in HICs and UMICs between 2023 and 2024, whereas speed in LMICs and LICs stagnated. Median fixed internet speeds in HICs and UMICs increased by more than 50 percent. In LMICs and LICs, marginal improvements kept median download speeds below 25 megabits per second for both mobile and fixed broadband in 2023 and 2024 (refer to figure 2.7). The disparity in internet speed translates into substantial differences in download times.\n\nWithin-country disparities in broadband speeds generally narrowed from 2019 to 2024 but remain pronounced, especially in SSA. The authors created a coefficient of variation (CV) of the logarithm of download speeds as a measure of within-country internet speed inequality. ³ Within-country speed inequality tends to be higher in LICs (refer to figure 2.8). In 2024, fixed broadband inequality was most pronounced in MENA, SAR, and SSA, with median CV values exceeding 0.15. Mobile speed inequality was highest in LAC and SSA. Although median inequality has generally declined, mobile broadband speed inequality rose in LAC and MENA. This reflects ongoing infrastructure projects that initially benefit certain areas, as seen in Cuba, where the introduction of the Arimao cable in early 2023 coincides with an inequality exceeding 0.5.\n\nGaps in data consumption across country income groups continue to widen, with HICs increasingly leaving others behind. In HICs and UMICs, annual fixed broadband traffic per capita is nearly 10 times the mobile traffic per capita (refer to figure 2.9). Total data traffic per capita in HICs reached nearly 1,400 gigabytes (GB) in 2023, around 400 GB in UMICs, 100 GB in LMICs, and only 5 GB in LICs.\n\n**FIGURE 2 .7 Median fixed and mobile download speed, by country income group, 2023 and 2024**\n\n0\n\n50\n\n100\n\n150\n\n0\n\n50\n\n100\n\n150\n\nHICs UMICs LMICs **Region** LICs\n\n**Fixed broadband speed (mbps) Mobile broadband speed (mbps)**\n\nFixed 2023 Fixed 2024 Mobile 2023 Mobile 2024\n\n_Sources:_ Original figure for this publication calculated using data from Ookla (https://www.speedtest.net/global-index). _Note:_ HICs = high-income countries; LICs = low-income countries; LMICs = lower-middle-income countries; mbps = megabits per second; UMICs = upper-middle -income countries.","publish_date":null,"title":"Internet speed increased significantly in HICs and UMICs between 2023 and 2024, whereas speed in LMICs and LICs stagnated. Median","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/P505350-59c98ca8-0803-4f23-b470-17f3dab010ab_p56.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/P505350-59c98ca8-0803-4f23-b470-17f3dab010ab_p56.pdf"},"request":{"example_id":"chart/P505350-59c98ca8-0803-4f23-b470-17f3dab010ab_p56","product_type":"parse","source_file_path":"P505350-59c98ca8-0803-4f23-b470-17f3dab010ab_p56.pdf"},"started_at":"2026-11-04T19:45:39.604291"} +{"completed_at":"2026-11-04T19:45:43.235113","latency_in_ms":1695,"output":{"example_id":"chart/PRO013216_91_Blackrock_Proxy-Statement-2025_p112","grounded_pages":[],"layout_pages":[],"markdown":"Equity Award Valuation Assumptions: BPIP award grant date fair values are calculated using the stock price as of the grant date assuming target performance. Adjustments to these values have been made using the stock price and projected vesting assumptions as of each subsequent measurement date, assuming actual performance relative to the applicable performance targets through the applicable measurement date and target performance for years in the performance period beyond the applicable measurement date (based on the award determination matrix for the BPIP award granted in the January following the applicable measurement date). RSU values represent fair values which are calculated using the stock price as of each measurement date. Stock options represent performance-based option awards with vesting contingent on the NEOs’ continued service and BlackRock’s achievement of a stock price hurdle and positive Organic Revenue Growth. Stock option fair values for measurement dates prior to when the price hurdle was achieved are calculated by a Monte Carlo simulation model with an embedded binomial lattice model, which accounts for the risk of non-achievement. For stock option fair values occurring after the achievement of the price hurdle, the values are calculated by a binomial lattice model only. The stock option binomial lattice modelling uses a consistent set of valuation inputs (including BlackRock stock price, volatility, and dividend rates, market risk free rate, and the options’ strike price and term) which have been updated as of each measurement date, as applicable.\n\n(3) Represents the cumulative Total Shareholder Return (“TSR”) of BlackRock for the periods ending on December 31, 2024, 2023,\n\n2022, 2021 and 2020, respectively, based on an initial fixed investment of $100 in BlackRock common stock on December 31, 2019. For 2024, represents the five-year TSR (2020-2024), for 2023, represents the four-year TSR (2020-2023), for 2022, represents the three-year TSR (2020-2022), for 2021, represents the two-year TSR (2020-2021) and for 2020 represents the one- year TSR (2020).\n\n(4) Represents the cumulative TSR of the S&P U.S. BMI Asset Management & Custody Banks Index (as reported in BlackRock’s Annual\n\nReport on Form 10-K) for each of the periods ending on December 31, 2024, 2023, 2022, 2021 and 2020, respectively, based on an initial fixed investment of $100 in BlackRock common stock on December 31, 2019. For 2024, represents the five-year TSR (2020-2024), for 2023, represents the four-year TSR (2020-2023), for 2022, represents the three-year TSR (2020-2022), for 2021, represents the two-year TSR (2020-2021) and for 2020 represents the one-year TSR (2020).\n\n(5) Reflects net income attributable to BlackRock, Inc. in the Company’s Consolidated Statements of Income included in BlackRock’s\n\nAnnual Reports on Form 10-K for each of the fiscal years ended December 31, 2024, 2023, 2022, 2021 and 2020, respectively.\n\n(6) BlackRock’s Company-Selected Measure is Operating Margin, as adjusted. For reconciliation with GAAP for 2022-2024, please see\n\nAnnex A. For reconciliation with GAAP for 2020 and 2021, please see **page 37** of BlackRock’s Annual Report on Form 10-K for the year ended December 31, 2022.\n\n### **Relationship between Compensation Actually Paid and Performance Measures Disclosed in the Pay Versus Performance Table**\n\nThe following chart shows that the “compensation actually paid” to our CEO and average “compensation actually paid” to our other NEOs for each of the 2020, 2021, 2022, 2023 and 2024 fiscal years align to trends with BlackRock’s and our peer group’s (S&P U.S. BMI Asset Management & Custody Banks Index) TSR over the same period. BlackRock maintained a relatively consistent TSR premium relative to our peer group.\n\n**Compensation Actually Paid vs. TSR**\n\n**Compensation Actually Paid ($ millions)**\n\n**Total Shareholder Return**\n\n**(FYE 2019 Indexed to $100)**\n\n**$56.9**\n\n**$65.3**\n\n**($5.8)**\n\n**$26.9**\n\n**$56.5**\n\n**$36.2 $39.2**\n\n**($12.5)**\n\n**$20.0**\n\n**$36.6**\n\n**$147.21**\n\n**$190.47**\n\n**$151.62**\n\n**$178.70**\n\n**$231.05**\n\n**$115.88**\n\n**$171.06**\n\n**$128.15**\n\n**$168.44**\n\n**$232.43**\n\n2020 2021 2022 2023 2024\n\n($20)\n\n$0\n\n$20\n\n$40\n\n$60\n\n$80\n\n$100\n\n$120\n\n$140\n\n$0.00\n\n$50.00\n\n$100.00\n\n$150.00\n\n$200.00\n\n$250.00\n\nCompensation Actually Paid to CEO\n\nAverage Compensation Actually Paid to Non-CEO NEOs\n\nTotal Shareholder Return\n\nPeer Group Total Shareholder Return\n\nExecutive Compensation\n\nBLACKROCK, INC. 2025 PROXY STATEMENT 105","pages":[{"line_numbers":[],"markdown":"Equity Award Valuation Assumptions: BPIP award grant date fair values are calculated using the stock price as of the grant date assuming target performance. Adjustments to these values have been made using the stock price and projected vesting assumptions as of each subsequent measurement date, assuming actual performance relative to the applicable performance targets through the applicable measurement date and target performance for years in the performance period beyond the applicable measurement date (based on the award determination matrix for the BPIP award granted in the January following the applicable measurement date). RSU values represent fair values which are calculated using the stock price as of each measurement date. Stock options represent performance-based option awards with vesting contingent on the NEOs’ continued service and BlackRock’s achievement of a stock price hurdle and positive Organic Revenue Growth. Stock option fair values for measurement dates prior to when the price hurdle was achieved are calculated by a Monte Carlo simulation model with an embedded binomial lattice model, which accounts for the risk of non-achievement. For stock option fair values occurring after the achievement of the price hurdle, the values are calculated by a binomial lattice model only. The stock option binomial lattice modelling uses a consistent set of valuation inputs (including BlackRock stock price, volatility, and dividend rates, market risk free rate, and the options’ strike price and term) which have been updated as of each measurement date, as applicable.\n\n(3) Represents the cumulative Total Shareholder Return (“TSR”) of BlackRock for the periods ending on December 31, 2024, 2023,\n\n2022, 2021 and 2020, respectively, based on an initial fixed investment of $100 in BlackRock common stock on December 31, 2019. For 2024, represents the five-year TSR (2020-2024), for 2023, represents the four-year TSR (2020-2023), for 2022, represents the three-year TSR (2020-2022), for 2021, represents the two-year TSR (2020-2021) and for 2020 represents the one- year TSR (2020).\n\n(4) Represents the cumulative TSR of the S&P U.S. BMI Asset Management & Custody Banks Index (as reported in BlackRock’s Annual\n\nReport on Form 10-K) for each of the periods ending on December 31, 2024, 2023, 2022, 2021 and 2020, respectively, based on an initial fixed investment of $100 in BlackRock common stock on December 31, 2019. For 2024, represents the five-year TSR (2020-2024), for 2023, represents the four-year TSR (2020-2023), for 2022, represents the three-year TSR (2020-2022), for 2021, represents the two-year TSR (2020-2021) and for 2020 represents the one-year TSR (2020).\n\n(5) Reflects net income attributable to BlackRock, Inc. in the Company’s Consolidated Statements of Income included in BlackRock’s\n\nAnnual Reports on Form 10-K for each of the fiscal years ended December 31, 2024, 2023, 2022, 2021 and 2020, respectively.\n\n(6) BlackRock’s Company-Selected Measure is Operating Margin, as adjusted. For reconciliation with GAAP for 2022-2024, please see\n\nAnnex A. For reconciliation with GAAP for 2020 and 2021, please see **page 37** of BlackRock’s Annual Report on Form 10-K for the year ended December 31, 2022.\n\n### **Relationship between Compensation Actually Paid and Performance Measures Disclosed in the Pay Versus Performance Table**\n\nThe following chart shows that the “compensation actually paid” to our CEO and average “compensation actually paid” to our other NEOs for each of the 2020, 2021, 2022, 2023 and 2024 fiscal years align to trends with BlackRock’s and our peer group’s (S&P U.S. BMI Asset Management & Custody Banks Index) TSR over the same period. BlackRock maintained a relatively consistent TSR premium relative to our peer group.\n\n**Compensation Actually Paid vs. TSR**\n\n**Compensation Actually Paid ($ millions)**\n\n**Total Shareholder Return**\n\n**(FYE 2019 Indexed to $100)**\n\n**$56.9**\n\n**$65.3**\n\n**($5.8)**\n\n**$26.9**\n\n**$56.5**\n\n**$36.2 $39.2**\n\n**($12.5)**\n\n**$20.0**\n\n**$36.6**\n\n**$147.21**\n\n**$190.47**\n\n**$151.62**\n\n**$178.70**\n\n**$231.05**\n\n**$115.88**\n\n**$171.06**\n\n**$128.15**\n\n**$168.44**\n\n**$232.43**\n\n2020 2021 2022 2023 2024\n\n($20)\n\n$0\n\n$20\n\n$40\n\n$60\n\n$80\n\n$100\n\n$120\n\n$140\n\n$0.00\n\n$50.00\n\n$100.00\n\n$150.00\n\n$200.00\n\n$250.00\n\nCompensation Actually Paid to CEO\n\nAverage Compensation Actually Paid to Non-CEO NEOs\n\nTotal Shareholder Return\n\nPeer Group Total Shareholder Return\n\nExecutive Compensation\n\nBLACKROCK, INC. 2025 PROXY STATEMENT 105","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Equity Award Valuation Assumptions: BPIP award grant date fair values are calculated using the stock price as of the grant date assuming target performance. Adjustments to these values have been made using the stock price and projected vesting assumptions as of each subsequent measurement date, assuming actual performance relative to the applicable performance targets through the applicable measurement date and target performance for years in the performance period beyond the applicable measurement date (based on the award determination matrix for the BPIP award granted in the January following the applicable measurement date). RSU values represent fair values which are calculated using the stock price as of each measurement date. Stock options represent performance-based option awards with vesting contingent on the NEOs’ continued service and BlackRock’s achievement of a stock price hurdle and positive Organic Revenue Growth. Stock option fair values for measurement dates prior to when the price hurdle was achieved are calculated by a Monte Carlo simulation model with an embedded binomial lattice model, which accounts for the risk of non-achievement. For stock option fair values occurring after the achievement of the price hurdle, the values are calculated by a binomial lattice model only. The stock option binomial lattice modelling uses a consistent set of valuation inputs (including BlackRock stock price, volatility, and dividend rates, market risk free rate, and the options’ strike price and term) which have been updated as of each measurement date, as applicable.\n\n(3) Represents the cumulative Total Shareholder Return (“TSR”) of BlackRock for the periods ending on December 31, 2024, 2023,\n\n2022, 2021 and 2020, respectively, based on an initial fixed investment of $100 in BlackRock common stock on December 31, 2019. For 2024, represents the five-year TSR (2020-2024), for 2023, represents the four-year TSR (2020-2023), for 2022, represents the three-year TSR (2020-2022), for 2021, represents the two-year TSR (2020-2021) and for 2020 represents the one- year TSR (2020).\n\n(4) Represents the cumulative TSR of the S&P U.S. BMI Asset Management & Custody Banks Index (as reported in BlackRock’s Annual\n\nReport on Form 10-K) for each of the periods ending on December 31, 2024, 2023, 2022, 2021 and 2020, respectively, based on an initial fixed investment of $100 in BlackRock common stock on December 31, 2019. For 2024, represents the five-year TSR (2020-2024), for 2023, represents the four-year TSR (2020-2023), for 2022, represents the three-year TSR (2020-2022), for 2021, represents the two-year TSR (2020-2021) and for 2020 represents the one-year TSR (2020).\n\n(5) Reflects net income attributable to BlackRock, Inc. in the Company’s Consolidated Statements of Income included in BlackRock’s\n\nAnnual Reports on Form 10-K for each of the fiscal years ended December 31, 2024, 2023, 2022, 2021 and 2020, respectively.\n\n(6) BlackRock’s Company-Selected Measure is Operating Margin, as adjusted. For reconciliation with GAAP for 2022-2024, please see\n\nAnnex A. For reconciliation with GAAP for 2020 and 2021, please see **page 37** of BlackRock’s Annual Report on Form 10-K for the year ended December 31, 2022.\n\n### **Relationship between Compensation Actually Paid and Performance Measures Disclosed in the Pay Versus Performance Table**\n\nThe following chart shows that the “compensation actually paid” to our CEO and average “compensation actually paid” to our other NEOs for each of the 2020, 2021, 2022, 2023 and 2024 fiscal years align to trends with BlackRock’s and our peer group’s (S&P U.S. BMI Asset Management & Custody Banks Index) TSR over the same period. BlackRock maintained a relatively consistent TSR premium relative to our peer group.\n\n**Compensation Actually Paid vs. TSR**\n\n**Compensation Actually Paid ($ millions)**\n\n**Total Shareholder Return**\n\n**(FYE 2019 Indexed to $100)**\n\n**$56.9**\n\n**$65.3**\n\n**($5.8)**\n\n**$26.9**\n\n**$56.5**\n\n**$36.2 $39.2**\n\n**($12.5)**\n\n**$20.0**\n\n**$36.6**\n\n**$147.21**\n\n**$190.47**\n\n**$151.62**\n\n**$178.70**\n\n**$231.05**\n\n**$115.88**\n\n**$171.06**\n\n**$128.15**\n\n**$168.44**\n\n**$232.43**\n\n2020 2021 2022 2023 2024\n\n($20)\n\n$0\n\n$20\n\n$40\n\n$60\n\n$80\n\n$100\n\n$120\n\n$140\n\n$0.00\n\n$50.00\n\n$100.00\n\n$150.00\n\n$200.00\n\n$250.00\n\nCompensation Actually Paid to CEO\n\nAverage Compensation Actually Paid to Non-CEO NEOs\n\nTotal Shareholder Return\n\nPeer Group Total Shareholder Return\n\nExecutive Compensation\n\nBLACKROCK, INC. 2025 PROXY STATEMENT 105","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/PRO013216_91_Blackrock_Proxy-Statement-2025_p112.pdf"]},"response":{"errors":[],"extract_id":"extract_20f8d5855ebcbb856e174e8a916beb5e","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Equity Award Valuation Assumptions: BPIP award grant date fair values are calculated using the stock price as of the grant date assuming target performance. Adjustments to these values have been made using the stock price and projected vesting assumptions as of each subsequent measurement date, assuming actual performance relative to the applicable performance targets through the applicable measurement date and target performance for years in the performance period beyond the applicable measurement date (based on the award determination matrix for the BPIP award granted in the January following the applicable measurement date). RSU values represent fair values which are calculated using the stock price as of each measurement date. Stock options represent performance-based option awards with vesting contingent on the NEOs’ continued service and BlackRock’s achievement of a stock price hurdle and positive Organic Revenue Growth. Stock option fair values for measurement dates prior to when the price hurdle was achieved are calculated by a Monte Carlo simulation model with an embedded binomial lattice model, which accounts for the risk of non-achievement. For stock option fair values occurring after the achievement of the price hurdle, the values are calculated by a binomial lattice model only. The stock option binomial lattice modelling uses a consistent set of valuation inputs (including BlackRock stock price, volatility, and dividend rates, market risk free rate, and the options’ strike price and term) which have been updated as of each measurement date, as applicable.\n\n(3) Represents the cumulative Total Shareholder Return (“TSR”) of BlackRock for the periods ending on December 31, 2024, 2023,\n\n2022, 2021 and 2020, respectively, based on an initial fixed investment of $100 in BlackRock common stock on December 31, 2019. For 2024, represents the five-year TSR (2020-2024), for 2023, represents the four-year TSR (2020-2023), for 2022, represents the three-year TSR (2020-2022), for 2021, represents the two-year TSR (2020-2021) and for 2020 represents the one- year TSR (2020).\n\n(4) Represents the cumulative TSR of the S&P U.S. BMI Asset Management & Custody Banks Index (as reported in BlackRock’s Annual\n\nReport on Form 10-K) for each of the periods ending on December 31, 2024, 2023, 2022, 2021 and 2020, respectively, based on an initial fixed investment of $100 in BlackRock common stock on December 31, 2019. For 2024, represents the five-year TSR (2020-2024), for 2023, represents the four-year TSR (2020-2023), for 2022, represents the three-year TSR (2020-2022), for 2021, represents the two-year TSR (2020-2021) and for 2020 represents the one-year TSR (2020).\n\n(5) Reflects net income attributable to BlackRock, Inc. in the Company’s Consolidated Statements of Income included in BlackRock’s\n\nAnnual Reports on Form 10-K for each of the fiscal years ended December 31, 2024, 2023, 2022, 2021 and 2020, respectively.\n\n(6) BlackRock’s Company-Selected Measure is Operating Margin, as adjusted. For reconciliation with GAAP for 2022-2024, please see\n\nAnnex A. For reconciliation with GAAP for 2020 and 2021, please see **page 37** of BlackRock’s Annual Report on Form 10-K for the year ended December 31, 2022.\n\n### **Relationship between Compensation Actually Paid and Performance Measures Disclosed in the Pay Versus Performance Table**\n\nThe following chart shows that the “compensation actually paid” to our CEO and average “compensation actually paid” to our other NEOs for each of the 2020, 2021, 2022, 2023 and 2024 fiscal years align to trends with BlackRock’s and our peer group’s (S&P U.S. BMI Asset Management & Custody Banks Index) TSR over the same period. BlackRock maintained a relatively consistent TSR premium relative to our peer group.\n\n**Compensation Actually Paid vs. TSR**\n\n**Compensation Actually Paid ($ millions)**\n\n**Total Shareholder Return**\n\n**(FYE 2019 Indexed to $100)**\n\n**$56.9**\n\n**$65.3**\n\n**($5.8)**\n\n**$26.9**\n\n**$56.5**\n\n**$36.2 $39.2**\n\n**($12.5)**\n\n**$20.0**\n\n**$36.6**\n\n**$147.21**\n\n**$190.47**\n\n**$151.62**\n\n**$178.70**\n\n**$231.05**\n\n**$115.88**\n\n**$171.06**\n\n**$128.15**\n\n**$168.44**\n\n**$232.43**\n\n2020 2021 2022 2023 2024\n\n($20)\n\n$0\n\n$20\n\n$40\n\n$60\n\n$80\n\n$100\n\n$120\n\n$140\n\n$0.00\n\n$50.00\n\n$100.00\n\n$150.00\n\n$200.00\n\n$250.00\n\nCompensation Actually Paid to CEO\n\nAverage Compensation Actually Paid to Non-CEO NEOs\n\nTotal Shareholder Return\n\nPeer Group Total Shareholder Return\n\nExecutive Compensation\n\nBLACKROCK, INC. 2025 PROXY STATEMENT 105"],"full_content":"Equity Award Valuation Assumptions: BPIP award grant date fair values are calculated using the stock price as of the grant date assuming target performance. Adjustments to these values have been made using the stock price and projected vesting assumptions as of each subsequent measurement date, assuming actual performance relative to the applicable performance targets through the applicable measurement date and target performance for years in the performance period beyond the applicable measurement date (based on the award determination matrix for the BPIP award granted in the January following the applicable measurement date). RSU values represent fair values which are calculated using the stock price as of each measurement date. Stock options represent performance-based option awards with vesting contingent on the NEOs’ continued service and BlackRock’s achievement of a stock price hurdle and positive Organic Revenue Growth. Stock option fair values for measurement dates prior to when the price hurdle was achieved are calculated by a Monte Carlo simulation model with an embedded binomial lattice model, which accounts for the risk of non-achievement. For stock option fair values occurring after the achievement of the price hurdle, the values are calculated by a binomial lattice model only. The stock option binomial lattice modelling uses a consistent set of valuation inputs (including BlackRock stock price, volatility, and dividend rates, market risk free rate, and the options’ strike price and term) which have been updated as of each measurement date, as applicable.\n\n(3) Represents the cumulative Total Shareholder Return (“TSR”) of BlackRock for the periods ending on December 31, 2024, 2023,\n\n2022, 2021 and 2020, respectively, based on an initial fixed investment of $100 in BlackRock common stock on December 31, 2019. For 2024, represents the five-year TSR (2020-2024), for 2023, represents the four-year TSR (2020-2023), for 2022, represents the three-year TSR (2020-2022), for 2021, represents the two-year TSR (2020-2021) and for 2020 represents the one- year TSR (2020).\n\n(4) Represents the cumulative TSR of the S&P U.S. BMI Asset Management & Custody Banks Index (as reported in BlackRock’s Annual\n\nReport on Form 10-K) for each of the periods ending on December 31, 2024, 2023, 2022, 2021 and 2020, respectively, based on an initial fixed investment of $100 in BlackRock common stock on December 31, 2019. For 2024, represents the five-year TSR (2020-2024), for 2023, represents the four-year TSR (2020-2023), for 2022, represents the three-year TSR (2020-2022), for 2021, represents the two-year TSR (2020-2021) and for 2020 represents the one-year TSR (2020).\n\n(5) Reflects net income attributable to BlackRock, Inc. in the Company’s Consolidated Statements of Income included in BlackRock’s\n\nAnnual Reports on Form 10-K for each of the fiscal years ended December 31, 2024, 2023, 2022, 2021 and 2020, respectively.\n\n(6) BlackRock’s Company-Selected Measure is Operating Margin, as adjusted. For reconciliation with GAAP for 2022-2024, please see\n\nAnnex A. For reconciliation with GAAP for 2020 and 2021, please see **page 37** of BlackRock’s Annual Report on Form 10-K for the year ended December 31, 2022.\n\n### **Relationship between Compensation Actually Paid and Performance Measures Disclosed in the Pay Versus Performance Table**\n\nThe following chart shows that the “compensation actually paid” to our CEO and average “compensation actually paid” to our other NEOs for each of the 2020, 2021, 2022, 2023 and 2024 fiscal years align to trends with BlackRock’s and our peer group’s (S&P U.S. BMI Asset Management & Custody Banks Index) TSR over the same period. BlackRock maintained a relatively consistent TSR premium relative to our peer group.\n\n**Compensation Actually Paid vs. TSR**\n\n**Compensation Actually Paid ($ millions)**\n\n**Total Shareholder Return**\n\n**(FYE 2019 Indexed to $100)**\n\n**$56.9**\n\n**$65.3**\n\n**($5.8)**\n\n**$26.9**\n\n**$56.5**\n\n**$36.2 $39.2**\n\n**($12.5)**\n\n**$20.0**\n\n**$36.6**\n\n**$147.21**\n\n**$190.47**\n\n**$151.62**\n\n**$178.70**\n\n**$231.05**\n\n**$115.88**\n\n**$171.06**\n\n**$128.15**\n\n**$168.44**\n\n**$232.43**\n\n2020 2021 2022 2023 2024\n\n($20)\n\n$0\n\n$20\n\n$40\n\n$60\n\n$80\n\n$100\n\n$120\n\n$140\n\n$0.00\n\n$50.00\n\n$100.00\n\n$150.00\n\n$200.00\n\n$250.00\n\nCompensation Actually Paid to CEO\n\nAverage Compensation Actually Paid to Non-CEO NEOs\n\nTotal Shareholder Return\n\nPeer Group Total Shareholder Return\n\nExecutive Compensation\n\nBLACKROCK, INC. 2025 PROXY STATEMENT 105","publish_date":null,"title":"Relationship between Compensation Actually Paid and Performance Measures Disclosed in the Pay Versus Performance Table","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/PRO013216_91_Blackrock_Proxy-Statement-2025_p112.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/PRO013216_91_Blackrock_Proxy-Statement-2025_p112.pdf"},"request":{"example_id":"chart/PRO013216_91_Blackrock_Proxy-Statement-2025_p112","product_type":"parse","source_file_path":"PRO013216_91_Blackrock_Proxy-Statement-2025_p112.pdf"},"started_at":"2026-11-04T19:45:41.539470"} +{"completed_at":"2026-11-04T19:45:43.934227","latency_in_ms":1195,"output":{"example_id":"chart/PRO013216_91_Blackrock_Proxy-Statement-2025_p71","grounded_pages":[],"layout_pages":[],"markdown":"## **Historical Outcomes – CEO and Other NEO Compensation Growth vs. BlackRock’s Financial Growth**\n\nThe contents of this section are supplemental to, and not intended to replace, the Pay Versus Performance disclosure made pursuant to Item 402(v) of Regulation S-K beginning on **page** **103** .\n\nOur assessment and pay determination process has resulted in disciplined pay levels that have been outpaced by BlackRock’s financial and market value growth over time. This is demonstrated by the 3% annualized growth in CEO Total Annual Compensation and 3% annualized growth in average non-CEO NEO Total Annual Compensation since 2014, during which time BlackRock’s Revenue, Operating Income, as adjusted, EPS, as adjusted, and Assets Under Management (AUM) have grown 6-10% per year. BlackRock has achieved annualized Total Shareholder Return of 14% per year over the period.\n\n### **10-Year BlackRock Annualized Growth Rates**\n\n### **3%**\n\nCEO Total Annual\n\nCompensation\n\n### **3%**\n\nAverage Non-CEO\n\nNEO Total\n\nAnnual Compensation\n\n### **6%**\n\nRevenue\n\n### **6%**\n\nOperating Income,\n\nas adjusted (1)\n\n### **8%**\n\nEPS, as\n\nadjusted (1)\n\n### **10%**\n\nAUM\n\n### **14%**\n\nTotal\n\nShareholder\n\nReturn\n\n(1) See Annex A for further information on as adjusted items and reconciliations to GAAP.\n\nThe graph below reflects BlackRock’s financial growth as well as CEO total compensation decisions during the period from 2015 to 2024.\n\n### **CEO Pay vs. Financial Growth**\n\n(1) Please see Annex A for reconciliations with GAAP for 2024, 2023 and 2022, our Form 10-K for the year ended December 31, 2023\n\nfor reconciliations with GAAP for 2021, 2020 and 2019 and our Form 10-K for the year ended December 31, 2019 for reconciliations with GAAP for 2018, 2017, 2016 and 2015.\n\nExecutive Compensation\n\n64 BLACKROCK, INC. 2025 PROXY STATEMENT","pages":[{"line_numbers":[],"markdown":"## **Historical Outcomes – CEO and Other NEO Compensation Growth vs. BlackRock’s Financial Growth**\n\nThe contents of this section are supplemental to, and not intended to replace, the Pay Versus Performance disclosure made pursuant to Item 402(v) of Regulation S-K beginning on **page** **103** .\n\nOur assessment and pay determination process has resulted in disciplined pay levels that have been outpaced by BlackRock’s financial and market value growth over time. This is demonstrated by the 3% annualized growth in CEO Total Annual Compensation and 3% annualized growth in average non-CEO NEO Total Annual Compensation since 2014, during which time BlackRock’s Revenue, Operating Income, as adjusted, EPS, as adjusted, and Assets Under Management (AUM) have grown 6-10% per year. BlackRock has achieved annualized Total Shareholder Return of 14% per year over the period.\n\n### **10-Year BlackRock Annualized Growth Rates**\n\n### **3%**\n\nCEO Total Annual\n\nCompensation\n\n### **3%**\n\nAverage Non-CEO\n\nNEO Total\n\nAnnual Compensation\n\n### **6%**\n\nRevenue\n\n### **6%**\n\nOperating Income,\n\nas adjusted (1)\n\n### **8%**\n\nEPS, as\n\nadjusted (1)\n\n### **10%**\n\nAUM\n\n### **14%**\n\nTotal\n\nShareholder\n\nReturn\n\n(1) See Annex A for further information on as adjusted items and reconciliations to GAAP.\n\nThe graph below reflects BlackRock’s financial growth as well as CEO total compensation decisions during the period from 2015 to 2024.\n\n### **CEO Pay vs. Financial Growth**\n\n(1) Please see Annex A for reconciliations with GAAP for 2024, 2023 and 2022, our Form 10-K for the year ended December 31, 2023\n\nfor reconciliations with GAAP for 2021, 2020 and 2019 and our Form 10-K for the year ended December 31, 2019 for reconciliations with GAAP for 2018, 2017, 2016 and 2015.\n\nExecutive Compensation\n\n64 BLACKROCK, INC. 2025 PROXY STATEMENT","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"## **Historical Outcomes – CEO and Other NEO Compensation Growth vs. BlackRock’s Financial Growth**\n\nThe contents of this section are supplemental to, and not intended to replace, the Pay Versus Performance disclosure made pursuant to Item 402(v) of Regulation S-K beginning on **page** **103** .\n\nOur assessment and pay determination process has resulted in disciplined pay levels that have been outpaced by BlackRock’s financial and market value growth over time. This is demonstrated by the 3% annualized growth in CEO Total Annual Compensation and 3% annualized growth in average non-CEO NEO Total Annual Compensation since 2014, during which time BlackRock’s Revenue, Operating Income, as adjusted, EPS, as adjusted, and Assets Under Management (AUM) have grown 6-10% per year. BlackRock has achieved annualized Total Shareholder Return of 14% per year over the period.\n\n### **10-Year BlackRock Annualized Growth Rates**\n\n### **3%**\n\nCEO Total Annual\n\nCompensation\n\n### **3%**\n\nAverage Non-CEO\n\nNEO Total\n\nAnnual Compensation\n\n### **6%**\n\nRevenue\n\n### **6%**\n\nOperating Income,\n\nas adjusted (1)\n\n### **8%**\n\nEPS, as\n\nadjusted (1)\n\n### **10%**\n\nAUM\n\n### **14%**\n\nTotal\n\nShareholder\n\nReturn\n\n(1) See Annex A for further information on as adjusted items and reconciliations to GAAP.\n\nThe graph below reflects BlackRock’s financial growth as well as CEO total compensation decisions during the period from 2015 to 2024.\n\n### **CEO Pay vs. Financial Growth**\n\n(1) Please see Annex A for reconciliations with GAAP for 2024, 2023 and 2022, our Form 10-K for the year ended December 31, 2023\n\nfor reconciliations with GAAP for 2021, 2020 and 2019 and our Form 10-K for the year ended December 31, 2019 for reconciliations with GAAP for 2018, 2017, 2016 and 2015.\n\nExecutive Compensation\n\n64 BLACKROCK, INC. 2025 PROXY STATEMENT","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/PRO013216_91_Blackrock_Proxy-Statement-2025_p71.pdf"]},"response":{"errors":[],"extract_id":"extract_944b6eeef1646fee5ba9581e7d9e0004","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["## **Historical Outcomes – CEO and Other NEO Compensation Growth vs. BlackRock’s Financial Growth**\n\nThe contents of this section are supplemental to, and not intended to replace, the Pay Versus Performance disclosure made pursuant to Item 402(v) of Regulation S-K beginning on **page** **103** .\n\nOur assessment and pay determination process has resulted in disciplined pay levels that have been outpaced by BlackRock’s financial and market value growth over time. This is demonstrated by the 3% annualized growth in CEO Total Annual Compensation and 3% annualized growth in average non-CEO NEO Total Annual Compensation since 2014, during which time BlackRock’s Revenue, Operating Income, as adjusted, EPS, as adjusted, and Assets Under Management (AUM) have grown 6-10% per year. BlackRock has achieved annualized Total Shareholder Return of 14% per year over the period.\n\n### **10-Year BlackRock Annualized Growth Rates**\n\n### **3%**\n\nCEO Total Annual\n\nCompensation\n\n### **3%**\n\nAverage Non-CEO\n\nNEO Total\n\nAnnual Compensation\n\n### **6%**\n\nRevenue\n\n### **6%**\n\nOperating Income,\n\nas adjusted (1)\n\n### **8%**\n\nEPS, as\n\nadjusted (1)\n\n### **10%**\n\nAUM\n\n### **14%**\n\nTotal\n\nShareholder\n\nReturn\n\n(1) See Annex A for further information on as adjusted items and reconciliations to GAAP.\n\nThe graph below reflects BlackRock’s financial growth as well as CEO total compensation decisions during the period from 2015 to 2024.\n\n### **CEO Pay vs. Financial Growth**\n\n(1) Please see Annex A for reconciliations with GAAP for 2024, 2023 and 2022, our Form 10-K for the year ended December 31, 2023\n\nfor reconciliations with GAAP for 2021, 2020 and 2019 and our Form 10-K for the year ended December 31, 2019 for reconciliations with GAAP for 2018, 2017, 2016 and 2015.\n\nExecutive Compensation\n\n64 BLACKROCK, INC. 2025 PROXY STATEMENT"],"full_content":"## **Historical Outcomes – CEO and Other NEO Compensation Growth vs. BlackRock’s Financial Growth**\n\nThe contents of this section are supplemental to, and not intended to replace, the Pay Versus Performance disclosure made pursuant to Item 402(v) of Regulation S-K beginning on **page** **103** .\n\nOur assessment and pay determination process has resulted in disciplined pay levels that have been outpaced by BlackRock’s financial and market value growth over time. This is demonstrated by the 3% annualized growth in CEO Total Annual Compensation and 3% annualized growth in average non-CEO NEO Total Annual Compensation since 2014, during which time BlackRock’s Revenue, Operating Income, as adjusted, EPS, as adjusted, and Assets Under Management (AUM) have grown 6-10% per year. BlackRock has achieved annualized Total Shareholder Return of 14% per year over the period.\n\n### **10-Year BlackRock Annualized Growth Rates**\n\n### **3%**\n\nCEO Total Annual\n\nCompensation\n\n### **3%**\n\nAverage Non-CEO\n\nNEO Total\n\nAnnual Compensation\n\n### **6%**\n\nRevenue\n\n### **6%**\n\nOperating Income,\n\nas adjusted (1)\n\n### **8%**\n\nEPS, as\n\nadjusted (1)\n\n### **10%**\n\nAUM\n\n### **14%**\n\nTotal\n\nShareholder\n\nReturn\n\n(1) See Annex A for further information on as adjusted items and reconciliations to GAAP.\n\nThe graph below reflects BlackRock’s financial growth as well as CEO total compensation decisions during the period from 2015 to 2024.\n\n### **CEO Pay vs. Financial Growth**\n\n(1) Please see Annex A for reconciliations with GAAP for 2024, 2023 and 2022, our Form 10-K for the year ended December 31, 2023\n\nfor reconciliations with GAAP for 2021, 2020 and 2019 and our Form 10-K for the year ended December 31, 2019 for reconciliations with GAAP for 2018, 2017, 2016 and 2015.\n\nExecutive Compensation\n\n64 BLACKROCK, INC. 2025 PROXY STATEMENT","publish_date":null,"title":"Historical Outcomes – CEO and Other NEO Compensation Growth vs. BlackRock’s Financial Growth","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/PRO013216_91_Blackrock_Proxy-Statement-2025_p71.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/PRO013216_91_Blackrock_Proxy-Statement-2025_p71.pdf"},"request":{"example_id":"chart/PRO013216_91_Blackrock_Proxy-Statement-2025_p71","product_type":"parse","source_file_path":"PRO013216_91_Blackrock_Proxy-Statement-2025_p71.pdf"},"started_at":"2026-11-04T19:45:42.738997"} +{"completed_at":"2026-11-04T19:45:48.102053","latency_in_ms":3325,"output":{"example_id":"chart/Quarterly-Sector-and-Investment-Research-Update-Q1-2025_p12","grounded_pages":[],"layout_pages":[],"markdown":"### Financials Price-to-Book Valuation Spreads\n\n# Financials’ performance may be durable this time\n\n## **Investors considering financials may be concerned about a repeat of 2016, when the sector underperformed despite low valuations and improving earnings. This time it looks different. The sector’s valuation spreads are near recessionary levels (left), suggesting the market is priced for bad news. Interest rates and credit spreads have declined—a good combination for the sector, historically (right).**\n\n**12**\n\n### **Financials have benefited from falling rates and credit spreads Valuation spreads are wide**\n\n### NTM Average Relative Performance when Credit Spreads (HYOAS) or Rates (1-Yr Yield) Are Rising or Falling, 1980–Present\n\n**Past performance is no guarantee of future results.** Sources: Haver Analytics, FactSet, Fidelity Investments, as of 11/30/24. LEFT: Valuation spread: The difference between the average price-to-book ratio of the most-expensive and least-expensive quartiles. Analysis based on Russell 3000. Data analyzed monthly from December 1990 to October 2024. RIGHT: NTM: Next 12 months. HYOAS: ICE BofA US High Yield Index Option-Adjusted Spread. Data analyzed monthly from January 1980 to October 2024.\n\n### 0\n\n### 50\n\n### 100\n\n### 150\n\n### 200\n\n### 250\n\n### 1990\n\n### 1992\n\n### 1994\n\n### 1995\n\n### 1997\n\n### 1998\n\n### 2000\n\n### 2002\n\n### 2003\n\n### 2005\n\n### 2006\n\n### 2008\n\n### 2009\n\n### 2011\n\n### 2013\n\n### 2014\n\n### 2016\n\n### 2017\n\n### 2019\n\n### 2021\n\n### 2022\n\n### 2024\n\n### \\-3%\n\n### \\-2%\n\n### \\-1%\n\n### 0%\n\n### 1%\n\n### 2%\n\n### 3%\n\n### 4%\n\n### 5%\n\n### Rates Rising/ Spreads Rising\n\n### Rates Rising/ Spreads Falling\n\n### Rates Falling/ Spreads Falling\n\n### Rates Falling/ Spreads Rising\n\n### **4%**\n\n### Price-to-book valuation spreads","pages":[{"line_numbers":[],"markdown":"### Financials Price-to-Book Valuation Spreads\n\n# Financials’ performance may be durable this time\n\n## **Investors considering financials may be concerned about a repeat of 2016, when the sector underperformed despite low valuations and improving earnings. This time it looks different. The sector’s valuation spreads are near recessionary levels (left), suggesting the market is priced for bad news. Interest rates and credit spreads have declined—a good combination for the sector, historically (right).**\n\n**12**\n\n### **Financials have benefited from falling rates and credit spreads Valuation spreads are wide**\n\n### NTM Average Relative Performance when Credit Spreads (HYOAS) or Rates (1-Yr Yield) Are Rising or Falling, 1980–Present\n\n**Past performance is no guarantee of future results.** Sources: Haver Analytics, FactSet, Fidelity Investments, as of 11/30/24. LEFT: Valuation spread: The difference between the average price-to-book ratio of the most-expensive and least-expensive quartiles. Analysis based on Russell 3000. Data analyzed monthly from December 1990 to October 2024. RIGHT: NTM: Next 12 months. HYOAS: ICE BofA US High Yield Index Option-Adjusted Spread. Data analyzed monthly from January 1980 to October 2024.\n\n### 0\n\n### 50\n\n### 100\n\n### 150\n\n### 200\n\n### 250\n\n### 1990\n\n### 1992\n\n### 1994\n\n### 1995\n\n### 1997\n\n### 1998\n\n### 2000\n\n### 2002\n\n### 2003\n\n### 2005\n\n### 2006\n\n### 2008\n\n### 2009\n\n### 2011\n\n### 2013\n\n### 2014\n\n### 2016\n\n### 2017\n\n### 2019\n\n### 2021\n\n### 2022\n\n### 2024\n\n### \\-3%\n\n### \\-2%\n\n### \\-1%\n\n### 0%\n\n### 1%\n\n### 2%\n\n### 3%\n\n### 4%\n\n### 5%\n\n### Rates Rising/ Spreads Rising\n\n### Rates Rising/ Spreads Falling\n\n### Rates Falling/ Spreads Falling\n\n### Rates Falling/ Spreads Rising\n\n### **4%**\n\n### Price-to-book valuation spreads","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"### Financials Price-to-Book Valuation Spreads\n\n# Financials’ performance may be durable this time\n\n## **Investors considering financials may be concerned about a repeat of 2016, when the sector underperformed despite low valuations and improving earnings. This time it looks different. The sector’s valuation spreads are near recessionary levels (left), suggesting the market is priced for bad news. Interest rates and credit spreads have declined—a good combination for the sector, historically (right).**\n\n**12**\n\n### **Financials have benefited from falling rates and credit spreads Valuation spreads are wide**\n\n### NTM Average Relative Performance when Credit Spreads (HYOAS) or Rates (1-Yr Yield) Are Rising or Falling, 1980–Present\n\n**Past performance is no guarantee of future results.** Sources: Haver Analytics, FactSet, Fidelity Investments, as of 11/30/24. LEFT: Valuation spread: The difference between the average price-to-book ratio of the most-expensive and least-expensive quartiles. Analysis based on Russell 3000. Data analyzed monthly from December 1990 to October 2024. RIGHT: NTM: Next 12 months. HYOAS: ICE BofA US High Yield Index Option-Adjusted Spread. Data analyzed monthly from January 1980 to October 2024.\n\n### 0\n\n### 50\n\n### 100\n\n### 150\n\n### 200\n\n### 250\n\n### 1990\n\n### 1992\n\n### 1994\n\n### 1995\n\n### 1997\n\n### 1998\n\n### 2000\n\n### 2002\n\n### 2003\n\n### 2005\n\n### 2006\n\n### 2008\n\n### 2009\n\n### 2011\n\n### 2013\n\n### 2014\n\n### 2016\n\n### 2017\n\n### 2019\n\n### 2021\n\n### 2022\n\n### 2024\n\n### \\-3%\n\n### \\-2%\n\n### \\-1%\n\n### 0%\n\n### 1%\n\n### 2%\n\n### 3%\n\n### 4%\n\n### 5%\n\n### Rates Rising/ Spreads Rising\n\n### Rates Rising/ Spreads Falling\n\n### Rates Falling/ Spreads Falling\n\n### Rates Falling/ Spreads Rising\n\n### **4%**\n\n### Price-to-book valuation spreads","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Quarterly-Sector-and-Investment-Research-Update-Q1-2025_p12.pdf"]},"response":{"errors":[],"extract_id":"extract_4bc335b4fc82c2cdfdc7feff9dfe92ab","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["### Financials Price-to-Book Valuation Spreads\n\n# Financials’ performance may be durable this time\n\n## **Investors considering financials may be concerned about a repeat of 2016, when the sector underperformed despite low valuations and improving earnings. This time it looks different. The sector’s valuation spreads are near recessionary levels (left), suggesting the market is priced for bad news. Interest rates and credit spreads have declined—a good combination for the sector, historically (right).**\n\n**12**\n\n### **Financials have benefited from falling rates and credit spreads Valuation spreads are wide**\n\n### NTM Average Relative Performance when Credit Spreads (HYOAS) or Rates (1-Yr Yield) Are Rising or Falling, 1980–Present\n\n**Past performance is no guarantee of future results.** Sources: Haver Analytics, FactSet, Fidelity Investments, as of 11/30/24. LEFT: Valuation spread: The difference between the average price-to-book ratio of the most-expensive and least-expensive quartiles. Analysis based on Russell 3000. Data analyzed monthly from December 1990 to October 2024. RIGHT: NTM: Next 12 months. HYOAS: ICE BofA US High Yield Index Option-Adjusted Spread. Data analyzed monthly from January 1980 to October 2024.\n\n### 0\n\n### 50\n\n### 100\n\n### 150\n\n### 200\n\n### 250\n\n### 1990\n\n### 1992\n\n### 1994\n\n### 1995\n\n### 1997\n\n### 1998\n\n### 2000\n\n### 2002\n\n### 2003\n\n### 2005\n\n### 2006\n\n### 2008\n\n### 2009\n\n### 2011\n\n### 2013\n\n### 2014\n\n### 2016\n\n### 2017\n\n### 2019\n\n### 2021\n\n### 2022\n\n### 2024\n\n### \\-3%\n\n### \\-2%\n\n### \\-1%\n\n### 0%\n\n### 1%\n\n### 2%\n\n### 3%\n\n### 4%\n\n### 5%\n\n### Rates Rising/ Spreads Rising\n\n### Rates Rising/ Spreads Falling\n\n### Rates Falling/ Spreads Falling\n\n### Rates Falling/ Spreads Rising\n\n### **4%**\n\n### Price-to-book valuation spreads"],"full_content":"### Financials Price-to-Book Valuation Spreads\n\n# Financials’ performance may be durable this time\n\n## **Investors considering financials may be concerned about a repeat of 2016, when the sector underperformed despite low valuations and improving earnings. This time it looks different. The sector’s valuation spreads are near recessionary levels (left), suggesting the market is priced for bad news. Interest rates and credit spreads have declined—a good combination for the sector, historically (right).**\n\n**12**\n\n### **Financials have benefited from falling rates and credit spreads Valuation spreads are wide**\n\n### NTM Average Relative Performance when Credit Spreads (HYOAS) or Rates (1-Yr Yield) Are Rising or Falling, 1980–Present\n\n**Past performance is no guarantee of future results.** Sources: Haver Analytics, FactSet, Fidelity Investments, as of 11/30/24. LEFT: Valuation spread: The difference between the average price-to-book ratio of the most-expensive and least-expensive quartiles. Analysis based on Russell 3000. Data analyzed monthly from December 1990 to October 2024. RIGHT: NTM: Next 12 months. HYOAS: ICE BofA US High Yield Index Option-Adjusted Spread. Data analyzed monthly from January 1980 to October 2024.\n\n### 0\n\n### 50\n\n### 100\n\n### 150\n\n### 200\n\n### 250\n\n### 1990\n\n### 1992\n\n### 1994\n\n### 1995\n\n### 1997\n\n### 1998\n\n### 2000\n\n### 2002\n\n### 2003\n\n### 2005\n\n### 2006\n\n### 2008\n\n### 2009\n\n### 2011\n\n### 2013\n\n### 2014\n\n### 2016\n\n### 2017\n\n### 2019\n\n### 2021\n\n### 2022\n\n### 2024\n\n### \\-3%\n\n### \\-2%\n\n### \\-1%\n\n### 0%\n\n### 1%\n\n### 2%\n\n### 3%\n\n### 4%\n\n### 5%\n\n### Rates Rising/ Spreads Rising\n\n### Rates Rising/ Spreads Falling\n\n### Rates Falling/ Spreads Falling\n\n### Rates Falling/ Spreads Rising\n\n### **4%**\n\n### Price-to-book valuation spreads","publish_date":null,"title":"Financials’ performance may be durable this time","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Quarterly-Sector-and-Investment-Research-Update-Q1-2025_p12.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Quarterly-Sector-and-Investment-Research-Update-Q1-2025_p12.pdf"},"request":{"example_id":"chart/Quarterly-Sector-and-Investment-Research-Update-Q1-2025_p12","product_type":"parse","source_file_path":"Quarterly-Sector-and-Investment-Research-Update-Q1-2025_p12.pdf"},"started_at":"2026-11-04T19:45:44.776638"} +{"completed_at":"2026-11-04T19:45:46.843056","latency_in_ms":928,"output":{"example_id":"chart/Quarterly-Sector-and-Investment-Research-Update-Q1-2025_p4","grounded_pages":[],"layout_pages":[],"markdown":"**4**\n\n### Average Annual CPI (January 2017–January 2020) with Period of 2018 Tariffs Highlighted\n\n**Past performance is no guarantee of future results.** Smoot-Hawley Tariff Act: U.S. legislation passed on June 17, 1930, that raised tariffs on agricultural imports. CPI is the Consumer Price Index, which seeks to measure the rate of change for price inflation among urban consumers. Sources: Haver Analytics, FactSet, Fidelity Investments, as of 11/30/24. **LEFT** : Data analyzed annually from 1923 through 1962. **RIGHT** : Data analyzed monthly from 2017 through January 2020.\n\n### **Inflation fell after the 2018 tariffs Deflation after the Smoot-Hawley Tariff Act**\n\n# Tariffs do not necessarily lead to higher inflation\n\n## **Many investors are concerned that the new Trump administration's potential imposition of tariffs could lead to higher inflation. They may want to be careful about investing based on that assumption. Consider the two major tariff episodes in U.S. history: The Smoot-Hawley Tariff Act of 1930 was followed by a period of deflation (left); and inflation declined after President Trump imposed tariffs in 2018 (right).**\n\n### Average Annual CPI (1923–1962) with Period of Smoot-Hawley Tariffs Highlighted\n\n### **Avg Annual Inflation 1930–Present**\n\n### \\-15%\n\n### \\-10%\n\n### \\-5%\n\n### 0%\n\n### 5%\n\n### 10%\n\n### 15%\n\n### 20%\n\n### 1920\n\n### 1922\n\n### 1924\n\n### 1926\n\n### 1928\n\n### 1930\n\n### 1932\n\n### 1934\n\n### 1936\n\n### 1938\n\n### 1940\n\n### 1942\n\n### 1944\n\n### 1946\n\n### 1948\n\n### 1950\n\n### 1952\n\n### 1954\n\n### 1956\n\n### 1958\n\n### 1960\n\n### 1962\n\n### 0\\.0%\n\n### 1\\.0%\n\n### 2\\.0%\n\n### 3\\.0%\n\n### 4\\.0%\n\n### 5\\.0%\n\n### Jan-17\n\n### Mar-17\n\n### May-17\n\n### Jul-17\n\n### Sep-17\n\n### Nov-17\n\n### Jan-18\n\n### Mar-18\n\n### May-18\n\n### Jul-18\n\n### Sep-18\n\n### Nov-18\n\n### Jan-19\n\n### Mar-19\n\n### May-19\n\n### Jul-19\n\n### Sep-19\n\n### Nov-19\n\n### Jan-20","pages":[{"line_numbers":[],"markdown":"**4**\n\n### Average Annual CPI (January 2017–January 2020) with Period of 2018 Tariffs Highlighted\n\n**Past performance is no guarantee of future results.** Smoot-Hawley Tariff Act: U.S. legislation passed on June 17, 1930, that raised tariffs on agricultural imports. CPI is the Consumer Price Index, which seeks to measure the rate of change for price inflation among urban consumers. Sources: Haver Analytics, FactSet, Fidelity Investments, as of 11/30/24. **LEFT** : Data analyzed annually from 1923 through 1962. **RIGHT** : Data analyzed monthly from 2017 through January 2020.\n\n### **Inflation fell after the 2018 tariffs Deflation after the Smoot-Hawley Tariff Act**\n\n# Tariffs do not necessarily lead to higher inflation\n\n## **Many investors are concerned that the new Trump administration's potential imposition of tariffs could lead to higher inflation. They may want to be careful about investing based on that assumption. Consider the two major tariff episodes in U.S. history: The Smoot-Hawley Tariff Act of 1930 was followed by a period of deflation (left); and inflation declined after President Trump imposed tariffs in 2018 (right).**\n\n### Average Annual CPI (1923–1962) with Period of Smoot-Hawley Tariffs Highlighted\n\n### **Avg Annual Inflation 1930–Present**\n\n### \\-15%\n\n### \\-10%\n\n### \\-5%\n\n### 0%\n\n### 5%\n\n### 10%\n\n### 15%\n\n### 20%\n\n### 1920\n\n### 1922\n\n### 1924\n\n### 1926\n\n### 1928\n\n### 1930\n\n### 1932\n\n### 1934\n\n### 1936\n\n### 1938\n\n### 1940\n\n### 1942\n\n### 1944\n\n### 1946\n\n### 1948\n\n### 1950\n\n### 1952\n\n### 1954\n\n### 1956\n\n### 1958\n\n### 1960\n\n### 1962\n\n### 0\\.0%\n\n### 1\\.0%\n\n### 2\\.0%\n\n### 3\\.0%\n\n### 4\\.0%\n\n### 5\\.0%\n\n### Jan-17\n\n### Mar-17\n\n### May-17\n\n### Jul-17\n\n### Sep-17\n\n### Nov-17\n\n### Jan-18\n\n### Mar-18\n\n### May-18\n\n### Jul-18\n\n### Sep-18\n\n### Nov-18\n\n### Jan-19\n\n### Mar-19\n\n### May-19\n\n### Jul-19\n\n### Sep-19\n\n### Nov-19\n\n### Jan-20","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**4**\n\n### Average Annual CPI (January 2017–January 2020) with Period of 2018 Tariffs Highlighted\n\n**Past performance is no guarantee of future results.** Smoot-Hawley Tariff Act: U.S. legislation passed on June 17, 1930, that raised tariffs on agricultural imports. CPI is the Consumer Price Index, which seeks to measure the rate of change for price inflation among urban consumers. Sources: Haver Analytics, FactSet, Fidelity Investments, as of 11/30/24. **LEFT** : Data analyzed annually from 1923 through 1962. **RIGHT** : Data analyzed monthly from 2017 through January 2020.\n\n### **Inflation fell after the 2018 tariffs Deflation after the Smoot-Hawley Tariff Act**\n\n# Tariffs do not necessarily lead to higher inflation\n\n## **Many investors are concerned that the new Trump administration's potential imposition of tariffs could lead to higher inflation. They may want to be careful about investing based on that assumption. Consider the two major tariff episodes in U.S. history: The Smoot-Hawley Tariff Act of 1930 was followed by a period of deflation (left); and inflation declined after President Trump imposed tariffs in 2018 (right).**\n\n### Average Annual CPI (1923–1962) with Period of Smoot-Hawley Tariffs Highlighted\n\n### **Avg Annual Inflation 1930–Present**\n\n### \\-15%\n\n### \\-10%\n\n### \\-5%\n\n### 0%\n\n### 5%\n\n### 10%\n\n### 15%\n\n### 20%\n\n### 1920\n\n### 1922\n\n### 1924\n\n### 1926\n\n### 1928\n\n### 1930\n\n### 1932\n\n### 1934\n\n### 1936\n\n### 1938\n\n### 1940\n\n### 1942\n\n### 1944\n\n### 1946\n\n### 1948\n\n### 1950\n\n### 1952\n\n### 1954\n\n### 1956\n\n### 1958\n\n### 1960\n\n### 1962\n\n### 0\\.0%\n\n### 1\\.0%\n\n### 2\\.0%\n\n### 3\\.0%\n\n### 4\\.0%\n\n### 5\\.0%\n\n### Jan-17\n\n### Mar-17\n\n### May-17\n\n### Jul-17\n\n### Sep-17\n\n### Nov-17\n\n### Jan-18\n\n### Mar-18\n\n### May-18\n\n### Jul-18\n\n### Sep-18\n\n### Nov-18\n\n### Jan-19\n\n### Mar-19\n\n### May-19\n\n### Jul-19\n\n### Sep-19\n\n### Nov-19\n\n### Jan-20","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Quarterly-Sector-and-Investment-Research-Update-Q1-2025_p4.pdf"]},"response":{"errors":[],"extract_id":"extract_9842bd6f64f285123bd9936a52dd1178","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**4**\n\n### Average Annual CPI (January 2017–January 2020) with Period of 2018 Tariffs Highlighted\n\n**Past performance is no guarantee of future results.** Smoot-Hawley Tariff Act: U.S. legislation passed on June 17, 1930, that raised tariffs on agricultural imports. CPI is the Consumer Price Index, which seeks to measure the rate of change for price inflation among urban consumers. Sources: Haver Analytics, FactSet, Fidelity Investments, as of 11/30/24. **LEFT** : Data analyzed annually from 1923 through 1962. **RIGHT** : Data analyzed monthly from 2017 through January 2020.\n\n### **Inflation fell after the 2018 tariffs Deflation after the Smoot-Hawley Tariff Act**\n\n# Tariffs do not necessarily lead to higher inflation\n\n## **Many investors are concerned that the new Trump administration's potential imposition of tariffs could lead to higher inflation. They may want to be careful about investing based on that assumption. Consider the two major tariff episodes in U.S. history: The Smoot-Hawley Tariff Act of 1930 was followed by a period of deflation (left); and inflation declined after President Trump imposed tariffs in 2018 (right).**\n\n### Average Annual CPI (1923–1962) with Period of Smoot-Hawley Tariffs Highlighted\n\n### **Avg Annual Inflation 1930–Present**\n\n### \\-15%\n\n### \\-10%\n\n### \\-5%\n\n### 0%\n\n### 5%\n\n### 10%\n\n### 15%\n\n### 20%\n\n### 1920\n\n### 1922\n\n### 1924\n\n### 1926\n\n### 1928\n\n### 1930\n\n### 1932\n\n### 1934\n\n### 1936\n\n### 1938\n\n### 1940\n\n### 1942\n\n### 1944\n\n### 1946\n\n### 1948\n\n### 1950\n\n### 1952\n\n### 1954\n\n### 1956\n\n### 1958\n\n### 1960\n\n### 1962\n\n### 0\\.0%\n\n### 1\\.0%\n\n### 2\\.0%\n\n### 3\\.0%\n\n### 4\\.0%\n\n### 5\\.0%\n\n### Jan-17\n\n### Mar-17\n\n### May-17\n\n### Jul-17\n\n### Sep-17\n\n### Nov-17\n\n### Jan-18\n\n### Mar-18\n\n### May-18\n\n### Jul-18\n\n### Sep-18\n\n### Nov-18\n\n### Jan-19\n\n### Mar-19\n\n### May-19\n\n### Jul-19\n\n### Sep-19\n\n### Nov-19\n\n### Jan-20"],"full_content":"**4**\n\n### Average Annual CPI (January 2017–January 2020) with Period of 2018 Tariffs Highlighted\n\n**Past performance is no guarantee of future results.** Smoot-Hawley Tariff Act: U.S. legislation passed on June 17, 1930, that raised tariffs on agricultural imports. CPI is the Consumer Price Index, which seeks to measure the rate of change for price inflation among urban consumers. Sources: Haver Analytics, FactSet, Fidelity Investments, as of 11/30/24. **LEFT** : Data analyzed annually from 1923 through 1962. **RIGHT** : Data analyzed monthly from 2017 through January 2020.\n\n### **Inflation fell after the 2018 tariffs Deflation after the Smoot-Hawley Tariff Act**\n\n# Tariffs do not necessarily lead to higher inflation\n\n## **Many investors are concerned that the new Trump administration's potential imposition of tariffs could lead to higher inflation. They may want to be careful about investing based on that assumption. Consider the two major tariff episodes in U.S. history: The Smoot-Hawley Tariff Act of 1930 was followed by a period of deflation (left); and inflation declined after President Trump imposed tariffs in 2018 (right).**\n\n### Average Annual CPI (1923–1962) with Period of Smoot-Hawley Tariffs Highlighted\n\n### **Avg Annual Inflation 1930–Present**\n\n### \\-15%\n\n### \\-10%\n\n### \\-5%\n\n### 0%\n\n### 5%\n\n### 10%\n\n### 15%\n\n### 20%\n\n### 1920\n\n### 1922\n\n### 1924\n\n### 1926\n\n### 1928\n\n### 1930\n\n### 1932\n\n### 1934\n\n### 1936\n\n### 1938\n\n### 1940\n\n### 1942\n\n### 1944\n\n### 1946\n\n### 1948\n\n### 1950\n\n### 1952\n\n### 1954\n\n### 1956\n\n### 1958\n\n### 1960\n\n### 1962\n\n### 0\\.0%\n\n### 1\\.0%\n\n### 2\\.0%\n\n### 3\\.0%\n\n### 4\\.0%\n\n### 5\\.0%\n\n### Jan-17\n\n### Mar-17\n\n### May-17\n\n### Jul-17\n\n### Sep-17\n\n### Nov-17\n\n### Jan-18\n\n### Mar-18\n\n### May-18\n\n### Jul-18\n\n### Sep-18\n\n### Nov-18\n\n### Jan-19\n\n### Mar-19\n\n### May-19\n\n### Jul-19\n\n### Sep-19\n\n### Nov-19\n\n### Jan-20","publish_date":null,"title":"Tariffs do not necessarily lead to higher inflation","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Quarterly-Sector-and-Investment-Research-Update-Q1-2025_p4.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Quarterly-Sector-and-Investment-Research-Update-Q1-2025_p4.pdf"},"request":{"example_id":"chart/Quarterly-Sector-and-Investment-Research-Update-Q1-2025_p4","product_type":"parse","source_file_path":"Quarterly-Sector-and-Investment-Research-Update-Q1-2025_p4.pdf"},"started_at":"2026-11-04T19:45:45.914762"} +{"completed_at":"2026-11-04T19:45:49.926283","latency_in_ms":2785,"output":{"example_id":"chart/Renewables2025_1_p110","grounded_pages":[],"layout_pages":[],"markdown":"Renewables 2025 Chapter 1. Renewable electricity Analysis and forecasts to 2030\n\nPAGE | 110\n\nI EA. CC BY 4.0.\n\n**Development of investment criteria in the past five years for new renewable generation and storage projects**\n\nIEA. CC BY 4.0.\n\nNotes: IRR = internal rate of return. WACC = weighted average cost of capital. The IRR to WACC spread represents the value created by the project over the minimum required return. Analysed companies (6): EDP S.A.; Enel SpA; ENGIE SA; Orsted A/S; RWE AG; SSE Renewables Limited. Source: IRR-WACC spread targets announced in companies’ annual reports, investment plans and publications of 2021- 2025.\n\n### Risk mitigation along the value chain: Focus on supply chains\n\nIn response to trade policy changes, industry players are redesigning activities along their value chains. For instance, major solar PV manufacturers are pushing for further vertical integration. By owning or managing more parts of the supply chain, they can increase control over costs and strategic planning. A prevalent strategy is to localise supply chain elements by establishing or boosting domestic production capacities. This can reduce costs by avoiding trade tariffs and increase resilience through self-sufficiency.\n\nFor wind manufactures, complex and strained logistics networks have created challenges in the past. Standardisation and reliance on proven solutions can reduce component-related costs, while designs adapted to transport-related infrastructure constraints (e.g. ports and shipping) can minimise bottlenecks.\n\n### Healthy Balance sheets: Steady debt and shareholder remuneration\n\nThe utilities and renewable IPPs we analysed often highlight their goal of maintaining a steady balance sheet and sustainable debt levels. Most of these large utilities and renewable IPPs have been (and plan to continue) sustaining a net debt-to-EBITDA ratio of less than four.\n\n0\n\n50\n\n100\n\n150\n\n200\n\n250\n\n300\n\n350\n\n400\n\n2021 2022 2023 2024 2025\n\nTarget IRR to WACC spreads (bps)\n\nAverage threshold\n\nHighest observed threshold\n\nLowest observed threshold","pages":[{"line_numbers":[],"markdown":"Renewables 2025 Chapter 1. Renewable electricity Analysis and forecasts to 2030\n\nPAGE | 110\n\nI EA. CC BY 4.0.\n\n**Development of investment criteria in the past five years for new renewable generation and storage projects**\n\nIEA. CC BY 4.0.\n\nNotes: IRR = internal rate of return. WACC = weighted average cost of capital. The IRR to WACC spread represents the value created by the project over the minimum required return. Analysed companies (6): EDP S.A.; Enel SpA; ENGIE SA; Orsted A/S; RWE AG; SSE Renewables Limited. Source: IRR-WACC spread targets announced in companies’ annual reports, investment plans and publications of 2021- 2025.\n\n### Risk mitigation along the value chain: Focus on supply chains\n\nIn response to trade policy changes, industry players are redesigning activities along their value chains. For instance, major solar PV manufacturers are pushing for further vertical integration. By owning or managing more parts of the supply chain, they can increase control over costs and strategic planning. A prevalent strategy is to localise supply chain elements by establishing or boosting domestic production capacities. This can reduce costs by avoiding trade tariffs and increase resilience through self-sufficiency.\n\nFor wind manufactures, complex and strained logistics networks have created challenges in the past. Standardisation and reliance on proven solutions can reduce component-related costs, while designs adapted to transport-related infrastructure constraints (e.g. ports and shipping) can minimise bottlenecks.\n\n### Healthy Balance sheets: Steady debt and shareholder remuneration\n\nThe utilities and renewable IPPs we analysed often highlight their goal of maintaining a steady balance sheet and sustainable debt levels. Most of these large utilities and renewable IPPs have been (and plan to continue) sustaining a net debt-to-EBITDA ratio of less than four.\n\n0\n\n50\n\n100\n\n150\n\n200\n\n250\n\n300\n\n350\n\n400\n\n2021 2022 2023 2024 2025\n\nTarget IRR to WACC spreads (bps)\n\nAverage threshold\n\nHighest observed threshold\n\nLowest observed threshold","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Renewables 2025 Chapter 1. Renewable electricity Analysis and forecasts to 2030\n\nPAGE | 110\n\nI EA. CC BY 4.0.\n\n**Development of investment criteria in the past five years for new renewable generation and storage projects**\n\nIEA. CC BY 4.0.\n\nNotes: IRR = internal rate of return. WACC = weighted average cost of capital. The IRR to WACC spread represents the value created by the project over the minimum required return. Analysed companies (6): EDP S.A.; Enel SpA; ENGIE SA; Orsted A/S; RWE AG; SSE Renewables Limited. Source: IRR-WACC spread targets announced in companies’ annual reports, investment plans and publications of 2021- 2025.\n\n### Risk mitigation along the value chain: Focus on supply chains\n\nIn response to trade policy changes, industry players are redesigning activities along their value chains. For instance, major solar PV manufacturers are pushing for further vertical integration. By owning or managing more parts of the supply chain, they can increase control over costs and strategic planning. A prevalent strategy is to localise supply chain elements by establishing or boosting domestic production capacities. This can reduce costs by avoiding trade tariffs and increase resilience through self-sufficiency.\n\nFor wind manufactures, complex and strained logistics networks have created challenges in the past. Standardisation and reliance on proven solutions can reduce component-related costs, while designs adapted to transport-related infrastructure constraints (e.g. ports and shipping) can minimise bottlenecks.\n\n### Healthy Balance sheets: Steady debt and shareholder remuneration\n\nThe utilities and renewable IPPs we analysed often highlight their goal of maintaining a steady balance sheet and sustainable debt levels. Most of these large utilities and renewable IPPs have been (and plan to continue) sustaining a net debt-to-EBITDA ratio of less than four.\n\n0\n\n50\n\n100\n\n150\n\n200\n\n250\n\n300\n\n350\n\n400\n\n2021 2022 2023 2024 2025\n\nTarget IRR to WACC spreads (bps)\n\nAverage threshold\n\nHighest observed threshold\n\nLowest observed threshold","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Renewables2025_1_p110.pdf"]},"response":{"errors":[],"extract_id":"extract_c8fe7f8f9e84f54d7f8c7262caddc850","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Renewables 2025 Chapter 1. Renewable electricity Analysis and forecasts to 2030\n\nPAGE | 110\n\nI EA. CC BY 4.0.\n\n**Development of investment criteria in the past five years for new renewable generation and storage projects**\n\nIEA. CC BY 4.0.\n\nNotes: IRR = internal rate of return. WACC = weighted average cost of capital. The IRR to WACC spread represents the value created by the project over the minimum required return. Analysed companies (6): EDP S.A.; Enel SpA; ENGIE SA; Orsted A/S; RWE AG; SSE Renewables Limited. Source: IRR-WACC spread targets announced in companies’ annual reports, investment plans and publications of 2021- 2025.\n\n### Risk mitigation along the value chain: Focus on supply chains\n\nIn response to trade policy changes, industry players are redesigning activities along their value chains. For instance, major solar PV manufacturers are pushing for further vertical integration. By owning or managing more parts of the supply chain, they can increase control over costs and strategic planning. A prevalent strategy is to localise supply chain elements by establishing or boosting domestic production capacities. This can reduce costs by avoiding trade tariffs and increase resilience through self-sufficiency.\n\nFor wind manufactures, complex and strained logistics networks have created challenges in the past. Standardisation and reliance on proven solutions can reduce component-related costs, while designs adapted to transport-related infrastructure constraints (e.g. ports and shipping) can minimise bottlenecks.\n\n### Healthy Balance sheets: Steady debt and shareholder remuneration\n\nThe utilities and renewable IPPs we analysed often highlight their goal of maintaining a steady balance sheet and sustainable debt levels. Most of these large utilities and renewable IPPs have been (and plan to continue) sustaining a net debt-to-EBITDA ratio of less than four.\n\n0\n\n50\n\n100\n\n150\n\n200\n\n250\n\n300\n\n350\n\n400\n\n2021 2022 2023 2024 2025\n\nTarget IRR to WACC spreads (bps)\n\nAverage threshold\n\nHighest observed threshold\n\nLowest observed threshold"],"full_content":"Renewables 2025 Chapter 1. Renewable electricity Analysis and forecasts to 2030\n\nPAGE | 110\n\nI EA. CC BY 4.0.\n\n**Development of investment criteria in the past five years for new renewable generation and storage projects**\n\nIEA. CC BY 4.0.\n\nNotes: IRR = internal rate of return. WACC = weighted average cost of capital. The IRR to WACC spread represents the value created by the project over the minimum required return. Analysed companies (6): EDP S.A.; Enel SpA; ENGIE SA; Orsted A/S; RWE AG; SSE Renewables Limited. Source: IRR-WACC spread targets announced in companies’ annual reports, investment plans and publications of 2021- 2025.\n\n### Risk mitigation along the value chain: Focus on supply chains\n\nIn response to trade policy changes, industry players are redesigning activities along their value chains. For instance, major solar PV manufacturers are pushing for further vertical integration. By owning or managing more parts of the supply chain, they can increase control over costs and strategic planning. A prevalent strategy is to localise supply chain elements by establishing or boosting domestic production capacities. This can reduce costs by avoiding trade tariffs and increase resilience through self-sufficiency.\n\nFor wind manufactures, complex and strained logistics networks have created challenges in the past. Standardisation and reliance on proven solutions can reduce component-related costs, while designs adapted to transport-related infrastructure constraints (e.g. ports and shipping) can minimise bottlenecks.\n\n### Healthy Balance sheets: Steady debt and shareholder remuneration\n\nThe utilities and renewable IPPs we analysed often highlight their goal of maintaining a steady balance sheet and sustainable debt levels. Most of these large utilities and renewable IPPs have been (and plan to continue) sustaining a net debt-to-EBITDA ratio of less than four.\n\n0\n\n50\n\n100\n\n150\n\n200\n\n250\n\n300\n\n350\n\n400\n\n2021 2022 2023 2024 2025\n\nTarget IRR to WACC spreads (bps)\n\nAverage threshold\n\nHighest observed threshold\n\nLowest observed threshold","publish_date":null,"title":"Risk mitigation along the value chain: Focus on supply chains Healthy Balance sheets: Steady debt and shareholder remuneration","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Renewables2025_1_p110.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Renewables2025_1_p110.pdf"},"request":{"example_id":"chart/Renewables2025_1_p110","product_type":"parse","source_file_path":"Renewables2025_1_p110.pdf"},"started_at":"2026-11-04T19:45:47.140982"} +{"completed_at":"2026-11-04T19:45:49.227624","latency_in_ms":1106,"output":{"example_id":"chart/Renewables2025_1_p111","grounded_pages":[],"layout_pages":[],"markdown":"Renewables 2025 Chapter 1. Renewable electricity Analysis and forecasts to 2030\n\nPAGE | 111\n\nI EA. CC BY 4.0.\n\nContinued growth ambitions are often coupled with announcements aimed at boosting investor confidence – either dividend payouts that are 5-15% higher than in 2024, or stock buyback plans. However, many still have a conservative payout strategy, retaining earnings for the investment pipeline.\n\nThe recent exit of six of the largest banks in the United States and the largest bank in the United Kingdom from the Net Zero Banking Alliance (NZBA) may also affect the use of financing to advance renewable energy deployment. In leaving the NZBA, these banks rescinded their commitment to align their lending and investment portfolios with climate targets and reduced the total assets of banks participating in the alliance by more than 20%.\n\n## **Profit and loss**\n\nBetween 2022 and 2024, prices for key solar PV components – including polysilicon, wafers, cells and modules – fell roughly 50%, mainly due to significant oversupply in China. Prices remained steadily low in the first half of 2025, placing continued pressure on manufacturer profitability throughout the entire supply chain.\n\n**Weighted average net margins of renewable energy companies and large utilities (Q1 2023-Q2 2025)**\n\nIEA. CC BY 4.0.\n\nNote: mfg = equipment manufacturing companies. Source: IEA analysis based on Bloomberg LP (2025), S&P Global Capital IQ (2025), annual reports, investment plans and publications for Q2 2025.\n\nDespite growing PV demand in the next five years, overcapacity is expected remain considerable through 2030 (see the section on supply chain security). The margins of many integrated Chinese manufacturers of polysilicon, wafers, cells and modules fell to around -10% in Q4 2024, and the cumulative net loss of\n\n\\-30%\n\n\\-20%\n\n\\-10%\n\n0%\n\n10%\n\n20%\n\n30%\n\nQ1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2\n\n2023 2024 2025\n\nWeighted average net margin\n\nPolysilicon Solar PV integrated mfg Utilities Wind integrated mfg\n\n**China**\n\n\\-30%\n\n\\-20%\n\n\\-10%\n\n0%\n\n10%\n\n20%\n\n30%\n\nQ1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2\n\n2023 2024 2025\n\n**Outside of China**","pages":[{"line_numbers":[],"markdown":"Renewables 2025 Chapter 1. Renewable electricity Analysis and forecasts to 2030\n\nPAGE | 111\n\nI EA. CC BY 4.0.\n\nContinued growth ambitions are often coupled with announcements aimed at boosting investor confidence – either dividend payouts that are 5-15% higher than in 2024, or stock buyback plans. However, many still have a conservative payout strategy, retaining earnings for the investment pipeline.\n\nThe recent exit of six of the largest banks in the United States and the largest bank in the United Kingdom from the Net Zero Banking Alliance (NZBA) may also affect the use of financing to advance renewable energy deployment. In leaving the NZBA, these banks rescinded their commitment to align their lending and investment portfolios with climate targets and reduced the total assets of banks participating in the alliance by more than 20%.\n\n## **Profit and loss**\n\nBetween 2022 and 2024, prices for key solar PV components – including polysilicon, wafers, cells and modules – fell roughly 50%, mainly due to significant oversupply in China. Prices remained steadily low in the first half of 2025, placing continued pressure on manufacturer profitability throughout the entire supply chain.\n\n**Weighted average net margins of renewable energy companies and large utilities (Q1 2023-Q2 2025)**\n\nIEA. CC BY 4.0.\n\nNote: mfg = equipment manufacturing companies. Source: IEA analysis based on Bloomberg LP (2025), S&P Global Capital IQ (2025), annual reports, investment plans and publications for Q2 2025.\n\nDespite growing PV demand in the next five years, overcapacity is expected remain considerable through 2030 (see the section on supply chain security). The margins of many integrated Chinese manufacturers of polysilicon, wafers, cells and modules fell to around -10% in Q4 2024, and the cumulative net loss of\n\n\\-30%\n\n\\-20%\n\n\\-10%\n\n0%\n\n10%\n\n20%\n\n30%\n\nQ1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2\n\n2023 2024 2025\n\nWeighted average net margin\n\nPolysilicon Solar PV integrated mfg Utilities Wind integrated mfg\n\n**China**\n\n\\-30%\n\n\\-20%\n\n\\-10%\n\n0%\n\n10%\n\n20%\n\n30%\n\nQ1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2\n\n2023 2024 2025\n\n**Outside of China**","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Renewables 2025 Chapter 1. Renewable electricity Analysis and forecasts to 2030\n\nPAGE | 111\n\nI EA. CC BY 4.0.\n\nContinued growth ambitions are often coupled with announcements aimed at boosting investor confidence – either dividend payouts that are 5-15% higher than in 2024, or stock buyback plans. However, many still have a conservative payout strategy, retaining earnings for the investment pipeline.\n\nThe recent exit of six of the largest banks in the United States and the largest bank in the United Kingdom from the Net Zero Banking Alliance (NZBA) may also affect the use of financing to advance renewable energy deployment. In leaving the NZBA, these banks rescinded their commitment to align their lending and investment portfolios with climate targets and reduced the total assets of banks participating in the alliance by more than 20%.\n\n## **Profit and loss**\n\nBetween 2022 and 2024, prices for key solar PV components – including polysilicon, wafers, cells and modules – fell roughly 50%, mainly due to significant oversupply in China. Prices remained steadily low in the first half of 2025, placing continued pressure on manufacturer profitability throughout the entire supply chain.\n\n**Weighted average net margins of renewable energy companies and large utilities (Q1 2023-Q2 2025)**\n\nIEA. CC BY 4.0.\n\nNote: mfg = equipment manufacturing companies. Source: IEA analysis based on Bloomberg LP (2025), S&P Global Capital IQ (2025), annual reports, investment plans and publications for Q2 2025.\n\nDespite growing PV demand in the next five years, overcapacity is expected remain considerable through 2030 (see the section on supply chain security). The margins of many integrated Chinese manufacturers of polysilicon, wafers, cells and modules fell to around -10% in Q4 2024, and the cumulative net loss of\n\n\\-30%\n\n\\-20%\n\n\\-10%\n\n0%\n\n10%\n\n20%\n\n30%\n\nQ1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2\n\n2023 2024 2025\n\nWeighted average net margin\n\nPolysilicon Solar PV integrated mfg Utilities Wind integrated mfg\n\n**China**\n\n\\-30%\n\n\\-20%\n\n\\-10%\n\n0%\n\n10%\n\n20%\n\n30%\n\nQ1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2\n\n2023 2024 2025\n\n**Outside of China**","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Renewables2025_1_p111.pdf"]},"response":{"errors":[],"extract_id":"extract_fa0535ac0a9360f242bfab9e2ed416f7","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Renewables 2025 Chapter 1. Renewable electricity Analysis and forecasts to 2030\n\nPAGE | 111\n\nI EA. CC BY 4.0.\n\nContinued growth ambitions are often coupled with announcements aimed at boosting investor confidence – either dividend payouts that are 5-15% higher than in 2024, or stock buyback plans. However, many still have a conservative payout strategy, retaining earnings for the investment pipeline.\n\nThe recent exit of six of the largest banks in the United States and the largest bank in the United Kingdom from the Net Zero Banking Alliance (NZBA) may also affect the use of financing to advance renewable energy deployment. In leaving the NZBA, these banks rescinded their commitment to align their lending and investment portfolios with climate targets and reduced the total assets of banks participating in the alliance by more than 20%.\n\n## **Profit and loss**\n\nBetween 2022 and 2024, prices for key solar PV components – including polysilicon, wafers, cells and modules – fell roughly 50%, mainly due to significant oversupply in China. Prices remained steadily low in the first half of 2025, placing continued pressure on manufacturer profitability throughout the entire supply chain.\n\n**Weighted average net margins of renewable energy companies and large utilities (Q1 2023-Q2 2025)**\n\nIEA. CC BY 4.0.\n\nNote: mfg = equipment manufacturing companies. Source: IEA analysis based on Bloomberg LP (2025), S&P Global Capital IQ (2025), annual reports, investment plans and publications for Q2 2025.\n\nDespite growing PV demand in the next five years, overcapacity is expected remain considerable through 2030 (see the section on supply chain security). The margins of many integrated Chinese manufacturers of polysilicon, wafers, cells and modules fell to around -10% in Q4 2024, and the cumulative net loss of\n\n\\-30%\n\n\\-20%\n\n\\-10%\n\n0%\n\n10%\n\n20%\n\n30%\n\nQ1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2\n\n2023 2024 2025\n\nWeighted average net margin\n\nPolysilicon Solar PV integrated mfg Utilities Wind integrated mfg\n\n**China**\n\n\\-30%\n\n\\-20%\n\n\\-10%\n\n0%\n\n10%\n\n20%\n\n30%\n\nQ1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2\n\n2023 2024 2025\n\n**Outside of China**"],"full_content":"Renewables 2025 Chapter 1. Renewable electricity Analysis and forecasts to 2030\n\nPAGE | 111\n\nI EA. CC BY 4.0.\n\nContinued growth ambitions are often coupled with announcements aimed at boosting investor confidence – either dividend payouts that are 5-15% higher than in 2024, or stock buyback plans. However, many still have a conservative payout strategy, retaining earnings for the investment pipeline.\n\nThe recent exit of six of the largest banks in the United States and the largest bank in the United Kingdom from the Net Zero Banking Alliance (NZBA) may also affect the use of financing to advance renewable energy deployment. In leaving the NZBA, these banks rescinded their commitment to align their lending and investment portfolios with climate targets and reduced the total assets of banks participating in the alliance by more than 20%.\n\n## **Profit and loss**\n\nBetween 2022 and 2024, prices for key solar PV components – including polysilicon, wafers, cells and modules – fell roughly 50%, mainly due to significant oversupply in China. Prices remained steadily low in the first half of 2025, placing continued pressure on manufacturer profitability throughout the entire supply chain.\n\n**Weighted average net margins of renewable energy companies and large utilities (Q1 2023-Q2 2025)**\n\nIEA. CC BY 4.0.\n\nNote: mfg = equipment manufacturing companies. Source: IEA analysis based on Bloomberg LP (2025), S&P Global Capital IQ (2025), annual reports, investment plans and publications for Q2 2025.\n\nDespite growing PV demand in the next five years, overcapacity is expected remain considerable through 2030 (see the section on supply chain security). The margins of many integrated Chinese manufacturers of polysilicon, wafers, cells and modules fell to around -10% in Q4 2024, and the cumulative net loss of\n\n\\-30%\n\n\\-20%\n\n\\-10%\n\n0%\n\n10%\n\n20%\n\n30%\n\nQ1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2\n\n2023 2024 2025\n\nWeighted average net margin\n\nPolysilicon Solar PV integrated mfg Utilities Wind integrated mfg\n\n**China**\n\n\\-30%\n\n\\-20%\n\n\\-10%\n\n0%\n\n10%\n\n20%\n\n30%\n\nQ1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2\n\n2023 2024 2025\n\n**Outside of China**","publish_date":null,"title":"Profit and loss","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Renewables2025_1_p111.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Renewables2025_1_p111.pdf"},"request":{"example_id":"chart/Renewables2025_1_p111","product_type":"parse","source_file_path":"Renewables2025_1_p111.pdf"},"started_at":"2026-11-04T19:45:48.120961"} +{"completed_at":"2026-11-04T19:45:50.921056","latency_in_ms":1290,"output":{"example_id":"chart/Renewables2025_1_p213","grounded_pages":[],"layout_pages":[],"markdown":"Chapter 3. Renewable heat Renewables 2025 Analysis and forecasts to 2030\n\nPAGE | 213\n\nI EA. CC BY 4.0.\n\nThe global solar heat for industrial processes (SHIP) market rebounded strongly in 2024, reaching a five-year high with a 30% increase in new installed capacity. A total of 106 new systems added 120 MWth, led by China with an 80-MWth parabolic trough project for a leisure park. Other highlights are Latin America’s first linear Fresnel plant in Mexico (111 kWth of capacity), producing 180°C steam; Germany’s demonstration plant, which began producing renewable fuels in September 2024 and has since secured several offtake agreements with airports and the shipping industry; Kenya’s 180-kW solar thermal system with 1 MWh of storage for a tea plantation; and a 4.1-MWth brewery in Spain that uses linear Fresnel collectors.\n\nIn Croatia, construction began in 2024 on a winning Innovation Fund solar thermal project (combined with thermal storage, a waste heat recovery system and two heat pumps). However, despite the overall rise in SHIP capacity, the total number of new projects dropped slightly, mainly due to a 30% subsidy cut in the Netherlands that led to several project cancellations.\n\nAt the end of 2024, the global SHIP market totalled 1 315 systems with a combined capacity of 1 071 MWth and 1.5 million m ² of collector area. The technology mix has also changed, with concentrating collectors making up 69% of new capacity, heavily influenced by the aforementioned large-scale project in China. The market is also trending towards larger systems, with average system size growing more than 40% in the past year alone.\n\n**Annual industrial solar thermal capacity increases and number of commissioned SHIP projects, 2018-2024 (left), and regional shares of total industrial solar thermal capacity, 2024 (right)**\n\nIEA. CC BY 4.0.\n\nNote: SHIP = solar heat for industrial processes. Source: IEA Technology Collaboration Programme (2023), Solar Industrial Heat Outlook 2023-2026.\n\n0\n\n20\n\n40\n\n60\n\n80\n\n100\n\n120\n\n0\n\n50\n\n100\n\n150\n\n200\n\n250\n\n300\n\n2018 2019 2020 2021 2022 2023 2024\n\nNumber of SHIP systems\n\nMW\n\nAnnual capacity increases and commissioned SHIP systems\n\nNewly installed SHIP capacity Number of commissioned SHIP systems\n\n0\n\n200\n\n400\n\n600\n\n800\n\n1 000\n\n1 200\n\nMW\n\nRegional shares of total\n\ncapacity, 2024\n\nRest of world\n\nCentral and South America South Africa\n\nChina","pages":[{"line_numbers":[],"markdown":"Chapter 3. Renewable heat Renewables 2025 Analysis and forecasts to 2030\n\nPAGE | 213\n\nI EA. CC BY 4.0.\n\nThe global solar heat for industrial processes (SHIP) market rebounded strongly in 2024, reaching a five-year high with a 30% increase in new installed capacity. A total of 106 new systems added 120 MWth, led by China with an 80-MWth parabolic trough project for a leisure park. Other highlights are Latin America’s first linear Fresnel plant in Mexico (111 kWth of capacity), producing 180°C steam; Germany’s demonstration plant, which began producing renewable fuels in September 2024 and has since secured several offtake agreements with airports and the shipping industry; Kenya’s 180-kW solar thermal system with 1 MWh of storage for a tea plantation; and a 4.1-MWth brewery in Spain that uses linear Fresnel collectors.\n\nIn Croatia, construction began in 2024 on a winning Innovation Fund solar thermal project (combined with thermal storage, a waste heat recovery system and two heat pumps). However, despite the overall rise in SHIP capacity, the total number of new projects dropped slightly, mainly due to a 30% subsidy cut in the Netherlands that led to several project cancellations.\n\nAt the end of 2024, the global SHIP market totalled 1 315 systems with a combined capacity of 1 071 MWth and 1.5 million m ² of collector area. The technology mix has also changed, with concentrating collectors making up 69% of new capacity, heavily influenced by the aforementioned large-scale project in China. The market is also trending towards larger systems, with average system size growing more than 40% in the past year alone.\n\n**Annual industrial solar thermal capacity increases and number of commissioned SHIP projects, 2018-2024 (left), and regional shares of total industrial solar thermal capacity, 2024 (right)**\n\nIEA. CC BY 4.0.\n\nNote: SHIP = solar heat for industrial processes. Source: IEA Technology Collaboration Programme (2023), Solar Industrial Heat Outlook 2023-2026.\n\n0\n\n20\n\n40\n\n60\n\n80\n\n100\n\n120\n\n0\n\n50\n\n100\n\n150\n\n200\n\n250\n\n300\n\n2018 2019 2020 2021 2022 2023 2024\n\nNumber of SHIP systems\n\nMW\n\nAnnual capacity increases and commissioned SHIP systems\n\nNewly installed SHIP capacity Number of commissioned SHIP systems\n\n0\n\n200\n\n400\n\n600\n\n800\n\n1 000\n\n1 200\n\nMW\n\nRegional shares of total\n\ncapacity, 2024\n\nRest of world\n\nCentral and South America South Africa\n\nChina","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Chapter 3. Renewable heat Renewables 2025 Analysis and forecasts to 2030\n\nPAGE | 213\n\nI EA. CC BY 4.0.\n\nThe global solar heat for industrial processes (SHIP) market rebounded strongly in 2024, reaching a five-year high with a 30% increase in new installed capacity. A total of 106 new systems added 120 MWth, led by China with an 80-MWth parabolic trough project for a leisure park. Other highlights are Latin America’s first linear Fresnel plant in Mexico (111 kWth of capacity), producing 180°C steam; Germany’s demonstration plant, which began producing renewable fuels in September 2024 and has since secured several offtake agreements with airports and the shipping industry; Kenya’s 180-kW solar thermal system with 1 MWh of storage for a tea plantation; and a 4.1-MWth brewery in Spain that uses linear Fresnel collectors.\n\nIn Croatia, construction began in 2024 on a winning Innovation Fund solar thermal project (combined with thermal storage, a waste heat recovery system and two heat pumps). However, despite the overall rise in SHIP capacity, the total number of new projects dropped slightly, mainly due to a 30% subsidy cut in the Netherlands that led to several project cancellations.\n\nAt the end of 2024, the global SHIP market totalled 1 315 systems with a combined capacity of 1 071 MWth and 1.5 million m ² of collector area. The technology mix has also changed, with concentrating collectors making up 69% of new capacity, heavily influenced by the aforementioned large-scale project in China. The market is also trending towards larger systems, with average system size growing more than 40% in the past year alone.\n\n**Annual industrial solar thermal capacity increases and number of commissioned SHIP projects, 2018-2024 (left), and regional shares of total industrial solar thermal capacity, 2024 (right)**\n\nIEA. CC BY 4.0.\n\nNote: SHIP = solar heat for industrial processes. Source: IEA Technology Collaboration Programme (2023), Solar Industrial Heat Outlook 2023-2026.\n\n0\n\n20\n\n40\n\n60\n\n80\n\n100\n\n120\n\n0\n\n50\n\n100\n\n150\n\n200\n\n250\n\n300\n\n2018 2019 2020 2021 2022 2023 2024\n\nNumber of SHIP systems\n\nMW\n\nAnnual capacity increases and commissioned SHIP systems\n\nNewly installed SHIP capacity Number of commissioned SHIP systems\n\n0\n\n200\n\n400\n\n600\n\n800\n\n1 000\n\n1 200\n\nMW\n\nRegional shares of total\n\ncapacity, 2024\n\nRest of world\n\nCentral and South America South Africa\n\nChina","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Renewables2025_1_p213.pdf"]},"response":{"errors":[],"extract_id":"extract_1772febf2310ea2c90fb3e2487743c77","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Chapter 3. Renewable heat Renewables 2025 Analysis and forecasts to 2030\n\nPAGE | 213\n\nI EA. CC BY 4.0.\n\nThe global solar heat for industrial processes (SHIP) market rebounded strongly in 2024, reaching a five-year high with a 30% increase in new installed capacity. A total of 106 new systems added 120 MWth, led by China with an 80-MWth parabolic trough project for a leisure park. Other highlights are Latin America’s first linear Fresnel plant in Mexico (111 kWth of capacity), producing 180°C steam; Germany’s demonstration plant, which began producing renewable fuels in September 2024 and has since secured several offtake agreements with airports and the shipping industry; Kenya’s 180-kW solar thermal system with 1 MWh of storage for a tea plantation; and a 4.1-MWth brewery in Spain that uses linear Fresnel collectors.\n\nIn Croatia, construction began in 2024 on a winning Innovation Fund solar thermal project (combined with thermal storage, a waste heat recovery system and two heat pumps). However, despite the overall rise in SHIP capacity, the total number of new projects dropped slightly, mainly due to a 30% subsidy cut in the Netherlands that led to several project cancellations.\n\nAt the end of 2024, the global SHIP market totalled 1 315 systems with a combined capacity of 1 071 MWth and 1.5 million m ² of collector area. The technology mix has also changed, with concentrating collectors making up 69% of new capacity, heavily influenced by the aforementioned large-scale project in China. The market is also trending towards larger systems, with average system size growing more than 40% in the past year alone.\n\n**Annual industrial solar thermal capacity increases and number of commissioned SHIP projects, 2018-2024 (left), and regional shares of total industrial solar thermal capacity, 2024 (right)**\n\nIEA. CC BY 4.0.\n\nNote: SHIP = solar heat for industrial processes. Source: IEA Technology Collaboration Programme (2023), Solar Industrial Heat Outlook 2023-2026.\n\n0\n\n20\n\n40\n\n60\n\n80\n\n100\n\n120\n\n0\n\n50\n\n100\n\n150\n\n200\n\n250\n\n300\n\n2018 2019 2020 2021 2022 2023 2024\n\nNumber of SHIP systems\n\nMW\n\nAnnual capacity increases and commissioned SHIP systems\n\nNewly installed SHIP capacity Number of commissioned SHIP systems\n\n0\n\n200\n\n400\n\n600\n\n800\n\n1 000\n\n1 200\n\nMW\n\nRegional shares of total\n\ncapacity, 2024\n\nRest of world\n\nCentral and South America South Africa\n\nChina"],"full_content":"Chapter 3. Renewable heat Renewables 2025 Analysis and forecasts to 2030\n\nPAGE | 213\n\nI EA. CC BY 4.0.\n\nThe global solar heat for industrial processes (SHIP) market rebounded strongly in 2024, reaching a five-year high with a 30% increase in new installed capacity. A total of 106 new systems added 120 MWth, led by China with an 80-MWth parabolic trough project for a leisure park. Other highlights are Latin America’s first linear Fresnel plant in Mexico (111 kWth of capacity), producing 180°C steam; Germany’s demonstration plant, which began producing renewable fuels in September 2024 and has since secured several offtake agreements with airports and the shipping industry; Kenya’s 180-kW solar thermal system with 1 MWh of storage for a tea plantation; and a 4.1-MWth brewery in Spain that uses linear Fresnel collectors.\n\nIn Croatia, construction began in 2024 on a winning Innovation Fund solar thermal project (combined with thermal storage, a waste heat recovery system and two heat pumps). However, despite the overall rise in SHIP capacity, the total number of new projects dropped slightly, mainly due to a 30% subsidy cut in the Netherlands that led to several project cancellations.\n\nAt the end of 2024, the global SHIP market totalled 1 315 systems with a combined capacity of 1 071 MWth and 1.5 million m ² of collector area. The technology mix has also changed, with concentrating collectors making up 69% of new capacity, heavily influenced by the aforementioned large-scale project in China. The market is also trending towards larger systems, with average system size growing more than 40% in the past year alone.\n\n**Annual industrial solar thermal capacity increases and number of commissioned SHIP projects, 2018-2024 (left), and regional shares of total industrial solar thermal capacity, 2024 (right)**\n\nIEA. CC BY 4.0.\n\nNote: SHIP = solar heat for industrial processes. Source: IEA Technology Collaboration Programme (2023), Solar Industrial Heat Outlook 2023-2026.\n\n0\n\n20\n\n40\n\n60\n\n80\n\n100\n\n120\n\n0\n\n50\n\n100\n\n150\n\n200\n\n250\n\n300\n\n2018 2019 2020 2021 2022 2023 2024\n\nNumber of SHIP systems\n\nMW\n\nAnnual capacity increases and commissioned SHIP systems\n\nNewly installed SHIP capacity Number of commissioned SHIP systems\n\n0\n\n200\n\n400\n\n600\n\n800\n\n1 000\n\n1 200\n\nMW\n\nRegional shares of total\n\ncapacity, 2024\n\nRest of world\n\nCentral and South America South Africa\n\nChina","publish_date":null,"title":"The global solar heat for industrial processes (SHIP) market rebounded strongly in 2024, reaching a five-year high with a 30%","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Renewables2025_1_p213.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Renewables2025_1_p213.pdf"},"request":{"example_id":"chart/Renewables2025_1_p213","product_type":"parse","source_file_path":"Renewables2025_1_p213.pdf"},"started_at":"2026-11-04T19:45:49.630591"} +{"completed_at":"2026-11-04T19:45:53.238350","latency_in_ms":2466,"output":{"example_id":"chart/Renewables2025_1_p24","grounded_pages":[],"layout_pages":[],"markdown":"Renewables 2025 Chapter 1. Renewable electricity Analysis and forecasts to 2030\n\nPAGE | 24\n\nI EA. CC BY 4.0.\n\npreviously to a trend-based assessment supported by historical data (see the section below on the role of wind and solar PV in power systems).\n\nSolar PV alone accounts for over 60% of the generation increase, followed by wind (32%). The share of renewables in global electricity generation is projected to rise from 32% in 2024 to 43% by 2030, while the share of variable renewable energy sources is set to almost double to 28%.\n\n**Global renewable power generation in 2024 and 2030, and change by technology**\n\nIEA. CC BY 4.0.\n\nNotes: VRE = variable renewable energy. “VRE share” includes solar PV and wind.\n\nHydropower electricity generation is expected to increase 7% over 2025-2030 as new projects become operational, mostly in emerging and developing countries. However, its share in global electricity generation decreases slightly, by almost one percentage point to 14% in 2030. Hydropower will account for just 30% of global renewable electricity generation in 2030, a sharp decline from over 80% two decades ago.\n\nIn contrast, the role of solar PV and wind increases drastically: in 2030, variable renewables account for almost two-thirds of global renewable electricity generation, rising from less than 46% today. Solar PV is expected to generate more electricity than hydropower by the end of the forecast period and to become the largest renewable energy source. Onshore and offshore wind generation are also expected to grow rapidly, together reaching over 4 500 TWh by 2030.\n\nClearly, the increased use of variable renewables raises the need for additional sources of power system flexibility. Nevertheless, bioenergy, geothermal and concentrated solar power expansions remain limited despite their critical role in integrating wind and solar PV generation into electricity systems around the world.\n\n15%\n\n28%\n\n32%\n\n43%\n\n0%\n\n10%\n\n20%\n\n30%\n\n40%\n\n50%\n\n60%\n\n70%\n\n80%\n\n90%\n\n0\n\n2 000\n\n4 000\n\n6 000\n\n8 000\n\n10 000\n\n12 000\n\n14 000\n\n16 000\n\n18 000\n\n2024 Hydropower Solar PV Wind Bioenergy Other\n\nrenewables\n\n2030\n\nGeneration Change Generation\n\nTWh/year\n\nVRE share Renewables share","pages":[{"line_numbers":[],"markdown":"Renewables 2025 Chapter 1. Renewable electricity Analysis and forecasts to 2030\n\nPAGE | 24\n\nI EA. CC BY 4.0.\n\npreviously to a trend-based assessment supported by historical data (see the section below on the role of wind and solar PV in power systems).\n\nSolar PV alone accounts for over 60% of the generation increase, followed by wind (32%). The share of renewables in global electricity generation is projected to rise from 32% in 2024 to 43% by 2030, while the share of variable renewable energy sources is set to almost double to 28%.\n\n**Global renewable power generation in 2024 and 2030, and change by technology**\n\nIEA. CC BY 4.0.\n\nNotes: VRE = variable renewable energy. “VRE share” includes solar PV and wind.\n\nHydropower electricity generation is expected to increase 7% over 2025-2030 as new projects become operational, mostly in emerging and developing countries. However, its share in global electricity generation decreases slightly, by almost one percentage point to 14% in 2030. Hydropower will account for just 30% of global renewable electricity generation in 2030, a sharp decline from over 80% two decades ago.\n\nIn contrast, the role of solar PV and wind increases drastically: in 2030, variable renewables account for almost two-thirds of global renewable electricity generation, rising from less than 46% today. Solar PV is expected to generate more electricity than hydropower by the end of the forecast period and to become the largest renewable energy source. Onshore and offshore wind generation are also expected to grow rapidly, together reaching over 4 500 TWh by 2030.\n\nClearly, the increased use of variable renewables raises the need for additional sources of power system flexibility. Nevertheless, bioenergy, geothermal and concentrated solar power expansions remain limited despite their critical role in integrating wind and solar PV generation into electricity systems around the world.\n\n15%\n\n28%\n\n32%\n\n43%\n\n0%\n\n10%\n\n20%\n\n30%\n\n40%\n\n50%\n\n60%\n\n70%\n\n80%\n\n90%\n\n0\n\n2 000\n\n4 000\n\n6 000\n\n8 000\n\n10 000\n\n12 000\n\n14 000\n\n16 000\n\n18 000\n\n2024 Hydropower Solar PV Wind Bioenergy Other\n\nrenewables\n\n2030\n\nGeneration Change Generation\n\nTWh/year\n\nVRE share Renewables share","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Renewables 2025 Chapter 1. Renewable electricity Analysis and forecasts to 2030\n\nPAGE | 24\n\nI EA. CC BY 4.0.\n\npreviously to a trend-based assessment supported by historical data (see the section below on the role of wind and solar PV in power systems).\n\nSolar PV alone accounts for over 60% of the generation increase, followed by wind (32%). The share of renewables in global electricity generation is projected to rise from 32% in 2024 to 43% by 2030, while the share of variable renewable energy sources is set to almost double to 28%.\n\n**Global renewable power generation in 2024 and 2030, and change by technology**\n\nIEA. CC BY 4.0.\n\nNotes: VRE = variable renewable energy. “VRE share” includes solar PV and wind.\n\nHydropower electricity generation is expected to increase 7% over 2025-2030 as new projects become operational, mostly in emerging and developing countries. However, its share in global electricity generation decreases slightly, by almost one percentage point to 14% in 2030. Hydropower will account for just 30% of global renewable electricity generation in 2030, a sharp decline from over 80% two decades ago.\n\nIn contrast, the role of solar PV and wind increases drastically: in 2030, variable renewables account for almost two-thirds of global renewable electricity generation, rising from less than 46% today. Solar PV is expected to generate more electricity than hydropower by the end of the forecast period and to become the largest renewable energy source. Onshore and offshore wind generation are also expected to grow rapidly, together reaching over 4 500 TWh by 2030.\n\nClearly, the increased use of variable renewables raises the need for additional sources of power system flexibility. Nevertheless, bioenergy, geothermal and concentrated solar power expansions remain limited despite their critical role in integrating wind and solar PV generation into electricity systems around the world.\n\n15%\n\n28%\n\n32%\n\n43%\n\n0%\n\n10%\n\n20%\n\n30%\n\n40%\n\n50%\n\n60%\n\n70%\n\n80%\n\n90%\n\n0\n\n2 000\n\n4 000\n\n6 000\n\n8 000\n\n10 000\n\n12 000\n\n14 000\n\n16 000\n\n18 000\n\n2024 Hydropower Solar PV Wind Bioenergy Other\n\nrenewables\n\n2030\n\nGeneration Change Generation\n\nTWh/year\n\nVRE share Renewables share","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Renewables2025_1_p24.pdf"]},"response":{"errors":[],"extract_id":"extract_2352326da5c8334b5d793b323231bcce","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Renewables 2025 Chapter 1. Renewable electricity Analysis and forecasts to 2030\n\nPAGE | 24\n\nI EA. CC BY 4.0.\n\npreviously to a trend-based assessment supported by historical data (see the section below on the role of wind and solar PV in power systems).\n\nSolar PV alone accounts for over 60% of the generation increase, followed by wind (32%). The share of renewables in global electricity generation is projected to rise from 32% in 2024 to 43% by 2030, while the share of variable renewable energy sources is set to almost double to 28%.\n\n**Global renewable power generation in 2024 and 2030, and change by technology**\n\nIEA. CC BY 4.0.\n\nNotes: VRE = variable renewable energy. “VRE share” includes solar PV and wind.\n\nHydropower electricity generation is expected to increase 7% over 2025-2030 as new projects become operational, mostly in emerging and developing countries. However, its share in global electricity generation decreases slightly, by almost one percentage point to 14% in 2030. Hydropower will account for just 30% of global renewable electricity generation in 2030, a sharp decline from over 80% two decades ago.\n\nIn contrast, the role of solar PV and wind increases drastically: in 2030, variable renewables account for almost two-thirds of global renewable electricity generation, rising from less than 46% today. Solar PV is expected to generate more electricity than hydropower by the end of the forecast period and to become the largest renewable energy source. Onshore and offshore wind generation are also expected to grow rapidly, together reaching over 4 500 TWh by 2030.\n\nClearly, the increased use of variable renewables raises the need for additional sources of power system flexibility. Nevertheless, bioenergy, geothermal and concentrated solar power expansions remain limited despite their critical role in integrating wind and solar PV generation into electricity systems around the world.\n\n15%\n\n28%\n\n32%\n\n43%\n\n0%\n\n10%\n\n20%\n\n30%\n\n40%\n\n50%\n\n60%\n\n70%\n\n80%\n\n90%\n\n0\n\n2 000\n\n4 000\n\n6 000\n\n8 000\n\n10 000\n\n12 000\n\n14 000\n\n16 000\n\n18 000\n\n2024 Hydropower Solar PV Wind Bioenergy Other\n\nrenewables\n\n2030\n\nGeneration Change Generation\n\nTWh/year\n\nVRE share Renewables share"],"full_content":"Renewables 2025 Chapter 1. Renewable electricity Analysis and forecasts to 2030\n\nPAGE | 24\n\nI EA. CC BY 4.0.\n\npreviously to a trend-based assessment supported by historical data (see the section below on the role of wind and solar PV in power systems).\n\nSolar PV alone accounts for over 60% of the generation increase, followed by wind (32%). The share of renewables in global electricity generation is projected to rise from 32% in 2024 to 43% by 2030, while the share of variable renewable energy sources is set to almost double to 28%.\n\n**Global renewable power generation in 2024 and 2030, and change by technology**\n\nIEA. CC BY 4.0.\n\nNotes: VRE = variable renewable energy. “VRE share” includes solar PV and wind.\n\nHydropower electricity generation is expected to increase 7% over 2025-2030 as new projects become operational, mostly in emerging and developing countries. However, its share in global electricity generation decreases slightly, by almost one percentage point to 14% in 2030. Hydropower will account for just 30% of global renewable electricity generation in 2030, a sharp decline from over 80% two decades ago.\n\nIn contrast, the role of solar PV and wind increases drastically: in 2030, variable renewables account for almost two-thirds of global renewable electricity generation, rising from less than 46% today. Solar PV is expected to generate more electricity than hydropower by the end of the forecast period and to become the largest renewable energy source. Onshore and offshore wind generation are also expected to grow rapidly, together reaching over 4 500 TWh by 2030.\n\nClearly, the increased use of variable renewables raises the need for additional sources of power system flexibility. Nevertheless, bioenergy, geothermal and concentrated solar power expansions remain limited despite their critical role in integrating wind and solar PV generation into electricity systems around the world.\n\n15%\n\n28%\n\n32%\n\n43%\n\n0%\n\n10%\n\n20%\n\n30%\n\n40%\n\n50%\n\n60%\n\n70%\n\n80%\n\n90%\n\n0\n\n2 000\n\n4 000\n\n6 000\n\n8 000\n\n10 000\n\n12 000\n\n14 000\n\n16 000\n\n18 000\n\n2024 Hydropower Solar PV Wind Bioenergy Other\n\nrenewables\n\n2030\n\nGeneration Change Generation\n\nTWh/year\n\nVRE share Renewables share","publish_date":null,"title":"previously to a trend-based assessment supported by historical data (see the section below on the role of wind and solar","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Renewables2025_1_p24.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Renewables2025_1_p24.pdf"},"request":{"example_id":"chart/Renewables2025_1_p24","product_type":"parse","source_file_path":"Renewables2025_1_p24.pdf"},"started_at":"2026-11-04T19:45:50.771643"} +{"completed_at":"2026-11-04T19:45:54.326282","latency_in_ms":1464,"output":{"example_id":"chart/Renewables2025_1_p30","grounded_pages":[],"layout_pages":[],"markdown":"Renewables 2025 Chapter 1. Renewable electricity Analysis and forecasts to 2030\n\nPAGE | 30\n\nI EA. CC BY 4.0.\n\nBelgium’s temporary VAT exemption. Our forecast underestimated the impacts of these changes on 2024 deployment, so we realigned our expectations with 2024 growth.\n\nIn other markets, policy changes included switching from net metering to net billing to achieve more cost-effective deployment of distributed PV. While this switch lowers the remuneration value for the consumer, it incentivises consumption during peak hours and also reduces grid congestion as well as the amount of grid costs passed on to consumers. Because our forecast underestimated the impacts of these changes on 2024 deployment in Poland and parts of Belgium, we have revised our expectations downwards for these countries as well as for the Netherlands, which recently announced plans to also transition by 2027.\n\n**Residential sector retail electricity prices and annual capacity additions in selected European countries, 2021-2024**\n\nIEA. CC BY 4.0.\n\nAdditional downward revisions reflect cuts to FITs in France, proposed tax-subsidy changes in Sweden, ongoing policy uncertainty in Switzerland, and weaker-than- expected H1-2025 deployment in Germany. These downward revisions outweigh upward adjustments to 17 countries for which we underestimated growth in 2024. In some cases, we simply undervalued the business case despite falling power prices (Hungary, Austria and Greece), while in others (e.g. Ireland and Portugal) elevated prices and continued policy support maintained stronger-than-expected growth.\n\nOffshore wind capacity is forecast to grow 57 GW by 2030; however, this is 9 GW (15%) less than what we projected in Renewables 2024. This reduction reflects an increasingly challenging business case for planned projects and extended project timelines. Rising costs, supply chain constraints and uncertainty around future electricity prices have raised concerns about project viability, impacting over 5 GW of the forecast.\n\n\\-4.0\n\n\\-3.0\n\n\\-2.0\n\n\\-1.0\n\n0\\.0\n\n1\\.0\n\n2\\.0\n\n3\\.0\n\n4\\.0\n\n0\n\n100\n\n200\n\n300\n\n400\n\n500\n\n600\n\n700\n\n800\n\nBelgium Germany Italy Netherlands Spain\n\nGW\n\nEUR/MWh\n\n2024 2023 2022 2021 Net additions (right axis)","pages":[{"line_numbers":[],"markdown":"Renewables 2025 Chapter 1. Renewable electricity Analysis and forecasts to 2030\n\nPAGE | 30\n\nI EA. CC BY 4.0.\n\nBelgium’s temporary VAT exemption. Our forecast underestimated the impacts of these changes on 2024 deployment, so we realigned our expectations with 2024 growth.\n\nIn other markets, policy changes included switching from net metering to net billing to achieve more cost-effective deployment of distributed PV. While this switch lowers the remuneration value for the consumer, it incentivises consumption during peak hours and also reduces grid congestion as well as the amount of grid costs passed on to consumers. Because our forecast underestimated the impacts of these changes on 2024 deployment in Poland and parts of Belgium, we have revised our expectations downwards for these countries as well as for the Netherlands, which recently announced plans to also transition by 2027.\n\n**Residential sector retail electricity prices and annual capacity additions in selected European countries, 2021-2024**\n\nIEA. CC BY 4.0.\n\nAdditional downward revisions reflect cuts to FITs in France, proposed tax-subsidy changes in Sweden, ongoing policy uncertainty in Switzerland, and weaker-than- expected H1-2025 deployment in Germany. These downward revisions outweigh upward adjustments to 17 countries for which we underestimated growth in 2024. In some cases, we simply undervalued the business case despite falling power prices (Hungary, Austria and Greece), while in others (e.g. Ireland and Portugal) elevated prices and continued policy support maintained stronger-than-expected growth.\n\nOffshore wind capacity is forecast to grow 57 GW by 2030; however, this is 9 GW (15%) less than what we projected in Renewables 2024. This reduction reflects an increasingly challenging business case for planned projects and extended project timelines. Rising costs, supply chain constraints and uncertainty around future electricity prices have raised concerns about project viability, impacting over 5 GW of the forecast.\n\n\\-4.0\n\n\\-3.0\n\n\\-2.0\n\n\\-1.0\n\n0\\.0\n\n1\\.0\n\n2\\.0\n\n3\\.0\n\n4\\.0\n\n0\n\n100\n\n200\n\n300\n\n400\n\n500\n\n600\n\n700\n\n800\n\nBelgium Germany Italy Netherlands Spain\n\nGW\n\nEUR/MWh\n\n2024 2023 2022 2021 Net additions (right axis)","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Renewables 2025 Chapter 1. Renewable electricity Analysis and forecasts to 2030\n\nPAGE | 30\n\nI EA. CC BY 4.0.\n\nBelgium’s temporary VAT exemption. Our forecast underestimated the impacts of these changes on 2024 deployment, so we realigned our expectations with 2024 growth.\n\nIn other markets, policy changes included switching from net metering to net billing to achieve more cost-effective deployment of distributed PV. While this switch lowers the remuneration value for the consumer, it incentivises consumption during peak hours and also reduces grid congestion as well as the amount of grid costs passed on to consumers. Because our forecast underestimated the impacts of these changes on 2024 deployment in Poland and parts of Belgium, we have revised our expectations downwards for these countries as well as for the Netherlands, which recently announced plans to also transition by 2027.\n\n**Residential sector retail electricity prices and annual capacity additions in selected European countries, 2021-2024**\n\nIEA. CC BY 4.0.\n\nAdditional downward revisions reflect cuts to FITs in France, proposed tax-subsidy changes in Sweden, ongoing policy uncertainty in Switzerland, and weaker-than- expected H1-2025 deployment in Germany. These downward revisions outweigh upward adjustments to 17 countries for which we underestimated growth in 2024. In some cases, we simply undervalued the business case despite falling power prices (Hungary, Austria and Greece), while in others (e.g. Ireland and Portugal) elevated prices and continued policy support maintained stronger-than-expected growth.\n\nOffshore wind capacity is forecast to grow 57 GW by 2030; however, this is 9 GW (15%) less than what we projected in Renewables 2024. This reduction reflects an increasingly challenging business case for planned projects and extended project timelines. Rising costs, supply chain constraints and uncertainty around future electricity prices have raised concerns about project viability, impacting over 5 GW of the forecast.\n\n\\-4.0\n\n\\-3.0\n\n\\-2.0\n\n\\-1.0\n\n0\\.0\n\n1\\.0\n\n2\\.0\n\n3\\.0\n\n4\\.0\n\n0\n\n100\n\n200\n\n300\n\n400\n\n500\n\n600\n\n700\n\n800\n\nBelgium Germany Italy Netherlands Spain\n\nGW\n\nEUR/MWh\n\n2024 2023 2022 2021 Net additions (right axis)","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Renewables2025_1_p30.pdf"]},"response":{"errors":[],"extract_id":"extract_71504837be863df01b2df53cec94973d","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Renewables 2025 Chapter 1. Renewable electricity Analysis and forecasts to 2030\n\nPAGE | 30\n\nI EA. CC BY 4.0.\n\nBelgium’s temporary VAT exemption. Our forecast underestimated the impacts of these changes on 2024 deployment, so we realigned our expectations with 2024 growth.\n\nIn other markets, policy changes included switching from net metering to net billing to achieve more cost-effective deployment of distributed PV. While this switch lowers the remuneration value for the consumer, it incentivises consumption during peak hours and also reduces grid congestion as well as the amount of grid costs passed on to consumers. Because our forecast underestimated the impacts of these changes on 2024 deployment in Poland and parts of Belgium, we have revised our expectations downwards for these countries as well as for the Netherlands, which recently announced plans to also transition by 2027.\n\n**Residential sector retail electricity prices and annual capacity additions in selected European countries, 2021-2024**\n\nIEA. CC BY 4.0.\n\nAdditional downward revisions reflect cuts to FITs in France, proposed tax-subsidy changes in Sweden, ongoing policy uncertainty in Switzerland, and weaker-than- expected H1-2025 deployment in Germany. These downward revisions outweigh upward adjustments to 17 countries for which we underestimated growth in 2024. In some cases, we simply undervalued the business case despite falling power prices (Hungary, Austria and Greece), while in others (e.g. Ireland and Portugal) elevated prices and continued policy support maintained stronger-than-expected growth.\n\nOffshore wind capacity is forecast to grow 57 GW by 2030; however, this is 9 GW (15%) less than what we projected in Renewables 2024. This reduction reflects an increasingly challenging business case for planned projects and extended project timelines. Rising costs, supply chain constraints and uncertainty around future electricity prices have raised concerns about project viability, impacting over 5 GW of the forecast.\n\n\\-4.0\n\n\\-3.0\n\n\\-2.0\n\n\\-1.0\n\n0\\.0\n\n1\\.0\n\n2\\.0\n\n3\\.0\n\n4\\.0\n\n0\n\n100\n\n200\n\n300\n\n400\n\n500\n\n600\n\n700\n\n800\n\nBelgium Germany Italy Netherlands Spain\n\nGW\n\nEUR/MWh\n\n2024 2023 2022 2021 Net additions (right axis)"],"full_content":"Renewables 2025 Chapter 1. Renewable electricity Analysis and forecasts to 2030\n\nPAGE | 30\n\nI EA. CC BY 4.0.\n\nBelgium’s temporary VAT exemption. Our forecast underestimated the impacts of these changes on 2024 deployment, so we realigned our expectations with 2024 growth.\n\nIn other markets, policy changes included switching from net metering to net billing to achieve more cost-effective deployment of distributed PV. While this switch lowers the remuneration value for the consumer, it incentivises consumption during peak hours and also reduces grid congestion as well as the amount of grid costs passed on to consumers. Because our forecast underestimated the impacts of these changes on 2024 deployment in Poland and parts of Belgium, we have revised our expectations downwards for these countries as well as for the Netherlands, which recently announced plans to also transition by 2027.\n\n**Residential sector retail electricity prices and annual capacity additions in selected European countries, 2021-2024**\n\nIEA. CC BY 4.0.\n\nAdditional downward revisions reflect cuts to FITs in France, proposed tax-subsidy changes in Sweden, ongoing policy uncertainty in Switzerland, and weaker-than- expected H1-2025 deployment in Germany. These downward revisions outweigh upward adjustments to 17 countries for which we underestimated growth in 2024. In some cases, we simply undervalued the business case despite falling power prices (Hungary, Austria and Greece), while in others (e.g. Ireland and Portugal) elevated prices and continued policy support maintained stronger-than-expected growth.\n\nOffshore wind capacity is forecast to grow 57 GW by 2030; however, this is 9 GW (15%) less than what we projected in Renewables 2024. This reduction reflects an increasingly challenging business case for planned projects and extended project timelines. Rising costs, supply chain constraints and uncertainty around future electricity prices have raised concerns about project viability, impacting over 5 GW of the forecast.\n\n\\-4.0\n\n\\-3.0\n\n\\-2.0\n\n\\-1.0\n\n0\\.0\n\n1\\.0\n\n2\\.0\n\n3\\.0\n\n4\\.0\n\n0\n\n100\n\n200\n\n300\n\n400\n\n500\n\n600\n\n700\n\n800\n\nBelgium Germany Italy Netherlands Spain\n\nGW\n\nEUR/MWh\n\n2024 2023 2022 2021 Net additions (right axis)","publish_date":null,"title":"Belgium’s temporary VAT exemption. Our forecast underestimated the impacts of these changes on 2024 deployment, so we realigned our expectations","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Renewables2025_1_p30.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Renewables2025_1_p30.pdf"},"request":{"example_id":"chart/Renewables2025_1_p30","product_type":"parse","source_file_path":"Renewables2025_1_p30.pdf"},"started_at":"2026-11-04T19:45:52.861528"} +{"completed_at":"2026-11-04T19:45:55.407497","latency_in_ms":1079,"output":{"example_id":"chart/Renewables2025_1_p32","grounded_pages":[],"layout_pages":[],"markdown":"Renewables 2025 Chapter 1. Renewable electricity Analysis and forecasts to 2030\n\nPAGE | 32\n\nI EA. CC BY 4.0.\n\n**EU installed renewable capacity in 2024 and 2030 vs REPowerEU 2030 targets (left) and shares of renewables in final energy consumption in 2030 (right)**\n\nIEA. CC BY 4.0.\n\nNote: RED = Renewable Energy Directive. NECP = National Energy and Climate Plan. Sources: REPowerEU ambitions for total renewable capacity are from REPowerEU Plan SWD (2022) 230 final; wind and solar PV are from Implementing the REpowerEU Action Plan. The solar PV ambition is in AC, as it is similar to the “almost 600 MW by 2030” target identified in the EU Solar Energy Strategy SWD (2022) 148 final. The 2030 EU offshore wind ambition is from Delivering on the EU Offshore Renewable Energy Ambitions, and the onshore wind aim is the difference between the 520-GW REPowerEU ambition and the 111-GW offshore wind target in Delivering on the EU Offshore Renewable Energy Ambitions. All solar PV values are in AC, including for the main and accelerated cases, and all solar PV totals are calculated in AC. The member-state ambition is estimated from NECPs.\n\nIn June 2024, member states submitted their final updated NECPs outlining their voluntary contributions towards the EU goal. However, the European Commission’s assessment of these final NECPs indicates that, in aggregate, they reach only a 40% renewable share – falling 2.5 percentage points short of the binding EU target and 5 percentage points below the REPowerEU ambition. This gap suggests that REPowerEU capacity aims are not aligned with final individual NECPs.\n\nThe second reason for the forecast shortfall in meeting EU-wide ambitions is that persistent challenges to faster solar and wind uptake require stronger policy attention. In the accelerated case, EU capacity reaches 1 255 GW ⁵ by 2030, on track to reach the REPower EU target. Achieving more rapid expansion will require four key policy actions:\n\n **Introduce auctions to provide revenue certainty.** Because auctions for long- term contracts provide revenue certainty, financing costs can drop. Clearer visibility over future auction schedules would help developers plan where to concentrate resources and co-ordinate future investment efforts.\n\n5 For this total, solar PV is calculated in AC instead of DC, and pumped storage is excluded for comparability with the 1 236-GW REPowerEU target. Total capacity in the main case with solar PV calculated in DC is 756 GW in 2024 and 1 123 GW in 2030; for the accelerated case it is 1 260 GW.\n\n0\n\n250\n\n500\n\n750\n\n1 000\n\n1 250\n\n1 500\n\n2024 2030 2030\n\nREPowerEU\n\nGW\n\nAccelerated Other Wind Solar PV\n\n25%\n\n30%\n\n35%\n\n40%\n\n45%\n\n50%\n\nRED II RED III\n\nBinding target Aspirtational target\n\nREPowerEU target Assessment of final updated NECPs","pages":[{"line_numbers":[],"markdown":"Renewables 2025 Chapter 1. Renewable electricity Analysis and forecasts to 2030\n\nPAGE | 32\n\nI EA. CC BY 4.0.\n\n**EU installed renewable capacity in 2024 and 2030 vs REPowerEU 2030 targets (left) and shares of renewables in final energy consumption in 2030 (right)**\n\nIEA. CC BY 4.0.\n\nNote: RED = Renewable Energy Directive. NECP = National Energy and Climate Plan. Sources: REPowerEU ambitions for total renewable capacity are from REPowerEU Plan SWD (2022) 230 final; wind and solar PV are from Implementing the REpowerEU Action Plan. The solar PV ambition is in AC, as it is similar to the “almost 600 MW by 2030” target identified in the EU Solar Energy Strategy SWD (2022) 148 final. The 2030 EU offshore wind ambition is from Delivering on the EU Offshore Renewable Energy Ambitions, and the onshore wind aim is the difference between the 520-GW REPowerEU ambition and the 111-GW offshore wind target in Delivering on the EU Offshore Renewable Energy Ambitions. All solar PV values are in AC, including for the main and accelerated cases, and all solar PV totals are calculated in AC. The member-state ambition is estimated from NECPs.\n\nIn June 2024, member states submitted their final updated NECPs outlining their voluntary contributions towards the EU goal. However, the European Commission’s assessment of these final NECPs indicates that, in aggregate, they reach only a 40% renewable share – falling 2.5 percentage points short of the binding EU target and 5 percentage points below the REPowerEU ambition. This gap suggests that REPowerEU capacity aims are not aligned with final individual NECPs.\n\nThe second reason for the forecast shortfall in meeting EU-wide ambitions is that persistent challenges to faster solar and wind uptake require stronger policy attention. In the accelerated case, EU capacity reaches 1 255 GW ⁵ by 2030, on track to reach the REPower EU target. Achieving more rapid expansion will require four key policy actions:\n\n **Introduce auctions to provide revenue certainty.** Because auctions for long- term contracts provide revenue certainty, financing costs can drop. Clearer visibility over future auction schedules would help developers plan where to concentrate resources and co-ordinate future investment efforts.\n\n5 For this total, solar PV is calculated in AC instead of DC, and pumped storage is excluded for comparability with the 1 236-GW REPowerEU target. Total capacity in the main case with solar PV calculated in DC is 756 GW in 2024 and 1 123 GW in 2030; for the accelerated case it is 1 260 GW.\n\n0\n\n250\n\n500\n\n750\n\n1 000\n\n1 250\n\n1 500\n\n2024 2030 2030\n\nREPowerEU\n\nGW\n\nAccelerated Other Wind Solar PV\n\n25%\n\n30%\n\n35%\n\n40%\n\n45%\n\n50%\n\nRED II RED III\n\nBinding target Aspirtational target\n\nREPowerEU target Assessment of final updated NECPs","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Renewables 2025 Chapter 1. Renewable electricity Analysis and forecasts to 2030\n\nPAGE | 32\n\nI EA. CC BY 4.0.\n\n**EU installed renewable capacity in 2024 and 2030 vs REPowerEU 2030 targets (left) and shares of renewables in final energy consumption in 2030 (right)**\n\nIEA. CC BY 4.0.\n\nNote: RED = Renewable Energy Directive. NECP = National Energy and Climate Plan. Sources: REPowerEU ambitions for total renewable capacity are from REPowerEU Plan SWD (2022) 230 final; wind and solar PV are from Implementing the REpowerEU Action Plan. The solar PV ambition is in AC, as it is similar to the “almost 600 MW by 2030” target identified in the EU Solar Energy Strategy SWD (2022) 148 final. The 2030 EU offshore wind ambition is from Delivering on the EU Offshore Renewable Energy Ambitions, and the onshore wind aim is the difference between the 520-GW REPowerEU ambition and the 111-GW offshore wind target in Delivering on the EU Offshore Renewable Energy Ambitions. All solar PV values are in AC, including for the main and accelerated cases, and all solar PV totals are calculated in AC. The member-state ambition is estimated from NECPs.\n\nIn June 2024, member states submitted their final updated NECPs outlining their voluntary contributions towards the EU goal. However, the European Commission’s assessment of these final NECPs indicates that, in aggregate, they reach only a 40% renewable share – falling 2.5 percentage points short of the binding EU target and 5 percentage points below the REPowerEU ambition. This gap suggests that REPowerEU capacity aims are not aligned with final individual NECPs.\n\nThe second reason for the forecast shortfall in meeting EU-wide ambitions is that persistent challenges to faster solar and wind uptake require stronger policy attention. In the accelerated case, EU capacity reaches 1 255 GW ⁵ by 2030, on track to reach the REPower EU target. Achieving more rapid expansion will require four key policy actions:\n\n **Introduce auctions to provide revenue certainty.** Because auctions for long- term contracts provide revenue certainty, financing costs can drop. Clearer visibility over future auction schedules would help developers plan where to concentrate resources and co-ordinate future investment efforts.\n\n5 For this total, solar PV is calculated in AC instead of DC, and pumped storage is excluded for comparability with the 1 236-GW REPowerEU target. Total capacity in the main case with solar PV calculated in DC is 756 GW in 2024 and 1 123 GW in 2030; for the accelerated case it is 1 260 GW.\n\n0\n\n250\n\n500\n\n750\n\n1 000\n\n1 250\n\n1 500\n\n2024 2030 2030\n\nREPowerEU\n\nGW\n\nAccelerated Other Wind Solar PV\n\n25%\n\n30%\n\n35%\n\n40%\n\n45%\n\n50%\n\nRED II RED III\n\nBinding target Aspirtational target\n\nREPowerEU target Assessment of final updated NECPs","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Renewables2025_1_p32.pdf"]},"response":{"errors":[],"extract_id":"extract_be9c9e3daa25e04e776a243c7fec821d","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Renewables 2025 Chapter 1. Renewable electricity Analysis and forecasts to 2030\n\nPAGE | 32\n\nI EA. CC BY 4.0.\n\n**EU installed renewable capacity in 2024 and 2030 vs REPowerEU 2030 targets (left) and shares of renewables in final energy consumption in 2030 (right)**\n\nIEA. CC BY 4.0.\n\nNote: RED = Renewable Energy Directive. NECP = National Energy and Climate Plan. Sources: REPowerEU ambitions for total renewable capacity are from REPowerEU Plan SWD (2022) 230 final; wind and solar PV are from Implementing the REpowerEU Action Plan. The solar PV ambition is in AC, as it is similar to the “almost 600 MW by 2030” target identified in the EU Solar Energy Strategy SWD (2022) 148 final. The 2030 EU offshore wind ambition is from Delivering on the EU Offshore Renewable Energy Ambitions, and the onshore wind aim is the difference between the 520-GW REPowerEU ambition and the 111-GW offshore wind target in Delivering on the EU Offshore Renewable Energy Ambitions. All solar PV values are in AC, including for the main and accelerated cases, and all solar PV totals are calculated in AC. The member-state ambition is estimated from NECPs.\n\nIn June 2024, member states submitted their final updated NECPs outlining their voluntary contributions towards the EU goal. However, the European Commission’s assessment of these final NECPs indicates that, in aggregate, they reach only a 40% renewable share – falling 2.5 percentage points short of the binding EU target and 5 percentage points below the REPowerEU ambition. This gap suggests that REPowerEU capacity aims are not aligned with final individual NECPs.\n\nThe second reason for the forecast shortfall in meeting EU-wide ambitions is that persistent challenges to faster solar and wind uptake require stronger policy attention. In the accelerated case, EU capacity reaches 1 255 GW ⁵ by 2030, on track to reach the REPower EU target. Achieving more rapid expansion will require four key policy actions:\n\n **Introduce auctions to provide revenue certainty.** Because auctions for long- term contracts provide revenue certainty, financing costs can drop. Clearer visibility over future auction schedules would help developers plan where to concentrate resources and co-ordinate future investment efforts.\n\n5 For this total, solar PV is calculated in AC instead of DC, and pumped storage is excluded for comparability with the 1 236-GW REPowerEU target. Total capacity in the main case with solar PV calculated in DC is 756 GW in 2024 and 1 123 GW in 2030; for the accelerated case it is 1 260 GW.\n\n0\n\n250\n\n500\n\n750\n\n1 000\n\n1 250\n\n1 500\n\n2024 2030 2030\n\nREPowerEU\n\nGW\n\nAccelerated Other Wind Solar PV\n\n25%\n\n30%\n\n35%\n\n40%\n\n45%\n\n50%\n\nRED II RED III\n\nBinding target Aspirtational target\n\nREPowerEU target Assessment of final updated NECPs"],"full_content":"Renewables 2025 Chapter 1. Renewable electricity Analysis and forecasts to 2030\n\nPAGE | 32\n\nI EA. CC BY 4.0.\n\n**EU installed renewable capacity in 2024 and 2030 vs REPowerEU 2030 targets (left) and shares of renewables in final energy consumption in 2030 (right)**\n\nIEA. CC BY 4.0.\n\nNote: RED = Renewable Energy Directive. NECP = National Energy and Climate Plan. Sources: REPowerEU ambitions for total renewable capacity are from REPowerEU Plan SWD (2022) 230 final; wind and solar PV are from Implementing the REpowerEU Action Plan. The solar PV ambition is in AC, as it is similar to the “almost 600 MW by 2030” target identified in the EU Solar Energy Strategy SWD (2022) 148 final. The 2030 EU offshore wind ambition is from Delivering on the EU Offshore Renewable Energy Ambitions, and the onshore wind aim is the difference between the 520-GW REPowerEU ambition and the 111-GW offshore wind target in Delivering on the EU Offshore Renewable Energy Ambitions. All solar PV values are in AC, including for the main and accelerated cases, and all solar PV totals are calculated in AC. The member-state ambition is estimated from NECPs.\n\nIn June 2024, member states submitted their final updated NECPs outlining their voluntary contributions towards the EU goal. However, the European Commission’s assessment of these final NECPs indicates that, in aggregate, they reach only a 40% renewable share – falling 2.5 percentage points short of the binding EU target and 5 percentage points below the REPowerEU ambition. This gap suggests that REPowerEU capacity aims are not aligned with final individual NECPs.\n\nThe second reason for the forecast shortfall in meeting EU-wide ambitions is that persistent challenges to faster solar and wind uptake require stronger policy attention. In the accelerated case, EU capacity reaches 1 255 GW ⁵ by 2030, on track to reach the REPower EU target. Achieving more rapid expansion will require four key policy actions:\n\n **Introduce auctions to provide revenue certainty.** Because auctions for long- term contracts provide revenue certainty, financing costs can drop. Clearer visibility over future auction schedules would help developers plan where to concentrate resources and co-ordinate future investment efforts.\n\n5 For this total, solar PV is calculated in AC instead of DC, and pumped storage is excluded for comparability with the 1 236-GW REPowerEU target. Total capacity in the main case with solar PV calculated in DC is 756 GW in 2024 and 1 123 GW in 2030; for the accelerated case it is 1 260 GW.\n\n0\n\n250\n\n500\n\n750\n\n1 000\n\n1 250\n\n1 500\n\n2024 2030 2030\n\nREPowerEU\n\nGW\n\nAccelerated Other Wind Solar PV\n\n25%\n\n30%\n\n35%\n\n40%\n\n45%\n\n50%\n\nRED II RED III\n\nBinding target Aspirtational target\n\nREPowerEU target Assessment of final updated NECPs","publish_date":null,"title":"In June 2024, member states submitted their final updated NECPs outlining their voluntary contributions towards the EU goal. However, the","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Renewables2025_1_p32.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Renewables2025_1_p32.pdf"},"request":{"example_id":"chart/Renewables2025_1_p32","product_type":"parse","source_file_path":"Renewables2025_1_p32.pdf"},"started_at":"2026-11-04T19:45:54.328060"} +{"completed_at":"2026-11-04T19:45:58.273727","latency_in_ms":2361,"output":{"example_id":"chart/Renewables2025_1_p36","grounded_pages":[],"layout_pages":[],"markdown":"Renewables 2025 Chapter 1. Renewable electricity Analysis and forecasts to 2030\n\nPAGE | 36\n\nI EA. CC BY 4.0.\n\nExpanded state-level auctions, the Capacity Investment Scheme and strong corporate demand support growth. These incentives, coupled with continued distributed solar PV expansion, will help Australia exceed its 2030 capacity ambitions.\n\nThe renewable electricity capacity forecast for the **ASEAN** region has been revised almost 15% upwards from last year, primarily owing to improved outlooks for Viet Nam, Thailand and the Philippines. Between 2025 and 2030, ASEAN countries are expected to add over 95 GW of renewable energy capacity – nearly double the deployment of the previous six-year period. More than half of these additions will come from solar PV.\n\nViet Nam is the region’s leader, accounting for over 40% of total growth, followed by Indonesia (20%). Both countries are anticipated to accelerate capacity expansion significantly towards 2030, particularly if remaining barriers are addressed under the accelerated case. Overall, the combination of more ambitious policy targets, strengthened implementation frameworks and increased adoption of grid flexibility solutions could enable renewable deployment in ASEAN to reach 70% above the main case – the highest upside potential of all major regions.\n\n**Net capacity additions in ASEAN, main and accelerated cases, 2020-2030**\n\nIEA. CC BY 4.0.\n\nNote: ASEAN = Association of Southeast Asian Nations.\n\n**Viet Nam** is expected to add 40 GW of renewable energy capacity over 2025- 2030, led by onshore wind and utility-scale PV. In April 2025, it updated its national Power Development Plan (PDP8) to significantly raise its 2030 ambitions, particularly for solar PV, prompting a 20% upwards revision to our forecast. The government also enabled direct PPAs between large consumers and generators, providing an additional boost to utility-scale deployment.\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n2020 2022 2024 2026 2028 2030\n\nGW\n\nMain case\n\nViet Nam Indonesia Philippines Thailand Rest of ASEAN\n\n2020 2022 2024 2026 2028 2030\n\nAccelerated case","pages":[{"line_numbers":[],"markdown":"Renewables 2025 Chapter 1. Renewable electricity Analysis and forecasts to 2030\n\nPAGE | 36\n\nI EA. CC BY 4.0.\n\nExpanded state-level auctions, the Capacity Investment Scheme and strong corporate demand support growth. These incentives, coupled with continued distributed solar PV expansion, will help Australia exceed its 2030 capacity ambitions.\n\nThe renewable electricity capacity forecast for the **ASEAN** region has been revised almost 15% upwards from last year, primarily owing to improved outlooks for Viet Nam, Thailand and the Philippines. Between 2025 and 2030, ASEAN countries are expected to add over 95 GW of renewable energy capacity – nearly double the deployment of the previous six-year period. More than half of these additions will come from solar PV.\n\nViet Nam is the region’s leader, accounting for over 40% of total growth, followed by Indonesia (20%). Both countries are anticipated to accelerate capacity expansion significantly towards 2030, particularly if remaining barriers are addressed under the accelerated case. Overall, the combination of more ambitious policy targets, strengthened implementation frameworks and increased adoption of grid flexibility solutions could enable renewable deployment in ASEAN to reach 70% above the main case – the highest upside potential of all major regions.\n\n**Net capacity additions in ASEAN, main and accelerated cases, 2020-2030**\n\nIEA. CC BY 4.0.\n\nNote: ASEAN = Association of Southeast Asian Nations.\n\n**Viet Nam** is expected to add 40 GW of renewable energy capacity over 2025- 2030, led by onshore wind and utility-scale PV. In April 2025, it updated its national Power Development Plan (PDP8) to significantly raise its 2030 ambitions, particularly for solar PV, prompting a 20% upwards revision to our forecast. The government also enabled direct PPAs between large consumers and generators, providing an additional boost to utility-scale deployment.\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n2020 2022 2024 2026 2028 2030\n\nGW\n\nMain case\n\nViet Nam Indonesia Philippines Thailand Rest of ASEAN\n\n2020 2022 2024 2026 2028 2030\n\nAccelerated case","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Renewables 2025 Chapter 1. Renewable electricity Analysis and forecasts to 2030\n\nPAGE | 36\n\nI EA. CC BY 4.0.\n\nExpanded state-level auctions, the Capacity Investment Scheme and strong corporate demand support growth. These incentives, coupled with continued distributed solar PV expansion, will help Australia exceed its 2030 capacity ambitions.\n\nThe renewable electricity capacity forecast for the **ASEAN** region has been revised almost 15% upwards from last year, primarily owing to improved outlooks for Viet Nam, Thailand and the Philippines. Between 2025 and 2030, ASEAN countries are expected to add over 95 GW of renewable energy capacity – nearly double the deployment of the previous six-year period. More than half of these additions will come from solar PV.\n\nViet Nam is the region’s leader, accounting for over 40% of total growth, followed by Indonesia (20%). Both countries are anticipated to accelerate capacity expansion significantly towards 2030, particularly if remaining barriers are addressed under the accelerated case. Overall, the combination of more ambitious policy targets, strengthened implementation frameworks and increased adoption of grid flexibility solutions could enable renewable deployment in ASEAN to reach 70% above the main case – the highest upside potential of all major regions.\n\n**Net capacity additions in ASEAN, main and accelerated cases, 2020-2030**\n\nIEA. CC BY 4.0.\n\nNote: ASEAN = Association of Southeast Asian Nations.\n\n**Viet Nam** is expected to add 40 GW of renewable energy capacity over 2025- 2030, led by onshore wind and utility-scale PV. In April 2025, it updated its national Power Development Plan (PDP8) to significantly raise its 2030 ambitions, particularly for solar PV, prompting a 20% upwards revision to our forecast. The government also enabled direct PPAs between large consumers and generators, providing an additional boost to utility-scale deployment.\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n2020 2022 2024 2026 2028 2030\n\nGW\n\nMain case\n\nViet Nam Indonesia Philippines Thailand Rest of ASEAN\n\n2020 2022 2024 2026 2028 2030\n\nAccelerated case","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Renewables2025_1_p36.pdf"]},"response":{"errors":[],"extract_id":"extract_7e230360d6a1c04557fd29051e285217","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Renewables 2025 Chapter 1. Renewable electricity Analysis and forecasts to 2030\n\nPAGE | 36\n\nI EA. CC BY 4.0.\n\nExpanded state-level auctions, the Capacity Investment Scheme and strong corporate demand support growth. These incentives, coupled with continued distributed solar PV expansion, will help Australia exceed its 2030 capacity ambitions.\n\nThe renewable electricity capacity forecast for the **ASEAN** region has been revised almost 15% upwards from last year, primarily owing to improved outlooks for Viet Nam, Thailand and the Philippines. Between 2025 and 2030, ASEAN countries are expected to add over 95 GW of renewable energy capacity – nearly double the deployment of the previous six-year period. More than half of these additions will come from solar PV.\n\nViet Nam is the region’s leader, accounting for over 40% of total growth, followed by Indonesia (20%). Both countries are anticipated to accelerate capacity expansion significantly towards 2030, particularly if remaining barriers are addressed under the accelerated case. Overall, the combination of more ambitious policy targets, strengthened implementation frameworks and increased adoption of grid flexibility solutions could enable renewable deployment in ASEAN to reach 70% above the main case – the highest upside potential of all major regions.\n\n**Net capacity additions in ASEAN, main and accelerated cases, 2020-2030**\n\nIEA. CC BY 4.0.\n\nNote: ASEAN = Association of Southeast Asian Nations.\n\n**Viet Nam** is expected to add 40 GW of renewable energy capacity over 2025- 2030, led by onshore wind and utility-scale PV. In April 2025, it updated its national Power Development Plan (PDP8) to significantly raise its 2030 ambitions, particularly for solar PV, prompting a 20% upwards revision to our forecast. The government also enabled direct PPAs between large consumers and generators, providing an additional boost to utility-scale deployment.\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n2020 2022 2024 2026 2028 2030\n\nGW\n\nMain case\n\nViet Nam Indonesia Philippines Thailand Rest of ASEAN\n\n2020 2022 2024 2026 2028 2030\n\nAccelerated case"],"full_content":"Renewables 2025 Chapter 1. Renewable electricity Analysis and forecasts to 2030\n\nPAGE | 36\n\nI EA. CC BY 4.0.\n\nExpanded state-level auctions, the Capacity Investment Scheme and strong corporate demand support growth. These incentives, coupled with continued distributed solar PV expansion, will help Australia exceed its 2030 capacity ambitions.\n\nThe renewable electricity capacity forecast for the **ASEAN** region has been revised almost 15% upwards from last year, primarily owing to improved outlooks for Viet Nam, Thailand and the Philippines. Between 2025 and 2030, ASEAN countries are expected to add over 95 GW of renewable energy capacity – nearly double the deployment of the previous six-year period. More than half of these additions will come from solar PV.\n\nViet Nam is the region’s leader, accounting for over 40% of total growth, followed by Indonesia (20%). Both countries are anticipated to accelerate capacity expansion significantly towards 2030, particularly if remaining barriers are addressed under the accelerated case. Overall, the combination of more ambitious policy targets, strengthened implementation frameworks and increased adoption of grid flexibility solutions could enable renewable deployment in ASEAN to reach 70% above the main case – the highest upside potential of all major regions.\n\n**Net capacity additions in ASEAN, main and accelerated cases, 2020-2030**\n\nIEA. CC BY 4.0.\n\nNote: ASEAN = Association of Southeast Asian Nations.\n\n**Viet Nam** is expected to add 40 GW of renewable energy capacity over 2025- 2030, led by onshore wind and utility-scale PV. In April 2025, it updated its national Power Development Plan (PDP8) to significantly raise its 2030 ambitions, particularly for solar PV, prompting a 20% upwards revision to our forecast. The government also enabled direct PPAs between large consumers and generators, providing an additional boost to utility-scale deployment.\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n2020 2022 2024 2026 2028 2030\n\nGW\n\nMain case\n\nViet Nam Indonesia Philippines Thailand Rest of ASEAN\n\n2020 2022 2024 2026 2028 2030\n\nAccelerated case","publish_date":null,"title":"Expanded state-level auctions, the Capacity Investment Scheme and strong corporate demand support growth. These incentives, coupled with continued distributed solar","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Renewables2025_1_p36.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Renewables2025_1_p36.pdf"},"request":{"example_id":"chart/Renewables2025_1_p36","product_type":"parse","source_file_path":"Renewables2025_1_p36.pdf"},"started_at":"2026-11-04T19:45:55.911734"} +{"completed_at":"2026-11-04T19:45:58.877986","latency_in_ms":958,"output":{"example_id":"chart/Renewables2025_1_p42","grounded_pages":[],"layout_pages":[],"markdown":"Renewables 2025 Chapter 1. Renewable electricity Analysis and forecasts to 2030\n\nPAGE | 42\n\nI EA. CC BY 4.0.\n\npressures due to anti-dumping and countervailing duties (AD/CVD) on modules imported from countries such as Cambodia, Malaysia and Viet Nam, along with increased tariffs on Chinese solar cells.\n\nThese tariffs could drive up costs just when federal subsidy support is ending. Plus, tariffs on non-energy-specific technologies and components may create additional financial pressure as the administration applies them to enabling infrastructure and construction materials, such as steel for transmission and distribution wires.\n\nDespite these headwinds, the accelerated-growth scenario anticipates nearly 24% higher capacity additions than in the main forecast. This optimistic outlook assumes the effective mitigation of current challenges through expanded corporate renewable energy procurement and stronger state-level renewable portfolio standards.\n\n## **Latin America and the Caribbean**\n\n### Solar PV is set to tie hydropower for largest installed capacity in 2030\n\nThe Latin America and Caribbean region is expected to add almost 160 GW of renewable energy capacity by 2030, with solar PV making up most of this growth. In fact, solar PV is projected to match hydropower in total installed capacity by 2030.\n\n**Total installed renewable capacity by technology in Latin America and the Caribbean, 2010-2030**\n\nIEA. CC BY 4.0.\n\nWhile both utility-scale and distributed solar PV are expanding rapidly and their capacities were roughly equal between 2023 and 2025, utility-scale installations\n\n0\n\n50\n\n100\n\n150\n\n200\n\n250\n\n2010 2012 2014 2016 2018 2020 2022 2024 2026 2028 2030\n\nInstalled capacity (GW)\n\nHydropower\n\nWind\n\nSolar PV\n\nUtility-scale PV\n\nDistributed PV\n\nOther renewables","pages":[{"line_numbers":[],"markdown":"Renewables 2025 Chapter 1. Renewable electricity Analysis and forecasts to 2030\n\nPAGE | 42\n\nI EA. CC BY 4.0.\n\npressures due to anti-dumping and countervailing duties (AD/CVD) on modules imported from countries such as Cambodia, Malaysia and Viet Nam, along with increased tariffs on Chinese solar cells.\n\nThese tariffs could drive up costs just when federal subsidy support is ending. Plus, tariffs on non-energy-specific technologies and components may create additional financial pressure as the administration applies them to enabling infrastructure and construction materials, such as steel for transmission and distribution wires.\n\nDespite these headwinds, the accelerated-growth scenario anticipates nearly 24% higher capacity additions than in the main forecast. This optimistic outlook assumes the effective mitigation of current challenges through expanded corporate renewable energy procurement and stronger state-level renewable portfolio standards.\n\n## **Latin America and the Caribbean**\n\n### Solar PV is set to tie hydropower for largest installed capacity in 2030\n\nThe Latin America and Caribbean region is expected to add almost 160 GW of renewable energy capacity by 2030, with solar PV making up most of this growth. In fact, solar PV is projected to match hydropower in total installed capacity by 2030.\n\n**Total installed renewable capacity by technology in Latin America and the Caribbean, 2010-2030**\n\nIEA. CC BY 4.0.\n\nWhile both utility-scale and distributed solar PV are expanding rapidly and their capacities were roughly equal between 2023 and 2025, utility-scale installations\n\n0\n\n50\n\n100\n\n150\n\n200\n\n250\n\n2010 2012 2014 2016 2018 2020 2022 2024 2026 2028 2030\n\nInstalled capacity (GW)\n\nHydropower\n\nWind\n\nSolar PV\n\nUtility-scale PV\n\nDistributed PV\n\nOther renewables","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Renewables 2025 Chapter 1. Renewable electricity Analysis and forecasts to 2030\n\nPAGE | 42\n\nI EA. CC BY 4.0.\n\npressures due to anti-dumping and countervailing duties (AD/CVD) on modules imported from countries such as Cambodia, Malaysia and Viet Nam, along with increased tariffs on Chinese solar cells.\n\nThese tariffs could drive up costs just when federal subsidy support is ending. Plus, tariffs on non-energy-specific technologies and components may create additional financial pressure as the administration applies them to enabling infrastructure and construction materials, such as steel for transmission and distribution wires.\n\nDespite these headwinds, the accelerated-growth scenario anticipates nearly 24% higher capacity additions than in the main forecast. This optimistic outlook assumes the effective mitigation of current challenges through expanded corporate renewable energy procurement and stronger state-level renewable portfolio standards.\n\n## **Latin America and the Caribbean**\n\n### Solar PV is set to tie hydropower for largest installed capacity in 2030\n\nThe Latin America and Caribbean region is expected to add almost 160 GW of renewable energy capacity by 2030, with solar PV making up most of this growth. In fact, solar PV is projected to match hydropower in total installed capacity by 2030.\n\n**Total installed renewable capacity by technology in Latin America and the Caribbean, 2010-2030**\n\nIEA. CC BY 4.0.\n\nWhile both utility-scale and distributed solar PV are expanding rapidly and their capacities were roughly equal between 2023 and 2025, utility-scale installations\n\n0\n\n50\n\n100\n\n150\n\n200\n\n250\n\n2010 2012 2014 2016 2018 2020 2022 2024 2026 2028 2030\n\nInstalled capacity (GW)\n\nHydropower\n\nWind\n\nSolar PV\n\nUtility-scale PV\n\nDistributed PV\n\nOther renewables","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Renewables2025_1_p42.pdf"]},"response":{"errors":[],"extract_id":"extract_b14733335012b120efa7d129f580da26","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Renewables 2025 Chapter 1. Renewable electricity Analysis and forecasts to 2030\n\nPAGE | 42\n\nI EA. CC BY 4.0.\n\npressures due to anti-dumping and countervailing duties (AD/CVD) on modules imported from countries such as Cambodia, Malaysia and Viet Nam, along with increased tariffs on Chinese solar cells.\n\nThese tariffs could drive up costs just when federal subsidy support is ending. Plus, tariffs on non-energy-specific technologies and components may create additional financial pressure as the administration applies them to enabling infrastructure and construction materials, such as steel for transmission and distribution wires.\n\nDespite these headwinds, the accelerated-growth scenario anticipates nearly 24% higher capacity additions than in the main forecast. This optimistic outlook assumes the effective mitigation of current challenges through expanded corporate renewable energy procurement and stronger state-level renewable portfolio standards.\n\n## **Latin America and the Caribbean**\n\n### Solar PV is set to tie hydropower for largest installed capacity in 2030\n\nThe Latin America and Caribbean region is expected to add almost 160 GW of renewable energy capacity by 2030, with solar PV making up most of this growth. In fact, solar PV is projected to match hydropower in total installed capacity by 2030.\n\n**Total installed renewable capacity by technology in Latin America and the Caribbean, 2010-2030**\n\nIEA. CC BY 4.0.\n\nWhile both utility-scale and distributed solar PV are expanding rapidly and their capacities were roughly equal between 2023 and 2025, utility-scale installations\n\n0\n\n50\n\n100\n\n150\n\n200\n\n250\n\n2010 2012 2014 2016 2018 2020 2022 2024 2026 2028 2030\n\nInstalled capacity (GW)\n\nHydropower\n\nWind\n\nSolar PV\n\nUtility-scale PV\n\nDistributed PV\n\nOther renewables"],"full_content":"Renewables 2025 Chapter 1. Renewable electricity Analysis and forecasts to 2030\n\nPAGE | 42\n\nI EA. CC BY 4.0.\n\npressures due to anti-dumping and countervailing duties (AD/CVD) on modules imported from countries such as Cambodia, Malaysia and Viet Nam, along with increased tariffs on Chinese solar cells.\n\nThese tariffs could drive up costs just when federal subsidy support is ending. Plus, tariffs on non-energy-specific technologies and components may create additional financial pressure as the administration applies them to enabling infrastructure and construction materials, such as steel for transmission and distribution wires.\n\nDespite these headwinds, the accelerated-growth scenario anticipates nearly 24% higher capacity additions than in the main forecast. This optimistic outlook assumes the effective mitigation of current challenges through expanded corporate renewable energy procurement and stronger state-level renewable portfolio standards.\n\n## **Latin America and the Caribbean**\n\n### Solar PV is set to tie hydropower for largest installed capacity in 2030\n\nThe Latin America and Caribbean region is expected to add almost 160 GW of renewable energy capacity by 2030, with solar PV making up most of this growth. In fact, solar PV is projected to match hydropower in total installed capacity by 2030.\n\n**Total installed renewable capacity by technology in Latin America and the Caribbean, 2010-2030**\n\nIEA. CC BY 4.0.\n\nWhile both utility-scale and distributed solar PV are expanding rapidly and their capacities were roughly equal between 2023 and 2025, utility-scale installations\n\n0\n\n50\n\n100\n\n150\n\n200\n\n250\n\n2010 2012 2014 2016 2018 2020 2022 2024 2026 2028 2030\n\nInstalled capacity (GW)\n\nHydropower\n\nWind\n\nSolar PV\n\nUtility-scale PV\n\nDistributed PV\n\nOther renewables","publish_date":null,"title":"Latin America and the Caribbean","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Renewables2025_1_p42.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Renewables2025_1_p42.pdf"},"request":{"example_id":"chart/Renewables2025_1_p42","product_type":"parse","source_file_path":"Renewables2025_1_p42.pdf"},"started_at":"2026-11-04T19:45:57.919453"} +{"completed_at":"2026-11-04T19:31:19.086354","latency_in_ms":1022,"output":{"example_id":"chart/Renewables2025_1_p43","grounded_pages":[],"layout_pages":[],"markdown":"Renewables 2025 Chapter 1. Renewable electricity Analysis and forecasts to 2030\n\nPAGE | 43\n\nI EA. CC BY 4.0.\n\nare forecast to outpace distributed systems in upcoming years. Solar PV capacity overtook wind in 2022, and by 2025 both utility-scale and distributed solar PV each individually match wind capacity.\n\nHydropower has long been the largest renewable electricity source in Latin America, and it continues to be critical in many countries across the region. Although the regional outlook for hydropower remains similar to last year, some markets still hold significant untapped potential. Colombia, for instance, is set to lead in new hydropower capacity, largely owing to completion of the 2.4-GW Ituango project. However, further large-scale hydropower development remains challenging due to environmental, social and financial constraints in multiple countries.\n\nIn recent years, solar PV has gained increasing importance in Latin America's energy transition. Throughout the 2020s, it has taken on a leading role, surpassing wind in capacity additions. It will be the primary driver of renewable energy growth, with utility-scale projects contributing 41% and distributed systems accounting for 33% of new capacity in 2025-2030.\n\n**Net renewable capacity additions by technology in Latin America and the Caribbean, 2010-2030**\n\nIEA. CC BY 4.0.\n\nThree countries are forecast to provide more than 80% of renewable capacity expansion in Latin America over the forecast period. Brazil leads regional expansion, installing about half of the new capacity, and Mexico and Chile each contribute roughly 15% to the region’s overall growth.\n\nOverall, the regional forecast has been revised down slightly (by 3%). While this revision is modest, it masks more significant shifts at the country and technology\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030\n\nNet additions (GW)\n\nHydropower Bioenergy Wind Solar PV Other renewables","pages":[{"line_numbers":[],"markdown":"Renewables 2025 Chapter 1. Renewable electricity Analysis and forecasts to 2030\n\nPAGE | 43\n\nI EA. CC BY 4.0.\n\nare forecast to outpace distributed systems in upcoming years. Solar PV capacity overtook wind in 2022, and by 2025 both utility-scale and distributed solar PV each individually match wind capacity.\n\nHydropower has long been the largest renewable electricity source in Latin America, and it continues to be critical in many countries across the region. Although the regional outlook for hydropower remains similar to last year, some markets still hold significant untapped potential. Colombia, for instance, is set to lead in new hydropower capacity, largely owing to completion of the 2.4-GW Ituango project. However, further large-scale hydropower development remains challenging due to environmental, social and financial constraints in multiple countries.\n\nIn recent years, solar PV has gained increasing importance in Latin America's energy transition. Throughout the 2020s, it has taken on a leading role, surpassing wind in capacity additions. It will be the primary driver of renewable energy growth, with utility-scale projects contributing 41% and distributed systems accounting for 33% of new capacity in 2025-2030.\n\n**Net renewable capacity additions by technology in Latin America and the Caribbean, 2010-2030**\n\nIEA. CC BY 4.0.\n\nThree countries are forecast to provide more than 80% of renewable capacity expansion in Latin America over the forecast period. Brazil leads regional expansion, installing about half of the new capacity, and Mexico and Chile each contribute roughly 15% to the region’s overall growth.\n\nOverall, the regional forecast has been revised down slightly (by 3%). While this revision is modest, it masks more significant shifts at the country and technology\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030\n\nNet additions (GW)\n\nHydropower Bioenergy Wind Solar PV Other renewables","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Renewables 2025 Chapter 1. Renewable electricity Analysis and forecasts to 2030\n\nPAGE | 43\n\nI EA. CC BY 4.0.\n\nare forecast to outpace distributed systems in upcoming years. Solar PV capacity overtook wind in 2022, and by 2025 both utility-scale and distributed solar PV each individually match wind capacity.\n\nHydropower has long been the largest renewable electricity source in Latin America, and it continues to be critical in many countries across the region. Although the regional outlook for hydropower remains similar to last year, some markets still hold significant untapped potential. Colombia, for instance, is set to lead in new hydropower capacity, largely owing to completion of the 2.4-GW Ituango project. However, further large-scale hydropower development remains challenging due to environmental, social and financial constraints in multiple countries.\n\nIn recent years, solar PV has gained increasing importance in Latin America's energy transition. Throughout the 2020s, it has taken on a leading role, surpassing wind in capacity additions. It will be the primary driver of renewable energy growth, with utility-scale projects contributing 41% and distributed systems accounting for 33% of new capacity in 2025-2030.\n\n**Net renewable capacity additions by technology in Latin America and the Caribbean, 2010-2030**\n\nIEA. CC BY 4.0.\n\nThree countries are forecast to provide more than 80% of renewable capacity expansion in Latin America over the forecast period. Brazil leads regional expansion, installing about half of the new capacity, and Mexico and Chile each contribute roughly 15% to the region’s overall growth.\n\nOverall, the regional forecast has been revised down slightly (by 3%). While this revision is modest, it masks more significant shifts at the country and technology\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030\n\nNet additions (GW)\n\nHydropower Bioenergy Wind Solar PV Other renewables","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Renewables2025_1_p43.pdf"]},"response":{"errors":[],"extract_id":"extract_a14366a356fe4eeafb5987fc58e95800","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Renewables 2025 Chapter 1. Renewable electricity Analysis and forecasts to 2030\n\nPAGE | 43\n\nI EA. CC BY 4.0.\n\nare forecast to outpace distributed systems in upcoming years. Solar PV capacity overtook wind in 2022, and by 2025 both utility-scale and distributed solar PV each individually match wind capacity.\n\nHydropower has long been the largest renewable electricity source in Latin America, and it continues to be critical in many countries across the region. Although the regional outlook for hydropower remains similar to last year, some markets still hold significant untapped potential. Colombia, for instance, is set to lead in new hydropower capacity, largely owing to completion of the 2.4-GW Ituango project. However, further large-scale hydropower development remains challenging due to environmental, social and financial constraints in multiple countries.\n\nIn recent years, solar PV has gained increasing importance in Latin America's energy transition. Throughout the 2020s, it has taken on a leading role, surpassing wind in capacity additions. It will be the primary driver of renewable energy growth, with utility-scale projects contributing 41% and distributed systems accounting for 33% of new capacity in 2025-2030.\n\n**Net renewable capacity additions by technology in Latin America and the Caribbean, 2010-2030**\n\nIEA. CC BY 4.0.\n\nThree countries are forecast to provide more than 80% of renewable capacity expansion in Latin America over the forecast period. Brazil leads regional expansion, installing about half of the new capacity, and Mexico and Chile each contribute roughly 15% to the region’s overall growth.\n\nOverall, the regional forecast has been revised down slightly (by 3%). While this revision is modest, it masks more significant shifts at the country and technology\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030\n\nNet additions (GW)\n\nHydropower Bioenergy Wind Solar PV Other renewables"],"full_content":"Renewables 2025 Chapter 1. Renewable electricity Analysis and forecasts to 2030\n\nPAGE | 43\n\nI EA. CC BY 4.0.\n\nare forecast to outpace distributed systems in upcoming years. Solar PV capacity overtook wind in 2022, and by 2025 both utility-scale and distributed solar PV each individually match wind capacity.\n\nHydropower has long been the largest renewable electricity source in Latin America, and it continues to be critical in many countries across the region. Although the regional outlook for hydropower remains similar to last year, some markets still hold significant untapped potential. Colombia, for instance, is set to lead in new hydropower capacity, largely owing to completion of the 2.4-GW Ituango project. However, further large-scale hydropower development remains challenging due to environmental, social and financial constraints in multiple countries.\n\nIn recent years, solar PV has gained increasing importance in Latin America's energy transition. Throughout the 2020s, it has taken on a leading role, surpassing wind in capacity additions. It will be the primary driver of renewable energy growth, with utility-scale projects contributing 41% and distributed systems accounting for 33% of new capacity in 2025-2030.\n\n**Net renewable capacity additions by technology in Latin America and the Caribbean, 2010-2030**\n\nIEA. CC BY 4.0.\n\nThree countries are forecast to provide more than 80% of renewable capacity expansion in Latin America over the forecast period. Brazil leads regional expansion, installing about half of the new capacity, and Mexico and Chile each contribute roughly 15% to the region’s overall growth.\n\nOverall, the regional forecast has been revised down slightly (by 3%). While this revision is modest, it masks more significant shifts at the country and technology\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030\n\nNet additions (GW)\n\nHydropower Bioenergy Wind Solar PV Other renewables","publish_date":null,"title":"are forecast to outpace distributed systems in upcoming years. Solar PV capacity overtook wind in 2022, and by 2025 both","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Renewables2025_1_p43.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Renewables2025_1_p43.pdf"},"request":{"example_id":"chart/Renewables2025_1_p43","product_type":"parse","source_file_path":"Renewables2025_1_p43.pdf"},"started_at":"2026-11-04T19:31:18.064141"} +{"completed_at":"2026-11-04T19:46:01.298735","latency_in_ms":1437,"output":{"example_id":"chart/Renewables2025_1_p44","grounded_pages":[],"layout_pages":[],"markdown":"Renewables 2025 Chapter 1. Renewable electricity Analysis and forecasts to 2030\n\nPAGE | 44\n\nI EA. CC BY 4.0.\n\nlevels. Projections have been reduced 16% for utility-scale solar PV and 9% for wind, whereas distributed solar PV has been revised up 30%, offsetting the decline.\n\nForecasts for utility-scale solar PV and wind reveal diverging trends across key Latin American markets. For Brazil, projections have been revised downwards because curtailment (particularly in the Northeast) and transmission bottlenecks are reducing project profitability, lengthening connection queues and extending project lead times. Curtailment also remains a key challenge in Chile, but the government is addressing it through grid expansion auctions and strong battery deployment, while a steady flow of permit requests indicates sustained interest in renewable power projects. Conversely, Mexico’s energy forecast has expanded since the country launched a USD 22-billion energy plan in early 2025, targeting 29 GW of new capacity – 6.4 GW from variable renewables – along with grid upgrades, storage expansion and revived projects.\n\nFor distributed PV, 30% more growth in the region is estimated in this year’s forecast than in last year’s. In Brazil, despite a 2023 policy change reducing remuneration under the net-metering scheme, distributed solar PV additions remain robust. Meanwhile, Mexico experienced record growth in distributed solar PV in 2024, adding 1 GW and reaching half a million users. This surge is largely the result of its attractive net metering programme, which is expected to enable an additional 13 GW of capacity by 2030, particularly benefiting commercial customers after the system-size threshold was increased from 0.5 MW to 0.7 MW.\n\n**Net renewable capacity additions by country and technology in Latin America, 2025- 2030**\n\nIEA. CC BY 4.0.\n\nSlow grid expansion continues to pose a significant challenge across Latin America. In Brazil, the increasing deployment of variable renewable energy has\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\n16\n\nNet additions (GW)\n\n**Utility-scale PV**\n\nBrazil Mexico Chile Argentina Colombia Rest of the region Renewables 2024 forecast\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n**Distributed PV**\n\n0\n\n1\n\n2\n\n3\n\n4\n\n5\n\n6\n\n**Wind**","pages":[{"line_numbers":[],"markdown":"Renewables 2025 Chapter 1. Renewable electricity Analysis and forecasts to 2030\n\nPAGE | 44\n\nI EA. CC BY 4.0.\n\nlevels. Projections have been reduced 16% for utility-scale solar PV and 9% for wind, whereas distributed solar PV has been revised up 30%, offsetting the decline.\n\nForecasts for utility-scale solar PV and wind reveal diverging trends across key Latin American markets. For Brazil, projections have been revised downwards because curtailment (particularly in the Northeast) and transmission bottlenecks are reducing project profitability, lengthening connection queues and extending project lead times. Curtailment also remains a key challenge in Chile, but the government is addressing it through grid expansion auctions and strong battery deployment, while a steady flow of permit requests indicates sustained interest in renewable power projects. Conversely, Mexico’s energy forecast has expanded since the country launched a USD 22-billion energy plan in early 2025, targeting 29 GW of new capacity – 6.4 GW from variable renewables – along with grid upgrades, storage expansion and revived projects.\n\nFor distributed PV, 30% more growth in the region is estimated in this year’s forecast than in last year’s. In Brazil, despite a 2023 policy change reducing remuneration under the net-metering scheme, distributed solar PV additions remain robust. Meanwhile, Mexico experienced record growth in distributed solar PV in 2024, adding 1 GW and reaching half a million users. This surge is largely the result of its attractive net metering programme, which is expected to enable an additional 13 GW of capacity by 2030, particularly benefiting commercial customers after the system-size threshold was increased from 0.5 MW to 0.7 MW.\n\n**Net renewable capacity additions by country and technology in Latin America, 2025- 2030**\n\nIEA. CC BY 4.0.\n\nSlow grid expansion continues to pose a significant challenge across Latin America. In Brazil, the increasing deployment of variable renewable energy has\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\n16\n\nNet additions (GW)\n\n**Utility-scale PV**\n\nBrazil Mexico Chile Argentina Colombia Rest of the region Renewables 2024 forecast\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n**Distributed PV**\n\n0\n\n1\n\n2\n\n3\n\n4\n\n5\n\n6\n\n**Wind**","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Renewables 2025 Chapter 1. Renewable electricity Analysis and forecasts to 2030\n\nPAGE | 44\n\nI EA. CC BY 4.0.\n\nlevels. Projections have been reduced 16% for utility-scale solar PV and 9% for wind, whereas distributed solar PV has been revised up 30%, offsetting the decline.\n\nForecasts for utility-scale solar PV and wind reveal diverging trends across key Latin American markets. For Brazil, projections have been revised downwards because curtailment (particularly in the Northeast) and transmission bottlenecks are reducing project profitability, lengthening connection queues and extending project lead times. Curtailment also remains a key challenge in Chile, but the government is addressing it through grid expansion auctions and strong battery deployment, while a steady flow of permit requests indicates sustained interest in renewable power projects. Conversely, Mexico’s energy forecast has expanded since the country launched a USD 22-billion energy plan in early 2025, targeting 29 GW of new capacity – 6.4 GW from variable renewables – along with grid upgrades, storage expansion and revived projects.\n\nFor distributed PV, 30% more growth in the region is estimated in this year’s forecast than in last year’s. In Brazil, despite a 2023 policy change reducing remuneration under the net-metering scheme, distributed solar PV additions remain robust. Meanwhile, Mexico experienced record growth in distributed solar PV in 2024, adding 1 GW and reaching half a million users. This surge is largely the result of its attractive net metering programme, which is expected to enable an additional 13 GW of capacity by 2030, particularly benefiting commercial customers after the system-size threshold was increased from 0.5 MW to 0.7 MW.\n\n**Net renewable capacity additions by country and technology in Latin America, 2025- 2030**\n\nIEA. CC BY 4.0.\n\nSlow grid expansion continues to pose a significant challenge across Latin America. In Brazil, the increasing deployment of variable renewable energy has\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\n16\n\nNet additions (GW)\n\n**Utility-scale PV**\n\nBrazil Mexico Chile Argentina Colombia Rest of the region Renewables 2024 forecast\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n**Distributed PV**\n\n0\n\n1\n\n2\n\n3\n\n4\n\n5\n\n6\n\n**Wind**","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Renewables2025_1_p44.pdf"]},"response":{"errors":[],"extract_id":"extract_e0f6d8e25ca7f4ab449dc2de68f82b3c","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Renewables 2025 Chapter 1. Renewable electricity Analysis and forecasts to 2030\n\nPAGE | 44\n\nI EA. CC BY 4.0.\n\nlevels. Projections have been reduced 16% for utility-scale solar PV and 9% for wind, whereas distributed solar PV has been revised up 30%, offsetting the decline.\n\nForecasts for utility-scale solar PV and wind reveal diverging trends across key Latin American markets. For Brazil, projections have been revised downwards because curtailment (particularly in the Northeast) and transmission bottlenecks are reducing project profitability, lengthening connection queues and extending project lead times. Curtailment also remains a key challenge in Chile, but the government is addressing it through grid expansion auctions and strong battery deployment, while a steady flow of permit requests indicates sustained interest in renewable power projects. Conversely, Mexico’s energy forecast has expanded since the country launched a USD 22-billion energy plan in early 2025, targeting 29 GW of new capacity – 6.4 GW from variable renewables – along with grid upgrades, storage expansion and revived projects.\n\nFor distributed PV, 30% more growth in the region is estimated in this year’s forecast than in last year’s. In Brazil, despite a 2023 policy change reducing remuneration under the net-metering scheme, distributed solar PV additions remain robust. Meanwhile, Mexico experienced record growth in distributed solar PV in 2024, adding 1 GW and reaching half a million users. This surge is largely the result of its attractive net metering programme, which is expected to enable an additional 13 GW of capacity by 2030, particularly benefiting commercial customers after the system-size threshold was increased from 0.5 MW to 0.7 MW.\n\n**Net renewable capacity additions by country and technology in Latin America, 2025- 2030**\n\nIEA. CC BY 4.0.\n\nSlow grid expansion continues to pose a significant challenge across Latin America. In Brazil, the increasing deployment of variable renewable energy has\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\n16\n\nNet additions (GW)\n\n**Utility-scale PV**\n\nBrazil Mexico Chile Argentina Colombia Rest of the region Renewables 2024 forecast\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n**Distributed PV**\n\n0\n\n1\n\n2\n\n3\n\n4\n\n5\n\n6\n\n**Wind**"],"full_content":"Renewables 2025 Chapter 1. Renewable electricity Analysis and forecasts to 2030\n\nPAGE | 44\n\nI EA. CC BY 4.0.\n\nlevels. Projections have been reduced 16% for utility-scale solar PV and 9% for wind, whereas distributed solar PV has been revised up 30%, offsetting the decline.\n\nForecasts for utility-scale solar PV and wind reveal diverging trends across key Latin American markets. For Brazil, projections have been revised downwards because curtailment (particularly in the Northeast) and transmission bottlenecks are reducing project profitability, lengthening connection queues and extending project lead times. Curtailment also remains a key challenge in Chile, but the government is addressing it through grid expansion auctions and strong battery deployment, while a steady flow of permit requests indicates sustained interest in renewable power projects. Conversely, Mexico’s energy forecast has expanded since the country launched a USD 22-billion energy plan in early 2025, targeting 29 GW of new capacity – 6.4 GW from variable renewables – along with grid upgrades, storage expansion and revived projects.\n\nFor distributed PV, 30% more growth in the region is estimated in this year’s forecast than in last year’s. In Brazil, despite a 2023 policy change reducing remuneration under the net-metering scheme, distributed solar PV additions remain robust. Meanwhile, Mexico experienced record growth in distributed solar PV in 2024, adding 1 GW and reaching half a million users. This surge is largely the result of its attractive net metering programme, which is expected to enable an additional 13 GW of capacity by 2030, particularly benefiting commercial customers after the system-size threshold was increased from 0.5 MW to 0.7 MW.\n\n**Net renewable capacity additions by country and technology in Latin America, 2025- 2030**\n\nIEA. CC BY 4.0.\n\nSlow grid expansion continues to pose a significant challenge across Latin America. In Brazil, the increasing deployment of variable renewable energy has\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\n16\n\nNet additions (GW)\n\n**Utility-scale PV**\n\nBrazil Mexico Chile Argentina Colombia Rest of the region Renewables 2024 forecast\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n**Distributed PV**\n\n0\n\n1\n\n2\n\n3\n\n4\n\n5\n\n6\n\n**Wind**","publish_date":null,"title":"levels. Projections have been reduced 16% for utility-scale solar PV and 9% for wind, whereas distributed solar PV has been","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Renewables2025_1_p44.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Renewables2025_1_p44.pdf"},"request":{"example_id":"chart/Renewables2025_1_p44","product_type":"parse","source_file_path":"Renewables2025_1_p44.pdf"},"started_at":"2026-11-04T19:45:59.860755"} +{"completed_at":"2026-11-04T19:31:20.126675","latency_in_ms":1038,"output":{"example_id":"chart/Renewables2025_1_p48","grounded_pages":[],"layout_pages":[],"markdown":"Renewables 2025 Chapter 1. Renewable electricity Analysis and forecasts to 2030\n\nPAGE | 48\n\nI EA. CC BY 4.0.\n\nOur review of the UNFCCC registry reveals that only 57 of the 196 Parties to the Paris Agreement had submitted NDC 3.0s as of 28 September 2025, accounting for only 23% of CO2 emissions from fuel combustion. Of these, 11 are advanced economies (19% of global CO2 emissions from fuel combustion), while the remaining 46 are developing and emerging countries (4%).\n\nEncouragingly, most submissions (54) mention renewable energy, recognising that it has a role to play in lowering GHG emissions. However, only 14 explicitly acknowledge the COP28 tripling pledge, and just 5 identify their renewable capacity ambitions for 2030, totalling 80 GW. This is less than 1% of the 11 500 GW needed to meet the global tripling target by the end of the decade.\n\n**Shares of fuel-combustion CO2 emissions in 2023 covered in NDC 3.0s (left) and country progress in submitting updates as of 28 September 2025 (right)**\n\nIEA. CC BY 4.0.\n\nNotes: NDC = Nationally Determined Contribution. On 20 January 2025, US Executive Order 14162 on Putting America First in International Environmental Agreements withdrew the United States from the Paris Climate Agreement, with official notification submitted to the UN on 27 January 2025, making the withdrawal effective on that day.\n\nHowever, the absence of a 2030 renewable capacity target does not necessarily indicate that a country lacks a plan to accelerate renewable capacity. Some countries have set their ambitions for 2035 instead of 2030, while others have expressed their renewable electricity targets in other units (e.g. share of power generation) or have embedded them within broader net-zero strategies.\n\nAdditionally, at the UN climate summit in New York on 24 September 2025, a further 10 countries verbally announced their intention to submit NDC 3.0s to the UNFCCC registry, but they have not yet done so. For example, China announced that it would increase its solar and wind capacity to three times the 2022 level (to around 3 600 GW) by 2035 under its planned NDC.\n\nNot submitted\n\n77%\n\nEmerging and\n\ndeveloping\n\n4%\n\nAdvanced\n\n19%\n\nSubmitted\n\n23%\n\n0\n\n50\n\n100\n\n150\n\n200\n\n250\n\nTotal Parties\n\nin Paris\n\nAgreement\n\nSubmitted NDC3.0s\n\nMentioned \"renewable\"\n\nenergy\n\nAcknowledged COP28 tripling\n\ntarget\n\nContained\n\n2030\n\nrenewable capacity\n\ntarget\n\nNumber of parties","pages":[{"line_numbers":[],"markdown":"Renewables 2025 Chapter 1. Renewable electricity Analysis and forecasts to 2030\n\nPAGE | 48\n\nI EA. CC BY 4.0.\n\nOur review of the UNFCCC registry reveals that only 57 of the 196 Parties to the Paris Agreement had submitted NDC 3.0s as of 28 September 2025, accounting for only 23% of CO2 emissions from fuel combustion. Of these, 11 are advanced economies (19% of global CO2 emissions from fuel combustion), while the remaining 46 are developing and emerging countries (4%).\n\nEncouragingly, most submissions (54) mention renewable energy, recognising that it has a role to play in lowering GHG emissions. However, only 14 explicitly acknowledge the COP28 tripling pledge, and just 5 identify their renewable capacity ambitions for 2030, totalling 80 GW. This is less than 1% of the 11 500 GW needed to meet the global tripling target by the end of the decade.\n\n**Shares of fuel-combustion CO2 emissions in 2023 covered in NDC 3.0s (left) and country progress in submitting updates as of 28 September 2025 (right)**\n\nIEA. CC BY 4.0.\n\nNotes: NDC = Nationally Determined Contribution. On 20 January 2025, US Executive Order 14162 on Putting America First in International Environmental Agreements withdrew the United States from the Paris Climate Agreement, with official notification submitted to the UN on 27 January 2025, making the withdrawal effective on that day.\n\nHowever, the absence of a 2030 renewable capacity target does not necessarily indicate that a country lacks a plan to accelerate renewable capacity. Some countries have set their ambitions for 2035 instead of 2030, while others have expressed their renewable electricity targets in other units (e.g. share of power generation) or have embedded them within broader net-zero strategies.\n\nAdditionally, at the UN climate summit in New York on 24 September 2025, a further 10 countries verbally announced their intention to submit NDC 3.0s to the UNFCCC registry, but they have not yet done so. For example, China announced that it would increase its solar and wind capacity to three times the 2022 level (to around 3 600 GW) by 2035 under its planned NDC.\n\nNot submitted\n\n77%\n\nEmerging and\n\ndeveloping\n\n4%\n\nAdvanced\n\n19%\n\nSubmitted\n\n23%\n\n0\n\n50\n\n100\n\n150\n\n200\n\n250\n\nTotal Parties\n\nin Paris\n\nAgreement\n\nSubmitted NDC3.0s\n\nMentioned \"renewable\"\n\nenergy\n\nAcknowledged COP28 tripling\n\ntarget\n\nContained\n\n2030\n\nrenewable capacity\n\ntarget\n\nNumber of parties","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Renewables 2025 Chapter 1. Renewable electricity Analysis and forecasts to 2030\n\nPAGE | 48\n\nI EA. CC BY 4.0.\n\nOur review of the UNFCCC registry reveals that only 57 of the 196 Parties to the Paris Agreement had submitted NDC 3.0s as of 28 September 2025, accounting for only 23% of CO2 emissions from fuel combustion. Of these, 11 are advanced economies (19% of global CO2 emissions from fuel combustion), while the remaining 46 are developing and emerging countries (4%).\n\nEncouragingly, most submissions (54) mention renewable energy, recognising that it has a role to play in lowering GHG emissions. However, only 14 explicitly acknowledge the COP28 tripling pledge, and just 5 identify their renewable capacity ambitions for 2030, totalling 80 GW. This is less than 1% of the 11 500 GW needed to meet the global tripling target by the end of the decade.\n\n**Shares of fuel-combustion CO2 emissions in 2023 covered in NDC 3.0s (left) and country progress in submitting updates as of 28 September 2025 (right)**\n\nIEA. CC BY 4.0.\n\nNotes: NDC = Nationally Determined Contribution. On 20 January 2025, US Executive Order 14162 on Putting America First in International Environmental Agreements withdrew the United States from the Paris Climate Agreement, with official notification submitted to the UN on 27 January 2025, making the withdrawal effective on that day.\n\nHowever, the absence of a 2030 renewable capacity target does not necessarily indicate that a country lacks a plan to accelerate renewable capacity. Some countries have set their ambitions for 2035 instead of 2030, while others have expressed their renewable electricity targets in other units (e.g. share of power generation) or have embedded them within broader net-zero strategies.\n\nAdditionally, at the UN climate summit in New York on 24 September 2025, a further 10 countries verbally announced their intention to submit NDC 3.0s to the UNFCCC registry, but they have not yet done so. For example, China announced that it would increase its solar and wind capacity to three times the 2022 level (to around 3 600 GW) by 2035 under its planned NDC.\n\nNot submitted\n\n77%\n\nEmerging and\n\ndeveloping\n\n4%\n\nAdvanced\n\n19%\n\nSubmitted\n\n23%\n\n0\n\n50\n\n100\n\n150\n\n200\n\n250\n\nTotal Parties\n\nin Paris\n\nAgreement\n\nSubmitted NDC3.0s\n\nMentioned \"renewable\"\n\nenergy\n\nAcknowledged COP28 tripling\n\ntarget\n\nContained\n\n2030\n\nrenewable capacity\n\ntarget\n\nNumber of parties","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Renewables2025_1_p48.pdf"]},"response":{"errors":[],"extract_id":"extract_5c7f91cd6563cd5b379af6a1e3db6020","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Renewables 2025 Chapter 1. Renewable electricity Analysis and forecasts to 2030\n\nPAGE | 48\n\nI EA. CC BY 4.0.\n\nOur review of the UNFCCC registry reveals that only 57 of the 196 Parties to the Paris Agreement had submitted NDC 3.0s as of 28 September 2025, accounting for only 23% of CO2 emissions from fuel combustion. Of these, 11 are advanced economies (19% of global CO2 emissions from fuel combustion), while the remaining 46 are developing and emerging countries (4%).\n\nEncouragingly, most submissions (54) mention renewable energy, recognising that it has a role to play in lowering GHG emissions. However, only 14 explicitly acknowledge the COP28 tripling pledge, and just 5 identify their renewable capacity ambitions for 2030, totalling 80 GW. This is less than 1% of the 11 500 GW needed to meet the global tripling target by the end of the decade.\n\n**Shares of fuel-combustion CO2 emissions in 2023 covered in NDC 3.0s (left) and country progress in submitting updates as of 28 September 2025 (right)**\n\nIEA. CC BY 4.0.\n\nNotes: NDC = Nationally Determined Contribution. On 20 January 2025, US Executive Order 14162 on Putting America First in International Environmental Agreements withdrew the United States from the Paris Climate Agreement, with official notification submitted to the UN on 27 January 2025, making the withdrawal effective on that day.\n\nHowever, the absence of a 2030 renewable capacity target does not necessarily indicate that a country lacks a plan to accelerate renewable capacity. Some countries have set their ambitions for 2035 instead of 2030, while others have expressed their renewable electricity targets in other units (e.g. share of power generation) or have embedded them within broader net-zero strategies.\n\nAdditionally, at the UN climate summit in New York on 24 September 2025, a further 10 countries verbally announced their intention to submit NDC 3.0s to the UNFCCC registry, but they have not yet done so. For example, China announced that it would increase its solar and wind capacity to three times the 2022 level (to around 3 600 GW) by 2035 under its planned NDC.\n\nNot submitted\n\n77%\n\nEmerging and\n\ndeveloping\n\n4%\n\nAdvanced\n\n19%\n\nSubmitted\n\n23%\n\n0\n\n50\n\n100\n\n150\n\n200\n\n250\n\nTotal Parties\n\nin Paris\n\nAgreement\n\nSubmitted NDC3.0s\n\nMentioned \"renewable\"\n\nenergy\n\nAcknowledged COP28 tripling\n\ntarget\n\nContained\n\n2030\n\nrenewable capacity\n\ntarget\n\nNumber of parties"],"full_content":"Renewables 2025 Chapter 1. Renewable electricity Analysis and forecasts to 2030\n\nPAGE | 48\n\nI EA. CC BY 4.0.\n\nOur review of the UNFCCC registry reveals that only 57 of the 196 Parties to the Paris Agreement had submitted NDC 3.0s as of 28 September 2025, accounting for only 23% of CO2 emissions from fuel combustion. Of these, 11 are advanced economies (19% of global CO2 emissions from fuel combustion), while the remaining 46 are developing and emerging countries (4%).\n\nEncouragingly, most submissions (54) mention renewable energy, recognising that it has a role to play in lowering GHG emissions. However, only 14 explicitly acknowledge the COP28 tripling pledge, and just 5 identify their renewable capacity ambitions for 2030, totalling 80 GW. This is less than 1% of the 11 500 GW needed to meet the global tripling target by the end of the decade.\n\n**Shares of fuel-combustion CO2 emissions in 2023 covered in NDC 3.0s (left) and country progress in submitting updates as of 28 September 2025 (right)**\n\nIEA. CC BY 4.0.\n\nNotes: NDC = Nationally Determined Contribution. On 20 January 2025, US Executive Order 14162 on Putting America First in International Environmental Agreements withdrew the United States from the Paris Climate Agreement, with official notification submitted to the UN on 27 January 2025, making the withdrawal effective on that day.\n\nHowever, the absence of a 2030 renewable capacity target does not necessarily indicate that a country lacks a plan to accelerate renewable capacity. Some countries have set their ambitions for 2035 instead of 2030, while others have expressed their renewable electricity targets in other units (e.g. share of power generation) or have embedded them within broader net-zero strategies.\n\nAdditionally, at the UN climate summit in New York on 24 September 2025, a further 10 countries verbally announced their intention to submit NDC 3.0s to the UNFCCC registry, but they have not yet done so. For example, China announced that it would increase its solar and wind capacity to three times the 2022 level (to around 3 600 GW) by 2035 under its planned NDC.\n\nNot submitted\n\n77%\n\nEmerging and\n\ndeveloping\n\n4%\n\nAdvanced\n\n19%\n\nSubmitted\n\n23%\n\n0\n\n50\n\n100\n\n150\n\n200\n\n250\n\nTotal Parties\n\nin Paris\n\nAgreement\n\nSubmitted NDC3.0s\n\nMentioned \"renewable\"\n\nenergy\n\nAcknowledged COP28 tripling\n\ntarget\n\nContained\n\n2030\n\nrenewable capacity\n\ntarget\n\nNumber of parties","publish_date":null,"title":"Our review of the UNFCCC registry reveals that only 57 of the 196 Parties to the Paris Agreement had submitted","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Renewables2025_1_p48.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Renewables2025_1_p48.pdf"},"request":{"example_id":"chart/Renewables2025_1_p48","product_type":"parse","source_file_path":"Renewables2025_1_p48.pdf"},"started_at":"2026-11-04T19:31:19.088007"} +{"completed_at":"2026-11-04T19:46:05.679872","latency_in_ms":2194,"output":{"example_id":"chart/Renewables2025_1_p49","grounded_pages":[],"layout_pages":[],"markdown":"Renewables 2025 Chapter 1. Renewable electricity Analysis and forecasts to 2030\n\nPAGE | 49\n\nI EA. CC BY 4.0.\n\n### While most countries are raising their renewable electricity goals outside of NDCs, overall global ambition is slightly lower than in last year’s analysis\n\nAlthough countries have so far been slow to submit their updated NDCs, many of their national plans already contain renewable capacity ambitions for 2030 or another indicator that can be used to estimate it, which could be used in their NDC 3.0. In fact, overall country ambitions for installed renewable power capacity in 2030 total roughly 7 500 GW. While this is 5% lower than our 2024 assessment, not all countries or technologies are following this trend.\n\nSince October 2024, nine countries have updated their national ambitions for 2030. However, only six made significant changes (greater than 1%). Of these six, three raised their ambitions, while the other three lowered them.\n\n**Global renewable electricity capacity ambitions by technology**\n\nIEA. CC BY 4.0.\n\nNotes: Ambition changes correspond to 23 countries that represent 83% of global installed renewable capacity in 2024.\n\nThe majority of the decrease in global ambition comes from the United States, where policy changes in January 2025 revoked the 2021 executive order that aimed for 100% clean energy by 2035. Other downward changes to 2030 goals come from Indonesia’s latest National Electricity Plan (RUKN 2025-2060), which is 20% less ambitious than its 2019 plan due to lower aims for solar PV. The latest energy plan for the Philippines (2023-2050) also envisions slightly lower installed capacity by 2030 than its previous plan, for both the reference and the clean- energy scenario. However, reductions in total capacity do not always reflect diminished ambition, as some declines stem from revised assumptions on demand growth or technology performance. For example, the Philippines lowered its capacity figures but still maintains its previous ambition for a 35% share of renewables by 2030.\n\n0\n\n1 000\n\n2 000\n\n3 000\n\n4 000\n\n5 000\n\n6 000\n\n7 000\n\n8 000\n\n9 000\n\nRenewables\n\n2024\n\nRaised ambition\n\nLowered ambition\n\nRenewables\n\n2025\n\nGW\n\nNot specified Others Wind Solar PV Hydropower\n\n0\n\n500\n\n1 000\n\n1 500\n\n2 000\n\n2 500\n\n3 000\n\nHydro- power\n\nSolar PV\n\nWind Others Not\n\nspecified\n\nRenewables 2024\n\nRenewables 2025","pages":[{"line_numbers":[],"markdown":"Renewables 2025 Chapter 1. Renewable electricity Analysis and forecasts to 2030\n\nPAGE | 49\n\nI EA. CC BY 4.0.\n\n### While most countries are raising their renewable electricity goals outside of NDCs, overall global ambition is slightly lower than in last year’s analysis\n\nAlthough countries have so far been slow to submit their updated NDCs, many of their national plans already contain renewable capacity ambitions for 2030 or another indicator that can be used to estimate it, which could be used in their NDC 3.0. In fact, overall country ambitions for installed renewable power capacity in 2030 total roughly 7 500 GW. While this is 5% lower than our 2024 assessment, not all countries or technologies are following this trend.\n\nSince October 2024, nine countries have updated their national ambitions for 2030. However, only six made significant changes (greater than 1%). Of these six, three raised their ambitions, while the other three lowered them.\n\n**Global renewable electricity capacity ambitions by technology**\n\nIEA. CC BY 4.0.\n\nNotes: Ambition changes correspond to 23 countries that represent 83% of global installed renewable capacity in 2024.\n\nThe majority of the decrease in global ambition comes from the United States, where policy changes in January 2025 revoked the 2021 executive order that aimed for 100% clean energy by 2035. Other downward changes to 2030 goals come from Indonesia’s latest National Electricity Plan (RUKN 2025-2060), which is 20% less ambitious than its 2019 plan due to lower aims for solar PV. The latest energy plan for the Philippines (2023-2050) also envisions slightly lower installed capacity by 2030 than its previous plan, for both the reference and the clean- energy scenario. However, reductions in total capacity do not always reflect diminished ambition, as some declines stem from revised assumptions on demand growth or technology performance. For example, the Philippines lowered its capacity figures but still maintains its previous ambition for a 35% share of renewables by 2030.\n\n0\n\n1 000\n\n2 000\n\n3 000\n\n4 000\n\n5 000\n\n6 000\n\n7 000\n\n8 000\n\n9 000\n\nRenewables\n\n2024\n\nRaised ambition\n\nLowered ambition\n\nRenewables\n\n2025\n\nGW\n\nNot specified Others Wind Solar PV Hydropower\n\n0\n\n500\n\n1 000\n\n1 500\n\n2 000\n\n2 500\n\n3 000\n\nHydro- power\n\nSolar PV\n\nWind Others Not\n\nspecified\n\nRenewables 2024\n\nRenewables 2025","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Renewables 2025 Chapter 1. Renewable electricity Analysis and forecasts to 2030\n\nPAGE | 49\n\nI EA. CC BY 4.0.\n\n### While most countries are raising their renewable electricity goals outside of NDCs, overall global ambition is slightly lower than in last year’s analysis\n\nAlthough countries have so far been slow to submit their updated NDCs, many of their national plans already contain renewable capacity ambitions for 2030 or another indicator that can be used to estimate it, which could be used in their NDC 3.0. In fact, overall country ambitions for installed renewable power capacity in 2030 total roughly 7 500 GW. While this is 5% lower than our 2024 assessment, not all countries or technologies are following this trend.\n\nSince October 2024, nine countries have updated their national ambitions for 2030. However, only six made significant changes (greater than 1%). Of these six, three raised their ambitions, while the other three lowered them.\n\n**Global renewable electricity capacity ambitions by technology**\n\nIEA. CC BY 4.0.\n\nNotes: Ambition changes correspond to 23 countries that represent 83% of global installed renewable capacity in 2024.\n\nThe majority of the decrease in global ambition comes from the United States, where policy changes in January 2025 revoked the 2021 executive order that aimed for 100% clean energy by 2035. Other downward changes to 2030 goals come from Indonesia’s latest National Electricity Plan (RUKN 2025-2060), which is 20% less ambitious than its 2019 plan due to lower aims for solar PV. The latest energy plan for the Philippines (2023-2050) also envisions slightly lower installed capacity by 2030 than its previous plan, for both the reference and the clean- energy scenario. However, reductions in total capacity do not always reflect diminished ambition, as some declines stem from revised assumptions on demand growth or technology performance. For example, the Philippines lowered its capacity figures but still maintains its previous ambition for a 35% share of renewables by 2030.\n\n0\n\n1 000\n\n2 000\n\n3 000\n\n4 000\n\n5 000\n\n6 000\n\n7 000\n\n8 000\n\n9 000\n\nRenewables\n\n2024\n\nRaised ambition\n\nLowered ambition\n\nRenewables\n\n2025\n\nGW\n\nNot specified Others Wind Solar PV Hydropower\n\n0\n\n500\n\n1 000\n\n1 500\n\n2 000\n\n2 500\n\n3 000\n\nHydro- power\n\nSolar PV\n\nWind Others Not\n\nspecified\n\nRenewables 2024\n\nRenewables 2025","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Renewables2025_1_p49.pdf"]},"response":{"errors":[],"extract_id":"extract_97d947f51f87a3fd9fe99dc9b558c530","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Renewables 2025 Chapter 1. Renewable electricity Analysis and forecasts to 2030\n\nPAGE | 49\n\nI EA. CC BY 4.0.\n\n### While most countries are raising their renewable electricity goals outside of NDCs, overall global ambition is slightly lower than in last year’s analysis\n\nAlthough countries have so far been slow to submit their updated NDCs, many of their national plans already contain renewable capacity ambitions for 2030 or another indicator that can be used to estimate it, which could be used in their NDC 3.0. In fact, overall country ambitions for installed renewable power capacity in 2030 total roughly 7 500 GW. While this is 5% lower than our 2024 assessment, not all countries or technologies are following this trend.\n\nSince October 2024, nine countries have updated their national ambitions for 2030. However, only six made significant changes (greater than 1%). Of these six, three raised their ambitions, while the other three lowered them.\n\n**Global renewable electricity capacity ambitions by technology**\n\nIEA. CC BY 4.0.\n\nNotes: Ambition changes correspond to 23 countries that represent 83% of global installed renewable capacity in 2024.\n\nThe majority of the decrease in global ambition comes from the United States, where policy changes in January 2025 revoked the 2021 executive order that aimed for 100% clean energy by 2035. Other downward changes to 2030 goals come from Indonesia’s latest National Electricity Plan (RUKN 2025-2060), which is 20% less ambitious than its 2019 plan due to lower aims for solar PV. The latest energy plan for the Philippines (2023-2050) also envisions slightly lower installed capacity by 2030 than its previous plan, for both the reference and the clean- energy scenario. However, reductions in total capacity do not always reflect diminished ambition, as some declines stem from revised assumptions on demand growth or technology performance. For example, the Philippines lowered its capacity figures but still maintains its previous ambition for a 35% share of renewables by 2030.\n\n0\n\n1 000\n\n2 000\n\n3 000\n\n4 000\n\n5 000\n\n6 000\n\n7 000\n\n8 000\n\n9 000\n\nRenewables\n\n2024\n\nRaised ambition\n\nLowered ambition\n\nRenewables\n\n2025\n\nGW\n\nNot specified Others Wind Solar PV Hydropower\n\n0\n\n500\n\n1 000\n\n1 500\n\n2 000\n\n2 500\n\n3 000\n\nHydro- power\n\nSolar PV\n\nWind Others Not\n\nspecified\n\nRenewables 2024\n\nRenewables 2025"],"full_content":"Renewables 2025 Chapter 1. Renewable electricity Analysis and forecasts to 2030\n\nPAGE | 49\n\nI EA. CC BY 4.0.\n\n### While most countries are raising their renewable electricity goals outside of NDCs, overall global ambition is slightly lower than in last year’s analysis\n\nAlthough countries have so far been slow to submit their updated NDCs, many of their national plans already contain renewable capacity ambitions for 2030 or another indicator that can be used to estimate it, which could be used in their NDC 3.0. In fact, overall country ambitions for installed renewable power capacity in 2030 total roughly 7 500 GW. While this is 5% lower than our 2024 assessment, not all countries or technologies are following this trend.\n\nSince October 2024, nine countries have updated their national ambitions for 2030. However, only six made significant changes (greater than 1%). Of these six, three raised their ambitions, while the other three lowered them.\n\n**Global renewable electricity capacity ambitions by technology**\n\nIEA. CC BY 4.0.\n\nNotes: Ambition changes correspond to 23 countries that represent 83% of global installed renewable capacity in 2024.\n\nThe majority of the decrease in global ambition comes from the United States, where policy changes in January 2025 revoked the 2021 executive order that aimed for 100% clean energy by 2035. Other downward changes to 2030 goals come from Indonesia’s latest National Electricity Plan (RUKN 2025-2060), which is 20% less ambitious than its 2019 plan due to lower aims for solar PV. The latest energy plan for the Philippines (2023-2050) also envisions slightly lower installed capacity by 2030 than its previous plan, for both the reference and the clean- energy scenario. However, reductions in total capacity do not always reflect diminished ambition, as some declines stem from revised assumptions on demand growth or technology performance. For example, the Philippines lowered its capacity figures but still maintains its previous ambition for a 35% share of renewables by 2030.\n\n0\n\n1 000\n\n2 000\n\n3 000\n\n4 000\n\n5 000\n\n6 000\n\n7 000\n\n8 000\n\n9 000\n\nRenewables\n\n2024\n\nRaised ambition\n\nLowered ambition\n\nRenewables\n\n2025\n\nGW\n\nNot specified Others Wind Solar PV Hydropower\n\n0\n\n500\n\n1 000\n\n1 500\n\n2 000\n\n2 500\n\n3 000\n\nHydro- power\n\nSolar PV\n\nWind Others Not\n\nspecified\n\nRenewables 2024\n\nRenewables 2025","publish_date":null,"title":"While most countries are raising their renewable electricity goals outside of NDCs, overall global ambition is slightly lower than in","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Renewables2025_1_p49.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Renewables2025_1_p49.pdf"},"request":{"example_id":"chart/Renewables2025_1_p49","product_type":"parse","source_file_path":"Renewables2025_1_p49.pdf"},"started_at":"2026-11-04T19:46:03.485507"} +{"completed_at":"2026-11-04T19:46:13.721872","latency_in_ms":8484,"output":{"example_id":"chart/Renewables2025_1_p75","grounded_pages":[],"layout_pages":[],"markdown":"Renewables 2025 Chapter 1. Renewable electricity Analysis and forecasts to 2030\n\nPAGE | 75\n\nI EA. CC BY 4.0.\n\n### Governments are reluctant to accept EACs in emissions-based trade rules\n\nWhile internationally recognised GHG accounting frameworks increasingly accommodate EAC use, national governments remain hesitant about accepting foreign certificates, particularly in regulatory and trade-related contexts.\n\nThe GHG Protocol, widely used by industry for corporate emissions reporting, defines two distinct approaches: location-based and market-based. The location- based method relies on the average emissions factor of the local electricity grid, while the market-based approach allows companies to report lower emissions by procuring EACs regardless of the grid’s underlying emissions intensity.\n\nCorporate initiatives and industry platforms, including RE100, have largely allowed the market-based approach. RE100 permits the use of most recognised EAC systems globally, enabling multinational companies to claim 100% renewable electricity sourcing. However, conditions may apply to avoid double counting or ensure environmental integrity. For example, since May 2025, companies in China have been required to purchase both green electricity certificates and carbon- offset certificates to meet RE100 requirements to avoid potential double counting.\n\n**Illustrative example of the impact of EACs on the carbon footprint of CAN fertilisers**\n\nIEA. CC BY 4.0.\n\nNotes: EAC = energy attribute certificate. CBAM = Carbon Border Adjustment Mechanism. We assumed a grid emissions factor of 561 g CO2-e/kWh, a price of EUR 70/t CO2-e for a CBAM certificate and EUR 0.3/MWh for an EAC.\n\nIn contrast, national governments have shown greater reluctance to incorporate foreign-issued EACs into official GHG accounting methodologies, especially in sectors tied to trade-related regulations. This is particularly relevant in the context of emerging emissions-based trade rules and product-specific emissions standards, such as those under discussion in the EU Carbon Border Adjustment Mechanism (CBAM) or in the new Batteries Regulation.\n\n0\n\n20\n\n40\n\n60\n\n80\n\n100\n\n120\n\n0\\.0\n\n0\\.2\n\n0\\.4\n\n0\\.6\n\n0\\.8\n\n1\\.0\n\n1\\.2\n\n1\\.4\n\n1\\.6\n\nWithout EACs With EACs\n\nFossil (coal-based) Grid-based\n\nEUR/t\n\nt CO2-e/t\n\nCarbon footprint\n\nCBAM costs\n\nEAC costs","pages":[{"line_numbers":[],"markdown":"Renewables 2025 Chapter 1. Renewable electricity Analysis and forecasts to 2030\n\nPAGE | 75\n\nI EA. CC BY 4.0.\n\n### Governments are reluctant to accept EACs in emissions-based trade rules\n\nWhile internationally recognised GHG accounting frameworks increasingly accommodate EAC use, national governments remain hesitant about accepting foreign certificates, particularly in regulatory and trade-related contexts.\n\nThe GHG Protocol, widely used by industry for corporate emissions reporting, defines two distinct approaches: location-based and market-based. The location- based method relies on the average emissions factor of the local electricity grid, while the market-based approach allows companies to report lower emissions by procuring EACs regardless of the grid’s underlying emissions intensity.\n\nCorporate initiatives and industry platforms, including RE100, have largely allowed the market-based approach. RE100 permits the use of most recognised EAC systems globally, enabling multinational companies to claim 100% renewable electricity sourcing. However, conditions may apply to avoid double counting or ensure environmental integrity. For example, since May 2025, companies in China have been required to purchase both green electricity certificates and carbon- offset certificates to meet RE100 requirements to avoid potential double counting.\n\n**Illustrative example of the impact of EACs on the carbon footprint of CAN fertilisers**\n\nIEA. CC BY 4.0.\n\nNotes: EAC = energy attribute certificate. CBAM = Carbon Border Adjustment Mechanism. We assumed a grid emissions factor of 561 g CO2-e/kWh, a price of EUR 70/t CO2-e for a CBAM certificate and EUR 0.3/MWh for an EAC.\n\nIn contrast, national governments have shown greater reluctance to incorporate foreign-issued EACs into official GHG accounting methodologies, especially in sectors tied to trade-related regulations. This is particularly relevant in the context of emerging emissions-based trade rules and product-specific emissions standards, such as those under discussion in the EU Carbon Border Adjustment Mechanism (CBAM) or in the new Batteries Regulation.\n\n0\n\n20\n\n40\n\n60\n\n80\n\n100\n\n120\n\n0\\.0\n\n0\\.2\n\n0\\.4\n\n0\\.6\n\n0\\.8\n\n1\\.0\n\n1\\.2\n\n1\\.4\n\n1\\.6\n\nWithout EACs With EACs\n\nFossil (coal-based) Grid-based\n\nEUR/t\n\nt CO2-e/t\n\nCarbon footprint\n\nCBAM costs\n\nEAC costs","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Renewables 2025 Chapter 1. Renewable electricity Analysis and forecasts to 2030\n\nPAGE | 75\n\nI EA. CC BY 4.0.\n\n### Governments are reluctant to accept EACs in emissions-based trade rules\n\nWhile internationally recognised GHG accounting frameworks increasingly accommodate EAC use, national governments remain hesitant about accepting foreign certificates, particularly in regulatory and trade-related contexts.\n\nThe GHG Protocol, widely used by industry for corporate emissions reporting, defines two distinct approaches: location-based and market-based. The location- based method relies on the average emissions factor of the local electricity grid, while the market-based approach allows companies to report lower emissions by procuring EACs regardless of the grid’s underlying emissions intensity.\n\nCorporate initiatives and industry platforms, including RE100, have largely allowed the market-based approach. RE100 permits the use of most recognised EAC systems globally, enabling multinational companies to claim 100% renewable electricity sourcing. However, conditions may apply to avoid double counting or ensure environmental integrity. For example, since May 2025, companies in China have been required to purchase both green electricity certificates and carbon- offset certificates to meet RE100 requirements to avoid potential double counting.\n\n**Illustrative example of the impact of EACs on the carbon footprint of CAN fertilisers**\n\nIEA. CC BY 4.0.\n\nNotes: EAC = energy attribute certificate. CBAM = Carbon Border Adjustment Mechanism. We assumed a grid emissions factor of 561 g CO2-e/kWh, a price of EUR 70/t CO2-e for a CBAM certificate and EUR 0.3/MWh for an EAC.\n\nIn contrast, national governments have shown greater reluctance to incorporate foreign-issued EACs into official GHG accounting methodologies, especially in sectors tied to trade-related regulations. This is particularly relevant in the context of emerging emissions-based trade rules and product-specific emissions standards, such as those under discussion in the EU Carbon Border Adjustment Mechanism (CBAM) or in the new Batteries Regulation.\n\n0\n\n20\n\n40\n\n60\n\n80\n\n100\n\n120\n\n0\\.0\n\n0\\.2\n\n0\\.4\n\n0\\.6\n\n0\\.8\n\n1\\.0\n\n1\\.2\n\n1\\.4\n\n1\\.6\n\nWithout EACs With EACs\n\nFossil (coal-based) Grid-based\n\nEUR/t\n\nt CO2-e/t\n\nCarbon footprint\n\nCBAM costs\n\nEAC costs","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Renewables2025_1_p75.pdf"]},"response":{"errors":[],"extract_id":"extract_300cd7017cfa90ac5a7b37955a2e1766","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Renewables 2025 Chapter 1. Renewable electricity Analysis and forecasts to 2030\n\nPAGE | 75\n\nI EA. CC BY 4.0.\n\n### Governments are reluctant to accept EACs in emissions-based trade rules\n\nWhile internationally recognised GHG accounting frameworks increasingly accommodate EAC use, national governments remain hesitant about accepting foreign certificates, particularly in regulatory and trade-related contexts.\n\nThe GHG Protocol, widely used by industry for corporate emissions reporting, defines two distinct approaches: location-based and market-based. The location- based method relies on the average emissions factor of the local electricity grid, while the market-based approach allows companies to report lower emissions by procuring EACs regardless of the grid’s underlying emissions intensity.\n\nCorporate initiatives and industry platforms, including RE100, have largely allowed the market-based approach. RE100 permits the use of most recognised EAC systems globally, enabling multinational companies to claim 100% renewable electricity sourcing. However, conditions may apply to avoid double counting or ensure environmental integrity. For example, since May 2025, companies in China have been required to purchase both green electricity certificates and carbon- offset certificates to meet RE100 requirements to avoid potential double counting.\n\n**Illustrative example of the impact of EACs on the carbon footprint of CAN fertilisers**\n\nIEA. CC BY 4.0.\n\nNotes: EAC = energy attribute certificate. CBAM = Carbon Border Adjustment Mechanism. We assumed a grid emissions factor of 561 g CO2-e/kWh, a price of EUR 70/t CO2-e for a CBAM certificate and EUR 0.3/MWh for an EAC.\n\nIn contrast, national governments have shown greater reluctance to incorporate foreign-issued EACs into official GHG accounting methodologies, especially in sectors tied to trade-related regulations. This is particularly relevant in the context of emerging emissions-based trade rules and product-specific emissions standards, such as those under discussion in the EU Carbon Border Adjustment Mechanism (CBAM) or in the new Batteries Regulation.\n\n0\n\n20\n\n40\n\n60\n\n80\n\n100\n\n120\n\n0\\.0\n\n0\\.2\n\n0\\.4\n\n0\\.6\n\n0\\.8\n\n1\\.0\n\n1\\.2\n\n1\\.4\n\n1\\.6\n\nWithout EACs With EACs\n\nFossil (coal-based) Grid-based\n\nEUR/t\n\nt CO2-e/t\n\nCarbon footprint\n\nCBAM costs\n\nEAC costs"],"full_content":"Renewables 2025 Chapter 1. Renewable electricity Analysis and forecasts to 2030\n\nPAGE | 75\n\nI EA. CC BY 4.0.\n\n### Governments are reluctant to accept EACs in emissions-based trade rules\n\nWhile internationally recognised GHG accounting frameworks increasingly accommodate EAC use, national governments remain hesitant about accepting foreign certificates, particularly in regulatory and trade-related contexts.\n\nThe GHG Protocol, widely used by industry for corporate emissions reporting, defines two distinct approaches: location-based and market-based. The location- based method relies on the average emissions factor of the local electricity grid, while the market-based approach allows companies to report lower emissions by procuring EACs regardless of the grid’s underlying emissions intensity.\n\nCorporate initiatives and industry platforms, including RE100, have largely allowed the market-based approach. RE100 permits the use of most recognised EAC systems globally, enabling multinational companies to claim 100% renewable electricity sourcing. However, conditions may apply to avoid double counting or ensure environmental integrity. For example, since May 2025, companies in China have been required to purchase both green electricity certificates and carbon- offset certificates to meet RE100 requirements to avoid potential double counting.\n\n**Illustrative example of the impact of EACs on the carbon footprint of CAN fertilisers**\n\nIEA. CC BY 4.0.\n\nNotes: EAC = energy attribute certificate. CBAM = Carbon Border Adjustment Mechanism. We assumed a grid emissions factor of 561 g CO2-e/kWh, a price of EUR 70/t CO2-e for a CBAM certificate and EUR 0.3/MWh for an EAC.\n\nIn contrast, national governments have shown greater reluctance to incorporate foreign-issued EACs into official GHG accounting methodologies, especially in sectors tied to trade-related regulations. This is particularly relevant in the context of emerging emissions-based trade rules and product-specific emissions standards, such as those under discussion in the EU Carbon Border Adjustment Mechanism (CBAM) or in the new Batteries Regulation.\n\n0\n\n20\n\n40\n\n60\n\n80\n\n100\n\n120\n\n0\\.0\n\n0\\.2\n\n0\\.4\n\n0\\.6\n\n0\\.8\n\n1\\.0\n\n1\\.2\n\n1\\.4\n\n1\\.6\n\nWithout EACs With EACs\n\nFossil (coal-based) Grid-based\n\nEUR/t\n\nt CO2-e/t\n\nCarbon footprint\n\nCBAM costs\n\nEAC costs","publish_date":null,"title":"Governments are reluctant to accept EACs in emissions-based trade rules","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Renewables2025_1_p75.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Renewables2025_1_p75.pdf"},"request":{"example_id":"chart/Renewables2025_1_p75","product_type":"parse","source_file_path":"Renewables2025_1_p75.pdf"},"started_at":"2026-11-04T19:46:05.237841"} +{"completed_at":"2026-11-04T19:31:14.488987","latency_in_ms":2594,"output":{"example_id":"chart/Renewables2025_1_p81","grounded_pages":[],"layout_pages":[],"markdown":"Renewables 2025 Chapter 1. Renewable electricity Analysis and forecasts to 2030\n\nPAGE | 81\n\nI EA. CC BY 4.0.\n\nand Greece, where the difference could have been as much as 30-50 percentage points. In the Low-RES scenario, the energy security challenges during the 2022 energy crisis would have been significantly more severe.\n\n**Net imports of coal and natural gas, actual and in Low-RES scenario, 2023**\n\nIEA. CC BY 4.0.\n\nNotes: RES = renewable energy source. The difference in volume of net imports between the Low-RES scenario and actual in Brazil is equal to 230%.\n\nThe United Kingdom expanded its non-hydro renewable electricity generation nearly sixfold between 2010 and 2023. As a result, the share of its electricity supply met by imported fossil fuels decreased from around 45% in 2013 to less than 25% in 2023, despite declining domestic coal and gas output. In the Low-RES scenario, import dependence would have approached 60% by 2023.\n\nIn China, despite massive domestic coal production, imports made up about 10% of the country’s total coal supply in 2023 – and nearly 40% of its natural gas. Without deployment of renewables over the past decade, China’ s fossil fuel- based electricity generation would have been more than 25% higher. This would have potentially required a doubling of fossil fuel imports, raising China’s electricity supply import dependence from 7% to nearly 25%.\n\nBrazil, with one of the lowest fossil fuel import dependencies among large economies – around 5% in 2023 – would also have experienced significant impacts. Without wind and solar energy deployment, fossil-based generation would have needed to rise by around 170 TWh, primarily from imported natural gas. Imports of natural gas would have increased nearly five-fold, pushing electricity import dependence to almost 30%, despite the country’s large hydropower base.\n\n0%\n\n20%\n\n40%\n\n60%\n\n80%\n\n100%\n\n120%\n\n0\n\n50\n\n100\n\n150\n\n200\n\n250\n\nActual\n\nLow RES\n\nActual\n\nLow RES\n\nActual\n\nLow RES\n\nActual\n\nLow RES\n\nActual\n\nLow RES\n\nActual\n\nLow RES\n\nActual\n\nLow RES\n\nActual\n\nLow RES\n\nActual\n\nLow RES\n\nJapan India Korea GermanyTürkiye Italy France UK Brazil\n\nMtoe\n\nNet imports for other use Net imports for power Difference (%)\n\n0%\n\n20%\n\n40%\n\n60%\n\n80%\n\n100%\n\n120%\n\n0\n\n100\n\n200\n\n300\n\n400\n\n500\n\n600\n\n700\n\n800\n\nActual\n\nLow RES\n\nActual\n\nLow RES\n\nChina EU\n\nMtoe","pages":[{"line_numbers":[],"markdown":"Renewables 2025 Chapter 1. Renewable electricity Analysis and forecasts to 2030\n\nPAGE | 81\n\nI EA. CC BY 4.0.\n\nand Greece, where the difference could have been as much as 30-50 percentage points. In the Low-RES scenario, the energy security challenges during the 2022 energy crisis would have been significantly more severe.\n\n**Net imports of coal and natural gas, actual and in Low-RES scenario, 2023**\n\nIEA. CC BY 4.0.\n\nNotes: RES = renewable energy source. The difference in volume of net imports between the Low-RES scenario and actual in Brazil is equal to 230%.\n\nThe United Kingdom expanded its non-hydro renewable electricity generation nearly sixfold between 2010 and 2023. As a result, the share of its electricity supply met by imported fossil fuels decreased from around 45% in 2013 to less than 25% in 2023, despite declining domestic coal and gas output. In the Low-RES scenario, import dependence would have approached 60% by 2023.\n\nIn China, despite massive domestic coal production, imports made up about 10% of the country’s total coal supply in 2023 – and nearly 40% of its natural gas. Without deployment of renewables over the past decade, China’ s fossil fuel- based electricity generation would have been more than 25% higher. This would have potentially required a doubling of fossil fuel imports, raising China’s electricity supply import dependence from 7% to nearly 25%.\n\nBrazil, with one of the lowest fossil fuel import dependencies among large economies – around 5% in 2023 – would also have experienced significant impacts. Without wind and solar energy deployment, fossil-based generation would have needed to rise by around 170 TWh, primarily from imported natural gas. Imports of natural gas would have increased nearly five-fold, pushing electricity import dependence to almost 30%, despite the country’s large hydropower base.\n\n0%\n\n20%\n\n40%\n\n60%\n\n80%\n\n100%\n\n120%\n\n0\n\n50\n\n100\n\n150\n\n200\n\n250\n\nActual\n\nLow RES\n\nActual\n\nLow RES\n\nActual\n\nLow RES\n\nActual\n\nLow RES\n\nActual\n\nLow RES\n\nActual\n\nLow RES\n\nActual\n\nLow RES\n\nActual\n\nLow RES\n\nActual\n\nLow RES\n\nJapan India Korea GermanyTürkiye Italy France UK Brazil\n\nMtoe\n\nNet imports for other use Net imports for power Difference (%)\n\n0%\n\n20%\n\n40%\n\n60%\n\n80%\n\n100%\n\n120%\n\n0\n\n100\n\n200\n\n300\n\n400\n\n500\n\n600\n\n700\n\n800\n\nActual\n\nLow RES\n\nActual\n\nLow RES\n\nChina EU\n\nMtoe","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Renewables 2025 Chapter 1. Renewable electricity Analysis and forecasts to 2030\n\nPAGE | 81\n\nI EA. CC BY 4.0.\n\nand Greece, where the difference could have been as much as 30-50 percentage points. In the Low-RES scenario, the energy security challenges during the 2022 energy crisis would have been significantly more severe.\n\n**Net imports of coal and natural gas, actual and in Low-RES scenario, 2023**\n\nIEA. CC BY 4.0.\n\nNotes: RES = renewable energy source. The difference in volume of net imports between the Low-RES scenario and actual in Brazil is equal to 230%.\n\nThe United Kingdom expanded its non-hydro renewable electricity generation nearly sixfold between 2010 and 2023. As a result, the share of its electricity supply met by imported fossil fuels decreased from around 45% in 2013 to less than 25% in 2023, despite declining domestic coal and gas output. In the Low-RES scenario, import dependence would have approached 60% by 2023.\n\nIn China, despite massive domestic coal production, imports made up about 10% of the country’s total coal supply in 2023 – and nearly 40% of its natural gas. Without deployment of renewables over the past decade, China’ s fossil fuel- based electricity generation would have been more than 25% higher. This would have potentially required a doubling of fossil fuel imports, raising China’s electricity supply import dependence from 7% to nearly 25%.\n\nBrazil, with one of the lowest fossil fuel import dependencies among large economies – around 5% in 2023 – would also have experienced significant impacts. Without wind and solar energy deployment, fossil-based generation would have needed to rise by around 170 TWh, primarily from imported natural gas. Imports of natural gas would have increased nearly five-fold, pushing electricity import dependence to almost 30%, despite the country’s large hydropower base.\n\n0%\n\n20%\n\n40%\n\n60%\n\n80%\n\n100%\n\n120%\n\n0\n\n50\n\n100\n\n150\n\n200\n\n250\n\nActual\n\nLow RES\n\nActual\n\nLow RES\n\nActual\n\nLow RES\n\nActual\n\nLow RES\n\nActual\n\nLow RES\n\nActual\n\nLow RES\n\nActual\n\nLow RES\n\nActual\n\nLow RES\n\nActual\n\nLow RES\n\nJapan India Korea GermanyTürkiye Italy France UK Brazil\n\nMtoe\n\nNet imports for other use Net imports for power Difference (%)\n\n0%\n\n20%\n\n40%\n\n60%\n\n80%\n\n100%\n\n120%\n\n0\n\n100\n\n200\n\n300\n\n400\n\n500\n\n600\n\n700\n\n800\n\nActual\n\nLow RES\n\nActual\n\nLow RES\n\nChina EU\n\nMtoe","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Renewables2025_1_p81.pdf"]},"response":{"errors":[],"extract_id":"extract_929f957b8cd6a25001cd362910592820","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Renewables 2025 Chapter 1. Renewable electricity Analysis and forecasts to 2030\n\nPAGE | 81\n\nI EA. CC BY 4.0.\n\nand Greece, where the difference could have been as much as 30-50 percentage points. In the Low-RES scenario, the energy security challenges during the 2022 energy crisis would have been significantly more severe.\n\n**Net imports of coal and natural gas, actual and in Low-RES scenario, 2023**\n\nIEA. CC BY 4.0.\n\nNotes: RES = renewable energy source. The difference in volume of net imports between the Low-RES scenario and actual in Brazil is equal to 230%.\n\nThe United Kingdom expanded its non-hydro renewable electricity generation nearly sixfold between 2010 and 2023. As a result, the share of its electricity supply met by imported fossil fuels decreased from around 45% in 2013 to less than 25% in 2023, despite declining domestic coal and gas output. In the Low-RES scenario, import dependence would have approached 60% by 2023.\n\nIn China, despite massive domestic coal production, imports made up about 10% of the country’s total coal supply in 2023 – and nearly 40% of its natural gas. Without deployment of renewables over the past decade, China’ s fossil fuel- based electricity generation would have been more than 25% higher. This would have potentially required a doubling of fossil fuel imports, raising China’s electricity supply import dependence from 7% to nearly 25%.\n\nBrazil, with one of the lowest fossil fuel import dependencies among large economies – around 5% in 2023 – would also have experienced significant impacts. Without wind and solar energy deployment, fossil-based generation would have needed to rise by around 170 TWh, primarily from imported natural gas. Imports of natural gas would have increased nearly five-fold, pushing electricity import dependence to almost 30%, despite the country’s large hydropower base.\n\n0%\n\n20%\n\n40%\n\n60%\n\n80%\n\n100%\n\n120%\n\n0\n\n50\n\n100\n\n150\n\n200\n\n250\n\nActual\n\nLow RES\n\nActual\n\nLow RES\n\nActual\n\nLow RES\n\nActual\n\nLow RES\n\nActual\n\nLow RES\n\nActual\n\nLow RES\n\nActual\n\nLow RES\n\nActual\n\nLow RES\n\nActual\n\nLow RES\n\nJapan India Korea GermanyTürkiye Italy France UK Brazil\n\nMtoe\n\nNet imports for other use Net imports for power Difference (%)\n\n0%\n\n20%\n\n40%\n\n60%\n\n80%\n\n100%\n\n120%\n\n0\n\n100\n\n200\n\n300\n\n400\n\n500\n\n600\n\n700\n\n800\n\nActual\n\nLow RES\n\nActual\n\nLow RES\n\nChina EU\n\nMtoe"],"full_content":"Renewables 2025 Chapter 1. Renewable electricity Analysis and forecasts to 2030\n\nPAGE | 81\n\nI EA. CC BY 4.0.\n\nand Greece, where the difference could have been as much as 30-50 percentage points. In the Low-RES scenario, the energy security challenges during the 2022 energy crisis would have been significantly more severe.\n\n**Net imports of coal and natural gas, actual and in Low-RES scenario, 2023**\n\nIEA. CC BY 4.0.\n\nNotes: RES = renewable energy source. The difference in volume of net imports between the Low-RES scenario and actual in Brazil is equal to 230%.\n\nThe United Kingdom expanded its non-hydro renewable electricity generation nearly sixfold between 2010 and 2023. As a result, the share of its electricity supply met by imported fossil fuels decreased from around 45% in 2013 to less than 25% in 2023, despite declining domestic coal and gas output. In the Low-RES scenario, import dependence would have approached 60% by 2023.\n\nIn China, despite massive domestic coal production, imports made up about 10% of the country’s total coal supply in 2023 – and nearly 40% of its natural gas. Without deployment of renewables over the past decade, China’ s fossil fuel- based electricity generation would have been more than 25% higher. This would have potentially required a doubling of fossil fuel imports, raising China’s electricity supply import dependence from 7% to nearly 25%.\n\nBrazil, with one of the lowest fossil fuel import dependencies among large economies – around 5% in 2023 – would also have experienced significant impacts. Without wind and solar energy deployment, fossil-based generation would have needed to rise by around 170 TWh, primarily from imported natural gas. Imports of natural gas would have increased nearly five-fold, pushing electricity import dependence to almost 30%, despite the country’s large hydropower base.\n\n0%\n\n20%\n\n40%\n\n60%\n\n80%\n\n100%\n\n120%\n\n0\n\n50\n\n100\n\n150\n\n200\n\n250\n\nActual\n\nLow RES\n\nActual\n\nLow RES\n\nActual\n\nLow RES\n\nActual\n\nLow RES\n\nActual\n\nLow RES\n\nActual\n\nLow RES\n\nActual\n\nLow RES\n\nActual\n\nLow RES\n\nActual\n\nLow RES\n\nJapan India Korea GermanyTürkiye Italy France UK Brazil\n\nMtoe\n\nNet imports for other use Net imports for power Difference (%)\n\n0%\n\n20%\n\n40%\n\n60%\n\n80%\n\n100%\n\n120%\n\n0\n\n100\n\n200\n\n300\n\n400\n\n500\n\n600\n\n700\n\n800\n\nActual\n\nLow RES\n\nActual\n\nLow RES\n\nChina EU\n\nMtoe","publish_date":null,"title":"and Greece, where the difference could have been as much as 30-50 percentage points. In the Low-RES scenario, the energy","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Renewables2025_1_p81.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Renewables2025_1_p81.pdf"},"request":{"example_id":"chart/Renewables2025_1_p81","product_type":"parse","source_file_path":"Renewables2025_1_p81.pdf"},"started_at":"2026-11-04T19:31:11.894151"} +{"completed_at":"2026-11-04T19:46:10.345360","latency_in_ms":2344,"output":{"example_id":"chart/SIA_State-of-Industry-Report_2024_final_091124_p11","grounded_pages":[],"layout_pages":[],"markdown":"**11** **2024 STATE OF THE U.S. SEMICONDUCTOR INDUSTRY**\n\n# **THE SEMICONDUCTOR WORKFORCE**\n\n### Having a competitive domestic workforce and resilient manufacturing capabilities are critical to America’s lead in semiconductors. In addition, a strong domestic semiconductor industry is essential to the U.S. economy.\n\n### **The semiconductor industry has a considerable economic footprint in the United States, with roughly 338,000 people working in the industry, including roles in chip design, electronic design automation (EDA), semiconductor manufacturing, and equipment manufacturing. Additionally, semiconductors enable over 300 downstream economic sectors accounting for over 26 million U.S. workers.**\n\n**SEMICONDUCTOR WORKFORCE BY FIELD, 2023**\n\n0 50,000 100,000 150,000 200,000\n\n**Machinery Manufacturing** 29,000 **Device Manufacturing** 200,000 **Design** 100,000 **EDA** 9,000\n\n338,000 **Total Jobs**\n\n250,000 300,000 350,000\n\n**Employment**\n\nSource: CES, QCEW, Oxford Economics","pages":[{"line_numbers":[],"markdown":"**11** **2024 STATE OF THE U.S. SEMICONDUCTOR INDUSTRY**\n\n# **THE SEMICONDUCTOR WORKFORCE**\n\n### Having a competitive domestic workforce and resilient manufacturing capabilities are critical to America’s lead in semiconductors. In addition, a strong domestic semiconductor industry is essential to the U.S. economy.\n\n### **The semiconductor industry has a considerable economic footprint in the United States, with roughly 338,000 people working in the industry, including roles in chip design, electronic design automation (EDA), semiconductor manufacturing, and equipment manufacturing. Additionally, semiconductors enable over 300 downstream economic sectors accounting for over 26 million U.S. workers.**\n\n**SEMICONDUCTOR WORKFORCE BY FIELD, 2023**\n\n0 50,000 100,000 150,000 200,000\n\n**Machinery Manufacturing** 29,000 **Device Manufacturing** 200,000 **Design** 100,000 **EDA** 9,000\n\n338,000 **Total Jobs**\n\n250,000 300,000 350,000\n\n**Employment**\n\nSource: CES, QCEW, Oxford Economics","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**11** **2024 STATE OF THE U.S. SEMICONDUCTOR INDUSTRY**\n\n# **THE SEMICONDUCTOR WORKFORCE**\n\n### Having a competitive domestic workforce and resilient manufacturing capabilities are critical to America’s lead in semiconductors. In addition, a strong domestic semiconductor industry is essential to the U.S. economy.\n\n### **The semiconductor industry has a considerable economic footprint in the United States, with roughly 338,000 people working in the industry, including roles in chip design, electronic design automation (EDA), semiconductor manufacturing, and equipment manufacturing. Additionally, semiconductors enable over 300 downstream economic sectors accounting for over 26 million U.S. workers.**\n\n**SEMICONDUCTOR WORKFORCE BY FIELD, 2023**\n\n0 50,000 100,000 150,000 200,000\n\n**Machinery Manufacturing** 29,000 **Device Manufacturing** 200,000 **Design** 100,000 **EDA** 9,000\n\n338,000 **Total Jobs**\n\n250,000 300,000 350,000\n\n**Employment**\n\nSource: CES, QCEW, Oxford Economics","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/SIA_State-of-Industry-Report_2024_final_091124_p11.pdf"]},"response":{"errors":[],"extract_id":"extract_7bef077d603a85a7151fe958a40b2175","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**11** **2024 STATE OF THE U.S. SEMICONDUCTOR INDUSTRY**\n\n# **THE SEMICONDUCTOR WORKFORCE**\n\n### Having a competitive domestic workforce and resilient manufacturing capabilities are critical to America’s lead in semiconductors. In addition, a strong domestic semiconductor industry is essential to the U.S. economy.\n\n### **The semiconductor industry has a considerable economic footprint in the United States, with roughly 338,000 people working in the industry, including roles in chip design, electronic design automation (EDA), semiconductor manufacturing, and equipment manufacturing. Additionally, semiconductors enable over 300 downstream economic sectors accounting for over 26 million U.S. workers.**\n\n**SEMICONDUCTOR WORKFORCE BY FIELD, 2023**\n\n0 50,000 100,000 150,000 200,000\n\n**Machinery Manufacturing** 29,000 **Device Manufacturing** 200,000 **Design** 100,000 **EDA** 9,000\n\n338,000 **Total Jobs**\n\n250,000 300,000 350,000\n\n**Employment**\n\nSource: CES, QCEW, Oxford Economics"],"full_content":"**11** **2024 STATE OF THE U.S. SEMICONDUCTOR INDUSTRY**\n\n# **THE SEMICONDUCTOR WORKFORCE**\n\n### Having a competitive domestic workforce and resilient manufacturing capabilities are critical to America’s lead in semiconductors. In addition, a strong domestic semiconductor industry is essential to the U.S. economy.\n\n### **The semiconductor industry has a considerable economic footprint in the United States, with roughly 338,000 people working in the industry, including roles in chip design, electronic design automation (EDA), semiconductor manufacturing, and equipment manufacturing. Additionally, semiconductors enable over 300 downstream economic sectors accounting for over 26 million U.S. workers.**\n\n**SEMICONDUCTOR WORKFORCE BY FIELD, 2023**\n\n0 50,000 100,000 150,000 200,000\n\n**Machinery Manufacturing** 29,000 **Device Manufacturing** 200,000 **Design** 100,000 **EDA** 9,000\n\n338,000 **Total Jobs**\n\n250,000 300,000 350,000\n\n**Employment**\n\nSource: CES, QCEW, Oxford Economics","publish_date":null,"title":"THE SEMICONDUCTOR WORKFORCE","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/SIA_State-of-Industry-Report_2024_final_091124_p11.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/SIA_State-of-Industry-Report_2024_final_091124_p11.pdf"},"request":{"example_id":"chart/SIA_State-of-Industry-Report_2024_final_091124_p11","product_type":"parse","source_file_path":"SIA_State-of-Industry-Report_2024_final_091124_p11.pdf"},"started_at":"2026-11-04T19:46:08.000811"} +{"completed_at":"2026-11-04T19:46:11.693861","latency_in_ms":2582,"output":{"example_id":"chart/SIA_State-of-Industry-Report_2024_final_091124_p23","grounded_pages":[],"layout_pages":[],"markdown":"**23** **2024 STATE OF THE U.S. SEMICONDUCTOR INDUSTRY**\n\n# **THE GLOBAL SEMICONDUCTOR MARKET**\n\n### Over the past three decades, the semiconductor industry has experienced rapid growth and delivered enormous economic impact.\n\n### **Chip performance and cost improvements made possible the evolution from mainframes to PCs in the 1990s, the Web and online services in the 2000s, and the smartphone revolution in the 2010s. Innovation in AI, electric vehicles, and industrial manufacturing will drive continued growth in the chip market through the next decade. By 2030, AI alone is projected to contribute more than $15 trillion to the global economy. Semiconductors have become essential to our modern world, which is why long-term market demand for semiconductors remains strong.**\n\nThe 2023 market was down through the first half of the year but strong sales to auto, industrial, and AI propelled a rebound in the second half of the year.\n\nFollowing a sales record year of $574.1 billion in 2022, global semiconductor sales in 2023 decreased by 8.2% to $526.9 billion. Sales were slow in the first half of 2023 but grew consistently in the second half of the year, largely due to increased demand for chips in AI, autos, and industrial applications. Sales\n\nduring the first half of 2024 increased by 19.2% compared to sales during the first half of 2023.\n\nEstimates from the World Semiconductor Trade Statistics (WSTS) project worldwide semiconductor industry sales will increase to $611 billion in 2024, a 16% increase compared to 2023. Notwithstanding the short-term cyclical downturn the market faced in 2023, long- term growth prospects for this foundational technology remain promising.\n\n0\n\n100\n\n200\n\n300\n\n400\n\n500\n\n600\n\n700\n\n'01 '02 '03 '04 '05 '06 '07 '08 '09 '10 '11 '12 '13 '14 '15 '16 '17 '18 '19 '20 '21 '22 '23 '24E\n\n\\-8.2% '22/'23\n\n**GLOBAL SEMICONDUCTOR SALES ($B)**","pages":[{"line_numbers":[],"markdown":"**23** **2024 STATE OF THE U.S. SEMICONDUCTOR INDUSTRY**\n\n# **THE GLOBAL SEMICONDUCTOR MARKET**\n\n### Over the past three decades, the semiconductor industry has experienced rapid growth and delivered enormous economic impact.\n\n### **Chip performance and cost improvements made possible the evolution from mainframes to PCs in the 1990s, the Web and online services in the 2000s, and the smartphone revolution in the 2010s. Innovation in AI, electric vehicles, and industrial manufacturing will drive continued growth in the chip market through the next decade. By 2030, AI alone is projected to contribute more than $15 trillion to the global economy. Semiconductors have become essential to our modern world, which is why long-term market demand for semiconductors remains strong.**\n\nThe 2023 market was down through the first half of the year but strong sales to auto, industrial, and AI propelled a rebound in the second half of the year.\n\nFollowing a sales record year of $574.1 billion in 2022, global semiconductor sales in 2023 decreased by 8.2% to $526.9 billion. Sales were slow in the first half of 2023 but grew consistently in the second half of the year, largely due to increased demand for chips in AI, autos, and industrial applications. Sales\n\nduring the first half of 2024 increased by 19.2% compared to sales during the first half of 2023.\n\nEstimates from the World Semiconductor Trade Statistics (WSTS) project worldwide semiconductor industry sales will increase to $611 billion in 2024, a 16% increase compared to 2023. Notwithstanding the short-term cyclical downturn the market faced in 2023, long- term growth prospects for this foundational technology remain promising.\n\n0\n\n100\n\n200\n\n300\n\n400\n\n500\n\n600\n\n700\n\n'01 '02 '03 '04 '05 '06 '07 '08 '09 '10 '11 '12 '13 '14 '15 '16 '17 '18 '19 '20 '21 '22 '23 '24E\n\n\\-8.2% '22/'23\n\n**GLOBAL SEMICONDUCTOR SALES ($B)**","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**23** **2024 STATE OF THE U.S. SEMICONDUCTOR INDUSTRY**\n\n# **THE GLOBAL SEMICONDUCTOR MARKET**\n\n### Over the past three decades, the semiconductor industry has experienced rapid growth and delivered enormous economic impact.\n\n### **Chip performance and cost improvements made possible the evolution from mainframes to PCs in the 1990s, the Web and online services in the 2000s, and the smartphone revolution in the 2010s. Innovation in AI, electric vehicles, and industrial manufacturing will drive continued growth in the chip market through the next decade. By 2030, AI alone is projected to contribute more than $15 trillion to the global economy. Semiconductors have become essential to our modern world, which is why long-term market demand for semiconductors remains strong.**\n\nThe 2023 market was down through the first half of the year but strong sales to auto, industrial, and AI propelled a rebound in the second half of the year.\n\nFollowing a sales record year of $574.1 billion in 2022, global semiconductor sales in 2023 decreased by 8.2% to $526.9 billion. Sales were slow in the first half of 2023 but grew consistently in the second half of the year, largely due to increased demand for chips in AI, autos, and industrial applications. Sales\n\nduring the first half of 2024 increased by 19.2% compared to sales during the first half of 2023.\n\nEstimates from the World Semiconductor Trade Statistics (WSTS) project worldwide semiconductor industry sales will increase to $611 billion in 2024, a 16% increase compared to 2023. Notwithstanding the short-term cyclical downturn the market faced in 2023, long- term growth prospects for this foundational technology remain promising.\n\n0\n\n100\n\n200\n\n300\n\n400\n\n500\n\n600\n\n700\n\n'01 '02 '03 '04 '05 '06 '07 '08 '09 '10 '11 '12 '13 '14 '15 '16 '17 '18 '19 '20 '21 '22 '23 '24E\n\n\\-8.2% '22/'23\n\n**GLOBAL SEMICONDUCTOR SALES ($B)**","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/SIA_State-of-Industry-Report_2024_final_091124_p23.pdf"]},"response":{"errors":[],"extract_id":"extract_399ae8573ede88d27485c2725d74583b","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**23** **2024 STATE OF THE U.S. SEMICONDUCTOR INDUSTRY**\n\n# **THE GLOBAL SEMICONDUCTOR MARKET**\n\n### Over the past three decades, the semiconductor industry has experienced rapid growth and delivered enormous economic impact.\n\n### **Chip performance and cost improvements made possible the evolution from mainframes to PCs in the 1990s, the Web and online services in the 2000s, and the smartphone revolution in the 2010s. Innovation in AI, electric vehicles, and industrial manufacturing will drive continued growth in the chip market through the next decade. By 2030, AI alone is projected to contribute more than $15 trillion to the global economy. Semiconductors have become essential to our modern world, which is why long-term market demand for semiconductors remains strong.**\n\nThe 2023 market was down through the first half of the year but strong sales to auto, industrial, and AI propelled a rebound in the second half of the year.\n\nFollowing a sales record year of $574.1 billion in 2022, global semiconductor sales in 2023 decreased by 8.2% to $526.9 billion. Sales were slow in the first half of 2023 but grew consistently in the second half of the year, largely due to increased demand for chips in AI, autos, and industrial applications. Sales\n\nduring the first half of 2024 increased by 19.2% compared to sales during the first half of 2023.\n\nEstimates from the World Semiconductor Trade Statistics (WSTS) project worldwide semiconductor industry sales will increase to $611 billion in 2024, a 16% increase compared to 2023. Notwithstanding the short-term cyclical downturn the market faced in 2023, long- term growth prospects for this foundational technology remain promising.\n\n0\n\n100\n\n200\n\n300\n\n400\n\n500\n\n600\n\n700\n\n'01 '02 '03 '04 '05 '06 '07 '08 '09 '10 '11 '12 '13 '14 '15 '16 '17 '18 '19 '20 '21 '22 '23 '24E\n\n\\-8.2% '22/'23\n\n**GLOBAL SEMICONDUCTOR SALES ($B)**"],"full_content":"**23** **2024 STATE OF THE U.S. SEMICONDUCTOR INDUSTRY**\n\n# **THE GLOBAL SEMICONDUCTOR MARKET**\n\n### Over the past three decades, the semiconductor industry has experienced rapid growth and delivered enormous economic impact.\n\n### **Chip performance and cost improvements made possible the evolution from mainframes to PCs in the 1990s, the Web and online services in the 2000s, and the smartphone revolution in the 2010s. Innovation in AI, electric vehicles, and industrial manufacturing will drive continued growth in the chip market through the next decade. By 2030, AI alone is projected to contribute more than $15 trillion to the global economy. Semiconductors have become essential to our modern world, which is why long-term market demand for semiconductors remains strong.**\n\nThe 2023 market was down through the first half of the year but strong sales to auto, industrial, and AI propelled a rebound in the second half of the year.\n\nFollowing a sales record year of $574.1 billion in 2022, global semiconductor sales in 2023 decreased by 8.2% to $526.9 billion. Sales were slow in the first half of 2023 but grew consistently in the second half of the year, largely due to increased demand for chips in AI, autos, and industrial applications. Sales\n\nduring the first half of 2024 increased by 19.2% compared to sales during the first half of 2023.\n\nEstimates from the World Semiconductor Trade Statistics (WSTS) project worldwide semiconductor industry sales will increase to $611 billion in 2024, a 16% increase compared to 2023. Notwithstanding the short-term cyclical downturn the market faced in 2023, long- term growth prospects for this foundational technology remain promising.\n\n0\n\n100\n\n200\n\n300\n\n400\n\n500\n\n600\n\n700\n\n'01 '02 '03 '04 '05 '06 '07 '08 '09 '10 '11 '12 '13 '14 '15 '16 '17 '18 '19 '20 '21 '22 '23 '24E\n\n\\-8.2% '22/'23\n\n**GLOBAL SEMICONDUCTOR SALES ($B)**","publish_date":null,"title":"THE GLOBAL SEMICONDUCTOR MARKET","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/SIA_State-of-Industry-Report_2024_final_091124_p23.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/SIA_State-of-Industry-Report_2024_final_091124_p23.pdf"},"request":{"example_id":"chart/SIA_State-of-Industry-Report_2024_final_091124_p23","product_type":"parse","source_file_path":"SIA_State-of-Industry-Report_2024_final_091124_p23.pdf"},"started_at":"2026-11-04T19:46:09.111461"} +{"completed_at":"2026-11-04T19:46:14.266902","latency_in_ms":2570,"output":{"example_id":"chart/SIA_State-of-Industry-Report_2024_final_091124_p28","grounded_pages":[],"layout_pages":[],"markdown":"**28** **2024 STATE OF THE U.S. SEMICONDUCTOR INDUSTRY**\n\n# **U.S. TECHNOLOGY COMPETITIVENESS**\n\n### U.S. semiconductor R&D expenditures are consistently high, reflecting the inherent link between U.S. market-share leadership, research investments, and continued innovation.\n\nIn 2023, overall U.S. semiconductor industry investment in R&D totaled $59.3 billion. The growth in R&D spending in 2023 represented an increase of 0.9% over 2022. R&D expenditures by U.S. semiconductor firms tend to be consistently high, regardless of cycles in annual sales, which reflects the importance to our industry of investing in innovation.\n\n**R&D EXPENDITURE ($B)**\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\n70\n\n'01 '02 '03 '04 '05 '06 '07 '08 '09 '10 '11 '12 '13 '14 '15 '16 '17 '18 '19 '20 '21 '22 '23\n\n**28 2024 STATE OF THE U.S. SEMICONDUCTOR INDUSTRY**","pages":[{"line_numbers":[],"markdown":"**28** **2024 STATE OF THE U.S. SEMICONDUCTOR INDUSTRY**\n\n# **U.S. TECHNOLOGY COMPETITIVENESS**\n\n### U.S. semiconductor R&D expenditures are consistently high, reflecting the inherent link between U.S. market-share leadership, research investments, and continued innovation.\n\nIn 2023, overall U.S. semiconductor industry investment in R&D totaled $59.3 billion. The growth in R&D spending in 2023 represented an increase of 0.9% over 2022. R&D expenditures by U.S. semiconductor firms tend to be consistently high, regardless of cycles in annual sales, which reflects the importance to our industry of investing in innovation.\n\n**R&D EXPENDITURE ($B)**\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\n70\n\n'01 '02 '03 '04 '05 '06 '07 '08 '09 '10 '11 '12 '13 '14 '15 '16 '17 '18 '19 '20 '21 '22 '23\n\n**28 2024 STATE OF THE U.S. SEMICONDUCTOR INDUSTRY**","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**28** **2024 STATE OF THE U.S. SEMICONDUCTOR INDUSTRY**\n\n# **U.S. TECHNOLOGY COMPETITIVENESS**\n\n### U.S. semiconductor R&D expenditures are consistently high, reflecting the inherent link between U.S. market-share leadership, research investments, and continued innovation.\n\nIn 2023, overall U.S. semiconductor industry investment in R&D totaled $59.3 billion. The growth in R&D spending in 2023 represented an increase of 0.9% over 2022. R&D expenditures by U.S. semiconductor firms tend to be consistently high, regardless of cycles in annual sales, which reflects the importance to our industry of investing in innovation.\n\n**R&D EXPENDITURE ($B)**\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\n70\n\n'01 '02 '03 '04 '05 '06 '07 '08 '09 '10 '11 '12 '13 '14 '15 '16 '17 '18 '19 '20 '21 '22 '23\n\n**28 2024 STATE OF THE U.S. SEMICONDUCTOR INDUSTRY**","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/SIA_State-of-Industry-Report_2024_final_091124_p28.pdf"]},"response":{"errors":[],"extract_id":"extract_4fce7d7930fcf26ec4916732e6dea331","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**28** **2024 STATE OF THE U.S. SEMICONDUCTOR INDUSTRY**\n\n# **U.S. TECHNOLOGY COMPETITIVENESS**\n\n### U.S. semiconductor R&D expenditures are consistently high, reflecting the inherent link between U.S. market-share leadership, research investments, and continued innovation.\n\nIn 2023, overall U.S. semiconductor industry investment in R&D totaled $59.3 billion. The growth in R&D spending in 2023 represented an increase of 0.9% over 2022. R&D expenditures by U.S. semiconductor firms tend to be consistently high, regardless of cycles in annual sales, which reflects the importance to our industry of investing in innovation.\n\n**R&D EXPENDITURE ($B)**\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\n70\n\n'01 '02 '03 '04 '05 '06 '07 '08 '09 '10 '11 '12 '13 '14 '15 '16 '17 '18 '19 '20 '21 '22 '23\n\n**28 2024 STATE OF THE U.S. SEMICONDUCTOR INDUSTRY**"],"full_content":"**28** **2024 STATE OF THE U.S. SEMICONDUCTOR INDUSTRY**\n\n# **U.S. TECHNOLOGY COMPETITIVENESS**\n\n### U.S. semiconductor R&D expenditures are consistently high, reflecting the inherent link between U.S. market-share leadership, research investments, and continued innovation.\n\nIn 2023, overall U.S. semiconductor industry investment in R&D totaled $59.3 billion. The growth in R&D spending in 2023 represented an increase of 0.9% over 2022. R&D expenditures by U.S. semiconductor firms tend to be consistently high, regardless of cycles in annual sales, which reflects the importance to our industry of investing in innovation.\n\n**R&D EXPENDITURE ($B)**\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\n70\n\n'01 '02 '03 '04 '05 '06 '07 '08 '09 '10 '11 '12 '13 '14 '15 '16 '17 '18 '19 '20 '21 '22 '23\n\n**28 2024 STATE OF THE U.S. SEMICONDUCTOR INDUSTRY**","publish_date":null,"title":"U.S. TECHNOLOGY COMPETITIVENESS","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/SIA_State-of-Industry-Report_2024_final_091124_p28.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/SIA_State-of-Industry-Report_2024_final_091124_p28.pdf"},"request":{"example_id":"chart/SIA_State-of-Industry-Report_2024_final_091124_p28","product_type":"parse","source_file_path":"SIA_State-of-Industry-Report_2024_final_091124_p28.pdf"},"started_at":"2026-11-04T19:46:11.696072"} +{"completed_at":"2026-11-04T19:46:17.444790","latency_in_ms":3174,"output":{"example_id":"chart/SPI_2025_Benchmark_Report_p120","grounded_pages":[],"layout_pages":[],"markdown":"### **2025 Professional Services Maturity** **™** **Benchmark Client Relationships Pillar**\n\nincreased revenue and profit margins. This is the first year SPI Research has analyzed maturity levels concerning these specific questions, so we can expect even more data in the next year.\n\n### **5-Year Client Relationships Trends**\n\nSPI Research identified three key trends in client relationships, as shown in Figure 25, which are critical performance indicators for the Client Relationships pillar. While it's encouraging that over 30% of the clients for the PSOs in this study were new, there was a decline in annual PS revenue growth in 2024, accompanied by increased discounts. Professional services experienced year-over-year revenue growth exceeding 10% for many years, but there have been challenges following the COVID-19 pandemic. It's essential to monitor these two KPI trends to prevent further deterioration in performance.\n\n**Figure 25: Client Relationships Trends of Note**\n\n**_Source: SPI Research, February 2025_**\n\nTable 117 provides an overview of benchmark surveys conducted over the past five years. It reveals that most Client Relationship metrics reached their peak performance in 2021. During this time, the market began to improve as PSOs adapted their marketing and sales strategies in response to the pandemic, which led to a surge in consulting demand.\n\nHowever, in 2023 and 2024, the market faced challenges due to economic uncertainty, the upcoming U.S. elections, and their potential impact on the country. While the deal pipeline showed improvement in 2024, indicating positive prospects for 2025, the percentage of deals won and the backlog at the start of the quarter declined. This suggests that Professional Services has not fully recovered in the market yet.\n\n**Table 117: Client Relationships Pillar 5-year Trend**\n\n**Key Performance Indicator (KPI) 5-year avg. 2020 2021 2022 2023 2024**\n\nYear-over-year change in PS revenue 8.7% 8.7% 10.6% 10.4% 7.8% 4.6%\n\nCurrent clients - existing services 55.5% 56\\.2% 58.9% 53.7% 52.3% 57.9%\n\nCurrent clients - new services 15.3% 15\\.6% 13.6% 17.0% 16.4% 12.8%\n\nNew logo clients - existing services 19.3% 18\\.1% 18.1% 19.0% 20.6% 21.0%","pages":[{"line_numbers":[],"markdown":"### **2025 Professional Services Maturity** **™** **Benchmark Client Relationships Pillar**\n\nincreased revenue and profit margins. This is the first year SPI Research has analyzed maturity levels concerning these specific questions, so we can expect even more data in the next year.\n\n### **5-Year Client Relationships Trends**\n\nSPI Research identified three key trends in client relationships, as shown in Figure 25, which are critical performance indicators for the Client Relationships pillar. While it's encouraging that over 30% of the clients for the PSOs in this study were new, there was a decline in annual PS revenue growth in 2024, accompanied by increased discounts. Professional services experienced year-over-year revenue growth exceeding 10% for many years, but there have been challenges following the COVID-19 pandemic. It's essential to monitor these two KPI trends to prevent further deterioration in performance.\n\n**Figure 25: Client Relationships Trends of Note**\n\n**_Source: SPI Research, February 2025_**\n\nTable 117 provides an overview of benchmark surveys conducted over the past five years. It reveals that most Client Relationship metrics reached their peak performance in 2021. During this time, the market began to improve as PSOs adapted their marketing and sales strategies in response to the pandemic, which led to a surge in consulting demand.\n\nHowever, in 2023 and 2024, the market faced challenges due to economic uncertainty, the upcoming U.S. elections, and their potential impact on the country. While the deal pipeline showed improvement in 2024, indicating positive prospects for 2025, the percentage of deals won and the backlog at the start of the quarter declined. This suggests that Professional Services has not fully recovered in the market yet.\n\n**Table 117: Client Relationships Pillar 5-year Trend**\n\n**Key Performance Indicator (KPI) 5-year avg. 2020 2021 2022 2023 2024**\n\nYear-over-year change in PS revenue 8.7% 8.7% 10.6% 10.4% 7.8% 4.6%\n\nCurrent clients - existing services 55.5% 56\\.2% 58.9% 53.7% 52.3% 57.9%\n\nCurrent clients - new services 15.3% 15\\.6% 13.6% 17.0% 16.4% 12.8%\n\nNew logo clients - existing services 19.3% 18\\.1% 18.1% 19.0% 20.6% 21.0%","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"### **2025 Professional Services Maturity** **™** **Benchmark Client Relationships Pillar**\n\nincreased revenue and profit margins. This is the first year SPI Research has analyzed maturity levels concerning these specific questions, so we can expect even more data in the next year.\n\n### **5-Year Client Relationships Trends**\n\nSPI Research identified three key trends in client relationships, as shown in Figure 25, which are critical performance indicators for the Client Relationships pillar. While it's encouraging that over 30% of the clients for the PSOs in this study were new, there was a decline in annual PS revenue growth in 2024, accompanied by increased discounts. Professional services experienced year-over-year revenue growth exceeding 10% for many years, but there have been challenges following the COVID-19 pandemic. It's essential to monitor these two KPI trends to prevent further deterioration in performance.\n\n**Figure 25: Client Relationships Trends of Note**\n\n**_Source: SPI Research, February 2025_**\n\nTable 117 provides an overview of benchmark surveys conducted over the past five years. It reveals that most Client Relationship metrics reached their peak performance in 2021. During this time, the market began to improve as PSOs adapted their marketing and sales strategies in response to the pandemic, which led to a surge in consulting demand.\n\nHowever, in 2023 and 2024, the market faced challenges due to economic uncertainty, the upcoming U.S. elections, and their potential impact on the country. While the deal pipeline showed improvement in 2024, indicating positive prospects for 2025, the percentage of deals won and the backlog at the start of the quarter declined. This suggests that Professional Services has not fully recovered in the market yet.\n\n**Table 117: Client Relationships Pillar 5-year Trend**\n\n**Key Performance Indicator (KPI) 5-year avg. 2020 2021 2022 2023 2024**\n\nYear-over-year change in PS revenue 8.7% 8.7% 10.6% 10.4% 7.8% 4.6%\n\nCurrent clients - existing services 55.5% 56\\.2% 58.9% 53.7% 52.3% 57.9%\n\nCurrent clients - new services 15.3% 15\\.6% 13.6% 17.0% 16.4% 12.8%\n\nNew logo clients - existing services 19.3% 18\\.1% 18.1% 19.0% 20.6% 21.0%","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/SPI_2025_Benchmark_Report_p120.pdf"]},"response":{"errors":[],"extract_id":"extract_d2b1d585adb4b97bcd881d0214badbe1","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["### **2025 Professional Services Maturity** **™** **Benchmark Client Relationships Pillar**\n\nincreased revenue and profit margins. This is the first year SPI Research has analyzed maturity levels concerning these specific questions, so we can expect even more data in the next year.\n\n### **5-Year Client Relationships Trends**\n\nSPI Research identified three key trends in client relationships, as shown in Figure 25, which are critical performance indicators for the Client Relationships pillar. While it's encouraging that over 30% of the clients for the PSOs in this study were new, there was a decline in annual PS revenue growth in 2024, accompanied by increased discounts. Professional services experienced year-over-year revenue growth exceeding 10% for many years, but there have been challenges following the COVID-19 pandemic. It's essential to monitor these two KPI trends to prevent further deterioration in performance.\n\n**Figure 25: Client Relationships Trends of Note**\n\n**_Source: SPI Research, February 2025_**\n\nTable 117 provides an overview of benchmark surveys conducted over the past five years. It reveals that most Client Relationship metrics reached their peak performance in 2021. During this time, the market began to improve as PSOs adapted their marketing and sales strategies in response to the pandemic, which led to a surge in consulting demand.\n\nHowever, in 2023 and 2024, the market faced challenges due to economic uncertainty, the upcoming U.S. elections, and their potential impact on the country. While the deal pipeline showed improvement in 2024, indicating positive prospects for 2025, the percentage of deals won and the backlog at the start of the quarter declined. This suggests that Professional Services has not fully recovered in the market yet.\n\n**Table 117: Client Relationships Pillar 5-year Trend**\n\n**Key Performance Indicator (KPI) 5-year avg. 2020 2021 2022 2023 2024**\n\nYear-over-year change in PS revenue 8.7% 8.7% 10.6% 10.4% 7.8% 4.6%\n\nCurrent clients - existing services 55.5% 56\\.2% 58.9% 53.7% 52.3% 57.9%\n\nCurrent clients - new services 15.3% 15\\.6% 13.6% 17.0% 16.4% 12.8%\n\nNew logo clients - existing services 19.3% 18\\.1% 18.1% 19.0% 20.6% 21.0%"],"full_content":"### **2025 Professional Services Maturity** **™** **Benchmark Client Relationships Pillar**\n\nincreased revenue and profit margins. This is the first year SPI Research has analyzed maturity levels concerning these specific questions, so we can expect even more data in the next year.\n\n### **5-Year Client Relationships Trends**\n\nSPI Research identified three key trends in client relationships, as shown in Figure 25, which are critical performance indicators for the Client Relationships pillar. While it's encouraging that over 30% of the clients for the PSOs in this study were new, there was a decline in annual PS revenue growth in 2024, accompanied by increased discounts. Professional services experienced year-over-year revenue growth exceeding 10% for many years, but there have been challenges following the COVID-19 pandemic. It's essential to monitor these two KPI trends to prevent further deterioration in performance.\n\n**Figure 25: Client Relationships Trends of Note**\n\n**_Source: SPI Research, February 2025_**\n\nTable 117 provides an overview of benchmark surveys conducted over the past five years. It reveals that most Client Relationship metrics reached their peak performance in 2021. During this time, the market began to improve as PSOs adapted their marketing and sales strategies in response to the pandemic, which led to a surge in consulting demand.\n\nHowever, in 2023 and 2024, the market faced challenges due to economic uncertainty, the upcoming U.S. elections, and their potential impact on the country. While the deal pipeline showed improvement in 2024, indicating positive prospects for 2025, the percentage of deals won and the backlog at the start of the quarter declined. This suggests that Professional Services has not fully recovered in the market yet.\n\n**Table 117: Client Relationships Pillar 5-year Trend**\n\n**Key Performance Indicator (KPI) 5-year avg. 2020 2021 2022 2023 2024**\n\nYear-over-year change in PS revenue 8.7% 8.7% 10.6% 10.4% 7.8% 4.6%\n\nCurrent clients - existing services 55.5% 56\\.2% 58.9% 53.7% 52.3% 57.9%\n\nCurrent clients - new services 15.3% 15\\.6% 13.6% 17.0% 16.4% 12.8%\n\nNew logo clients - existing services 19.3% 18\\.1% 18.1% 19.0% 20.6% 21.0%","publish_date":null,"title":"5-Year Client Relationships Trends","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/SPI_2025_Benchmark_Report_p120.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/SPI_2025_Benchmark_Report_p120.pdf"},"request":{"example_id":"chart/SPI_2025_Benchmark_Report_p120","product_type":"parse","source_file_path":"SPI_2025_Benchmark_Report_p120.pdf"},"started_at":"2026-11-04T19:46:14.269971"} +{"completed_at":"2026-11-04T19:46:17.452246","latency_in_ms":1355,"output":{"example_id":"chart/SPI_2025_Benchmark_Report_p148","grounded_pages":[],"layout_pages":[],"markdown":"### **2025 Professional Services Maturity** **™** **Benchmark Talent Pillar**\n\n### **TALENT PILLAR**\n\nThe Talent Pillar is the central element in Service Performance Insight’s Professional Services Maturity™ Model. It encompasses all Professional Services Organization (PSO) workforce strategy aspects. This pillar focuses on the people, processes, and systems needed to recruit, attract, train, motivate, and retain a high-quality consulting workforce. Given the changing dynamics of the workforce, it requires more attention than ever.\n\nTalent remains a key challenge and a primary area for improvement in professional services. In 2024, factors such as rising inflation, high interest rates, global conflicts, and uncertainties surrounding the US election have impacted the global economy. Additionally, the skills required due to the emergence of AI in both professional and personal environments have transformed the workplace. A significant portion of the consulting workforce operates virtually, with 68% of consulting hours delivered off-site, up from 58% in 2023. Today’s consulting work relies on a multilingual, multigenerational, multicultural, technically skilled, and project-based workforce. The demand for horizontal applications and technical skills remains high while emphasis is shifting toward business processes and vertical expertise due to technological advancements.\n\nThe key takeaway from the Talent Pillar is that implementing best practices—such as providing clear roles and job descriptions, timely performance reviews, fair compensation, career planning, and skill development—results in significant benefits for employee satisfaction, billable utilization, and on-time project completion. A well-executed talent strategy can lead to lasting advantages, so creating a focused list for any maturity improvement plans is essential.\n\n**Figure 26: Talent Trends of Note**\n\n**_Source: SPI Research, February 2025_**\n\nFigure 26 highlights some of the trends associated with talent. At the same time, the numbers aren't significantly different from 2023. They do show a lengthened recruiting cycle and increased cost per consultant. These two KPIs are correlated, and PS executives must plan for the additional time and expense as talent becomes scarce. This change is especially true considering artificial intelligence, and PSOs will need the best consultants available to drive revenue. SPI Research was also concerned with the","pages":[{"line_numbers":[],"markdown":"### **2025 Professional Services Maturity** **™** **Benchmark Talent Pillar**\n\n### **TALENT PILLAR**\n\nThe Talent Pillar is the central element in Service Performance Insight’s Professional Services Maturity™ Model. It encompasses all Professional Services Organization (PSO) workforce strategy aspects. This pillar focuses on the people, processes, and systems needed to recruit, attract, train, motivate, and retain a high-quality consulting workforce. Given the changing dynamics of the workforce, it requires more attention than ever.\n\nTalent remains a key challenge and a primary area for improvement in professional services. In 2024, factors such as rising inflation, high interest rates, global conflicts, and uncertainties surrounding the US election have impacted the global economy. Additionally, the skills required due to the emergence of AI in both professional and personal environments have transformed the workplace. A significant portion of the consulting workforce operates virtually, with 68% of consulting hours delivered off-site, up from 58% in 2023. Today’s consulting work relies on a multilingual, multigenerational, multicultural, technically skilled, and project-based workforce. The demand for horizontal applications and technical skills remains high while emphasis is shifting toward business processes and vertical expertise due to technological advancements.\n\nThe key takeaway from the Talent Pillar is that implementing best practices—such as providing clear roles and job descriptions, timely performance reviews, fair compensation, career planning, and skill development—results in significant benefits for employee satisfaction, billable utilization, and on-time project completion. A well-executed talent strategy can lead to lasting advantages, so creating a focused list for any maturity improvement plans is essential.\n\n**Figure 26: Talent Trends of Note**\n\n**_Source: SPI Research, February 2025_**\n\nFigure 26 highlights some of the trends associated with talent. At the same time, the numbers aren't significantly different from 2023. They do show a lengthened recruiting cycle and increased cost per consultant. These two KPIs are correlated, and PS executives must plan for the additional time and expense as talent becomes scarce. This change is especially true considering artificial intelligence, and PSOs will need the best consultants available to drive revenue. SPI Research was also concerned with the","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"### **2025 Professional Services Maturity** **™** **Benchmark Talent Pillar**\n\n### **TALENT PILLAR**\n\nThe Talent Pillar is the central element in Service Performance Insight’s Professional Services Maturity™ Model. It encompasses all Professional Services Organization (PSO) workforce strategy aspects. This pillar focuses on the people, processes, and systems needed to recruit, attract, train, motivate, and retain a high-quality consulting workforce. Given the changing dynamics of the workforce, it requires more attention than ever.\n\nTalent remains a key challenge and a primary area for improvement in professional services. In 2024, factors such as rising inflation, high interest rates, global conflicts, and uncertainties surrounding the US election have impacted the global economy. Additionally, the skills required due to the emergence of AI in both professional and personal environments have transformed the workplace. A significant portion of the consulting workforce operates virtually, with 68% of consulting hours delivered off-site, up from 58% in 2023. Today’s consulting work relies on a multilingual, multigenerational, multicultural, technically skilled, and project-based workforce. The demand for horizontal applications and technical skills remains high while emphasis is shifting toward business processes and vertical expertise due to technological advancements.\n\nThe key takeaway from the Talent Pillar is that implementing best practices—such as providing clear roles and job descriptions, timely performance reviews, fair compensation, career planning, and skill development—results in significant benefits for employee satisfaction, billable utilization, and on-time project completion. A well-executed talent strategy can lead to lasting advantages, so creating a focused list for any maturity improvement plans is essential.\n\n**Figure 26: Talent Trends of Note**\n\n**_Source: SPI Research, February 2025_**\n\nFigure 26 highlights some of the trends associated with talent. At the same time, the numbers aren't significantly different from 2023. They do show a lengthened recruiting cycle and increased cost per consultant. These two KPIs are correlated, and PS executives must plan for the additional time and expense as talent becomes scarce. This change is especially true considering artificial intelligence, and PSOs will need the best consultants available to drive revenue. SPI Research was also concerned with the","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/SPI_2025_Benchmark_Report_p148.pdf"]},"response":{"errors":[],"extract_id":"extract_2eec4521a27831d9262828aff09622f8","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["### **2025 Professional Services Maturity** **™** **Benchmark Talent Pillar**\n\n### **TALENT PILLAR**\n\nThe Talent Pillar is the central element in Service Performance Insight’s Professional Services Maturity™ Model. It encompasses all Professional Services Organization (PSO) workforce strategy aspects. This pillar focuses on the people, processes, and systems needed to recruit, attract, train, motivate, and retain a high-quality consulting workforce. Given the changing dynamics of the workforce, it requires more attention than ever.\n\nTalent remains a key challenge and a primary area for improvement in professional services. In 2024, factors such as rising inflation, high interest rates, global conflicts, and uncertainties surrounding the US election have impacted the global economy. Additionally, the skills required due to the emergence of AI in both professional and personal environments have transformed the workplace. A significant portion of the consulting workforce operates virtually, with 68% of consulting hours delivered off-site, up from 58% in 2023. Today’s consulting work relies on a multilingual, multigenerational, multicultural, technically skilled, and project-based workforce. The demand for horizontal applications and technical skills remains high while emphasis is shifting toward business processes and vertical expertise due to technological advancements.\n\nThe key takeaway from the Talent Pillar is that implementing best practices—such as providing clear roles and job descriptions, timely performance reviews, fair compensation, career planning, and skill development—results in significant benefits for employee satisfaction, billable utilization, and on-time project completion. A well-executed talent strategy can lead to lasting advantages, so creating a focused list for any maturity improvement plans is essential.\n\n**Figure 26: Talent Trends of Note**\n\n**_Source: SPI Research, February 2025_**\n\nFigure 26 highlights some of the trends associated with talent. At the same time, the numbers aren't significantly different from 2023. They do show a lengthened recruiting cycle and increased cost per consultant. These two KPIs are correlated, and PS executives must plan for the additional time and expense as talent becomes scarce. This change is especially true considering artificial intelligence, and PSOs will need the best consultants available to drive revenue. SPI Research was also concerned with the"],"full_content":"### **2025 Professional Services Maturity** **™** **Benchmark Talent Pillar**\n\n### **TALENT PILLAR**\n\nThe Talent Pillar is the central element in Service Performance Insight’s Professional Services Maturity™ Model. It encompasses all Professional Services Organization (PSO) workforce strategy aspects. This pillar focuses on the people, processes, and systems needed to recruit, attract, train, motivate, and retain a high-quality consulting workforce. Given the changing dynamics of the workforce, it requires more attention than ever.\n\nTalent remains a key challenge and a primary area for improvement in professional services. In 2024, factors such as rising inflation, high interest rates, global conflicts, and uncertainties surrounding the US election have impacted the global economy. Additionally, the skills required due to the emergence of AI in both professional and personal environments have transformed the workplace. A significant portion of the consulting workforce operates virtually, with 68% of consulting hours delivered off-site, up from 58% in 2023. Today’s consulting work relies on a multilingual, multigenerational, multicultural, technically skilled, and project-based workforce. The demand for horizontal applications and technical skills remains high while emphasis is shifting toward business processes and vertical expertise due to technological advancements.\n\nThe key takeaway from the Talent Pillar is that implementing best practices—such as providing clear roles and job descriptions, timely performance reviews, fair compensation, career planning, and skill development—results in significant benefits for employee satisfaction, billable utilization, and on-time project completion. A well-executed talent strategy can lead to lasting advantages, so creating a focused list for any maturity improvement plans is essential.\n\n**Figure 26: Talent Trends of Note**\n\n**_Source: SPI Research, February 2025_**\n\nFigure 26 highlights some of the trends associated with talent. At the same time, the numbers aren't significantly different from 2023. They do show a lengthened recruiting cycle and increased cost per consultant. These two KPIs are correlated, and PS executives must plan for the additional time and expense as talent becomes scarce. This change is especially true considering artificial intelligence, and PSOs will need the best consultants available to drive revenue. SPI Research was also concerned with the","publish_date":null,"title":"TALENT PILLAR","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/SPI_2025_Benchmark_Report_p148.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/SPI_2025_Benchmark_Report_p148.pdf"},"request":{"example_id":"chart/SPI_2025_Benchmark_Report_p148","product_type":"parse","source_file_path":"SPI_2025_Benchmark_Report_p148.pdf"},"started_at":"2026-11-04T19:46:16.096278"} +{"completed_at":"2026-11-04T19:46:19.943048","latency_in_ms":2496,"output":{"example_id":"chart/SPI_2025_Benchmark_Report_p171","grounded_pages":[],"layout_pages":[],"markdown":"### **2025 Professional Services Maturity** **™** **Benchmark Service Execution Pillar**\n\n### **SERVICE EXECUTION PILLAR**\n\nThe Service Execution Pillar assesses the quality, efficiency, and repeatability of service delivery. It emphasizes the core activities involved in planning, scheduling, and executing service engagements. Regardless of how mature other areas of the PSO may be, success hinges on the ability to deliver services effectively and profitably, with a strong focus on quality, timeliness, and customer value.\n\nThe Service Execution pillar is essential for generating revenue in professional services. To maximize project margins, work must be scoped, priced, sold, delivered, and invoiced. Critical to this process is the alignment of sales, human resources, service, and finance functions. Capturing all project-related information—such as time, expenses, project details, and knowledge—is vital for proper invoicing and for enhancing future service delivery.\n\nSPI Research has identified key trends within the Service Execution pillar, as shown in Figure 29, which are crucial for successful project delivery. As previously mentioned, billable utilization is one of the most critical Key Performance Indicators (KPIs) in professional services. Unfortunately, recent data shows a decline in billable utilization, on-time project delivery, and project margins. These downward trends could adversely affect the PSO’s growth potential in 2025.\n\n**Figure 29: Service Execution Trends of Note**\n\n**_Source: SPI Research, February 2025_**\n\nIn an increasingly competitive consulting landscape, operational excellence often determines success. PSOs must implement visibility and management controls to ensure effective resource and project management. When done correctly, gross project margins exceeding 60% are achievable. Conversely, poor execution can lead to project margins in the single digits or even negative figures.\n\nFor decades, SPI has analyzed various aspects of service delivery, and this area has seen significant improvements as firms recognize the importance of resource specialization and comprehensive visibility in project delivery. PSOs have integrated automation, virtualization, analytics, and cohesive business applications into their service delivery operations. Advances in technology have played a significant role in","pages":[{"line_numbers":[],"markdown":"### **2025 Professional Services Maturity** **™** **Benchmark Service Execution Pillar**\n\n### **SERVICE EXECUTION PILLAR**\n\nThe Service Execution Pillar assesses the quality, efficiency, and repeatability of service delivery. It emphasizes the core activities involved in planning, scheduling, and executing service engagements. Regardless of how mature other areas of the PSO may be, success hinges on the ability to deliver services effectively and profitably, with a strong focus on quality, timeliness, and customer value.\n\nThe Service Execution pillar is essential for generating revenue in professional services. To maximize project margins, work must be scoped, priced, sold, delivered, and invoiced. Critical to this process is the alignment of sales, human resources, service, and finance functions. Capturing all project-related information—such as time, expenses, project details, and knowledge—is vital for proper invoicing and for enhancing future service delivery.\n\nSPI Research has identified key trends within the Service Execution pillar, as shown in Figure 29, which are crucial for successful project delivery. As previously mentioned, billable utilization is one of the most critical Key Performance Indicators (KPIs) in professional services. Unfortunately, recent data shows a decline in billable utilization, on-time project delivery, and project margins. These downward trends could adversely affect the PSO’s growth potential in 2025.\n\n**Figure 29: Service Execution Trends of Note**\n\n**_Source: SPI Research, February 2025_**\n\nIn an increasingly competitive consulting landscape, operational excellence often determines success. PSOs must implement visibility and management controls to ensure effective resource and project management. When done correctly, gross project margins exceeding 60% are achievable. Conversely, poor execution can lead to project margins in the single digits or even negative figures.\n\nFor decades, SPI has analyzed various aspects of service delivery, and this area has seen significant improvements as firms recognize the importance of resource specialization and comprehensive visibility in project delivery. PSOs have integrated automation, virtualization, analytics, and cohesive business applications into their service delivery operations. Advances in technology have played a significant role in","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"### **2025 Professional Services Maturity** **™** **Benchmark Service Execution Pillar**\n\n### **SERVICE EXECUTION PILLAR**\n\nThe Service Execution Pillar assesses the quality, efficiency, and repeatability of service delivery. It emphasizes the core activities involved in planning, scheduling, and executing service engagements. Regardless of how mature other areas of the PSO may be, success hinges on the ability to deliver services effectively and profitably, with a strong focus on quality, timeliness, and customer value.\n\nThe Service Execution pillar is essential for generating revenue in professional services. To maximize project margins, work must be scoped, priced, sold, delivered, and invoiced. Critical to this process is the alignment of sales, human resources, service, and finance functions. Capturing all project-related information—such as time, expenses, project details, and knowledge—is vital for proper invoicing and for enhancing future service delivery.\n\nSPI Research has identified key trends within the Service Execution pillar, as shown in Figure 29, which are crucial for successful project delivery. As previously mentioned, billable utilization is one of the most critical Key Performance Indicators (KPIs) in professional services. Unfortunately, recent data shows a decline in billable utilization, on-time project delivery, and project margins. These downward trends could adversely affect the PSO’s growth potential in 2025.\n\n**Figure 29: Service Execution Trends of Note**\n\n**_Source: SPI Research, February 2025_**\n\nIn an increasingly competitive consulting landscape, operational excellence often determines success. PSOs must implement visibility and management controls to ensure effective resource and project management. When done correctly, gross project margins exceeding 60% are achievable. Conversely, poor execution can lead to project margins in the single digits or even negative figures.\n\nFor decades, SPI has analyzed various aspects of service delivery, and this area has seen significant improvements as firms recognize the importance of resource specialization and comprehensive visibility in project delivery. PSOs have integrated automation, virtualization, analytics, and cohesive business applications into their service delivery operations. Advances in technology have played a significant role in","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/SPI_2025_Benchmark_Report_p171.pdf"]},"response":{"errors":[],"extract_id":"extract_8cb7fa9e9de326a53750ad4865b7b2cc","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["### **2025 Professional Services Maturity** **™** **Benchmark Service Execution Pillar**\n\n### **SERVICE EXECUTION PILLAR**\n\nThe Service Execution Pillar assesses the quality, efficiency, and repeatability of service delivery. It emphasizes the core activities involved in planning, scheduling, and executing service engagements. Regardless of how mature other areas of the PSO may be, success hinges on the ability to deliver services effectively and profitably, with a strong focus on quality, timeliness, and customer value.\n\nThe Service Execution pillar is essential for generating revenue in professional services. To maximize project margins, work must be scoped, priced, sold, delivered, and invoiced. Critical to this process is the alignment of sales, human resources, service, and finance functions. Capturing all project-related information—such as time, expenses, project details, and knowledge—is vital for proper invoicing and for enhancing future service delivery.\n\nSPI Research has identified key trends within the Service Execution pillar, as shown in Figure 29, which are crucial for successful project delivery. As previously mentioned, billable utilization is one of the most critical Key Performance Indicators (KPIs) in professional services. Unfortunately, recent data shows a decline in billable utilization, on-time project delivery, and project margins. These downward trends could adversely affect the PSO’s growth potential in 2025.\n\n**Figure 29: Service Execution Trends of Note**\n\n**_Source: SPI Research, February 2025_**\n\nIn an increasingly competitive consulting landscape, operational excellence often determines success. PSOs must implement visibility and management controls to ensure effective resource and project management. When done correctly, gross project margins exceeding 60% are achievable. Conversely, poor execution can lead to project margins in the single digits or even negative figures.\n\nFor decades, SPI has analyzed various aspects of service delivery, and this area has seen significant improvements as firms recognize the importance of resource specialization and comprehensive visibility in project delivery. PSOs have integrated automation, virtualization, analytics, and cohesive business applications into their service delivery operations. Advances in technology have played a significant role in"],"full_content":"### **2025 Professional Services Maturity** **™** **Benchmark Service Execution Pillar**\n\n### **SERVICE EXECUTION PILLAR**\n\nThe Service Execution Pillar assesses the quality, efficiency, and repeatability of service delivery. It emphasizes the core activities involved in planning, scheduling, and executing service engagements. Regardless of how mature other areas of the PSO may be, success hinges on the ability to deliver services effectively and profitably, with a strong focus on quality, timeliness, and customer value.\n\nThe Service Execution pillar is essential for generating revenue in professional services. To maximize project margins, work must be scoped, priced, sold, delivered, and invoiced. Critical to this process is the alignment of sales, human resources, service, and finance functions. Capturing all project-related information—such as time, expenses, project details, and knowledge—is vital for proper invoicing and for enhancing future service delivery.\n\nSPI Research has identified key trends within the Service Execution pillar, as shown in Figure 29, which are crucial for successful project delivery. As previously mentioned, billable utilization is one of the most critical Key Performance Indicators (KPIs) in professional services. Unfortunately, recent data shows a decline in billable utilization, on-time project delivery, and project margins. These downward trends could adversely affect the PSO’s growth potential in 2025.\n\n**Figure 29: Service Execution Trends of Note**\n\n**_Source: SPI Research, February 2025_**\n\nIn an increasingly competitive consulting landscape, operational excellence often determines success. PSOs must implement visibility and management controls to ensure effective resource and project management. When done correctly, gross project margins exceeding 60% are achievable. Conversely, poor execution can lead to project margins in the single digits or even negative figures.\n\nFor decades, SPI has analyzed various aspects of service delivery, and this area has seen significant improvements as firms recognize the importance of resource specialization and comprehensive visibility in project delivery. PSOs have integrated automation, virtualization, analytics, and cohesive business applications into their service delivery operations. Advances in technology have played a significant role in","publish_date":null,"title":"SERVICE EXECUTION PILLAR","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/SPI_2025_Benchmark_Report_p171.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/SPI_2025_Benchmark_Report_p171.pdf"},"request":{"example_id":"chart/SPI_2025_Benchmark_Report_p171","product_type":"parse","source_file_path":"SPI_2025_Benchmark_Report_p171.pdf"},"started_at":"2026-11-04T19:46:17.446537"} +{"completed_at":"2026-11-04T19:46:20.586892","latency_in_ms":1563,"output":{"example_id":"chart/SPI_2025_Benchmark_Report_p198","grounded_pages":[],"layout_pages":[],"markdown":"### **2025 Professional Services Maturity** **™** **Benchmark Finance & Operations Pillar**\n\n### **FINANCE AND OPERATIONS PILLAR**\n\nThe Finance and Operations pillar encompasses the responsibilities typical of the CFO in prominent PSOs. This aspect is integral to the role of the Chief Service Executive across all PS organizations, regardless of their size. In this service performance pillar, SPI Research evaluated 38 key performance indicators related to revenue, margin achievement, and operating expenses. The analysis includes detailed profit and loss statements and expense ratios categorized by organization size, geography, and industry.\n\nSPI Research has noted operational trends in Figure 31, indicating a deterioration of financial and operational performance. The ability to meet revenue targets (margins) also declined in 2024 as problems related to sales and service delivery became more pronounced. Achieving targets and maintaining high revenue levels per consultant is crucial, and these metrics are heavily influenced by operational success. The data emphasizes the necessity for PSOs to reassess their performance in sales and delivery and make necessary adjustments to enhance both areas in 2025.\n\n**Figure 31: Finance & Operations Trends of Note**\n\n**_Source: SPI Research, February 2025_**\n\n### **Finance and Operations Maturity**\n\nFigure 32 illustrates the significance of reaching higher levels of performance. The gap between Level 1 and Level 5 performance is substantial. Profitability at Level 5 is notably higher, which should capture the attention of executives. However, since numerous factors influence profitability, they must evaluate improvements across the organization. Level 5 organizations operate at a much higher financial standard than their peers, with much of their success attributed to structured service delivery and enhanced visibility throughout the organization. Rapid changes in the industry mean that PSOs capable of efficiently preparing for and managing these changes will thrive in the long term.","pages":[{"line_numbers":[],"markdown":"### **2025 Professional Services Maturity** **™** **Benchmark Finance & Operations Pillar**\n\n### **FINANCE AND OPERATIONS PILLAR**\n\nThe Finance and Operations pillar encompasses the responsibilities typical of the CFO in prominent PSOs. This aspect is integral to the role of the Chief Service Executive across all PS organizations, regardless of their size. In this service performance pillar, SPI Research evaluated 38 key performance indicators related to revenue, margin achievement, and operating expenses. The analysis includes detailed profit and loss statements and expense ratios categorized by organization size, geography, and industry.\n\nSPI Research has noted operational trends in Figure 31, indicating a deterioration of financial and operational performance. The ability to meet revenue targets (margins) also declined in 2024 as problems related to sales and service delivery became more pronounced. Achieving targets and maintaining high revenue levels per consultant is crucial, and these metrics are heavily influenced by operational success. The data emphasizes the necessity for PSOs to reassess their performance in sales and delivery and make necessary adjustments to enhance both areas in 2025.\n\n**Figure 31: Finance & Operations Trends of Note**\n\n**_Source: SPI Research, February 2025_**\n\n### **Finance and Operations Maturity**\n\nFigure 32 illustrates the significance of reaching higher levels of performance. The gap between Level 1 and Level 5 performance is substantial. Profitability at Level 5 is notably higher, which should capture the attention of executives. However, since numerous factors influence profitability, they must evaluate improvements across the organization. Level 5 organizations operate at a much higher financial standard than their peers, with much of their success attributed to structured service delivery and enhanced visibility throughout the organization. Rapid changes in the industry mean that PSOs capable of efficiently preparing for and managing these changes will thrive in the long term.","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"### **2025 Professional Services Maturity** **™** **Benchmark Finance & Operations Pillar**\n\n### **FINANCE AND OPERATIONS PILLAR**\n\nThe Finance and Operations pillar encompasses the responsibilities typical of the CFO in prominent PSOs. This aspect is integral to the role of the Chief Service Executive across all PS organizations, regardless of their size. In this service performance pillar, SPI Research evaluated 38 key performance indicators related to revenue, margin achievement, and operating expenses. The analysis includes detailed profit and loss statements and expense ratios categorized by organization size, geography, and industry.\n\nSPI Research has noted operational trends in Figure 31, indicating a deterioration of financial and operational performance. The ability to meet revenue targets (margins) also declined in 2024 as problems related to sales and service delivery became more pronounced. Achieving targets and maintaining high revenue levels per consultant is crucial, and these metrics are heavily influenced by operational success. The data emphasizes the necessity for PSOs to reassess their performance in sales and delivery and make necessary adjustments to enhance both areas in 2025.\n\n**Figure 31: Finance & Operations Trends of Note**\n\n**_Source: SPI Research, February 2025_**\n\n### **Finance and Operations Maturity**\n\nFigure 32 illustrates the significance of reaching higher levels of performance. The gap between Level 1 and Level 5 performance is substantial. Profitability at Level 5 is notably higher, which should capture the attention of executives. However, since numerous factors influence profitability, they must evaluate improvements across the organization. Level 5 organizations operate at a much higher financial standard than their peers, with much of their success attributed to structured service delivery and enhanced visibility throughout the organization. Rapid changes in the industry mean that PSOs capable of efficiently preparing for and managing these changes will thrive in the long term.","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/SPI_2025_Benchmark_Report_p198.pdf"]},"response":{"errors":[],"extract_id":"extract_d51020d4d675503cea4e2b954d228303","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["### **2025 Professional Services Maturity** **™** **Benchmark Finance & Operations Pillar**\n\n### **FINANCE AND OPERATIONS PILLAR**\n\nThe Finance and Operations pillar encompasses the responsibilities typical of the CFO in prominent PSOs. This aspect is integral to the role of the Chief Service Executive across all PS organizations, regardless of their size. In this service performance pillar, SPI Research evaluated 38 key performance indicators related to revenue, margin achievement, and operating expenses. The analysis includes detailed profit and loss statements and expense ratios categorized by organization size, geography, and industry.\n\nSPI Research has noted operational trends in Figure 31, indicating a deterioration of financial and operational performance. The ability to meet revenue targets (margins) also declined in 2024 as problems related to sales and service delivery became more pronounced. Achieving targets and maintaining high revenue levels per consultant is crucial, and these metrics are heavily influenced by operational success. The data emphasizes the necessity for PSOs to reassess their performance in sales and delivery and make necessary adjustments to enhance both areas in 2025.\n\n**Figure 31: Finance & Operations Trends of Note**\n\n**_Source: SPI Research, February 2025_**\n\n### **Finance and Operations Maturity**\n\nFigure 32 illustrates the significance of reaching higher levels of performance. The gap between Level 1 and Level 5 performance is substantial. Profitability at Level 5 is notably higher, which should capture the attention of executives. However, since numerous factors influence profitability, they must evaluate improvements across the organization. Level 5 organizations operate at a much higher financial standard than their peers, with much of their success attributed to structured service delivery and enhanced visibility throughout the organization. Rapid changes in the industry mean that PSOs capable of efficiently preparing for and managing these changes will thrive in the long term."],"full_content":"### **2025 Professional Services Maturity** **™** **Benchmark Finance & Operations Pillar**\n\n### **FINANCE AND OPERATIONS PILLAR**\n\nThe Finance and Operations pillar encompasses the responsibilities typical of the CFO in prominent PSOs. This aspect is integral to the role of the Chief Service Executive across all PS organizations, regardless of their size. In this service performance pillar, SPI Research evaluated 38 key performance indicators related to revenue, margin achievement, and operating expenses. The analysis includes detailed profit and loss statements and expense ratios categorized by organization size, geography, and industry.\n\nSPI Research has noted operational trends in Figure 31, indicating a deterioration of financial and operational performance. The ability to meet revenue targets (margins) also declined in 2024 as problems related to sales and service delivery became more pronounced. Achieving targets and maintaining high revenue levels per consultant is crucial, and these metrics are heavily influenced by operational success. The data emphasizes the necessity for PSOs to reassess their performance in sales and delivery and make necessary adjustments to enhance both areas in 2025.\n\n**Figure 31: Finance & Operations Trends of Note**\n\n**_Source: SPI Research, February 2025_**\n\n### **Finance and Operations Maturity**\n\nFigure 32 illustrates the significance of reaching higher levels of performance. The gap between Level 1 and Level 5 performance is substantial. Profitability at Level 5 is notably higher, which should capture the attention of executives. However, since numerous factors influence profitability, they must evaluate improvements across the organization. Level 5 organizations operate at a much higher financial standard than their peers, with much of their success attributed to structured service delivery and enhanced visibility throughout the organization. Rapid changes in the industry mean that PSOs capable of efficiently preparing for and managing these changes will thrive in the long term.","publish_date":null,"title":"FINANCE AND OPERATIONS PILLAR Finance and Operations Maturity","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/SPI_2025_Benchmark_Report_p198.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/SPI_2025_Benchmark_Report_p198.pdf"},"request":{"example_id":"chart/SPI_2025_Benchmark_Report_p198","product_type":"parse","source_file_path":"SPI_2025_Benchmark_Report_p198.pdf"},"started_at":"2026-11-04T19:46:19.023334"} +{"completed_at":"2026-11-04T19:46:22.452407","latency_in_ms":1227,"output":{"example_id":"chart/SPI_2025_Benchmark_Report_p78","grounded_pages":[],"layout_pages":[],"markdown":"### **2025 Professional Services Maturity** **™** **Benchmark PS Business Applications**\n\nrevenue per project and consultant, and a higher profit figure than those organizations that did not use project-based ERP. This trend will be necessary as we advance, as the project-based ERP vendors are offering so much more than just financials.\n\n### **PS Solution Adoption**\n\nIn this year’s survey, commercial adoption decreased in all the primary business applications except for Planning / Business Intelligence (Table 64). SPI’s clients emphasized the importance of business and capacity planning tools to help them navigate a challenging year. HCM usage went up in 2024 and continues an upward trend in use as more and more firms realize the benefits of integrating payroll with recruiting and skill building. The integration of HCM and PSA gives employers and employees visibility to upcoming projects and the skills they need to be part of them. The abundance of high quality, affordable cloud-based solutions has enabled greater numbers of PSOs to adopt commercial business applications, yet a surprisingly large number of firms still rely on antiquated homegrown applications and spreadsheets. Excel remains the most-used business application.\n\n**Figure 19: Commercial Solution Adoption**\n\n**_Source: SPI Research, February 2025_**\n\n**Table 64: Commercial Solution Adoption**\n\n**Solution 2022 2023 2024**\n\nPlanning (PL) / Business Intelligence (BI) 53.1% 54.1% 47.9%\n\nClient Relationship Management (CRM) 85.1% 79.6% 83.9%\n\nHuman Capital Management (HCM) 70.5% 67.8% 69.0%\n\nProfessional Services Automation (PSA) 78.4% 74.3% 71.3%\n\nCorporate Financial Management (CFM) 95\\.5% 94.7% 91.5%\n\n**_Source: SPI Research, February 2025_**","pages":[{"line_numbers":[],"markdown":"### **2025 Professional Services Maturity** **™** **Benchmark PS Business Applications**\n\nrevenue per project and consultant, and a higher profit figure than those organizations that did not use project-based ERP. This trend will be necessary as we advance, as the project-based ERP vendors are offering so much more than just financials.\n\n### **PS Solution Adoption**\n\nIn this year’s survey, commercial adoption decreased in all the primary business applications except for Planning / Business Intelligence (Table 64). SPI’s clients emphasized the importance of business and capacity planning tools to help them navigate a challenging year. HCM usage went up in 2024 and continues an upward trend in use as more and more firms realize the benefits of integrating payroll with recruiting and skill building. The integration of HCM and PSA gives employers and employees visibility to upcoming projects and the skills they need to be part of them. The abundance of high quality, affordable cloud-based solutions has enabled greater numbers of PSOs to adopt commercial business applications, yet a surprisingly large number of firms still rely on antiquated homegrown applications and spreadsheets. Excel remains the most-used business application.\n\n**Figure 19: Commercial Solution Adoption**\n\n**_Source: SPI Research, February 2025_**\n\n**Table 64: Commercial Solution Adoption**\n\n**Solution 2022 2023 2024**\n\nPlanning (PL) / Business Intelligence (BI) 53.1% 54.1% 47.9%\n\nClient Relationship Management (CRM) 85.1% 79.6% 83.9%\n\nHuman Capital Management (HCM) 70.5% 67.8% 69.0%\n\nProfessional Services Automation (PSA) 78.4% 74.3% 71.3%\n\nCorporate Financial Management (CFM) 95\\.5% 94.7% 91.5%\n\n**_Source: SPI Research, February 2025_**","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"### **2025 Professional Services Maturity** **™** **Benchmark PS Business Applications**\n\nrevenue per project and consultant, and a higher profit figure than those organizations that did not use project-based ERP. This trend will be necessary as we advance, as the project-based ERP vendors are offering so much more than just financials.\n\n### **PS Solution Adoption**\n\nIn this year’s survey, commercial adoption decreased in all the primary business applications except for Planning / Business Intelligence (Table 64). SPI’s clients emphasized the importance of business and capacity planning tools to help them navigate a challenging year. HCM usage went up in 2024 and continues an upward trend in use as more and more firms realize the benefits of integrating payroll with recruiting and skill building. The integration of HCM and PSA gives employers and employees visibility to upcoming projects and the skills they need to be part of them. The abundance of high quality, affordable cloud-based solutions has enabled greater numbers of PSOs to adopt commercial business applications, yet a surprisingly large number of firms still rely on antiquated homegrown applications and spreadsheets. Excel remains the most-used business application.\n\n**Figure 19: Commercial Solution Adoption**\n\n**_Source: SPI Research, February 2025_**\n\n**Table 64: Commercial Solution Adoption**\n\n**Solution 2022 2023 2024**\n\nPlanning (PL) / Business Intelligence (BI) 53.1% 54.1% 47.9%\n\nClient Relationship Management (CRM) 85.1% 79.6% 83.9%\n\nHuman Capital Management (HCM) 70.5% 67.8% 69.0%\n\nProfessional Services Automation (PSA) 78.4% 74.3% 71.3%\n\nCorporate Financial Management (CFM) 95\\.5% 94.7% 91.5%\n\n**_Source: SPI Research, February 2025_**","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/SPI_2025_Benchmark_Report_p78.pdf"]},"response":{"errors":[],"extract_id":"extract_bb4e0140e80e1bf64d28b658cfe78741","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["### **2025 Professional Services Maturity** **™** **Benchmark PS Business Applications**\n\nrevenue per project and consultant, and a higher profit figure than those organizations that did not use project-based ERP. This trend will be necessary as we advance, as the project-based ERP vendors are offering so much more than just financials.\n\n### **PS Solution Adoption**\n\nIn this year’s survey, commercial adoption decreased in all the primary business applications except for Planning / Business Intelligence (Table 64). SPI’s clients emphasized the importance of business and capacity planning tools to help them navigate a challenging year. HCM usage went up in 2024 and continues an upward trend in use as more and more firms realize the benefits of integrating payroll with recruiting and skill building. The integration of HCM and PSA gives employers and employees visibility to upcoming projects and the skills they need to be part of them. The abundance of high quality, affordable cloud-based solutions has enabled greater numbers of PSOs to adopt commercial business applications, yet a surprisingly large number of firms still rely on antiquated homegrown applications and spreadsheets. Excel remains the most-used business application.\n\n**Figure 19: Commercial Solution Adoption**\n\n**_Source: SPI Research, February 2025_**\n\n**Table 64: Commercial Solution Adoption**\n\n**Solution 2022 2023 2024**\n\nPlanning (PL) / Business Intelligence (BI) 53.1% 54.1% 47.9%\n\nClient Relationship Management (CRM) 85.1% 79.6% 83.9%\n\nHuman Capital Management (HCM) 70.5% 67.8% 69.0%\n\nProfessional Services Automation (PSA) 78.4% 74.3% 71.3%\n\nCorporate Financial Management (CFM) 95\\.5% 94.7% 91.5%\n\n**_Source: SPI Research, February 2025_**"],"full_content":"### **2025 Professional Services Maturity** **™** **Benchmark PS Business Applications**\n\nrevenue per project and consultant, and a higher profit figure than those organizations that did not use project-based ERP. This trend will be necessary as we advance, as the project-based ERP vendors are offering so much more than just financials.\n\n### **PS Solution Adoption**\n\nIn this year’s survey, commercial adoption decreased in all the primary business applications except for Planning / Business Intelligence (Table 64). SPI’s clients emphasized the importance of business and capacity planning tools to help them navigate a challenging year. HCM usage went up in 2024 and continues an upward trend in use as more and more firms realize the benefits of integrating payroll with recruiting and skill building. The integration of HCM and PSA gives employers and employees visibility to upcoming projects and the skills they need to be part of them. The abundance of high quality, affordable cloud-based solutions has enabled greater numbers of PSOs to adopt commercial business applications, yet a surprisingly large number of firms still rely on antiquated homegrown applications and spreadsheets. Excel remains the most-used business application.\n\n**Figure 19: Commercial Solution Adoption**\n\n**_Source: SPI Research, February 2025_**\n\n**Table 64: Commercial Solution Adoption**\n\n**Solution 2022 2023 2024**\n\nPlanning (PL) / Business Intelligence (BI) 53.1% 54.1% 47.9%\n\nClient Relationship Management (CRM) 85.1% 79.6% 83.9%\n\nHuman Capital Management (HCM) 70.5% 67.8% 69.0%\n\nProfessional Services Automation (PSA) 78.4% 74.3% 71.3%\n\nCorporate Financial Management (CFM) 95\\.5% 94.7% 91.5%\n\n**_Source: SPI Research, February 2025_**","publish_date":null,"title":"PS Solution Adoption","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/SPI_2025_Benchmark_Report_p78.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/SPI_2025_Benchmark_Report_p78.pdf"},"request":{"example_id":"chart/SPI_2025_Benchmark_Report_p78","product_type":"parse","source_file_path":"SPI_2025_Benchmark_Report_p78.pdf"},"started_at":"2026-11-04T19:46:21.224853"} +{"completed_at":"2026-11-04T19:46:24.481534","latency_in_ms":1384,"output":{"example_id":"chart/SPI_2025_Benchmark_Report_p97","grounded_pages":[],"layout_pages":[],"markdown":"### **2025 Professional Services Maturity** **™** **Benchmark Leadership Pillar**\n\nleadership challenges in embedded PSOs are charter conflict and forging cross-functional relationships. Embedded PS executives are tasked with developing a high-quality consulting business, but consulting is subordinate to product proliferation and adoption. A new, more strategic role is emerging to drive client adoption, optimization, and renewals. This role requires significantly greater alignment with sales, support, and product development, so collaboration and team-building skills are paramount.\n\n### **5-Year Leadership Trends**\n\nFigure 20 shows the degradation of three important leadership KPIs over the past five years. What makes this figure noteworthy is the deterioration. Leadership is by far the most critical aspect of success in professional services. As employees lose faith in leadership and it becomes harder to move forward, the professional services firm will suffer. SPI Research hopes PS leaders take heed of these figures and graphs and begin to help reestablish confidence and communication.\n\n**Figure 20: Leadership Trends of Note**\n\n**_Source: SPI Research, February 2025_**\n\nTable 78 shows the five-year trend of all the leadership KPIs in tabular form. Like those in the figure above, most have regressed over the past five years and sit below their five-year average. Few would doubt that PSO employees are not working hard. However, taking a step back occasionally to communicate and rebuild the culture, trust, and agility will help drive PSOs forward. The stress of COVID, coupled with geopolitical and economic tensions, has most likely hurt employees in every industry. But the Professional Services market is the leader and, therefore, must be a beacon across industries.\n\n**Table 78: Leadership Pillar 5-year Trends**\n\n**Key Performance Indicator (KPI) 5-year avg. 2020 2021 2022 2023 2024**\n\nWell understood vision, mission, and strategy 3.86 3.99 3.97 3.89 3.81 3.77\n\nConfidence in PS leadership 4.04 4\\.21 4.17 4.08 3.95 3.94\n\nEase of getting things done 3.73 3\\.88 3.85 3.72 3.70 3.63\n\nGoals and measurement alignment 3.74 3\\.84 3.81 3.68 3.75 3.73\n\nEmployees have confidence in PSO's future 3.91 4\\.07 4.08 3.93 3.80 3.78","pages":[{"line_numbers":[],"markdown":"### **2025 Professional Services Maturity** **™** **Benchmark Leadership Pillar**\n\nleadership challenges in embedded PSOs are charter conflict and forging cross-functional relationships. Embedded PS executives are tasked with developing a high-quality consulting business, but consulting is subordinate to product proliferation and adoption. A new, more strategic role is emerging to drive client adoption, optimization, and renewals. This role requires significantly greater alignment with sales, support, and product development, so collaboration and team-building skills are paramount.\n\n### **5-Year Leadership Trends**\n\nFigure 20 shows the degradation of three important leadership KPIs over the past five years. What makes this figure noteworthy is the deterioration. Leadership is by far the most critical aspect of success in professional services. As employees lose faith in leadership and it becomes harder to move forward, the professional services firm will suffer. SPI Research hopes PS leaders take heed of these figures and graphs and begin to help reestablish confidence and communication.\n\n**Figure 20: Leadership Trends of Note**\n\n**_Source: SPI Research, February 2025_**\n\nTable 78 shows the five-year trend of all the leadership KPIs in tabular form. Like those in the figure above, most have regressed over the past five years and sit below their five-year average. Few would doubt that PSO employees are not working hard. However, taking a step back occasionally to communicate and rebuild the culture, trust, and agility will help drive PSOs forward. The stress of COVID, coupled with geopolitical and economic tensions, has most likely hurt employees in every industry. But the Professional Services market is the leader and, therefore, must be a beacon across industries.\n\n**Table 78: Leadership Pillar 5-year Trends**\n\n**Key Performance Indicator (KPI) 5-year avg. 2020 2021 2022 2023 2024**\n\nWell understood vision, mission, and strategy 3.86 3.99 3.97 3.89 3.81 3.77\n\nConfidence in PS leadership 4.04 4\\.21 4.17 4.08 3.95 3.94\n\nEase of getting things done 3.73 3\\.88 3.85 3.72 3.70 3.63\n\nGoals and measurement alignment 3.74 3\\.84 3.81 3.68 3.75 3.73\n\nEmployees have confidence in PSO's future 3.91 4\\.07 4.08 3.93 3.80 3.78","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"### **2025 Professional Services Maturity** **™** **Benchmark Leadership Pillar**\n\nleadership challenges in embedded PSOs are charter conflict and forging cross-functional relationships. Embedded PS executives are tasked with developing a high-quality consulting business, but consulting is subordinate to product proliferation and adoption. A new, more strategic role is emerging to drive client adoption, optimization, and renewals. This role requires significantly greater alignment with sales, support, and product development, so collaboration and team-building skills are paramount.\n\n### **5-Year Leadership Trends**\n\nFigure 20 shows the degradation of three important leadership KPIs over the past five years. What makes this figure noteworthy is the deterioration. Leadership is by far the most critical aspect of success in professional services. As employees lose faith in leadership and it becomes harder to move forward, the professional services firm will suffer. SPI Research hopes PS leaders take heed of these figures and graphs and begin to help reestablish confidence and communication.\n\n**Figure 20: Leadership Trends of Note**\n\n**_Source: SPI Research, February 2025_**\n\nTable 78 shows the five-year trend of all the leadership KPIs in tabular form. Like those in the figure above, most have regressed over the past five years and sit below their five-year average. Few would doubt that PSO employees are not working hard. However, taking a step back occasionally to communicate and rebuild the culture, trust, and agility will help drive PSOs forward. The stress of COVID, coupled with geopolitical and economic tensions, has most likely hurt employees in every industry. But the Professional Services market is the leader and, therefore, must be a beacon across industries.\n\n**Table 78: Leadership Pillar 5-year Trends**\n\n**Key Performance Indicator (KPI) 5-year avg. 2020 2021 2022 2023 2024**\n\nWell understood vision, mission, and strategy 3.86 3.99 3.97 3.89 3.81 3.77\n\nConfidence in PS leadership 4.04 4\\.21 4.17 4.08 3.95 3.94\n\nEase of getting things done 3.73 3\\.88 3.85 3.72 3.70 3.63\n\nGoals and measurement alignment 3.74 3\\.84 3.81 3.68 3.75 3.73\n\nEmployees have confidence in PSO's future 3.91 4\\.07 4.08 3.93 3.80 3.78","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/SPI_2025_Benchmark_Report_p97.pdf"]},"response":{"errors":[],"extract_id":"extract_9da22dc0022473e7cbb5aa05a7e8fcec","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["### **2025 Professional Services Maturity** **™** **Benchmark Leadership Pillar**\n\nleadership challenges in embedded PSOs are charter conflict and forging cross-functional relationships. Embedded PS executives are tasked with developing a high-quality consulting business, but consulting is subordinate to product proliferation and adoption. A new, more strategic role is emerging to drive client adoption, optimization, and renewals. This role requires significantly greater alignment with sales, support, and product development, so collaboration and team-building skills are paramount.\n\n### **5-Year Leadership Trends**\n\nFigure 20 shows the degradation of three important leadership KPIs over the past five years. What makes this figure noteworthy is the deterioration. Leadership is by far the most critical aspect of success in professional services. As employees lose faith in leadership and it becomes harder to move forward, the professional services firm will suffer. SPI Research hopes PS leaders take heed of these figures and graphs and begin to help reestablish confidence and communication.\n\n**Figure 20: Leadership Trends of Note**\n\n**_Source: SPI Research, February 2025_**\n\nTable 78 shows the five-year trend of all the leadership KPIs in tabular form. Like those in the figure above, most have regressed over the past five years and sit below their five-year average. Few would doubt that PSO employees are not working hard. However, taking a step back occasionally to communicate and rebuild the culture, trust, and agility will help drive PSOs forward. The stress of COVID, coupled with geopolitical and economic tensions, has most likely hurt employees in every industry. But the Professional Services market is the leader and, therefore, must be a beacon across industries.\n\n**Table 78: Leadership Pillar 5-year Trends**\n\n**Key Performance Indicator (KPI) 5-year avg. 2020 2021 2022 2023 2024**\n\nWell understood vision, mission, and strategy 3.86 3.99 3.97 3.89 3.81 3.77\n\nConfidence in PS leadership 4.04 4\\.21 4.17 4.08 3.95 3.94\n\nEase of getting things done 3.73 3\\.88 3.85 3.72 3.70 3.63\n\nGoals and measurement alignment 3.74 3\\.84 3.81 3.68 3.75 3.73\n\nEmployees have confidence in PSO's future 3.91 4\\.07 4.08 3.93 3.80 3.78"],"full_content":"### **2025 Professional Services Maturity** **™** **Benchmark Leadership Pillar**\n\nleadership challenges in embedded PSOs are charter conflict and forging cross-functional relationships. Embedded PS executives are tasked with developing a high-quality consulting business, but consulting is subordinate to product proliferation and adoption. A new, more strategic role is emerging to drive client adoption, optimization, and renewals. This role requires significantly greater alignment with sales, support, and product development, so collaboration and team-building skills are paramount.\n\n### **5-Year Leadership Trends**\n\nFigure 20 shows the degradation of three important leadership KPIs over the past five years. What makes this figure noteworthy is the deterioration. Leadership is by far the most critical aspect of success in professional services. As employees lose faith in leadership and it becomes harder to move forward, the professional services firm will suffer. SPI Research hopes PS leaders take heed of these figures and graphs and begin to help reestablish confidence and communication.\n\n**Figure 20: Leadership Trends of Note**\n\n**_Source: SPI Research, February 2025_**\n\nTable 78 shows the five-year trend of all the leadership KPIs in tabular form. Like those in the figure above, most have regressed over the past five years and sit below their five-year average. Few would doubt that PSO employees are not working hard. However, taking a step back occasionally to communicate and rebuild the culture, trust, and agility will help drive PSOs forward. The stress of COVID, coupled with geopolitical and economic tensions, has most likely hurt employees in every industry. But the Professional Services market is the leader and, therefore, must be a beacon across industries.\n\n**Table 78: Leadership Pillar 5-year Trends**\n\n**Key Performance Indicator (KPI) 5-year avg. 2020 2021 2022 2023 2024**\n\nWell understood vision, mission, and strategy 3.86 3.99 3.97 3.89 3.81 3.77\n\nConfidence in PS leadership 4.04 4\\.21 4.17 4.08 3.95 3.94\n\nEase of getting things done 3.73 3\\.88 3.85 3.72 3.70 3.63\n\nGoals and measurement alignment 3.74 3\\.84 3.81 3.68 3.75 3.73\n\nEmployees have confidence in PSO's future 3.91 4\\.07 4.08 3.93 3.80 3.78","publish_date":null,"title":"5-Year Leadership Trends","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/SPI_2025_Benchmark_Report_p97.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/SPI_2025_Benchmark_Report_p97.pdf"},"request":{"example_id":"chart/SPI_2025_Benchmark_Report_p97","product_type":"parse","source_file_path":"SPI_2025_Benchmark_Report_p97.pdf"},"started_at":"2026-11-04T19:46:23.097179"} +{"completed_at":"2026-11-04T19:30:00.217419","latency_in_ms":2771,"output":{"example_id":"chart/SRI-Insights-August 2025_media_embargo_p1","grounded_pages":[],"layout_pages":[],"markdown":"First-quarter 2025 fires trigger second highest first- half natural catastrophe insured losses\n\nSwiss Re Institute’s preliminary estimates see global insured losses from natural catastrophes reaching USD 80 billion in the first half of this year (see Figure 1). The main drivers were the wildfires that swept through parts of Los Angeles (LA) County in January, these alone accounting for close to 70% of the global insured losses in the first quarter (see Figure 2 opposite). Most of the first-half losses came in January-March, mounting to more than double the first-quarter average of the previous five years. Losses moderated to below historical average in the second quarter, but not enough to offset the severe first quarter outcome.\n\nFirst-half insured losses of USD 80 billion would make for the industry’s second highest first-half loss tally ever after 2011, when earthquakes (EQ) in New Zealand (February) and Japan (March) triggered record high claims in the first six months of that year. It will also make 2025 the fifth in succession in which first-half insured losses have exceeded USD 50 billion. There were no other major insured loss-making events in the first half and absent the unseasonal LA County fires, global insured losses would have been well below trend. With changing climates, unexpected and unseasonal weather conditions may occur more frequently, making loss outcomes more volatile and difficult to predict.\n\n# Unseasonal fires trigger above-trend catastrophe insured losses first half 2025\n\n### We estimate that global insured catastrophe losses reached USD 80 billion in the first half of 2025, above trend. Unprecedented losses from wildfires in California and large thunderstorms in the US led to the second costliest first half ever (after the first half of 2011). Wildfire risk is evolving amid settlement trends and lengthening of fire seasons, with changing climates compounding the loss threat that fires present. This adds volatility to global natural catastrophe losses, making the latter more difficult to predict. In the second half, the Atlantic hurricane season approaching its peak remains a key swing factor for catastrophe activity although other perils, including wildfire, may also add to the full-year insured loss total.\n\n**Figure 1:** Global insured losses from natural catastrophes in 1H (USD bn, 2025 prices)\n\nSource: Swiss Re Institute\n\n0\n\n20\n\n40\n\n60\n\n80\n\n100\n\n120\n\n140\n\nFirst-half insured losses\n\nTrend\n\n2025 2021 2017 2013 2009 2005 2001 1997\n\nEarthquake, Chile\n\nEarthquakes, Japan & NZ\n\nWildfires, US\n\nTrend First half insured losses\n\n**Figure 2:** Global insured losses from natural catastrophes in 1Q and 2Q 2025 and previous years averages (USD bn, 2025 prices)\n\nSource: Swiss Re Institute\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\n30\n\n15\n\n5-\n\n20\n\n20\n\n2Q 1Q\n\n30-year average (1995–2024) 15-year average (2010–2024)\n\n5-year average (2020–2024) 2025 wildfires 2025 others\n\n### _sigma_ insights August 2025\n\n# Under embargo until Wednesday 6 August 2025, 10 am CEST","pages":[{"line_numbers":[],"markdown":"First-quarter 2025 fires trigger second highest first- half natural catastrophe insured losses\n\nSwiss Re Institute’s preliminary estimates see global insured losses from natural catastrophes reaching USD 80 billion in the first half of this year (see Figure 1). The main drivers were the wildfires that swept through parts of Los Angeles (LA) County in January, these alone accounting for close to 70% of the global insured losses in the first quarter (see Figure 2 opposite). Most of the first-half losses came in January-March, mounting to more than double the first-quarter average of the previous five years. Losses moderated to below historical average in the second quarter, but not enough to offset the severe first quarter outcome.\n\nFirst-half insured losses of USD 80 billion would make for the industry’s second highest first-half loss tally ever after 2011, when earthquakes (EQ) in New Zealand (February) and Japan (March) triggered record high claims in the first six months of that year. It will also make 2025 the fifth in succession in which first-half insured losses have exceeded USD 50 billion. There were no other major insured loss-making events in the first half and absent the unseasonal LA County fires, global insured losses would have been well below trend. With changing climates, unexpected and unseasonal weather conditions may occur more frequently, making loss outcomes more volatile and difficult to predict.\n\n# Unseasonal fires trigger above-trend catastrophe insured losses first half 2025\n\n### We estimate that global insured catastrophe losses reached USD 80 billion in the first half of 2025, above trend. Unprecedented losses from wildfires in California and large thunderstorms in the US led to the second costliest first half ever (after the first half of 2011). Wildfire risk is evolving amid settlement trends and lengthening of fire seasons, with changing climates compounding the loss threat that fires present. This adds volatility to global natural catastrophe losses, making the latter more difficult to predict. In the second half, the Atlantic hurricane season approaching its peak remains a key swing factor for catastrophe activity although other perils, including wildfire, may also add to the full-year insured loss total.\n\n**Figure 1:** Global insured losses from natural catastrophes in 1H (USD bn, 2025 prices)\n\nSource: Swiss Re Institute\n\n0\n\n20\n\n40\n\n60\n\n80\n\n100\n\n120\n\n140\n\nFirst-half insured losses\n\nTrend\n\n2025 2021 2017 2013 2009 2005 2001 1997\n\nEarthquake, Chile\n\nEarthquakes, Japan & NZ\n\nWildfires, US\n\nTrend First half insured losses\n\n**Figure 2:** Global insured losses from natural catastrophes in 1Q and 2Q 2025 and previous years averages (USD bn, 2025 prices)\n\nSource: Swiss Re Institute\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\n30\n\n15\n\n5-\n\n20\n\n20\n\n2Q 1Q\n\n30-year average (1995–2024) 15-year average (2010–2024)\n\n5-year average (2020–2024) 2025 wildfires 2025 others\n\n### _sigma_ insights August 2025\n\n# Under embargo until Wednesday 6 August 2025, 10 am CEST","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"First-quarter 2025 fires trigger second highest first- half natural catastrophe insured losses\n\nSwiss Re Institute’s preliminary estimates see global insured losses from natural catastrophes reaching USD 80 billion in the first half of this year (see Figure 1). The main drivers were the wildfires that swept through parts of Los Angeles (LA) County in January, these alone accounting for close to 70% of the global insured losses in the first quarter (see Figure 2 opposite). Most of the first-half losses came in January-March, mounting to more than double the first-quarter average of the previous five years. Losses moderated to below historical average in the second quarter, but not enough to offset the severe first quarter outcome.\n\nFirst-half insured losses of USD 80 billion would make for the industry’s second highest first-half loss tally ever after 2011, when earthquakes (EQ) in New Zealand (February) and Japan (March) triggered record high claims in the first six months of that year. It will also make 2025 the fifth in succession in which first-half insured losses have exceeded USD 50 billion. There were no other major insured loss-making events in the first half and absent the unseasonal LA County fires, global insured losses would have been well below trend. With changing climates, unexpected and unseasonal weather conditions may occur more frequently, making loss outcomes more volatile and difficult to predict.\n\n# Unseasonal fires trigger above-trend catastrophe insured losses first half 2025\n\n### We estimate that global insured catastrophe losses reached USD 80 billion in the first half of 2025, above trend. Unprecedented losses from wildfires in California and large thunderstorms in the US led to the second costliest first half ever (after the first half of 2011). Wildfire risk is evolving amid settlement trends and lengthening of fire seasons, with changing climates compounding the loss threat that fires present. This adds volatility to global natural catastrophe losses, making the latter more difficult to predict. In the second half, the Atlantic hurricane season approaching its peak remains a key swing factor for catastrophe activity although other perils, including wildfire, may also add to the full-year insured loss total.\n\n**Figure 1:** Global insured losses from natural catastrophes in 1H (USD bn, 2025 prices)\n\nSource: Swiss Re Institute\n\n0\n\n20\n\n40\n\n60\n\n80\n\n100\n\n120\n\n140\n\nFirst-half insured losses\n\nTrend\n\n2025 2021 2017 2013 2009 2005 2001 1997\n\nEarthquake, Chile\n\nEarthquakes, Japan & NZ\n\nWildfires, US\n\nTrend First half insured losses\n\n**Figure 2:** Global insured losses from natural catastrophes in 1Q and 2Q 2025 and previous years averages (USD bn, 2025 prices)\n\nSource: Swiss Re Institute\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\n30\n\n15\n\n5-\n\n20\n\n20\n\n2Q 1Q\n\n30-year average (1995–2024) 15-year average (2010–2024)\n\n5-year average (2020–2024) 2025 wildfires 2025 others\n\n### _sigma_ insights August 2025\n\n# Under embargo until Wednesday 6 August 2025, 10 am CEST","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/SRI-Insights-August%202025_media_embargo_p1.pdf"]},"response":{"errors":[],"extract_id":"extract_6750f34eda17775882cd1b519a6bc4bb","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["First-quarter 2025 fires trigger second highest first- half natural catastrophe insured losses\n\nSwiss Re Institute’s preliminary estimates see global insured losses from natural catastrophes reaching USD 80 billion in the first half of this year (see Figure 1). The main drivers were the wildfires that swept through parts of Los Angeles (LA) County in January, these alone accounting for close to 70% of the global insured losses in the first quarter (see Figure 2 opposite). Most of the first-half losses came in January-March, mounting to more than double the first-quarter average of the previous five years. Losses moderated to below historical average in the second quarter, but not enough to offset the severe first quarter outcome.\n\nFirst-half insured losses of USD 80 billion would make for the industry’s second highest first-half loss tally ever after 2011, when earthquakes (EQ) in New Zealand (February) and Japan (March) triggered record high claims in the first six months of that year. It will also make 2025 the fifth in succession in which first-half insured losses have exceeded USD 50 billion. There were no other major insured loss-making events in the first half and absent the unseasonal LA County fires, global insured losses would have been well below trend. With changing climates, unexpected and unseasonal weather conditions may occur more frequently, making loss outcomes more volatile and difficult to predict.\n\n# Unseasonal fires trigger above-trend catastrophe insured losses first half 2025\n\n### We estimate that global insured catastrophe losses reached USD 80 billion in the first half of 2025, above trend. Unprecedented losses from wildfires in California and large thunderstorms in the US led to the second costliest first half ever (after the first half of 2011). Wildfire risk is evolving amid settlement trends and lengthening of fire seasons, with changing climates compounding the loss threat that fires present. This adds volatility to global natural catastrophe losses, making the latter more difficult to predict. In the second half, the Atlantic hurricane season approaching its peak remains a key swing factor for catastrophe activity although other perils, including wildfire, may also add to the full-year insured loss total.\n\n**Figure 1:** Global insured losses from natural catastrophes in 1H (USD bn, 2025 prices)\n\nSource: Swiss Re Institute\n\n0\n\n20\n\n40\n\n60\n\n80\n\n100\n\n120\n\n140\n\nFirst-half insured losses\n\nTrend\n\n2025 2021 2017 2013 2009 2005 2001 1997\n\nEarthquake, Chile\n\nEarthquakes, Japan & NZ\n\nWildfires, US\n\nTrend First half insured losses\n\n**Figure 2:** Global insured losses from natural catastrophes in 1Q and 2Q 2025 and previous years averages (USD bn, 2025 prices)\n\nSource: Swiss Re Institute\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\n30\n\n15\n\n5-\n\n20\n\n20\n\n2Q 1Q\n\n30-year average (1995–2024) 15-year average (2010–2024)\n\n5-year average (2020–2024) 2025 wildfires 2025 others\n\n### _sigma_ insights August 2025\n\n# Under embargo until Wednesday 6 August 2025, 10 am CEST"],"full_content":"First-quarter 2025 fires trigger second highest first- half natural catastrophe insured losses\n\nSwiss Re Institute’s preliminary estimates see global insured losses from natural catastrophes reaching USD 80 billion in the first half of this year (see Figure 1). The main drivers were the wildfires that swept through parts of Los Angeles (LA) County in January, these alone accounting for close to 70% of the global insured losses in the first quarter (see Figure 2 opposite). Most of the first-half losses came in January-March, mounting to more than double the first-quarter average of the previous five years. Losses moderated to below historical average in the second quarter, but not enough to offset the severe first quarter outcome.\n\nFirst-half insured losses of USD 80 billion would make for the industry’s second highest first-half loss tally ever after 2011, when earthquakes (EQ) in New Zealand (February) and Japan (March) triggered record high claims in the first six months of that year. It will also make 2025 the fifth in succession in which first-half insured losses have exceeded USD 50 billion. There were no other major insured loss-making events in the first half and absent the unseasonal LA County fires, global insured losses would have been well below trend. With changing climates, unexpected and unseasonal weather conditions may occur more frequently, making loss outcomes more volatile and difficult to predict.\n\n# Unseasonal fires trigger above-trend catastrophe insured losses first half 2025\n\n### We estimate that global insured catastrophe losses reached USD 80 billion in the first half of 2025, above trend. Unprecedented losses from wildfires in California and large thunderstorms in the US led to the second costliest first half ever (after the first half of 2011). Wildfire risk is evolving amid settlement trends and lengthening of fire seasons, with changing climates compounding the loss threat that fires present. This adds volatility to global natural catastrophe losses, making the latter more difficult to predict. In the second half, the Atlantic hurricane season approaching its peak remains a key swing factor for catastrophe activity although other perils, including wildfire, may also add to the full-year insured loss total.\n\n**Figure 1:** Global insured losses from natural catastrophes in 1H (USD bn, 2025 prices)\n\nSource: Swiss Re Institute\n\n0\n\n20\n\n40\n\n60\n\n80\n\n100\n\n120\n\n140\n\nFirst-half insured losses\n\nTrend\n\n2025 2021 2017 2013 2009 2005 2001 1997\n\nEarthquake, Chile\n\nEarthquakes, Japan & NZ\n\nWildfires, US\n\nTrend First half insured losses\n\n**Figure 2:** Global insured losses from natural catastrophes in 1Q and 2Q 2025 and previous years averages (USD bn, 2025 prices)\n\nSource: Swiss Re Institute\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\n30\n\n15\n\n5-\n\n20\n\n20\n\n2Q 1Q\n\n30-year average (1995–2024) 15-year average (2010–2024)\n\n5-year average (2020–2024) 2025 wildfires 2025 others\n\n### _sigma_ insights August 2025\n\n# Under embargo until Wednesday 6 August 2025, 10 am CEST","publish_date":null,"title":"Unseasonal fires trigger above-trend catastrophe insured losses first half 2025","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/SRI-Insights-August%202025_media_embargo_p1.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/SRI-Insights-August%202025_media_embargo_p1.pdf"},"request":{"example_id":"chart/SRI-Insights-August 2025_media_embargo_p1","product_type":"parse","source_file_path":"SRI-Insights-August 2025_media_embargo_p1.pdf"},"started_at":"2026-11-04T19:29:57.445879"} +{"completed_at":"2026-11-04T19:46:27.367017","latency_in_ms":1573,"output":{"example_id":"chart/SRI-Insights-August 2025_media_embargo_p2","grounded_pages":[],"layout_pages":[],"markdown":"2 **Swiss Re Institute** Unseasonal fires trigger above-trend first-half 2025 natural catastrophe insured losses\n\nThe Palisades and Eaton wildfires in LA County were the costliest events of the first half, resulting in combined insured losses of USD 40 billion (preliminary estimates). That’s half of global first- half losses insured losses. For all other major perils, also severe convective storms (SCS), losses were instead mostly below the January-June average of the previous five years (see Figure 3).\n\n**Figure 3:** Global insured losses from natural catastrophes by peril in 1H 2025 and previous half-year averages (USD bn, 2025 prices)\n\nSource: Swiss Re Institute\n\n0 10 20 30 40 50 60 70 80\n\nDroughts\n\nEuropean winter storms\n\nWildfires\n\nOther secondary perils\n\nFloods\n\nEarthquakes\n\nTropical cyclones\n\nSevere convective storms\n\n2025\n\n5-year average (2020–2024)\n\n10-year average (2015–2024)\n\n30-year average\n\n(1995–2024)\n\nDroughts\n\nOther secondary perils European winter storms Wildfires\n\nSevere convective storms Floods Tropical cyclones Earthquakes\n\nEconomic losses from natural catastrophes in the first half of 2025 were USD 135 billion, of which insurance covered about 59% (see Figure 4). That’s higher than the previous 10-year average of 42% and reflects that much of the damage caused by disasters was in US and other regions of relatively high insurance penetration, and also that wildfire is widely covered in personal property covers.\n\n**Figure 4:** Global annual insured and uninsured losses from natural catastrophes in 1H (USD bn, 2025 prices)\n\nSource: Swiss Re Institute\n\n0\n\n30\n\n60\n\n90\n\n120\n\n150\n\n180\n\nUninsured losses (left)\n\nInsured losses (left)\n\n2017 2018 2019 2020 2021 2022 2023 2024 2025 2016\n\n0%\n\n20%\n\n40%\n\n60%\n\nInsured losses as a percentage of total economic losses (right)\n\n37%\n\n43% 43%\n\n33%\n\n47% 50%\n\n39% 39%\n\n51%\n\n59%\n\nUninsured losses (left) Insured losses (left)\n\nInsured losses as a percentage of total economic losses (right)\n\nThe LA County fires were not the only major disaster in the first half. Another catastrophe was the magnitude 7.7 earthquake that struck Myanmar in March, claiming around 3 900 human lives, more than any other disaster. ¹ The quake struck densely populated regions of the country including Mandalay, its second largest city, and total economic losses are estimated to have been around USD 14 billion. There is no information on insured losses but, with low insurance penetration, these will likely be low relative to regions of higher risk protection uptake. The shock waves from the earthquake were felt as far away as in India, China and Thailand. In Bangkok, the seismic shakes caused a high-rise building under construction to collapse, claiming more lives. Bangkok sits on a basin of soft sediment that acts like as amplifier, increasing the intensity and duration of shaking and making even far-off earthquakes feel intense. For this reason, even though far away from active faults, Bangkok is exposed to loss potential from seismic activity. Preliminary estimates from the Thai General Insurance Association indicate insured losses in the country of USD 1.5 billion as a result of the earthquake.\n\nWith earthquakes, location and ground conditions matter as much as magnitude with respect to damage caused. Many cities around the world including Mexico City, LA, Beijing and Tokyo are built on sedimentary basins and face elevated seismic risk on account of potential for amplified ground motions. Without proper building codes and adequate enforcement thereof, the increased shaking can cause severe structural damage and loss of life. Modern seismic hazard assessment including localised effects is needed to accurately assess the resilience of buildings to ground shakes, and the stringency of building codes required. Strong building codes reduce damage but are not always rigorously enforced. Risk models must account for basin effects to avoid underestimating loss potential.\n\nNon-seasonal wildfires account for half of the January- June losses\n\nIn January 2025, two major wildfires burned close to 40 000 acres in LA County and destroyed more than 16 000 structures in the Palisades and Eaton neighbourhoods. ² A prolonged and dry Santa Ana wind with gusts reaching speeds of almost 100 mph, coupled with a lack of rainfall, created unseasonal and extreme fire-fuelling weather conditions. The area affected by the Palisades fire contains some of the densest concentration of high-value single-family residential property in the US. The resulting fires covered locations of high exposure, leading to exceptional losses. A very large number of structures were destroyed, more than 400% of the long- term annual average, causing high losses in what was otherwise in terms of physical events an unremarkable fire season. For instance, in year to 19 June 2025, the number of acres burned in the California and the number of fire outbreaks were just 5% and 38% of the respective long-term yearly averages.\n\nAccording to Swiss Re estimates, as of July 2025 combined insured losses from the Palisades and Eaton firestorm(s) were USD 40 billion, by far the largest global insured wildfire loss event(s) to date, and among the top 10 costliest natural catastrophe insured loss events of all perils on _sigma_ records. Total first-half insured losses for wildfire were the highest ever, surpassing even the previous full-year peaks for this peril by a significant margin (see Figure 5 below).","pages":[{"line_numbers":[],"markdown":"2 **Swiss Re Institute** Unseasonal fires trigger above-trend first-half 2025 natural catastrophe insured losses\n\nThe Palisades and Eaton wildfires in LA County were the costliest events of the first half, resulting in combined insured losses of USD 40 billion (preliminary estimates). That’s half of global first- half losses insured losses. For all other major perils, also severe convective storms (SCS), losses were instead mostly below the January-June average of the previous five years (see Figure 3).\n\n**Figure 3:** Global insured losses from natural catastrophes by peril in 1H 2025 and previous half-year averages (USD bn, 2025 prices)\n\nSource: Swiss Re Institute\n\n0 10 20 30 40 50 60 70 80\n\nDroughts\n\nEuropean winter storms\n\nWildfires\n\nOther secondary perils\n\nFloods\n\nEarthquakes\n\nTropical cyclones\n\nSevere convective storms\n\n2025\n\n5-year average (2020–2024)\n\n10-year average (2015–2024)\n\n30-year average\n\n(1995–2024)\n\nDroughts\n\nOther secondary perils European winter storms Wildfires\n\nSevere convective storms Floods Tropical cyclones Earthquakes\n\nEconomic losses from natural catastrophes in the first half of 2025 were USD 135 billion, of which insurance covered about 59% (see Figure 4). That’s higher than the previous 10-year average of 42% and reflects that much of the damage caused by disasters was in US and other regions of relatively high insurance penetration, and also that wildfire is widely covered in personal property covers.\n\n**Figure 4:** Global annual insured and uninsured losses from natural catastrophes in 1H (USD bn, 2025 prices)\n\nSource: Swiss Re Institute\n\n0\n\n30\n\n60\n\n90\n\n120\n\n150\n\n180\n\nUninsured losses (left)\n\nInsured losses (left)\n\n2017 2018 2019 2020 2021 2022 2023 2024 2025 2016\n\n0%\n\n20%\n\n40%\n\n60%\n\nInsured losses as a percentage of total economic losses (right)\n\n37%\n\n43% 43%\n\n33%\n\n47% 50%\n\n39% 39%\n\n51%\n\n59%\n\nUninsured losses (left) Insured losses (left)\n\nInsured losses as a percentage of total economic losses (right)\n\nThe LA County fires were not the only major disaster in the first half. Another catastrophe was the magnitude 7.7 earthquake that struck Myanmar in March, claiming around 3 900 human lives, more than any other disaster. ¹ The quake struck densely populated regions of the country including Mandalay, its second largest city, and total economic losses are estimated to have been around USD 14 billion. There is no information on insured losses but, with low insurance penetration, these will likely be low relative to regions of higher risk protection uptake. The shock waves from the earthquake were felt as far away as in India, China and Thailand. In Bangkok, the seismic shakes caused a high-rise building under construction to collapse, claiming more lives. Bangkok sits on a basin of soft sediment that acts like as amplifier, increasing the intensity and duration of shaking and making even far-off earthquakes feel intense. For this reason, even though far away from active faults, Bangkok is exposed to loss potential from seismic activity. Preliminary estimates from the Thai General Insurance Association indicate insured losses in the country of USD 1.5 billion as a result of the earthquake.\n\nWith earthquakes, location and ground conditions matter as much as magnitude with respect to damage caused. Many cities around the world including Mexico City, LA, Beijing and Tokyo are built on sedimentary basins and face elevated seismic risk on account of potential for amplified ground motions. Without proper building codes and adequate enforcement thereof, the increased shaking can cause severe structural damage and loss of life. Modern seismic hazard assessment including localised effects is needed to accurately assess the resilience of buildings to ground shakes, and the stringency of building codes required. Strong building codes reduce damage but are not always rigorously enforced. Risk models must account for basin effects to avoid underestimating loss potential.\n\nNon-seasonal wildfires account for half of the January- June losses\n\nIn January 2025, two major wildfires burned close to 40 000 acres in LA County and destroyed more than 16 000 structures in the Palisades and Eaton neighbourhoods. ² A prolonged and dry Santa Ana wind with gusts reaching speeds of almost 100 mph, coupled with a lack of rainfall, created unseasonal and extreme fire-fuelling weather conditions. The area affected by the Palisades fire contains some of the densest concentration of high-value single-family residential property in the US. The resulting fires covered locations of high exposure, leading to exceptional losses. A very large number of structures were destroyed, more than 400% of the long- term annual average, causing high losses in what was otherwise in terms of physical events an unremarkable fire season. For instance, in year to 19 June 2025, the number of acres burned in the California and the number of fire outbreaks were just 5% and 38% of the respective long-term yearly averages.\n\nAccording to Swiss Re estimates, as of July 2025 combined insured losses from the Palisades and Eaton firestorm(s) were USD 40 billion, by far the largest global insured wildfire loss event(s) to date, and among the top 10 costliest natural catastrophe insured loss events of all perils on _sigma_ records. Total first-half insured losses for wildfire were the highest ever, surpassing even the previous full-year peaks for this peril by a significant margin (see Figure 5 below).","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"2 **Swiss Re Institute** Unseasonal fires trigger above-trend first-half 2025 natural catastrophe insured losses\n\nThe Palisades and Eaton wildfires in LA County were the costliest events of the first half, resulting in combined insured losses of USD 40 billion (preliminary estimates). That’s half of global first- half losses insured losses. For all other major perils, also severe convective storms (SCS), losses were instead mostly below the January-June average of the previous five years (see Figure 3).\n\n**Figure 3:** Global insured losses from natural catastrophes by peril in 1H 2025 and previous half-year averages (USD bn, 2025 prices)\n\nSource: Swiss Re Institute\n\n0 10 20 30 40 50 60 70 80\n\nDroughts\n\nEuropean winter storms\n\nWildfires\n\nOther secondary perils\n\nFloods\n\nEarthquakes\n\nTropical cyclones\n\nSevere convective storms\n\n2025\n\n5-year average (2020–2024)\n\n10-year average (2015–2024)\n\n30-year average\n\n(1995–2024)\n\nDroughts\n\nOther secondary perils European winter storms Wildfires\n\nSevere convective storms Floods Tropical cyclones Earthquakes\n\nEconomic losses from natural catastrophes in the first half of 2025 were USD 135 billion, of which insurance covered about 59% (see Figure 4). That’s higher than the previous 10-year average of 42% and reflects that much of the damage caused by disasters was in US and other regions of relatively high insurance penetration, and also that wildfire is widely covered in personal property covers.\n\n**Figure 4:** Global annual insured and uninsured losses from natural catastrophes in 1H (USD bn, 2025 prices)\n\nSource: Swiss Re Institute\n\n0\n\n30\n\n60\n\n90\n\n120\n\n150\n\n180\n\nUninsured losses (left)\n\nInsured losses (left)\n\n2017 2018 2019 2020 2021 2022 2023 2024 2025 2016\n\n0%\n\n20%\n\n40%\n\n60%\n\nInsured losses as a percentage of total economic losses (right)\n\n37%\n\n43% 43%\n\n33%\n\n47% 50%\n\n39% 39%\n\n51%\n\n59%\n\nUninsured losses (left) Insured losses (left)\n\nInsured losses as a percentage of total economic losses (right)\n\nThe LA County fires were not the only major disaster in the first half. Another catastrophe was the magnitude 7.7 earthquake that struck Myanmar in March, claiming around 3 900 human lives, more than any other disaster. ¹ The quake struck densely populated regions of the country including Mandalay, its second largest city, and total economic losses are estimated to have been around USD 14 billion. There is no information on insured losses but, with low insurance penetration, these will likely be low relative to regions of higher risk protection uptake. The shock waves from the earthquake were felt as far away as in India, China and Thailand. In Bangkok, the seismic shakes caused a high-rise building under construction to collapse, claiming more lives. Bangkok sits on a basin of soft sediment that acts like as amplifier, increasing the intensity and duration of shaking and making even far-off earthquakes feel intense. For this reason, even though far away from active faults, Bangkok is exposed to loss potential from seismic activity. Preliminary estimates from the Thai General Insurance Association indicate insured losses in the country of USD 1.5 billion as a result of the earthquake.\n\nWith earthquakes, location and ground conditions matter as much as magnitude with respect to damage caused. Many cities around the world including Mexico City, LA, Beijing and Tokyo are built on sedimentary basins and face elevated seismic risk on account of potential for amplified ground motions. Without proper building codes and adequate enforcement thereof, the increased shaking can cause severe structural damage and loss of life. Modern seismic hazard assessment including localised effects is needed to accurately assess the resilience of buildings to ground shakes, and the stringency of building codes required. Strong building codes reduce damage but are not always rigorously enforced. Risk models must account for basin effects to avoid underestimating loss potential.\n\nNon-seasonal wildfires account for half of the January- June losses\n\nIn January 2025, two major wildfires burned close to 40 000 acres in LA County and destroyed more than 16 000 structures in the Palisades and Eaton neighbourhoods. ² A prolonged and dry Santa Ana wind with gusts reaching speeds of almost 100 mph, coupled with a lack of rainfall, created unseasonal and extreme fire-fuelling weather conditions. The area affected by the Palisades fire contains some of the densest concentration of high-value single-family residential property in the US. The resulting fires covered locations of high exposure, leading to exceptional losses. A very large number of structures were destroyed, more than 400% of the long- term annual average, causing high losses in what was otherwise in terms of physical events an unremarkable fire season. For instance, in year to 19 June 2025, the number of acres burned in the California and the number of fire outbreaks were just 5% and 38% of the respective long-term yearly averages.\n\nAccording to Swiss Re estimates, as of July 2025 combined insured losses from the Palisades and Eaton firestorm(s) were USD 40 billion, by far the largest global insured wildfire loss event(s) to date, and among the top 10 costliest natural catastrophe insured loss events of all perils on _sigma_ records. Total first-half insured losses for wildfire were the highest ever, surpassing even the previous full-year peaks for this peril by a significant margin (see Figure 5 below).","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/SRI-Insights-August%202025_media_embargo_p2.pdf"]},"response":{"errors":[],"extract_id":"extract_862742e4a2c02774e0065b19b052a2d2","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["2 **Swiss Re Institute** Unseasonal fires trigger above-trend first-half 2025 natural catastrophe insured losses\n\nThe Palisades and Eaton wildfires in LA County were the costliest events of the first half, resulting in combined insured losses of USD 40 billion (preliminary estimates). That’s half of global first- half losses insured losses. For all other major perils, also severe convective storms (SCS), losses were instead mostly below the January-June average of the previous five years (see Figure 3).\n\n**Figure 3:** Global insured losses from natural catastrophes by peril in 1H 2025 and previous half-year averages (USD bn, 2025 prices)\n\nSource: Swiss Re Institute\n\n0 10 20 30 40 50 60 70 80\n\nDroughts\n\nEuropean winter storms\n\nWildfires\n\nOther secondary perils\n\nFloods\n\nEarthquakes\n\nTropical cyclones\n\nSevere convective storms\n\n2025\n\n5-year average (2020–2024)\n\n10-year average (2015–2024)\n\n30-year average\n\n(1995–2024)\n\nDroughts\n\nOther secondary perils European winter storms Wildfires\n\nSevere convective storms Floods Tropical cyclones Earthquakes\n\nEconomic losses from natural catastrophes in the first half of 2025 were USD 135 billion, of which insurance covered about 59% (see Figure 4). That’s higher than the previous 10-year average of 42% and reflects that much of the damage caused by disasters was in US and other regions of relatively high insurance penetration, and also that wildfire is widely covered in personal property covers.\n\n**Figure 4:** Global annual insured and uninsured losses from natural catastrophes in 1H (USD bn, 2025 prices)\n\nSource: Swiss Re Institute\n\n0\n\n30\n\n60\n\n90\n\n120\n\n150\n\n180\n\nUninsured losses (left)\n\nInsured losses (left)\n\n2017 2018 2019 2020 2021 2022 2023 2024 2025 2016\n\n0%\n\n20%\n\n40%\n\n60%\n\nInsured losses as a percentage of total economic losses (right)\n\n37%\n\n43% 43%\n\n33%\n\n47% 50%\n\n39% 39%\n\n51%\n\n59%\n\nUninsured losses (left) Insured losses (left)\n\nInsured losses as a percentage of total economic losses (right)\n\nThe LA County fires were not the only major disaster in the first half. Another catastrophe was the magnitude 7.7 earthquake that struck Myanmar in March, claiming around 3 900 human lives, more than any other disaster. ¹ The quake struck densely populated regions of the country including Mandalay, its second largest city, and total economic losses are estimated to have been around USD 14 billion. There is no information on insured losses but, with low insurance penetration, these will likely be low relative to regions of higher risk protection uptake. The shock waves from the earthquake were felt as far away as in India, China and Thailand. In Bangkok, the seismic shakes caused a high-rise building under construction to collapse, claiming more lives. Bangkok sits on a basin of soft sediment that acts like as amplifier, increasing the intensity and duration of shaking and making even far-off earthquakes feel intense. For this reason, even though far away from active faults, Bangkok is exposed to loss potential from seismic activity. Preliminary estimates from the Thai General Insurance Association indicate insured losses in the country of USD 1.5 billion as a result of the earthquake.\n\nWith earthquakes, location and ground conditions matter as much as magnitude with respect to damage caused. Many cities around the world including Mexico City, LA, Beijing and Tokyo are built on sedimentary basins and face elevated seismic risk on account of potential for amplified ground motions. Without proper building codes and adequate enforcement thereof, the increased shaking can cause severe structural damage and loss of life. Modern seismic hazard assessment including localised effects is needed to accurately assess the resilience of buildings to ground shakes, and the stringency of building codes required. Strong building codes reduce damage but are not always rigorously enforced. Risk models must account for basin effects to avoid underestimating loss potential.\n\nNon-seasonal wildfires account for half of the January- June losses\n\nIn January 2025, two major wildfires burned close to 40 000 acres in LA County and destroyed more than 16 000 structures in the Palisades and Eaton neighbourhoods. ² A prolonged and dry Santa Ana wind with gusts reaching speeds of almost 100 mph, coupled with a lack of rainfall, created unseasonal and extreme fire-fuelling weather conditions. The area affected by the Palisades fire contains some of the densest concentration of high-value single-family residential property in the US. The resulting fires covered locations of high exposure, leading to exceptional losses. A very large number of structures were destroyed, more than 400% of the long- term annual average, causing high losses in what was otherwise in terms of physical events an unremarkable fire season. For instance, in year to 19 June 2025, the number of acres burned in the California and the number of fire outbreaks were just 5% and 38% of the respective long-term yearly averages.\n\nAccording to Swiss Re estimates, as of July 2025 combined insured losses from the Palisades and Eaton firestorm(s) were USD 40 billion, by far the largest global insured wildfire loss event(s) to date, and among the top 10 costliest natural catastrophe insured loss events of all perils on _sigma_ records. Total first-half insured losses for wildfire were the highest ever, surpassing even the previous full-year peaks for this peril by a significant margin (see Figure 5 below)."],"full_content":"2 **Swiss Re Institute** Unseasonal fires trigger above-trend first-half 2025 natural catastrophe insured losses\n\nThe Palisades and Eaton wildfires in LA County were the costliest events of the first half, resulting in combined insured losses of USD 40 billion (preliminary estimates). That’s half of global first- half losses insured losses. For all other major perils, also severe convective storms (SCS), losses were instead mostly below the January-June average of the previous five years (see Figure 3).\n\n**Figure 3:** Global insured losses from natural catastrophes by peril in 1H 2025 and previous half-year averages (USD bn, 2025 prices)\n\nSource: Swiss Re Institute\n\n0 10 20 30 40 50 60 70 80\n\nDroughts\n\nEuropean winter storms\n\nWildfires\n\nOther secondary perils\n\nFloods\n\nEarthquakes\n\nTropical cyclones\n\nSevere convective storms\n\n2025\n\n5-year average (2020–2024)\n\n10-year average (2015–2024)\n\n30-year average\n\n(1995–2024)\n\nDroughts\n\nOther secondary perils European winter storms Wildfires\n\nSevere convective storms Floods Tropical cyclones Earthquakes\n\nEconomic losses from natural catastrophes in the first half of 2025 were USD 135 billion, of which insurance covered about 59% (see Figure 4). That’s higher than the previous 10-year average of 42% and reflects that much of the damage caused by disasters was in US and other regions of relatively high insurance penetration, and also that wildfire is widely covered in personal property covers.\n\n**Figure 4:** Global annual insured and uninsured losses from natural catastrophes in 1H (USD bn, 2025 prices)\n\nSource: Swiss Re Institute\n\n0\n\n30\n\n60\n\n90\n\n120\n\n150\n\n180\n\nUninsured losses (left)\n\nInsured losses (left)\n\n2017 2018 2019 2020 2021 2022 2023 2024 2025 2016\n\n0%\n\n20%\n\n40%\n\n60%\n\nInsured losses as a percentage of total economic losses (right)\n\n37%\n\n43% 43%\n\n33%\n\n47% 50%\n\n39% 39%\n\n51%\n\n59%\n\nUninsured losses (left) Insured losses (left)\n\nInsured losses as a percentage of total economic losses (right)\n\nThe LA County fires were not the only major disaster in the first half. Another catastrophe was the magnitude 7.7 earthquake that struck Myanmar in March, claiming around 3 900 human lives, more than any other disaster. ¹ The quake struck densely populated regions of the country including Mandalay, its second largest city, and total economic losses are estimated to have been around USD 14 billion. There is no information on insured losses but, with low insurance penetration, these will likely be low relative to regions of higher risk protection uptake. The shock waves from the earthquake were felt as far away as in India, China and Thailand. In Bangkok, the seismic shakes caused a high-rise building under construction to collapse, claiming more lives. Bangkok sits on a basin of soft sediment that acts like as amplifier, increasing the intensity and duration of shaking and making even far-off earthquakes feel intense. For this reason, even though far away from active faults, Bangkok is exposed to loss potential from seismic activity. Preliminary estimates from the Thai General Insurance Association indicate insured losses in the country of USD 1.5 billion as a result of the earthquake.\n\nWith earthquakes, location and ground conditions matter as much as magnitude with respect to damage caused. Many cities around the world including Mexico City, LA, Beijing and Tokyo are built on sedimentary basins and face elevated seismic risk on account of potential for amplified ground motions. Without proper building codes and adequate enforcement thereof, the increased shaking can cause severe structural damage and loss of life. Modern seismic hazard assessment including localised effects is needed to accurately assess the resilience of buildings to ground shakes, and the stringency of building codes required. Strong building codes reduce damage but are not always rigorously enforced. Risk models must account for basin effects to avoid underestimating loss potential.\n\nNon-seasonal wildfires account for half of the January- June losses\n\nIn January 2025, two major wildfires burned close to 40 000 acres in LA County and destroyed more than 16 000 structures in the Palisades and Eaton neighbourhoods. ² A prolonged and dry Santa Ana wind with gusts reaching speeds of almost 100 mph, coupled with a lack of rainfall, created unseasonal and extreme fire-fuelling weather conditions. The area affected by the Palisades fire contains some of the densest concentration of high-value single-family residential property in the US. The resulting fires covered locations of high exposure, leading to exceptional losses. A very large number of structures were destroyed, more than 400% of the long- term annual average, causing high losses in what was otherwise in terms of physical events an unremarkable fire season. For instance, in year to 19 June 2025, the number of acres burned in the California and the number of fire outbreaks were just 5% and 38% of the respective long-term yearly averages.\n\nAccording to Swiss Re estimates, as of July 2025 combined insured losses from the Palisades and Eaton firestorm(s) were USD 40 billion, by far the largest global insured wildfire loss event(s) to date, and among the top 10 costliest natural catastrophe insured loss events of all perils on _sigma_ records. Total first-half insured losses for wildfire were the highest ever, surpassing even the previous full-year peaks for this peril by a significant margin (see Figure 5 below).","publish_date":null,"title":"Non-seasonal wildfires account for half of the January- June losses","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/SRI-Insights-August%202025_media_embargo_p2.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/SRI-Insights-August%202025_media_embargo_p2.pdf"},"request":{"example_id":"chart/SRI-Insights-August 2025_media_embargo_p2","product_type":"parse","source_file_path":"SRI-Insights-August 2025_media_embargo_p2.pdf"},"started_at":"2026-11-04T19:46:25.793574"} +{"completed_at":"2026-11-04T19:46:28.981038","latency_in_ms":1612,"output":{"example_id":"chart/SRI-Insights-August 2025_media_embargo_p3","grounded_pages":[],"layout_pages":[],"markdown":"3 **Swiss Re Institute** Unseasonal fires trigger above-trend first-half 2025 natural catastrophe insured losses\n\nOverall, losses from wildfires have been rising significantly in the past decade. Prior to 2015, on average wildfires across the world contributed around 1% of the total insured losses from all natural catastrophes. In the last 10 years, this has risen to 7%, the costliest periods being a two-year stretch of 2017‒18, and to a lesser extent 2020. Eight of the 10 costliest wildfire insured loss events have all occurred in the last decade, after adjusting for inflation. Wildfires are an ever-present hazard in hot and dry regions with large areas of vegetation, such as there are in North America. The single biggest loss growth driver for this peril is increased exposure in such hazardous regions. Due to the combination of high hazard and high value asset concentration, most fire losses originate in the US and particularly in California, where expansion in said regions has been high. Since 1990, exposure growth in the high-risk wildland urban interface (WUI) zones has outpaced exposure growth in non-WUI zones by a factor of 1.8 in the US, and by a factor of 1.9 in California. ³\n\nWildfires are typical for southern California after an extended period of low precipitation and in the presence of strong, dry winds. They are a natural phenomenon of the forest ecosystem, essential for restoring nutrients to the soil, clearing out decay and helping plants to reproduce. Occasionally forest fires turn into urban conflagration when they spread to urban areas, threatening human safety and property. Suburban sprawl with communities expanding into wildland increases the risk of such events. The close proximity of structures to dense vegetation allows for easier transfer of flames and embers to nearby buildings, increasing the vulnerability of the suburban environment risk. In addition, WUI development adds additional sources of ignition (sparks from machinery or trains, campfires, arcing from electrical utility lines, as well as arson).\n\nOngoing warming and drying trends, coupled with changing rainfall patterns, contribute to drier and more flammable fuels, creating conditions more conducive for ignition and spread. Rising temperatures are also extending the fire seasons, lengthening the overlap of the peak fuel dryness and the height of the Santa Ana winds season in California. The January fires in California were unusual in that they occurred in the winter season, typically the state’s wet season. Historically, the biggest wildfire loss events in California have occurred during the hot and dry late summer and autumn months, after which cooler temperatures and rains serve to\n\ndampen fire risk. Based on _sigma_ records, 99% of the wildfire losses that originate in California occur in the second half of the year.\n\nThe 2025 fires give the measure of the scale of devastation and loss potential when all ingredients for major conflagration come together: weather (drought and winds); dense urban or suburban environment; and fuel (flammable structures and dry grassland that connects the structures, providing a pathway of fire). There have been large wildfire-loss events outside of the US in recent years too, such as in Australia (as in 2019 and 2009), Canada (as in 2016), and also Europe (eg, Portugal in 2017), with all ingredients for urban conflagration present. Droughts are becoming more frequent in Europe also, extending the wildfire season and exacerbating water shortages, which hinder firefighting efforts.\n\nWildfire is a complex peril to model. Human actions influence the occurrence and damage potential of wildfires to a much greater extent than other perils. To date traditional risk evaluation methods – such as predictive or probabilistic models – have not been fully captured the complex interactions between drivers of wildfire risk, changes in fire behaviour and inter-annual variability. Wildfire remains a volatile peril and in our view, more and rising associated insured losses can be expected. ⁴\n\nSevere convective storm losses remain high though below trend\n\nBetween March and June, a number of severe thunderstorms with large hail and tornado outbreaks in the US led to global SCS-related insured losses of USD 31 billion in the first half of 2025 (see Figure 6 below). Though below the trend estimate of USD 35 billion and also the first-half loss tallies for this peril in 2023 and 2024, those SCS losses are still the 4 th costliest ever. Only a decade ago insured losses triggered by SCS were trending at around USD 20 billion.\n\nEntangled with high year-on-year volatility, SCS drive a substantial portion of global natural catastrophe insured losses. The main factors pushing losses higher are increased exposure values due to economic and population growth, urbanisation (expanding urban footprints in exposed regions) and inflation (mainly construction\n\n**Figure 5:** Global annual insured losses from wildfires (USD bn, 2025 prices)\n\nSource: Swiss Re Institute\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n35\n\n40\n\n45\n\n1H2025 2023 2021 2019 2017 2015 2013 2011 2009 2007 2005 2003 2001 1999 1997\n\nEditable field 2023 - with outline\n\nEditable field 2023 - without outline\n\nRest of the world US\n\nCedar & Old, US\n\nFort McMurray Canada\n\nTubbs, Atlas, Thomas, US\n\nPalisades, Eaton, US\n\nCamp, Woolsey US\n\nGlass, US","pages":[{"line_numbers":[],"markdown":"3 **Swiss Re Institute** Unseasonal fires trigger above-trend first-half 2025 natural catastrophe insured losses\n\nOverall, losses from wildfires have been rising significantly in the past decade. Prior to 2015, on average wildfires across the world contributed around 1% of the total insured losses from all natural catastrophes. In the last 10 years, this has risen to 7%, the costliest periods being a two-year stretch of 2017‒18, and to a lesser extent 2020. Eight of the 10 costliest wildfire insured loss events have all occurred in the last decade, after adjusting for inflation. Wildfires are an ever-present hazard in hot and dry regions with large areas of vegetation, such as there are in North America. The single biggest loss growth driver for this peril is increased exposure in such hazardous regions. Due to the combination of high hazard and high value asset concentration, most fire losses originate in the US and particularly in California, where expansion in said regions has been high. Since 1990, exposure growth in the high-risk wildland urban interface (WUI) zones has outpaced exposure growth in non-WUI zones by a factor of 1.8 in the US, and by a factor of 1.9 in California. ³\n\nWildfires are typical for southern California after an extended period of low precipitation and in the presence of strong, dry winds. They are a natural phenomenon of the forest ecosystem, essential for restoring nutrients to the soil, clearing out decay and helping plants to reproduce. Occasionally forest fires turn into urban conflagration when they spread to urban areas, threatening human safety and property. Suburban sprawl with communities expanding into wildland increases the risk of such events. The close proximity of structures to dense vegetation allows for easier transfer of flames and embers to nearby buildings, increasing the vulnerability of the suburban environment risk. In addition, WUI development adds additional sources of ignition (sparks from machinery or trains, campfires, arcing from electrical utility lines, as well as arson).\n\nOngoing warming and drying trends, coupled with changing rainfall patterns, contribute to drier and more flammable fuels, creating conditions more conducive for ignition and spread. Rising temperatures are also extending the fire seasons, lengthening the overlap of the peak fuel dryness and the height of the Santa Ana winds season in California. The January fires in California were unusual in that they occurred in the winter season, typically the state’s wet season. Historically, the biggest wildfire loss events in California have occurred during the hot and dry late summer and autumn months, after which cooler temperatures and rains serve to\n\ndampen fire risk. Based on _sigma_ records, 99% of the wildfire losses that originate in California occur in the second half of the year.\n\nThe 2025 fires give the measure of the scale of devastation and loss potential when all ingredients for major conflagration come together: weather (drought and winds); dense urban or suburban environment; and fuel (flammable structures and dry grassland that connects the structures, providing a pathway of fire). There have been large wildfire-loss events outside of the US in recent years too, such as in Australia (as in 2019 and 2009), Canada (as in 2016), and also Europe (eg, Portugal in 2017), with all ingredients for urban conflagration present. Droughts are becoming more frequent in Europe also, extending the wildfire season and exacerbating water shortages, which hinder firefighting efforts.\n\nWildfire is a complex peril to model. Human actions influence the occurrence and damage potential of wildfires to a much greater extent than other perils. To date traditional risk evaluation methods – such as predictive or probabilistic models – have not been fully captured the complex interactions between drivers of wildfire risk, changes in fire behaviour and inter-annual variability. Wildfire remains a volatile peril and in our view, more and rising associated insured losses can be expected. ⁴\n\nSevere convective storm losses remain high though below trend\n\nBetween March and June, a number of severe thunderstorms with large hail and tornado outbreaks in the US led to global SCS-related insured losses of USD 31 billion in the first half of 2025 (see Figure 6 below). Though below the trend estimate of USD 35 billion and also the first-half loss tallies for this peril in 2023 and 2024, those SCS losses are still the 4 th costliest ever. Only a decade ago insured losses triggered by SCS were trending at around USD 20 billion.\n\nEntangled with high year-on-year volatility, SCS drive a substantial portion of global natural catastrophe insured losses. The main factors pushing losses higher are increased exposure values due to economic and population growth, urbanisation (expanding urban footprints in exposed regions) and inflation (mainly construction\n\n**Figure 5:** Global annual insured losses from wildfires (USD bn, 2025 prices)\n\nSource: Swiss Re Institute\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n35\n\n40\n\n45\n\n1H2025 2023 2021 2019 2017 2015 2013 2011 2009 2007 2005 2003 2001 1999 1997\n\nEditable field 2023 - with outline\n\nEditable field 2023 - without outline\n\nRest of the world US\n\nCedar & Old, US\n\nFort McMurray Canada\n\nTubbs, Atlas, Thomas, US\n\nPalisades, Eaton, US\n\nCamp, Woolsey US\n\nGlass, US","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"3 **Swiss Re Institute** Unseasonal fires trigger above-trend first-half 2025 natural catastrophe insured losses\n\nOverall, losses from wildfires have been rising significantly in the past decade. Prior to 2015, on average wildfires across the world contributed around 1% of the total insured losses from all natural catastrophes. In the last 10 years, this has risen to 7%, the costliest periods being a two-year stretch of 2017‒18, and to a lesser extent 2020. Eight of the 10 costliest wildfire insured loss events have all occurred in the last decade, after adjusting for inflation. Wildfires are an ever-present hazard in hot and dry regions with large areas of vegetation, such as there are in North America. The single biggest loss growth driver for this peril is increased exposure in such hazardous regions. Due to the combination of high hazard and high value asset concentration, most fire losses originate in the US and particularly in California, where expansion in said regions has been high. Since 1990, exposure growth in the high-risk wildland urban interface (WUI) zones has outpaced exposure growth in non-WUI zones by a factor of 1.8 in the US, and by a factor of 1.9 in California. ³\n\nWildfires are typical for southern California after an extended period of low precipitation and in the presence of strong, dry winds. They are a natural phenomenon of the forest ecosystem, essential for restoring nutrients to the soil, clearing out decay and helping plants to reproduce. Occasionally forest fires turn into urban conflagration when they spread to urban areas, threatening human safety and property. Suburban sprawl with communities expanding into wildland increases the risk of such events. The close proximity of structures to dense vegetation allows for easier transfer of flames and embers to nearby buildings, increasing the vulnerability of the suburban environment risk. In addition, WUI development adds additional sources of ignition (sparks from machinery or trains, campfires, arcing from electrical utility lines, as well as arson).\n\nOngoing warming and drying trends, coupled with changing rainfall patterns, contribute to drier and more flammable fuels, creating conditions more conducive for ignition and spread. Rising temperatures are also extending the fire seasons, lengthening the overlap of the peak fuel dryness and the height of the Santa Ana winds season in California. The January fires in California were unusual in that they occurred in the winter season, typically the state’s wet season. Historically, the biggest wildfire loss events in California have occurred during the hot and dry late summer and autumn months, after which cooler temperatures and rains serve to\n\ndampen fire risk. Based on _sigma_ records, 99% of the wildfire losses that originate in California occur in the second half of the year.\n\nThe 2025 fires give the measure of the scale of devastation and loss potential when all ingredients for major conflagration come together: weather (drought and winds); dense urban or suburban environment; and fuel (flammable structures and dry grassland that connects the structures, providing a pathway of fire). There have been large wildfire-loss events outside of the US in recent years too, such as in Australia (as in 2019 and 2009), Canada (as in 2016), and also Europe (eg, Portugal in 2017), with all ingredients for urban conflagration present. Droughts are becoming more frequent in Europe also, extending the wildfire season and exacerbating water shortages, which hinder firefighting efforts.\n\nWildfire is a complex peril to model. Human actions influence the occurrence and damage potential of wildfires to a much greater extent than other perils. To date traditional risk evaluation methods – such as predictive or probabilistic models – have not been fully captured the complex interactions between drivers of wildfire risk, changes in fire behaviour and inter-annual variability. Wildfire remains a volatile peril and in our view, more and rising associated insured losses can be expected. ⁴\n\nSevere convective storm losses remain high though below trend\n\nBetween March and June, a number of severe thunderstorms with large hail and tornado outbreaks in the US led to global SCS-related insured losses of USD 31 billion in the first half of 2025 (see Figure 6 below). Though below the trend estimate of USD 35 billion and also the first-half loss tallies for this peril in 2023 and 2024, those SCS losses are still the 4 th costliest ever. Only a decade ago insured losses triggered by SCS were trending at around USD 20 billion.\n\nEntangled with high year-on-year volatility, SCS drive a substantial portion of global natural catastrophe insured losses. The main factors pushing losses higher are increased exposure values due to economic and population growth, urbanisation (expanding urban footprints in exposed regions) and inflation (mainly construction\n\n**Figure 5:** Global annual insured losses from wildfires (USD bn, 2025 prices)\n\nSource: Swiss Re Institute\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n35\n\n40\n\n45\n\n1H2025 2023 2021 2019 2017 2015 2013 2011 2009 2007 2005 2003 2001 1999 1997\n\nEditable field 2023 - with outline\n\nEditable field 2023 - without outline\n\nRest of the world US\n\nCedar & Old, US\n\nFort McMurray Canada\n\nTubbs, Atlas, Thomas, US\n\nPalisades, Eaton, US\n\nCamp, Woolsey US\n\nGlass, US","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/SRI-Insights-August%202025_media_embargo_p3.pdf"]},"response":{"errors":[],"extract_id":"extract_2a4546d5fa16c45ec3427a12cbbe6b36","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["3 **Swiss Re Institute** Unseasonal fires trigger above-trend first-half 2025 natural catastrophe insured losses\n\nOverall, losses from wildfires have been rising significantly in the past decade. Prior to 2015, on average wildfires across the world contributed around 1% of the total insured losses from all natural catastrophes. In the last 10 years, this has risen to 7%, the costliest periods being a two-year stretch of 2017‒18, and to a lesser extent 2020. Eight of the 10 costliest wildfire insured loss events have all occurred in the last decade, after adjusting for inflation. Wildfires are an ever-present hazard in hot and dry regions with large areas of vegetation, such as there are in North America. The single biggest loss growth driver for this peril is increased exposure in such hazardous regions. Due to the combination of high hazard and high value asset concentration, most fire losses originate in the US and particularly in California, where expansion in said regions has been high. Since 1990, exposure growth in the high-risk wildland urban interface (WUI) zones has outpaced exposure growth in non-WUI zones by a factor of 1.8 in the US, and by a factor of 1.9 in California. ³\n\nWildfires are typical for southern California after an extended period of low precipitation and in the presence of strong, dry winds. They are a natural phenomenon of the forest ecosystem, essential for restoring nutrients to the soil, clearing out decay and helping plants to reproduce. Occasionally forest fires turn into urban conflagration when they spread to urban areas, threatening human safety and property. Suburban sprawl with communities expanding into wildland increases the risk of such events. The close proximity of structures to dense vegetation allows for easier transfer of flames and embers to nearby buildings, increasing the vulnerability of the suburban environment risk. In addition, WUI development adds additional sources of ignition (sparks from machinery or trains, campfires, arcing from electrical utility lines, as well as arson).\n\nOngoing warming and drying trends, coupled with changing rainfall patterns, contribute to drier and more flammable fuels, creating conditions more conducive for ignition and spread. Rising temperatures are also extending the fire seasons, lengthening the overlap of the peak fuel dryness and the height of the Santa Ana winds season in California. The January fires in California were unusual in that they occurred in the winter season, typically the state’s wet season. Historically, the biggest wildfire loss events in California have occurred during the hot and dry late summer and autumn months, after which cooler temperatures and rains serve to\n\ndampen fire risk. Based on _sigma_ records, 99% of the wildfire losses that originate in California occur in the second half of the year.\n\nThe 2025 fires give the measure of the scale of devastation and loss potential when all ingredients for major conflagration come together: weather (drought and winds); dense urban or suburban environment; and fuel (flammable structures and dry grassland that connects the structures, providing a pathway of fire). There have been large wildfire-loss events outside of the US in recent years too, such as in Australia (as in 2019 and 2009), Canada (as in 2016), and also Europe (eg, Portugal in 2017), with all ingredients for urban conflagration present. Droughts are becoming more frequent in Europe also, extending the wildfire season and exacerbating water shortages, which hinder firefighting efforts.\n\nWildfire is a complex peril to model. Human actions influence the occurrence and damage potential of wildfires to a much greater extent than other perils. To date traditional risk evaluation methods – such as predictive or probabilistic models – have not been fully captured the complex interactions between drivers of wildfire risk, changes in fire behaviour and inter-annual variability. Wildfire remains a volatile peril and in our view, more and rising associated insured losses can be expected. ⁴\n\nSevere convective storm losses remain high though below trend\n\nBetween March and June, a number of severe thunderstorms with large hail and tornado outbreaks in the US led to global SCS-related insured losses of USD 31 billion in the first half of 2025 (see Figure 6 below). Though below the trend estimate of USD 35 billion and also the first-half loss tallies for this peril in 2023 and 2024, those SCS losses are still the 4 th costliest ever. Only a decade ago insured losses triggered by SCS were trending at around USD 20 billion.\n\nEntangled with high year-on-year volatility, SCS drive a substantial portion of global natural catastrophe insured losses. The main factors pushing losses higher are increased exposure values due to economic and population growth, urbanisation (expanding urban footprints in exposed regions) and inflation (mainly construction\n\n**Figure 5:** Global annual insured losses from wildfires (USD bn, 2025 prices)\n\nSource: Swiss Re Institute\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n35\n\n40\n\n45\n\n1H2025 2023 2021 2019 2017 2015 2013 2011 2009 2007 2005 2003 2001 1999 1997\n\nEditable field 2023 - with outline\n\nEditable field 2023 - without outline\n\nRest of the world US\n\nCedar & Old, US\n\nFort McMurray Canada\n\nTubbs, Atlas, Thomas, US\n\nPalisades, Eaton, US\n\nCamp, Woolsey US\n\nGlass, US"],"full_content":"3 **Swiss Re Institute** Unseasonal fires trigger above-trend first-half 2025 natural catastrophe insured losses\n\nOverall, losses from wildfires have been rising significantly in the past decade. Prior to 2015, on average wildfires across the world contributed around 1% of the total insured losses from all natural catastrophes. In the last 10 years, this has risen to 7%, the costliest periods being a two-year stretch of 2017‒18, and to a lesser extent 2020. Eight of the 10 costliest wildfire insured loss events have all occurred in the last decade, after adjusting for inflation. Wildfires are an ever-present hazard in hot and dry regions with large areas of vegetation, such as there are in North America. The single biggest loss growth driver for this peril is increased exposure in such hazardous regions. Due to the combination of high hazard and high value asset concentration, most fire losses originate in the US and particularly in California, where expansion in said regions has been high. Since 1990, exposure growth in the high-risk wildland urban interface (WUI) zones has outpaced exposure growth in non-WUI zones by a factor of 1.8 in the US, and by a factor of 1.9 in California. ³\n\nWildfires are typical for southern California after an extended period of low precipitation and in the presence of strong, dry winds. They are a natural phenomenon of the forest ecosystem, essential for restoring nutrients to the soil, clearing out decay and helping plants to reproduce. Occasionally forest fires turn into urban conflagration when they spread to urban areas, threatening human safety and property. Suburban sprawl with communities expanding into wildland increases the risk of such events. The close proximity of structures to dense vegetation allows for easier transfer of flames and embers to nearby buildings, increasing the vulnerability of the suburban environment risk. In addition, WUI development adds additional sources of ignition (sparks from machinery or trains, campfires, arcing from electrical utility lines, as well as arson).\n\nOngoing warming and drying trends, coupled with changing rainfall patterns, contribute to drier and more flammable fuels, creating conditions more conducive for ignition and spread. Rising temperatures are also extending the fire seasons, lengthening the overlap of the peak fuel dryness and the height of the Santa Ana winds season in California. The January fires in California were unusual in that they occurred in the winter season, typically the state’s wet season. Historically, the biggest wildfire loss events in California have occurred during the hot and dry late summer and autumn months, after which cooler temperatures and rains serve to\n\ndampen fire risk. Based on _sigma_ records, 99% of the wildfire losses that originate in California occur in the second half of the year.\n\nThe 2025 fires give the measure of the scale of devastation and loss potential when all ingredients for major conflagration come together: weather (drought and winds); dense urban or suburban environment; and fuel (flammable structures and dry grassland that connects the structures, providing a pathway of fire). There have been large wildfire-loss events outside of the US in recent years too, such as in Australia (as in 2019 and 2009), Canada (as in 2016), and also Europe (eg, Portugal in 2017), with all ingredients for urban conflagration present. Droughts are becoming more frequent in Europe also, extending the wildfire season and exacerbating water shortages, which hinder firefighting efforts.\n\nWildfire is a complex peril to model. Human actions influence the occurrence and damage potential of wildfires to a much greater extent than other perils. To date traditional risk evaluation methods – such as predictive or probabilistic models – have not been fully captured the complex interactions between drivers of wildfire risk, changes in fire behaviour and inter-annual variability. Wildfire remains a volatile peril and in our view, more and rising associated insured losses can be expected. ⁴\n\nSevere convective storm losses remain high though below trend\n\nBetween March and June, a number of severe thunderstorms with large hail and tornado outbreaks in the US led to global SCS-related insured losses of USD 31 billion in the first half of 2025 (see Figure 6 below). Though below the trend estimate of USD 35 billion and also the first-half loss tallies for this peril in 2023 and 2024, those SCS losses are still the 4 th costliest ever. Only a decade ago insured losses triggered by SCS were trending at around USD 20 billion.\n\nEntangled with high year-on-year volatility, SCS drive a substantial portion of global natural catastrophe insured losses. The main factors pushing losses higher are increased exposure values due to economic and population growth, urbanisation (expanding urban footprints in exposed regions) and inflation (mainly construction\n\n**Figure 5:** Global annual insured losses from wildfires (USD bn, 2025 prices)\n\nSource: Swiss Re Institute\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n35\n\n40\n\n45\n\n1H2025 2023 2021 2019 2017 2015 2013 2011 2009 2007 2005 2003 2001 1999 1997\n\nEditable field 2023 - with outline\n\nEditable field 2023 - without outline\n\nRest of the world US\n\nCedar & Old, US\n\nFort McMurray Canada\n\nTubbs, Atlas, Thomas, US\n\nPalisades, Eaton, US\n\nCamp, Woolsey US\n\nGlass, US","publish_date":null,"title":"Severe convective storm losses remain high though below trend","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/SRI-Insights-August%202025_media_embargo_p3.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/SRI-Insights-August%202025_media_embargo_p3.pdf"},"request":{"example_id":"chart/SRI-Insights-August 2025_media_embargo_p3","product_type":"parse","source_file_path":"SRI-Insights-August 2025_media_embargo_p3.pdf"},"started_at":"2026-11-04T19:46:27.368680"} +{"completed_at":"2026-11-04T19:46:30.153244","latency_in_ms":1563,"output":{"example_id":"chart/SRI-Insights-August 2025_media_embargo_p4","grounded_pages":[],"layout_pages":[],"markdown":"4 **Swiss Re Institute** Unseasonal fires trigger above-trend first-half 2025 natural catastrophe insured losses\n\ncosts). Post-pandemic inflation made reconstruction significantly more expensive. For example, in the US, construction costs rose by 35.64% from January 2020 to June 2025, directly impacting property claims costs. ⁴ There is also evidence that SCS activity has been increasing in specific regions and another factor contributing to losses is vulnerability of exposed assets. ⁵ Like solar power systems on roof tops, the number of installations of which have risen rapidly, increasing exposure vulnerability. ⁶ Although SCS losses have so far been below trend in 2025, there is no reason to believe that they will not continue to rise over the long run.\n\nJune marks the end of the peak of severe thunderstorm activity across the central US. On average at least 70% of the losses from this peril originate in the first half of the year and historically, first- half losses from SCS have never been below 50% of the full-year total (see Figure 7 opposite). That said, significant severe weather activity might continue through the year and large events can happen in the second half too, also outside of the US. The US is particularly prone to SCS due to its geography, but large-scale cost events can happen in Europe too, as in the case of Italy in July 2023.\n\nNot over yet: all it takes is one major hit\n\nThe second half of the year started with the lingering warming effects of a major heat dome that in late June brought temperatures in excess of 40°C in western and central Europe. ⁷ These led to wildfire outbreaks in many countries, also going into the second half. Meanwhile, torrential rains in the US led to catastrophic flash flooding in central Texas in July, another signal of what severe weather conditions could bring in the second half of the year. With the US season for SCS past its peak, the focus for the second half shifts to the North Atlantic hurricane season (TCNA), the high point typically coming in early September. Historically, second-half year insured losses have been higher but also more volatile than in the first. In the past 30 years, 60% of the total natural catastrophe insured losses originated in the second half on average (see Figure 8 opposite), mostly on account of costly TCNA activity in the third quarter (2004, 2005, 2008 and 2017 and 2022).\n\nGlobal insured losses from natural catastrophes have been growing at a long-term trend rate of 5‒7% annually in real terms (see Figure 9 below). ⁸ If this trend holds, we estimate that insured losses from natural disasters this year will approach USD 150\n\nbillion, in 2025 prices. ⁹ Our estimate of the first-half loss is past the mid-point of this outcome. With first-half insured losses above trend and catastrophe activity typically higher in the second half,\n\ntotal insured losses for the full year could exceed our projection.\n\nSecond-half losses can deviate from trend in any year depending foremost on the intensity of storms in the TCNA season. Latest meteorological forecasts predict near- to above-average activity in 2025, with three to five major (Saffir-Simpson Category 3 and above) hurricanes, compared to a long-term average (since 1950) of 3. Generally speaking, higher activity increases the chance of a severe hit. However, for insurers (and exposed communities), the key factor determining the scale of losses is _where_ a hurricane hits. A storm that hits coastal locations with high value concentration can happen in any season, regardless of how much activity is predicted. And it can take just one severe event to cause significant damage, as was the experience with Hurricane Andrew in 1992, a year of otherwise limited activity. This year’s 20 th anniversary of Hurricane Katrina serves as a further reminder that tropical cyclones, in particular major hurricanes, pose substantial risk to the east and Gulf Coasts of North America, as\n\n**Figure 6:** Global insured losses from SCS in 1H (USD bn, 2025 prices)\n\nSource: Swiss Re Institute\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\nFirst-half SCS\n\nTrend\n\n2025 2021 2017 2013 2009 2005 2001 1997\n\nEditable field 2023 - with outline\n\nEditable field 2023 - without outline\n\nTrend First half SCS\n\n**Figure 7:** Global insured losses from SCS in 1H and 2H, %\n\nSource: Swiss Re Institute\n\n0\n\n20\n\n40\n\n60\n\n80\n\n100\n\nSe\n\nFi\n\n2024 2020 2016 2012 2008 2004 2000 1996\n\nEditable field 2023 - w\n\nEditable field 2023 - w\n\nFirst half Second half\n\n**Figure 8:** Average global insured losses (USD billion), and share of losses (%), from natural catastrophes by quarter and peril (1995‒2024; 2025 prices)\n\nSource: Swiss Re Institute\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n35\n\n4Q 3Q 2Q 1Q\n\n16%\n\n44%\n\n22%\n\n18%\n\nEuropean winter storms Wildfires Other secondary\n\nFloods Earthquakes Cyclones Convective storms","pages":[{"line_numbers":[],"markdown":"4 **Swiss Re Institute** Unseasonal fires trigger above-trend first-half 2025 natural catastrophe insured losses\n\ncosts). Post-pandemic inflation made reconstruction significantly more expensive. For example, in the US, construction costs rose by 35.64% from January 2020 to June 2025, directly impacting property claims costs. ⁴ There is also evidence that SCS activity has been increasing in specific regions and another factor contributing to losses is vulnerability of exposed assets. ⁵ Like solar power systems on roof tops, the number of installations of which have risen rapidly, increasing exposure vulnerability. ⁶ Although SCS losses have so far been below trend in 2025, there is no reason to believe that they will not continue to rise over the long run.\n\nJune marks the end of the peak of severe thunderstorm activity across the central US. On average at least 70% of the losses from this peril originate in the first half of the year and historically, first- half losses from SCS have never been below 50% of the full-year total (see Figure 7 opposite). That said, significant severe weather activity might continue through the year and large events can happen in the second half too, also outside of the US. The US is particularly prone to SCS due to its geography, but large-scale cost events can happen in Europe too, as in the case of Italy in July 2023.\n\nNot over yet: all it takes is one major hit\n\nThe second half of the year started with the lingering warming effects of a major heat dome that in late June brought temperatures in excess of 40°C in western and central Europe. ⁷ These led to wildfire outbreaks in many countries, also going into the second half. Meanwhile, torrential rains in the US led to catastrophic flash flooding in central Texas in July, another signal of what severe weather conditions could bring in the second half of the year. With the US season for SCS past its peak, the focus for the second half shifts to the North Atlantic hurricane season (TCNA), the high point typically coming in early September. Historically, second-half year insured losses have been higher but also more volatile than in the first. In the past 30 years, 60% of the total natural catastrophe insured losses originated in the second half on average (see Figure 8 opposite), mostly on account of costly TCNA activity in the third quarter (2004, 2005, 2008 and 2017 and 2022).\n\nGlobal insured losses from natural catastrophes have been growing at a long-term trend rate of 5‒7% annually in real terms (see Figure 9 below). ⁸ If this trend holds, we estimate that insured losses from natural disasters this year will approach USD 150\n\nbillion, in 2025 prices. ⁹ Our estimate of the first-half loss is past the mid-point of this outcome. With first-half insured losses above trend and catastrophe activity typically higher in the second half,\n\ntotal insured losses for the full year could exceed our projection.\n\nSecond-half losses can deviate from trend in any year depending foremost on the intensity of storms in the TCNA season. Latest meteorological forecasts predict near- to above-average activity in 2025, with three to five major (Saffir-Simpson Category 3 and above) hurricanes, compared to a long-term average (since 1950) of 3. Generally speaking, higher activity increases the chance of a severe hit. However, for insurers (and exposed communities), the key factor determining the scale of losses is _where_ a hurricane hits. A storm that hits coastal locations with high value concentration can happen in any season, regardless of how much activity is predicted. And it can take just one severe event to cause significant damage, as was the experience with Hurricane Andrew in 1992, a year of otherwise limited activity. This year’s 20 th anniversary of Hurricane Katrina serves as a further reminder that tropical cyclones, in particular major hurricanes, pose substantial risk to the east and Gulf Coasts of North America, as\n\n**Figure 6:** Global insured losses from SCS in 1H (USD bn, 2025 prices)\n\nSource: Swiss Re Institute\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\nFirst-half SCS\n\nTrend\n\n2025 2021 2017 2013 2009 2005 2001 1997\n\nEditable field 2023 - with outline\n\nEditable field 2023 - without outline\n\nTrend First half SCS\n\n**Figure 7:** Global insured losses from SCS in 1H and 2H, %\n\nSource: Swiss Re Institute\n\n0\n\n20\n\n40\n\n60\n\n80\n\n100\n\nSe\n\nFi\n\n2024 2020 2016 2012 2008 2004 2000 1996\n\nEditable field 2023 - w\n\nEditable field 2023 - w\n\nFirst half Second half\n\n**Figure 8:** Average global insured losses (USD billion), and share of losses (%), from natural catastrophes by quarter and peril (1995‒2024; 2025 prices)\n\nSource: Swiss Re Institute\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n35\n\n4Q 3Q 2Q 1Q\n\n16%\n\n44%\n\n22%\n\n18%\n\nEuropean winter storms Wildfires Other secondary\n\nFloods Earthquakes Cyclones Convective storms","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"4 **Swiss Re Institute** Unseasonal fires trigger above-trend first-half 2025 natural catastrophe insured losses\n\ncosts). Post-pandemic inflation made reconstruction significantly more expensive. For example, in the US, construction costs rose by 35.64% from January 2020 to June 2025, directly impacting property claims costs. ⁴ There is also evidence that SCS activity has been increasing in specific regions and another factor contributing to losses is vulnerability of exposed assets. ⁵ Like solar power systems on roof tops, the number of installations of which have risen rapidly, increasing exposure vulnerability. ⁶ Although SCS losses have so far been below trend in 2025, there is no reason to believe that they will not continue to rise over the long run.\n\nJune marks the end of the peak of severe thunderstorm activity across the central US. On average at least 70% of the losses from this peril originate in the first half of the year and historically, first- half losses from SCS have never been below 50% of the full-year total (see Figure 7 opposite). That said, significant severe weather activity might continue through the year and large events can happen in the second half too, also outside of the US. The US is particularly prone to SCS due to its geography, but large-scale cost events can happen in Europe too, as in the case of Italy in July 2023.\n\nNot over yet: all it takes is one major hit\n\nThe second half of the year started with the lingering warming effects of a major heat dome that in late June brought temperatures in excess of 40°C in western and central Europe. ⁷ These led to wildfire outbreaks in many countries, also going into the second half. Meanwhile, torrential rains in the US led to catastrophic flash flooding in central Texas in July, another signal of what severe weather conditions could bring in the second half of the year. With the US season for SCS past its peak, the focus for the second half shifts to the North Atlantic hurricane season (TCNA), the high point typically coming in early September. Historically, second-half year insured losses have been higher but also more volatile than in the first. In the past 30 years, 60% of the total natural catastrophe insured losses originated in the second half on average (see Figure 8 opposite), mostly on account of costly TCNA activity in the third quarter (2004, 2005, 2008 and 2017 and 2022).\n\nGlobal insured losses from natural catastrophes have been growing at a long-term trend rate of 5‒7% annually in real terms (see Figure 9 below). ⁸ If this trend holds, we estimate that insured losses from natural disasters this year will approach USD 150\n\nbillion, in 2025 prices. ⁹ Our estimate of the first-half loss is past the mid-point of this outcome. With first-half insured losses above trend and catastrophe activity typically higher in the second half,\n\ntotal insured losses for the full year could exceed our projection.\n\nSecond-half losses can deviate from trend in any year depending foremost on the intensity of storms in the TCNA season. Latest meteorological forecasts predict near- to above-average activity in 2025, with three to five major (Saffir-Simpson Category 3 and above) hurricanes, compared to a long-term average (since 1950) of 3. Generally speaking, higher activity increases the chance of a severe hit. However, for insurers (and exposed communities), the key factor determining the scale of losses is _where_ a hurricane hits. A storm that hits coastal locations with high value concentration can happen in any season, regardless of how much activity is predicted. And it can take just one severe event to cause significant damage, as was the experience with Hurricane Andrew in 1992, a year of otherwise limited activity. This year’s 20 th anniversary of Hurricane Katrina serves as a further reminder that tropical cyclones, in particular major hurricanes, pose substantial risk to the east and Gulf Coasts of North America, as\n\n**Figure 6:** Global insured losses from SCS in 1H (USD bn, 2025 prices)\n\nSource: Swiss Re Institute\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\nFirst-half SCS\n\nTrend\n\n2025 2021 2017 2013 2009 2005 2001 1997\n\nEditable field 2023 - with outline\n\nEditable field 2023 - without outline\n\nTrend First half SCS\n\n**Figure 7:** Global insured losses from SCS in 1H and 2H, %\n\nSource: Swiss Re Institute\n\n0\n\n20\n\n40\n\n60\n\n80\n\n100\n\nSe\n\nFi\n\n2024 2020 2016 2012 2008 2004 2000 1996\n\nEditable field 2023 - w\n\nEditable field 2023 - w\n\nFirst half Second half\n\n**Figure 8:** Average global insured losses (USD billion), and share of losses (%), from natural catastrophes by quarter and peril (1995‒2024; 2025 prices)\n\nSource: Swiss Re Institute\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n35\n\n4Q 3Q 2Q 1Q\n\n16%\n\n44%\n\n22%\n\n18%\n\nEuropean winter storms Wildfires Other secondary\n\nFloods Earthquakes Cyclones Convective storms","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/SRI-Insights-August%202025_media_embargo_p4.pdf"]},"response":{"errors":[],"extract_id":"extract_2a6a9cecbc353beb2aefdcf763167f84","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["4 **Swiss Re Institute** Unseasonal fires trigger above-trend first-half 2025 natural catastrophe insured losses\n\ncosts). Post-pandemic inflation made reconstruction significantly more expensive. For example, in the US, construction costs rose by 35.64% from January 2020 to June 2025, directly impacting property claims costs. ⁴ There is also evidence that SCS activity has been increasing in specific regions and another factor contributing to losses is vulnerability of exposed assets. ⁵ Like solar power systems on roof tops, the number of installations of which have risen rapidly, increasing exposure vulnerability. ⁶ Although SCS losses have so far been below trend in 2025, there is no reason to believe that they will not continue to rise over the long run.\n\nJune marks the end of the peak of severe thunderstorm activity across the central US. On average at least 70% of the losses from this peril originate in the first half of the year and historically, first- half losses from SCS have never been below 50% of the full-year total (see Figure 7 opposite). That said, significant severe weather activity might continue through the year and large events can happen in the second half too, also outside of the US. The US is particularly prone to SCS due to its geography, but large-scale cost events can happen in Europe too, as in the case of Italy in July 2023.\n\nNot over yet: all it takes is one major hit\n\nThe second half of the year started with the lingering warming effects of a major heat dome that in late June brought temperatures in excess of 40°C in western and central Europe. ⁷ These led to wildfire outbreaks in many countries, also going into the second half. Meanwhile, torrential rains in the US led to catastrophic flash flooding in central Texas in July, another signal of what severe weather conditions could bring in the second half of the year. With the US season for SCS past its peak, the focus for the second half shifts to the North Atlantic hurricane season (TCNA), the high point typically coming in early September. Historically, second-half year insured losses have been higher but also more volatile than in the first. In the past 30 years, 60% of the total natural catastrophe insured losses originated in the second half on average (see Figure 8 opposite), mostly on account of costly TCNA activity in the third quarter (2004, 2005, 2008 and 2017 and 2022).\n\nGlobal insured losses from natural catastrophes have been growing at a long-term trend rate of 5‒7% annually in real terms (see Figure 9 below). ⁸ If this trend holds, we estimate that insured losses from natural disasters this year will approach USD 150\n\nbillion, in 2025 prices. ⁹ Our estimate of the first-half loss is past the mid-point of this outcome. With first-half insured losses above trend and catastrophe activity typically higher in the second half,\n\ntotal insured losses for the full year could exceed our projection.\n\nSecond-half losses can deviate from trend in any year depending foremost on the intensity of storms in the TCNA season. Latest meteorological forecasts predict near- to above-average activity in 2025, with three to five major (Saffir-Simpson Category 3 and above) hurricanes, compared to a long-term average (since 1950) of 3. Generally speaking, higher activity increases the chance of a severe hit. However, for insurers (and exposed communities), the key factor determining the scale of losses is _where_ a hurricane hits. A storm that hits coastal locations with high value concentration can happen in any season, regardless of how much activity is predicted. And it can take just one severe event to cause significant damage, as was the experience with Hurricane Andrew in 1992, a year of otherwise limited activity. This year’s 20 th anniversary of Hurricane Katrina serves as a further reminder that tropical cyclones, in particular major hurricanes, pose substantial risk to the east and Gulf Coasts of North America, as\n\n**Figure 6:** Global insured losses from SCS in 1H (USD bn, 2025 prices)\n\nSource: Swiss Re Institute\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\nFirst-half SCS\n\nTrend\n\n2025 2021 2017 2013 2009 2005 2001 1997\n\nEditable field 2023 - with outline\n\nEditable field 2023 - without outline\n\nTrend First half SCS\n\n**Figure 7:** Global insured losses from SCS in 1H and 2H, %\n\nSource: Swiss Re Institute\n\n0\n\n20\n\n40\n\n60\n\n80\n\n100\n\nSe\n\nFi\n\n2024 2020 2016 2012 2008 2004 2000 1996\n\nEditable field 2023 - w\n\nEditable field 2023 - w\n\nFirst half Second half\n\n**Figure 8:** Average global insured losses (USD billion), and share of losses (%), from natural catastrophes by quarter and peril (1995‒2024; 2025 prices)\n\nSource: Swiss Re Institute\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n35\n\n4Q 3Q 2Q 1Q\n\n16%\n\n44%\n\n22%\n\n18%\n\nEuropean winter storms Wildfires Other secondary\n\nFloods Earthquakes Cyclones Convective storms"],"full_content":"4 **Swiss Re Institute** Unseasonal fires trigger above-trend first-half 2025 natural catastrophe insured losses\n\ncosts). Post-pandemic inflation made reconstruction significantly more expensive. For example, in the US, construction costs rose by 35.64% from January 2020 to June 2025, directly impacting property claims costs. ⁴ There is also evidence that SCS activity has been increasing in specific regions and another factor contributing to losses is vulnerability of exposed assets. ⁵ Like solar power systems on roof tops, the number of installations of which have risen rapidly, increasing exposure vulnerability. ⁶ Although SCS losses have so far been below trend in 2025, there is no reason to believe that they will not continue to rise over the long run.\n\nJune marks the end of the peak of severe thunderstorm activity across the central US. On average at least 70% of the losses from this peril originate in the first half of the year and historically, first- half losses from SCS have never been below 50% of the full-year total (see Figure 7 opposite). That said, significant severe weather activity might continue through the year and large events can happen in the second half too, also outside of the US. The US is particularly prone to SCS due to its geography, but large-scale cost events can happen in Europe too, as in the case of Italy in July 2023.\n\nNot over yet: all it takes is one major hit\n\nThe second half of the year started with the lingering warming effects of a major heat dome that in late June brought temperatures in excess of 40°C in western and central Europe. ⁷ These led to wildfire outbreaks in many countries, also going into the second half. Meanwhile, torrential rains in the US led to catastrophic flash flooding in central Texas in July, another signal of what severe weather conditions could bring in the second half of the year. With the US season for SCS past its peak, the focus for the second half shifts to the North Atlantic hurricane season (TCNA), the high point typically coming in early September. Historically, second-half year insured losses have been higher but also more volatile than in the first. In the past 30 years, 60% of the total natural catastrophe insured losses originated in the second half on average (see Figure 8 opposite), mostly on account of costly TCNA activity in the third quarter (2004, 2005, 2008 and 2017 and 2022).\n\nGlobal insured losses from natural catastrophes have been growing at a long-term trend rate of 5‒7% annually in real terms (see Figure 9 below). ⁸ If this trend holds, we estimate that insured losses from natural disasters this year will approach USD 150\n\nbillion, in 2025 prices. ⁹ Our estimate of the first-half loss is past the mid-point of this outcome. With first-half insured losses above trend and catastrophe activity typically higher in the second half,\n\ntotal insured losses for the full year could exceed our projection.\n\nSecond-half losses can deviate from trend in any year depending foremost on the intensity of storms in the TCNA season. Latest meteorological forecasts predict near- to above-average activity in 2025, with three to five major (Saffir-Simpson Category 3 and above) hurricanes, compared to a long-term average (since 1950) of 3. Generally speaking, higher activity increases the chance of a severe hit. However, for insurers (and exposed communities), the key factor determining the scale of losses is _where_ a hurricane hits. A storm that hits coastal locations with high value concentration can happen in any season, regardless of how much activity is predicted. And it can take just one severe event to cause significant damage, as was the experience with Hurricane Andrew in 1992, a year of otherwise limited activity. This year’s 20 th anniversary of Hurricane Katrina serves as a further reminder that tropical cyclones, in particular major hurricanes, pose substantial risk to the east and Gulf Coasts of North America, as\n\n**Figure 6:** Global insured losses from SCS in 1H (USD bn, 2025 prices)\n\nSource: Swiss Re Institute\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\nFirst-half SCS\n\nTrend\n\n2025 2021 2017 2013 2009 2005 2001 1997\n\nEditable field 2023 - with outline\n\nEditable field 2023 - without outline\n\nTrend First half SCS\n\n**Figure 7:** Global insured losses from SCS in 1H and 2H, %\n\nSource: Swiss Re Institute\n\n0\n\n20\n\n40\n\n60\n\n80\n\n100\n\nSe\n\nFi\n\n2024 2020 2016 2012 2008 2004 2000 1996\n\nEditable field 2023 - w\n\nEditable field 2023 - w\n\nFirst half Second half\n\n**Figure 8:** Average global insured losses (USD billion), and share of losses (%), from natural catastrophes by quarter and peril (1995‒2024; 2025 prices)\n\nSource: Swiss Re Institute\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n35\n\n4Q 3Q 2Q 1Q\n\n16%\n\n44%\n\n22%\n\n18%\n\nEuropean winter storms Wildfires Other secondary\n\nFloods Earthquakes Cyclones Convective storms","publish_date":null,"title":"Not over yet: all it takes is one major hit","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/SRI-Insights-August%202025_media_embargo_p4.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/SRI-Insights-August%202025_media_embargo_p4.pdf"},"request":{"example_id":"chart/SRI-Insights-August 2025_media_embargo_p4","product_type":"parse","source_file_path":"SRI-Insights-August 2025_media_embargo_p4.pdf"},"started_at":"2026-11-04T19:46:28.589833"} +{"completed_at":"2026-11-04T19:46:30.997413","latency_in_ms":1450,"output":{"example_id":"chart/SRI-Insights-August 2025_media_embargo_p5","grounded_pages":[],"layout_pages":[],"markdown":"5 **Swiss Re Institute** Unseasonal fires trigger above-trend first-half 2025 natural catastrophe insured losses\n\nwell as the Caribbean. ¹⁰ For coastal communities, preparation and advance disaster resilience are essential to minimise the impact.\n\nBut that is not all that can lead to deviation from trend. There could be major losses from the full spectrum of perils, such as summer/autumn large scale floods across the world. Current Mediterranean sea surface temperatures are 3°C above the historic average, the warmest on record. This could lead to higher-than-usual rainfall levels in autumn, potentially triggering large-scale floods. ¹¹ There may also be more wildfires in North America and in other regions.\n\nHistorically, the great majority of wildfire losses occur in the second half when dry fuel on the ground is at its peak: as already stated, the LA County fires were unseasonal and to this end, an exception. Catastrophe losses fluctuate greatly year-on- year, mostly due to random natural weather variability. Amid year- on-year loss volatility, exposures are still rising. Long-term growth of catastrophe insured losses is mostly driven by economic development and urbanisation, often in hazardous regions such as by WUI and along the coast, and also construction costs rising more rapidly than overall inflation. Hazard intensification in changing climates like longer and harsher wildfire seasons, and more intense rainfall may alter loss seasonality patterns and add to the overall losses. All of those factors will continue to underpin steep growth in insured losses over the longer term, with some\n\nabove-average years when unpredictable severe weather hits in high exposure locations, but also some below-average ones.\n\n**Authors** Dr Chandan Banerjee Lucia Bevere Dr Shilpa Gahlot Dr Mahesh H Puttaiah The authors would like to thank Dr Erik Lindgren, Elisabeth Viktor and Dr Rogier de Jong for their contributions to the report.\n\n**Deputy managing editor** Dr Thomas Holzheu **Managing editor** Dr Jerome Jean Haegeli\n\nSwiss Re. All rights reserved. Although all the information used in this study was taken from reliable sources, Swiss Re does not accept any responsibility for the accuracy or comprehensiveness of the information given or forward looking statements made. The information provided and forward-looking statements made are for informational purposes only and in no way constitute or should be taken to reflect Swiss Re’s position, in particular in relation to any ongoing or future dispute. In no event shall Swiss Re be liable for any loss or damage arising in connection with the use of this information and readers are cautioned not to place undue reliance on forward-looking statements. Swiss Re undertakes no obligation to publicly revise or update any forward-looking statements, whether as a result of new information, future events or otherwise.\n\n**References**\n\n1 _SITUATION UPDATE No. 10 - M7.7 Mandalay Earthquake,_ AHA Centre, 23\n\nApril 2025.\n\n2 See _2025 Fire Season Incident Archive,_ CAL FIRE\n\n3 _Mapping Change in the Wildland Urban Interface_ , 1990-2020, Silivis Lab, 2025,\n\nbased on Radeloff et. al. _Rapid growth of US wildland-urban interface raises wildfire risk_ , Proceedings of the National Academy of Sciences, vol. 115, 2018.\n\n4 _7 lessons learned from the California wildfires for European insurers_ , Swiss Re,\n\n7 July 2025.\n\n5 _Producer Price Indexes_ , Bureau of Labor Statistics, June 2025.\n\n6 _The effects of climate change on hail storms,_ nature reviews earth &\n\nenvironment, vol 2, 9 February 2021.\n\n7 _sigma_ 1/2024, Natural catastrophes in 2023: gearing up for today’s and\n\ntomorrow’s weather risks, Swiss Re Institute.\n\n8 _Extreme heat grips Europe,_ World Meteorological Organization, 3 July 2025.\n\nThe term “heat dome” refers to the phenomenon of high-pressure systems that trap hot air towards the surface preventing it from rising and cooling.\n\n9 _sigma_ 1/2025, Natural catastrophes: insured losses on trend to USD 145 billion\n\nin 2025, Swiss Re Institute.\n\n10 The current trend estimate of USD 150 billion for 2025 full-year losses is at 2025\n\nprices. In _sigma_ 1/2025, published in April, we cited a figure of USD 145 billion, based on 2024 prices.\n\n11 _Hurricane Katrina: a watershed event for insurance_ , Swiss Re Institute, 23 June\n\n2025\\.\n\n**Figure 9:** Global insured losses from natural catastrophes (USD bn, 2025 prices)\n\nSource: Swiss Re Institute\n\n0\n\n50\n\n100\n\n150\n\n200\n\n2025 2023 2021 2019 2017 2015 2013 2011 2009 2007 2005 2003 2001 1999 1997\n\nEditable field 2023 - with outline\n\nEditable field 2023 - without outline\n\nFY Trend Second half First half\n\nHurricanes Harvey, Irma, Maria Earthquakes Japan & NZ, Thailand flood Hurricanes Katrina, Rita, Wilma","pages":[{"line_numbers":[],"markdown":"5 **Swiss Re Institute** Unseasonal fires trigger above-trend first-half 2025 natural catastrophe insured losses\n\nwell as the Caribbean. ¹⁰ For coastal communities, preparation and advance disaster resilience are essential to minimise the impact.\n\nBut that is not all that can lead to deviation from trend. There could be major losses from the full spectrum of perils, such as summer/autumn large scale floods across the world. Current Mediterranean sea surface temperatures are 3°C above the historic average, the warmest on record. This could lead to higher-than-usual rainfall levels in autumn, potentially triggering large-scale floods. ¹¹ There may also be more wildfires in North America and in other regions.\n\nHistorically, the great majority of wildfire losses occur in the second half when dry fuel on the ground is at its peak: as already stated, the LA County fires were unseasonal and to this end, an exception. Catastrophe losses fluctuate greatly year-on- year, mostly due to random natural weather variability. Amid year- on-year loss volatility, exposures are still rising. Long-term growth of catastrophe insured losses is mostly driven by economic development and urbanisation, often in hazardous regions such as by WUI and along the coast, and also construction costs rising more rapidly than overall inflation. Hazard intensification in changing climates like longer and harsher wildfire seasons, and more intense rainfall may alter loss seasonality patterns and add to the overall losses. All of those factors will continue to underpin steep growth in insured losses over the longer term, with some\n\nabove-average years when unpredictable severe weather hits in high exposure locations, but also some below-average ones.\n\n**Authors** Dr Chandan Banerjee Lucia Bevere Dr Shilpa Gahlot Dr Mahesh H Puttaiah The authors would like to thank Dr Erik Lindgren, Elisabeth Viktor and Dr Rogier de Jong for their contributions to the report.\n\n**Deputy managing editor** Dr Thomas Holzheu **Managing editor** Dr Jerome Jean Haegeli\n\nSwiss Re. All rights reserved. Although all the information used in this study was taken from reliable sources, Swiss Re does not accept any responsibility for the accuracy or comprehensiveness of the information given or forward looking statements made. The information provided and forward-looking statements made are for informational purposes only and in no way constitute or should be taken to reflect Swiss Re’s position, in particular in relation to any ongoing or future dispute. In no event shall Swiss Re be liable for any loss or damage arising in connection with the use of this information and readers are cautioned not to place undue reliance on forward-looking statements. Swiss Re undertakes no obligation to publicly revise or update any forward-looking statements, whether as a result of new information, future events or otherwise.\n\n**References**\n\n1 _SITUATION UPDATE No. 10 - M7.7 Mandalay Earthquake,_ AHA Centre, 23\n\nApril 2025.\n\n2 See _2025 Fire Season Incident Archive,_ CAL FIRE\n\n3 _Mapping Change in the Wildland Urban Interface_ , 1990-2020, Silivis Lab, 2025,\n\nbased on Radeloff et. al. _Rapid growth of US wildland-urban interface raises wildfire risk_ , Proceedings of the National Academy of Sciences, vol. 115, 2018.\n\n4 _7 lessons learned from the California wildfires for European insurers_ , Swiss Re,\n\n7 July 2025.\n\n5 _Producer Price Indexes_ , Bureau of Labor Statistics, June 2025.\n\n6 _The effects of climate change on hail storms,_ nature reviews earth &\n\nenvironment, vol 2, 9 February 2021.\n\n7 _sigma_ 1/2024, Natural catastrophes in 2023: gearing up for today’s and\n\ntomorrow’s weather risks, Swiss Re Institute.\n\n8 _Extreme heat grips Europe,_ World Meteorological Organization, 3 July 2025.\n\nThe term “heat dome” refers to the phenomenon of high-pressure systems that trap hot air towards the surface preventing it from rising and cooling.\n\n9 _sigma_ 1/2025, Natural catastrophes: insured losses on trend to USD 145 billion\n\nin 2025, Swiss Re Institute.\n\n10 The current trend estimate of USD 150 billion for 2025 full-year losses is at 2025\n\nprices. In _sigma_ 1/2025, published in April, we cited a figure of USD 145 billion, based on 2024 prices.\n\n11 _Hurricane Katrina: a watershed event for insurance_ , Swiss Re Institute, 23 June\n\n2025\\.\n\n**Figure 9:** Global insured losses from natural catastrophes (USD bn, 2025 prices)\n\nSource: Swiss Re Institute\n\n0\n\n50\n\n100\n\n150\n\n200\n\n2025 2023 2021 2019 2017 2015 2013 2011 2009 2007 2005 2003 2001 1999 1997\n\nEditable field 2023 - with outline\n\nEditable field 2023 - without outline\n\nFY Trend Second half First half\n\nHurricanes Harvey, Irma, Maria Earthquakes Japan & NZ, Thailand flood Hurricanes Katrina, Rita, Wilma","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"5 **Swiss Re Institute** Unseasonal fires trigger above-trend first-half 2025 natural catastrophe insured losses\n\nwell as the Caribbean. ¹⁰ For coastal communities, preparation and advance disaster resilience are essential to minimise the impact.\n\nBut that is not all that can lead to deviation from trend. There could be major losses from the full spectrum of perils, such as summer/autumn large scale floods across the world. Current Mediterranean sea surface temperatures are 3°C above the historic average, the warmest on record. This could lead to higher-than-usual rainfall levels in autumn, potentially triggering large-scale floods. ¹¹ There may also be more wildfires in North America and in other regions.\n\nHistorically, the great majority of wildfire losses occur in the second half when dry fuel on the ground is at its peak: as already stated, the LA County fires were unseasonal and to this end, an exception. Catastrophe losses fluctuate greatly year-on- year, mostly due to random natural weather variability. Amid year- on-year loss volatility, exposures are still rising. Long-term growth of catastrophe insured losses is mostly driven by economic development and urbanisation, often in hazardous regions such as by WUI and along the coast, and also construction costs rising more rapidly than overall inflation. Hazard intensification in changing climates like longer and harsher wildfire seasons, and more intense rainfall may alter loss seasonality patterns and add to the overall losses. All of those factors will continue to underpin steep growth in insured losses over the longer term, with some\n\nabove-average years when unpredictable severe weather hits in high exposure locations, but also some below-average ones.\n\n**Authors** Dr Chandan Banerjee Lucia Bevere Dr Shilpa Gahlot Dr Mahesh H Puttaiah The authors would like to thank Dr Erik Lindgren, Elisabeth Viktor and Dr Rogier de Jong for their contributions to the report.\n\n**Deputy managing editor** Dr Thomas Holzheu **Managing editor** Dr Jerome Jean Haegeli\n\nSwiss Re. All rights reserved. Although all the information used in this study was taken from reliable sources, Swiss Re does not accept any responsibility for the accuracy or comprehensiveness of the information given or forward looking statements made. The information provided and forward-looking statements made are for informational purposes only and in no way constitute or should be taken to reflect Swiss Re’s position, in particular in relation to any ongoing or future dispute. In no event shall Swiss Re be liable for any loss or damage arising in connection with the use of this information and readers are cautioned not to place undue reliance on forward-looking statements. Swiss Re undertakes no obligation to publicly revise or update any forward-looking statements, whether as a result of new information, future events or otherwise.\n\n**References**\n\n1 _SITUATION UPDATE No. 10 - M7.7 Mandalay Earthquake,_ AHA Centre, 23\n\nApril 2025.\n\n2 See _2025 Fire Season Incident Archive,_ CAL FIRE\n\n3 _Mapping Change in the Wildland Urban Interface_ , 1990-2020, Silivis Lab, 2025,\n\nbased on Radeloff et. al. _Rapid growth of US wildland-urban interface raises wildfire risk_ , Proceedings of the National Academy of Sciences, vol. 115, 2018.\n\n4 _7 lessons learned from the California wildfires for European insurers_ , Swiss Re,\n\n7 July 2025.\n\n5 _Producer Price Indexes_ , Bureau of Labor Statistics, June 2025.\n\n6 _The effects of climate change on hail storms,_ nature reviews earth &\n\nenvironment, vol 2, 9 February 2021.\n\n7 _sigma_ 1/2024, Natural catastrophes in 2023: gearing up for today’s and\n\ntomorrow’s weather risks, Swiss Re Institute.\n\n8 _Extreme heat grips Europe,_ World Meteorological Organization, 3 July 2025.\n\nThe term “heat dome” refers to the phenomenon of high-pressure systems that trap hot air towards the surface preventing it from rising and cooling.\n\n9 _sigma_ 1/2025, Natural catastrophes: insured losses on trend to USD 145 billion\n\nin 2025, Swiss Re Institute.\n\n10 The current trend estimate of USD 150 billion for 2025 full-year losses is at 2025\n\nprices. In _sigma_ 1/2025, published in April, we cited a figure of USD 145 billion, based on 2024 prices.\n\n11 _Hurricane Katrina: a watershed event for insurance_ , Swiss Re Institute, 23 June\n\n2025\\.\n\n**Figure 9:** Global insured losses from natural catastrophes (USD bn, 2025 prices)\n\nSource: Swiss Re Institute\n\n0\n\n50\n\n100\n\n150\n\n200\n\n2025 2023 2021 2019 2017 2015 2013 2011 2009 2007 2005 2003 2001 1999 1997\n\nEditable field 2023 - with outline\n\nEditable field 2023 - without outline\n\nFY Trend Second half First half\n\nHurricanes Harvey, Irma, Maria Earthquakes Japan & NZ, Thailand flood Hurricanes Katrina, Rita, Wilma","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/SRI-Insights-August%202025_media_embargo_p5.pdf"]},"response":{"errors":[],"extract_id":"extract_e35e730489f713404e9d0effbd6b888b","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["5 **Swiss Re Institute** Unseasonal fires trigger above-trend first-half 2025 natural catastrophe insured losses\n\nwell as the Caribbean. ¹⁰ For coastal communities, preparation and advance disaster resilience are essential to minimise the impact.\n\nBut that is not all that can lead to deviation from trend. There could be major losses from the full spectrum of perils, such as summer/autumn large scale floods across the world. Current Mediterranean sea surface temperatures are 3°C above the historic average, the warmest on record. This could lead to higher-than-usual rainfall levels in autumn, potentially triggering large-scale floods. ¹¹ There may also be more wildfires in North America and in other regions.\n\nHistorically, the great majority of wildfire losses occur in the second half when dry fuel on the ground is at its peak: as already stated, the LA County fires were unseasonal and to this end, an exception. Catastrophe losses fluctuate greatly year-on- year, mostly due to random natural weather variability. Amid year- on-year loss volatility, exposures are still rising. Long-term growth of catastrophe insured losses is mostly driven by economic development and urbanisation, often in hazardous regions such as by WUI and along the coast, and also construction costs rising more rapidly than overall inflation. Hazard intensification in changing climates like longer and harsher wildfire seasons, and more intense rainfall may alter loss seasonality patterns and add to the overall losses. All of those factors will continue to underpin steep growth in insured losses over the longer term, with some\n\nabove-average years when unpredictable severe weather hits in high exposure locations, but also some below-average ones.\n\n**Authors** Dr Chandan Banerjee Lucia Bevere Dr Shilpa Gahlot Dr Mahesh H Puttaiah The authors would like to thank Dr Erik Lindgren, Elisabeth Viktor and Dr Rogier de Jong for their contributions to the report.\n\n**Deputy managing editor** Dr Thomas Holzheu **Managing editor** Dr Jerome Jean Haegeli\n\nSwiss Re. All rights reserved. Although all the information used in this study was taken from reliable sources, Swiss Re does not accept any responsibility for the accuracy or comprehensiveness of the information given or forward looking statements made. The information provided and forward-looking statements made are for informational purposes only and in no way constitute or should be taken to reflect Swiss Re’s position, in particular in relation to any ongoing or future dispute. In no event shall Swiss Re be liable for any loss or damage arising in connection with the use of this information and readers are cautioned not to place undue reliance on forward-looking statements. Swiss Re undertakes no obligation to publicly revise or update any forward-looking statements, whether as a result of new information, future events or otherwise.\n\n**References**\n\n1 _SITUATION UPDATE No. 10 - M7.7 Mandalay Earthquake,_ AHA Centre, 23\n\nApril 2025.\n\n2 See _2025 Fire Season Incident Archive,_ CAL FIRE\n\n3 _Mapping Change in the Wildland Urban Interface_ , 1990-2020, Silivis Lab, 2025,\n\nbased on Radeloff et. al. _Rapid growth of US wildland-urban interface raises wildfire risk_ , Proceedings of the National Academy of Sciences, vol. 115, 2018.\n\n4 _7 lessons learned from the California wildfires for European insurers_ , Swiss Re,\n\n7 July 2025.\n\n5 _Producer Price Indexes_ , Bureau of Labor Statistics, June 2025.\n\n6 _The effects of climate change on hail storms,_ nature reviews earth &\n\nenvironment, vol 2, 9 February 2021.\n\n7 _sigma_ 1/2024, Natural catastrophes in 2023: gearing up for today’s and\n\ntomorrow’s weather risks, Swiss Re Institute.\n\n8 _Extreme heat grips Europe,_ World Meteorological Organization, 3 July 2025.\n\nThe term “heat dome” refers to the phenomenon of high-pressure systems that trap hot air towards the surface preventing it from rising and cooling.\n\n9 _sigma_ 1/2025, Natural catastrophes: insured losses on trend to USD 145 billion\n\nin 2025, Swiss Re Institute.\n\n10 The current trend estimate of USD 150 billion for 2025 full-year losses is at 2025\n\nprices. In _sigma_ 1/2025, published in April, we cited a figure of USD 145 billion, based on 2024 prices.\n\n11 _Hurricane Katrina: a watershed event for insurance_ , Swiss Re Institute, 23 June\n\n2025\\.\n\n**Figure 9:** Global insured losses from natural catastrophes (USD bn, 2025 prices)\n\nSource: Swiss Re Institute\n\n0\n\n50\n\n100\n\n150\n\n200\n\n2025 2023 2021 2019 2017 2015 2013 2011 2009 2007 2005 2003 2001 1999 1997\n\nEditable field 2023 - with outline\n\nEditable field 2023 - without outline\n\nFY Trend Second half First half\n\nHurricanes Harvey, Irma, Maria Earthquakes Japan & NZ, Thailand flood Hurricanes Katrina, Rita, Wilma"],"full_content":"5 **Swiss Re Institute** Unseasonal fires trigger above-trend first-half 2025 natural catastrophe insured losses\n\nwell as the Caribbean. ¹⁰ For coastal communities, preparation and advance disaster resilience are essential to minimise the impact.\n\nBut that is not all that can lead to deviation from trend. There could be major losses from the full spectrum of perils, such as summer/autumn large scale floods across the world. Current Mediterranean sea surface temperatures are 3°C above the historic average, the warmest on record. This could lead to higher-than-usual rainfall levels in autumn, potentially triggering large-scale floods. ¹¹ There may also be more wildfires in North America and in other regions.\n\nHistorically, the great majority of wildfire losses occur in the second half when dry fuel on the ground is at its peak: as already stated, the LA County fires were unseasonal and to this end, an exception. Catastrophe losses fluctuate greatly year-on- year, mostly due to random natural weather variability. Amid year- on-year loss volatility, exposures are still rising. Long-term growth of catastrophe insured losses is mostly driven by economic development and urbanisation, often in hazardous regions such as by WUI and along the coast, and also construction costs rising more rapidly than overall inflation. Hazard intensification in changing climates like longer and harsher wildfire seasons, and more intense rainfall may alter loss seasonality patterns and add to the overall losses. All of those factors will continue to underpin steep growth in insured losses over the longer term, with some\n\nabove-average years when unpredictable severe weather hits in high exposure locations, but also some below-average ones.\n\n**Authors** Dr Chandan Banerjee Lucia Bevere Dr Shilpa Gahlot Dr Mahesh H Puttaiah The authors would like to thank Dr Erik Lindgren, Elisabeth Viktor and Dr Rogier de Jong for their contributions to the report.\n\n**Deputy managing editor** Dr Thomas Holzheu **Managing editor** Dr Jerome Jean Haegeli\n\nSwiss Re. All rights reserved. Although all the information used in this study was taken from reliable sources, Swiss Re does not accept any responsibility for the accuracy or comprehensiveness of the information given or forward looking statements made. The information provided and forward-looking statements made are for informational purposes only and in no way constitute or should be taken to reflect Swiss Re’s position, in particular in relation to any ongoing or future dispute. In no event shall Swiss Re be liable for any loss or damage arising in connection with the use of this information and readers are cautioned not to place undue reliance on forward-looking statements. Swiss Re undertakes no obligation to publicly revise or update any forward-looking statements, whether as a result of new information, future events or otherwise.\n\n**References**\n\n1 _SITUATION UPDATE No. 10 - M7.7 Mandalay Earthquake,_ AHA Centre, 23\n\nApril 2025.\n\n2 See _2025 Fire Season Incident Archive,_ CAL FIRE\n\n3 _Mapping Change in the Wildland Urban Interface_ , 1990-2020, Silivis Lab, 2025,\n\nbased on Radeloff et. al. _Rapid growth of US wildland-urban interface raises wildfire risk_ , Proceedings of the National Academy of Sciences, vol. 115, 2018.\n\n4 _7 lessons learned from the California wildfires for European insurers_ , Swiss Re,\n\n7 July 2025.\n\n5 _Producer Price Indexes_ , Bureau of Labor Statistics, June 2025.\n\n6 _The effects of climate change on hail storms,_ nature reviews earth &\n\nenvironment, vol 2, 9 February 2021.\n\n7 _sigma_ 1/2024, Natural catastrophes in 2023: gearing up for today’s and\n\ntomorrow’s weather risks, Swiss Re Institute.\n\n8 _Extreme heat grips Europe,_ World Meteorological Organization, 3 July 2025.\n\nThe term “heat dome” refers to the phenomenon of high-pressure systems that trap hot air towards the surface preventing it from rising and cooling.\n\n9 _sigma_ 1/2025, Natural catastrophes: insured losses on trend to USD 145 billion\n\nin 2025, Swiss Re Institute.\n\n10 The current trend estimate of USD 150 billion for 2025 full-year losses is at 2025\n\nprices. In _sigma_ 1/2025, published in April, we cited a figure of USD 145 billion, based on 2024 prices.\n\n11 _Hurricane Katrina: a watershed event for insurance_ , Swiss Re Institute, 23 June\n\n2025\\.\n\n**Figure 9:** Global insured losses from natural catastrophes (USD bn, 2025 prices)\n\nSource: Swiss Re Institute\n\n0\n\n50\n\n100\n\n150\n\n200\n\n2025 2023 2021 2019 2017 2015 2013 2011 2009 2007 2005 2003 2001 1999 1997\n\nEditable field 2023 - with outline\n\nEditable field 2023 - without outline\n\nFY Trend Second half First half\n\nHurricanes Harvey, Irma, Maria Earthquakes Japan & NZ, Thailand flood Hurricanes Katrina, Rita, Wilma","publish_date":null,"title":"and Dr Rogier de Jong for their contributions to the report.","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/SRI-Insights-August%202025_media_embargo_p5.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/SRI-Insights-August%202025_media_embargo_p5.pdf"},"request":{"example_id":"chart/SRI-Insights-August 2025_media_embargo_p5","product_type":"parse","source_file_path":"SRI-Insights-August 2025_media_embargo_p5.pdf"},"started_at":"2026-11-04T19:46:29.546799"} +{"completed_at":"2026-11-04T19:46:34.985872","latency_in_ms":3230,"output":{"example_id":"chart/SSIF-Visa-Report_p6","grounded_pages":[],"layout_pages":[],"markdown":"Sprott Student Investment Fund|6\n\nBUYER POWER | HIGH 4 The global payments industry is highly competitive, and the limited number of buyers are constantly looking for the best fees and services a company can offer. Institutions do not face high switching costs when changing providers. With companies offering rebates and incentives to gain market share, Visa will have to offer similar services to not lose their customers. This will put downward pressure on Visa’s margins in the long run. With Visa’s growing network effect along with product and service innovation, Visa can continue to lower the bargaining power of buyers. More than that, product innovation will also lock in buyers, reducing their bargaining power. SUPPLIER POWER | LOW 2 The supplier power Visa faces is low. Visa has most of its software developed internally, meaning there is little dependence on outside firms. For the suppliers Visa does depend on, products are similar, and switching costs are relatively low. Moreover, suppliers do not hold Visa back from future innovation and growth. MARKET TRENDS\n\nINDUSTRY TRENDS\n\nKEY TREND I | CONTINUED MOVE AWAY FROM CASH The COVID-19 pandemic has accelerated the move away from cash. Across all age-groups, cash use has continued to decrease relative to prior years. Furthermore, cash as a preferred payment\n\ninstrument across all age groups has declined 5% year over year in 2020. Focusing in on who predominately uses cash, it is the 55+ demographic. As millennials and Generation Z become the largest spending demographic, the gross volume of cash transactions will continue to decrease. Additionally, the 55+ age demographic holds the most cash in their pocket, purse, or wallet at once. The number of in- person bill payments decreased significantly over the pandemic. From 2016 to 2019, in-person payments dropped from 92% to 87%, a decrease of 5%. In contrast over 2020, in-person payments dropped 7% alone down to 80%. Less in-person spending is a direct hit to the amount of cash being used to complete transactions. This drastic shift came down to two factors: lockdowns resulting in less in-person shopping, and fewer number of overall in-person payments a consumer made. ²³\n\n23 https://www.frbsf.org/cash/publications/fed-notes/2021/may/2021-findings-from-the-diary-of-consumer-payment-choice/\n\n0%\n\n5%\n\n10%\n\n15%\n\n20%\n\n25%\n\n30%\n\n35%\n\n40%\n\n45%\n\n50%\n\n2016 2017 2018 2019 2020\n\n### Payment Instruments Preffered for\n\n### Payment\n\nCash Credit Debit\n\nSource: FRBSF","pages":[{"line_numbers":[],"markdown":"Sprott Student Investment Fund|6\n\nBUYER POWER | HIGH 4 The global payments industry is highly competitive, and the limited number of buyers are constantly looking for the best fees and services a company can offer. Institutions do not face high switching costs when changing providers. With companies offering rebates and incentives to gain market share, Visa will have to offer similar services to not lose their customers. This will put downward pressure on Visa’s margins in the long run. With Visa’s growing network effect along with product and service innovation, Visa can continue to lower the bargaining power of buyers. More than that, product innovation will also lock in buyers, reducing their bargaining power. SUPPLIER POWER | LOW 2 The supplier power Visa faces is low. Visa has most of its software developed internally, meaning there is little dependence on outside firms. For the suppliers Visa does depend on, products are similar, and switching costs are relatively low. Moreover, suppliers do not hold Visa back from future innovation and growth. MARKET TRENDS\n\nINDUSTRY TRENDS\n\nKEY TREND I | CONTINUED MOVE AWAY FROM CASH The COVID-19 pandemic has accelerated the move away from cash. Across all age-groups, cash use has continued to decrease relative to prior years. Furthermore, cash as a preferred payment\n\ninstrument across all age groups has declined 5% year over year in 2020. Focusing in on who predominately uses cash, it is the 55+ demographic. As millennials and Generation Z become the largest spending demographic, the gross volume of cash transactions will continue to decrease. Additionally, the 55+ age demographic holds the most cash in their pocket, purse, or wallet at once. The number of in- person bill payments decreased significantly over the pandemic. From 2016 to 2019, in-person payments dropped from 92% to 87%, a decrease of 5%. In contrast over 2020, in-person payments dropped 7% alone down to 80%. Less in-person spending is a direct hit to the amount of cash being used to complete transactions. This drastic shift came down to two factors: lockdowns resulting in less in-person shopping, and fewer number of overall in-person payments a consumer made. ²³\n\n23 https://www.frbsf.org/cash/publications/fed-notes/2021/may/2021-findings-from-the-diary-of-consumer-payment-choice/\n\n0%\n\n5%\n\n10%\n\n15%\n\n20%\n\n25%\n\n30%\n\n35%\n\n40%\n\n45%\n\n50%\n\n2016 2017 2018 2019 2020\n\n### Payment Instruments Preffered for\n\n### Payment\n\nCash Credit Debit\n\nSource: FRBSF","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Sprott Student Investment Fund|6\n\nBUYER POWER | HIGH 4 The global payments industry is highly competitive, and the limited number of buyers are constantly looking for the best fees and services a company can offer. Institutions do not face high switching costs when changing providers. With companies offering rebates and incentives to gain market share, Visa will have to offer similar services to not lose their customers. This will put downward pressure on Visa’s margins in the long run. With Visa’s growing network effect along with product and service innovation, Visa can continue to lower the bargaining power of buyers. More than that, product innovation will also lock in buyers, reducing their bargaining power. SUPPLIER POWER | LOW 2 The supplier power Visa faces is low. Visa has most of its software developed internally, meaning there is little dependence on outside firms. For the suppliers Visa does depend on, products are similar, and switching costs are relatively low. Moreover, suppliers do not hold Visa back from future innovation and growth. MARKET TRENDS\n\nINDUSTRY TRENDS\n\nKEY TREND I | CONTINUED MOVE AWAY FROM CASH The COVID-19 pandemic has accelerated the move away from cash. Across all age-groups, cash use has continued to decrease relative to prior years. Furthermore, cash as a preferred payment\n\ninstrument across all age groups has declined 5% year over year in 2020. Focusing in on who predominately uses cash, it is the 55+ demographic. As millennials and Generation Z become the largest spending demographic, the gross volume of cash transactions will continue to decrease. Additionally, the 55+ age demographic holds the most cash in their pocket, purse, or wallet at once. The number of in- person bill payments decreased significantly over the pandemic. From 2016 to 2019, in-person payments dropped from 92% to 87%, a decrease of 5%. In contrast over 2020, in-person payments dropped 7% alone down to 80%. Less in-person spending is a direct hit to the amount of cash being used to complete transactions. This drastic shift came down to two factors: lockdowns resulting in less in-person shopping, and fewer number of overall in-person payments a consumer made. ²³\n\n23 https://www.frbsf.org/cash/publications/fed-notes/2021/may/2021-findings-from-the-diary-of-consumer-payment-choice/\n\n0%\n\n5%\n\n10%\n\n15%\n\n20%\n\n25%\n\n30%\n\n35%\n\n40%\n\n45%\n\n50%\n\n2016 2017 2018 2019 2020\n\n### Payment Instruments Preffered for\n\n### Payment\n\nCash Credit Debit\n\nSource: FRBSF","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/SSIF-Visa-Report_p6.pdf"]},"response":{"errors":[],"extract_id":"extract_c3fcef6ee0a89debc8dd7d4bc236e4f8","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Sprott Student Investment Fund|6\n\nBUYER POWER | HIGH 4 The global payments industry is highly competitive, and the limited number of buyers are constantly looking for the best fees and services a company can offer. Institutions do not face high switching costs when changing providers. With companies offering rebates and incentives to gain market share, Visa will have to offer similar services to not lose their customers. This will put downward pressure on Visa’s margins in the long run. With Visa’s growing network effect along with product and service innovation, Visa can continue to lower the bargaining power of buyers. More than that, product innovation will also lock in buyers, reducing their bargaining power. SUPPLIER POWER | LOW 2 The supplier power Visa faces is low. Visa has most of its software developed internally, meaning there is little dependence on outside firms. For the suppliers Visa does depend on, products are similar, and switching costs are relatively low. Moreover, suppliers do not hold Visa back from future innovation and growth. MARKET TRENDS\n\nINDUSTRY TRENDS\n\nKEY TREND I | CONTINUED MOVE AWAY FROM CASH The COVID-19 pandemic has accelerated the move away from cash. Across all age-groups, cash use has continued to decrease relative to prior years. Furthermore, cash as a preferred payment\n\ninstrument across all age groups has declined 5% year over year in 2020. Focusing in on who predominately uses cash, it is the 55+ demographic. As millennials and Generation Z become the largest spending demographic, the gross volume of cash transactions will continue to decrease. Additionally, the 55+ age demographic holds the most cash in their pocket, purse, or wallet at once. The number of in- person bill payments decreased significantly over the pandemic. From 2016 to 2019, in-person payments dropped from 92% to 87%, a decrease of 5%. In contrast over 2020, in-person payments dropped 7% alone down to 80%. Less in-person spending is a direct hit to the amount of cash being used to complete transactions. This drastic shift came down to two factors: lockdowns resulting in less in-person shopping, and fewer number of overall in-person payments a consumer made. ²³\n\n23 https://www.frbsf.org/cash/publications/fed-notes/2021/may/2021-findings-from-the-diary-of-consumer-payment-choice/\n\n0%\n\n5%\n\n10%\n\n15%\n\n20%\n\n25%\n\n30%\n\n35%\n\n40%\n\n45%\n\n50%\n\n2016 2017 2018 2019 2020\n\n### Payment Instruments Preffered for\n\n### Payment\n\nCash Credit Debit\n\nSource: FRBSF"],"full_content":"Sprott Student Investment Fund|6\n\nBUYER POWER | HIGH 4 The global payments industry is highly competitive, and the limited number of buyers are constantly looking for the best fees and services a company can offer. Institutions do not face high switching costs when changing providers. With companies offering rebates and incentives to gain market share, Visa will have to offer similar services to not lose their customers. This will put downward pressure on Visa’s margins in the long run. With Visa’s growing network effect along with product and service innovation, Visa can continue to lower the bargaining power of buyers. More than that, product innovation will also lock in buyers, reducing their bargaining power. SUPPLIER POWER | LOW 2 The supplier power Visa faces is low. Visa has most of its software developed internally, meaning there is little dependence on outside firms. For the suppliers Visa does depend on, products are similar, and switching costs are relatively low. Moreover, suppliers do not hold Visa back from future innovation and growth. MARKET TRENDS\n\nINDUSTRY TRENDS\n\nKEY TREND I | CONTINUED MOVE AWAY FROM CASH The COVID-19 pandemic has accelerated the move away from cash. Across all age-groups, cash use has continued to decrease relative to prior years. Furthermore, cash as a preferred payment\n\ninstrument across all age groups has declined 5% year over year in 2020. Focusing in on who predominately uses cash, it is the 55+ demographic. As millennials and Generation Z become the largest spending demographic, the gross volume of cash transactions will continue to decrease. Additionally, the 55+ age demographic holds the most cash in their pocket, purse, or wallet at once. The number of in- person bill payments decreased significantly over the pandemic. From 2016 to 2019, in-person payments dropped from 92% to 87%, a decrease of 5%. In contrast over 2020, in-person payments dropped 7% alone down to 80%. Less in-person spending is a direct hit to the amount of cash being used to complete transactions. This drastic shift came down to two factors: lockdowns resulting in less in-person shopping, and fewer number of overall in-person payments a consumer made. ²³\n\n23 https://www.frbsf.org/cash/publications/fed-notes/2021/may/2021-findings-from-the-diary-of-consumer-payment-choice/\n\n0%\n\n5%\n\n10%\n\n15%\n\n20%\n\n25%\n\n30%\n\n35%\n\n40%\n\n45%\n\n50%\n\n2016 2017 2018 2019 2020\n\n### Payment Instruments Preffered for\n\n### Payment\n\nCash Credit Debit\n\nSource: FRBSF","publish_date":null,"title":"Payment Instruments Preffered for Payment","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/SSIF-Visa-Report_p6.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/SSIF-Visa-Report_p6.pdf"},"request":{"example_id":"chart/SSIF-Visa-Report_p6","product_type":"parse","source_file_path":"SSIF-Visa-Report_p6.pdf"},"started_at":"2026-11-04T19:46:31.755574"} +{"completed_at":"2026-11-04T19:46:36.867407","latency_in_ms":3891,"output":{"example_id":"chart/SSIF-Visa-Report_p7","grounded_pages":[],"layout_pages":[],"markdown":"Sprott Student Investment Fund|7\n\nKEY TREND II | ECOMMERCE GROWTH Expanding upon Key Trend I, we have seen tremendous growth in the Ecommerce space over recent years. Ecommerce sales reached nearly $4.3 trillion in 2020 and are projected to climb to $4.9 trillion in 2021. With public health measures such as lockdowns and social distancing measures making it difficult to shop in store, Ecommerce has seen a euphoric rise over the past 18 months. From 2019 to 2020, Ecommerce volumes grew 27.6% and as an overall percentage of retail sales grew from 13.6% in 2019 to 18.0% in 2020. By 2021, this number is projected to reach 19.5%, a 45.8% increase in only two years. While this growth is not expected to continue at the same rate, the trend is not stopping. From 2021-2024, Ecommerce is expected to grow at an annualized rate of 10.55%. By 2024, retail Ecommerce is expected to generate $6.4 trillion in sales worldwide which equates to 21.8% of total retail sales. This trend is also being seen around the globe. Latin American Ecommerce sales have grown 36.7% in 2020, North American sales have jumped 31.8%, and sales within Asia are up 26.4%. Mobile commerce is growing at an even faster rate with a compound annual growth rate forecast of 25.5%. Three quarters of consumers report they buy from their smartphone because it takes less time. ²⁴\n\nSource: Shopify Global Ecommerce Statistics\n\nKEY TREND III | INCREASE IN DIGITAL FRAUD ATTEMPTS Digital fraud attempts have increased a staggering 25% in the US and 23% in Canada over the first four months of 2021 in comparison to the last four months of 2020. While a variety of industries were hit by this increase, it is the Financial Services industry hit the hardest with an increase of nearly 150%. This industry is sought after as it is where the highest value transactions take place. The most common types of fraud in the industry are identity theft, first-party application fraud, and account takeover. With this trend increasing as more and more consumers move online to make purchases, Visa will be required to increase their security and fraud services. This added expense could potentially cut into Visa’s margins and hurt their bottom line. ²⁵\n\n24 https://www.shopify.ca/enterprise/global-ecommerce-statistics 25 https://www.transunion.com/blog/global-fraud-trends-Q2-2021?utm\\_campaign=q2-quarterly-fraud- report&utm\\_content=blog&utm\\_medium=press-release&utm\\_source=press-release&utmsource=press-release\n\n$3,354\n\n$4,280\n\n$4,891\n\n$5,424\n\n$5,908\n\n$6,388\n\n13\\.6%\n\n18\\.0% 19.5% 20.4% 21.1% 21.8%\n\n0\\.0%\n\n10\\.0%\n\n20\\.0%\n\n30\\.0%\n\n40\\.0%\n\n50\\.0%\n\n60\\.0%\n\n70\\.0%\n\n$-\n\n$1,000\n\n$2,000\n\n$3,000\n\n$4,000\n\n$5,000\n\n$6,000\n\n$7,000\n\n2019 2020 2021 2022 2023 2024\n\n### Retail Ecommerce Sales Worldwide, 2019-2024\n\n### (in trillions, and % of total retail sales)\n\nRetail Ecommerce Sales % of total retail sales","pages":[{"line_numbers":[],"markdown":"Sprott Student Investment Fund|7\n\nKEY TREND II | ECOMMERCE GROWTH Expanding upon Key Trend I, we have seen tremendous growth in the Ecommerce space over recent years. Ecommerce sales reached nearly $4.3 trillion in 2020 and are projected to climb to $4.9 trillion in 2021. With public health measures such as lockdowns and social distancing measures making it difficult to shop in store, Ecommerce has seen a euphoric rise over the past 18 months. From 2019 to 2020, Ecommerce volumes grew 27.6% and as an overall percentage of retail sales grew from 13.6% in 2019 to 18.0% in 2020. By 2021, this number is projected to reach 19.5%, a 45.8% increase in only two years. While this growth is not expected to continue at the same rate, the trend is not stopping. From 2021-2024, Ecommerce is expected to grow at an annualized rate of 10.55%. By 2024, retail Ecommerce is expected to generate $6.4 trillion in sales worldwide which equates to 21.8% of total retail sales. This trend is also being seen around the globe. Latin American Ecommerce sales have grown 36.7% in 2020, North American sales have jumped 31.8%, and sales within Asia are up 26.4%. Mobile commerce is growing at an even faster rate with a compound annual growth rate forecast of 25.5%. Three quarters of consumers report they buy from their smartphone because it takes less time. ²⁴\n\nSource: Shopify Global Ecommerce Statistics\n\nKEY TREND III | INCREASE IN DIGITAL FRAUD ATTEMPTS Digital fraud attempts have increased a staggering 25% in the US and 23% in Canada over the first four months of 2021 in comparison to the last four months of 2020. While a variety of industries were hit by this increase, it is the Financial Services industry hit the hardest with an increase of nearly 150%. This industry is sought after as it is where the highest value transactions take place. The most common types of fraud in the industry are identity theft, first-party application fraud, and account takeover. With this trend increasing as more and more consumers move online to make purchases, Visa will be required to increase their security and fraud services. This added expense could potentially cut into Visa’s margins and hurt their bottom line. ²⁵\n\n24 https://www.shopify.ca/enterprise/global-ecommerce-statistics 25 https://www.transunion.com/blog/global-fraud-trends-Q2-2021?utm\\_campaign=q2-quarterly-fraud- report&utm\\_content=blog&utm\\_medium=press-release&utm\\_source=press-release&utmsource=press-release\n\n$3,354\n\n$4,280\n\n$4,891\n\n$5,424\n\n$5,908\n\n$6,388\n\n13\\.6%\n\n18\\.0% 19.5% 20.4% 21.1% 21.8%\n\n0\\.0%\n\n10\\.0%\n\n20\\.0%\n\n30\\.0%\n\n40\\.0%\n\n50\\.0%\n\n60\\.0%\n\n70\\.0%\n\n$-\n\n$1,000\n\n$2,000\n\n$3,000\n\n$4,000\n\n$5,000\n\n$6,000\n\n$7,000\n\n2019 2020 2021 2022 2023 2024\n\n### Retail Ecommerce Sales Worldwide, 2019-2024\n\n### (in trillions, and % of total retail sales)\n\nRetail Ecommerce Sales % of total retail sales","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Sprott Student Investment Fund|7\n\nKEY TREND II | ECOMMERCE GROWTH Expanding upon Key Trend I, we have seen tremendous growth in the Ecommerce space over recent years. Ecommerce sales reached nearly $4.3 trillion in 2020 and are projected to climb to $4.9 trillion in 2021. With public health measures such as lockdowns and social distancing measures making it difficult to shop in store, Ecommerce has seen a euphoric rise over the past 18 months. From 2019 to 2020, Ecommerce volumes grew 27.6% and as an overall percentage of retail sales grew from 13.6% in 2019 to 18.0% in 2020. By 2021, this number is projected to reach 19.5%, a 45.8% increase in only two years. While this growth is not expected to continue at the same rate, the trend is not stopping. From 2021-2024, Ecommerce is expected to grow at an annualized rate of 10.55%. By 2024, retail Ecommerce is expected to generate $6.4 trillion in sales worldwide which equates to 21.8% of total retail sales. This trend is also being seen around the globe. Latin American Ecommerce sales have grown 36.7% in 2020, North American sales have jumped 31.8%, and sales within Asia are up 26.4%. Mobile commerce is growing at an even faster rate with a compound annual growth rate forecast of 25.5%. Three quarters of consumers report they buy from their smartphone because it takes less time. ²⁴\n\nSource: Shopify Global Ecommerce Statistics\n\nKEY TREND III | INCREASE IN DIGITAL FRAUD ATTEMPTS Digital fraud attempts have increased a staggering 25% in the US and 23% in Canada over the first four months of 2021 in comparison to the last four months of 2020. While a variety of industries were hit by this increase, it is the Financial Services industry hit the hardest with an increase of nearly 150%. This industry is sought after as it is where the highest value transactions take place. The most common types of fraud in the industry are identity theft, first-party application fraud, and account takeover. With this trend increasing as more and more consumers move online to make purchases, Visa will be required to increase their security and fraud services. This added expense could potentially cut into Visa’s margins and hurt their bottom line. ²⁵\n\n24 https://www.shopify.ca/enterprise/global-ecommerce-statistics 25 https://www.transunion.com/blog/global-fraud-trends-Q2-2021?utm\\_campaign=q2-quarterly-fraud- report&utm\\_content=blog&utm\\_medium=press-release&utm\\_source=press-release&utmsource=press-release\n\n$3,354\n\n$4,280\n\n$4,891\n\n$5,424\n\n$5,908\n\n$6,388\n\n13\\.6%\n\n18\\.0% 19.5% 20.4% 21.1% 21.8%\n\n0\\.0%\n\n10\\.0%\n\n20\\.0%\n\n30\\.0%\n\n40\\.0%\n\n50\\.0%\n\n60\\.0%\n\n70\\.0%\n\n$-\n\n$1,000\n\n$2,000\n\n$3,000\n\n$4,000\n\n$5,000\n\n$6,000\n\n$7,000\n\n2019 2020 2021 2022 2023 2024\n\n### Retail Ecommerce Sales Worldwide, 2019-2024\n\n### (in trillions, and % of total retail sales)\n\nRetail Ecommerce Sales % of total retail sales","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/SSIF-Visa-Report_p7.pdf"]},"response":{"errors":[],"extract_id":"extract_8f33fef2c92cdf95e9e4be0e7f321b7c","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Sprott Student Investment Fund|7\n\nKEY TREND II | ECOMMERCE GROWTH Expanding upon Key Trend I, we have seen tremendous growth in the Ecommerce space over recent years. Ecommerce sales reached nearly $4.3 trillion in 2020 and are projected to climb to $4.9 trillion in 2021. With public health measures such as lockdowns and social distancing measures making it difficult to shop in store, Ecommerce has seen a euphoric rise over the past 18 months. From 2019 to 2020, Ecommerce volumes grew 27.6% and as an overall percentage of retail sales grew from 13.6% in 2019 to 18.0% in 2020. By 2021, this number is projected to reach 19.5%, a 45.8% increase in only two years. While this growth is not expected to continue at the same rate, the trend is not stopping. From 2021-2024, Ecommerce is expected to grow at an annualized rate of 10.55%. By 2024, retail Ecommerce is expected to generate $6.4 trillion in sales worldwide which equates to 21.8% of total retail sales. This trend is also being seen around the globe. Latin American Ecommerce sales have grown 36.7% in 2020, North American sales have jumped 31.8%, and sales within Asia are up 26.4%. Mobile commerce is growing at an even faster rate with a compound annual growth rate forecast of 25.5%. Three quarters of consumers report they buy from their smartphone because it takes less time. ²⁴\n\nSource: Shopify Global Ecommerce Statistics\n\nKEY TREND III | INCREASE IN DIGITAL FRAUD ATTEMPTS Digital fraud attempts have increased a staggering 25% in the US and 23% in Canada over the first four months of 2021 in comparison to the last four months of 2020. While a variety of industries were hit by this increase, it is the Financial Services industry hit the hardest with an increase of nearly 150%. This industry is sought after as it is where the highest value transactions take place. The most common types of fraud in the industry are identity theft, first-party application fraud, and account takeover. With this trend increasing as more and more consumers move online to make purchases, Visa will be required to increase their security and fraud services. This added expense could potentially cut into Visa’s margins and hurt their bottom line. ²⁵\n\n24 https://www.shopify.ca/enterprise/global-ecommerce-statistics 25 https://www.transunion.com/blog/global-fraud-trends-Q2-2021?utm\\_campaign=q2-quarterly-fraud- report&utm\\_content=blog&utm\\_medium=press-release&utm\\_source=press-release&utmsource=press-release\n\n$3,354\n\n$4,280\n\n$4,891\n\n$5,424\n\n$5,908\n\n$6,388\n\n13\\.6%\n\n18\\.0% 19.5% 20.4% 21.1% 21.8%\n\n0\\.0%\n\n10\\.0%\n\n20\\.0%\n\n30\\.0%\n\n40\\.0%\n\n50\\.0%\n\n60\\.0%\n\n70\\.0%\n\n$-\n\n$1,000\n\n$2,000\n\n$3,000\n\n$4,000\n\n$5,000\n\n$6,000\n\n$7,000\n\n2019 2020 2021 2022 2023 2024\n\n### Retail Ecommerce Sales Worldwide, 2019-2024\n\n### (in trillions, and % of total retail sales)\n\nRetail Ecommerce Sales % of total retail sales"],"full_content":"Sprott Student Investment Fund|7\n\nKEY TREND II | ECOMMERCE GROWTH Expanding upon Key Trend I, we have seen tremendous growth in the Ecommerce space over recent years. Ecommerce sales reached nearly $4.3 trillion in 2020 and are projected to climb to $4.9 trillion in 2021. With public health measures such as lockdowns and social distancing measures making it difficult to shop in store, Ecommerce has seen a euphoric rise over the past 18 months. From 2019 to 2020, Ecommerce volumes grew 27.6% and as an overall percentage of retail sales grew from 13.6% in 2019 to 18.0% in 2020. By 2021, this number is projected to reach 19.5%, a 45.8% increase in only two years. While this growth is not expected to continue at the same rate, the trend is not stopping. From 2021-2024, Ecommerce is expected to grow at an annualized rate of 10.55%. By 2024, retail Ecommerce is expected to generate $6.4 trillion in sales worldwide which equates to 21.8% of total retail sales. This trend is also being seen around the globe. Latin American Ecommerce sales have grown 36.7% in 2020, North American sales have jumped 31.8%, and sales within Asia are up 26.4%. Mobile commerce is growing at an even faster rate with a compound annual growth rate forecast of 25.5%. Three quarters of consumers report they buy from their smartphone because it takes less time. ²⁴\n\nSource: Shopify Global Ecommerce Statistics\n\nKEY TREND III | INCREASE IN DIGITAL FRAUD ATTEMPTS Digital fraud attempts have increased a staggering 25% in the US and 23% in Canada over the first four months of 2021 in comparison to the last four months of 2020. While a variety of industries were hit by this increase, it is the Financial Services industry hit the hardest with an increase of nearly 150%. This industry is sought after as it is where the highest value transactions take place. The most common types of fraud in the industry are identity theft, first-party application fraud, and account takeover. With this trend increasing as more and more consumers move online to make purchases, Visa will be required to increase their security and fraud services. This added expense could potentially cut into Visa’s margins and hurt their bottom line. ²⁵\n\n24 https://www.shopify.ca/enterprise/global-ecommerce-statistics 25 https://www.transunion.com/blog/global-fraud-trends-Q2-2021?utm\\_campaign=q2-quarterly-fraud- report&utm\\_content=blog&utm\\_medium=press-release&utm\\_source=press-release&utmsource=press-release\n\n$3,354\n\n$4,280\n\n$4,891\n\n$5,424\n\n$5,908\n\n$6,388\n\n13\\.6%\n\n18\\.0% 19.5% 20.4% 21.1% 21.8%\n\n0\\.0%\n\n10\\.0%\n\n20\\.0%\n\n30\\.0%\n\n40\\.0%\n\n50\\.0%\n\n60\\.0%\n\n70\\.0%\n\n$-\n\n$1,000\n\n$2,000\n\n$3,000\n\n$4,000\n\n$5,000\n\n$6,000\n\n$7,000\n\n2019 2020 2021 2022 2023 2024\n\n### Retail Ecommerce Sales Worldwide, 2019-2024\n\n### (in trillions, and % of total retail sales)\n\nRetail Ecommerce Sales % of total retail sales","publish_date":null,"title":"Retail Ecommerce Sales Worldwide, 2019-2024 (in trillions, and % of total retail sales)","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/SSIF-Visa-Report_p7.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/SSIF-Visa-Report_p7.pdf"},"request":{"example_id":"chart/SSIF-Visa-Report_p7","product_type":"parse","source_file_path":"SSIF-Visa-Report_p7.pdf"},"started_at":"2026-11-04T19:46:32.975742"} +{"completed_at":"2026-11-04T19:46:35.796418","latency_in_ms":1141,"output":{"example_id":"chart/She-figures_p105","grounded_pages":[],"layout_pages":[],"markdown":"103\n\n_Figure_ _3\\.5 Proportion (%) of_ _employed population in KIA among total employment, by sex, 2021_\n\n**Source: Eurostat** **–** **Human resources in science and technology (online data code: hrst\\_st\\_ncat) and Eurostat** **–** **EU- LFS** **–** **Active population by sex, age and citizenship (online data code: lfsa\\_agan).**\n\nNotes: Definition differs, see metadata: ES, FR; Break in time series: EU-27, BE, BG, CZ, DK, DE, EE, IE, EL, ES, FR, HR, IT, CY, LV, LT, LU, HU, MT, NL, AT, PL, PT, RO, SI, SK, FI, SE, IS, NO, RS; Reference year differs: ME: 2020, MK: 2020, TR: 2020, UK: 2019.\n\n17\\.0\n\n22\\.5\n\n21\\.3\n\n21\\.8\n\n21\\.5\n\n28\\.4\n\n27\\.0\n\n22\\.3\n\n21\\.1\n\n25\\.7\n\n25\\.6\n\n30\\.8\n\n31\\.2\n\n24\\.9\n\n29\\.4\n\n22\\.3\n\n40\\.4\n\n30\\.8\n\n26\\.9\n\n23\\.1\n\n31\\.4\n\n33\\.5\n\n30\\.5\n\n33\\.4\n\n26\\.2\n\n35\\.1\n\n35\\.0\n\n37\\.4\n\n31\\.0\n\n37\\.9\n\n39\\.3\n\n38\\.3\n\n32\\.0\n\n40\\.8\n\n56\\.8\n\n32\\.1\n\n32\\.5\n\n35\\.0\n\n36\\.5\n\n37\\.5\n\n41\\.6\n\n42\\.5\n\n42\\.7\n\n42\\.7\n\n44\\.4\n\n44\\.8\n\n45\\.2\n\n45\\.2\n\n45\\.4\n\n45\\.5\n\n46\\.1\n\n46\\.1\n\n46\\.1\n\n46\\.2\n\n46\\.2\n\n46\\.3\n\n47\\.5\n\n47\\.5\n\n47\\.8\n\n48\\.4\n\n48\\.5\n\n48\\.7\n\n50\\.4\n\n52\\.8\n\n53\\.0\n\n54\\.2\n\n54\\.8\n\n56\\.3\n\n58\\.2\n\n59\\.7\n\n23\\.3\n\n25\\.6\n\n26\\.9\n\n28\\.3\n\n29\\.0\n\n34\\.5\n\n33\\.6\n\n31\\.7\n\n30\\.9\n\n34\\.0\n\n34\\.1\n\n38\\.0\n\n38\\.0\n\n35\\.0\n\n36\\.9\n\n34\\.1\n\n43\\.1\n\n38\\.0\n\n35\\.9\n\n33\\.9\n\n37\\.7\n\n40\\.2\n\n38\\.5\n\n40\\.4\n\n37\\.0\n\n41\\.5\n\n41\\.5\n\n42\\.8\n\n41\\.0\n\n45\\.1\n\n46\\.3\n\n46\\.1\n\n43\\.3\n\n49\\.1\n\n58\\.1\n\n0 10 20 30 40 50 60 70\n\nRO\n\nTR\n\nMK\n\nRS\n\nBG\n\nES\n\nIT\n\nHR\n\nPL\n\nCZ\n\nME\n\nPT\n\nFI\n\nLT\n\nEU-27\n\nLV\n\nNL\n\nAT\n\nHU\n\nSK\n\nEL\n\nDK\n\nDE\n\nFR\n\nEE\n\nNO\n\nCY\n\nMT\n\nSI\n\nUK\n\nIE\n\nBE\n\nIS\n\nSE\n\nLU\n\nTotal\n\nWomen\n\nMen","pages":[{"line_numbers":[],"markdown":"103\n\n_Figure_ _3\\.5 Proportion (%) of_ _employed population in KIA among total employment, by sex, 2021_\n\n**Source: Eurostat** **–** **Human resources in science and technology (online data code: hrst\\_st\\_ncat) and Eurostat** **–** **EU- LFS** **–** **Active population by sex, age and citizenship (online data code: lfsa\\_agan).**\n\nNotes: Definition differs, see metadata: ES, FR; Break in time series: EU-27, BE, BG, CZ, DK, DE, EE, IE, EL, ES, FR, HR, IT, CY, LV, LT, LU, HU, MT, NL, AT, PL, PT, RO, SI, SK, FI, SE, IS, NO, RS; Reference year differs: ME: 2020, MK: 2020, TR: 2020, UK: 2019.\n\n17\\.0\n\n22\\.5\n\n21\\.3\n\n21\\.8\n\n21\\.5\n\n28\\.4\n\n27\\.0\n\n22\\.3\n\n21\\.1\n\n25\\.7\n\n25\\.6\n\n30\\.8\n\n31\\.2\n\n24\\.9\n\n29\\.4\n\n22\\.3\n\n40\\.4\n\n30\\.8\n\n26\\.9\n\n23\\.1\n\n31\\.4\n\n33\\.5\n\n30\\.5\n\n33\\.4\n\n26\\.2\n\n35\\.1\n\n35\\.0\n\n37\\.4\n\n31\\.0\n\n37\\.9\n\n39\\.3\n\n38\\.3\n\n32\\.0\n\n40\\.8\n\n56\\.8\n\n32\\.1\n\n32\\.5\n\n35\\.0\n\n36\\.5\n\n37\\.5\n\n41\\.6\n\n42\\.5\n\n42\\.7\n\n42\\.7\n\n44\\.4\n\n44\\.8\n\n45\\.2\n\n45\\.2\n\n45\\.4\n\n45\\.5\n\n46\\.1\n\n46\\.1\n\n46\\.1\n\n46\\.2\n\n46\\.2\n\n46\\.3\n\n47\\.5\n\n47\\.5\n\n47\\.8\n\n48\\.4\n\n48\\.5\n\n48\\.7\n\n50\\.4\n\n52\\.8\n\n53\\.0\n\n54\\.2\n\n54\\.8\n\n56\\.3\n\n58\\.2\n\n59\\.7\n\n23\\.3\n\n25\\.6\n\n26\\.9\n\n28\\.3\n\n29\\.0\n\n34\\.5\n\n33\\.6\n\n31\\.7\n\n30\\.9\n\n34\\.0\n\n34\\.1\n\n38\\.0\n\n38\\.0\n\n35\\.0\n\n36\\.9\n\n34\\.1\n\n43\\.1\n\n38\\.0\n\n35\\.9\n\n33\\.9\n\n37\\.7\n\n40\\.2\n\n38\\.5\n\n40\\.4\n\n37\\.0\n\n41\\.5\n\n41\\.5\n\n42\\.8\n\n41\\.0\n\n45\\.1\n\n46\\.3\n\n46\\.1\n\n43\\.3\n\n49\\.1\n\n58\\.1\n\n0 10 20 30 40 50 60 70\n\nRO\n\nTR\n\nMK\n\nRS\n\nBG\n\nES\n\nIT\n\nHR\n\nPL\n\nCZ\n\nME\n\nPT\n\nFI\n\nLT\n\nEU-27\n\nLV\n\nNL\n\nAT\n\nHU\n\nSK\n\nEL\n\nDK\n\nDE\n\nFR\n\nEE\n\nNO\n\nCY\n\nMT\n\nSI\n\nUK\n\nIE\n\nBE\n\nIS\n\nSE\n\nLU\n\nTotal\n\nWomen\n\nMen","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"103\n\n_Figure_ _3\\.5 Proportion (%) of_ _employed population in KIA among total employment, by sex, 2021_\n\n**Source: Eurostat** **–** **Human resources in science and technology (online data code: hrst\\_st\\_ncat) and Eurostat** **–** **EU- LFS** **–** **Active population by sex, age and citizenship (online data code: lfsa\\_agan).**\n\nNotes: Definition differs, see metadata: ES, FR; Break in time series: EU-27, BE, BG, CZ, DK, DE, EE, IE, EL, ES, FR, HR, IT, CY, LV, LT, LU, HU, MT, NL, AT, PL, PT, RO, SI, SK, FI, SE, IS, NO, RS; Reference year differs: ME: 2020, MK: 2020, TR: 2020, UK: 2019.\n\n17\\.0\n\n22\\.5\n\n21\\.3\n\n21\\.8\n\n21\\.5\n\n28\\.4\n\n27\\.0\n\n22\\.3\n\n21\\.1\n\n25\\.7\n\n25\\.6\n\n30\\.8\n\n31\\.2\n\n24\\.9\n\n29\\.4\n\n22\\.3\n\n40\\.4\n\n30\\.8\n\n26\\.9\n\n23\\.1\n\n31\\.4\n\n33\\.5\n\n30\\.5\n\n33\\.4\n\n26\\.2\n\n35\\.1\n\n35\\.0\n\n37\\.4\n\n31\\.0\n\n37\\.9\n\n39\\.3\n\n38\\.3\n\n32\\.0\n\n40\\.8\n\n56\\.8\n\n32\\.1\n\n32\\.5\n\n35\\.0\n\n36\\.5\n\n37\\.5\n\n41\\.6\n\n42\\.5\n\n42\\.7\n\n42\\.7\n\n44\\.4\n\n44\\.8\n\n45\\.2\n\n45\\.2\n\n45\\.4\n\n45\\.5\n\n46\\.1\n\n46\\.1\n\n46\\.1\n\n46\\.2\n\n46\\.2\n\n46\\.3\n\n47\\.5\n\n47\\.5\n\n47\\.8\n\n48\\.4\n\n48\\.5\n\n48\\.7\n\n50\\.4\n\n52\\.8\n\n53\\.0\n\n54\\.2\n\n54\\.8\n\n56\\.3\n\n58\\.2\n\n59\\.7\n\n23\\.3\n\n25\\.6\n\n26\\.9\n\n28\\.3\n\n29\\.0\n\n34\\.5\n\n33\\.6\n\n31\\.7\n\n30\\.9\n\n34\\.0\n\n34\\.1\n\n38\\.0\n\n38\\.0\n\n35\\.0\n\n36\\.9\n\n34\\.1\n\n43\\.1\n\n38\\.0\n\n35\\.9\n\n33\\.9\n\n37\\.7\n\n40\\.2\n\n38\\.5\n\n40\\.4\n\n37\\.0\n\n41\\.5\n\n41\\.5\n\n42\\.8\n\n41\\.0\n\n45\\.1\n\n46\\.3\n\n46\\.1\n\n43\\.3\n\n49\\.1\n\n58\\.1\n\n0 10 20 30 40 50 60 70\n\nRO\n\nTR\n\nMK\n\nRS\n\nBG\n\nES\n\nIT\n\nHR\n\nPL\n\nCZ\n\nME\n\nPT\n\nFI\n\nLT\n\nEU-27\n\nLV\n\nNL\n\nAT\n\nHU\n\nSK\n\nEL\n\nDK\n\nDE\n\nFR\n\nEE\n\nNO\n\nCY\n\nMT\n\nSI\n\nUK\n\nIE\n\nBE\n\nIS\n\nSE\n\nLU\n\nTotal\n\nWomen\n\nMen","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/She-figures_p105.pdf"]},"response":{"errors":[],"extract_id":"extract_e0e422539b1dcc5edad4902708879251","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["103\n\n_Figure_ _3\\.5 Proportion (%) of_ _employed population in KIA among total employment, by sex, 2021_\n\n**Source: Eurostat** **–** **Human resources in science and technology (online data code: hrst\\_st\\_ncat) and Eurostat** **–** **EU- LFS** **–** **Active population by sex, age and citizenship (online data code: lfsa\\_agan).**\n\nNotes: Definition differs, see metadata: ES, FR; Break in time series: EU-27, BE, BG, CZ, DK, DE, EE, IE, EL, ES, FR, HR, IT, CY, LV, LT, LU, HU, MT, NL, AT, PL, PT, RO, SI, SK, FI, SE, IS, NO, RS; Reference year differs: ME: 2020, MK: 2020, TR: 2020, UK: 2019.\n\n17\\.0\n\n22\\.5\n\n21\\.3\n\n21\\.8\n\n21\\.5\n\n28\\.4\n\n27\\.0\n\n22\\.3\n\n21\\.1\n\n25\\.7\n\n25\\.6\n\n30\\.8\n\n31\\.2\n\n24\\.9\n\n29\\.4\n\n22\\.3\n\n40\\.4\n\n30\\.8\n\n26\\.9\n\n23\\.1\n\n31\\.4\n\n33\\.5\n\n30\\.5\n\n33\\.4\n\n26\\.2\n\n35\\.1\n\n35\\.0\n\n37\\.4\n\n31\\.0\n\n37\\.9\n\n39\\.3\n\n38\\.3\n\n32\\.0\n\n40\\.8\n\n56\\.8\n\n32\\.1\n\n32\\.5\n\n35\\.0\n\n36\\.5\n\n37\\.5\n\n41\\.6\n\n42\\.5\n\n42\\.7\n\n42\\.7\n\n44\\.4\n\n44\\.8\n\n45\\.2\n\n45\\.2\n\n45\\.4\n\n45\\.5\n\n46\\.1\n\n46\\.1\n\n46\\.1\n\n46\\.2\n\n46\\.2\n\n46\\.3\n\n47\\.5\n\n47\\.5\n\n47\\.8\n\n48\\.4\n\n48\\.5\n\n48\\.7\n\n50\\.4\n\n52\\.8\n\n53\\.0\n\n54\\.2\n\n54\\.8\n\n56\\.3\n\n58\\.2\n\n59\\.7\n\n23\\.3\n\n25\\.6\n\n26\\.9\n\n28\\.3\n\n29\\.0\n\n34\\.5\n\n33\\.6\n\n31\\.7\n\n30\\.9\n\n34\\.0\n\n34\\.1\n\n38\\.0\n\n38\\.0\n\n35\\.0\n\n36\\.9\n\n34\\.1\n\n43\\.1\n\n38\\.0\n\n35\\.9\n\n33\\.9\n\n37\\.7\n\n40\\.2\n\n38\\.5\n\n40\\.4\n\n37\\.0\n\n41\\.5\n\n41\\.5\n\n42\\.8\n\n41\\.0\n\n45\\.1\n\n46\\.3\n\n46\\.1\n\n43\\.3\n\n49\\.1\n\n58\\.1\n\n0 10 20 30 40 50 60 70\n\nRO\n\nTR\n\nMK\n\nRS\n\nBG\n\nES\n\nIT\n\nHR\n\nPL\n\nCZ\n\nME\n\nPT\n\nFI\n\nLT\n\nEU-27\n\nLV\n\nNL\n\nAT\n\nHU\n\nSK\n\nEL\n\nDK\n\nDE\n\nFR\n\nEE\n\nNO\n\nCY\n\nMT\n\nSI\n\nUK\n\nIE\n\nBE\n\nIS\n\nSE\n\nLU\n\nTotal\n\nWomen\n\nMen"],"full_content":"103\n\n_Figure_ _3\\.5 Proportion (%) of_ _employed population in KIA among total employment, by sex, 2021_\n\n**Source: Eurostat** **–** **Human resources in science and technology (online data code: hrst\\_st\\_ncat) and Eurostat** **–** **EU- LFS** **–** **Active population by sex, age and citizenship (online data code: lfsa\\_agan).**\n\nNotes: Definition differs, see metadata: ES, FR; Break in time series: EU-27, BE, BG, CZ, DK, DE, EE, IE, EL, ES, FR, HR, IT, CY, LV, LT, LU, HU, MT, NL, AT, PL, PT, RO, SI, SK, FI, SE, IS, NO, RS; Reference year differs: ME: 2020, MK: 2020, TR: 2020, UK: 2019.\n\n17\\.0\n\n22\\.5\n\n21\\.3\n\n21\\.8\n\n21\\.5\n\n28\\.4\n\n27\\.0\n\n22\\.3\n\n21\\.1\n\n25\\.7\n\n25\\.6\n\n30\\.8\n\n31\\.2\n\n24\\.9\n\n29\\.4\n\n22\\.3\n\n40\\.4\n\n30\\.8\n\n26\\.9\n\n23\\.1\n\n31\\.4\n\n33\\.5\n\n30\\.5\n\n33\\.4\n\n26\\.2\n\n35\\.1\n\n35\\.0\n\n37\\.4\n\n31\\.0\n\n37\\.9\n\n39\\.3\n\n38\\.3\n\n32\\.0\n\n40\\.8\n\n56\\.8\n\n32\\.1\n\n32\\.5\n\n35\\.0\n\n36\\.5\n\n37\\.5\n\n41\\.6\n\n42\\.5\n\n42\\.7\n\n42\\.7\n\n44\\.4\n\n44\\.8\n\n45\\.2\n\n45\\.2\n\n45\\.4\n\n45\\.5\n\n46\\.1\n\n46\\.1\n\n46\\.1\n\n46\\.2\n\n46\\.2\n\n46\\.3\n\n47\\.5\n\n47\\.5\n\n47\\.8\n\n48\\.4\n\n48\\.5\n\n48\\.7\n\n50\\.4\n\n52\\.8\n\n53\\.0\n\n54\\.2\n\n54\\.8\n\n56\\.3\n\n58\\.2\n\n59\\.7\n\n23\\.3\n\n25\\.6\n\n26\\.9\n\n28\\.3\n\n29\\.0\n\n34\\.5\n\n33\\.6\n\n31\\.7\n\n30\\.9\n\n34\\.0\n\n34\\.1\n\n38\\.0\n\n38\\.0\n\n35\\.0\n\n36\\.9\n\n34\\.1\n\n43\\.1\n\n38\\.0\n\n35\\.9\n\n33\\.9\n\n37\\.7\n\n40\\.2\n\n38\\.5\n\n40\\.4\n\n37\\.0\n\n41\\.5\n\n41\\.5\n\n42\\.8\n\n41\\.0\n\n45\\.1\n\n46\\.3\n\n46\\.1\n\n43\\.3\n\n49\\.1\n\n58\\.1\n\n0 10 20 30 40 50 60 70\n\nRO\n\nTR\n\nMK\n\nRS\n\nBG\n\nES\n\nIT\n\nHR\n\nPL\n\nCZ\n\nME\n\nPT\n\nFI\n\nLT\n\nEU-27\n\nLV\n\nNL\n\nAT\n\nHU\n\nSK\n\nEL\n\nDK\n\nDE\n\nFR\n\nEE\n\nNO\n\nCY\n\nMT\n\nSI\n\nUK\n\nIE\n\nBE\n\nIS\n\nSE\n\nLU\n\nTotal\n\nWomen\n\nMen","publish_date":null,"title":"Figure 3.5 Proportion (%) of employed population in KIA among total employment, by sex, 2021 LU SE IS BE IE","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/She-figures_p105.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/She-figures_p105.pdf"},"request":{"example_id":"chart/She-figures_p105","product_type":"parse","source_file_path":"She-figures_p105.pdf"},"started_at":"2026-11-04T19:46:34.655265"} +{"completed_at":"2026-11-04T19:46:38.512264","latency_in_ms":1739,"output":{"example_id":"chart/She-figures_p115","grounded_pages":[],"layout_pages":[],"markdown":"113\n\n_Figure 3.9 Distribution (%) of R&D personnel across occupations, all sectors (BES, GOV, HES), by sex, 2021_\n\n**Source: Eurostat** **–** **Research and development statistics (online data code: rd\\_p\\_persocc) and OECD** **–** **R&D personnel by sector and function.**\n\nNotes: Definition differs, see metadata: FI (2021 - Researchers, ‘Technicians and other supporting staff’, Total - Women, Total); Break in time series: BE, SE (2021 - Researchers, ‘Technicians and other supporting staff’, Total - Women, Total); Estimated: EU-27 (2021 - All professional positions - Women, Total), UK (2018 - All professional positions - Women, Total); Provisional: DK (2019 - Researchers, ‘Technicians and other supporting staff’, Total - Women, Total); Difference in methodology: JP (2021 - All professional positions - Women, Total); Reference year differs: DK: 2019, UK: 2018, ME: 2019, MK: 2020, RU: 2020; Data not available EU-27, BE, DK, EL, FR, LV, LT, NL, FI, SE, NO, RU (Technicians, ‘Other supporting staff’ - Women, Men, Total), US, IL (All professional positions - Women, Men, Total).\n\n0 10 20 30 40 50 60 70 80 90 100","pages":[{"line_numbers":[],"markdown":"113\n\n_Figure 3.9 Distribution (%) of R&D personnel across occupations, all sectors (BES, GOV, HES), by sex, 2021_\n\n**Source: Eurostat** **–** **Research and development statistics (online data code: rd\\_p\\_persocc) and OECD** **–** **R&D personnel by sector and function.**\n\nNotes: Definition differs, see metadata: FI (2021 - Researchers, ‘Technicians and other supporting staff’, Total - Women, Total); Break in time series: BE, SE (2021 - Researchers, ‘Technicians and other supporting staff’, Total - Women, Total); Estimated: EU-27 (2021 - All professional positions - Women, Total), UK (2018 - All professional positions - Women, Total); Provisional: DK (2019 - Researchers, ‘Technicians and other supporting staff’, Total - Women, Total); Difference in methodology: JP (2021 - All professional positions - Women, Total); Reference year differs: DK: 2019, UK: 2018, ME: 2019, MK: 2020, RU: 2020; Data not available EU-27, BE, DK, EL, FR, LV, LT, NL, FI, SE, NO, RU (Technicians, ‘Other supporting staff’ - Women, Men, Total), US, IL (All professional positions - Women, Men, Total).\n\n0 10 20 30 40 50 60 70 80 90 100","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"113\n\n_Figure 3.9 Distribution (%) of R&D personnel across occupations, all sectors (BES, GOV, HES), by sex, 2021_\n\n**Source: Eurostat** **–** **Research and development statistics (online data code: rd\\_p\\_persocc) and OECD** **–** **R&D personnel by sector and function.**\n\nNotes: Definition differs, see metadata: FI (2021 - Researchers, ‘Technicians and other supporting staff’, Total - Women, Total); Break in time series: BE, SE (2021 - Researchers, ‘Technicians and other supporting staff’, Total - Women, Total); Estimated: EU-27 (2021 - All professional positions - Women, Total), UK (2018 - All professional positions - Women, Total); Provisional: DK (2019 - Researchers, ‘Technicians and other supporting staff’, Total - Women, Total); Difference in methodology: JP (2021 - All professional positions - Women, Total); Reference year differs: DK: 2019, UK: 2018, ME: 2019, MK: 2020, RU: 2020; Data not available EU-27, BE, DK, EL, FR, LV, LT, NL, FI, SE, NO, RU (Technicians, ‘Other supporting staff’ - Women, Men, Total), US, IL (All professional positions - Women, Men, Total).\n\n0 10 20 30 40 50 60 70 80 90 100","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/She-figures_p115.pdf"]},"response":{"errors":[],"extract_id":"extract_373d5730feae94a4b6d39a05acaaaa1a","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["113\n\n_Figure 3.9 Distribution (%) of R&D personnel across occupations, all sectors (BES, GOV, HES), by sex, 2021_\n\n**Source: Eurostat** **–** **Research and development statistics (online data code: rd\\_p\\_persocc) and OECD** **–** **R&D personnel by sector and function.**\n\nNotes: Definition differs, see metadata: FI (2021 - Researchers, ‘Technicians and other supporting staff’, Total - Women, Total); Break in time series: BE, SE (2021 - Researchers, ‘Technicians and other supporting staff’, Total - Women, Total); Estimated: EU-27 (2021 - All professional positions - Women, Total), UK (2018 - All professional positions - Women, Total); Provisional: DK (2019 - Researchers, ‘Technicians and other supporting staff’, Total - Women, Total); Difference in methodology: JP (2021 - All professional positions - Women, Total); Reference year differs: DK: 2019, UK: 2018, ME: 2019, MK: 2020, RU: 2020; Data not available EU-27, BE, DK, EL, FR, LV, LT, NL, FI, SE, NO, RU (Technicians, ‘Other supporting staff’ - Women, Men, Total), US, IL (All professional positions - Women, Men, Total).\n\n0 10 20 30 40 50 60 70 80 90 100"],"full_content":"113\n\n_Figure 3.9 Distribution (%) of R&D personnel across occupations, all sectors (BES, GOV, HES), by sex, 2021_\n\n**Source: Eurostat** **–** **Research and development statistics (online data code: rd\\_p\\_persocc) and OECD** **–** **R&D personnel by sector and function.**\n\nNotes: Definition differs, see metadata: FI (2021 - Researchers, ‘Technicians and other supporting staff’, Total - Women, Total); Break in time series: BE, SE (2021 - Researchers, ‘Technicians and other supporting staff’, Total - Women, Total); Estimated: EU-27 (2021 - All professional positions - Women, Total), UK (2018 - All professional positions - Women, Total); Provisional: DK (2019 - Researchers, ‘Technicians and other supporting staff’, Total - Women, Total); Difference in methodology: JP (2021 - All professional positions - Women, Total); Reference year differs: DK: 2019, UK: 2018, ME: 2019, MK: 2020, RU: 2020; Data not available EU-27, BE, DK, EL, FR, LV, LT, NL, FI, SE, NO, RU (Technicians, ‘Other supporting staff’ - Women, Men, Total), US, IL (All professional positions - Women, Men, Total).\n\n0 10 20 30 40 50 60 70 80 90 100","publish_date":null,"title":"Figure 3.9 Distribution (%) of R&D personnel across occupations, all sectors (BES, GOV, HES), by sex, 2021","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/She-figures_p115.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/She-figures_p115.pdf"},"request":{"example_id":"chart/She-figures_p115","product_type":"parse","source_file_path":"She-figures_p115.pdf"},"started_at":"2026-11-04T19:46:36.772293"} +{"completed_at":"2026-11-04T19:46:40.353143","latency_in_ms":2343,"output":{"example_id":"chart/She-figures_p278","grounded_pages":[],"layout_pages":[],"markdown":"276\n\n_Figure 7.1 Proportion (%) of women among all authors in all fields of R&D, by seniority level, 2018-2022_\n\n**Source: Scopus.**\n\nNotes: Data not available: XK.\n\n0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%\n\nWLD\n\nEU-27\n\nBE BG CZ DK DE EE IE EL ES FR HR\n\nIT\n\nCY LV LT LU HU MT NL AT PL PT RO\n\nSI SK FI SE IS\n\nNO UK BA ME MD MK AL RS TR UA GE AM FO\n\nIL\n\nTN AR AU BR CA\n\nCN\\_X\\_HK\n\nHK IN ID JP MX RU ZA KR US\n\nLess than 5 Years Between 5 and 10 Years More than 10 Years\n\nGender balance","pages":[{"line_numbers":[],"markdown":"276\n\n_Figure 7.1 Proportion (%) of women among all authors in all fields of R&D, by seniority level, 2018-2022_\n\n**Source: Scopus.**\n\nNotes: Data not available: XK.\n\n0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%\n\nWLD\n\nEU-27\n\nBE BG CZ DK DE EE IE EL ES FR HR\n\nIT\n\nCY LV LT LU HU MT NL AT PL PT RO\n\nSI SK FI SE IS\n\nNO UK BA ME MD MK AL RS TR UA GE AM FO\n\nIL\n\nTN AR AU BR CA\n\nCN\\_X\\_HK\n\nHK IN ID JP MX RU ZA KR US\n\nLess than 5 Years Between 5 and 10 Years More than 10 Years\n\nGender balance","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"276\n\n_Figure 7.1 Proportion (%) of women among all authors in all fields of R&D, by seniority level, 2018-2022_\n\n**Source: Scopus.**\n\nNotes: Data not available: XK.\n\n0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%\n\nWLD\n\nEU-27\n\nBE BG CZ DK DE EE IE EL ES FR HR\n\nIT\n\nCY LV LT LU HU MT NL AT PL PT RO\n\nSI SK FI SE IS\n\nNO UK BA ME MD MK AL RS TR UA GE AM FO\n\nIL\n\nTN AR AU BR CA\n\nCN\\_X\\_HK\n\nHK IN ID JP MX RU ZA KR US\n\nLess than 5 Years Between 5 and 10 Years More than 10 Years\n\nGender balance","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/She-figures_p278.pdf"]},"response":{"errors":[],"extract_id":"extract_30dfc9d44f4ed092f92143ea360e7049","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["276\n\n_Figure 7.1 Proportion (%) of women among all authors in all fields of R&D, by seniority level, 2018-2022_\n\n**Source: Scopus.**\n\nNotes: Data not available: XK.\n\n0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%\n\nWLD\n\nEU-27\n\nBE BG CZ DK DE EE IE EL ES FR HR\n\nIT\n\nCY LV LT LU HU MT NL AT PL PT RO\n\nSI SK FI SE IS\n\nNO UK BA ME MD MK AL RS TR UA GE AM FO\n\nIL\n\nTN AR AU BR CA\n\nCN\\_X\\_HK\n\nHK IN ID JP MX RU ZA KR US\n\nLess than 5 Years Between 5 and 10 Years More than 10 Years\n\nGender balance"],"full_content":"276\n\n_Figure 7.1 Proportion (%) of women among all authors in all fields of R&D, by seniority level, 2018-2022_\n\n**Source: Scopus.**\n\nNotes: Data not available: XK.\n\n0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%\n\nWLD\n\nEU-27\n\nBE BG CZ DK DE EE IE EL ES FR HR\n\nIT\n\nCY LV LT LU HU MT NL AT PL PT RO\n\nSI SK FI SE IS\n\nNO UK BA ME MD MK AL RS TR UA GE AM FO\n\nIL\n\nTN AR AU BR CA\n\nCN\\_X\\_HK\n\nHK IN ID JP MX RU ZA KR US\n\nLess than 5 Years Between 5 and 10 Years More than 10 Years\n\nGender balance","publish_date":null,"title":"Figure 7.1 Proportion (%) of women among all authors in all fields of R&D, by seniority level, 2018-2022","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/She-figures_p278.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/She-figures_p278.pdf"},"request":{"example_id":"chart/She-figures_p278","product_type":"parse","source_file_path":"She-figures_p278.pdf"},"started_at":"2026-11-04T19:46:38.009263"} +{"completed_at":"2026-11-04T19:46:40.569001","latency_in_ms":1136,"output":{"example_id":"chart/She-figures_p88","grounded_pages":[],"layout_pages":[],"markdown":"86\n\n_Figure 3.1 Proportion (%) of women in the EU among total employment, population of tertiary- educated HRSTCs, and population of S&Es, 2017-2021_\n\n**Source: Eurostat** **–** **EU-LFS (online data code: lfsa\\_egan) & Human Resources in science and technology (online data**\n\n**code: hrst\\_st\\_ncat).**\n\nNotes: Age group: 25-64 (for all indicators). Blue shading indicates gender balance (i.e. where women comprise between\n\n40 % and 60 % of a given population).\n\n53\\.3% 53.5% 53.7% 54.0% 54.1%\n\n40\\.5% 41.0% 41.3% 41.2% 40.7%\n\n46\\.0% 46.1% 46.2% 46.2% 46.4%\n\n0%\n\n10%\n\n20%\n\n30%\n\n40%\n\n50%\n\n60%\n\n2017 2018 2019 2020 2021\n\nGender balance\n\nTertiary educated and\n\nemployed as professionals or technicians (HRSTC)\n\nEmployed scientists and engineers\n\nTotal employment","pages":[{"line_numbers":[],"markdown":"86\n\n_Figure 3.1 Proportion (%) of women in the EU among total employment, population of tertiary- educated HRSTCs, and population of S&Es, 2017-2021_\n\n**Source: Eurostat** **–** **EU-LFS (online data code: lfsa\\_egan) & Human Resources in science and technology (online data**\n\n**code: hrst\\_st\\_ncat).**\n\nNotes: Age group: 25-64 (for all indicators). Blue shading indicates gender balance (i.e. where women comprise between\n\n40 % and 60 % of a given population).\n\n53\\.3% 53.5% 53.7% 54.0% 54.1%\n\n40\\.5% 41.0% 41.3% 41.2% 40.7%\n\n46\\.0% 46.1% 46.2% 46.2% 46.4%\n\n0%\n\n10%\n\n20%\n\n30%\n\n40%\n\n50%\n\n60%\n\n2017 2018 2019 2020 2021\n\nGender balance\n\nTertiary educated and\n\nemployed as professionals or technicians (HRSTC)\n\nEmployed scientists and engineers\n\nTotal employment","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"86\n\n_Figure 3.1 Proportion (%) of women in the EU among total employment, population of tertiary- educated HRSTCs, and population of S&Es, 2017-2021_\n\n**Source: Eurostat** **–** **EU-LFS (online data code: lfsa\\_egan) & Human Resources in science and technology (online data**\n\n**code: hrst\\_st\\_ncat).**\n\nNotes: Age group: 25-64 (for all indicators). Blue shading indicates gender balance (i.e. where women comprise between\n\n40 % and 60 % of a given population).\n\n53\\.3% 53.5% 53.7% 54.0% 54.1%\n\n40\\.5% 41.0% 41.3% 41.2% 40.7%\n\n46\\.0% 46.1% 46.2% 46.2% 46.4%\n\n0%\n\n10%\n\n20%\n\n30%\n\n40%\n\n50%\n\n60%\n\n2017 2018 2019 2020 2021\n\nGender balance\n\nTertiary educated and\n\nemployed as professionals or technicians (HRSTC)\n\nEmployed scientists and engineers\n\nTotal employment","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/She-figures_p88.pdf"]},"response":{"errors":[],"extract_id":"extract_aea46d8360e377203623774f607ee686","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["86\n\n_Figure 3.1 Proportion (%) of women in the EU among total employment, population of tertiary- educated HRSTCs, and population of S&Es, 2017-2021_\n\n**Source: Eurostat** **–** **EU-LFS (online data code: lfsa\\_egan) & Human Resources in science and technology (online data**\n\n**code: hrst\\_st\\_ncat).**\n\nNotes: Age group: 25-64 (for all indicators). Blue shading indicates gender balance (i.e. where women comprise between\n\n40 % and 60 % of a given population).\n\n53\\.3% 53.5% 53.7% 54.0% 54.1%\n\n40\\.5% 41.0% 41.3% 41.2% 40.7%\n\n46\\.0% 46.1% 46.2% 46.2% 46.4%\n\n0%\n\n10%\n\n20%\n\n30%\n\n40%\n\n50%\n\n60%\n\n2017 2018 2019 2020 2021\n\nGender balance\n\nTertiary educated and\n\nemployed as professionals or technicians (HRSTC)\n\nEmployed scientists and engineers\n\nTotal employment"],"full_content":"86\n\n_Figure 3.1 Proportion (%) of women in the EU among total employment, population of tertiary- educated HRSTCs, and population of S&Es, 2017-2021_\n\n**Source: Eurostat** **–** **EU-LFS (online data code: lfsa\\_egan) & Human Resources in science and technology (online data**\n\n**code: hrst\\_st\\_ncat).**\n\nNotes: Age group: 25-64 (for all indicators). Blue shading indicates gender balance (i.e. where women comprise between\n\n40 % and 60 % of a given population).\n\n53\\.3% 53.5% 53.7% 54.0% 54.1%\n\n40\\.5% 41.0% 41.3% 41.2% 40.7%\n\n46\\.0% 46.1% 46.2% 46.2% 46.4%\n\n0%\n\n10%\n\n20%\n\n30%\n\n40%\n\n50%\n\n60%\n\n2017 2018 2019 2020 2021\n\nGender balance\n\nTertiary educated and\n\nemployed as professionals or technicians (HRSTC)\n\nEmployed scientists and engineers\n\nTotal employment","publish_date":null,"title":"Figure 3.1 Proportion (%) of women in the EU among total employment, population of tertiary- educated HRSTCs, and population of","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/She-figures_p88.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/She-figures_p88.pdf"},"request":{"example_id":"chart/She-figures_p88","product_type":"parse","source_file_path":"She-figures_p88.pdf"},"started_at":"2026-11-04T19:46:39.432397"} +{"completed_at":"2026-11-04T19:46:42.139214","latency_in_ms":1243,"output":{"example_id":"chart/Snapshot-of-Global-PV-Markets_2025_p10","grounded_pages":[],"layout_pages":[],"markdown":"**Task 1 Strategic PV Analysis and Outreach** – 2025 Snapshot of Global PV Markets\n\n9\n\nbalanced by slow-downs in Spain and the Netherlands. Of particular note, the Turkish market grew over the past two years and is expected to remain strong as local manufacturing develops and electricity prices remain high following the Lira’s depreciation. Several GW were installed and commissioned over the end of 2023 and early 2024, impacting how data was recorded in previous reports.\n\nThe 2024 Indian market benefited from both low-cost imports and a cut-off date for using them, as well as corporate green mandates. Not shown in Figure 2, Pakistan had very strong additions to new capacity. Data quality is uncertain when it comes to installed volumes, but large volumes of modules were imported in 2023 and 2024 (24 GW over 2 years according to the China Chamber of Commerce), and larger volumes were installed in 2024 with reported new capacity more than 13 times 2023 volumes at 17 GW. Policy changes have impacted deployment budgets in South Korea, whilst stabilised consumer electricity prices have contributed to the Spanish reduced market.\n\nThese developments have not been significant enough to impact the regional distribution shares of new capacity in major markets as China continues to dominate (Figure 3).\n\n## **1\\.2 Impact of over-capacity in manufacturing**\n\nIn 2023, manufacturing capacity largely outstripped the market’s ability to absorb new module availability. Module stocks in China and Europe grew from already high 2022 values to reach an estimated 150 GW by the end of 2023. In 2024, stocks remained high as relatively stable markets in Europe and China failed to absorb these historical stocks.\n\nManufacturers continued to deliver low-price modules to markets in an effort to secure cash flows, and so module prices continued to decrease across 2024. Concerted action by Chinese manufacturing led to module prices stabilising in Q1 2025 due to controlled production cuts and upwards movement in upstream costs.\n\nThese low prices, combined with different local contexts stimulated markets in China and a selection of countries including India, Pakistan, Brazil and supported modest growth in some countries in Europe.\n\nHowever, the pressure on the industry remains tremendous and solutions are quite complex: with all segments of the PV value chain in significant overcapacity, the global development of\n\n0\n\n100\n\n200\n\n300\n\n400\n\n500\n\n600\n\n2019 2020 2021 2022 2023 2024\n\nGWp\n\n### FIGURE 3: GROWTH OF NEW ANNUAL CAPACITY IN MAJOR MARKETS\n\nOther Non IEA PVPS Countries India Other IEA PVPS Countries Japan European Union USA China\n\n_3Source: IEA PVPS_","pages":[{"line_numbers":[],"markdown":"**Task 1 Strategic PV Analysis and Outreach** – 2025 Snapshot of Global PV Markets\n\n9\n\nbalanced by slow-downs in Spain and the Netherlands. Of particular note, the Turkish market grew over the past two years and is expected to remain strong as local manufacturing develops and electricity prices remain high following the Lira’s depreciation. Several GW were installed and commissioned over the end of 2023 and early 2024, impacting how data was recorded in previous reports.\n\nThe 2024 Indian market benefited from both low-cost imports and a cut-off date for using them, as well as corporate green mandates. Not shown in Figure 2, Pakistan had very strong additions to new capacity. Data quality is uncertain when it comes to installed volumes, but large volumes of modules were imported in 2023 and 2024 (24 GW over 2 years according to the China Chamber of Commerce), and larger volumes were installed in 2024 with reported new capacity more than 13 times 2023 volumes at 17 GW. Policy changes have impacted deployment budgets in South Korea, whilst stabilised consumer electricity prices have contributed to the Spanish reduced market.\n\nThese developments have not been significant enough to impact the regional distribution shares of new capacity in major markets as China continues to dominate (Figure 3).\n\n## **1\\.2 Impact of over-capacity in manufacturing**\n\nIn 2023, manufacturing capacity largely outstripped the market’s ability to absorb new module availability. Module stocks in China and Europe grew from already high 2022 values to reach an estimated 150 GW by the end of 2023. In 2024, stocks remained high as relatively stable markets in Europe and China failed to absorb these historical stocks.\n\nManufacturers continued to deliver low-price modules to markets in an effort to secure cash flows, and so module prices continued to decrease across 2024. Concerted action by Chinese manufacturing led to module prices stabilising in Q1 2025 due to controlled production cuts and upwards movement in upstream costs.\n\nThese low prices, combined with different local contexts stimulated markets in China and a selection of countries including India, Pakistan, Brazil and supported modest growth in some countries in Europe.\n\nHowever, the pressure on the industry remains tremendous and solutions are quite complex: with all segments of the PV value chain in significant overcapacity, the global development of\n\n0\n\n100\n\n200\n\n300\n\n400\n\n500\n\n600\n\n2019 2020 2021 2022 2023 2024\n\nGWp\n\n### FIGURE 3: GROWTH OF NEW ANNUAL CAPACITY IN MAJOR MARKETS\n\nOther Non IEA PVPS Countries India Other IEA PVPS Countries Japan European Union USA China\n\n_3Source: IEA PVPS_","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**Task 1 Strategic PV Analysis and Outreach** – 2025 Snapshot of Global PV Markets\n\n9\n\nbalanced by slow-downs in Spain and the Netherlands. Of particular note, the Turkish market grew over the past two years and is expected to remain strong as local manufacturing develops and electricity prices remain high following the Lira’s depreciation. Several GW were installed and commissioned over the end of 2023 and early 2024, impacting how data was recorded in previous reports.\n\nThe 2024 Indian market benefited from both low-cost imports and a cut-off date for using them, as well as corporate green mandates. Not shown in Figure 2, Pakistan had very strong additions to new capacity. Data quality is uncertain when it comes to installed volumes, but large volumes of modules were imported in 2023 and 2024 (24 GW over 2 years according to the China Chamber of Commerce), and larger volumes were installed in 2024 with reported new capacity more than 13 times 2023 volumes at 17 GW. Policy changes have impacted deployment budgets in South Korea, whilst stabilised consumer electricity prices have contributed to the Spanish reduced market.\n\nThese developments have not been significant enough to impact the regional distribution shares of new capacity in major markets as China continues to dominate (Figure 3).\n\n## **1\\.2 Impact of over-capacity in manufacturing**\n\nIn 2023, manufacturing capacity largely outstripped the market’s ability to absorb new module availability. Module stocks in China and Europe grew from already high 2022 values to reach an estimated 150 GW by the end of 2023. In 2024, stocks remained high as relatively stable markets in Europe and China failed to absorb these historical stocks.\n\nManufacturers continued to deliver low-price modules to markets in an effort to secure cash flows, and so module prices continued to decrease across 2024. Concerted action by Chinese manufacturing led to module prices stabilising in Q1 2025 due to controlled production cuts and upwards movement in upstream costs.\n\nThese low prices, combined with different local contexts stimulated markets in China and a selection of countries including India, Pakistan, Brazil and supported modest growth in some countries in Europe.\n\nHowever, the pressure on the industry remains tremendous and solutions are quite complex: with all segments of the PV value chain in significant overcapacity, the global development of\n\n0\n\n100\n\n200\n\n300\n\n400\n\n500\n\n600\n\n2019 2020 2021 2022 2023 2024\n\nGWp\n\n### FIGURE 3: GROWTH OF NEW ANNUAL CAPACITY IN MAJOR MARKETS\n\nOther Non IEA PVPS Countries India Other IEA PVPS Countries Japan European Union USA China\n\n_3Source: IEA PVPS_","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Snapshot-of-Global-PV-Markets_2025_p10.pdf"]},"response":{"errors":[],"extract_id":"extract_d7043976d6a3ba8428030f2f85d297bd","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**Task 1 Strategic PV Analysis and Outreach** – 2025 Snapshot of Global PV Markets\n\n9\n\nbalanced by slow-downs in Spain and the Netherlands. Of particular note, the Turkish market grew over the past two years and is expected to remain strong as local manufacturing develops and electricity prices remain high following the Lira’s depreciation. Several GW were installed and commissioned over the end of 2023 and early 2024, impacting how data was recorded in previous reports.\n\nThe 2024 Indian market benefited from both low-cost imports and a cut-off date for using them, as well as corporate green mandates. Not shown in Figure 2, Pakistan had very strong additions to new capacity. Data quality is uncertain when it comes to installed volumes, but large volumes of modules were imported in 2023 and 2024 (24 GW over 2 years according to the China Chamber of Commerce), and larger volumes were installed in 2024 with reported new capacity more than 13 times 2023 volumes at 17 GW. Policy changes have impacted deployment budgets in South Korea, whilst stabilised consumer electricity prices have contributed to the Spanish reduced market.\n\nThese developments have not been significant enough to impact the regional distribution shares of new capacity in major markets as China continues to dominate (Figure 3).\n\n## **1\\.2 Impact of over-capacity in manufacturing**\n\nIn 2023, manufacturing capacity largely outstripped the market’s ability to absorb new module availability. Module stocks in China and Europe grew from already high 2022 values to reach an estimated 150 GW by the end of 2023. In 2024, stocks remained high as relatively stable markets in Europe and China failed to absorb these historical stocks.\n\nManufacturers continued to deliver low-price modules to markets in an effort to secure cash flows, and so module prices continued to decrease across 2024. Concerted action by Chinese manufacturing led to module prices stabilising in Q1 2025 due to controlled production cuts and upwards movement in upstream costs.\n\nThese low prices, combined with different local contexts stimulated markets in China and a selection of countries including India, Pakistan, Brazil and supported modest growth in some countries in Europe.\n\nHowever, the pressure on the industry remains tremendous and solutions are quite complex: with all segments of the PV value chain in significant overcapacity, the global development of\n\n0\n\n100\n\n200\n\n300\n\n400\n\n500\n\n600\n\n2019 2020 2021 2022 2023 2024\n\nGWp\n\n### FIGURE 3: GROWTH OF NEW ANNUAL CAPACITY IN MAJOR MARKETS\n\nOther Non IEA PVPS Countries India Other IEA PVPS Countries Japan European Union USA China\n\n_3Source: IEA PVPS_"],"full_content":"**Task 1 Strategic PV Analysis and Outreach** – 2025 Snapshot of Global PV Markets\n\n9\n\nbalanced by slow-downs in Spain and the Netherlands. Of particular note, the Turkish market grew over the past two years and is expected to remain strong as local manufacturing develops and electricity prices remain high following the Lira’s depreciation. Several GW were installed and commissioned over the end of 2023 and early 2024, impacting how data was recorded in previous reports.\n\nThe 2024 Indian market benefited from both low-cost imports and a cut-off date for using them, as well as corporate green mandates. Not shown in Figure 2, Pakistan had very strong additions to new capacity. Data quality is uncertain when it comes to installed volumes, but large volumes of modules were imported in 2023 and 2024 (24 GW over 2 years according to the China Chamber of Commerce), and larger volumes were installed in 2024 with reported new capacity more than 13 times 2023 volumes at 17 GW. Policy changes have impacted deployment budgets in South Korea, whilst stabilised consumer electricity prices have contributed to the Spanish reduced market.\n\nThese developments have not been significant enough to impact the regional distribution shares of new capacity in major markets as China continues to dominate (Figure 3).\n\n## **1\\.2 Impact of over-capacity in manufacturing**\n\nIn 2023, manufacturing capacity largely outstripped the market’s ability to absorb new module availability. Module stocks in China and Europe grew from already high 2022 values to reach an estimated 150 GW by the end of 2023. In 2024, stocks remained high as relatively stable markets in Europe and China failed to absorb these historical stocks.\n\nManufacturers continued to deliver low-price modules to markets in an effort to secure cash flows, and so module prices continued to decrease across 2024. Concerted action by Chinese manufacturing led to module prices stabilising in Q1 2025 due to controlled production cuts and upwards movement in upstream costs.\n\nThese low prices, combined with different local contexts stimulated markets in China and a selection of countries including India, Pakistan, Brazil and supported modest growth in some countries in Europe.\n\nHowever, the pressure on the industry remains tremendous and solutions are quite complex: with all segments of the PV value chain in significant overcapacity, the global development of\n\n0\n\n100\n\n200\n\n300\n\n400\n\n500\n\n600\n\n2019 2020 2021 2022 2023 2024\n\nGWp\n\n### FIGURE 3: GROWTH OF NEW ANNUAL CAPACITY IN MAJOR MARKETS\n\nOther Non IEA PVPS Countries India Other IEA PVPS Countries Japan European Union USA China\n\n_3Source: IEA PVPS_","publish_date":null,"title":"1.2 Impact of over-capacity in manufacturing","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Snapshot-of-Global-PV-Markets_2025_p10.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Snapshot-of-Global-PV-Markets_2025_p10.pdf"},"request":{"example_id":"chart/Snapshot-of-Global-PV-Markets_2025_p10","product_type":"parse","source_file_path":"Snapshot-of-Global-PV-Markets_2025_p10.pdf"},"started_at":"2026-11-04T19:46:40.896061"} +{"completed_at":"2026-11-04T19:29:59.990352","latency_in_ms":2541,"output":{"example_id":"chart/Snapshot-of-Global-PV-Markets_2025_p12","grounded_pages":[],"layout_pages":[],"markdown":"**Task 1 Strategic PV Analysis and Outreach** – 2025 Snapshot of Global PV Markets\n\n11\n\n(Brazil, Germany, Türkiye, Italy and France for example), the sheer volume that can be installed in individual utility scale systems is leading to this segment outpacing distributed growth around the world. Newer applications such as floating PV and linear PV are more often within the centralised market, whilst agriPV and parking canopies exist across both the centralised and decentralised segments. The increase in deployment of prosumer and self- consumption remuneration models tends to be limited to smaller capacity systems, with a larger number of unit systems but much lower capacities.\n\n## **2 CUMULATIVE INSTALLED CAPACITY IN THE WORLD**\n\nIn 2024, the **global cumulative installed capacity reached 2.25 TW** , as shown in Figure 5. It took more than 40 years to reach a cumulative capacity of 1.18 TW (in 2022), but just 2 years to double this. China now has nearly 50% of cumulative worldwide capacity. Growth in cumulative capacity remained over 35% — above average for the past 10 years. Within the Top Ten of total cumulative installed capacities (see Table 1 above), where last year the Chinese cumulative capacity was just over double that of Europe, this year it is triple — the faster growth rates in annual capacity leaving the EU and other countries lagging. There is still a long gap before Japan, who slipping to 5 th place and has a slowing market, will be overtaken in cumulative capacity— the next markets would have to improve on this year’s annual volumes over at least 5 years to catch up the more than 40 GW gap.\n\nAs India and the UK join the IEA PVPS programme in 2025, the cumulative volume of installations outside of IEA PVPS will become marginal.\n\n0\n\n100\n\n200\n\n300\n\n400\n\n500\n\n600\n\n2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024\n\nGWp\n\n### FIGURE 4: SEGMENTATION OF PV INSTALLATION 2014-2024\n\nROOFTOP UTILITY SCALE\n\n_5Source: IEA PVPS, Becquerel Institute_","pages":[{"line_numbers":[],"markdown":"**Task 1 Strategic PV Analysis and Outreach** – 2025 Snapshot of Global PV Markets\n\n11\n\n(Brazil, Germany, Türkiye, Italy and France for example), the sheer volume that can be installed in individual utility scale systems is leading to this segment outpacing distributed growth around the world. Newer applications such as floating PV and linear PV are more often within the centralised market, whilst agriPV and parking canopies exist across both the centralised and decentralised segments. The increase in deployment of prosumer and self- consumption remuneration models tends to be limited to smaller capacity systems, with a larger number of unit systems but much lower capacities.\n\n## **2 CUMULATIVE INSTALLED CAPACITY IN THE WORLD**\n\nIn 2024, the **global cumulative installed capacity reached 2.25 TW** , as shown in Figure 5. It took more than 40 years to reach a cumulative capacity of 1.18 TW (in 2022), but just 2 years to double this. China now has nearly 50% of cumulative worldwide capacity. Growth in cumulative capacity remained over 35% — above average for the past 10 years. Within the Top Ten of total cumulative installed capacities (see Table 1 above), where last year the Chinese cumulative capacity was just over double that of Europe, this year it is triple — the faster growth rates in annual capacity leaving the EU and other countries lagging. There is still a long gap before Japan, who slipping to 5 th place and has a slowing market, will be overtaken in cumulative capacity— the next markets would have to improve on this year’s annual volumes over at least 5 years to catch up the more than 40 GW gap.\n\nAs India and the UK join the IEA PVPS programme in 2025, the cumulative volume of installations outside of IEA PVPS will become marginal.\n\n0\n\n100\n\n200\n\n300\n\n400\n\n500\n\n600\n\n2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024\n\nGWp\n\n### FIGURE 4: SEGMENTATION OF PV INSTALLATION 2014-2024\n\nROOFTOP UTILITY SCALE\n\n_5Source: IEA PVPS, Becquerel Institute_","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**Task 1 Strategic PV Analysis and Outreach** – 2025 Snapshot of Global PV Markets\n\n11\n\n(Brazil, Germany, Türkiye, Italy and France for example), the sheer volume that can be installed in individual utility scale systems is leading to this segment outpacing distributed growth around the world. Newer applications such as floating PV and linear PV are more often within the centralised market, whilst agriPV and parking canopies exist across both the centralised and decentralised segments. The increase in deployment of prosumer and self- consumption remuneration models tends to be limited to smaller capacity systems, with a larger number of unit systems but much lower capacities.\n\n## **2 CUMULATIVE INSTALLED CAPACITY IN THE WORLD**\n\nIn 2024, the **global cumulative installed capacity reached 2.25 TW** , as shown in Figure 5. It took more than 40 years to reach a cumulative capacity of 1.18 TW (in 2022), but just 2 years to double this. China now has nearly 50% of cumulative worldwide capacity. Growth in cumulative capacity remained over 35% — above average for the past 10 years. Within the Top Ten of total cumulative installed capacities (see Table 1 above), where last year the Chinese cumulative capacity was just over double that of Europe, this year it is triple — the faster growth rates in annual capacity leaving the EU and other countries lagging. There is still a long gap before Japan, who slipping to 5 th place and has a slowing market, will be overtaken in cumulative capacity— the next markets would have to improve on this year’s annual volumes over at least 5 years to catch up the more than 40 GW gap.\n\nAs India and the UK join the IEA PVPS programme in 2025, the cumulative volume of installations outside of IEA PVPS will become marginal.\n\n0\n\n100\n\n200\n\n300\n\n400\n\n500\n\n600\n\n2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024\n\nGWp\n\n### FIGURE 4: SEGMENTATION OF PV INSTALLATION 2014-2024\n\nROOFTOP UTILITY SCALE\n\n_5Source: IEA PVPS, Becquerel Institute_","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Snapshot-of-Global-PV-Markets_2025_p12.pdf"]},"response":{"errors":[],"extract_id":"extract_bb81102dd79c558af88fe0ae6bf54062","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**Task 1 Strategic PV Analysis and Outreach** – 2025 Snapshot of Global PV Markets\n\n11\n\n(Brazil, Germany, Türkiye, Italy and France for example), the sheer volume that can be installed in individual utility scale systems is leading to this segment outpacing distributed growth around the world. Newer applications such as floating PV and linear PV are more often within the centralised market, whilst agriPV and parking canopies exist across both the centralised and decentralised segments. The increase in deployment of prosumer and self- consumption remuneration models tends to be limited to smaller capacity systems, with a larger number of unit systems but much lower capacities.\n\n## **2 CUMULATIVE INSTALLED CAPACITY IN THE WORLD**\n\nIn 2024, the **global cumulative installed capacity reached 2.25 TW** , as shown in Figure 5. It took more than 40 years to reach a cumulative capacity of 1.18 TW (in 2022), but just 2 years to double this. China now has nearly 50% of cumulative worldwide capacity. Growth in cumulative capacity remained over 35% — above average for the past 10 years. Within the Top Ten of total cumulative installed capacities (see Table 1 above), where last year the Chinese cumulative capacity was just over double that of Europe, this year it is triple — the faster growth rates in annual capacity leaving the EU and other countries lagging. There is still a long gap before Japan, who slipping to 5 th place and has a slowing market, will be overtaken in cumulative capacity— the next markets would have to improve on this year’s annual volumes over at least 5 years to catch up the more than 40 GW gap.\n\nAs India and the UK join the IEA PVPS programme in 2025, the cumulative volume of installations outside of IEA PVPS will become marginal.\n\n0\n\n100\n\n200\n\n300\n\n400\n\n500\n\n600\n\n2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024\n\nGWp\n\n### FIGURE 4: SEGMENTATION OF PV INSTALLATION 2014-2024\n\nROOFTOP UTILITY SCALE\n\n_5Source: IEA PVPS, Becquerel Institute_"],"full_content":"**Task 1 Strategic PV Analysis and Outreach** – 2025 Snapshot of Global PV Markets\n\n11\n\n(Brazil, Germany, Türkiye, Italy and France for example), the sheer volume that can be installed in individual utility scale systems is leading to this segment outpacing distributed growth around the world. Newer applications such as floating PV and linear PV are more often within the centralised market, whilst agriPV and parking canopies exist across both the centralised and decentralised segments. The increase in deployment of prosumer and self- consumption remuneration models tends to be limited to smaller capacity systems, with a larger number of unit systems but much lower capacities.\n\n## **2 CUMULATIVE INSTALLED CAPACITY IN THE WORLD**\n\nIn 2024, the **global cumulative installed capacity reached 2.25 TW** , as shown in Figure 5. It took more than 40 years to reach a cumulative capacity of 1.18 TW (in 2022), but just 2 years to double this. China now has nearly 50% of cumulative worldwide capacity. Growth in cumulative capacity remained over 35% — above average for the past 10 years. Within the Top Ten of total cumulative installed capacities (see Table 1 above), where last year the Chinese cumulative capacity was just over double that of Europe, this year it is triple — the faster growth rates in annual capacity leaving the EU and other countries lagging. There is still a long gap before Japan, who slipping to 5 th place and has a slowing market, will be overtaken in cumulative capacity— the next markets would have to improve on this year’s annual volumes over at least 5 years to catch up the more than 40 GW gap.\n\nAs India and the UK join the IEA PVPS programme in 2025, the cumulative volume of installations outside of IEA PVPS will become marginal.\n\n0\n\n100\n\n200\n\n300\n\n400\n\n500\n\n600\n\n2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024\n\nGWp\n\n### FIGURE 4: SEGMENTATION OF PV INSTALLATION 2014-2024\n\nROOFTOP UTILITY SCALE\n\n_5Source: IEA PVPS, Becquerel Institute_","publish_date":null,"title":"2 CUMULATIVE INSTALLED CAPACITY IN THE WORLD","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Snapshot-of-Global-PV-Markets_2025_p12.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Snapshot-of-Global-PV-Markets_2025_p12.pdf"},"request":{"example_id":"chart/Snapshot-of-Global-PV-Markets_2025_p12","product_type":"parse","source_file_path":"Snapshot-of-Global-PV-Markets_2025_p12.pdf"},"started_at":"2026-11-04T19:29:57.448996"} +{"completed_at":"2026-11-04T19:44:49.041395","latency_in_ms":2894,"output":{"example_id":"chart/Snapshot-of-Global-PV-Markets_2025_p25","grounded_pages":[],"layout_pages":[],"markdown":"**Task 1 Strategic PV Analysis and Outreach** – 2025 Snapshot of Global PV Markets\n\n24\n\nPart of the electrification of uses is the move towards electric vehicles (EVs) - the electrification of transport is well underway, with practical examples of charging EVs during peak load for grid management, a practice enabled by smart meters, whilst concepts such as virtual self- consumption, dependent on metering and secure exchange frameworks remain an attractive possibility to provide a framework for EVs as mobile storage for excess PV generation. The continued and sustained growth in EV sales across the world is an important indicator of state and consumer engagement in the transition to cleaner electricity-based economies.\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\n16\n\n18\n\n20\n\n0\n\n100\n\n200\n\n300\n\n400\n\n500\n\n600\n\n2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024\n\nEV sales (millions)\n\nPV installation (GWp)\n\n### FIGURE 11: EVOLUTION OF EV AND PV ANNUAL GROWTH\n\nPV installation (GW) EV sales (millions)\n\n_13Source : IEA PVPS, EV Volumes, Reuters_","pages":[{"line_numbers":[],"markdown":"**Task 1 Strategic PV Analysis and Outreach** – 2025 Snapshot of Global PV Markets\n\n24\n\nPart of the electrification of uses is the move towards electric vehicles (EVs) - the electrification of transport is well underway, with practical examples of charging EVs during peak load for grid management, a practice enabled by smart meters, whilst concepts such as virtual self- consumption, dependent on metering and secure exchange frameworks remain an attractive possibility to provide a framework for EVs as mobile storage for excess PV generation. The continued and sustained growth in EV sales across the world is an important indicator of state and consumer engagement in the transition to cleaner electricity-based economies.\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\n16\n\n18\n\n20\n\n0\n\n100\n\n200\n\n300\n\n400\n\n500\n\n600\n\n2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024\n\nEV sales (millions)\n\nPV installation (GWp)\n\n### FIGURE 11: EVOLUTION OF EV AND PV ANNUAL GROWTH\n\nPV installation (GW) EV sales (millions)\n\n_13Source : IEA PVPS, EV Volumes, Reuters_","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**Task 1 Strategic PV Analysis and Outreach** – 2025 Snapshot of Global PV Markets\n\n24\n\nPart of the electrification of uses is the move towards electric vehicles (EVs) - the electrification of transport is well underway, with practical examples of charging EVs during peak load for grid management, a practice enabled by smart meters, whilst concepts such as virtual self- consumption, dependent on metering and secure exchange frameworks remain an attractive possibility to provide a framework for EVs as mobile storage for excess PV generation. The continued and sustained growth in EV sales across the world is an important indicator of state and consumer engagement in the transition to cleaner electricity-based economies.\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\n16\n\n18\n\n20\n\n0\n\n100\n\n200\n\n300\n\n400\n\n500\n\n600\n\n2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024\n\nEV sales (millions)\n\nPV installation (GWp)\n\n### FIGURE 11: EVOLUTION OF EV AND PV ANNUAL GROWTH\n\nPV installation (GW) EV sales (millions)\n\n_13Source : IEA PVPS, EV Volumes, Reuters_","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Snapshot-of-Global-PV-Markets_2025_p25.pdf"]},"response":{"errors":[],"extract_id":"extract_d47ce1986ae404163ef15e7d08d5ffac","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**Task 1 Strategic PV Analysis and Outreach** – 2025 Snapshot of Global PV Markets\n\n24\n\nPart of the electrification of uses is the move towards electric vehicles (EVs) - the electrification of transport is well underway, with practical examples of charging EVs during peak load for grid management, a practice enabled by smart meters, whilst concepts such as virtual self- consumption, dependent on metering and secure exchange frameworks remain an attractive possibility to provide a framework for EVs as mobile storage for excess PV generation. The continued and sustained growth in EV sales across the world is an important indicator of state and consumer engagement in the transition to cleaner electricity-based economies.\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\n16\n\n18\n\n20\n\n0\n\n100\n\n200\n\n300\n\n400\n\n500\n\n600\n\n2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024\n\nEV sales (millions)\n\nPV installation (GWp)\n\n### FIGURE 11: EVOLUTION OF EV AND PV ANNUAL GROWTH\n\nPV installation (GW) EV sales (millions)\n\n_13Source : IEA PVPS, EV Volumes, Reuters_"],"full_content":"**Task 1 Strategic PV Analysis and Outreach** – 2025 Snapshot of Global PV Markets\n\n24\n\nPart of the electrification of uses is the move towards electric vehicles (EVs) - the electrification of transport is well underway, with practical examples of charging EVs during peak load for grid management, a practice enabled by smart meters, whilst concepts such as virtual self- consumption, dependent on metering and secure exchange frameworks remain an attractive possibility to provide a framework for EVs as mobile storage for excess PV generation. The continued and sustained growth in EV sales across the world is an important indicator of state and consumer engagement in the transition to cleaner electricity-based economies.\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\n16\n\n18\n\n20\n\n0\n\n100\n\n200\n\n300\n\n400\n\n500\n\n600\n\n2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024\n\nEV sales (millions)\n\nPV installation (GWp)\n\n### FIGURE 11: EVOLUTION OF EV AND PV ANNUAL GROWTH\n\nPV installation (GW) EV sales (millions)\n\n_13Source : IEA PVPS, EV Volumes, Reuters_","publish_date":null,"title":"FIGURE 11: EVOLUTION OF EV AND PV ANNUAL GROWTH","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Snapshot-of-Global-PV-Markets_2025_p25.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Snapshot-of-Global-PV-Markets_2025_p25.pdf"},"request":{"example_id":"chart/Snapshot-of-Global-PV-Markets_2025_p25","product_type":"parse","source_file_path":"Snapshot-of-Global-PV-Markets_2025_p25.pdf"},"started_at":"2026-11-04T19:44:46.146512"} +{"completed_at":"2026-11-04T19:44:48.889022","latency_in_ms":2739,"output":{"example_id":"chart/TRS241_Web_p49","grounded_pages":[],"layout_pages":[],"markdown":"POWER PLANNING: PART OF OVERALL PLAN 25\n\nB\\*\n\n_V_\n\nC.H\n\n**I Sourc«9: IAEA En«ray and Economic Oafo Bonkl**\n\n**1983 World** **_BortV_** **Atloi I**\n\n**U.S**\n\n**industrial**\n\n**market**\n\n**economies**\n\n**AR: BR: CA: CH: CN: DC: ES: FR: IN: IT: JP: KR: MY: PH: SE: US: YU:**\n\n**ARGENTINA BRAZIL CANADA SWITZERLAND CHINA GERMANY, FEDERAL REPUBLIC OF SPAIN FRANCE INDIA ITALY JAPAN KOREA, REPUBLIC OF MALAYSIA PHILIPPINES SWEDEN UNITED STATES OF AMERICA YUGOSLAVIA**\n\n**middle income**\n\n**economies**\n\n**low income economies**\n\n**200 300**\n\n1981 per capita energy _use_ (GJ)\n\n_FIG.2.2. Relation of GDP to energy consumption (both in per capita terms)._\n\nthe case of overall energy demand. For several decades before 1973, the demand for electrical energy grew more rapidly than the demand for total energy. While the growth in total energy demand slowed considerably after 1973, electrical energy demand was only slightly deflected from its historic growth path. Figure 2.3 shows this quite well by comparing the per capita growth rates of total energy, electrical energy and non-electrical energy relative to the growth of GDP in six developed nations. The growth in non-electrical energy relative to GDP dropped sharply after 1973, while average growth of electrical energy relative to GDP was generally much the same from 1960 to 1980.\n\nFrom cross-sectional data of many countries at various stages of economic\n\ndevelopment, it can be seen that the energy/GDP relationship is not fixed as might at first be expected. It differs widely even among nations with similar economic conditions. Moreover, a nation moving through various stages of development — from an agrarian to an industrial to a post-industrial service economy — will have a quite different level of energy use per unit of output.","pages":[{"line_numbers":[],"markdown":"POWER PLANNING: PART OF OVERALL PLAN 25\n\nB\\*\n\n_V_\n\nC.H\n\n**I Sourc«9: IAEA En«ray and Economic Oafo Bonkl**\n\n**1983 World** **_BortV_** **Atloi I**\n\n**U.S**\n\n**industrial**\n\n**market**\n\n**economies**\n\n**AR: BR: CA: CH: CN: DC: ES: FR: IN: IT: JP: KR: MY: PH: SE: US: YU:**\n\n**ARGENTINA BRAZIL CANADA SWITZERLAND CHINA GERMANY, FEDERAL REPUBLIC OF SPAIN FRANCE INDIA ITALY JAPAN KOREA, REPUBLIC OF MALAYSIA PHILIPPINES SWEDEN UNITED STATES OF AMERICA YUGOSLAVIA**\n\n**middle income**\n\n**economies**\n\n**low income economies**\n\n**200 300**\n\n1981 per capita energy _use_ (GJ)\n\n_FIG.2.2. Relation of GDP to energy consumption (both in per capita terms)._\n\nthe case of overall energy demand. For several decades before 1973, the demand for electrical energy grew more rapidly than the demand for total energy. While the growth in total energy demand slowed considerably after 1973, electrical energy demand was only slightly deflected from its historic growth path. Figure 2.3 shows this quite well by comparing the per capita growth rates of total energy, electrical energy and non-electrical energy relative to the growth of GDP in six developed nations. The growth in non-electrical energy relative to GDP dropped sharply after 1973, while average growth of electrical energy relative to GDP was generally much the same from 1960 to 1980.\n\nFrom cross-sectional data of many countries at various stages of economic\n\ndevelopment, it can be seen that the energy/GDP relationship is not fixed as might at first be expected. It differs widely even among nations with similar economic conditions. Moreover, a nation moving through various stages of development — from an agrarian to an industrial to a post-industrial service economy — will have a quite different level of energy use per unit of output.","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"POWER PLANNING: PART OF OVERALL PLAN 25\n\nB\\*\n\n_V_\n\nC.H\n\n**I Sourc«9: IAEA En«ray and Economic Oafo Bonkl**\n\n**1983 World** **_BortV_** **Atloi I**\n\n**U.S**\n\n**industrial**\n\n**market**\n\n**economies**\n\n**AR: BR: CA: CH: CN: DC: ES: FR: IN: IT: JP: KR: MY: PH: SE: US: YU:**\n\n**ARGENTINA BRAZIL CANADA SWITZERLAND CHINA GERMANY, FEDERAL REPUBLIC OF SPAIN FRANCE INDIA ITALY JAPAN KOREA, REPUBLIC OF MALAYSIA PHILIPPINES SWEDEN UNITED STATES OF AMERICA YUGOSLAVIA**\n\n**middle income**\n\n**economies**\n\n**low income economies**\n\n**200 300**\n\n1981 per capita energy _use_ (GJ)\n\n_FIG.2.2. Relation of GDP to energy consumption (both in per capita terms)._\n\nthe case of overall energy demand. For several decades before 1973, the demand for electrical energy grew more rapidly than the demand for total energy. While the growth in total energy demand slowed considerably after 1973, electrical energy demand was only slightly deflected from its historic growth path. Figure 2.3 shows this quite well by comparing the per capita growth rates of total energy, electrical energy and non-electrical energy relative to the growth of GDP in six developed nations. The growth in non-electrical energy relative to GDP dropped sharply after 1973, while average growth of electrical energy relative to GDP was generally much the same from 1960 to 1980.\n\nFrom cross-sectional data of many countries at various stages of economic\n\ndevelopment, it can be seen that the energy/GDP relationship is not fixed as might at first be expected. It differs widely even among nations with similar economic conditions. Moreover, a nation moving through various stages of development — from an agrarian to an industrial to a post-industrial service economy — will have a quite different level of energy use per unit of output.","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/TRS241_Web_p49.pdf"]},"response":{"errors":[],"extract_id":"extract_6f0d62a5152220f02e562e0c38e30e83","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["POWER PLANNING: PART OF OVERALL PLAN 25\n\nB\\*\n\n_V_\n\nC.H\n\n**I Sourc«9: IAEA En«ray and Economic Oafo Bonkl**\n\n**1983 World** **_BortV_** **Atloi I**\n\n**U.S**\n\n**industrial**\n\n**market**\n\n**economies**\n\n**AR: BR: CA: CH: CN: DC: ES: FR: IN: IT: JP: KR: MY: PH: SE: US: YU:**\n\n**ARGENTINA BRAZIL CANADA SWITZERLAND CHINA GERMANY, FEDERAL REPUBLIC OF SPAIN FRANCE INDIA ITALY JAPAN KOREA, REPUBLIC OF MALAYSIA PHILIPPINES SWEDEN UNITED STATES OF AMERICA YUGOSLAVIA**\n\n**middle income**\n\n**economies**\n\n**low income economies**\n\n**200 300**\n\n1981 per capita energy _use_ (GJ)\n\n_FIG.2.2. Relation of GDP to energy consumption (both in per capita terms)._\n\nthe case of overall energy demand. For several decades before 1973, the demand for electrical energy grew more rapidly than the demand for total energy. While the growth in total energy demand slowed considerably after 1973, electrical energy demand was only slightly deflected from its historic growth path. Figure 2.3 shows this quite well by comparing the per capita growth rates of total energy, electrical energy and non-electrical energy relative to the growth of GDP in six developed nations. The growth in non-electrical energy relative to GDP dropped sharply after 1973, while average growth of electrical energy relative to GDP was generally much the same from 1960 to 1980.\n\nFrom cross-sectional data of many countries at various stages of economic\n\ndevelopment, it can be seen that the energy/GDP relationship is not fixed as might at first be expected. It differs widely even among nations with similar economic conditions. Moreover, a nation moving through various stages of development — from an agrarian to an industrial to a post-industrial service economy — will have a quite different level of energy use per unit of output."],"full_content":"POWER PLANNING: PART OF OVERALL PLAN 25\n\nB\\*\n\n_V_\n\nC.H\n\n**I Sourc«9: IAEA En«ray and Economic Oafo Bonkl**\n\n**1983 World** **_BortV_** **Atloi I**\n\n**U.S**\n\n**industrial**\n\n**market**\n\n**economies**\n\n**AR: BR: CA: CH: CN: DC: ES: FR: IN: IT: JP: KR: MY: PH: SE: US: YU:**\n\n**ARGENTINA BRAZIL CANADA SWITZERLAND CHINA GERMANY, FEDERAL REPUBLIC OF SPAIN FRANCE INDIA ITALY JAPAN KOREA, REPUBLIC OF MALAYSIA PHILIPPINES SWEDEN UNITED STATES OF AMERICA YUGOSLAVIA**\n\n**middle income**\n\n**economies**\n\n**low income economies**\n\n**200 300**\n\n1981 per capita energy _use_ (GJ)\n\n_FIG.2.2. Relation of GDP to energy consumption (both in per capita terms)._\n\nthe case of overall energy demand. For several decades before 1973, the demand for electrical energy grew more rapidly than the demand for total energy. While the growth in total energy demand slowed considerably after 1973, electrical energy demand was only slightly deflected from its historic growth path. Figure 2.3 shows this quite well by comparing the per capita growth rates of total energy, electrical energy and non-electrical energy relative to the growth of GDP in six developed nations. The growth in non-electrical energy relative to GDP dropped sharply after 1973, while average growth of electrical energy relative to GDP was generally much the same from 1960 to 1980.\n\nFrom cross-sectional data of many countries at various stages of economic\n\ndevelopment, it can be seen that the energy/GDP relationship is not fixed as might at first be expected. It differs widely even among nations with similar economic conditions. Moreover, a nation moving through various stages of development — from an agrarian to an industrial to a post-industrial service economy — will have a quite different level of energy use per unit of output.","publish_date":null,"title":"25","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/TRS241_Web_p49.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/TRS241_Web_p49.pdf"},"request":{"example_id":"chart/TRS241_Web_p49","product_type":"parse","source_file_path":"TRS241_Web_p49.pdf"},"started_at":"2026-11-04T19:44:46.149498"} +{"completed_at":"2026-11-04T19:44:51.541537","latency_in_ms":2798,"output":{"example_id":"chart/TRS241_Web_p50","grounded_pages":[],"layout_pages":[],"markdown":"26 CHAPTER 2\n\nQ. **O**\n\n**Q\\_**\n\n**Q** O\n\n<<\n\n**O.** oo\n\n950,000 Production\n\n## Berlin Model Y 375,000 Production\n\n## Texas Model Y >250,000 Production\n\n## Cybertruck >125,000 Pilot production\n\n## Nevada Tesla Semi - Pilot production\n\n## Various Next Gen Platform - In development\n\n## TBD Roadster - In development\n\nInstalled capacity ≠ current production rate and there may be limitations discovered as production rates approach capacity. Production rates depend on a variety of factors, including equipment uptime, component supply, downtime related to factory upgrades, regulatory considerations and other factors.\n\n## Market share of Tesla vehicles by region (TTM)\n\n### Source: Tesla estimates based on latest available data from ACEA; Autonews.com; CAAM – light- duty vehicles only TTM = Trailing twelve months 7\n\n## During the quarter we brought down several production lines for upgrades at various factories, which led to a sequential decline in production volumes. We made further progress smoothing out the delivery rate across the quarter, with September accounting for ~40% of Q3 deliveries this year, compared to September accounting for ~65% of Q3 deliveries in 2022.\n\n## US: California, Nevada and Texas\n\n## At Gigafactory Texas, we began pilot production of the Cybertruck, which remains on track for initial deliveries this year. We are expecting the Model Y production rate in Texas to grow very gradually from its current level as we ramp additional supply chain needs in a cost-efficient manner. Production of our higher density 4680 cell is progressing as planned, and we continue building capacity for cathode production and lithium refining in the U.S.\n\n## China: Shanghai\n\n## Other than scheduled downtime in Q3, our Shanghai factory has been successfully running near full capacity for several quarters, and we do not expect a meaningful increase in weekly production run rate. Giga Shanghai remains our main export hub.\n\n## Europe: Berlin-Brandenburg\n\n## Model Y remained the best-selling vehicle of any kind in Europe year-to-date (based on the latest available data as of August). Similar to Texas, further production ramp of Model Y will be gradual.\n\n### 0%\n\n### 1%\n\n### 2%\n\n### 3%\n\n### 4%\n\n### US/Canada Europe China","pages":[{"line_numbers":[],"markdown":"## V E H I C L E C A P A C I T Y\n\n## Current Installed Annual Vehicle Capacity\n\n## Region Model Capacity Status\n\n## California Model S / Model X 100,000 Production\n\n## Model 3 / Model Y 550,000 Production\n\n## Shanghai Model 3 / Model Y >950,000 Production\n\n## Berlin Model Y 375,000 Production\n\n## Texas Model Y >250,000 Production\n\n## Cybertruck >125,000 Pilot production\n\n## Nevada Tesla Semi - Pilot production\n\n## Various Next Gen Platform - In development\n\n## TBD Roadster - In development\n\nInstalled capacity ≠ current production rate and there may be limitations discovered as production rates approach capacity. Production rates depend on a variety of factors, including equipment uptime, component supply, downtime related to factory upgrades, regulatory considerations and other factors.\n\n## Market share of Tesla vehicles by region (TTM)\n\n### Source: Tesla estimates based on latest available data from ACEA; Autonews.com; CAAM – light- duty vehicles only TTM = Trailing twelve months 7\n\n## During the quarter we brought down several production lines for upgrades at various factories, which led to a sequential decline in production volumes. We made further progress smoothing out the delivery rate across the quarter, with September accounting for ~40% of Q3 deliveries this year, compared to September accounting for ~65% of Q3 deliveries in 2022.\n\n## US: California, Nevada and Texas\n\n## At Gigafactory Texas, we began pilot production of the Cybertruck, which remains on track for initial deliveries this year. We are expecting the Model Y production rate in Texas to grow very gradually from its current level as we ramp additional supply chain needs in a cost-efficient manner. Production of our higher density 4680 cell is progressing as planned, and we continue building capacity for cathode production and lithium refining in the U.S.\n\n## China: Shanghai\n\n## Other than scheduled downtime in Q3, our Shanghai factory has been successfully running near full capacity for several quarters, and we do not expect a meaningful increase in weekly production run rate. Giga Shanghai remains our main export hub.\n\n## Europe: Berlin-Brandenburg\n\n## Model Y remained the best-selling vehicle of any kind in Europe year-to-date (based on the latest available data as of August). Similar to Texas, further production ramp of Model Y will be gradual.\n\n### 0%\n\n### 1%\n\n### 2%\n\n### 3%\n\n### 4%\n\n### US/Canada Europe China","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"## V E H I C L E C A P A C I T Y\n\n## Current Installed Annual Vehicle Capacity\n\n## Region Model Capacity Status\n\n## California Model S / Model X 100,000 Production\n\n## Model 3 / Model Y 550,000 Production\n\n## Shanghai Model 3 / Model Y >950,000 Production\n\n## Berlin Model Y 375,000 Production\n\n## Texas Model Y >250,000 Production\n\n## Cybertruck >125,000 Pilot production\n\n## Nevada Tesla Semi - Pilot production\n\n## Various Next Gen Platform - In development\n\n## TBD Roadster - In development\n\nInstalled capacity ≠ current production rate and there may be limitations discovered as production rates approach capacity. Production rates depend on a variety of factors, including equipment uptime, component supply, downtime related to factory upgrades, regulatory considerations and other factors.\n\n## Market share of Tesla vehicles by region (TTM)\n\n### Source: Tesla estimates based on latest available data from ACEA; Autonews.com; CAAM – light- duty vehicles only TTM = Trailing twelve months 7\n\n## During the quarter we brought down several production lines for upgrades at various factories, which led to a sequential decline in production volumes. We made further progress smoothing out the delivery rate across the quarter, with September accounting for ~40% of Q3 deliveries this year, compared to September accounting for ~65% of Q3 deliveries in 2022.\n\n## US: California, Nevada and Texas\n\n## At Gigafactory Texas, we began pilot production of the Cybertruck, which remains on track for initial deliveries this year. We are expecting the Model Y production rate in Texas to grow very gradually from its current level as we ramp additional supply chain needs in a cost-efficient manner. Production of our higher density 4680 cell is progressing as planned, and we continue building capacity for cathode production and lithium refining in the U.S.\n\n## China: Shanghai\n\n## Other than scheduled downtime in Q3, our Shanghai factory has been successfully running near full capacity for several quarters, and we do not expect a meaningful increase in weekly production run rate. Giga Shanghai remains our main export hub.\n\n## Europe: Berlin-Brandenburg\n\n## Model Y remained the best-selling vehicle of any kind in Europe year-to-date (based on the latest available data as of August). Similar to Texas, further production ramp of Model Y will be gradual.\n\n### 0%\n\n### 1%\n\n### 2%\n\n### 3%\n\n### 4%\n\n### US/Canada Europe China","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/TSLA-Q3-2023-Update-3_p7.pdf"]},"response":{"errors":[],"extract_id":"extract_02fe79337e3df56808b7482df4a3f4df","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["## V E H I C L E C A P A C I T Y\n\n## Current Installed Annual Vehicle Capacity\n\n## Region Model Capacity Status\n\n## California Model S / Model X 100,000 Production\n\n## Model 3 / Model Y 550,000 Production\n\n## Shanghai Model 3 / Model Y >950,000 Production\n\n## Berlin Model Y 375,000 Production\n\n## Texas Model Y >250,000 Production\n\n## Cybertruck >125,000 Pilot production\n\n## Nevada Tesla Semi - Pilot production\n\n## Various Next Gen Platform - In development\n\n## TBD Roadster - In development\n\nInstalled capacity ≠ current production rate and there may be limitations discovered as production rates approach capacity. Production rates depend on a variety of factors, including equipment uptime, component supply, downtime related to factory upgrades, regulatory considerations and other factors.\n\n## Market share of Tesla vehicles by region (TTM)\n\n### Source: Tesla estimates based on latest available data from ACEA; Autonews.com; CAAM – light- duty vehicles only TTM = Trailing twelve months 7\n\n## During the quarter we brought down several production lines for upgrades at various factories, which led to a sequential decline in production volumes. We made further progress smoothing out the delivery rate across the quarter, with September accounting for ~40% of Q3 deliveries this year, compared to September accounting for ~65% of Q3 deliveries in 2022.\n\n## US: California, Nevada and Texas\n\n## At Gigafactory Texas, we began pilot production of the Cybertruck, which remains on track for initial deliveries this year. We are expecting the Model Y production rate in Texas to grow very gradually from its current level as we ramp additional supply chain needs in a cost-efficient manner. Production of our higher density 4680 cell is progressing as planned, and we continue building capacity for cathode production and lithium refining in the U.S.\n\n## China: Shanghai\n\n## Other than scheduled downtime in Q3, our Shanghai factory has been successfully running near full capacity for several quarters, and we do not expect a meaningful increase in weekly production run rate. Giga Shanghai remains our main export hub.\n\n## Europe: Berlin-Brandenburg\n\n## Model Y remained the best-selling vehicle of any kind in Europe year-to-date (based on the latest available data as of August). Similar to Texas, further production ramp of Model Y will be gradual.\n\n### 0%\n\n### 1%\n\n### 2%\n\n### 3%\n\n### 4%\n\n### US/Canada Europe China"],"full_content":"## V E H I C L E C A P A C I T Y\n\n## Current Installed Annual Vehicle Capacity\n\n## Region Model Capacity Status\n\n## California Model S / Model X 100,000 Production\n\n## Model 3 / Model Y 550,000 Production\n\n## Shanghai Model 3 / Model Y >950,000 Production\n\n## Berlin Model Y 375,000 Production\n\n## Texas Model Y >250,000 Production\n\n## Cybertruck >125,000 Pilot production\n\n## Nevada Tesla Semi - Pilot production\n\n## Various Next Gen Platform - In development\n\n## TBD Roadster - In development\n\nInstalled capacity ≠ current production rate and there may be limitations discovered as production rates approach capacity. Production rates depend on a variety of factors, including equipment uptime, component supply, downtime related to factory upgrades, regulatory considerations and other factors.\n\n## Market share of Tesla vehicles by region (TTM)\n\n### Source: Tesla estimates based on latest available data from ACEA; Autonews.com; CAAM – light- duty vehicles only TTM = Trailing twelve months 7\n\n## During the quarter we brought down several production lines for upgrades at various factories, which led to a sequential decline in production volumes. We made further progress smoothing out the delivery rate across the quarter, with September accounting for ~40% of Q3 deliveries this year, compared to September accounting for ~65% of Q3 deliveries in 2022.\n\n## US: California, Nevada and Texas\n\n## At Gigafactory Texas, we began pilot production of the Cybertruck, which remains on track for initial deliveries this year. We are expecting the Model Y production rate in Texas to grow very gradually from its current level as we ramp additional supply chain needs in a cost-efficient manner. Production of our higher density 4680 cell is progressing as planned, and we continue building capacity for cathode production and lithium refining in the U.S.\n\n## China: Shanghai\n\n## Other than scheduled downtime in Q3, our Shanghai factory has been successfully running near full capacity for several quarters, and we do not expect a meaningful increase in weekly production run rate. Giga Shanghai remains our main export hub.\n\n## Europe: Berlin-Brandenburg\n\n## Model Y remained the best-selling vehicle of any kind in Europe year-to-date (based on the latest available data as of August). Similar to Texas, further production ramp of Model Y will be gradual.\n\n### 0%\n\n### 1%\n\n### 2%\n\n### 3%\n\n### 4%\n\n### US/Canada Europe China","publish_date":null,"title":"V E H I C L E C A P A C I T Y Current Installed Annual Vehicle Capacity","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/TSLA-Q3-2023-Update-3_p7.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/TSLA-Q3-2023-Update-3_p7.pdf"},"request":{"example_id":"chart/TSLA-Q3-2023-Update-3_p7","product_type":"parse","source_file_path":"TSLA-Q3-2023-Update-3_p7.pdf"},"started_at":"2026-11-04T19:44:54.322322"} +{"completed_at":"2026-11-04T19:44:58.774388","latency_in_ms":2856,"output":{"example_id":"chart/TSLA-Q3-2023-Update-3_p8","grounded_pages":[],"layout_pages":[],"markdown":"## Cost of goods sold per vehicle\n\n## C O R E T E C H N O L O G Y\n\n## Cumulative miles driven with FSD Beta (millions)\n\n### 8\n\n## Artificial Intelligence Software and Hardware\n\n## Software that safely performs tasks in the real world is the key focus of our AI development efforts. We have commissioned one of the world's largest supercomputers to accelerate the pace of our AI development, with compute capacity more than doubling compared to Q2. Our large installed base of vehicles continues to generate anonymized video and other data used to develop our FSD Capability features.\n\n## Vehicle and Other Software\n\n## All Tesla rentals through Hertz in the U.S. and Canada now allow Tesla app access, allowing renters to use keyless lock/unlock via phone key, remotely precondition the cabin, track charge status and more. Customers who already have a Tesla Profile will have their settings and preferences seamlessly applied, making the rental car feel like their own. The in-app service experience was also redesigned to allow customers to schedule service, access their loaner, track service progress, pay and manage drop- off/pickup. Prospective customers can similarly schedule, locate and test drive a demo vehicle.\n\n## Battery, Powertrain & Manufacturing\n\n## Despite macroeconomic headwinds, our planned factory shutdowns in Q3 and ongoing ramp at new factories, our average vehicle cost was ~$37,500, and we continue to work to reduce the cost further. For very heavy vehicles, a high voltage powertrain architecture brings notable cost savings, which is why Cybertruck will adopt an 800-volt architecture.\n\n### 0\n\n### 100\n\n### 200\n\n### 300\n\n### 400\n\n### 500\n\n### 600\n\n### $36,000\n\n### $36,500\n\n### $37,000\n\n### $37,500\n\n### $38,000\n\n### $38,500\n\n### $39,000\n\n### $39,500\n\n### $40,000\n\n### Q3 2022 Q4 2022 Q1 2023 Q2 2023 Q3 2023","pages":[{"line_numbers":[],"markdown":"## Cost of goods sold per vehicle\n\n## C O R E T E C H N O L O G Y\n\n## Cumulative miles driven with FSD Beta (millions)\n\n### 8\n\n## Artificial Intelligence Software and Hardware\n\n## Software that safely performs tasks in the real world is the key focus of our AI development efforts. We have commissioned one of the world's largest supercomputers to accelerate the pace of our AI development, with compute capacity more than doubling compared to Q2. Our large installed base of vehicles continues to generate anonymized video and other data used to develop our FSD Capability features.\n\n## Vehicle and Other Software\n\n## All Tesla rentals through Hertz in the U.S. and Canada now allow Tesla app access, allowing renters to use keyless lock/unlock via phone key, remotely precondition the cabin, track charge status and more. Customers who already have a Tesla Profile will have their settings and preferences seamlessly applied, making the rental car feel like their own. The in-app service experience was also redesigned to allow customers to schedule service, access their loaner, track service progress, pay and manage drop- off/pickup. Prospective customers can similarly schedule, locate and test drive a demo vehicle.\n\n## Battery, Powertrain & Manufacturing\n\n## Despite macroeconomic headwinds, our planned factory shutdowns in Q3 and ongoing ramp at new factories, our average vehicle cost was ~$37,500, and we continue to work to reduce the cost further. For very heavy vehicles, a high voltage powertrain architecture brings notable cost savings, which is why Cybertruck will adopt an 800-volt architecture.\n\n### 0\n\n### 100\n\n### 200\n\n### 300\n\n### 400\n\n### 500\n\n### 600\n\n### $36,000\n\n### $36,500\n\n### $37,000\n\n### $37,500\n\n### $38,000\n\n### $38,500\n\n### $39,000\n\n### $39,500\n\n### $40,000\n\n### Q3 2022 Q4 2022 Q1 2023 Q2 2023 Q3 2023","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"## Cost of goods sold per vehicle\n\n## C O R E T E C H N O L O G Y\n\n## Cumulative miles driven with FSD Beta (millions)\n\n### 8\n\n## Artificial Intelligence Software and Hardware\n\n## Software that safely performs tasks in the real world is the key focus of our AI development efforts. We have commissioned one of the world's largest supercomputers to accelerate the pace of our AI development, with compute capacity more than doubling compared to Q2. Our large installed base of vehicles continues to generate anonymized video and other data used to develop our FSD Capability features.\n\n## Vehicle and Other Software\n\n## All Tesla rentals through Hertz in the U.S. and Canada now allow Tesla app access, allowing renters to use keyless lock/unlock via phone key, remotely precondition the cabin, track charge status and more. Customers who already have a Tesla Profile will have their settings and preferences seamlessly applied, making the rental car feel like their own. The in-app service experience was also redesigned to allow customers to schedule service, access their loaner, track service progress, pay and manage drop- off/pickup. Prospective customers can similarly schedule, locate and test drive a demo vehicle.\n\n## Battery, Powertrain & Manufacturing\n\n## Despite macroeconomic headwinds, our planned factory shutdowns in Q3 and ongoing ramp at new factories, our average vehicle cost was ~$37,500, and we continue to work to reduce the cost further. For very heavy vehicles, a high voltage powertrain architecture brings notable cost savings, which is why Cybertruck will adopt an 800-volt architecture.\n\n### 0\n\n### 100\n\n### 200\n\n### 300\n\n### 400\n\n### 500\n\n### 600\n\n### $36,000\n\n### $36,500\n\n### $37,000\n\n### $37,500\n\n### $38,000\n\n### $38,500\n\n### $39,000\n\n### $39,500\n\n### $40,000\n\n### Q3 2022 Q4 2022 Q1 2023 Q2 2023 Q3 2023","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/TSLA-Q3-2023-Update-3_p8.pdf"]},"response":{"errors":[],"extract_id":"extract_d58f817862c3133390359ab85a1137eb","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["## Cost of goods sold per vehicle\n\n## C O R E T E C H N O L O G Y\n\n## Cumulative miles driven with FSD Beta (millions)\n\n### 8\n\n## Artificial Intelligence Software and Hardware\n\n## Software that safely performs tasks in the real world is the key focus of our AI development efforts. We have commissioned one of the world's largest supercomputers to accelerate the pace of our AI development, with compute capacity more than doubling compared to Q2. Our large installed base of vehicles continues to generate anonymized video and other data used to develop our FSD Capability features.\n\n## Vehicle and Other Software\n\n## All Tesla rentals through Hertz in the U.S. and Canada now allow Tesla app access, allowing renters to use keyless lock/unlock via phone key, remotely precondition the cabin, track charge status and more. Customers who already have a Tesla Profile will have their settings and preferences seamlessly applied, making the rental car feel like their own. The in-app service experience was also redesigned to allow customers to schedule service, access their loaner, track service progress, pay and manage drop- off/pickup. Prospective customers can similarly schedule, locate and test drive a demo vehicle.\n\n## Battery, Powertrain & Manufacturing\n\n## Despite macroeconomic headwinds, our planned factory shutdowns in Q3 and ongoing ramp at new factories, our average vehicle cost was ~$37,500, and we continue to work to reduce the cost further. For very heavy vehicles, a high voltage powertrain architecture brings notable cost savings, which is why Cybertruck will adopt an 800-volt architecture.\n\n### 0\n\n### 100\n\n### 200\n\n### 300\n\n### 400\n\n### 500\n\n### 600\n\n### $36,000\n\n### $36,500\n\n### $37,000\n\n### $37,500\n\n### $38,000\n\n### $38,500\n\n### $39,000\n\n### $39,500\n\n### $40,000\n\n### Q3 2022 Q4 2022 Q1 2023 Q2 2023 Q3 2023"],"full_content":"## Cost of goods sold per vehicle\n\n## C O R E T E C H N O L O G Y\n\n## Cumulative miles driven with FSD Beta (millions)\n\n### 8\n\n## Artificial Intelligence Software and Hardware\n\n## Software that safely performs tasks in the real world is the key focus of our AI development efforts. We have commissioned one of the world's largest supercomputers to accelerate the pace of our AI development, with compute capacity more than doubling compared to Q2. Our large installed base of vehicles continues to generate anonymized video and other data used to develop our FSD Capability features.\n\n## Vehicle and Other Software\n\n## All Tesla rentals through Hertz in the U.S. and Canada now allow Tesla app access, allowing renters to use keyless lock/unlock via phone key, remotely precondition the cabin, track charge status and more. Customers who already have a Tesla Profile will have their settings and preferences seamlessly applied, making the rental car feel like their own. The in-app service experience was also redesigned to allow customers to schedule service, access their loaner, track service progress, pay and manage drop- off/pickup. Prospective customers can similarly schedule, locate and test drive a demo vehicle.\n\n## Battery, Powertrain & Manufacturing\n\n## Despite macroeconomic headwinds, our planned factory shutdowns in Q3 and ongoing ramp at new factories, our average vehicle cost was ~$37,500, and we continue to work to reduce the cost further. For very heavy vehicles, a high voltage powertrain architecture brings notable cost savings, which is why Cybertruck will adopt an 800-volt architecture.\n\n### 0\n\n### 100\n\n### 200\n\n### 300\n\n### 400\n\n### 500\n\n### 600\n\n### $36,000\n\n### $36,500\n\n### $37,000\n\n### $37,500\n\n### $38,000\n\n### $38,500\n\n### $39,000\n\n### $39,500\n\n### $40,000\n\n### Q3 2022 Q4 2022 Q1 2023 Q2 2023 Q3 2023","publish_date":null,"title":"C O R E T E C H N O L O G Y Artificial Intelligence Software and Hardware Software","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/TSLA-Q3-2023-Update-3_p8.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/TSLA-Q3-2023-Update-3_p8.pdf"},"request":{"example_id":"chart/TSLA-Q3-2023-Update-3_p8","product_type":"parse","source_file_path":"TSLA-Q3-2023-Update-3_p8.pdf"},"started_at":"2026-11-04T19:44:55.918085"} +{"completed_at":"2026-11-04T19:44:59.399349","latency_in_ms":2595,"output":{"example_id":"chart/TSLA-Q4-2024-Update_p26","grounded_pages":[],"layout_pages":[],"markdown":"## Vehicle Deliveries (millions of units)\n\n## **K E Y M E T R I C S Q U A R T E R L Y** (Unaudited)\n\n## Operating Cash Flow ($B) Free Cash Flow ($B)\n\n## Net Income ($B) Adjusted EBITDA ($B)\n\n### 26\n\n## 0\\.0\n\n## 0\\.1\n\n## 0\\.2\n\n## 0\\.3\n\n## 0\\.4\n\n## 0\\.5\n\n## 1Q-2022\n\n## 2Q-2022\n\n## 3Q-2022\n\n## 4Q-2022\n\n## 1Q-2023\n\n## 2Q-2023\n\n## 3Q-2023\n\n## 4Q-2023\n\n## 1Q-2024\n\n## 2Q-2024\n\n## 3Q-2024\n\n## 4Q-2024\n\n## \\-3\n\n## \\-2\n\n## \\-1\n\n## 0\n\n## 1\n\n## 2\n\n## 3\n\n## 4\n\n## 5\n\n## 6\n\n## 7\n\n## 1Q-2022\n\n## 2Q-2022\n\n## 3Q-2022\n\n## 4Q-2022\n\n## 1Q-2023\n\n## 2Q-2023\n\n## 3Q-2023\n\n## 4Q-2023\n\n## 1Q-2024\n\n## 2Q-2024\n\n## 3Q-2024\n\n## 4Q-2024\n\n## 0\n\n## 1\n\n## 2\n\n## 3\n\n## 4\n\n## 5\n\n## 6\n\n## 7\n\n## 8\n\n## 1Q-2022\n\n## 2Q-2022\n\n## 3Q-2022\n\n## 4Q-2022\n\n## 1Q-2023\n\n## 2Q-2023\n\n## 3Q-2023\n\n## 4Q-2023\n\n## 1Q-2024\n\n## 2Q-2024\n\n## 3Q-2024\n\n## 4Q-2024","pages":[{"line_numbers":[],"markdown":"## Vehicle Deliveries (millions of units)\n\n## **K E Y M E T R I C S Q U A R T E R L Y** (Unaudited)\n\n## Operating Cash Flow ($B) Free Cash Flow ($B)\n\n## Net Income ($B) Adjusted EBITDA ($B)\n\n### 26\n\n## 0\\.0\n\n## 0\\.1\n\n## 0\\.2\n\n## 0\\.3\n\n## 0\\.4\n\n## 0\\.5\n\n## 1Q-2022\n\n## 2Q-2022\n\n## 3Q-2022\n\n## 4Q-2022\n\n## 1Q-2023\n\n## 2Q-2023\n\n## 3Q-2023\n\n## 4Q-2023\n\n## 1Q-2024\n\n## 2Q-2024\n\n## 3Q-2024\n\n## 4Q-2024\n\n## \\-3\n\n## \\-2\n\n## \\-1\n\n## 0\n\n## 1\n\n## 2\n\n## 3\n\n## 4\n\n## 5\n\n## 6\n\n## 7\n\n## 1Q-2022\n\n## 2Q-2022\n\n## 3Q-2022\n\n## 4Q-2022\n\n## 1Q-2023\n\n## 2Q-2023\n\n## 3Q-2023\n\n## 4Q-2023\n\n## 1Q-2024\n\n## 2Q-2024\n\n## 3Q-2024\n\n## 4Q-2024\n\n## 0\n\n## 1\n\n## 2\n\n## 3\n\n## 4\n\n## 5\n\n## 6\n\n## 7\n\n## 8\n\n## 1Q-2022\n\n## 2Q-2022\n\n## 3Q-2022\n\n## 4Q-2022\n\n## 1Q-2023\n\n## 2Q-2023\n\n## 3Q-2023\n\n## 4Q-2023\n\n## 1Q-2024\n\n## 2Q-2024\n\n## 3Q-2024\n\n## 4Q-2024","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"## Vehicle Deliveries (millions of units)\n\n## **K E Y M E T R I C S Q U A R T E R L Y** (Unaudited)\n\n## Operating Cash Flow ($B) Free Cash Flow ($B)\n\n## Net Income ($B) Adjusted EBITDA ($B)\n\n### 26\n\n## 0\\.0\n\n## 0\\.1\n\n## 0\\.2\n\n## 0\\.3\n\n## 0\\.4\n\n## 0\\.5\n\n## 1Q-2022\n\n## 2Q-2022\n\n## 3Q-2022\n\n## 4Q-2022\n\n## 1Q-2023\n\n## 2Q-2023\n\n## 3Q-2023\n\n## 4Q-2023\n\n## 1Q-2024\n\n## 2Q-2024\n\n## 3Q-2024\n\n## 4Q-2024\n\n## \\-3\n\n## \\-2\n\n## \\-1\n\n## 0\n\n## 1\n\n## 2\n\n## 3\n\n## 4\n\n## 5\n\n## 6\n\n## 7\n\n## 1Q-2022\n\n## 2Q-2022\n\n## 3Q-2022\n\n## 4Q-2022\n\n## 1Q-2023\n\n## 2Q-2023\n\n## 3Q-2023\n\n## 4Q-2023\n\n## 1Q-2024\n\n## 2Q-2024\n\n## 3Q-2024\n\n## 4Q-2024\n\n## 0\n\n## 1\n\n## 2\n\n## 3\n\n## 4\n\n## 5\n\n## 6\n\n## 7\n\n## 8\n\n## 1Q-2022\n\n## 2Q-2022\n\n## 3Q-2022\n\n## 4Q-2022\n\n## 1Q-2023\n\n## 2Q-2023\n\n## 3Q-2023\n\n## 4Q-2023\n\n## 1Q-2024\n\n## 2Q-2024\n\n## 3Q-2024\n\n## 4Q-2024","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/TSLA-Q4-2024-Update_p26.pdf"]},"response":{"errors":[],"extract_id":"extract_a2e6e3cdb59b65c6f98c022cf2cbbb7b","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["## Vehicle Deliveries (millions of units)\n\n## **K E Y M E T R I C S Q U A R T E R L Y** (Unaudited)\n\n## Operating Cash Flow ($B) Free Cash Flow ($B)\n\n## Net Income ($B) Adjusted EBITDA ($B)\n\n### 26\n\n## 0\\.0\n\n## 0\\.1\n\n## 0\\.2\n\n## 0\\.3\n\n## 0\\.4\n\n## 0\\.5\n\n## 1Q-2022\n\n## 2Q-2022\n\n## 3Q-2022\n\n## 4Q-2022\n\n## 1Q-2023\n\n## 2Q-2023\n\n## 3Q-2023\n\n## 4Q-2023\n\n## 1Q-2024\n\n## 2Q-2024\n\n## 3Q-2024\n\n## 4Q-2024\n\n## \\-3\n\n## \\-2\n\n## \\-1\n\n## 0\n\n## 1\n\n## 2\n\n## 3\n\n## 4\n\n## 5\n\n## 6\n\n## 7\n\n## 1Q-2022\n\n## 2Q-2022\n\n## 3Q-2022\n\n## 4Q-2022\n\n## 1Q-2023\n\n## 2Q-2023\n\n## 3Q-2023\n\n## 4Q-2023\n\n## 1Q-2024\n\n## 2Q-2024\n\n## 3Q-2024\n\n## 4Q-2024\n\n## 0\n\n## 1\n\n## 2\n\n## 3\n\n## 4\n\n## 5\n\n## 6\n\n## 7\n\n## 8\n\n## 1Q-2022\n\n## 2Q-2022\n\n## 3Q-2022\n\n## 4Q-2022\n\n## 1Q-2023\n\n## 2Q-2023\n\n## 3Q-2023\n\n## 4Q-2023\n\n## 1Q-2024\n\n## 2Q-2024\n\n## 3Q-2024\n\n## 4Q-2024"],"full_content":"## Vehicle Deliveries (millions of units)\n\n## **K E Y M E T R I C S Q U A R T E R L Y** (Unaudited)\n\n## Operating Cash Flow ($B) Free Cash Flow ($B)\n\n## Net Income ($B) Adjusted EBITDA ($B)\n\n### 26\n\n## 0\\.0\n\n## 0\\.1\n\n## 0\\.2\n\n## 0\\.3\n\n## 0\\.4\n\n## 0\\.5\n\n## 1Q-2022\n\n## 2Q-2022\n\n## 3Q-2022\n\n## 4Q-2022\n\n## 1Q-2023\n\n## 2Q-2023\n\n## 3Q-2023\n\n## 4Q-2023\n\n## 1Q-2024\n\n## 2Q-2024\n\n## 3Q-2024\n\n## 4Q-2024\n\n## \\-3\n\n## \\-2\n\n## \\-1\n\n## 0\n\n## 1\n\n## 2\n\n## 3\n\n## 4\n\n## 5\n\n## 6\n\n## 7\n\n## 1Q-2022\n\n## 2Q-2022\n\n## 3Q-2022\n\n## 4Q-2022\n\n## 1Q-2023\n\n## 2Q-2023\n\n## 3Q-2023\n\n## 4Q-2023\n\n## 1Q-2024\n\n## 2Q-2024\n\n## 3Q-2024\n\n## 4Q-2024\n\n## 0\n\n## 1\n\n## 2\n\n## 3\n\n## 4\n\n## 5\n\n## 6\n\n## 7\n\n## 8\n\n## 1Q-2022\n\n## 2Q-2022\n\n## 3Q-2022\n\n## 4Q-2022\n\n## 1Q-2023\n\n## 2Q-2023\n\n## 3Q-2023\n\n## 4Q-2023\n\n## 1Q-2024\n\n## 2Q-2024\n\n## 3Q-2024\n\n## 4Q-2024","publish_date":null,"title":"K E Y M E T R I C S Q U A R T E R L Y (Unaudited)","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/TSLA-Q4-2024-Update_p26.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/TSLA-Q4-2024-Update_p26.pdf"},"request":{"example_id":"chart/TSLA-Q4-2024-Update_p26","product_type":"parse","source_file_path":"TSLA-Q4-2024-Update_p26.pdf"},"started_at":"2026-11-04T19:44:56.803997"} +{"completed_at":"2026-11-04T19:44:59.700260","latency_in_ms":1861,"output":{"example_id":"chart/TSLA-Q4-2024-Update_p27","grounded_pages":[],"layout_pages":[],"markdown":"## Operating Cash Flow ($B) Free Cash Flow ($B)\n\n## **K E Y M E T R I C S T R A I L I N G 1 2 M O N T H S ( T T M )** (Unaudited)\n\n## Net Income ($B) Adjusted EBITDA ($B)\n\n## Vehicle Deliveries (millions of units)\n\n### 27\n\n## 0\\.0\n\n## 0\\.2\n\n## 0\\.4\n\n## 0\\.6\n\n## 0\\.8\n\n## 1\\.0\n\n## 1\\.2\n\n## 1\\.4\n\n## 1\\.6\n\n## 1\\.8\n\n## 2\\.0\n\n## 1Q-2022\n\n## 2Q-2022\n\n## 3Q-2022\n\n## 4Q-2022\n\n## 1Q-2023\n\n## 2Q-2023\n\n## 3Q-2023\n\n## 4Q-2023\n\n## 1Q-2024\n\n## 2Q-2024\n\n## 3Q-2024\n\n## 4Q-2024\n\n## 0\n\n## 2\n\n## 4\n\n## 6\n\n## 8\n\n## 10\n\n## 12\n\n## 14\n\n## 16\n\n## 18\n\n## 20\n\n## 1Q-2022\n\n## 2Q-2022\n\n## 3Q-2022\n\n## 4Q-2022\n\n## 1Q-2023\n\n## 2Q-2023\n\n## 3Q-2023\n\n## 4Q-2023\n\n## 1Q-2024\n\n## 2Q-2024\n\n## 3Q-2024\n\n## 4Q-2024\n\n## 0\n\n## 2\n\n## 4\n\n## 6\n\n## 8\n\n## 10\n\n## 12\n\n## 14\n\n## 16\n\n## 18\n\n## 20\n\n## 1Q-2022\n\n## 2Q-2022\n\n## 3Q-2022\n\n## 4Q-2022\n\n## 1Q-2023\n\n## 2Q-2023\n\n## 3Q-2023\n\n## 4Q-2023\n\n## 1Q-2024\n\n## 2Q-2024\n\n## 3Q-2024\n\n## 4Q-2024","pages":[{"line_numbers":[],"markdown":"## Operating Cash Flow ($B) Free Cash Flow ($B)\n\n## **K E Y M E T R I C S T R A I L I N G 1 2 M O N T H S ( T T M )** (Unaudited)\n\n## Net Income ($B) Adjusted EBITDA ($B)\n\n## Vehicle Deliveries (millions of units)\n\n### 27\n\n## 0\\.0\n\n## 0\\.2\n\n## 0\\.4\n\n## 0\\.6\n\n## 0\\.8\n\n## 1\\.0\n\n## 1\\.2\n\n## 1\\.4\n\n## 1\\.6\n\n## 1\\.8\n\n## 2\\.0\n\n## 1Q-2022\n\n## 2Q-2022\n\n## 3Q-2022\n\n## 4Q-2022\n\n## 1Q-2023\n\n## 2Q-2023\n\n## 3Q-2023\n\n## 4Q-2023\n\n## 1Q-2024\n\n## 2Q-2024\n\n## 3Q-2024\n\n## 4Q-2024\n\n## 0\n\n## 2\n\n## 4\n\n## 6\n\n## 8\n\n## 10\n\n## 12\n\n## 14\n\n## 16\n\n## 18\n\n## 20\n\n## 1Q-2022\n\n## 2Q-2022\n\n## 3Q-2022\n\n## 4Q-2022\n\n## 1Q-2023\n\n## 2Q-2023\n\n## 3Q-2023\n\n## 4Q-2023\n\n## 1Q-2024\n\n## 2Q-2024\n\n## 3Q-2024\n\n## 4Q-2024\n\n## 0\n\n## 2\n\n## 4\n\n## 6\n\n## 8\n\n## 10\n\n## 12\n\n## 14\n\n## 16\n\n## 18\n\n## 20\n\n## 1Q-2022\n\n## 2Q-2022\n\n## 3Q-2022\n\n## 4Q-2022\n\n## 1Q-2023\n\n## 2Q-2023\n\n## 3Q-2023\n\n## 4Q-2023\n\n## 1Q-2024\n\n## 2Q-2024\n\n## 3Q-2024\n\n## 4Q-2024","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"## Operating Cash Flow ($B) Free Cash Flow ($B)\n\n## **K E Y M E T R I C S T R A I L I N G 1 2 M O N T H S ( T T M )** (Unaudited)\n\n## Net Income ($B) Adjusted EBITDA ($B)\n\n## Vehicle Deliveries (millions of units)\n\n### 27\n\n## 0\\.0\n\n## 0\\.2\n\n## 0\\.4\n\n## 0\\.6\n\n## 0\\.8\n\n## 1\\.0\n\n## 1\\.2\n\n## 1\\.4\n\n## 1\\.6\n\n## 1\\.8\n\n## 2\\.0\n\n## 1Q-2022\n\n## 2Q-2022\n\n## 3Q-2022\n\n## 4Q-2022\n\n## 1Q-2023\n\n## 2Q-2023\n\n## 3Q-2023\n\n## 4Q-2023\n\n## 1Q-2024\n\n## 2Q-2024\n\n## 3Q-2024\n\n## 4Q-2024\n\n## 0\n\n## 2\n\n## 4\n\n## 6\n\n## 8\n\n## 10\n\n## 12\n\n## 14\n\n## 16\n\n## 18\n\n## 20\n\n## 1Q-2022\n\n## 2Q-2022\n\n## 3Q-2022\n\n## 4Q-2022\n\n## 1Q-2023\n\n## 2Q-2023\n\n## 3Q-2023\n\n## 4Q-2023\n\n## 1Q-2024\n\n## 2Q-2024\n\n## 3Q-2024\n\n## 4Q-2024\n\n## 0\n\n## 2\n\n## 4\n\n## 6\n\n## 8\n\n## 10\n\n## 12\n\n## 14\n\n## 16\n\n## 18\n\n## 20\n\n## 1Q-2022\n\n## 2Q-2022\n\n## 3Q-2022\n\n## 4Q-2022\n\n## 1Q-2023\n\n## 2Q-2023\n\n## 3Q-2023\n\n## 4Q-2023\n\n## 1Q-2024\n\n## 2Q-2024\n\n## 3Q-2024\n\n## 4Q-2024","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/TSLA-Q4-2024-Update_p27.pdf"]},"response":{"errors":[],"extract_id":"extract_7ff5ee88eaa81d6354253ee3e01f9f39","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["## Operating Cash Flow ($B) Free Cash Flow ($B)\n\n## **K E Y M E T R I C S T R A I L I N G 1 2 M O N T H S ( T T M )** (Unaudited)\n\n## Net Income ($B) Adjusted EBITDA ($B)\n\n## Vehicle Deliveries (millions of units)\n\n### 27\n\n## 0\\.0\n\n## 0\\.2\n\n## 0\\.4\n\n## 0\\.6\n\n## 0\\.8\n\n## 1\\.0\n\n## 1\\.2\n\n## 1\\.4\n\n## 1\\.6\n\n## 1\\.8\n\n## 2\\.0\n\n## 1Q-2022\n\n## 2Q-2022\n\n## 3Q-2022\n\n## 4Q-2022\n\n## 1Q-2023\n\n## 2Q-2023\n\n## 3Q-2023\n\n## 4Q-2023\n\n## 1Q-2024\n\n## 2Q-2024\n\n## 3Q-2024\n\n## 4Q-2024\n\n## 0\n\n## 2\n\n## 4\n\n## 6\n\n## 8\n\n## 10\n\n## 12\n\n## 14\n\n## 16\n\n## 18\n\n## 20\n\n## 1Q-2022\n\n## 2Q-2022\n\n## 3Q-2022\n\n## 4Q-2022\n\n## 1Q-2023\n\n## 2Q-2023\n\n## 3Q-2023\n\n## 4Q-2023\n\n## 1Q-2024\n\n## 2Q-2024\n\n## 3Q-2024\n\n## 4Q-2024\n\n## 0\n\n## 2\n\n## 4\n\n## 6\n\n## 8\n\n## 10\n\n## 12\n\n## 14\n\n## 16\n\n## 18\n\n## 20\n\n## 1Q-2022\n\n## 2Q-2022\n\n## 3Q-2022\n\n## 4Q-2022\n\n## 1Q-2023\n\n## 2Q-2023\n\n## 3Q-2023\n\n## 4Q-2023\n\n## 1Q-2024\n\n## 2Q-2024\n\n## 3Q-2024\n\n## 4Q-2024"],"full_content":"## Operating Cash Flow ($B) Free Cash Flow ($B)\n\n## **K E Y M E T R I C S T R A I L I N G 1 2 M O N T H S ( T T M )** (Unaudited)\n\n## Net Income ($B) Adjusted EBITDA ($B)\n\n## Vehicle Deliveries (millions of units)\n\n### 27\n\n## 0\\.0\n\n## 0\\.2\n\n## 0\\.4\n\n## 0\\.6\n\n## 0\\.8\n\n## 1\\.0\n\n## 1\\.2\n\n## 1\\.4\n\n## 1\\.6\n\n## 1\\.8\n\n## 2\\.0\n\n## 1Q-2022\n\n## 2Q-2022\n\n## 3Q-2022\n\n## 4Q-2022\n\n## 1Q-2023\n\n## 2Q-2023\n\n## 3Q-2023\n\n## 4Q-2023\n\n## 1Q-2024\n\n## 2Q-2024\n\n## 3Q-2024\n\n## 4Q-2024\n\n## 0\n\n## 2\n\n## 4\n\n## 6\n\n## 8\n\n## 10\n\n## 12\n\n## 14\n\n## 16\n\n## 18\n\n## 20\n\n## 1Q-2022\n\n## 2Q-2022\n\n## 3Q-2022\n\n## 4Q-2022\n\n## 1Q-2023\n\n## 2Q-2023\n\n## 3Q-2023\n\n## 4Q-2023\n\n## 1Q-2024\n\n## 2Q-2024\n\n## 3Q-2024\n\n## 4Q-2024\n\n## 0\n\n## 2\n\n## 4\n\n## 6\n\n## 8\n\n## 10\n\n## 12\n\n## 14\n\n## 16\n\n## 18\n\n## 20\n\n## 1Q-2022\n\n## 2Q-2022\n\n## 3Q-2022\n\n## 4Q-2022\n\n## 1Q-2023\n\n## 2Q-2023\n\n## 3Q-2023\n\n## 4Q-2023\n\n## 1Q-2024\n\n## 2Q-2024\n\n## 3Q-2024\n\n## 4Q-2024","publish_date":null,"title":"K E Y M E T R I C S T R A I L I N G 1 2","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/TSLA-Q4-2024-Update_p27.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/TSLA-Q4-2024-Update_p27.pdf"},"request":{"example_id":"chart/TSLA-Q4-2024-Update_p27","product_type":"parse","source_file_path":"TSLA-Q4-2024-Update_p27.pdf"},"started_at":"2026-11-04T19:44:57.838717"} +{"completed_at":"2026-11-04T19:45:01.208845","latency_in_ms":1804,"output":{"example_id":"chart/TSLA-Q4-2024-Update_p9","grounded_pages":[],"layout_pages":[],"markdown":"## V E H I C L E C A P A C I T Y\n\n## **Current Installed Annual Vehicle Capacity**\n\n## Region Model Capacity Status\n\n## California **Model S / Model X** 100,000 Production\n\n## **Model 3 / Model Y** >550,000 Production\n\n## Shanghai **Model 3 / Model Y** >950,000 Production\n\n## Berlin **Model Y** >375,000 Production\n\n## Texas **Model Y** >250,000 Production\n\n## **Cybertruck** >125,000 Production\n\n## **Cybercab** \\- In development\n\n## Nevada **Tesla Semi** \\- Pilot production\n\n## TBD **Roadster** \\- In development\n\n### Installed capacity ≠ current production rate and there may be limitations discovered as production rates approach capacity. Production rates depend on a variety of factors, including equipment uptime, component supply, downtime related to factory upgrades, regulatory considerations and other factors.\n\n## **Market share of Tesla vehicles by region (TTM)**\n\n### Source: Tesla estimates based on latest available data from ACEA; Autonews.com; CAAM – light-duty vehicles only; TTM = Trailing twelve months\n\n## Preparations are underway across our existing factories for the launch of new products in 2025, including more affordable models. New Model Y has launched on the configurator globally with production ramp and deliveries beginning later in Q1.\n\n## **US: California, Nevada and Texas**\n\n## Semi Factory construction continued in Q4 and recently completed roof and wall enclosure of the main building area. First truck builds are scheduled to start by end of 2025 with ramp beginning in early 2026. Preparation is underway for Cybercab lines at Gigafactory Texas with volume production planned for 2026.\n\n## Given its unique functionality – including Powershare, stainless steel exterior, enhanced durability and bioweapon defense mode – we deployed a fleet of Cybertrucks equipped with Starlink in the Los Angeles area to help first responders and those impacted by the fires get access to electricity and Wi-Fi. We expect Cybertruck to be eligible for the IRA consumer tax credit, helping to improve affordability and access for even more customers.\n\n## **APAC: Shanghai**\n\n## In Q4, we achieved record deliveries in China as Model Y became the best-selling vehicle for the full year. Tesla also became the fastest growing brand in South Korea, and we launched vehicle sales in the Philippines.\n\n## **Europe: Berlin-Brandenburg**\n\n## In 2024, Model Y was the best-selling vehicle of any type in Denmark, Norway, Sweden, Switzerland and the Netherlands and we expect Model Y to have been the second best-selling vehicle of any type in Europe. Tesla was the most sold brand in Norway for the fourth year in a row with Model Y and Model 3 the best- and second-best selling cars of any type in 2024.\n\n### 9\n\n### 0%\n\n### 1%\n\n### 2%\n\n### 3%\n\n### 4%\n\n### US/Canada Europe China","pages":[{"line_numbers":[],"markdown":"## V E H I C L E C A P A C I T Y\n\n## **Current Installed Annual Vehicle Capacity**\n\n## Region Model Capacity Status\n\n## California **Model S / Model X** 100,000 Production\n\n## **Model 3 / Model Y** >550,000 Production\n\n## Shanghai **Model 3 / Model Y** >950,000 Production\n\n## Berlin **Model Y** >375,000 Production\n\n## Texas **Model Y** >250,000 Production\n\n## **Cybertruck** >125,000 Production\n\n## **Cybercab** \\- In development\n\n## Nevada **Tesla Semi** \\- Pilot production\n\n## TBD **Roadster** \\- In development\n\n### Installed capacity ≠ current production rate and there may be limitations discovered as production rates approach capacity. Production rates depend on a variety of factors, including equipment uptime, component supply, downtime related to factory upgrades, regulatory considerations and other factors.\n\n## **Market share of Tesla vehicles by region (TTM)**\n\n### Source: Tesla estimates based on latest available data from ACEA; Autonews.com; CAAM – light-duty vehicles only; TTM = Trailing twelve months\n\n## Preparations are underway across our existing factories for the launch of new products in 2025, including more affordable models. New Model Y has launched on the configurator globally with production ramp and deliveries beginning later in Q1.\n\n## **US: California, Nevada and Texas**\n\n## Semi Factory construction continued in Q4 and recently completed roof and wall enclosure of the main building area. First truck builds are scheduled to start by end of 2025 with ramp beginning in early 2026. Preparation is underway for Cybercab lines at Gigafactory Texas with volume production planned for 2026.\n\n## Given its unique functionality – including Powershare, stainless steel exterior, enhanced durability and bioweapon defense mode – we deployed a fleet of Cybertrucks equipped with Starlink in the Los Angeles area to help first responders and those impacted by the fires get access to electricity and Wi-Fi. We expect Cybertruck to be eligible for the IRA consumer tax credit, helping to improve affordability and access for even more customers.\n\n## **APAC: Shanghai**\n\n## In Q4, we achieved record deliveries in China as Model Y became the best-selling vehicle for the full year. Tesla also became the fastest growing brand in South Korea, and we launched vehicle sales in the Philippines.\n\n## **Europe: Berlin-Brandenburg**\n\n## In 2024, Model Y was the best-selling vehicle of any type in Denmark, Norway, Sweden, Switzerland and the Netherlands and we expect Model Y to have been the second best-selling vehicle of any type in Europe. Tesla was the most sold brand in Norway for the fourth year in a row with Model Y and Model 3 the best- and second-best selling cars of any type in 2024.\n\n### 9\n\n### 0%\n\n### 1%\n\n### 2%\n\n### 3%\n\n### 4%\n\n### US/Canada Europe China","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"## V E H I C L E C A P A C I T Y\n\n## **Current Installed Annual Vehicle Capacity**\n\n## Region Model Capacity Status\n\n## California **Model S / Model X** 100,000 Production\n\n## **Model 3 / Model Y** >550,000 Production\n\n## Shanghai **Model 3 / Model Y** >950,000 Production\n\n## Berlin **Model Y** >375,000 Production\n\n## Texas **Model Y** >250,000 Production\n\n## **Cybertruck** >125,000 Production\n\n## **Cybercab** \\- In development\n\n## Nevada **Tesla Semi** \\- Pilot production\n\n## TBD **Roadster** \\- In development\n\n### Installed capacity ≠ current production rate and there may be limitations discovered as production rates approach capacity. Production rates depend on a variety of factors, including equipment uptime, component supply, downtime related to factory upgrades, regulatory considerations and other factors.\n\n## **Market share of Tesla vehicles by region (TTM)**\n\n### Source: Tesla estimates based on latest available data from ACEA; Autonews.com; CAAM – light-duty vehicles only; TTM = Trailing twelve months\n\n## Preparations are underway across our existing factories for the launch of new products in 2025, including more affordable models. New Model Y has launched on the configurator globally with production ramp and deliveries beginning later in Q1.\n\n## **US: California, Nevada and Texas**\n\n## Semi Factory construction continued in Q4 and recently completed roof and wall enclosure of the main building area. First truck builds are scheduled to start by end of 2025 with ramp beginning in early 2026. Preparation is underway for Cybercab lines at Gigafactory Texas with volume production planned for 2026.\n\n## Given its unique functionality – including Powershare, stainless steel exterior, enhanced durability and bioweapon defense mode – we deployed a fleet of Cybertrucks equipped with Starlink in the Los Angeles area to help first responders and those impacted by the fires get access to electricity and Wi-Fi. We expect Cybertruck to be eligible for the IRA consumer tax credit, helping to improve affordability and access for even more customers.\n\n## **APAC: Shanghai**\n\n## In Q4, we achieved record deliveries in China as Model Y became the best-selling vehicle for the full year. Tesla also became the fastest growing brand in South Korea, and we launched vehicle sales in the Philippines.\n\n## **Europe: Berlin-Brandenburg**\n\n## In 2024, Model Y was the best-selling vehicle of any type in Denmark, Norway, Sweden, Switzerland and the Netherlands and we expect Model Y to have been the second best-selling vehicle of any type in Europe. Tesla was the most sold brand in Norway for the fourth year in a row with Model Y and Model 3 the best- and second-best selling cars of any type in 2024.\n\n### 9\n\n### 0%\n\n### 1%\n\n### 2%\n\n### 3%\n\n### 4%\n\n### US/Canada Europe China","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/TSLA-Q4-2024-Update_p9.pdf"]},"response":{"errors":[],"extract_id":"extract_df5005251104672452346ce9908f8eba","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["## V E H I C L E C A P A C I T Y\n\n## **Current Installed Annual Vehicle Capacity**\n\n## Region Model Capacity Status\n\n## California **Model S / Model X** 100,000 Production\n\n## **Model 3 / Model Y** >550,000 Production\n\n## Shanghai **Model 3 / Model Y** >950,000 Production\n\n## Berlin **Model Y** >375,000 Production\n\n## Texas **Model Y** >250,000 Production\n\n## **Cybertruck** >125,000 Production\n\n## **Cybercab** \\- In development\n\n## Nevada **Tesla Semi** \\- Pilot production\n\n## TBD **Roadster** \\- In development\n\n### Installed capacity ≠ current production rate and there may be limitations discovered as production rates approach capacity. Production rates depend on a variety of factors, including equipment uptime, component supply, downtime related to factory upgrades, regulatory considerations and other factors.\n\n## **Market share of Tesla vehicles by region (TTM)**\n\n### Source: Tesla estimates based on latest available data from ACEA; Autonews.com; CAAM – light-duty vehicles only; TTM = Trailing twelve months\n\n## Preparations are underway across our existing factories for the launch of new products in 2025, including more affordable models. New Model Y has launched on the configurator globally with production ramp and deliveries beginning later in Q1.\n\n## **US: California, Nevada and Texas**\n\n## Semi Factory construction continued in Q4 and recently completed roof and wall enclosure of the main building area. First truck builds are scheduled to start by end of 2025 with ramp beginning in early 2026. Preparation is underway for Cybercab lines at Gigafactory Texas with volume production planned for 2026.\n\n## Given its unique functionality – including Powershare, stainless steel exterior, enhanced durability and bioweapon defense mode – we deployed a fleet of Cybertrucks equipped with Starlink in the Los Angeles area to help first responders and those impacted by the fires get access to electricity and Wi-Fi. We expect Cybertruck to be eligible for the IRA consumer tax credit, helping to improve affordability and access for even more customers.\n\n## **APAC: Shanghai**\n\n## In Q4, we achieved record deliveries in China as Model Y became the best-selling vehicle for the full year. Tesla also became the fastest growing brand in South Korea, and we launched vehicle sales in the Philippines.\n\n## **Europe: Berlin-Brandenburg**\n\n## In 2024, Model Y was the best-selling vehicle of any type in Denmark, Norway, Sweden, Switzerland and the Netherlands and we expect Model Y to have been the second best-selling vehicle of any type in Europe. Tesla was the most sold brand in Norway for the fourth year in a row with Model Y and Model 3 the best- and second-best selling cars of any type in 2024.\n\n### 9\n\n### 0%\n\n### 1%\n\n### 2%\n\n### 3%\n\n### 4%\n\n### US/Canada Europe China"],"full_content":"## V E H I C L E C A P A C I T Y\n\n## **Current Installed Annual Vehicle Capacity**\n\n## Region Model Capacity Status\n\n## California **Model S / Model X** 100,000 Production\n\n## **Model 3 / Model Y** >550,000 Production\n\n## Shanghai **Model 3 / Model Y** >950,000 Production\n\n## Berlin **Model Y** >375,000 Production\n\n## Texas **Model Y** >250,000 Production\n\n## **Cybertruck** >125,000 Production\n\n## **Cybercab** \\- In development\n\n## Nevada **Tesla Semi** \\- Pilot production\n\n## TBD **Roadster** \\- In development\n\n### Installed capacity ≠ current production rate and there may be limitations discovered as production rates approach capacity. Production rates depend on a variety of factors, including equipment uptime, component supply, downtime related to factory upgrades, regulatory considerations and other factors.\n\n## **Market share of Tesla vehicles by region (TTM)**\n\n### Source: Tesla estimates based on latest available data from ACEA; Autonews.com; CAAM – light-duty vehicles only; TTM = Trailing twelve months\n\n## Preparations are underway across our existing factories for the launch of new products in 2025, including more affordable models. New Model Y has launched on the configurator globally with production ramp and deliveries beginning later in Q1.\n\n## **US: California, Nevada and Texas**\n\n## Semi Factory construction continued in Q4 and recently completed roof and wall enclosure of the main building area. First truck builds are scheduled to start by end of 2025 with ramp beginning in early 2026. Preparation is underway for Cybercab lines at Gigafactory Texas with volume production planned for 2026.\n\n## Given its unique functionality – including Powershare, stainless steel exterior, enhanced durability and bioweapon defense mode – we deployed a fleet of Cybertrucks equipped with Starlink in the Los Angeles area to help first responders and those impacted by the fires get access to electricity and Wi-Fi. We expect Cybertruck to be eligible for the IRA consumer tax credit, helping to improve affordability and access for even more customers.\n\n## **APAC: Shanghai**\n\n## In Q4, we achieved record deliveries in China as Model Y became the best-selling vehicle for the full year. Tesla also became the fastest growing brand in South Korea, and we launched vehicle sales in the Philippines.\n\n## **Europe: Berlin-Brandenburg**\n\n## In 2024, Model Y was the best-selling vehicle of any type in Denmark, Norway, Sweden, Switzerland and the Netherlands and we expect Model Y to have been the second best-selling vehicle of any type in Europe. Tesla was the most sold brand in Norway for the fourth year in a row with Model Y and Model 3 the best- and second-best selling cars of any type in 2024.\n\n### 9\n\n### 0%\n\n### 1%\n\n### 2%\n\n### 3%\n\n### 4%\n\n### US/Canada Europe China","publish_date":null,"title":"V E H I C L E C A P A C I T Y Current Installed Annual Vehicle Capacity","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/TSLA-Q4-2024-Update_p9.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/TSLA-Q4-2024-Update_p9.pdf"},"request":{"example_id":"chart/TSLA-Q4-2024-Update_p9","product_type":"parse","source_file_path":"TSLA-Q4-2024-Update_p9.pdf"},"started_at":"2026-11-04T19:44:59.404277"} +{"completed_at":"2026-11-04T19:45:02.027301","latency_in_ms":1334,"output":{"example_id":"chart/Taking-Stock-2025_p20","grounded_pages":[],"layout_pages":[],"markdown":"RHODIUM GROUP | ENERGY & CLIMATE TAKING STOCK 2025\n\n20\n\nFIGURE 9 Liquified natural gas exports Trillion cubic feet\n\nSource: Rhodium Group Note: The high, mid, and low ranges reflect uncertainty around future fossil fuel prices, economic growth, clean energy technology costs, and growth in LNG export capacity.\n\nInternational trade has an even bigger effect on US production of oil and other liquid fuels. Domestic consumption varies little, mostly due to the slow electrification of the vehicle fleet. Across the economy, domestic demand for liquid fuels increases by 1-5% in 2030 compared to 2024 levels and changes by -5% to +5% in 2040 (Figure 10). Domestic production levels vary widely across scenarios, from a 21% decrease relative to 2024 levels in 2040 in the low emissions case to a 40% increase in 2040 in the high emissions case. The level of net exports mostly causes this variation in production.\n\nMid: 10 Tcf, +129%\n\nLow: 8 Tcf, +94% from 2024 levels\n\nHigh: 11 Tcf, +150%\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n2024 2026 2028 2030 2032 2034 2036 2038 2040","pages":[{"line_numbers":[],"markdown":"RHODIUM GROUP | ENERGY & CLIMATE TAKING STOCK 2025\n\n20\n\nFIGURE 9 Liquified natural gas exports Trillion cubic feet\n\nSource: Rhodium Group Note: The high, mid, and low ranges reflect uncertainty around future fossil fuel prices, economic growth, clean energy technology costs, and growth in LNG export capacity.\n\nInternational trade has an even bigger effect on US production of oil and other liquid fuels. Domestic consumption varies little, mostly due to the slow electrification of the vehicle fleet. Across the economy, domestic demand for liquid fuels increases by 1-5% in 2030 compared to 2024 levels and changes by -5% to +5% in 2040 (Figure 10). Domestic production levels vary widely across scenarios, from a 21% decrease relative to 2024 levels in 2040 in the low emissions case to a 40% increase in 2040 in the high emissions case. The level of net exports mostly causes this variation in production.\n\nMid: 10 Tcf, +129%\n\nLow: 8 Tcf, +94% from 2024 levels\n\nHigh: 11 Tcf, +150%\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n2024 2026 2028 2030 2032 2034 2036 2038 2040","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"RHODIUM GROUP | ENERGY & CLIMATE TAKING STOCK 2025\n\n20\n\nFIGURE 9 Liquified natural gas exports Trillion cubic feet\n\nSource: Rhodium Group Note: The high, mid, and low ranges reflect uncertainty around future fossil fuel prices, economic growth, clean energy technology costs, and growth in LNG export capacity.\n\nInternational trade has an even bigger effect on US production of oil and other liquid fuels. Domestic consumption varies little, mostly due to the slow electrification of the vehicle fleet. Across the economy, domestic demand for liquid fuels increases by 1-5% in 2030 compared to 2024 levels and changes by -5% to +5% in 2040 (Figure 10). Domestic production levels vary widely across scenarios, from a 21% decrease relative to 2024 levels in 2040 in the low emissions case to a 40% increase in 2040 in the high emissions case. The level of net exports mostly causes this variation in production.\n\nMid: 10 Tcf, +129%\n\nLow: 8 Tcf, +94% from 2024 levels\n\nHigh: 11 Tcf, +150%\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n2024 2026 2028 2030 2032 2034 2036 2038 2040","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Taking-Stock-2025_p20.pdf"]},"response":{"errors":[],"extract_id":"extract_cea3e4ad15ca97c78d79b93032cb72f2","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["RHODIUM GROUP | ENERGY & CLIMATE TAKING STOCK 2025\n\n20\n\nFIGURE 9 Liquified natural gas exports Trillion cubic feet\n\nSource: Rhodium Group Note: The high, mid, and low ranges reflect uncertainty around future fossil fuel prices, economic growth, clean energy technology costs, and growth in LNG export capacity.\n\nInternational trade has an even bigger effect on US production of oil and other liquid fuels. Domestic consumption varies little, mostly due to the slow electrification of the vehicle fleet. Across the economy, domestic demand for liquid fuels increases by 1-5% in 2030 compared to 2024 levels and changes by -5% to +5% in 2040 (Figure 10). Domestic production levels vary widely across scenarios, from a 21% decrease relative to 2024 levels in 2040 in the low emissions case to a 40% increase in 2040 in the high emissions case. The level of net exports mostly causes this variation in production.\n\nMid: 10 Tcf, +129%\n\nLow: 8 Tcf, +94% from 2024 levels\n\nHigh: 11 Tcf, +150%\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n2024 2026 2028 2030 2032 2034 2036 2038 2040"],"full_content":"RHODIUM GROUP | ENERGY & CLIMATE TAKING STOCK 2025\n\n20\n\nFIGURE 9 Liquified natural gas exports Trillion cubic feet\n\nSource: Rhodium Group Note: The high, mid, and low ranges reflect uncertainty around future fossil fuel prices, economic growth, clean energy technology costs, and growth in LNG export capacity.\n\nInternational trade has an even bigger effect on US production of oil and other liquid fuels. Domestic consumption varies little, mostly due to the slow electrification of the vehicle fleet. Across the economy, domestic demand for liquid fuels increases by 1-5% in 2030 compared to 2024 levels and changes by -5% to +5% in 2040 (Figure 10). Domestic production levels vary widely across scenarios, from a 21% decrease relative to 2024 levels in 2040 in the low emissions case to a 40% increase in 2040 in the high emissions case. The level of net exports mostly causes this variation in production.\n\nMid: 10 Tcf, +129%\n\nLow: 8 Tcf, +94% from 2024 levels\n\nHigh: 11 Tcf, +150%\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n2024 2026 2028 2030 2032 2034 2036 2038 2040","publish_date":null,"title":"Liquified natural gas exports","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Taking-Stock-2025_p20.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Taking-Stock-2025_p20.pdf"},"request":{"example_id":"chart/Taking-Stock-2025_p20","product_type":"parse","source_file_path":"Taking-Stock-2025_p20.pdf"},"started_at":"2026-11-04T19:45:00.692496"} +{"completed_at":"2026-11-04T19:45:04.467483","latency_in_ms":2438,"output":{"example_id":"chart/Technology ecosystem diversity 2H25 update_p10","grounded_pages":[],"layout_pages":[],"markdown":"### 10\n\nOmdia commissioned research, sponsored by Ericsson\n\nFigure 3: HHI by markets, share of revenue\n\nSource: Omdia","pages":[{"line_numbers":[],"markdown":"### 10\n\nOmdia commissioned research, sponsored by Ericsson\n\nFigure 3: HHI by markets, share of revenue\n\nSource: Omdia","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"### 10\n\nOmdia commissioned research, sponsored by Ericsson\n\nFigure 3: HHI by markets, share of revenue\n\nSource: Omdia","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Technology%20ecosystem%20diversity%202H25%20update_p10.pdf"]},"response":{"errors":[],"extract_id":"extract_de3a4e70745f1d88c7c563652ae67795","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["### 10\n\nOmdia commissioned research, sponsored by Ericsson\n\nFigure 3: HHI by markets, share of revenue\n\nSource: Omdia"],"full_content":"### 10\n\nOmdia commissioned research, sponsored by Ericsson\n\nFigure 3: HHI by markets, share of revenue\n\nSource: Omdia","publish_date":null,"title":"10","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Technology%20ecosystem%20diversity%202H25%20update_p10.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Technology%20ecosystem%20diversity%202H25%20update_p10.pdf"},"request":{"example_id":"chart/Technology ecosystem diversity 2H25 update_p10","product_type":"parse","source_file_path":"Technology ecosystem diversity 2H25 update_p10.pdf"},"started_at":"2026-11-04T19:45:02.028535"} +{"completed_at":"2026-11-04T19:45:04.517588","latency_in_ms":1196,"output":{"example_id":"chart/Technology ecosystem diversity 2H25 update_p13","grounded_pages":[],"layout_pages":[],"markdown":"### 13\n\nOmdia commissioned research, sponsored by Ericsson\n\n### Several vendors provide products and solutions across several mobile network domains. Vendors that operate in more than one domain are counted for each domain where we have identified an active partnership.\n\n### Overall, as of June 2025, the US and the UK had 44 active vendor partnerships across the four mobile network domains; Japan had 30, China and India 23, South Korea 21, Canada 19, and Australia 17. The ecosystem of vendors remains largely the same as in 2024, but a few new active vendors were identified in Australia, the US, the UK, and Canada.\n\nFigure 4: Active vendors in the mobile ecosystem, by domain and by country\n\nSource: Omdia\n\n### Omdia conducted a similar assessment in the same countries for three fixed network domains: copper and fiber access, switching and routing, and optical networking. As in mobile domains, in fixed domains there were, unsurprisingly, more active vendors in the larger markets (the US, China, India, and Japan) than in the smaller markets.\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n35\n\n40\n\n45\n\n50\n\nAustralia Canada Japan South Korea\n\nUS China UK India\n\nRAN Mobile core OSS/BSS MEC\n\n**Number of active vendors**","pages":[{"line_numbers":[],"markdown":"### 13\n\nOmdia commissioned research, sponsored by Ericsson\n\n### Several vendors provide products and solutions across several mobile network domains. Vendors that operate in more than one domain are counted for each domain where we have identified an active partnership.\n\n### Overall, as of June 2025, the US and the UK had 44 active vendor partnerships across the four mobile network domains; Japan had 30, China and India 23, South Korea 21, Canada 19, and Australia 17. The ecosystem of vendors remains largely the same as in 2024, but a few new active vendors were identified in Australia, the US, the UK, and Canada.\n\nFigure 4: Active vendors in the mobile ecosystem, by domain and by country\n\nSource: Omdia\n\n### Omdia conducted a similar assessment in the same countries for three fixed network domains: copper and fiber access, switching and routing, and optical networking. As in mobile domains, in fixed domains there were, unsurprisingly, more active vendors in the larger markets (the US, China, India, and Japan) than in the smaller markets.\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n35\n\n40\n\n45\n\n50\n\nAustralia Canada Japan South Korea\n\nUS China UK India\n\nRAN Mobile core OSS/BSS MEC\n\n**Number of active vendors**","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"### 13\n\nOmdia commissioned research, sponsored by Ericsson\n\n### Several vendors provide products and solutions across several mobile network domains. Vendors that operate in more than one domain are counted for each domain where we have identified an active partnership.\n\n### Overall, as of June 2025, the US and the UK had 44 active vendor partnerships across the four mobile network domains; Japan had 30, China and India 23, South Korea 21, Canada 19, and Australia 17. The ecosystem of vendors remains largely the same as in 2024, but a few new active vendors were identified in Australia, the US, the UK, and Canada.\n\nFigure 4: Active vendors in the mobile ecosystem, by domain and by country\n\nSource: Omdia\n\n### Omdia conducted a similar assessment in the same countries for three fixed network domains: copper and fiber access, switching and routing, and optical networking. As in mobile domains, in fixed domains there were, unsurprisingly, more active vendors in the larger markets (the US, China, India, and Japan) than in the smaller markets.\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n35\n\n40\n\n45\n\n50\n\nAustralia Canada Japan South Korea\n\nUS China UK India\n\nRAN Mobile core OSS/BSS MEC\n\n**Number of active vendors**","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Technology%20ecosystem%20diversity%202H25%20update_p13.pdf"]},"response":{"errors":[],"extract_id":"extract_6f779302668e81d9cf9044bc21fc01b7","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["### 13\n\nOmdia commissioned research, sponsored by Ericsson\n\n### Several vendors provide products and solutions across several mobile network domains. Vendors that operate in more than one domain are counted for each domain where we have identified an active partnership.\n\n### Overall, as of June 2025, the US and the UK had 44 active vendor partnerships across the four mobile network domains; Japan had 30, China and India 23, South Korea 21, Canada 19, and Australia 17. The ecosystem of vendors remains largely the same as in 2024, but a few new active vendors were identified in Australia, the US, the UK, and Canada.\n\nFigure 4: Active vendors in the mobile ecosystem, by domain and by country\n\nSource: Omdia\n\n### Omdia conducted a similar assessment in the same countries for three fixed network domains: copper and fiber access, switching and routing, and optical networking. As in mobile domains, in fixed domains there were, unsurprisingly, more active vendors in the larger markets (the US, China, India, and Japan) than in the smaller markets.\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n35\n\n40\n\n45\n\n50\n\nAustralia Canada Japan South Korea\n\nUS China UK India\n\nRAN Mobile core OSS/BSS MEC\n\n**Number of active vendors**"],"full_content":"### 13\n\nOmdia commissioned research, sponsored by Ericsson\n\n### Several vendors provide products and solutions across several mobile network domains. Vendors that operate in more than one domain are counted for each domain where we have identified an active partnership.\n\n### Overall, as of June 2025, the US and the UK had 44 active vendor partnerships across the four mobile network domains; Japan had 30, China and India 23, South Korea 21, Canada 19, and Australia 17. The ecosystem of vendors remains largely the same as in 2024, but a few new active vendors were identified in Australia, the US, the UK, and Canada.\n\nFigure 4: Active vendors in the mobile ecosystem, by domain and by country\n\nSource: Omdia\n\n### Omdia conducted a similar assessment in the same countries for three fixed network domains: copper and fiber access, switching and routing, and optical networking. As in mobile domains, in fixed domains there were, unsurprisingly, more active vendors in the larger markets (the US, China, India, and Japan) than in the smaller markets.\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n35\n\n40\n\n45\n\n50\n\nAustralia Canada Japan South Korea\n\nUS China UK India\n\nRAN Mobile core OSS/BSS MEC\n\n**Number of active vendors**","publish_date":null,"title":"13 Several vendors provide products and solutions across several mobile network domains. Vendors that operate in more than one domain","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Technology%20ecosystem%20diversity%202H25%20update_p13.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Technology%20ecosystem%20diversity%202H25%20update_p13.pdf"},"request":{"example_id":"chart/Technology ecosystem diversity 2H25 update_p13","product_type":"parse","source_file_path":"Technology ecosystem diversity 2H25 update_p13.pdf"},"started_at":"2026-11-04T19:45:03.321304"} +{"completed_at":"2026-11-04T19:45:06.132480","latency_in_ms":1640,"output":{"example_id":"chart/Technopak_Industry_Report_p12","grounded_pages":[],"layout_pages":[],"markdown":"12\n\nprovided under the Scheme. Under PMAY-U, 1.18 crore houses have been sanctioned while more than 85.5 lakh houses have already been constructed and delivered to the beneficiaries.\n\n**Urbanisation and Smart Cities** In the last decade, the urban population in India has grown at an annual rate of 2.4%. The urban population of India is projected to reach approximately 41% by CY2030 from 31% in CY2010. Further, the number of metropolitan cities in India are estimated to increase from 46 as per Census 2011 to 68 in CY2030. The Smart City Project was launched in June 2015 by GoI with a plan to build 100 smart cities across India,. As of 3 rd July 2024, the 100 Cities have completed 7,188 projects (90% of total projects) amounting to INR 1,44,237 crore as part of the mission. The balance 830 projects amounting to INR 19,926 crore are also in advanced stages of completion. On the financial progress, the mission has an allocated GoI budget of INR 48,000 crore for the 100 Cities. As on date, GOI has released INR 46,585 crore (~97% of the allocated GoI budget) to 100 cities.. **Climate Change and Built Environment** At the UN Climate Change Conference in 2021, India committed to achieving net-zero emissions by 2070, aiming to reduce projected carbon emissions by 1 billion tonnes and decrease the carbon intensity of its economy by 45% by 2030. Transforming the built environment, which contributes nearly 40% of global energy-related greenhouse gas emissions, is central to achieving these targets. With rising heat levels and pollution in India driving the demand for consumer durables, the industry is pushing towards innovation and managing escalating energy needs sustainably. With half of the buildings projected to exist in India by 2050 yet to be constructed, there is a significant opportunity to incorporate climate-friendly, energy-saving solutions into new developments. This alignment with sustainability goals highlights the critical role of consumer durables in addressing evolving consumer needs while mitigating climate impacts. One of the most effective solutions to combat climate change is to reduce greenhouse gas emissions and decrease reliance on fossil fuels. Harnessing renewable energy offers significant benefits in addressing climate change issues. India ranks 4th globally in renewable energy capacity, with notable achievements in wind and solar power. The country has set an ambitious target of 500 GW of non-fossil fuel-based energy by 2030, representing the world’s largest renewable energy expansion plan.India’s solar capacity has surged 30-fold in the past nine years, reaching 90.76 GW by September 2024, with a potential of 748 GW. The country’s renewable energy capacity has grown by 128% since FY2014, supported by initiatives such as the \"Development of Solar Parks\" scheme.\n\n_Exhibit 2.4: Solar Energy Capacity Forecast (FY)_\n\n_Source: Secondary Research_ **Policy Reforms** Various policy reforms have been incorporated by GoI to boost production, exports and promote ease of doing business. Some of these policies include:\n\n• **PLI Scheme:** Under PLI scheme, GoI aims at achieving large-scale electronics manufacturing in India. Investment of INR 1.23 lakh crore has been realized till March 2024. There is a PLI Scheme specifically for white goods manufacturers in India under which, GoI proposes a financial incentive to boost domestic manufacturing and attract large investments in white goods manufacturing value chain, with an objective of removing sectoral disabilities, creating economies of scale, enhancing exports, creating a robust component ecosystem and employment generation. This scheme is expected to be implemented\n\n**34\\.6 40.1**\n\n**54\\.0 65.0**\n\n**90\\.8**\n\n**113\\.3**\n\n**145\\.0**\n\n**185\\.6**\n\n2020 2021 2022 2023 2024 2025P 2026P 2027P","pages":[{"line_numbers":[],"markdown":"12\n\nprovided under the Scheme. Under PMAY-U, 1.18 crore houses have been sanctioned while more than 85.5 lakh houses have already been constructed and delivered to the beneficiaries.\n\n**Urbanisation and Smart Cities** In the last decade, the urban population in India has grown at an annual rate of 2.4%. The urban population of India is projected to reach approximately 41% by CY2030 from 31% in CY2010. Further, the number of metropolitan cities in India are estimated to increase from 46 as per Census 2011 to 68 in CY2030. The Smart City Project was launched in June 2015 by GoI with a plan to build 100 smart cities across India,. As of 3 rd July 2024, the 100 Cities have completed 7,188 projects (90% of total projects) amounting to INR 1,44,237 crore as part of the mission. The balance 830 projects amounting to INR 19,926 crore are also in advanced stages of completion. On the financial progress, the mission has an allocated GoI budget of INR 48,000 crore for the 100 Cities. As on date, GOI has released INR 46,585 crore (~97% of the allocated GoI budget) to 100 cities.. **Climate Change and Built Environment** At the UN Climate Change Conference in 2021, India committed to achieving net-zero emissions by 2070, aiming to reduce projected carbon emissions by 1 billion tonnes and decrease the carbon intensity of its economy by 45% by 2030. Transforming the built environment, which contributes nearly 40% of global energy-related greenhouse gas emissions, is central to achieving these targets. With rising heat levels and pollution in India driving the demand for consumer durables, the industry is pushing towards innovation and managing escalating energy needs sustainably. With half of the buildings projected to exist in India by 2050 yet to be constructed, there is a significant opportunity to incorporate climate-friendly, energy-saving solutions into new developments. This alignment with sustainability goals highlights the critical role of consumer durables in addressing evolving consumer needs while mitigating climate impacts. One of the most effective solutions to combat climate change is to reduce greenhouse gas emissions and decrease reliance on fossil fuels. Harnessing renewable energy offers significant benefits in addressing climate change issues. India ranks 4th globally in renewable energy capacity, with notable achievements in wind and solar power. The country has set an ambitious target of 500 GW of non-fossil fuel-based energy by 2030, representing the world’s largest renewable energy expansion plan.India’s solar capacity has surged 30-fold in the past nine years, reaching 90.76 GW by September 2024, with a potential of 748 GW. The country’s renewable energy capacity has grown by 128% since FY2014, supported by initiatives such as the \"Development of Solar Parks\" scheme.\n\n_Exhibit 2.4: Solar Energy Capacity Forecast (FY)_\n\n_Source: Secondary Research_ **Policy Reforms** Various policy reforms have been incorporated by GoI to boost production, exports and promote ease of doing business. Some of these policies include:\n\n• **PLI Scheme:** Under PLI scheme, GoI aims at achieving large-scale electronics manufacturing in India. Investment of INR 1.23 lakh crore has been realized till March 2024. There is a PLI Scheme specifically for white goods manufacturers in India under which, GoI proposes a financial incentive to boost domestic manufacturing and attract large investments in white goods manufacturing value chain, with an objective of removing sectoral disabilities, creating economies of scale, enhancing exports, creating a robust component ecosystem and employment generation. This scheme is expected to be implemented\n\n**34\\.6 40.1**\n\n**54\\.0 65.0**\n\n**90\\.8**\n\n**113\\.3**\n\n**145\\.0**\n\n**185\\.6**\n\n2020 2021 2022 2023 2024 2025P 2026P 2027P","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"12\n\nprovided under the Scheme. Under PMAY-U, 1.18 crore houses have been sanctioned while more than 85.5 lakh houses have already been constructed and delivered to the beneficiaries.\n\n**Urbanisation and Smart Cities** In the last decade, the urban population in India has grown at an annual rate of 2.4%. The urban population of India is projected to reach approximately 41% by CY2030 from 31% in CY2010. Further, the number of metropolitan cities in India are estimated to increase from 46 as per Census 2011 to 68 in CY2030. The Smart City Project was launched in June 2015 by GoI with a plan to build 100 smart cities across India,. As of 3 rd July 2024, the 100 Cities have completed 7,188 projects (90% of total projects) amounting to INR 1,44,237 crore as part of the mission. The balance 830 projects amounting to INR 19,926 crore are also in advanced stages of completion. On the financial progress, the mission has an allocated GoI budget of INR 48,000 crore for the 100 Cities. As on date, GOI has released INR 46,585 crore (~97% of the allocated GoI budget) to 100 cities.. **Climate Change and Built Environment** At the UN Climate Change Conference in 2021, India committed to achieving net-zero emissions by 2070, aiming to reduce projected carbon emissions by 1 billion tonnes and decrease the carbon intensity of its economy by 45% by 2030. Transforming the built environment, which contributes nearly 40% of global energy-related greenhouse gas emissions, is central to achieving these targets. With rising heat levels and pollution in India driving the demand for consumer durables, the industry is pushing towards innovation and managing escalating energy needs sustainably. With half of the buildings projected to exist in India by 2050 yet to be constructed, there is a significant opportunity to incorporate climate-friendly, energy-saving solutions into new developments. This alignment with sustainability goals highlights the critical role of consumer durables in addressing evolving consumer needs while mitigating climate impacts. One of the most effective solutions to combat climate change is to reduce greenhouse gas emissions and decrease reliance on fossil fuels. Harnessing renewable energy offers significant benefits in addressing climate change issues. India ranks 4th globally in renewable energy capacity, with notable achievements in wind and solar power. The country has set an ambitious target of 500 GW of non-fossil fuel-based energy by 2030, representing the world’s largest renewable energy expansion plan.India’s solar capacity has surged 30-fold in the past nine years, reaching 90.76 GW by September 2024, with a potential of 748 GW. The country’s renewable energy capacity has grown by 128% since FY2014, supported by initiatives such as the \"Development of Solar Parks\" scheme.\n\n_Exhibit 2.4: Solar Energy Capacity Forecast (FY)_\n\n_Source: Secondary Research_ **Policy Reforms** Various policy reforms have been incorporated by GoI to boost production, exports and promote ease of doing business. Some of these policies include:\n\n• **PLI Scheme:** Under PLI scheme, GoI aims at achieving large-scale electronics manufacturing in India. Investment of INR 1.23 lakh crore has been realized till March 2024. There is a PLI Scheme specifically for white goods manufacturers in India under which, GoI proposes a financial incentive to boost domestic manufacturing and attract large investments in white goods manufacturing value chain, with an objective of removing sectoral disabilities, creating economies of scale, enhancing exports, creating a robust component ecosystem and employment generation. This scheme is expected to be implemented\n\n**34\\.6 40.1**\n\n**54\\.0 65.0**\n\n**90\\.8**\n\n**113\\.3**\n\n**145\\.0**\n\n**185\\.6**\n\n2020 2021 2022 2023 2024 2025P 2026P 2027P","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Technopak_Industry_Report_p12.pdf"]},"response":{"errors":[],"extract_id":"extract_d40933760ef31aa3f1eb24cb23a92d44","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["12\n\nprovided under the Scheme. Under PMAY-U, 1.18 crore houses have been sanctioned while more than 85.5 lakh houses have already been constructed and delivered to the beneficiaries.\n\n**Urbanisation and Smart Cities** In the last decade, the urban population in India has grown at an annual rate of 2.4%. The urban population of India is projected to reach approximately 41% by CY2030 from 31% in CY2010. Further, the number of metropolitan cities in India are estimated to increase from 46 as per Census 2011 to 68 in CY2030. The Smart City Project was launched in June 2015 by GoI with a plan to build 100 smart cities across India,. As of 3 rd July 2024, the 100 Cities have completed 7,188 projects (90% of total projects) amounting to INR 1,44,237 crore as part of the mission. The balance 830 projects amounting to INR 19,926 crore are also in advanced stages of completion. On the financial progress, the mission has an allocated GoI budget of INR 48,000 crore for the 100 Cities. As on date, GOI has released INR 46,585 crore (~97% of the allocated GoI budget) to 100 cities.. **Climate Change and Built Environment** At the UN Climate Change Conference in 2021, India committed to achieving net-zero emissions by 2070, aiming to reduce projected carbon emissions by 1 billion tonnes and decrease the carbon intensity of its economy by 45% by 2030. Transforming the built environment, which contributes nearly 40% of global energy-related greenhouse gas emissions, is central to achieving these targets. With rising heat levels and pollution in India driving the demand for consumer durables, the industry is pushing towards innovation and managing escalating energy needs sustainably. With half of the buildings projected to exist in India by 2050 yet to be constructed, there is a significant opportunity to incorporate climate-friendly, energy-saving solutions into new developments. This alignment with sustainability goals highlights the critical role of consumer durables in addressing evolving consumer needs while mitigating climate impacts. One of the most effective solutions to combat climate change is to reduce greenhouse gas emissions and decrease reliance on fossil fuels. Harnessing renewable energy offers significant benefits in addressing climate change issues. India ranks 4th globally in renewable energy capacity, with notable achievements in wind and solar power. The country has set an ambitious target of 500 GW of non-fossil fuel-based energy by 2030, representing the world’s largest renewable energy expansion plan.India’s solar capacity has surged 30-fold in the past nine years, reaching 90.76 GW by September 2024, with a potential of 748 GW. The country’s renewable energy capacity has grown by 128% since FY2014, supported by initiatives such as the \"Development of Solar Parks\" scheme.\n\n_Exhibit 2.4: Solar Energy Capacity Forecast (FY)_\n\n_Source: Secondary Research_ **Policy Reforms** Various policy reforms have been incorporated by GoI to boost production, exports and promote ease of doing business. Some of these policies include:\n\n• **PLI Scheme:** Under PLI scheme, GoI aims at achieving large-scale electronics manufacturing in India. Investment of INR 1.23 lakh crore has been realized till March 2024. There is a PLI Scheme specifically for white goods manufacturers in India under which, GoI proposes a financial incentive to boost domestic manufacturing and attract large investments in white goods manufacturing value chain, with an objective of removing sectoral disabilities, creating economies of scale, enhancing exports, creating a robust component ecosystem and employment generation. This scheme is expected to be implemented\n\n**34\\.6 40.1**\n\n**54\\.0 65.0**\n\n**90\\.8**\n\n**113\\.3**\n\n**145\\.0**\n\n**185\\.6**\n\n2020 2021 2022 2023 2024 2025P 2026P 2027P"],"full_content":"12\n\nprovided under the Scheme. Under PMAY-U, 1.18 crore houses have been sanctioned while more than 85.5 lakh houses have already been constructed and delivered to the beneficiaries.\n\n**Urbanisation and Smart Cities** In the last decade, the urban population in India has grown at an annual rate of 2.4%. The urban population of India is projected to reach approximately 41% by CY2030 from 31% in CY2010. Further, the number of metropolitan cities in India are estimated to increase from 46 as per Census 2011 to 68 in CY2030. The Smart City Project was launched in June 2015 by GoI with a plan to build 100 smart cities across India,. As of 3 rd July 2024, the 100 Cities have completed 7,188 projects (90% of total projects) amounting to INR 1,44,237 crore as part of the mission. The balance 830 projects amounting to INR 19,926 crore are also in advanced stages of completion. On the financial progress, the mission has an allocated GoI budget of INR 48,000 crore for the 100 Cities. As on date, GOI has released INR 46,585 crore (~97% of the allocated GoI budget) to 100 cities.. **Climate Change and Built Environment** At the UN Climate Change Conference in 2021, India committed to achieving net-zero emissions by 2070, aiming to reduce projected carbon emissions by 1 billion tonnes and decrease the carbon intensity of its economy by 45% by 2030. Transforming the built environment, which contributes nearly 40% of global energy-related greenhouse gas emissions, is central to achieving these targets. With rising heat levels and pollution in India driving the demand for consumer durables, the industry is pushing towards innovation and managing escalating energy needs sustainably. With half of the buildings projected to exist in India by 2050 yet to be constructed, there is a significant opportunity to incorporate climate-friendly, energy-saving solutions into new developments. This alignment with sustainability goals highlights the critical role of consumer durables in addressing evolving consumer needs while mitigating climate impacts. One of the most effective solutions to combat climate change is to reduce greenhouse gas emissions and decrease reliance on fossil fuels. Harnessing renewable energy offers significant benefits in addressing climate change issues. India ranks 4th globally in renewable energy capacity, with notable achievements in wind and solar power. The country has set an ambitious target of 500 GW of non-fossil fuel-based energy by 2030, representing the world’s largest renewable energy expansion plan.India’s solar capacity has surged 30-fold in the past nine years, reaching 90.76 GW by September 2024, with a potential of 748 GW. The country’s renewable energy capacity has grown by 128% since FY2014, supported by initiatives such as the \"Development of Solar Parks\" scheme.\n\n_Exhibit 2.4: Solar Energy Capacity Forecast (FY)_\n\n_Source: Secondary Research_ **Policy Reforms** Various policy reforms have been incorporated by GoI to boost production, exports and promote ease of doing business. Some of these policies include:\n\n• **PLI Scheme:** Under PLI scheme, GoI aims at achieving large-scale electronics manufacturing in India. Investment of INR 1.23 lakh crore has been realized till March 2024. There is a PLI Scheme specifically for white goods manufacturers in India under which, GoI proposes a financial incentive to boost domestic manufacturing and attract large investments in white goods manufacturing value chain, with an objective of removing sectoral disabilities, creating economies of scale, enhancing exports, creating a robust component ecosystem and employment generation. This scheme is expected to be implemented\n\n**34\\.6 40.1**\n\n**54\\.0 65.0**\n\n**90\\.8**\n\n**113\\.3**\n\n**145\\.0**\n\n**185\\.6**\n\n2020 2021 2022 2023 2024 2025P 2026P 2027P","publish_date":null,"title":"12","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Technopak_Industry_Report_p12.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Technopak_Industry_Report_p12.pdf"},"request":{"example_id":"chart/Technopak_Industry_Report_p12","product_type":"parse","source_file_path":"Technopak_Industry_Report_p12.pdf"},"started_at":"2026-11-04T19:45:04.491698"} +{"completed_at":"2026-11-04T19:45:07.104306","latency_in_ms":1080,"output":{"example_id":"chart/Technopak_Industry_Report_p14","grounded_pages":[],"layout_pages":[],"markdown":"14\n\n_Exhibit 2.5: Number of Villages Electrified in India (CY)_\n\n_Source: Press Information Bureau (PIB)_ Electrification in India has increased over time with the support from schemes like DDUGJY, Saubhagya and IPDS, which has given the electrical market a major boost. It has opened new markets for companies in the white goods and FMEG industry, that can now sell their products into the rural parts of India as well. The expansion of rural electrification is expected to significantly increase the demand of electrical products, as new power lines and infrastructure need to be set up and maintained. **2\\.4 Potential Growth Areas Rural Markets** The rural market represents a significant opportunity for consumer durables due to its large demographic and emerging economic potential. Comprising approximately 65% of India's population, rural areas are poised to see substantial growth in both household income and consumption. Projections indicate that the average rural household income will reach INR 2.4 lakh by FY2025-26, reflecting an increase in purchasing power. Moreover, rural consumption is expected to grow at a CAGR of ~9.7% between CY2020 and CY2030, further underscoring the untapped potential of this market for companies in the consumer durables sector. As improving economic conditions enhance the financial capacity of rural consumers, businesses have a unique opportunity to cater to their evolving needs and preferences. **Export Markets** Indian consumer durables and appliances exports have grown significantly across key categories between CY2019 and CY2023. Air conditioners led with a ~14% CAGR, reaching USD 314.2 million in CY2023. Washing machines and small kitchen appliances followed with CAGRs of ~15% and ~18%, respectively, while refrigerators grew modestly to USD 231.7 million. Fans recorded an ~11% CAGR, and water purifiers reached USD 167.9 million, showcasing India’s growing footprint in the global consumer durables market.\n\nThe USA offers immense growth potential for Indian exports in consumer durables, particularly electronic goods, supported by a robust bilateral trade relationship. In FY2023, trade between the two nations reached a record USD 128.78 billion. India’s total exports to the USA were valued at USD 78.54 billion in CY2023. Exports of consumer durables and appliances to the USA were notable, with air conditioners accounting for ~14% of India’s total air conditioner exports, while washing machines and fans represented ~31% and ~11%, respectively, of total exports in their categories. The consistent increase in exports to the USA underscores its critical role as a lucrative market for India, offering significant opportunities for further expansion.\n\n**3,061 21,759 73,739**\n\n**2,49,799**\n\n**4,70,838 5,12,153**\n\n**5,97,464**\n\n1950 1961 1969 1980 1990 2002 2022","pages":[{"line_numbers":[],"markdown":"14\n\n_Exhibit 2.5: Number of Villages Electrified in India (CY)_\n\n_Source: Press Information Bureau (PIB)_ Electrification in India has increased over time with the support from schemes like DDUGJY, Saubhagya and IPDS, which has given the electrical market a major boost. It has opened new markets for companies in the white goods and FMEG industry, that can now sell their products into the rural parts of India as well. The expansion of rural electrification is expected to significantly increase the demand of electrical products, as new power lines and infrastructure need to be set up and maintained. **2\\.4 Potential Growth Areas Rural Markets** The rural market represents a significant opportunity for consumer durables due to its large demographic and emerging economic potential. Comprising approximately 65% of India's population, rural areas are poised to see substantial growth in both household income and consumption. Projections indicate that the average rural household income will reach INR 2.4 lakh by FY2025-26, reflecting an increase in purchasing power. Moreover, rural consumption is expected to grow at a CAGR of ~9.7% between CY2020 and CY2030, further underscoring the untapped potential of this market for companies in the consumer durables sector. As improving economic conditions enhance the financial capacity of rural consumers, businesses have a unique opportunity to cater to their evolving needs and preferences. **Export Markets** Indian consumer durables and appliances exports have grown significantly across key categories between CY2019 and CY2023. Air conditioners led with a ~14% CAGR, reaching USD 314.2 million in CY2023. Washing machines and small kitchen appliances followed with CAGRs of ~15% and ~18%, respectively, while refrigerators grew modestly to USD 231.7 million. Fans recorded an ~11% CAGR, and water purifiers reached USD 167.9 million, showcasing India’s growing footprint in the global consumer durables market.\n\nThe USA offers immense growth potential for Indian exports in consumer durables, particularly electronic goods, supported by a robust bilateral trade relationship. In FY2023, trade between the two nations reached a record USD 128.78 billion. India’s total exports to the USA were valued at USD 78.54 billion in CY2023. Exports of consumer durables and appliances to the USA were notable, with air conditioners accounting for ~14% of India’s total air conditioner exports, while washing machines and fans represented ~31% and ~11%, respectively, of total exports in their categories. The consistent increase in exports to the USA underscores its critical role as a lucrative market for India, offering significant opportunities for further expansion.\n\n**3,061 21,759 73,739**\n\n**2,49,799**\n\n**4,70,838 5,12,153**\n\n**5,97,464**\n\n1950 1961 1969 1980 1990 2002 2022","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"14\n\n_Exhibit 2.5: Number of Villages Electrified in India (CY)_\n\n_Source: Press Information Bureau (PIB)_ Electrification in India has increased over time with the support from schemes like DDUGJY, Saubhagya and IPDS, which has given the electrical market a major boost. It has opened new markets for companies in the white goods and FMEG industry, that can now sell their products into the rural parts of India as well. The expansion of rural electrification is expected to significantly increase the demand of electrical products, as new power lines and infrastructure need to be set up and maintained. **2\\.4 Potential Growth Areas Rural Markets** The rural market represents a significant opportunity for consumer durables due to its large demographic and emerging economic potential. Comprising approximately 65% of India's population, rural areas are poised to see substantial growth in both household income and consumption. Projections indicate that the average rural household income will reach INR 2.4 lakh by FY2025-26, reflecting an increase in purchasing power. Moreover, rural consumption is expected to grow at a CAGR of ~9.7% between CY2020 and CY2030, further underscoring the untapped potential of this market for companies in the consumer durables sector. As improving economic conditions enhance the financial capacity of rural consumers, businesses have a unique opportunity to cater to their evolving needs and preferences. **Export Markets** Indian consumer durables and appliances exports have grown significantly across key categories between CY2019 and CY2023. Air conditioners led with a ~14% CAGR, reaching USD 314.2 million in CY2023. Washing machines and small kitchen appliances followed with CAGRs of ~15% and ~18%, respectively, while refrigerators grew modestly to USD 231.7 million. Fans recorded an ~11% CAGR, and water purifiers reached USD 167.9 million, showcasing India’s growing footprint in the global consumer durables market.\n\nThe USA offers immense growth potential for Indian exports in consumer durables, particularly electronic goods, supported by a robust bilateral trade relationship. In FY2023, trade between the two nations reached a record USD 128.78 billion. India’s total exports to the USA were valued at USD 78.54 billion in CY2023. Exports of consumer durables and appliances to the USA were notable, with air conditioners accounting for ~14% of India’s total air conditioner exports, while washing machines and fans represented ~31% and ~11%, respectively, of total exports in their categories. The consistent increase in exports to the USA underscores its critical role as a lucrative market for India, offering significant opportunities for further expansion.\n\n**3,061 21,759 73,739**\n\n**2,49,799**\n\n**4,70,838 5,12,153**\n\n**5,97,464**\n\n1950 1961 1969 1980 1990 2002 2022","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Technopak_Industry_Report_p14.pdf"]},"response":{"errors":[],"extract_id":"extract_0b5b3e78d4f1ff10c033bd035e39f2d8","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["14\n\n_Exhibit 2.5: Number of Villages Electrified in India (CY)_\n\n_Source: Press Information Bureau (PIB)_ Electrification in India has increased over time with the support from schemes like DDUGJY, Saubhagya and IPDS, which has given the electrical market a major boost. It has opened new markets for companies in the white goods and FMEG industry, that can now sell their products into the rural parts of India as well. The expansion of rural electrification is expected to significantly increase the demand of electrical products, as new power lines and infrastructure need to be set up and maintained. **2\\.4 Potential Growth Areas Rural Markets** The rural market represents a significant opportunity for consumer durables due to its large demographic and emerging economic potential. Comprising approximately 65% of India's population, rural areas are poised to see substantial growth in both household income and consumption. Projections indicate that the average rural household income will reach INR 2.4 lakh by FY2025-26, reflecting an increase in purchasing power. Moreover, rural consumption is expected to grow at a CAGR of ~9.7% between CY2020 and CY2030, further underscoring the untapped potential of this market for companies in the consumer durables sector. As improving economic conditions enhance the financial capacity of rural consumers, businesses have a unique opportunity to cater to their evolving needs and preferences. **Export Markets** Indian consumer durables and appliances exports have grown significantly across key categories between CY2019 and CY2023. Air conditioners led with a ~14% CAGR, reaching USD 314.2 million in CY2023. Washing machines and small kitchen appliances followed with CAGRs of ~15% and ~18%, respectively, while refrigerators grew modestly to USD 231.7 million. Fans recorded an ~11% CAGR, and water purifiers reached USD 167.9 million, showcasing India’s growing footprint in the global consumer durables market.\n\nThe USA offers immense growth potential for Indian exports in consumer durables, particularly electronic goods, supported by a robust bilateral trade relationship. In FY2023, trade between the two nations reached a record USD 128.78 billion. India’s total exports to the USA were valued at USD 78.54 billion in CY2023. Exports of consumer durables and appliances to the USA were notable, with air conditioners accounting for ~14% of India’s total air conditioner exports, while washing machines and fans represented ~31% and ~11%, respectively, of total exports in their categories. The consistent increase in exports to the USA underscores its critical role as a lucrative market for India, offering significant opportunities for further expansion.\n\n**3,061 21,759 73,739**\n\n**2,49,799**\n\n**4,70,838 5,12,153**\n\n**5,97,464**\n\n1950 1961 1969 1980 1990 2002 2022"],"full_content":"14\n\n_Exhibit 2.5: Number of Villages Electrified in India (CY)_\n\n_Source: Press Information Bureau (PIB)_ Electrification in India has increased over time with the support from schemes like DDUGJY, Saubhagya and IPDS, which has given the electrical market a major boost. It has opened new markets for companies in the white goods and FMEG industry, that can now sell their products into the rural parts of India as well. The expansion of rural electrification is expected to significantly increase the demand of electrical products, as new power lines and infrastructure need to be set up and maintained. **2\\.4 Potential Growth Areas Rural Markets** The rural market represents a significant opportunity for consumer durables due to its large demographic and emerging economic potential. Comprising approximately 65% of India's population, rural areas are poised to see substantial growth in both household income and consumption. Projections indicate that the average rural household income will reach INR 2.4 lakh by FY2025-26, reflecting an increase in purchasing power. Moreover, rural consumption is expected to grow at a CAGR of ~9.7% between CY2020 and CY2030, further underscoring the untapped potential of this market for companies in the consumer durables sector. As improving economic conditions enhance the financial capacity of rural consumers, businesses have a unique opportunity to cater to their evolving needs and preferences. **Export Markets** Indian consumer durables and appliances exports have grown significantly across key categories between CY2019 and CY2023. Air conditioners led with a ~14% CAGR, reaching USD 314.2 million in CY2023. Washing machines and small kitchen appliances followed with CAGRs of ~15% and ~18%, respectively, while refrigerators grew modestly to USD 231.7 million. Fans recorded an ~11% CAGR, and water purifiers reached USD 167.9 million, showcasing India’s growing footprint in the global consumer durables market.\n\nThe USA offers immense growth potential for Indian exports in consumer durables, particularly electronic goods, supported by a robust bilateral trade relationship. In FY2023, trade between the two nations reached a record USD 128.78 billion. India’s total exports to the USA were valued at USD 78.54 billion in CY2023. Exports of consumer durables and appliances to the USA were notable, with air conditioners accounting for ~14% of India’s total air conditioner exports, while washing machines and fans represented ~31% and ~11%, respectively, of total exports in their categories. The consistent increase in exports to the USA underscores its critical role as a lucrative market for India, offering significant opportunities for further expansion.\n\n**3,061 21,759 73,739**\n\n**2,49,799**\n\n**4,70,838 5,12,153**\n\n**5,97,464**\n\n1950 1961 1969 1980 1990 2002 2022","publish_date":null,"title":"2.4 Potential Growth Areas 14","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Technopak_Industry_Report_p14.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Technopak_Industry_Report_p14.pdf"},"request":{"example_id":"chart/Technopak_Industry_Report_p14","product_type":"parse","source_file_path":"Technopak_Industry_Report_p14.pdf"},"started_at":"2026-11-04T19:45:06.023965"} +{"completed_at":"2026-11-04T19:45:09.193491","latency_in_ms":1899,"output":{"example_id":"chart/Technopak_Industry_Report_p17","grounded_pages":[],"layout_pages":[],"markdown":"17\n\n**3\\.3 Water Purifier Market Size** The Indian water purifier market has exhibited continuous growth over the years. It has grown at a CAGR of ~4.9% from INR 6,960 crore in FY2019 to INR 8,860 crore\\* in FY2024 and is further expected to grow at a CAGR of ~10.1% till FY2029 to reach a market value of INR 14,350 crore. _\\*All the market value figures in this report are at the consumer price level, unless otherwise specified._ The high growth rate of the water purifier market can be attributed to various factors, including increasing population, rising disposable income, growing water contamination, a surge in cases of water-borne diseases, increasing health consciousness among individuals, and the growth in electrification. As awareness of water contamination and the benefits of purified water increases, it is expected that more households will invest in water purifiers, driving market growth. In particular, Tier-2 cities in India are expected to experience higher growth in the near future. _Exhibit 3.3: Indian Water Purifier Market- By Value (in INR crore) (FY)_\n\n_Source: Technopak Analysis Note: This does not include exports. Market size at consumer price level is the addition of mark up of industry level channel margins to the market size at wholesale level, which_ _is on the manufacturer’s realization._ **3\\.3.1 Water Purifier (Product) Market in India** The Indian water purifier product market has exhibited continuous growth over the years. It has grown at a CAGR of ~5% from INR 3,800 crore in FY2019 to INR 4,860 crore in FY2024 and is further expected to grow at a CAGR of ~10% till FY2029 to reach a market value of INR 7,830 crore. _Exhibit 3.4: Indian Water Purifier Product Market- By Value (in INR crore) (FY)_\n\n_Source: Technopak Analysis_\n\n**3,800 4,860**\n\n**7,830**\n\n**3,160**\n\n**4,000**\n\n**7,110**\n\n2019 2024 2029 P\n\nService\n\nProduct\n\n**3,800 4,860**\n\n**7,830**\n\n2019 2024 2029 P\n\n**6,960**\n\n**8,860**\n\n**14,350**","pages":[{"line_numbers":[],"markdown":"17\n\n**3\\.3 Water Purifier Market Size** The Indian water purifier market has exhibited continuous growth over the years. It has grown at a CAGR of ~4.9% from INR 6,960 crore in FY2019 to INR 8,860 crore\\* in FY2024 and is further expected to grow at a CAGR of ~10.1% till FY2029 to reach a market value of INR 14,350 crore. _\\*All the market value figures in this report are at the consumer price level, unless otherwise specified._ The high growth rate of the water purifier market can be attributed to various factors, including increasing population, rising disposable income, growing water contamination, a surge in cases of water-borne diseases, increasing health consciousness among individuals, and the growth in electrification. As awareness of water contamination and the benefits of purified water increases, it is expected that more households will invest in water purifiers, driving market growth. In particular, Tier-2 cities in India are expected to experience higher growth in the near future. _Exhibit 3.3: Indian Water Purifier Market- By Value (in INR crore) (FY)_\n\n_Source: Technopak Analysis Note: This does not include exports. Market size at consumer price level is the addition of mark up of industry level channel margins to the market size at wholesale level, which_ _is on the manufacturer’s realization._ **3\\.3.1 Water Purifier (Product) Market in India** The Indian water purifier product market has exhibited continuous growth over the years. It has grown at a CAGR of ~5% from INR 3,800 crore in FY2019 to INR 4,860 crore in FY2024 and is further expected to grow at a CAGR of ~10% till FY2029 to reach a market value of INR 7,830 crore. _Exhibit 3.4: Indian Water Purifier Product Market- By Value (in INR crore) (FY)_\n\n_Source: Technopak Analysis_\n\n**3,800 4,860**\n\n**7,830**\n\n**3,160**\n\n**4,000**\n\n**7,110**\n\n2019 2024 2029 P\n\nService\n\nProduct\n\n**3,800 4,860**\n\n**7,830**\n\n2019 2024 2029 P\n\n**6,960**\n\n**8,860**\n\n**14,350**","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"17\n\n**3\\.3 Water Purifier Market Size** The Indian water purifier market has exhibited continuous growth over the years. It has grown at a CAGR of ~4.9% from INR 6,960 crore in FY2019 to INR 8,860 crore\\* in FY2024 and is further expected to grow at a CAGR of ~10.1% till FY2029 to reach a market value of INR 14,350 crore. _\\*All the market value figures in this report are at the consumer price level, unless otherwise specified._ The high growth rate of the water purifier market can be attributed to various factors, including increasing population, rising disposable income, growing water contamination, a surge in cases of water-borne diseases, increasing health consciousness among individuals, and the growth in electrification. As awareness of water contamination and the benefits of purified water increases, it is expected that more households will invest in water purifiers, driving market growth. In particular, Tier-2 cities in India are expected to experience higher growth in the near future. _Exhibit 3.3: Indian Water Purifier Market- By Value (in INR crore) (FY)_\n\n_Source: Technopak Analysis Note: This does not include exports. Market size at consumer price level is the addition of mark up of industry level channel margins to the market size at wholesale level, which_ _is on the manufacturer’s realization._ **3\\.3.1 Water Purifier (Product) Market in India** The Indian water purifier product market has exhibited continuous growth over the years. It has grown at a CAGR of ~5% from INR 3,800 crore in FY2019 to INR 4,860 crore in FY2024 and is further expected to grow at a CAGR of ~10% till FY2029 to reach a market value of INR 7,830 crore. _Exhibit 3.4: Indian Water Purifier Product Market- By Value (in INR crore) (FY)_\n\n_Source: Technopak Analysis_\n\n**3,800 4,860**\n\n**7,830**\n\n**3,160**\n\n**4,000**\n\n**7,110**\n\n2019 2024 2029 P\n\nService\n\nProduct\n\n**3,800 4,860**\n\n**7,830**\n\n2019 2024 2029 P\n\n**6,960**\n\n**8,860**\n\n**14,350**","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Technopak_Industry_Report_p17.pdf"]},"response":{"errors":[],"extract_id":"extract_3e5ed3f1771a37a411d8f8e6a763d091","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["17\n\n**3\\.3 Water Purifier Market Size** The Indian water purifier market has exhibited continuous growth over the years. It has grown at a CAGR of ~4.9% from INR 6,960 crore in FY2019 to INR 8,860 crore\\* in FY2024 and is further expected to grow at a CAGR of ~10.1% till FY2029 to reach a market value of INR 14,350 crore. _\\*All the market value figures in this report are at the consumer price level, unless otherwise specified._ The high growth rate of the water purifier market can be attributed to various factors, including increasing population, rising disposable income, growing water contamination, a surge in cases of water-borne diseases, increasing health consciousness among individuals, and the growth in electrification. As awareness of water contamination and the benefits of purified water increases, it is expected that more households will invest in water purifiers, driving market growth. In particular, Tier-2 cities in India are expected to experience higher growth in the near future. _Exhibit 3.3: Indian Water Purifier Market- By Value (in INR crore) (FY)_\n\n_Source: Technopak Analysis Note: This does not include exports. Market size at consumer price level is the addition of mark up of industry level channel margins to the market size at wholesale level, which_ _is on the manufacturer’s realization._ **3\\.3.1 Water Purifier (Product) Market in India** The Indian water purifier product market has exhibited continuous growth over the years. It has grown at a CAGR of ~5% from INR 3,800 crore in FY2019 to INR 4,860 crore in FY2024 and is further expected to grow at a CAGR of ~10% till FY2029 to reach a market value of INR 7,830 crore. _Exhibit 3.4: Indian Water Purifier Product Market- By Value (in INR crore) (FY)_\n\n_Source: Technopak Analysis_\n\n**3,800 4,860**\n\n**7,830**\n\n**3,160**\n\n**4,000**\n\n**7,110**\n\n2019 2024 2029 P\n\nService\n\nProduct\n\n**3,800 4,860**\n\n**7,830**\n\n2019 2024 2029 P\n\n**6,960**\n\n**8,860**\n\n**14,350**"],"full_content":"17\n\n**3\\.3 Water Purifier Market Size** The Indian water purifier market has exhibited continuous growth over the years. It has grown at a CAGR of ~4.9% from INR 6,960 crore in FY2019 to INR 8,860 crore\\* in FY2024 and is further expected to grow at a CAGR of ~10.1% till FY2029 to reach a market value of INR 14,350 crore. _\\*All the market value figures in this report are at the consumer price level, unless otherwise specified._ The high growth rate of the water purifier market can be attributed to various factors, including increasing population, rising disposable income, growing water contamination, a surge in cases of water-borne diseases, increasing health consciousness among individuals, and the growth in electrification. As awareness of water contamination and the benefits of purified water increases, it is expected that more households will invest in water purifiers, driving market growth. In particular, Tier-2 cities in India are expected to experience higher growth in the near future. _Exhibit 3.3: Indian Water Purifier Market- By Value (in INR crore) (FY)_\n\n_Source: Technopak Analysis Note: This does not include exports. Market size at consumer price level is the addition of mark up of industry level channel margins to the market size at wholesale level, which_ _is on the manufacturer’s realization._ **3\\.3.1 Water Purifier (Product) Market in India** The Indian water purifier product market has exhibited continuous growth over the years. It has grown at a CAGR of ~5% from INR 3,800 crore in FY2019 to INR 4,860 crore in FY2024 and is further expected to grow at a CAGR of ~10% till FY2029 to reach a market value of INR 7,830 crore. _Exhibit 3.4: Indian Water Purifier Product Market- By Value (in INR crore) (FY)_\n\n_Source: Technopak Analysis_\n\n**3,800 4,860**\n\n**7,830**\n\n**3,160**\n\n**4,000**\n\n**7,110**\n\n2019 2024 2029 P\n\nService\n\nProduct\n\n**3,800 4,860**\n\n**7,830**\n\n2019 2024 2029 P\n\n**6,960**\n\n**8,860**\n\n**14,350**","publish_date":null,"title":"3.3 Water Purifier Market Size 3.3.1 Water Purifier (Product) Market in India 17","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Technopak_Industry_Report_p17.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Technopak_Industry_Report_p17.pdf"},"request":{"example_id":"chart/Technopak_Industry_Report_p17","product_type":"parse","source_file_path":"Technopak_Industry_Report_p17.pdf"},"started_at":"2026-11-04T19:45:07.293975"} +{"completed_at":"2026-11-04T19:45:09.793861","latency_in_ms":1238,"output":{"example_id":"chart/Technopak_Industry_Report_p18","grounded_pages":[],"layout_pages":[],"markdown":"18\n\nThe water purifier product market in India has steadily transitioned towards branded play, with branded players commanding a substantial 83% market share as of FY2024. Within this branded market, the top 5-7 leading players hold close to 90% market share. The shift towards branded play over unbranded play is influenced by factors on both the demand and supply sides. On the demand side, increasing health awareness among consumers and increase in availability and access to piped water are key drivers of this shift. On the supply side, technological interventions, brand-building initiatives by major players, the implementation of GST, and robust distribution network servicing extensive retail footprints contribute to the transformation towards branded play. As of FY2024, branded play accounted for nearly 83% (~INR 4,035 crore) of the water purifier product market in India. This marks a significant increase from the market share of around 78% (~ INR 2,960 crore) recorded in FY2019, showcasing a notable growth trajectory for the branded market. The branded play is estimated to capture ~88% (~INR 6,890 crore) market share by FY2029. _Exhibit 3.5: Share of Branded Play in Indian Water Purifier (Product) Market- By Value (in %) (FY)_\n\n_Source: Technopak Analysis_ **Key Sub-categories of Water Purifiers** **By Need for Electricity** The Indian water purifier market can be segmented based on the need for electricity into Electric and Non-Electric categories. Electric water purifiers require electricity to operate and utilize RO, UV, or a combination of both RO+UV technology to provide pure and clean drinking water. RO water purifiers reduce Total Dissolved Solids (TDS) and enhance taste, making them suitable for areas with medium to high TDS levels in the fresh water supply. On the other hand, UV water purifiers remove sediments, microbials, and improve the water odor, making them applicable in areas with a low level of TDS in the fresh water supply. Filters of electric water purifiers typically need replacement once a year. Non-electric water purifiers operate without electricity and employ gravity-based mechanisms to purify water. They use a variety of filters like sediment filters, carbon filters etc., UF (Ultra Filtration) and/or chemical technology to remove impurities and bacteria, ensuring water purification. Non-electric water purifiers are particularly preferred in situations where inline water supply and electricity are not available. While these water purifiers do not enhance the taste and color of the water, they are suitable for use when the TDS level is low, i.e., within consumable limits. Filters of non-electric water purifiers typically require replacement every 3-6 months. Electric water purifiers exhibit superior water purification efficiency compared to non-electric water purifiers. As of FY2024, electric and non-electric water purifiers accounted for ~97% (INR 4,715 crore) and ~3% (INR 145 crore) respectively, of the overall Indian water purifier product market by value. It is projected that by FY2029, electric and non-electric water purifiers will constitute ~99% (INR 7,750 crore) and ~1% (INR 80 crore) respectively, of the overall Indian water purifier product market by value. In terms of volume, the market size of electric and non-electric water purifiers was ~3.6 and ~0.6 million units as of FY2024. By FY2029, the market size of electric and non-electric water purifiers is projected to be ~5.1 and ~0.2 million units, respectively.\n\n**78%**\n\n**22%**\n\n**2019**\n\n**83%**\n\n**17%**\n\n**2024**\n\nBranded Play Unbranded Play\n\n**88%**\n\n**12%**\n\n**2029P**","pages":[{"line_numbers":[],"markdown":"18\n\nThe water purifier product market in India has steadily transitioned towards branded play, with branded players commanding a substantial 83% market share as of FY2024. Within this branded market, the top 5-7 leading players hold close to 90% market share. The shift towards branded play over unbranded play is influenced by factors on both the demand and supply sides. On the demand side, increasing health awareness among consumers and increase in availability and access to piped water are key drivers of this shift. On the supply side, technological interventions, brand-building initiatives by major players, the implementation of GST, and robust distribution network servicing extensive retail footprints contribute to the transformation towards branded play. As of FY2024, branded play accounted for nearly 83% (~INR 4,035 crore) of the water purifier product market in India. This marks a significant increase from the market share of around 78% (~ INR 2,960 crore) recorded in FY2019, showcasing a notable growth trajectory for the branded market. The branded play is estimated to capture ~88% (~INR 6,890 crore) market share by FY2029. _Exhibit 3.5: Share of Branded Play in Indian Water Purifier (Product) Market- By Value (in %) (FY)_\n\n_Source: Technopak Analysis_ **Key Sub-categories of Water Purifiers** **By Need for Electricity** The Indian water purifier market can be segmented based on the need for electricity into Electric and Non-Electric categories. Electric water purifiers require electricity to operate and utilize RO, UV, or a combination of both RO+UV technology to provide pure and clean drinking water. RO water purifiers reduce Total Dissolved Solids (TDS) and enhance taste, making them suitable for areas with medium to high TDS levels in the fresh water supply. On the other hand, UV water purifiers remove sediments, microbials, and improve the water odor, making them applicable in areas with a low level of TDS in the fresh water supply. Filters of electric water purifiers typically need replacement once a year. Non-electric water purifiers operate without electricity and employ gravity-based mechanisms to purify water. They use a variety of filters like sediment filters, carbon filters etc., UF (Ultra Filtration) and/or chemical technology to remove impurities and bacteria, ensuring water purification. Non-electric water purifiers are particularly preferred in situations where inline water supply and electricity are not available. While these water purifiers do not enhance the taste and color of the water, they are suitable for use when the TDS level is low, i.e., within consumable limits. Filters of non-electric water purifiers typically require replacement every 3-6 months. Electric water purifiers exhibit superior water purification efficiency compared to non-electric water purifiers. As of FY2024, electric and non-electric water purifiers accounted for ~97% (INR 4,715 crore) and ~3% (INR 145 crore) respectively, of the overall Indian water purifier product market by value. It is projected that by FY2029, electric and non-electric water purifiers will constitute ~99% (INR 7,750 crore) and ~1% (INR 80 crore) respectively, of the overall Indian water purifier product market by value. In terms of volume, the market size of electric and non-electric water purifiers was ~3.6 and ~0.6 million units as of FY2024. By FY2029, the market size of electric and non-electric water purifiers is projected to be ~5.1 and ~0.2 million units, respectively.\n\n**78%**\n\n**22%**\n\n**2019**\n\n**83%**\n\n**17%**\n\n**2024**\n\nBranded Play Unbranded Play\n\n**88%**\n\n**12%**\n\n**2029P**","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"18\n\nThe water purifier product market in India has steadily transitioned towards branded play, with branded players commanding a substantial 83% market share as of FY2024. Within this branded market, the top 5-7 leading players hold close to 90% market share. The shift towards branded play over unbranded play is influenced by factors on both the demand and supply sides. On the demand side, increasing health awareness among consumers and increase in availability and access to piped water are key drivers of this shift. On the supply side, technological interventions, brand-building initiatives by major players, the implementation of GST, and robust distribution network servicing extensive retail footprints contribute to the transformation towards branded play. As of FY2024, branded play accounted for nearly 83% (~INR 4,035 crore) of the water purifier product market in India. This marks a significant increase from the market share of around 78% (~ INR 2,960 crore) recorded in FY2019, showcasing a notable growth trajectory for the branded market. The branded play is estimated to capture ~88% (~INR 6,890 crore) market share by FY2029. _Exhibit 3.5: Share of Branded Play in Indian Water Purifier (Product) Market- By Value (in %) (FY)_\n\n_Source: Technopak Analysis_ **Key Sub-categories of Water Purifiers** **By Need for Electricity** The Indian water purifier market can be segmented based on the need for electricity into Electric and Non-Electric categories. Electric water purifiers require electricity to operate and utilize RO, UV, or a combination of both RO+UV technology to provide pure and clean drinking water. RO water purifiers reduce Total Dissolved Solids (TDS) and enhance taste, making them suitable for areas with medium to high TDS levels in the fresh water supply. On the other hand, UV water purifiers remove sediments, microbials, and improve the water odor, making them applicable in areas with a low level of TDS in the fresh water supply. Filters of electric water purifiers typically need replacement once a year. Non-electric water purifiers operate without electricity and employ gravity-based mechanisms to purify water. They use a variety of filters like sediment filters, carbon filters etc., UF (Ultra Filtration) and/or chemical technology to remove impurities and bacteria, ensuring water purification. Non-electric water purifiers are particularly preferred in situations where inline water supply and electricity are not available. While these water purifiers do not enhance the taste and color of the water, they are suitable for use when the TDS level is low, i.e., within consumable limits. Filters of non-electric water purifiers typically require replacement every 3-6 months. Electric water purifiers exhibit superior water purification efficiency compared to non-electric water purifiers. As of FY2024, electric and non-electric water purifiers accounted for ~97% (INR 4,715 crore) and ~3% (INR 145 crore) respectively, of the overall Indian water purifier product market by value. It is projected that by FY2029, electric and non-electric water purifiers will constitute ~99% (INR 7,750 crore) and ~1% (INR 80 crore) respectively, of the overall Indian water purifier product market by value. In terms of volume, the market size of electric and non-electric water purifiers was ~3.6 and ~0.6 million units as of FY2024. By FY2029, the market size of electric and non-electric water purifiers is projected to be ~5.1 and ~0.2 million units, respectively.\n\n**78%**\n\n**22%**\n\n**2019**\n\n**83%**\n\n**17%**\n\n**2024**\n\nBranded Play Unbranded Play\n\n**88%**\n\n**12%**\n\n**2029P**","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Technopak_Industry_Report_p18.pdf"]},"response":{"errors":[],"extract_id":"extract_7b3a51ac790702554e74d9a02c2d8304","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["18\n\nThe water purifier product market in India has steadily transitioned towards branded play, with branded players commanding a substantial 83% market share as of FY2024. Within this branded market, the top 5-7 leading players hold close to 90% market share. The shift towards branded play over unbranded play is influenced by factors on both the demand and supply sides. On the demand side, increasing health awareness among consumers and increase in availability and access to piped water are key drivers of this shift. On the supply side, technological interventions, brand-building initiatives by major players, the implementation of GST, and robust distribution network servicing extensive retail footprints contribute to the transformation towards branded play. As of FY2024, branded play accounted for nearly 83% (~INR 4,035 crore) of the water purifier product market in India. This marks a significant increase from the market share of around 78% (~ INR 2,960 crore) recorded in FY2019, showcasing a notable growth trajectory for the branded market. The branded play is estimated to capture ~88% (~INR 6,890 crore) market share by FY2029. _Exhibit 3.5: Share of Branded Play in Indian Water Purifier (Product) Market- By Value (in %) (FY)_\n\n_Source: Technopak Analysis_ **Key Sub-categories of Water Purifiers** **By Need for Electricity** The Indian water purifier market can be segmented based on the need for electricity into Electric and Non-Electric categories. Electric water purifiers require electricity to operate and utilize RO, UV, or a combination of both RO+UV technology to provide pure and clean drinking water. RO water purifiers reduce Total Dissolved Solids (TDS) and enhance taste, making them suitable for areas with medium to high TDS levels in the fresh water supply. On the other hand, UV water purifiers remove sediments, microbials, and improve the water odor, making them applicable in areas with a low level of TDS in the fresh water supply. Filters of electric water purifiers typically need replacement once a year. Non-electric water purifiers operate without electricity and employ gravity-based mechanisms to purify water. They use a variety of filters like sediment filters, carbon filters etc., UF (Ultra Filtration) and/or chemical technology to remove impurities and bacteria, ensuring water purification. Non-electric water purifiers are particularly preferred in situations where inline water supply and electricity are not available. While these water purifiers do not enhance the taste and color of the water, they are suitable for use when the TDS level is low, i.e., within consumable limits. Filters of non-electric water purifiers typically require replacement every 3-6 months. Electric water purifiers exhibit superior water purification efficiency compared to non-electric water purifiers. As of FY2024, electric and non-electric water purifiers accounted for ~97% (INR 4,715 crore) and ~3% (INR 145 crore) respectively, of the overall Indian water purifier product market by value. It is projected that by FY2029, electric and non-electric water purifiers will constitute ~99% (INR 7,750 crore) and ~1% (INR 80 crore) respectively, of the overall Indian water purifier product market by value. In terms of volume, the market size of electric and non-electric water purifiers was ~3.6 and ~0.6 million units as of FY2024. By FY2029, the market size of electric and non-electric water purifiers is projected to be ~5.1 and ~0.2 million units, respectively.\n\n**78%**\n\n**22%**\n\n**2019**\n\n**83%**\n\n**17%**\n\n**2024**\n\nBranded Play Unbranded Play\n\n**88%**\n\n**12%**\n\n**2029P**"],"full_content":"18\n\nThe water purifier product market in India has steadily transitioned towards branded play, with branded players commanding a substantial 83% market share as of FY2024. Within this branded market, the top 5-7 leading players hold close to 90% market share. The shift towards branded play over unbranded play is influenced by factors on both the demand and supply sides. On the demand side, increasing health awareness among consumers and increase in availability and access to piped water are key drivers of this shift. On the supply side, technological interventions, brand-building initiatives by major players, the implementation of GST, and robust distribution network servicing extensive retail footprints contribute to the transformation towards branded play. As of FY2024, branded play accounted for nearly 83% (~INR 4,035 crore) of the water purifier product market in India. This marks a significant increase from the market share of around 78% (~ INR 2,960 crore) recorded in FY2019, showcasing a notable growth trajectory for the branded market. The branded play is estimated to capture ~88% (~INR 6,890 crore) market share by FY2029. _Exhibit 3.5: Share of Branded Play in Indian Water Purifier (Product) Market- By Value (in %) (FY)_\n\n_Source: Technopak Analysis_ **Key Sub-categories of Water Purifiers** **By Need for Electricity** The Indian water purifier market can be segmented based on the need for electricity into Electric and Non-Electric categories. Electric water purifiers require electricity to operate and utilize RO, UV, or a combination of both RO+UV technology to provide pure and clean drinking water. RO water purifiers reduce Total Dissolved Solids (TDS) and enhance taste, making them suitable for areas with medium to high TDS levels in the fresh water supply. On the other hand, UV water purifiers remove sediments, microbials, and improve the water odor, making them applicable in areas with a low level of TDS in the fresh water supply. Filters of electric water purifiers typically need replacement once a year. Non-electric water purifiers operate without electricity and employ gravity-based mechanisms to purify water. They use a variety of filters like sediment filters, carbon filters etc., UF (Ultra Filtration) and/or chemical technology to remove impurities and bacteria, ensuring water purification. Non-electric water purifiers are particularly preferred in situations where inline water supply and electricity are not available. While these water purifiers do not enhance the taste and color of the water, they are suitable for use when the TDS level is low, i.e., within consumable limits. Filters of non-electric water purifiers typically require replacement every 3-6 months. Electric water purifiers exhibit superior water purification efficiency compared to non-electric water purifiers. As of FY2024, electric and non-electric water purifiers accounted for ~97% (INR 4,715 crore) and ~3% (INR 145 crore) respectively, of the overall Indian water purifier product market by value. It is projected that by FY2029, electric and non-electric water purifiers will constitute ~99% (INR 7,750 crore) and ~1% (INR 80 crore) respectively, of the overall Indian water purifier product market by value. In terms of volume, the market size of electric and non-electric water purifiers was ~3.6 and ~0.6 million units as of FY2024. By FY2029, the market size of electric and non-electric water purifiers is projected to be ~5.1 and ~0.2 million units, respectively.\n\n**78%**\n\n**22%**\n\n**2019**\n\n**83%**\n\n**17%**\n\n**2024**\n\nBranded Play Unbranded Play\n\n**88%**\n\n**12%**\n\n**2029P**","publish_date":null,"title":"Key Sub-categories of Water Purifiers 18","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Technopak_Industry_Report_p18.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Technopak_Industry_Report_p18.pdf"},"request":{"example_id":"chart/Technopak_Industry_Report_p18","product_type":"parse","source_file_path":"Technopak_Industry_Report_p18.pdf"},"started_at":"2026-11-04T19:45:08.555391"} +{"completed_at":"2026-11-04T19:45:13.451058","latency_in_ms":3655,"output":{"example_id":"chart/Technopak_Industry_Report_p4","grounded_pages":[],"layout_pages":[],"markdown":"4\n\n**Inflation** Inflation, measured by the consumer price index (CPI), reflects changes in the prices of commonly purchased goods and services. Global inflation spiked to ~8.7% in CY2022. The increase in global crude oil and commodity prices, along with COVID-19 supply chain disruptions, have driven inflation, prompting countries to raise interest rates in response. Global inflation has eased to ~6.8% in CY2023, with a projected rate of ~5.9% for CY2024. Pre-pandemic inflation was ~3.5% in CY2019. India’s CPI inflation is expected to drop from ~5.4% in CY2023 to ~4.6% in CY2024, and ~4.2% in CY2025. In the USA and Germany, inflation is projected to fall from ~4.1% and ~6.0% in CY2023 to ~2.9% and ~2.4% in CY2024, reaching ~2.0% in CY2025. China’s inflation may rise from ~0.2% in CY2023 to ~2.0% in CY2025. _Exhibit 1.3: Global Inflation Rate, Average Consumer Price Index (%) of Key Economies (CY)_\n\n_Source: IMF projections, Technopak Analysis Note: For India, CY2017 data refers to FY2018 and so on_ **1\\.2 Overview of Indian Economy** **India’s GDP and GDP Growth (Real and Nominal)** **\\- Historical, Current & Projected Trajectory** India is ranked fifth in the world in terms of nominal gross domestic product (\"GDP\") for FY2024 and is the third- largest economy in the world in terms of purchasing power parity (\"PPP\"). India is projected to be a USD 6.5 trillion economy by FY2029 and is anticipated to become the third largest economy, surpassing Germany, and Japan in terms of Nominal GDP. _Exhibit.1.4: India’s GDP at Current Prices (Nominal GDP) (in USD trillion) and GDP Growth Rate (%) (FY)_\n\n_Source: RBI, Technopak Analysis Note: 1 USD = INR 80_\n\n**1\\.1**\n\n**1\\.2**\n\n**1\\.4**\n\n**1\\.6**\n\n**1\\.7**\n\n**1\\.9**\n\n**2\\.1**\n\n**2\\.4**\n\n**2\\.5**\n\n**2\\.5**\n\n**2\\.9**\n\n**3\\.4**\n\n**3\\.7**\n\n**5\\.8**\n\n**6\\.5**\n\n**13\\.8% 12.9% 11.4%**\n\n**10\\.3%**\n\n**11\\.6%** **11\\.5%** **10\\.3%**\n\n**6\\.4%**\n\n**\\-1.2%**\n\n**16\\.9% 16.9%**\n\n**8\\.8%**\n\n**12\\.0% 12.1%**\n\n\\-5.0%\n\n0\\.0%\n\n5\\.0%\n\n10\\.0%\n\n15\\.0%\n\n20\\.0%\n\n0\\.0\n\n1\\.0\n\n2\\.0\n\n3\\.0\n\n4\\.0\n\n5\\.0\n\n6\\.0\n\n7\\.0\n\n2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024E 2028P 2029P\n\nGDP (USD Tn) GDP Growth Rate(%)\n\n**Country 2017 2018 2019 2020 2021 2022 2023 2024E 2025P 2026P 2027P 2028P**\n\nUSA 2.1% 2.4% 1.8% 1.2% 4.7% 8.0% 4.1% 2.9% 2.0% 2.1% 2.1% 2.1%\n\nChina 1.6% 2.1% 2.9% 2.5% 0.9% 2.0% 0.2% 1.0% 2.0% 2.0% 2.0% 2.0%\n\nJapan 0.5% 1.0% 0.5% 0.0% (0.2%) 2.5% 3.3% 2.2% 2.1% 2.0% 2.0% 2.0%\n\nGermany 1.7% 1.9% 1.4% 0.4% 3.2% 8.7% 6.0% 2.4% 2.0% 2.0% 2.0% 2.0%\n\nIndia 3.6% 3.4% 4.8% 6.2% 5.5% 6.7% 5.4% 4.6% 4.2% 4.1% 4.0% 4.0%\n\nUK 2.7% 2.5% 1.8% 0.9% 2.6% 9.1% 7.3% 2.5% 2.0% 2.0% 2.0% 2.0%\n\nFrance 1.2% 2.1% 1.3% 0.5% 2.1% 5.9% 5.7% 2.4% 1.8% 1.8% 1.8% 1.8%\n\nBrazil 3.4% 3.7% 3.7% 3.2% 8.3% 9.3% 4.6% 4.1% 3.0% 3.1% 3.0% 3.0%\n\nAustralia 2.0% 1.9% 1.6% 0.9% 2.8% 6.6% 5.6% 3.5% 3.0% 2.7% 2.7% 2.6%\n\n**World 3.3% 3.6% 3.5% 3.2% 4.7% 8.7% 6.8% 5.9% 4.5% 3.7% 3.5% 3.4%**","pages":[{"line_numbers":[],"markdown":"4\n\n**Inflation** Inflation, measured by the consumer price index (CPI), reflects changes in the prices of commonly purchased goods and services. Global inflation spiked to ~8.7% in CY2022. The increase in global crude oil and commodity prices, along with COVID-19 supply chain disruptions, have driven inflation, prompting countries to raise interest rates in response. Global inflation has eased to ~6.8% in CY2023, with a projected rate of ~5.9% for CY2024. Pre-pandemic inflation was ~3.5% in CY2019. India’s CPI inflation is expected to drop from ~5.4% in CY2023 to ~4.6% in CY2024, and ~4.2% in CY2025. In the USA and Germany, inflation is projected to fall from ~4.1% and ~6.0% in CY2023 to ~2.9% and ~2.4% in CY2024, reaching ~2.0% in CY2025. China’s inflation may rise from ~0.2% in CY2023 to ~2.0% in CY2025. _Exhibit 1.3: Global Inflation Rate, Average Consumer Price Index (%) of Key Economies (CY)_\n\n_Source: IMF projections, Technopak Analysis Note: For India, CY2017 data refers to FY2018 and so on_ **1\\.2 Overview of Indian Economy** **India’s GDP and GDP Growth (Real and Nominal)** **\\- Historical, Current & Projected Trajectory** India is ranked fifth in the world in terms of nominal gross domestic product (\"GDP\") for FY2024 and is the third- largest economy in the world in terms of purchasing power parity (\"PPP\"). India is projected to be a USD 6.5 trillion economy by FY2029 and is anticipated to become the third largest economy, surpassing Germany, and Japan in terms of Nominal GDP. _Exhibit.1.4: India’s GDP at Current Prices (Nominal GDP) (in USD trillion) and GDP Growth Rate (%) (FY)_\n\n_Source: RBI, Technopak Analysis Note: 1 USD = INR 80_\n\n**1\\.1**\n\n**1\\.2**\n\n**1\\.4**\n\n**1\\.6**\n\n**1\\.7**\n\n**1\\.9**\n\n**2\\.1**\n\n**2\\.4**\n\n**2\\.5**\n\n**2\\.5**\n\n**2\\.9**\n\n**3\\.4**\n\n**3\\.7**\n\n**5\\.8**\n\n**6\\.5**\n\n**13\\.8% 12.9% 11.4%**\n\n**10\\.3%**\n\n**11\\.6%** **11\\.5%** **10\\.3%**\n\n**6\\.4%**\n\n**\\-1.2%**\n\n**16\\.9% 16.9%**\n\n**8\\.8%**\n\n**12\\.0% 12.1%**\n\n\\-5.0%\n\n0\\.0%\n\n5\\.0%\n\n10\\.0%\n\n15\\.0%\n\n20\\.0%\n\n0\\.0\n\n1\\.0\n\n2\\.0\n\n3\\.0\n\n4\\.0\n\n5\\.0\n\n6\\.0\n\n7\\.0\n\n2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024E 2028P 2029P\n\nGDP (USD Tn) GDP Growth Rate(%)\n\n**Country 2017 2018 2019 2020 2021 2022 2023 2024E 2025P 2026P 2027P 2028P**\n\nUSA 2.1% 2.4% 1.8% 1.2% 4.7% 8.0% 4.1% 2.9% 2.0% 2.1% 2.1% 2.1%\n\nChina 1.6% 2.1% 2.9% 2.5% 0.9% 2.0% 0.2% 1.0% 2.0% 2.0% 2.0% 2.0%\n\nJapan 0.5% 1.0% 0.5% 0.0% (0.2%) 2.5% 3.3% 2.2% 2.1% 2.0% 2.0% 2.0%\n\nGermany 1.7% 1.9% 1.4% 0.4% 3.2% 8.7% 6.0% 2.4% 2.0% 2.0% 2.0% 2.0%\n\nIndia 3.6% 3.4% 4.8% 6.2% 5.5% 6.7% 5.4% 4.6% 4.2% 4.1% 4.0% 4.0%\n\nUK 2.7% 2.5% 1.8% 0.9% 2.6% 9.1% 7.3% 2.5% 2.0% 2.0% 2.0% 2.0%\n\nFrance 1.2% 2.1% 1.3% 0.5% 2.1% 5.9% 5.7% 2.4% 1.8% 1.8% 1.8% 1.8%\n\nBrazil 3.4% 3.7% 3.7% 3.2% 8.3% 9.3% 4.6% 4.1% 3.0% 3.1% 3.0% 3.0%\n\nAustralia 2.0% 1.9% 1.6% 0.9% 2.8% 6.6% 5.6% 3.5% 3.0% 2.7% 2.7% 2.6%\n\n**World 3.3% 3.6% 3.5% 3.2% 4.7% 8.7% 6.8% 5.9% 4.5% 3.7% 3.5% 3.4%**","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"4\n\n**Inflation** Inflation, measured by the consumer price index (CPI), reflects changes in the prices of commonly purchased goods and services. Global inflation spiked to ~8.7% in CY2022. The increase in global crude oil and commodity prices, along with COVID-19 supply chain disruptions, have driven inflation, prompting countries to raise interest rates in response. Global inflation has eased to ~6.8% in CY2023, with a projected rate of ~5.9% for CY2024. Pre-pandemic inflation was ~3.5% in CY2019. India’s CPI inflation is expected to drop from ~5.4% in CY2023 to ~4.6% in CY2024, and ~4.2% in CY2025. In the USA and Germany, inflation is projected to fall from ~4.1% and ~6.0% in CY2023 to ~2.9% and ~2.4% in CY2024, reaching ~2.0% in CY2025. China’s inflation may rise from ~0.2% in CY2023 to ~2.0% in CY2025. _Exhibit 1.3: Global Inflation Rate, Average Consumer Price Index (%) of Key Economies (CY)_\n\n_Source: IMF projections, Technopak Analysis Note: For India, CY2017 data refers to FY2018 and so on_ **1\\.2 Overview of Indian Economy** **India’s GDP and GDP Growth (Real and Nominal)** **\\- Historical, Current & Projected Trajectory** India is ranked fifth in the world in terms of nominal gross domestic product (\"GDP\") for FY2024 and is the third- largest economy in the world in terms of purchasing power parity (\"PPP\"). India is projected to be a USD 6.5 trillion economy by FY2029 and is anticipated to become the third largest economy, surpassing Germany, and Japan in terms of Nominal GDP. _Exhibit.1.4: India’s GDP at Current Prices (Nominal GDP) (in USD trillion) and GDP Growth Rate (%) (FY)_\n\n_Source: RBI, Technopak Analysis Note: 1 USD = INR 80_\n\n**1\\.1**\n\n**1\\.2**\n\n**1\\.4**\n\n**1\\.6**\n\n**1\\.7**\n\n**1\\.9**\n\n**2\\.1**\n\n**2\\.4**\n\n**2\\.5**\n\n**2\\.5**\n\n**2\\.9**\n\n**3\\.4**\n\n**3\\.7**\n\n**5\\.8**\n\n**6\\.5**\n\n**13\\.8% 12.9% 11.4%**\n\n**10\\.3%**\n\n**11\\.6%** **11\\.5%** **10\\.3%**\n\n**6\\.4%**\n\n**\\-1.2%**\n\n**16\\.9% 16.9%**\n\n**8\\.8%**\n\n**12\\.0% 12.1%**\n\n\\-5.0%\n\n0\\.0%\n\n5\\.0%\n\n10\\.0%\n\n15\\.0%\n\n20\\.0%\n\n0\\.0\n\n1\\.0\n\n2\\.0\n\n3\\.0\n\n4\\.0\n\n5\\.0\n\n6\\.0\n\n7\\.0\n\n2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024E 2028P 2029P\n\nGDP (USD Tn) GDP Growth Rate(%)\n\n**Country 2017 2018 2019 2020 2021 2022 2023 2024E 2025P 2026P 2027P 2028P**\n\nUSA 2.1% 2.4% 1.8% 1.2% 4.7% 8.0% 4.1% 2.9% 2.0% 2.1% 2.1% 2.1%\n\nChina 1.6% 2.1% 2.9% 2.5% 0.9% 2.0% 0.2% 1.0% 2.0% 2.0% 2.0% 2.0%\n\nJapan 0.5% 1.0% 0.5% 0.0% (0.2%) 2.5% 3.3% 2.2% 2.1% 2.0% 2.0% 2.0%\n\nGermany 1.7% 1.9% 1.4% 0.4% 3.2% 8.7% 6.0% 2.4% 2.0% 2.0% 2.0% 2.0%\n\nIndia 3.6% 3.4% 4.8% 6.2% 5.5% 6.7% 5.4% 4.6% 4.2% 4.1% 4.0% 4.0%\n\nUK 2.7% 2.5% 1.8% 0.9% 2.6% 9.1% 7.3% 2.5% 2.0% 2.0% 2.0% 2.0%\n\nFrance 1.2% 2.1% 1.3% 0.5% 2.1% 5.9% 5.7% 2.4% 1.8% 1.8% 1.8% 1.8%\n\nBrazil 3.4% 3.7% 3.7% 3.2% 8.3% 9.3% 4.6% 4.1% 3.0% 3.1% 3.0% 3.0%\n\nAustralia 2.0% 1.9% 1.6% 0.9% 2.8% 6.6% 5.6% 3.5% 3.0% 2.7% 2.7% 2.6%\n\n**World 3.3% 3.6% 3.5% 3.2% 4.7% 8.7% 6.8% 5.9% 4.5% 3.7% 3.5% 3.4%**","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Technopak_Industry_Report_p4.pdf"]},"response":{"errors":[],"extract_id":"extract_91953307f32d73991e56666f8584c8fa","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["4\n\n**Inflation** Inflation, measured by the consumer price index (CPI), reflects changes in the prices of commonly purchased goods and services. Global inflation spiked to ~8.7% in CY2022. The increase in global crude oil and commodity prices, along with COVID-19 supply chain disruptions, have driven inflation, prompting countries to raise interest rates in response. Global inflation has eased to ~6.8% in CY2023, with a projected rate of ~5.9% for CY2024. Pre-pandemic inflation was ~3.5% in CY2019. India’s CPI inflation is expected to drop from ~5.4% in CY2023 to ~4.6% in CY2024, and ~4.2% in CY2025. In the USA and Germany, inflation is projected to fall from ~4.1% and ~6.0% in CY2023 to ~2.9% and ~2.4% in CY2024, reaching ~2.0% in CY2025. China’s inflation may rise from ~0.2% in CY2023 to ~2.0% in CY2025. _Exhibit 1.3: Global Inflation Rate, Average Consumer Price Index (%) of Key Economies (CY)_\n\n_Source: IMF projections, Technopak Analysis Note: For India, CY2017 data refers to FY2018 and so on_ **1\\.2 Overview of Indian Economy** **India’s GDP and GDP Growth (Real and Nominal)** **\\- Historical, Current & Projected Trajectory** India is ranked fifth in the world in terms of nominal gross domestic product (\"GDP\") for FY2024 and is the third- largest economy in the world in terms of purchasing power parity (\"PPP\"). India is projected to be a USD 6.5 trillion economy by FY2029 and is anticipated to become the third largest economy, surpassing Germany, and Japan in terms of Nominal GDP. _Exhibit.1.4: India’s GDP at Current Prices (Nominal GDP) (in USD trillion) and GDP Growth Rate (%) (FY)_\n\n_Source: RBI, Technopak Analysis Note: 1 USD = INR 80_\n\n**1\\.1**\n\n**1\\.2**\n\n**1\\.4**\n\n**1\\.6**\n\n**1\\.7**\n\n**1\\.9**\n\n**2\\.1**\n\n**2\\.4**\n\n**2\\.5**\n\n**2\\.5**\n\n**2\\.9**\n\n**3\\.4**\n\n**3\\.7**\n\n**5\\.8**\n\n**6\\.5**\n\n**13\\.8% 12.9% 11.4%**\n\n**10\\.3%**\n\n**11\\.6%** **11\\.5%** **10\\.3%**\n\n**6\\.4%**\n\n**\\-1.2%**\n\n**16\\.9% 16.9%**\n\n**8\\.8%**\n\n**12\\.0% 12.1%**\n\n\\-5.0%\n\n0\\.0%\n\n5\\.0%\n\n10\\.0%\n\n15\\.0%\n\n20\\.0%\n\n0\\.0\n\n1\\.0\n\n2\\.0\n\n3\\.0\n\n4\\.0\n\n5\\.0\n\n6\\.0\n\n7\\.0\n\n2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024E 2028P 2029P\n\nGDP (USD Tn) GDP Growth Rate(%)\n\n**Country 2017 2018 2019 2020 2021 2022 2023 2024E 2025P 2026P 2027P 2028P**\n\nUSA 2.1% 2.4% 1.8% 1.2% 4.7% 8.0% 4.1% 2.9% 2.0% 2.1% 2.1% 2.1%\n\nChina 1.6% 2.1% 2.9% 2.5% 0.9% 2.0% 0.2% 1.0% 2.0% 2.0% 2.0% 2.0%\n\nJapan 0.5% 1.0% 0.5% 0.0% (0.2%) 2.5% 3.3% 2.2% 2.1% 2.0% 2.0% 2.0%\n\nGermany 1.7% 1.9% 1.4% 0.4% 3.2% 8.7% 6.0% 2.4% 2.0% 2.0% 2.0% 2.0%\n\nIndia 3.6% 3.4% 4.8% 6.2% 5.5% 6.7% 5.4% 4.6% 4.2% 4.1% 4.0% 4.0%\n\nUK 2.7% 2.5% 1.8% 0.9% 2.6% 9.1% 7.3% 2.5% 2.0% 2.0% 2.0% 2.0%\n\nFrance 1.2% 2.1% 1.3% 0.5% 2.1% 5.9% 5.7% 2.4% 1.8% 1.8% 1.8% 1.8%\n\nBrazil 3.4% 3.7% 3.7% 3.2% 8.3% 9.3% 4.6% 4.1% 3.0% 3.1% 3.0% 3.0%\n\nAustralia 2.0% 1.9% 1.6% 0.9% 2.8% 6.6% 5.6% 3.5% 3.0% 2.7% 2.7% 2.6%\n\n**World 3.3% 3.6% 3.5% 3.2% 4.7% 8.7% 6.8% 5.9% 4.5% 3.7% 3.5% 3.4%**"],"full_content":"4\n\n**Inflation** Inflation, measured by the consumer price index (CPI), reflects changes in the prices of commonly purchased goods and services. Global inflation spiked to ~8.7% in CY2022. The increase in global crude oil and commodity prices, along with COVID-19 supply chain disruptions, have driven inflation, prompting countries to raise interest rates in response. Global inflation has eased to ~6.8% in CY2023, with a projected rate of ~5.9% for CY2024. Pre-pandemic inflation was ~3.5% in CY2019. India’s CPI inflation is expected to drop from ~5.4% in CY2023 to ~4.6% in CY2024, and ~4.2% in CY2025. In the USA and Germany, inflation is projected to fall from ~4.1% and ~6.0% in CY2023 to ~2.9% and ~2.4% in CY2024, reaching ~2.0% in CY2025. China’s inflation may rise from ~0.2% in CY2023 to ~2.0% in CY2025. _Exhibit 1.3: Global Inflation Rate, Average Consumer Price Index (%) of Key Economies (CY)_\n\n_Source: IMF projections, Technopak Analysis Note: For India, CY2017 data refers to FY2018 and so on_ **1\\.2 Overview of Indian Economy** **India’s GDP and GDP Growth (Real and Nominal)** **\\- Historical, Current & Projected Trajectory** India is ranked fifth in the world in terms of nominal gross domestic product (\"GDP\") for FY2024 and is the third- largest economy in the world in terms of purchasing power parity (\"PPP\"). India is projected to be a USD 6.5 trillion economy by FY2029 and is anticipated to become the third largest economy, surpassing Germany, and Japan in terms of Nominal GDP. _Exhibit.1.4: India’s GDP at Current Prices (Nominal GDP) (in USD trillion) and GDP Growth Rate (%) (FY)_\n\n_Source: RBI, Technopak Analysis Note: 1 USD = INR 80_\n\n**1\\.1**\n\n**1\\.2**\n\n**1\\.4**\n\n**1\\.6**\n\n**1\\.7**\n\n**1\\.9**\n\n**2\\.1**\n\n**2\\.4**\n\n**2\\.5**\n\n**2\\.5**\n\n**2\\.9**\n\n**3\\.4**\n\n**3\\.7**\n\n**5\\.8**\n\n**6\\.5**\n\n**13\\.8% 12.9% 11.4%**\n\n**10\\.3%**\n\n**11\\.6%** **11\\.5%** **10\\.3%**\n\n**6\\.4%**\n\n**\\-1.2%**\n\n**16\\.9% 16.9%**\n\n**8\\.8%**\n\n**12\\.0% 12.1%**\n\n\\-5.0%\n\n0\\.0%\n\n5\\.0%\n\n10\\.0%\n\n15\\.0%\n\n20\\.0%\n\n0\\.0\n\n1\\.0\n\n2\\.0\n\n3\\.0\n\n4\\.0\n\n5\\.0\n\n6\\.0\n\n7\\.0\n\n2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024E 2028P 2029P\n\nGDP (USD Tn) GDP Growth Rate(%)\n\n**Country 2017 2018 2019 2020 2021 2022 2023 2024E 2025P 2026P 2027P 2028P**\n\nUSA 2.1% 2.4% 1.8% 1.2% 4.7% 8.0% 4.1% 2.9% 2.0% 2.1% 2.1% 2.1%\n\nChina 1.6% 2.1% 2.9% 2.5% 0.9% 2.0% 0.2% 1.0% 2.0% 2.0% 2.0% 2.0%\n\nJapan 0.5% 1.0% 0.5% 0.0% (0.2%) 2.5% 3.3% 2.2% 2.1% 2.0% 2.0% 2.0%\n\nGermany 1.7% 1.9% 1.4% 0.4% 3.2% 8.7% 6.0% 2.4% 2.0% 2.0% 2.0% 2.0%\n\nIndia 3.6% 3.4% 4.8% 6.2% 5.5% 6.7% 5.4% 4.6% 4.2% 4.1% 4.0% 4.0%\n\nUK 2.7% 2.5% 1.8% 0.9% 2.6% 9.1% 7.3% 2.5% 2.0% 2.0% 2.0% 2.0%\n\nFrance 1.2% 2.1% 1.3% 0.5% 2.1% 5.9% 5.7% 2.4% 1.8% 1.8% 1.8% 1.8%\n\nBrazil 3.4% 3.7% 3.7% 3.2% 8.3% 9.3% 4.6% 4.1% 3.0% 3.1% 3.0% 3.0%\n\nAustralia 2.0% 1.9% 1.6% 0.9% 2.8% 6.6% 5.6% 3.5% 3.0% 2.7% 2.7% 2.6%\n\n**World 3.3% 3.6% 3.5% 3.2% 4.7% 8.7% 6.8% 5.9% 4.5% 3.7% 3.5% 3.4%**","publish_date":null,"title":"Inflation 1.2 Overview of Indian Economy 4","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Technopak_Industry_Report_p4.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Technopak_Industry_Report_p4.pdf"},"request":{"example_id":"chart/Technopak_Industry_Report_p4","product_type":"parse","source_file_path":"Technopak_Industry_Report_p4.pdf"},"started_at":"2026-11-04T19:45:09.795346"} +{"completed_at":"2026-11-04T19:45:13.535400","latency_in_ms":1475,"output":{"example_id":"chart/Technopak_Industry_Report_p6","grounded_pages":[],"layout_pages":[],"markdown":"6\n\n_Exhibit 1.7:_ _India’s Per Capita Consumption Expenditure (Current Prices) and Growth (%) (in INR) (FY)_\n\n_Source: Ministry of Statistics and Program Implementation , Technopak Analysis_ **Demographic Profile of India** India has one of the youngest populations globally compared to other leading economies. The median age in India was 29.5 years for CY2023, as compared to 38.5 years and 39.8 years in the USA and China respectively and is expected to remain under 30 years until CY2030. The younger population is naturally predisposed to adopting the latest trends and exploration, given their educational profile and exposure to media and technology. This presents an opportunity for domestic consumption in the form of branded products and organized retail. _Exhibit 1.8: Median Age: Key Emerging & Developed Economies (CY2023)_\n\n**Country India China USA Singapore Russia Brazil Vietnam UK**\n\nMedian Age (Yrs.) 29.5 39.8 38.5 38.9 41.5 34.7 32.7 40.6\n\n_Source: World Population Review Note: For India, Data for CY2023 refers to FY2024_ **_More than half of India’s population falls in the 15_** **_\\-49-year age bracket_**\n\nAs of April 2024, India, the most populous country in the world with approximately 1.44 billion people, has a demographic composition that highlights its potential as a dynamic consumer market. The population is distributed as 25% under the age of 15, 68% within the working-age group of 15-65 years, and 7% aged 65 and above. This youthful and predominantly working-age population positions India with a significant demographic advantage. The large base of younger consumers is not only more informed and open to experimentation but also early adopters of brands, setting trends across various industries. Despite varying levels of disposable income, this group prioritizes spending on experiences and lifestyle-enhancing products, fueling demand and driving innovation in the market. _Exhibit 1.9: Population Distribution of India by Age (%) (CY2023)_\n\n_Source: World Bank and Technopak Estimates Note: For India, CY2023 refers to FY2024 data_\n\n**57,402**\n\n**63,065**\n\n**70,175**\n\n**76,794**\n\n**84,808**\n\n**91,254**\n\n**88,775**\n\n**1,05,092**\n\n**1,18,755**\n\n**1,27,760**\n\n**1,35,762**\n\n**1,77,956**\n\n**9\\.9% 11.3%**\n\n**9\\.4% 10.4%**\n\n**7\\.6%**\n\n**\\-2.7%**\n\n**18\\.4%**\n\n**13\\.0% 7.6%**\n\n**6\\.3%** **7\\.00%**\n\n\\-5.0%\n\n0\\.0%\n\n5\\.0%\n\n10\\.0%\n\n15\\.0%\n\n20\\.0%\n\n0\n\n50,000\n\n1,00,000\n\n1,50,000\n\n2,00,000\n\n2015 2016 2017 2018 2019 2020 2021 2022 2023 2024E 2025P 2029P\n\nPer Capita PFCE Per Capita PFCE Growth Rate(%)\n\n**25%**\n\n**68%**\n\n**7%**\n\n**<15 15-65 >65+**","pages":[{"line_numbers":[],"markdown":"6\n\n_Exhibit 1.7:_ _India’s Per Capita Consumption Expenditure (Current Prices) and Growth (%) (in INR) (FY)_\n\n_Source: Ministry of Statistics and Program Implementation , Technopak Analysis_ **Demographic Profile of India** India has one of the youngest populations globally compared to other leading economies. The median age in India was 29.5 years for CY2023, as compared to 38.5 years and 39.8 years in the USA and China respectively and is expected to remain under 30 years until CY2030. The younger population is naturally predisposed to adopting the latest trends and exploration, given their educational profile and exposure to media and technology. This presents an opportunity for domestic consumption in the form of branded products and organized retail. _Exhibit 1.8: Median Age: Key Emerging & Developed Economies (CY2023)_\n\n**Country India China USA Singapore Russia Brazil Vietnam UK**\n\nMedian Age (Yrs.) 29.5 39.8 38.5 38.9 41.5 34.7 32.7 40.6\n\n_Source: World Population Review Note: For India, Data for CY2023 refers to FY2024_ **_More than half of India’s population falls in the 15_** **_\\-49-year age bracket_**\n\nAs of April 2024, India, the most populous country in the world with approximately 1.44 billion people, has a demographic composition that highlights its potential as a dynamic consumer market. The population is distributed as 25% under the age of 15, 68% within the working-age group of 15-65 years, and 7% aged 65 and above. This youthful and predominantly working-age population positions India with a significant demographic advantage. The large base of younger consumers is not only more informed and open to experimentation but also early adopters of brands, setting trends across various industries. Despite varying levels of disposable income, this group prioritizes spending on experiences and lifestyle-enhancing products, fueling demand and driving innovation in the market. _Exhibit 1.9: Population Distribution of India by Age (%) (CY2023)_\n\n_Source: World Bank and Technopak Estimates Note: For India, CY2023 refers to FY2024 data_\n\n**57,402**\n\n**63,065**\n\n**70,175**\n\n**76,794**\n\n**84,808**\n\n**91,254**\n\n**88,775**\n\n**1,05,092**\n\n**1,18,755**\n\n**1,27,760**\n\n**1,35,762**\n\n**1,77,956**\n\n**9\\.9% 11.3%**\n\n**9\\.4% 10.4%**\n\n**7\\.6%**\n\n**\\-2.7%**\n\n**18\\.4%**\n\n**13\\.0% 7.6%**\n\n**6\\.3%** **7\\.00%**\n\n\\-5.0%\n\n0\\.0%\n\n5\\.0%\n\n10\\.0%\n\n15\\.0%\n\n20\\.0%\n\n0\n\n50,000\n\n1,00,000\n\n1,50,000\n\n2,00,000\n\n2015 2016 2017 2018 2019 2020 2021 2022 2023 2024E 2025P 2029P\n\nPer Capita PFCE Per Capita PFCE Growth Rate(%)\n\n**25%**\n\n**68%**\n\n**7%**\n\n**<15 15-65 >65+**","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"6\n\n_Exhibit 1.7:_ _India’s Per Capita Consumption Expenditure (Current Prices) and Growth (%) (in INR) (FY)_\n\n_Source: Ministry of Statistics and Program Implementation , Technopak Analysis_ **Demographic Profile of India** India has one of the youngest populations globally compared to other leading economies. The median age in India was 29.5 years for CY2023, as compared to 38.5 years and 39.8 years in the USA and China respectively and is expected to remain under 30 years until CY2030. The younger population is naturally predisposed to adopting the latest trends and exploration, given their educational profile and exposure to media and technology. This presents an opportunity for domestic consumption in the form of branded products and organized retail. _Exhibit 1.8: Median Age: Key Emerging & Developed Economies (CY2023)_\n\n**Country India China USA Singapore Russia Brazil Vietnam UK**\n\nMedian Age (Yrs.) 29.5 39.8 38.5 38.9 41.5 34.7 32.7 40.6\n\n_Source: World Population Review Note: For India, Data for CY2023 refers to FY2024_ **_More than half of India’s population falls in the 15_** **_\\-49-year age bracket_**\n\nAs of April 2024, India, the most populous country in the world with approximately 1.44 billion people, has a demographic composition that highlights its potential as a dynamic consumer market. The population is distributed as 25% under the age of 15, 68% within the working-age group of 15-65 years, and 7% aged 65 and above. This youthful and predominantly working-age population positions India with a significant demographic advantage. The large base of younger consumers is not only more informed and open to experimentation but also early adopters of brands, setting trends across various industries. Despite varying levels of disposable income, this group prioritizes spending on experiences and lifestyle-enhancing products, fueling demand and driving innovation in the market. _Exhibit 1.9: Population Distribution of India by Age (%) (CY2023)_\n\n_Source: World Bank and Technopak Estimates Note: For India, CY2023 refers to FY2024 data_\n\n**57,402**\n\n**63,065**\n\n**70,175**\n\n**76,794**\n\n**84,808**\n\n**91,254**\n\n**88,775**\n\n**1,05,092**\n\n**1,18,755**\n\n**1,27,760**\n\n**1,35,762**\n\n**1,77,956**\n\n**9\\.9% 11.3%**\n\n**9\\.4% 10.4%**\n\n**7\\.6%**\n\n**\\-2.7%**\n\n**18\\.4%**\n\n**13\\.0% 7.6%**\n\n**6\\.3%** **7\\.00%**\n\n\\-5.0%\n\n0\\.0%\n\n5\\.0%\n\n10\\.0%\n\n15\\.0%\n\n20\\.0%\n\n0\n\n50,000\n\n1,00,000\n\n1,50,000\n\n2,00,000\n\n2015 2016 2017 2018 2019 2020 2021 2022 2023 2024E 2025P 2029P\n\nPer Capita PFCE Per Capita PFCE Growth Rate(%)\n\n**25%**\n\n**68%**\n\n**7%**\n\n**<15 15-65 >65+**","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Technopak_Industry_Report_p6.pdf"]},"response":{"errors":[],"extract_id":"extract_303191d0ea553192b0e051a4d2bc5659","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["6\n\n_Exhibit 1.7:_ _India’s Per Capita Consumption Expenditure (Current Prices) and Growth (%) (in INR) (FY)_\n\n_Source: Ministry of Statistics and Program Implementation , Technopak Analysis_ **Demographic Profile of India** India has one of the youngest populations globally compared to other leading economies. The median age in India was 29.5 years for CY2023, as compared to 38.5 years and 39.8 years in the USA and China respectively and is expected to remain under 30 years until CY2030. The younger population is naturally predisposed to adopting the latest trends and exploration, given their educational profile and exposure to media and technology. This presents an opportunity for domestic consumption in the form of branded products and organized retail. _Exhibit 1.8: Median Age: Key Emerging & Developed Economies (CY2023)_\n\n**Country India China USA Singapore Russia Brazil Vietnam UK**\n\nMedian Age (Yrs.) 29.5 39.8 38.5 38.9 41.5 34.7 32.7 40.6\n\n_Source: World Population Review Note: For India, Data for CY2023 refers to FY2024_ **_More than half of India’s population falls in the 15_** **_\\-49-year age bracket_**\n\nAs of April 2024, India, the most populous country in the world with approximately 1.44 billion people, has a demographic composition that highlights its potential as a dynamic consumer market. The population is distributed as 25% under the age of 15, 68% within the working-age group of 15-65 years, and 7% aged 65 and above. This youthful and predominantly working-age population positions India with a significant demographic advantage. The large base of younger consumers is not only more informed and open to experimentation but also early adopters of brands, setting trends across various industries. Despite varying levels of disposable income, this group prioritizes spending on experiences and lifestyle-enhancing products, fueling demand and driving innovation in the market. _Exhibit 1.9: Population Distribution of India by Age (%) (CY2023)_\n\n_Source: World Bank and Technopak Estimates Note: For India, CY2023 refers to FY2024 data_\n\n**57,402**\n\n**63,065**\n\n**70,175**\n\n**76,794**\n\n**84,808**\n\n**91,254**\n\n**88,775**\n\n**1,05,092**\n\n**1,18,755**\n\n**1,27,760**\n\n**1,35,762**\n\n**1,77,956**\n\n**9\\.9% 11.3%**\n\n**9\\.4% 10.4%**\n\n**7\\.6%**\n\n**\\-2.7%**\n\n**18\\.4%**\n\n**13\\.0% 7.6%**\n\n**6\\.3%** **7\\.00%**\n\n\\-5.0%\n\n0\\.0%\n\n5\\.0%\n\n10\\.0%\n\n15\\.0%\n\n20\\.0%\n\n0\n\n50,000\n\n1,00,000\n\n1,50,000\n\n2,00,000\n\n2015 2016 2017 2018 2019 2020 2021 2022 2023 2024E 2025P 2029P\n\nPer Capita PFCE Per Capita PFCE Growth Rate(%)\n\n**25%**\n\n**68%**\n\n**7%**\n\n**<15 15-65 >65+**"],"full_content":"6\n\n_Exhibit 1.7:_ _India’s Per Capita Consumption Expenditure (Current Prices) and Growth (%) (in INR) (FY)_\n\n_Source: Ministry of Statistics and Program Implementation , Technopak Analysis_ **Demographic Profile of India** India has one of the youngest populations globally compared to other leading economies. The median age in India was 29.5 years for CY2023, as compared to 38.5 years and 39.8 years in the USA and China respectively and is expected to remain under 30 years until CY2030. The younger population is naturally predisposed to adopting the latest trends and exploration, given their educational profile and exposure to media and technology. This presents an opportunity for domestic consumption in the form of branded products and organized retail. _Exhibit 1.8: Median Age: Key Emerging & Developed Economies (CY2023)_\n\n**Country India China USA Singapore Russia Brazil Vietnam UK**\n\nMedian Age (Yrs.) 29.5 39.8 38.5 38.9 41.5 34.7 32.7 40.6\n\n_Source: World Population Review Note: For India, Data for CY2023 refers to FY2024_ **_More than half of India’s population falls in the 15_** **_\\-49-year age bracket_**\n\nAs of April 2024, India, the most populous country in the world with approximately 1.44 billion people, has a demographic composition that highlights its potential as a dynamic consumer market. The population is distributed as 25% under the age of 15, 68% within the working-age group of 15-65 years, and 7% aged 65 and above. This youthful and predominantly working-age population positions India with a significant demographic advantage. The large base of younger consumers is not only more informed and open to experimentation but also early adopters of brands, setting trends across various industries. Despite varying levels of disposable income, this group prioritizes spending on experiences and lifestyle-enhancing products, fueling demand and driving innovation in the market. _Exhibit 1.9: Population Distribution of India by Age (%) (CY2023)_\n\n_Source: World Bank and Technopak Estimates Note: For India, CY2023 refers to FY2024 data_\n\n**57,402**\n\n**63,065**\n\n**70,175**\n\n**76,794**\n\n**84,808**\n\n**91,254**\n\n**88,775**\n\n**1,05,092**\n\n**1,18,755**\n\n**1,27,760**\n\n**1,35,762**\n\n**1,77,956**\n\n**9\\.9% 11.3%**\n\n**9\\.4% 10.4%**\n\n**7\\.6%**\n\n**\\-2.7%**\n\n**18\\.4%**\n\n**13\\.0% 7.6%**\n\n**6\\.3%** **7\\.00%**\n\n\\-5.0%\n\n0\\.0%\n\n5\\.0%\n\n10\\.0%\n\n15\\.0%\n\n20\\.0%\n\n0\n\n50,000\n\n1,00,000\n\n1,50,000\n\n2,00,000\n\n2015 2016 2017 2018 2019 2020 2021 2022 2023 2024E 2025P 2029P\n\nPer Capita PFCE Per Capita PFCE Growth Rate(%)\n\n**25%**\n\n**68%**\n\n**7%**\n\n**<15 15-65 >65+**","publish_date":null,"title":"6","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Technopak_Industry_Report_p6.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Technopak_Industry_Report_p6.pdf"},"request":{"example_id":"chart/Technopak_Industry_Report_p6","product_type":"parse","source_file_path":"Technopak_Industry_Report_p6.pdf"},"started_at":"2026-11-04T19:45:12.059803"} +{"completed_at":"2026-11-04T19:45:14.364964","latency_in_ms":1025,"output":{"example_id":"chart/The_State_of_UK_Competition_Report_2024_p21","grounded_pages":[],"layout_pages":[],"markdown":"19\n\n### **_Figure 5: Average markups in GB have risen since 1997_**\n\n### _Economy-wide average markup estimates. Baseline measure: Ordinary Least Square (OLS) estimation of a translog production function with materials as flexible input. Firm-level estimates are aggregated and weighted by turnover. Data from Annual Respondents Database X (ARDx) 1997-2020 and Annual Business Survey (ABS) 2021. GB only_\n\n### **Services have been the major contributor to the rise in markups**\n\n### 2\\.20 The average rise in markups has been driven predominantly by the service sector. Markups in administrative and support services, professional, technical, and scientific services, and arts and entertainment (in more recent years) have increased particularly rapidly. In contrast, markups in manufacturing, construction as well as accommodation and food services have seen milder increases in the last two decades.\n\n### 2\\.21 Figure 6 compares the rise in sectoral markups to the whole-economy average. It illustrates that the rise in markups varies widely. Markups have been rising in professional services over the entire time period, whereas the rise in administrative and support services and arts and entertainment only begins","pages":[{"line_numbers":[],"markdown":"19\n\n### **_Figure 5: Average markups in GB have risen since 1997_**\n\n### _Economy-wide average markup estimates. Baseline measure: Ordinary Least Square (OLS) estimation of a translog production function with materials as flexible input. Firm-level estimates are aggregated and weighted by turnover. Data from Annual Respondents Database X (ARDx) 1997-2020 and Annual Business Survey (ABS) 2021. GB only_\n\n### **Services have been the major contributor to the rise in markups**\n\n### 2\\.20 The average rise in markups has been driven predominantly by the service sector. Markups in administrative and support services, professional, technical, and scientific services, and arts and entertainment (in more recent years) have increased particularly rapidly. In contrast, markups in manufacturing, construction as well as accommodation and food services have seen milder increases in the last two decades.\n\n### 2\\.21 Figure 6 compares the rise in sectoral markups to the whole-economy average. It illustrates that the rise in markups varies widely. Markups have been rising in professional services over the entire time period, whereas the rise in administrative and support services and arts and entertainment only begins","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"19\n\n### **_Figure 5: Average markups in GB have risen since 1997_**\n\n### _Economy-wide average markup estimates. Baseline measure: Ordinary Least Square (OLS) estimation of a translog production function with materials as flexible input. Firm-level estimates are aggregated and weighted by turnover. Data from Annual Respondents Database X (ARDx) 1997-2020 and Annual Business Survey (ABS) 2021. GB only_\n\n### **Services have been the major contributor to the rise in markups**\n\n### 2\\.20 The average rise in markups has been driven predominantly by the service sector. Markups in administrative and support services, professional, technical, and scientific services, and arts and entertainment (in more recent years) have increased particularly rapidly. In contrast, markups in manufacturing, construction as well as accommodation and food services have seen milder increases in the last two decades.\n\n### 2\\.21 Figure 6 compares the rise in sectoral markups to the whole-economy average. It illustrates that the rise in markups varies widely. Markups have been rising in professional services over the entire time period, whereas the rise in administrative and support services and arts and entertainment only begins","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/The_State_of_UK_Competition_Report_2024_p21.pdf"]},"response":{"errors":[],"extract_id":"extract_06555de597938bf5026de85afd59e62a","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["19\n\n### **_Figure 5: Average markups in GB have risen since 1997_**\n\n### _Economy-wide average markup estimates. Baseline measure: Ordinary Least Square (OLS) estimation of a translog production function with materials as flexible input. Firm-level estimates are aggregated and weighted by turnover. Data from Annual Respondents Database X (ARDx) 1997-2020 and Annual Business Survey (ABS) 2021. GB only_\n\n### **Services have been the major contributor to the rise in markups**\n\n### 2\\.20 The average rise in markups has been driven predominantly by the service sector. Markups in administrative and support services, professional, technical, and scientific services, and arts and entertainment (in more recent years) have increased particularly rapidly. In contrast, markups in manufacturing, construction as well as accommodation and food services have seen milder increases in the last two decades.\n\n### 2\\.21 Figure 6 compares the rise in sectoral markups to the whole-economy average. It illustrates that the rise in markups varies widely. Markups have been rising in professional services over the entire time period, whereas the rise in administrative and support services and arts and entertainment only begins"],"full_content":"19\n\n### **_Figure 5: Average markups in GB have risen since 1997_**\n\n### _Economy-wide average markup estimates. Baseline measure: Ordinary Least Square (OLS) estimation of a translog production function with materials as flexible input. Firm-level estimates are aggregated and weighted by turnover. Data from Annual Respondents Database X (ARDx) 1997-2020 and Annual Business Survey (ABS) 2021. GB only_\n\n### **Services have been the major contributor to the rise in markups**\n\n### 2\\.20 The average rise in markups has been driven predominantly by the service sector. Markups in administrative and support services, professional, technical, and scientific services, and arts and entertainment (in more recent years) have increased particularly rapidly. In contrast, markups in manufacturing, construction as well as accommodation and food services have seen milder increases in the last two decades.\n\n### 2\\.21 Figure 6 compares the rise in sectoral markups to the whole-economy average. It illustrates that the rise in markups varies widely. Markups have been rising in professional services over the entire time period, whereas the rise in administrative and support services and arts and entertainment only begins","publish_date":null,"title":"Services have been the major contributor to the rise in markups","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/The_State_of_UK_Competition_Report_2024_p21.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/The_State_of_UK_Competition_Report_2024_p21.pdf"},"request":{"example_id":"chart/The_State_of_UK_Competition_Report_2024_p21","product_type":"parse","source_file_path":"The_State_of_UK_Competition_Report_2024_p21.pdf"},"started_at":"2026-11-04T19:45:13.339724"} +{"completed_at":"2026-11-04T19:45:16.019669","latency_in_ms":1250,"output":{"example_id":"chart/The_State_of_UK_Competition_Report_2024_p22","grounded_pages":[],"layout_pages":[],"markdown":"20\n\n### after the Great Financial Crisis. Manufacturing markups have not risen over the past twenty-five years.\n\n### **_Figure 6: Some service sectors have experienced the steepest rise in average markups_**\n\n### _Average markups in Standard Industrial Classification (SIC) sectors. Ordinary Least Square (OLS) estimation of a Translog production function with materials as flexible input. Firm-level estimates are aggregated weighting by turnover. Data from Annual Respondents Database X (ARDx) 1997-2020 and Annual Business Survey (ABS) 2021. GB only_","pages":[{"line_numbers":[],"markdown":"20\n\n### after the Great Financial Crisis. Manufacturing markups have not risen over the past twenty-five years.\n\n### **_Figure 6: Some service sectors have experienced the steepest rise in average markups_**\n\n### _Average markups in Standard Industrial Classification (SIC) sectors. Ordinary Least Square (OLS) estimation of a Translog production function with materials as flexible input. Firm-level estimates are aggregated weighting by turnover. Data from Annual Respondents Database X (ARDx) 1997-2020 and Annual Business Survey (ABS) 2021. GB only_","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"20\n\n### after the Great Financial Crisis. Manufacturing markups have not risen over the past twenty-five years.\n\n### **_Figure 6: Some service sectors have experienced the steepest rise in average markups_**\n\n### _Average markups in Standard Industrial Classification (SIC) sectors. Ordinary Least Square (OLS) estimation of a Translog production function with materials as flexible input. Firm-level estimates are aggregated weighting by turnover. Data from Annual Respondents Database X (ARDx) 1997-2020 and Annual Business Survey (ABS) 2021. GB only_","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/The_State_of_UK_Competition_Report_2024_p22.pdf"]},"response":{"errors":[],"extract_id":"extract_18f19172d9d3e1fc7eddb06247bcc877","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["20\n\n### after the Great Financial Crisis. Manufacturing markups have not risen over the past twenty-five years.\n\n### **_Figure 6: Some service sectors have experienced the steepest rise in average markups_**\n\n### _Average markups in Standard Industrial Classification (SIC) sectors. Ordinary Least Square (OLS) estimation of a Translog production function with materials as flexible input. Firm-level estimates are aggregated weighting by turnover. Data from Annual Respondents Database X (ARDx) 1997-2020 and Annual Business Survey (ABS) 2021. GB only_"],"full_content":"20\n\n### after the Great Financial Crisis. Manufacturing markups have not risen over the past twenty-five years.\n\n### **_Figure 6: Some service sectors have experienced the steepest rise in average markups_**\n\n### _Average markups in Standard Industrial Classification (SIC) sectors. Ordinary Least Square (OLS) estimation of a Translog production function with materials as flexible input. Firm-level estimates are aggregated weighting by turnover. Data from Annual Respondents Database X (ARDx) 1997-2020 and Annual Business Survey (ABS) 2021. GB only_","publish_date":null,"title":"after the Great Financial Crisis. Manufacturing markups have not risen over the past twenty-five years. Figure 6: Some service sectors","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/The_State_of_UK_Competition_Report_2024_p22.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/The_State_of_UK_Competition_Report_2024_p22.pdf"},"request":{"example_id":"chart/The_State_of_UK_Competition_Report_2024_p22","product_type":"parse","source_file_path":"The_State_of_UK_Competition_Report_2024_p22.pdf"},"started_at":"2026-11-04T19:45:14.769415"} +{"completed_at":"2026-11-04T19:45:18.306705","latency_in_ms":2285,"output":{"example_id":"chart/The_State_of_UK_Competition_Report_2024_p28","grounded_pages":[],"layout_pages":[],"markdown":"26\n\n### **_Figure 10: Markups have risen more at the top of the distribution than elsewhere_**\n\n### _Evolution of average markups in different quantiles of the markup distribution. Data from Annual Respondents Database X (ARDx) 1997-2020 and Annual Business Survey (ABS) 2021. GB only_\n\n### 2\\.41 We first show how key firm characteristics like age, size, profit shares and labour shares are related to markups across the economy. The profit share measures what fraction of a firm’s sales are retained as profits, and the labour share measures the fraction of sales that are used to meet employment costs.\n\n### 2\\.42 As markups represent the profit on each additional sale, we expect markups to be positively correlated with the profit share. The closeness of this relationship depends on fixed costs and the returns to scale of the production technology. For instance, if firms incur large fixed costs such as marketing overheads, they might have large markups but a small profit share.\n\n### 2\\.43 As for the labour share, recent work suggests higher concentration in output markets is associated with lower labour shares (e.g., Autor, Dorn, Katz, Patterson, Van Reenen, 2020; De Loecker, Eeckhout and Unger, 2020; Deb, Eeckhout, Patel and Warren, 2023). Higher markups require firms to produce less than in a competitive market, so they hire less labour. Therefore, absent","pages":[{"line_numbers":[],"markdown":"26\n\n### **_Figure 10: Markups have risen more at the top of the distribution than elsewhere_**\n\n### _Evolution of average markups in different quantiles of the markup distribution. Data from Annual Respondents Database X (ARDx) 1997-2020 and Annual Business Survey (ABS) 2021. GB only_\n\n### 2\\.41 We first show how key firm characteristics like age, size, profit shares and labour shares are related to markups across the economy. The profit share measures what fraction of a firm’s sales are retained as profits, and the labour share measures the fraction of sales that are used to meet employment costs.\n\n### 2\\.42 As markups represent the profit on each additional sale, we expect markups to be positively correlated with the profit share. The closeness of this relationship depends on fixed costs and the returns to scale of the production technology. For instance, if firms incur large fixed costs such as marketing overheads, they might have large markups but a small profit share.\n\n### 2\\.43 As for the labour share, recent work suggests higher concentration in output markets is associated with lower labour shares (e.g., Autor, Dorn, Katz, Patterson, Van Reenen, 2020; De Loecker, Eeckhout and Unger, 2020; Deb, Eeckhout, Patel and Warren, 2023). Higher markups require firms to produce less than in a competitive market, so they hire less labour. Therefore, absent","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"26\n\n### **_Figure 10: Markups have risen more at the top of the distribution than elsewhere_**\n\n### _Evolution of average markups in different quantiles of the markup distribution. Data from Annual Respondents Database X (ARDx) 1997-2020 and Annual Business Survey (ABS) 2021. GB only_\n\n### 2\\.41 We first show how key firm characteristics like age, size, profit shares and labour shares are related to markups across the economy. The profit share measures what fraction of a firm’s sales are retained as profits, and the labour share measures the fraction of sales that are used to meet employment costs.\n\n### 2\\.42 As markups represent the profit on each additional sale, we expect markups to be positively correlated with the profit share. The closeness of this relationship depends on fixed costs and the returns to scale of the production technology. For instance, if firms incur large fixed costs such as marketing overheads, they might have large markups but a small profit share.\n\n### 2\\.43 As for the labour share, recent work suggests higher concentration in output markets is associated with lower labour shares (e.g., Autor, Dorn, Katz, Patterson, Van Reenen, 2020; De Loecker, Eeckhout and Unger, 2020; Deb, Eeckhout, Patel and Warren, 2023). Higher markups require firms to produce less than in a competitive market, so they hire less labour. Therefore, absent","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/The_State_of_UK_Competition_Report_2024_p28.pdf"]},"response":{"errors":[],"extract_id":"extract_b3a2e04dc9ee9169b53802734d7bc8cc","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["26\n\n### **_Figure 10: Markups have risen more at the top of the distribution than elsewhere_**\n\n### _Evolution of average markups in different quantiles of the markup distribution. Data from Annual Respondents Database X (ARDx) 1997-2020 and Annual Business Survey (ABS) 2021. GB only_\n\n### 2\\.41 We first show how key firm characteristics like age, size, profit shares and labour shares are related to markups across the economy. The profit share measures what fraction of a firm’s sales are retained as profits, and the labour share measures the fraction of sales that are used to meet employment costs.\n\n### 2\\.42 As markups represent the profit on each additional sale, we expect markups to be positively correlated with the profit share. The closeness of this relationship depends on fixed costs and the returns to scale of the production technology. For instance, if firms incur large fixed costs such as marketing overheads, they might have large markups but a small profit share.\n\n### 2\\.43 As for the labour share, recent work suggests higher concentration in output markets is associated with lower labour shares (e.g., Autor, Dorn, Katz, Patterson, Van Reenen, 2020; De Loecker, Eeckhout and Unger, 2020; Deb, Eeckhout, Patel and Warren, 2023). Higher markups require firms to produce less than in a competitive market, so they hire less labour. Therefore, absent"],"full_content":"26\n\n### **_Figure 10: Markups have risen more at the top of the distribution than elsewhere_**\n\n### _Evolution of average markups in different quantiles of the markup distribution. Data from Annual Respondents Database X (ARDx) 1997-2020 and Annual Business Survey (ABS) 2021. GB only_\n\n### 2\\.41 We first show how key firm characteristics like age, size, profit shares and labour shares are related to markups across the economy. The profit share measures what fraction of a firm’s sales are retained as profits, and the labour share measures the fraction of sales that are used to meet employment costs.\n\n### 2\\.42 As markups represent the profit on each additional sale, we expect markups to be positively correlated with the profit share. The closeness of this relationship depends on fixed costs and the returns to scale of the production technology. For instance, if firms incur large fixed costs such as marketing overheads, they might have large markups but a small profit share.\n\n### 2\\.43 As for the labour share, recent work suggests higher concentration in output markets is associated with lower labour shares (e.g., Autor, Dorn, Katz, Patterson, Van Reenen, 2020; De Loecker, Eeckhout and Unger, 2020; Deb, Eeckhout, Patel and Warren, 2023). Higher markups require firms to produce less than in a competitive market, so they hire less labour. Therefore, absent","publish_date":null,"title":"Figure 10: Markups have risen more at the top of the distribution than elsewhere Evolution of average markups in different","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/The_State_of_UK_Competition_Report_2024_p28.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/The_State_of_UK_Competition_Report_2024_p28.pdf"},"request":{"example_id":"chart/The_State_of_UK_Competition_Report_2024_p28","product_type":"parse","source_file_path":"The_State_of_UK_Competition_Report_2024_p28.pdf"},"started_at":"2026-11-04T19:45:16.021545"} +{"completed_at":"2026-11-04T19:45:18.955515","latency_in_ms":2033,"output":{"example_id":"chart/The_State_of_UK_Competition_Report_2024_p31","grounded_pages":[],"layout_pages":[],"markdown":"29\n\n### However, our primary data source does not contain balance sheet information that would allow us to replicate this analysis.\n\n### 2\\.49 We have therefore conducted supplementary analysis using Bureau van Dijk’s FAME dataset and the methodology that we used in the CMA’s previous State of Competition report (2022).\n\n### 2\\.50 While our baseline methodological approach is more robust, Figure E.1 in the appendix shows that overall markup trends using this alternative data source and methodology are similar to our preferred baseline. Figure 13 below also shows that the relationship between markups and the labour share (which we can also calculate using FAME data) is comparable to what we find in our baseline data.\n\n### **_Figure 13: Markups are consistently positively related to profits, overall debt, and the labour share, and negatively related to short-term liquidity_**\n\n### _Regression coefficients of markups on Berry ratio, debt ratio, liquidity ratio and labour share under different model specifications. Controls include year fixed effects, firm fixed effects, and observations weighted by turnover. Data from Bureau van Dijk’s FAME 2013-2023. UK_","pages":[{"line_numbers":[],"markdown":"29\n\n### However, our primary data source does not contain balance sheet information that would allow us to replicate this analysis.\n\n### 2\\.49 We have therefore conducted supplementary analysis using Bureau van Dijk’s FAME dataset and the methodology that we used in the CMA’s previous State of Competition report (2022).\n\n### 2\\.50 While our baseline methodological approach is more robust, Figure E.1 in the appendix shows that overall markup trends using this alternative data source and methodology are similar to our preferred baseline. Figure 13 below also shows that the relationship between markups and the labour share (which we can also calculate using FAME data) is comparable to what we find in our baseline data.\n\n### **_Figure 13: Markups are consistently positively related to profits, overall debt, and the labour share, and negatively related to short-term liquidity_**\n\n### _Regression coefficients of markups on Berry ratio, debt ratio, liquidity ratio and labour share under different model specifications. Controls include year fixed effects, firm fixed effects, and observations weighted by turnover. Data from Bureau van Dijk’s FAME 2013-2023. UK_","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"29\n\n### However, our primary data source does not contain balance sheet information that would allow us to replicate this analysis.\n\n### 2\\.49 We have therefore conducted supplementary analysis using Bureau van Dijk’s FAME dataset and the methodology that we used in the CMA’s previous State of Competition report (2022).\n\n### 2\\.50 While our baseline methodological approach is more robust, Figure E.1 in the appendix shows that overall markup trends using this alternative data source and methodology are similar to our preferred baseline. Figure 13 below also shows that the relationship between markups and the labour share (which we can also calculate using FAME data) is comparable to what we find in our baseline data.\n\n### **_Figure 13: Markups are consistently positively related to profits, overall debt, and the labour share, and negatively related to short-term liquidity_**\n\n### _Regression coefficients of markups on Berry ratio, debt ratio, liquidity ratio and labour share under different model specifications. Controls include year fixed effects, firm fixed effects, and observations weighted by turnover. Data from Bureau van Dijk’s FAME 2013-2023. UK_","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/The_State_of_UK_Competition_Report_2024_p31.pdf"]},"response":{"errors":[],"extract_id":"extract_bd60c93c0c2252eb20d7d0582b142eb2","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["29\n\n### However, our primary data source does not contain balance sheet information that would allow us to replicate this analysis.\n\n### 2\\.49 We have therefore conducted supplementary analysis using Bureau van Dijk’s FAME dataset and the methodology that we used in the CMA’s previous State of Competition report (2022).\n\n### 2\\.50 While our baseline methodological approach is more robust, Figure E.1 in the appendix shows that overall markup trends using this alternative data source and methodology are similar to our preferred baseline. Figure 13 below also shows that the relationship between markups and the labour share (which we can also calculate using FAME data) is comparable to what we find in our baseline data.\n\n### **_Figure 13: Markups are consistently positively related to profits, overall debt, and the labour share, and negatively related to short-term liquidity_**\n\n### _Regression coefficients of markups on Berry ratio, debt ratio, liquidity ratio and labour share under different model specifications. Controls include year fixed effects, firm fixed effects, and observations weighted by turnover. Data from Bureau van Dijk’s FAME 2013-2023. UK_"],"full_content":"29\n\n### However, our primary data source does not contain balance sheet information that would allow us to replicate this analysis.\n\n### 2\\.49 We have therefore conducted supplementary analysis using Bureau van Dijk’s FAME dataset and the methodology that we used in the CMA’s previous State of Competition report (2022).\n\n### 2\\.50 While our baseline methodological approach is more robust, Figure E.1 in the appendix shows that overall markup trends using this alternative data source and methodology are similar to our preferred baseline. Figure 13 below also shows that the relationship between markups and the labour share (which we can also calculate using FAME data) is comparable to what we find in our baseline data.\n\n### **_Figure 13: Markups are consistently positively related to profits, overall debt, and the labour share, and negatively related to short-term liquidity_**\n\n### _Regression coefficients of markups on Berry ratio, debt ratio, liquidity ratio and labour share under different model specifications. Controls include year fixed effects, firm fixed effects, and observations weighted by turnover. Data from Bureau van Dijk’s FAME 2013-2023. UK_","publish_date":null,"title":"However, our primary data source does not contain balance sheet information that would allow us to replicate this analysis. 2.49","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/The_State_of_UK_Competition_Report_2024_p31.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/The_State_of_UK_Competition_Report_2024_p31.pdf"},"request":{"example_id":"chart/The_State_of_UK_Competition_Report_2024_p31","product_type":"parse","source_file_path":"The_State_of_UK_Competition_Report_2024_p31.pdf"},"started_at":"2026-11-04T19:45:16.922077"} +{"completed_at":"2026-11-04T19:45:21.073560","latency_in_ms":2237,"output":{"example_id":"chart/The_State_of_UK_Competition_Report_2024_p34","grounded_pages":[],"layout_pages":[],"markdown":"32\n\n### measure is called the Herfindahl-Hirschman Index (or HHI for short). It ranges from zero for a perfectly competitive market to 10,000 for a monopoly.\n\n### 3\\.10 Competition agencies sometimes define markets with an HHI of above 1,500 as moderately concentrated, and those with an HHI of above 2,500 as highly concentrated.\n\n### **Concentration has remained relatively stable on average**\n\n### 3\\.11 Figure 14 plots the average concentration ratio for the five and ten largest firms in each four-digit Standard Industrial Classification (SIC) industry between 1997 and 2022.\n\n### **_Figure 14: Mean CR5 and CR10 have remained roughly stable after the Great Financial Crisis_**\n\n### _Mean concentration ratios (CR5 and CR10) computed at the four-digit Standard Industial Classification (SIC) level and aggregated by weighting for industry turnover. Data from Business Structure Database (BSD) 1997 – 2022. UK_\n\n### 3\\.12 The overall change in average concentration ratios over the past twenty-five years has been small. The ratios declined sharply in the early 2000s then peaked around 2010. Both the CR5 and the CR10 have stabilised in more","pages":[{"line_numbers":[],"markdown":"32\n\n### measure is called the Herfindahl-Hirschman Index (or HHI for short). It ranges from zero for a perfectly competitive market to 10,000 for a monopoly.\n\n### 3\\.10 Competition agencies sometimes define markets with an HHI of above 1,500 as moderately concentrated, and those with an HHI of above 2,500 as highly concentrated.\n\n### **Concentration has remained relatively stable on average**\n\n### 3\\.11 Figure 14 plots the average concentration ratio for the five and ten largest firms in each four-digit Standard Industrial Classification (SIC) industry between 1997 and 2022.\n\n### **_Figure 14: Mean CR5 and CR10 have remained roughly stable after the Great Financial Crisis_**\n\n### _Mean concentration ratios (CR5 and CR10) computed at the four-digit Standard Industial Classification (SIC) level and aggregated by weighting for industry turnover. Data from Business Structure Database (BSD) 1997 – 2022. UK_\n\n### 3\\.12 The overall change in average concentration ratios over the past twenty-five years has been small. The ratios declined sharply in the early 2000s then peaked around 2010. Both the CR5 and the CR10 have stabilised in more","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"32\n\n### measure is called the Herfindahl-Hirschman Index (or HHI for short). It ranges from zero for a perfectly competitive market to 10,000 for a monopoly.\n\n### 3\\.10 Competition agencies sometimes define markets with an HHI of above 1,500 as moderately concentrated, and those with an HHI of above 2,500 as highly concentrated.\n\n### **Concentration has remained relatively stable on average**\n\n### 3\\.11 Figure 14 plots the average concentration ratio for the five and ten largest firms in each four-digit Standard Industrial Classification (SIC) industry between 1997 and 2022.\n\n### **_Figure 14: Mean CR5 and CR10 have remained roughly stable after the Great Financial Crisis_**\n\n### _Mean concentration ratios (CR5 and CR10) computed at the four-digit Standard Industial Classification (SIC) level and aggregated by weighting for industry turnover. Data from Business Structure Database (BSD) 1997 – 2022. UK_\n\n### 3\\.12 The overall change in average concentration ratios over the past twenty-five years has been small. The ratios declined sharply in the early 2000s then peaked around 2010. Both the CR5 and the CR10 have stabilised in more","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/The_State_of_UK_Competition_Report_2024_p34.pdf"]},"response":{"errors":[],"extract_id":"extract_9a026cc6db59f589a4809a44191e70a5","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["32\n\n### measure is called the Herfindahl-Hirschman Index (or HHI for short). It ranges from zero for a perfectly competitive market to 10,000 for a monopoly.\n\n### 3\\.10 Competition agencies sometimes define markets with an HHI of above 1,500 as moderately concentrated, and those with an HHI of above 2,500 as highly concentrated.\n\n### **Concentration has remained relatively stable on average**\n\n### 3\\.11 Figure 14 plots the average concentration ratio for the five and ten largest firms in each four-digit Standard Industrial Classification (SIC) industry between 1997 and 2022.\n\n### **_Figure 14: Mean CR5 and CR10 have remained roughly stable after the Great Financial Crisis_**\n\n### _Mean concentration ratios (CR5 and CR10) computed at the four-digit Standard Industial Classification (SIC) level and aggregated by weighting for industry turnover. Data from Business Structure Database (BSD) 1997 – 2022. UK_\n\n### 3\\.12 The overall change in average concentration ratios over the past twenty-five years has been small. The ratios declined sharply in the early 2000s then peaked around 2010. Both the CR5 and the CR10 have stabilised in more"],"full_content":"32\n\n### measure is called the Herfindahl-Hirschman Index (or HHI for short). It ranges from zero for a perfectly competitive market to 10,000 for a monopoly.\n\n### 3\\.10 Competition agencies sometimes define markets with an HHI of above 1,500 as moderately concentrated, and those with an HHI of above 2,500 as highly concentrated.\n\n### **Concentration has remained relatively stable on average**\n\n### 3\\.11 Figure 14 plots the average concentration ratio for the five and ten largest firms in each four-digit Standard Industrial Classification (SIC) industry between 1997 and 2022.\n\n### **_Figure 14: Mean CR5 and CR10 have remained roughly stable after the Great Financial Crisis_**\n\n### _Mean concentration ratios (CR5 and CR10) computed at the four-digit Standard Industial Classification (SIC) level and aggregated by weighting for industry turnover. Data from Business Structure Database (BSD) 1997 – 2022. UK_\n\n### 3\\.12 The overall change in average concentration ratios over the past twenty-five years has been small. The ratios declined sharply in the early 2000s then peaked around 2010. Both the CR5 and the CR10 have stabilised in more","publish_date":null,"title":"Concentration has remained relatively stable on average","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/The_State_of_UK_Competition_Report_2024_p34.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/The_State_of_UK_Competition_Report_2024_p34.pdf"},"request":{"example_id":"chart/The_State_of_UK_Competition_Report_2024_p34","product_type":"parse","source_file_path":"The_State_of_UK_Competition_Report_2024_p34.pdf"},"started_at":"2026-11-04T19:45:18.836200"} +{"completed_at":"2026-11-04T19:45:21.052332","latency_in_ms":1403,"output":{"example_id":"chart/Treasury_Bulletin_2025_06_p13","grounded_pages":[],"layout_pages":[],"markdown":"7\n\nthose consistent with a 2.0 percent target rate for total inflation, they are well down from the 0.7 percent to 0.8 percent monthly rates seen in the latter half of 2022. Over the year through April 2025, rent of housing inflation was 4.2 percent, the slowest pace since January 2022. Inflation for non-housing core services has been an integral driver of elevated core inflation. Significantly, inflation in this metric slowed to 0.2 percent during the first quarter of 2025, from an average 0.3 percent during the latter half of 2024. The slowing pace of core non-housing services inflation largely reflected an easing of prices in several categories, including transportation services and hotel lodgings. From 2017 to 2019, annual non-housing services inflation averaged 2.1 percent, versus 4.0 percent over the year ending in December 2024. As of April 2025, core non-housing services inflation was 0.2 percent, and over the year through April, was 2.8 percent, the slowest twelve-month pace since March 2021. Inflation as measured by the PCE price index is the Federal Reserve’s preferred measure of inflation: the FOMC’s 2 percent inflation target is expressed in terms of headline PCE, and core PCE inflation is also taken into consideration. There are notable differences in weights and methodologies between the CPI and the PCE measures. Historically, twelve-month CPI inflation has exceeded PCE inflation by about 0.4 percentage points on average. During 2024, however, the wedge widened, averaging 0.6 percentage points in the year’s first half. As core services price growth slowed, the wedge narrowed noticeably, averaging 0.3 percentage points during the latter half. Headline PCE inflation slowed during 2024 and was 2.1 percent over the twelve months through September—or 0.1 percentage point above the FOMC inflation target—while core PCE inflation was 2.7 percent over that same period. By February 2025, headline PCE inflation had accelerated to 0.7 percentage points above the target. Over the year through March 2025 (latest data available), headline PCE was 2.3 percent—0.3 percentage points above the Fed target percent—and the core reading was 2.6 percent.\n\n**Consumer Prices**\n\n(Percent change from a year earlier)","pages":[{"line_numbers":[],"markdown":"7\n\nthose consistent with a 2.0 percent target rate for total inflation, they are well down from the 0.7 percent to 0.8 percent monthly rates seen in the latter half of 2022. Over the year through April 2025, rent of housing inflation was 4.2 percent, the slowest pace since January 2022. Inflation for non-housing core services has been an integral driver of elevated core inflation. Significantly, inflation in this metric slowed to 0.2 percent during the first quarter of 2025, from an average 0.3 percent during the latter half of 2024. The slowing pace of core non-housing services inflation largely reflected an easing of prices in several categories, including transportation services and hotel lodgings. From 2017 to 2019, annual non-housing services inflation averaged 2.1 percent, versus 4.0 percent over the year ending in December 2024. As of April 2025, core non-housing services inflation was 0.2 percent, and over the year through April, was 2.8 percent, the slowest twelve-month pace since March 2021. Inflation as measured by the PCE price index is the Federal Reserve’s preferred measure of inflation: the FOMC’s 2 percent inflation target is expressed in terms of headline PCE, and core PCE inflation is also taken into consideration. There are notable differences in weights and methodologies between the CPI and the PCE measures. Historically, twelve-month CPI inflation has exceeded PCE inflation by about 0.4 percentage points on average. During 2024, however, the wedge widened, averaging 0.6 percentage points in the year’s first half. As core services price growth slowed, the wedge narrowed noticeably, averaging 0.3 percentage points during the latter half. Headline PCE inflation slowed during 2024 and was 2.1 percent over the twelve months through September—or 0.1 percentage point above the FOMC inflation target—while core PCE inflation was 2.7 percent over that same period. By February 2025, headline PCE inflation had accelerated to 0.7 percentage points above the target. Over the year through March 2025 (latest data available), headline PCE was 2.3 percent—0.3 percentage points above the Fed target percent—and the core reading was 2.6 percent.\n\n**Consumer Prices**\n\n(Percent change from a year earlier)","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"7\n\nthose consistent with a 2.0 percent target rate for total inflation, they are well down from the 0.7 percent to 0.8 percent monthly rates seen in the latter half of 2022. Over the year through April 2025, rent of housing inflation was 4.2 percent, the slowest pace since January 2022. Inflation for non-housing core services has been an integral driver of elevated core inflation. Significantly, inflation in this metric slowed to 0.2 percent during the first quarter of 2025, from an average 0.3 percent during the latter half of 2024. The slowing pace of core non-housing services inflation largely reflected an easing of prices in several categories, including transportation services and hotel lodgings. From 2017 to 2019, annual non-housing services inflation averaged 2.1 percent, versus 4.0 percent over the year ending in December 2024. As of April 2025, core non-housing services inflation was 0.2 percent, and over the year through April, was 2.8 percent, the slowest twelve-month pace since March 2021. Inflation as measured by the PCE price index is the Federal Reserve’s preferred measure of inflation: the FOMC’s 2 percent inflation target is expressed in terms of headline PCE, and core PCE inflation is also taken into consideration. There are notable differences in weights and methodologies between the CPI and the PCE measures. Historically, twelve-month CPI inflation has exceeded PCE inflation by about 0.4 percentage points on average. During 2024, however, the wedge widened, averaging 0.6 percentage points in the year’s first half. As core services price growth slowed, the wedge narrowed noticeably, averaging 0.3 percentage points during the latter half. Headline PCE inflation slowed during 2024 and was 2.1 percent over the twelve months through September—or 0.1 percentage point above the FOMC inflation target—while core PCE inflation was 2.7 percent over that same period. By February 2025, headline PCE inflation had accelerated to 0.7 percentage points above the target. Over the year through March 2025 (latest data available), headline PCE was 2.3 percent—0.3 percentage points above the Fed target percent—and the core reading was 2.6 percent.\n\n**Consumer Prices**\n\n(Percent change from a year earlier)","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Treasury_Bulletin_2025_06_p13.pdf"]},"response":{"errors":[],"extract_id":"extract_b2e9e0f7dddcab8c8171a88ab14ca89f","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["7\n\nthose consistent with a 2.0 percent target rate for total inflation, they are well down from the 0.7 percent to 0.8 percent monthly rates seen in the latter half of 2022. Over the year through April 2025, rent of housing inflation was 4.2 percent, the slowest pace since January 2022. Inflation for non-housing core services has been an integral driver of elevated core inflation. Significantly, inflation in this metric slowed to 0.2 percent during the first quarter of 2025, from an average 0.3 percent during the latter half of 2024. The slowing pace of core non-housing services inflation largely reflected an easing of prices in several categories, including transportation services and hotel lodgings. From 2017 to 2019, annual non-housing services inflation averaged 2.1 percent, versus 4.0 percent over the year ending in December 2024. As of April 2025, core non-housing services inflation was 0.2 percent, and over the year through April, was 2.8 percent, the slowest twelve-month pace since March 2021. Inflation as measured by the PCE price index is the Federal Reserve’s preferred measure of inflation: the FOMC’s 2 percent inflation target is expressed in terms of headline PCE, and core PCE inflation is also taken into consideration. There are notable differences in weights and methodologies between the CPI and the PCE measures. Historically, twelve-month CPI inflation has exceeded PCE inflation by about 0.4 percentage points on average. During 2024, however, the wedge widened, averaging 0.6 percentage points in the year’s first half. As core services price growth slowed, the wedge narrowed noticeably, averaging 0.3 percentage points during the latter half. Headline PCE inflation slowed during 2024 and was 2.1 percent over the twelve months through September—or 0.1 percentage point above the FOMC inflation target—while core PCE inflation was 2.7 percent over that same period. By February 2025, headline PCE inflation had accelerated to 0.7 percentage points above the target. Over the year through March 2025 (latest data available), headline PCE was 2.3 percent—0.3 percentage points above the Fed target percent—and the core reading was 2.6 percent.\n\n**Consumer Prices**\n\n(Percent change from a year earlier)"],"full_content":"7\n\nthose consistent with a 2.0 percent target rate for total inflation, they are well down from the 0.7 percent to 0.8 percent monthly rates seen in the latter half of 2022. Over the year through April 2025, rent of housing inflation was 4.2 percent, the slowest pace since January 2022. Inflation for non-housing core services has been an integral driver of elevated core inflation. Significantly, inflation in this metric slowed to 0.2 percent during the first quarter of 2025, from an average 0.3 percent during the latter half of 2024. The slowing pace of core non-housing services inflation largely reflected an easing of prices in several categories, including transportation services and hotel lodgings. From 2017 to 2019, annual non-housing services inflation averaged 2.1 percent, versus 4.0 percent over the year ending in December 2024. As of April 2025, core non-housing services inflation was 0.2 percent, and over the year through April, was 2.8 percent, the slowest twelve-month pace since March 2021. Inflation as measured by the PCE price index is the Federal Reserve’s preferred measure of inflation: the FOMC’s 2 percent inflation target is expressed in terms of headline PCE, and core PCE inflation is also taken into consideration. There are notable differences in weights and methodologies between the CPI and the PCE measures. Historically, twelve-month CPI inflation has exceeded PCE inflation by about 0.4 percentage points on average. During 2024, however, the wedge widened, averaging 0.6 percentage points in the year’s first half. As core services price growth slowed, the wedge narrowed noticeably, averaging 0.3 percentage points during the latter half. Headline PCE inflation slowed during 2024 and was 2.1 percent over the twelve months through September—or 0.1 percentage point above the FOMC inflation target—while core PCE inflation was 2.7 percent over that same period. By February 2025, headline PCE inflation had accelerated to 0.7 percentage points above the target. Over the year through March 2025 (latest data available), headline PCE was 2.3 percent—0.3 percentage points above the Fed target percent—and the core reading was 2.6 percent.\n\n**Consumer Prices**\n\n(Percent change from a year earlier)","publish_date":null,"title":"those consistent with a 2.0 percent target rate for total inflation, they are well down from the 0.7 percent to","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Treasury_Bulletin_2025_06_p13.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Treasury_Bulletin_2025_06_p13.pdf"},"request":{"example_id":"chart/Treasury_Bulletin_2025_06_p13","product_type":"parse","source_file_path":"Treasury_Bulletin_2025_06_p13.pdf"},"started_at":"2026-11-04T19:45:19.648763"} +{"completed_at":"2026-11-04T19:45:22.834245","latency_in_ms":2603,"output":{"example_id":"chart/Treasury_Bulletin_2025_06_p21","grounded_pages":[],"layout_pages":[],"markdown":"**1** **4**","pages":[{"line_numbers":[],"markdown":"**1** **4**","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**1** **4**","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Treasury_Bulletin_2025_06_p21.pdf"]},"response":{"errors":[],"extract_id":"extract_d11ec351f24aaf1955856da1749b49e0","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**1** **4**"],"full_content":"**1** **4**","publish_date":null,"title":"4 1","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Treasury_Bulletin_2025_06_p21.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Treasury_Bulletin_2025_06_p21.pdf"},"request":{"example_id":"chart/Treasury_Bulletin_2025_06_p21","product_type":"parse","source_file_path":"Treasury_Bulletin_2025_06_p21.pdf"},"started_at":"2026-11-04T19:45:20.230836"} +{"completed_at":"2026-11-04T19:45:23.165415","latency_in_ms":1046,"output":{"example_id":"chart/Treasury_Bulletin_2025_06_p9","grounded_pages":[],"layout_pages":[],"markdown":"3\n\n### **Growth of Real GDP**\n\n(Quarterly percent change at annual rate)\n\n### **Labor Markets and Wages**\n\n### Labor supply and demand in the economy largely returned to normal balance by the end of 2024 and have remained stable this year. Payroll job growth has gradually slowed from the rapid rates that dominated 2021 and 2022—monthly average gains of roughly 600,000 and 380,000 jobs, respectively—to an average monthly gain of 209,000 jobs in the fourth quarter of 2024, and then to 133,000 per month on average in this year’s first quarter. The latest labor report for April 2025 showed payrolls increased by 177,000 over the month, in line with estimates of the break- even rates needed to maintain a steady unemployment rate given population growth.\n\n### The unemployment rate has been remarkably stable for the past several months: since May 2024, it has hovered in the narrow range of 4.0 and 4.2 percent and stood at 4.2 percent as of April 2025. A broader measure, which captures underemployment of the workforce, declined during the latter half of 2024 to end the year at 7.5 percent, where it remained in January 2025. Since then, however, it has drifted up to a high of 8.0 percent in February, before declining to 7.8 percent in April, still 0.3 percentage points above the level at the end of last year. Even so, both the unemployment and underemployment rates are still relatively low by historical standards. More timely data also suggest that unemployment remains low. As of the week ending May 10, the level of initial unemployment insurance claims had increased only slightly from the end of December 2024, and as of the week ending May 3, the level of continuing unemployment claims was little changed over this timeframe. These readings are only","pages":[{"line_numbers":[],"markdown":"3\n\n### **Growth of Real GDP**\n\n(Quarterly percent change at annual rate)\n\n### **Labor Markets and Wages**\n\n### Labor supply and demand in the economy largely returned to normal balance by the end of 2024 and have remained stable this year. Payroll job growth has gradually slowed from the rapid rates that dominated 2021 and 2022—monthly average gains of roughly 600,000 and 380,000 jobs, respectively—to an average monthly gain of 209,000 jobs in the fourth quarter of 2024, and then to 133,000 per month on average in this year’s first quarter. The latest labor report for April 2025 showed payrolls increased by 177,000 over the month, in line with estimates of the break- even rates needed to maintain a steady unemployment rate given population growth.\n\n### The unemployment rate has been remarkably stable for the past several months: since May 2024, it has hovered in the narrow range of 4.0 and 4.2 percent and stood at 4.2 percent as of April 2025. A broader measure, which captures underemployment of the workforce, declined during the latter half of 2024 to end the year at 7.5 percent, where it remained in January 2025. Since then, however, it has drifted up to a high of 8.0 percent in February, before declining to 7.8 percent in April, still 0.3 percentage points above the level at the end of last year. Even so, both the unemployment and underemployment rates are still relatively low by historical standards. More timely data also suggest that unemployment remains low. As of the week ending May 10, the level of initial unemployment insurance claims had increased only slightly from the end of December 2024, and as of the week ending May 3, the level of continuing unemployment claims was little changed over this timeframe. These readings are only","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"3\n\n### **Growth of Real GDP**\n\n(Quarterly percent change at annual rate)\n\n### **Labor Markets and Wages**\n\n### Labor supply and demand in the economy largely returned to normal balance by the end of 2024 and have remained stable this year. Payroll job growth has gradually slowed from the rapid rates that dominated 2021 and 2022—monthly average gains of roughly 600,000 and 380,000 jobs, respectively—to an average monthly gain of 209,000 jobs in the fourth quarter of 2024, and then to 133,000 per month on average in this year’s first quarter. The latest labor report for April 2025 showed payrolls increased by 177,000 over the month, in line with estimates of the break- even rates needed to maintain a steady unemployment rate given population growth.\n\n### The unemployment rate has been remarkably stable for the past several months: since May 2024, it has hovered in the narrow range of 4.0 and 4.2 percent and stood at 4.2 percent as of April 2025. A broader measure, which captures underemployment of the workforce, declined during the latter half of 2024 to end the year at 7.5 percent, where it remained in January 2025. Since then, however, it has drifted up to a high of 8.0 percent in February, before declining to 7.8 percent in April, still 0.3 percentage points above the level at the end of last year. Even so, both the unemployment and underemployment rates are still relatively low by historical standards. More timely data also suggest that unemployment remains low. As of the week ending May 10, the level of initial unemployment insurance claims had increased only slightly from the end of December 2024, and as of the week ending May 3, the level of continuing unemployment claims was little changed over this timeframe. These readings are only","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Treasury_Bulletin_2025_06_p9.pdf"]},"response":{"errors":[],"extract_id":"extract_fdc45a49e0090112582ee03f2a7a2957","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["3\n\n### **Growth of Real GDP**\n\n(Quarterly percent change at annual rate)\n\n### **Labor Markets and Wages**\n\n### Labor supply and demand in the economy largely returned to normal balance by the end of 2024 and have remained stable this year. Payroll job growth has gradually slowed from the rapid rates that dominated 2021 and 2022—monthly average gains of roughly 600,000 and 380,000 jobs, respectively—to an average monthly gain of 209,000 jobs in the fourth quarter of 2024, and then to 133,000 per month on average in this year’s first quarter. The latest labor report for April 2025 showed payrolls increased by 177,000 over the month, in line with estimates of the break- even rates needed to maintain a steady unemployment rate given population growth.\n\n### The unemployment rate has been remarkably stable for the past several months: since May 2024, it has hovered in the narrow range of 4.0 and 4.2 percent and stood at 4.2 percent as of April 2025. A broader measure, which captures underemployment of the workforce, declined during the latter half of 2024 to end the year at 7.5 percent, where it remained in January 2025. Since then, however, it has drifted up to a high of 8.0 percent in February, before declining to 7.8 percent in April, still 0.3 percentage points above the level at the end of last year. Even so, both the unemployment and underemployment rates are still relatively low by historical standards. More timely data also suggest that unemployment remains low. As of the week ending May 10, the level of initial unemployment insurance claims had increased only slightly from the end of December 2024, and as of the week ending May 3, the level of continuing unemployment claims was little changed over this timeframe. These readings are only"],"full_content":"3\n\n### **Growth of Real GDP**\n\n(Quarterly percent change at annual rate)\n\n### **Labor Markets and Wages**\n\n### Labor supply and demand in the economy largely returned to normal balance by the end of 2024 and have remained stable this year. Payroll job growth has gradually slowed from the rapid rates that dominated 2021 and 2022—monthly average gains of roughly 600,000 and 380,000 jobs, respectively—to an average monthly gain of 209,000 jobs in the fourth quarter of 2024, and then to 133,000 per month on average in this year’s first quarter. The latest labor report for April 2025 showed payrolls increased by 177,000 over the month, in line with estimates of the break- even rates needed to maintain a steady unemployment rate given population growth.\n\n### The unemployment rate has been remarkably stable for the past several months: since May 2024, it has hovered in the narrow range of 4.0 and 4.2 percent and stood at 4.2 percent as of April 2025. A broader measure, which captures underemployment of the workforce, declined during the latter half of 2024 to end the year at 7.5 percent, where it remained in January 2025. Since then, however, it has drifted up to a high of 8.0 percent in February, before declining to 7.8 percent in April, still 0.3 percentage points above the level at the end of last year. Even so, both the unemployment and underemployment rates are still relatively low by historical standards. More timely data also suggest that unemployment remains low. As of the week ending May 10, the level of initial unemployment insurance claims had increased only slightly from the end of December 2024, and as of the week ending May 3, the level of continuing unemployment claims was little changed over this timeframe. These readings are only","publish_date":null,"title":"Growth of Real GDP Labor Markets and Wages Labor supply and demand in the economy largely returned to normal balance","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Treasury_Bulletin_2025_06_p9.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Treasury_Bulletin_2025_06_p9.pdf"},"request":{"example_id":"chart/Treasury_Bulletin_2025_06_p9","product_type":"parse","source_file_path":"Treasury_Bulletin_2025_06_p9.pdf"},"started_at":"2026-11-04T19:45:22.118789"} +{"completed_at":"2026-11-04T19:45:25.500551","latency_in_ms":2002,"output":{"example_id":"chart/US50419123-White-Paper-Signature-Sept-2025_p16","grounded_pages":[],"layout_pages":[],"markdown":"**16** White Paper, sponsored by Google Cloud April 2023 | IDC #US50419123\n\nThe Digital Customer Journey for Manufacturers\n\nTable of Contents\n\n■ **Very limited** ■ **Early stages** ■ **Preconfigured** ■ **End to end** ■ **Orchestrated**\n\n10% 40% 20% 30% 60% 50%\n\n**Breaking down organizational silos** **. . . . . . . . .**\n\n**Integrating offline data/CRM** **. . . . . . . . . . . . . . . . .**\n\n**Accessing partner/channel and ecosystem data** **. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .**\n\n**Offering digitally centric products and services** **. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .**\n\n**Addressing cybersecurity concerns** **. . . . . . . . .**\n\n**Upskilling/reskilling teams** **. . . . . . . . . . . . . . . . . . .**\n\n**Using data to create or adopt new business models** **. . . . . . . . . . . . . . . . . . . . . . . . .**\n\n**Using real-time data and feedback to help drive more business** **. . . . . . . . . . . . . . . . . .**\n\n**FIGURE 4**\n\n### **First-Party Data/Integration Priorities by Maturity Level**\n\n**What are your top priorities around data and integration to drive better manufacturing outcomes?** (% of respondents)\n\nn = 1,514; Source: IDC’s _Manufacturing Customer Experience Survey_ , September 2022 For an accessible version of the data in this figure, see **Figure 4 Supplemental Data** in Appendix 2.","pages":[{"line_numbers":[],"markdown":"**16** White Paper, sponsored by Google Cloud April 2023 | IDC #US50419123\n\nThe Digital Customer Journey for Manufacturers\n\nTable of Contents\n\n■ **Very limited** ■ **Early stages** ■ **Preconfigured** ■ **End to end** ■ **Orchestrated**\n\n10% 40% 20% 30% 60% 50%\n\n**Breaking down organizational silos** **. . . . . . . . .**\n\n**Integrating offline data/CRM** **. . . . . . . . . . . . . . . . .**\n\n**Accessing partner/channel and ecosystem data** **. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .**\n\n**Offering digitally centric products and services** **. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .**\n\n**Addressing cybersecurity concerns** **. . . . . . . . .**\n\n**Upskilling/reskilling teams** **. . . . . . . . . . . . . . . . . . .**\n\n**Using data to create or adopt new business models** **. . . . . . . . . . . . . . . . . . . . . . . . .**\n\n**Using real-time data and feedback to help drive more business** **. . . . . . . . . . . . . . . . . .**\n\n**FIGURE 4**\n\n### **First-Party Data/Integration Priorities by Maturity Level**\n\n**What are your top priorities around data and integration to drive better manufacturing outcomes?** (% of respondents)\n\nn = 1,514; Source: IDC’s _Manufacturing Customer Experience Survey_ , September 2022 For an accessible version of the data in this figure, see **Figure 4 Supplemental Data** in Appendix 2.","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**16** White Paper, sponsored by Google Cloud April 2023 | IDC #US50419123\n\nThe Digital Customer Journey for Manufacturers\n\nTable of Contents\n\n■ **Very limited** ■ **Early stages** ■ **Preconfigured** ■ **End to end** ■ **Orchestrated**\n\n10% 40% 20% 30% 60% 50%\n\n**Breaking down organizational silos** **. . . . . . . . .**\n\n**Integrating offline data/CRM** **. . . . . . . . . . . . . . . . .**\n\n**Accessing partner/channel and ecosystem data** **. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .**\n\n**Offering digitally centric products and services** **. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .**\n\n**Addressing cybersecurity concerns** **. . . . . . . . .**\n\n**Upskilling/reskilling teams** **. . . . . . . . . . . . . . . . . . .**\n\n**Using data to create or adopt new business models** **. . . . . . . . . . . . . . . . . . . . . . . . .**\n\n**Using real-time data and feedback to help drive more business** **. . . . . . . . . . . . . . . . . .**\n\n**FIGURE 4**\n\n### **First-Party Data/Integration Priorities by Maturity Level**\n\n**What are your top priorities around data and integration to drive better manufacturing outcomes?** (% of respondents)\n\nn = 1,514; Source: IDC’s _Manufacturing Customer Experience Survey_ , September 2022 For an accessible version of the data in this figure, see **Figure 4 Supplemental Data** in Appendix 2.","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/US50419123-White-Paper-Signature-Sept-2025_p16.pdf"]},"response":{"errors":[],"extract_id":"extract_e624fc8c4cbe1d9479a0ca8f74fcefe6","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**16** White Paper, sponsored by Google Cloud April 2023 | IDC #US50419123\n\nThe Digital Customer Journey for Manufacturers\n\nTable of Contents\n\n■ **Very limited** ■ **Early stages** ■ **Preconfigured** ■ **End to end** ■ **Orchestrated**\n\n10% 40% 20% 30% 60% 50%\n\n**Breaking down organizational silos** **. . . . . . . . .**\n\n**Integrating offline data/CRM** **. . . . . . . . . . . . . . . . .**\n\n**Accessing partner/channel and ecosystem data** **. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .**\n\n**Offering digitally centric products and services** **. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .**\n\n**Addressing cybersecurity concerns** **. . . . . . . . .**\n\n**Upskilling/reskilling teams** **. . . . . . . . . . . . . . . . . . .**\n\n**Using data to create or adopt new business models** **. . . . . . . . . . . . . . . . . . . . . . . . .**\n\n**Using real-time data and feedback to help drive more business** **. . . . . . . . . . . . . . . . . .**\n\n**FIGURE 4**\n\n### **First-Party Data/Integration Priorities by Maturity Level**\n\n**What are your top priorities around data and integration to drive better manufacturing outcomes?** (% of respondents)\n\nn = 1,514; Source: IDC’s _Manufacturing Customer Experience Survey_ , September 2022 For an accessible version of the data in this figure, see **Figure 4 Supplemental Data** in Appendix 2."],"full_content":"**16** White Paper, sponsored by Google Cloud April 2023 | IDC #US50419123\n\nThe Digital Customer Journey for Manufacturers\n\nTable of Contents\n\n■ **Very limited** ■ **Early stages** ■ **Preconfigured** ■ **End to end** ■ **Orchestrated**\n\n10% 40% 20% 30% 60% 50%\n\n**Breaking down organizational silos** **. . . . . . . . .**\n\n**Integrating offline data/CRM** **. . . . . . . . . . . . . . . . .**\n\n**Accessing partner/channel and ecosystem data** **. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .**\n\n**Offering digitally centric products and services** **. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .**\n\n**Addressing cybersecurity concerns** **. . . . . . . . .**\n\n**Upskilling/reskilling teams** **. . . . . . . . . . . . . . . . . . .**\n\n**Using data to create or adopt new business models** **. . . . . . . . . . . . . . . . . . . . . . . . .**\n\n**Using real-time data and feedback to help drive more business** **. . . . . . . . . . . . . . . . . .**\n\n**FIGURE 4**\n\n### **First-Party Data/Integration Priorities by Maturity Level**\n\n**What are your top priorities around data and integration to drive better manufacturing outcomes?** (% of respondents)\n\nn = 1,514; Source: IDC’s _Manufacturing Customer Experience Survey_ , September 2022 For an accessible version of the data in this figure, see **Figure 4 Supplemental Data** in Appendix 2.","publish_date":null,"title":"■ ■ ■ ■ ■","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/US50419123-White-Paper-Signature-Sept-2025_p16.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/US50419123-White-Paper-Signature-Sept-2025_p16.pdf"},"request":{"example_id":"chart/US50419123-White-Paper-Signature-Sept-2025_p16","product_type":"parse","source_file_path":"US50419123-White-Paper-Signature-Sept-2025_p16.pdf"},"started_at":"2026-11-04T19:45:23.497933"} +{"completed_at":"2026-11-04T19:29:59.989812","latency_in_ms":1300,"output":{"example_id":"chart/US_Professional_Services_Partner_Compensation_Survey_2024_p11","grounded_pages":[],"layout_pages":[],"markdown":"**Partner turnover (%)**\n\n## Partner turnover\n\nIn response to a slower 2023, wherein people voluntarily or involuntarily chose to leave, respondents reported partner turnover was higher in 2024 compared to 2022: 49% of respondents said partner turnover was higher or significantly higher compared to their expectations at the beginning of the year; only 30% said the same in 2022.\n\n# **Year-over-year hiring trends**\n\nNote: Numbers may not sum to 100%, because of rounding.\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=144\n\nSignificantly lower 5\n\n9\n\nLower 14\n\n13\n\nAbout the same 45\n\n27\n\nSignificantly higher 6\n\n14\n\nPrefer not to answer 6\n\n1\n\nHigher 24\n\n35\n\n**In your most recent fiscal year, how much partner turnover was there at your firm compared to your expectations at the beginning of the fiscal year?**\n\n2024 2022\n\nDiverse region, inclusive workforces: Diversity and inclusion policy and practice in Asia Pacific **HEIDRICK & STRUGGLES**\n\n**11**","pages":[{"line_numbers":[],"markdown":"**Partner turnover (%)**\n\n## Partner turnover\n\nIn response to a slower 2023, wherein people voluntarily or involuntarily chose to leave, respondents reported partner turnover was higher in 2024 compared to 2022: 49% of respondents said partner turnover was higher or significantly higher compared to their expectations at the beginning of the year; only 30% said the same in 2022.\n\n# **Year-over-year hiring trends**\n\nNote: Numbers may not sum to 100%, because of rounding.\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=144\n\nSignificantly lower 5\n\n9\n\nLower 14\n\n13\n\nAbout the same 45\n\n27\n\nSignificantly higher 6\n\n14\n\nPrefer not to answer 6\n\n1\n\nHigher 24\n\n35\n\n**In your most recent fiscal year, how much partner turnover was there at your firm compared to your expectations at the beginning of the fiscal year?**\n\n2024 2022\n\nDiverse region, inclusive workforces: Diversity and inclusion policy and practice in Asia Pacific **HEIDRICK & STRUGGLES**\n\n**11**","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**Partner turnover (%)**\n\n## Partner turnover\n\nIn response to a slower 2023, wherein people voluntarily or involuntarily chose to leave, respondents reported partner turnover was higher in 2024 compared to 2022: 49% of respondents said partner turnover was higher or significantly higher compared to their expectations at the beginning of the year; only 30% said the same in 2022.\n\n# **Year-over-year hiring trends**\n\nNote: Numbers may not sum to 100%, because of rounding.\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=144\n\nSignificantly lower 5\n\n9\n\nLower 14\n\n13\n\nAbout the same 45\n\n27\n\nSignificantly higher 6\n\n14\n\nPrefer not to answer 6\n\n1\n\nHigher 24\n\n35\n\n**In your most recent fiscal year, how much partner turnover was there at your firm compared to your expectations at the beginning of the fiscal year?**\n\n2024 2022\n\nDiverse region, inclusive workforces: Diversity and inclusion policy and practice in Asia Pacific **HEIDRICK & STRUGGLES**\n\n**11**","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/US_Professional_Services_Partner_Compensation_Survey_2024_p11.pdf"]},"response":{"errors":[],"extract_id":"extract_85659ae20b6634f19aa921908b0e46e2","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**Partner turnover (%)**\n\n## Partner turnover\n\nIn response to a slower 2023, wherein people voluntarily or involuntarily chose to leave, respondents reported partner turnover was higher in 2024 compared to 2022: 49% of respondents said partner turnover was higher or significantly higher compared to their expectations at the beginning of the year; only 30% said the same in 2022.\n\n# **Year-over-year hiring trends**\n\nNote: Numbers may not sum to 100%, because of rounding.\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=144\n\nSignificantly lower 5\n\n9\n\nLower 14\n\n13\n\nAbout the same 45\n\n27\n\nSignificantly higher 6\n\n14\n\nPrefer not to answer 6\n\n1\n\nHigher 24\n\n35\n\n**In your most recent fiscal year, how much partner turnover was there at your firm compared to your expectations at the beginning of the fiscal year?**\n\n2024 2022\n\nDiverse region, inclusive workforces: Diversity and inclusion policy and practice in Asia Pacific **HEIDRICK & STRUGGLES**\n\n**11**"],"full_content":"**Partner turnover (%)**\n\n## Partner turnover\n\nIn response to a slower 2023, wherein people voluntarily or involuntarily chose to leave, respondents reported partner turnover was higher in 2024 compared to 2022: 49% of respondents said partner turnover was higher or significantly higher compared to their expectations at the beginning of the year; only 30% said the same in 2022.\n\n# **Year-over-year hiring trends**\n\nNote: Numbers may not sum to 100%, because of rounding.\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=144\n\nSignificantly lower 5\n\n9\n\nLower 14\n\n13\n\nAbout the same 45\n\n27\n\nSignificantly higher 6\n\n14\n\nPrefer not to answer 6\n\n1\n\nHigher 24\n\n35\n\n**In your most recent fiscal year, how much partner turnover was there at your firm compared to your expectations at the beginning of the fiscal year?**\n\n2024 2022\n\nDiverse region, inclusive workforces: Diversity and inclusion policy and practice in Asia Pacific **HEIDRICK & STRUGGLES**\n\n**11**","publish_date":null,"title":"Year-over-year hiring trends","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/US_Professional_Services_Partner_Compensation_Survey_2024_p11.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/US_Professional_Services_Partner_Compensation_Survey_2024_p11.pdf"},"request":{"example_id":"chart/US_Professional_Services_Partner_Compensation_Survey_2024_p11","product_type":"parse","source_file_path":"US_Professional_Services_Partner_Compensation_Survey_2024_p11.pdf"},"started_at":"2026-11-04T19:29:58.689016"} +{"completed_at":"2026-11-04T19:45:29.436386","latency_in_ms":3320,"output":{"example_id":"chart/US_Professional_Services_Partner_Compensation_Survey_2024_p12","grounded_pages":[],"layout_pages":[],"markdown":"Broadly, there was higher turnover across all sectors, but with notable spikes at IT consulting firms, where more than 60% of respondents from those firms saw higher or significantly higher turnover than expected.\n\nAcross all firms, nearly half, 49%, saw higher or significantly higher than expected turnover. It is possible that the high turnover rate can be in part attributed to layoffs in the industry.\n\n**In your most recent fiscal year, how much partner turnover was there at your firm compared to your expectations at the beginning of the fiscal year?**\n\nNote: Numbers may not sum to 100%, because of rounding.\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=144\n\nNote: Numbers may not sum to 100%, because of rounding. For those at firms with less than $1.00bn, 6% of respondents preferred not to answer.\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=144\n\nOverall\n\nManagement/ advisory consulting\n\nOverall\n\n$1.00bn–$5.00bn\n\n$5.01bn–$20.00bn\n\nMore than $20.00bn\n\nLess than $1.00bn 48 18 6\n\n27 35 14 9 13\n\n27 35 14 9 13\n\nOther 25 38 19 6 13\n\n13 35 26 26\n\nStrategy consulting 30 45 5 5 11\n\n4 12 35 38 12\n\nTechnology/IT consulting 20 48 16 4 12\n\n14 45 26 2 14\n\n29 22 19 15 15\n\n15 6\n\n**By firm type**\n\n**By firm revenue**\n\n**In your most recent fiscal year, how much partner turnover was there at your firm compared to your expectations at the beginning of the fiscal year?**\n\nSignificantly lower\n\nSignificantly lower\n\nLower\n\nLower\n\nAbout the same\n\nAbout the same\n\nHigher\n\nHigher\n\nSignificantly higher\n\nSignificantly higher\n\n**Partner turnover (%)**\n\n**2024 US PROFESSIONAL SERVICES PARTNER COMPENSATION SURVEY**\n\n**12**","pages":[{"line_numbers":[],"markdown":"Broadly, there was higher turnover across all sectors, but with notable spikes at IT consulting firms, where more than 60% of respondents from those firms saw higher or significantly higher turnover than expected.\n\nAcross all firms, nearly half, 49%, saw higher or significantly higher than expected turnover. It is possible that the high turnover rate can be in part attributed to layoffs in the industry.\n\n**In your most recent fiscal year, how much partner turnover was there at your firm compared to your expectations at the beginning of the fiscal year?**\n\nNote: Numbers may not sum to 100%, because of rounding.\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=144\n\nNote: Numbers may not sum to 100%, because of rounding. For those at firms with less than $1.00bn, 6% of respondents preferred not to answer.\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=144\n\nOverall\n\nManagement/ advisory consulting\n\nOverall\n\n$1.00bn–$5.00bn\n\n$5.01bn–$20.00bn\n\nMore than $20.00bn\n\nLess than $1.00bn 48 18 6\n\n27 35 14 9 13\n\n27 35 14 9 13\n\nOther 25 38 19 6 13\n\n13 35 26 26\n\nStrategy consulting 30 45 5 5 11\n\n4 12 35 38 12\n\nTechnology/IT consulting 20 48 16 4 12\n\n14 45 26 2 14\n\n29 22 19 15 15\n\n15 6\n\n**By firm type**\n\n**By firm revenue**\n\n**In your most recent fiscal year, how much partner turnover was there at your firm compared to your expectations at the beginning of the fiscal year?**\n\nSignificantly lower\n\nSignificantly lower\n\nLower\n\nLower\n\nAbout the same\n\nAbout the same\n\nHigher\n\nHigher\n\nSignificantly higher\n\nSignificantly higher\n\n**Partner turnover (%)**\n\n**2024 US PROFESSIONAL SERVICES PARTNER COMPENSATION SURVEY**\n\n**12**","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Broadly, there was higher turnover across all sectors, but with notable spikes at IT consulting firms, where more than 60% of respondents from those firms saw higher or significantly higher turnover than expected.\n\nAcross all firms, nearly half, 49%, saw higher or significantly higher than expected turnover. It is possible that the high turnover rate can be in part attributed to layoffs in the industry.\n\n**In your most recent fiscal year, how much partner turnover was there at your firm compared to your expectations at the beginning of the fiscal year?**\n\nNote: Numbers may not sum to 100%, because of rounding.\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=144\n\nNote: Numbers may not sum to 100%, because of rounding. For those at firms with less than $1.00bn, 6% of respondents preferred not to answer.\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=144\n\nOverall\n\nManagement/ advisory consulting\n\nOverall\n\n$1.00bn–$5.00bn\n\n$5.01bn–$20.00bn\n\nMore than $20.00bn\n\nLess than $1.00bn 48 18 6\n\n27 35 14 9 13\n\n27 35 14 9 13\n\nOther 25 38 19 6 13\n\n13 35 26 26\n\nStrategy consulting 30 45 5 5 11\n\n4 12 35 38 12\n\nTechnology/IT consulting 20 48 16 4 12\n\n14 45 26 2 14\n\n29 22 19 15 15\n\n15 6\n\n**By firm type**\n\n**By firm revenue**\n\n**In your most recent fiscal year, how much partner turnover was there at your firm compared to your expectations at the beginning of the fiscal year?**\n\nSignificantly lower\n\nSignificantly lower\n\nLower\n\nLower\n\nAbout the same\n\nAbout the same\n\nHigher\n\nHigher\n\nSignificantly higher\n\nSignificantly higher\n\n**Partner turnover (%)**\n\n**2024 US PROFESSIONAL SERVICES PARTNER COMPENSATION SURVEY**\n\n**12**","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/US_Professional_Services_Partner_Compensation_Survey_2024_p12.pdf"]},"response":{"errors":[],"extract_id":"extract_02e990b34af0c9b7b02474976cfab3bf","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Broadly, there was higher turnover across all sectors, but with notable spikes at IT consulting firms, where more than 60% of respondents from those firms saw higher or significantly higher turnover than expected.\n\nAcross all firms, nearly half, 49%, saw higher or significantly higher than expected turnover. It is possible that the high turnover rate can be in part attributed to layoffs in the industry.\n\n**In your most recent fiscal year, how much partner turnover was there at your firm compared to your expectations at the beginning of the fiscal year?**\n\nNote: Numbers may not sum to 100%, because of rounding.\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=144\n\nNote: Numbers may not sum to 100%, because of rounding. For those at firms with less than $1.00bn, 6% of respondents preferred not to answer.\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=144\n\nOverall\n\nManagement/ advisory consulting\n\nOverall\n\n$1.00bn–$5.00bn\n\n$5.01bn–$20.00bn\n\nMore than $20.00bn\n\nLess than $1.00bn 48 18 6\n\n27 35 14 9 13\n\n27 35 14 9 13\n\nOther 25 38 19 6 13\n\n13 35 26 26\n\nStrategy consulting 30 45 5 5 11\n\n4 12 35 38 12\n\nTechnology/IT consulting 20 48 16 4 12\n\n14 45 26 2 14\n\n29 22 19 15 15\n\n15 6\n\n**By firm type**\n\n**By firm revenue**\n\n**In your most recent fiscal year, how much partner turnover was there at your firm compared to your expectations at the beginning of the fiscal year?**\n\nSignificantly lower\n\nSignificantly lower\n\nLower\n\nLower\n\nAbout the same\n\nAbout the same\n\nHigher\n\nHigher\n\nSignificantly higher\n\nSignificantly higher\n\n**Partner turnover (%)**\n\n**2024 US PROFESSIONAL SERVICES PARTNER COMPENSATION SURVEY**\n\n**12**"],"full_content":"Broadly, there was higher turnover across all sectors, but with notable spikes at IT consulting firms, where more than 60% of respondents from those firms saw higher or significantly higher turnover than expected.\n\nAcross all firms, nearly half, 49%, saw higher or significantly higher than expected turnover. It is possible that the high turnover rate can be in part attributed to layoffs in the industry.\n\n**In your most recent fiscal year, how much partner turnover was there at your firm compared to your expectations at the beginning of the fiscal year?**\n\nNote: Numbers may not sum to 100%, because of rounding.\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=144\n\nNote: Numbers may not sum to 100%, because of rounding. For those at firms with less than $1.00bn, 6% of respondents preferred not to answer.\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=144\n\nOverall\n\nManagement/ advisory consulting\n\nOverall\n\n$1.00bn–$5.00bn\n\n$5.01bn–$20.00bn\n\nMore than $20.00bn\n\nLess than $1.00bn 48 18 6\n\n27 35 14 9 13\n\n27 35 14 9 13\n\nOther 25 38 19 6 13\n\n13 35 26 26\n\nStrategy consulting 30 45 5 5 11\n\n4 12 35 38 12\n\nTechnology/IT consulting 20 48 16 4 12\n\n14 45 26 2 14\n\n29 22 19 15 15\n\n15 6\n\n**By firm type**\n\n**By firm revenue**\n\n**In your most recent fiscal year, how much partner turnover was there at your firm compared to your expectations at the beginning of the fiscal year?**\n\nSignificantly lower\n\nSignificantly lower\n\nLower\n\nLower\n\nAbout the same\n\nAbout the same\n\nHigher\n\nHigher\n\nSignificantly higher\n\nSignificantly higher\n\n**Partner turnover (%)**\n\n**2024 US PROFESSIONAL SERVICES PARTNER COMPENSATION SURVEY**\n\n**12**","publish_date":null,"title":"Partner turnover (%)","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/US_Professional_Services_Partner_Compensation_Survey_2024_p12.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/US_Professional_Services_Partner_Compensation_Survey_2024_p12.pdf"},"request":{"example_id":"chart/US_Professional_Services_Partner_Compensation_Survey_2024_p12","product_type":"parse","source_file_path":"US_Professional_Services_Partner_Compensation_Survey_2024_p12.pdf"},"started_at":"2026-11-04T19:45:26.116194"} +{"completed_at":"2026-11-04T19:45:30.276461","latency_in_ms":3298,"output":{"example_id":"chart/US_Professional_Services_Partner_Compensation_Survey_2024_p13","grounded_pages":[],"layout_pages":[],"markdown":"**By firm revenue**\n\n## Delivery support\n\nIt is likely that larger firms were not investing as much in their most recent fiscal year. In terms of delivery support, 40% of all respondents at firms with more than $20 billion in revenue said they had less support, and nearly as many at companies with $5.01 billion to $20.00 billion said the same.\n\nNote: Numbers may not sum to 100%, because of rounding.\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=144\n\nOverall\n\n$1.00bn–$5.00bn\n\n$5.01bn–$20.00bn\n\nMore than $20.00bn\n\nLess than $1.00bn\n\n34 13 8 7 38\n\n22 22 17 4 35\n\n38 12 8 38 4\n\n40 16 7 5 33\n\n30 9 6 3 52\n\n**In your most recent fiscal year, did the average delivery support ratio at your firm change?**\n\nYes, partners had less support\n\nNo Yes, partners had more support\n\nIt varied too much to say\n\nDon’t know/prefer not to answer\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=143\n\nNo 4\n\n38\n\nYes, to attract them 8\n\n6\n\nYes, to retain them 13\n\n8\n\n8\n\n6 Don’t know/prefer not to answer\n\nIt varied too much to say\n\n6\n\n2\n\nYes, for both attraction and retention 67\n\n34\n\n**In your most recent fiscal year, did your firm have to pay junior consultants notably more than in prior years?**\n\n2024 2022\n\n**Junior talent (%)**\n\n## Attracting junior talent\n\nIn 2024, we saw substantial changes in firms’ ability to attract junior consultants. Two years ago, when the market was significantly more competitive and with more work requiring more delivery support, 88% of respondents said that their firms had to pay junior consultants notably more than in prior years—for attraction, retention, or both—while only 4% said that they did not have to do so at all.\n\nIn 2024, these shares have shifted dramatically, with nearly 40% saying that they did _not_ have to pay junior consultants notably more.\n\n**Delivery support (%)**\n\nDiverse region, inclusive workforces: Diversity and inclusion policy and practice in Asia Pacific **HEIDRICK & STRUGGLES**\n\n**13**","pages":[{"line_numbers":[],"markdown":"**By firm revenue**\n\n## Delivery support\n\nIt is likely that larger firms were not investing as much in their most recent fiscal year. In terms of delivery support, 40% of all respondents at firms with more than $20 billion in revenue said they had less support, and nearly as many at companies with $5.01 billion to $20.00 billion said the same.\n\nNote: Numbers may not sum to 100%, because of rounding.\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=144\n\nOverall\n\n$1.00bn–$5.00bn\n\n$5.01bn–$20.00bn\n\nMore than $20.00bn\n\nLess than $1.00bn\n\n34 13 8 7 38\n\n22 22 17 4 35\n\n38 12 8 38 4\n\n40 16 7 5 33\n\n30 9 6 3 52\n\n**In your most recent fiscal year, did the average delivery support ratio at your firm change?**\n\nYes, partners had less support\n\nNo Yes, partners had more support\n\nIt varied too much to say\n\nDon’t know/prefer not to answer\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=143\n\nNo 4\n\n38\n\nYes, to attract them 8\n\n6\n\nYes, to retain them 13\n\n8\n\n8\n\n6 Don’t know/prefer not to answer\n\nIt varied too much to say\n\n6\n\n2\n\nYes, for both attraction and retention 67\n\n34\n\n**In your most recent fiscal year, did your firm have to pay junior consultants notably more than in prior years?**\n\n2024 2022\n\n**Junior talent (%)**\n\n## Attracting junior talent\n\nIn 2024, we saw substantial changes in firms’ ability to attract junior consultants. Two years ago, when the market was significantly more competitive and with more work requiring more delivery support, 88% of respondents said that their firms had to pay junior consultants notably more than in prior years—for attraction, retention, or both—while only 4% said that they did not have to do so at all.\n\nIn 2024, these shares have shifted dramatically, with nearly 40% saying that they did _not_ have to pay junior consultants notably more.\n\n**Delivery support (%)**\n\nDiverse region, inclusive workforces: Diversity and inclusion policy and practice in Asia Pacific **HEIDRICK & STRUGGLES**\n\n**13**","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**By firm revenue**\n\n## Delivery support\n\nIt is likely that larger firms were not investing as much in their most recent fiscal year. In terms of delivery support, 40% of all respondents at firms with more than $20 billion in revenue said they had less support, and nearly as many at companies with $5.01 billion to $20.00 billion said the same.\n\nNote: Numbers may not sum to 100%, because of rounding.\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=144\n\nOverall\n\n$1.00bn–$5.00bn\n\n$5.01bn–$20.00bn\n\nMore than $20.00bn\n\nLess than $1.00bn\n\n34 13 8 7 38\n\n22 22 17 4 35\n\n38 12 8 38 4\n\n40 16 7 5 33\n\n30 9 6 3 52\n\n**In your most recent fiscal year, did the average delivery support ratio at your firm change?**\n\nYes, partners had less support\n\nNo Yes, partners had more support\n\nIt varied too much to say\n\nDon’t know/prefer not to answer\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=143\n\nNo 4\n\n38\n\nYes, to attract them 8\n\n6\n\nYes, to retain them 13\n\n8\n\n8\n\n6 Don’t know/prefer not to answer\n\nIt varied too much to say\n\n6\n\n2\n\nYes, for both attraction and retention 67\n\n34\n\n**In your most recent fiscal year, did your firm have to pay junior consultants notably more than in prior years?**\n\n2024 2022\n\n**Junior talent (%)**\n\n## Attracting junior talent\n\nIn 2024, we saw substantial changes in firms’ ability to attract junior consultants. Two years ago, when the market was significantly more competitive and with more work requiring more delivery support, 88% of respondents said that their firms had to pay junior consultants notably more than in prior years—for attraction, retention, or both—while only 4% said that they did not have to do so at all.\n\nIn 2024, these shares have shifted dramatically, with nearly 40% saying that they did _not_ have to pay junior consultants notably more.\n\n**Delivery support (%)**\n\nDiverse region, inclusive workforces: Diversity and inclusion policy and practice in Asia Pacific **HEIDRICK & STRUGGLES**\n\n**13**","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/US_Professional_Services_Partner_Compensation_Survey_2024_p13.pdf"]},"response":{"errors":[],"extract_id":"extract_7d6ed8e945bad10437fd9a9dbf3adc29","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**By firm revenue**\n\n## Delivery support\n\nIt is likely that larger firms were not investing as much in their most recent fiscal year. In terms of delivery support, 40% of all respondents at firms with more than $20 billion in revenue said they had less support, and nearly as many at companies with $5.01 billion to $20.00 billion said the same.\n\nNote: Numbers may not sum to 100%, because of rounding.\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=144\n\nOverall\n\n$1.00bn–$5.00bn\n\n$5.01bn–$20.00bn\n\nMore than $20.00bn\n\nLess than $1.00bn\n\n34 13 8 7 38\n\n22 22 17 4 35\n\n38 12 8 38 4\n\n40 16 7 5 33\n\n30 9 6 3 52\n\n**In your most recent fiscal year, did the average delivery support ratio at your firm change?**\n\nYes, partners had less support\n\nNo Yes, partners had more support\n\nIt varied too much to say\n\nDon’t know/prefer not to answer\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=143\n\nNo 4\n\n38\n\nYes, to attract them 8\n\n6\n\nYes, to retain them 13\n\n8\n\n8\n\n6 Don’t know/prefer not to answer\n\nIt varied too much to say\n\n6\n\n2\n\nYes, for both attraction and retention 67\n\n34\n\n**In your most recent fiscal year, did your firm have to pay junior consultants notably more than in prior years?**\n\n2024 2022\n\n**Junior talent (%)**\n\n## Attracting junior talent\n\nIn 2024, we saw substantial changes in firms’ ability to attract junior consultants. Two years ago, when the market was significantly more competitive and with more work requiring more delivery support, 88% of respondents said that their firms had to pay junior consultants notably more than in prior years—for attraction, retention, or both—while only 4% said that they did not have to do so at all.\n\nIn 2024, these shares have shifted dramatically, with nearly 40% saying that they did _not_ have to pay junior consultants notably more.\n\n**Delivery support (%)**\n\nDiverse region, inclusive workforces: Diversity and inclusion policy and practice in Asia Pacific **HEIDRICK & STRUGGLES**\n\n**13**"],"full_content":"**By firm revenue**\n\n## Delivery support\n\nIt is likely that larger firms were not investing as much in their most recent fiscal year. In terms of delivery support, 40% of all respondents at firms with more than $20 billion in revenue said they had less support, and nearly as many at companies with $5.01 billion to $20.00 billion said the same.\n\nNote: Numbers may not sum to 100%, because of rounding.\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=144\n\nOverall\n\n$1.00bn–$5.00bn\n\n$5.01bn–$20.00bn\n\nMore than $20.00bn\n\nLess than $1.00bn\n\n34 13 8 7 38\n\n22 22 17 4 35\n\n38 12 8 38 4\n\n40 16 7 5 33\n\n30 9 6 3 52\n\n**In your most recent fiscal year, did the average delivery support ratio at your firm change?**\n\nYes, partners had less support\n\nNo Yes, partners had more support\n\nIt varied too much to say\n\nDon’t know/prefer not to answer\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=143\n\nNo 4\n\n38\n\nYes, to attract them 8\n\n6\n\nYes, to retain them 13\n\n8\n\n8\n\n6 Don’t know/prefer not to answer\n\nIt varied too much to say\n\n6\n\n2\n\nYes, for both attraction and retention 67\n\n34\n\n**In your most recent fiscal year, did your firm have to pay junior consultants notably more than in prior years?**\n\n2024 2022\n\n**Junior talent (%)**\n\n## Attracting junior talent\n\nIn 2024, we saw substantial changes in firms’ ability to attract junior consultants. Two years ago, when the market was significantly more competitive and with more work requiring more delivery support, 88% of respondents said that their firms had to pay junior consultants notably more than in prior years—for attraction, retention, or both—while only 4% said that they did not have to do so at all.\n\nIn 2024, these shares have shifted dramatically, with nearly 40% saying that they did _not_ have to pay junior consultants notably more.\n\n**Delivery support (%)**\n\nDiverse region, inclusive workforces: Diversity and inclusion policy and practice in Asia Pacific **HEIDRICK & STRUGGLES**\n\n**13**","publish_date":null,"title":"Delivery support Attracting junior talent","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/US_Professional_Services_Partner_Compensation_Survey_2024_p13.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/US_Professional_Services_Partner_Compensation_Survey_2024_p13.pdf"},"request":{"example_id":"chart/US_Professional_Services_Partner_Compensation_Survey_2024_p13","product_type":"parse","source_file_path":"US_Professional_Services_Partner_Compensation_Survey_2024_p13.pdf"},"started_at":"2026-11-04T19:45:26.978325"} +{"completed_at":"2026-11-04T19:30:00.313692","latency_in_ms":1130,"output":{"example_id":"chart/US_Professional_Services_Partner_Compensation_Survey_2024_p14","grounded_pages":[],"layout_pages":[],"markdown":"Technology/IT consulting 20 4 16 60\n\n**By firm type**\n\nNote: Numbers may not sum to 100%, because of rounding.\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=143\n\nOverall\n\n$1.00bn–$5.00bn\n\n$5.01bn–$20.00bn\n\nMore than $20.00bn\n\nLess than $1.00bn\n\n38 6 8 48\n\n39\n\n28\n\n9\n\n8\n\n4 48\n\n64\n\n43 5 9 43\n\n36 9 9 45\n\nNo Yes, to attract and/or retain them It varied too much to say Don’t know/prefer not to say\n\n**By firm revenue**\n\nOverall 38 6 8 48\n\nOther 31 19 50\n\nStrategy consulting 49 7 14 30\n\nNo Yes, to attract and/or retain them It varied too much to say Don’t know/prefer not to say\n\nStrategy consulting respondents least often said they had to pay junior consultants more, while technology and IT consulting respondents most often said they had to pay junior consultants more.\n\n**Junior talent (%)**\n\n**In your most recent fiscal year, did your firm have to pay junior consultants notably more than in prior years?**\n\n**In your most recent fiscal year, did your firm have to pay junior consultants notably more than in prior years?** Respondents at the largest firms least often said they needed to pay junior consultants more; still, two- thirds of respondents at companies with revenue between $5 billion and $20 billion said that they did have to pay junior consultants notably more compared to prior years.\n\nManagement/advisory consulting 41 54 3 2\n\n**2024 US PROFESSIONAL SERVICES PARTNER COMPENSATION SURVEY**\n\n**14**","pages":[{"line_numbers":[],"markdown":"Technology/IT consulting 20 4 16 60\n\n**By firm type**\n\nNote: Numbers may not sum to 100%, because of rounding.\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=143\n\nOverall\n\n$1.00bn–$5.00bn\n\n$5.01bn–$20.00bn\n\nMore than $20.00bn\n\nLess than $1.00bn\n\n38 6 8 48\n\n39\n\n28\n\n9\n\n8\n\n4 48\n\n64\n\n43 5 9 43\n\n36 9 9 45\n\nNo Yes, to attract and/or retain them It varied too much to say Don’t know/prefer not to say\n\n**By firm revenue**\n\nOverall 38 6 8 48\n\nOther 31 19 50\n\nStrategy consulting 49 7 14 30\n\nNo Yes, to attract and/or retain them It varied too much to say Don’t know/prefer not to say\n\nStrategy consulting respondents least often said they had to pay junior consultants more, while technology and IT consulting respondents most often said they had to pay junior consultants more.\n\n**Junior talent (%)**\n\n**In your most recent fiscal year, did your firm have to pay junior consultants notably more than in prior years?**\n\n**In your most recent fiscal year, did your firm have to pay junior consultants notably more than in prior years?** Respondents at the largest firms least often said they needed to pay junior consultants more; still, two- thirds of respondents at companies with revenue between $5 billion and $20 billion said that they did have to pay junior consultants notably more compared to prior years.\n\nManagement/advisory consulting 41 54 3 2\n\n**2024 US PROFESSIONAL SERVICES PARTNER COMPENSATION SURVEY**\n\n**14**","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Technology/IT consulting 20 4 16 60\n\n**By firm type**\n\nNote: Numbers may not sum to 100%, because of rounding.\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=143\n\nOverall\n\n$1.00bn–$5.00bn\n\n$5.01bn–$20.00bn\n\nMore than $20.00bn\n\nLess than $1.00bn\n\n38 6 8 48\n\n39\n\n28\n\n9\n\n8\n\n4 48\n\n64\n\n43 5 9 43\n\n36 9 9 45\n\nNo Yes, to attract and/or retain them It varied too much to say Don’t know/prefer not to say\n\n**By firm revenue**\n\nOverall 38 6 8 48\n\nOther 31 19 50\n\nStrategy consulting 49 7 14 30\n\nNo Yes, to attract and/or retain them It varied too much to say Don’t know/prefer not to say\n\nStrategy consulting respondents least often said they had to pay junior consultants more, while technology and IT consulting respondents most often said they had to pay junior consultants more.\n\n**Junior talent (%)**\n\n**In your most recent fiscal year, did your firm have to pay junior consultants notably more than in prior years?**\n\n**In your most recent fiscal year, did your firm have to pay junior consultants notably more than in prior years?** Respondents at the largest firms least often said they needed to pay junior consultants more; still, two- thirds of respondents at companies with revenue between $5 billion and $20 billion said that they did have to pay junior consultants notably more compared to prior years.\n\nManagement/advisory consulting 41 54 3 2\n\n**2024 US PROFESSIONAL SERVICES PARTNER COMPENSATION SURVEY**\n\n**14**","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/US_Professional_Services_Partner_Compensation_Survey_2024_p14.pdf"]},"response":{"errors":[],"extract_id":"extract_53a98aab226469e7d0603d9ee2e0f039","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Technology/IT consulting 20 4 16 60\n\n**By firm type**\n\nNote: Numbers may not sum to 100%, because of rounding.\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=143\n\nOverall\n\n$1.00bn–$5.00bn\n\n$5.01bn–$20.00bn\n\nMore than $20.00bn\n\nLess than $1.00bn\n\n38 6 8 48\n\n39\n\n28\n\n9\n\n8\n\n4 48\n\n64\n\n43 5 9 43\n\n36 9 9 45\n\nNo Yes, to attract and/or retain them It varied too much to say Don’t know/prefer not to say\n\n**By firm revenue**\n\nOverall 38 6 8 48\n\nOther 31 19 50\n\nStrategy consulting 49 7 14 30\n\nNo Yes, to attract and/or retain them It varied too much to say Don’t know/prefer not to say\n\nStrategy consulting respondents least often said they had to pay junior consultants more, while technology and IT consulting respondents most often said they had to pay junior consultants more.\n\n**Junior talent (%)**\n\n**In your most recent fiscal year, did your firm have to pay junior consultants notably more than in prior years?**\n\n**In your most recent fiscal year, did your firm have to pay junior consultants notably more than in prior years?** Respondents at the largest firms least often said they needed to pay junior consultants more; still, two- thirds of respondents at companies with revenue between $5 billion and $20 billion said that they did have to pay junior consultants notably more compared to prior years.\n\nManagement/advisory consulting 41 54 3 2\n\n**2024 US PROFESSIONAL SERVICES PARTNER COMPENSATION SURVEY**\n\n**14**"],"full_content":"Technology/IT consulting 20 4 16 60\n\n**By firm type**\n\nNote: Numbers may not sum to 100%, because of rounding.\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=143\n\nOverall\n\n$1.00bn–$5.00bn\n\n$5.01bn–$20.00bn\n\nMore than $20.00bn\n\nLess than $1.00bn\n\n38 6 8 48\n\n39\n\n28\n\n9\n\n8\n\n4 48\n\n64\n\n43 5 9 43\n\n36 9 9 45\n\nNo Yes, to attract and/or retain them It varied too much to say Don’t know/prefer not to say\n\n**By firm revenue**\n\nOverall 38 6 8 48\n\nOther 31 19 50\n\nStrategy consulting 49 7 14 30\n\nNo Yes, to attract and/or retain them It varied too much to say Don’t know/prefer not to say\n\nStrategy consulting respondents least often said they had to pay junior consultants more, while technology and IT consulting respondents most often said they had to pay junior consultants more.\n\n**Junior talent (%)**\n\n**In your most recent fiscal year, did your firm have to pay junior consultants notably more than in prior years?**\n\n**In your most recent fiscal year, did your firm have to pay junior consultants notably more than in prior years?** Respondents at the largest firms least often said they needed to pay junior consultants more; still, two- thirds of respondents at companies with revenue between $5 billion and $20 billion said that they did have to pay junior consultants notably more compared to prior years.\n\nManagement/advisory consulting 41 54 3 2\n\n**2024 US PROFESSIONAL SERVICES PARTNER COMPENSATION SURVEY**\n\n**14**","publish_date":null,"title":"Junior talent (%)","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/US_Professional_Services_Partner_Compensation_Survey_2024_p14.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/US_Professional_Services_Partner_Compensation_Survey_2024_p14.pdf"},"request":{"example_id":"chart/US_Professional_Services_Partner_Compensation_Survey_2024_p14","product_type":"parse","source_file_path":"US_Professional_Services_Partner_Compensation_Survey_2024_p14.pdf"},"started_at":"2026-11-04T19:29:59.182936"} +{"completed_at":"2026-11-04T19:45:30.872374","latency_in_ms":1272,"output":{"example_id":"chart/US_Professional_Services_Partner_Compensation_Survey_2024_p15","grounded_pages":[],"layout_pages":[],"markdown":"Note: Numbers may not sum to 100%, because of rounding.\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=144\n\nYes 49\n\n27\n\nNo 19\n\n40\n\nIt varied too much to say 8\n\n12\n\nDon’t know/prefer not to answer 23\n\n21\n\n**In your most recent fiscal year, did your firm pay a premium to attract or retain diverse consultants at any level?**\n\n2024 2022\n\n**Diverse talent (%)**\n\n## Attracting diverse talent\n\nMeanwhile, significantly fewer respondents said they’re paying a premium to attract or retain diverse consultants at any level.\n\nWe found the results of this year’s survey disappointing, if not surprising. As firms face more uncertain market conditions, we are seeing the focus on diversity and inclusion wane slightly across industries as firms retreat to more traditional hiring practices. More forward-looking firms, however, will find themselves best positioned for future success by both continuing to prioritize competition for a diverse group of partners and cultivating a diverse leadership pipeline of more junior talent.\n\nRespondents at strategy consulting firms more often said that they did not pay a premium for diverse talent at any level.\n\nOverall 27 12 21 40\n\nOther 31 19 19 31\n\nManagement/advisory consulting 29 12 19 41\n\nStrategy consulting 18 7 20 55\n\nTechnology/IT consulting 36 16 28 20\n\nYes No It varied too much to say Don’t know/prefer not to say\n\n**By firm type**\n\nNote: Numbers may not sum to 100%, because of rounding.\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=144\n\n**In your most recent fiscal year, did your firm pay a premium to attract or retain diverse consultants at any level?**\n\nDiverse region, inclusive workforces: Diversity and inclusion policy and practice in Asia Pacific **HEIDRICK & STRUGGLES**\n\n**15**","pages":[{"line_numbers":[],"markdown":"Note: Numbers may not sum to 100%, because of rounding.\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=144\n\nYes 49\n\n27\n\nNo 19\n\n40\n\nIt varied too much to say 8\n\n12\n\nDon’t know/prefer not to answer 23\n\n21\n\n**In your most recent fiscal year, did your firm pay a premium to attract or retain diverse consultants at any level?**\n\n2024 2022\n\n**Diverse talent (%)**\n\n## Attracting diverse talent\n\nMeanwhile, significantly fewer respondents said they’re paying a premium to attract or retain diverse consultants at any level.\n\nWe found the results of this year’s survey disappointing, if not surprising. As firms face more uncertain market conditions, we are seeing the focus on diversity and inclusion wane slightly across industries as firms retreat to more traditional hiring practices. More forward-looking firms, however, will find themselves best positioned for future success by both continuing to prioritize competition for a diverse group of partners and cultivating a diverse leadership pipeline of more junior talent.\n\nRespondents at strategy consulting firms more often said that they did not pay a premium for diverse talent at any level.\n\nOverall 27 12 21 40\n\nOther 31 19 19 31\n\nManagement/advisory consulting 29 12 19 41\n\nStrategy consulting 18 7 20 55\n\nTechnology/IT consulting 36 16 28 20\n\nYes No It varied too much to say Don’t know/prefer not to say\n\n**By firm type**\n\nNote: Numbers may not sum to 100%, because of rounding.\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=144\n\n**In your most recent fiscal year, did your firm pay a premium to attract or retain diverse consultants at any level?**\n\nDiverse region, inclusive workforces: Diversity and inclusion policy and practice in Asia Pacific **HEIDRICK & STRUGGLES**\n\n**15**","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Note: Numbers may not sum to 100%, because of rounding.\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=144\n\nYes 49\n\n27\n\nNo 19\n\n40\n\nIt varied too much to say 8\n\n12\n\nDon’t know/prefer not to answer 23\n\n21\n\n**In your most recent fiscal year, did your firm pay a premium to attract or retain diverse consultants at any level?**\n\n2024 2022\n\n**Diverse talent (%)**\n\n## Attracting diverse talent\n\nMeanwhile, significantly fewer respondents said they’re paying a premium to attract or retain diverse consultants at any level.\n\nWe found the results of this year’s survey disappointing, if not surprising. As firms face more uncertain market conditions, we are seeing the focus on diversity and inclusion wane slightly across industries as firms retreat to more traditional hiring practices. More forward-looking firms, however, will find themselves best positioned for future success by both continuing to prioritize competition for a diverse group of partners and cultivating a diverse leadership pipeline of more junior talent.\n\nRespondents at strategy consulting firms more often said that they did not pay a premium for diverse talent at any level.\n\nOverall 27 12 21 40\n\nOther 31 19 19 31\n\nManagement/advisory consulting 29 12 19 41\n\nStrategy consulting 18 7 20 55\n\nTechnology/IT consulting 36 16 28 20\n\nYes No It varied too much to say Don’t know/prefer not to say\n\n**By firm type**\n\nNote: Numbers may not sum to 100%, because of rounding.\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=144\n\n**In your most recent fiscal year, did your firm pay a premium to attract or retain diverse consultants at any level?**\n\nDiverse region, inclusive workforces: Diversity and inclusion policy and practice in Asia Pacific **HEIDRICK & STRUGGLES**\n\n**15**","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/US_Professional_Services_Partner_Compensation_Survey_2024_p15.pdf"]},"response":{"errors":[],"extract_id":"extract_6a4ff7c3d2b65c8f649f972c8131c227","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Note: Numbers may not sum to 100%, because of rounding.\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=144\n\nYes 49\n\n27\n\nNo 19\n\n40\n\nIt varied too much to say 8\n\n12\n\nDon’t know/prefer not to answer 23\n\n21\n\n**In your most recent fiscal year, did your firm pay a premium to attract or retain diverse consultants at any level?**\n\n2024 2022\n\n**Diverse talent (%)**\n\n## Attracting diverse talent\n\nMeanwhile, significantly fewer respondents said they’re paying a premium to attract or retain diverse consultants at any level.\n\nWe found the results of this year’s survey disappointing, if not surprising. As firms face more uncertain market conditions, we are seeing the focus on diversity and inclusion wane slightly across industries as firms retreat to more traditional hiring practices. More forward-looking firms, however, will find themselves best positioned for future success by both continuing to prioritize competition for a diverse group of partners and cultivating a diverse leadership pipeline of more junior talent.\n\nRespondents at strategy consulting firms more often said that they did not pay a premium for diverse talent at any level.\n\nOverall 27 12 21 40\n\nOther 31 19 19 31\n\nManagement/advisory consulting 29 12 19 41\n\nStrategy consulting 18 7 20 55\n\nTechnology/IT consulting 36 16 28 20\n\nYes No It varied too much to say Don’t know/prefer not to say\n\n**By firm type**\n\nNote: Numbers may not sum to 100%, because of rounding.\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=144\n\n**In your most recent fiscal year, did your firm pay a premium to attract or retain diverse consultants at any level?**\n\nDiverse region, inclusive workforces: Diversity and inclusion policy and practice in Asia Pacific **HEIDRICK & STRUGGLES**\n\n**15**"],"full_content":"Note: Numbers may not sum to 100%, because of rounding.\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=144\n\nYes 49\n\n27\n\nNo 19\n\n40\n\nIt varied too much to say 8\n\n12\n\nDon’t know/prefer not to answer 23\n\n21\n\n**In your most recent fiscal year, did your firm pay a premium to attract or retain diverse consultants at any level?**\n\n2024 2022\n\n**Diverse talent (%)**\n\n## Attracting diverse talent\n\nMeanwhile, significantly fewer respondents said they’re paying a premium to attract or retain diverse consultants at any level.\n\nWe found the results of this year’s survey disappointing, if not surprising. As firms face more uncertain market conditions, we are seeing the focus on diversity and inclusion wane slightly across industries as firms retreat to more traditional hiring practices. More forward-looking firms, however, will find themselves best positioned for future success by both continuing to prioritize competition for a diverse group of partners and cultivating a diverse leadership pipeline of more junior talent.\n\nRespondents at strategy consulting firms more often said that they did not pay a premium for diverse talent at any level.\n\nOverall 27 12 21 40\n\nOther 31 19 19 31\n\nManagement/advisory consulting 29 12 19 41\n\nStrategy consulting 18 7 20 55\n\nTechnology/IT consulting 36 16 28 20\n\nYes No It varied too much to say Don’t know/prefer not to say\n\n**By firm type**\n\nNote: Numbers may not sum to 100%, because of rounding.\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=144\n\n**In your most recent fiscal year, did your firm pay a premium to attract or retain diverse consultants at any level?**\n\nDiverse region, inclusive workforces: Diversity and inclusion policy and practice in Asia Pacific **HEIDRICK & STRUGGLES**\n\n**15**","publish_date":null,"title":"We found the results of this year’s","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/US_Professional_Services_Partner_Compensation_Survey_2024_p15.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/US_Professional_Services_Partner_Compensation_Survey_2024_p15.pdf"},"request":{"example_id":"chart/US_Professional_Services_Partner_Compensation_Survey_2024_p15","product_type":"parse","source_file_path":"US_Professional_Services_Partner_Compensation_Survey_2024_p15.pdf"},"started_at":"2026-11-04T19:45:29.599564"} +{"completed_at":"2026-11-04T19:45:32.024535","latency_in_ms":1390,"output":{"example_id":"chart/US_Professional_Services_Partner_Compensation_Survey_2024_p16","grounded_pages":[],"layout_pages":[],"markdown":"Yes No It varied too much to say Don’t know/prefer not to say\n\nNote: Numbers may not sum to 100%, because of rounding.\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=144\n\nOverall\n\n$1.00bn–$5.00bn\n\n$5.01bn–$20.00bn\n\nMore than $20.00bn\n\nLess than $1.00bn\n\n27 12 21 40\n\n39 26 9 26\n\n42 15 31 12\n\n24 9 29 38\n\n15 9 18 58\n\n**By firm revenue**\n\n**In your most recent fiscal year, did your firm pay a premium to attract or retain diverse consultants at any level?**\n\n**Diverse talent (%)** There was an overall decline in reported premiums paid to attract diverse consultants compared to 2022. This decrease in the investment in diverse hiring is disappointing, and we hope to see the trend reverse itself in future surveys.\n\n**2024 US PROFESSIONAL SERVICES PARTNER COMPENSATION SURVEY**\n\n**16**","pages":[{"line_numbers":[],"markdown":"Yes No It varied too much to say Don’t know/prefer not to say\n\nNote: Numbers may not sum to 100%, because of rounding.\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=144\n\nOverall\n\n$1.00bn–$5.00bn\n\n$5.01bn–$20.00bn\n\nMore than $20.00bn\n\nLess than $1.00bn\n\n27 12 21 40\n\n39 26 9 26\n\n42 15 31 12\n\n24 9 29 38\n\n15 9 18 58\n\n**By firm revenue**\n\n**In your most recent fiscal year, did your firm pay a premium to attract or retain diverse consultants at any level?**\n\n**Diverse talent (%)** There was an overall decline in reported premiums paid to attract diverse consultants compared to 2022. This decrease in the investment in diverse hiring is disappointing, and we hope to see the trend reverse itself in future surveys.\n\n**2024 US PROFESSIONAL SERVICES PARTNER COMPENSATION SURVEY**\n\n**16**","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Yes No It varied too much to say Don’t know/prefer not to say\n\nNote: Numbers may not sum to 100%, because of rounding.\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=144\n\nOverall\n\n$1.00bn–$5.00bn\n\n$5.01bn–$20.00bn\n\nMore than $20.00bn\n\nLess than $1.00bn\n\n27 12 21 40\n\n39 26 9 26\n\n42 15 31 12\n\n24 9 29 38\n\n15 9 18 58\n\n**By firm revenue**\n\n**In your most recent fiscal year, did your firm pay a premium to attract or retain diverse consultants at any level?**\n\n**Diverse talent (%)** There was an overall decline in reported premiums paid to attract diverse consultants compared to 2022. This decrease in the investment in diverse hiring is disappointing, and we hope to see the trend reverse itself in future surveys.\n\n**2024 US PROFESSIONAL SERVICES PARTNER COMPENSATION SURVEY**\n\n**16**","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/US_Professional_Services_Partner_Compensation_Survey_2024_p16.pdf"]},"response":{"errors":[],"extract_id":"extract_7c6dfeeeb2f3a6ea79192f5bdfd6b898","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Yes No It varied too much to say Don’t know/prefer not to say\n\nNote: Numbers may not sum to 100%, because of rounding.\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=144\n\nOverall\n\n$1.00bn–$5.00bn\n\n$5.01bn–$20.00bn\n\nMore than $20.00bn\n\nLess than $1.00bn\n\n27 12 21 40\n\n39 26 9 26\n\n42 15 31 12\n\n24 9 29 38\n\n15 9 18 58\n\n**By firm revenue**\n\n**In your most recent fiscal year, did your firm pay a premium to attract or retain diverse consultants at any level?**\n\n**Diverse talent (%)** There was an overall decline in reported premiums paid to attract diverse consultants compared to 2022. This decrease in the investment in diverse hiring is disappointing, and we hope to see the trend reverse itself in future surveys.\n\n**2024 US PROFESSIONAL SERVICES PARTNER COMPENSATION SURVEY**\n\n**16**"],"full_content":"Yes No It varied too much to say Don’t know/prefer not to say\n\nNote: Numbers may not sum to 100%, because of rounding.\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=144\n\nOverall\n\n$1.00bn–$5.00bn\n\n$5.01bn–$20.00bn\n\nMore than $20.00bn\n\nLess than $1.00bn\n\n27 12 21 40\n\n39 26 9 26\n\n42 15 31 12\n\n24 9 29 38\n\n15 9 18 58\n\n**By firm revenue**\n\n**In your most recent fiscal year, did your firm pay a premium to attract or retain diverse consultants at any level?**\n\n**Diverse talent (%)** There was an overall decline in reported premiums paid to attract diverse consultants compared to 2022. This decrease in the investment in diverse hiring is disappointing, and we hope to see the trend reverse itself in future surveys.\n\n**2024 US PROFESSIONAL SERVICES PARTNER COMPENSATION SURVEY**\n\n**16**","publish_date":null,"title":"Diverse talent (%)","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/US_Professional_Services_Partner_Compensation_Survey_2024_p16.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/US_Professional_Services_Partner_Compensation_Survey_2024_p16.pdf"},"request":{"example_id":"chart/US_Professional_Services_Partner_Compensation_Survey_2024_p16","product_type":"parse","source_file_path":"US_Professional_Services_Partner_Compensation_Survey_2024_p16.pdf"},"started_at":"2026-11-04T19:45:30.634313"} +{"completed_at":"2026-11-04T19:45:32.614170","latency_in_ms":1225,"output":{"example_id":"chart/US_Professional_Services_Partner_Compensation_Survey_2024_p17","grounded_pages":[],"layout_pages":[],"markdown":"## Outlook for this fiscal year\n\nRespondents’ average outlook for their firm’s fiscal year compared to the previous fiscal year skews slightly optimistic, with 44% who are optimistic.\n\nVery pessimistic\n\nSomewhat pessimistic\n\nNeutral\n\nSomewhat optimistic\n\nVery optimistic\n\nInterest rates and their impact on the economy\n\nUS presidential elections/geopolitical uncertainty\n\nRise of AI\n\nEffects of hybrid work models on our organization\n\nIncreasing threat of cyberattacks\n\nRise of industrial automation\n\nRise of green resources and sustainable technology\n\nCommercial real estate market\n\nOther\n\n44% are optimistic\n\n26% are pessimistic\n\n**What is your outlook on your firm’s overall performance this fiscal year compared to last fiscal year?**\n\n**What are the key external factors influencing your overall outlook for this fiscal year? (Select all that apply)**\n\n**Outlook for current fiscal year (%)**\n\n**Key external factors influencing outlook (%)**\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=142\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=144\n\n3\n\n80\n\n23\n\n56\n\n30\n\n52\n\n8\n\n34\n\n22\n\n7\n\n10\n\n11\n\n11\n\n9\n\n# **Future outlook and expectations**\n\nFor this year’s report, given the unpredictability of the economy and professional services market, we asked respondents to share their expectations for the current fiscal year—whether they are optimistic, neutral, or pessimistic, and what external factors are most contributing to that outlook.\n\nUnsurprisingly, interest rates, AI, and the US presidential election and\n\ngeopolitical uncertainty seem to be top of mind for everyone, regardless of their general outlook. What was more notable was respondents’ outlook regarding the effects of hybrid work models on their organizations. Those with a pessimistic outlook cited the effects of hybrid work only 13% of the time, while those with an optimistic outlook cited hybrid 27% of the time. It is encouraging to see that new working\n\nnorms are more often a cause for a positive outlook, especially following so many years of uncertainty and conflict.\n\nRegardless of these factors, our survey respondents’ average outlook for their firm’s fiscal year compared to the last year skews slightly optimistic. Respondents at smaller firms, those with between $1 billion and $5 billion in revenue, were the most optimistic.\n\n## External factors contributing to outlook\n\nRegardless of respondents’ general outlook, interest rates and their impact on the economy is the top key external factor influencing respondents’ outlook. This is unsurprising given the effect interest rates have on all of consulting.\n\nIt is also notable that the top three external factors influencing respondents’\n\nDiverse region, inclusive workforces: Diversity and inclusion policy and practice in Asia Pacific **HEIDRICK & STRUGGLES**\n\n**17**","pages":[{"line_numbers":[],"markdown":"## Outlook for this fiscal year\n\nRespondents’ average outlook for their firm’s fiscal year compared to the previous fiscal year skews slightly optimistic, with 44% who are optimistic.\n\nVery pessimistic\n\nSomewhat pessimistic\n\nNeutral\n\nSomewhat optimistic\n\nVery optimistic\n\nInterest rates and their impact on the economy\n\nUS presidential elections/geopolitical uncertainty\n\nRise of AI\n\nEffects of hybrid work models on our organization\n\nIncreasing threat of cyberattacks\n\nRise of industrial automation\n\nRise of green resources and sustainable technology\n\nCommercial real estate market\n\nOther\n\n44% are optimistic\n\n26% are pessimistic\n\n**What is your outlook on your firm’s overall performance this fiscal year compared to last fiscal year?**\n\n**What are the key external factors influencing your overall outlook for this fiscal year? (Select all that apply)**\n\n**Outlook for current fiscal year (%)**\n\n**Key external factors influencing outlook (%)**\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=142\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=144\n\n3\n\n80\n\n23\n\n56\n\n30\n\n52\n\n8\n\n34\n\n22\n\n7\n\n10\n\n11\n\n11\n\n9\n\n# **Future outlook and expectations**\n\nFor this year’s report, given the unpredictability of the economy and professional services market, we asked respondents to share their expectations for the current fiscal year—whether they are optimistic, neutral, or pessimistic, and what external factors are most contributing to that outlook.\n\nUnsurprisingly, interest rates, AI, and the US presidential election and\n\ngeopolitical uncertainty seem to be top of mind for everyone, regardless of their general outlook. What was more notable was respondents’ outlook regarding the effects of hybrid work models on their organizations. Those with a pessimistic outlook cited the effects of hybrid work only 13% of the time, while those with an optimistic outlook cited hybrid 27% of the time. It is encouraging to see that new working\n\nnorms are more often a cause for a positive outlook, especially following so many years of uncertainty and conflict.\n\nRegardless of these factors, our survey respondents’ average outlook for their firm’s fiscal year compared to the last year skews slightly optimistic. Respondents at smaller firms, those with between $1 billion and $5 billion in revenue, were the most optimistic.\n\n## External factors contributing to outlook\n\nRegardless of respondents’ general outlook, interest rates and their impact on the economy is the top key external factor influencing respondents’ outlook. This is unsurprising given the effect interest rates have on all of consulting.\n\nIt is also notable that the top three external factors influencing respondents’\n\nDiverse region, inclusive workforces: Diversity and inclusion policy and practice in Asia Pacific **HEIDRICK & STRUGGLES**\n\n**17**","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"## Outlook for this fiscal year\n\nRespondents’ average outlook for their firm’s fiscal year compared to the previous fiscal year skews slightly optimistic, with 44% who are optimistic.\n\nVery pessimistic\n\nSomewhat pessimistic\n\nNeutral\n\nSomewhat optimistic\n\nVery optimistic\n\nInterest rates and their impact on the economy\n\nUS presidential elections/geopolitical uncertainty\n\nRise of AI\n\nEffects of hybrid work models on our organization\n\nIncreasing threat of cyberattacks\n\nRise of industrial automation\n\nRise of green resources and sustainable technology\n\nCommercial real estate market\n\nOther\n\n44% are optimistic\n\n26% are pessimistic\n\n**What is your outlook on your firm’s overall performance this fiscal year compared to last fiscal year?**\n\n**What are the key external factors influencing your overall outlook for this fiscal year? (Select all that apply)**\n\n**Outlook for current fiscal year (%)**\n\n**Key external factors influencing outlook (%)**\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=142\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=144\n\n3\n\n80\n\n23\n\n56\n\n30\n\n52\n\n8\n\n34\n\n22\n\n7\n\n10\n\n11\n\n11\n\n9\n\n# **Future outlook and expectations**\n\nFor this year’s report, given the unpredictability of the economy and professional services market, we asked respondents to share their expectations for the current fiscal year—whether they are optimistic, neutral, or pessimistic, and what external factors are most contributing to that outlook.\n\nUnsurprisingly, interest rates, AI, and the US presidential election and\n\ngeopolitical uncertainty seem to be top of mind for everyone, regardless of their general outlook. What was more notable was respondents’ outlook regarding the effects of hybrid work models on their organizations. Those with a pessimistic outlook cited the effects of hybrid work only 13% of the time, while those with an optimistic outlook cited hybrid 27% of the time. It is encouraging to see that new working\n\nnorms are more often a cause for a positive outlook, especially following so many years of uncertainty and conflict.\n\nRegardless of these factors, our survey respondents’ average outlook for their firm’s fiscal year compared to the last year skews slightly optimistic. Respondents at smaller firms, those with between $1 billion and $5 billion in revenue, were the most optimistic.\n\n## External factors contributing to outlook\n\nRegardless of respondents’ general outlook, interest rates and their impact on the economy is the top key external factor influencing respondents’ outlook. This is unsurprising given the effect interest rates have on all of consulting.\n\nIt is also notable that the top three external factors influencing respondents’\n\nDiverse region, inclusive workforces: Diversity and inclusion policy and practice in Asia Pacific **HEIDRICK & STRUGGLES**\n\n**17**","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/US_Professional_Services_Partner_Compensation_Survey_2024_p17.pdf"]},"response":{"errors":[],"extract_id":"extract_a76003bb9a021f7ff4c73738f2357259","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["## Outlook for this fiscal year\n\nRespondents’ average outlook for their firm’s fiscal year compared to the previous fiscal year skews slightly optimistic, with 44% who are optimistic.\n\nVery pessimistic\n\nSomewhat pessimistic\n\nNeutral\n\nSomewhat optimistic\n\nVery optimistic\n\nInterest rates and their impact on the economy\n\nUS presidential elections/geopolitical uncertainty\n\nRise of AI\n\nEffects of hybrid work models on our organization\n\nIncreasing threat of cyberattacks\n\nRise of industrial automation\n\nRise of green resources and sustainable technology\n\nCommercial real estate market\n\nOther\n\n44% are optimistic\n\n26% are pessimistic\n\n**What is your outlook on your firm’s overall performance this fiscal year compared to last fiscal year?**\n\n**What are the key external factors influencing your overall outlook for this fiscal year? (Select all that apply)**\n\n**Outlook for current fiscal year (%)**\n\n**Key external factors influencing outlook (%)**\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=142\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=144\n\n3\n\n80\n\n23\n\n56\n\n30\n\n52\n\n8\n\n34\n\n22\n\n7\n\n10\n\n11\n\n11\n\n9\n\n# **Future outlook and expectations**\n\nFor this year’s report, given the unpredictability of the economy and professional services market, we asked respondents to share their expectations for the current fiscal year—whether they are optimistic, neutral, or pessimistic, and what external factors are most contributing to that outlook.\n\nUnsurprisingly, interest rates, AI, and the US presidential election and\n\ngeopolitical uncertainty seem to be top of mind for everyone, regardless of their general outlook. What was more notable was respondents’ outlook regarding the effects of hybrid work models on their organizations. Those with a pessimistic outlook cited the effects of hybrid work only 13% of the time, while those with an optimistic outlook cited hybrid 27% of the time. It is encouraging to see that new working\n\nnorms are more often a cause for a positive outlook, especially following so many years of uncertainty and conflict.\n\nRegardless of these factors, our survey respondents’ average outlook for their firm’s fiscal year compared to the last year skews slightly optimistic. Respondents at smaller firms, those with between $1 billion and $5 billion in revenue, were the most optimistic.\n\n## External factors contributing to outlook\n\nRegardless of respondents’ general outlook, interest rates and their impact on the economy is the top key external factor influencing respondents’ outlook. This is unsurprising given the effect interest rates have on all of consulting.\n\nIt is also notable that the top three external factors influencing respondents’\n\nDiverse region, inclusive workforces: Diversity and inclusion policy and practice in Asia Pacific **HEIDRICK & STRUGGLES**\n\n**17**"],"full_content":"## Outlook for this fiscal year\n\nRespondents’ average outlook for their firm’s fiscal year compared to the previous fiscal year skews slightly optimistic, with 44% who are optimistic.\n\nVery pessimistic\n\nSomewhat pessimistic\n\nNeutral\n\nSomewhat optimistic\n\nVery optimistic\n\nInterest rates and their impact on the economy\n\nUS presidential elections/geopolitical uncertainty\n\nRise of AI\n\nEffects of hybrid work models on our organization\n\nIncreasing threat of cyberattacks\n\nRise of industrial automation\n\nRise of green resources and sustainable technology\n\nCommercial real estate market\n\nOther\n\n44% are optimistic\n\n26% are pessimistic\n\n**What is your outlook on your firm’s overall performance this fiscal year compared to last fiscal year?**\n\n**What are the key external factors influencing your overall outlook for this fiscal year? (Select all that apply)**\n\n**Outlook for current fiscal year (%)**\n\n**Key external factors influencing outlook (%)**\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=142\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=144\n\n3\n\n80\n\n23\n\n56\n\n30\n\n52\n\n8\n\n34\n\n22\n\n7\n\n10\n\n11\n\n11\n\n9\n\n# **Future outlook and expectations**\n\nFor this year’s report, given the unpredictability of the economy and professional services market, we asked respondents to share their expectations for the current fiscal year—whether they are optimistic, neutral, or pessimistic, and what external factors are most contributing to that outlook.\n\nUnsurprisingly, interest rates, AI, and the US presidential election and\n\ngeopolitical uncertainty seem to be top of mind for everyone, regardless of their general outlook. What was more notable was respondents’ outlook regarding the effects of hybrid work models on their organizations. Those with a pessimistic outlook cited the effects of hybrid work only 13% of the time, while those with an optimistic outlook cited hybrid 27% of the time. It is encouraging to see that new working\n\nnorms are more often a cause for a positive outlook, especially following so many years of uncertainty and conflict.\n\nRegardless of these factors, our survey respondents’ average outlook for their firm’s fiscal year compared to the last year skews slightly optimistic. Respondents at smaller firms, those with between $1 billion and $5 billion in revenue, were the most optimistic.\n\n## External factors contributing to outlook\n\nRegardless of respondents’ general outlook, interest rates and their impact on the economy is the top key external factor influencing respondents’ outlook. This is unsurprising given the effect interest rates have on all of consulting.\n\nIt is also notable that the top three external factors influencing respondents’\n\nDiverse region, inclusive workforces: Diversity and inclusion policy and practice in Asia Pacific **HEIDRICK & STRUGGLES**\n\n**17**","publish_date":null,"title":"Future outlook and expectations","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/US_Professional_Services_Partner_Compensation_Survey_2024_p17.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/US_Professional_Services_Partner_Compensation_Survey_2024_p17.pdf"},"request":{"example_id":"chart/US_Professional_Services_Partner_Compensation_Survey_2024_p17","product_type":"parse","source_file_path":"US_Professional_Services_Partner_Compensation_Survey_2024_p17.pdf"},"started_at":"2026-11-04T19:45:31.388287"} +{"completed_at":"2026-11-04T19:45:33.416665","latency_in_ms":1134,"output":{"example_id":"chart/US_Professional_Services_Partner_Compensation_Survey_2024_p19","grounded_pages":[],"layout_pages":[],"markdown":"When compared to respondents at other types of firms, those at management consulting firms least often cited AI as a key external factor—unsurprising, given that management consulting is still in the early stages of planning for and implementing AI technologies into their products, services, and processes.\n\nWe also find it unsurprising that respondents at strategy consulting firms reported the most concern about the US presidential election and geopolitical issues, given that much of work is at the board level. It makes sense that these issues are top of mind for board members.\n\nFor respondents in IT consulting, it also makes sense that there was slightly less overall focus on interest rates and their impact on the economy, given that their services are in many ways a commodity for organizations and are therefore less likely to be as significantly impacted by economic shifts.\n\n**Key external factors influencing outlook for current fiscal year, by firm type (Select all that apply)**\n\n65\n\n51\n\n60\n\n40\n\n53\n\n44\n\n60\n\n67\n\n81 81\n\n72\n\n80\n\n12\n\n8\n\n4 0\n\n5\n\n12\n\n20\n\n7\n\n7 10\n\n16\n\n0\n\n14\n\n19\n\n32\n\n40\n\n7\n\n15\n\n4\n\n20\n\nInterest rates and their impact on the economy\n\nUS presidential elections/geopolitical uncertainty\n\nRise of AI\n\nEffects of hybrid work models on our organization\n\nIncreasing threat of cyberattacks\n\nRise of industrial automation\n\nRise of green resources and sustainable technology\n\nCommercial real estate market\n\nOther\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=142\n\nStrategy consulting Management/advisory consulting Technology/IT consulting Other\n\n12 0\n\n13\n\n2\n\n**Key external factors influencing outlook (%)**\n\nDiverse region, inclusive workforces: Diversity and inclusion policy and practice in Asia Pacific **HEIDRICK & STRUGGLES**\n\n**19**","pages":[{"line_numbers":[],"markdown":"When compared to respondents at other types of firms, those at management consulting firms least often cited AI as a key external factor—unsurprising, given that management consulting is still in the early stages of planning for and implementing AI technologies into their products, services, and processes.\n\nWe also find it unsurprising that respondents at strategy consulting firms reported the most concern about the US presidential election and geopolitical issues, given that much of work is at the board level. It makes sense that these issues are top of mind for board members.\n\nFor respondents in IT consulting, it also makes sense that there was slightly less overall focus on interest rates and their impact on the economy, given that their services are in many ways a commodity for organizations and are therefore less likely to be as significantly impacted by economic shifts.\n\n**Key external factors influencing outlook for current fiscal year, by firm type (Select all that apply)**\n\n65\n\n51\n\n60\n\n40\n\n53\n\n44\n\n60\n\n67\n\n81 81\n\n72\n\n80\n\n12\n\n8\n\n4 0\n\n5\n\n12\n\n20\n\n7\n\n7 10\n\n16\n\n0\n\n14\n\n19\n\n32\n\n40\n\n7\n\n15\n\n4\n\n20\n\nInterest rates and their impact on the economy\n\nUS presidential elections/geopolitical uncertainty\n\nRise of AI\n\nEffects of hybrid work models on our organization\n\nIncreasing threat of cyberattacks\n\nRise of industrial automation\n\nRise of green resources and sustainable technology\n\nCommercial real estate market\n\nOther\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=142\n\nStrategy consulting Management/advisory consulting Technology/IT consulting Other\n\n12 0\n\n13\n\n2\n\n**Key external factors influencing outlook (%)**\n\nDiverse region, inclusive workforces: Diversity and inclusion policy and practice in Asia Pacific **HEIDRICK & STRUGGLES**\n\n**19**","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"When compared to respondents at other types of firms, those at management consulting firms least often cited AI as a key external factor—unsurprising, given that management consulting is still in the early stages of planning for and implementing AI technologies into their products, services, and processes.\n\nWe also find it unsurprising that respondents at strategy consulting firms reported the most concern about the US presidential election and geopolitical issues, given that much of work is at the board level. It makes sense that these issues are top of mind for board members.\n\nFor respondents in IT consulting, it also makes sense that there was slightly less overall focus on interest rates and their impact on the economy, given that their services are in many ways a commodity for organizations and are therefore less likely to be as significantly impacted by economic shifts.\n\n**Key external factors influencing outlook for current fiscal year, by firm type (Select all that apply)**\n\n65\n\n51\n\n60\n\n40\n\n53\n\n44\n\n60\n\n67\n\n81 81\n\n72\n\n80\n\n12\n\n8\n\n4 0\n\n5\n\n12\n\n20\n\n7\n\n7 10\n\n16\n\n0\n\n14\n\n19\n\n32\n\n40\n\n7\n\n15\n\n4\n\n20\n\nInterest rates and their impact on the economy\n\nUS presidential elections/geopolitical uncertainty\n\nRise of AI\n\nEffects of hybrid work models on our organization\n\nIncreasing threat of cyberattacks\n\nRise of industrial automation\n\nRise of green resources and sustainable technology\n\nCommercial real estate market\n\nOther\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=142\n\nStrategy consulting Management/advisory consulting Technology/IT consulting Other\n\n12 0\n\n13\n\n2\n\n**Key external factors influencing outlook (%)**\n\nDiverse region, inclusive workforces: Diversity and inclusion policy and practice in Asia Pacific **HEIDRICK & STRUGGLES**\n\n**19**","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/US_Professional_Services_Partner_Compensation_Survey_2024_p19.pdf"]},"response":{"errors":[],"extract_id":"extract_9db764809f2bd1b3487b44e04f87c40b","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["When compared to respondents at other types of firms, those at management consulting firms least often cited AI as a key external factor—unsurprising, given that management consulting is still in the early stages of planning for and implementing AI technologies into their products, services, and processes.\n\nWe also find it unsurprising that respondents at strategy consulting firms reported the most concern about the US presidential election and geopolitical issues, given that much of work is at the board level. It makes sense that these issues are top of mind for board members.\n\nFor respondents in IT consulting, it also makes sense that there was slightly less overall focus on interest rates and their impact on the economy, given that their services are in many ways a commodity for organizations and are therefore less likely to be as significantly impacted by economic shifts.\n\n**Key external factors influencing outlook for current fiscal year, by firm type (Select all that apply)**\n\n65\n\n51\n\n60\n\n40\n\n53\n\n44\n\n60\n\n67\n\n81 81\n\n72\n\n80\n\n12\n\n8\n\n4 0\n\n5\n\n12\n\n20\n\n7\n\n7 10\n\n16\n\n0\n\n14\n\n19\n\n32\n\n40\n\n7\n\n15\n\n4\n\n20\n\nInterest rates and their impact on the economy\n\nUS presidential elections/geopolitical uncertainty\n\nRise of AI\n\nEffects of hybrid work models on our organization\n\nIncreasing threat of cyberattacks\n\nRise of industrial automation\n\nRise of green resources and sustainable technology\n\nCommercial real estate market\n\nOther\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=142\n\nStrategy consulting Management/advisory consulting Technology/IT consulting Other\n\n12 0\n\n13\n\n2\n\n**Key external factors influencing outlook (%)**\n\nDiverse region, inclusive workforces: Diversity and inclusion policy and practice in Asia Pacific **HEIDRICK & STRUGGLES**\n\n**19**"],"full_content":"When compared to respondents at other types of firms, those at management consulting firms least often cited AI as a key external factor—unsurprising, given that management consulting is still in the early stages of planning for and implementing AI technologies into their products, services, and processes.\n\nWe also find it unsurprising that respondents at strategy consulting firms reported the most concern about the US presidential election and geopolitical issues, given that much of work is at the board level. It makes sense that these issues are top of mind for board members.\n\nFor respondents in IT consulting, it also makes sense that there was slightly less overall focus on interest rates and their impact on the economy, given that their services are in many ways a commodity for organizations and are therefore less likely to be as significantly impacted by economic shifts.\n\n**Key external factors influencing outlook for current fiscal year, by firm type (Select all that apply)**\n\n65\n\n51\n\n60\n\n40\n\n53\n\n44\n\n60\n\n67\n\n81 81\n\n72\n\n80\n\n12\n\n8\n\n4 0\n\n5\n\n12\n\n20\n\n7\n\n7 10\n\n16\n\n0\n\n14\n\n19\n\n32\n\n40\n\n7\n\n15\n\n4\n\n20\n\nInterest rates and their impact on the economy\n\nUS presidential elections/geopolitical uncertainty\n\nRise of AI\n\nEffects of hybrid work models on our organization\n\nIncreasing threat of cyberattacks\n\nRise of industrial automation\n\nRise of green resources and sustainable technology\n\nCommercial real estate market\n\nOther\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=142\n\nStrategy consulting Management/advisory consulting Technology/IT consulting Other\n\n12 0\n\n13\n\n2\n\n**Key external factors influencing outlook (%)**\n\nDiverse region, inclusive workforces: Diversity and inclusion policy and practice in Asia Pacific **HEIDRICK & STRUGGLES**\n\n**19**","publish_date":null,"title":"Key external factors influencing outlook (%)","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/US_Professional_Services_Partner_Compensation_Survey_2024_p19.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/US_Professional_Services_Partner_Compensation_Survey_2024_p19.pdf"},"request":{"example_id":"chart/US_Professional_Services_Partner_Compensation_Survey_2024_p19","product_type":"parse","source_file_path":"US_Professional_Services_Partner_Compensation_Survey_2024_p19.pdf"},"started_at":"2026-11-04T19:45:32.281720"} +{"completed_at":"2026-11-04T19:45:36.277528","latency_in_ms":2858,"output":{"example_id":"chart/US_Professional_Services_Partner_Compensation_Survey_2024_p20","grounded_pages":[],"layout_pages":[],"markdown":"By firm type, technology and IT firms were least often optimistic—at only 28%—and by far the most often neutral.\n\nRespondents from management and advisory consulting firms were the most split—with a higher share of optimistic respondents than those at both strategy and technology and IT firms, and with the highest share of pessimists: one- third said they are pessimistic, which is 11 percentage points or more higher than respondents at any other firm type.\n\nBy firm revenue, respondents at smaller firms, those with between $1 billion and $5 billion in revenue, were the most optimistic—83% reported feeling optimistic, 22 percentage points higher than the next closest group, the smallest firms, and more than double the share of respondents at firms with $5 billion to $20 billion in revenue.\n\nRespondents at the largest firms, those with $20 billion or more in revenue, were least often optimistic, and most often neutral.\n\nNote: Numbers may not sum to 100%, because of rounding.\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=142\n\nLess than $1.00bn\n\n$5.01bn–$20.00bn\n\nMore than $20.00bn\n\n$1.00bn–$5.00bn\n\n18 61 21\n\n31 35 35\n\n36 26 38\n\n9 83 9\n\n**Key external factors influencing outlook for current fiscal year, by firm type**\n\n**Key external factors influencing outlook for current fiscal year, by firm revenue**\n\nPessimistic Neutral\n\nNote: Numbers may not sum to 100%, because of rounding.\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=142\n\nStrategy consulting\n\nTechnology/IT consulting\n\nOther\n\nManagement/ advisory consulting\n\n23 41 36\n\n20 28 52\n\n19 69 13\n\n34 46 20\n\nPessimistic Neutral Optimistic\n\nOptimistic\n\n**Key external factors influencing outlook (%)**\n\n**2024 US PROFESSIONAL SERVICES PARTNER COMPENSATION SURVEY**\n\n**20**","pages":[{"line_numbers":[],"markdown":"By firm type, technology and IT firms were least often optimistic—at only 28%—and by far the most often neutral.\n\nRespondents from management and advisory consulting firms were the most split—with a higher share of optimistic respondents than those at both strategy and technology and IT firms, and with the highest share of pessimists: one- third said they are pessimistic, which is 11 percentage points or more higher than respondents at any other firm type.\n\nBy firm revenue, respondents at smaller firms, those with between $1 billion and $5 billion in revenue, were the most optimistic—83% reported feeling optimistic, 22 percentage points higher than the next closest group, the smallest firms, and more than double the share of respondents at firms with $5 billion to $20 billion in revenue.\n\nRespondents at the largest firms, those with $20 billion or more in revenue, were least often optimistic, and most often neutral.\n\nNote: Numbers may not sum to 100%, because of rounding.\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=142\n\nLess than $1.00bn\n\n$5.01bn–$20.00bn\n\nMore than $20.00bn\n\n$1.00bn–$5.00bn\n\n18 61 21\n\n31 35 35\n\n36 26 38\n\n9 83 9\n\n**Key external factors influencing outlook for current fiscal year, by firm type**\n\n**Key external factors influencing outlook for current fiscal year, by firm revenue**\n\nPessimistic Neutral\n\nNote: Numbers may not sum to 100%, because of rounding.\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=142\n\nStrategy consulting\n\nTechnology/IT consulting\n\nOther\n\nManagement/ advisory consulting\n\n23 41 36\n\n20 28 52\n\n19 69 13\n\n34 46 20\n\nPessimistic Neutral Optimistic\n\nOptimistic\n\n**Key external factors influencing outlook (%)**\n\n**2024 US PROFESSIONAL SERVICES PARTNER COMPENSATION SURVEY**\n\n**20**","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"By firm type, technology and IT firms were least often optimistic—at only 28%—and by far the most often neutral.\n\nRespondents from management and advisory consulting firms were the most split—with a higher share of optimistic respondents than those at both strategy and technology and IT firms, and with the highest share of pessimists: one- third said they are pessimistic, which is 11 percentage points or more higher than respondents at any other firm type.\n\nBy firm revenue, respondents at smaller firms, those with between $1 billion and $5 billion in revenue, were the most optimistic—83% reported feeling optimistic, 22 percentage points higher than the next closest group, the smallest firms, and more than double the share of respondents at firms with $5 billion to $20 billion in revenue.\n\nRespondents at the largest firms, those with $20 billion or more in revenue, were least often optimistic, and most often neutral.\n\nNote: Numbers may not sum to 100%, because of rounding.\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=142\n\nLess than $1.00bn\n\n$5.01bn–$20.00bn\n\nMore than $20.00bn\n\n$1.00bn–$5.00bn\n\n18 61 21\n\n31 35 35\n\n36 26 38\n\n9 83 9\n\n**Key external factors influencing outlook for current fiscal year, by firm type**\n\n**Key external factors influencing outlook for current fiscal year, by firm revenue**\n\nPessimistic Neutral\n\nNote: Numbers may not sum to 100%, because of rounding.\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=142\n\nStrategy consulting\n\nTechnology/IT consulting\n\nOther\n\nManagement/ advisory consulting\n\n23 41 36\n\n20 28 52\n\n19 69 13\n\n34 46 20\n\nPessimistic Neutral Optimistic\n\nOptimistic\n\n**Key external factors influencing outlook (%)**\n\n**2024 US PROFESSIONAL SERVICES PARTNER COMPENSATION SURVEY**\n\n**20**","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/US_Professional_Services_Partner_Compensation_Survey_2024_p20.pdf"]},"response":{"errors":[],"extract_id":"extract_1d15e60693ee66fc2a4819733c8ea466","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["By firm type, technology and IT firms were least often optimistic—at only 28%—and by far the most often neutral.\n\nRespondents from management and advisory consulting firms were the most split—with a higher share of optimistic respondents than those at both strategy and technology and IT firms, and with the highest share of pessimists: one- third said they are pessimistic, which is 11 percentage points or more higher than respondents at any other firm type.\n\nBy firm revenue, respondents at smaller firms, those with between $1 billion and $5 billion in revenue, were the most optimistic—83% reported feeling optimistic, 22 percentage points higher than the next closest group, the smallest firms, and more than double the share of respondents at firms with $5 billion to $20 billion in revenue.\n\nRespondents at the largest firms, those with $20 billion or more in revenue, were least often optimistic, and most often neutral.\n\nNote: Numbers may not sum to 100%, because of rounding.\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=142\n\nLess than $1.00bn\n\n$5.01bn–$20.00bn\n\nMore than $20.00bn\n\n$1.00bn–$5.00bn\n\n18 61 21\n\n31 35 35\n\n36 26 38\n\n9 83 9\n\n**Key external factors influencing outlook for current fiscal year, by firm type**\n\n**Key external factors influencing outlook for current fiscal year, by firm revenue**\n\nPessimistic Neutral\n\nNote: Numbers may not sum to 100%, because of rounding.\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=142\n\nStrategy consulting\n\nTechnology/IT consulting\n\nOther\n\nManagement/ advisory consulting\n\n23 41 36\n\n20 28 52\n\n19 69 13\n\n34 46 20\n\nPessimistic Neutral Optimistic\n\nOptimistic\n\n**Key external factors influencing outlook (%)**\n\n**2024 US PROFESSIONAL SERVICES PARTNER COMPENSATION SURVEY**\n\n**20**"],"full_content":"By firm type, technology and IT firms were least often optimistic—at only 28%—and by far the most often neutral.\n\nRespondents from management and advisory consulting firms were the most split—with a higher share of optimistic respondents than those at both strategy and technology and IT firms, and with the highest share of pessimists: one- third said they are pessimistic, which is 11 percentage points or more higher than respondents at any other firm type.\n\nBy firm revenue, respondents at smaller firms, those with between $1 billion and $5 billion in revenue, were the most optimistic—83% reported feeling optimistic, 22 percentage points higher than the next closest group, the smallest firms, and more than double the share of respondents at firms with $5 billion to $20 billion in revenue.\n\nRespondents at the largest firms, those with $20 billion or more in revenue, were least often optimistic, and most often neutral.\n\nNote: Numbers may not sum to 100%, because of rounding.\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=142\n\nLess than $1.00bn\n\n$5.01bn–$20.00bn\n\nMore than $20.00bn\n\n$1.00bn–$5.00bn\n\n18 61 21\n\n31 35 35\n\n36 26 38\n\n9 83 9\n\n**Key external factors influencing outlook for current fiscal year, by firm type**\n\n**Key external factors influencing outlook for current fiscal year, by firm revenue**\n\nPessimistic Neutral\n\nNote: Numbers may not sum to 100%, because of rounding.\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=142\n\nStrategy consulting\n\nTechnology/IT consulting\n\nOther\n\nManagement/ advisory consulting\n\n23 41 36\n\n20 28 52\n\n19 69 13\n\n34 46 20\n\nPessimistic Neutral Optimistic\n\nOptimistic\n\n**Key external factors influencing outlook (%)**\n\n**2024 US PROFESSIONAL SERVICES PARTNER COMPENSATION SURVEY**\n\n**20**","publish_date":null,"title":"Key external factors influencing outlook (%)","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/US_Professional_Services_Partner_Compensation_Survey_2024_p20.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/US_Professional_Services_Partner_Compensation_Survey_2024_p20.pdf"},"request":{"example_id":"chart/US_Professional_Services_Partner_Compensation_Survey_2024_p20","product_type":"parse","source_file_path":"US_Professional_Services_Partner_Compensation_Survey_2024_p20.pdf"},"started_at":"2026-11-04T19:45:33.418848"} +{"completed_at":"2026-11-04T19:45:37.653540","latency_in_ms":2955,"output":{"example_id":"chart/US_Professional_Services_Partner_Compensation_Survey_2024_p21","grounded_pages":[],"layout_pages":[],"markdown":"## Average total cash compensation\n\nOverall 2023 average total cash compensation for partners at professional services firms was $672,000. Despite almost a quarter being pessimistic (see chart at the top of page 20), partners at strategy consulting firms saw the highest total cash compensation, almost double the average total cash compensation of their peers in management consulting.\n\nUnsurprisingly, total cash compensation rose by tenure as a partner, though partners with current firm tenure at between three and five years saw the lowest total cash compensation.\n\nAverage total cash compensation was relatively steady across gender and racial or ethnic groups—with white people and men only slightly outearning their lesser-represented counterparts.\n\n**Average total cash compensation ($ thousands)**\n\n# **State of professional services partner compensation**\n\nOverall (n= 155)\n\nStrategy consulting (n= 47)\n\nManagement/advisory consulting (n= 64)\n\nTechnology/IT consulting (n= 27)\n\nOther (n= 17)\n\nLess than $1.00bn (n= 37)\n\n$1.00bn–$5.00bn (n= 23)\n\n$5.01bn–$20.00bn (n= 29)\n\nMore than $20.00bn (n= 61)\n\nLess than 6 years (n= 34)\n\n6–10 years (n= 56)\n\n11–15 years (n= 26)\n\nMore than 15 years (n= 31)\n\nLess than 3 years (n= 37)\n\n3–5 years (n= 43)\n\n6–10 years (n= 30)\n\nMore than 10 years (n= 42)\n\nLess than 2 years (n= 28)\n\n2–3 years (n= 49)\n\n4–5 years (n= 32)\n\nMore than 5 years (n= 41)\n\nMen (n= 119)\n\nWomen (n= 19)\n\nRacial or ethnic minority (n= 39)\n\nWhite (n= 116)\n\nOverall Firm type Firm revenue Tenure as partner Tenure at firm Tenure in current role Gender Race/ ethnicity\n\n1,069\n\n421\n\n415\n\n555\n\n521\n\n598\n\n868\n\n920\n\n449\n\n1,039\n\n776\n\n650\n\n656\n\n936\n\n641\n\n472\n\n651\n\n908\n\n727\n\n497\n\n688\n\n622\n\n646\n\n638\n\n672\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024\n\nDiverse region, inclusive workforces: Diversity and inclusion policy and practice in Asia Pacific **HEIDRICK & STRUGGLES**\n\n**21**","pages":[{"line_numbers":[],"markdown":"## Average total cash compensation\n\nOverall 2023 average total cash compensation for partners at professional services firms was $672,000. Despite almost a quarter being pessimistic (see chart at the top of page 20), partners at strategy consulting firms saw the highest total cash compensation, almost double the average total cash compensation of their peers in management consulting.\n\nUnsurprisingly, total cash compensation rose by tenure as a partner, though partners with current firm tenure at between three and five years saw the lowest total cash compensation.\n\nAverage total cash compensation was relatively steady across gender and racial or ethnic groups—with white people and men only slightly outearning their lesser-represented counterparts.\n\n**Average total cash compensation ($ thousands)**\n\n# **State of professional services partner compensation**\n\nOverall (n= 155)\n\nStrategy consulting (n= 47)\n\nManagement/advisory consulting (n= 64)\n\nTechnology/IT consulting (n= 27)\n\nOther (n= 17)\n\nLess than $1.00bn (n= 37)\n\n$1.00bn–$5.00bn (n= 23)\n\n$5.01bn–$20.00bn (n= 29)\n\nMore than $20.00bn (n= 61)\n\nLess than 6 years (n= 34)\n\n6–10 years (n= 56)\n\n11–15 years (n= 26)\n\nMore than 15 years (n= 31)\n\nLess than 3 years (n= 37)\n\n3–5 years (n= 43)\n\n6–10 years (n= 30)\n\nMore than 10 years (n= 42)\n\nLess than 2 years (n= 28)\n\n2–3 years (n= 49)\n\n4–5 years (n= 32)\n\nMore than 5 years (n= 41)\n\nMen (n= 119)\n\nWomen (n= 19)\n\nRacial or ethnic minority (n= 39)\n\nWhite (n= 116)\n\nOverall Firm type Firm revenue Tenure as partner Tenure at firm Tenure in current role Gender Race/ ethnicity\n\n1,069\n\n421\n\n415\n\n555\n\n521\n\n598\n\n868\n\n920\n\n449\n\n1,039\n\n776\n\n650\n\n656\n\n936\n\n641\n\n472\n\n651\n\n908\n\n727\n\n497\n\n688\n\n622\n\n646\n\n638\n\n672\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024\n\nDiverse region, inclusive workforces: Diversity and inclusion policy and practice in Asia Pacific **HEIDRICK & STRUGGLES**\n\n**21**","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"## Average total cash compensation\n\nOverall 2023 average total cash compensation for partners at professional services firms was $672,000. Despite almost a quarter being pessimistic (see chart at the top of page 20), partners at strategy consulting firms saw the highest total cash compensation, almost double the average total cash compensation of their peers in management consulting.\n\nUnsurprisingly, total cash compensation rose by tenure as a partner, though partners with current firm tenure at between three and five years saw the lowest total cash compensation.\n\nAverage total cash compensation was relatively steady across gender and racial or ethnic groups—with white people and men only slightly outearning their lesser-represented counterparts.\n\n**Average total cash compensation ($ thousands)**\n\n# **State of professional services partner compensation**\n\nOverall (n= 155)\n\nStrategy consulting (n= 47)\n\nManagement/advisory consulting (n= 64)\n\nTechnology/IT consulting (n= 27)\n\nOther (n= 17)\n\nLess than $1.00bn (n= 37)\n\n$1.00bn–$5.00bn (n= 23)\n\n$5.01bn–$20.00bn (n= 29)\n\nMore than $20.00bn (n= 61)\n\nLess than 6 years (n= 34)\n\n6–10 years (n= 56)\n\n11–15 years (n= 26)\n\nMore than 15 years (n= 31)\n\nLess than 3 years (n= 37)\n\n3–5 years (n= 43)\n\n6–10 years (n= 30)\n\nMore than 10 years (n= 42)\n\nLess than 2 years (n= 28)\n\n2–3 years (n= 49)\n\n4–5 years (n= 32)\n\nMore than 5 years (n= 41)\n\nMen (n= 119)\n\nWomen (n= 19)\n\nRacial or ethnic minority (n= 39)\n\nWhite (n= 116)\n\nOverall Firm type Firm revenue Tenure as partner Tenure at firm Tenure in current role Gender Race/ ethnicity\n\n1,069\n\n421\n\n415\n\n555\n\n521\n\n598\n\n868\n\n920\n\n449\n\n1,039\n\n776\n\n650\n\n656\n\n936\n\n641\n\n472\n\n651\n\n908\n\n727\n\n497\n\n688\n\n622\n\n646\n\n638\n\n672\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024\n\nDiverse region, inclusive workforces: Diversity and inclusion policy and practice in Asia Pacific **HEIDRICK & STRUGGLES**\n\n**21**","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/US_Professional_Services_Partner_Compensation_Survey_2024_p21.pdf"]},"response":{"errors":[],"extract_id":"extract_23919e20fa9ed4956cc6cb7e5a9966b3","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["## Average total cash compensation\n\nOverall 2023 average total cash compensation for partners at professional services firms was $672,000. Despite almost a quarter being pessimistic (see chart at the top of page 20), partners at strategy consulting firms saw the highest total cash compensation, almost double the average total cash compensation of their peers in management consulting.\n\nUnsurprisingly, total cash compensation rose by tenure as a partner, though partners with current firm tenure at between three and five years saw the lowest total cash compensation.\n\nAverage total cash compensation was relatively steady across gender and racial or ethnic groups—with white people and men only slightly outearning their lesser-represented counterparts.\n\n**Average total cash compensation ($ thousands)**\n\n# **State of professional services partner compensation**\n\nOverall (n= 155)\n\nStrategy consulting (n= 47)\n\nManagement/advisory consulting (n= 64)\n\nTechnology/IT consulting (n= 27)\n\nOther (n= 17)\n\nLess than $1.00bn (n= 37)\n\n$1.00bn–$5.00bn (n= 23)\n\n$5.01bn–$20.00bn (n= 29)\n\nMore than $20.00bn (n= 61)\n\nLess than 6 years (n= 34)\n\n6–10 years (n= 56)\n\n11–15 years (n= 26)\n\nMore than 15 years (n= 31)\n\nLess than 3 years (n= 37)\n\n3–5 years (n= 43)\n\n6–10 years (n= 30)\n\nMore than 10 years (n= 42)\n\nLess than 2 years (n= 28)\n\n2–3 years (n= 49)\n\n4–5 years (n= 32)\n\nMore than 5 years (n= 41)\n\nMen (n= 119)\n\nWomen (n= 19)\n\nRacial or ethnic minority (n= 39)\n\nWhite (n= 116)\n\nOverall Firm type Firm revenue Tenure as partner Tenure at firm Tenure in current role Gender Race/ ethnicity\n\n1,069\n\n421\n\n415\n\n555\n\n521\n\n598\n\n868\n\n920\n\n449\n\n1,039\n\n776\n\n650\n\n656\n\n936\n\n641\n\n472\n\n651\n\n908\n\n727\n\n497\n\n688\n\n622\n\n646\n\n638\n\n672\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024\n\nDiverse region, inclusive workforces: Diversity and inclusion policy and practice in Asia Pacific **HEIDRICK & STRUGGLES**\n\n**21**"],"full_content":"## Average total cash compensation\n\nOverall 2023 average total cash compensation for partners at professional services firms was $672,000. Despite almost a quarter being pessimistic (see chart at the top of page 20), partners at strategy consulting firms saw the highest total cash compensation, almost double the average total cash compensation of their peers in management consulting.\n\nUnsurprisingly, total cash compensation rose by tenure as a partner, though partners with current firm tenure at between three and five years saw the lowest total cash compensation.\n\nAverage total cash compensation was relatively steady across gender and racial or ethnic groups—with white people and men only slightly outearning their lesser-represented counterparts.\n\n**Average total cash compensation ($ thousands)**\n\n# **State of professional services partner compensation**\n\nOverall (n= 155)\n\nStrategy consulting (n= 47)\n\nManagement/advisory consulting (n= 64)\n\nTechnology/IT consulting (n= 27)\n\nOther (n= 17)\n\nLess than $1.00bn (n= 37)\n\n$1.00bn–$5.00bn (n= 23)\n\n$5.01bn–$20.00bn (n= 29)\n\nMore than $20.00bn (n= 61)\n\nLess than 6 years (n= 34)\n\n6–10 years (n= 56)\n\n11–15 years (n= 26)\n\nMore than 15 years (n= 31)\n\nLess than 3 years (n= 37)\n\n3–5 years (n= 43)\n\n6–10 years (n= 30)\n\nMore than 10 years (n= 42)\n\nLess than 2 years (n= 28)\n\n2–3 years (n= 49)\n\n4–5 years (n= 32)\n\nMore than 5 years (n= 41)\n\nMen (n= 119)\n\nWomen (n= 19)\n\nRacial or ethnic minority (n= 39)\n\nWhite (n= 116)\n\nOverall Firm type Firm revenue Tenure as partner Tenure at firm Tenure in current role Gender Race/ ethnicity\n\n1,069\n\n421\n\n415\n\n555\n\n521\n\n598\n\n868\n\n920\n\n449\n\n1,039\n\n776\n\n650\n\n656\n\n936\n\n641\n\n472\n\n651\n\n908\n\n727\n\n497\n\n688\n\n622\n\n646\n\n638\n\n672\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024\n\nDiverse region, inclusive workforces: Diversity and inclusion policy and practice in Asia Pacific **HEIDRICK & STRUGGLES**\n\n**21**","publish_date":null,"title":"State of professional services partner compensation","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/US_Professional_Services_Partner_Compensation_Survey_2024_p21.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/US_Professional_Services_Partner_Compensation_Survey_2024_p21.pdf"},"request":{"example_id":"chart/US_Professional_Services_Partner_Compensation_Survey_2024_p21","product_type":"parse","source_file_path":"US_Professional_Services_Partner_Compensation_Survey_2024_p21.pdf"},"started_at":"2026-11-04T19:45:34.698210"} +{"completed_at":"2026-11-04T19:45:37.996048","latency_in_ms":1872,"output":{"example_id":"chart/US_Professional_Services_Partner_Compensation_Survey_2024_p22","grounded_pages":[],"layout_pages":[],"markdown":"By firm type, though respondents at strategy consulting firms saw the highest average total cash compensation, and most often reported total cash compensation of more than $1 million, respondents at management and advisory firms saw the largest increase in total cash compensation from our last survey; 49% reported total cash compensation of more than $1 million in our 2024 survey, an 8 percentage- point increase from the 2022 survey.\n\n**Total cash compensation, by firm type (%)**\n\n**Strategy Management/advisory Technology/IT**\n\n5\n\n13\n\n22\n\n9\n\n12\n\n30 $400,000 or less\n\n$401,000–$600,000\n\n$601,000–$800,000\n\n$801,000–$1.00m\n\n$1.01m–$2.00m\n\n$2.01m–$3.00m\n\nMore than $3.00m\n\n16\n\n15\n\n10\n\n23\n\n20\n\n22\n\n13\n\n13\n\n10\n\n9\n\n12\n\n7\n\n12\n\n0\n\n18\n\n9\n\n27\n\n4\n\n19\n\n21\n\n12 8\n\n4\n\n29\n\n32\n\n29\n\n45\n\n22\n\n33\n\n6\n\n6\n\n0\n\n2\n\n2\n\n0\n\n0\n\nNote: Numbers may not sum to 100%, because of rounding.\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=155\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2022, n=192\n\n2024 2022\n\n**2024 US PROFESSIONAL SERVICES PARTNER COMPENSATION SURVEY**\n\n**22**","pages":[{"line_numbers":[],"markdown":"By firm type, though respondents at strategy consulting firms saw the highest average total cash compensation, and most often reported total cash compensation of more than $1 million, respondents at management and advisory firms saw the largest increase in total cash compensation from our last survey; 49% reported total cash compensation of more than $1 million in our 2024 survey, an 8 percentage- point increase from the 2022 survey.\n\n**Total cash compensation, by firm type (%)**\n\n**Strategy Management/advisory Technology/IT**\n\n5\n\n13\n\n22\n\n9\n\n12\n\n30 $400,000 or less\n\n$401,000–$600,000\n\n$601,000–$800,000\n\n$801,000–$1.00m\n\n$1.01m–$2.00m\n\n$2.01m–$3.00m\n\nMore than $3.00m\n\n16\n\n15\n\n10\n\n23\n\n20\n\n22\n\n13\n\n13\n\n10\n\n9\n\n12\n\n7\n\n12\n\n0\n\n18\n\n9\n\n27\n\n4\n\n19\n\n21\n\n12 8\n\n4\n\n29\n\n32\n\n29\n\n45\n\n22\n\n33\n\n6\n\n6\n\n0\n\n2\n\n2\n\n0\n\n0\n\nNote: Numbers may not sum to 100%, because of rounding.\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=155\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2022, n=192\n\n2024 2022\n\n**2024 US PROFESSIONAL SERVICES PARTNER COMPENSATION SURVEY**\n\n**22**","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"By firm type, though respondents at strategy consulting firms saw the highest average total cash compensation, and most often reported total cash compensation of more than $1 million, respondents at management and advisory firms saw the largest increase in total cash compensation from our last survey; 49% reported total cash compensation of more than $1 million in our 2024 survey, an 8 percentage- point increase from the 2022 survey.\n\n**Total cash compensation, by firm type (%)**\n\n**Strategy Management/advisory Technology/IT**\n\n5\n\n13\n\n22\n\n9\n\n12\n\n30 $400,000 or less\n\n$401,000–$600,000\n\n$601,000–$800,000\n\n$801,000–$1.00m\n\n$1.01m–$2.00m\n\n$2.01m–$3.00m\n\nMore than $3.00m\n\n16\n\n15\n\n10\n\n23\n\n20\n\n22\n\n13\n\n13\n\n10\n\n9\n\n12\n\n7\n\n12\n\n0\n\n18\n\n9\n\n27\n\n4\n\n19\n\n21\n\n12 8\n\n4\n\n29\n\n32\n\n29\n\n45\n\n22\n\n33\n\n6\n\n6\n\n0\n\n2\n\n2\n\n0\n\n0\n\nNote: Numbers may not sum to 100%, because of rounding.\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=155\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2022, n=192\n\n2024 2022\n\n**2024 US PROFESSIONAL SERVICES PARTNER COMPENSATION SURVEY**\n\n**22**","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/US_Professional_Services_Partner_Compensation_Survey_2024_p22.pdf"]},"response":{"errors":[],"extract_id":"extract_33dee81d0256e8e36b6e3220e34f1d0a","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["By firm type, though respondents at strategy consulting firms saw the highest average total cash compensation, and most often reported total cash compensation of more than $1 million, respondents at management and advisory firms saw the largest increase in total cash compensation from our last survey; 49% reported total cash compensation of more than $1 million in our 2024 survey, an 8 percentage- point increase from the 2022 survey.\n\n**Total cash compensation, by firm type (%)**\n\n**Strategy Management/advisory Technology/IT**\n\n5\n\n13\n\n22\n\n9\n\n12\n\n30 $400,000 or less\n\n$401,000–$600,000\n\n$601,000–$800,000\n\n$801,000–$1.00m\n\n$1.01m–$2.00m\n\n$2.01m–$3.00m\n\nMore than $3.00m\n\n16\n\n15\n\n10\n\n23\n\n20\n\n22\n\n13\n\n13\n\n10\n\n9\n\n12\n\n7\n\n12\n\n0\n\n18\n\n9\n\n27\n\n4\n\n19\n\n21\n\n12 8\n\n4\n\n29\n\n32\n\n29\n\n45\n\n22\n\n33\n\n6\n\n6\n\n0\n\n2\n\n2\n\n0\n\n0\n\nNote: Numbers may not sum to 100%, because of rounding.\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=155\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2022, n=192\n\n2024 2022\n\n**2024 US PROFESSIONAL SERVICES PARTNER COMPENSATION SURVEY**\n\n**22**"],"full_content":"By firm type, though respondents at strategy consulting firms saw the highest average total cash compensation, and most often reported total cash compensation of more than $1 million, respondents at management and advisory firms saw the largest increase in total cash compensation from our last survey; 49% reported total cash compensation of more than $1 million in our 2024 survey, an 8 percentage- point increase from the 2022 survey.\n\n**Total cash compensation, by firm type (%)**\n\n**Strategy Management/advisory Technology/IT**\n\n5\n\n13\n\n22\n\n9\n\n12\n\n30 $400,000 or less\n\n$401,000–$600,000\n\n$601,000–$800,000\n\n$801,000–$1.00m\n\n$1.01m–$2.00m\n\n$2.01m–$3.00m\n\nMore than $3.00m\n\n16\n\n15\n\n10\n\n23\n\n20\n\n22\n\n13\n\n13\n\n10\n\n9\n\n12\n\n7\n\n12\n\n0\n\n18\n\n9\n\n27\n\n4\n\n19\n\n21\n\n12 8\n\n4\n\n29\n\n32\n\n29\n\n45\n\n22\n\n33\n\n6\n\n6\n\n0\n\n2\n\n2\n\n0\n\n0\n\nNote: Numbers may not sum to 100%, because of rounding.\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=155\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2022, n=192\n\n2024 2022\n\n**2024 US PROFESSIONAL SERVICES PARTNER COMPENSATION SURVEY**\n\n**22**","publish_date":null,"title":"Total cash compensation, by firm type (%)","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/US_Professional_Services_Partner_Compensation_Survey_2024_p22.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/US_Professional_Services_Partner_Compensation_Survey_2024_p22.pdf"},"request":{"example_id":"chart/US_Professional_Services_Partner_Compensation_Survey_2024_p22","product_type":"parse","source_file_path":"US_Professional_Services_Partner_Compensation_Survey_2024_p22.pdf"},"started_at":"2026-11-04T19:45:36.123442"} +{"completed_at":"2026-11-04T19:45:39.601772","latency_in_ms":2236,"output":{"example_id":"chart/US_Professional_Services_Partner_Compensation_Survey_2024_p24","grounded_pages":[],"layout_pages":[],"markdown":"**Total cash compensation, by tenure as partner (%)**\n\n**Less than 6 years 6–10 years More than 10 years**\n\n$400,000 or less\n\n$401,000–$600,000\n\n$601,000–$800,000\n\n$801,000–$1.00m\n\n$1.01m–$2.00m\n\n$2.01m–$3.00m\n\nMore than $3.00m\n\n15\n\n26\n\n17\n\n35\n\n15\n\n9\n\n21\n\n9\n\n29\n\n15\n\n2 2\n\n2\n\n2\n\n3\n\n3\n\n3\n\n4\n\n7\n\n16\n\n13\n\n12\n\n14\n\n4\n\n30\n\n16\n\n30\n\n46\n\n4\n\n4\n\n5\n\n11\n\n25\n\n16\n\n9\n\n16\n\n25\n\n4\n\n7\n\n7\n\n27\n\n45\n\nNote: Numbers may not sum to 100%, because of rounding.\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=155\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2022, n=192\n\n2024 2022\n\nBy tenure as a partner in any firm, total cash compensation is down for those with fewer years of tenure, but it has increased for those with longer partner tenure, with a notable exception: partners with more than 10 years’ tenure have a lower share of extremely high earners (that is, those earning more than $2 million) in 2024 compared to 2022.\n\n**2024 US PROFESSIONAL SERVICES PARTNER COMPENSATION SURVEY**\n\n**24**","pages":[{"line_numbers":[],"markdown":"**Total cash compensation, by tenure as partner (%)**\n\n**Less than 6 years 6–10 years More than 10 years**\n\n$400,000 or less\n\n$401,000–$600,000\n\n$601,000–$800,000\n\n$801,000–$1.00m\n\n$1.01m–$2.00m\n\n$2.01m–$3.00m\n\nMore than $3.00m\n\n15\n\n26\n\n17\n\n35\n\n15\n\n9\n\n21\n\n9\n\n29\n\n15\n\n2 2\n\n2\n\n2\n\n3\n\n3\n\n3\n\n4\n\n7\n\n16\n\n13\n\n12\n\n14\n\n4\n\n30\n\n16\n\n30\n\n46\n\n4\n\n4\n\n5\n\n11\n\n25\n\n16\n\n9\n\n16\n\n25\n\n4\n\n7\n\n7\n\n27\n\n45\n\nNote: Numbers may not sum to 100%, because of rounding.\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=155\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2022, n=192\n\n2024 2022\n\nBy tenure as a partner in any firm, total cash compensation is down for those with fewer years of tenure, but it has increased for those with longer partner tenure, with a notable exception: partners with more than 10 years’ tenure have a lower share of extremely high earners (that is, those earning more than $2 million) in 2024 compared to 2022.\n\n**2024 US PROFESSIONAL SERVICES PARTNER COMPENSATION SURVEY**\n\n**24**","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**Total cash compensation, by tenure as partner (%)**\n\n**Less than 6 years 6–10 years More than 10 years**\n\n$400,000 or less\n\n$401,000–$600,000\n\n$601,000–$800,000\n\n$801,000–$1.00m\n\n$1.01m–$2.00m\n\n$2.01m–$3.00m\n\nMore than $3.00m\n\n15\n\n26\n\n17\n\n35\n\n15\n\n9\n\n21\n\n9\n\n29\n\n15\n\n2 2\n\n2\n\n2\n\n3\n\n3\n\n3\n\n4\n\n7\n\n16\n\n13\n\n12\n\n14\n\n4\n\n30\n\n16\n\n30\n\n46\n\n4\n\n4\n\n5\n\n11\n\n25\n\n16\n\n9\n\n16\n\n25\n\n4\n\n7\n\n7\n\n27\n\n45\n\nNote: Numbers may not sum to 100%, because of rounding.\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=155\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2022, n=192\n\n2024 2022\n\nBy tenure as a partner in any firm, total cash compensation is down for those with fewer years of tenure, but it has increased for those with longer partner tenure, with a notable exception: partners with more than 10 years’ tenure have a lower share of extremely high earners (that is, those earning more than $2 million) in 2024 compared to 2022.\n\n**2024 US PROFESSIONAL SERVICES PARTNER COMPENSATION SURVEY**\n\n**24**","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/US_Professional_Services_Partner_Compensation_Survey_2024_p24.pdf"]},"response":{"errors":[],"extract_id":"extract_ff7dc5f5b8e6ae7c60d2f2fbdf4e8bf4","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**Total cash compensation, by tenure as partner (%)**\n\n**Less than 6 years 6–10 years More than 10 years**\n\n$400,000 or less\n\n$401,000–$600,000\n\n$601,000–$800,000\n\n$801,000–$1.00m\n\n$1.01m–$2.00m\n\n$2.01m–$3.00m\n\nMore than $3.00m\n\n15\n\n26\n\n17\n\n35\n\n15\n\n9\n\n21\n\n9\n\n29\n\n15\n\n2 2\n\n2\n\n2\n\n3\n\n3\n\n3\n\n4\n\n7\n\n16\n\n13\n\n12\n\n14\n\n4\n\n30\n\n16\n\n30\n\n46\n\n4\n\n4\n\n5\n\n11\n\n25\n\n16\n\n9\n\n16\n\n25\n\n4\n\n7\n\n7\n\n27\n\n45\n\nNote: Numbers may not sum to 100%, because of rounding.\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=155\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2022, n=192\n\n2024 2022\n\nBy tenure as a partner in any firm, total cash compensation is down for those with fewer years of tenure, but it has increased for those with longer partner tenure, with a notable exception: partners with more than 10 years’ tenure have a lower share of extremely high earners (that is, those earning more than $2 million) in 2024 compared to 2022.\n\n**2024 US PROFESSIONAL SERVICES PARTNER COMPENSATION SURVEY**\n\n**24**"],"full_content":"**Total cash compensation, by tenure as partner (%)**\n\n**Less than 6 years 6–10 years More than 10 years**\n\n$400,000 or less\n\n$401,000–$600,000\n\n$601,000–$800,000\n\n$801,000–$1.00m\n\n$1.01m–$2.00m\n\n$2.01m–$3.00m\n\nMore than $3.00m\n\n15\n\n26\n\n17\n\n35\n\n15\n\n9\n\n21\n\n9\n\n29\n\n15\n\n2 2\n\n2\n\n2\n\n3\n\n3\n\n3\n\n4\n\n7\n\n16\n\n13\n\n12\n\n14\n\n4\n\n30\n\n16\n\n30\n\n46\n\n4\n\n4\n\n5\n\n11\n\n25\n\n16\n\n9\n\n16\n\n25\n\n4\n\n7\n\n7\n\n27\n\n45\n\nNote: Numbers may not sum to 100%, because of rounding.\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=155\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2022, n=192\n\n2024 2022\n\nBy tenure as a partner in any firm, total cash compensation is down for those with fewer years of tenure, but it has increased for those with longer partner tenure, with a notable exception: partners with more than 10 years’ tenure have a lower share of extremely high earners (that is, those earning more than $2 million) in 2024 compared to 2022.\n\n**2024 US PROFESSIONAL SERVICES PARTNER COMPENSATION SURVEY**\n\n**24**","publish_date":null,"title":"Total cash compensation, by tenure as partner (%)","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/US_Professional_Services_Partner_Compensation_Survey_2024_p24.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/US_Professional_Services_Partner_Compensation_Survey_2024_p24.pdf"},"request":{"example_id":"chart/US_Professional_Services_Partner_Compensation_Survey_2024_p24","product_type":"parse","source_file_path":"US_Professional_Services_Partner_Compensation_Survey_2024_p24.pdf"},"started_at":"2026-11-04T19:45:37.365618"} +{"completed_at":"2026-11-04T19:44:49.279306","latency_in_ms":3132,"output":{"example_id":"chart/US_Professional_Services_Partner_Compensation_Survey_2024_p25","grounded_pages":[],"layout_pages":[],"markdown":"**Total cash compensation, by tenure at current firm (%)**\n\n$400,000 or less\n\n$401,000–$600,000\n\n$601,000–$800,000\n\n$801,000–$1.00m\n\n$1.01m–$2.00m\n\n$2.01m–$3.00m\n\nMore than $3.00m\n\nNote: Numbers may not sum to 100%, because of rounding.\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=155\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2022, n=192\n\n**Less than 3 years**\n\n22\n\n30\n\n15\n\n16\n\n11\n\n16\n\n20\n\n6\n\n11\n\n26\n\n22\n\n0\n\n3\n\n3\n\n**3–5 years**\n\n3\n\n2\n\n2\n\n13\n\n12\n\n19\n\n33\n\n6\n\n16\n\n22\n\n6\n\n5\n\n31\n\n30\n\n**6–10 years More than 10 years**\n\n2\n\n5\n\n5\n\n4\n\n7\n\n13\n\n19\n\n15\n\n10\n\n15\n\n22\n\n17\n\n27\n\n40\n\n5\n\n7\n\n14\n\n13\n\n16\n\n7\n\n19\n\n13\n\n16\n\n27\n\n57\n\n3\n\n3\n\n0\n\n2024 2022\n\nBy tenure at current firm, total cash compensation has also, on the whole, decreased for those with fewer years of tenure and increased for those with six or more years of tenure. For longer-tenured partners, there were more respondents with compensation in the $2 million to $3 million range in 2022, whereas in 2024 there are more respondents in the $1 million to $2 million range. From this, we infer that this is likely a market-driven drop in performance bonuses.\n\nDiverse region, inclusive workforces: Diversity and inclusion policy and practice in Asia Pacific **HEIDRICK & STRUGGLES**\n\n**25**","pages":[{"line_numbers":[],"markdown":"**Total cash compensation, by tenure at current firm (%)**\n\n$400,000 or less\n\n$401,000–$600,000\n\n$601,000–$800,000\n\n$801,000–$1.00m\n\n$1.01m–$2.00m\n\n$2.01m–$3.00m\n\nMore than $3.00m\n\nNote: Numbers may not sum to 100%, because of rounding.\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=155\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2022, n=192\n\n**Less than 3 years**\n\n22\n\n30\n\n15\n\n16\n\n11\n\n16\n\n20\n\n6\n\n11\n\n26\n\n22\n\n0\n\n3\n\n3\n\n**3–5 years**\n\n3\n\n2\n\n2\n\n13\n\n12\n\n19\n\n33\n\n6\n\n16\n\n22\n\n6\n\n5\n\n31\n\n30\n\n**6–10 years More than 10 years**\n\n2\n\n5\n\n5\n\n4\n\n7\n\n13\n\n19\n\n15\n\n10\n\n15\n\n22\n\n17\n\n27\n\n40\n\n5\n\n7\n\n14\n\n13\n\n16\n\n7\n\n19\n\n13\n\n16\n\n27\n\n57\n\n3\n\n3\n\n0\n\n2024 2022\n\nBy tenure at current firm, total cash compensation has also, on the whole, decreased for those with fewer years of tenure and increased for those with six or more years of tenure. For longer-tenured partners, there were more respondents with compensation in the $2 million to $3 million range in 2022, whereas in 2024 there are more respondents in the $1 million to $2 million range. From this, we infer that this is likely a market-driven drop in performance bonuses.\n\nDiverse region, inclusive workforces: Diversity and inclusion policy and practice in Asia Pacific **HEIDRICK & STRUGGLES**\n\n**25**","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**Total cash compensation, by tenure at current firm (%)**\n\n$400,000 or less\n\n$401,000–$600,000\n\n$601,000–$800,000\n\n$801,000–$1.00m\n\n$1.01m–$2.00m\n\n$2.01m–$3.00m\n\nMore than $3.00m\n\nNote: Numbers may not sum to 100%, because of rounding.\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=155\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2022, n=192\n\n**Less than 3 years**\n\n22\n\n30\n\n15\n\n16\n\n11\n\n16\n\n20\n\n6\n\n11\n\n26\n\n22\n\n0\n\n3\n\n3\n\n**3–5 years**\n\n3\n\n2\n\n2\n\n13\n\n12\n\n19\n\n33\n\n6\n\n16\n\n22\n\n6\n\n5\n\n31\n\n30\n\n**6–10 years More than 10 years**\n\n2\n\n5\n\n5\n\n4\n\n7\n\n13\n\n19\n\n15\n\n10\n\n15\n\n22\n\n17\n\n27\n\n40\n\n5\n\n7\n\n14\n\n13\n\n16\n\n7\n\n19\n\n13\n\n16\n\n27\n\n57\n\n3\n\n3\n\n0\n\n2024 2022\n\nBy tenure at current firm, total cash compensation has also, on the whole, decreased for those with fewer years of tenure and increased for those with six or more years of tenure. For longer-tenured partners, there were more respondents with compensation in the $2 million to $3 million range in 2022, whereas in 2024 there are more respondents in the $1 million to $2 million range. From this, we infer that this is likely a market-driven drop in performance bonuses.\n\nDiverse region, inclusive workforces: Diversity and inclusion policy and practice in Asia Pacific **HEIDRICK & STRUGGLES**\n\n**25**","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/US_Professional_Services_Partner_Compensation_Survey_2024_p25.pdf"]},"response":{"errors":[],"extract_id":"extract_1c4274f0f4b201a51f0bf1f157b02b1c","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**Total cash compensation, by tenure at current firm (%)**\n\n$400,000 or less\n\n$401,000–$600,000\n\n$601,000–$800,000\n\n$801,000–$1.00m\n\n$1.01m–$2.00m\n\n$2.01m–$3.00m\n\nMore than $3.00m\n\nNote: Numbers may not sum to 100%, because of rounding.\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=155\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2022, n=192\n\n**Less than 3 years**\n\n22\n\n30\n\n15\n\n16\n\n11\n\n16\n\n20\n\n6\n\n11\n\n26\n\n22\n\n0\n\n3\n\n3\n\n**3–5 years**\n\n3\n\n2\n\n2\n\n13\n\n12\n\n19\n\n33\n\n6\n\n16\n\n22\n\n6\n\n5\n\n31\n\n30\n\n**6–10 years More than 10 years**\n\n2\n\n5\n\n5\n\n4\n\n7\n\n13\n\n19\n\n15\n\n10\n\n15\n\n22\n\n17\n\n27\n\n40\n\n5\n\n7\n\n14\n\n13\n\n16\n\n7\n\n19\n\n13\n\n16\n\n27\n\n57\n\n3\n\n3\n\n0\n\n2024 2022\n\nBy tenure at current firm, total cash compensation has also, on the whole, decreased for those with fewer years of tenure and increased for those with six or more years of tenure. For longer-tenured partners, there were more respondents with compensation in the $2 million to $3 million range in 2022, whereas in 2024 there are more respondents in the $1 million to $2 million range. From this, we infer that this is likely a market-driven drop in performance bonuses.\n\nDiverse region, inclusive workforces: Diversity and inclusion policy and practice in Asia Pacific **HEIDRICK & STRUGGLES**\n\n**25**"],"full_content":"**Total cash compensation, by tenure at current firm (%)**\n\n$400,000 or less\n\n$401,000–$600,000\n\n$601,000–$800,000\n\n$801,000–$1.00m\n\n$1.01m–$2.00m\n\n$2.01m–$3.00m\n\nMore than $3.00m\n\nNote: Numbers may not sum to 100%, because of rounding.\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=155\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2022, n=192\n\n**Less than 3 years**\n\n22\n\n30\n\n15\n\n16\n\n11\n\n16\n\n20\n\n6\n\n11\n\n26\n\n22\n\n0\n\n3\n\n3\n\n**3–5 years**\n\n3\n\n2\n\n2\n\n13\n\n12\n\n19\n\n33\n\n6\n\n16\n\n22\n\n6\n\n5\n\n31\n\n30\n\n**6–10 years More than 10 years**\n\n2\n\n5\n\n5\n\n4\n\n7\n\n13\n\n19\n\n15\n\n10\n\n15\n\n22\n\n17\n\n27\n\n40\n\n5\n\n7\n\n14\n\n13\n\n16\n\n7\n\n19\n\n13\n\n16\n\n27\n\n57\n\n3\n\n3\n\n0\n\n2024 2022\n\nBy tenure at current firm, total cash compensation has also, on the whole, decreased for those with fewer years of tenure and increased for those with six or more years of tenure. For longer-tenured partners, there were more respondents with compensation in the $2 million to $3 million range in 2022, whereas in 2024 there are more respondents in the $1 million to $2 million range. From this, we infer that this is likely a market-driven drop in performance bonuses.\n\nDiverse region, inclusive workforces: Diversity and inclusion policy and practice in Asia Pacific **HEIDRICK & STRUGGLES**\n\n**25**","publish_date":null,"title":"Total cash compensation, by tenure at current firm (%)","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/US_Professional_Services_Partner_Compensation_Survey_2024_p25.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/US_Professional_Services_Partner_Compensation_Survey_2024_p25.pdf"},"request":{"example_id":"chart/US_Professional_Services_Partner_Compensation_Survey_2024_p25","product_type":"parse","source_file_path":"US_Professional_Services_Partner_Compensation_Survey_2024_p25.pdf"},"started_at":"2026-11-04T19:44:46.146748"} +{"completed_at":"2026-11-04T19:44:48.740985","latency_in_ms":960,"output":{"example_id":"chart/US_Professional_Services_Partner_Compensation_Survey_2024_p26","grounded_pages":[],"layout_pages":[],"markdown":"**Total cash compensation, by gender (%)**\n\n$400,000 or less\n\n$401,000–$600,000\n\n$601,000–$800,000\n\n$801,000–$1.00m\n\n$1.01m–$2.00m\n\n$2.01m–$3.00m\n\nMore than $3.00m\n\nNote: Numbers may not sum to 100%, because of rounding.\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=155\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2022, n=192\n\n**Women**\n\n20\n\n11\n\n13\n\n37\n\n7\n\n5\n\n40\n\n7\n\n0\n\n13\n\n42\n\n0\n\n0\n\n5\n\n**Men**\n\n2\n\n2\n\n10\n\n13\n\n15\n\n18\n\n13\n\n15\n\n16\n\n5\n\n14\n\n11\n\n29\n\n36\n\n2024 2022\n\nBy gender, total cash compensation overall trended higher for women in 2024 compared to 2022. We saw a sharp spike in total cash compensation for women in the $1 million to $2 million range, and a sharp decline for men in the $800,000 to $1 million range.\n\n**2024 US PROFESSIONAL SERVICES PARTNER COMPENSATION SURVEY**\n\n**26**","pages":[{"line_numbers":[],"markdown":"**Total cash compensation, by gender (%)**\n\n$400,000 or less\n\n$401,000–$600,000\n\n$601,000–$800,000\n\n$801,000–$1.00m\n\n$1.01m–$2.00m\n\n$2.01m–$3.00m\n\nMore than $3.00m\n\nNote: Numbers may not sum to 100%, because of rounding.\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=155\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2022, n=192\n\n**Women**\n\n20\n\n11\n\n13\n\n37\n\n7\n\n5\n\n40\n\n7\n\n0\n\n13\n\n42\n\n0\n\n0\n\n5\n\n**Men**\n\n2\n\n2\n\n10\n\n13\n\n15\n\n18\n\n13\n\n15\n\n16\n\n5\n\n14\n\n11\n\n29\n\n36\n\n2024 2022\n\nBy gender, total cash compensation overall trended higher for women in 2024 compared to 2022. We saw a sharp spike in total cash compensation for women in the $1 million to $2 million range, and a sharp decline for men in the $800,000 to $1 million range.\n\n**2024 US PROFESSIONAL SERVICES PARTNER COMPENSATION SURVEY**\n\n**26**","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**Total cash compensation, by gender (%)**\n\n$400,000 or less\n\n$401,000–$600,000\n\n$601,000–$800,000\n\n$801,000–$1.00m\n\n$1.01m–$2.00m\n\n$2.01m–$3.00m\n\nMore than $3.00m\n\nNote: Numbers may not sum to 100%, because of rounding.\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=155\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2022, n=192\n\n**Women**\n\n20\n\n11\n\n13\n\n37\n\n7\n\n5\n\n40\n\n7\n\n0\n\n13\n\n42\n\n0\n\n0\n\n5\n\n**Men**\n\n2\n\n2\n\n10\n\n13\n\n15\n\n18\n\n13\n\n15\n\n16\n\n5\n\n14\n\n11\n\n29\n\n36\n\n2024 2022\n\nBy gender, total cash compensation overall trended higher for women in 2024 compared to 2022. We saw a sharp spike in total cash compensation for women in the $1 million to $2 million range, and a sharp decline for men in the $800,000 to $1 million range.\n\n**2024 US PROFESSIONAL SERVICES PARTNER COMPENSATION SURVEY**\n\n**26**","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/US_Professional_Services_Partner_Compensation_Survey_2024_p26.pdf"]},"response":{"errors":[],"extract_id":"extract_ef86e476c3392f5064a53a30a3483273","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**Total cash compensation, by gender (%)**\n\n$400,000 or less\n\n$401,000–$600,000\n\n$601,000–$800,000\n\n$801,000–$1.00m\n\n$1.01m–$2.00m\n\n$2.01m–$3.00m\n\nMore than $3.00m\n\nNote: Numbers may not sum to 100%, because of rounding.\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=155\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2022, n=192\n\n**Women**\n\n20\n\n11\n\n13\n\n37\n\n7\n\n5\n\n40\n\n7\n\n0\n\n13\n\n42\n\n0\n\n0\n\n5\n\n**Men**\n\n2\n\n2\n\n10\n\n13\n\n15\n\n18\n\n13\n\n15\n\n16\n\n5\n\n14\n\n11\n\n29\n\n36\n\n2024 2022\n\nBy gender, total cash compensation overall trended higher for women in 2024 compared to 2022. We saw a sharp spike in total cash compensation for women in the $1 million to $2 million range, and a sharp decline for men in the $800,000 to $1 million range.\n\n**2024 US PROFESSIONAL SERVICES PARTNER COMPENSATION SURVEY**\n\n**26**"],"full_content":"**Total cash compensation, by gender (%)**\n\n$400,000 or less\n\n$401,000–$600,000\n\n$601,000–$800,000\n\n$801,000–$1.00m\n\n$1.01m–$2.00m\n\n$2.01m–$3.00m\n\nMore than $3.00m\n\nNote: Numbers may not sum to 100%, because of rounding.\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=155\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2022, n=192\n\n**Women**\n\n20\n\n11\n\n13\n\n37\n\n7\n\n5\n\n40\n\n7\n\n0\n\n13\n\n42\n\n0\n\n0\n\n5\n\n**Men**\n\n2\n\n2\n\n10\n\n13\n\n15\n\n18\n\n13\n\n15\n\n16\n\n5\n\n14\n\n11\n\n29\n\n36\n\n2024 2022\n\nBy gender, total cash compensation overall trended higher for women in 2024 compared to 2022. We saw a sharp spike in total cash compensation for women in the $1 million to $2 million range, and a sharp decline for men in the $800,000 to $1 million range.\n\n**2024 US PROFESSIONAL SERVICES PARTNER COMPENSATION SURVEY**\n\n**26**","publish_date":null,"title":"Total cash compensation, by gender (%)","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/US_Professional_Services_Partner_Compensation_Survey_2024_p26.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/US_Professional_Services_Partner_Compensation_Survey_2024_p26.pdf"},"request":{"example_id":"chart/US_Professional_Services_Partner_Compensation_Survey_2024_p26","product_type":"parse","source_file_path":"US_Professional_Services_Partner_Compensation_Survey_2024_p26.pdf"},"started_at":"2026-11-04T19:44:47.780382"} +{"completed_at":"2026-11-04T19:44:51.038738","latency_in_ms":2148,"output":{"example_id":"chart/US_Professional_Services_Partner_Compensation_Survey_2024_p27","grounded_pages":[],"layout_pages":[],"markdown":"**Total cash compensation, by race/ethnicity (%)**\n\n$400,000 or less\n\n$401,000–$600,000\n\n$601,000–$800,000\n\n$801,000–$1.00m\n\n$1.01m–$2.00m\n\n$2.01m–$3.00m\n\nMore than $3.00m\n\nNote: Numbers may not sum to 100%, because of rounding.\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=155\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2022, n=192\n\n**Racial or ethnic minority White**\n\n8\n\n21\n\n12\n\n13\n\n12\n\n21\n\n29\n\n12\n\n24\n\n31\n\n0\n\n0\n\n2\n\n15\n\n3\n\n3\n\n12\n\n11\n\n19\n\n25\n\n9\n\n9\n\n16\n\n6\n\n12\n\n7\n\n30\n\n38\n\n2024 2022\n\nFor respondents who represent racial or ethnic minorities, total cash compensation was split: 46% of respondents of racial or ethnic minority groups reported compensation of more than $1 million, up from 38% in 2022. However, 55% reported compensation of $800,000 or less, a substantial increase from 32% in 2022.\n\n## Equity or non-cash compensation\n\nThe share of respondents with an equity or non-cash element to their compensation decreased from 66% in 2022 to 59% in 2024. For those who do receive non-cash compensation, more than half, 56%, report that it is 25% or less of their total compensation.\n\n**Equity or non-cash compensation (%)**\n\n**Do you have an equity or non-cash element to your compensation? As a percentage of your total cash and non- cash compensation, what share is non-cash?**\n\n1%–25%\n\n26%–50%\n\n51%–75%\n\n76%–100%\n\n101%–150%\n\nMore than 150%\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=155\n\n56\n\n32\n\n5\n\n4\n\n1\n\n2\n\n2024\n\n2022\n\n66\n\n59\n\nYes\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=84\n\nDiverse region, inclusive workforces: Diversity and inclusion policy and practice in Asia Pacific **HEIDRICK & STRUGGLES**\n\n**27**","pages":[{"line_numbers":[],"markdown":"**Total cash compensation, by race/ethnicity (%)**\n\n$400,000 or less\n\n$401,000–$600,000\n\n$601,000–$800,000\n\n$801,000–$1.00m\n\n$1.01m–$2.00m\n\n$2.01m–$3.00m\n\nMore than $3.00m\n\nNote: Numbers may not sum to 100%, because of rounding.\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=155\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2022, n=192\n\n**Racial or ethnic minority White**\n\n8\n\n21\n\n12\n\n13\n\n12\n\n21\n\n29\n\n12\n\n24\n\n31\n\n0\n\n0\n\n2\n\n15\n\n3\n\n3\n\n12\n\n11\n\n19\n\n25\n\n9\n\n9\n\n16\n\n6\n\n12\n\n7\n\n30\n\n38\n\n2024 2022\n\nFor respondents who represent racial or ethnic minorities, total cash compensation was split: 46% of respondents of racial or ethnic minority groups reported compensation of more than $1 million, up from 38% in 2022. However, 55% reported compensation of $800,000 or less, a substantial increase from 32% in 2022.\n\n## Equity or non-cash compensation\n\nThe share of respondents with an equity or non-cash element to their compensation decreased from 66% in 2022 to 59% in 2024. For those who do receive non-cash compensation, more than half, 56%, report that it is 25% or less of their total compensation.\n\n**Equity or non-cash compensation (%)**\n\n**Do you have an equity or non-cash element to your compensation? As a percentage of your total cash and non- cash compensation, what share is non-cash?**\n\n1%–25%\n\n26%–50%\n\n51%–75%\n\n76%–100%\n\n101%–150%\n\nMore than 150%\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=155\n\n56\n\n32\n\n5\n\n4\n\n1\n\n2\n\n2024\n\n2022\n\n66\n\n59\n\nYes\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=84\n\nDiverse region, inclusive workforces: Diversity and inclusion policy and practice in Asia Pacific **HEIDRICK & STRUGGLES**\n\n**27**","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**Total cash compensation, by race/ethnicity (%)**\n\n$400,000 or less\n\n$401,000–$600,000\n\n$601,000–$800,000\n\n$801,000–$1.00m\n\n$1.01m–$2.00m\n\n$2.01m–$3.00m\n\nMore than $3.00m\n\nNote: Numbers may not sum to 100%, because of rounding.\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=155\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2022, n=192\n\n**Racial or ethnic minority White**\n\n8\n\n21\n\n12\n\n13\n\n12\n\n21\n\n29\n\n12\n\n24\n\n31\n\n0\n\n0\n\n2\n\n15\n\n3\n\n3\n\n12\n\n11\n\n19\n\n25\n\n9\n\n9\n\n16\n\n6\n\n12\n\n7\n\n30\n\n38\n\n2024 2022\n\nFor respondents who represent racial or ethnic minorities, total cash compensation was split: 46% of respondents of racial or ethnic minority groups reported compensation of more than $1 million, up from 38% in 2022. However, 55% reported compensation of $800,000 or less, a substantial increase from 32% in 2022.\n\n## Equity or non-cash compensation\n\nThe share of respondents with an equity or non-cash element to their compensation decreased from 66% in 2022 to 59% in 2024. For those who do receive non-cash compensation, more than half, 56%, report that it is 25% or less of their total compensation.\n\n**Equity or non-cash compensation (%)**\n\n**Do you have an equity or non-cash element to your compensation? As a percentage of your total cash and non- cash compensation, what share is non-cash?**\n\n1%–25%\n\n26%–50%\n\n51%–75%\n\n76%–100%\n\n101%–150%\n\nMore than 150%\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=155\n\n56\n\n32\n\n5\n\n4\n\n1\n\n2\n\n2024\n\n2022\n\n66\n\n59\n\nYes\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=84\n\nDiverse region, inclusive workforces: Diversity and inclusion policy and practice in Asia Pacific **HEIDRICK & STRUGGLES**\n\n**27**","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/US_Professional_Services_Partner_Compensation_Survey_2024_p27.pdf"]},"response":{"errors":[],"extract_id":"extract_e229ab483ea757c5f65595122fc20139","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**Total cash compensation, by race/ethnicity (%)**\n\n$400,000 or less\n\n$401,000–$600,000\n\n$601,000–$800,000\n\n$801,000–$1.00m\n\n$1.01m–$2.00m\n\n$2.01m–$3.00m\n\nMore than $3.00m\n\nNote: Numbers may not sum to 100%, because of rounding.\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=155\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2022, n=192\n\n**Racial or ethnic minority White**\n\n8\n\n21\n\n12\n\n13\n\n12\n\n21\n\n29\n\n12\n\n24\n\n31\n\n0\n\n0\n\n2\n\n15\n\n3\n\n3\n\n12\n\n11\n\n19\n\n25\n\n9\n\n9\n\n16\n\n6\n\n12\n\n7\n\n30\n\n38\n\n2024 2022\n\nFor respondents who represent racial or ethnic minorities, total cash compensation was split: 46% of respondents of racial or ethnic minority groups reported compensation of more than $1 million, up from 38% in 2022. However, 55% reported compensation of $800,000 or less, a substantial increase from 32% in 2022.\n\n## Equity or non-cash compensation\n\nThe share of respondents with an equity or non-cash element to their compensation decreased from 66% in 2022 to 59% in 2024. For those who do receive non-cash compensation, more than half, 56%, report that it is 25% or less of their total compensation.\n\n**Equity or non-cash compensation (%)**\n\n**Do you have an equity or non-cash element to your compensation? As a percentage of your total cash and non- cash compensation, what share is non-cash?**\n\n1%–25%\n\n26%–50%\n\n51%–75%\n\n76%–100%\n\n101%–150%\n\nMore than 150%\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=155\n\n56\n\n32\n\n5\n\n4\n\n1\n\n2\n\n2024\n\n2022\n\n66\n\n59\n\nYes\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=84\n\nDiverse region, inclusive workforces: Diversity and inclusion policy and practice in Asia Pacific **HEIDRICK & STRUGGLES**\n\n**27**"],"full_content":"**Total cash compensation, by race/ethnicity (%)**\n\n$400,000 or less\n\n$401,000–$600,000\n\n$601,000–$800,000\n\n$801,000–$1.00m\n\n$1.01m–$2.00m\n\n$2.01m–$3.00m\n\nMore than $3.00m\n\nNote: Numbers may not sum to 100%, because of rounding.\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=155\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2022, n=192\n\n**Racial or ethnic minority White**\n\n8\n\n21\n\n12\n\n13\n\n12\n\n21\n\n29\n\n12\n\n24\n\n31\n\n0\n\n0\n\n2\n\n15\n\n3\n\n3\n\n12\n\n11\n\n19\n\n25\n\n9\n\n9\n\n16\n\n6\n\n12\n\n7\n\n30\n\n38\n\n2024 2022\n\nFor respondents who represent racial or ethnic minorities, total cash compensation was split: 46% of respondents of racial or ethnic minority groups reported compensation of more than $1 million, up from 38% in 2022. However, 55% reported compensation of $800,000 or less, a substantial increase from 32% in 2022.\n\n## Equity or non-cash compensation\n\nThe share of respondents with an equity or non-cash element to their compensation decreased from 66% in 2022 to 59% in 2024. For those who do receive non-cash compensation, more than half, 56%, report that it is 25% or less of their total compensation.\n\n**Equity or non-cash compensation (%)**\n\n**Do you have an equity or non-cash element to your compensation? As a percentage of your total cash and non- cash compensation, what share is non-cash?**\n\n1%–25%\n\n26%–50%\n\n51%–75%\n\n76%–100%\n\n101%–150%\n\nMore than 150%\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=155\n\n56\n\n32\n\n5\n\n4\n\n1\n\n2\n\n2024\n\n2022\n\n66\n\n59\n\nYes\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=84\n\nDiverse region, inclusive workforces: Diversity and inclusion policy and practice in Asia Pacific **HEIDRICK & STRUGGLES**\n\n**27**","publish_date":null,"title":"Equity or non-cash compensation","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/US_Professional_Services_Partner_Compensation_Survey_2024_p27.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/US_Professional_Services_Partner_Compensation_Survey_2024_p27.pdf"},"request":{"example_id":"chart/US_Professional_Services_Partner_Compensation_Survey_2024_p27","product_type":"parse","source_file_path":"US_Professional_Services_Partner_Compensation_Survey_2024_p27.pdf"},"started_at":"2026-11-04T19:44:48.890200"} +{"completed_at":"2026-11-04T19:44:51.754321","latency_in_ms":1515,"output":{"example_id":"chart/US_Professional_Services_Partner_Compensation_Survey_2024_p7","grounded_pages":[],"layout_pages":[],"markdown":"By firm size, we saw a correlation between personal revenue targets and firm size (the smaller the firm size, the smaller the revenue target), with the exception of the largest firms; for respondents at firms with more than $20 billion in revenue, there was more of an even distribution of revenue targets.\n\n46\n\n27\n\n13\n\n19\n\n22\n\n14\n\n26\n\n9\n\n16\n\n24\n\n10\n\n18\n\n29\n\n11\n\n10\n\n16\n\n22\n\n9\n\n21\n\n28\n\n20\n\n10\n\n20\n\n21\n\n14\n\n13\n\n7\n\n7\n\n10\n\n7\n\n7\n\n7\n\n7\n\n7\n\n13\n\n14\n\n14 8\n\n8\n\n8\n\n17\n\n14\n\n28\n\n20\n\n10\n\n7\n\nLess than $1.00bn\n\nLess than 3 years\n\n$5.01bn– $20.00bn\n\n6–10 years\n\nMore than $20.00bn\n\nMore than 10 years\n\n$1.01bn– $5.00bn\n\n3–5 years\n\n**Personal revenue targets, by firm size**\n\n**Personal revenue targets, by tenure at firm**\n\nLess than $1.00m\n\n$1.00m–$3.00m\n\n$3.01m–$6.00m\n\n$6.01m–$10.00m\n\n$10.01m–$15.00m\n\n$15.01m–$20.00m\n\n$20.01m–$25.00m\n\n$25.01m–$30.00m\n\nMore than $30.00m\n\nPrefer not to answer\n\nLess than $1.00m\n\n$1.00m–$3.00m\n\n$3.01m–$6.00m\n\n$6.01m–$10.00m\n\n$10.01m–$15.00m\n\n$15.01m–$20.00m\n\n$20.01m–$25.00m\n\n$25.01m–$30.00m\n\nMore than $30.00m\n\nPrefer not to answer\n\nNote: Numbers may not sum to 100%, because of rounding.\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=155\n\nNote: Numbers may not sum to 100%, because of rounding. Source: Heidrick & Struggles US professional services partner compensation survey, 2024, n=155\n\n0\n\n0\n\n0 0\n\n0\n\n0\n\n0\n\n0\n\n0 0\n\n5\n\n5\n\n5\n\n5\n\n5\n\n5\n\n5\n\n5\n\n5 5\n\n3\n\n3\n\n3\n\n3\n\n2\n\n2\n\n2\n\n2\n\n2\n\n2\n\n3\n\n3\n\n4\n\n4\n\nWe also saw a positive correlation between personal revenue targets and tenure at current firm, with those who have been at their firm longer having higher targets than their less-tenured peers.\n\n**Personal revenue targets (%)**\n\nDiverse region, inclusive workforces: Diversity and inclusion policy and practice in Asia Pacific **HEIDRICK & STRUGGLES**\n\n**7**","pages":[{"line_numbers":[],"markdown":"By firm size, we saw a correlation between personal revenue targets and firm size (the smaller the firm size, the smaller the revenue target), with the exception of the largest firms; for respondents at firms with more than $20 billion in revenue, there was more of an even distribution of revenue targets.\n\n46\n\n27\n\n13\n\n19\n\n22\n\n14\n\n26\n\n9\n\n16\n\n24\n\n10\n\n18\n\n29\n\n11\n\n10\n\n16\n\n22\n\n9\n\n21\n\n28\n\n20\n\n10\n\n20\n\n21\n\n14\n\n13\n\n7\n\n7\n\n10\n\n7\n\n7\n\n7\n\n7\n\n7\n\n13\n\n14\n\n14 8\n\n8\n\n8\n\n17\n\n14\n\n28\n\n20\n\n10\n\n7\n\nLess than $1.00bn\n\nLess than 3 years\n\n$5.01bn– $20.00bn\n\n6–10 years\n\nMore than $20.00bn\n\nMore than 10 years\n\n$1.01bn– $5.00bn\n\n3–5 years\n\n**Personal revenue targets, by firm size**\n\n**Personal revenue targets, by tenure at firm**\n\nLess than $1.00m\n\n$1.00m–$3.00m\n\n$3.01m–$6.00m\n\n$6.01m–$10.00m\n\n$10.01m–$15.00m\n\n$15.01m–$20.00m\n\n$20.01m–$25.00m\n\n$25.01m–$30.00m\n\nMore than $30.00m\n\nPrefer not to answer\n\nLess than $1.00m\n\n$1.00m–$3.00m\n\n$3.01m–$6.00m\n\n$6.01m–$10.00m\n\n$10.01m–$15.00m\n\n$15.01m–$20.00m\n\n$20.01m–$25.00m\n\n$25.01m–$30.00m\n\nMore than $30.00m\n\nPrefer not to answer\n\nNote: Numbers may not sum to 100%, because of rounding.\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=155\n\nNote: Numbers may not sum to 100%, because of rounding. Source: Heidrick & Struggles US professional services partner compensation survey, 2024, n=155\n\n0\n\n0\n\n0 0\n\n0\n\n0\n\n0\n\n0\n\n0 0\n\n5\n\n5\n\n5\n\n5\n\n5\n\n5\n\n5\n\n5\n\n5 5\n\n3\n\n3\n\n3\n\n3\n\n2\n\n2\n\n2\n\n2\n\n2\n\n2\n\n3\n\n3\n\n4\n\n4\n\nWe also saw a positive correlation between personal revenue targets and tenure at current firm, with those who have been at their firm longer having higher targets than their less-tenured peers.\n\n**Personal revenue targets (%)**\n\nDiverse region, inclusive workforces: Diversity and inclusion policy and practice in Asia Pacific **HEIDRICK & STRUGGLES**\n\n**7**","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"By firm size, we saw a correlation between personal revenue targets and firm size (the smaller the firm size, the smaller the revenue target), with the exception of the largest firms; for respondents at firms with more than $20 billion in revenue, there was more of an even distribution of revenue targets.\n\n46\n\n27\n\n13\n\n19\n\n22\n\n14\n\n26\n\n9\n\n16\n\n24\n\n10\n\n18\n\n29\n\n11\n\n10\n\n16\n\n22\n\n9\n\n21\n\n28\n\n20\n\n10\n\n20\n\n21\n\n14\n\n13\n\n7\n\n7\n\n10\n\n7\n\n7\n\n7\n\n7\n\n7\n\n13\n\n14\n\n14 8\n\n8\n\n8\n\n17\n\n14\n\n28\n\n20\n\n10\n\n7\n\nLess than $1.00bn\n\nLess than 3 years\n\n$5.01bn– $20.00bn\n\n6–10 years\n\nMore than $20.00bn\n\nMore than 10 years\n\n$1.01bn– $5.00bn\n\n3–5 years\n\n**Personal revenue targets, by firm size**\n\n**Personal revenue targets, by tenure at firm**\n\nLess than $1.00m\n\n$1.00m–$3.00m\n\n$3.01m–$6.00m\n\n$6.01m–$10.00m\n\n$10.01m–$15.00m\n\n$15.01m–$20.00m\n\n$20.01m–$25.00m\n\n$25.01m–$30.00m\n\nMore than $30.00m\n\nPrefer not to answer\n\nLess than $1.00m\n\n$1.00m–$3.00m\n\n$3.01m–$6.00m\n\n$6.01m–$10.00m\n\n$10.01m–$15.00m\n\n$15.01m–$20.00m\n\n$20.01m–$25.00m\n\n$25.01m–$30.00m\n\nMore than $30.00m\n\nPrefer not to answer\n\nNote: Numbers may not sum to 100%, because of rounding.\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=155\n\nNote: Numbers may not sum to 100%, because of rounding. Source: Heidrick & Struggles US professional services partner compensation survey, 2024, n=155\n\n0\n\n0\n\n0 0\n\n0\n\n0\n\n0\n\n0\n\n0 0\n\n5\n\n5\n\n5\n\n5\n\n5\n\n5\n\n5\n\n5\n\n5 5\n\n3\n\n3\n\n3\n\n3\n\n2\n\n2\n\n2\n\n2\n\n2\n\n2\n\n3\n\n3\n\n4\n\n4\n\nWe also saw a positive correlation between personal revenue targets and tenure at current firm, with those who have been at their firm longer having higher targets than their less-tenured peers.\n\n**Personal revenue targets (%)**\n\nDiverse region, inclusive workforces: Diversity and inclusion policy and practice in Asia Pacific **HEIDRICK & STRUGGLES**\n\n**7**","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/US_Professional_Services_Partner_Compensation_Survey_2024_p7.pdf"]},"response":{"errors":[],"extract_id":"extract_38710d350b44aa5ee128279c83357fc7","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["By firm size, we saw a correlation between personal revenue targets and firm size (the smaller the firm size, the smaller the revenue target), with the exception of the largest firms; for respondents at firms with more than $20 billion in revenue, there was more of an even distribution of revenue targets.\n\n46\n\n27\n\n13\n\n19\n\n22\n\n14\n\n26\n\n9\n\n16\n\n24\n\n10\n\n18\n\n29\n\n11\n\n10\n\n16\n\n22\n\n9\n\n21\n\n28\n\n20\n\n10\n\n20\n\n21\n\n14\n\n13\n\n7\n\n7\n\n10\n\n7\n\n7\n\n7\n\n7\n\n7\n\n13\n\n14\n\n14 8\n\n8\n\n8\n\n17\n\n14\n\n28\n\n20\n\n10\n\n7\n\nLess than $1.00bn\n\nLess than 3 years\n\n$5.01bn– $20.00bn\n\n6–10 years\n\nMore than $20.00bn\n\nMore than 10 years\n\n$1.01bn– $5.00bn\n\n3–5 years\n\n**Personal revenue targets, by firm size**\n\n**Personal revenue targets, by tenure at firm**\n\nLess than $1.00m\n\n$1.00m–$3.00m\n\n$3.01m–$6.00m\n\n$6.01m–$10.00m\n\n$10.01m–$15.00m\n\n$15.01m–$20.00m\n\n$20.01m–$25.00m\n\n$25.01m–$30.00m\n\nMore than $30.00m\n\nPrefer not to answer\n\nLess than $1.00m\n\n$1.00m–$3.00m\n\n$3.01m–$6.00m\n\n$6.01m–$10.00m\n\n$10.01m–$15.00m\n\n$15.01m–$20.00m\n\n$20.01m–$25.00m\n\n$25.01m–$30.00m\n\nMore than $30.00m\n\nPrefer not to answer\n\nNote: Numbers may not sum to 100%, because of rounding.\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=155\n\nNote: Numbers may not sum to 100%, because of rounding. Source: Heidrick & Struggles US professional services partner compensation survey, 2024, n=155\n\n0\n\n0\n\n0 0\n\n0\n\n0\n\n0\n\n0\n\n0 0\n\n5\n\n5\n\n5\n\n5\n\n5\n\n5\n\n5\n\n5\n\n5 5\n\n3\n\n3\n\n3\n\n3\n\n2\n\n2\n\n2\n\n2\n\n2\n\n2\n\n3\n\n3\n\n4\n\n4\n\nWe also saw a positive correlation between personal revenue targets and tenure at current firm, with those who have been at their firm longer having higher targets than their less-tenured peers.\n\n**Personal revenue targets (%)**\n\nDiverse region, inclusive workforces: Diversity and inclusion policy and practice in Asia Pacific **HEIDRICK & STRUGGLES**\n\n**7**"],"full_content":"By firm size, we saw a correlation between personal revenue targets and firm size (the smaller the firm size, the smaller the revenue target), with the exception of the largest firms; for respondents at firms with more than $20 billion in revenue, there was more of an even distribution of revenue targets.\n\n46\n\n27\n\n13\n\n19\n\n22\n\n14\n\n26\n\n9\n\n16\n\n24\n\n10\n\n18\n\n29\n\n11\n\n10\n\n16\n\n22\n\n9\n\n21\n\n28\n\n20\n\n10\n\n20\n\n21\n\n14\n\n13\n\n7\n\n7\n\n10\n\n7\n\n7\n\n7\n\n7\n\n7\n\n13\n\n14\n\n14 8\n\n8\n\n8\n\n17\n\n14\n\n28\n\n20\n\n10\n\n7\n\nLess than $1.00bn\n\nLess than 3 years\n\n$5.01bn– $20.00bn\n\n6–10 years\n\nMore than $20.00bn\n\nMore than 10 years\n\n$1.01bn– $5.00bn\n\n3–5 years\n\n**Personal revenue targets, by firm size**\n\n**Personal revenue targets, by tenure at firm**\n\nLess than $1.00m\n\n$1.00m–$3.00m\n\n$3.01m–$6.00m\n\n$6.01m–$10.00m\n\n$10.01m–$15.00m\n\n$15.01m–$20.00m\n\n$20.01m–$25.00m\n\n$25.01m–$30.00m\n\nMore than $30.00m\n\nPrefer not to answer\n\nLess than $1.00m\n\n$1.00m–$3.00m\n\n$3.01m–$6.00m\n\n$6.01m–$10.00m\n\n$10.01m–$15.00m\n\n$15.01m–$20.00m\n\n$20.01m–$25.00m\n\n$25.01m–$30.00m\n\nMore than $30.00m\n\nPrefer not to answer\n\nNote: Numbers may not sum to 100%, because of rounding.\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=155\n\nNote: Numbers may not sum to 100%, because of rounding. Source: Heidrick & Struggles US professional services partner compensation survey, 2024, n=155\n\n0\n\n0\n\n0 0\n\n0\n\n0\n\n0\n\n0\n\n0 0\n\n5\n\n5\n\n5\n\n5\n\n5\n\n5\n\n5\n\n5\n\n5 5\n\n3\n\n3\n\n3\n\n3\n\n2\n\n2\n\n2\n\n2\n\n2\n\n2\n\n3\n\n3\n\n4\n\n4\n\nWe also saw a positive correlation between personal revenue targets and tenure at current firm, with those who have been at their firm longer having higher targets than their less-tenured peers.\n\n**Personal revenue targets (%)**\n\nDiverse region, inclusive workforces: Diversity and inclusion policy and practice in Asia Pacific **HEIDRICK & STRUGGLES**\n\n**7**","publish_date":null,"title":"Personal revenue targets (%)","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/US_Professional_Services_Partner_Compensation_Survey_2024_p7.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/US_Professional_Services_Partner_Compensation_Survey_2024_p7.pdf"},"request":{"example_id":"chart/US_Professional_Services_Partner_Compensation_Survey_2024_p7","product_type":"parse","source_file_path":"US_Professional_Services_Partner_Compensation_Survey_2024_p7.pdf"},"started_at":"2026-11-04T19:44:50.239210"} +{"completed_at":"2026-11-04T19:44:52.879835","latency_in_ms":1336,"output":{"example_id":"chart/US_Professional_Services_Partner_Compensation_Survey_2024_p8","grounded_pages":[],"layout_pages":[],"markdown":"By tenure in their current role, there is a correlation between revenue targets and tenure in role for those who are either new in their role or are more tenured. More-junior respondents identified lower personal targets, while more-senior respondents identified higher targets.\n\nRespondents in their current role for two to three years have slightly higher personal revenue targets than their counterparts with tenure between four and five years: 48% of respondents with two to three years’ tenure reported personal revenue targets below $10 million, while 63% of respondents with four to five years of experience reported the same. And 28% of respondents with two to three years’ tenure in their current role reported personal revenue targets of $20 million or more, while only 19% of respondents with four to five years of experience reported the same.\n\n0\n\n18 10\n\n7\n\n21 14\n\n21 24\n\n11 20\n\n7 12\n\n8\n\n8 7\n\nLess than 2 years 4–5 years More than 5 years 2–3 years\n\n**Personal revenue targets, by tenure in role**\n\nLess than $1.00m\n\n$1.00m–$3.00m\n\n$3.01m–$6.00m\n\n$6.01m–$10.00m\n\n$10.01m–$15.00m\n\n$15.01m–$20.00m\n\n$20.01m–$25.00m\n\n$25.01m–$30.00m\n\nMore than $30.00m\n\nPrefer not to answer\n\nNote: Numbers may not sum to 100%, because of rounding.\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=155\n\n4 2\n\n4\n\n0\n\n22\n\n19\n\n19\n\n13\n\n13\n\n6\n\n3\n\n3\n\n3\n\n0\n\n0\n\n7\n\n0\n\n12\n\n10\n\n10\n\n15\n\n24\n\n7\n\n12\n\n2\n\nPractice leader\n\nRegional leader\n\nOffice leader\n\nNo\n\n**What is your current role? Do you have any additional firm leadership roles?**\n\n9\n\n54\n\n7\n\n30\n\n**Current role information (%)**\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=187\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=169\n\n## Respondent demographics\n\nCompared to 2022, there is a slightly higher share of respondents (70%) who say they have an additional firm leadership role. The vast majority of respondents are at the partner level.\n\nPartner (or partner-equivalent)\n\nPre-partner\n\n7\n\n93\n\n**Personal revenue targets (%)**\n\n**2024 US PROFESSIONAL SERVICES PARTNER COMPENSATION SURVEY**\n\n**8**","pages":[{"line_numbers":[],"markdown":"By tenure in their current role, there is a correlation between revenue targets and tenure in role for those who are either new in their role or are more tenured. More-junior respondents identified lower personal targets, while more-senior respondents identified higher targets.\n\nRespondents in their current role for two to three years have slightly higher personal revenue targets than their counterparts with tenure between four and five years: 48% of respondents with two to three years’ tenure reported personal revenue targets below $10 million, while 63% of respondents with four to five years of experience reported the same. And 28% of respondents with two to three years’ tenure in their current role reported personal revenue targets of $20 million or more, while only 19% of respondents with four to five years of experience reported the same.\n\n0\n\n18 10\n\n7\n\n21 14\n\n21 24\n\n11 20\n\n7 12\n\n8\n\n8 7\n\nLess than 2 years 4–5 years More than 5 years 2–3 years\n\n**Personal revenue targets, by tenure in role**\n\nLess than $1.00m\n\n$1.00m–$3.00m\n\n$3.01m–$6.00m\n\n$6.01m–$10.00m\n\n$10.01m–$15.00m\n\n$15.01m–$20.00m\n\n$20.01m–$25.00m\n\n$25.01m–$30.00m\n\nMore than $30.00m\n\nPrefer not to answer\n\nNote: Numbers may not sum to 100%, because of rounding.\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=155\n\n4 2\n\n4\n\n0\n\n22\n\n19\n\n19\n\n13\n\n13\n\n6\n\n3\n\n3\n\n3\n\n0\n\n0\n\n7\n\n0\n\n12\n\n10\n\n10\n\n15\n\n24\n\n7\n\n12\n\n2\n\nPractice leader\n\nRegional leader\n\nOffice leader\n\nNo\n\n**What is your current role? Do you have any additional firm leadership roles?**\n\n9\n\n54\n\n7\n\n30\n\n**Current role information (%)**\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=187\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=169\n\n## Respondent demographics\n\nCompared to 2022, there is a slightly higher share of respondents (70%) who say they have an additional firm leadership role. The vast majority of respondents are at the partner level.\n\nPartner (or partner-equivalent)\n\nPre-partner\n\n7\n\n93\n\n**Personal revenue targets (%)**\n\n**2024 US PROFESSIONAL SERVICES PARTNER COMPENSATION SURVEY**\n\n**8**","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"By tenure in their current role, there is a correlation between revenue targets and tenure in role for those who are either new in their role or are more tenured. More-junior respondents identified lower personal targets, while more-senior respondents identified higher targets.\n\nRespondents in their current role for two to three years have slightly higher personal revenue targets than their counterparts with tenure between four and five years: 48% of respondents with two to three years’ tenure reported personal revenue targets below $10 million, while 63% of respondents with four to five years of experience reported the same. And 28% of respondents with two to three years’ tenure in their current role reported personal revenue targets of $20 million or more, while only 19% of respondents with four to five years of experience reported the same.\n\n0\n\n18 10\n\n7\n\n21 14\n\n21 24\n\n11 20\n\n7 12\n\n8\n\n8 7\n\nLess than 2 years 4–5 years More than 5 years 2–3 years\n\n**Personal revenue targets, by tenure in role**\n\nLess than $1.00m\n\n$1.00m–$3.00m\n\n$3.01m–$6.00m\n\n$6.01m–$10.00m\n\n$10.01m–$15.00m\n\n$15.01m–$20.00m\n\n$20.01m–$25.00m\n\n$25.01m–$30.00m\n\nMore than $30.00m\n\nPrefer not to answer\n\nNote: Numbers may not sum to 100%, because of rounding.\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=155\n\n4 2\n\n4\n\n0\n\n22\n\n19\n\n19\n\n13\n\n13\n\n6\n\n3\n\n3\n\n3\n\n0\n\n0\n\n7\n\n0\n\n12\n\n10\n\n10\n\n15\n\n24\n\n7\n\n12\n\n2\n\nPractice leader\n\nRegional leader\n\nOffice leader\n\nNo\n\n**What is your current role? Do you have any additional firm leadership roles?**\n\n9\n\n54\n\n7\n\n30\n\n**Current role information (%)**\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=187\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=169\n\n## Respondent demographics\n\nCompared to 2022, there is a slightly higher share of respondents (70%) who say they have an additional firm leadership role. The vast majority of respondents are at the partner level.\n\nPartner (or partner-equivalent)\n\nPre-partner\n\n7\n\n93\n\n**Personal revenue targets (%)**\n\n**2024 US PROFESSIONAL SERVICES PARTNER COMPENSATION SURVEY**\n\n**8**","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/US_Professional_Services_Partner_Compensation_Survey_2024_p8.pdf"]},"response":{"errors":[],"extract_id":"extract_a8932c68170da57f6fa14052b552e983","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["By tenure in their current role, there is a correlation between revenue targets and tenure in role for those who are either new in their role or are more tenured. More-junior respondents identified lower personal targets, while more-senior respondents identified higher targets.\n\nRespondents in their current role for two to three years have slightly higher personal revenue targets than their counterparts with tenure between four and five years: 48% of respondents with two to three years’ tenure reported personal revenue targets below $10 million, while 63% of respondents with four to five years of experience reported the same. And 28% of respondents with two to three years’ tenure in their current role reported personal revenue targets of $20 million or more, while only 19% of respondents with four to five years of experience reported the same.\n\n0\n\n18 10\n\n7\n\n21 14\n\n21 24\n\n11 20\n\n7 12\n\n8\n\n8 7\n\nLess than 2 years 4–5 years More than 5 years 2–3 years\n\n**Personal revenue targets, by tenure in role**\n\nLess than $1.00m\n\n$1.00m–$3.00m\n\n$3.01m–$6.00m\n\n$6.01m–$10.00m\n\n$10.01m–$15.00m\n\n$15.01m–$20.00m\n\n$20.01m–$25.00m\n\n$25.01m–$30.00m\n\nMore than $30.00m\n\nPrefer not to answer\n\nNote: Numbers may not sum to 100%, because of rounding.\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=155\n\n4 2\n\n4\n\n0\n\n22\n\n19\n\n19\n\n13\n\n13\n\n6\n\n3\n\n3\n\n3\n\n0\n\n0\n\n7\n\n0\n\n12\n\n10\n\n10\n\n15\n\n24\n\n7\n\n12\n\n2\n\nPractice leader\n\nRegional leader\n\nOffice leader\n\nNo\n\n**What is your current role? Do you have any additional firm leadership roles?**\n\n9\n\n54\n\n7\n\n30\n\n**Current role information (%)**\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=187\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=169\n\n## Respondent demographics\n\nCompared to 2022, there is a slightly higher share of respondents (70%) who say they have an additional firm leadership role. The vast majority of respondents are at the partner level.\n\nPartner (or partner-equivalent)\n\nPre-partner\n\n7\n\n93\n\n**Personal revenue targets (%)**\n\n**2024 US PROFESSIONAL SERVICES PARTNER COMPENSATION SURVEY**\n\n**8**"],"full_content":"By tenure in their current role, there is a correlation between revenue targets and tenure in role for those who are either new in their role or are more tenured. More-junior respondents identified lower personal targets, while more-senior respondents identified higher targets.\n\nRespondents in their current role for two to three years have slightly higher personal revenue targets than their counterparts with tenure between four and five years: 48% of respondents with two to three years’ tenure reported personal revenue targets below $10 million, while 63% of respondents with four to five years of experience reported the same. And 28% of respondents with two to three years’ tenure in their current role reported personal revenue targets of $20 million or more, while only 19% of respondents with four to five years of experience reported the same.\n\n0\n\n18 10\n\n7\n\n21 14\n\n21 24\n\n11 20\n\n7 12\n\n8\n\n8 7\n\nLess than 2 years 4–5 years More than 5 years 2–3 years\n\n**Personal revenue targets, by tenure in role**\n\nLess than $1.00m\n\n$1.00m–$3.00m\n\n$3.01m–$6.00m\n\n$6.01m–$10.00m\n\n$10.01m–$15.00m\n\n$15.01m–$20.00m\n\n$20.01m–$25.00m\n\n$25.01m–$30.00m\n\nMore than $30.00m\n\nPrefer not to answer\n\nNote: Numbers may not sum to 100%, because of rounding.\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=155\n\n4 2\n\n4\n\n0\n\n22\n\n19\n\n19\n\n13\n\n13\n\n6\n\n3\n\n3\n\n3\n\n0\n\n0\n\n7\n\n0\n\n12\n\n10\n\n10\n\n15\n\n24\n\n7\n\n12\n\n2\n\nPractice leader\n\nRegional leader\n\nOffice leader\n\nNo\n\n**What is your current role? Do you have any additional firm leadership roles?**\n\n9\n\n54\n\n7\n\n30\n\n**Current role information (%)**\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=187\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=169\n\n## Respondent demographics\n\nCompared to 2022, there is a slightly higher share of respondents (70%) who say they have an additional firm leadership role. The vast majority of respondents are at the partner level.\n\nPartner (or partner-equivalent)\n\nPre-partner\n\n7\n\n93\n\n**Personal revenue targets (%)**\n\n**2024 US PROFESSIONAL SERVICES PARTNER COMPENSATION SURVEY**\n\n**8**","publish_date":null,"title":"Respondent demographics","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/US_Professional_Services_Partner_Compensation_Survey_2024_p8.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/US_Professional_Services_Partner_Compensation_Survey_2024_p8.pdf"},"request":{"example_id":"chart/US_Professional_Services_Partner_Compensation_Survey_2024_p8","product_type":"parse","source_file_path":"US_Professional_Services_Partner_Compensation_Survey_2024_p8.pdf"},"started_at":"2026-11-04T19:44:51.543006"} +{"completed_at":"2026-11-04T19:44:53.787359","latency_in_ms":1324,"output":{"example_id":"chart/US_Professional_Services_Partner_Compensation_Survey_2024_p9","grounded_pages":[],"layout_pages":[],"markdown":"Note: Numbers may not sum to 100%, because of rounding.\n\nIn 2024, on average, the time from when respondents joined their firm to when they started their current role is 3.2 years, which suggests that these respondents joined their firms in more-senior roles. One- third of respondents joined their current firm in their current role.\n\n**Current role information (%)**\n\n**How many years have you been a partner at your current firm or any other firm?**\n\n**How many years have you been with your current firm?**\n\n**How many years have you been in your current role?**\n\nLess than 1 year\n\n1–3 years\n\n4–5 years\n\n6–10 years\n\nMore than 10 years\n\n2\n\n6\n\n13\n\n9\n\n18\n\n12\n\n24\n\n38\n\n39\n\n39\n\n2\n\nN/A\n\n11\n\n8\n\n22\n\n26\n\n17\n\n19\n\n19\n\n20\n\n29\n\n27\n\nN/A\n\n1\n\n16\n\n9\n\n42\n\n42\n\n23\n\n22\n\n10\n\n21\n\n7\n\n7\n\nPrefer not to answer\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=149\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=153\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=151\n\n2024 2022\n\nDiverse region, inclusive workforces: Diversity and inclusion policy and practice in Asia Pacific **HEIDRICK & STRUGGLES**\n\n**9**","pages":[{"line_numbers":[],"markdown":"Note: Numbers may not sum to 100%, because of rounding.\n\nIn 2024, on average, the time from when respondents joined their firm to when they started their current role is 3.2 years, which suggests that these respondents joined their firms in more-senior roles. One- third of respondents joined their current firm in their current role.\n\n**Current role information (%)**\n\n**How many years have you been a partner at your current firm or any other firm?**\n\n**How many years have you been with your current firm?**\n\n**How many years have you been in your current role?**\n\nLess than 1 year\n\n1–3 years\n\n4–5 years\n\n6–10 years\n\nMore than 10 years\n\n2\n\n6\n\n13\n\n9\n\n18\n\n12\n\n24\n\n38\n\n39\n\n39\n\n2\n\nN/A\n\n11\n\n8\n\n22\n\n26\n\n17\n\n19\n\n19\n\n20\n\n29\n\n27\n\nN/A\n\n1\n\n16\n\n9\n\n42\n\n42\n\n23\n\n22\n\n10\n\n21\n\n7\n\n7\n\nPrefer not to answer\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=149\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=153\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=151\n\n2024 2022\n\nDiverse region, inclusive workforces: Diversity and inclusion policy and practice in Asia Pacific **HEIDRICK & STRUGGLES**\n\n**9**","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Note: Numbers may not sum to 100%, because of rounding.\n\nIn 2024, on average, the time from when respondents joined their firm to when they started their current role is 3.2 years, which suggests that these respondents joined their firms in more-senior roles. One- third of respondents joined their current firm in their current role.\n\n**Current role information (%)**\n\n**How many years have you been a partner at your current firm or any other firm?**\n\n**How many years have you been with your current firm?**\n\n**How many years have you been in your current role?**\n\nLess than 1 year\n\n1–3 years\n\n4–5 years\n\n6–10 years\n\nMore than 10 years\n\n2\n\n6\n\n13\n\n9\n\n18\n\n12\n\n24\n\n38\n\n39\n\n39\n\n2\n\nN/A\n\n11\n\n8\n\n22\n\n26\n\n17\n\n19\n\n19\n\n20\n\n29\n\n27\n\nN/A\n\n1\n\n16\n\n9\n\n42\n\n42\n\n23\n\n22\n\n10\n\n21\n\n7\n\n7\n\nPrefer not to answer\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=149\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=153\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=151\n\n2024 2022\n\nDiverse region, inclusive workforces: Diversity and inclusion policy and practice in Asia Pacific **HEIDRICK & STRUGGLES**\n\n**9**","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/US_Professional_Services_Partner_Compensation_Survey_2024_p9.pdf"]},"response":{"errors":[],"extract_id":"extract_3e3b35910a5ada95e93b7779fb1c0b70","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Note: Numbers may not sum to 100%, because of rounding.\n\nIn 2024, on average, the time from when respondents joined their firm to when they started their current role is 3.2 years, which suggests that these respondents joined their firms in more-senior roles. One- third of respondents joined their current firm in their current role.\n\n**Current role information (%)**\n\n**How many years have you been a partner at your current firm or any other firm?**\n\n**How many years have you been with your current firm?**\n\n**How many years have you been in your current role?**\n\nLess than 1 year\n\n1–3 years\n\n4–5 years\n\n6–10 years\n\nMore than 10 years\n\n2\n\n6\n\n13\n\n9\n\n18\n\n12\n\n24\n\n38\n\n39\n\n39\n\n2\n\nN/A\n\n11\n\n8\n\n22\n\n26\n\n17\n\n19\n\n19\n\n20\n\n29\n\n27\n\nN/A\n\n1\n\n16\n\n9\n\n42\n\n42\n\n23\n\n22\n\n10\n\n21\n\n7\n\n7\n\nPrefer not to answer\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=149\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=153\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=151\n\n2024 2022\n\nDiverse region, inclusive workforces: Diversity and inclusion policy and practice in Asia Pacific **HEIDRICK & STRUGGLES**\n\n**9**"],"full_content":"Note: Numbers may not sum to 100%, because of rounding.\n\nIn 2024, on average, the time from when respondents joined their firm to when they started their current role is 3.2 years, which suggests that these respondents joined their firms in more-senior roles. One- third of respondents joined their current firm in their current role.\n\n**Current role information (%)**\n\n**How many years have you been a partner at your current firm or any other firm?**\n\n**How many years have you been with your current firm?**\n\n**How many years have you been in your current role?**\n\nLess than 1 year\n\n1–3 years\n\n4–5 years\n\n6–10 years\n\nMore than 10 years\n\n2\n\n6\n\n13\n\n9\n\n18\n\n12\n\n24\n\n38\n\n39\n\n39\n\n2\n\nN/A\n\n11\n\n8\n\n22\n\n26\n\n17\n\n19\n\n19\n\n20\n\n29\n\n27\n\nN/A\n\n1\n\n16\n\n9\n\n42\n\n42\n\n23\n\n22\n\n10\n\n21\n\n7\n\n7\n\nPrefer not to answer\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=149\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=153\n\nSource: Heidrick & Struggles US professional services partner compensation survey, 2024, n=151\n\n2024 2022\n\nDiverse region, inclusive workforces: Diversity and inclusion policy and practice in Asia Pacific **HEIDRICK & STRUGGLES**\n\n**9**","publish_date":null,"title":"Current role information (%)","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/US_Professional_Services_Partner_Compensation_Survey_2024_p9.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/US_Professional_Services_Partner_Compensation_Survey_2024_p9.pdf"},"request":{"example_id":"chart/US_Professional_Services_Partner_Compensation_Survey_2024_p9","product_type":"parse","source_file_path":"US_Professional_Services_Partner_Compensation_Survey_2024_p9.pdf"},"started_at":"2026-11-04T19:44:52.463147"} +{"completed_at":"2026-11-04T19:44:56.409135","latency_in_ms":3055,"output":{"example_id":"chart/Unit4-PSO-SPI-2023-Infographic-BR230203INT_p1","grounded_pages":[],"layout_pages":[],"markdown":"**unit4.com**\n\nAll rights reserved. The information contained in this document is intended for general information only, as it is summary in nature and subject to change. Any third-party brand names and/or trademarks referenced are either registered or unregistered trademarks of their respective owners. IG230203INT\n\n**Copyright** **© Unit4 N.V.**\n\n**For more information, go to:**\n\n# **The Professional Services Market**\n\n### **The view from 2023**\n\n**It’s a new year, and it’s time for a new snapshot of the professional services industry. Service Performance Insight’s annual** **Professional Services Maturity Benchmark™** **\\- now in its 16** **th** **year - is the leading analysis of the sector. Here’s some of the most important topline findings from the 2023 report.**\n\n### Many KPIs have deteriorated since the previous year. Although these are minor slips, they should serve as a wake up call to leadership.\n\n## **PSOs worked harder in 2022 – but not necessarily smarter**\n\n### Deal pipelines relative to bookings forecast: **↓ 21%**\n\n### Employee billable utilization: **↓ 2.5%**\n\n### Projects delivered on time: **↓ 4%**\n\n### Project margin: **↓ 1.5%**\n\n### Annual billable revenue per consultant: **↓ 1%**\n\n### % of annual revenue target reached: **↓ 3.5%**\n\n### % of annual margin target reached: **↓ 3.3%**\n\n### Further, the movement to remote working has stalled at around **two-thirds of billable hours.** (Although this may be the optimum balance for a highly people-focused sector.)\n\n### But some subsectors are performing better than others: IT consulting leads the pack. Management consulting continues its downward growth trends – and advertising agencies, which peaked in 2020, have returned to pre-pandemic growth levels.\n\n## **Revenue growth is also down across the sector**\n\n### Despite redundancies upping the available talent pool, it still takes **10 days longer than in 2021** to hire a new fee earner and for them to become productive.\n\n### Many firms are investing heavily in moving to a **people-centric model** to satisfy both their consultants and their clients as a result. But attrition rates still remain higher than their 5-year average baseline.\n\n## **Talent is still a major challenge**\n\n### The importance of **real-time information** is now generally understood throughout the industry thanks to its ability to **enable employees to perform at a higher level.** Newer technologies are being implemented in professional services: capabilities such as collaboration tools, machine learning (ML), artificial intelligence (AI), and automation are now increasingly being used to boost operational visibility, automation efciency, performance improvement, and goal achievement.\n\n## **All aspects of maturity still depend on technology**\n\n### For all of SPI’s insights into how professional services businesses perform – and how to perform better – download the Professional Services Maturity Benchmark™ report below.\n\n## **Want to know more?**\n\n# **Click here**\n\n**25%**\n\n**20%**\n\n**15%**\n\n**10%**\n\n**5%**\n\n**0%**\n\n**Annual Revenue Growth**\n\n**IT Consult Mgmt Con. Soft. PS SaaS PS AE Advertise.**\n\n**2020 2021 2022**\n\n**13\\.4%**\n\n**10\\.6%**\n\n**23\\.4%**\n\n**18\\.3%**\n\n**15\\.8%**\n\n**10%**\n\n**12\\.4%**\n\n**9\\.1%**\n\n**9\\.5%**\n\n**13\\.2%**\n\n**7\\.9%**\n\n**7\\.3%**\n\n**13\\.1%**\n\n**7\\.4%**\n\n**10\\.6%**\n\n**10\\.2%**\n\n**9\\.4%**\n\n**6%**","pages":[{"line_numbers":[],"markdown":"**unit4.com**\n\nAll rights reserved. The information contained in this document is intended for general information only, as it is summary in nature and subject to change. Any third-party brand names and/or trademarks referenced are either registered or unregistered trademarks of their respective owners. IG230203INT\n\n**Copyright** **© Unit4 N.V.**\n\n**For more information, go to:**\n\n# **The Professional Services Market**\n\n### **The view from 2023**\n\n**It’s a new year, and it’s time for a new snapshot of the professional services industry. Service Performance Insight’s annual** **Professional Services Maturity Benchmark™** **\\- now in its 16** **th** **year - is the leading analysis of the sector. Here’s some of the most important topline findings from the 2023 report.**\n\n### Many KPIs have deteriorated since the previous year. Although these are minor slips, they should serve as a wake up call to leadership.\n\n## **PSOs worked harder in 2022 – but not necessarily smarter**\n\n### Deal pipelines relative to bookings forecast: **↓ 21%**\n\n### Employee billable utilization: **↓ 2.5%**\n\n### Projects delivered on time: **↓ 4%**\n\n### Project margin: **↓ 1.5%**\n\n### Annual billable revenue per consultant: **↓ 1%**\n\n### % of annual revenue target reached: **↓ 3.5%**\n\n### % of annual margin target reached: **↓ 3.3%**\n\n### Further, the movement to remote working has stalled at around **two-thirds of billable hours.** (Although this may be the optimum balance for a highly people-focused sector.)\n\n### But some subsectors are performing better than others: IT consulting leads the pack. Management consulting continues its downward growth trends – and advertising agencies, which peaked in 2020, have returned to pre-pandemic growth levels.\n\n## **Revenue growth is also down across the sector**\n\n### Despite redundancies upping the available talent pool, it still takes **10 days longer than in 2021** to hire a new fee earner and for them to become productive.\n\n### Many firms are investing heavily in moving to a **people-centric model** to satisfy both their consultants and their clients as a result. But attrition rates still remain higher than their 5-year average baseline.\n\n## **Talent is still a major challenge**\n\n### The importance of **real-time information** is now generally understood throughout the industry thanks to its ability to **enable employees to perform at a higher level.** Newer technologies are being implemented in professional services: capabilities such as collaboration tools, machine learning (ML), artificial intelligence (AI), and automation are now increasingly being used to boost operational visibility, automation efciency, performance improvement, and goal achievement.\n\n## **All aspects of maturity still depend on technology**\n\n### For all of SPI’s insights into how professional services businesses perform – and how to perform better – download the Professional Services Maturity Benchmark™ report below.\n\n## **Want to know more?**\n\n# **Click here**\n\n**25%**\n\n**20%**\n\n**15%**\n\n**10%**\n\n**5%**\n\n**0%**\n\n**Annual Revenue Growth**\n\n**IT Consult Mgmt Con. Soft. PS SaaS PS AE Advertise.**\n\n**2020 2021 2022**\n\n**13\\.4%**\n\n**10\\.6%**\n\n**23\\.4%**\n\n**18\\.3%**\n\n**15\\.8%**\n\n**10%**\n\n**12\\.4%**\n\n**9\\.1%**\n\n**9\\.5%**\n\n**13\\.2%**\n\n**7\\.9%**\n\n**7\\.3%**\n\n**13\\.1%**\n\n**7\\.4%**\n\n**10\\.6%**\n\n**10\\.2%**\n\n**9\\.4%**\n\n**6%**","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**unit4.com**\n\nAll rights reserved. The information contained in this document is intended for general information only, as it is summary in nature and subject to change. Any third-party brand names and/or trademarks referenced are either registered or unregistered trademarks of their respective owners. IG230203INT\n\n**Copyright** **© Unit4 N.V.**\n\n**For more information, go to:**\n\n# **The Professional Services Market**\n\n### **The view from 2023**\n\n**It’s a new year, and it’s time for a new snapshot of the professional services industry. Service Performance Insight’s annual** **Professional Services Maturity Benchmark™** **\\- now in its 16** **th** **year - is the leading analysis of the sector. Here’s some of the most important topline findings from the 2023 report.**\n\n### Many KPIs have deteriorated since the previous year. Although these are minor slips, they should serve as a wake up call to leadership.\n\n## **PSOs worked harder in 2022 – but not necessarily smarter**\n\n### Deal pipelines relative to bookings forecast: **↓ 21%**\n\n### Employee billable utilization: **↓ 2.5%**\n\n### Projects delivered on time: **↓ 4%**\n\n### Project margin: **↓ 1.5%**\n\n### Annual billable revenue per consultant: **↓ 1%**\n\n### % of annual revenue target reached: **↓ 3.5%**\n\n### % of annual margin target reached: **↓ 3.3%**\n\n### Further, the movement to remote working has stalled at around **two-thirds of billable hours.** (Although this may be the optimum balance for a highly people-focused sector.)\n\n### But some subsectors are performing better than others: IT consulting leads the pack. Management consulting continues its downward growth trends – and advertising agencies, which peaked in 2020, have returned to pre-pandemic growth levels.\n\n## **Revenue growth is also down across the sector**\n\n### Despite redundancies upping the available talent pool, it still takes **10 days longer than in 2021** to hire a new fee earner and for them to become productive.\n\n### Many firms are investing heavily in moving to a **people-centric model** to satisfy both their consultants and their clients as a result. But attrition rates still remain higher than their 5-year average baseline.\n\n## **Talent is still a major challenge**\n\n### The importance of **real-time information** is now generally understood throughout the industry thanks to its ability to **enable employees to perform at a higher level.** Newer technologies are being implemented in professional services: capabilities such as collaboration tools, machine learning (ML), artificial intelligence (AI), and automation are now increasingly being used to boost operational visibility, automation efciency, performance improvement, and goal achievement.\n\n## **All aspects of maturity still depend on technology**\n\n### For all of SPI’s insights into how professional services businesses perform – and how to perform better – download the Professional Services Maturity Benchmark™ report below.\n\n## **Want to know more?**\n\n# **Click here**\n\n**25%**\n\n**20%**\n\n**15%**\n\n**10%**\n\n**5%**\n\n**0%**\n\n**Annual Revenue Growth**\n\n**IT Consult Mgmt Con. Soft. PS SaaS PS AE Advertise.**\n\n**2020 2021 2022**\n\n**13\\.4%**\n\n**10\\.6%**\n\n**23\\.4%**\n\n**18\\.3%**\n\n**15\\.8%**\n\n**10%**\n\n**12\\.4%**\n\n**9\\.1%**\n\n**9\\.5%**\n\n**13\\.2%**\n\n**7\\.9%**\n\n**7\\.3%**\n\n**13\\.1%**\n\n**7\\.4%**\n\n**10\\.6%**\n\n**10\\.2%**\n\n**9\\.4%**\n\n**6%**","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Unit4-PSO-SPI-2023-Infographic-BR230203INT_p1.pdf"]},"response":{"errors":[],"extract_id":"extract_ed25fcfccf29d2738d45a88b311f4954","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**unit4.com**\n\nAll rights reserved. The information contained in this document is intended for general information only, as it is summary in nature and subject to change. Any third-party brand names and/or trademarks referenced are either registered or unregistered trademarks of their respective owners. IG230203INT\n\n**Copyright** **© Unit4 N.V.**\n\n**For more information, go to:**\n\n# **The Professional Services Market**\n\n### **The view from 2023**\n\n**It’s a new year, and it’s time for a new snapshot of the professional services industry. Service Performance Insight’s annual** **Professional Services Maturity Benchmark™** **\\- now in its 16** **th** **year - is the leading analysis of the sector. Here’s some of the most important topline findings from the 2023 report.**\n\n### Many KPIs have deteriorated since the previous year. Although these are minor slips, they should serve as a wake up call to leadership.\n\n## **PSOs worked harder in 2022 – but not necessarily smarter**\n\n### Deal pipelines relative to bookings forecast: **↓ 21%**\n\n### Employee billable utilization: **↓ 2.5%**\n\n### Projects delivered on time: **↓ 4%**\n\n### Project margin: **↓ 1.5%**\n\n### Annual billable revenue per consultant: **↓ 1%**\n\n### % of annual revenue target reached: **↓ 3.5%**\n\n### % of annual margin target reached: **↓ 3.3%**\n\n### Further, the movement to remote working has stalled at around **two-thirds of billable hours.** (Although this may be the optimum balance for a highly people-focused sector.)\n\n### But some subsectors are performing better than others: IT consulting leads the pack. Management consulting continues its downward growth trends – and advertising agencies, which peaked in 2020, have returned to pre-pandemic growth levels.\n\n## **Revenue growth is also down across the sector**\n\n### Despite redundancies upping the available talent pool, it still takes **10 days longer than in 2021** to hire a new fee earner and for them to become productive.\n\n### Many firms are investing heavily in moving to a **people-centric model** to satisfy both their consultants and their clients as a result. But attrition rates still remain higher than their 5-year average baseline.\n\n## **Talent is still a major challenge**\n\n### The importance of **real-time information** is now generally understood throughout the industry thanks to its ability to **enable employees to perform at a higher level.** Newer technologies are being implemented in professional services: capabilities such as collaboration tools, machine learning (ML), artificial intelligence (AI), and automation are now increasingly being used to boost operational visibility, automation efciency, performance improvement, and goal achievement.\n\n## **All aspects of maturity still depend on technology**\n\n### For all of SPI’s insights into how professional services businesses perform – and how to perform better – download the Professional Services Maturity Benchmark™ report below.\n\n## **Want to know more?**\n\n# **Click here**\n\n**25%**\n\n**20%**\n\n**15%**\n\n**10%**\n\n**5%**\n\n**0%**\n\n**Annual Revenue Growth**\n\n**IT Consult Mgmt Con. Soft. PS SaaS PS AE Advertise.**\n\n**2020 2021 2022**\n\n**13\\.4%**\n\n**10\\.6%**\n\n**23\\.4%**\n\n**18\\.3%**\n\n**15\\.8%**\n\n**10%**\n\n**12\\.4%**\n\n**9\\.1%**\n\n**9\\.5%**\n\n**13\\.2%**\n\n**7\\.9%**\n\n**7\\.3%**\n\n**13\\.1%**\n\n**7\\.4%**\n\n**10\\.6%**\n\n**10\\.2%**\n\n**9\\.4%**\n\n**6%**"],"full_content":"**unit4.com**\n\nAll rights reserved. The information contained in this document is intended for general information only, as it is summary in nature and subject to change. Any third-party brand names and/or trademarks referenced are either registered or unregistered trademarks of their respective owners. IG230203INT\n\n**Copyright** **© Unit4 N.V.**\n\n**For more information, go to:**\n\n# **The Professional Services Market**\n\n### **The view from 2023**\n\n**It’s a new year, and it’s time for a new snapshot of the professional services industry. Service Performance Insight’s annual** **Professional Services Maturity Benchmark™** **\\- now in its 16** **th** **year - is the leading analysis of the sector. Here’s some of the most important topline findings from the 2023 report.**\n\n### Many KPIs have deteriorated since the previous year. Although these are minor slips, they should serve as a wake up call to leadership.\n\n## **PSOs worked harder in 2022 – but not necessarily smarter**\n\n### Deal pipelines relative to bookings forecast: **↓ 21%**\n\n### Employee billable utilization: **↓ 2.5%**\n\n### Projects delivered on time: **↓ 4%**\n\n### Project margin: **↓ 1.5%**\n\n### Annual billable revenue per consultant: **↓ 1%**\n\n### % of annual revenue target reached: **↓ 3.5%**\n\n### % of annual margin target reached: **↓ 3.3%**\n\n### Further, the movement to remote working has stalled at around **two-thirds of billable hours.** (Although this may be the optimum balance for a highly people-focused sector.)\n\n### But some subsectors are performing better than others: IT consulting leads the pack. Management consulting continues its downward growth trends – and advertising agencies, which peaked in 2020, have returned to pre-pandemic growth levels.\n\n## **Revenue growth is also down across the sector**\n\n### Despite redundancies upping the available talent pool, it still takes **10 days longer than in 2021** to hire a new fee earner and for them to become productive.\n\n### Many firms are investing heavily in moving to a **people-centric model** to satisfy both their consultants and their clients as a result. But attrition rates still remain higher than their 5-year average baseline.\n\n## **Talent is still a major challenge**\n\n### The importance of **real-time information** is now generally understood throughout the industry thanks to its ability to **enable employees to perform at a higher level.** Newer technologies are being implemented in professional services: capabilities such as collaboration tools, machine learning (ML), artificial intelligence (AI), and automation are now increasingly being used to boost operational visibility, automation efciency, performance improvement, and goal achievement.\n\n## **All aspects of maturity still depend on technology**\n\n### For all of SPI’s insights into how professional services businesses perform – and how to perform better – download the Professional Services Maturity Benchmark™ report below.\n\n## **Want to know more?**\n\n# **Click here**\n\n**25%**\n\n**20%**\n\n**15%**\n\n**10%**\n\n**5%**\n\n**0%**\n\n**Annual Revenue Growth**\n\n**IT Consult Mgmt Con. Soft. PS SaaS PS AE Advertise.**\n\n**2020 2021 2022**\n\n**13\\.4%**\n\n**10\\.6%**\n\n**23\\.4%**\n\n**18\\.3%**\n\n**15\\.8%**\n\n**10%**\n\n**12\\.4%**\n\n**9\\.1%**\n\n**9\\.5%**\n\n**13\\.2%**\n\n**7\\.9%**\n\n**7\\.3%**\n\n**13\\.1%**\n\n**7\\.4%**\n\n**10\\.6%**\n\n**10\\.2%**\n\n**9\\.4%**\n\n**6%**","publish_date":null,"title":"The Professional Services Market","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Unit4-PSO-SPI-2023-Infographic-BR230203INT_p1.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Unit4-PSO-SPI-2023-Infographic-BR230203INT_p1.pdf"},"request":{"example_id":"chart/Unit4-PSO-SPI-2023-Infographic-BR230203INT_p1","product_type":"parse","source_file_path":"Unit4-PSO-SPI-2023-Infographic-BR230203INT_p1.pdf"},"started_at":"2026-11-04T19:44:53.353183"} +{"completed_at":"2026-11-04T19:44:55.549733","latency_in_ms":1132,"output":{"example_id":"chart/VPEG6_SIV_Information_Memorandum__June_2025__p11","grounded_pages":[],"layout_pages":[],"markdown":"VANTAGE PRIVATE EQUITY GROWTH 6, LP INFORMATION MEMORANDUM 11\n\n### **VANTAGE HAS GENERATED GLOBAL TOP QUARTILE PRIVATE EQUITY RETURNS**\n\nVantage has generated Top Quartile returns against direct buyout funds, with investors receiving the added benefits of diversification and lower risk through a Fund of Funds approach.\n\n**Vantage has Generated Top Quartile Returns when Compared to Global Direct Private Equity Funds**\n\n**Vantage vs Global Private Equity (Direct Buyout Funds) TVPI - 2009-2018**\n\n3\\.0x\n\n2\\.5x\n\n2\\.0x\n\n1\\.5x\n\n1\\.0x\n\n0\\.5x\n\nQuartile 1 Quartile 2 Quartile 3 Quartile 4\n\nSource: Based on global private equity returns on a Total Value Paid In basis from 2009-2018, Persistence in Alternative Strategies: Private Equity Buyout, Preqin. Vantage TVPI based on VPEG2 and VPEG3 TVPI as at 31 March 2025.\n\n**Vantage has generated a 2.1x platform wide Net Multiple on all funds post GFC, this is Top Quartile for Direct Buyout**\n\nVANTAGE\n\nGLOBAL PRIVATE EQUITY","pages":[{"line_numbers":[],"markdown":"VANTAGE PRIVATE EQUITY GROWTH 6, LP INFORMATION MEMORANDUM 11\n\n### **VANTAGE HAS GENERATED GLOBAL TOP QUARTILE PRIVATE EQUITY RETURNS**\n\nVantage has generated Top Quartile returns against direct buyout funds, with investors receiving the added benefits of diversification and lower risk through a Fund of Funds approach.\n\n**Vantage has Generated Top Quartile Returns when Compared to Global Direct Private Equity Funds**\n\n**Vantage vs Global Private Equity (Direct Buyout Funds) TVPI - 2009-2018**\n\n3\\.0x\n\n2\\.5x\n\n2\\.0x\n\n1\\.5x\n\n1\\.0x\n\n0\\.5x\n\nQuartile 1 Quartile 2 Quartile 3 Quartile 4\n\nSource: Based on global private equity returns on a Total Value Paid In basis from 2009-2018, Persistence in Alternative Strategies: Private Equity Buyout, Preqin. Vantage TVPI based on VPEG2 and VPEG3 TVPI as at 31 March 2025.\n\n**Vantage has generated a 2.1x platform wide Net Multiple on all funds post GFC, this is Top Quartile for Direct Buyout**\n\nVANTAGE\n\nGLOBAL PRIVATE EQUITY","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"VANTAGE PRIVATE EQUITY GROWTH 6, LP INFORMATION MEMORANDUM 11\n\n### **VANTAGE HAS GENERATED GLOBAL TOP QUARTILE PRIVATE EQUITY RETURNS**\n\nVantage has generated Top Quartile returns against direct buyout funds, with investors receiving the added benefits of diversification and lower risk through a Fund of Funds approach.\n\n**Vantage has Generated Top Quartile Returns when Compared to Global Direct Private Equity Funds**\n\n**Vantage vs Global Private Equity (Direct Buyout Funds) TVPI - 2009-2018**\n\n3\\.0x\n\n2\\.5x\n\n2\\.0x\n\n1\\.5x\n\n1\\.0x\n\n0\\.5x\n\nQuartile 1 Quartile 2 Quartile 3 Quartile 4\n\nSource: Based on global private equity returns on a Total Value Paid In basis from 2009-2018, Persistence in Alternative Strategies: Private Equity Buyout, Preqin. Vantage TVPI based on VPEG2 and VPEG3 TVPI as at 31 March 2025.\n\n**Vantage has generated a 2.1x platform wide Net Multiple on all funds post GFC, this is Top Quartile for Direct Buyout**\n\nVANTAGE\n\nGLOBAL PRIVATE EQUITY","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/VPEG6_SIV_Information_Memorandum__June_2025__p11.pdf"]},"response":{"errors":[],"extract_id":"extract_58898a858d267de4a70dbb8e29139d88","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["VANTAGE PRIVATE EQUITY GROWTH 6, LP INFORMATION MEMORANDUM 11\n\n### **VANTAGE HAS GENERATED GLOBAL TOP QUARTILE PRIVATE EQUITY RETURNS**\n\nVantage has generated Top Quartile returns against direct buyout funds, with investors receiving the added benefits of diversification and lower risk through a Fund of Funds approach.\n\n**Vantage has Generated Top Quartile Returns when Compared to Global Direct Private Equity Funds**\n\n**Vantage vs Global Private Equity (Direct Buyout Funds) TVPI - 2009-2018**\n\n3\\.0x\n\n2\\.5x\n\n2\\.0x\n\n1\\.5x\n\n1\\.0x\n\n0\\.5x\n\nQuartile 1 Quartile 2 Quartile 3 Quartile 4\n\nSource: Based on global private equity returns on a Total Value Paid In basis from 2009-2018, Persistence in Alternative Strategies: Private Equity Buyout, Preqin. Vantage TVPI based on VPEG2 and VPEG3 TVPI as at 31 March 2025.\n\n**Vantage has generated a 2.1x platform wide Net Multiple on all funds post GFC, this is Top Quartile for Direct Buyout**\n\nVANTAGE\n\nGLOBAL PRIVATE EQUITY"],"full_content":"VANTAGE PRIVATE EQUITY GROWTH 6, LP INFORMATION MEMORANDUM 11\n\n### **VANTAGE HAS GENERATED GLOBAL TOP QUARTILE PRIVATE EQUITY RETURNS**\n\nVantage has generated Top Quartile returns against direct buyout funds, with investors receiving the added benefits of diversification and lower risk through a Fund of Funds approach.\n\n**Vantage has Generated Top Quartile Returns when Compared to Global Direct Private Equity Funds**\n\n**Vantage vs Global Private Equity (Direct Buyout Funds) TVPI - 2009-2018**\n\n3\\.0x\n\n2\\.5x\n\n2\\.0x\n\n1\\.5x\n\n1\\.0x\n\n0\\.5x\n\nQuartile 1 Quartile 2 Quartile 3 Quartile 4\n\nSource: Based on global private equity returns on a Total Value Paid In basis from 2009-2018, Persistence in Alternative Strategies: Private Equity Buyout, Preqin. Vantage TVPI based on VPEG2 and VPEG3 TVPI as at 31 March 2025.\n\n**Vantage has generated a 2.1x platform wide Net Multiple on all funds post GFC, this is Top Quartile for Direct Buyout**\n\nVANTAGE\n\nGLOBAL PRIVATE EQUITY","publish_date":null,"title":"VANTAGE HAS GENERATED GLOBAL TOP QUARTILE PRIVATE EQUITY RETURNS","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/VPEG6_SIV_Information_Memorandum__June_2025__p11.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/VPEG6_SIV_Information_Memorandum__June_2025__p11.pdf"},"request":{"example_id":"chart/VPEG6_SIV_Information_Memorandum__June_2025__p11","product_type":"parse","source_file_path":"VPEG6_SIV_Information_Memorandum__June_2025__p11.pdf"},"started_at":"2026-11-04T19:44:54.417113"} +{"completed_at":"2026-11-04T19:44:57.352547","latency_in_ms":1221,"output":{"example_id":"chart/VPEG6_SIV_Information_Memorandum__June_2025__p20","grounded_pages":[],"layout_pages":[],"markdown":"VANTAGE PRIVATE EQUITY GROWTH 6, LP INFORMATION MEMORANDUM 20\n\n### **SUPERIOR RETURNS AT LOWER RISK LEVELS REFLECT LOWER LEVERAGE AND ENTRY VALUATIONS IN AUSTRALIA**\n\nAnalysis by IFM Investors of investment valuations and leverage as a proxy for risk shows that in addition to Australian Private Equity delivering higher returns historically, it’s also done so with relatively lower risk.\n\nThe median entry valuations in Australian Private Equity is significa tly lower than in the US and Europe over an extended period of time. This lower pricing contributes to lower debt, a factor helping to reduce investment risk.\n\nLower pricing and lower debt are factors that have supported Australian Private Equity performance and contributed to Australia outperforming all other regions.\n\n**Materially lower entry valuations in Australia (Median entry EV/EBITDA valuation) Lower leverage reduces risk in Australian investment (Median entry Net Debt/EBITDA valuation)**\n\n18x\n\n15x\n\n12x\n\n9x\n\n6x\n\n3x\n\n2006 2008 2010 2012 2014 2016 2018 2020 2022\n\nSource: IFM Investors, Pitchbook Annual PE Breakdown for US and EU data; AVCAL for ANZ data. AVCAL, now AIC, stopped collecting deal metric data in 2018.\n\n9x\n\n7\\.5x\n\n6x\n\n4\\.5x\n\n3x\n\n1\\.5x\n\n2006 2008 2010 2012 2014 2016 2018 2020 2022\n\nSource: IFM Investors, Pitchbook Annual PE Breakdown for US and EU data; AVCAL for ANZ data. AVCAL, now AIC, stopped collecting deal metric data in 2018.\n\nUS PE\n\nEUROPE PE\n\nAUSTRALIA PE\n\n### SECTION 2","pages":[{"line_numbers":[],"markdown":"VANTAGE PRIVATE EQUITY GROWTH 6, LP INFORMATION MEMORANDUM 20\n\n### **SUPERIOR RETURNS AT LOWER RISK LEVELS REFLECT LOWER LEVERAGE AND ENTRY VALUATIONS IN AUSTRALIA**\n\nAnalysis by IFM Investors of investment valuations and leverage as a proxy for risk shows that in addition to Australian Private Equity delivering higher returns historically, it’s also done so with relatively lower risk.\n\nThe median entry valuations in Australian Private Equity is significa tly lower than in the US and Europe over an extended period of time. This lower pricing contributes to lower debt, a factor helping to reduce investment risk.\n\nLower pricing and lower debt are factors that have supported Australian Private Equity performance and contributed to Australia outperforming all other regions.\n\n**Materially lower entry valuations in Australia (Median entry EV/EBITDA valuation) Lower leverage reduces risk in Australian investment (Median entry Net Debt/EBITDA valuation)**\n\n18x\n\n15x\n\n12x\n\n9x\n\n6x\n\n3x\n\n2006 2008 2010 2012 2014 2016 2018 2020 2022\n\nSource: IFM Investors, Pitchbook Annual PE Breakdown for US and EU data; AVCAL for ANZ data. AVCAL, now AIC, stopped collecting deal metric data in 2018.\n\n9x\n\n7\\.5x\n\n6x\n\n4\\.5x\n\n3x\n\n1\\.5x\n\n2006 2008 2010 2012 2014 2016 2018 2020 2022\n\nSource: IFM Investors, Pitchbook Annual PE Breakdown for US and EU data; AVCAL for ANZ data. AVCAL, now AIC, stopped collecting deal metric data in 2018.\n\nUS PE\n\nEUROPE PE\n\nAUSTRALIA PE\n\n### SECTION 2","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"VANTAGE PRIVATE EQUITY GROWTH 6, LP INFORMATION MEMORANDUM 20\n\n### **SUPERIOR RETURNS AT LOWER RISK LEVELS REFLECT LOWER LEVERAGE AND ENTRY VALUATIONS IN AUSTRALIA**\n\nAnalysis by IFM Investors of investment valuations and leverage as a proxy for risk shows that in addition to Australian Private Equity delivering higher returns historically, it’s also done so with relatively lower risk.\n\nThe median entry valuations in Australian Private Equity is significa tly lower than in the US and Europe over an extended period of time. This lower pricing contributes to lower debt, a factor helping to reduce investment risk.\n\nLower pricing and lower debt are factors that have supported Australian Private Equity performance and contributed to Australia outperforming all other regions.\n\n**Materially lower entry valuations in Australia (Median entry EV/EBITDA valuation) Lower leverage reduces risk in Australian investment (Median entry Net Debt/EBITDA valuation)**\n\n18x\n\n15x\n\n12x\n\n9x\n\n6x\n\n3x\n\n2006 2008 2010 2012 2014 2016 2018 2020 2022\n\nSource: IFM Investors, Pitchbook Annual PE Breakdown for US and EU data; AVCAL for ANZ data. AVCAL, now AIC, stopped collecting deal metric data in 2018.\n\n9x\n\n7\\.5x\n\n6x\n\n4\\.5x\n\n3x\n\n1\\.5x\n\n2006 2008 2010 2012 2014 2016 2018 2020 2022\n\nSource: IFM Investors, Pitchbook Annual PE Breakdown for US and EU data; AVCAL for ANZ data. AVCAL, now AIC, stopped collecting deal metric data in 2018.\n\nUS PE\n\nEUROPE PE\n\nAUSTRALIA PE\n\n### SECTION 2","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/VPEG6_SIV_Information_Memorandum__June_2025__p20.pdf"]},"response":{"errors":[],"extract_id":"extract_68198d734b54530523eb6c74943d2449","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["VANTAGE PRIVATE EQUITY GROWTH 6, LP INFORMATION MEMORANDUM 20\n\n### **SUPERIOR RETURNS AT LOWER RISK LEVELS REFLECT LOWER LEVERAGE AND ENTRY VALUATIONS IN AUSTRALIA**\n\nAnalysis by IFM Investors of investment valuations and leverage as a proxy for risk shows that in addition to Australian Private Equity delivering higher returns historically, it’s also done so with relatively lower risk.\n\nThe median entry valuations in Australian Private Equity is significa tly lower than in the US and Europe over an extended period of time. This lower pricing contributes to lower debt, a factor helping to reduce investment risk.\n\nLower pricing and lower debt are factors that have supported Australian Private Equity performance and contributed to Australia outperforming all other regions.\n\n**Materially lower entry valuations in Australia (Median entry EV/EBITDA valuation) Lower leverage reduces risk in Australian investment (Median entry Net Debt/EBITDA valuation)**\n\n18x\n\n15x\n\n12x\n\n9x\n\n6x\n\n3x\n\n2006 2008 2010 2012 2014 2016 2018 2020 2022\n\nSource: IFM Investors, Pitchbook Annual PE Breakdown for US and EU data; AVCAL for ANZ data. AVCAL, now AIC, stopped collecting deal metric data in 2018.\n\n9x\n\n7\\.5x\n\n6x\n\n4\\.5x\n\n3x\n\n1\\.5x\n\n2006 2008 2010 2012 2014 2016 2018 2020 2022\n\nSource: IFM Investors, Pitchbook Annual PE Breakdown for US and EU data; AVCAL for ANZ data. AVCAL, now AIC, stopped collecting deal metric data in 2018.\n\nUS PE\n\nEUROPE PE\n\nAUSTRALIA PE\n\n### SECTION 2"],"full_content":"VANTAGE PRIVATE EQUITY GROWTH 6, LP INFORMATION MEMORANDUM 20\n\n### **SUPERIOR RETURNS AT LOWER RISK LEVELS REFLECT LOWER LEVERAGE AND ENTRY VALUATIONS IN AUSTRALIA**\n\nAnalysis by IFM Investors of investment valuations and leverage as a proxy for risk shows that in addition to Australian Private Equity delivering higher returns historically, it’s also done so with relatively lower risk.\n\nThe median entry valuations in Australian Private Equity is significa tly lower than in the US and Europe over an extended period of time. This lower pricing contributes to lower debt, a factor helping to reduce investment risk.\n\nLower pricing and lower debt are factors that have supported Australian Private Equity performance and contributed to Australia outperforming all other regions.\n\n**Materially lower entry valuations in Australia (Median entry EV/EBITDA valuation) Lower leverage reduces risk in Australian investment (Median entry Net Debt/EBITDA valuation)**\n\n18x\n\n15x\n\n12x\n\n9x\n\n6x\n\n3x\n\n2006 2008 2010 2012 2014 2016 2018 2020 2022\n\nSource: IFM Investors, Pitchbook Annual PE Breakdown for US and EU data; AVCAL for ANZ data. AVCAL, now AIC, stopped collecting deal metric data in 2018.\n\n9x\n\n7\\.5x\n\n6x\n\n4\\.5x\n\n3x\n\n1\\.5x\n\n2006 2008 2010 2012 2014 2016 2018 2020 2022\n\nSource: IFM Investors, Pitchbook Annual PE Breakdown for US and EU data; AVCAL for ANZ data. AVCAL, now AIC, stopped collecting deal metric data in 2018.\n\nUS PE\n\nEUROPE PE\n\nAUSTRALIA PE\n\n### SECTION 2","publish_date":null,"title":"SECTION 2","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/VPEG6_SIV_Information_Memorandum__June_2025__p20.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/VPEG6_SIV_Information_Memorandum__June_2025__p20.pdf"},"request":{"example_id":"chart/VPEG6_SIV_Information_Memorandum__June_2025__p20","product_type":"parse","source_file_path":"VPEG6_SIV_Information_Memorandum__June_2025__p20.pdf"},"started_at":"2026-11-04T19:44:56.130583"} +{"completed_at":"2026-11-04T19:44:58.714921","latency_in_ms":1586,"output":{"example_id":"chart/Visa-Fiscal-2025-Annual-Report_p10","grounded_pages":[],"layout_pages":[],"markdown":"CEO Letter\n\n**8 | Visa**\n\n**Farewell, guest checkout**\n\nSince the rise of the consumer internet 30 years ago, ecommerce checkout has been cumbersome, requiring consumers to manually enter card details, retype shipping information, remember passwords and navigate multiple screens. That era is ending. Manual entry guest checkout is headed toward extinction, replaced by seamless, secure and frictionless payment experiences.\n\nThe share of Visa ecommerce transactions using manual entry guest\n\ncheckout has dropped from 44 percent in\n\n2019 to just 16 percent in 2025, and among our top 25 ecommerce sellers, it is less than four percent. Just as smartphones replaced the need to remember phone numbers and search engines replaced remembering URLs, manual entry is giving way to faster and easier options.\n\nCard-on-file commerce, where merchants store credentials for recurring payments, and \"button\" checkout via digital wallets from technology platforms and financial\n\ninstitutions now account for 84 percent of Visa ecommerce transactions.9\n\n**Share of Visa ecommerce transactions by checkout type**\n\n**44%**\n\n**56%**\n\n**2019**\n\n**35%**\n\n**65%**\n\n**2020**\n\n**30%**\n\n**70%**\n\n**2021**\n\n**26%**\n\n**74%**\n\n**2022**\n\n**24%**\n\n**76%**\n\n**2023**\n\n**19%**\n\n**81%**\n\n**2024**\n\n**16%**\n\n**84%**\n\n**2025**\n\nCard-on-file & “button” checkout\n\nManual entry guest checkout\n\n_Note: Reflects Visa processed transactions, among card-not-present approved transactions. Card-on-file includes both tokenized and non-tokenized card-on-file transactions. Data represents share of transactions in September of each year. Excludes Russia, China and Visa Direct in Peru._\n\n**Hello, tokenization**\n\nThis shift away from manual entry guest checkout is powered by Visa’s tokenization technology. Tokenization replaces sensitive card information with\n\na unique, secure digital identifier that can only be used in the specific context it was issued. Even if intercepted, it is useless to anyone else, keeping payments safe. It is the technology that enables both digital wallet checkout experiences and secure card-on-file systems. For more than a decade, Visa has been the driving force behind the use of tokenization in payments. Today, more than 50 percent of ecommerce transactions in our ecosystem are tokenized, driving closer to our ultimate goal of 100 percent tokenization.9 This technology delivered an authorization lift of nearly five percent for ecommerce sellers, resulting in more than $110 billion in incremental sales in 2024.10 In addition, transactions using Visa tokens had more than 35 percent lower fraud, saving $1.1 billion for the year.10 We are augmenting tokens with enhanced data, passkeys and other innovations to continue improving the speed and security of ecommerce transactions.\n\n**Tokenized share of Visa ecommerce transactions**\n\nNon-tokenized transactions\n\nTokenized transactions\n\n**2019 2020 2021 2022 2023 2024 2025**\n\n**93%**\n\n**13%**\n\n**87%**\n\n**26%**\n\n**74%**\n\n**36%**\n\n**64%**\n\n**45%**\n\n**55%**\n\n**54%**\n\n**46% 96%**\n\n**44% 7%**\n\n_Note: Reflects Visa processed transactions, among card-not-present approved transactions. Tokenized transactions include both \"button\" checkout and tokenized card-on-file transactions. Data represents share of transactions in September of each year. Excludes Russia, China and Visa Direct in Peru._ _9_ _Among Visa processed card-not-present transactions._ _Excludes Russia, China and Visa Direct in Peru._\n\n_10 Calendar year_ _2024\\. Excludes prepaid volumes._","pages":[{"line_numbers":[],"markdown":"CEO Letter\n\n**8 | Visa**\n\n**Farewell, guest checkout**\n\nSince the rise of the consumer internet 30 years ago, ecommerce checkout has been cumbersome, requiring consumers to manually enter card details, retype shipping information, remember passwords and navigate multiple screens. That era is ending. Manual entry guest checkout is headed toward extinction, replaced by seamless, secure and frictionless payment experiences.\n\nThe share of Visa ecommerce transactions using manual entry guest\n\ncheckout has dropped from 44 percent in\n\n2019 to just 16 percent in 2025, and among our top 25 ecommerce sellers, it is less than four percent. Just as smartphones replaced the need to remember phone numbers and search engines replaced remembering URLs, manual entry is giving way to faster and easier options.\n\nCard-on-file commerce, where merchants store credentials for recurring payments, and \"button\" checkout via digital wallets from technology platforms and financial\n\ninstitutions now account for 84 percent of Visa ecommerce transactions.9\n\n**Share of Visa ecommerce transactions by checkout type**\n\n**44%**\n\n**56%**\n\n**2019**\n\n**35%**\n\n**65%**\n\n**2020**\n\n**30%**\n\n**70%**\n\n**2021**\n\n**26%**\n\n**74%**\n\n**2022**\n\n**24%**\n\n**76%**\n\n**2023**\n\n**19%**\n\n**81%**\n\n**2024**\n\n**16%**\n\n**84%**\n\n**2025**\n\nCard-on-file & “button” checkout\n\nManual entry guest checkout\n\n_Note: Reflects Visa processed transactions, among card-not-present approved transactions. Card-on-file includes both tokenized and non-tokenized card-on-file transactions. Data represents share of transactions in September of each year. Excludes Russia, China and Visa Direct in Peru._\n\n**Hello, tokenization**\n\nThis shift away from manual entry guest checkout is powered by Visa’s tokenization technology. Tokenization replaces sensitive card information with\n\na unique, secure digital identifier that can only be used in the specific context it was issued. Even if intercepted, it is useless to anyone else, keeping payments safe. It is the technology that enables both digital wallet checkout experiences and secure card-on-file systems. For more than a decade, Visa has been the driving force behind the use of tokenization in payments. Today, more than 50 percent of ecommerce transactions in our ecosystem are tokenized, driving closer to our ultimate goal of 100 percent tokenization.9 This technology delivered an authorization lift of nearly five percent for ecommerce sellers, resulting in more than $110 billion in incremental sales in 2024.10 In addition, transactions using Visa tokens had more than 35 percent lower fraud, saving $1.1 billion for the year.10 We are augmenting tokens with enhanced data, passkeys and other innovations to continue improving the speed and security of ecommerce transactions.\n\n**Tokenized share of Visa ecommerce transactions**\n\nNon-tokenized transactions\n\nTokenized transactions\n\n**2019 2020 2021 2022 2023 2024 2025**\n\n**93%**\n\n**13%**\n\n**87%**\n\n**26%**\n\n**74%**\n\n**36%**\n\n**64%**\n\n**45%**\n\n**55%**\n\n**54%**\n\n**46% 96%**\n\n**44% 7%**\n\n_Note: Reflects Visa processed transactions, among card-not-present approved transactions. Tokenized transactions include both \"button\" checkout and tokenized card-on-file transactions. Data represents share of transactions in September of each year. Excludes Russia, China and Visa Direct in Peru._ _9_ _Among Visa processed card-not-present transactions._ _Excludes Russia, China and Visa Direct in Peru._\n\n_10 Calendar year_ _2024\\. Excludes prepaid volumes._","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"CEO Letter\n\n**8 | Visa**\n\n**Farewell, guest checkout**\n\nSince the rise of the consumer internet 30 years ago, ecommerce checkout has been cumbersome, requiring consumers to manually enter card details, retype shipping information, remember passwords and navigate multiple screens. That era is ending. Manual entry guest checkout is headed toward extinction, replaced by seamless, secure and frictionless payment experiences.\n\nThe share of Visa ecommerce transactions using manual entry guest\n\ncheckout has dropped from 44 percent in\n\n2019 to just 16 percent in 2025, and among our top 25 ecommerce sellers, it is less than four percent. Just as smartphones replaced the need to remember phone numbers and search engines replaced remembering URLs, manual entry is giving way to faster and easier options.\n\nCard-on-file commerce, where merchants store credentials for recurring payments, and \"button\" checkout via digital wallets from technology platforms and financial\n\ninstitutions now account for 84 percent of Visa ecommerce transactions.9\n\n**Share of Visa ecommerce transactions by checkout type**\n\n**44%**\n\n**56%**\n\n**2019**\n\n**35%**\n\n**65%**\n\n**2020**\n\n**30%**\n\n**70%**\n\n**2021**\n\n**26%**\n\n**74%**\n\n**2022**\n\n**24%**\n\n**76%**\n\n**2023**\n\n**19%**\n\n**81%**\n\n**2024**\n\n**16%**\n\n**84%**\n\n**2025**\n\nCard-on-file & “button” checkout\n\nManual entry guest checkout\n\n_Note: Reflects Visa processed transactions, among card-not-present approved transactions. Card-on-file includes both tokenized and non-tokenized card-on-file transactions. Data represents share of transactions in September of each year. Excludes Russia, China and Visa Direct in Peru._\n\n**Hello, tokenization**\n\nThis shift away from manual entry guest checkout is powered by Visa’s tokenization technology. Tokenization replaces sensitive card information with\n\na unique, secure digital identifier that can only be used in the specific context it was issued. Even if intercepted, it is useless to anyone else, keeping payments safe. It is the technology that enables both digital wallet checkout experiences and secure card-on-file systems. For more than a decade, Visa has been the driving force behind the use of tokenization in payments. Today, more than 50 percent of ecommerce transactions in our ecosystem are tokenized, driving closer to our ultimate goal of 100 percent tokenization.9 This technology delivered an authorization lift of nearly five percent for ecommerce sellers, resulting in more than $110 billion in incremental sales in 2024.10 In addition, transactions using Visa tokens had more than 35 percent lower fraud, saving $1.1 billion for the year.10 We are augmenting tokens with enhanced data, passkeys and other innovations to continue improving the speed and security of ecommerce transactions.\n\n**Tokenized share of Visa ecommerce transactions**\n\nNon-tokenized transactions\n\nTokenized transactions\n\n**2019 2020 2021 2022 2023 2024 2025**\n\n**93%**\n\n**13%**\n\n**87%**\n\n**26%**\n\n**74%**\n\n**36%**\n\n**64%**\n\n**45%**\n\n**55%**\n\n**54%**\n\n**46% 96%**\n\n**44% 7%**\n\n_Note: Reflects Visa processed transactions, among card-not-present approved transactions. Tokenized transactions include both \"button\" checkout and tokenized card-on-file transactions. Data represents share of transactions in September of each year. Excludes Russia, China and Visa Direct in Peru._ _9_ _Among Visa processed card-not-present transactions._ _Excludes Russia, China and Visa Direct in Peru._\n\n_10 Calendar year_ _2024\\. Excludes prepaid volumes._","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Visa-Fiscal-2025-Annual-Report_p10.pdf"]},"response":{"errors":[],"extract_id":"extract_22009c49fb7d714d4bc1bb16ceee67d8","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["CEO Letter\n\n**8 | Visa**\n\n**Farewell, guest checkout**\n\nSince the rise of the consumer internet 30 years ago, ecommerce checkout has been cumbersome, requiring consumers to manually enter card details, retype shipping information, remember passwords and navigate multiple screens. That era is ending. Manual entry guest checkout is headed toward extinction, replaced by seamless, secure and frictionless payment experiences.\n\nThe share of Visa ecommerce transactions using manual entry guest\n\ncheckout has dropped from 44 percent in\n\n2019 to just 16 percent in 2025, and among our top 25 ecommerce sellers, it is less than four percent. Just as smartphones replaced the need to remember phone numbers and search engines replaced remembering URLs, manual entry is giving way to faster and easier options.\n\nCard-on-file commerce, where merchants store credentials for recurring payments, and \"button\" checkout via digital wallets from technology platforms and financial\n\ninstitutions now account for 84 percent of Visa ecommerce transactions.9\n\n**Share of Visa ecommerce transactions by checkout type**\n\n**44%**\n\n**56%**\n\n**2019**\n\n**35%**\n\n**65%**\n\n**2020**\n\n**30%**\n\n**70%**\n\n**2021**\n\n**26%**\n\n**74%**\n\n**2022**\n\n**24%**\n\n**76%**\n\n**2023**\n\n**19%**\n\n**81%**\n\n**2024**\n\n**16%**\n\n**84%**\n\n**2025**\n\nCard-on-file & “button” checkout\n\nManual entry guest checkout\n\n_Note: Reflects Visa processed transactions, among card-not-present approved transactions. Card-on-file includes both tokenized and non-tokenized card-on-file transactions. Data represents share of transactions in September of each year. Excludes Russia, China and Visa Direct in Peru._\n\n**Hello, tokenization**\n\nThis shift away from manual entry guest checkout is powered by Visa’s tokenization technology. Tokenization replaces sensitive card information with\n\na unique, secure digital identifier that can only be used in the specific context it was issued. Even if intercepted, it is useless to anyone else, keeping payments safe. It is the technology that enables both digital wallet checkout experiences and secure card-on-file systems. For more than a decade, Visa has been the driving force behind the use of tokenization in payments. Today, more than 50 percent of ecommerce transactions in our ecosystem are tokenized, driving closer to our ultimate goal of 100 percent tokenization.9 This technology delivered an authorization lift of nearly five percent for ecommerce sellers, resulting in more than $110 billion in incremental sales in 2024.10 In addition, transactions using Visa tokens had more than 35 percent lower fraud, saving $1.1 billion for the year.10 We are augmenting tokens with enhanced data, passkeys and other innovations to continue improving the speed and security of ecommerce transactions.\n\n**Tokenized share of Visa ecommerce transactions**\n\nNon-tokenized transactions\n\nTokenized transactions\n\n**2019 2020 2021 2022 2023 2024 2025**\n\n**93%**\n\n**13%**\n\n**87%**\n\n**26%**\n\n**74%**\n\n**36%**\n\n**64%**\n\n**45%**\n\n**55%**\n\n**54%**\n\n**46% 96%**\n\n**44% 7%**\n\n_Note: Reflects Visa processed transactions, among card-not-present approved transactions. Tokenized transactions include both \"button\" checkout and tokenized card-on-file transactions. Data represents share of transactions in September of each year. Excludes Russia, China and Visa Direct in Peru._ _9_ _Among Visa processed card-not-present transactions._ _Excludes Russia, China and Visa Direct in Peru._\n\n_10 Calendar year_ _2024\\. Excludes prepaid volumes._"],"full_content":"CEO Letter\n\n**8 | Visa**\n\n**Farewell, guest checkout**\n\nSince the rise of the consumer internet 30 years ago, ecommerce checkout has been cumbersome, requiring consumers to manually enter card details, retype shipping information, remember passwords and navigate multiple screens. That era is ending. Manual entry guest checkout is headed toward extinction, replaced by seamless, secure and frictionless payment experiences.\n\nThe share of Visa ecommerce transactions using manual entry guest\n\ncheckout has dropped from 44 percent in\n\n2019 to just 16 percent in 2025, and among our top 25 ecommerce sellers, it is less than four percent. Just as smartphones replaced the need to remember phone numbers and search engines replaced remembering URLs, manual entry is giving way to faster and easier options.\n\nCard-on-file commerce, where merchants store credentials for recurring payments, and \"button\" checkout via digital wallets from technology platforms and financial\n\ninstitutions now account for 84 percent of Visa ecommerce transactions.9\n\n**Share of Visa ecommerce transactions by checkout type**\n\n**44%**\n\n**56%**\n\n**2019**\n\n**35%**\n\n**65%**\n\n**2020**\n\n**30%**\n\n**70%**\n\n**2021**\n\n**26%**\n\n**74%**\n\n**2022**\n\n**24%**\n\n**76%**\n\n**2023**\n\n**19%**\n\n**81%**\n\n**2024**\n\n**16%**\n\n**84%**\n\n**2025**\n\nCard-on-file & “button” checkout\n\nManual entry guest checkout\n\n_Note: Reflects Visa processed transactions, among card-not-present approved transactions. Card-on-file includes both tokenized and non-tokenized card-on-file transactions. Data represents share of transactions in September of each year. Excludes Russia, China and Visa Direct in Peru._\n\n**Hello, tokenization**\n\nThis shift away from manual entry guest checkout is powered by Visa’s tokenization technology. Tokenization replaces sensitive card information with\n\na unique, secure digital identifier that can only be used in the specific context it was issued. Even if intercepted, it is useless to anyone else, keeping payments safe. It is the technology that enables both digital wallet checkout experiences and secure card-on-file systems. For more than a decade, Visa has been the driving force behind the use of tokenization in payments. Today, more than 50 percent of ecommerce transactions in our ecosystem are tokenized, driving closer to our ultimate goal of 100 percent tokenization.9 This technology delivered an authorization lift of nearly five percent for ecommerce sellers, resulting in more than $110 billion in incremental sales in 2024.10 In addition, transactions using Visa tokens had more than 35 percent lower fraud, saving $1.1 billion for the year.10 We are augmenting tokens with enhanced data, passkeys and other innovations to continue improving the speed and security of ecommerce transactions.\n\n**Tokenized share of Visa ecommerce transactions**\n\nNon-tokenized transactions\n\nTokenized transactions\n\n**2019 2020 2021 2022 2023 2024 2025**\n\n**93%**\n\n**13%**\n\n**87%**\n\n**26%**\n\n**74%**\n\n**36%**\n\n**64%**\n\n**45%**\n\n**55%**\n\n**54%**\n\n**46% 96%**\n\n**44% 7%**\n\n_Note: Reflects Visa processed transactions, among card-not-present approved transactions. Tokenized transactions include both \"button\" checkout and tokenized card-on-file transactions. Data represents share of transactions in September of each year. Excludes Russia, China and Visa Direct in Peru._ _9_ _Among Visa processed card-not-present transactions._ _Excludes Russia, China and Visa Direct in Peru._\n\n_10 Calendar year_ _2024\\. Excludes prepaid volumes._","publish_date":null,"title":"Share of Visa ecommerce transactions by checkout type Tokenized share of Visa ecommerce transactions","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Visa-Fiscal-2025-Annual-Report_p10.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Visa-Fiscal-2025-Annual-Report_p10.pdf"},"request":{"example_id":"chart/Visa-Fiscal-2025-Annual-Report_p10","product_type":"parse","source_file_path":"Visa-Fiscal-2025-Annual-Report_p10.pdf"},"started_at":"2026-11-04T19:44:57.128601"} +{"completed_at":"2026-11-04T19:45:01.457486","latency_in_ms":2921,"output":{"example_id":"chart/WEF_Asset_Tokenization_in_Financial_Markets_2025_p22","grounded_pages":[],"layout_pages":[],"markdown":"# Contents\n\n## 22\n\n# Tokenized assets\n\n# 3\\.2 Asset issuance\n\n# Equities\n\n# Public equities are shares of publicly traded companies listed on stock exchanges and offered to a broad investor base. The global public equity market was valued at nearly\n\n## $115 trillion in 2023. ¹⁰ The market\n\n# capitalization of tokenized public stocks was estimated at nearly $16 million\n\n## by March 2025. ¹¹\n\n# Current public equities markets in AEs are already highly efficient and benefit from decades of technology modernization and proven intermediary chains delivering financial services. Due to existing efficiencies in AEs, tokenizing public equities may not be\n\n## prioritized over other asset classes. ¹²\n\n# However, opportunities exist, particularly in EMEs, to improve efficiencies and provide\n\n## access to equity markets. ¹³\n\n## Source: RWA.xyz (2025)\n\n## F I G U R E 7\n\n# Tokenized stocks\n\n## Backed-issued stocks bCSPX, bCOIN and bNVDA represent almost 90% of tokenized stocks in value.\n\n## bCSPX bCOIN bNVDA sMSTR TSLA.d Others\n\n## $18M\n\n## $16M\n\n## $14M\n\n## $12M\n\n## $10M\n\n## $8M\n\n## $6M\n\n## $4M\n\n## $2M\n\n## $0M Jan 23 Mar 23 May 23 Jul 23 Sep 23 Nov 23 Jan 24 Mar 24 May 24 Jul 24 Sep 24 Nov 24 Jan 25 Mar 25 Sep 24","pages":[{"line_numbers":[],"markdown":"# Contents\n\n## 22\n\n# Tokenized assets\n\n# 3\\.2 Asset issuance\n\n# Equities\n\n# Public equities are shares of publicly traded companies listed on stock exchanges and offered to a broad investor base. The global public equity market was valued at nearly\n\n## $115 trillion in 2023. ¹⁰ The market\n\n# capitalization of tokenized public stocks was estimated at nearly $16 million\n\n## by March 2025. ¹¹\n\n# Current public equities markets in AEs are already highly efficient and benefit from decades of technology modernization and proven intermediary chains delivering financial services. Due to existing efficiencies in AEs, tokenizing public equities may not be\n\n## prioritized over other asset classes. ¹²\n\n# However, opportunities exist, particularly in EMEs, to improve efficiencies and provide\n\n## access to equity markets. ¹³\n\n## Source: RWA.xyz (2025)\n\n## F I G U R E 7\n\n# Tokenized stocks\n\n## Backed-issued stocks bCSPX, bCOIN and bNVDA represent almost 90% of tokenized stocks in value.\n\n## bCSPX bCOIN bNVDA sMSTR TSLA.d Others\n\n## $18M\n\n## $16M\n\n## $14M\n\n## $12M\n\n## $10M\n\n## $8M\n\n## $6M\n\n## $4M\n\n## $2M\n\n## $0M Jan 23 Mar 23 May 23 Jul 23 Sep 23 Nov 23 Jan 24 Mar 24 May 24 Jul 24 Sep 24 Nov 24 Jan 25 Mar 25 Sep 24","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"# Contents\n\n## 22\n\n# Tokenized assets\n\n# 3\\.2 Asset issuance\n\n# Equities\n\n# Public equities are shares of publicly traded companies listed on stock exchanges and offered to a broad investor base. The global public equity market was valued at nearly\n\n## $115 trillion in 2023. ¹⁰ The market\n\n# capitalization of tokenized public stocks was estimated at nearly $16 million\n\n## by March 2025. ¹¹\n\n# Current public equities markets in AEs are already highly efficient and benefit from decades of technology modernization and proven intermediary chains delivering financial services. Due to existing efficiencies in AEs, tokenizing public equities may not be\n\n## prioritized over other asset classes. ¹²\n\n# However, opportunities exist, particularly in EMEs, to improve efficiencies and provide\n\n## access to equity markets. ¹³\n\n## Source: RWA.xyz (2025)\n\n## F I G U R E 7\n\n# Tokenized stocks\n\n## Backed-issued stocks bCSPX, bCOIN and bNVDA represent almost 90% of tokenized stocks in value.\n\n## bCSPX bCOIN bNVDA sMSTR TSLA.d Others\n\n## $18M\n\n## $16M\n\n## $14M\n\n## $12M\n\n## $10M\n\n## $8M\n\n## $6M\n\n## $4M\n\n## $2M\n\n## $0M Jan 23 Mar 23 May 23 Jul 23 Sep 23 Nov 23 Jan 24 Mar 24 May 24 Jul 24 Sep 24 Nov 24 Jan 25 Mar 25 Sep 24","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/WEF_Asset_Tokenization_in_Financial_Markets_2025_p22.pdf"]},"response":{"errors":[],"extract_id":"extract_c517b7d287f1ab4710d7af03e5f0ff30","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["# Contents\n\n## 22\n\n# Tokenized assets\n\n# 3\\.2 Asset issuance\n\n# Equities\n\n# Public equities are shares of publicly traded companies listed on stock exchanges and offered to a broad investor base. The global public equity market was valued at nearly\n\n## $115 trillion in 2023. ¹⁰ The market\n\n# capitalization of tokenized public stocks was estimated at nearly $16 million\n\n## by March 2025. ¹¹\n\n# Current public equities markets in AEs are already highly efficient and benefit from decades of technology modernization and proven intermediary chains delivering financial services. Due to existing efficiencies in AEs, tokenizing public equities may not be\n\n## prioritized over other asset classes. ¹²\n\n# However, opportunities exist, particularly in EMEs, to improve efficiencies and provide\n\n## access to equity markets. ¹³\n\n## Source: RWA.xyz (2025)\n\n## F I G U R E 7\n\n# Tokenized stocks\n\n## Backed-issued stocks bCSPX, bCOIN and bNVDA represent almost 90% of tokenized stocks in value.\n\n## bCSPX bCOIN bNVDA sMSTR TSLA.d Others\n\n## $18M\n\n## $16M\n\n## $14M\n\n## $12M\n\n## $10M\n\n## $8M\n\n## $6M\n\n## $4M\n\n## $2M\n\n## $0M Jan 23 Mar 23 May 23 Jul 23 Sep 23 Nov 23 Jan 24 Mar 24 May 24 Jul 24 Sep 24 Nov 24 Jan 25 Mar 25 Sep 24"],"full_content":"# Contents\n\n## 22\n\n# Tokenized assets\n\n# 3\\.2 Asset issuance\n\n# Equities\n\n# Public equities are shares of publicly traded companies listed on stock exchanges and offered to a broad investor base. The global public equity market was valued at nearly\n\n## $115 trillion in 2023. ¹⁰ The market\n\n# capitalization of tokenized public stocks was estimated at nearly $16 million\n\n## by March 2025. ¹¹\n\n# Current public equities markets in AEs are already highly efficient and benefit from decades of technology modernization and proven intermediary chains delivering financial services. Due to existing efficiencies in AEs, tokenizing public equities may not be\n\n## prioritized over other asset classes. ¹²\n\n# However, opportunities exist, particularly in EMEs, to improve efficiencies and provide\n\n## access to equity markets. ¹³\n\n## Source: RWA.xyz (2025)\n\n## F I G U R E 7\n\n# Tokenized stocks\n\n## Backed-issued stocks bCSPX, bCOIN and bNVDA represent almost 90% of tokenized stocks in value.\n\n## bCSPX bCOIN bNVDA sMSTR TSLA.d Others\n\n## $18M\n\n## $16M\n\n## $14M\n\n## $12M\n\n## $10M\n\n## $8M\n\n## $6M\n\n## $4M\n\n## $2M\n\n## $0M Jan 23 Mar 23 May 23 Jul 23 Sep 23 Nov 23 Jan 24 Mar 24 May 24 Jul 24 Sep 24 Nov 24 Jan 25 Mar 25 Sep 24","publish_date":null,"title":"3.2 Asset issuance","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/WEF_Asset_Tokenization_in_Financial_Markets_2025_p22.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/WEF_Asset_Tokenization_in_Financial_Markets_2025_p22.pdf"},"request":{"example_id":"chart/WEF_Asset_Tokenization_in_Financial_Markets_2025_p22","product_type":"parse","source_file_path":"WEF_Asset_Tokenization_in_Financial_Markets_2025_p22.pdf"},"started_at":"2026-11-04T19:44:58.536048"} +{"completed_at":"2026-11-04T19:45:00.605757","latency_in_ms":1105,"output":{"example_id":"chart/WEF_Asset_Tokenization_in_Financial_Markets_2025_p24","grounded_pages":[],"layout_pages":[],"markdown":"# Contents\n\n## 24\n\n# Tokenized assets\n\n## Source: Aldasoro, I., Cornelli, G. Frost, J., Koo Wilkens, P., & Shreeti, V. (2025). Tokenisation of government bonds, mimeo\n\n## Based on the subset of tokenized bonds with an available International Securities Identification Number (ISIN).\n\n# Sovereign bonds\n\n# Government or sovereign bonds are debt securities issued by national governments, often considered low-risk investments and used as benchmarks for interest rates. They are critical for funding government operations and old debt or interest. They also play an integral role in institutional portfolios, providing predictable returns and aiding in\n\n## retirement planning. ²⁴\n\n# Government authorities are gradually endorsing tokenized public debt, evidenced by His Majesty’s Treasury’s plans for a digital gilts pilot in the UK Finance’s comprehensive\n\n## roadmap, ²⁵ the US Commodity Futures\n\n# Trading Commission acceptance of tokenized\n\n## treasuries as non-cash collateral ²⁶ and the US\n\n# Treasury’s acknowledgment of potential\n\n## operational benefits. ²⁷ In Hong Kong, the\n\n# government has committed to the issuance of tokenized bonds as standard practice. The Hong Kong Monetary Authority (HKMA) is preparing to issue the third tranche of tokenized bonds, while actively exploring\n\n## tokenizing existing bonds. ²⁸ Collectively, there\n\n# is a progressive shift towards adopting tokenization for sovereign securities.\n\n## Cumulative amount issued (left-hand side):\n\n## Commercial\n\n## Sovereign, supranational and agency\n\n## Cumulative issuances (right-hand side):\n\n## $ billion\n\n## No of bonds\n\n## 2019\n\n## 0\n\n## 1\n\n## 2\n\n## 3\n\n## 4\n\n## 0\n\n## 5\n\n## 10\n\n## 15\n\n## 20\n\n## 2020 2021 2022 2023 2024\n\n## F I G U R E 8\n\n# A growing number of issuers are experimenting with tokenized bonds\n\n## Commercial vs. sovereign, supranational and agency tokenized bonds\n\n# Fixed income\n\n# Fixed-income securities are debt instruments issued by a government, corporation or other entity to finance their operations. They provide investors with a return in the form of fixed periodic payments and the eventual return of\n\n## the principal at maturity. ²⁰ The global fixed-\n\n# income, or bond, market was valued at\n\n## approximately $140.7 trillion in 2023. ²¹ Fixed-\n\n# income tokenization has seen notable advances in sovereign, corporate and municipal bonds.\n\n# According to an Official Monetary and Financial Institutions Forum (OMFIF) survey of 26 financial institutions, 65% believed that\n\n## bonds were the most likely to be tokenized. ²²\n\n# For example, the European Investment Bank (EIB) has issued several bonds on-chain, proving the feasibility of tokenization. Since 2021, with its inaugural digital bond issued on Ethereum, the EIB has issued five bonds, most recently using HSBC’s Orion platform for issuance and the Banque de France’s DL3S\n\n## platform for settlement. ²³","pages":[{"line_numbers":[],"markdown":"# Contents\n\n## 24\n\n# Tokenized assets\n\n## Source: Aldasoro, I., Cornelli, G. Frost, J., Koo Wilkens, P., & Shreeti, V. (2025). Tokenisation of government bonds, mimeo\n\n## Based on the subset of tokenized bonds with an available International Securities Identification Number (ISIN).\n\n# Sovereign bonds\n\n# Government or sovereign bonds are debt securities issued by national governments, often considered low-risk investments and used as benchmarks for interest rates. They are critical for funding government operations and old debt or interest. They also play an integral role in institutional portfolios, providing predictable returns and aiding in\n\n## retirement planning. ²⁴\n\n# Government authorities are gradually endorsing tokenized public debt, evidenced by His Majesty’s Treasury’s plans for a digital gilts pilot in the UK Finance’s comprehensive\n\n## roadmap, ²⁵ the US Commodity Futures\n\n# Trading Commission acceptance of tokenized\n\n## treasuries as non-cash collateral ²⁶ and the US\n\n# Treasury’s acknowledgment of potential\n\n## operational benefits. ²⁷ In Hong Kong, the\n\n# government has committed to the issuance of tokenized bonds as standard practice. The Hong Kong Monetary Authority (HKMA) is preparing to issue the third tranche of tokenized bonds, while actively exploring\n\n## tokenizing existing bonds. ²⁸ Collectively, there\n\n# is a progressive shift towards adopting tokenization for sovereign securities.\n\n## Cumulative amount issued (left-hand side):\n\n## Commercial\n\n## Sovereign, supranational and agency\n\n## Cumulative issuances (right-hand side):\n\n## $ billion\n\n## No of bonds\n\n## 2019\n\n## 0\n\n## 1\n\n## 2\n\n## 3\n\n## 4\n\n## 0\n\n## 5\n\n## 10\n\n## 15\n\n## 20\n\n## 2020 2021 2022 2023 2024\n\n## F I G U R E 8\n\n# A growing number of issuers are experimenting with tokenized bonds\n\n## Commercial vs. sovereign, supranational and agency tokenized bonds\n\n# Fixed income\n\n# Fixed-income securities are debt instruments issued by a government, corporation or other entity to finance their operations. They provide investors with a return in the form of fixed periodic payments and the eventual return of\n\n## the principal at maturity. ²⁰ The global fixed-\n\n# income, or bond, market was valued at\n\n## approximately $140.7 trillion in 2023. ²¹ Fixed-\n\n# income tokenization has seen notable advances in sovereign, corporate and municipal bonds.\n\n# According to an Official Monetary and Financial Institutions Forum (OMFIF) survey of 26 financial institutions, 65% believed that\n\n## bonds were the most likely to be tokenized. ²²\n\n# For example, the European Investment Bank (EIB) has issued several bonds on-chain, proving the feasibility of tokenization. Since 2021, with its inaugural digital bond issued on Ethereum, the EIB has issued five bonds, most recently using HSBC’s Orion platform for issuance and the Banque de France’s DL3S\n\n## platform for settlement. ²³","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"# Contents\n\n## 24\n\n# Tokenized assets\n\n## Source: Aldasoro, I., Cornelli, G. Frost, J., Koo Wilkens, P., & Shreeti, V. (2025). Tokenisation of government bonds, mimeo\n\n## Based on the subset of tokenized bonds with an available International Securities Identification Number (ISIN).\n\n# Sovereign bonds\n\n# Government or sovereign bonds are debt securities issued by national governments, often considered low-risk investments and used as benchmarks for interest rates. They are critical for funding government operations and old debt or interest. They also play an integral role in institutional portfolios, providing predictable returns and aiding in\n\n## retirement planning. ²⁴\n\n# Government authorities are gradually endorsing tokenized public debt, evidenced by His Majesty’s Treasury’s plans for a digital gilts pilot in the UK Finance’s comprehensive\n\n## roadmap, ²⁵ the US Commodity Futures\n\n# Trading Commission acceptance of tokenized\n\n## treasuries as non-cash collateral ²⁶ and the US\n\n# Treasury’s acknowledgment of potential\n\n## operational benefits. ²⁷ In Hong Kong, the\n\n# government has committed to the issuance of tokenized bonds as standard practice. The Hong Kong Monetary Authority (HKMA) is preparing to issue the third tranche of tokenized bonds, while actively exploring\n\n## tokenizing existing bonds. ²⁸ Collectively, there\n\n# is a progressive shift towards adopting tokenization for sovereign securities.\n\n## Cumulative amount issued (left-hand side):\n\n## Commercial\n\n## Sovereign, supranational and agency\n\n## Cumulative issuances (right-hand side):\n\n## $ billion\n\n## No of bonds\n\n## 2019\n\n## 0\n\n## 1\n\n## 2\n\n## 3\n\n## 4\n\n## 0\n\n## 5\n\n## 10\n\n## 15\n\n## 20\n\n## 2020 2021 2022 2023 2024\n\n## F I G U R E 8\n\n# A growing number of issuers are experimenting with tokenized bonds\n\n## Commercial vs. sovereign, supranational and agency tokenized bonds\n\n# Fixed income\n\n# Fixed-income securities are debt instruments issued by a government, corporation or other entity to finance their operations. They provide investors with a return in the form of fixed periodic payments and the eventual return of\n\n## the principal at maturity. ²⁰ The global fixed-\n\n# income, or bond, market was valued at\n\n## approximately $140.7 trillion in 2023. ²¹ Fixed-\n\n# income tokenization has seen notable advances in sovereign, corporate and municipal bonds.\n\n# According to an Official Monetary and Financial Institutions Forum (OMFIF) survey of 26 financial institutions, 65% believed that\n\n## bonds were the most likely to be tokenized. ²²\n\n# For example, the European Investment Bank (EIB) has issued several bonds on-chain, proving the feasibility of tokenization. Since 2021, with its inaugural digital bond issued on Ethereum, the EIB has issued five bonds, most recently using HSBC’s Orion platform for issuance and the Banque de France’s DL3S\n\n## platform for settlement. ²³","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/WEF_Asset_Tokenization_in_Financial_Markets_2025_p24.pdf"]},"response":{"errors":[],"extract_id":"extract_4f7947989484cc99d1fd806537d3e917","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["# Contents\n\n## 24\n\n# Tokenized assets\n\n## Source: Aldasoro, I., Cornelli, G. Frost, J., Koo Wilkens, P., & Shreeti, V. (2025). Tokenisation of government bonds, mimeo\n\n## Based on the subset of tokenized bonds with an available International Securities Identification Number (ISIN).\n\n# Sovereign bonds\n\n# Government or sovereign bonds are debt securities issued by national governments, often considered low-risk investments and used as benchmarks for interest rates. They are critical for funding government operations and old debt or interest. They also play an integral role in institutional portfolios, providing predictable returns and aiding in\n\n## retirement planning. ²⁴\n\n# Government authorities are gradually endorsing tokenized public debt, evidenced by His Majesty’s Treasury’s plans for a digital gilts pilot in the UK Finance’s comprehensive\n\n## roadmap, ²⁵ the US Commodity Futures\n\n# Trading Commission acceptance of tokenized\n\n## treasuries as non-cash collateral ²⁶ and the US\n\n# Treasury’s acknowledgment of potential\n\n## operational benefits. ²⁷ In Hong Kong, the\n\n# government has committed to the issuance of tokenized bonds as standard practice. The Hong Kong Monetary Authority (HKMA) is preparing to issue the third tranche of tokenized bonds, while actively exploring\n\n## tokenizing existing bonds. ²⁸ Collectively, there\n\n# is a progressive shift towards adopting tokenization for sovereign securities.\n\n## Cumulative amount issued (left-hand side):\n\n## Commercial\n\n## Sovereign, supranational and agency\n\n## Cumulative issuances (right-hand side):\n\n## $ billion\n\n## No of bonds\n\n## 2019\n\n## 0\n\n## 1\n\n## 2\n\n## 3\n\n## 4\n\n## 0\n\n## 5\n\n## 10\n\n## 15\n\n## 20\n\n## 2020 2021 2022 2023 2024\n\n## F I G U R E 8\n\n# A growing number of issuers are experimenting with tokenized bonds\n\n## Commercial vs. sovereign, supranational and agency tokenized bonds\n\n# Fixed income\n\n# Fixed-income securities are debt instruments issued by a government, corporation or other entity to finance their operations. They provide investors with a return in the form of fixed periodic payments and the eventual return of\n\n## the principal at maturity. ²⁰ The global fixed-\n\n# income, or bond, market was valued at\n\n## approximately $140.7 trillion in 2023. ²¹ Fixed-\n\n# income tokenization has seen notable advances in sovereign, corporate and municipal bonds.\n\n# According to an Official Monetary and Financial Institutions Forum (OMFIF) survey of 26 financial institutions, 65% believed that\n\n## bonds were the most likely to be tokenized. ²²\n\n# For example, the European Investment Bank (EIB) has issued several bonds on-chain, proving the feasibility of tokenization. Since 2021, with its inaugural digital bond issued on Ethereum, the EIB has issued five bonds, most recently using HSBC’s Orion platform for issuance and the Banque de France’s DL3S\n\n## platform for settlement. ²³"],"full_content":"# Contents\n\n## 24\n\n# Tokenized assets\n\n## Source: Aldasoro, I., Cornelli, G. Frost, J., Koo Wilkens, P., & Shreeti, V. (2025). Tokenisation of government bonds, mimeo\n\n## Based on the subset of tokenized bonds with an available International Securities Identification Number (ISIN).\n\n# Sovereign bonds\n\n# Government or sovereign bonds are debt securities issued by national governments, often considered low-risk investments and used as benchmarks for interest rates. They are critical for funding government operations and old debt or interest. They also play an integral role in institutional portfolios, providing predictable returns and aiding in\n\n## retirement planning. ²⁴\n\n# Government authorities are gradually endorsing tokenized public debt, evidenced by His Majesty’s Treasury’s plans for a digital gilts pilot in the UK Finance’s comprehensive\n\n## roadmap, ²⁵ the US Commodity Futures\n\n# Trading Commission acceptance of tokenized\n\n## treasuries as non-cash collateral ²⁶ and the US\n\n# Treasury’s acknowledgment of potential\n\n## operational benefits. ²⁷ In Hong Kong, the\n\n# government has committed to the issuance of tokenized bonds as standard practice. The Hong Kong Monetary Authority (HKMA) is preparing to issue the third tranche of tokenized bonds, while actively exploring\n\n## tokenizing existing bonds. ²⁸ Collectively, there\n\n# is a progressive shift towards adopting tokenization for sovereign securities.\n\n## Cumulative amount issued (left-hand side):\n\n## Commercial\n\n## Sovereign, supranational and agency\n\n## Cumulative issuances (right-hand side):\n\n## $ billion\n\n## No of bonds\n\n## 2019\n\n## 0\n\n## 1\n\n## 2\n\n## 3\n\n## 4\n\n## 0\n\n## 5\n\n## 10\n\n## 15\n\n## 20\n\n## 2020 2021 2022 2023 2024\n\n## F I G U R E 8\n\n# A growing number of issuers are experimenting with tokenized bonds\n\n## Commercial vs. sovereign, supranational and agency tokenized bonds\n\n# Fixed income\n\n# Fixed-income securities are debt instruments issued by a government, corporation or other entity to finance their operations. They provide investors with a return in the form of fixed periodic payments and the eventual return of\n\n## the principal at maturity. ²⁰ The global fixed-\n\n# income, or bond, market was valued at\n\n## approximately $140.7 trillion in 2023. ²¹ Fixed-\n\n# income tokenization has seen notable advances in sovereign, corporate and municipal bonds.\n\n# According to an Official Monetary and Financial Institutions Forum (OMFIF) survey of 26 financial institutions, 65% believed that\n\n## bonds were the most likely to be tokenized. ²²\n\n# For example, the European Investment Bank (EIB) has issued several bonds on-chain, proving the feasibility of tokenization. Since 2021, with its inaugural digital bond issued on Ethereum, the EIB has issued five bonds, most recently using HSBC’s Orion platform for issuance and the Banque de France’s DL3S\n\n## platform for settlement. ²³","publish_date":null,"title":"Fixed income","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/WEF_Asset_Tokenization_in_Financial_Markets_2025_p24.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/WEF_Asset_Tokenization_in_Financial_Markets_2025_p24.pdf"},"request":{"example_id":"chart/WEF_Asset_Tokenization_in_Financial_Markets_2025_p24","product_type":"parse","source_file_path":"WEF_Asset_Tokenization_in_Financial_Markets_2025_p24.pdf"},"started_at":"2026-11-04T19:44:59.500277"} +{"completed_at":"2026-11-04T19:45:02.173712","latency_in_ms":1285,"output":{"example_id":"chart/WEF_Asset_Tokenization_in_Financial_Markets_2025_p27","grounded_pages":[],"layout_pages":[],"markdown":"# Contents\n\n## 27\n\n# Private debt\n\n# The terms “private debt” and “private credit” are used interchangeably and refer to non- bank lending or bonds outside traditional\n\n## markets used to fund private businesses. ⁴⁴\n\n# The global private credit market has surpassed $3 trillion in assets under management\n\n## (AUM). ⁴⁵ This asset class consists of higher-\n\n# risk loans with higher-interest rates than conventional debt. Unlike public debt, private debt is issued by non-bank entities or individuals to raise funds or form capital.\n\n# Private debt can attract investors, offering higher returns and risk. Private-sector leaders believe private debt will be the first asset class\n\n## to be tokenized and routinely traded. ⁴⁶ About\n\n# 70% of private-debt investors are institutional investors or entities investing on behalf of an\n\n## individual, typically a private credit fund. ⁴⁷\n\n# Most private loans on-chain are direct loans to institutional crypto investors. These loans mostly appeal to high-net-worth individuals and accredited investors as yields range from 9% to 20%. However, they carry risks in that\n\n## the collateral posted is limited in diversity. ⁴⁸\n\n# Because crypto-assets such as Ethereum back most loans, significant market corrections of these assets could affect the lending positions of the holders.\n\n# Examples of the benefits of tokenization for private debt include:\n\n# 1\\. Enhances information symmetry Transaction flow would be in real time, providing data on underlying assets\n\n## at all times. ⁴⁹\n\n# 2\\. Facilitates operational efficiency Smart contracts reduce and mitigate back- office costs and lower transaction costs and management fees.\n\n# 3\\. Expands accessibility Easing the process of buying into private credit funds and of trading shares can increase the range of investors. For example, in 2022, KKR collaborated with Securitize to tokenize a segment of its $4 billion healthcare fund on Avalanche, reducing investment minimums and onboarding accredited investors,\n\n## increasing reach. ⁵⁰\n\n# On-chain activity for private debt/credit is increasing but is largely limited to crypto- assets and select corporate bond issuances. Protocols such as Maple, Goldfinch and Centrifuge have driven more than $13 billion in on-chain loans in public networks, including\n\n## Ethereum and Solana. ⁵¹\n\n# Tokenized assets\n\n## F I G U R E 1 0\n\n# Tokenized private credit by issuer\n\n## Figure represents almost 80% of the value of outstanding tokenized private loans\n\n## Source: RWA.xyz (2025)\n\n## Figure Tradable Maple PACT MB | Mercado Bitcoin Others\n\n## $14M\n\n## $12M\n\n## $10M\n\n## $8M\n\n## $6M\n\n## $4M\n\n## $2M\n\n## $0M\n\n## Jan 23 Mar 23 May 23\n\n## Jul 23 Sep 23 Nov 23 Jan 24 Mar 24 May 24 Jul 24 Sep 24 Nov 24 Jan 25 Mar 25","pages":[{"line_numbers":[],"markdown":"# Contents\n\n## 27\n\n# Private debt\n\n# The terms “private debt” and “private credit” are used interchangeably and refer to non- bank lending or bonds outside traditional\n\n## markets used to fund private businesses. ⁴⁴\n\n# The global private credit market has surpassed $3 trillion in assets under management\n\n## (AUM). ⁴⁵ This asset class consists of higher-\n\n# risk loans with higher-interest rates than conventional debt. Unlike public debt, private debt is issued by non-bank entities or individuals to raise funds or form capital.\n\n# Private debt can attract investors, offering higher returns and risk. Private-sector leaders believe private debt will be the first asset class\n\n## to be tokenized and routinely traded. ⁴⁶ About\n\n# 70% of private-debt investors are institutional investors or entities investing on behalf of an\n\n## individual, typically a private credit fund. ⁴⁷\n\n# Most private loans on-chain are direct loans to institutional crypto investors. These loans mostly appeal to high-net-worth individuals and accredited investors as yields range from 9% to 20%. However, they carry risks in that\n\n## the collateral posted is limited in diversity. ⁴⁸\n\n# Because crypto-assets such as Ethereum back most loans, significant market corrections of these assets could affect the lending positions of the holders.\n\n# Examples of the benefits of tokenization for private debt include:\n\n# 1\\. Enhances information symmetry Transaction flow would be in real time, providing data on underlying assets\n\n## at all times. ⁴⁹\n\n# 2\\. Facilitates operational efficiency Smart contracts reduce and mitigate back- office costs and lower transaction costs and management fees.\n\n# 3\\. Expands accessibility Easing the process of buying into private credit funds and of trading shares can increase the range of investors. For example, in 2022, KKR collaborated with Securitize to tokenize a segment of its $4 billion healthcare fund on Avalanche, reducing investment minimums and onboarding accredited investors,\n\n## increasing reach. ⁵⁰\n\n# On-chain activity for private debt/credit is increasing but is largely limited to crypto- assets and select corporate bond issuances. Protocols such as Maple, Goldfinch and Centrifuge have driven more than $13 billion in on-chain loans in public networks, including\n\n## Ethereum and Solana. ⁵¹\n\n# Tokenized assets\n\n## F I G U R E 1 0\n\n# Tokenized private credit by issuer\n\n## Figure represents almost 80% of the value of outstanding tokenized private loans\n\n## Source: RWA.xyz (2025)\n\n## Figure Tradable Maple PACT MB | Mercado Bitcoin Others\n\n## $14M\n\n## $12M\n\n## $10M\n\n## $8M\n\n## $6M\n\n## $4M\n\n## $2M\n\n## $0M\n\n## Jan 23 Mar 23 May 23\n\n## Jul 23 Sep 23 Nov 23 Jan 24 Mar 24 May 24 Jul 24 Sep 24 Nov 24 Jan 25 Mar 25","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"# Contents\n\n## 27\n\n# Private debt\n\n# The terms “private debt” and “private credit” are used interchangeably and refer to non- bank lending or bonds outside traditional\n\n## markets used to fund private businesses. ⁴⁴\n\n# The global private credit market has surpassed $3 trillion in assets under management\n\n## (AUM). ⁴⁵ This asset class consists of higher-\n\n# risk loans with higher-interest rates than conventional debt. Unlike public debt, private debt is issued by non-bank entities or individuals to raise funds or form capital.\n\n# Private debt can attract investors, offering higher returns and risk. Private-sector leaders believe private debt will be the first asset class\n\n## to be tokenized and routinely traded. ⁴⁶ About\n\n# 70% of private-debt investors are institutional investors or entities investing on behalf of an\n\n## individual, typically a private credit fund. ⁴⁷\n\n# Most private loans on-chain are direct loans to institutional crypto investors. These loans mostly appeal to high-net-worth individuals and accredited investors as yields range from 9% to 20%. However, they carry risks in that\n\n## the collateral posted is limited in diversity. ⁴⁸\n\n# Because crypto-assets such as Ethereum back most loans, significant market corrections of these assets could affect the lending positions of the holders.\n\n# Examples of the benefits of tokenization for private debt include:\n\n# 1\\. Enhances information symmetry Transaction flow would be in real time, providing data on underlying assets\n\n## at all times. ⁴⁹\n\n# 2\\. Facilitates operational efficiency Smart contracts reduce and mitigate back- office costs and lower transaction costs and management fees.\n\n# 3\\. Expands accessibility Easing the process of buying into private credit funds and of trading shares can increase the range of investors. For example, in 2022, KKR collaborated with Securitize to tokenize a segment of its $4 billion healthcare fund on Avalanche, reducing investment minimums and onboarding accredited investors,\n\n## increasing reach. ⁵⁰\n\n# On-chain activity for private debt/credit is increasing but is largely limited to crypto- assets and select corporate bond issuances. Protocols such as Maple, Goldfinch and Centrifuge have driven more than $13 billion in on-chain loans in public networks, including\n\n## Ethereum and Solana. ⁵¹\n\n# Tokenized assets\n\n## F I G U R E 1 0\n\n# Tokenized private credit by issuer\n\n## Figure represents almost 80% of the value of outstanding tokenized private loans\n\n## Source: RWA.xyz (2025)\n\n## Figure Tradable Maple PACT MB | Mercado Bitcoin Others\n\n## $14M\n\n## $12M\n\n## $10M\n\n## $8M\n\n## $6M\n\n## $4M\n\n## $2M\n\n## $0M\n\n## Jan 23 Mar 23 May 23\n\n## Jul 23 Sep 23 Nov 23 Jan 24 Mar 24 May 24 Jul 24 Sep 24 Nov 24 Jan 25 Mar 25","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/WEF_Asset_Tokenization_in_Financial_Markets_2025_p27.pdf"]},"response":{"errors":[],"extract_id":"extract_a17acc94a8868fdcb37880d620a96e78","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["# Contents\n\n## 27\n\n# Private debt\n\n# The terms “private debt” and “private credit” are used interchangeably and refer to non- bank lending or bonds outside traditional\n\n## markets used to fund private businesses. ⁴⁴\n\n# The global private credit market has surpassed $3 trillion in assets under management\n\n## (AUM). ⁴⁵ This asset class consists of higher-\n\n# risk loans with higher-interest rates than conventional debt. Unlike public debt, private debt is issued by non-bank entities or individuals to raise funds or form capital.\n\n# Private debt can attract investors, offering higher returns and risk. Private-sector leaders believe private debt will be the first asset class\n\n## to be tokenized and routinely traded. ⁴⁶ About\n\n# 70% of private-debt investors are institutional investors or entities investing on behalf of an\n\n## individual, typically a private credit fund. ⁴⁷\n\n# Most private loans on-chain are direct loans to institutional crypto investors. These loans mostly appeal to high-net-worth individuals and accredited investors as yields range from 9% to 20%. However, they carry risks in that\n\n## the collateral posted is limited in diversity. ⁴⁸\n\n# Because crypto-assets such as Ethereum back most loans, significant market corrections of these assets could affect the lending positions of the holders.\n\n# Examples of the benefits of tokenization for private debt include:\n\n# 1\\. Enhances information symmetry Transaction flow would be in real time, providing data on underlying assets\n\n## at all times. ⁴⁹\n\n# 2\\. Facilitates operational efficiency Smart contracts reduce and mitigate back- office costs and lower transaction costs and management fees.\n\n# 3\\. Expands accessibility Easing the process of buying into private credit funds and of trading shares can increase the range of investors. For example, in 2022, KKR collaborated with Securitize to tokenize a segment of its $4 billion healthcare fund on Avalanche, reducing investment minimums and onboarding accredited investors,\n\n## increasing reach. ⁵⁰\n\n# On-chain activity for private debt/credit is increasing but is largely limited to crypto- assets and select corporate bond issuances. Protocols such as Maple, Goldfinch and Centrifuge have driven more than $13 billion in on-chain loans in public networks, including\n\n## Ethereum and Solana. ⁵¹\n\n# Tokenized assets\n\n## F I G U R E 1 0\n\n# Tokenized private credit by issuer\n\n## Figure represents almost 80% of the value of outstanding tokenized private loans\n\n## Source: RWA.xyz (2025)\n\n## Figure Tradable Maple PACT MB | Mercado Bitcoin Others\n\n## $14M\n\n## $12M\n\n## $10M\n\n## $8M\n\n## $6M\n\n## $4M\n\n## $2M\n\n## $0M\n\n## Jan 23 Mar 23 May 23\n\n## Jul 23 Sep 23 Nov 23 Jan 24 Mar 24 May 24 Jul 24 Sep 24 Nov 24 Jan 25 Mar 25"],"full_content":"# Contents\n\n## 27\n\n# Private debt\n\n# The terms “private debt” and “private credit” are used interchangeably and refer to non- bank lending or bonds outside traditional\n\n## markets used to fund private businesses. ⁴⁴\n\n# The global private credit market has surpassed $3 trillion in assets under management\n\n## (AUM). ⁴⁵ This asset class consists of higher-\n\n# risk loans with higher-interest rates than conventional debt. Unlike public debt, private debt is issued by non-bank entities or individuals to raise funds or form capital.\n\n# Private debt can attract investors, offering higher returns and risk. Private-sector leaders believe private debt will be the first asset class\n\n## to be tokenized and routinely traded. ⁴⁶ About\n\n# 70% of private-debt investors are institutional investors or entities investing on behalf of an\n\n## individual, typically a private credit fund. ⁴⁷\n\n# Most private loans on-chain are direct loans to institutional crypto investors. These loans mostly appeal to high-net-worth individuals and accredited investors as yields range from 9% to 20%. However, they carry risks in that\n\n## the collateral posted is limited in diversity. ⁴⁸\n\n# Because crypto-assets such as Ethereum back most loans, significant market corrections of these assets could affect the lending positions of the holders.\n\n# Examples of the benefits of tokenization for private debt include:\n\n# 1\\. Enhances information symmetry Transaction flow would be in real time, providing data on underlying assets\n\n## at all times. ⁴⁹\n\n# 2\\. Facilitates operational efficiency Smart contracts reduce and mitigate back- office costs and lower transaction costs and management fees.\n\n# 3\\. Expands accessibility Easing the process of buying into private credit funds and of trading shares can increase the range of investors. For example, in 2022, KKR collaborated with Securitize to tokenize a segment of its $4 billion healthcare fund on Avalanche, reducing investment minimums and onboarding accredited investors,\n\n## increasing reach. ⁵⁰\n\n# On-chain activity for private debt/credit is increasing but is largely limited to crypto- assets and select corporate bond issuances. Protocols such as Maple, Goldfinch and Centrifuge have driven more than $13 billion in on-chain loans in public networks, including\n\n## Ethereum and Solana. ⁵¹\n\n# Tokenized assets\n\n## F I G U R E 1 0\n\n# Tokenized private credit by issuer\n\n## Figure represents almost 80% of the value of outstanding tokenized private loans\n\n## Source: RWA.xyz (2025)\n\n## Figure Tradable Maple PACT MB | Mercado Bitcoin Others\n\n## $14M\n\n## $12M\n\n## $10M\n\n## $8M\n\n## $6M\n\n## $4M\n\n## $2M\n\n## $0M\n\n## Jan 23 Mar 23 May 23\n\n## Jul 23 Sep 23 Nov 23 Jan 24 Mar 24 May 24 Jul 24 Sep 24 Nov 24 Jan 25 Mar 25","publish_date":null,"title":"Tokenized private credit by issuer","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/WEF_Asset_Tokenization_in_Financial_Markets_2025_p27.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/WEF_Asset_Tokenization_in_Financial_Markets_2025_p27.pdf"},"request":{"example_id":"chart/WEF_Asset_Tokenization_in_Financial_Markets_2025_p27","product_type":"parse","source_file_path":"WEF_Asset_Tokenization_in_Financial_Markets_2025_p27.pdf"},"started_at":"2026-11-04T19:45:00.888080"} +{"completed_at":"2026-11-04T19:45:03.931119","latency_in_ms":1755,"output":{"example_id":"chart/WEF_Navigating_Global_Financial_System_Fragmentation_2025_p12","grounded_pages":[],"layout_pages":[],"markdown":"F I G U R E 4 Share of global exchange reserves, global export invoicing and foreign transaction by currency\n\nWhile the US dollar will likely continue to play a significant role as a reserve and funding currency, it is becoming less dominant as a medium of exchange for financing goods and services trade. Amid a trend towards greater trade regionalization, the renminbi and euro are expected to become more important within Asian and European trade. ¹³ Data from October 2024 shows that the renminbi is the second largest global currency in the trade finance market, with a share of 5.8%, just ahead of the euro but behind the US dollar, which has a share of around 83%. ¹⁴ Digital currencies, including wholesale central\n\nbank digital currencies (CBDCs), could provide alternative cross-border payment rails, creating a more diverse global financial landscape.\n\nMoreover, rising debt levels in the United States raise questions about debt sustainability, while growing domestic political polarization could undermine the independence of US institutions. Over time, these developments could reduce confidence in the dollar and lead to a more multipolar currency system.\n\nEven as financial integration continued to deepen in the 2000s, challenges to the system began to emerge. A crucial turning point came after the terrorist attacks on 11 September 2001, which shifted some governments’ view of the financial system not just as economic infrastructure, but as a powerful tool of economic statecraft.\n\nEconomic statecraft broadly refers to the use of economic tools and policies by a state to achieve its foreign-policy and domestic objectives. Such measures can include punitive steps such as sanctions, designed to pressure norms-violating countries and to protect the issuing government’s interests, as well as inducements, such as trade agreements aimed at improving relationships and fostering cooperation.\n\n## 1\\.3 Use of the financial system for economic statecraft\n\nGlobal exchange reserves Share of export invoicing Share of daily foreign transactions\n\nBillions of US dollars %\n\nGBP EUR RMB JPY USD\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\n70\n\n80\n\n90\n\n100\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\n70\n\n80\n\n90\n\n100\n\n58%\n\n20%\n\n6%\n\n2%\n\n48%\n\n$88\n\n$31\n\n$13\n\n$17\n\n$7\n\n22%\n\n7% 4% 4% 5%\n\nNote: To calculate share of export invoicing, project team used the Bloomberg Intelligence FX Strategy Dashboard and Global SWIFT data as a proxy using inbound and outbound traffic processed via SWIFT. To calculate share of foreign transactions, project team used the turnover of OTC foreign exchange instruments by currency, “net-net” basis, 2022 daily averages in billions of USD.\n\nSource: Share of global foreign exchange reserves: IMF’s Currency Composition of Official Foreign Exchange Reserves Q2 2024. Share of export invoicing: Bloomberg Intelligence FX Strategy Dashboard. Share of foreign transactions: BIS Triennial Survey, Data Portal. Oliver Wyman analysis\n\nEconomic statecraft broadly refers to the use of economic tools and policies by a state to achieve its foreign- policy and domestic objectives.\n\nNavigating Global Financial System Fragmentation 12","pages":[{"line_numbers":[],"markdown":"F I G U R E 4 Share of global exchange reserves, global export invoicing and foreign transaction by currency\n\nWhile the US dollar will likely continue to play a significant role as a reserve and funding currency, it is becoming less dominant as a medium of exchange for financing goods and services trade. Amid a trend towards greater trade regionalization, the renminbi and euro are expected to become more important within Asian and European trade. ¹³ Data from October 2024 shows that the renminbi is the second largest global currency in the trade finance market, with a share of 5.8%, just ahead of the euro but behind the US dollar, which has a share of around 83%. ¹⁴ Digital currencies, including wholesale central\n\nbank digital currencies (CBDCs), could provide alternative cross-border payment rails, creating a more diverse global financial landscape.\n\nMoreover, rising debt levels in the United States raise questions about debt sustainability, while growing domestic political polarization could undermine the independence of US institutions. Over time, these developments could reduce confidence in the dollar and lead to a more multipolar currency system.\n\nEven as financial integration continued to deepen in the 2000s, challenges to the system began to emerge. A crucial turning point came after the terrorist attacks on 11 September 2001, which shifted some governments’ view of the financial system not just as economic infrastructure, but as a powerful tool of economic statecraft.\n\nEconomic statecraft broadly refers to the use of economic tools and policies by a state to achieve its foreign-policy and domestic objectives. Such measures can include punitive steps such as sanctions, designed to pressure norms-violating countries and to protect the issuing government’s interests, as well as inducements, such as trade agreements aimed at improving relationships and fostering cooperation.\n\n## 1\\.3 Use of the financial system for economic statecraft\n\nGlobal exchange reserves Share of export invoicing Share of daily foreign transactions\n\nBillions of US dollars %\n\nGBP EUR RMB JPY USD\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\n70\n\n80\n\n90\n\n100\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\n70\n\n80\n\n90\n\n100\n\n58%\n\n20%\n\n6%\n\n2%\n\n48%\n\n$88\n\n$31\n\n$13\n\n$17\n\n$7\n\n22%\n\n7% 4% 4% 5%\n\nNote: To calculate share of export invoicing, project team used the Bloomberg Intelligence FX Strategy Dashboard and Global SWIFT data as a proxy using inbound and outbound traffic processed via SWIFT. To calculate share of foreign transactions, project team used the turnover of OTC foreign exchange instruments by currency, “net-net” basis, 2022 daily averages in billions of USD.\n\nSource: Share of global foreign exchange reserves: IMF’s Currency Composition of Official Foreign Exchange Reserves Q2 2024. Share of export invoicing: Bloomberg Intelligence FX Strategy Dashboard. Share of foreign transactions: BIS Triennial Survey, Data Portal. Oliver Wyman analysis\n\nEconomic statecraft broadly refers to the use of economic tools and policies by a state to achieve its foreign- policy and domestic objectives.\n\nNavigating Global Financial System Fragmentation 12","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"F I G U R E 4 Share of global exchange reserves, global export invoicing and foreign transaction by currency\n\nWhile the US dollar will likely continue to play a significant role as a reserve and funding currency, it is becoming less dominant as a medium of exchange for financing goods and services trade. Amid a trend towards greater trade regionalization, the renminbi and euro are expected to become more important within Asian and European trade. ¹³ Data from October 2024 shows that the renminbi is the second largest global currency in the trade finance market, with a share of 5.8%, just ahead of the euro but behind the US dollar, which has a share of around 83%. ¹⁴ Digital currencies, including wholesale central\n\nbank digital currencies (CBDCs), could provide alternative cross-border payment rails, creating a more diverse global financial landscape.\n\nMoreover, rising debt levels in the United States raise questions about debt sustainability, while growing domestic political polarization could undermine the independence of US institutions. Over time, these developments could reduce confidence in the dollar and lead to a more multipolar currency system.\n\nEven as financial integration continued to deepen in the 2000s, challenges to the system began to emerge. A crucial turning point came after the terrorist attacks on 11 September 2001, which shifted some governments’ view of the financial system not just as economic infrastructure, but as a powerful tool of economic statecraft.\n\nEconomic statecraft broadly refers to the use of economic tools and policies by a state to achieve its foreign-policy and domestic objectives. Such measures can include punitive steps such as sanctions, designed to pressure norms-violating countries and to protect the issuing government’s interests, as well as inducements, such as trade agreements aimed at improving relationships and fostering cooperation.\n\n## 1\\.3 Use of the financial system for economic statecraft\n\nGlobal exchange reserves Share of export invoicing Share of daily foreign transactions\n\nBillions of US dollars %\n\nGBP EUR RMB JPY USD\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\n70\n\n80\n\n90\n\n100\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\n70\n\n80\n\n90\n\n100\n\n58%\n\n20%\n\n6%\n\n2%\n\n48%\n\n$88\n\n$31\n\n$13\n\n$17\n\n$7\n\n22%\n\n7% 4% 4% 5%\n\nNote: To calculate share of export invoicing, project team used the Bloomberg Intelligence FX Strategy Dashboard and Global SWIFT data as a proxy using inbound and outbound traffic processed via SWIFT. To calculate share of foreign transactions, project team used the turnover of OTC foreign exchange instruments by currency, “net-net” basis, 2022 daily averages in billions of USD.\n\nSource: Share of global foreign exchange reserves: IMF’s Currency Composition of Official Foreign Exchange Reserves Q2 2024. Share of export invoicing: Bloomberg Intelligence FX Strategy Dashboard. Share of foreign transactions: BIS Triennial Survey, Data Portal. Oliver Wyman analysis\n\nEconomic statecraft broadly refers to the use of economic tools and policies by a state to achieve its foreign- policy and domestic objectives.\n\nNavigating Global Financial System Fragmentation 12","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/WEF_Navigating_Global_Financial_System_Fragmentation_2025_p12.pdf"]},"response":{"errors":[],"extract_id":"extract_58bcfdd45ab83ce0e8eaf51a89434996","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["F I G U R E 4 Share of global exchange reserves, global export invoicing and foreign transaction by currency\n\nWhile the US dollar will likely continue to play a significant role as a reserve and funding currency, it is becoming less dominant as a medium of exchange for financing goods and services trade. Amid a trend towards greater trade regionalization, the renminbi and euro are expected to become more important within Asian and European trade. ¹³ Data from October 2024 shows that the renminbi is the second largest global currency in the trade finance market, with a share of 5.8%, just ahead of the euro but behind the US dollar, which has a share of around 83%. ¹⁴ Digital currencies, including wholesale central\n\nbank digital currencies (CBDCs), could provide alternative cross-border payment rails, creating a more diverse global financial landscape.\n\nMoreover, rising debt levels in the United States raise questions about debt sustainability, while growing domestic political polarization could undermine the independence of US institutions. Over time, these developments could reduce confidence in the dollar and lead to a more multipolar currency system.\n\nEven as financial integration continued to deepen in the 2000s, challenges to the system began to emerge. A crucial turning point came after the terrorist attacks on 11 September 2001, which shifted some governments’ view of the financial system not just as economic infrastructure, but as a powerful tool of economic statecraft.\n\nEconomic statecraft broadly refers to the use of economic tools and policies by a state to achieve its foreign-policy and domestic objectives. Such measures can include punitive steps such as sanctions, designed to pressure norms-violating countries and to protect the issuing government’s interests, as well as inducements, such as trade agreements aimed at improving relationships and fostering cooperation.\n\n## 1\\.3 Use of the financial system for economic statecraft\n\nGlobal exchange reserves Share of export invoicing Share of daily foreign transactions\n\nBillions of US dollars %\n\nGBP EUR RMB JPY USD\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\n70\n\n80\n\n90\n\n100\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\n70\n\n80\n\n90\n\n100\n\n58%\n\n20%\n\n6%\n\n2%\n\n48%\n\n$88\n\n$31\n\n$13\n\n$17\n\n$7\n\n22%\n\n7% 4% 4% 5%\n\nNote: To calculate share of export invoicing, project team used the Bloomberg Intelligence FX Strategy Dashboard and Global SWIFT data as a proxy using inbound and outbound traffic processed via SWIFT. To calculate share of foreign transactions, project team used the turnover of OTC foreign exchange instruments by currency, “net-net” basis, 2022 daily averages in billions of USD.\n\nSource: Share of global foreign exchange reserves: IMF’s Currency Composition of Official Foreign Exchange Reserves Q2 2024. Share of export invoicing: Bloomberg Intelligence FX Strategy Dashboard. Share of foreign transactions: BIS Triennial Survey, Data Portal. Oliver Wyman analysis\n\nEconomic statecraft broadly refers to the use of economic tools and policies by a state to achieve its foreign- policy and domestic objectives.\n\nNavigating Global Financial System Fragmentation 12"],"full_content":"F I G U R E 4 Share of global exchange reserves, global export invoicing and foreign transaction by currency\n\nWhile the US dollar will likely continue to play a significant role as a reserve and funding currency, it is becoming less dominant as a medium of exchange for financing goods and services trade. Amid a trend towards greater trade regionalization, the renminbi and euro are expected to become more important within Asian and European trade. ¹³ Data from October 2024 shows that the renminbi is the second largest global currency in the trade finance market, with a share of 5.8%, just ahead of the euro but behind the US dollar, which has a share of around 83%. ¹⁴ Digital currencies, including wholesale central\n\nbank digital currencies (CBDCs), could provide alternative cross-border payment rails, creating a more diverse global financial landscape.\n\nMoreover, rising debt levels in the United States raise questions about debt sustainability, while growing domestic political polarization could undermine the independence of US institutions. Over time, these developments could reduce confidence in the dollar and lead to a more multipolar currency system.\n\nEven as financial integration continued to deepen in the 2000s, challenges to the system began to emerge. A crucial turning point came after the terrorist attacks on 11 September 2001, which shifted some governments’ view of the financial system not just as economic infrastructure, but as a powerful tool of economic statecraft.\n\nEconomic statecraft broadly refers to the use of economic tools and policies by a state to achieve its foreign-policy and domestic objectives. Such measures can include punitive steps such as sanctions, designed to pressure norms-violating countries and to protect the issuing government’s interests, as well as inducements, such as trade agreements aimed at improving relationships and fostering cooperation.\n\n## 1\\.3 Use of the financial system for economic statecraft\n\nGlobal exchange reserves Share of export invoicing Share of daily foreign transactions\n\nBillions of US dollars %\n\nGBP EUR RMB JPY USD\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\n70\n\n80\n\n90\n\n100\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\n70\n\n80\n\n90\n\n100\n\n58%\n\n20%\n\n6%\n\n2%\n\n48%\n\n$88\n\n$31\n\n$13\n\n$17\n\n$7\n\n22%\n\n7% 4% 4% 5%\n\nNote: To calculate share of export invoicing, project team used the Bloomberg Intelligence FX Strategy Dashboard and Global SWIFT data as a proxy using inbound and outbound traffic processed via SWIFT. To calculate share of foreign transactions, project team used the turnover of OTC foreign exchange instruments by currency, “net-net” basis, 2022 daily averages in billions of USD.\n\nSource: Share of global foreign exchange reserves: IMF’s Currency Composition of Official Foreign Exchange Reserves Q2 2024. Share of export invoicing: Bloomberg Intelligence FX Strategy Dashboard. Share of foreign transactions: BIS Triennial Survey, Data Portal. Oliver Wyman analysis\n\nEconomic statecraft broadly refers to the use of economic tools and policies by a state to achieve its foreign- policy and domestic objectives.\n\nNavigating Global Financial System Fragmentation 12","publish_date":null,"title":"1.3 Use of the financial system for economic statecraft","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/WEF_Navigating_Global_Financial_System_Fragmentation_2025_p12.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/WEF_Navigating_Global_Financial_System_Fragmentation_2025_p12.pdf"},"request":{"example_id":"chart/WEF_Navigating_Global_Financial_System_Fragmentation_2025_p12","product_type":"parse","source_file_path":"WEF_Navigating_Global_Financial_System_Fragmentation_2025_p12.pdf"},"started_at":"2026-11-04T19:45:02.175639"} +{"completed_at":"2026-11-04T19:45:06.107447","latency_in_ms":2643,"output":{"example_id":"chart/WEF_Navigating_Global_Financial_System_Fragmentation_2025_p18","grounded_pages":[],"layout_pages":[],"markdown":"F I G U R E 7 Short-run impact of financial fragmentation on gross domestic product across geopolitical blocs\n\n\\-0.6%\n\n\\-1.4%\n\n0\\.0%\n\n\\-0.6%\n\n0%\n\n\\-2%\n\n\\-4%\n\n\\-6%\n\n\\-8%\n\n\\-10%\n\n\\-12%\n\nScenario 1 Low fragmentation\n\nMarginal change in GDP growth (deviation from baseline in %)\n\nScenario 3 High fragmentation Scenario 2 Moderate fragmentation Scenario 4 Very high fragmentation\n\n\\-1.8%\n\n\\-3.9% -3.5%\n\n\\-10.5%\n\n\\-5.5%\n\n\\-3.2% -2.8%\n\n\\-2.3%\n\n\\-2.8%\n\n\\-4.6% \\-4.5%\n\n\\-3.6%\n\nSimilar GDP decrease as the Great Recession’s impact on the US in 2009\n\nOver 2x larger GDP decrease than the global impact of the COVID-19 pandemic in 2020\n\nNeutrals East World West\n\nSource: NERA analysis based on the multi-country, multisector model of Baqaee & Farhi (2024). ³² Data from 2013 World Input–Output Database and Asian Development Bank’s 2023 Input–Output Tables. Short-run impact is defined as the impact measured one year after the shocks and is based on applying the one- year to 10-year trade elasticity ratio found in Boehm et al. (2023) ³³ to the elasticities in Baqaee & Farhi (2024), as in Bolhuis et al. (2023) ³⁴\n\nassumes that three distinct economic blocs emerge based on today’s geopolitical environment: the “West” (the US, the EU, Australia, Canada, Japan, South Korea and the United Kingdom), the\n\n“East” (China and Russia) and “Neutrals” (Brazil, India, Indonesia, Mexico, Taiwan, Türkiye and a composite of additional remaining countries).\n\nScenario 1 – Low fragmentation: In this scenario, countries restrict capital and trade flows only in certain sensitive areas and encourage unimpeded activity in all other parts of the economy. The model envisions a ban on trading sensitive technology between the Eastern and Western blocs similar to the “small yard and high fence” approach, which limits investment restrictions to sensitive industries or technologies without impeding capital flows to other parts of the economy. ³⁵ The overall impact on global output is notable, but moderate, at -0.6% after one year. Inflation increases by 0.6% globally.\n\nScenario 2 – Moderate fragmentation: This scenario introduces economic statecraft restrictions on all economic exchanges between the three blocs, with less severe measures between Neutrals and the other two blocs. Accordingly, the model shows that there are more significant tariff and financial productivity losses, driving a 2.3% decline in global economic growth in the short run. The effect on inflation is also more significant in this scenario, with a projected increase of 2.3%.\n\nNavigating Global Financial System Fragmentation 18","pages":[{"line_numbers":[],"markdown":"F I G U R E 7 Short-run impact of financial fragmentation on gross domestic product across geopolitical blocs\n\n\\-0.6%\n\n\\-1.4%\n\n0\\.0%\n\n\\-0.6%\n\n0%\n\n\\-2%\n\n\\-4%\n\n\\-6%\n\n\\-8%\n\n\\-10%\n\n\\-12%\n\nScenario 1 Low fragmentation\n\nMarginal change in GDP growth (deviation from baseline in %)\n\nScenario 3 High fragmentation Scenario 2 Moderate fragmentation Scenario 4 Very high fragmentation\n\n\\-1.8%\n\n\\-3.9% -3.5%\n\n\\-10.5%\n\n\\-5.5%\n\n\\-3.2% -2.8%\n\n\\-2.3%\n\n\\-2.8%\n\n\\-4.6% \\-4.5%\n\n\\-3.6%\n\nSimilar GDP decrease as the Great Recession’s impact on the US in 2009\n\nOver 2x larger GDP decrease than the global impact of the COVID-19 pandemic in 2020\n\nNeutrals East World West\n\nSource: NERA analysis based on the multi-country, multisector model of Baqaee & Farhi (2024). ³² Data from 2013 World Input–Output Database and Asian Development Bank’s 2023 Input–Output Tables. Short-run impact is defined as the impact measured one year after the shocks and is based on applying the one- year to 10-year trade elasticity ratio found in Boehm et al. (2023) ³³ to the elasticities in Baqaee & Farhi (2024), as in Bolhuis et al. (2023) ³⁴\n\nassumes that three distinct economic blocs emerge based on today’s geopolitical environment: the “West” (the US, the EU, Australia, Canada, Japan, South Korea and the United Kingdom), the\n\n“East” (China and Russia) and “Neutrals” (Brazil, India, Indonesia, Mexico, Taiwan, Türkiye and a composite of additional remaining countries).\n\nScenario 1 – Low fragmentation: In this scenario, countries restrict capital and trade flows only in certain sensitive areas and encourage unimpeded activity in all other parts of the economy. The model envisions a ban on trading sensitive technology between the Eastern and Western blocs similar to the “small yard and high fence” approach, which limits investment restrictions to sensitive industries or technologies without impeding capital flows to other parts of the economy. ³⁵ The overall impact on global output is notable, but moderate, at -0.6% after one year. Inflation increases by 0.6% globally.\n\nScenario 2 – Moderate fragmentation: This scenario introduces economic statecraft restrictions on all economic exchanges between the three blocs, with less severe measures between Neutrals and the other two blocs. Accordingly, the model shows that there are more significant tariff and financial productivity losses, driving a 2.3% decline in global economic growth in the short run. The effect on inflation is also more significant in this scenario, with a projected increase of 2.3%.\n\nNavigating Global Financial System Fragmentation 18","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"F I G U R E 7 Short-run impact of financial fragmentation on gross domestic product across geopolitical blocs\n\n\\-0.6%\n\n\\-1.4%\n\n0\\.0%\n\n\\-0.6%\n\n0%\n\n\\-2%\n\n\\-4%\n\n\\-6%\n\n\\-8%\n\n\\-10%\n\n\\-12%\n\nScenario 1 Low fragmentation\n\nMarginal change in GDP growth (deviation from baseline in %)\n\nScenario 3 High fragmentation Scenario 2 Moderate fragmentation Scenario 4 Very high fragmentation\n\n\\-1.8%\n\n\\-3.9% -3.5%\n\n\\-10.5%\n\n\\-5.5%\n\n\\-3.2% -2.8%\n\n\\-2.3%\n\n\\-2.8%\n\n\\-4.6% \\-4.5%\n\n\\-3.6%\n\nSimilar GDP decrease as the Great Recession’s impact on the US in 2009\n\nOver 2x larger GDP decrease than the global impact of the COVID-19 pandemic in 2020\n\nNeutrals East World West\n\nSource: NERA analysis based on the multi-country, multisector model of Baqaee & Farhi (2024). ³² Data from 2013 World Input–Output Database and Asian Development Bank’s 2023 Input–Output Tables. Short-run impact is defined as the impact measured one year after the shocks and is based on applying the one- year to 10-year trade elasticity ratio found in Boehm et al. (2023) ³³ to the elasticities in Baqaee & Farhi (2024), as in Bolhuis et al. (2023) ³⁴\n\nassumes that three distinct economic blocs emerge based on today’s geopolitical environment: the “West” (the US, the EU, Australia, Canada, Japan, South Korea and the United Kingdom), the\n\n“East” (China and Russia) and “Neutrals” (Brazil, India, Indonesia, Mexico, Taiwan, Türkiye and a composite of additional remaining countries).\n\nScenario 1 – Low fragmentation: In this scenario, countries restrict capital and trade flows only in certain sensitive areas and encourage unimpeded activity in all other parts of the economy. The model envisions a ban on trading sensitive technology between the Eastern and Western blocs similar to the “small yard and high fence” approach, which limits investment restrictions to sensitive industries or technologies without impeding capital flows to other parts of the economy. ³⁵ The overall impact on global output is notable, but moderate, at -0.6% after one year. Inflation increases by 0.6% globally.\n\nScenario 2 – Moderate fragmentation: This scenario introduces economic statecraft restrictions on all economic exchanges between the three blocs, with less severe measures between Neutrals and the other two blocs. Accordingly, the model shows that there are more significant tariff and financial productivity losses, driving a 2.3% decline in global economic growth in the short run. The effect on inflation is also more significant in this scenario, with a projected increase of 2.3%.\n\nNavigating Global Financial System Fragmentation 18","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/WEF_Navigating_Global_Financial_System_Fragmentation_2025_p18.pdf"]},"response":{"errors":[],"extract_id":"extract_995dd6729ef4759d809ad733615eaf0f","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["F I G U R E 7 Short-run impact of financial fragmentation on gross domestic product across geopolitical blocs\n\n\\-0.6%\n\n\\-1.4%\n\n0\\.0%\n\n\\-0.6%\n\n0%\n\n\\-2%\n\n\\-4%\n\n\\-6%\n\n\\-8%\n\n\\-10%\n\n\\-12%\n\nScenario 1 Low fragmentation\n\nMarginal change in GDP growth (deviation from baseline in %)\n\nScenario 3 High fragmentation Scenario 2 Moderate fragmentation Scenario 4 Very high fragmentation\n\n\\-1.8%\n\n\\-3.9% -3.5%\n\n\\-10.5%\n\n\\-5.5%\n\n\\-3.2% -2.8%\n\n\\-2.3%\n\n\\-2.8%\n\n\\-4.6% \\-4.5%\n\n\\-3.6%\n\nSimilar GDP decrease as the Great Recession’s impact on the US in 2009\n\nOver 2x larger GDP decrease than the global impact of the COVID-19 pandemic in 2020\n\nNeutrals East World West\n\nSource: NERA analysis based on the multi-country, multisector model of Baqaee & Farhi (2024). ³² Data from 2013 World Input–Output Database and Asian Development Bank’s 2023 Input–Output Tables. Short-run impact is defined as the impact measured one year after the shocks and is based on applying the one- year to 10-year trade elasticity ratio found in Boehm et al. (2023) ³³ to the elasticities in Baqaee & Farhi (2024), as in Bolhuis et al. (2023) ³⁴\n\nassumes that three distinct economic blocs emerge based on today’s geopolitical environment: the “West” (the US, the EU, Australia, Canada, Japan, South Korea and the United Kingdom), the\n\n“East” (China and Russia) and “Neutrals” (Brazil, India, Indonesia, Mexico, Taiwan, Türkiye and a composite of additional remaining countries).\n\nScenario 1 – Low fragmentation: In this scenario, countries restrict capital and trade flows only in certain sensitive areas and encourage unimpeded activity in all other parts of the economy. The model envisions a ban on trading sensitive technology between the Eastern and Western blocs similar to the “small yard and high fence” approach, which limits investment restrictions to sensitive industries or technologies without impeding capital flows to other parts of the economy. ³⁵ The overall impact on global output is notable, but moderate, at -0.6% after one year. Inflation increases by 0.6% globally.\n\nScenario 2 – Moderate fragmentation: This scenario introduces economic statecraft restrictions on all economic exchanges between the three blocs, with less severe measures between Neutrals and the other two blocs. Accordingly, the model shows that there are more significant tariff and financial productivity losses, driving a 2.3% decline in global economic growth in the short run. The effect on inflation is also more significant in this scenario, with a projected increase of 2.3%.\n\nNavigating Global Financial System Fragmentation 18"],"full_content":"F I G U R E 7 Short-run impact of financial fragmentation on gross domestic product across geopolitical blocs\n\n\\-0.6%\n\n\\-1.4%\n\n0\\.0%\n\n\\-0.6%\n\n0%\n\n\\-2%\n\n\\-4%\n\n\\-6%\n\n\\-8%\n\n\\-10%\n\n\\-12%\n\nScenario 1 Low fragmentation\n\nMarginal change in GDP growth (deviation from baseline in %)\n\nScenario 3 High fragmentation Scenario 2 Moderate fragmentation Scenario 4 Very high fragmentation\n\n\\-1.8%\n\n\\-3.9% -3.5%\n\n\\-10.5%\n\n\\-5.5%\n\n\\-3.2% -2.8%\n\n\\-2.3%\n\n\\-2.8%\n\n\\-4.6% \\-4.5%\n\n\\-3.6%\n\nSimilar GDP decrease as the Great Recession’s impact on the US in 2009\n\nOver 2x larger GDP decrease than the global impact of the COVID-19 pandemic in 2020\n\nNeutrals East World West\n\nSource: NERA analysis based on the multi-country, multisector model of Baqaee & Farhi (2024). ³² Data from 2013 World Input–Output Database and Asian Development Bank’s 2023 Input–Output Tables. Short-run impact is defined as the impact measured one year after the shocks and is based on applying the one- year to 10-year trade elasticity ratio found in Boehm et al. (2023) ³³ to the elasticities in Baqaee & Farhi (2024), as in Bolhuis et al. (2023) ³⁴\n\nassumes that three distinct economic blocs emerge based on today’s geopolitical environment: the “West” (the US, the EU, Australia, Canada, Japan, South Korea and the United Kingdom), the\n\n“East” (China and Russia) and “Neutrals” (Brazil, India, Indonesia, Mexico, Taiwan, Türkiye and a composite of additional remaining countries).\n\nScenario 1 – Low fragmentation: In this scenario, countries restrict capital and trade flows only in certain sensitive areas and encourage unimpeded activity in all other parts of the economy. The model envisions a ban on trading sensitive technology between the Eastern and Western blocs similar to the “small yard and high fence” approach, which limits investment restrictions to sensitive industries or technologies without impeding capital flows to other parts of the economy. ³⁵ The overall impact on global output is notable, but moderate, at -0.6% after one year. Inflation increases by 0.6% globally.\n\nScenario 2 – Moderate fragmentation: This scenario introduces economic statecraft restrictions on all economic exchanges between the three blocs, with less severe measures between Neutrals and the other two blocs. Accordingly, the model shows that there are more significant tariff and financial productivity losses, driving a 2.3% decline in global economic growth in the short run. The effect on inflation is also more significant in this scenario, with a projected increase of 2.3%.\n\nNavigating Global Financial System Fragmentation 18","publish_date":null,"title":"FIGURE 7 Short-run impact of financial fragmentation on gross domestic product across geopolitical blocs","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/WEF_Navigating_Global_Financial_System_Fragmentation_2025_p18.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/WEF_Navigating_Global_Financial_System_Fragmentation_2025_p18.pdf"},"request":{"example_id":"chart/WEF_Navigating_Global_Financial_System_Fragmentation_2025_p18","product_type":"parse","source_file_path":"WEF_Navigating_Global_Financial_System_Fragmentation_2025_p18.pdf"},"started_at":"2026-11-04T19:45:03.463517"} +{"completed_at":"2026-11-04T19:45:07.292412","latency_in_ms":2773,"output":{"example_id":"chart/WEF_The_Global_Public_Impact_of_GovTech_2025_p29","grounded_pages":[],"layout_pages":[],"markdown":"## A3 Methodology\n\nThe study employs a structured approach to estimate the GovTech market size (1) and its GovTech opportunity for the public sector (2) by 2034. The methodology integrates an analysis of public sector IT spending, expert opinions and established models to produce reliable estimates of both market growth and societal impact.\n\n1\\. Assessment of the GovTech market size\n\na. Determination of sector focus: Calculation of the market comprises four sectors: public services, education, healthcare/life sciences and transport.\n\nb. Projecting IT spending: Growth projections extended to 2034 by applying the average annual growth rate of 8% (based on 2022-2028 data) (see Figure 5). 13,14\n\nIT spending in various sectors (in $ billions in 2024 and 2034) with an estimate growth rate of 8% per annum F I G U R E 5\n\n$1,570\n\n## 2034\n\n$220\n\n$170\n\n$70\n\n## 2024\n\n$40\n\n$90\n\n$70\n\n$670\n\nPublic sector Education Healthcare and life science Transport\n\n## A2 Additional information for the country cases\n\nAzerbaijan: For more information, see: Centre for the Fourth Industrial Revolution Azerbaijan\n\nBahrain: For more information, see: Information & eGovernment Authority\n\nBrazil: For more information, see: Ministry of Management and Innovation in Public Services\n\nGermany: For more information, see: CIO Bund\n\nMalaysia: For more information, see: MyDIGITAL\n\nRwanda: For more information, see: IremboGov\n\nUkraine: For more information, see: Ministry of Digital Transformation\n\nThe Global Public Impact of GovTech: A $9.8 Trillion Opportunity 29","pages":[{"line_numbers":[],"markdown":"## A3 Methodology\n\nThe study employs a structured approach to estimate the GovTech market size (1) and its GovTech opportunity for the public sector (2) by 2034. The methodology integrates an analysis of public sector IT spending, expert opinions and established models to produce reliable estimates of both market growth and societal impact.\n\n1\\. Assessment of the GovTech market size\n\na. Determination of sector focus: Calculation of the market comprises four sectors: public services, education, healthcare/life sciences and transport.\n\nb. Projecting IT spending: Growth projections extended to 2034 by applying the average annual growth rate of 8% (based on 2022-2028 data) (see Figure 5). 13,14\n\nIT spending in various sectors (in $ billions in 2024 and 2034) with an estimate growth rate of 8% per annum F I G U R E 5\n\n$1,570\n\n## 2034\n\n$220\n\n$170\n\n$70\n\n## 2024\n\n$40\n\n$90\n\n$70\n\n$670\n\nPublic sector Education Healthcare and life science Transport\n\n## A2 Additional information for the country cases\n\nAzerbaijan: For more information, see: Centre for the Fourth Industrial Revolution Azerbaijan\n\nBahrain: For more information, see: Information & eGovernment Authority\n\nBrazil: For more information, see: Ministry of Management and Innovation in Public Services\n\nGermany: For more information, see: CIO Bund\n\nMalaysia: For more information, see: MyDIGITAL\n\nRwanda: For more information, see: IremboGov\n\nUkraine: For more information, see: Ministry of Digital Transformation\n\nThe Global Public Impact of GovTech: A $9.8 Trillion Opportunity 29","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"## A3 Methodology\n\nThe study employs a structured approach to estimate the GovTech market size (1) and its GovTech opportunity for the public sector (2) by 2034. The methodology integrates an analysis of public sector IT spending, expert opinions and established models to produce reliable estimates of both market growth and societal impact.\n\n1\\. Assessment of the GovTech market size\n\na. Determination of sector focus: Calculation of the market comprises four sectors: public services, education, healthcare/life sciences and transport.\n\nb. Projecting IT spending: Growth projections extended to 2034 by applying the average annual growth rate of 8% (based on 2022-2028 data) (see Figure 5). 13,14\n\nIT spending in various sectors (in $ billions in 2024 and 2034) with an estimate growth rate of 8% per annum F I G U R E 5\n\n$1,570\n\n## 2034\n\n$220\n\n$170\n\n$70\n\n## 2024\n\n$40\n\n$90\n\n$70\n\n$670\n\nPublic sector Education Healthcare and life science Transport\n\n## A2 Additional information for the country cases\n\nAzerbaijan: For more information, see: Centre for the Fourth Industrial Revolution Azerbaijan\n\nBahrain: For more information, see: Information & eGovernment Authority\n\nBrazil: For more information, see: Ministry of Management and Innovation in Public Services\n\nGermany: For more information, see: CIO Bund\n\nMalaysia: For more information, see: MyDIGITAL\n\nRwanda: For more information, see: IremboGov\n\nUkraine: For more information, see: Ministry of Digital Transformation\n\nThe Global Public Impact of GovTech: A $9.8 Trillion Opportunity 29","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/WEF_The_Global_Public_Impact_of_GovTech_2025_p29.pdf"]},"response":{"errors":[],"extract_id":"extract_453b9e98a16500631e10b18d289081ef","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["## A3 Methodology\n\nThe study employs a structured approach to estimate the GovTech market size (1) and its GovTech opportunity for the public sector (2) by 2034. The methodology integrates an analysis of public sector IT spending, expert opinions and established models to produce reliable estimates of both market growth and societal impact.\n\n1\\. Assessment of the GovTech market size\n\na. Determination of sector focus: Calculation of the market comprises four sectors: public services, education, healthcare/life sciences and transport.\n\nb. Projecting IT spending: Growth projections extended to 2034 by applying the average annual growth rate of 8% (based on 2022-2028 data) (see Figure 5). 13,14\n\nIT spending in various sectors (in $ billions in 2024 and 2034) with an estimate growth rate of 8% per annum F I G U R E 5\n\n$1,570\n\n## 2034\n\n$220\n\n$170\n\n$70\n\n## 2024\n\n$40\n\n$90\n\n$70\n\n$670\n\nPublic sector Education Healthcare and life science Transport\n\n## A2 Additional information for the country cases\n\nAzerbaijan: For more information, see: Centre for the Fourth Industrial Revolution Azerbaijan\n\nBahrain: For more information, see: Information & eGovernment Authority\n\nBrazil: For more information, see: Ministry of Management and Innovation in Public Services\n\nGermany: For more information, see: CIO Bund\n\nMalaysia: For more information, see: MyDIGITAL\n\nRwanda: For more information, see: IremboGov\n\nUkraine: For more information, see: Ministry of Digital Transformation\n\nThe Global Public Impact of GovTech: A $9.8 Trillion Opportunity 29"],"full_content":"## A3 Methodology\n\nThe study employs a structured approach to estimate the GovTech market size (1) and its GovTech opportunity for the public sector (2) by 2034. The methodology integrates an analysis of public sector IT spending, expert opinions and established models to produce reliable estimates of both market growth and societal impact.\n\n1\\. Assessment of the GovTech market size\n\na. Determination of sector focus: Calculation of the market comprises four sectors: public services, education, healthcare/life sciences and transport.\n\nb. Projecting IT spending: Growth projections extended to 2034 by applying the average annual growth rate of 8% (based on 2022-2028 data) (see Figure 5). 13,14\n\nIT spending in various sectors (in $ billions in 2024 and 2034) with an estimate growth rate of 8% per annum F I G U R E 5\n\n$1,570\n\n## 2034\n\n$220\n\n$170\n\n$70\n\n## 2024\n\n$40\n\n$90\n\n$70\n\n$670\n\nPublic sector Education Healthcare and life science Transport\n\n## A2 Additional information for the country cases\n\nAzerbaijan: For more information, see: Centre for the Fourth Industrial Revolution Azerbaijan\n\nBahrain: For more information, see: Information & eGovernment Authority\n\nBrazil: For more information, see: Ministry of Management and Innovation in Public Services\n\nGermany: For more information, see: CIO Bund\n\nMalaysia: For more information, see: MyDIGITAL\n\nRwanda: For more information, see: IremboGov\n\nUkraine: For more information, see: Ministry of Digital Transformation\n\nThe Global Public Impact of GovTech: A $9.8 Trillion Opportunity 29","publish_date":null,"title":"2034 2024","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/WEF_The_Global_Public_Impact_of_GovTech_2025_p29.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/WEF_The_Global_Public_Impact_of_GovTech_2025_p29.pdf"},"request":{"example_id":"chart/WEF_The_Global_Public_Impact_of_GovTech_2025_p29","product_type":"parse","source_file_path":"WEF_The_Global_Public_Impact_of_GovTech_2025_p29.pdf"},"started_at":"2026-11-04T19:45:04.518730"} +{"completed_at":"2026-11-04T19:45:07.458874","latency_in_ms":1349,"output":{"example_id":"chart/Whatnextfortheglobalcarindustry_p20","grounded_pages":[],"layout_pages":[],"markdown":"What Next for the Global Car Industry? Chapter 1: The global car industry An Energy Technology Perspectives Special Report in context\n\n20\n\nI EA. CC BY 4.0.\n\nlower cost. As the contribution of the electric motor is limited, and fuel savings are generally low (between 5% and 15% depending on motor voltage and hybrid architecture), these vehicles are often classified as ICE vehicles.\n\n Full hybrid electric vehicles ( **FHEV** ) use high-voltage electric motors (typically over 200 V) that can power the vehicle with the engine off for short distances. These are typically paired with a 1-2 kWh battery, and their higher electric power supports stronger regenerative braking and delivers more substantial fuel savings compared to conventional cars, depending on driving mode (20-30% on average under the Worldwide Harmonized Light Vehicles Test Cycle (WLTC) and up to 40% under urban driving conditions).\n\nThe fuel-saving potential of HEVs means they can be used by carmakers to comply with CO2 emission and fuel economy standards in many car markets. With a lower purchase price than electric cars, hybrid electric cars appeal to many consumers looking for an efficient vehicle with a small purchase price premium that can be used with conventional refuelling infrastructure. In 2024, hybrid electric car sales (including both FHEVs and MHEVs) reached 12.5 million – more than 15% of global car sales. Sales of MHEVs, which have a lower purchase price than equivalent FHEVs, have grown markedly in recent years. In 2024, they accounted for 45% of global HEV sales.\n\n**Car market shares by powertrain in selected markets, 2019 and 2024**\n\nIEA. CC BY 4.0.\n\nNotes: ROW = Rest of World; BEV = battery electric vehicle; PHEV = plug-in hybrid vehicle; FHEV = full hybrid electric vehicle; MHEV = mild-hybrid electric vehicle. “Others” includes internal combustion engine cars and fuel cell electric cars. Sources: IEA analysis based on Marklines and EV Volumes.\n\nMarket uptake of full and mild-hybrid electric cars has historically been highest in Japan, where they represented more than half of domestic sales in 2024, up from 35% in 2019. In Europe, carmakers have been offering an increasing number of hybrid electric models in their lineups, with 80 FHEV and 150 MHEV models available in 2024, over twice the level seen in 2019. As a result, their market share\n\n0%\n\n20%\n\n40%\n\n60%\n\n80%\n\n100%\n\n2019 2024 2019 2024 2019 2024 2019 2024 2019 2024\n\nChina Europe United States Japan ROW\n\nBEV PHEV FHEV MHEV Others","pages":[{"line_numbers":[],"markdown":"What Next for the Global Car Industry? Chapter 1: The global car industry An Energy Technology Perspectives Special Report in context\n\n20\n\nI EA. CC BY 4.0.\n\nlower cost. As the contribution of the electric motor is limited, and fuel savings are generally low (between 5% and 15% depending on motor voltage and hybrid architecture), these vehicles are often classified as ICE vehicles.\n\n Full hybrid electric vehicles ( **FHEV** ) use high-voltage electric motors (typically over 200 V) that can power the vehicle with the engine off for short distances. These are typically paired with a 1-2 kWh battery, and their higher electric power supports stronger regenerative braking and delivers more substantial fuel savings compared to conventional cars, depending on driving mode (20-30% on average under the Worldwide Harmonized Light Vehicles Test Cycle (WLTC) and up to 40% under urban driving conditions).\n\nThe fuel-saving potential of HEVs means they can be used by carmakers to comply with CO2 emission and fuel economy standards in many car markets. With a lower purchase price than electric cars, hybrid electric cars appeal to many consumers looking for an efficient vehicle with a small purchase price premium that can be used with conventional refuelling infrastructure. In 2024, hybrid electric car sales (including both FHEVs and MHEVs) reached 12.5 million – more than 15% of global car sales. Sales of MHEVs, which have a lower purchase price than equivalent FHEVs, have grown markedly in recent years. In 2024, they accounted for 45% of global HEV sales.\n\n**Car market shares by powertrain in selected markets, 2019 and 2024**\n\nIEA. CC BY 4.0.\n\nNotes: ROW = Rest of World; BEV = battery electric vehicle; PHEV = plug-in hybrid vehicle; FHEV = full hybrid electric vehicle; MHEV = mild-hybrid electric vehicle. “Others” includes internal combustion engine cars and fuel cell electric cars. Sources: IEA analysis based on Marklines and EV Volumes.\n\nMarket uptake of full and mild-hybrid electric cars has historically been highest in Japan, where they represented more than half of domestic sales in 2024, up from 35% in 2019. In Europe, carmakers have been offering an increasing number of hybrid electric models in their lineups, with 80 FHEV and 150 MHEV models available in 2024, over twice the level seen in 2019. As a result, their market share\n\n0%\n\n20%\n\n40%\n\n60%\n\n80%\n\n100%\n\n2019 2024 2019 2024 2019 2024 2019 2024 2019 2024\n\nChina Europe United States Japan ROW\n\nBEV PHEV FHEV MHEV Others","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"What Next for the Global Car Industry? Chapter 1: The global car industry An Energy Technology Perspectives Special Report in context\n\n20\n\nI EA. CC BY 4.0.\n\nlower cost. As the contribution of the electric motor is limited, and fuel savings are generally low (between 5% and 15% depending on motor voltage and hybrid architecture), these vehicles are often classified as ICE vehicles.\n\n Full hybrid electric vehicles ( **FHEV** ) use high-voltage electric motors (typically over 200 V) that can power the vehicle with the engine off for short distances. These are typically paired with a 1-2 kWh battery, and their higher electric power supports stronger regenerative braking and delivers more substantial fuel savings compared to conventional cars, depending on driving mode (20-30% on average under the Worldwide Harmonized Light Vehicles Test Cycle (WLTC) and up to 40% under urban driving conditions).\n\nThe fuel-saving potential of HEVs means they can be used by carmakers to comply with CO2 emission and fuel economy standards in many car markets. With a lower purchase price than electric cars, hybrid electric cars appeal to many consumers looking for an efficient vehicle with a small purchase price premium that can be used with conventional refuelling infrastructure. In 2024, hybrid electric car sales (including both FHEVs and MHEVs) reached 12.5 million – more than 15% of global car sales. Sales of MHEVs, which have a lower purchase price than equivalent FHEVs, have grown markedly in recent years. In 2024, they accounted for 45% of global HEV sales.\n\n**Car market shares by powertrain in selected markets, 2019 and 2024**\n\nIEA. CC BY 4.0.\n\nNotes: ROW = Rest of World; BEV = battery electric vehicle; PHEV = plug-in hybrid vehicle; FHEV = full hybrid electric vehicle; MHEV = mild-hybrid electric vehicle. “Others” includes internal combustion engine cars and fuel cell electric cars. Sources: IEA analysis based on Marklines and EV Volumes.\n\nMarket uptake of full and mild-hybrid electric cars has historically been highest in Japan, where they represented more than half of domestic sales in 2024, up from 35% in 2019. In Europe, carmakers have been offering an increasing number of hybrid electric models in their lineups, with 80 FHEV and 150 MHEV models available in 2024, over twice the level seen in 2019. As a result, their market share\n\n0%\n\n20%\n\n40%\n\n60%\n\n80%\n\n100%\n\n2019 2024 2019 2024 2019 2024 2019 2024 2019 2024\n\nChina Europe United States Japan ROW\n\nBEV PHEV FHEV MHEV Others","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Whatnextfortheglobalcarindustry_p20.pdf"]},"response":{"errors":[],"extract_id":"extract_78479da728a217084608da4b4609825f","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["What Next for the Global Car Industry? Chapter 1: The global car industry An Energy Technology Perspectives Special Report in context\n\n20\n\nI EA. CC BY 4.0.\n\nlower cost. As the contribution of the electric motor is limited, and fuel savings are generally low (between 5% and 15% depending on motor voltage and hybrid architecture), these vehicles are often classified as ICE vehicles.\n\n Full hybrid electric vehicles ( **FHEV** ) use high-voltage electric motors (typically over 200 V) that can power the vehicle with the engine off for short distances. These are typically paired with a 1-2 kWh battery, and their higher electric power supports stronger regenerative braking and delivers more substantial fuel savings compared to conventional cars, depending on driving mode (20-30% on average under the Worldwide Harmonized Light Vehicles Test Cycle (WLTC) and up to 40% under urban driving conditions).\n\nThe fuel-saving potential of HEVs means they can be used by carmakers to comply with CO2 emission and fuel economy standards in many car markets. With a lower purchase price than electric cars, hybrid electric cars appeal to many consumers looking for an efficient vehicle with a small purchase price premium that can be used with conventional refuelling infrastructure. In 2024, hybrid electric car sales (including both FHEVs and MHEVs) reached 12.5 million – more than 15% of global car sales. Sales of MHEVs, which have a lower purchase price than equivalent FHEVs, have grown markedly in recent years. In 2024, they accounted for 45% of global HEV sales.\n\n**Car market shares by powertrain in selected markets, 2019 and 2024**\n\nIEA. CC BY 4.0.\n\nNotes: ROW = Rest of World; BEV = battery electric vehicle; PHEV = plug-in hybrid vehicle; FHEV = full hybrid electric vehicle; MHEV = mild-hybrid electric vehicle. “Others” includes internal combustion engine cars and fuel cell electric cars. Sources: IEA analysis based on Marklines and EV Volumes.\n\nMarket uptake of full and mild-hybrid electric cars has historically been highest in Japan, where they represented more than half of domestic sales in 2024, up from 35% in 2019. In Europe, carmakers have been offering an increasing number of hybrid electric models in their lineups, with 80 FHEV and 150 MHEV models available in 2024, over twice the level seen in 2019. As a result, their market share\n\n0%\n\n20%\n\n40%\n\n60%\n\n80%\n\n100%\n\n2019 2024 2019 2024 2019 2024 2019 2024 2019 2024\n\nChina Europe United States Japan ROW\n\nBEV PHEV FHEV MHEV Others"],"full_content":"What Next for the Global Car Industry? Chapter 1: The global car industry An Energy Technology Perspectives Special Report in context\n\n20\n\nI EA. CC BY 4.0.\n\nlower cost. As the contribution of the electric motor is limited, and fuel savings are generally low (between 5% and 15% depending on motor voltage and hybrid architecture), these vehicles are often classified as ICE vehicles.\n\n Full hybrid electric vehicles ( **FHEV** ) use high-voltage electric motors (typically over 200 V) that can power the vehicle with the engine off for short distances. These are typically paired with a 1-2 kWh battery, and their higher electric power supports stronger regenerative braking and delivers more substantial fuel savings compared to conventional cars, depending on driving mode (20-30% on average under the Worldwide Harmonized Light Vehicles Test Cycle (WLTC) and up to 40% under urban driving conditions).\n\nThe fuel-saving potential of HEVs means they can be used by carmakers to comply with CO2 emission and fuel economy standards in many car markets. With a lower purchase price than electric cars, hybrid electric cars appeal to many consumers looking for an efficient vehicle with a small purchase price premium that can be used with conventional refuelling infrastructure. In 2024, hybrid electric car sales (including both FHEVs and MHEVs) reached 12.5 million – more than 15% of global car sales. Sales of MHEVs, which have a lower purchase price than equivalent FHEVs, have grown markedly in recent years. In 2024, they accounted for 45% of global HEV sales.\n\n**Car market shares by powertrain in selected markets, 2019 and 2024**\n\nIEA. CC BY 4.0.\n\nNotes: ROW = Rest of World; BEV = battery electric vehicle; PHEV = plug-in hybrid vehicle; FHEV = full hybrid electric vehicle; MHEV = mild-hybrid electric vehicle. “Others” includes internal combustion engine cars and fuel cell electric cars. Sources: IEA analysis based on Marklines and EV Volumes.\n\nMarket uptake of full and mild-hybrid electric cars has historically been highest in Japan, where they represented more than half of domestic sales in 2024, up from 35% in 2019. In Europe, carmakers have been offering an increasing number of hybrid electric models in their lineups, with 80 FHEV and 150 MHEV models available in 2024, over twice the level seen in 2019. As a result, their market share\n\n0%\n\n20%\n\n40%\n\n60%\n\n80%\n\n100%\n\n2019 2024 2019 2024 2019 2024 2019 2024 2019 2024\n\nChina Europe United States Japan ROW\n\nBEV PHEV FHEV MHEV Others","publish_date":null,"title":"lower cost. As the contribution of the electric motor is limited, and fuel savings are generally low (between 5% and","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Whatnextfortheglobalcarindustry_p20.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Whatnextfortheglobalcarindustry_p20.pdf"},"request":{"example_id":"chart/Whatnextfortheglobalcarindustry_p20","product_type":"parse","source_file_path":"Whatnextfortheglobalcarindustry_p20.pdf"},"started_at":"2026-11-04T19:45:06.108956"} +{"completed_at":"2026-11-04T19:45:09.658826","latency_in_ms":2198,"output":{"example_id":"chart/Whatnextfortheglobalcarindustry_p22","grounded_pages":[],"layout_pages":[],"markdown":"What Next for the Global Car Industry? Chapter 1: The global car industry An Energy Technology Perspectives Special Report in context\n\n22\n\nI EA. CC BY 4.0.\n\n**Figure 1.2 Sales share of SUVs and pick-up trucks by market, 2010-2024**\n\nIEA. CC BY 4.0.\n\nNotes: Sales share includes segments SUV-A to SUV-E and pick-up trucks. Source: IEA analysis based on Marklines.\n\nBeyond consumer preferences for larger cars, regulatory frameworks have also contributed to SUV sales growth. For example, the US Corporate Average Fuel Economy (CAFE) Standards set separate requirements for passenger cars and light trucks based on the footprint of the vehicle. Under this framework, SUVs are classified as light trucks, and the standards for model years 2012-2016 required greater efficiency improvements for passenger cars than for light trucks. Similarly, EU CO2 standards for cars also included emissions target adjustments for carmakers selling heavier-than-average cars until 2025. Differentiated emission targets and the higher profit margins on larger models help explain why automakers have increasingly chosen to produce bigger cars. However, as the EU CO2 standards enter the new enforcement period for 2025-2029, specific manufacturer CO2 targets have been revised based on the latest market data. As a result, previous CO2 emission allowances granted to manufacturers of heavier- than-average cars have been cancelled. See the box Cars are getting heavier despite material substitution efforts below for more on the impact of car size increases.\n\n## **The share of global car manufacturing located in China has grown substantially**\n\nThe car market is highly globalised: in most countries, consumers can buy cars from all major original equipment manufacturers (OEMs), even though different carmakers have typically focused on different markets and powertrains. Nearly all the cars sold worldwide today are manufactured by carmakers with headquarters in either the European Union, the United States, Japan, China or Korea, and are manufactured in those countries or regions. In 2008, the majority of global car production took place in Europe (25% of global car output), North America (25%),\n\n0%\n\n20%\n\n40%\n\n60%\n\n80%\n\n2010 2012 2014 2016 2018 2020 2022 2024\n\nGlobal China\n\nNorth America Other advanced economies\n\nEurope Other emerging and developing economies","pages":[{"line_numbers":[],"markdown":"What Next for the Global Car Industry? Chapter 1: The global car industry An Energy Technology Perspectives Special Report in context\n\n22\n\nI EA. CC BY 4.0.\n\n**Figure 1.2 Sales share of SUVs and pick-up trucks by market, 2010-2024**\n\nIEA. CC BY 4.0.\n\nNotes: Sales share includes segments SUV-A to SUV-E and pick-up trucks. Source: IEA analysis based on Marklines.\n\nBeyond consumer preferences for larger cars, regulatory frameworks have also contributed to SUV sales growth. For example, the US Corporate Average Fuel Economy (CAFE) Standards set separate requirements for passenger cars and light trucks based on the footprint of the vehicle. Under this framework, SUVs are classified as light trucks, and the standards for model years 2012-2016 required greater efficiency improvements for passenger cars than for light trucks. Similarly, EU CO2 standards for cars also included emissions target adjustments for carmakers selling heavier-than-average cars until 2025. Differentiated emission targets and the higher profit margins on larger models help explain why automakers have increasingly chosen to produce bigger cars. However, as the EU CO2 standards enter the new enforcement period for 2025-2029, specific manufacturer CO2 targets have been revised based on the latest market data. As a result, previous CO2 emission allowances granted to manufacturers of heavier- than-average cars have been cancelled. See the box Cars are getting heavier despite material substitution efforts below for more on the impact of car size increases.\n\n## **The share of global car manufacturing located in China has grown substantially**\n\nThe car market is highly globalised: in most countries, consumers can buy cars from all major original equipment manufacturers (OEMs), even though different carmakers have typically focused on different markets and powertrains. Nearly all the cars sold worldwide today are manufactured by carmakers with headquarters in either the European Union, the United States, Japan, China or Korea, and are manufactured in those countries or regions. In 2008, the majority of global car production took place in Europe (25% of global car output), North America (25%),\n\n0%\n\n20%\n\n40%\n\n60%\n\n80%\n\n2010 2012 2014 2016 2018 2020 2022 2024\n\nGlobal China\n\nNorth America Other advanced economies\n\nEurope Other emerging and developing economies","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"What Next for the Global Car Industry? Chapter 1: The global car industry An Energy Technology Perspectives Special Report in context\n\n22\n\nI EA. CC BY 4.0.\n\n**Figure 1.2 Sales share of SUVs and pick-up trucks by market, 2010-2024**\n\nIEA. CC BY 4.0.\n\nNotes: Sales share includes segments SUV-A to SUV-E and pick-up trucks. Source: IEA analysis based on Marklines.\n\nBeyond consumer preferences for larger cars, regulatory frameworks have also contributed to SUV sales growth. For example, the US Corporate Average Fuel Economy (CAFE) Standards set separate requirements for passenger cars and light trucks based on the footprint of the vehicle. Under this framework, SUVs are classified as light trucks, and the standards for model years 2012-2016 required greater efficiency improvements for passenger cars than for light trucks. Similarly, EU CO2 standards for cars also included emissions target adjustments for carmakers selling heavier-than-average cars until 2025. Differentiated emission targets and the higher profit margins on larger models help explain why automakers have increasingly chosen to produce bigger cars. However, as the EU CO2 standards enter the new enforcement period for 2025-2029, specific manufacturer CO2 targets have been revised based on the latest market data. As a result, previous CO2 emission allowances granted to manufacturers of heavier- than-average cars have been cancelled. See the box Cars are getting heavier despite material substitution efforts below for more on the impact of car size increases.\n\n## **The share of global car manufacturing located in China has grown substantially**\n\nThe car market is highly globalised: in most countries, consumers can buy cars from all major original equipment manufacturers (OEMs), even though different carmakers have typically focused on different markets and powertrains. Nearly all the cars sold worldwide today are manufactured by carmakers with headquarters in either the European Union, the United States, Japan, China or Korea, and are manufactured in those countries or regions. In 2008, the majority of global car production took place in Europe (25% of global car output), North America (25%),\n\n0%\n\n20%\n\n40%\n\n60%\n\n80%\n\n2010 2012 2014 2016 2018 2020 2022 2024\n\nGlobal China\n\nNorth America Other advanced economies\n\nEurope Other emerging and developing economies","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Whatnextfortheglobalcarindustry_p22.pdf"]},"response":{"errors":[],"extract_id":"extract_5ef84be38da9ac95401486f5e6288256","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["What Next for the Global Car Industry? Chapter 1: The global car industry An Energy Technology Perspectives Special Report in context\n\n22\n\nI EA. CC BY 4.0.\n\n**Figure 1.2 Sales share of SUVs and pick-up trucks by market, 2010-2024**\n\nIEA. CC BY 4.0.\n\nNotes: Sales share includes segments SUV-A to SUV-E and pick-up trucks. Source: IEA analysis based on Marklines.\n\nBeyond consumer preferences for larger cars, regulatory frameworks have also contributed to SUV sales growth. For example, the US Corporate Average Fuel Economy (CAFE) Standards set separate requirements for passenger cars and light trucks based on the footprint of the vehicle. Under this framework, SUVs are classified as light trucks, and the standards for model years 2012-2016 required greater efficiency improvements for passenger cars than for light trucks. Similarly, EU CO2 standards for cars also included emissions target adjustments for carmakers selling heavier-than-average cars until 2025. Differentiated emission targets and the higher profit margins on larger models help explain why automakers have increasingly chosen to produce bigger cars. However, as the EU CO2 standards enter the new enforcement period for 2025-2029, specific manufacturer CO2 targets have been revised based on the latest market data. As a result, previous CO2 emission allowances granted to manufacturers of heavier- than-average cars have been cancelled. See the box Cars are getting heavier despite material substitution efforts below for more on the impact of car size increases.\n\n## **The share of global car manufacturing located in China has grown substantially**\n\nThe car market is highly globalised: in most countries, consumers can buy cars from all major original equipment manufacturers (OEMs), even though different carmakers have typically focused on different markets and powertrains. Nearly all the cars sold worldwide today are manufactured by carmakers with headquarters in either the European Union, the United States, Japan, China or Korea, and are manufactured in those countries or regions. In 2008, the majority of global car production took place in Europe (25% of global car output), North America (25%),\n\n0%\n\n20%\n\n40%\n\n60%\n\n80%\n\n2010 2012 2014 2016 2018 2020 2022 2024\n\nGlobal China\n\nNorth America Other advanced economies\n\nEurope Other emerging and developing economies"],"full_content":"What Next for the Global Car Industry? Chapter 1: The global car industry An Energy Technology Perspectives Special Report in context\n\n22\n\nI EA. CC BY 4.0.\n\n**Figure 1.2 Sales share of SUVs and pick-up trucks by market, 2010-2024**\n\nIEA. CC BY 4.0.\n\nNotes: Sales share includes segments SUV-A to SUV-E and pick-up trucks. Source: IEA analysis based on Marklines.\n\nBeyond consumer preferences for larger cars, regulatory frameworks have also contributed to SUV sales growth. For example, the US Corporate Average Fuel Economy (CAFE) Standards set separate requirements for passenger cars and light trucks based on the footprint of the vehicle. Under this framework, SUVs are classified as light trucks, and the standards for model years 2012-2016 required greater efficiency improvements for passenger cars than for light trucks. Similarly, EU CO2 standards for cars also included emissions target adjustments for carmakers selling heavier-than-average cars until 2025. Differentiated emission targets and the higher profit margins on larger models help explain why automakers have increasingly chosen to produce bigger cars. However, as the EU CO2 standards enter the new enforcement period for 2025-2029, specific manufacturer CO2 targets have been revised based on the latest market data. As a result, previous CO2 emission allowances granted to manufacturers of heavier- than-average cars have been cancelled. See the box Cars are getting heavier despite material substitution efforts below for more on the impact of car size increases.\n\n## **The share of global car manufacturing located in China has grown substantially**\n\nThe car market is highly globalised: in most countries, consumers can buy cars from all major original equipment manufacturers (OEMs), even though different carmakers have typically focused on different markets and powertrains. Nearly all the cars sold worldwide today are manufactured by carmakers with headquarters in either the European Union, the United States, Japan, China or Korea, and are manufactured in those countries or regions. In 2008, the majority of global car production took place in Europe (25% of global car output), North America (25%),\n\n0%\n\n20%\n\n40%\n\n60%\n\n80%\n\n2010 2012 2014 2016 2018 2020 2022 2024\n\nGlobal China\n\nNorth America Other advanced economies\n\nEurope Other emerging and developing economies","publish_date":null,"title":"The share of global car manufacturing located in China has grown substantially","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Whatnextfortheglobalcarindustry_p22.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Whatnextfortheglobalcarindustry_p22.pdf"},"request":{"example_id":"chart/Whatnextfortheglobalcarindustry_p22","product_type":"parse","source_file_path":"Whatnextfortheglobalcarindustry_p22.pdf"},"started_at":"2026-11-04T19:45:07.460539"} +{"completed_at":"2026-11-04T19:45:12.278616","latency_in_ms":3119,"output":{"example_id":"chart/Whatnextfortheglobalcarindustry_p31","grounded_pages":[],"layout_pages":[],"markdown":"What Next for the Global Car Industry? Chapter 1: The global car industry An Energy Technology Perspectives Special Report in context\n\n31\n\nI EA. CC BY 4.0.\n\ndomestically. An important reason for this is that Japanese automakers were technologically superior: they had developed “lean manufacturing” techniques which guaranteed lower costs and higher quality compared to the older manufacturing systems employed by US carmakers at the time.\n\nThe affordability and superior energy efficiency led to a surge in imports of cars from Japan. US carmakers were slow to adapt their vehicle offering and technology to cater for the higher oil price environment, leading to losses of over USD 6 billion across the Big Three by 1980. In 1980, more than one in five cars sold in the United States was imported from Japan, and Japan accounted for nearly two-thirds of imported cars. In response, the US government negotiated a “voluntary export restriction” with the Japanese government, effectively setting a quota of cars that could be imported to the United States. This measure remained in place until 1994, and coincided with a real term increase in the price of new cars in the United States.\n\nJapanese carmakers have built manufacturing capacity in the United States since the 1980s. The first Honda factory began operations in 1982, and, in 1984, the landmark NUMMI plant – a joint venture between GM and Toyota – was set up. The investments of Japanese OEMs led to technology transfer, as US carmakers started to adopt some of the new manufacturing techniques pioneered by Japanese carmakers, and improved their productivity and quality over the following decade. The fuel economy of domestically made cars also improved by over 10% from 1980 to 1990, closing the gap with imported cars. To this day, Japanese OEMs account for around one-third of cars produced in the United States.\n\n**Japanese car imports in the United States, oil price and political milestones, 1964-1984**\n\nNotes: JV = joint venture. WTI = West Texas Intermediate. Real oil prices are derived using the Consumer Price Index (CPI). Sources: IEA analysis based on USITC (1985), FRED (2025).\n\n0\n\n20\n\n40\n\n60\n\n80\n\n100\n\n120\n\n140\n\n160\n\n0%\n\n5%\n\n10%\n\n15%\n\n20%\n\n25%\n\n1964 1966 1968 1970 1972 1974 1976 1978 1980 1982 1984\n\nUSD (2024)/bbl\n\nJapanese imports as share of demand WTI price\n\n10% Emergency Import Surcharge\n\nVoluntary Export\n\nRestrictions\n\nHonda Factory\n\nToyota-GM JV","pages":[{"line_numbers":[],"markdown":"What Next for the Global Car Industry? Chapter 1: The global car industry An Energy Technology Perspectives Special Report in context\n\n31\n\nI EA. CC BY 4.0.\n\ndomestically. An important reason for this is that Japanese automakers were technologically superior: they had developed “lean manufacturing” techniques which guaranteed lower costs and higher quality compared to the older manufacturing systems employed by US carmakers at the time.\n\nThe affordability and superior energy efficiency led to a surge in imports of cars from Japan. US carmakers were slow to adapt their vehicle offering and technology to cater for the higher oil price environment, leading to losses of over USD 6 billion across the Big Three by 1980. In 1980, more than one in five cars sold in the United States was imported from Japan, and Japan accounted for nearly two-thirds of imported cars. In response, the US government negotiated a “voluntary export restriction” with the Japanese government, effectively setting a quota of cars that could be imported to the United States. This measure remained in place until 1994, and coincided with a real term increase in the price of new cars in the United States.\n\nJapanese carmakers have built manufacturing capacity in the United States since the 1980s. The first Honda factory began operations in 1982, and, in 1984, the landmark NUMMI plant – a joint venture between GM and Toyota – was set up. The investments of Japanese OEMs led to technology transfer, as US carmakers started to adopt some of the new manufacturing techniques pioneered by Japanese carmakers, and improved their productivity and quality over the following decade. The fuel economy of domestically made cars also improved by over 10% from 1980 to 1990, closing the gap with imported cars. To this day, Japanese OEMs account for around one-third of cars produced in the United States.\n\n**Japanese car imports in the United States, oil price and political milestones, 1964-1984**\n\nNotes: JV = joint venture. WTI = West Texas Intermediate. Real oil prices are derived using the Consumer Price Index (CPI). Sources: IEA analysis based on USITC (1985), FRED (2025).\n\n0\n\n20\n\n40\n\n60\n\n80\n\n100\n\n120\n\n140\n\n160\n\n0%\n\n5%\n\n10%\n\n15%\n\n20%\n\n25%\n\n1964 1966 1968 1970 1972 1974 1976 1978 1980 1982 1984\n\nUSD (2024)/bbl\n\nJapanese imports as share of demand WTI price\n\n10% Emergency Import Surcharge\n\nVoluntary Export\n\nRestrictions\n\nHonda Factory\n\nToyota-GM JV","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"What Next for the Global Car Industry? Chapter 1: The global car industry An Energy Technology Perspectives Special Report in context\n\n31\n\nI EA. CC BY 4.0.\n\ndomestically. An important reason for this is that Japanese automakers were technologically superior: they had developed “lean manufacturing” techniques which guaranteed lower costs and higher quality compared to the older manufacturing systems employed by US carmakers at the time.\n\nThe affordability and superior energy efficiency led to a surge in imports of cars from Japan. US carmakers were slow to adapt their vehicle offering and technology to cater for the higher oil price environment, leading to losses of over USD 6 billion across the Big Three by 1980. In 1980, more than one in five cars sold in the United States was imported from Japan, and Japan accounted for nearly two-thirds of imported cars. In response, the US government negotiated a “voluntary export restriction” with the Japanese government, effectively setting a quota of cars that could be imported to the United States. This measure remained in place until 1994, and coincided with a real term increase in the price of new cars in the United States.\n\nJapanese carmakers have built manufacturing capacity in the United States since the 1980s. The first Honda factory began operations in 1982, and, in 1984, the landmark NUMMI plant – a joint venture between GM and Toyota – was set up. The investments of Japanese OEMs led to technology transfer, as US carmakers started to adopt some of the new manufacturing techniques pioneered by Japanese carmakers, and improved their productivity and quality over the following decade. The fuel economy of domestically made cars also improved by over 10% from 1980 to 1990, closing the gap with imported cars. To this day, Japanese OEMs account for around one-third of cars produced in the United States.\n\n**Japanese car imports in the United States, oil price and political milestones, 1964-1984**\n\nNotes: JV = joint venture. WTI = West Texas Intermediate. Real oil prices are derived using the Consumer Price Index (CPI). Sources: IEA analysis based on USITC (1985), FRED (2025).\n\n0\n\n20\n\n40\n\n60\n\n80\n\n100\n\n120\n\n140\n\n160\n\n0%\n\n5%\n\n10%\n\n15%\n\n20%\n\n25%\n\n1964 1966 1968 1970 1972 1974 1976 1978 1980 1982 1984\n\nUSD (2024)/bbl\n\nJapanese imports as share of demand WTI price\n\n10% Emergency Import Surcharge\n\nVoluntary Export\n\nRestrictions\n\nHonda Factory\n\nToyota-GM JV","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Whatnextfortheglobalcarindustry_p31.pdf"]},"response":{"errors":[],"extract_id":"extract_712fd13af9446e824bfe94bb179c8fd7","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["What Next for the Global Car Industry? Chapter 1: The global car industry An Energy Technology Perspectives Special Report in context\n\n31\n\nI EA. CC BY 4.0.\n\ndomestically. An important reason for this is that Japanese automakers were technologically superior: they had developed “lean manufacturing” techniques which guaranteed lower costs and higher quality compared to the older manufacturing systems employed by US carmakers at the time.\n\nThe affordability and superior energy efficiency led to a surge in imports of cars from Japan. US carmakers were slow to adapt their vehicle offering and technology to cater for the higher oil price environment, leading to losses of over USD 6 billion across the Big Three by 1980. In 1980, more than one in five cars sold in the United States was imported from Japan, and Japan accounted for nearly two-thirds of imported cars. In response, the US government negotiated a “voluntary export restriction” with the Japanese government, effectively setting a quota of cars that could be imported to the United States. This measure remained in place until 1994, and coincided with a real term increase in the price of new cars in the United States.\n\nJapanese carmakers have built manufacturing capacity in the United States since the 1980s. The first Honda factory began operations in 1982, and, in 1984, the landmark NUMMI plant – a joint venture between GM and Toyota – was set up. The investments of Japanese OEMs led to technology transfer, as US carmakers started to adopt some of the new manufacturing techniques pioneered by Japanese carmakers, and improved their productivity and quality over the following decade. The fuel economy of domestically made cars also improved by over 10% from 1980 to 1990, closing the gap with imported cars. To this day, Japanese OEMs account for around one-third of cars produced in the United States.\n\n**Japanese car imports in the United States, oil price and political milestones, 1964-1984**\n\nNotes: JV = joint venture. WTI = West Texas Intermediate. Real oil prices are derived using the Consumer Price Index (CPI). Sources: IEA analysis based on USITC (1985), FRED (2025).\n\n0\n\n20\n\n40\n\n60\n\n80\n\n100\n\n120\n\n140\n\n160\n\n0%\n\n5%\n\n10%\n\n15%\n\n20%\n\n25%\n\n1964 1966 1968 1970 1972 1974 1976 1978 1980 1982 1984\n\nUSD (2024)/bbl\n\nJapanese imports as share of demand WTI price\n\n10% Emergency Import Surcharge\n\nVoluntary Export\n\nRestrictions\n\nHonda Factory\n\nToyota-GM JV"],"full_content":"What Next for the Global Car Industry? Chapter 1: The global car industry An Energy Technology Perspectives Special Report in context\n\n31\n\nI EA. CC BY 4.0.\n\ndomestically. An important reason for this is that Japanese automakers were technologically superior: they had developed “lean manufacturing” techniques which guaranteed lower costs and higher quality compared to the older manufacturing systems employed by US carmakers at the time.\n\nThe affordability and superior energy efficiency led to a surge in imports of cars from Japan. US carmakers were slow to adapt their vehicle offering and technology to cater for the higher oil price environment, leading to losses of over USD 6 billion across the Big Three by 1980. In 1980, more than one in five cars sold in the United States was imported from Japan, and Japan accounted for nearly two-thirds of imported cars. In response, the US government negotiated a “voluntary export restriction” with the Japanese government, effectively setting a quota of cars that could be imported to the United States. This measure remained in place until 1994, and coincided with a real term increase in the price of new cars in the United States.\n\nJapanese carmakers have built manufacturing capacity in the United States since the 1980s. The first Honda factory began operations in 1982, and, in 1984, the landmark NUMMI plant – a joint venture between GM and Toyota – was set up. The investments of Japanese OEMs led to technology transfer, as US carmakers started to adopt some of the new manufacturing techniques pioneered by Japanese carmakers, and improved their productivity and quality over the following decade. The fuel economy of domestically made cars also improved by over 10% from 1980 to 1990, closing the gap with imported cars. To this day, Japanese OEMs account for around one-third of cars produced in the United States.\n\n**Japanese car imports in the United States, oil price and political milestones, 1964-1984**\n\nNotes: JV = joint venture. WTI = West Texas Intermediate. Real oil prices are derived using the Consumer Price Index (CPI). Sources: IEA analysis based on USITC (1985), FRED (2025).\n\n0\n\n20\n\n40\n\n60\n\n80\n\n100\n\n120\n\n140\n\n160\n\n0%\n\n5%\n\n10%\n\n15%\n\n20%\n\n25%\n\n1964 1966 1968 1970 1972 1974 1976 1978 1980 1982 1984\n\nUSD (2024)/bbl\n\nJapanese imports as share of demand WTI price\n\n10% Emergency Import Surcharge\n\nVoluntary Export\n\nRestrictions\n\nHonda Factory\n\nToyota-GM JV","publish_date":null,"title":"domestically. An important reason for this is that Japanese automakers were technologically superior: they had developed “lean manufacturing” techniques which","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Whatnextfortheglobalcarindustry_p31.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Whatnextfortheglobalcarindustry_p31.pdf"},"request":{"example_id":"chart/Whatnextfortheglobalcarindustry_p31","product_type":"parse","source_file_path":"Whatnextfortheglobalcarindustry_p31.pdf"},"started_at":"2026-11-04T19:45:09.159236"} +{"completed_at":"2026-11-04T19:45:14.131892","latency_in_ms":3302,"output":{"example_id":"chart/Whatnextfortheglobalcarindustry_p36","grounded_pages":[],"layout_pages":[],"markdown":"What Next for the Global Car Industry? Chapter 1: The global car industry An Energy Technology Perspectives Special Report in context\n\n36\n\nI EA. CC BY 4.0.\n\n**Figure 1.10 Revenues and profits for global original equipment manufacturers, 2016- 2024**\n\nIEA. CC BY 4.0.\n\nNotes: OEM = original equipment manufacturer. Data taken from a sample of the 26 largest automakers with revenues accounting for nearly three-quarters of global sales. China: BYD, Changan, Dongfeng, GAC, Geely, Great Wall, Leapmotor, Li Auto, SAIC, Seres Group. North America: GM, Ford. Europe: BMW Group, Mercedes-Benz, Renault-Nissan Alliance, Stellantis, VW Group. Japan: Honda, Mazda, Mitsubishi, Subaru, Suzuki, Toyota. Other: Tata Group (India), Hyundai (Korea). Chery and FAW (China) are excluded since these are state-owned and there is no financial data reported after 2020. The sales from these companies account for around 80% of the global car market. Numbers are adjusted for inflation to 2024. Source: IEA analysis based on Bloomberg Terminal.\n\nAutomotive suppliers tend to have profit margins that are lower than those experienced by OEMs in advanced economies, partly as a result of strong competition. Chinese automotive suppliers are nearly all related to battery manufacturing and have been leaders in terms of net profit margins, with several companies reaching margins of over 20% in the past 10 years. ⁵ The average profitability of Chinese suppliers has been lower, at around 6-7% for most years since 2016, although this trend has been broken in the past 2 years to reach more than 10% in 2024. In addition, while North American and European suppliers have not recovered their pre-pandemic profitability, the average profitability of Chinese suppliers has grown, led by CATL, the world’s largest battery manufacturer. CATL has experienced unprecedented growth since it first started generating revenues in 2015, becoming a market leader in 2024. The year 2024 was the first in which CATL experienced a decline in total revenues – driven by declining battery prices – after experiencing sixfold growth between 2020 and 2022. CATL’s net profit margin has been around or above 10% ever since 2015. This is very high compared to other automotive component suppliers: since 2020, the average profitability of the five largest diversified suppliers by revenue has been around 2.5%. In contrast, among specialised companies (such as those supplying\n\n5 CATL in 2016-2017, Gotion High-tech in 2014-2016 and Eve Energy in 2019-2020.\n\n0\n\n500\n\n1 000\n\n1 500\n\n2 000\n\n2 500\n\n3 000\n\n2016 2018 2020 2022 2024\n\nBillion USD (2024, MER)\n\n**Global OEMs' revenues per region**\n\nEurope North America Japan China Other Total (nominal)\n\n0\n\n50\n\n100\n\n150\n\n200\n\n250\n\n300\n\n2016 2018 2020 2022 2024\n\n**Global OEMs' profits per region**","pages":[{"line_numbers":[],"markdown":"What Next for the Global Car Industry? Chapter 1: The global car industry An Energy Technology Perspectives Special Report in context\n\n36\n\nI EA. CC BY 4.0.\n\n**Figure 1.10 Revenues and profits for global original equipment manufacturers, 2016- 2024**\n\nIEA. CC BY 4.0.\n\nNotes: OEM = original equipment manufacturer. Data taken from a sample of the 26 largest automakers with revenues accounting for nearly three-quarters of global sales. China: BYD, Changan, Dongfeng, GAC, Geely, Great Wall, Leapmotor, Li Auto, SAIC, Seres Group. North America: GM, Ford. Europe: BMW Group, Mercedes-Benz, Renault-Nissan Alliance, Stellantis, VW Group. Japan: Honda, Mazda, Mitsubishi, Subaru, Suzuki, Toyota. Other: Tata Group (India), Hyundai (Korea). Chery and FAW (China) are excluded since these are state-owned and there is no financial data reported after 2020. The sales from these companies account for around 80% of the global car market. Numbers are adjusted for inflation to 2024. Source: IEA analysis based on Bloomberg Terminal.\n\nAutomotive suppliers tend to have profit margins that are lower than those experienced by OEMs in advanced economies, partly as a result of strong competition. Chinese automotive suppliers are nearly all related to battery manufacturing and have been leaders in terms of net profit margins, with several companies reaching margins of over 20% in the past 10 years. ⁵ The average profitability of Chinese suppliers has been lower, at around 6-7% for most years since 2016, although this trend has been broken in the past 2 years to reach more than 10% in 2024. In addition, while North American and European suppliers have not recovered their pre-pandemic profitability, the average profitability of Chinese suppliers has grown, led by CATL, the world’s largest battery manufacturer. CATL has experienced unprecedented growth since it first started generating revenues in 2015, becoming a market leader in 2024. The year 2024 was the first in which CATL experienced a decline in total revenues – driven by declining battery prices – after experiencing sixfold growth between 2020 and 2022. CATL’s net profit margin has been around or above 10% ever since 2015. This is very high compared to other automotive component suppliers: since 2020, the average profitability of the five largest diversified suppliers by revenue has been around 2.5%. In contrast, among specialised companies (such as those supplying\n\n5 CATL in 2016-2017, Gotion High-tech in 2014-2016 and Eve Energy in 2019-2020.\n\n0\n\n500\n\n1 000\n\n1 500\n\n2 000\n\n2 500\n\n3 000\n\n2016 2018 2020 2022 2024\n\nBillion USD (2024, MER)\n\n**Global OEMs' revenues per region**\n\nEurope North America Japan China Other Total (nominal)\n\n0\n\n50\n\n100\n\n150\n\n200\n\n250\n\n300\n\n2016 2018 2020 2022 2024\n\n**Global OEMs' profits per region**","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"What Next for the Global Car Industry? Chapter 1: The global car industry An Energy Technology Perspectives Special Report in context\n\n36\n\nI EA. CC BY 4.0.\n\n**Figure 1.10 Revenues and profits for global original equipment manufacturers, 2016- 2024**\n\nIEA. CC BY 4.0.\n\nNotes: OEM = original equipment manufacturer. Data taken from a sample of the 26 largest automakers with revenues accounting for nearly three-quarters of global sales. China: BYD, Changan, Dongfeng, GAC, Geely, Great Wall, Leapmotor, Li Auto, SAIC, Seres Group. North America: GM, Ford. Europe: BMW Group, Mercedes-Benz, Renault-Nissan Alliance, Stellantis, VW Group. Japan: Honda, Mazda, Mitsubishi, Subaru, Suzuki, Toyota. Other: Tata Group (India), Hyundai (Korea). Chery and FAW (China) are excluded since these are state-owned and there is no financial data reported after 2020. The sales from these companies account for around 80% of the global car market. Numbers are adjusted for inflation to 2024. Source: IEA analysis based on Bloomberg Terminal.\n\nAutomotive suppliers tend to have profit margins that are lower than those experienced by OEMs in advanced economies, partly as a result of strong competition. Chinese automotive suppliers are nearly all related to battery manufacturing and have been leaders in terms of net profit margins, with several companies reaching margins of over 20% in the past 10 years. ⁵ The average profitability of Chinese suppliers has been lower, at around 6-7% for most years since 2016, although this trend has been broken in the past 2 years to reach more than 10% in 2024. In addition, while North American and European suppliers have not recovered their pre-pandemic profitability, the average profitability of Chinese suppliers has grown, led by CATL, the world’s largest battery manufacturer. CATL has experienced unprecedented growth since it first started generating revenues in 2015, becoming a market leader in 2024. The year 2024 was the first in which CATL experienced a decline in total revenues – driven by declining battery prices – after experiencing sixfold growth between 2020 and 2022. CATL’s net profit margin has been around or above 10% ever since 2015. This is very high compared to other automotive component suppliers: since 2020, the average profitability of the five largest diversified suppliers by revenue has been around 2.5%. In contrast, among specialised companies (such as those supplying\n\n5 CATL in 2016-2017, Gotion High-tech in 2014-2016 and Eve Energy in 2019-2020.\n\n0\n\n500\n\n1 000\n\n1 500\n\n2 000\n\n2 500\n\n3 000\n\n2016 2018 2020 2022 2024\n\nBillion USD (2024, MER)\n\n**Global OEMs' revenues per region**\n\nEurope North America Japan China Other Total (nominal)\n\n0\n\n50\n\n100\n\n150\n\n200\n\n250\n\n300\n\n2016 2018 2020 2022 2024\n\n**Global OEMs' profits per region**","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Whatnextfortheglobalcarindustry_p36.pdf"]},"response":{"errors":[],"extract_id":"extract_57744152254d6f9c4df4b6e43c196bd6","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["What Next for the Global Car Industry? Chapter 1: The global car industry An Energy Technology Perspectives Special Report in context\n\n36\n\nI EA. CC BY 4.0.\n\n**Figure 1.10 Revenues and profits for global original equipment manufacturers, 2016- 2024**\n\nIEA. CC BY 4.0.\n\nNotes: OEM = original equipment manufacturer. Data taken from a sample of the 26 largest automakers with revenues accounting for nearly three-quarters of global sales. China: BYD, Changan, Dongfeng, GAC, Geely, Great Wall, Leapmotor, Li Auto, SAIC, Seres Group. North America: GM, Ford. Europe: BMW Group, Mercedes-Benz, Renault-Nissan Alliance, Stellantis, VW Group. Japan: Honda, Mazda, Mitsubishi, Subaru, Suzuki, Toyota. Other: Tata Group (India), Hyundai (Korea). Chery and FAW (China) are excluded since these are state-owned and there is no financial data reported after 2020. The sales from these companies account for around 80% of the global car market. Numbers are adjusted for inflation to 2024. Source: IEA analysis based on Bloomberg Terminal.\n\nAutomotive suppliers tend to have profit margins that are lower than those experienced by OEMs in advanced economies, partly as a result of strong competition. Chinese automotive suppliers are nearly all related to battery manufacturing and have been leaders in terms of net profit margins, with several companies reaching margins of over 20% in the past 10 years. ⁵ The average profitability of Chinese suppliers has been lower, at around 6-7% for most years since 2016, although this trend has been broken in the past 2 years to reach more than 10% in 2024. In addition, while North American and European suppliers have not recovered their pre-pandemic profitability, the average profitability of Chinese suppliers has grown, led by CATL, the world’s largest battery manufacturer. CATL has experienced unprecedented growth since it first started generating revenues in 2015, becoming a market leader in 2024. The year 2024 was the first in which CATL experienced a decline in total revenues – driven by declining battery prices – after experiencing sixfold growth between 2020 and 2022. CATL’s net profit margin has been around or above 10% ever since 2015. This is very high compared to other automotive component suppliers: since 2020, the average profitability of the five largest diversified suppliers by revenue has been around 2.5%. In contrast, among specialised companies (such as those supplying\n\n5 CATL in 2016-2017, Gotion High-tech in 2014-2016 and Eve Energy in 2019-2020.\n\n0\n\n500\n\n1 000\n\n1 500\n\n2 000\n\n2 500\n\n3 000\n\n2016 2018 2020 2022 2024\n\nBillion USD (2024, MER)\n\n**Global OEMs' revenues per region**\n\nEurope North America Japan China Other Total (nominal)\n\n0\n\n50\n\n100\n\n150\n\n200\n\n250\n\n300\n\n2016 2018 2020 2022 2024\n\n**Global OEMs' profits per region**"],"full_content":"What Next for the Global Car Industry? Chapter 1: The global car industry An Energy Technology Perspectives Special Report in context\n\n36\n\nI EA. CC BY 4.0.\n\n**Figure 1.10 Revenues and profits for global original equipment manufacturers, 2016- 2024**\n\nIEA. CC BY 4.0.\n\nNotes: OEM = original equipment manufacturer. Data taken from a sample of the 26 largest automakers with revenues accounting for nearly three-quarters of global sales. China: BYD, Changan, Dongfeng, GAC, Geely, Great Wall, Leapmotor, Li Auto, SAIC, Seres Group. North America: GM, Ford. Europe: BMW Group, Mercedes-Benz, Renault-Nissan Alliance, Stellantis, VW Group. Japan: Honda, Mazda, Mitsubishi, Subaru, Suzuki, Toyota. Other: Tata Group (India), Hyundai (Korea). Chery and FAW (China) are excluded since these are state-owned and there is no financial data reported after 2020. The sales from these companies account for around 80% of the global car market. Numbers are adjusted for inflation to 2024. Source: IEA analysis based on Bloomberg Terminal.\n\nAutomotive suppliers tend to have profit margins that are lower than those experienced by OEMs in advanced economies, partly as a result of strong competition. Chinese automotive suppliers are nearly all related to battery manufacturing and have been leaders in terms of net profit margins, with several companies reaching margins of over 20% in the past 10 years. ⁵ The average profitability of Chinese suppliers has been lower, at around 6-7% for most years since 2016, although this trend has been broken in the past 2 years to reach more than 10% in 2024. In addition, while North American and European suppliers have not recovered their pre-pandemic profitability, the average profitability of Chinese suppliers has grown, led by CATL, the world’s largest battery manufacturer. CATL has experienced unprecedented growth since it first started generating revenues in 2015, becoming a market leader in 2024. The year 2024 was the first in which CATL experienced a decline in total revenues – driven by declining battery prices – after experiencing sixfold growth between 2020 and 2022. CATL’s net profit margin has been around or above 10% ever since 2015. This is very high compared to other automotive component suppliers: since 2020, the average profitability of the five largest diversified suppliers by revenue has been around 2.5%. In contrast, among specialised companies (such as those supplying\n\n5 CATL in 2016-2017, Gotion High-tech in 2014-2016 and Eve Energy in 2019-2020.\n\n0\n\n500\n\n1 000\n\n1 500\n\n2 000\n\n2 500\n\n3 000\n\n2016 2018 2020 2022 2024\n\nBillion USD (2024, MER)\n\n**Global OEMs' revenues per region**\n\nEurope North America Japan China Other Total (nominal)\n\n0\n\n50\n\n100\n\n150\n\n200\n\n250\n\n300\n\n2016 2018 2020 2022 2024\n\n**Global OEMs' profits per region**","publish_date":null,"title":"Automotive suppliers tend to have profit margins that are lower than those experienced by OEMs in advanced economies, partly as","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Whatnextfortheglobalcarindustry_p36.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/Whatnextfortheglobalcarindustry_p36.pdf"},"request":{"example_id":"chart/Whatnextfortheglobalcarindustry_p36","product_type":"parse","source_file_path":"Whatnextfortheglobalcarindustry_p36.pdf"},"started_at":"2026-11-04T19:45:10.829077"} +{"completed_at":"2026-11-04T19:45:14.765939","latency_in_ms":2485,"output":{"example_id":"chart/World_Inequality_Report_2026_p105","grounded_pages":[],"layout_pages":[],"markdown":"Chapter 5. Exorbitant Privilege\n\n**Figure 5.1. The U.S. exorbitant privilege has evolved into a structural privilege of the rich world**\n\nThe centrality of the U.S. dollar, and now other currencies, in the global financial system allows rich economies to borrow at very low rates.\n\n−5% −4% −3% −2% −1%\n\n0% 1% 2% 3% 4% 5% 6% 7%\n\n1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 2020 2025\n\nExcess yield as %\n\nof country GDP, MER\n\nBRICS\n\nChina\n\nEurozone\n\nJapan\n\nRussia\n\nUnited States\n\n**Excess yield (assets−liabilities) as % of country GDP, 1970−2025**\n\n**Interpretation.** This graph shows excess yield income, defined as the difference between the return on foreign assets and liabilities, as a share of national GDP. The figure shows that the exorbitant privilege once exclusive to the United States has become a broader rich−world phenomenon. The United States maintains a substantial privilege of 2.2% in 2025. The Eurozone follows with 1% by 2025. Japan stands out with a privilege of 5.9% by 2025. In contrast, BRICS countries face a consistent burden of around 2.1%, highlighting their role as net providers of capital to wealthier economies. **Notes.** Positive values represent income gains from financial privilege; negative values represent financial burden. BRICS countries comprise Brazil, Russia, India, China, and South Africa. **Sources and series:** Nievas and Sodano (2025) and wir2026.wid.world/methodology.\n\n20% of countries consistently record positive income flows. Meanwhile, the bottom 80% are persistent net debtors and face negative income balances, reinforcing their disadvantage.\n\n**Rich countries are global financial rentiers by political design, not because of market dynamics**\n\n**Figure 5.5** to **Figure 5.7** reveal why the exorbitant privilege has persisted and extended: the global financial and monetary system has been deliberately structured to favor advanced economies. Their role as issuers of reserve currencies, the composition of their external portfolios, and the cost asymmetries between assets and liabilities combine to make them global financial rentiers.\n\n**Figure 5.5** documents a foundation of this privilege: currency dominance. Over the last few decades, the U.S. dollar has remained the predominant medium of trade invoicing, financial asset denomination, and central bank reserves. The euro has\n\nalso become, to a lesser extent, a major player since its creation. Other currencies play only marginal roles. This institutional inequality ensures persistent demand for dollar- and euro-denominated assets. This leads to persistently lower borrowing costs for the U.S. and Eurozone, whereas other economies are more exposed to debt in foreign currencies and vulnerable to exchange rate fluctuations.\n\n**Figure 5.6** highlights the composition of cross-border portfolios. Rich countries hold equity and foreign direct investment on the asset side, which typically have higher returns, while their liabilities are predominantly low-cost debt securities. Poorer countries show the mirror image: they hold large shares of reserves, safe but low-yielding, while issuing liabilities in the form of high-cost debt and inward foreign direct investment (FDI). This asymmetry means that even when poorer countries save and accumulate foreign assets, those assets generate little return, while their liabilities remain costly.\n\n**Figure 5.7** complements this picture by\n\n105","pages":[{"line_numbers":[],"markdown":"Chapter 5. Exorbitant Privilege\n\n**Figure 5.1. The U.S. exorbitant privilege has evolved into a structural privilege of the rich world**\n\nThe centrality of the U.S. dollar, and now other currencies, in the global financial system allows rich economies to borrow at very low rates.\n\n−5% −4% −3% −2% −1%\n\n0% 1% 2% 3% 4% 5% 6% 7%\n\n1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 2020 2025\n\nExcess yield as %\n\nof country GDP, MER\n\nBRICS\n\nChina\n\nEurozone\n\nJapan\n\nRussia\n\nUnited States\n\n**Excess yield (assets−liabilities) as % of country GDP, 1970−2025**\n\n**Interpretation.** This graph shows excess yield income, defined as the difference between the return on foreign assets and liabilities, as a share of national GDP. The figure shows that the exorbitant privilege once exclusive to the United States has become a broader rich−world phenomenon. The United States maintains a substantial privilege of 2.2% in 2025. The Eurozone follows with 1% by 2025. Japan stands out with a privilege of 5.9% by 2025. In contrast, BRICS countries face a consistent burden of around 2.1%, highlighting their role as net providers of capital to wealthier economies. **Notes.** Positive values represent income gains from financial privilege; negative values represent financial burden. BRICS countries comprise Brazil, Russia, India, China, and South Africa. **Sources and series:** Nievas and Sodano (2025) and wir2026.wid.world/methodology.\n\n20% of countries consistently record positive income flows. Meanwhile, the bottom 80% are persistent net debtors and face negative income balances, reinforcing their disadvantage.\n\n**Rich countries are global financial rentiers by political design, not because of market dynamics**\n\n**Figure 5.5** to **Figure 5.7** reveal why the exorbitant privilege has persisted and extended: the global financial and monetary system has been deliberately structured to favor advanced economies. Their role as issuers of reserve currencies, the composition of their external portfolios, and the cost asymmetries between assets and liabilities combine to make them global financial rentiers.\n\n**Figure 5.5** documents a foundation of this privilege: currency dominance. Over the last few decades, the U.S. dollar has remained the predominant medium of trade invoicing, financial asset denomination, and central bank reserves. The euro has\n\nalso become, to a lesser extent, a major player since its creation. Other currencies play only marginal roles. This institutional inequality ensures persistent demand for dollar- and euro-denominated assets. This leads to persistently lower borrowing costs for the U.S. and Eurozone, whereas other economies are more exposed to debt in foreign currencies and vulnerable to exchange rate fluctuations.\n\n**Figure 5.6** highlights the composition of cross-border portfolios. Rich countries hold equity and foreign direct investment on the asset side, which typically have higher returns, while their liabilities are predominantly low-cost debt securities. Poorer countries show the mirror image: they hold large shares of reserves, safe but low-yielding, while issuing liabilities in the form of high-cost debt and inward foreign direct investment (FDI). This asymmetry means that even when poorer countries save and accumulate foreign assets, those assets generate little return, while their liabilities remain costly.\n\n**Figure 5.7** complements this picture by\n\n105","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Chapter 5. Exorbitant Privilege\n\n**Figure 5.1. The U.S. exorbitant privilege has evolved into a structural privilege of the rich world**\n\nThe centrality of the U.S. dollar, and now other currencies, in the global financial system allows rich economies to borrow at very low rates.\n\n−5% −4% −3% −2% −1%\n\n0% 1% 2% 3% 4% 5% 6% 7%\n\n1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 2020 2025\n\nExcess yield as %\n\nof country GDP, MER\n\nBRICS\n\nChina\n\nEurozone\n\nJapan\n\nRussia\n\nUnited States\n\n**Excess yield (assets−liabilities) as % of country GDP, 1970−2025**\n\n**Interpretation.** This graph shows excess yield income, defined as the difference between the return on foreign assets and liabilities, as a share of national GDP. The figure shows that the exorbitant privilege once exclusive to the United States has become a broader rich−world phenomenon. The United States maintains a substantial privilege of 2.2% in 2025. The Eurozone follows with 1% by 2025. Japan stands out with a privilege of 5.9% by 2025. In contrast, BRICS countries face a consistent burden of around 2.1%, highlighting their role as net providers of capital to wealthier economies. **Notes.** Positive values represent income gains from financial privilege; negative values represent financial burden. BRICS countries comprise Brazil, Russia, India, China, and South Africa. **Sources and series:** Nievas and Sodano (2025) and wir2026.wid.world/methodology.\n\n20% of countries consistently record positive income flows. Meanwhile, the bottom 80% are persistent net debtors and face negative income balances, reinforcing their disadvantage.\n\n**Rich countries are global financial rentiers by political design, not because of market dynamics**\n\n**Figure 5.5** to **Figure 5.7** reveal why the exorbitant privilege has persisted and extended: the global financial and monetary system has been deliberately structured to favor advanced economies. Their role as issuers of reserve currencies, the composition of their external portfolios, and the cost asymmetries between assets and liabilities combine to make them global financial rentiers.\n\n**Figure 5.5** documents a foundation of this privilege: currency dominance. Over the last few decades, the U.S. dollar has remained the predominant medium of trade invoicing, financial asset denomination, and central bank reserves. The euro has\n\nalso become, to a lesser extent, a major player since its creation. Other currencies play only marginal roles. This institutional inequality ensures persistent demand for dollar- and euro-denominated assets. This leads to persistently lower borrowing costs for the U.S. and Eurozone, whereas other economies are more exposed to debt in foreign currencies and vulnerable to exchange rate fluctuations.\n\n**Figure 5.6** highlights the composition of cross-border portfolios. Rich countries hold equity and foreign direct investment on the asset side, which typically have higher returns, while their liabilities are predominantly low-cost debt securities. Poorer countries show the mirror image: they hold large shares of reserves, safe but low-yielding, while issuing liabilities in the form of high-cost debt and inward foreign direct investment (FDI). This asymmetry means that even when poorer countries save and accumulate foreign assets, those assets generate little return, while their liabilities remain costly.\n\n**Figure 5.7** complements this picture by\n\n105","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p105.pdf"]},"response":{"errors":[],"extract_id":"extract_b413af071a1357e1663535327ef41e6f","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Chapter 5. Exorbitant Privilege\n\n**Figure 5.1. The U.S. exorbitant privilege has evolved into a structural privilege of the rich world**\n\nThe centrality of the U.S. dollar, and now other currencies, in the global financial system allows rich economies to borrow at very low rates.\n\n−5% −4% −3% −2% −1%\n\n0% 1% 2% 3% 4% 5% 6% 7%\n\n1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 2020 2025\n\nExcess yield as %\n\nof country GDP, MER\n\nBRICS\n\nChina\n\nEurozone\n\nJapan\n\nRussia\n\nUnited States\n\n**Excess yield (assets−liabilities) as % of country GDP, 1970−2025**\n\n**Interpretation.** This graph shows excess yield income, defined as the difference between the return on foreign assets and liabilities, as a share of national GDP. The figure shows that the exorbitant privilege once exclusive to the United States has become a broader rich−world phenomenon. The United States maintains a substantial privilege of 2.2% in 2025. The Eurozone follows with 1% by 2025. Japan stands out with a privilege of 5.9% by 2025. In contrast, BRICS countries face a consistent burden of around 2.1%, highlighting their role as net providers of capital to wealthier economies. **Notes.** Positive values represent income gains from financial privilege; negative values represent financial burden. BRICS countries comprise Brazil, Russia, India, China, and South Africa. **Sources and series:** Nievas and Sodano (2025) and wir2026.wid.world/methodology.\n\n20% of countries consistently record positive income flows. Meanwhile, the bottom 80% are persistent net debtors and face negative income balances, reinforcing their disadvantage.\n\n**Rich countries are global financial rentiers by political design, not because of market dynamics**\n\n**Figure 5.5** to **Figure 5.7** reveal why the exorbitant privilege has persisted and extended: the global financial and monetary system has been deliberately structured to favor advanced economies. Their role as issuers of reserve currencies, the composition of their external portfolios, and the cost asymmetries between assets and liabilities combine to make them global financial rentiers.\n\n**Figure 5.5** documents a foundation of this privilege: currency dominance. Over the last few decades, the U.S. dollar has remained the predominant medium of trade invoicing, financial asset denomination, and central bank reserves. The euro has\n\nalso become, to a lesser extent, a major player since its creation. Other currencies play only marginal roles. This institutional inequality ensures persistent demand for dollar- and euro-denominated assets. This leads to persistently lower borrowing costs for the U.S. and Eurozone, whereas other economies are more exposed to debt in foreign currencies and vulnerable to exchange rate fluctuations.\n\n**Figure 5.6** highlights the composition of cross-border portfolios. Rich countries hold equity and foreign direct investment on the asset side, which typically have higher returns, while their liabilities are predominantly low-cost debt securities. Poorer countries show the mirror image: they hold large shares of reserves, safe but low-yielding, while issuing liabilities in the form of high-cost debt and inward foreign direct investment (FDI). This asymmetry means that even when poorer countries save and accumulate foreign assets, those assets generate little return, while their liabilities remain costly.\n\n**Figure 5.7** complements this picture by\n\n105"],"full_content":"Chapter 5. Exorbitant Privilege\n\n**Figure 5.1. The U.S. exorbitant privilege has evolved into a structural privilege of the rich world**\n\nThe centrality of the U.S. dollar, and now other currencies, in the global financial system allows rich economies to borrow at very low rates.\n\n−5% −4% −3% −2% −1%\n\n0% 1% 2% 3% 4% 5% 6% 7%\n\n1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 2020 2025\n\nExcess yield as %\n\nof country GDP, MER\n\nBRICS\n\nChina\n\nEurozone\n\nJapan\n\nRussia\n\nUnited States\n\n**Excess yield (assets−liabilities) as % of country GDP, 1970−2025**\n\n**Interpretation.** This graph shows excess yield income, defined as the difference between the return on foreign assets and liabilities, as a share of national GDP. The figure shows that the exorbitant privilege once exclusive to the United States has become a broader rich−world phenomenon. The United States maintains a substantial privilege of 2.2% in 2025. The Eurozone follows with 1% by 2025. Japan stands out with a privilege of 5.9% by 2025. In contrast, BRICS countries face a consistent burden of around 2.1%, highlighting their role as net providers of capital to wealthier economies. **Notes.** Positive values represent income gains from financial privilege; negative values represent financial burden. BRICS countries comprise Brazil, Russia, India, China, and South Africa. **Sources and series:** Nievas and Sodano (2025) and wir2026.wid.world/methodology.\n\n20% of countries consistently record positive income flows. Meanwhile, the bottom 80% are persistent net debtors and face negative income balances, reinforcing their disadvantage.\n\n**Rich countries are global financial rentiers by political design, not because of market dynamics**\n\n**Figure 5.5** to **Figure 5.7** reveal why the exorbitant privilege has persisted and extended: the global financial and monetary system has been deliberately structured to favor advanced economies. Their role as issuers of reserve currencies, the composition of their external portfolios, and the cost asymmetries between assets and liabilities combine to make them global financial rentiers.\n\n**Figure 5.5** documents a foundation of this privilege: currency dominance. Over the last few decades, the U.S. dollar has remained the predominant medium of trade invoicing, financial asset denomination, and central bank reserves. The euro has\n\nalso become, to a lesser extent, a major player since its creation. Other currencies play only marginal roles. This institutional inequality ensures persistent demand for dollar- and euro-denominated assets. This leads to persistently lower borrowing costs for the U.S. and Eurozone, whereas other economies are more exposed to debt in foreign currencies and vulnerable to exchange rate fluctuations.\n\n**Figure 5.6** highlights the composition of cross-border portfolios. Rich countries hold equity and foreign direct investment on the asset side, which typically have higher returns, while their liabilities are predominantly low-cost debt securities. Poorer countries show the mirror image: they hold large shares of reserves, safe but low-yielding, while issuing liabilities in the form of high-cost debt and inward foreign direct investment (FDI). This asymmetry means that even when poorer countries save and accumulate foreign assets, those assets generate little return, while their liabilities remain costly.\n\n**Figure 5.7** complements this picture by\n\n105","publish_date":null,"title":"Chapter 5. Exorbitant Privilege Figure5.1. TheU.S.exorbitantprivilegehasevolvedintoa structuralprivilegeoftherichworld 20% of countries consistently record also become, to a lesser extent, a major","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p105.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p105.pdf"},"request":{"example_id":"chart/World_Inequality_Report_2026_p105","product_type":"parse","source_file_path":"World_Inequality_Report_2026_p105.pdf"},"started_at":"2026-11-04T19:45:12.280724"} +{"completed_at":"2026-11-04T19:45:15.891804","latency_in_ms":2354,"output":{"example_id":"chart/World_Inequality_Report_2026_p106","grounded_pages":[],"layout_pages":[],"markdown":"Chapter 5. Exorbitant Privilege\n\n**Figure 5.2. Rich countries receive 1% of their GDP from the rest of the world due to financial privilege**\n\n−6% −5% −4% −3% −2% −1%\n\n0% 1% 2% 3%\n\n1980 1990 2000 2010 2020\n\nExcess yield\n\n(% of group GDP), MER\n\nBottom 20%\n\n20%−40%\n\n40%−60%\n\n60%−80%\n\nTop 20%\n\n**By income group**\n\n−6% −5% −4% −3% −2% −1%\n\n0% 1% 2% 3%\n\n1980 1990 2000 2010 2020\n\nChina\n\nEASA (excl. China)\n\nEurope\n\nLATA\n\nMENA\n\nNAOC\n\nRUCA\n\nSSAF\n\nSSEA\n\n**By world region**\n\n**Interpretation.** These two panels show excess yield income (privilege), defined as the difference between the return on foreign assets and liabilities, as a share of national GDP across different country groupings. The left panel presents data by per capita income quintiles. It shows that only the top 20% richest countries enjoy consistently positive excess yield income (1% of their combined GDP by 2025). This privilege stems from the centrality of these rich countries in the monetary and financial system. The right panel shows this pattern by world regions. Financial privilege is overwhelmingly concentrated in East Asia, North America & Oceania, and Europe, while the other regions face consistent financial burden relative to their GDP. **Sources and series:** wir2026.wid.world/methodology and Nievas and Sodano (2025).\n\n**Excess yield (assets−liabilities) as % of group GDP, 1970−2025**\n\ncomparing returns on investments directly. The richest 20% consistently earn more on their assets abroad and pay less on their liabilities. Over the last half-century, global returns on assets have fallen for all, but the decline has been steepest for poorer countries. More importantly, the liability costs have remained high or even increased for the poorer countries. Only the richest 20% have experienced a large decrease in liability costs. The result is a structural advantage for rich countries: they are “charged less” on what they owe. These patterns are not the accidental outcome of market forces. They stem from policy design and institutional dominance. For instance, regulatory standards such as Basel III increased the demand for “safe” assets, consolidating the role of U.S. Treasuries and European sovereign bonds. Credit rating agencies, largely based in advanced economies, reinforce the perception of safety for rich-country debt and risk for poorer-country debt. Central banks worldwide accumulate reserves in dollars and euros, further entrenching the system. The broader implication is that the\n\nrichest economies do not simply benefit from privilege; they actively shape and maintain it. By controlling the currencies, rules, and institutions at the center of global finance, they secure a rentier position that channels income from the rest of the world, thereby exacerbating inequality across countries.\n\n**Barriers for reducing inequality across countries**\n\nThe financial asymmetries documented in this chapter are not only technical imbalances; they translate directly into barriers for development. **Figure 5.8**\n\nillustrates how poorer countries systematically transfer resources to richer ones, constraining their fiscal capacity and long-term growth prospects. The bottom 80% of countries devote a significant share of their GDP to net income outflows, which can be seen as the cost of financing the privilege for the top 20%. These outflows, averaging 2–3% of GDP each year, represent resources that could otherwise be invested in schools, hospitals, or infrastructure. The cost is\n\n106","pages":[{"line_numbers":[],"markdown":"Chapter 5. Exorbitant Privilege\n\n**Figure 5.2. Rich countries receive 1% of their GDP from the rest of the world due to financial privilege**\n\n−6% −5% −4% −3% −2% −1%\n\n0% 1% 2% 3%\n\n1980 1990 2000 2010 2020\n\nExcess yield\n\n(% of group GDP), MER\n\nBottom 20%\n\n20%−40%\n\n40%−60%\n\n60%−80%\n\nTop 20%\n\n**By income group**\n\n−6% −5% −4% −3% −2% −1%\n\n0% 1% 2% 3%\n\n1980 1990 2000 2010 2020\n\nChina\n\nEASA (excl. China)\n\nEurope\n\nLATA\n\nMENA\n\nNAOC\n\nRUCA\n\nSSAF\n\nSSEA\n\n**By world region**\n\n**Interpretation.** These two panels show excess yield income (privilege), defined as the difference between the return on foreign assets and liabilities, as a share of national GDP across different country groupings. The left panel presents data by per capita income quintiles. It shows that only the top 20% richest countries enjoy consistently positive excess yield income (1% of their combined GDP by 2025). This privilege stems from the centrality of these rich countries in the monetary and financial system. The right panel shows this pattern by world regions. Financial privilege is overwhelmingly concentrated in East Asia, North America & Oceania, and Europe, while the other regions face consistent financial burden relative to their GDP. **Sources and series:** wir2026.wid.world/methodology and Nievas and Sodano (2025).\n\n**Excess yield (assets−liabilities) as % of group GDP, 1970−2025**\n\ncomparing returns on investments directly. The richest 20% consistently earn more on their assets abroad and pay less on their liabilities. Over the last half-century, global returns on assets have fallen for all, but the decline has been steepest for poorer countries. More importantly, the liability costs have remained high or even increased for the poorer countries. Only the richest 20% have experienced a large decrease in liability costs. The result is a structural advantage for rich countries: they are “charged less” on what they owe. These patterns are not the accidental outcome of market forces. They stem from policy design and institutional dominance. For instance, regulatory standards such as Basel III increased the demand for “safe” assets, consolidating the role of U.S. Treasuries and European sovereign bonds. Credit rating agencies, largely based in advanced economies, reinforce the perception of safety for rich-country debt and risk for poorer-country debt. Central banks worldwide accumulate reserves in dollars and euros, further entrenching the system. The broader implication is that the\n\nrichest economies do not simply benefit from privilege; they actively shape and maintain it. By controlling the currencies, rules, and institutions at the center of global finance, they secure a rentier position that channels income from the rest of the world, thereby exacerbating inequality across countries.\n\n**Barriers for reducing inequality across countries**\n\nThe financial asymmetries documented in this chapter are not only technical imbalances; they translate directly into barriers for development. **Figure 5.8**\n\nillustrates how poorer countries systematically transfer resources to richer ones, constraining their fiscal capacity and long-term growth prospects. The bottom 80% of countries devote a significant share of their GDP to net income outflows, which can be seen as the cost of financing the privilege for the top 20%. These outflows, averaging 2–3% of GDP each year, represent resources that could otherwise be invested in schools, hospitals, or infrastructure. The cost is\n\n106","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Chapter 5. Exorbitant Privilege\n\n**Figure 5.2. Rich countries receive 1% of their GDP from the rest of the world due to financial privilege**\n\n−6% −5% −4% −3% −2% −1%\n\n0% 1% 2% 3%\n\n1980 1990 2000 2010 2020\n\nExcess yield\n\n(% of group GDP), MER\n\nBottom 20%\n\n20%−40%\n\n40%−60%\n\n60%−80%\n\nTop 20%\n\n**By income group**\n\n−6% −5% −4% −3% −2% −1%\n\n0% 1% 2% 3%\n\n1980 1990 2000 2010 2020\n\nChina\n\nEASA (excl. China)\n\nEurope\n\nLATA\n\nMENA\n\nNAOC\n\nRUCA\n\nSSAF\n\nSSEA\n\n**By world region**\n\n**Interpretation.** These two panels show excess yield income (privilege), defined as the difference between the return on foreign assets and liabilities, as a share of national GDP across different country groupings. The left panel presents data by per capita income quintiles. It shows that only the top 20% richest countries enjoy consistently positive excess yield income (1% of their combined GDP by 2025). This privilege stems from the centrality of these rich countries in the monetary and financial system. The right panel shows this pattern by world regions. Financial privilege is overwhelmingly concentrated in East Asia, North America & Oceania, and Europe, while the other regions face consistent financial burden relative to their GDP. **Sources and series:** wir2026.wid.world/methodology and Nievas and Sodano (2025).\n\n**Excess yield (assets−liabilities) as % of group GDP, 1970−2025**\n\ncomparing returns on investments directly. The richest 20% consistently earn more on their assets abroad and pay less on their liabilities. Over the last half-century, global returns on assets have fallen for all, but the decline has been steepest for poorer countries. More importantly, the liability costs have remained high or even increased for the poorer countries. Only the richest 20% have experienced a large decrease in liability costs. The result is a structural advantage for rich countries: they are “charged less” on what they owe. These patterns are not the accidental outcome of market forces. They stem from policy design and institutional dominance. For instance, regulatory standards such as Basel III increased the demand for “safe” assets, consolidating the role of U.S. Treasuries and European sovereign bonds. Credit rating agencies, largely based in advanced economies, reinforce the perception of safety for rich-country debt and risk for poorer-country debt. Central banks worldwide accumulate reserves in dollars and euros, further entrenching the system. The broader implication is that the\n\nrichest economies do not simply benefit from privilege; they actively shape and maintain it. By controlling the currencies, rules, and institutions at the center of global finance, they secure a rentier position that channels income from the rest of the world, thereby exacerbating inequality across countries.\n\n**Barriers for reducing inequality across countries**\n\nThe financial asymmetries documented in this chapter are not only technical imbalances; they translate directly into barriers for development. **Figure 5.8**\n\nillustrates how poorer countries systematically transfer resources to richer ones, constraining their fiscal capacity and long-term growth prospects. The bottom 80% of countries devote a significant share of their GDP to net income outflows, which can be seen as the cost of financing the privilege for the top 20%. These outflows, averaging 2–3% of GDP each year, represent resources that could otherwise be invested in schools, hospitals, or infrastructure. The cost is\n\n106","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p106.pdf"]},"response":{"errors":[],"extract_id":"extract_e042c633c70083413648f7e780cd93f4","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Chapter 5. Exorbitant Privilege\n\n**Figure 5.2. Rich countries receive 1% of their GDP from the rest of the world due to financial privilege**\n\n−6% −5% −4% −3% −2% −1%\n\n0% 1% 2% 3%\n\n1980 1990 2000 2010 2020\n\nExcess yield\n\n(% of group GDP), MER\n\nBottom 20%\n\n20%−40%\n\n40%−60%\n\n60%−80%\n\nTop 20%\n\n**By income group**\n\n−6% −5% −4% −3% −2% −1%\n\n0% 1% 2% 3%\n\n1980 1990 2000 2010 2020\n\nChina\n\nEASA (excl. China)\n\nEurope\n\nLATA\n\nMENA\n\nNAOC\n\nRUCA\n\nSSAF\n\nSSEA\n\n**By world region**\n\n**Interpretation.** These two panels show excess yield income (privilege), defined as the difference between the return on foreign assets and liabilities, as a share of national GDP across different country groupings. The left panel presents data by per capita income quintiles. It shows that only the top 20% richest countries enjoy consistently positive excess yield income (1% of their combined GDP by 2025). This privilege stems from the centrality of these rich countries in the monetary and financial system. The right panel shows this pattern by world regions. Financial privilege is overwhelmingly concentrated in East Asia, North America & Oceania, and Europe, while the other regions face consistent financial burden relative to their GDP. **Sources and series:** wir2026.wid.world/methodology and Nievas and Sodano (2025).\n\n**Excess yield (assets−liabilities) as % of group GDP, 1970−2025**\n\ncomparing returns on investments directly. The richest 20% consistently earn more on their assets abroad and pay less on their liabilities. Over the last half-century, global returns on assets have fallen for all, but the decline has been steepest for poorer countries. More importantly, the liability costs have remained high or even increased for the poorer countries. Only the richest 20% have experienced a large decrease in liability costs. The result is a structural advantage for rich countries: they are “charged less” on what they owe. These patterns are not the accidental outcome of market forces. They stem from policy design and institutional dominance. For instance, regulatory standards such as Basel III increased the demand for “safe” assets, consolidating the role of U.S. Treasuries and European sovereign bonds. Credit rating agencies, largely based in advanced economies, reinforce the perception of safety for rich-country debt and risk for poorer-country debt. Central banks worldwide accumulate reserves in dollars and euros, further entrenching the system. The broader implication is that the\n\nrichest economies do not simply benefit from privilege; they actively shape and maintain it. By controlling the currencies, rules, and institutions at the center of global finance, they secure a rentier position that channels income from the rest of the world, thereby exacerbating inequality across countries.\n\n**Barriers for reducing inequality across countries**\n\nThe financial asymmetries documented in this chapter are not only technical imbalances; they translate directly into barriers for development. **Figure 5.8**\n\nillustrates how poorer countries systematically transfer resources to richer ones, constraining their fiscal capacity and long-term growth prospects. The bottom 80% of countries devote a significant share of their GDP to net income outflows, which can be seen as the cost of financing the privilege for the top 20%. These outflows, averaging 2–3% of GDP each year, represent resources that could otherwise be invested in schools, hospitals, or infrastructure. The cost is\n\n106"],"full_content":"Chapter 5. Exorbitant Privilege\n\n**Figure 5.2. Rich countries receive 1% of their GDP from the rest of the world due to financial privilege**\n\n−6% −5% −4% −3% −2% −1%\n\n0% 1% 2% 3%\n\n1980 1990 2000 2010 2020\n\nExcess yield\n\n(% of group GDP), MER\n\nBottom 20%\n\n20%−40%\n\n40%−60%\n\n60%−80%\n\nTop 20%\n\n**By income group**\n\n−6% −5% −4% −3% −2% −1%\n\n0% 1% 2% 3%\n\n1980 1990 2000 2010 2020\n\nChina\n\nEASA (excl. China)\n\nEurope\n\nLATA\n\nMENA\n\nNAOC\n\nRUCA\n\nSSAF\n\nSSEA\n\n**By world region**\n\n**Interpretation.** These two panels show excess yield income (privilege), defined as the difference between the return on foreign assets and liabilities, as a share of national GDP across different country groupings. The left panel presents data by per capita income quintiles. It shows that only the top 20% richest countries enjoy consistently positive excess yield income (1% of their combined GDP by 2025). This privilege stems from the centrality of these rich countries in the monetary and financial system. The right panel shows this pattern by world regions. Financial privilege is overwhelmingly concentrated in East Asia, North America & Oceania, and Europe, while the other regions face consistent financial burden relative to their GDP. **Sources and series:** wir2026.wid.world/methodology and Nievas and Sodano (2025).\n\n**Excess yield (assets−liabilities) as % of group GDP, 1970−2025**\n\ncomparing returns on investments directly. The richest 20% consistently earn more on their assets abroad and pay less on their liabilities. Over the last half-century, global returns on assets have fallen for all, but the decline has been steepest for poorer countries. More importantly, the liability costs have remained high or even increased for the poorer countries. Only the richest 20% have experienced a large decrease in liability costs. The result is a structural advantage for rich countries: they are “charged less” on what they owe. These patterns are not the accidental outcome of market forces. They stem from policy design and institutional dominance. For instance, regulatory standards such as Basel III increased the demand for “safe” assets, consolidating the role of U.S. Treasuries and European sovereign bonds. Credit rating agencies, largely based in advanced economies, reinforce the perception of safety for rich-country debt and risk for poorer-country debt. Central banks worldwide accumulate reserves in dollars and euros, further entrenching the system. The broader implication is that the\n\nrichest economies do not simply benefit from privilege; they actively shape and maintain it. By controlling the currencies, rules, and institutions at the center of global finance, they secure a rentier position that channels income from the rest of the world, thereby exacerbating inequality across countries.\n\n**Barriers for reducing inequality across countries**\n\nThe financial asymmetries documented in this chapter are not only technical imbalances; they translate directly into barriers for development. **Figure 5.8**\n\nillustrates how poorer countries systematically transfer resources to richer ones, constraining their fiscal capacity and long-term growth prospects. The bottom 80% of countries devote a significant share of their GDP to net income outflows, which can be seen as the cost of financing the privilege for the top 20%. These outflows, averaging 2–3% of GDP each year, represent resources that could otherwise be invested in schools, hospitals, or infrastructure. The cost is\n\n106","publish_date":null,"title":"Chapter 5. Exorbitant Privilege Figure5.2. Richcountriesreceive1%oftheirGDPfromtherest oftheworldduetofinancialprivilege comparing returns on investments directly. richesteconomiesdonotsimplybenefitfrom privilege; they actively shape and maintain The","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p106.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p106.pdf"},"request":{"example_id":"chart/World_Inequality_Report_2026_p106","product_type":"parse","source_file_path":"World_Inequality_Report_2026_p106.pdf"},"started_at":"2026-11-04T19:45:13.537720"} +{"completed_at":"2026-11-04T19:45:16.104107","latency_in_ms":1334,"output":{"example_id":"chart/World_Inequality_Report_2026_p107","grounded_pages":[],"layout_pages":[],"markdown":"Chapter 5. Exorbitant Privilege\n\n**Figure 5.3. Privilege persists for the U.S. (and its region) despite negative net foreign asset positions**\n\n−6%\n\n−4%\n\n−2%\n\n0%\n\n2%\n\n4%\n\n6%\n\n1970 1980 1990 2000 2010 2020\n\nNet foreign assets\n\n(% of world GDP), MER\n\nBottom 20%\n\n20%−40%\n\n40%−60%\n\n60%−80%\n\nTop 20%\n\n**By income group**\n\n−20%\n\n−16%\n\n−12%\n\n−8%\n\n−4%\n\n0%\n\n4%\n\n8%\n\n1970 1980 1990 2000 2010 2020\n\nChina\n\nEASA (excl. China)\n\nEurope\n\nLATA\n\nMENA\n\nNAOC\n\nRUCA\n\nSSAF\n\nSSEA\n\n**By world region**\n\n**Interpretation.** These panels show net foreign assets (NFA) as a share of world GDP by income group (left) and world region (right). Global net asset positions remain deeply unequal. The top 20% richest countries have maintained a positive NFA position equivalent to nearly 2% of world GDP in 2025, while the bottom 80% of the world population have held mainly negative positions, deteriorated by net investment income outflows and valuation losses. **Sources and series:** Nievas and Sodano (2025) and wir2026.wid.world/methodology.\n\n**Net foreign assets as % of world GDP, 1970−2025**\n\nparticularly heavy for low-income regions, where financing privilege often demands higher budgets than health expenditure. By contrast, rich countries receive steady inflows, reinforcing their ability to sustain higher living standards. The implication is stark. The current financial system perpetuates global inequality by design. In many ways, these income transfers function as a modern form of unequal exchange, subtler than colonial extraction, but no less constraining for the development paths of poorer nations.\n\n**Need for reforms in the international financial, trade, and monetary systems**\n\n**Figure 5.9** synthesizes two centuries of evidence on how global asymmetries in trade, finance, and income have been structured (see Nievas and Piketty (2025)). Taken together, its four panels reveal not just fluctuations in balances but enduring patterns of power and privilege in the international financial system. Panel (a) traces net foreign income\n\nbalances. It highlights Europe’s remarkable capacity in the 19th century to enjoy positive income flows despite persistent trade deficits. By the eve of World War I, these inflows amounted to about 1.5% of world GDP, an unprecedented record. Panels (b) and (c) explain how this was possible: Europe’s foreign assets, concentrated in colonies, peaked at nearly one-third of world GDP, allowing it to convert deficits in goods and services into surpluses in income. A striking modern parallel emerges in North America & Oceania. Panel (c) shows the region today holds large negative net foreign assets, while panel (d) confirms a persistent trade deficit. Yet panel (a) reveals that North America & Oceania still records positive net foreign income. The explanation lies in panel (b): excess yield. Like Europe in the colonial era, North America & Oceania, led by the United States, systematically earns higher returns on its assets abroad than it pays on its liabilities. This exorbitant privilege allows the region to live with persistent trade and net foreign asset deficits while continuing to obtain positive net income from the rest of the world.\n\n107","pages":[{"line_numbers":[],"markdown":"Chapter 5. Exorbitant Privilege\n\n**Figure 5.3. Privilege persists for the U.S. (and its region) despite negative net foreign asset positions**\n\n−6%\n\n−4%\n\n−2%\n\n0%\n\n2%\n\n4%\n\n6%\n\n1970 1980 1990 2000 2010 2020\n\nNet foreign assets\n\n(% of world GDP), MER\n\nBottom 20%\n\n20%−40%\n\n40%−60%\n\n60%−80%\n\nTop 20%\n\n**By income group**\n\n−20%\n\n−16%\n\n−12%\n\n−8%\n\n−4%\n\n0%\n\n4%\n\n8%\n\n1970 1980 1990 2000 2010 2020\n\nChina\n\nEASA (excl. China)\n\nEurope\n\nLATA\n\nMENA\n\nNAOC\n\nRUCA\n\nSSAF\n\nSSEA\n\n**By world region**\n\n**Interpretation.** These panels show net foreign assets (NFA) as a share of world GDP by income group (left) and world region (right). Global net asset positions remain deeply unequal. The top 20% richest countries have maintained a positive NFA position equivalent to nearly 2% of world GDP in 2025, while the bottom 80% of the world population have held mainly negative positions, deteriorated by net investment income outflows and valuation losses. **Sources and series:** Nievas and Sodano (2025) and wir2026.wid.world/methodology.\n\n**Net foreign assets as % of world GDP, 1970−2025**\n\nparticularly heavy for low-income regions, where financing privilege often demands higher budgets than health expenditure. By contrast, rich countries receive steady inflows, reinforcing their ability to sustain higher living standards. The implication is stark. The current financial system perpetuates global inequality by design. In many ways, these income transfers function as a modern form of unequal exchange, subtler than colonial extraction, but no less constraining for the development paths of poorer nations.\n\n**Need for reforms in the international financial, trade, and monetary systems**\n\n**Figure 5.9** synthesizes two centuries of evidence on how global asymmetries in trade, finance, and income have been structured (see Nievas and Piketty (2025)). Taken together, its four panels reveal not just fluctuations in balances but enduring patterns of power and privilege in the international financial system. Panel (a) traces net foreign income\n\nbalances. It highlights Europe’s remarkable capacity in the 19th century to enjoy positive income flows despite persistent trade deficits. By the eve of World War I, these inflows amounted to about 1.5% of world GDP, an unprecedented record. Panels (b) and (c) explain how this was possible: Europe’s foreign assets, concentrated in colonies, peaked at nearly one-third of world GDP, allowing it to convert deficits in goods and services into surpluses in income. A striking modern parallel emerges in North America & Oceania. Panel (c) shows the region today holds large negative net foreign assets, while panel (d) confirms a persistent trade deficit. Yet panel (a) reveals that North America & Oceania still records positive net foreign income. The explanation lies in panel (b): excess yield. Like Europe in the colonial era, North America & Oceania, led by the United States, systematically earns higher returns on its assets abroad than it pays on its liabilities. This exorbitant privilege allows the region to live with persistent trade and net foreign asset deficits while continuing to obtain positive net income from the rest of the world.\n\n107","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Chapter 5. Exorbitant Privilege\n\n**Figure 5.3. Privilege persists for the U.S. (and its region) despite negative net foreign asset positions**\n\n−6%\n\n−4%\n\n−2%\n\n0%\n\n2%\n\n4%\n\n6%\n\n1970 1980 1990 2000 2010 2020\n\nNet foreign assets\n\n(% of world GDP), MER\n\nBottom 20%\n\n20%−40%\n\n40%−60%\n\n60%−80%\n\nTop 20%\n\n**By income group**\n\n−20%\n\n−16%\n\n−12%\n\n−8%\n\n−4%\n\n0%\n\n4%\n\n8%\n\n1970 1980 1990 2000 2010 2020\n\nChina\n\nEASA (excl. China)\n\nEurope\n\nLATA\n\nMENA\n\nNAOC\n\nRUCA\n\nSSAF\n\nSSEA\n\n**By world region**\n\n**Interpretation.** These panels show net foreign assets (NFA) as a share of world GDP by income group (left) and world region (right). Global net asset positions remain deeply unequal. The top 20% richest countries have maintained a positive NFA position equivalent to nearly 2% of world GDP in 2025, while the bottom 80% of the world population have held mainly negative positions, deteriorated by net investment income outflows and valuation losses. **Sources and series:** Nievas and Sodano (2025) and wir2026.wid.world/methodology.\n\n**Net foreign assets as % of world GDP, 1970−2025**\n\nparticularly heavy for low-income regions, where financing privilege often demands higher budgets than health expenditure. By contrast, rich countries receive steady inflows, reinforcing their ability to sustain higher living standards. The implication is stark. The current financial system perpetuates global inequality by design. In many ways, these income transfers function as a modern form of unequal exchange, subtler than colonial extraction, but no less constraining for the development paths of poorer nations.\n\n**Need for reforms in the international financial, trade, and monetary systems**\n\n**Figure 5.9** synthesizes two centuries of evidence on how global asymmetries in trade, finance, and income have been structured (see Nievas and Piketty (2025)). Taken together, its four panels reveal not just fluctuations in balances but enduring patterns of power and privilege in the international financial system. Panel (a) traces net foreign income\n\nbalances. It highlights Europe’s remarkable capacity in the 19th century to enjoy positive income flows despite persistent trade deficits. By the eve of World War I, these inflows amounted to about 1.5% of world GDP, an unprecedented record. Panels (b) and (c) explain how this was possible: Europe’s foreign assets, concentrated in colonies, peaked at nearly one-third of world GDP, allowing it to convert deficits in goods and services into surpluses in income. A striking modern parallel emerges in North America & Oceania. Panel (c) shows the region today holds large negative net foreign assets, while panel (d) confirms a persistent trade deficit. Yet panel (a) reveals that North America & Oceania still records positive net foreign income. The explanation lies in panel (b): excess yield. Like Europe in the colonial era, North America & Oceania, led by the United States, systematically earns higher returns on its assets abroad than it pays on its liabilities. This exorbitant privilege allows the region to live with persistent trade and net foreign asset deficits while continuing to obtain positive net income from the rest of the world.\n\n107","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p107.pdf"]},"response":{"errors":[],"extract_id":"extract_96f45a34059fee8ba1cb24317003adb5","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Chapter 5. Exorbitant Privilege\n\n**Figure 5.3. Privilege persists for the U.S. (and its region) despite negative net foreign asset positions**\n\n−6%\n\n−4%\n\n−2%\n\n0%\n\n2%\n\n4%\n\n6%\n\n1970 1980 1990 2000 2010 2020\n\nNet foreign assets\n\n(% of world GDP), MER\n\nBottom 20%\n\n20%−40%\n\n40%−60%\n\n60%−80%\n\nTop 20%\n\n**By income group**\n\n−20%\n\n−16%\n\n−12%\n\n−8%\n\n−4%\n\n0%\n\n4%\n\n8%\n\n1970 1980 1990 2000 2010 2020\n\nChina\n\nEASA (excl. China)\n\nEurope\n\nLATA\n\nMENA\n\nNAOC\n\nRUCA\n\nSSAF\n\nSSEA\n\n**By world region**\n\n**Interpretation.** These panels show net foreign assets (NFA) as a share of world GDP by income group (left) and world region (right). Global net asset positions remain deeply unequal. The top 20% richest countries have maintained a positive NFA position equivalent to nearly 2% of world GDP in 2025, while the bottom 80% of the world population have held mainly negative positions, deteriorated by net investment income outflows and valuation losses. **Sources and series:** Nievas and Sodano (2025) and wir2026.wid.world/methodology.\n\n**Net foreign assets as % of world GDP, 1970−2025**\n\nparticularly heavy for low-income regions, where financing privilege often demands higher budgets than health expenditure. By contrast, rich countries receive steady inflows, reinforcing their ability to sustain higher living standards. The implication is stark. The current financial system perpetuates global inequality by design. In many ways, these income transfers function as a modern form of unequal exchange, subtler than colonial extraction, but no less constraining for the development paths of poorer nations.\n\n**Need for reforms in the international financial, trade, and monetary systems**\n\n**Figure 5.9** synthesizes two centuries of evidence on how global asymmetries in trade, finance, and income have been structured (see Nievas and Piketty (2025)). Taken together, its four panels reveal not just fluctuations in balances but enduring patterns of power and privilege in the international financial system. Panel (a) traces net foreign income\n\nbalances. It highlights Europe’s remarkable capacity in the 19th century to enjoy positive income flows despite persistent trade deficits. By the eve of World War I, these inflows amounted to about 1.5% of world GDP, an unprecedented record. Panels (b) and (c) explain how this was possible: Europe’s foreign assets, concentrated in colonies, peaked at nearly one-third of world GDP, allowing it to convert deficits in goods and services into surpluses in income. A striking modern parallel emerges in North America & Oceania. Panel (c) shows the region today holds large negative net foreign assets, while panel (d) confirms a persistent trade deficit. Yet panel (a) reveals that North America & Oceania still records positive net foreign income. The explanation lies in panel (b): excess yield. Like Europe in the colonial era, North America & Oceania, led by the United States, systematically earns higher returns on its assets abroad than it pays on its liabilities. This exorbitant privilege allows the region to live with persistent trade and net foreign asset deficits while continuing to obtain positive net income from the rest of the world.\n\n107"],"full_content":"Chapter 5. Exorbitant Privilege\n\n**Figure 5.3. Privilege persists for the U.S. (and its region) despite negative net foreign asset positions**\n\n−6%\n\n−4%\n\n−2%\n\n0%\n\n2%\n\n4%\n\n6%\n\n1970 1980 1990 2000 2010 2020\n\nNet foreign assets\n\n(% of world GDP), MER\n\nBottom 20%\n\n20%−40%\n\n40%−60%\n\n60%−80%\n\nTop 20%\n\n**By income group**\n\n−20%\n\n−16%\n\n−12%\n\n−8%\n\n−4%\n\n0%\n\n4%\n\n8%\n\n1970 1980 1990 2000 2010 2020\n\nChina\n\nEASA (excl. China)\n\nEurope\n\nLATA\n\nMENA\n\nNAOC\n\nRUCA\n\nSSAF\n\nSSEA\n\n**By world region**\n\n**Interpretation.** These panels show net foreign assets (NFA) as a share of world GDP by income group (left) and world region (right). Global net asset positions remain deeply unequal. The top 20% richest countries have maintained a positive NFA position equivalent to nearly 2% of world GDP in 2025, while the bottom 80% of the world population have held mainly negative positions, deteriorated by net investment income outflows and valuation losses. **Sources and series:** Nievas and Sodano (2025) and wir2026.wid.world/methodology.\n\n**Net foreign assets as % of world GDP, 1970−2025**\n\nparticularly heavy for low-income regions, where financing privilege often demands higher budgets than health expenditure. By contrast, rich countries receive steady inflows, reinforcing their ability to sustain higher living standards. The implication is stark. The current financial system perpetuates global inequality by design. In many ways, these income transfers function as a modern form of unequal exchange, subtler than colonial extraction, but no less constraining for the development paths of poorer nations.\n\n**Need for reforms in the international financial, trade, and monetary systems**\n\n**Figure 5.9** synthesizes two centuries of evidence on how global asymmetries in trade, finance, and income have been structured (see Nievas and Piketty (2025)). Taken together, its four panels reveal not just fluctuations in balances but enduring patterns of power and privilege in the international financial system. Panel (a) traces net foreign income\n\nbalances. It highlights Europe’s remarkable capacity in the 19th century to enjoy positive income flows despite persistent trade deficits. By the eve of World War I, these inflows amounted to about 1.5% of world GDP, an unprecedented record. Panels (b) and (c) explain how this was possible: Europe’s foreign assets, concentrated in colonies, peaked at nearly one-third of world GDP, allowing it to convert deficits in goods and services into surpluses in income. A striking modern parallel emerges in North America & Oceania. Panel (c) shows the region today holds large negative net foreign assets, while panel (d) confirms a persistent trade deficit. Yet panel (a) reveals that North America & Oceania still records positive net foreign income. The explanation lies in panel (b): excess yield. Like Europe in the colonial era, North America & Oceania, led by the United States, systematically earns higher returns on its assets abroad than it pays on its liabilities. This exorbitant privilege allows the region to live with persistent trade and net foreign asset deficits while continuing to obtain positive net income from the rest of the world.\n\n107","publish_date":null,"title":"Chapter 5. Exorbitant Privilege Figure5.3. PrivilegepersistsfortheU.S.(anditsregion)despite negativenetforeignassetpositions particularly heavy for low-income regions, balances. It highlights Europe’s remarkable where financing privilege","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p107.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p107.pdf"},"request":{"example_id":"chart/World_Inequality_Report_2026_p107","product_type":"parse","source_file_path":"World_Inequality_Report_2026_p107.pdf"},"started_at":"2026-11-04T19:45:14.769681"} +{"completed_at":"2026-11-04T19:45:17.754343","latency_in_ms":1648,"output":{"example_id":"chart/World_Inequality_Report_2026_p108","grounded_pages":[],"layout_pages":[],"markdown":"Chapter 5. Exorbitant Privilege\n\n**Figure 5.4. There is a persistent net income transfer from poor to rich**\n\n−6% −5% −4% −3% −2% −1%\n\n0% 1% 2% 3% 4% 5% 6%\n\n1980 1990 2000 2010 2020\n\nNet foreign capital income\n\n(% of group GDP), MER\n\nBottom 20%\n\n20%−40%\n\n40%−60%\n\n60%−80%\n\nTop 20%\n\n**By income group**\n\n−6% −5% −4% −3% −2% −1%\n\n0% 1% 2% 3% 4% 5% 6%\n\n1980 1990 2000 2010 2020\n\nChina\n\nEASA (excl. China)\n\nEurope\n\nLATA\n\nMENA\n\nNAOC\n\nRUCA\n\nSSAF\n\nSSEA\n\n**By world region**\n\n**Interpretation.** These panels show net foreign capital income as a share of GDP by income group (left) and region (right). The top 20% richest countries consistently earn positive net income from abroad, while poorer countries face persistent deficits. This reflects structural asymmetries: rich countries invest in high−return assets and issue low−cost liabilities, whereas poorer countries hold low−yield reserves and pay high returns on debt. At the regional level, East Asia (excluding China), Europe, and North America & Oceania capture the gains, while the rest of the world bear the costs. **Sources and series:** Nievas and Sodano (2025) and wir2026.wid.world/methodology.\n\n**Net foreign capital income as % of group GDP, 1970−2025**\n\nEast Asia, by contrast, follows a more “textbook” path. Its rising creditor position since the 1980s has been built on persistent surpluses in trade, not on privileged yields. This comparison underscores how structural asymmetries and unequal exchange differ in geography but not in essence: colonial Europe relied mainly on extraction and colonial rents; today’s North America & Oceania relies mainly on financial dominance and institutionalized excess yields. This shows that global imbalances are not corrected by market forces. They are sustained by entrenched hierarchies of finance, trade, and monetary power. Addressing such asymmetries requires systemic reform. A meaningful reform of the global monetary and trade system will require a new mix of rules and institutions. Proposals in Nievas and Piketty (2025) include options such as pegged exchange rates closer to purchasing power parities, expanded use of special drawing rights (SDRs, an international reserve asset created by the International Monetary Fund (IMF)), creation of a global currency, centralized credit and debit systems, and even corrective\n\ntaxes on excessive surpluses. The underlying message is clear: global economic relations are shaped less by self-correcting markets than by persistent power asymmetries and structural imbalances. Without bold reforms, the skewed logic of “exorbitant privilege” will continue, locking the Global South into unequal exchange and constraining its development.\n\n**Main takeaways**\n\nThis chapter has shown that what began as the United States’ “exorbitant privilege” has become a structural privilege of the rich world. Advanced economies borrow cheaply, lend profitably, and secure income inflows, while poorer countries face the opposite reality: costly liabilities, low-yield assets, and a persistent outflow of income. These patterns are not the result of market efficiency but of institutional design that places reserve currency issuers and financial centers at the core of the global system. The evidence demonstrates that this\n\n108","pages":[{"line_numbers":[],"markdown":"Chapter 5. Exorbitant Privilege\n\n**Figure 5.4. There is a persistent net income transfer from poor to rich**\n\n−6% −5% −4% −3% −2% −1%\n\n0% 1% 2% 3% 4% 5% 6%\n\n1980 1990 2000 2010 2020\n\nNet foreign capital income\n\n(% of group GDP), MER\n\nBottom 20%\n\n20%−40%\n\n40%−60%\n\n60%−80%\n\nTop 20%\n\n**By income group**\n\n−6% −5% −4% −3% −2% −1%\n\n0% 1% 2% 3% 4% 5% 6%\n\n1980 1990 2000 2010 2020\n\nChina\n\nEASA (excl. China)\n\nEurope\n\nLATA\n\nMENA\n\nNAOC\n\nRUCA\n\nSSAF\n\nSSEA\n\n**By world region**\n\n**Interpretation.** These panels show net foreign capital income as a share of GDP by income group (left) and region (right). The top 20% richest countries consistently earn positive net income from abroad, while poorer countries face persistent deficits. This reflects structural asymmetries: rich countries invest in high−return assets and issue low−cost liabilities, whereas poorer countries hold low−yield reserves and pay high returns on debt. At the regional level, East Asia (excluding China), Europe, and North America & Oceania capture the gains, while the rest of the world bear the costs. **Sources and series:** Nievas and Sodano (2025) and wir2026.wid.world/methodology.\n\n**Net foreign capital income as % of group GDP, 1970−2025**\n\nEast Asia, by contrast, follows a more “textbook” path. Its rising creditor position since the 1980s has been built on persistent surpluses in trade, not on privileged yields. This comparison underscores how structural asymmetries and unequal exchange differ in geography but not in essence: colonial Europe relied mainly on extraction and colonial rents; today’s North America & Oceania relies mainly on financial dominance and institutionalized excess yields. This shows that global imbalances are not corrected by market forces. They are sustained by entrenched hierarchies of finance, trade, and monetary power. Addressing such asymmetries requires systemic reform. A meaningful reform of the global monetary and trade system will require a new mix of rules and institutions. Proposals in Nievas and Piketty (2025) include options such as pegged exchange rates closer to purchasing power parities, expanded use of special drawing rights (SDRs, an international reserve asset created by the International Monetary Fund (IMF)), creation of a global currency, centralized credit and debit systems, and even corrective\n\ntaxes on excessive surpluses. The underlying message is clear: global economic relations are shaped less by self-correcting markets than by persistent power asymmetries and structural imbalances. Without bold reforms, the skewed logic of “exorbitant privilege” will continue, locking the Global South into unequal exchange and constraining its development.\n\n**Main takeaways**\n\nThis chapter has shown that what began as the United States’ “exorbitant privilege” has become a structural privilege of the rich world. Advanced economies borrow cheaply, lend profitably, and secure income inflows, while poorer countries face the opposite reality: costly liabilities, low-yield assets, and a persistent outflow of income. These patterns are not the result of market efficiency but of institutional design that places reserve currency issuers and financial centers at the core of the global system. The evidence demonstrates that this\n\n108","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Chapter 5. Exorbitant Privilege\n\n**Figure 5.4. There is a persistent net income transfer from poor to rich**\n\n−6% −5% −4% −3% −2% −1%\n\n0% 1% 2% 3% 4% 5% 6%\n\n1980 1990 2000 2010 2020\n\nNet foreign capital income\n\n(% of group GDP), MER\n\nBottom 20%\n\n20%−40%\n\n40%−60%\n\n60%−80%\n\nTop 20%\n\n**By income group**\n\n−6% −5% −4% −3% −2% −1%\n\n0% 1% 2% 3% 4% 5% 6%\n\n1980 1990 2000 2010 2020\n\nChina\n\nEASA (excl. China)\n\nEurope\n\nLATA\n\nMENA\n\nNAOC\n\nRUCA\n\nSSAF\n\nSSEA\n\n**By world region**\n\n**Interpretation.** These panels show net foreign capital income as a share of GDP by income group (left) and region (right). The top 20% richest countries consistently earn positive net income from abroad, while poorer countries face persistent deficits. This reflects structural asymmetries: rich countries invest in high−return assets and issue low−cost liabilities, whereas poorer countries hold low−yield reserves and pay high returns on debt. At the regional level, East Asia (excluding China), Europe, and North America & Oceania capture the gains, while the rest of the world bear the costs. **Sources and series:** Nievas and Sodano (2025) and wir2026.wid.world/methodology.\n\n**Net foreign capital income as % of group GDP, 1970−2025**\n\nEast Asia, by contrast, follows a more “textbook” path. Its rising creditor position since the 1980s has been built on persistent surpluses in trade, not on privileged yields. This comparison underscores how structural asymmetries and unequal exchange differ in geography but not in essence: colonial Europe relied mainly on extraction and colonial rents; today’s North America & Oceania relies mainly on financial dominance and institutionalized excess yields. This shows that global imbalances are not corrected by market forces. They are sustained by entrenched hierarchies of finance, trade, and monetary power. Addressing such asymmetries requires systemic reform. A meaningful reform of the global monetary and trade system will require a new mix of rules and institutions. Proposals in Nievas and Piketty (2025) include options such as pegged exchange rates closer to purchasing power parities, expanded use of special drawing rights (SDRs, an international reserve asset created by the International Monetary Fund (IMF)), creation of a global currency, centralized credit and debit systems, and even corrective\n\ntaxes on excessive surpluses. The underlying message is clear: global economic relations are shaped less by self-correcting markets than by persistent power asymmetries and structural imbalances. Without bold reforms, the skewed logic of “exorbitant privilege” will continue, locking the Global South into unequal exchange and constraining its development.\n\n**Main takeaways**\n\nThis chapter has shown that what began as the United States’ “exorbitant privilege” has become a structural privilege of the rich world. Advanced economies borrow cheaply, lend profitably, and secure income inflows, while poorer countries face the opposite reality: costly liabilities, low-yield assets, and a persistent outflow of income. These patterns are not the result of market efficiency but of institutional design that places reserve currency issuers and financial centers at the core of the global system. The evidence demonstrates that this\n\n108","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p108.pdf"]},"response":{"errors":[],"extract_id":"extract_45183aca8cf9f7984305b760c01fd582","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Chapter 5. Exorbitant Privilege\n\n**Figure 5.4. There is a persistent net income transfer from poor to rich**\n\n−6% −5% −4% −3% −2% −1%\n\n0% 1% 2% 3% 4% 5% 6%\n\n1980 1990 2000 2010 2020\n\nNet foreign capital income\n\n(% of group GDP), MER\n\nBottom 20%\n\n20%−40%\n\n40%−60%\n\n60%−80%\n\nTop 20%\n\n**By income group**\n\n−6% −5% −4% −3% −2% −1%\n\n0% 1% 2% 3% 4% 5% 6%\n\n1980 1990 2000 2010 2020\n\nChina\n\nEASA (excl. China)\n\nEurope\n\nLATA\n\nMENA\n\nNAOC\n\nRUCA\n\nSSAF\n\nSSEA\n\n**By world region**\n\n**Interpretation.** These panels show net foreign capital income as a share of GDP by income group (left) and region (right). The top 20% richest countries consistently earn positive net income from abroad, while poorer countries face persistent deficits. This reflects structural asymmetries: rich countries invest in high−return assets and issue low−cost liabilities, whereas poorer countries hold low−yield reserves and pay high returns on debt. At the regional level, East Asia (excluding China), Europe, and North America & Oceania capture the gains, while the rest of the world bear the costs. **Sources and series:** Nievas and Sodano (2025) and wir2026.wid.world/methodology.\n\n**Net foreign capital income as % of group GDP, 1970−2025**\n\nEast Asia, by contrast, follows a more “textbook” path. Its rising creditor position since the 1980s has been built on persistent surpluses in trade, not on privileged yields. This comparison underscores how structural asymmetries and unequal exchange differ in geography but not in essence: colonial Europe relied mainly on extraction and colonial rents; today’s North America & Oceania relies mainly on financial dominance and institutionalized excess yields. This shows that global imbalances are not corrected by market forces. They are sustained by entrenched hierarchies of finance, trade, and monetary power. Addressing such asymmetries requires systemic reform. A meaningful reform of the global monetary and trade system will require a new mix of rules and institutions. Proposals in Nievas and Piketty (2025) include options such as pegged exchange rates closer to purchasing power parities, expanded use of special drawing rights (SDRs, an international reserve asset created by the International Monetary Fund (IMF)), creation of a global currency, centralized credit and debit systems, and even corrective\n\ntaxes on excessive surpluses. The underlying message is clear: global economic relations are shaped less by self-correcting markets than by persistent power asymmetries and structural imbalances. Without bold reforms, the skewed logic of “exorbitant privilege” will continue, locking the Global South into unequal exchange and constraining its development.\n\n**Main takeaways**\n\nThis chapter has shown that what began as the United States’ “exorbitant privilege” has become a structural privilege of the rich world. Advanced economies borrow cheaply, lend profitably, and secure income inflows, while poorer countries face the opposite reality: costly liabilities, low-yield assets, and a persistent outflow of income. These patterns are not the result of market efficiency but of institutional design that places reserve currency issuers and financial centers at the core of the global system. The evidence demonstrates that this\n\n108"],"full_content":"Chapter 5. Exorbitant Privilege\n\n**Figure 5.4. There is a persistent net income transfer from poor to rich**\n\n−6% −5% −4% −3% −2% −1%\n\n0% 1% 2% 3% 4% 5% 6%\n\n1980 1990 2000 2010 2020\n\nNet foreign capital income\n\n(% of group GDP), MER\n\nBottom 20%\n\n20%−40%\n\n40%−60%\n\n60%−80%\n\nTop 20%\n\n**By income group**\n\n−6% −5% −4% −3% −2% −1%\n\n0% 1% 2% 3% 4% 5% 6%\n\n1980 1990 2000 2010 2020\n\nChina\n\nEASA (excl. China)\n\nEurope\n\nLATA\n\nMENA\n\nNAOC\n\nRUCA\n\nSSAF\n\nSSEA\n\n**By world region**\n\n**Interpretation.** These panels show net foreign capital income as a share of GDP by income group (left) and region (right). The top 20% richest countries consistently earn positive net income from abroad, while poorer countries face persistent deficits. This reflects structural asymmetries: rich countries invest in high−return assets and issue low−cost liabilities, whereas poorer countries hold low−yield reserves and pay high returns on debt. At the regional level, East Asia (excluding China), Europe, and North America & Oceania capture the gains, while the rest of the world bear the costs. **Sources and series:** Nievas and Sodano (2025) and wir2026.wid.world/methodology.\n\n**Net foreign capital income as % of group GDP, 1970−2025**\n\nEast Asia, by contrast, follows a more “textbook” path. Its rising creditor position since the 1980s has been built on persistent surpluses in trade, not on privileged yields. This comparison underscores how structural asymmetries and unequal exchange differ in geography but not in essence: colonial Europe relied mainly on extraction and colonial rents; today’s North America & Oceania relies mainly on financial dominance and institutionalized excess yields. This shows that global imbalances are not corrected by market forces. They are sustained by entrenched hierarchies of finance, trade, and monetary power. Addressing such asymmetries requires systemic reform. A meaningful reform of the global monetary and trade system will require a new mix of rules and institutions. Proposals in Nievas and Piketty (2025) include options such as pegged exchange rates closer to purchasing power parities, expanded use of special drawing rights (SDRs, an international reserve asset created by the International Monetary Fund (IMF)), creation of a global currency, centralized credit and debit systems, and even corrective\n\ntaxes on excessive surpluses. The underlying message is clear: global economic relations are shaped less by self-correcting markets than by persistent power asymmetries and structural imbalances. Without bold reforms, the skewed logic of “exorbitant privilege” will continue, locking the Global South into unequal exchange and constraining its development.\n\n**Main takeaways**\n\nThis chapter has shown that what began as the United States’ “exorbitant privilege” has become a structural privilege of the rich world. Advanced economies borrow cheaply, lend profitably, and secure income inflows, while poorer countries face the opposite reality: costly liabilities, low-yield assets, and a persistent outflow of income. These patterns are not the result of market efficiency but of institutional design that places reserve currency issuers and financial centers at the core of the global system. The evidence demonstrates that this\n\n108","publish_date":null,"title":"Chapter 5. Exorbitant Privilege Figure5.4. Thereisapersistentnetincometransferfrompoor torich East Asia, by contrast, follows a more taxes on excessive surpluses. “textbook” path.","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p108.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p108.pdf"},"request":{"example_id":"chart/World_Inequality_Report_2026_p108","product_type":"parse","source_file_path":"World_Inequality_Report_2026_p108.pdf"},"started_at":"2026-11-04T19:45:16.105437"} +{"completed_at":"2026-11-04T19:45:18.834628","latency_in_ms":1076,"output":{"example_id":"chart/World_Inequality_Report_2026_p109","grounded_pages":[],"layout_pages":[],"markdown":"Chapter 5. Exorbitant Privilege\n\n**Figure 5.5. Rich countries are global financial rentiers by political design, not because of market dynamics**\n\n0%\n\n15%\n\n30%\n\n45%\n\n60%\n\n75%\n\n90%\n\n1990 2000 2010 2020\n\nShare of global trade by\n\ncurrency invoiced in\n\n**Trade by currency**\n\n0%\n\n10%\n\n20%\n\n30%\n\n40%\n\n50%\n\n1990 2000 2010 2020\n\nShare of global assets\n\nby currency\n\nU.S. dollar\n\nEuro\n\nPound sterling\n\nJapanese yen\n\nChinese renminbi\n\nOther currencies\n\nSwiss franc\n\n**Assets by currency**\n\n0%\n\n15%\n\n30%\n\n45%\n\n60%\n\n75%\n\n90%\n\n1990 2000 2010 2020\n\nShare of global reserves\n\nby currency\n\n**Reserves by currency**\n\n**Interpretation.** The left panel shows the dominance of the U.S. dollar and the euro in global trade invoicing, driving global demand for deposits and assets in these currencies to hedge against default and exchange rate risks. The center panel shows that this dominance extends to global foreign asset portfolios: both private and public investors worldwide accumulate USD− and EUR−denominated assets for safety, liquidity, and regulatory reasons, especially since Basel III rules boosted demand for low−risk instruments. The right panel confirms that these currencies also dominate central bank reserves, locking in persistent demand. **Sources and series:** Nievas and Sodano (2025) and wir2026.wid.world/methodology.\n\n**Share of global trade, assets, and reserves by currency, 1990−2025**\n\nprivilege translates into a continuous transfer of resources from poorer to richer countries. Far from narrowing gaps, financial globalization has increased them. This amounts to a modern form of unequal exchange: colonial powers once relied on resource extraction to transform deficits into surpluses; today, advanced economies achieve the same through excess yields. The burden falls on developing countries, reducing their capacity to invest in education, health, and infrastructure. By constraining human capital formation and fiscal space, the system limits their ability to reduce inequality across countries. Without structural reform, global inequality will persist.\n\n109","pages":[{"line_numbers":[],"markdown":"Chapter 5. Exorbitant Privilege\n\n**Figure 5.5. Rich countries are global financial rentiers by political design, not because of market dynamics**\n\n0%\n\n15%\n\n30%\n\n45%\n\n60%\n\n75%\n\n90%\n\n1990 2000 2010 2020\n\nShare of global trade by\n\ncurrency invoiced in\n\n**Trade by currency**\n\n0%\n\n10%\n\n20%\n\n30%\n\n40%\n\n50%\n\n1990 2000 2010 2020\n\nShare of global assets\n\nby currency\n\nU.S. dollar\n\nEuro\n\nPound sterling\n\nJapanese yen\n\nChinese renminbi\n\nOther currencies\n\nSwiss franc\n\n**Assets by currency**\n\n0%\n\n15%\n\n30%\n\n45%\n\n60%\n\n75%\n\n90%\n\n1990 2000 2010 2020\n\nShare of global reserves\n\nby currency\n\n**Reserves by currency**\n\n**Interpretation.** The left panel shows the dominance of the U.S. dollar and the euro in global trade invoicing, driving global demand for deposits and assets in these currencies to hedge against default and exchange rate risks. The center panel shows that this dominance extends to global foreign asset portfolios: both private and public investors worldwide accumulate USD− and EUR−denominated assets for safety, liquidity, and regulatory reasons, especially since Basel III rules boosted demand for low−risk instruments. The right panel confirms that these currencies also dominate central bank reserves, locking in persistent demand. **Sources and series:** Nievas and Sodano (2025) and wir2026.wid.world/methodology.\n\n**Share of global trade, assets, and reserves by currency, 1990−2025**\n\nprivilege translates into a continuous transfer of resources from poorer to richer countries. Far from narrowing gaps, financial globalization has increased them. This amounts to a modern form of unequal exchange: colonial powers once relied on resource extraction to transform deficits into surpluses; today, advanced economies achieve the same through excess yields. The burden falls on developing countries, reducing their capacity to invest in education, health, and infrastructure. By constraining human capital formation and fiscal space, the system limits their ability to reduce inequality across countries. Without structural reform, global inequality will persist.\n\n109","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Chapter 5. Exorbitant Privilege\n\n**Figure 5.5. Rich countries are global financial rentiers by political design, not because of market dynamics**\n\n0%\n\n15%\n\n30%\n\n45%\n\n60%\n\n75%\n\n90%\n\n1990 2000 2010 2020\n\nShare of global trade by\n\ncurrency invoiced in\n\n**Trade by currency**\n\n0%\n\n10%\n\n20%\n\n30%\n\n40%\n\n50%\n\n1990 2000 2010 2020\n\nShare of global assets\n\nby currency\n\nU.S. dollar\n\nEuro\n\nPound sterling\n\nJapanese yen\n\nChinese renminbi\n\nOther currencies\n\nSwiss franc\n\n**Assets by currency**\n\n0%\n\n15%\n\n30%\n\n45%\n\n60%\n\n75%\n\n90%\n\n1990 2000 2010 2020\n\nShare of global reserves\n\nby currency\n\n**Reserves by currency**\n\n**Interpretation.** The left panel shows the dominance of the U.S. dollar and the euro in global trade invoicing, driving global demand for deposits and assets in these currencies to hedge against default and exchange rate risks. The center panel shows that this dominance extends to global foreign asset portfolios: both private and public investors worldwide accumulate USD− and EUR−denominated assets for safety, liquidity, and regulatory reasons, especially since Basel III rules boosted demand for low−risk instruments. The right panel confirms that these currencies also dominate central bank reserves, locking in persistent demand. **Sources and series:** Nievas and Sodano (2025) and wir2026.wid.world/methodology.\n\n**Share of global trade, assets, and reserves by currency, 1990−2025**\n\nprivilege translates into a continuous transfer of resources from poorer to richer countries. Far from narrowing gaps, financial globalization has increased them. This amounts to a modern form of unequal exchange: colonial powers once relied on resource extraction to transform deficits into surpluses; today, advanced economies achieve the same through excess yields. The burden falls on developing countries, reducing their capacity to invest in education, health, and infrastructure. By constraining human capital formation and fiscal space, the system limits their ability to reduce inequality across countries. Without structural reform, global inequality will persist.\n\n109","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p109.pdf"]},"response":{"errors":[],"extract_id":"extract_80799990172b5df4fac210fe0d745112","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Chapter 5. Exorbitant Privilege\n\n**Figure 5.5. Rich countries are global financial rentiers by political design, not because of market dynamics**\n\n0%\n\n15%\n\n30%\n\n45%\n\n60%\n\n75%\n\n90%\n\n1990 2000 2010 2020\n\nShare of global trade by\n\ncurrency invoiced in\n\n**Trade by currency**\n\n0%\n\n10%\n\n20%\n\n30%\n\n40%\n\n50%\n\n1990 2000 2010 2020\n\nShare of global assets\n\nby currency\n\nU.S. dollar\n\nEuro\n\nPound sterling\n\nJapanese yen\n\nChinese renminbi\n\nOther currencies\n\nSwiss franc\n\n**Assets by currency**\n\n0%\n\n15%\n\n30%\n\n45%\n\n60%\n\n75%\n\n90%\n\n1990 2000 2010 2020\n\nShare of global reserves\n\nby currency\n\n**Reserves by currency**\n\n**Interpretation.** The left panel shows the dominance of the U.S. dollar and the euro in global trade invoicing, driving global demand for deposits and assets in these currencies to hedge against default and exchange rate risks. The center panel shows that this dominance extends to global foreign asset portfolios: both private and public investors worldwide accumulate USD− and EUR−denominated assets for safety, liquidity, and regulatory reasons, especially since Basel III rules boosted demand for low−risk instruments. The right panel confirms that these currencies also dominate central bank reserves, locking in persistent demand. **Sources and series:** Nievas and Sodano (2025) and wir2026.wid.world/methodology.\n\n**Share of global trade, assets, and reserves by currency, 1990−2025**\n\nprivilege translates into a continuous transfer of resources from poorer to richer countries. Far from narrowing gaps, financial globalization has increased them. This amounts to a modern form of unequal exchange: colonial powers once relied on resource extraction to transform deficits into surpluses; today, advanced economies achieve the same through excess yields. The burden falls on developing countries, reducing their capacity to invest in education, health, and infrastructure. By constraining human capital formation and fiscal space, the system limits their ability to reduce inequality across countries. Without structural reform, global inequality will persist.\n\n109"],"full_content":"Chapter 5. Exorbitant Privilege\n\n**Figure 5.5. Rich countries are global financial rentiers by political design, not because of market dynamics**\n\n0%\n\n15%\n\n30%\n\n45%\n\n60%\n\n75%\n\n90%\n\n1990 2000 2010 2020\n\nShare of global trade by\n\ncurrency invoiced in\n\n**Trade by currency**\n\n0%\n\n10%\n\n20%\n\n30%\n\n40%\n\n50%\n\n1990 2000 2010 2020\n\nShare of global assets\n\nby currency\n\nU.S. dollar\n\nEuro\n\nPound sterling\n\nJapanese yen\n\nChinese renminbi\n\nOther currencies\n\nSwiss franc\n\n**Assets by currency**\n\n0%\n\n15%\n\n30%\n\n45%\n\n60%\n\n75%\n\n90%\n\n1990 2000 2010 2020\n\nShare of global reserves\n\nby currency\n\n**Reserves by currency**\n\n**Interpretation.** The left panel shows the dominance of the U.S. dollar and the euro in global trade invoicing, driving global demand for deposits and assets in these currencies to hedge against default and exchange rate risks. The center panel shows that this dominance extends to global foreign asset portfolios: both private and public investors worldwide accumulate USD− and EUR−denominated assets for safety, liquidity, and regulatory reasons, especially since Basel III rules boosted demand for low−risk instruments. The right panel confirms that these currencies also dominate central bank reserves, locking in persistent demand. **Sources and series:** Nievas and Sodano (2025) and wir2026.wid.world/methodology.\n\n**Share of global trade, assets, and reserves by currency, 1990−2025**\n\nprivilege translates into a continuous transfer of resources from poorer to richer countries. Far from narrowing gaps, financial globalization has increased them. This amounts to a modern form of unequal exchange: colonial powers once relied on resource extraction to transform deficits into surpluses; today, advanced economies achieve the same through excess yields. The burden falls on developing countries, reducing their capacity to invest in education, health, and infrastructure. By constraining human capital formation and fiscal space, the system limits their ability to reduce inequality across countries. Without structural reform, global inequality will persist.\n\n109","publish_date":null,"title":"Chapter 5. Exorbitant Privilege Figure5.5. Richcountriesareglobalfinancialrentiersbypolitical design,notbecauseofmarketdynamics privilege translates into a continuous transfer of resources from poorer to richer countries.","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p109.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p109.pdf"},"request":{"example_id":"chart/World_Inequality_Report_2026_p109","product_type":"parse","source_file_path":"World_Inequality_Report_2026_p109.pdf"},"started_at":"2026-11-04T19:45:17.757832"} +{"completed_at":"2026-11-04T19:45:20.123439","latency_in_ms":1274,"output":{"example_id":"chart/World_Inequality_Report_2026_p110","grounded_pages":[],"layout_pages":[],"markdown":"Chapter 5. Exorbitant Privilege\n\n**Figure 5.6. Poor countries finance the privilege through asymmetric portfolios**\n\nBottom 20% 20−40% 40−60% 60−80% Top 20%\n\n0%\n\n10% 20% 30% 40% 50% 60% 70% 80% 90%\n\n100%\n\nA L A L A L A L A L\n\n**1970–1999**\n\nBottom 20% 20−40% 40−60% 60−80% Top 20%\n\n0%\n\n10% 20% 30% 40% 50% 60% 70% 80% 90%\n\n100%\n\nA L A L A L A L A L\n\n**2000–2025**\n\nDebt Equity FDI Reserves\n\n**Interpretation.** This figure shows the decomposition of foreign assets (A) and liabilities (L) across country income groups in two periods: 1970–1999 and 2000–2025. Rich countries (top 20%) hold fewer foreign exchange reserves and issue fewer foreign direct investment (FDI) liabilities, both of which are low−return components. Instead, they have increased their share of equity and FDI assets, which typically yield higher returns. On the liability side, they continue to rely on debt issuance, which is safer and lower−cost due to their strong credit ratings and reserve currency status. In contrast, poorer countries (bottom 80%) have portfolios skewed toward reserves as assets and FDI as liabilities, both associated with lower net returns. **Sources and series:** Nievas and Sodano (2025) and wir2026.wid.world/methodology.\n\n**Decomposition of assets (A) and liabilities (L)**\n\n**Figure 5.7. Poorer countries face lower asset returns and higher liability costs**\n\n0% 1% 2% 3% 4% 5% 6% 7% 8% 9%\n\n10% 11%\n\n1980 1990 2000 2010 2020\n\nAverage return per income\n\ngroup (%), MER\n\nWorld\n\nBottom 20%\n\n20%−40%\n\n40%−60%\n\n60%−80%\n\nTop 20%\n\n**Returns on foreign assets**\n\n0% 1% 2% 3% 4% 5% 6% 7% 8% 9%\n\n10% 11%\n\n1980 1990 2000 2010 2020\n\nWorld\n\nBottom 20%\n\n20%−40%\n\n40%−60%\n\n60%−80%\n\nTop 20%\n\n**Returns on foreign liabilities**\n\n**Interpretation.** These two panels compare the average return on foreign assets (left) and liabilities (right) across countries grouped by per capita income levels. The top 20% richest countries consistently achieve higher returns on their assets abroad, while facing lower average returns on their liabilities. In contrast, the bottom 80% of countries not only face lower returns on assets but also pay significantly higher returns on liabilities. Only the top 20% have managed to reduce liability costs significantly. All of this results in a positive return differential for the richest countries and a negative differential for poorer countries, structurally transferring income from poor to rich through global financial flows. **Sources and series:** Nievas and Sodano (2025) and wir2026.wid.world/methodology.\n\n**Return to foreign assets and liabilities, 1970−2025**\n\n110","pages":[{"line_numbers":[],"markdown":"Chapter 5. Exorbitant Privilege\n\n**Figure 5.6. Poor countries finance the privilege through asymmetric portfolios**\n\nBottom 20% 20−40% 40−60% 60−80% Top 20%\n\n0%\n\n10% 20% 30% 40% 50% 60% 70% 80% 90%\n\n100%\n\nA L A L A L A L A L\n\n**1970–1999**\n\nBottom 20% 20−40% 40−60% 60−80% Top 20%\n\n0%\n\n10% 20% 30% 40% 50% 60% 70% 80% 90%\n\n100%\n\nA L A L A L A L A L\n\n**2000–2025**\n\nDebt Equity FDI Reserves\n\n**Interpretation.** This figure shows the decomposition of foreign assets (A) and liabilities (L) across country income groups in two periods: 1970–1999 and 2000–2025. Rich countries (top 20%) hold fewer foreign exchange reserves and issue fewer foreign direct investment (FDI) liabilities, both of which are low−return components. Instead, they have increased their share of equity and FDI assets, which typically yield higher returns. On the liability side, they continue to rely on debt issuance, which is safer and lower−cost due to their strong credit ratings and reserve currency status. In contrast, poorer countries (bottom 80%) have portfolios skewed toward reserves as assets and FDI as liabilities, both associated with lower net returns. **Sources and series:** Nievas and Sodano (2025) and wir2026.wid.world/methodology.\n\n**Decomposition of assets (A) and liabilities (L)**\n\n**Figure 5.7. Poorer countries face lower asset returns and higher liability costs**\n\n0% 1% 2% 3% 4% 5% 6% 7% 8% 9%\n\n10% 11%\n\n1980 1990 2000 2010 2020\n\nAverage return per income\n\ngroup (%), MER\n\nWorld\n\nBottom 20%\n\n20%−40%\n\n40%−60%\n\n60%−80%\n\nTop 20%\n\n**Returns on foreign assets**\n\n0% 1% 2% 3% 4% 5% 6% 7% 8% 9%\n\n10% 11%\n\n1980 1990 2000 2010 2020\n\nWorld\n\nBottom 20%\n\n20%−40%\n\n40%−60%\n\n60%−80%\n\nTop 20%\n\n**Returns on foreign liabilities**\n\n**Interpretation.** These two panels compare the average return on foreign assets (left) and liabilities (right) across countries grouped by per capita income levels. The top 20% richest countries consistently achieve higher returns on their assets abroad, while facing lower average returns on their liabilities. In contrast, the bottom 80% of countries not only face lower returns on assets but also pay significantly higher returns on liabilities. Only the top 20% have managed to reduce liability costs significantly. All of this results in a positive return differential for the richest countries and a negative differential for poorer countries, structurally transferring income from poor to rich through global financial flows. **Sources and series:** Nievas and Sodano (2025) and wir2026.wid.world/methodology.\n\n**Return to foreign assets and liabilities, 1970−2025**\n\n110","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Chapter 5. Exorbitant Privilege\n\n**Figure 5.6. Poor countries finance the privilege through asymmetric portfolios**\n\nBottom 20% 20−40% 40−60% 60−80% Top 20%\n\n0%\n\n10% 20% 30% 40% 50% 60% 70% 80% 90%\n\n100%\n\nA L A L A L A L A L\n\n**1970–1999**\n\nBottom 20% 20−40% 40−60% 60−80% Top 20%\n\n0%\n\n10% 20% 30% 40% 50% 60% 70% 80% 90%\n\n100%\n\nA L A L A L A L A L\n\n**2000–2025**\n\nDebt Equity FDI Reserves\n\n**Interpretation.** This figure shows the decomposition of foreign assets (A) and liabilities (L) across country income groups in two periods: 1970–1999 and 2000–2025. Rich countries (top 20%) hold fewer foreign exchange reserves and issue fewer foreign direct investment (FDI) liabilities, both of which are low−return components. Instead, they have increased their share of equity and FDI assets, which typically yield higher returns. On the liability side, they continue to rely on debt issuance, which is safer and lower−cost due to their strong credit ratings and reserve currency status. In contrast, poorer countries (bottom 80%) have portfolios skewed toward reserves as assets and FDI as liabilities, both associated with lower net returns. **Sources and series:** Nievas and Sodano (2025) and wir2026.wid.world/methodology.\n\n**Decomposition of assets (A) and liabilities (L)**\n\n**Figure 5.7. Poorer countries face lower asset returns and higher liability costs**\n\n0% 1% 2% 3% 4% 5% 6% 7% 8% 9%\n\n10% 11%\n\n1980 1990 2000 2010 2020\n\nAverage return per income\n\ngroup (%), MER\n\nWorld\n\nBottom 20%\n\n20%−40%\n\n40%−60%\n\n60%−80%\n\nTop 20%\n\n**Returns on foreign assets**\n\n0% 1% 2% 3% 4% 5% 6% 7% 8% 9%\n\n10% 11%\n\n1980 1990 2000 2010 2020\n\nWorld\n\nBottom 20%\n\n20%−40%\n\n40%−60%\n\n60%−80%\n\nTop 20%\n\n**Returns on foreign liabilities**\n\n**Interpretation.** These two panels compare the average return on foreign assets (left) and liabilities (right) across countries grouped by per capita income levels. The top 20% richest countries consistently achieve higher returns on their assets abroad, while facing lower average returns on their liabilities. In contrast, the bottom 80% of countries not only face lower returns on assets but also pay significantly higher returns on liabilities. Only the top 20% have managed to reduce liability costs significantly. All of this results in a positive return differential for the richest countries and a negative differential for poorer countries, structurally transferring income from poor to rich through global financial flows. **Sources and series:** Nievas and Sodano (2025) and wir2026.wid.world/methodology.\n\n**Return to foreign assets and liabilities, 1970−2025**\n\n110","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p110.pdf"]},"response":{"errors":[],"extract_id":"extract_11d4e8769b0cbd8fc794dd22d4698dd3","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Chapter 5. Exorbitant Privilege\n\n**Figure 5.6. Poor countries finance the privilege through asymmetric portfolios**\n\nBottom 20% 20−40% 40−60% 60−80% Top 20%\n\n0%\n\n10% 20% 30% 40% 50% 60% 70% 80% 90%\n\n100%\n\nA L A L A L A L A L\n\n**1970–1999**\n\nBottom 20% 20−40% 40−60% 60−80% Top 20%\n\n0%\n\n10% 20% 30% 40% 50% 60% 70% 80% 90%\n\n100%\n\nA L A L A L A L A L\n\n**2000–2025**\n\nDebt Equity FDI Reserves\n\n**Interpretation.** This figure shows the decomposition of foreign assets (A) and liabilities (L) across country income groups in two periods: 1970–1999 and 2000–2025. Rich countries (top 20%) hold fewer foreign exchange reserves and issue fewer foreign direct investment (FDI) liabilities, both of which are low−return components. Instead, they have increased their share of equity and FDI assets, which typically yield higher returns. On the liability side, they continue to rely on debt issuance, which is safer and lower−cost due to their strong credit ratings and reserve currency status. In contrast, poorer countries (bottom 80%) have portfolios skewed toward reserves as assets and FDI as liabilities, both associated with lower net returns. **Sources and series:** Nievas and Sodano (2025) and wir2026.wid.world/methodology.\n\n**Decomposition of assets (A) and liabilities (L)**\n\n**Figure 5.7. Poorer countries face lower asset returns and higher liability costs**\n\n0% 1% 2% 3% 4% 5% 6% 7% 8% 9%\n\n10% 11%\n\n1980 1990 2000 2010 2020\n\nAverage return per income\n\ngroup (%), MER\n\nWorld\n\nBottom 20%\n\n20%−40%\n\n40%−60%\n\n60%−80%\n\nTop 20%\n\n**Returns on foreign assets**\n\n0% 1% 2% 3% 4% 5% 6% 7% 8% 9%\n\n10% 11%\n\n1980 1990 2000 2010 2020\n\nWorld\n\nBottom 20%\n\n20%−40%\n\n40%−60%\n\n60%−80%\n\nTop 20%\n\n**Returns on foreign liabilities**\n\n**Interpretation.** These two panels compare the average return on foreign assets (left) and liabilities (right) across countries grouped by per capita income levels. The top 20% richest countries consistently achieve higher returns on their assets abroad, while facing lower average returns on their liabilities. In contrast, the bottom 80% of countries not only face lower returns on assets but also pay significantly higher returns on liabilities. Only the top 20% have managed to reduce liability costs significantly. All of this results in a positive return differential for the richest countries and a negative differential for poorer countries, structurally transferring income from poor to rich through global financial flows. **Sources and series:** Nievas and Sodano (2025) and wir2026.wid.world/methodology.\n\n**Return to foreign assets and liabilities, 1970−2025**\n\n110"],"full_content":"Chapter 5. Exorbitant Privilege\n\n**Figure 5.6. Poor countries finance the privilege through asymmetric portfolios**\n\nBottom 20% 20−40% 40−60% 60−80% Top 20%\n\n0%\n\n10% 20% 30% 40% 50% 60% 70% 80% 90%\n\n100%\n\nA L A L A L A L A L\n\n**1970–1999**\n\nBottom 20% 20−40% 40−60% 60−80% Top 20%\n\n0%\n\n10% 20% 30% 40% 50% 60% 70% 80% 90%\n\n100%\n\nA L A L A L A L A L\n\n**2000–2025**\n\nDebt Equity FDI Reserves\n\n**Interpretation.** This figure shows the decomposition of foreign assets (A) and liabilities (L) across country income groups in two periods: 1970–1999 and 2000–2025. Rich countries (top 20%) hold fewer foreign exchange reserves and issue fewer foreign direct investment (FDI) liabilities, both of which are low−return components. Instead, they have increased their share of equity and FDI assets, which typically yield higher returns. On the liability side, they continue to rely on debt issuance, which is safer and lower−cost due to their strong credit ratings and reserve currency status. In contrast, poorer countries (bottom 80%) have portfolios skewed toward reserves as assets and FDI as liabilities, both associated with lower net returns. **Sources and series:** Nievas and Sodano (2025) and wir2026.wid.world/methodology.\n\n**Decomposition of assets (A) and liabilities (L)**\n\n**Figure 5.7. Poorer countries face lower asset returns and higher liability costs**\n\n0% 1% 2% 3% 4% 5% 6% 7% 8% 9%\n\n10% 11%\n\n1980 1990 2000 2010 2020\n\nAverage return per income\n\ngroup (%), MER\n\nWorld\n\nBottom 20%\n\n20%−40%\n\n40%−60%\n\n60%−80%\n\nTop 20%\n\n**Returns on foreign assets**\n\n0% 1% 2% 3% 4% 5% 6% 7% 8% 9%\n\n10% 11%\n\n1980 1990 2000 2010 2020\n\nWorld\n\nBottom 20%\n\n20%−40%\n\n40%−60%\n\n60%−80%\n\nTop 20%\n\n**Returns on foreign liabilities**\n\n**Interpretation.** These two panels compare the average return on foreign assets (left) and liabilities (right) across countries grouped by per capita income levels. The top 20% richest countries consistently achieve higher returns on their assets abroad, while facing lower average returns on their liabilities. In contrast, the bottom 80% of countries not only face lower returns on assets but also pay significantly higher returns on liabilities. Only the top 20% have managed to reduce liability costs significantly. All of this results in a positive return differential for the richest countries and a negative differential for poorer countries, structurally transferring income from poor to rich through global financial flows. **Sources and series:** Nievas and Sodano (2025) and wir2026.wid.world/methodology.\n\n**Return to foreign assets and liabilities, 1970−2025**\n\n110","publish_date":null,"title":"Chapter 5. Exorbitant Privilege Figure5.6. Poorcountriesfinancetheprivilegethrough asymmetricportfolios Figure5.7. Poorercountriesfacelowerassetreturnsandhigher liabilitycosts 110","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p110.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p110.pdf"},"request":{"example_id":"chart/World_Inequality_Report_2026_p110","product_type":"parse","source_file_path":"World_Inequality_Report_2026_p110.pdf"},"started_at":"2026-11-04T19:45:18.849022"} +{"completed_at":"2026-11-04T19:45:21.687823","latency_in_ms":1841,"output":{"example_id":"chart/World_Inequality_Report_2026_p111","grounded_pages":[],"layout_pages":[],"markdown":"Chapter 5. Exorbitant Privilege\n\n**Figure 5.8. Poorer countries can spend less on public services, exacerbating inequality**\n\n0%\n\n1%\n\n2%\n\n3%\n\n4%\n\n5%\n\n6%\n\nBottom 20% 20%−40% 40%−60% 60%−80%\n\nExpenditure\n\n% of GDP\n\nPrivilege Health Education\n\n**Interpretation.** This figure illustrates the cost of the global “privilege” system for the bottom 80% of the income distribution in 2022. The blue bars represent the share of GDP that each group effectively transfers to the top 20% richest countries through net income outflows (what can be seen as the cost of financing the privilege of the global top 20%). In many cases, these outflows are comparable to or even exceed the public investment these groups can make in health or education. For example, the 20%–40% group loses more in privilege outflows than it can allocate to health, a key driver of inequality reduction. This underscores how the bottom 80% bear a significant burden in sustaining global financial hierarchies, often at the expense of investments in their own human capital and exacerbating inequalities. **Sources and series:** Nievas and Sodano (2025).\n\n**Expenditure as a share of GDP, 2022**\n\n111","pages":[{"line_numbers":[],"markdown":"Chapter 5. Exorbitant Privilege\n\n**Figure 5.8. Poorer countries can spend less on public services, exacerbating inequality**\n\n0%\n\n1%\n\n2%\n\n3%\n\n4%\n\n5%\n\n6%\n\nBottom 20% 20%−40% 40%−60% 60%−80%\n\nExpenditure\n\n% of GDP\n\nPrivilege Health Education\n\n**Interpretation.** This figure illustrates the cost of the global “privilege” system for the bottom 80% of the income distribution in 2022. The blue bars represent the share of GDP that each group effectively transfers to the top 20% richest countries through net income outflows (what can be seen as the cost of financing the privilege of the global top 20%). In many cases, these outflows are comparable to or even exceed the public investment these groups can make in health or education. For example, the 20%–40% group loses more in privilege outflows than it can allocate to health, a key driver of inequality reduction. This underscores how the bottom 80% bear a significant burden in sustaining global financial hierarchies, often at the expense of investments in their own human capital and exacerbating inequalities. **Sources and series:** Nievas and Sodano (2025).\n\n**Expenditure as a share of GDP, 2022**\n\n111","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Chapter 5. Exorbitant Privilege\n\n**Figure 5.8. Poorer countries can spend less on public services, exacerbating inequality**\n\n0%\n\n1%\n\n2%\n\n3%\n\n4%\n\n5%\n\n6%\n\nBottom 20% 20%−40% 40%−60% 60%−80%\n\nExpenditure\n\n% of GDP\n\nPrivilege Health Education\n\n**Interpretation.** This figure illustrates the cost of the global “privilege” system for the bottom 80% of the income distribution in 2022. The blue bars represent the share of GDP that each group effectively transfers to the top 20% richest countries through net income outflows (what can be seen as the cost of financing the privilege of the global top 20%). In many cases, these outflows are comparable to or even exceed the public investment these groups can make in health or education. For example, the 20%–40% group loses more in privilege outflows than it can allocate to health, a key driver of inequality reduction. This underscores how the bottom 80% bear a significant burden in sustaining global financial hierarchies, often at the expense of investments in their own human capital and exacerbating inequalities. **Sources and series:** Nievas and Sodano (2025).\n\n**Expenditure as a share of GDP, 2022**\n\n111","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p111.pdf"]},"response":{"errors":[],"extract_id":"extract_e6b76a59c59c21c50f877f8835e4475c","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Chapter 5. Exorbitant Privilege\n\n**Figure 5.8. Poorer countries can spend less on public services, exacerbating inequality**\n\n0%\n\n1%\n\n2%\n\n3%\n\n4%\n\n5%\n\n6%\n\nBottom 20% 20%−40% 40%−60% 60%−80%\n\nExpenditure\n\n% of GDP\n\nPrivilege Health Education\n\n**Interpretation.** This figure illustrates the cost of the global “privilege” system for the bottom 80% of the income distribution in 2022. The blue bars represent the share of GDP that each group effectively transfers to the top 20% richest countries through net income outflows (what can be seen as the cost of financing the privilege of the global top 20%). In many cases, these outflows are comparable to or even exceed the public investment these groups can make in health or education. For example, the 20%–40% group loses more in privilege outflows than it can allocate to health, a key driver of inequality reduction. This underscores how the bottom 80% bear a significant burden in sustaining global financial hierarchies, often at the expense of investments in their own human capital and exacerbating inequalities. **Sources and series:** Nievas and Sodano (2025).\n\n**Expenditure as a share of GDP, 2022**\n\n111"],"full_content":"Chapter 5. Exorbitant Privilege\n\n**Figure 5.8. Poorer countries can spend less on public services, exacerbating inequality**\n\n0%\n\n1%\n\n2%\n\n3%\n\n4%\n\n5%\n\n6%\n\nBottom 20% 20%−40% 40%−60% 60%−80%\n\nExpenditure\n\n% of GDP\n\nPrivilege Health Education\n\n**Interpretation.** This figure illustrates the cost of the global “privilege” system for the bottom 80% of the income distribution in 2022. The blue bars represent the share of GDP that each group effectively transfers to the top 20% richest countries through net income outflows (what can be seen as the cost of financing the privilege of the global top 20%). In many cases, these outflows are comparable to or even exceed the public investment these groups can make in health or education. For example, the 20%–40% group loses more in privilege outflows than it can allocate to health, a key driver of inequality reduction. This underscores how the bottom 80% bear a significant burden in sustaining global financial hierarchies, often at the expense of investments in their own human capital and exacerbating inequalities. **Sources and series:** Nievas and Sodano (2025).\n\n**Expenditure as a share of GDP, 2022**\n\n111","publish_date":null,"title":"Chapter 5. Exorbitant Privilege Figure5.8. Poorercountriescanspendlessonpublicservices, exacerbatinginequality 111","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p111.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p111.pdf"},"request":{"example_id":"chart/World_Inequality_Report_2026_p111","product_type":"parse","source_file_path":"World_Inequality_Report_2026_p111.pdf"},"started_at":"2026-11-04T19:45:19.846090"} +{"completed_at":"2026-11-04T19:45:22.117091","latency_in_ms":1063,"output":{"example_id":"chart/World_Inequality_Report_2026_p118","grounded_pages":[],"layout_pages":[],"markdown":"Chapter 6. Climate, a Capital Problem\n\n**Figure 6.1. Triple climate inequality: the poorest lose the most, contribute the least, and lack the means to act**\n\n74%\n\n10%\n\n2%\n\n22%\n\n43%\n\n24%\n\n3%\n\n47%\n\n74%\n\n0%\n\n10%\n\n20%\n\n30%\n\n40%\n\n50%\n\n60%\n\n70%\n\n80%\n\nBottom 50% Middle 40% Top 10%\n\nShare in global totals\n\nRelative losses Emissions (consumption−based) Capacity to finance (wealth)\n\n**Global climate inequality: relative losses, emissions, and capacity to finance**\n\n**Interpretation.** The figure illustrates three dimensions of global climate inequality. Projected relative income losses from climate change are taken from Bothe et al. (2025). They represent percentage reductions in income compared with a business−as−usual scenario. The global bottom 50% concentrates 74% of these percentage reductions. The distribution of emissions is based on Bruckner et al. (2022). The distribution of wealth shares comes from WID (2025). Groups are defined by income for losses, by emitters for emissions, and by wealth for the wealth distribution, but all three distributions are highly correlated. For another paper on emissions inequalities by income groups, see Kartha et al. (2020), who find similar concentration levels. **Sources and series:** Bothe et al. (2025), Bruckner et al. (2022), and WID (2025).\n\nconcentration. In France, Germany, and the United States, the carbon footprint of the wealthiest 10% is three to five times higher when private ownership–based emissions are included. In the United States, the top 10% accounts for 24% of consumption-based emissions but 72% of ownership-based emissions. The share of the top 1% rises from 6% (consumption-based) to nearly 43% (ownership-based). At the global scale, the contrast is even sharper. The top 1% accounts for 41% of all greenhouse gas emissions under ownership-based accounting, compared with 15% under the consumption approach. Conversely, the contribution of the bottom 50% drops from 10% to 3% ( **Figure 6.2** ). In other words, the average individual in the top 1% emits more than twenty-five times as much carbon as the global average citizen. Their share of emissions even exceeds their share of global wealth—estimated at 36% in 2022 (Chancel and Rehm (2025a)). The extreme concentration of private ownership–based emissions stems from both the amount of wealth owned\n\nand the investment choices made. Wealthy individuals not only hold larger asset portfolios but also allocate them disproportionately toward high-carbon sectors. As shown in **Figure 6.3** , every $1 million invested in business assets in the United States corresponds to roughly 143 tonnes of carbon emissions, compared with 75 tonnes for equities (Chancel and Rehm (2025a)). Similar patterns emerge in France and Germany. The global top 10% allocates about half of their wealth to such carbon-intensive holdings, often seeking higher-risk, higher-return investments that coincide with higher emissions. Hsu, Li, and Tsou (2023) find that high-emission companies yield, on average, 4.4 percentage points more in annual excess returns than low-emission peers—an implicit “pollution premium” that further incentivizes carbon-heavy investments. From this ownership perspective, the nature of emissions changes across the wealth distribution. For low- and\n\n118","pages":[{"line_numbers":[],"markdown":"Chapter 6. Climate, a Capital Problem\n\n**Figure 6.1. Triple climate inequality: the poorest lose the most, contribute the least, and lack the means to act**\n\n74%\n\n10%\n\n2%\n\n22%\n\n43%\n\n24%\n\n3%\n\n47%\n\n74%\n\n0%\n\n10%\n\n20%\n\n30%\n\n40%\n\n50%\n\n60%\n\n70%\n\n80%\n\nBottom 50% Middle 40% Top 10%\n\nShare in global totals\n\nRelative losses Emissions (consumption−based) Capacity to finance (wealth)\n\n**Global climate inequality: relative losses, emissions, and capacity to finance**\n\n**Interpretation.** The figure illustrates three dimensions of global climate inequality. Projected relative income losses from climate change are taken from Bothe et al. (2025). They represent percentage reductions in income compared with a business−as−usual scenario. The global bottom 50% concentrates 74% of these percentage reductions. The distribution of emissions is based on Bruckner et al. (2022). The distribution of wealth shares comes from WID (2025). Groups are defined by income for losses, by emitters for emissions, and by wealth for the wealth distribution, but all three distributions are highly correlated. For another paper on emissions inequalities by income groups, see Kartha et al. (2020), who find similar concentration levels. **Sources and series:** Bothe et al. (2025), Bruckner et al. (2022), and WID (2025).\n\nconcentration. In France, Germany, and the United States, the carbon footprint of the wealthiest 10% is three to five times higher when private ownership–based emissions are included. In the United States, the top 10% accounts for 24% of consumption-based emissions but 72% of ownership-based emissions. The share of the top 1% rises from 6% (consumption-based) to nearly 43% (ownership-based). At the global scale, the contrast is even sharper. The top 1% accounts for 41% of all greenhouse gas emissions under ownership-based accounting, compared with 15% under the consumption approach. Conversely, the contribution of the bottom 50% drops from 10% to 3% ( **Figure 6.2** ). In other words, the average individual in the top 1% emits more than twenty-five times as much carbon as the global average citizen. Their share of emissions even exceeds their share of global wealth—estimated at 36% in 2022 (Chancel and Rehm (2025a)). The extreme concentration of private ownership–based emissions stems from both the amount of wealth owned\n\nand the investment choices made. Wealthy individuals not only hold larger asset portfolios but also allocate them disproportionately toward high-carbon sectors. As shown in **Figure 6.3** , every $1 million invested in business assets in the United States corresponds to roughly 143 tonnes of carbon emissions, compared with 75 tonnes for equities (Chancel and Rehm (2025a)). Similar patterns emerge in France and Germany. The global top 10% allocates about half of their wealth to such carbon-intensive holdings, often seeking higher-risk, higher-return investments that coincide with higher emissions. Hsu, Li, and Tsou (2023) find that high-emission companies yield, on average, 4.4 percentage points more in annual excess returns than low-emission peers—an implicit “pollution premium” that further incentivizes carbon-heavy investments. From this ownership perspective, the nature of emissions changes across the wealth distribution. For low- and\n\n118","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Chapter 6. Climate, a Capital Problem\n\n**Figure 6.1. Triple climate inequality: the poorest lose the most, contribute the least, and lack the means to act**\n\n74%\n\n10%\n\n2%\n\n22%\n\n43%\n\n24%\n\n3%\n\n47%\n\n74%\n\n0%\n\n10%\n\n20%\n\n30%\n\n40%\n\n50%\n\n60%\n\n70%\n\n80%\n\nBottom 50% Middle 40% Top 10%\n\nShare in global totals\n\nRelative losses Emissions (consumption−based) Capacity to finance (wealth)\n\n**Global climate inequality: relative losses, emissions, and capacity to finance**\n\n**Interpretation.** The figure illustrates three dimensions of global climate inequality. Projected relative income losses from climate change are taken from Bothe et al. (2025). They represent percentage reductions in income compared with a business−as−usual scenario. The global bottom 50% concentrates 74% of these percentage reductions. The distribution of emissions is based on Bruckner et al. (2022). The distribution of wealth shares comes from WID (2025). Groups are defined by income for losses, by emitters for emissions, and by wealth for the wealth distribution, but all three distributions are highly correlated. For another paper on emissions inequalities by income groups, see Kartha et al. (2020), who find similar concentration levels. **Sources and series:** Bothe et al. (2025), Bruckner et al. (2022), and WID (2025).\n\nconcentration. In France, Germany, and the United States, the carbon footprint of the wealthiest 10% is three to five times higher when private ownership–based emissions are included. In the United States, the top 10% accounts for 24% of consumption-based emissions but 72% of ownership-based emissions. The share of the top 1% rises from 6% (consumption-based) to nearly 43% (ownership-based). At the global scale, the contrast is even sharper. The top 1% accounts for 41% of all greenhouse gas emissions under ownership-based accounting, compared with 15% under the consumption approach. Conversely, the contribution of the bottom 50% drops from 10% to 3% ( **Figure 6.2** ). In other words, the average individual in the top 1% emits more than twenty-five times as much carbon as the global average citizen. Their share of emissions even exceeds their share of global wealth—estimated at 36% in 2022 (Chancel and Rehm (2025a)). The extreme concentration of private ownership–based emissions stems from both the amount of wealth owned\n\nand the investment choices made. Wealthy individuals not only hold larger asset portfolios but also allocate them disproportionately toward high-carbon sectors. As shown in **Figure 6.3** , every $1 million invested in business assets in the United States corresponds to roughly 143 tonnes of carbon emissions, compared with 75 tonnes for equities (Chancel and Rehm (2025a)). Similar patterns emerge in France and Germany. The global top 10% allocates about half of their wealth to such carbon-intensive holdings, often seeking higher-risk, higher-return investments that coincide with higher emissions. Hsu, Li, and Tsou (2023) find that high-emission companies yield, on average, 4.4 percentage points more in annual excess returns than low-emission peers—an implicit “pollution premium” that further incentivizes carbon-heavy investments. From this ownership perspective, the nature of emissions changes across the wealth distribution. For low- and\n\n118","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p118.pdf"]},"response":{"errors":[],"extract_id":"extract_c3eeb3c7448896ee6b02e91d5ee035d4","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Chapter 6. Climate, a Capital Problem\n\n**Figure 6.1. Triple climate inequality: the poorest lose the most, contribute the least, and lack the means to act**\n\n74%\n\n10%\n\n2%\n\n22%\n\n43%\n\n24%\n\n3%\n\n47%\n\n74%\n\n0%\n\n10%\n\n20%\n\n30%\n\n40%\n\n50%\n\n60%\n\n70%\n\n80%\n\nBottom 50% Middle 40% Top 10%\n\nShare in global totals\n\nRelative losses Emissions (consumption−based) Capacity to finance (wealth)\n\n**Global climate inequality: relative losses, emissions, and capacity to finance**\n\n**Interpretation.** The figure illustrates three dimensions of global climate inequality. Projected relative income losses from climate change are taken from Bothe et al. (2025). They represent percentage reductions in income compared with a business−as−usual scenario. The global bottom 50% concentrates 74% of these percentage reductions. The distribution of emissions is based on Bruckner et al. (2022). The distribution of wealth shares comes from WID (2025). Groups are defined by income for losses, by emitters for emissions, and by wealth for the wealth distribution, but all three distributions are highly correlated. For another paper on emissions inequalities by income groups, see Kartha et al. (2020), who find similar concentration levels. **Sources and series:** Bothe et al. (2025), Bruckner et al. (2022), and WID (2025).\n\nconcentration. In France, Germany, and the United States, the carbon footprint of the wealthiest 10% is three to five times higher when private ownership–based emissions are included. In the United States, the top 10% accounts for 24% of consumption-based emissions but 72% of ownership-based emissions. The share of the top 1% rises from 6% (consumption-based) to nearly 43% (ownership-based). At the global scale, the contrast is even sharper. The top 1% accounts for 41% of all greenhouse gas emissions under ownership-based accounting, compared with 15% under the consumption approach. Conversely, the contribution of the bottom 50% drops from 10% to 3% ( **Figure 6.2** ). In other words, the average individual in the top 1% emits more than twenty-five times as much carbon as the global average citizen. Their share of emissions even exceeds their share of global wealth—estimated at 36% in 2022 (Chancel and Rehm (2025a)). The extreme concentration of private ownership–based emissions stems from both the amount of wealth owned\n\nand the investment choices made. Wealthy individuals not only hold larger asset portfolios but also allocate them disproportionately toward high-carbon sectors. As shown in **Figure 6.3** , every $1 million invested in business assets in the United States corresponds to roughly 143 tonnes of carbon emissions, compared with 75 tonnes for equities (Chancel and Rehm (2025a)). Similar patterns emerge in France and Germany. The global top 10% allocates about half of their wealth to such carbon-intensive holdings, often seeking higher-risk, higher-return investments that coincide with higher emissions. Hsu, Li, and Tsou (2023) find that high-emission companies yield, on average, 4.4 percentage points more in annual excess returns than low-emission peers—an implicit “pollution premium” that further incentivizes carbon-heavy investments. From this ownership perspective, the nature of emissions changes across the wealth distribution. For low- and\n\n118"],"full_content":"Chapter 6. Climate, a Capital Problem\n\n**Figure 6.1. Triple climate inequality: the poorest lose the most, contribute the least, and lack the means to act**\n\n74%\n\n10%\n\n2%\n\n22%\n\n43%\n\n24%\n\n3%\n\n47%\n\n74%\n\n0%\n\n10%\n\n20%\n\n30%\n\n40%\n\n50%\n\n60%\n\n70%\n\n80%\n\nBottom 50% Middle 40% Top 10%\n\nShare in global totals\n\nRelative losses Emissions (consumption−based) Capacity to finance (wealth)\n\n**Global climate inequality: relative losses, emissions, and capacity to finance**\n\n**Interpretation.** The figure illustrates three dimensions of global climate inequality. Projected relative income losses from climate change are taken from Bothe et al. (2025). They represent percentage reductions in income compared with a business−as−usual scenario. The global bottom 50% concentrates 74% of these percentage reductions. The distribution of emissions is based on Bruckner et al. (2022). The distribution of wealth shares comes from WID (2025). Groups are defined by income for losses, by emitters for emissions, and by wealth for the wealth distribution, but all three distributions are highly correlated. For another paper on emissions inequalities by income groups, see Kartha et al. (2020), who find similar concentration levels. **Sources and series:** Bothe et al. (2025), Bruckner et al. (2022), and WID (2025).\n\nconcentration. In France, Germany, and the United States, the carbon footprint of the wealthiest 10% is three to five times higher when private ownership–based emissions are included. In the United States, the top 10% accounts for 24% of consumption-based emissions but 72% of ownership-based emissions. The share of the top 1% rises from 6% (consumption-based) to nearly 43% (ownership-based). At the global scale, the contrast is even sharper. The top 1% accounts for 41% of all greenhouse gas emissions under ownership-based accounting, compared with 15% under the consumption approach. Conversely, the contribution of the bottom 50% drops from 10% to 3% ( **Figure 6.2** ). In other words, the average individual in the top 1% emits more than twenty-five times as much carbon as the global average citizen. Their share of emissions even exceeds their share of global wealth—estimated at 36% in 2022 (Chancel and Rehm (2025a)). The extreme concentration of private ownership–based emissions stems from both the amount of wealth owned\n\nand the investment choices made. Wealthy individuals not only hold larger asset portfolios but also allocate them disproportionately toward high-carbon sectors. As shown in **Figure 6.3** , every $1 million invested in business assets in the United States corresponds to roughly 143 tonnes of carbon emissions, compared with 75 tonnes for equities (Chancel and Rehm (2025a)). Similar patterns emerge in France and Germany. The global top 10% allocates about half of their wealth to such carbon-intensive holdings, often seeking higher-risk, higher-return investments that coincide with higher emissions. Hsu, Li, and Tsou (2023) find that high-emission companies yield, on average, 4.4 percentage points more in annual excess returns than low-emission peers—an implicit “pollution premium” that further incentivizes carbon-heavy investments. From this ownership perspective, the nature of emissions changes across the wealth distribution. For low- and\n\n118","publish_date":null,"title":"Chapter 6. Climate, a Capital Problem Figure6.1. Tripleclimateinequality: thepoorestlosethemost, contributetheleast,andlackthemeanstoact concentration. In France, Germany, and and the investment choices made.","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p118.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p118.pdf"},"request":{"example_id":"chart/World_Inequality_Report_2026_p118","product_type":"parse","source_file_path":"World_Inequality_Report_2026_p118.pdf"},"started_at":"2026-11-04T19:45:21.054016"} +{"completed_at":"2026-11-04T19:45:23.640196","latency_in_ms":1068,"output":{"example_id":"chart/World_Inequality_Report_2026_p119","grounded_pages":[],"layout_pages":[],"markdown":"Chapter 6. Climate, a Capital Problem\n\n**Figure 6.2. Emissions are highly concentrated among the rich, especially when looking at ownership**\n\n10%\n\n3%\n\n43%\n\n20%\n\n47%\n\n77%\n\n15%\n\n41%\n\n0%\n\n10% 20% 30% 40% 50% 60% 70% 80% 90%\n\nBottom 50% Middle 40% Top 10% Top 1%\n\nShare in total emissions\n\nConsumption Private ownership\n\n**Emissions shares by groups**\n\n**Interpretation.** The figure shows the share of global GHG emissions attributable to the bottom 50% and the top 1% of the world population. Emissions are separated into consumption−based (emissions from production attributed to final consumers) and ownership−based (scope 1 emissions from firms and assets owned by individuals). Private ownership−based emissions (representing around 60% of total emissions) do not include government−owned or direct household emissions. The total volume of emissions covered by the ownership−based approach is relatively close to that explicitly accounted for in the consumption−based approach presented here. The latter assumes that emissions associated with government activities and investments, typically representing 30%–40% of total emissions are distribution−neutral (Bruckner et al. (2022)). Groups are defined by consumption−based emissions and wealth respectively, but both distributions are highly correlated. **Sources and series:** Bruckner et al. (2022) and Chancel and Rehm (2025b).\n\nmiddle-income groups, nearly all emissions are linked to essential consumption — transportation, heating, or electricity. For the top 10%, and especially for the top 1%, emissions from capital ownership dominate, accounting for 75–95% of their total footprint in France, Germany, and the United States. This also means that the wealthiest have a far greater capacity to reduce emissions without compromising their living standards. Taking a global view, net ownership positions reveal how investors in high-income countries profit from pollution abroad.\n\n**Figure 6.4** shows that major European economies, Japan, and South Korea exhibit large positive net ownership-emission positions. In France, adjusting for international investment raises national emissions by 36%, reflecting the fact that French investors own polluting production facilities abroad whose emissions exceed those generated domestically by foreign-owned firms. By contrast, many middle- and low-income countries show negative net positions: part\n\nof the emissions from their domestic production is ultimately attributable to foreign investors in richer countries. These patterns point to a crucial implication: climate regulations and taxation should take asset ownership into account. Evidently, stronger regulations on high-carbon investments are necessary. In addition, a carbon tax on wealth, based on the carbon content of owned assets or of investments, would arguably be significantly more progressive than a consumption-based levy. It would ensure that those who profit most from carbon-intensive activities, often across borders, contribute their fair share to the transition to a greener economy.\n\n119","pages":[{"line_numbers":[],"markdown":"Chapter 6. Climate, a Capital Problem\n\n**Figure 6.2. Emissions are highly concentrated among the rich, especially when looking at ownership**\n\n10%\n\n3%\n\n43%\n\n20%\n\n47%\n\n77%\n\n15%\n\n41%\n\n0%\n\n10% 20% 30% 40% 50% 60% 70% 80% 90%\n\nBottom 50% Middle 40% Top 10% Top 1%\n\nShare in total emissions\n\nConsumption Private ownership\n\n**Emissions shares by groups**\n\n**Interpretation.** The figure shows the share of global GHG emissions attributable to the bottom 50% and the top 1% of the world population. Emissions are separated into consumption−based (emissions from production attributed to final consumers) and ownership−based (scope 1 emissions from firms and assets owned by individuals). Private ownership−based emissions (representing around 60% of total emissions) do not include government−owned or direct household emissions. The total volume of emissions covered by the ownership−based approach is relatively close to that explicitly accounted for in the consumption−based approach presented here. The latter assumes that emissions associated with government activities and investments, typically representing 30%–40% of total emissions are distribution−neutral (Bruckner et al. (2022)). Groups are defined by consumption−based emissions and wealth respectively, but both distributions are highly correlated. **Sources and series:** Bruckner et al. (2022) and Chancel and Rehm (2025b).\n\nmiddle-income groups, nearly all emissions are linked to essential consumption — transportation, heating, or electricity. For the top 10%, and especially for the top 1%, emissions from capital ownership dominate, accounting for 75–95% of their total footprint in France, Germany, and the United States. This also means that the wealthiest have a far greater capacity to reduce emissions without compromising their living standards. Taking a global view, net ownership positions reveal how investors in high-income countries profit from pollution abroad.\n\n**Figure 6.4** shows that major European economies, Japan, and South Korea exhibit large positive net ownership-emission positions. In France, adjusting for international investment raises national emissions by 36%, reflecting the fact that French investors own polluting production facilities abroad whose emissions exceed those generated domestically by foreign-owned firms. By contrast, many middle- and low-income countries show negative net positions: part\n\nof the emissions from their domestic production is ultimately attributable to foreign investors in richer countries. These patterns point to a crucial implication: climate regulations and taxation should take asset ownership into account. Evidently, stronger regulations on high-carbon investments are necessary. In addition, a carbon tax on wealth, based on the carbon content of owned assets or of investments, would arguably be significantly more progressive than a consumption-based levy. It would ensure that those who profit most from carbon-intensive activities, often across borders, contribute their fair share to the transition to a greener economy.\n\n119","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Chapter 6. Climate, a Capital Problem\n\n**Figure 6.2. Emissions are highly concentrated among the rich, especially when looking at ownership**\n\n10%\n\n3%\n\n43%\n\n20%\n\n47%\n\n77%\n\n15%\n\n41%\n\n0%\n\n10% 20% 30% 40% 50% 60% 70% 80% 90%\n\nBottom 50% Middle 40% Top 10% Top 1%\n\nShare in total emissions\n\nConsumption Private ownership\n\n**Emissions shares by groups**\n\n**Interpretation.** The figure shows the share of global GHG emissions attributable to the bottom 50% and the top 1% of the world population. Emissions are separated into consumption−based (emissions from production attributed to final consumers) and ownership−based (scope 1 emissions from firms and assets owned by individuals). Private ownership−based emissions (representing around 60% of total emissions) do not include government−owned or direct household emissions. The total volume of emissions covered by the ownership−based approach is relatively close to that explicitly accounted for in the consumption−based approach presented here. The latter assumes that emissions associated with government activities and investments, typically representing 30%–40% of total emissions are distribution−neutral (Bruckner et al. (2022)). Groups are defined by consumption−based emissions and wealth respectively, but both distributions are highly correlated. **Sources and series:** Bruckner et al. (2022) and Chancel and Rehm (2025b).\n\nmiddle-income groups, nearly all emissions are linked to essential consumption — transportation, heating, or electricity. For the top 10%, and especially for the top 1%, emissions from capital ownership dominate, accounting for 75–95% of their total footprint in France, Germany, and the United States. This also means that the wealthiest have a far greater capacity to reduce emissions without compromising their living standards. Taking a global view, net ownership positions reveal how investors in high-income countries profit from pollution abroad.\n\n**Figure 6.4** shows that major European economies, Japan, and South Korea exhibit large positive net ownership-emission positions. In France, adjusting for international investment raises national emissions by 36%, reflecting the fact that French investors own polluting production facilities abroad whose emissions exceed those generated domestically by foreign-owned firms. By contrast, many middle- and low-income countries show negative net positions: part\n\nof the emissions from their domestic production is ultimately attributable to foreign investors in richer countries. These patterns point to a crucial implication: climate regulations and taxation should take asset ownership into account. Evidently, stronger regulations on high-carbon investments are necessary. In addition, a carbon tax on wealth, based on the carbon content of owned assets or of investments, would arguably be significantly more progressive than a consumption-based levy. It would ensure that those who profit most from carbon-intensive activities, often across borders, contribute their fair share to the transition to a greener economy.\n\n119","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p119.pdf"]},"response":{"errors":[],"extract_id":"extract_62b062ac4c12f080355eff8cb6bfb1bb","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Chapter 6. Climate, a Capital Problem\n\n**Figure 6.2. Emissions are highly concentrated among the rich, especially when looking at ownership**\n\n10%\n\n3%\n\n43%\n\n20%\n\n47%\n\n77%\n\n15%\n\n41%\n\n0%\n\n10% 20% 30% 40% 50% 60% 70% 80% 90%\n\nBottom 50% Middle 40% Top 10% Top 1%\n\nShare in total emissions\n\nConsumption Private ownership\n\n**Emissions shares by groups**\n\n**Interpretation.** The figure shows the share of global GHG emissions attributable to the bottom 50% and the top 1% of the world population. Emissions are separated into consumption−based (emissions from production attributed to final consumers) and ownership−based (scope 1 emissions from firms and assets owned by individuals). Private ownership−based emissions (representing around 60% of total emissions) do not include government−owned or direct household emissions. The total volume of emissions covered by the ownership−based approach is relatively close to that explicitly accounted for in the consumption−based approach presented here. The latter assumes that emissions associated with government activities and investments, typically representing 30%–40% of total emissions are distribution−neutral (Bruckner et al. (2022)). Groups are defined by consumption−based emissions and wealth respectively, but both distributions are highly correlated. **Sources and series:** Bruckner et al. (2022) and Chancel and Rehm (2025b).\n\nmiddle-income groups, nearly all emissions are linked to essential consumption — transportation, heating, or electricity. For the top 10%, and especially for the top 1%, emissions from capital ownership dominate, accounting for 75–95% of their total footprint in France, Germany, and the United States. This also means that the wealthiest have a far greater capacity to reduce emissions without compromising their living standards. Taking a global view, net ownership positions reveal how investors in high-income countries profit from pollution abroad.\n\n**Figure 6.4** shows that major European economies, Japan, and South Korea exhibit large positive net ownership-emission positions. In France, adjusting for international investment raises national emissions by 36%, reflecting the fact that French investors own polluting production facilities abroad whose emissions exceed those generated domestically by foreign-owned firms. By contrast, many middle- and low-income countries show negative net positions: part\n\nof the emissions from their domestic production is ultimately attributable to foreign investors in richer countries. These patterns point to a crucial implication: climate regulations and taxation should take asset ownership into account. Evidently, stronger regulations on high-carbon investments are necessary. In addition, a carbon tax on wealth, based on the carbon content of owned assets or of investments, would arguably be significantly more progressive than a consumption-based levy. It would ensure that those who profit most from carbon-intensive activities, often across borders, contribute their fair share to the transition to a greener economy.\n\n119"],"full_content":"Chapter 6. Climate, a Capital Problem\n\n**Figure 6.2. Emissions are highly concentrated among the rich, especially when looking at ownership**\n\n10%\n\n3%\n\n43%\n\n20%\n\n47%\n\n77%\n\n15%\n\n41%\n\n0%\n\n10% 20% 30% 40% 50% 60% 70% 80% 90%\n\nBottom 50% Middle 40% Top 10% Top 1%\n\nShare in total emissions\n\nConsumption Private ownership\n\n**Emissions shares by groups**\n\n**Interpretation.** The figure shows the share of global GHG emissions attributable to the bottom 50% and the top 1% of the world population. Emissions are separated into consumption−based (emissions from production attributed to final consumers) and ownership−based (scope 1 emissions from firms and assets owned by individuals). Private ownership−based emissions (representing around 60% of total emissions) do not include government−owned or direct household emissions. The total volume of emissions covered by the ownership−based approach is relatively close to that explicitly accounted for in the consumption−based approach presented here. The latter assumes that emissions associated with government activities and investments, typically representing 30%–40% of total emissions are distribution−neutral (Bruckner et al. (2022)). Groups are defined by consumption−based emissions and wealth respectively, but both distributions are highly correlated. **Sources and series:** Bruckner et al. (2022) and Chancel and Rehm (2025b).\n\nmiddle-income groups, nearly all emissions are linked to essential consumption — transportation, heating, or electricity. For the top 10%, and especially for the top 1%, emissions from capital ownership dominate, accounting for 75–95% of their total footprint in France, Germany, and the United States. This also means that the wealthiest have a far greater capacity to reduce emissions without compromising their living standards. Taking a global view, net ownership positions reveal how investors in high-income countries profit from pollution abroad.\n\n**Figure 6.4** shows that major European economies, Japan, and South Korea exhibit large positive net ownership-emission positions. In France, adjusting for international investment raises national emissions by 36%, reflecting the fact that French investors own polluting production facilities abroad whose emissions exceed those generated domestically by foreign-owned firms. By contrast, many middle- and low-income countries show negative net positions: part\n\nof the emissions from their domestic production is ultimately attributable to foreign investors in richer countries. These patterns point to a crucial implication: climate regulations and taxation should take asset ownership into account. Evidently, stronger regulations on high-carbon investments are necessary. In addition, a carbon tax on wealth, based on the carbon content of owned assets or of investments, would arguably be significantly more progressive than a consumption-based levy. It would ensure that those who profit most from carbon-intensive activities, often across borders, contribute their fair share to the transition to a greener economy.\n\n119","publish_date":null,"title":"Chapter 6. Climate, a Capital Problem Figure6.2. Emissionsarehighlyconcentratedamongtherich, especiallywhenlookingatownership middle-income groups, nearly all emissions of the emissions from their domestic","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p119.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p119.pdf"},"request":{"example_id":"chart/World_Inequality_Report_2026_p119","product_type":"parse","source_file_path":"World_Inequality_Report_2026_p119.pdf"},"started_at":"2026-11-04T19:45:22.571483"} +{"completed_at":"2026-11-04T19:45:26.303775","latency_in_ms":2661,"output":{"example_id":"chart/World_Inequality_Report_2026_p131","grounded_pages":[],"layout_pages":[],"markdown":"Chapter 7. Global Taxation of Multi-Millionaires\n\n**Figure 7.1. A more progressive tax system is needed in order to reduce political capture by the very rich**\n\n0% 2% 4% 6% 8%\n\n10% 12% 14% 16% 18% 20% 22% 24%\n\n1960\n\n1970\n\n1980\n\n1990\n\n2000\n\n2010\n\nCharitable donation share of\n\nthe top 0.01% in the U.S.\n\nUnited States\n\n**United States**\n\n0%\n\n10%\n\n20%\n\n30%\n\n40%\n\n50%\n\n60%\n\nP0−P10\n\nP10−P20\n\nP20−P30\n\nP30−P40\n\nP40−P50\n\nP50−P60\n\nP60−P70\n\nP70−P80\n\nP80−P90\n\nP90−P100\n\nPercentage of total political\n\ndonations by income decile\n\nFrance South Korea\n\n**France and South Korea**\n\n**Interpretation.** Private giving is increasingly concentrated at the top. In the U.S., the top 0.01% have starkly increased their share of charitable contributions since 1960, reaching more than 20% in 2012. In France and South Korea (2013–2021), political donations are dominated by the richest 10% who donate much more than any other income group. These patterns suggest rising top−end inequality translates into unequal influence over philanthropy and politics. **Sources and series:** Cagé (2024).\n\n**Donations in the U.S. (1960−2012) and France and South Korea (2013−2021)**\n\n**Regressivity at the top**\n\nOne of the most important facts highlighted in this report is that tax progressivity breaks down precisely where it should matter most: at the very top of the distribution. While tax systems are designed to appear progressive, effective rates often fall sharply for multi-millionaires and billionaires once we compare total taxes paid against their full economic income. Large parts of top incomes escape taxation.\n\n**Figure 7.2** illustrates one of the main paradoxes of modern tax systems: while tax codes in high-income countries are often designed to be progressive, they become regressive at the very top of the distribution (see Zucman (2024)). Regressivity emerges because the income tax fails at the top. In France and the Netherlands, billionaires’ effective income tax rate drops to near zero because of tax avoidance. In the United States, anti-avoidance rules keep billionaire rates somewhat higher, but they still fall sharply compared to upper-middle groups. Tax avoidance primarily operates through two channels (see Alstadsæter et al. (2023)\n\nand Zucman (2024)): (i) delaying or avoiding dividend distributions and capital gains realizations, and (ii) using holding companies and similar legal structures to accumulate earnings tax-free.\n\n**Figure 7.3** situates this regressivity in historical perspective. Over the past three decades, global multi-millionaire wealth has soared, tripling relative to world income (see Alstadsæter et al. (2023)). In 1995, the global top 0.001% held assets equivalent to 12% of global income. Three decades later, their share has nearly tripled, reaching 33% by 2025. Put differently, about sixty thousand individuals now control wealth worth one-third of the world’s combined income. Tax systems meant to fund public goods and reduce inequality instead reinforce concentration at the very top. Unless corrected, regressivity at the top will continue to erode both fiscal capacity and democratic cohesion.\n\n131","pages":[{"line_numbers":[],"markdown":"Chapter 7. Global Taxation of Multi-Millionaires\n\n**Figure 7.1. A more progressive tax system is needed in order to reduce political capture by the very rich**\n\n0% 2% 4% 6% 8%\n\n10% 12% 14% 16% 18% 20% 22% 24%\n\n1960\n\n1970\n\n1980\n\n1990\n\n2000\n\n2010\n\nCharitable donation share of\n\nthe top 0.01% in the U.S.\n\nUnited States\n\n**United States**\n\n0%\n\n10%\n\n20%\n\n30%\n\n40%\n\n50%\n\n60%\n\nP0−P10\n\nP10−P20\n\nP20−P30\n\nP30−P40\n\nP40−P50\n\nP50−P60\n\nP60−P70\n\nP70−P80\n\nP80−P90\n\nP90−P100\n\nPercentage of total political\n\ndonations by income decile\n\nFrance South Korea\n\n**France and South Korea**\n\n**Interpretation.** Private giving is increasingly concentrated at the top. In the U.S., the top 0.01% have starkly increased their share of charitable contributions since 1960, reaching more than 20% in 2012. In France and South Korea (2013–2021), political donations are dominated by the richest 10% who donate much more than any other income group. These patterns suggest rising top−end inequality translates into unequal influence over philanthropy and politics. **Sources and series:** Cagé (2024).\n\n**Donations in the U.S. (1960−2012) and France and South Korea (2013−2021)**\n\n**Regressivity at the top**\n\nOne of the most important facts highlighted in this report is that tax progressivity breaks down precisely where it should matter most: at the very top of the distribution. While tax systems are designed to appear progressive, effective rates often fall sharply for multi-millionaires and billionaires once we compare total taxes paid against their full economic income. Large parts of top incomes escape taxation.\n\n**Figure 7.2** illustrates one of the main paradoxes of modern tax systems: while tax codes in high-income countries are often designed to be progressive, they become regressive at the very top of the distribution (see Zucman (2024)). Regressivity emerges because the income tax fails at the top. In France and the Netherlands, billionaires’ effective income tax rate drops to near zero because of tax avoidance. In the United States, anti-avoidance rules keep billionaire rates somewhat higher, but they still fall sharply compared to upper-middle groups. Tax avoidance primarily operates through two channels (see Alstadsæter et al. (2023)\n\nand Zucman (2024)): (i) delaying or avoiding dividend distributions and capital gains realizations, and (ii) using holding companies and similar legal structures to accumulate earnings tax-free.\n\n**Figure 7.3** situates this regressivity in historical perspective. Over the past three decades, global multi-millionaire wealth has soared, tripling relative to world income (see Alstadsæter et al. (2023)). In 1995, the global top 0.001% held assets equivalent to 12% of global income. Three decades later, their share has nearly tripled, reaching 33% by 2025. Put differently, about sixty thousand individuals now control wealth worth one-third of the world’s combined income. Tax systems meant to fund public goods and reduce inequality instead reinforce concentration at the very top. Unless corrected, regressivity at the top will continue to erode both fiscal capacity and democratic cohesion.\n\n131","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Chapter 7. Global Taxation of Multi-Millionaires\n\n**Figure 7.1. A more progressive tax system is needed in order to reduce political capture by the very rich**\n\n0% 2% 4% 6% 8%\n\n10% 12% 14% 16% 18% 20% 22% 24%\n\n1960\n\n1970\n\n1980\n\n1990\n\n2000\n\n2010\n\nCharitable donation share of\n\nthe top 0.01% in the U.S.\n\nUnited States\n\n**United States**\n\n0%\n\n10%\n\n20%\n\n30%\n\n40%\n\n50%\n\n60%\n\nP0−P10\n\nP10−P20\n\nP20−P30\n\nP30−P40\n\nP40−P50\n\nP50−P60\n\nP60−P70\n\nP70−P80\n\nP80−P90\n\nP90−P100\n\nPercentage of total political\n\ndonations by income decile\n\nFrance South Korea\n\n**France and South Korea**\n\n**Interpretation.** Private giving is increasingly concentrated at the top. In the U.S., the top 0.01% have starkly increased their share of charitable contributions since 1960, reaching more than 20% in 2012. In France and South Korea (2013–2021), political donations are dominated by the richest 10% who donate much more than any other income group. These patterns suggest rising top−end inequality translates into unequal influence over philanthropy and politics. **Sources and series:** Cagé (2024).\n\n**Donations in the U.S. (1960−2012) and France and South Korea (2013−2021)**\n\n**Regressivity at the top**\n\nOne of the most important facts highlighted in this report is that tax progressivity breaks down precisely where it should matter most: at the very top of the distribution. While tax systems are designed to appear progressive, effective rates often fall sharply for multi-millionaires and billionaires once we compare total taxes paid against their full economic income. Large parts of top incomes escape taxation.\n\n**Figure 7.2** illustrates one of the main paradoxes of modern tax systems: while tax codes in high-income countries are often designed to be progressive, they become regressive at the very top of the distribution (see Zucman (2024)). Regressivity emerges because the income tax fails at the top. In France and the Netherlands, billionaires’ effective income tax rate drops to near zero because of tax avoidance. In the United States, anti-avoidance rules keep billionaire rates somewhat higher, but they still fall sharply compared to upper-middle groups. Tax avoidance primarily operates through two channels (see Alstadsæter et al. (2023)\n\nand Zucman (2024)): (i) delaying or avoiding dividend distributions and capital gains realizations, and (ii) using holding companies and similar legal structures to accumulate earnings tax-free.\n\n**Figure 7.3** situates this regressivity in historical perspective. Over the past three decades, global multi-millionaire wealth has soared, tripling relative to world income (see Alstadsæter et al. (2023)). In 1995, the global top 0.001% held assets equivalent to 12% of global income. Three decades later, their share has nearly tripled, reaching 33% by 2025. Put differently, about sixty thousand individuals now control wealth worth one-third of the world’s combined income. Tax systems meant to fund public goods and reduce inequality instead reinforce concentration at the very top. Unless corrected, regressivity at the top will continue to erode both fiscal capacity and democratic cohesion.\n\n131","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p131.pdf"]},"response":{"errors":[],"extract_id":"extract_7b848c1c13b8115f85c8786cb271a472","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Chapter 7. Global Taxation of Multi-Millionaires\n\n**Figure 7.1. A more progressive tax system is needed in order to reduce political capture by the very rich**\n\n0% 2% 4% 6% 8%\n\n10% 12% 14% 16% 18% 20% 22% 24%\n\n1960\n\n1970\n\n1980\n\n1990\n\n2000\n\n2010\n\nCharitable donation share of\n\nthe top 0.01% in the U.S.\n\nUnited States\n\n**United States**\n\n0%\n\n10%\n\n20%\n\n30%\n\n40%\n\n50%\n\n60%\n\nP0−P10\n\nP10−P20\n\nP20−P30\n\nP30−P40\n\nP40−P50\n\nP50−P60\n\nP60−P70\n\nP70−P80\n\nP80−P90\n\nP90−P100\n\nPercentage of total political\n\ndonations by income decile\n\nFrance South Korea\n\n**France and South Korea**\n\n**Interpretation.** Private giving is increasingly concentrated at the top. In the U.S., the top 0.01% have starkly increased their share of charitable contributions since 1960, reaching more than 20% in 2012. In France and South Korea (2013–2021), political donations are dominated by the richest 10% who donate much more than any other income group. These patterns suggest rising top−end inequality translates into unequal influence over philanthropy and politics. **Sources and series:** Cagé (2024).\n\n**Donations in the U.S. (1960−2012) and France and South Korea (2013−2021)**\n\n**Regressivity at the top**\n\nOne of the most important facts highlighted in this report is that tax progressivity breaks down precisely where it should matter most: at the very top of the distribution. While tax systems are designed to appear progressive, effective rates often fall sharply for multi-millionaires and billionaires once we compare total taxes paid against their full economic income. Large parts of top incomes escape taxation.\n\n**Figure 7.2** illustrates one of the main paradoxes of modern tax systems: while tax codes in high-income countries are often designed to be progressive, they become regressive at the very top of the distribution (see Zucman (2024)). Regressivity emerges because the income tax fails at the top. In France and the Netherlands, billionaires’ effective income tax rate drops to near zero because of tax avoidance. In the United States, anti-avoidance rules keep billionaire rates somewhat higher, but they still fall sharply compared to upper-middle groups. Tax avoidance primarily operates through two channels (see Alstadsæter et al. (2023)\n\nand Zucman (2024)): (i) delaying or avoiding dividend distributions and capital gains realizations, and (ii) using holding companies and similar legal structures to accumulate earnings tax-free.\n\n**Figure 7.3** situates this regressivity in historical perspective. Over the past three decades, global multi-millionaire wealth has soared, tripling relative to world income (see Alstadsæter et al. (2023)). In 1995, the global top 0.001% held assets equivalent to 12% of global income. Three decades later, their share has nearly tripled, reaching 33% by 2025. Put differently, about sixty thousand individuals now control wealth worth one-third of the world’s combined income. Tax systems meant to fund public goods and reduce inequality instead reinforce concentration at the very top. Unless corrected, regressivity at the top will continue to erode both fiscal capacity and democratic cohesion.\n\n131"],"full_content":"Chapter 7. Global Taxation of Multi-Millionaires\n\n**Figure 7.1. A more progressive tax system is needed in order to reduce political capture by the very rich**\n\n0% 2% 4% 6% 8%\n\n10% 12% 14% 16% 18% 20% 22% 24%\n\n1960\n\n1970\n\n1980\n\n1990\n\n2000\n\n2010\n\nCharitable donation share of\n\nthe top 0.01% in the U.S.\n\nUnited States\n\n**United States**\n\n0%\n\n10%\n\n20%\n\n30%\n\n40%\n\n50%\n\n60%\n\nP0−P10\n\nP10−P20\n\nP20−P30\n\nP30−P40\n\nP40−P50\n\nP50−P60\n\nP60−P70\n\nP70−P80\n\nP80−P90\n\nP90−P100\n\nPercentage of total political\n\ndonations by income decile\n\nFrance South Korea\n\n**France and South Korea**\n\n**Interpretation.** Private giving is increasingly concentrated at the top. In the U.S., the top 0.01% have starkly increased their share of charitable contributions since 1960, reaching more than 20% in 2012. In France and South Korea (2013–2021), political donations are dominated by the richest 10% who donate much more than any other income group. These patterns suggest rising top−end inequality translates into unequal influence over philanthropy and politics. **Sources and series:** Cagé (2024).\n\n**Donations in the U.S. (1960−2012) and France and South Korea (2013−2021)**\n\n**Regressivity at the top**\n\nOne of the most important facts highlighted in this report is that tax progressivity breaks down precisely where it should matter most: at the very top of the distribution. While tax systems are designed to appear progressive, effective rates often fall sharply for multi-millionaires and billionaires once we compare total taxes paid against their full economic income. Large parts of top incomes escape taxation.\n\n**Figure 7.2** illustrates one of the main paradoxes of modern tax systems: while tax codes in high-income countries are often designed to be progressive, they become regressive at the very top of the distribution (see Zucman (2024)). Regressivity emerges because the income tax fails at the top. In France and the Netherlands, billionaires’ effective income tax rate drops to near zero because of tax avoidance. In the United States, anti-avoidance rules keep billionaire rates somewhat higher, but they still fall sharply compared to upper-middle groups. Tax avoidance primarily operates through two channels (see Alstadsæter et al. (2023)\n\nand Zucman (2024)): (i) delaying or avoiding dividend distributions and capital gains realizations, and (ii) using holding companies and similar legal structures to accumulate earnings tax-free.\n\n**Figure 7.3** situates this regressivity in historical perspective. Over the past three decades, global multi-millionaire wealth has soared, tripling relative to world income (see Alstadsæter et al. (2023)). In 1995, the global top 0.001% held assets equivalent to 12% of global income. Three decades later, their share has nearly tripled, reaching 33% by 2025. Put differently, about sixty thousand individuals now control wealth worth one-third of the world’s combined income. Tax systems meant to fund public goods and reduce inequality instead reinforce concentration at the very top. Unless corrected, regressivity at the top will continue to erode both fiscal capacity and democratic cohesion.\n\n131","publish_date":null,"title":"Chapter 7. Global Taxation of Multi-Millionaires Figure7.1. Amoreprogressivetaxsystemisneededinorderto reducepoliticalcapturebytheveryrich Donations in the U.S. (1960−2012) and France and South Korea (2013−2021)","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p131.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p131.pdf"},"request":{"example_id":"chart/World_Inequality_Report_2026_p131","product_type":"parse","source_file_path":"World_Inequality_Report_2026_p131.pdf"},"started_at":"2026-11-04T19:45:23.641947"} +{"completed_at":"2026-11-04T19:45:26.976705","latency_in_ms":1468,"output":{"example_id":"chart/World_Inequality_Report_2026_p132","grounded_pages":[],"layout_pages":[],"markdown":"Chapter 7. Global Taxation of Multi-Millionaires\n\n**Figure 7.2. The super-rich pay proportionately less than others**\n\n99% of the population are located in the shaded area.\n\n0%\n\n5%\n\n10%\n\n15%\n\n20%\n\n25%\n\nP0−10\n\nP10−20\n\nP20−30\n\nP30−40\n\nP40−50\n\nP50−60\n\nP60−70\n\nP70−80\n\nP80−90\n\nP90−95\n\nP95−99\n\nP99−99.9\n\nP99.9−99.99\n\nP99.99−P99.999\n\nP99.999−P99.9999\n\nBillionaires\n\nEffective income tax rates by\n\nincome groups and for\n\nbillionaires (% of pre−tax\n\nincome)\n\nBrazil France Netherlands Spain United States\n\n**Interpretation.** This figure shows effective income tax rates by pre−tax income group and for U.S. dollar billionaires in Brazil, France, the Netherlands, Spain, and the United States. Income tax rates include only individual income taxes and equivalent levies. All values are expressed as a share of pre−tax income, defined as all national income before taxes and transfers, after pensions. P0−10 denotes the bottom 10% of the income distribution, P10−20 the next decile, etc. **Sources and series:** Artola et al. (2022), Bozio et al. (2024), Bozio et al. (2020), Bruil et al. (2024), Palomo et al. (2025), Saez and Zucman (2019), and Zucman (2024).\n\n**Effective income tax rates by income groups**\n\n**Safeguarding progressivity at the top**\n\nIf today’s tax systems collapse into regressivity at the very top, the crucial question to ask is: how do we restore fairness? **Figure 7.4** provides a straightforward answer: introduce a minimum wealth tax on centi-millionaires and billionaires (see Zucman (2024)). The figure simulates what would happen if governments in high-income countries implemented a tax of 1%, 2%, or 3% of wealth annually. The results are striking. With no reform, billionaires face effective tax rates around 20%, well below the burden of households with lower incomes. A 1% wealth tax would modestly increase their contribution, but regressivity would remain. At 2%, the decline is essentially neutralized; centi-millionaires and billionaires would be brought back up to roughly the same tax burden as other taxpayers. At 3%, the system becomes modestly progressive again, with the very richest paying more than the rest. A 2% tax rate is the minimum benchmark for safeguarding non-regressivity. This proposal builds on work by the EU Tax Observatory and the report titled “ _A_\n\n_Blueprint for a Coordinated Minimum Effective Taxation Standard for Ultra-High-Net-Worth Individuals_ ”, prepared by Gabriel Zucman and commissioned by the Brazilian G20 presidency in 2024. The report shows that such a measure is technically feasible, administratively manageable, and politically transformative. A 2% minimum tax on global billionaires could raise between $200 and $250 billion annually from about 3,000 individuals worldwide, funds equivalent to the entire health budgets of many low-income countries combined. Momentum for this idea has accelerated in several countries. Some examples are Brazil, South Africa, Spain, and France. Brazil placed the billionaire tax on the G20 agenda during its 2024 presidency. Spain has also taken a leadership role internationally, co-launching with Brazil and South Africa in 2025 a platform at the UN to build support for a global billionaire tax. France debated a 2% tax on fortunes above €100 million earlier in 2025: the National Assembly approved it, but the Senate rejected the bill. The issue in France is at the center of the political debate. In the United States, Senator Elizabeth Warren and others continue to\n\n132","pages":[{"line_numbers":[],"markdown":"Chapter 7. Global Taxation of Multi-Millionaires\n\n**Figure 7.2. The super-rich pay proportionately less than others**\n\n99% of the population are located in the shaded area.\n\n0%\n\n5%\n\n10%\n\n15%\n\n20%\n\n25%\n\nP0−10\n\nP10−20\n\nP20−30\n\nP30−40\n\nP40−50\n\nP50−60\n\nP60−70\n\nP70−80\n\nP80−90\n\nP90−95\n\nP95−99\n\nP99−99.9\n\nP99.9−99.99\n\nP99.99−P99.999\n\nP99.999−P99.9999\n\nBillionaires\n\nEffective income tax rates by\n\nincome groups and for\n\nbillionaires (% of pre−tax\n\nincome)\n\nBrazil France Netherlands Spain United States\n\n**Interpretation.** This figure shows effective income tax rates by pre−tax income group and for U.S. dollar billionaires in Brazil, France, the Netherlands, Spain, and the United States. Income tax rates include only individual income taxes and equivalent levies. All values are expressed as a share of pre−tax income, defined as all national income before taxes and transfers, after pensions. P0−10 denotes the bottom 10% of the income distribution, P10−20 the next decile, etc. **Sources and series:** Artola et al. (2022), Bozio et al. (2024), Bozio et al. (2020), Bruil et al. (2024), Palomo et al. (2025), Saez and Zucman (2019), and Zucman (2024).\n\n**Effective income tax rates by income groups**\n\n**Safeguarding progressivity at the top**\n\nIf today’s tax systems collapse into regressivity at the very top, the crucial question to ask is: how do we restore fairness? **Figure 7.4** provides a straightforward answer: introduce a minimum wealth tax on centi-millionaires and billionaires (see Zucman (2024)). The figure simulates what would happen if governments in high-income countries implemented a tax of 1%, 2%, or 3% of wealth annually. The results are striking. With no reform, billionaires face effective tax rates around 20%, well below the burden of households with lower incomes. A 1% wealth tax would modestly increase their contribution, but regressivity would remain. At 2%, the decline is essentially neutralized; centi-millionaires and billionaires would be brought back up to roughly the same tax burden as other taxpayers. At 3%, the system becomes modestly progressive again, with the very richest paying more than the rest. A 2% tax rate is the minimum benchmark for safeguarding non-regressivity. This proposal builds on work by the EU Tax Observatory and the report titled “ _A_\n\n_Blueprint for a Coordinated Minimum Effective Taxation Standard for Ultra-High-Net-Worth Individuals_ ”, prepared by Gabriel Zucman and commissioned by the Brazilian G20 presidency in 2024. The report shows that such a measure is technically feasible, administratively manageable, and politically transformative. A 2% minimum tax on global billionaires could raise between $200 and $250 billion annually from about 3,000 individuals worldwide, funds equivalent to the entire health budgets of many low-income countries combined. Momentum for this idea has accelerated in several countries. Some examples are Brazil, South Africa, Spain, and France. Brazil placed the billionaire tax on the G20 agenda during its 2024 presidency. Spain has also taken a leadership role internationally, co-launching with Brazil and South Africa in 2025 a platform at the UN to build support for a global billionaire tax. France debated a 2% tax on fortunes above €100 million earlier in 2025: the National Assembly approved it, but the Senate rejected the bill. The issue in France is at the center of the political debate. In the United States, Senator Elizabeth Warren and others continue to\n\n132","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Chapter 7. Global Taxation of Multi-Millionaires\n\n**Figure 7.2. The super-rich pay proportionately less than others**\n\n99% of the population are located in the shaded area.\n\n0%\n\n5%\n\n10%\n\n15%\n\n20%\n\n25%\n\nP0−10\n\nP10−20\n\nP20−30\n\nP30−40\n\nP40−50\n\nP50−60\n\nP60−70\n\nP70−80\n\nP80−90\n\nP90−95\n\nP95−99\n\nP99−99.9\n\nP99.9−99.99\n\nP99.99−P99.999\n\nP99.999−P99.9999\n\nBillionaires\n\nEffective income tax rates by\n\nincome groups and for\n\nbillionaires (% of pre−tax\n\nincome)\n\nBrazil France Netherlands Spain United States\n\n**Interpretation.** This figure shows effective income tax rates by pre−tax income group and for U.S. dollar billionaires in Brazil, France, the Netherlands, Spain, and the United States. Income tax rates include only individual income taxes and equivalent levies. All values are expressed as a share of pre−tax income, defined as all national income before taxes and transfers, after pensions. P0−10 denotes the bottom 10% of the income distribution, P10−20 the next decile, etc. **Sources and series:** Artola et al. (2022), Bozio et al. (2024), Bozio et al. (2020), Bruil et al. (2024), Palomo et al. (2025), Saez and Zucman (2019), and Zucman (2024).\n\n**Effective income tax rates by income groups**\n\n**Safeguarding progressivity at the top**\n\nIf today’s tax systems collapse into regressivity at the very top, the crucial question to ask is: how do we restore fairness? **Figure 7.4** provides a straightforward answer: introduce a minimum wealth tax on centi-millionaires and billionaires (see Zucman (2024)). The figure simulates what would happen if governments in high-income countries implemented a tax of 1%, 2%, or 3% of wealth annually. The results are striking. With no reform, billionaires face effective tax rates around 20%, well below the burden of households with lower incomes. A 1% wealth tax would modestly increase their contribution, but regressivity would remain. At 2%, the decline is essentially neutralized; centi-millionaires and billionaires would be brought back up to roughly the same tax burden as other taxpayers. At 3%, the system becomes modestly progressive again, with the very richest paying more than the rest. A 2% tax rate is the minimum benchmark for safeguarding non-regressivity. This proposal builds on work by the EU Tax Observatory and the report titled “ _A_\n\n_Blueprint for a Coordinated Minimum Effective Taxation Standard for Ultra-High-Net-Worth Individuals_ ”, prepared by Gabriel Zucman and commissioned by the Brazilian G20 presidency in 2024. The report shows that such a measure is technically feasible, administratively manageable, and politically transformative. A 2% minimum tax on global billionaires could raise between $200 and $250 billion annually from about 3,000 individuals worldwide, funds equivalent to the entire health budgets of many low-income countries combined. Momentum for this idea has accelerated in several countries. Some examples are Brazil, South Africa, Spain, and France. Brazil placed the billionaire tax on the G20 agenda during its 2024 presidency. Spain has also taken a leadership role internationally, co-launching with Brazil and South Africa in 2025 a platform at the UN to build support for a global billionaire tax. France debated a 2% tax on fortunes above €100 million earlier in 2025: the National Assembly approved it, but the Senate rejected the bill. The issue in France is at the center of the political debate. In the United States, Senator Elizabeth Warren and others continue to\n\n132","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p132.pdf"]},"response":{"errors":[],"extract_id":"extract_2c44b8a5abe7a345ab8cd0c5128010ad","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Chapter 7. Global Taxation of Multi-Millionaires\n\n**Figure 7.2. The super-rich pay proportionately less than others**\n\n99% of the population are located in the shaded area.\n\n0%\n\n5%\n\n10%\n\n15%\n\n20%\n\n25%\n\nP0−10\n\nP10−20\n\nP20−30\n\nP30−40\n\nP40−50\n\nP50−60\n\nP60−70\n\nP70−80\n\nP80−90\n\nP90−95\n\nP95−99\n\nP99−99.9\n\nP99.9−99.99\n\nP99.99−P99.999\n\nP99.999−P99.9999\n\nBillionaires\n\nEffective income tax rates by\n\nincome groups and for\n\nbillionaires (% of pre−tax\n\nincome)\n\nBrazil France Netherlands Spain United States\n\n**Interpretation.** This figure shows effective income tax rates by pre−tax income group and for U.S. dollar billionaires in Brazil, France, the Netherlands, Spain, and the United States. Income tax rates include only individual income taxes and equivalent levies. All values are expressed as a share of pre−tax income, defined as all national income before taxes and transfers, after pensions. P0−10 denotes the bottom 10% of the income distribution, P10−20 the next decile, etc. **Sources and series:** Artola et al. (2022), Bozio et al. (2024), Bozio et al. (2020), Bruil et al. (2024), Palomo et al. (2025), Saez and Zucman (2019), and Zucman (2024).\n\n**Effective income tax rates by income groups**\n\n**Safeguarding progressivity at the top**\n\nIf today’s tax systems collapse into regressivity at the very top, the crucial question to ask is: how do we restore fairness? **Figure 7.4** provides a straightforward answer: introduce a minimum wealth tax on centi-millionaires and billionaires (see Zucman (2024)). The figure simulates what would happen if governments in high-income countries implemented a tax of 1%, 2%, or 3% of wealth annually. The results are striking. With no reform, billionaires face effective tax rates around 20%, well below the burden of households with lower incomes. A 1% wealth tax would modestly increase their contribution, but regressivity would remain. At 2%, the decline is essentially neutralized; centi-millionaires and billionaires would be brought back up to roughly the same tax burden as other taxpayers. At 3%, the system becomes modestly progressive again, with the very richest paying more than the rest. A 2% tax rate is the minimum benchmark for safeguarding non-regressivity. This proposal builds on work by the EU Tax Observatory and the report titled “ _A_\n\n_Blueprint for a Coordinated Minimum Effective Taxation Standard for Ultra-High-Net-Worth Individuals_ ”, prepared by Gabriel Zucman and commissioned by the Brazilian G20 presidency in 2024. The report shows that such a measure is technically feasible, administratively manageable, and politically transformative. A 2% minimum tax on global billionaires could raise between $200 and $250 billion annually from about 3,000 individuals worldwide, funds equivalent to the entire health budgets of many low-income countries combined. Momentum for this idea has accelerated in several countries. Some examples are Brazil, South Africa, Spain, and France. Brazil placed the billionaire tax on the G20 agenda during its 2024 presidency. Spain has also taken a leadership role internationally, co-launching with Brazil and South Africa in 2025 a platform at the UN to build support for a global billionaire tax. France debated a 2% tax on fortunes above €100 million earlier in 2025: the National Assembly approved it, but the Senate rejected the bill. The issue in France is at the center of the political debate. In the United States, Senator Elizabeth Warren and others continue to\n\n132"],"full_content":"Chapter 7. Global Taxation of Multi-Millionaires\n\n**Figure 7.2. The super-rich pay proportionately less than others**\n\n99% of the population are located in the shaded area.\n\n0%\n\n5%\n\n10%\n\n15%\n\n20%\n\n25%\n\nP0−10\n\nP10−20\n\nP20−30\n\nP30−40\n\nP40−50\n\nP50−60\n\nP60−70\n\nP70−80\n\nP80−90\n\nP90−95\n\nP95−99\n\nP99−99.9\n\nP99.9−99.99\n\nP99.99−P99.999\n\nP99.999−P99.9999\n\nBillionaires\n\nEffective income tax rates by\n\nincome groups and for\n\nbillionaires (% of pre−tax\n\nincome)\n\nBrazil France Netherlands Spain United States\n\n**Interpretation.** This figure shows effective income tax rates by pre−tax income group and for U.S. dollar billionaires in Brazil, France, the Netherlands, Spain, and the United States. Income tax rates include only individual income taxes and equivalent levies. All values are expressed as a share of pre−tax income, defined as all national income before taxes and transfers, after pensions. P0−10 denotes the bottom 10% of the income distribution, P10−20 the next decile, etc. **Sources and series:** Artola et al. (2022), Bozio et al. (2024), Bozio et al. (2020), Bruil et al. (2024), Palomo et al. (2025), Saez and Zucman (2019), and Zucman (2024).\n\n**Effective income tax rates by income groups**\n\n**Safeguarding progressivity at the top**\n\nIf today’s tax systems collapse into regressivity at the very top, the crucial question to ask is: how do we restore fairness? **Figure 7.4** provides a straightforward answer: introduce a minimum wealth tax on centi-millionaires and billionaires (see Zucman (2024)). The figure simulates what would happen if governments in high-income countries implemented a tax of 1%, 2%, or 3% of wealth annually. The results are striking. With no reform, billionaires face effective tax rates around 20%, well below the burden of households with lower incomes. A 1% wealth tax would modestly increase their contribution, but regressivity would remain. At 2%, the decline is essentially neutralized; centi-millionaires and billionaires would be brought back up to roughly the same tax burden as other taxpayers. At 3%, the system becomes modestly progressive again, with the very richest paying more than the rest. A 2% tax rate is the minimum benchmark for safeguarding non-regressivity. This proposal builds on work by the EU Tax Observatory and the report titled “ _A_\n\n_Blueprint for a Coordinated Minimum Effective Taxation Standard for Ultra-High-Net-Worth Individuals_ ”, prepared by Gabriel Zucman and commissioned by the Brazilian G20 presidency in 2024. The report shows that such a measure is technically feasible, administratively manageable, and politically transformative. A 2% minimum tax on global billionaires could raise between $200 and $250 billion annually from about 3,000 individuals worldwide, funds equivalent to the entire health budgets of many low-income countries combined. Momentum for this idea has accelerated in several countries. Some examples are Brazil, South Africa, Spain, and France. Brazil placed the billionaire tax on the G20 agenda during its 2024 presidency. Spain has also taken a leadership role internationally, co-launching with Brazil and South Africa in 2025 a platform at the UN to build support for a global billionaire tax. France debated a 2% tax on fortunes above €100 million earlier in 2025: the National Assembly approved it, but the Senate rejected the bill. The issue in France is at the center of the political debate. In the United States, Senator Elizabeth Warren and others continue to\n\n132","publish_date":null,"title":"Chapter 7. Global Taxation of Multi-Millionaires Figure7.2. Thesuper-richpayproportionatelylessthanothers Effective income tax rates by income groups Blueprint for a Coordinated Minimum","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p132.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p132.pdf"},"request":{"example_id":"chart/World_Inequality_Report_2026_p132","product_type":"parse","source_file_path":"World_Inequality_Report_2026_p132.pdf"},"started_at":"2026-11-04T19:45:25.507723"} +{"completed_at":"2026-11-04T19:45:27.694487","latency_in_ms":1386,"output":{"example_id":"chart/World_Inequality_Report_2026_p145","grounded_pages":[],"layout_pages":[],"markdown":"Chapter 8. Political Cleavages\n\n**Figure 8.1. Working class representation has always been low and has further deteriorated in recent decades**\n\n0% 2% 4% 6% 8%\n\n10% 12% 14% 16% 18% 20%\n\n1900\n\n1905\n\n1910\n\n1915\n\n1920\n\n1925\n\n1930\n\n1935\n\n1940\n\n1945\n\n1950\n\n1955\n\n1960\n\n1965\n\n1970\n\n1975\n\n1980\n\n1985\n\n1990\n\n1995\n\n2000\n\n2005\n\n2010\n\n2015\n\n2020\n\n2025\n\nShare of working−class MPs (%)\n\nFrance United Kingdom United States\n\n**Working class representation in Western democracies, 1900−2025**\n\n**Interpretation.** The long−run decline in the share of working−class members of parliament (MPs): evidence from France, the United Kingdom, and the United States. The figure plots the evolution of the share of working−class MPs—measured as the number of MPs whose former occupation just before the elections was a “manual” occupation (United Kingdom), a “blue−collar” occupation (United States), or an occupation as “employés et ouvriers” (France)—over the total number of MPs in each country. The share of working−class occupations in the total labor force is usually around 50%–60% or more. **Sources and series:** Cagé (2024).\n\nwith the bottom 90% has remained negative: top-income voters are consistently less likely to support the left. This transition highlights a shift from “class-based” to “multi-elite” party systems, which now feature opposing camps comprising an educated “Brahmin left” and an affluent “merchant right” (see Gethin, Martínez-Toledano, and Piketty (2022);\n\nGethin, Martínez-Toledano, and Piketty\n\n(2021); Gethin and Martínez-Toledano\n\n(2025)). This structural transformation is closely related to educational expansion and the ensuing “complexification” of the occupational structure. By way of illustration, many high-degree but relatively low-income voters (e.g., teachers or nurses) currently vote for the left, while many voters with lower degrees but relatively higher income (e.g., self-employed or truck drivers) tend to vote for the right. Importantly, this transformation has not been restricted to a few countries. It has unfolded across almost all twenty-one Western democracies included in the _World Political Cleavages and Inequality_\n\n_Database_ , despite wide differences in their histories, institutions, and party structures. This common transformation has been closely linked not only to the rise of a new sociocultural axis of political conflict, but also to the convergence of parties on economic policy, political fragmentation, economic development, and educational progress (see Gethin and Martínez-Toledano (2025)).\n\n**Figure 8.3** documents the remarkable reversal of educational divides across different Western democracies. In the postwar decades, voters with low levels of education were the backbone of left-wing parties, while the highly educated leaned right. Today, the opposite holds: highly educated voters now disproportionately vote more for the left, while the less educated often support conservative parties. This reversal has transformed left-wing parties from representing the industrial working class in the past into coalitions anchored in the educated middle and upper-middle classes in the present. These left parties increasingly attract highly educated “sociocultural professionals”, such\n\n145","pages":[{"line_numbers":[],"markdown":"Chapter 8. Political Cleavages\n\n**Figure 8.1. Working class representation has always been low and has further deteriorated in recent decades**\n\n0% 2% 4% 6% 8%\n\n10% 12% 14% 16% 18% 20%\n\n1900\n\n1905\n\n1910\n\n1915\n\n1920\n\n1925\n\n1930\n\n1935\n\n1940\n\n1945\n\n1950\n\n1955\n\n1960\n\n1965\n\n1970\n\n1975\n\n1980\n\n1985\n\n1990\n\n1995\n\n2000\n\n2005\n\n2010\n\n2015\n\n2020\n\n2025\n\nShare of working−class MPs (%)\n\nFrance United Kingdom United States\n\n**Working class representation in Western democracies, 1900−2025**\n\n**Interpretation.** The long−run decline in the share of working−class members of parliament (MPs): evidence from France, the United Kingdom, and the United States. The figure plots the evolution of the share of working−class MPs—measured as the number of MPs whose former occupation just before the elections was a “manual” occupation (United Kingdom), a “blue−collar” occupation (United States), or an occupation as “employés et ouvriers” (France)—over the total number of MPs in each country. The share of working−class occupations in the total labor force is usually around 50%–60% or more. **Sources and series:** Cagé (2024).\n\nwith the bottom 90% has remained negative: top-income voters are consistently less likely to support the left. This transition highlights a shift from “class-based” to “multi-elite” party systems, which now feature opposing camps comprising an educated “Brahmin left” and an affluent “merchant right” (see Gethin, Martínez-Toledano, and Piketty (2022);\n\nGethin, Martínez-Toledano, and Piketty\n\n(2021); Gethin and Martínez-Toledano\n\n(2025)). This structural transformation is closely related to educational expansion and the ensuing “complexification” of the occupational structure. By way of illustration, many high-degree but relatively low-income voters (e.g., teachers or nurses) currently vote for the left, while many voters with lower degrees but relatively higher income (e.g., self-employed or truck drivers) tend to vote for the right. Importantly, this transformation has not been restricted to a few countries. It has unfolded across almost all twenty-one Western democracies included in the _World Political Cleavages and Inequality_\n\n_Database_ , despite wide differences in their histories, institutions, and party structures. This common transformation has been closely linked not only to the rise of a new sociocultural axis of political conflict, but also to the convergence of parties on economic policy, political fragmentation, economic development, and educational progress (see Gethin and Martínez-Toledano (2025)).\n\n**Figure 8.3** documents the remarkable reversal of educational divides across different Western democracies. In the postwar decades, voters with low levels of education were the backbone of left-wing parties, while the highly educated leaned right. Today, the opposite holds: highly educated voters now disproportionately vote more for the left, while the less educated often support conservative parties. This reversal has transformed left-wing parties from representing the industrial working class in the past into coalitions anchored in the educated middle and upper-middle classes in the present. These left parties increasingly attract highly educated “sociocultural professionals”, such\n\n145","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Chapter 8. Political Cleavages\n\n**Figure 8.1. Working class representation has always been low and has further deteriorated in recent decades**\n\n0% 2% 4% 6% 8%\n\n10% 12% 14% 16% 18% 20%\n\n1900\n\n1905\n\n1910\n\n1915\n\n1920\n\n1925\n\n1930\n\n1935\n\n1940\n\n1945\n\n1950\n\n1955\n\n1960\n\n1965\n\n1970\n\n1975\n\n1980\n\n1985\n\n1990\n\n1995\n\n2000\n\n2005\n\n2010\n\n2015\n\n2020\n\n2025\n\nShare of working−class MPs (%)\n\nFrance United Kingdom United States\n\n**Working class representation in Western democracies, 1900−2025**\n\n**Interpretation.** The long−run decline in the share of working−class members of parliament (MPs): evidence from France, the United Kingdom, and the United States. The figure plots the evolution of the share of working−class MPs—measured as the number of MPs whose former occupation just before the elections was a “manual” occupation (United Kingdom), a “blue−collar” occupation (United States), or an occupation as “employés et ouvriers” (France)—over the total number of MPs in each country. The share of working−class occupations in the total labor force is usually around 50%–60% or more. **Sources and series:** Cagé (2024).\n\nwith the bottom 90% has remained negative: top-income voters are consistently less likely to support the left. This transition highlights a shift from “class-based” to “multi-elite” party systems, which now feature opposing camps comprising an educated “Brahmin left” and an affluent “merchant right” (see Gethin, Martínez-Toledano, and Piketty (2022);\n\nGethin, Martínez-Toledano, and Piketty\n\n(2021); Gethin and Martínez-Toledano\n\n(2025)). This structural transformation is closely related to educational expansion and the ensuing “complexification” of the occupational structure. By way of illustration, many high-degree but relatively low-income voters (e.g., teachers or nurses) currently vote for the left, while many voters with lower degrees but relatively higher income (e.g., self-employed or truck drivers) tend to vote for the right. Importantly, this transformation has not been restricted to a few countries. It has unfolded across almost all twenty-one Western democracies included in the _World Political Cleavages and Inequality_\n\n_Database_ , despite wide differences in their histories, institutions, and party structures. This common transformation has been closely linked not only to the rise of a new sociocultural axis of political conflict, but also to the convergence of parties on economic policy, political fragmentation, economic development, and educational progress (see Gethin and Martínez-Toledano (2025)).\n\n**Figure 8.3** documents the remarkable reversal of educational divides across different Western democracies. In the postwar decades, voters with low levels of education were the backbone of left-wing parties, while the highly educated leaned right. Today, the opposite holds: highly educated voters now disproportionately vote more for the left, while the less educated often support conservative parties. This reversal has transformed left-wing parties from representing the industrial working class in the past into coalitions anchored in the educated middle and upper-middle classes in the present. These left parties increasingly attract highly educated “sociocultural professionals”, such\n\n145","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p145.pdf"]},"response":{"errors":[],"extract_id":"extract_1b4fa94345e1eee2015356fc039423f0","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Chapter 8. Political Cleavages\n\n**Figure 8.1. Working class representation has always been low and has further deteriorated in recent decades**\n\n0% 2% 4% 6% 8%\n\n10% 12% 14% 16% 18% 20%\n\n1900\n\n1905\n\n1910\n\n1915\n\n1920\n\n1925\n\n1930\n\n1935\n\n1940\n\n1945\n\n1950\n\n1955\n\n1960\n\n1965\n\n1970\n\n1975\n\n1980\n\n1985\n\n1990\n\n1995\n\n2000\n\n2005\n\n2010\n\n2015\n\n2020\n\n2025\n\nShare of working−class MPs (%)\n\nFrance United Kingdom United States\n\n**Working class representation in Western democracies, 1900−2025**\n\n**Interpretation.** The long−run decline in the share of working−class members of parliament (MPs): evidence from France, the United Kingdom, and the United States. The figure plots the evolution of the share of working−class MPs—measured as the number of MPs whose former occupation just before the elections was a “manual” occupation (United Kingdom), a “blue−collar” occupation (United States), or an occupation as “employés et ouvriers” (France)—over the total number of MPs in each country. The share of working−class occupations in the total labor force is usually around 50%–60% or more. **Sources and series:** Cagé (2024).\n\nwith the bottom 90% has remained negative: top-income voters are consistently less likely to support the left. This transition highlights a shift from “class-based” to “multi-elite” party systems, which now feature opposing camps comprising an educated “Brahmin left” and an affluent “merchant right” (see Gethin, Martínez-Toledano, and Piketty (2022);\n\nGethin, Martínez-Toledano, and Piketty\n\n(2021); Gethin and Martínez-Toledano\n\n(2025)). This structural transformation is closely related to educational expansion and the ensuing “complexification” of the occupational structure. By way of illustration, many high-degree but relatively low-income voters (e.g., teachers or nurses) currently vote for the left, while many voters with lower degrees but relatively higher income (e.g., self-employed or truck drivers) tend to vote for the right. Importantly, this transformation has not been restricted to a few countries. It has unfolded across almost all twenty-one Western democracies included in the _World Political Cleavages and Inequality_\n\n_Database_ , despite wide differences in their histories, institutions, and party structures. This common transformation has been closely linked not only to the rise of a new sociocultural axis of political conflict, but also to the convergence of parties on economic policy, political fragmentation, economic development, and educational progress (see Gethin and Martínez-Toledano (2025)).\n\n**Figure 8.3** documents the remarkable reversal of educational divides across different Western democracies. In the postwar decades, voters with low levels of education were the backbone of left-wing parties, while the highly educated leaned right. Today, the opposite holds: highly educated voters now disproportionately vote more for the left, while the less educated often support conservative parties. This reversal has transformed left-wing parties from representing the industrial working class in the past into coalitions anchored in the educated middle and upper-middle classes in the present. These left parties increasingly attract highly educated “sociocultural professionals”, such\n\n145"],"full_content":"Chapter 8. Political Cleavages\n\n**Figure 8.1. Working class representation has always been low and has further deteriorated in recent decades**\n\n0% 2% 4% 6% 8%\n\n10% 12% 14% 16% 18% 20%\n\n1900\n\n1905\n\n1910\n\n1915\n\n1920\n\n1925\n\n1930\n\n1935\n\n1940\n\n1945\n\n1950\n\n1955\n\n1960\n\n1965\n\n1970\n\n1975\n\n1980\n\n1985\n\n1990\n\n1995\n\n2000\n\n2005\n\n2010\n\n2015\n\n2020\n\n2025\n\nShare of working−class MPs (%)\n\nFrance United Kingdom United States\n\n**Working class representation in Western democracies, 1900−2025**\n\n**Interpretation.** The long−run decline in the share of working−class members of parliament (MPs): evidence from France, the United Kingdom, and the United States. The figure plots the evolution of the share of working−class MPs—measured as the number of MPs whose former occupation just before the elections was a “manual” occupation (United Kingdom), a “blue−collar” occupation (United States), or an occupation as “employés et ouvriers” (France)—over the total number of MPs in each country. The share of working−class occupations in the total labor force is usually around 50%–60% or more. **Sources and series:** Cagé (2024).\n\nwith the bottom 90% has remained negative: top-income voters are consistently less likely to support the left. This transition highlights a shift from “class-based” to “multi-elite” party systems, which now feature opposing camps comprising an educated “Brahmin left” and an affluent “merchant right” (see Gethin, Martínez-Toledano, and Piketty (2022);\n\nGethin, Martínez-Toledano, and Piketty\n\n(2021); Gethin and Martínez-Toledano\n\n(2025)). This structural transformation is closely related to educational expansion and the ensuing “complexification” of the occupational structure. By way of illustration, many high-degree but relatively low-income voters (e.g., teachers or nurses) currently vote for the left, while many voters with lower degrees but relatively higher income (e.g., self-employed or truck drivers) tend to vote for the right. Importantly, this transformation has not been restricted to a few countries. It has unfolded across almost all twenty-one Western democracies included in the _World Political Cleavages and Inequality_\n\n_Database_ , despite wide differences in their histories, institutions, and party structures. This common transformation has been closely linked not only to the rise of a new sociocultural axis of political conflict, but also to the convergence of parties on economic policy, political fragmentation, economic development, and educational progress (see Gethin and Martínez-Toledano (2025)).\n\n**Figure 8.3** documents the remarkable reversal of educational divides across different Western democracies. In the postwar decades, voters with low levels of education were the backbone of left-wing parties, while the highly educated leaned right. Today, the opposite holds: highly educated voters now disproportionately vote more for the left, while the less educated often support conservative parties. This reversal has transformed left-wing parties from representing the industrial working class in the past into coalitions anchored in the educated middle and upper-middle classes in the present. These left parties increasingly attract highly educated “sociocultural professionals”, such\n\n145","publish_date":null,"title":"Chapter 8. Political Cleavages Figure8.1. Workingclassrepresentationhasalwaysbeenlow andhasfurtherdeterioratedinrecentdecades Working class representation in Western democracies, 1900−2025 withthebottom90%hasremainednegative: , despite wide differences in","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p145.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p145.pdf"},"request":{"example_id":"chart/World_Inequality_Report_2026_p145","product_type":"parse","source_file_path":"World_Inequality_Report_2026_p145.pdf"},"started_at":"2026-11-04T19:45:26.307539"} +{"completed_at":"2026-11-04T19:45:28.879706","latency_in_ms":1408,"output":{"example_id":"chart/World_Inequality_Report_2026_p146","grounded_pages":[],"layout_pages":[],"markdown":"Chapter 8. Political Cleavages\n\n**Figure 8.2. Educated voters increasingly support the left, while high-income voters continue leaning right**\n\nHigher−educated voters voting for left−wing\n\nparties (democratic, labor, social− democratic, socialist, green, etc.)\n\nTop−income voters voting for right−wing\n\nparties (other parties) −20%\n\n−15%\n\n−10%\n\n−5%\n\n0%\n\n5%\n\n10%\n\n15%\n\n1960−64\n\n1965−69\n\n1970−74\n\n1975−79\n\n1980−84\n\n1985−89\n\n1990−94\n\n1995−99\n\n2000−04\n\n2005−09\n\n2010−14\n\n2015−19\n\n2020−25\n\nDifference (pp): top 10% minus\n\nbottom 90% voting left\n\n(% of top 10% educated voting left) minus (% of bottom 90% educated voting left)\n\n(% of top 10% earners voting left) minus (% of bottom 90% earners voting left)\n\n**Education and income divides in Western democracies, 1960−2025**\n\n**Interpretation.** In the 1960s, both higher−educated and high−income voters were less likely to vote for left−wing (democratic / labor / social−democratic / socialist / green) parties than lower−educated and low−income voters by more than 10 percentage points. The left vote has gradually become associated with higher education voters, giving rising to a multi−elite party system. Figures correspond to five−year averages for Australia, Britain, Canada, Denmark, France, Germany, Italy, the Netherlands, Norway, Sweden, Switzerland, and the U.S. Estimates control for income/education, age, gender, religion, church attendance, rural/urban, region, race/ethnicity, employment status, and marital status (in country−years for which these variables are available). **Sources and series:** Gethin et al. (2021) and _World Political Cleavages and Inequality Database_ (wpid.world).\n\nas education and healthcare workers, public sector employees, and urban elites, whose concerns and political priorities extend beyond redistribution to issues such as climate change, minority rights, and gender equality. Meanwhile, less educated voters have turned toward conservative platforms. This pattern is visible in nearly all Western democracies, though with varying intensity. Portugal and Ireland remain partial exceptions, with smaller or delayed reversals. Yet the broader trend is clear: education, once a strong determinant of support for conservative parties, has become a defining predictor of left voting in most Western democracies. Contrary to education, the income divide has remained relatively stable. **Figure 8.4**\n\nshows that higher-income voters continue to lean right. This enduring divide has meant that even as educational elites shifted left, affluent groups have largely remained on the right. Yet the income divide has not been static. In many Western democracies, the influence of income on the vote declined during the\n\nlate 20th century, as party competition increasingly shifted toward cultural issues. A striking case is the United States. In recent elections, high-income voters have shifted to the Democratic Party, becoming more likely than low-income groups to support the Democrats. This shift represents a historic reversal of the classic postwar pattern, which highlights how far the U.S. has moved toward a “high-income-left” coalition and illustrates the weight of sociocultural divides in shaping political conflict.\n\n**Figure 8.5** brings these dynamics together by plotting income and education gradients simultaneously. In the 1960s, parties lined up along a diagonal: left-wing parties were supported by low-income and less educated voters, while right-wing parties drew support from high-income and highly educated voters. By 2000–2025, this alignment has fractured. Parties now cluster across different quadrants. Green parties occupy the high-education section but do not distinguish themselves in terms of income. Anti-immigration movements dominate the low-education space, appealing to both low- and high-income groups. Conservatives\n\n146","pages":[{"line_numbers":[],"markdown":"Chapter 8. Political Cleavages\n\n**Figure 8.2. Educated voters increasingly support the left, while high-income voters continue leaning right**\n\nHigher−educated voters voting for left−wing\n\nparties (democratic, labor, social− democratic, socialist, green, etc.)\n\nTop−income voters voting for right−wing\n\nparties (other parties) −20%\n\n−15%\n\n−10%\n\n−5%\n\n0%\n\n5%\n\n10%\n\n15%\n\n1960−64\n\n1965−69\n\n1970−74\n\n1975−79\n\n1980−84\n\n1985−89\n\n1990−94\n\n1995−99\n\n2000−04\n\n2005−09\n\n2010−14\n\n2015−19\n\n2020−25\n\nDifference (pp): top 10% minus\n\nbottom 90% voting left\n\n(% of top 10% educated voting left) minus (% of bottom 90% educated voting left)\n\n(% of top 10% earners voting left) minus (% of bottom 90% earners voting left)\n\n**Education and income divides in Western democracies, 1960−2025**\n\n**Interpretation.** In the 1960s, both higher−educated and high−income voters were less likely to vote for left−wing (democratic / labor / social−democratic / socialist / green) parties than lower−educated and low−income voters by more than 10 percentage points. The left vote has gradually become associated with higher education voters, giving rising to a multi−elite party system. Figures correspond to five−year averages for Australia, Britain, Canada, Denmark, France, Germany, Italy, the Netherlands, Norway, Sweden, Switzerland, and the U.S. Estimates control for income/education, age, gender, religion, church attendance, rural/urban, region, race/ethnicity, employment status, and marital status (in country−years for which these variables are available). **Sources and series:** Gethin et al. (2021) and _World Political Cleavages and Inequality Database_ (wpid.world).\n\nas education and healthcare workers, public sector employees, and urban elites, whose concerns and political priorities extend beyond redistribution to issues such as climate change, minority rights, and gender equality. Meanwhile, less educated voters have turned toward conservative platforms. This pattern is visible in nearly all Western democracies, though with varying intensity. Portugal and Ireland remain partial exceptions, with smaller or delayed reversals. Yet the broader trend is clear: education, once a strong determinant of support for conservative parties, has become a defining predictor of left voting in most Western democracies. Contrary to education, the income divide has remained relatively stable. **Figure 8.4**\n\nshows that higher-income voters continue to lean right. This enduring divide has meant that even as educational elites shifted left, affluent groups have largely remained on the right. Yet the income divide has not been static. In many Western democracies, the influence of income on the vote declined during the\n\nlate 20th century, as party competition increasingly shifted toward cultural issues. A striking case is the United States. In recent elections, high-income voters have shifted to the Democratic Party, becoming more likely than low-income groups to support the Democrats. This shift represents a historic reversal of the classic postwar pattern, which highlights how far the U.S. has moved toward a “high-income-left” coalition and illustrates the weight of sociocultural divides in shaping political conflict.\n\n**Figure 8.5** brings these dynamics together by plotting income and education gradients simultaneously. In the 1960s, parties lined up along a diagonal: left-wing parties were supported by low-income and less educated voters, while right-wing parties drew support from high-income and highly educated voters. By 2000–2025, this alignment has fractured. Parties now cluster across different quadrants. Green parties occupy the high-education section but do not distinguish themselves in terms of income. Anti-immigration movements dominate the low-education space, appealing to both low- and high-income groups. Conservatives\n\n146","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Chapter 8. Political Cleavages\n\n**Figure 8.2. Educated voters increasingly support the left, while high-income voters continue leaning right**\n\nHigher−educated voters voting for left−wing\n\nparties (democratic, labor, social− democratic, socialist, green, etc.)\n\nTop−income voters voting for right−wing\n\nparties (other parties) −20%\n\n−15%\n\n−10%\n\n−5%\n\n0%\n\n5%\n\n10%\n\n15%\n\n1960−64\n\n1965−69\n\n1970−74\n\n1975−79\n\n1980−84\n\n1985−89\n\n1990−94\n\n1995−99\n\n2000−04\n\n2005−09\n\n2010−14\n\n2015−19\n\n2020−25\n\nDifference (pp): top 10% minus\n\nbottom 90% voting left\n\n(% of top 10% educated voting left) minus (% of bottom 90% educated voting left)\n\n(% of top 10% earners voting left) minus (% of bottom 90% earners voting left)\n\n**Education and income divides in Western democracies, 1960−2025**\n\n**Interpretation.** In the 1960s, both higher−educated and high−income voters were less likely to vote for left−wing (democratic / labor / social−democratic / socialist / green) parties than lower−educated and low−income voters by more than 10 percentage points. The left vote has gradually become associated with higher education voters, giving rising to a multi−elite party system. Figures correspond to five−year averages for Australia, Britain, Canada, Denmark, France, Germany, Italy, the Netherlands, Norway, Sweden, Switzerland, and the U.S. Estimates control for income/education, age, gender, religion, church attendance, rural/urban, region, race/ethnicity, employment status, and marital status (in country−years for which these variables are available). **Sources and series:** Gethin et al. (2021) and _World Political Cleavages and Inequality Database_ (wpid.world).\n\nas education and healthcare workers, public sector employees, and urban elites, whose concerns and political priorities extend beyond redistribution to issues such as climate change, minority rights, and gender equality. Meanwhile, less educated voters have turned toward conservative platforms. This pattern is visible in nearly all Western democracies, though with varying intensity. Portugal and Ireland remain partial exceptions, with smaller or delayed reversals. Yet the broader trend is clear: education, once a strong determinant of support for conservative parties, has become a defining predictor of left voting in most Western democracies. Contrary to education, the income divide has remained relatively stable. **Figure 8.4**\n\nshows that higher-income voters continue to lean right. This enduring divide has meant that even as educational elites shifted left, affluent groups have largely remained on the right. Yet the income divide has not been static. In many Western democracies, the influence of income on the vote declined during the\n\nlate 20th century, as party competition increasingly shifted toward cultural issues. A striking case is the United States. In recent elections, high-income voters have shifted to the Democratic Party, becoming more likely than low-income groups to support the Democrats. This shift represents a historic reversal of the classic postwar pattern, which highlights how far the U.S. has moved toward a “high-income-left” coalition and illustrates the weight of sociocultural divides in shaping political conflict.\n\n**Figure 8.5** brings these dynamics together by plotting income and education gradients simultaneously. In the 1960s, parties lined up along a diagonal: left-wing parties were supported by low-income and less educated voters, while right-wing parties drew support from high-income and highly educated voters. By 2000–2025, this alignment has fractured. Parties now cluster across different quadrants. Green parties occupy the high-education section but do not distinguish themselves in terms of income. Anti-immigration movements dominate the low-education space, appealing to both low- and high-income groups. Conservatives\n\n146","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p146.pdf"]},"response":{"errors":[],"extract_id":"extract_6ea6dbdd63969fc25aa5be7a251c781c","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Chapter 8. Political Cleavages\n\n**Figure 8.2. Educated voters increasingly support the left, while high-income voters continue leaning right**\n\nHigher−educated voters voting for left−wing\n\nparties (democratic, labor, social− democratic, socialist, green, etc.)\n\nTop−income voters voting for right−wing\n\nparties (other parties) −20%\n\n−15%\n\n−10%\n\n−5%\n\n0%\n\n5%\n\n10%\n\n15%\n\n1960−64\n\n1965−69\n\n1970−74\n\n1975−79\n\n1980−84\n\n1985−89\n\n1990−94\n\n1995−99\n\n2000−04\n\n2005−09\n\n2010−14\n\n2015−19\n\n2020−25\n\nDifference (pp): top 10% minus\n\nbottom 90% voting left\n\n(% of top 10% educated voting left) minus (% of bottom 90% educated voting left)\n\n(% of top 10% earners voting left) minus (% of bottom 90% earners voting left)\n\n**Education and income divides in Western democracies, 1960−2025**\n\n**Interpretation.** In the 1960s, both higher−educated and high−income voters were less likely to vote for left−wing (democratic / labor / social−democratic / socialist / green) parties than lower−educated and low−income voters by more than 10 percentage points. The left vote has gradually become associated with higher education voters, giving rising to a multi−elite party system. Figures correspond to five−year averages for Australia, Britain, Canada, Denmark, France, Germany, Italy, the Netherlands, Norway, Sweden, Switzerland, and the U.S. Estimates control for income/education, age, gender, religion, church attendance, rural/urban, region, race/ethnicity, employment status, and marital status (in country−years for which these variables are available). **Sources and series:** Gethin et al. (2021) and _World Political Cleavages and Inequality Database_ (wpid.world).\n\nas education and healthcare workers, public sector employees, and urban elites, whose concerns and political priorities extend beyond redistribution to issues such as climate change, minority rights, and gender equality. Meanwhile, less educated voters have turned toward conservative platforms. This pattern is visible in nearly all Western democracies, though with varying intensity. Portugal and Ireland remain partial exceptions, with smaller or delayed reversals. Yet the broader trend is clear: education, once a strong determinant of support for conservative parties, has become a defining predictor of left voting in most Western democracies. Contrary to education, the income divide has remained relatively stable. **Figure 8.4**\n\nshows that higher-income voters continue to lean right. This enduring divide has meant that even as educational elites shifted left, affluent groups have largely remained on the right. Yet the income divide has not been static. In many Western democracies, the influence of income on the vote declined during the\n\nlate 20th century, as party competition increasingly shifted toward cultural issues. A striking case is the United States. In recent elections, high-income voters have shifted to the Democratic Party, becoming more likely than low-income groups to support the Democrats. This shift represents a historic reversal of the classic postwar pattern, which highlights how far the U.S. has moved toward a “high-income-left” coalition and illustrates the weight of sociocultural divides in shaping political conflict.\n\n**Figure 8.5** brings these dynamics together by plotting income and education gradients simultaneously. In the 1960s, parties lined up along a diagonal: left-wing parties were supported by low-income and less educated voters, while right-wing parties drew support from high-income and highly educated voters. By 2000–2025, this alignment has fractured. Parties now cluster across different quadrants. Green parties occupy the high-education section but do not distinguish themselves in terms of income. Anti-immigration movements dominate the low-education space, appealing to both low- and high-income groups. Conservatives\n\n146"],"full_content":"Chapter 8. Political Cleavages\n\n**Figure 8.2. Educated voters increasingly support the left, while high-income voters continue leaning right**\n\nHigher−educated voters voting for left−wing\n\nparties (democratic, labor, social− democratic, socialist, green, etc.)\n\nTop−income voters voting for right−wing\n\nparties (other parties) −20%\n\n−15%\n\n−10%\n\n−5%\n\n0%\n\n5%\n\n10%\n\n15%\n\n1960−64\n\n1965−69\n\n1970−74\n\n1975−79\n\n1980−84\n\n1985−89\n\n1990−94\n\n1995−99\n\n2000−04\n\n2005−09\n\n2010−14\n\n2015−19\n\n2020−25\n\nDifference (pp): top 10% minus\n\nbottom 90% voting left\n\n(% of top 10% educated voting left) minus (% of bottom 90% educated voting left)\n\n(% of top 10% earners voting left) minus (% of bottom 90% earners voting left)\n\n**Education and income divides in Western democracies, 1960−2025**\n\n**Interpretation.** In the 1960s, both higher−educated and high−income voters were less likely to vote for left−wing (democratic / labor / social−democratic / socialist / green) parties than lower−educated and low−income voters by more than 10 percentage points. The left vote has gradually become associated with higher education voters, giving rising to a multi−elite party system. Figures correspond to five−year averages for Australia, Britain, Canada, Denmark, France, Germany, Italy, the Netherlands, Norway, Sweden, Switzerland, and the U.S. Estimates control for income/education, age, gender, religion, church attendance, rural/urban, region, race/ethnicity, employment status, and marital status (in country−years for which these variables are available). **Sources and series:** Gethin et al. (2021) and _World Political Cleavages and Inequality Database_ (wpid.world).\n\nas education and healthcare workers, public sector employees, and urban elites, whose concerns and political priorities extend beyond redistribution to issues such as climate change, minority rights, and gender equality. Meanwhile, less educated voters have turned toward conservative platforms. This pattern is visible in nearly all Western democracies, though with varying intensity. Portugal and Ireland remain partial exceptions, with smaller or delayed reversals. Yet the broader trend is clear: education, once a strong determinant of support for conservative parties, has become a defining predictor of left voting in most Western democracies. Contrary to education, the income divide has remained relatively stable. **Figure 8.4**\n\nshows that higher-income voters continue to lean right. This enduring divide has meant that even as educational elites shifted left, affluent groups have largely remained on the right. Yet the income divide has not been static. In many Western democracies, the influence of income on the vote declined during the\n\nlate 20th century, as party competition increasingly shifted toward cultural issues. A striking case is the United States. In recent elections, high-income voters have shifted to the Democratic Party, becoming more likely than low-income groups to support the Democrats. This shift represents a historic reversal of the classic postwar pattern, which highlights how far the U.S. has moved toward a “high-income-left” coalition and illustrates the weight of sociocultural divides in shaping political conflict.\n\n**Figure 8.5** brings these dynamics together by plotting income and education gradients simultaneously. In the 1960s, parties lined up along a diagonal: left-wing parties were supported by low-income and less educated voters, while right-wing parties drew support from high-income and highly educated voters. By 2000–2025, this alignment has fractured. Parties now cluster across different quadrants. Green parties occupy the high-education section but do not distinguish themselves in terms of income. Anti-immigration movements dominate the low-education space, appealing to both low- and high-income groups. Conservatives\n\n146","publish_date":null,"title":"Chapter 8. Political Cleavages Figure8.2. Educatedvotersincreasinglysupporttheleft,while high-incomevoterscontinueleaningright Education and income divides in Western democracies, 1960−2025 as education and healthcare workers,","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p146.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p146.pdf"},"request":{"example_id":"chart/World_Inequality_Report_2026_p146","product_type":"parse","source_file_path":"World_Inequality_Report_2026_p146.pdf"},"started_at":"2026-11-04T19:45:27.471631"} +{"completed_at":"2026-11-04T19:45:29.606842","latency_in_ms":1209,"output":{"example_id":"chart/World_Inequality_Report_2026_p150","grounded_pages":[],"layout_pages":[],"markdown":"Chapter 8. Political Cleavages\n\n**Figure 8.3. The reversal of educational divides in Western democracies**\n\n−40% −35% −30% −25% −20% −15% −10%\n\n−5%\n\n0% 5%\n\n10% 15% 20% 25% 30%\n\n1950−59 1960−69 1970−79 1980−89 1990−99 2000−09 2010−19 2020−25\n\nTime period\n\nDifference between (% of top 10%\n\neducated voting left) and (% of\n\nbottom 90% educated voting left)\n\nAustralia Britain Canada Denmark Finland Iceland\n\nIreland New Zealand Norway Sweden United States Average\n\n**English−speaking and Northern European countries**\n\n−40% −35% −30% −25% −20% −15% −10%\n\n−5%\n\n0% 5%\n\n10% 15% 20% 25% 30%\n\n1950−59 1960−69 1970−79 1980−89 1990−99 2000−09 2010−19 2020−25\n\nTime period\n\nDifference between (% of top 10%\n\neducated voting left) and (% of\n\nbottom 90% educated voting left)\n\nAustria Belgium Germany Italy Luxembourg Netherlands\n\nPortugal Spain Switzerland France Average\n\n**Continental and Southern European countries**\n\n**Interpretation.** In these countries, higher−educated voters used to be significantly more likely to vote for conservative parties and have gradually become less likely to vote for these parties. Estimates control for income, age, gender, religion, church attendance, rural / urban, region, race / ethnicity, employment status, and marital status (in country−years for which these variables are available). **Sources and series:** Gethin et al. (2021) and _World Political Cleavages and Inequality Database_ (wpid.world).\n\n**Education divides in Western democracies, 1960−2025**\n\n150","pages":[{"line_numbers":[],"markdown":"Chapter 8. Political Cleavages\n\n**Figure 8.3. The reversal of educational divides in Western democracies**\n\n−40% −35% −30% −25% −20% −15% −10%\n\n−5%\n\n0% 5%\n\n10% 15% 20% 25% 30%\n\n1950−59 1960−69 1970−79 1980−89 1990−99 2000−09 2010−19 2020−25\n\nTime period\n\nDifference between (% of top 10%\n\neducated voting left) and (% of\n\nbottom 90% educated voting left)\n\nAustralia Britain Canada Denmark Finland Iceland\n\nIreland New Zealand Norway Sweden United States Average\n\n**English−speaking and Northern European countries**\n\n−40% −35% −30% −25% −20% −15% −10%\n\n−5%\n\n0% 5%\n\n10% 15% 20% 25% 30%\n\n1950−59 1960−69 1970−79 1980−89 1990−99 2000−09 2010−19 2020−25\n\nTime period\n\nDifference between (% of top 10%\n\neducated voting left) and (% of\n\nbottom 90% educated voting left)\n\nAustria Belgium Germany Italy Luxembourg Netherlands\n\nPortugal Spain Switzerland France Average\n\n**Continental and Southern European countries**\n\n**Interpretation.** In these countries, higher−educated voters used to be significantly more likely to vote for conservative parties and have gradually become less likely to vote for these parties. Estimates control for income, age, gender, religion, church attendance, rural / urban, region, race / ethnicity, employment status, and marital status (in country−years for which these variables are available). **Sources and series:** Gethin et al. (2021) and _World Political Cleavages and Inequality Database_ (wpid.world).\n\n**Education divides in Western democracies, 1960−2025**\n\n150","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Chapter 8. Political Cleavages\n\n**Figure 8.3. The reversal of educational divides in Western democracies**\n\n−40% −35% −30% −25% −20% −15% −10%\n\n−5%\n\n0% 5%\n\n10% 15% 20% 25% 30%\n\n1950−59 1960−69 1970−79 1980−89 1990−99 2000−09 2010−19 2020−25\n\nTime period\n\nDifference between (% of top 10%\n\neducated voting left) and (% of\n\nbottom 90% educated voting left)\n\nAustralia Britain Canada Denmark Finland Iceland\n\nIreland New Zealand Norway Sweden United States Average\n\n**English−speaking and Northern European countries**\n\n−40% −35% −30% −25% −20% −15% −10%\n\n−5%\n\n0% 5%\n\n10% 15% 20% 25% 30%\n\n1950−59 1960−69 1970−79 1980−89 1990−99 2000−09 2010−19 2020−25\n\nTime period\n\nDifference between (% of top 10%\n\neducated voting left) and (% of\n\nbottom 90% educated voting left)\n\nAustria Belgium Germany Italy Luxembourg Netherlands\n\nPortugal Spain Switzerland France Average\n\n**Continental and Southern European countries**\n\n**Interpretation.** In these countries, higher−educated voters used to be significantly more likely to vote for conservative parties and have gradually become less likely to vote for these parties. Estimates control for income, age, gender, religion, church attendance, rural / urban, region, race / ethnicity, employment status, and marital status (in country−years for which these variables are available). **Sources and series:** Gethin et al. (2021) and _World Political Cleavages and Inequality Database_ (wpid.world).\n\n**Education divides in Western democracies, 1960−2025**\n\n150","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p150.pdf"]},"response":{"errors":[],"extract_id":"extract_5d0992d6b8431e0c4e8cf28f8efe5342","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Chapter 8. Political Cleavages\n\n**Figure 8.3. The reversal of educational divides in Western democracies**\n\n−40% −35% −30% −25% −20% −15% −10%\n\n−5%\n\n0% 5%\n\n10% 15% 20% 25% 30%\n\n1950−59 1960−69 1970−79 1980−89 1990−99 2000−09 2010−19 2020−25\n\nTime period\n\nDifference between (% of top 10%\n\neducated voting left) and (% of\n\nbottom 90% educated voting left)\n\nAustralia Britain Canada Denmark Finland Iceland\n\nIreland New Zealand Norway Sweden United States Average\n\n**English−speaking and Northern European countries**\n\n−40% −35% −30% −25% −20% −15% −10%\n\n−5%\n\n0% 5%\n\n10% 15% 20% 25% 30%\n\n1950−59 1960−69 1970−79 1980−89 1990−99 2000−09 2010−19 2020−25\n\nTime period\n\nDifference between (% of top 10%\n\neducated voting left) and (% of\n\nbottom 90% educated voting left)\n\nAustria Belgium Germany Italy Luxembourg Netherlands\n\nPortugal Spain Switzerland France Average\n\n**Continental and Southern European countries**\n\n**Interpretation.** In these countries, higher−educated voters used to be significantly more likely to vote for conservative parties and have gradually become less likely to vote for these parties. Estimates control for income, age, gender, religion, church attendance, rural / urban, region, race / ethnicity, employment status, and marital status (in country−years for which these variables are available). **Sources and series:** Gethin et al. (2021) and _World Political Cleavages and Inequality Database_ (wpid.world).\n\n**Education divides in Western democracies, 1960−2025**\n\n150"],"full_content":"Chapter 8. Political Cleavages\n\n**Figure 8.3. The reversal of educational divides in Western democracies**\n\n−40% −35% −30% −25% −20% −15% −10%\n\n−5%\n\n0% 5%\n\n10% 15% 20% 25% 30%\n\n1950−59 1960−69 1970−79 1980−89 1990−99 2000−09 2010−19 2020−25\n\nTime period\n\nDifference between (% of top 10%\n\neducated voting left) and (% of\n\nbottom 90% educated voting left)\n\nAustralia Britain Canada Denmark Finland Iceland\n\nIreland New Zealand Norway Sweden United States Average\n\n**English−speaking and Northern European countries**\n\n−40% −35% −30% −25% −20% −15% −10%\n\n−5%\n\n0% 5%\n\n10% 15% 20% 25% 30%\n\n1950−59 1960−69 1970−79 1980−89 1990−99 2000−09 2010−19 2020−25\n\nTime period\n\nDifference between (% of top 10%\n\neducated voting left) and (% of\n\nbottom 90% educated voting left)\n\nAustria Belgium Germany Italy Luxembourg Netherlands\n\nPortugal Spain Switzerland France Average\n\n**Continental and Southern European countries**\n\n**Interpretation.** In these countries, higher−educated voters used to be significantly more likely to vote for conservative parties and have gradually become less likely to vote for these parties. Estimates control for income, age, gender, religion, church attendance, rural / urban, region, race / ethnicity, employment status, and marital status (in country−years for which these variables are available). **Sources and series:** Gethin et al. (2021) and _World Political Cleavages and Inequality Database_ (wpid.world).\n\n**Education divides in Western democracies, 1960−2025**\n\n150","publish_date":null,"title":"Chapter 8. Political Cleavages Figure8.3. ThereversalofeducationaldividesinWestern democracies Education divides in Western democracies, 1960−2025 150","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p150.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p150.pdf"},"request":{"example_id":"chart/World_Inequality_Report_2026_p150","product_type":"parse","source_file_path":"World_Inequality_Report_2026_p150.pdf"},"started_at":"2026-11-04T19:45:28.397770"} +{"completed_at":"2026-11-04T19:45:30.831475","latency_in_ms":1223,"output":{"example_id":"chart/World_Inequality_Report_2026_p151","grounded_pages":[],"layout_pages":[],"markdown":"Chapter 8. Political Cleavages\n\n**Figure 8.4. The stability/decline of income divides in Western democracies**\n\n−40% −35% −30% −25% −20% −15% −10%\n\n−5%\n\n0% 5%\n\n10% 15% 20% 25% 30%\n\n1950−59 1960−69 1970−79 1980−89 1990−99 2000−09 2010−19 2020−25\n\nTime period\n\nDifference between (% of top 10%\n\nearners voting left) and (% of\n\nbottom 90% earners voting left)\n\nAustralia Britain Canada Denmark Finland Iceland\n\nIreland New Zealand Norway Sweden United States Average\n\n**English−speaking and Northern European countries**\n\n−40% −35% −30% −25% −20% −15% −10%\n\n−5%\n\n0% 5%\n\n10% 15% 20% 25% 30%\n\n1950−59 1960−69 1970−79 1980−89 1990−99 2000−09 2010−19 2020−25\n\nTime period\n\nDifference between (% of top 10%\n\nearners voting left) and (% of\n\nbottom 90% earners voting left)\n\nAustria Belgium Germany Italy Luxembourg Netherlands\n\nPortugal Spain Switzerland France Average\n\n**Continental and Southern European countries**\n\n**Interpretation.** In these countries, top−income voters have remained significantly more likely to vote for conservative parties than low−income voters. Estimates control for income, age, gender, religion, church attendance, rural / urban, region, race / ethnicity, employment status, and marital status (in country−years for which these variables are available). **Sources and series:** Gethin et al. (2021) and _World Political Cleavages and Inequality Database_ (wpid.world).\n\n**Income divides in Western democracies, 1960−2025**\n\n151","pages":[{"line_numbers":[],"markdown":"Chapter 8. Political Cleavages\n\n**Figure 8.4. The stability/decline of income divides in Western democracies**\n\n−40% −35% −30% −25% −20% −15% −10%\n\n−5%\n\n0% 5%\n\n10% 15% 20% 25% 30%\n\n1950−59 1960−69 1970−79 1980−89 1990−99 2000−09 2010−19 2020−25\n\nTime period\n\nDifference between (% of top 10%\n\nearners voting left) and (% of\n\nbottom 90% earners voting left)\n\nAustralia Britain Canada Denmark Finland Iceland\n\nIreland New Zealand Norway Sweden United States Average\n\n**English−speaking and Northern European countries**\n\n−40% −35% −30% −25% −20% −15% −10%\n\n−5%\n\n0% 5%\n\n10% 15% 20% 25% 30%\n\n1950−59 1960−69 1970−79 1980−89 1990−99 2000−09 2010−19 2020−25\n\nTime period\n\nDifference between (% of top 10%\n\nearners voting left) and (% of\n\nbottom 90% earners voting left)\n\nAustria Belgium Germany Italy Luxembourg Netherlands\n\nPortugal Spain Switzerland France Average\n\n**Continental and Southern European countries**\n\n**Interpretation.** In these countries, top−income voters have remained significantly more likely to vote for conservative parties than low−income voters. Estimates control for income, age, gender, religion, church attendance, rural / urban, region, race / ethnicity, employment status, and marital status (in country−years for which these variables are available). **Sources and series:** Gethin et al. (2021) and _World Political Cleavages and Inequality Database_ (wpid.world).\n\n**Income divides in Western democracies, 1960−2025**\n\n151","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Chapter 8. Political Cleavages\n\n**Figure 8.4. The stability/decline of income divides in Western democracies**\n\n−40% −35% −30% −25% −20% −15% −10%\n\n−5%\n\n0% 5%\n\n10% 15% 20% 25% 30%\n\n1950−59 1960−69 1970−79 1980−89 1990−99 2000−09 2010−19 2020−25\n\nTime period\n\nDifference between (% of top 10%\n\nearners voting left) and (% of\n\nbottom 90% earners voting left)\n\nAustralia Britain Canada Denmark Finland Iceland\n\nIreland New Zealand Norway Sweden United States Average\n\n**English−speaking and Northern European countries**\n\n−40% −35% −30% −25% −20% −15% −10%\n\n−5%\n\n0% 5%\n\n10% 15% 20% 25% 30%\n\n1950−59 1960−69 1970−79 1980−89 1990−99 2000−09 2010−19 2020−25\n\nTime period\n\nDifference between (% of top 10%\n\nearners voting left) and (% of\n\nbottom 90% earners voting left)\n\nAustria Belgium Germany Italy Luxembourg Netherlands\n\nPortugal Spain Switzerland France Average\n\n**Continental and Southern European countries**\n\n**Interpretation.** In these countries, top−income voters have remained significantly more likely to vote for conservative parties than low−income voters. Estimates control for income, age, gender, religion, church attendance, rural / urban, region, race / ethnicity, employment status, and marital status (in country−years for which these variables are available). **Sources and series:** Gethin et al. (2021) and _World Political Cleavages and Inequality Database_ (wpid.world).\n\n**Income divides in Western democracies, 1960−2025**\n\n151","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p151.pdf"]},"response":{"errors":[],"extract_id":"extract_9be2e6b3a6010a99b080337c2f85abcd","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Chapter 8. Political Cleavages\n\n**Figure 8.4. The stability/decline of income divides in Western democracies**\n\n−40% −35% −30% −25% −20% −15% −10%\n\n−5%\n\n0% 5%\n\n10% 15% 20% 25% 30%\n\n1950−59 1960−69 1970−79 1980−89 1990−99 2000−09 2010−19 2020−25\n\nTime period\n\nDifference between (% of top 10%\n\nearners voting left) and (% of\n\nbottom 90% earners voting left)\n\nAustralia Britain Canada Denmark Finland Iceland\n\nIreland New Zealand Norway Sweden United States Average\n\n**English−speaking and Northern European countries**\n\n−40% −35% −30% −25% −20% −15% −10%\n\n−5%\n\n0% 5%\n\n10% 15% 20% 25% 30%\n\n1950−59 1960−69 1970−79 1980−89 1990−99 2000−09 2010−19 2020−25\n\nTime period\n\nDifference between (% of top 10%\n\nearners voting left) and (% of\n\nbottom 90% earners voting left)\n\nAustria Belgium Germany Italy Luxembourg Netherlands\n\nPortugal Spain Switzerland France Average\n\n**Continental and Southern European countries**\n\n**Interpretation.** In these countries, top−income voters have remained significantly more likely to vote for conservative parties than low−income voters. Estimates control for income, age, gender, religion, church attendance, rural / urban, region, race / ethnicity, employment status, and marital status (in country−years for which these variables are available). **Sources and series:** Gethin et al. (2021) and _World Political Cleavages and Inequality Database_ (wpid.world).\n\n**Income divides in Western democracies, 1960−2025**\n\n151"],"full_content":"Chapter 8. Political Cleavages\n\n**Figure 8.4. The stability/decline of income divides in Western democracies**\n\n−40% −35% −30% −25% −20% −15% −10%\n\n−5%\n\n0% 5%\n\n10% 15% 20% 25% 30%\n\n1950−59 1960−69 1970−79 1980−89 1990−99 2000−09 2010−19 2020−25\n\nTime period\n\nDifference between (% of top 10%\n\nearners voting left) and (% of\n\nbottom 90% earners voting left)\n\nAustralia Britain Canada Denmark Finland Iceland\n\nIreland New Zealand Norway Sweden United States Average\n\n**English−speaking and Northern European countries**\n\n−40% −35% −30% −25% −20% −15% −10%\n\n−5%\n\n0% 5%\n\n10% 15% 20% 25% 30%\n\n1950−59 1960−69 1970−79 1980−89 1990−99 2000−09 2010−19 2020−25\n\nTime period\n\nDifference between (% of top 10%\n\nearners voting left) and (% of\n\nbottom 90% earners voting left)\n\nAustria Belgium Germany Italy Luxembourg Netherlands\n\nPortugal Spain Switzerland France Average\n\n**Continental and Southern European countries**\n\n**Interpretation.** In these countries, top−income voters have remained significantly more likely to vote for conservative parties than low−income voters. Estimates control for income, age, gender, religion, church attendance, rural / urban, region, race / ethnicity, employment status, and marital status (in country−years for which these variables are available). **Sources and series:** Gethin et al. (2021) and _World Political Cleavages and Inequality Database_ (wpid.world).\n\n**Income divides in Western democracies, 1960−2025**\n\n151","publish_date":null,"title":"Chapter 8. Political Cleavages Figure8.4. Thestability/declineofincomedividesinWestern democracies Income divides in Western democracies, 1960−2025 151","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p151.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p151.pdf"},"request":{"example_id":"chart/World_Inequality_Report_2026_p151","product_type":"parse","source_file_path":"World_Inequality_Report_2026_p151.pdf"},"started_at":"2026-11-04T19:45:29.608430"} +{"completed_at":"2026-11-04T19:45:32.620227","latency_in_ms":1880,"output":{"example_id":"chart/World_Inequality_Report_2026_p17","grounded_pages":[],"layout_pages":[],"markdown":"Executive Summary\n\n**Figure 5. Women persistently receive lower labor income than men everywhere**\n\n40%\n\n36%\n\n16%\n\n40% 37%\n\n20%\n\n28%\n\n34%\n\nGender parity\n\n0%\n\n10%\n\n20%\n\n30%\n\n40%\n\n50%\n\n60%\n\nMiddle East\n\n& North\n\nAfrica\n\nSouth & Southeast\n\nAsia\n\nSub−\n\nSaharan\n\nAfrica\n\nEast Asia\n\nLatin\n\nAmerica\n\nRussia & Central\n\nAsia\n\nEurope North America & Oceania\n\nFemale labor income share (%)\n\n1990 2000 2010 2020 2025\n\n**Female labor income shares, 1990−2025**\n\n**Interpretation.** This figure shows the evolution of the female labor income share between 1990 and 2025 across world regions. In 2025, female workers earn about 16% of total labor income in the Middle East & North Africa, but about 40% in North America & Oceania and Europe. At the global level, women earned 27.8% of labor income in 1990 and 28.2% in 2025. While some progress has been made, gender parity remains distant in all regions. **Sources and series:** Neef and Robilliard (2021), Gabrielli et al. (2024), and wir2026.wid.world/methodology.\n\n**Political cleavages and democracy**\n\nEconomic divides do not stop at the marketplace; they spill directly into politics. Inequality shapes who is represented, whose voices carry weight, and how coalitions are built, or fail to be built. **Figure 15** shows how the traditional class-based alignment of politics in Western democracies has broken down. ³ In the mid-20th century, lower-income and less educated voters largely supported left-wing parties, while wealthier and more educated groups leaned right, creating a clear class divide and rising redistribution. Today, that pattern has fractured. First, education and income now point in different directions (see **Figure 15** ), making broad coalitions for redistribution far harder to sustain. This evolution can be accounted for by the fact that educational expansion has come with a complexification of the class structure. For example, many high-degree but relatively low-income voters (e.g., teachers or nurses) currently vote for the left, while many voters with lower degrees but relatively higher income (e.g.,\n\nself-employed or truck drivers) tend to vote for the right. The even more striking evolution is the rise of territorial divides within countries. In many advanced democracies, gaps in political affiliations between large metropolitan centers and smaller towns have reached levels unseen in a century (see **Figure 16** ). Unequal access to public services (education, health, transportation, and other infrastructures), job opportunities, and exposure to trade shocks has fractured social cohesion and weakened the coalitions necessary for redistributive reform. As a consequence, working-class voters are now fragmented across parties on both sides of the aisle or left without strong representation, which limits their political influence and entrenches inequality. In order to reactivate the redistributive coalitions of the postwar era, it is critical to design more ambitious policy platforms benefiting all territories, as they successfully did in the past. This fragmentation erodes the political foundations needed to tackle inequality\n\n17","pages":[{"line_numbers":[],"markdown":"Executive Summary\n\n**Figure 5. Women persistently receive lower labor income than men everywhere**\n\n40%\n\n36%\n\n16%\n\n40% 37%\n\n20%\n\n28%\n\n34%\n\nGender parity\n\n0%\n\n10%\n\n20%\n\n30%\n\n40%\n\n50%\n\n60%\n\nMiddle East\n\n& North\n\nAfrica\n\nSouth & Southeast\n\nAsia\n\nSub−\n\nSaharan\n\nAfrica\n\nEast Asia\n\nLatin\n\nAmerica\n\nRussia & Central\n\nAsia\n\nEurope North America & Oceania\n\nFemale labor income share (%)\n\n1990 2000 2010 2020 2025\n\n**Female labor income shares, 1990−2025**\n\n**Interpretation.** This figure shows the evolution of the female labor income share between 1990 and 2025 across world regions. In 2025, female workers earn about 16% of total labor income in the Middle East & North Africa, but about 40% in North America & Oceania and Europe. At the global level, women earned 27.8% of labor income in 1990 and 28.2% in 2025. While some progress has been made, gender parity remains distant in all regions. **Sources and series:** Neef and Robilliard (2021), Gabrielli et al. (2024), and wir2026.wid.world/methodology.\n\n**Political cleavages and democracy**\n\nEconomic divides do not stop at the marketplace; they spill directly into politics. Inequality shapes who is represented, whose voices carry weight, and how coalitions are built, or fail to be built. **Figure 15** shows how the traditional class-based alignment of politics in Western democracies has broken down. ³ In the mid-20th century, lower-income and less educated voters largely supported left-wing parties, while wealthier and more educated groups leaned right, creating a clear class divide and rising redistribution. Today, that pattern has fractured. First, education and income now point in different directions (see **Figure 15** ), making broad coalitions for redistribution far harder to sustain. This evolution can be accounted for by the fact that educational expansion has come with a complexification of the class structure. For example, many high-degree but relatively low-income voters (e.g., teachers or nurses) currently vote for the left, while many voters with lower degrees but relatively higher income (e.g.,\n\nself-employed or truck drivers) tend to vote for the right. The even more striking evolution is the rise of territorial divides within countries. In many advanced democracies, gaps in political affiliations between large metropolitan centers and smaller towns have reached levels unseen in a century (see **Figure 16** ). Unequal access to public services (education, health, transportation, and other infrastructures), job opportunities, and exposure to trade shocks has fractured social cohesion and weakened the coalitions necessary for redistributive reform. As a consequence, working-class voters are now fragmented across parties on both sides of the aisle or left without strong representation, which limits their political influence and entrenches inequality. In order to reactivate the redistributive coalitions of the postwar era, it is critical to design more ambitious policy platforms benefiting all territories, as they successfully did in the past. This fragmentation erodes the political foundations needed to tackle inequality\n\n17","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Executive Summary\n\n**Figure 5. Women persistently receive lower labor income than men everywhere**\n\n40%\n\n36%\n\n16%\n\n40% 37%\n\n20%\n\n28%\n\n34%\n\nGender parity\n\n0%\n\n10%\n\n20%\n\n30%\n\n40%\n\n50%\n\n60%\n\nMiddle East\n\n& North\n\nAfrica\n\nSouth & Southeast\n\nAsia\n\nSub−\n\nSaharan\n\nAfrica\n\nEast Asia\n\nLatin\n\nAmerica\n\nRussia & Central\n\nAsia\n\nEurope North America & Oceania\n\nFemale labor income share (%)\n\n1990 2000 2010 2020 2025\n\n**Female labor income shares, 1990−2025**\n\n**Interpretation.** This figure shows the evolution of the female labor income share between 1990 and 2025 across world regions. In 2025, female workers earn about 16% of total labor income in the Middle East & North Africa, but about 40% in North America & Oceania and Europe. At the global level, women earned 27.8% of labor income in 1990 and 28.2% in 2025. While some progress has been made, gender parity remains distant in all regions. **Sources and series:** Neef and Robilliard (2021), Gabrielli et al. (2024), and wir2026.wid.world/methodology.\n\n**Political cleavages and democracy**\n\nEconomic divides do not stop at the marketplace; they spill directly into politics. Inequality shapes who is represented, whose voices carry weight, and how coalitions are built, or fail to be built. **Figure 15** shows how the traditional class-based alignment of politics in Western democracies has broken down. ³ In the mid-20th century, lower-income and less educated voters largely supported left-wing parties, while wealthier and more educated groups leaned right, creating a clear class divide and rising redistribution. Today, that pattern has fractured. First, education and income now point in different directions (see **Figure 15** ), making broad coalitions for redistribution far harder to sustain. This evolution can be accounted for by the fact that educational expansion has come with a complexification of the class structure. For example, many high-degree but relatively low-income voters (e.g., teachers or nurses) currently vote for the left, while many voters with lower degrees but relatively higher income (e.g.,\n\nself-employed or truck drivers) tend to vote for the right. The even more striking evolution is the rise of territorial divides within countries. In many advanced democracies, gaps in political affiliations between large metropolitan centers and smaller towns have reached levels unseen in a century (see **Figure 16** ). Unequal access to public services (education, health, transportation, and other infrastructures), job opportunities, and exposure to trade shocks has fractured social cohesion and weakened the coalitions necessary for redistributive reform. As a consequence, working-class voters are now fragmented across parties on both sides of the aisle or left without strong representation, which limits their political influence and entrenches inequality. In order to reactivate the redistributive coalitions of the postwar era, it is critical to design more ambitious policy platforms benefiting all territories, as they successfully did in the past. This fragmentation erodes the political foundations needed to tackle inequality\n\n17","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p17.pdf"]},"response":{"errors":[],"extract_id":"extract_cd7578ee6067c4209f548d62334204be","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Executive Summary\n\n**Figure 5. Women persistently receive lower labor income than men everywhere**\n\n40%\n\n36%\n\n16%\n\n40% 37%\n\n20%\n\n28%\n\n34%\n\nGender parity\n\n0%\n\n10%\n\n20%\n\n30%\n\n40%\n\n50%\n\n60%\n\nMiddle East\n\n& North\n\nAfrica\n\nSouth & Southeast\n\nAsia\n\nSub−\n\nSaharan\n\nAfrica\n\nEast Asia\n\nLatin\n\nAmerica\n\nRussia & Central\n\nAsia\n\nEurope North America & Oceania\n\nFemale labor income share (%)\n\n1990 2000 2010 2020 2025\n\n**Female labor income shares, 1990−2025**\n\n**Interpretation.** This figure shows the evolution of the female labor income share between 1990 and 2025 across world regions. In 2025, female workers earn about 16% of total labor income in the Middle East & North Africa, but about 40% in North America & Oceania and Europe. At the global level, women earned 27.8% of labor income in 1990 and 28.2% in 2025. While some progress has been made, gender parity remains distant in all regions. **Sources and series:** Neef and Robilliard (2021), Gabrielli et al. (2024), and wir2026.wid.world/methodology.\n\n**Political cleavages and democracy**\n\nEconomic divides do not stop at the marketplace; they spill directly into politics. Inequality shapes who is represented, whose voices carry weight, and how coalitions are built, or fail to be built. **Figure 15** shows how the traditional class-based alignment of politics in Western democracies has broken down. ³ In the mid-20th century, lower-income and less educated voters largely supported left-wing parties, while wealthier and more educated groups leaned right, creating a clear class divide and rising redistribution. Today, that pattern has fractured. First, education and income now point in different directions (see **Figure 15** ), making broad coalitions for redistribution far harder to sustain. This evolution can be accounted for by the fact that educational expansion has come with a complexification of the class structure. For example, many high-degree but relatively low-income voters (e.g., teachers or nurses) currently vote for the left, while many voters with lower degrees but relatively higher income (e.g.,\n\nself-employed or truck drivers) tend to vote for the right. The even more striking evolution is the rise of territorial divides within countries. In many advanced democracies, gaps in political affiliations between large metropolitan centers and smaller towns have reached levels unseen in a century (see **Figure 16** ). Unequal access to public services (education, health, transportation, and other infrastructures), job opportunities, and exposure to trade shocks has fractured social cohesion and weakened the coalitions necessary for redistributive reform. As a consequence, working-class voters are now fragmented across parties on both sides of the aisle or left without strong representation, which limits their political influence and entrenches inequality. In order to reactivate the redistributive coalitions of the postwar era, it is critical to design more ambitious policy platforms benefiting all territories, as they successfully did in the past. This fragmentation erodes the political foundations needed to tackle inequality\n\n17"],"full_content":"Executive Summary\n\n**Figure 5. Women persistently receive lower labor income than men everywhere**\n\n40%\n\n36%\n\n16%\n\n40% 37%\n\n20%\n\n28%\n\n34%\n\nGender parity\n\n0%\n\n10%\n\n20%\n\n30%\n\n40%\n\n50%\n\n60%\n\nMiddle East\n\n& North\n\nAfrica\n\nSouth & Southeast\n\nAsia\n\nSub−\n\nSaharan\n\nAfrica\n\nEast Asia\n\nLatin\n\nAmerica\n\nRussia & Central\n\nAsia\n\nEurope North America & Oceania\n\nFemale labor income share (%)\n\n1990 2000 2010 2020 2025\n\n**Female labor income shares, 1990−2025**\n\n**Interpretation.** This figure shows the evolution of the female labor income share between 1990 and 2025 across world regions. In 2025, female workers earn about 16% of total labor income in the Middle East & North Africa, but about 40% in North America & Oceania and Europe. At the global level, women earned 27.8% of labor income in 1990 and 28.2% in 2025. While some progress has been made, gender parity remains distant in all regions. **Sources and series:** Neef and Robilliard (2021), Gabrielli et al. (2024), and wir2026.wid.world/methodology.\n\n**Political cleavages and democracy**\n\nEconomic divides do not stop at the marketplace; they spill directly into politics. Inequality shapes who is represented, whose voices carry weight, and how coalitions are built, or fail to be built. **Figure 15** shows how the traditional class-based alignment of politics in Western democracies has broken down. ³ In the mid-20th century, lower-income and less educated voters largely supported left-wing parties, while wealthier and more educated groups leaned right, creating a clear class divide and rising redistribution. Today, that pattern has fractured. First, education and income now point in different directions (see **Figure 15** ), making broad coalitions for redistribution far harder to sustain. This evolution can be accounted for by the fact that educational expansion has come with a complexification of the class structure. For example, many high-degree but relatively low-income voters (e.g., teachers or nurses) currently vote for the left, while many voters with lower degrees but relatively higher income (e.g.,\n\nself-employed or truck drivers) tend to vote for the right. The even more striking evolution is the rise of territorial divides within countries. In many advanced democracies, gaps in political affiliations between large metropolitan centers and smaller towns have reached levels unseen in a century (see **Figure 16** ). Unequal access to public services (education, health, transportation, and other infrastructures), job opportunities, and exposure to trade shocks has fractured social cohesion and weakened the coalitions necessary for redistributive reform. As a consequence, working-class voters are now fragmented across parties on both sides of the aisle or left without strong representation, which limits their political influence and entrenches inequality. In order to reactivate the redistributive coalitions of the postwar era, it is critical to design more ambitious policy platforms benefiting all territories, as they successfully did in the past. This fragmentation erodes the political foundations needed to tackle inequality\n\n17","publish_date":null,"title":"Executive Summary Figure5. Womenpersistentlyreceivelowerlaborincomethan meneverywhere self-employed or truck drivers) tend to vote for the right. Economic divides do not stop","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p17.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p17.pdf"},"request":{"example_id":"chart/World_Inequality_Report_2026_p17","product_type":"parse","source_file_path":"World_Inequality_Report_2026_p17.pdf"},"started_at":"2026-11-04T19:45:30.740167"} +{"completed_at":"2026-11-04T19:45:32.619451","latency_in_ms":1078,"output":{"example_id":"chart/World_Inequality_Report_2026_p18","grounded_pages":[],"layout_pages":[],"markdown":"Executive Summary\n\n**Figure 6. After including domestic labor, women earn only 32% of men’s hourly income**\n\n53\n\n43\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\n70\n\nWomen Men\n\nAverage labor time (hours per week)\n\n(15−to−64−year−old) (2020−2025)\n\n**Hours worked**\n\n**per week**\n\n61%\n\n32%\n\n0%\n\n10%\n\n20%\n\n30%\n\n40%\n\n50%\n\n60%\n\n70%\n\nConventional Real\n\nFemale hourly income\n\n(% male hourly income) (2020–2025)\n\n**Women's hourly income**\n\n**(% of men's)**\n\nDomestic labor Economic labor Excluding domestic labor Including domestic labor\n\n**Interpretation.** The left panel shows that, globally, women work more hours per week than men once both economic and domestic labor are counted. The right panel shows that women’s hourly income is substantially lower than men’s: the measured gap (39%=100%−61%) is smaller when only economic labor is considered, but becomes much larger once domestic labor hours are included (68%=100%−32%). Together, the two figures highlight the double burden women face: more total work time combined with lower hourly returns to their labor. **Notes.** Economic labor includes paid activities recorded in national accounts. Domestic labor includes household tasks, cooking, and care work. Calculations from Andreescu et al. (2025) using global time−use and income data. **Sources and series:** Andreescu et al. (2025).\n\n**Gender gap including domestic labor hours, 2020−2025**\n\nand prevents the implementation of redistributive policies. Meanwhile, the influence of wealth in politics compounds the inequality in political influence. **Figure 17** shows how campaign financing is heavily concentrated among the top earners: in countries like France and South Korea, the richest 10% of citizens disproportionately provide the majority of political donations. This concentration of financial power amplifies elite voices, narrows the space for equitable policymaking, and further marginalizes the working majority. Reducing inequality is a political choice. But fragmented electorates, underrepresentation of workers, and the outsized influence of wealth all work against the coalitions needed for reform. This reality can change. It reflects political choices about campaign finance rules, party strategies, and institutional design that can be reshaped with sufficient will. Building the conditions for consensus is therefore as central to reducing inequality as any specific policy instrument.\n\n**Policy directions**\n\nThe evidence makes one conclusion clear: inequality can be reduced. There are a range of policies that, in different ways, have proven effective in narrowing gaps. One important avenue is through public investments in education and health. These are among the most powerful equalizers, yet access to these basic services remains uneven and stratified. Public investment in free, high-quality schools, universal healthcare, childcare, and nutrition programs can reduce early-life disparities and foster lifelong learning opportunities. By ensuring that talent and effort, rather than background, determine life chances, such investments build more inclusive and resilient societies. Another path is through redistributive programs. Cash transfers, pensions, unemployment benefits, and targeted support for vulnerable households can directly shift resources from the top to the bottom of the distribution. Where well designed, such measures have narrowed\n\n18","pages":[{"line_numbers":[],"markdown":"Executive Summary\n\n**Figure 6. After including domestic labor, women earn only 32% of men’s hourly income**\n\n53\n\n43\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\n70\n\nWomen Men\n\nAverage labor time (hours per week)\n\n(15−to−64−year−old) (2020−2025)\n\n**Hours worked**\n\n**per week**\n\n61%\n\n32%\n\n0%\n\n10%\n\n20%\n\n30%\n\n40%\n\n50%\n\n60%\n\n70%\n\nConventional Real\n\nFemale hourly income\n\n(% male hourly income) (2020–2025)\n\n**Women's hourly income**\n\n**(% of men's)**\n\nDomestic labor Economic labor Excluding domestic labor Including domestic labor\n\n**Interpretation.** The left panel shows that, globally, women work more hours per week than men once both economic and domestic labor are counted. The right panel shows that women’s hourly income is substantially lower than men’s: the measured gap (39%=100%−61%) is smaller when only economic labor is considered, but becomes much larger once domestic labor hours are included (68%=100%−32%). Together, the two figures highlight the double burden women face: more total work time combined with lower hourly returns to their labor. **Notes.** Economic labor includes paid activities recorded in national accounts. Domestic labor includes household tasks, cooking, and care work. Calculations from Andreescu et al. (2025) using global time−use and income data. **Sources and series:** Andreescu et al. (2025).\n\n**Gender gap including domestic labor hours, 2020−2025**\n\nand prevents the implementation of redistributive policies. Meanwhile, the influence of wealth in politics compounds the inequality in political influence. **Figure 17** shows how campaign financing is heavily concentrated among the top earners: in countries like France and South Korea, the richest 10% of citizens disproportionately provide the majority of political donations. This concentration of financial power amplifies elite voices, narrows the space for equitable policymaking, and further marginalizes the working majority. Reducing inequality is a political choice. But fragmented electorates, underrepresentation of workers, and the outsized influence of wealth all work against the coalitions needed for reform. This reality can change. It reflects political choices about campaign finance rules, party strategies, and institutional design that can be reshaped with sufficient will. Building the conditions for consensus is therefore as central to reducing inequality as any specific policy instrument.\n\n**Policy directions**\n\nThe evidence makes one conclusion clear: inequality can be reduced. There are a range of policies that, in different ways, have proven effective in narrowing gaps. One important avenue is through public investments in education and health. These are among the most powerful equalizers, yet access to these basic services remains uneven and stratified. Public investment in free, high-quality schools, universal healthcare, childcare, and nutrition programs can reduce early-life disparities and foster lifelong learning opportunities. By ensuring that talent and effort, rather than background, determine life chances, such investments build more inclusive and resilient societies. Another path is through redistributive programs. Cash transfers, pensions, unemployment benefits, and targeted support for vulnerable households can directly shift resources from the top to the bottom of the distribution. Where well designed, such measures have narrowed\n\n18","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Executive Summary\n\n**Figure 6. After including domestic labor, women earn only 32% of men’s hourly income**\n\n53\n\n43\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\n70\n\nWomen Men\n\nAverage labor time (hours per week)\n\n(15−to−64−year−old) (2020−2025)\n\n**Hours worked**\n\n**per week**\n\n61%\n\n32%\n\n0%\n\n10%\n\n20%\n\n30%\n\n40%\n\n50%\n\n60%\n\n70%\n\nConventional Real\n\nFemale hourly income\n\n(% male hourly income) (2020–2025)\n\n**Women's hourly income**\n\n**(% of men's)**\n\nDomestic labor Economic labor Excluding domestic labor Including domestic labor\n\n**Interpretation.** The left panel shows that, globally, women work more hours per week than men once both economic and domestic labor are counted. The right panel shows that women’s hourly income is substantially lower than men’s: the measured gap (39%=100%−61%) is smaller when only economic labor is considered, but becomes much larger once domestic labor hours are included (68%=100%−32%). Together, the two figures highlight the double burden women face: more total work time combined with lower hourly returns to their labor. **Notes.** Economic labor includes paid activities recorded in national accounts. Domestic labor includes household tasks, cooking, and care work. Calculations from Andreescu et al. (2025) using global time−use and income data. **Sources and series:** Andreescu et al. (2025).\n\n**Gender gap including domestic labor hours, 2020−2025**\n\nand prevents the implementation of redistributive policies. Meanwhile, the influence of wealth in politics compounds the inequality in political influence. **Figure 17** shows how campaign financing is heavily concentrated among the top earners: in countries like France and South Korea, the richest 10% of citizens disproportionately provide the majority of political donations. This concentration of financial power amplifies elite voices, narrows the space for equitable policymaking, and further marginalizes the working majority. Reducing inequality is a political choice. But fragmented electorates, underrepresentation of workers, and the outsized influence of wealth all work against the coalitions needed for reform. This reality can change. It reflects political choices about campaign finance rules, party strategies, and institutional design that can be reshaped with sufficient will. Building the conditions for consensus is therefore as central to reducing inequality as any specific policy instrument.\n\n**Policy directions**\n\nThe evidence makes one conclusion clear: inequality can be reduced. There are a range of policies that, in different ways, have proven effective in narrowing gaps. One important avenue is through public investments in education and health. These are among the most powerful equalizers, yet access to these basic services remains uneven and stratified. Public investment in free, high-quality schools, universal healthcare, childcare, and nutrition programs can reduce early-life disparities and foster lifelong learning opportunities. By ensuring that talent and effort, rather than background, determine life chances, such investments build more inclusive and resilient societies. Another path is through redistributive programs. Cash transfers, pensions, unemployment benefits, and targeted support for vulnerable households can directly shift resources from the top to the bottom of the distribution. Where well designed, such measures have narrowed\n\n18","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p18.pdf"]},"response":{"errors":[],"extract_id":"extract_321d5f1f0b12fb1f1de547c359a91555","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Executive Summary\n\n**Figure 6. After including domestic labor, women earn only 32% of men’s hourly income**\n\n53\n\n43\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\n70\n\nWomen Men\n\nAverage labor time (hours per week)\n\n(15−to−64−year−old) (2020−2025)\n\n**Hours worked**\n\n**per week**\n\n61%\n\n32%\n\n0%\n\n10%\n\n20%\n\n30%\n\n40%\n\n50%\n\n60%\n\n70%\n\nConventional Real\n\nFemale hourly income\n\n(% male hourly income) (2020–2025)\n\n**Women's hourly income**\n\n**(% of men's)**\n\nDomestic labor Economic labor Excluding domestic labor Including domestic labor\n\n**Interpretation.** The left panel shows that, globally, women work more hours per week than men once both economic and domestic labor are counted. The right panel shows that women’s hourly income is substantially lower than men’s: the measured gap (39%=100%−61%) is smaller when only economic labor is considered, but becomes much larger once domestic labor hours are included (68%=100%−32%). Together, the two figures highlight the double burden women face: more total work time combined with lower hourly returns to their labor. **Notes.** Economic labor includes paid activities recorded in national accounts. Domestic labor includes household tasks, cooking, and care work. Calculations from Andreescu et al. (2025) using global time−use and income data. **Sources and series:** Andreescu et al. (2025).\n\n**Gender gap including domestic labor hours, 2020−2025**\n\nand prevents the implementation of redistributive policies. Meanwhile, the influence of wealth in politics compounds the inequality in political influence. **Figure 17** shows how campaign financing is heavily concentrated among the top earners: in countries like France and South Korea, the richest 10% of citizens disproportionately provide the majority of political donations. This concentration of financial power amplifies elite voices, narrows the space for equitable policymaking, and further marginalizes the working majority. Reducing inequality is a political choice. But fragmented electorates, underrepresentation of workers, and the outsized influence of wealth all work against the coalitions needed for reform. This reality can change. It reflects political choices about campaign finance rules, party strategies, and institutional design that can be reshaped with sufficient will. Building the conditions for consensus is therefore as central to reducing inequality as any specific policy instrument.\n\n**Policy directions**\n\nThe evidence makes one conclusion clear: inequality can be reduced. There are a range of policies that, in different ways, have proven effective in narrowing gaps. One important avenue is through public investments in education and health. These are among the most powerful equalizers, yet access to these basic services remains uneven and stratified. Public investment in free, high-quality schools, universal healthcare, childcare, and nutrition programs can reduce early-life disparities and foster lifelong learning opportunities. By ensuring that talent and effort, rather than background, determine life chances, such investments build more inclusive and resilient societies. Another path is through redistributive programs. Cash transfers, pensions, unemployment benefits, and targeted support for vulnerable households can directly shift resources from the top to the bottom of the distribution. Where well designed, such measures have narrowed\n\n18"],"full_content":"Executive Summary\n\n**Figure 6. After including domestic labor, women earn only 32% of men’s hourly income**\n\n53\n\n43\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\n70\n\nWomen Men\n\nAverage labor time (hours per week)\n\n(15−to−64−year−old) (2020−2025)\n\n**Hours worked**\n\n**per week**\n\n61%\n\n32%\n\n0%\n\n10%\n\n20%\n\n30%\n\n40%\n\n50%\n\n60%\n\n70%\n\nConventional Real\n\nFemale hourly income\n\n(% male hourly income) (2020–2025)\n\n**Women's hourly income**\n\n**(% of men's)**\n\nDomestic labor Economic labor Excluding domestic labor Including domestic labor\n\n**Interpretation.** The left panel shows that, globally, women work more hours per week than men once both economic and domestic labor are counted. The right panel shows that women’s hourly income is substantially lower than men’s: the measured gap (39%=100%−61%) is smaller when only economic labor is considered, but becomes much larger once domestic labor hours are included (68%=100%−32%). Together, the two figures highlight the double burden women face: more total work time combined with lower hourly returns to their labor. **Notes.** Economic labor includes paid activities recorded in national accounts. Domestic labor includes household tasks, cooking, and care work. Calculations from Andreescu et al. (2025) using global time−use and income data. **Sources and series:** Andreescu et al. (2025).\n\n**Gender gap including domestic labor hours, 2020−2025**\n\nand prevents the implementation of redistributive policies. Meanwhile, the influence of wealth in politics compounds the inequality in political influence. **Figure 17** shows how campaign financing is heavily concentrated among the top earners: in countries like France and South Korea, the richest 10% of citizens disproportionately provide the majority of political donations. This concentration of financial power amplifies elite voices, narrows the space for equitable policymaking, and further marginalizes the working majority. Reducing inequality is a political choice. But fragmented electorates, underrepresentation of workers, and the outsized influence of wealth all work against the coalitions needed for reform. This reality can change. It reflects political choices about campaign finance rules, party strategies, and institutional design that can be reshaped with sufficient will. Building the conditions for consensus is therefore as central to reducing inequality as any specific policy instrument.\n\n**Policy directions**\n\nThe evidence makes one conclusion clear: inequality can be reduced. There are a range of policies that, in different ways, have proven effective in narrowing gaps. One important avenue is through public investments in education and health. These are among the most powerful equalizers, yet access to these basic services remains uneven and stratified. Public investment in free, high-quality schools, universal healthcare, childcare, and nutrition programs can reduce early-life disparities and foster lifelong learning opportunities. By ensuring that talent and effort, rather than background, determine life chances, such investments build more inclusive and resilient societies. Another path is through redistributive programs. Cash transfers, pensions, unemployment benefits, and targeted support for vulnerable households can directly shift resources from the top to the bottom of the distribution. Where well designed, such measures have narrowed\n\n18","publish_date":null,"title":"Executive Summary Figure6. Afterincludingdomesticlabor,womenearnonly32% ofmen’shourlyincome Gender gap including domestic labor hours, 2020−2025 and prevents the implementation of redistributive policies. Meanwhile,","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p18.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p18.pdf"},"request":{"example_id":"chart/World_Inequality_Report_2026_p18","product_type":"parse","source_file_path":"World_Inequality_Report_2026_p18.pdf"},"started_at":"2026-11-04T19:45:31.540552"} +{"completed_at":"2026-11-04T19:45:35.068193","latency_in_ms":2445,"output":{"example_id":"chart/World_Inequality_Report_2026_p198","grounded_pages":[],"layout_pages":[],"markdown":"Country-sheets\n\n# **VIETNAM**\n\n101,598,527 **||** e761 (avg. monthly income, PPP)\n\n**Inequality in Vietnam**\n\nIn Vietnam, inequality is moderate and has remained broadly stable over the past decade. The top 10% of earners receive 43% of total income, while the bottom 50% account for 16%. Wealth is more concentrated, with the richest 10% holding 59% of total wealth and the top 1% holding 25%. Average income per capita is 9,100 euros (PPP), and average wealth stands at 38,700 euros (PPP). The income gap between the top 10% and the bottom 50% increased slightly from 26.7 to 27.2 between 2014 and 2024, indicating stable but persistent disparities. Female labor participation remained nearly unchanged, moving from 39.6% to 38.8%. Overall, Vietnam shows stable income distribution alongside a persistent concentration of wealth at the top. ‘\n\n**Table 1:** **_Inequality outlook – Vietnam_**\n\n**Avg.**\n\n**Income** **(PPP €)**\n\n**Share**\n\n**of**\n\n**total (%)**\n\n**Avg.**\n\n**Wealth** **(PPP €)**\n\n**Share**\n\n**of**\n\n**total (%)**\n\n**Full pop.** 9,126 100.0% 38,655 100.0%\n\n**Bottom**\n\n**50%** 1,448 15.8% 1,778 4.6%\n\n**Middle**\n\n**40%** 9,368 41.1% 35,079 36.3%\n\n**Top 10%** 39,382 43.2% 228,062 59.0%\n\n**Top 1%** 141,660 15.5% 954,768 24.7%\n\n**2014 2024**\n\n26\\.7 27.2\n\n39\\.6% 38.8%\n\n**Income Wealth**\n\n**Year**\n\n**Top 10% to Bot. 50%**\n\n**Income gap**\n\n**Female labor share**\n\n**_Interpretation:_** Country has a transparency index of 5/20. All values are estimated at per capita (full population) level. See glossary for all definitions of concepts and indicators. **_Sources and series:_** wir2026.wid.world/methodology.\n\n**Figure 1:** **_Top 10% and bottom 50% income shares in Vietnam, 1980-2024_**\n\n**_Interpretation:_** The Top 10% income share is equal to 43% in 2024. Income is measured after the operation of pensions and unemployment insurance systems and before income tax. **_Sources and series:_** wir2026.wid.world/methodology.\n\n198","pages":[{"line_numbers":[],"markdown":"Country-sheets\n\n# **VIETNAM**\n\n101,598,527 **||** e761 (avg. monthly income, PPP)\n\n**Inequality in Vietnam**\n\nIn Vietnam, inequality is moderate and has remained broadly stable over the past decade. The top 10% of earners receive 43% of total income, while the bottom 50% account for 16%. Wealth is more concentrated, with the richest 10% holding 59% of total wealth and the top 1% holding 25%. Average income per capita is 9,100 euros (PPP), and average wealth stands at 38,700 euros (PPP). The income gap between the top 10% and the bottom 50% increased slightly from 26.7 to 27.2 between 2014 and 2024, indicating stable but persistent disparities. Female labor participation remained nearly unchanged, moving from 39.6% to 38.8%. Overall, Vietnam shows stable income distribution alongside a persistent concentration of wealth at the top. ‘\n\n**Table 1:** **_Inequality outlook – Vietnam_**\n\n**Avg.**\n\n**Income** **(PPP €)**\n\n**Share**\n\n**of**\n\n**total (%)**\n\n**Avg.**\n\n**Wealth** **(PPP €)**\n\n**Share**\n\n**of**\n\n**total (%)**\n\n**Full pop.** 9,126 100.0% 38,655 100.0%\n\n**Bottom**\n\n**50%** 1,448 15.8% 1,778 4.6%\n\n**Middle**\n\n**40%** 9,368 41.1% 35,079 36.3%\n\n**Top 10%** 39,382 43.2% 228,062 59.0%\n\n**Top 1%** 141,660 15.5% 954,768 24.7%\n\n**2014 2024**\n\n26\\.7 27.2\n\n39\\.6% 38.8%\n\n**Income Wealth**\n\n**Year**\n\n**Top 10% to Bot. 50%**\n\n**Income gap**\n\n**Female labor share**\n\n**_Interpretation:_** Country has a transparency index of 5/20. All values are estimated at per capita (full population) level. See glossary for all definitions of concepts and indicators. **_Sources and series:_** wir2026.wid.world/methodology.\n\n**Figure 1:** **_Top 10% and bottom 50% income shares in Vietnam, 1980-2024_**\n\n**_Interpretation:_** The Top 10% income share is equal to 43% in 2024. Income is measured after the operation of pensions and unemployment insurance systems and before income tax. **_Sources and series:_** wir2026.wid.world/methodology.\n\n198","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Country-sheets\n\n# **VIETNAM**\n\n101,598,527 **||** e761 (avg. monthly income, PPP)\n\n**Inequality in Vietnam**\n\nIn Vietnam, inequality is moderate and has remained broadly stable over the past decade. The top 10% of earners receive 43% of total income, while the bottom 50% account for 16%. Wealth is more concentrated, with the richest 10% holding 59% of total wealth and the top 1% holding 25%. Average income per capita is 9,100 euros (PPP), and average wealth stands at 38,700 euros (PPP). The income gap between the top 10% and the bottom 50% increased slightly from 26.7 to 27.2 between 2014 and 2024, indicating stable but persistent disparities. Female labor participation remained nearly unchanged, moving from 39.6% to 38.8%. Overall, Vietnam shows stable income distribution alongside a persistent concentration of wealth at the top. ‘\n\n**Table 1:** **_Inequality outlook – Vietnam_**\n\n**Avg.**\n\n**Income** **(PPP €)**\n\n**Share**\n\n**of**\n\n**total (%)**\n\n**Avg.**\n\n**Wealth** **(PPP €)**\n\n**Share**\n\n**of**\n\n**total (%)**\n\n**Full pop.** 9,126 100.0% 38,655 100.0%\n\n**Bottom**\n\n**50%** 1,448 15.8% 1,778 4.6%\n\n**Middle**\n\n**40%** 9,368 41.1% 35,079 36.3%\n\n**Top 10%** 39,382 43.2% 228,062 59.0%\n\n**Top 1%** 141,660 15.5% 954,768 24.7%\n\n**2014 2024**\n\n26\\.7 27.2\n\n39\\.6% 38.8%\n\n**Income Wealth**\n\n**Year**\n\n**Top 10% to Bot. 50%**\n\n**Income gap**\n\n**Female labor share**\n\n**_Interpretation:_** Country has a transparency index of 5/20. All values are estimated at per capita (full population) level. See glossary for all definitions of concepts and indicators. **_Sources and series:_** wir2026.wid.world/methodology.\n\n**Figure 1:** **_Top 10% and bottom 50% income shares in Vietnam, 1980-2024_**\n\n**_Interpretation:_** The Top 10% income share is equal to 43% in 2024. Income is measured after the operation of pensions and unemployment insurance systems and before income tax. **_Sources and series:_** wir2026.wid.world/methodology.\n\n198","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p198.pdf"]},"response":{"errors":[],"extract_id":"extract_78a859569f9c3681e8f29c17785bf192","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Country-sheets\n\n# **VIETNAM**\n\n101,598,527 **||** e761 (avg. monthly income, PPP)\n\n**Inequality in Vietnam**\n\nIn Vietnam, inequality is moderate and has remained broadly stable over the past decade. The top 10% of earners receive 43% of total income, while the bottom 50% account for 16%. Wealth is more concentrated, with the richest 10% holding 59% of total wealth and the top 1% holding 25%. Average income per capita is 9,100 euros (PPP), and average wealth stands at 38,700 euros (PPP). The income gap between the top 10% and the bottom 50% increased slightly from 26.7 to 27.2 between 2014 and 2024, indicating stable but persistent disparities. Female labor participation remained nearly unchanged, moving from 39.6% to 38.8%. Overall, Vietnam shows stable income distribution alongside a persistent concentration of wealth at the top. ‘\n\n**Table 1:** **_Inequality outlook – Vietnam_**\n\n**Avg.**\n\n**Income** **(PPP €)**\n\n**Share**\n\n**of**\n\n**total (%)**\n\n**Avg.**\n\n**Wealth** **(PPP €)**\n\n**Share**\n\n**of**\n\n**total (%)**\n\n**Full pop.** 9,126 100.0% 38,655 100.0%\n\n**Bottom**\n\n**50%** 1,448 15.8% 1,778 4.6%\n\n**Middle**\n\n**40%** 9,368 41.1% 35,079 36.3%\n\n**Top 10%** 39,382 43.2% 228,062 59.0%\n\n**Top 1%** 141,660 15.5% 954,768 24.7%\n\n**2014 2024**\n\n26\\.7 27.2\n\n39\\.6% 38.8%\n\n**Income Wealth**\n\n**Year**\n\n**Top 10% to Bot. 50%**\n\n**Income gap**\n\n**Female labor share**\n\n**_Interpretation:_** Country has a transparency index of 5/20. All values are estimated at per capita (full population) level. See glossary for all definitions of concepts and indicators. **_Sources and series:_** wir2026.wid.world/methodology.\n\n**Figure 1:** **_Top 10% and bottom 50% income shares in Vietnam, 1980-2024_**\n\n**_Interpretation:_** The Top 10% income share is equal to 43% in 2024. Income is measured after the operation of pensions and unemployment insurance systems and before income tax. **_Sources and series:_** wir2026.wid.world/methodology.\n\n198"],"full_content":"Country-sheets\n\n# **VIETNAM**\n\n101,598,527 **||** e761 (avg. monthly income, PPP)\n\n**Inequality in Vietnam**\n\nIn Vietnam, inequality is moderate and has remained broadly stable over the past decade. The top 10% of earners receive 43% of total income, while the bottom 50% account for 16%. Wealth is more concentrated, with the richest 10% holding 59% of total wealth and the top 1% holding 25%. Average income per capita is 9,100 euros (PPP), and average wealth stands at 38,700 euros (PPP). The income gap between the top 10% and the bottom 50% increased slightly from 26.7 to 27.2 between 2014 and 2024, indicating stable but persistent disparities. Female labor participation remained nearly unchanged, moving from 39.6% to 38.8%. Overall, Vietnam shows stable income distribution alongside a persistent concentration of wealth at the top. ‘\n\n**Table 1:** **_Inequality outlook – Vietnam_**\n\n**Avg.**\n\n**Income** **(PPP €)**\n\n**Share**\n\n**of**\n\n**total (%)**\n\n**Avg.**\n\n**Wealth** **(PPP €)**\n\n**Share**\n\n**of**\n\n**total (%)**\n\n**Full pop.** 9,126 100.0% 38,655 100.0%\n\n**Bottom**\n\n**50%** 1,448 15.8% 1,778 4.6%\n\n**Middle**\n\n**40%** 9,368 41.1% 35,079 36.3%\n\n**Top 10%** 39,382 43.2% 228,062 59.0%\n\n**Top 1%** 141,660 15.5% 954,768 24.7%\n\n**2014 2024**\n\n26\\.7 27.2\n\n39\\.6% 38.8%\n\n**Income Wealth**\n\n**Year**\n\n**Top 10% to Bot. 50%**\n\n**Income gap**\n\n**Female labor share**\n\n**_Interpretation:_** Country has a transparency index of 5/20. All values are estimated at per capita (full population) level. See glossary for all definitions of concepts and indicators. **_Sources and series:_** wir2026.wid.world/methodology.\n\n**Figure 1:** **_Top 10% and bottom 50% income shares in Vietnam, 1980-2024_**\n\n**_Interpretation:_** The Top 10% income share is equal to 43% in 2024. Income is measured after the operation of pensions and unemployment insurance systems and before income tax. **_Sources and series:_** wir2026.wid.world/methodology.\n\n198","publish_date":null,"title":"VIETNAM","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p198.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p198.pdf"},"request":{"example_id":"chart/World_Inequality_Report_2026_p198","product_type":"parse","source_file_path":"World_Inequality_Report_2026_p198.pdf"},"started_at":"2026-11-04T19:45:32.622304"} +{"completed_at":"2026-11-04T19:45:34.696417","latency_in_ms":1065,"output":{"example_id":"chart/World_Inequality_Report_2026_p20","grounded_pages":[],"layout_pages":[],"markdown":"Executive Summary\n\n**Figure 8. Income and, even more, wealth are extremely concentrated at the top in every region**\n\n0%\n\n10% 20% 30% 40% 50% 60% 70% 80%\n\nEURO\n\nEASA\n\nNAOC\n\nRUCA\n\nSSAF\n\nSSEA\n\nMENA\n\nLATA\n\nShare of national income (%)\n\n**Income**\n\n0%\n\n10% 20% 30% 40% 50% 60% 70% 80%\n\nEURO\n\nSSEA\n\nNAOC\n\nLATA\n\nEASA\n\nSSAF\n\nMENA\n\nRUCA\n\nShare of personal wealth (%)\n\n**Wealth**\n\nBottom 50% Middle 40% Top 10%\n\n**Interpretation.** In every region, income and wealth are distributed very unequally within regions. Wealth is much more concentrated at the top than income.The figures are arranged according to top 10% shares. Income is measured after pension and unemployment benefits are received by individuals, but before income taxes and other transfers. Net personal wealth is the sum of financial (e.g., equity, bonds) and non−financial assets (e.g., housing, land) owned by individuals, net of debts. **Notes.** EASA: East Asia, EURO: Europe, LATA: Latin America, MENA: Middle East & North Africa, NAOC: North America & Oceania, SSEA: South & Southeast Asia, SSAF: Sub−Saharan Africa, and RUCA: Russia & Central Asia. **Sources and series:** wir2026.wid.world/methodology.\n\n**Inequality within regions, 2025**\n\n**Conclusion**\n\nInequality is a political choice. It is the result of our policies, institutions, and governance structures. The costs of escalating inequality are clear: widening divides, fragile democracies, and a climate crisis borne most heavily by those least responsible. But the possibilities of reform are equally clear. Where redistribution is strong, taxation is fair, and social investment is prioritized, inequality narrows. The tools exist. The challenge is political will. The choices we make in the coming years will determine whether the global economy continues down a path of extreme concentration or moves toward shared prosperity.\n\n20","pages":[{"line_numbers":[],"markdown":"Executive Summary\n\n**Figure 8. Income and, even more, wealth are extremely concentrated at the top in every region**\n\n0%\n\n10% 20% 30% 40% 50% 60% 70% 80%\n\nEURO\n\nEASA\n\nNAOC\n\nRUCA\n\nSSAF\n\nSSEA\n\nMENA\n\nLATA\n\nShare of national income (%)\n\n**Income**\n\n0%\n\n10% 20% 30% 40% 50% 60% 70% 80%\n\nEURO\n\nSSEA\n\nNAOC\n\nLATA\n\nEASA\n\nSSAF\n\nMENA\n\nRUCA\n\nShare of personal wealth (%)\n\n**Wealth**\n\nBottom 50% Middle 40% Top 10%\n\n**Interpretation.** In every region, income and wealth are distributed very unequally within regions. Wealth is much more concentrated at the top than income.The figures are arranged according to top 10% shares. Income is measured after pension and unemployment benefits are received by individuals, but before income taxes and other transfers. Net personal wealth is the sum of financial (e.g., equity, bonds) and non−financial assets (e.g., housing, land) owned by individuals, net of debts. **Notes.** EASA: East Asia, EURO: Europe, LATA: Latin America, MENA: Middle East & North Africa, NAOC: North America & Oceania, SSEA: South & Southeast Asia, SSAF: Sub−Saharan Africa, and RUCA: Russia & Central Asia. **Sources and series:** wir2026.wid.world/methodology.\n\n**Inequality within regions, 2025**\n\n**Conclusion**\n\nInequality is a political choice. It is the result of our policies, institutions, and governance structures. The costs of escalating inequality are clear: widening divides, fragile democracies, and a climate crisis borne most heavily by those least responsible. But the possibilities of reform are equally clear. Where redistribution is strong, taxation is fair, and social investment is prioritized, inequality narrows. The tools exist. The challenge is political will. The choices we make in the coming years will determine whether the global economy continues down a path of extreme concentration or moves toward shared prosperity.\n\n20","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Executive Summary\n\n**Figure 8. Income and, even more, wealth are extremely concentrated at the top in every region**\n\n0%\n\n10% 20% 30% 40% 50% 60% 70% 80%\n\nEURO\n\nEASA\n\nNAOC\n\nRUCA\n\nSSAF\n\nSSEA\n\nMENA\n\nLATA\n\nShare of national income (%)\n\n**Income**\n\n0%\n\n10% 20% 30% 40% 50% 60% 70% 80%\n\nEURO\n\nSSEA\n\nNAOC\n\nLATA\n\nEASA\n\nSSAF\n\nMENA\n\nRUCA\n\nShare of personal wealth (%)\n\n**Wealth**\n\nBottom 50% Middle 40% Top 10%\n\n**Interpretation.** In every region, income and wealth are distributed very unequally within regions. Wealth is much more concentrated at the top than income.The figures are arranged according to top 10% shares. Income is measured after pension and unemployment benefits are received by individuals, but before income taxes and other transfers. Net personal wealth is the sum of financial (e.g., equity, bonds) and non−financial assets (e.g., housing, land) owned by individuals, net of debts. **Notes.** EASA: East Asia, EURO: Europe, LATA: Latin America, MENA: Middle East & North Africa, NAOC: North America & Oceania, SSEA: South & Southeast Asia, SSAF: Sub−Saharan Africa, and RUCA: Russia & Central Asia. **Sources and series:** wir2026.wid.world/methodology.\n\n**Inequality within regions, 2025**\n\n**Conclusion**\n\nInequality is a political choice. It is the result of our policies, institutions, and governance structures. The costs of escalating inequality are clear: widening divides, fragile democracies, and a climate crisis borne most heavily by those least responsible. But the possibilities of reform are equally clear. Where redistribution is strong, taxation is fair, and social investment is prioritized, inequality narrows. The tools exist. The challenge is political will. The choices we make in the coming years will determine whether the global economy continues down a path of extreme concentration or moves toward shared prosperity.\n\n20","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p20.pdf"]},"response":{"errors":[],"extract_id":"extract_e8744ede84449d0ec4091a52bef81e63","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Executive Summary\n\n**Figure 8. Income and, even more, wealth are extremely concentrated at the top in every region**\n\n0%\n\n10% 20% 30% 40% 50% 60% 70% 80%\n\nEURO\n\nEASA\n\nNAOC\n\nRUCA\n\nSSAF\n\nSSEA\n\nMENA\n\nLATA\n\nShare of national income (%)\n\n**Income**\n\n0%\n\n10% 20% 30% 40% 50% 60% 70% 80%\n\nEURO\n\nSSEA\n\nNAOC\n\nLATA\n\nEASA\n\nSSAF\n\nMENA\n\nRUCA\n\nShare of personal wealth (%)\n\n**Wealth**\n\nBottom 50% Middle 40% Top 10%\n\n**Interpretation.** In every region, income and wealth are distributed very unequally within regions. Wealth is much more concentrated at the top than income.The figures are arranged according to top 10% shares. Income is measured after pension and unemployment benefits are received by individuals, but before income taxes and other transfers. Net personal wealth is the sum of financial (e.g., equity, bonds) and non−financial assets (e.g., housing, land) owned by individuals, net of debts. **Notes.** EASA: East Asia, EURO: Europe, LATA: Latin America, MENA: Middle East & North Africa, NAOC: North America & Oceania, SSEA: South & Southeast Asia, SSAF: Sub−Saharan Africa, and RUCA: Russia & Central Asia. **Sources and series:** wir2026.wid.world/methodology.\n\n**Inequality within regions, 2025**\n\n**Conclusion**\n\nInequality is a political choice. It is the result of our policies, institutions, and governance structures. The costs of escalating inequality are clear: widening divides, fragile democracies, and a climate crisis borne most heavily by those least responsible. But the possibilities of reform are equally clear. Where redistribution is strong, taxation is fair, and social investment is prioritized, inequality narrows. The tools exist. The challenge is political will. The choices we make in the coming years will determine whether the global economy continues down a path of extreme concentration or moves toward shared prosperity.\n\n20"],"full_content":"Executive Summary\n\n**Figure 8. Income and, even more, wealth are extremely concentrated at the top in every region**\n\n0%\n\n10% 20% 30% 40% 50% 60% 70% 80%\n\nEURO\n\nEASA\n\nNAOC\n\nRUCA\n\nSSAF\n\nSSEA\n\nMENA\n\nLATA\n\nShare of national income (%)\n\n**Income**\n\n0%\n\n10% 20% 30% 40% 50% 60% 70% 80%\n\nEURO\n\nSSEA\n\nNAOC\n\nLATA\n\nEASA\n\nSSAF\n\nMENA\n\nRUCA\n\nShare of personal wealth (%)\n\n**Wealth**\n\nBottom 50% Middle 40% Top 10%\n\n**Interpretation.** In every region, income and wealth are distributed very unequally within regions. Wealth is much more concentrated at the top than income.The figures are arranged according to top 10% shares. Income is measured after pension and unemployment benefits are received by individuals, but before income taxes and other transfers. Net personal wealth is the sum of financial (e.g., equity, bonds) and non−financial assets (e.g., housing, land) owned by individuals, net of debts. **Notes.** EASA: East Asia, EURO: Europe, LATA: Latin America, MENA: Middle East & North Africa, NAOC: North America & Oceania, SSEA: South & Southeast Asia, SSAF: Sub−Saharan Africa, and RUCA: Russia & Central Asia. **Sources and series:** wir2026.wid.world/methodology.\n\n**Inequality within regions, 2025**\n\n**Conclusion**\n\nInequality is a political choice. It is the result of our policies, institutions, and governance structures. The costs of escalating inequality are clear: widening divides, fragile democracies, and a climate crisis borne most heavily by those least responsible. But the possibilities of reform are equally clear. Where redistribution is strong, taxation is fair, and social investment is prioritized, inequality narrows. The tools exist. The challenge is political will. The choices we make in the coming years will determine whether the global economy continues down a path of extreme concentration or moves toward shared prosperity.\n\n20","publish_date":null,"title":"Executive Summary Figure8. Incomeand,evenmore,wealthareextremely concentratedatthetopineveryregion O O O O M M Inequality is a political choice. It is the result","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p20.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p20.pdf"},"request":{"example_id":"chart/World_Inequality_Report_2026_p20","product_type":"parse","source_file_path":"World_Inequality_Report_2026_p20.pdf"},"started_at":"2026-11-04T19:45:33.631157"} +{"completed_at":"2026-11-04T19:45:39.145401","latency_in_ms":4203,"output":{"example_id":"chart/World_Inequality_Report_2026_p21","grounded_pages":[],"layout_pages":[],"markdown":"Executive Summary\n\n**Figure 9. Some countries face the double burden of low incomes and very high inequality**\n\nTop 10/bottom 50 income gaps across the world, 2025\n\n09 – 19 19 – 28 28 – 34 34 – 41 41 – 103\n\n**Top 10/bottom 50 income gaps across the world, 2025**\n\n**Interpretation.** This map shows the ratio between the average income of the top 10% and the average income of the bottom 50% of the population in each country in 2025. Income is measured after pension and unemployment benefits are received by individuals, but before other taxes they pay and transfers they receive. **Sources and series:** wir2026.wid.world/methodology and Chancel and Piketty (2021).\n\n**Figure 10. Inequality can be reduced with progressive taxation and transfers**\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n1980 1985 1990 1995 2000 2005 2010 2015 2020\n\nRedistribution: reduction in\n\ntop 10% to bottom 50% income ratio\n\nthrough taxes and transfers (%)\n\nEast Asia\n\nEurope\n\nLatin America\n\nMENA\n\nNorth America & Oceania\n\nRussia & Central Asia\n\nSouth & Southeast Asia\n\nSub−Saharan Africa\n\n**Intepretation.** The figure shows the impact of taxes and transfers on inequality across regions, measured by the reduction in the top 10% to bottom 50% income ratio (a positive value indicates inequality reduction). Tax and transfer systems reduce inequality in all regions, but the extent of redistribution varies greatly. **Sources and series:** wir2026.wid.world/methodology and Fisher−Post and Gethin (2025).\n\n**Redistribution, 1980−2025: reduction in top 10/bottom 50**\n\n**income gaps through taxes and transfers**\n\n21","pages":[{"line_numbers":[],"markdown":"Executive Summary\n\n**Figure 9. Some countries face the double burden of low incomes and very high inequality**\n\nTop 10/bottom 50 income gaps across the world, 2025\n\n09 – 19 19 – 28 28 – 34 34 – 41 41 – 103\n\n**Top 10/bottom 50 income gaps across the world, 2025**\n\n**Interpretation.** This map shows the ratio between the average income of the top 10% and the average income of the bottom 50% of the population in each country in 2025. Income is measured after pension and unemployment benefits are received by individuals, but before other taxes they pay and transfers they receive. **Sources and series:** wir2026.wid.world/methodology and Chancel and Piketty (2021).\n\n**Figure 10. Inequality can be reduced with progressive taxation and transfers**\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n1980 1985 1990 1995 2000 2005 2010 2015 2020\n\nRedistribution: reduction in\n\ntop 10% to bottom 50% income ratio\n\nthrough taxes and transfers (%)\n\nEast Asia\n\nEurope\n\nLatin America\n\nMENA\n\nNorth America & Oceania\n\nRussia & Central Asia\n\nSouth & Southeast Asia\n\nSub−Saharan Africa\n\n**Intepretation.** The figure shows the impact of taxes and transfers on inequality across regions, measured by the reduction in the top 10% to bottom 50% income ratio (a positive value indicates inequality reduction). Tax and transfer systems reduce inequality in all regions, but the extent of redistribution varies greatly. **Sources and series:** wir2026.wid.world/methodology and Fisher−Post and Gethin (2025).\n\n**Redistribution, 1980−2025: reduction in top 10/bottom 50**\n\n**income gaps through taxes and transfers**\n\n21","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Executive Summary\n\n**Figure 9. Some countries face the double burden of low incomes and very high inequality**\n\nTop 10/bottom 50 income gaps across the world, 2025\n\n09 – 19 19 – 28 28 – 34 34 – 41 41 – 103\n\n**Top 10/bottom 50 income gaps across the world, 2025**\n\n**Interpretation.** This map shows the ratio between the average income of the top 10% and the average income of the bottom 50% of the population in each country in 2025. Income is measured after pension and unemployment benefits are received by individuals, but before other taxes they pay and transfers they receive. **Sources and series:** wir2026.wid.world/methodology and Chancel and Piketty (2021).\n\n**Figure 10. Inequality can be reduced with progressive taxation and transfers**\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n1980 1985 1990 1995 2000 2005 2010 2015 2020\n\nRedistribution: reduction in\n\ntop 10% to bottom 50% income ratio\n\nthrough taxes and transfers (%)\n\nEast Asia\n\nEurope\n\nLatin America\n\nMENA\n\nNorth America & Oceania\n\nRussia & Central Asia\n\nSouth & Southeast Asia\n\nSub−Saharan Africa\n\n**Intepretation.** The figure shows the impact of taxes and transfers on inequality across regions, measured by the reduction in the top 10% to bottom 50% income ratio (a positive value indicates inequality reduction). Tax and transfer systems reduce inequality in all regions, but the extent of redistribution varies greatly. **Sources and series:** wir2026.wid.world/methodology and Fisher−Post and Gethin (2025).\n\n**Redistribution, 1980−2025: reduction in top 10/bottom 50**\n\n**income gaps through taxes and transfers**\n\n21","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p21.pdf"]},"response":{"errors":[],"extract_id":"extract_df59f77e428f6028d40f4bfb28e4ef75","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Executive Summary\n\n**Figure 9. Some countries face the double burden of low incomes and very high inequality**\n\nTop 10/bottom 50 income gaps across the world, 2025\n\n09 – 19 19 – 28 28 – 34 34 – 41 41 – 103\n\n**Top 10/bottom 50 income gaps across the world, 2025**\n\n**Interpretation.** This map shows the ratio between the average income of the top 10% and the average income of the bottom 50% of the population in each country in 2025. Income is measured after pension and unemployment benefits are received by individuals, but before other taxes they pay and transfers they receive. **Sources and series:** wir2026.wid.world/methodology and Chancel and Piketty (2021).\n\n**Figure 10. Inequality can be reduced with progressive taxation and transfers**\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n1980 1985 1990 1995 2000 2005 2010 2015 2020\n\nRedistribution: reduction in\n\ntop 10% to bottom 50% income ratio\n\nthrough taxes and transfers (%)\n\nEast Asia\n\nEurope\n\nLatin America\n\nMENA\n\nNorth America & Oceania\n\nRussia & Central Asia\n\nSouth & Southeast Asia\n\nSub−Saharan Africa\n\n**Intepretation.** The figure shows the impact of taxes and transfers on inequality across regions, measured by the reduction in the top 10% to bottom 50% income ratio (a positive value indicates inequality reduction). Tax and transfer systems reduce inequality in all regions, but the extent of redistribution varies greatly. **Sources and series:** wir2026.wid.world/methodology and Fisher−Post and Gethin (2025).\n\n**Redistribution, 1980−2025: reduction in top 10/bottom 50**\n\n**income gaps through taxes and transfers**\n\n21"],"full_content":"Executive Summary\n\n**Figure 9. Some countries face the double burden of low incomes and very high inequality**\n\nTop 10/bottom 50 income gaps across the world, 2025\n\n09 – 19 19 – 28 28 – 34 34 – 41 41 – 103\n\n**Top 10/bottom 50 income gaps across the world, 2025**\n\n**Interpretation.** This map shows the ratio between the average income of the top 10% and the average income of the bottom 50% of the population in each country in 2025. Income is measured after pension and unemployment benefits are received by individuals, but before other taxes they pay and transfers they receive. **Sources and series:** wir2026.wid.world/methodology and Chancel and Piketty (2021).\n\n**Figure 10. Inequality can be reduced with progressive taxation and transfers**\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n1980 1985 1990 1995 2000 2005 2010 2015 2020\n\nRedistribution: reduction in\n\ntop 10% to bottom 50% income ratio\n\nthrough taxes and transfers (%)\n\nEast Asia\n\nEurope\n\nLatin America\n\nMENA\n\nNorth America & Oceania\n\nRussia & Central Asia\n\nSouth & Southeast Asia\n\nSub−Saharan Africa\n\n**Intepretation.** The figure shows the impact of taxes and transfers on inequality across regions, measured by the reduction in the top 10% to bottom 50% income ratio (a positive value indicates inequality reduction). Tax and transfer systems reduce inequality in all regions, but the extent of redistribution varies greatly. **Sources and series:** wir2026.wid.world/methodology and Fisher−Post and Gethin (2025).\n\n**Redistribution, 1980−2025: reduction in top 10/bottom 50**\n\n**income gaps through taxes and transfers**\n\n21","publish_date":null,"title":"Executive Summary Figure9. Somecountriesfacethedoubleburdenoflowincomes andveryhighinequality Figure10. Inequalitycanbereducedwithprogressivetaxation andtransfers 21","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p21.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p21.pdf"},"request":{"example_id":"chart/World_Inequality_Report_2026_p21","product_type":"parse","source_file_path":"World_Inequality_Report_2026_p21.pdf"},"started_at":"2026-11-04T19:45:34.941422"} +{"completed_at":"2026-11-04T19:45:38.694030","latency_in_ms":2519,"output":{"example_id":"chart/World_Inequality_Report_2026_p22","grounded_pages":[],"layout_pages":[],"markdown":"Executive Summary\n\n**Figure 11. Large inequality of opportunity across regions**\n\n220\n\n1,444 1,823\n\n2,518\n\n7,433\n\n593\n\n2,941\n\n1,642\n\n9,025\n\n0\n\n1,000\n\n2,000\n\n3,000\n\n4,000\n\n5,000\n\n6,000\n\n7,000\n\n8,000\n\n9,000\n\n10,000\n\nSub−\n\nSaharan\n\nAfrica\n\nSouth &\n\nSoutheast\n\nAsia\n\nMiddle East &\n\nNorth Africa\n\nWorld Latin\n\nAmerica\n\nRussia & Central\n\nAsia\n\nEast Asia\n\nEurope North America & Oceania\n\nPublic education expenditure (PPP € 2025)\n\nper school−age individual (0–24)\n\n**Public education expenditure per school−age individual (0−24), 2025**\n\n**Interpretation.** In 2025, average public education expenditure per school−age individual (0−to−24−year−old) varies enormously across world regions, from €220 in Sub−Saharan Africa to €9,025 in North America & Oceania (PPP € 2025), i.e., a gap of almost 1 to 41. If we were using market exchange rates (MERs) rather than PPPs, the gaps would be 2–3 times larger. **Sources and series:** _World Human Capital Expenditure Database_ (whce.world) and Bharti et al. (2025).\n\n**Figure 12. The ultra-rich escape progressive taxation**\n\n99% of the population are located in the shaded area.\n\n0%\n\n5%\n\n10%\n\n15%\n\n20%\n\n25%\n\nP0−10\n\nP10−20\n\nP20−30\n\nP30−40\n\nP40−50\n\nP50−60\n\nP60−70\n\nP70−80\n\nP80−90\n\nP90−95\n\nP95−99\n\nP99−99.9\n\nP99.9−99.99\n\nP99.99−P99.999\n\nP99.999−P99.9999\n\nBillionaires\n\nEffective income tax rates by\n\nincome groups and for\n\nbillionaires (% of pre−tax\n\nincome)\n\nBrazil France Netherlands Spain United States\n\n**Interpretation.** This figure shows effective income tax rates by pre−tax income group and for U.S. dollar billionaires in Brazil, France, the Netherlands, Spain, and the United States. Income tax rates include only individual income taxes and equivalent levies. All values are expressed as a share of pre−tax income, defined as all national income before taxes and transfers, after pensions. P0−10 denotes the bottom 10% of the income distribution, P10−20 the next decile, etc. **Sources and series:** Artola et al. (2022), Bozio et al. (2024), Bozio et al. (2020), Bruil et al. (2024), Palomo et al. (2025), Saez and Zucman (2019), and Zucman (2024).\n\n**Effective income tax rates by income groups**\n\n22","pages":[{"line_numbers":[],"markdown":"Executive Summary\n\n**Figure 11. Large inequality of opportunity across regions**\n\n220\n\n1,444 1,823\n\n2,518\n\n7,433\n\n593\n\n2,941\n\n1,642\n\n9,025\n\n0\n\n1,000\n\n2,000\n\n3,000\n\n4,000\n\n5,000\n\n6,000\n\n7,000\n\n8,000\n\n9,000\n\n10,000\n\nSub−\n\nSaharan\n\nAfrica\n\nSouth &\n\nSoutheast\n\nAsia\n\nMiddle East &\n\nNorth Africa\n\nWorld Latin\n\nAmerica\n\nRussia & Central\n\nAsia\n\nEast Asia\n\nEurope North America & Oceania\n\nPublic education expenditure (PPP € 2025)\n\nper school−age individual (0–24)\n\n**Public education expenditure per school−age individual (0−24), 2025**\n\n**Interpretation.** In 2025, average public education expenditure per school−age individual (0−to−24−year−old) varies enormously across world regions, from €220 in Sub−Saharan Africa to €9,025 in North America & Oceania (PPP € 2025), i.e., a gap of almost 1 to 41. If we were using market exchange rates (MERs) rather than PPPs, the gaps would be 2–3 times larger. **Sources and series:** _World Human Capital Expenditure Database_ (whce.world) and Bharti et al. (2025).\n\n**Figure 12. The ultra-rich escape progressive taxation**\n\n99% of the population are located in the shaded area.\n\n0%\n\n5%\n\n10%\n\n15%\n\n20%\n\n25%\n\nP0−10\n\nP10−20\n\nP20−30\n\nP30−40\n\nP40−50\n\nP50−60\n\nP60−70\n\nP70−80\n\nP80−90\n\nP90−95\n\nP95−99\n\nP99−99.9\n\nP99.9−99.99\n\nP99.99−P99.999\n\nP99.999−P99.9999\n\nBillionaires\n\nEffective income tax rates by\n\nincome groups and for\n\nbillionaires (% of pre−tax\n\nincome)\n\nBrazil France Netherlands Spain United States\n\n**Interpretation.** This figure shows effective income tax rates by pre−tax income group and for U.S. dollar billionaires in Brazil, France, the Netherlands, Spain, and the United States. Income tax rates include only individual income taxes and equivalent levies. All values are expressed as a share of pre−tax income, defined as all national income before taxes and transfers, after pensions. P0−10 denotes the bottom 10% of the income distribution, P10−20 the next decile, etc. **Sources and series:** Artola et al. (2022), Bozio et al. (2024), Bozio et al. (2020), Bruil et al. (2024), Palomo et al. (2025), Saez and Zucman (2019), and Zucman (2024).\n\n**Effective income tax rates by income groups**\n\n22","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Executive Summary\n\n**Figure 11. Large inequality of opportunity across regions**\n\n220\n\n1,444 1,823\n\n2,518\n\n7,433\n\n593\n\n2,941\n\n1,642\n\n9,025\n\n0\n\n1,000\n\n2,000\n\n3,000\n\n4,000\n\n5,000\n\n6,000\n\n7,000\n\n8,000\n\n9,000\n\n10,000\n\nSub−\n\nSaharan\n\nAfrica\n\nSouth &\n\nSoutheast\n\nAsia\n\nMiddle East &\n\nNorth Africa\n\nWorld Latin\n\nAmerica\n\nRussia & Central\n\nAsia\n\nEast Asia\n\nEurope North America & Oceania\n\nPublic education expenditure (PPP € 2025)\n\nper school−age individual (0–24)\n\n**Public education expenditure per school−age individual (0−24), 2025**\n\n**Interpretation.** In 2025, average public education expenditure per school−age individual (0−to−24−year−old) varies enormously across world regions, from €220 in Sub−Saharan Africa to €9,025 in North America & Oceania (PPP € 2025), i.e., a gap of almost 1 to 41. If we were using market exchange rates (MERs) rather than PPPs, the gaps would be 2–3 times larger. **Sources and series:** _World Human Capital Expenditure Database_ (whce.world) and Bharti et al. (2025).\n\n**Figure 12. The ultra-rich escape progressive taxation**\n\n99% of the population are located in the shaded area.\n\n0%\n\n5%\n\n10%\n\n15%\n\n20%\n\n25%\n\nP0−10\n\nP10−20\n\nP20−30\n\nP30−40\n\nP40−50\n\nP50−60\n\nP60−70\n\nP70−80\n\nP80−90\n\nP90−95\n\nP95−99\n\nP99−99.9\n\nP99.9−99.99\n\nP99.99−P99.999\n\nP99.999−P99.9999\n\nBillionaires\n\nEffective income tax rates by\n\nincome groups and for\n\nbillionaires (% of pre−tax\n\nincome)\n\nBrazil France Netherlands Spain United States\n\n**Interpretation.** This figure shows effective income tax rates by pre−tax income group and for U.S. dollar billionaires in Brazil, France, the Netherlands, Spain, and the United States. Income tax rates include only individual income taxes and equivalent levies. All values are expressed as a share of pre−tax income, defined as all national income before taxes and transfers, after pensions. P0−10 denotes the bottom 10% of the income distribution, P10−20 the next decile, etc. **Sources and series:** Artola et al. (2022), Bozio et al. (2024), Bozio et al. (2020), Bruil et al. (2024), Palomo et al. (2025), Saez and Zucman (2019), and Zucman (2024).\n\n**Effective income tax rates by income groups**\n\n22","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p22.pdf"]},"response":{"errors":[],"extract_id":"extract_7745c24b1e292f08c336fac204570970","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Executive Summary\n\n**Figure 11. Large inequality of opportunity across regions**\n\n220\n\n1,444 1,823\n\n2,518\n\n7,433\n\n593\n\n2,941\n\n1,642\n\n9,025\n\n0\n\n1,000\n\n2,000\n\n3,000\n\n4,000\n\n5,000\n\n6,000\n\n7,000\n\n8,000\n\n9,000\n\n10,000\n\nSub−\n\nSaharan\n\nAfrica\n\nSouth &\n\nSoutheast\n\nAsia\n\nMiddle East &\n\nNorth Africa\n\nWorld Latin\n\nAmerica\n\nRussia & Central\n\nAsia\n\nEast Asia\n\nEurope North America & Oceania\n\nPublic education expenditure (PPP € 2025)\n\nper school−age individual (0–24)\n\n**Public education expenditure per school−age individual (0−24), 2025**\n\n**Interpretation.** In 2025, average public education expenditure per school−age individual (0−to−24−year−old) varies enormously across world regions, from €220 in Sub−Saharan Africa to €9,025 in North America & Oceania (PPP € 2025), i.e., a gap of almost 1 to 41. If we were using market exchange rates (MERs) rather than PPPs, the gaps would be 2–3 times larger. **Sources and series:** _World Human Capital Expenditure Database_ (whce.world) and Bharti et al. (2025).\n\n**Figure 12. The ultra-rich escape progressive taxation**\n\n99% of the population are located in the shaded area.\n\n0%\n\n5%\n\n10%\n\n15%\n\n20%\n\n25%\n\nP0−10\n\nP10−20\n\nP20−30\n\nP30−40\n\nP40−50\n\nP50−60\n\nP60−70\n\nP70−80\n\nP80−90\n\nP90−95\n\nP95−99\n\nP99−99.9\n\nP99.9−99.99\n\nP99.99−P99.999\n\nP99.999−P99.9999\n\nBillionaires\n\nEffective income tax rates by\n\nincome groups and for\n\nbillionaires (% of pre−tax\n\nincome)\n\nBrazil France Netherlands Spain United States\n\n**Interpretation.** This figure shows effective income tax rates by pre−tax income group and for U.S. dollar billionaires in Brazil, France, the Netherlands, Spain, and the United States. Income tax rates include only individual income taxes and equivalent levies. All values are expressed as a share of pre−tax income, defined as all national income before taxes and transfers, after pensions. P0−10 denotes the bottom 10% of the income distribution, P10−20 the next decile, etc. **Sources and series:** Artola et al. (2022), Bozio et al. (2024), Bozio et al. (2020), Bruil et al. (2024), Palomo et al. (2025), Saez and Zucman (2019), and Zucman (2024).\n\n**Effective income tax rates by income groups**\n\n22"],"full_content":"Executive Summary\n\n**Figure 11. Large inequality of opportunity across regions**\n\n220\n\n1,444 1,823\n\n2,518\n\n7,433\n\n593\n\n2,941\n\n1,642\n\n9,025\n\n0\n\n1,000\n\n2,000\n\n3,000\n\n4,000\n\n5,000\n\n6,000\n\n7,000\n\n8,000\n\n9,000\n\n10,000\n\nSub−\n\nSaharan\n\nAfrica\n\nSouth &\n\nSoutheast\n\nAsia\n\nMiddle East &\n\nNorth Africa\n\nWorld Latin\n\nAmerica\n\nRussia & Central\n\nAsia\n\nEast Asia\n\nEurope North America & Oceania\n\nPublic education expenditure (PPP € 2025)\n\nper school−age individual (0–24)\n\n**Public education expenditure per school−age individual (0−24), 2025**\n\n**Interpretation.** In 2025, average public education expenditure per school−age individual (0−to−24−year−old) varies enormously across world regions, from €220 in Sub−Saharan Africa to €9,025 in North America & Oceania (PPP € 2025), i.e., a gap of almost 1 to 41. If we were using market exchange rates (MERs) rather than PPPs, the gaps would be 2–3 times larger. **Sources and series:** _World Human Capital Expenditure Database_ (whce.world) and Bharti et al. (2025).\n\n**Figure 12. The ultra-rich escape progressive taxation**\n\n99% of the population are located in the shaded area.\n\n0%\n\n5%\n\n10%\n\n15%\n\n20%\n\n25%\n\nP0−10\n\nP10−20\n\nP20−30\n\nP30−40\n\nP40−50\n\nP50−60\n\nP60−70\n\nP70−80\n\nP80−90\n\nP90−95\n\nP95−99\n\nP99−99.9\n\nP99.9−99.99\n\nP99.99−P99.999\n\nP99.999−P99.9999\n\nBillionaires\n\nEffective income tax rates by\n\nincome groups and for\n\nbillionaires (% of pre−tax\n\nincome)\n\nBrazil France Netherlands Spain United States\n\n**Interpretation.** This figure shows effective income tax rates by pre−tax income group and for U.S. dollar billionaires in Brazil, France, the Netherlands, Spain, and the United States. Income tax rates include only individual income taxes and equivalent levies. All values are expressed as a share of pre−tax income, defined as all national income before taxes and transfers, after pensions. P0−10 denotes the bottom 10% of the income distribution, P10−20 the next decile, etc. **Sources and series:** Artola et al. (2022), Bozio et al. (2024), Bozio et al. (2020), Bruil et al. (2024), Palomo et al. (2025), Saez and Zucman (2019), and Zucman (2024).\n\n**Effective income tax rates by income groups**\n\n22","publish_date":null,"title":"Executive Summary Figure11. Largeinequalityofopportunityacrossregions Public education expenditure per school−age individual (0−24), 2025 Figure12. Theultra-richescapeprogressivetaxation Effective income tax rates by income","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p22.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p22.pdf"},"request":{"example_id":"chart/World_Inequality_Report_2026_p22","product_type":"parse","source_file_path":"World_Inequality_Report_2026_p22.pdf"},"started_at":"2026-11-04T19:45:36.174478"} +{"completed_at":"2026-11-04T19:45:41.542794","latency_in_ms":4083,"output":{"example_id":"chart/World_Inequality_Report_2026_p23","grounded_pages":[],"layout_pages":[],"markdown":"Executive Summary\n\n**Figure 13. The international financial system generates more inequality**\n\nThe centrality of the U.S. dollar, and now other currencies, in the global financial system allows rich economies to borrow at very low rates.\n\n−5% −4% −3% −2% −1%\n\n0% 1% 2% 3% 4% 5% 6% 7%\n\n1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 2020 2025\n\nExcess yield as %\n\nof country GDP, MER\n\nBRICS\n\nChina\n\nEurozone\n\nJapan\n\nRussia\n\nUnited States\n\n**Excess yield (assets−liabilities) as % of country GDP, 1970−2025**\n\n**Interpretation.** This graph shows excess yield income, defined as the difference between the return on foreign assets and liabilities, as a share of national GDP. The figure shows that the exorbitant privilege once exclusive to the United States has become a broader rich−world phenomenon. The United States maintains a substantial privilege of 2.2% in 2025. The Eurozone follows with 1% by 2025. Japan stands out with a privilege of 5.9% by 2025. In contrast, BRICS countries face a consistent burden of around 2.1%, highlighting their role as net providers of capital to wealthier economies. **Notes.** Positive values represent income gains from financial privilege; negative values represent financial burden. BRICS countries comprise Brazil, Russia, India, China, and South Africa. **Sources and series:** Nievas and Sodano (2025) and wir2026.wid.world/methodology.\n\n**Figure 14. Privileged countries face lower liability costs by political design, not market dynamics**\n\n0%\n\n10%\n\n20%\n\n30%\n\n40%\n\n50%\n\n60%\n\n70%\n\n80%\n\n90%\n\n1976 1986 1996 2006 2016\n\nShare of global reserves by currency\n\nU.S. dollar\n\nEuro\n\nPound sterling\n\nJapanese yen\n\nChinese renminbi\n\nOther currencies\n\nSwiss franc\n\n**Interpretation.** Rich countries are the issuers of international reserve currencies, which are then used in international transactions and as a reserve of value around the globe.These currencies dominate central bank reserves due to international financial rules like Basel III, locking in persistent demand. This leads to persistently lower borrowing costs. **Sources and series:** Nievas and Sodano (2025) and wir2026.wid.world/methodology.\n\n**Share of global reserves by currency, 1976−2022**\n\n23","pages":[{"line_numbers":[],"markdown":"Executive Summary\n\n**Figure 13. The international financial system generates more inequality**\n\nThe centrality of the U.S. dollar, and now other currencies, in the global financial system allows rich economies to borrow at very low rates.\n\n−5% −4% −3% −2% −1%\n\n0% 1% 2% 3% 4% 5% 6% 7%\n\n1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 2020 2025\n\nExcess yield as %\n\nof country GDP, MER\n\nBRICS\n\nChina\n\nEurozone\n\nJapan\n\nRussia\n\nUnited States\n\n**Excess yield (assets−liabilities) as % of country GDP, 1970−2025**\n\n**Interpretation.** This graph shows excess yield income, defined as the difference between the return on foreign assets and liabilities, as a share of national GDP. The figure shows that the exorbitant privilege once exclusive to the United States has become a broader rich−world phenomenon. The United States maintains a substantial privilege of 2.2% in 2025. The Eurozone follows with 1% by 2025. Japan stands out with a privilege of 5.9% by 2025. In contrast, BRICS countries face a consistent burden of around 2.1%, highlighting their role as net providers of capital to wealthier economies. **Notes.** Positive values represent income gains from financial privilege; negative values represent financial burden. BRICS countries comprise Brazil, Russia, India, China, and South Africa. **Sources and series:** Nievas and Sodano (2025) and wir2026.wid.world/methodology.\n\n**Figure 14. Privileged countries face lower liability costs by political design, not market dynamics**\n\n0%\n\n10%\n\n20%\n\n30%\n\n40%\n\n50%\n\n60%\n\n70%\n\n80%\n\n90%\n\n1976 1986 1996 2006 2016\n\nShare of global reserves by currency\n\nU.S. dollar\n\nEuro\n\nPound sterling\n\nJapanese yen\n\nChinese renminbi\n\nOther currencies\n\nSwiss franc\n\n**Interpretation.** Rich countries are the issuers of international reserve currencies, which are then used in international transactions and as a reserve of value around the globe.These currencies dominate central bank reserves due to international financial rules like Basel III, locking in persistent demand. This leads to persistently lower borrowing costs. **Sources and series:** Nievas and Sodano (2025) and wir2026.wid.world/methodology.\n\n**Share of global reserves by currency, 1976−2022**\n\n23","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Executive Summary\n\n**Figure 13. The international financial system generates more inequality**\n\nThe centrality of the U.S. dollar, and now other currencies, in the global financial system allows rich economies to borrow at very low rates.\n\n−5% −4% −3% −2% −1%\n\n0% 1% 2% 3% 4% 5% 6% 7%\n\n1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 2020 2025\n\nExcess yield as %\n\nof country GDP, MER\n\nBRICS\n\nChina\n\nEurozone\n\nJapan\n\nRussia\n\nUnited States\n\n**Excess yield (assets−liabilities) as % of country GDP, 1970−2025**\n\n**Interpretation.** This graph shows excess yield income, defined as the difference between the return on foreign assets and liabilities, as a share of national GDP. The figure shows that the exorbitant privilege once exclusive to the United States has become a broader rich−world phenomenon. The United States maintains a substantial privilege of 2.2% in 2025. The Eurozone follows with 1% by 2025. Japan stands out with a privilege of 5.9% by 2025. In contrast, BRICS countries face a consistent burden of around 2.1%, highlighting their role as net providers of capital to wealthier economies. **Notes.** Positive values represent income gains from financial privilege; negative values represent financial burden. BRICS countries comprise Brazil, Russia, India, China, and South Africa. **Sources and series:** Nievas and Sodano (2025) and wir2026.wid.world/methodology.\n\n**Figure 14. Privileged countries face lower liability costs by political design, not market dynamics**\n\n0%\n\n10%\n\n20%\n\n30%\n\n40%\n\n50%\n\n60%\n\n70%\n\n80%\n\n90%\n\n1976 1986 1996 2006 2016\n\nShare of global reserves by currency\n\nU.S. dollar\n\nEuro\n\nPound sterling\n\nJapanese yen\n\nChinese renminbi\n\nOther currencies\n\nSwiss franc\n\n**Interpretation.** Rich countries are the issuers of international reserve currencies, which are then used in international transactions and as a reserve of value around the globe.These currencies dominate central bank reserves due to international financial rules like Basel III, locking in persistent demand. This leads to persistently lower borrowing costs. **Sources and series:** Nievas and Sodano (2025) and wir2026.wid.world/methodology.\n\n**Share of global reserves by currency, 1976−2022**\n\n23","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p23.pdf"]},"response":{"errors":[],"extract_id":"extract_081cef632f9de92b82c5a4b7495d2eae","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Executive Summary\n\n**Figure 13. The international financial system generates more inequality**\n\nThe centrality of the U.S. dollar, and now other currencies, in the global financial system allows rich economies to borrow at very low rates.\n\n−5% −4% −3% −2% −1%\n\n0% 1% 2% 3% 4% 5% 6% 7%\n\n1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 2020 2025\n\nExcess yield as %\n\nof country GDP, MER\n\nBRICS\n\nChina\n\nEurozone\n\nJapan\n\nRussia\n\nUnited States\n\n**Excess yield (assets−liabilities) as % of country GDP, 1970−2025**\n\n**Interpretation.** This graph shows excess yield income, defined as the difference between the return on foreign assets and liabilities, as a share of national GDP. The figure shows that the exorbitant privilege once exclusive to the United States has become a broader rich−world phenomenon. The United States maintains a substantial privilege of 2.2% in 2025. The Eurozone follows with 1% by 2025. Japan stands out with a privilege of 5.9% by 2025. In contrast, BRICS countries face a consistent burden of around 2.1%, highlighting their role as net providers of capital to wealthier economies. **Notes.** Positive values represent income gains from financial privilege; negative values represent financial burden. BRICS countries comprise Brazil, Russia, India, China, and South Africa. **Sources and series:** Nievas and Sodano (2025) and wir2026.wid.world/methodology.\n\n**Figure 14. Privileged countries face lower liability costs by political design, not market dynamics**\n\n0%\n\n10%\n\n20%\n\n30%\n\n40%\n\n50%\n\n60%\n\n70%\n\n80%\n\n90%\n\n1976 1986 1996 2006 2016\n\nShare of global reserves by currency\n\nU.S. dollar\n\nEuro\n\nPound sterling\n\nJapanese yen\n\nChinese renminbi\n\nOther currencies\n\nSwiss franc\n\n**Interpretation.** Rich countries are the issuers of international reserve currencies, which are then used in international transactions and as a reserve of value around the globe.These currencies dominate central bank reserves due to international financial rules like Basel III, locking in persistent demand. This leads to persistently lower borrowing costs. **Sources and series:** Nievas and Sodano (2025) and wir2026.wid.world/methodology.\n\n**Share of global reserves by currency, 1976−2022**\n\n23"],"full_content":"Executive Summary\n\n**Figure 13. The international financial system generates more inequality**\n\nThe centrality of the U.S. dollar, and now other currencies, in the global financial system allows rich economies to borrow at very low rates.\n\n−5% −4% −3% −2% −1%\n\n0% 1% 2% 3% 4% 5% 6% 7%\n\n1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 2020 2025\n\nExcess yield as %\n\nof country GDP, MER\n\nBRICS\n\nChina\n\nEurozone\n\nJapan\n\nRussia\n\nUnited States\n\n**Excess yield (assets−liabilities) as % of country GDP, 1970−2025**\n\n**Interpretation.** This graph shows excess yield income, defined as the difference between the return on foreign assets and liabilities, as a share of national GDP. The figure shows that the exorbitant privilege once exclusive to the United States has become a broader rich−world phenomenon. The United States maintains a substantial privilege of 2.2% in 2025. The Eurozone follows with 1% by 2025. Japan stands out with a privilege of 5.9% by 2025. In contrast, BRICS countries face a consistent burden of around 2.1%, highlighting their role as net providers of capital to wealthier economies. **Notes.** Positive values represent income gains from financial privilege; negative values represent financial burden. BRICS countries comprise Brazil, Russia, India, China, and South Africa. **Sources and series:** Nievas and Sodano (2025) and wir2026.wid.world/methodology.\n\n**Figure 14. Privileged countries face lower liability costs by political design, not market dynamics**\n\n0%\n\n10%\n\n20%\n\n30%\n\n40%\n\n50%\n\n60%\n\n70%\n\n80%\n\n90%\n\n1976 1986 1996 2006 2016\n\nShare of global reserves by currency\n\nU.S. dollar\n\nEuro\n\nPound sterling\n\nJapanese yen\n\nChinese renminbi\n\nOther currencies\n\nSwiss franc\n\n**Interpretation.** Rich countries are the issuers of international reserve currencies, which are then used in international transactions and as a reserve of value around the globe.These currencies dominate central bank reserves due to international financial rules like Basel III, locking in persistent demand. This leads to persistently lower borrowing costs. **Sources and series:** Nievas and Sodano (2025) and wir2026.wid.world/methodology.\n\n**Share of global reserves by currency, 1976−2022**\n\n23","publish_date":null,"title":"Executive Summary Figure13. Theinternationalfinancialsystemgeneratesmore inequality Figure14. Privilegedcountriesfacelowerliabilitycostsby politicaldesign,notmarketdynamics 23","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p23.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p23.pdf"},"request":{"example_id":"chart/World_Inequality_Report_2026_p23","product_type":"parse","source_file_path":"World_Inequality_Report_2026_p23.pdf"},"started_at":"2026-11-04T19:45:37.459525"} +{"completed_at":"2026-11-04T19:45:41.135707","latency_in_ms":2439,"output":{"example_id":"chart/World_Inequality_Report_2026_p24","grounded_pages":[],"layout_pages":[],"markdown":"Executive Summary\n\n**Figure 15. We need political action but political coalitions are difficult to form**\n\nHigher−educated voters voting for left−wing\n\nparties (democratic, labor, social− democratic, socialist, green, etc.)\n\nTop−income voters voting for right−wing\n\nparties (other parties) −20%\n\n−15%\n\n−10%\n\n−5%\n\n0%\n\n5%\n\n10%\n\n15%\n\n1960−64\n\n1965−69\n\n1970−74\n\n1975−79\n\n1980−84\n\n1985−89\n\n1990−94\n\n1995−99\n\n2000−04\n\n2005−09\n\n2010−14\n\n2015−19\n\n2020−25\n\nDifference (pp): top 10% minus\n\nbottom 90% voting left\n\n(% of top 10% educated voting left) minus (% of bottom 90% educated voting left)\n\n(% of top 10% earners voting left) minus (% of bottom 90% earners voting left)\n\n**Education and income divides in Western democracies, 1960−2025**\n\n**Interpretation.** In the 1960s, both higher−educated and high−income voters were less likely to vote for left−wing (democratic / labor / social−democratic / socialist / green) parties than lower−educated and low−income voters by more than 10 percentage points. The left vote has gradually become associated with higher education voters, giving rising to a multi−elite party system. Figures correspond to five−year averages for Australia, Britain, Canada, Denmark, France, Germany, Italy, the Netherlands, Norway, Sweden, Switzerland, and the U.S. Estimates control for income/education, age, gender, religion, church attendance, rural/urban, region, race/ethnicity, employment status, and marital status (in country−years for which these variables are available). **Sources and series:** Gethin et al. (2021) and _World Political Cleavages and Inequality Database_ (wpid.world).\n\n**Figure 16. Divides between large cities and smaller towns have reached levels unseen in a century**\n\n**Urban voters vote more for left parties**\n\n80% 90%\n\n100% 110% 120% 130% 140% 150% 160% 170% 180%\n\n1840\n\n1860\n\n1880\n\n1900\n\n1920\n\n1940\n\n1960\n\n1980\n\n2000\n\n2020\n\nLeft vote:\n\nurban 50% vs. rural 50%\n\n(% of national average)\n\nEuropean elections 1994–2024\n\nFrench legislative elections 1848–2022\n\n**Interpretation.** This panel shows the ratio of the left−wing vote in urban areas to that in rural areas. It compares the 50% most urban with the 50% most rural (by agglomeration size). In both European elections (1994–2024) and legislative elections (1848–2022), the urban–rural gap widens markedly from the mid−1990s onward, with a sharp rise in the 2024 European election. **Sources and series:** Cagé and Piketty (2025) and unehistoiredunconflitpolitique.fr.\n\n**Ratio of left wing vote beween urban and rural areas, 1848−2022**\n\n24","pages":[{"line_numbers":[],"markdown":"Executive Summary\n\n**Figure 15. We need political action but political coalitions are difficult to form**\n\nHigher−educated voters voting for left−wing\n\nparties (democratic, labor, social− democratic, socialist, green, etc.)\n\nTop−income voters voting for right−wing\n\nparties (other parties) −20%\n\n−15%\n\n−10%\n\n−5%\n\n0%\n\n5%\n\n10%\n\n15%\n\n1960−64\n\n1965−69\n\n1970−74\n\n1975−79\n\n1980−84\n\n1985−89\n\n1990−94\n\n1995−99\n\n2000−04\n\n2005−09\n\n2010−14\n\n2015−19\n\n2020−25\n\nDifference (pp): top 10% minus\n\nbottom 90% voting left\n\n(% of top 10% educated voting left) minus (% of bottom 90% educated voting left)\n\n(% of top 10% earners voting left) minus (% of bottom 90% earners voting left)\n\n**Education and income divides in Western democracies, 1960−2025**\n\n**Interpretation.** In the 1960s, both higher−educated and high−income voters were less likely to vote for left−wing (democratic / labor / social−democratic / socialist / green) parties than lower−educated and low−income voters by more than 10 percentage points. The left vote has gradually become associated with higher education voters, giving rising to a multi−elite party system. Figures correspond to five−year averages for Australia, Britain, Canada, Denmark, France, Germany, Italy, the Netherlands, Norway, Sweden, Switzerland, and the U.S. Estimates control for income/education, age, gender, religion, church attendance, rural/urban, region, race/ethnicity, employment status, and marital status (in country−years for which these variables are available). **Sources and series:** Gethin et al. (2021) and _World Political Cleavages and Inequality Database_ (wpid.world).\n\n**Figure 16. Divides between large cities and smaller towns have reached levels unseen in a century**\n\n**Urban voters vote more for left parties**\n\n80% 90%\n\n100% 110% 120% 130% 140% 150% 160% 170% 180%\n\n1840\n\n1860\n\n1880\n\n1900\n\n1920\n\n1940\n\n1960\n\n1980\n\n2000\n\n2020\n\nLeft vote:\n\nurban 50% vs. rural 50%\n\n(% of national average)\n\nEuropean elections 1994–2024\n\nFrench legislative elections 1848–2022\n\n**Interpretation.** This panel shows the ratio of the left−wing vote in urban areas to that in rural areas. It compares the 50% most urban with the 50% most rural (by agglomeration size). In both European elections (1994–2024) and legislative elections (1848–2022), the urban–rural gap widens markedly from the mid−1990s onward, with a sharp rise in the 2024 European election. **Sources and series:** Cagé and Piketty (2025) and unehistoiredunconflitpolitique.fr.\n\n**Ratio of left wing vote beween urban and rural areas, 1848−2022**\n\n24","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Executive Summary\n\n**Figure 15. We need political action but political coalitions are difficult to form**\n\nHigher−educated voters voting for left−wing\n\nparties (democratic, labor, social− democratic, socialist, green, etc.)\n\nTop−income voters voting for right−wing\n\nparties (other parties) −20%\n\n−15%\n\n−10%\n\n−5%\n\n0%\n\n5%\n\n10%\n\n15%\n\n1960−64\n\n1965−69\n\n1970−74\n\n1975−79\n\n1980−84\n\n1985−89\n\n1990−94\n\n1995−99\n\n2000−04\n\n2005−09\n\n2010−14\n\n2015−19\n\n2020−25\n\nDifference (pp): top 10% minus\n\nbottom 90% voting left\n\n(% of top 10% educated voting left) minus (% of bottom 90% educated voting left)\n\n(% of top 10% earners voting left) minus (% of bottom 90% earners voting left)\n\n**Education and income divides in Western democracies, 1960−2025**\n\n**Interpretation.** In the 1960s, both higher−educated and high−income voters were less likely to vote for left−wing (democratic / labor / social−democratic / socialist / green) parties than lower−educated and low−income voters by more than 10 percentage points. The left vote has gradually become associated with higher education voters, giving rising to a multi−elite party system. Figures correspond to five−year averages for Australia, Britain, Canada, Denmark, France, Germany, Italy, the Netherlands, Norway, Sweden, Switzerland, and the U.S. Estimates control for income/education, age, gender, religion, church attendance, rural/urban, region, race/ethnicity, employment status, and marital status (in country−years for which these variables are available). **Sources and series:** Gethin et al. (2021) and _World Political Cleavages and Inequality Database_ (wpid.world).\n\n**Figure 16. Divides between large cities and smaller towns have reached levels unseen in a century**\n\n**Urban voters vote more for left parties**\n\n80% 90%\n\n100% 110% 120% 130% 140% 150% 160% 170% 180%\n\n1840\n\n1860\n\n1880\n\n1900\n\n1920\n\n1940\n\n1960\n\n1980\n\n2000\n\n2020\n\nLeft vote:\n\nurban 50% vs. rural 50%\n\n(% of national average)\n\nEuropean elections 1994–2024\n\nFrench legislative elections 1848–2022\n\n**Interpretation.** This panel shows the ratio of the left−wing vote in urban areas to that in rural areas. It compares the 50% most urban with the 50% most rural (by agglomeration size). In both European elections (1994–2024) and legislative elections (1848–2022), the urban–rural gap widens markedly from the mid−1990s onward, with a sharp rise in the 2024 European election. **Sources and series:** Cagé and Piketty (2025) and unehistoiredunconflitpolitique.fr.\n\n**Ratio of left wing vote beween urban and rural areas, 1848−2022**\n\n24","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p24.pdf"]},"response":{"errors":[],"extract_id":"extract_a483728e6d9cab57e603008486bb8ec5","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Executive Summary\n\n**Figure 15. We need political action but political coalitions are difficult to form**\n\nHigher−educated voters voting for left−wing\n\nparties (democratic, labor, social− democratic, socialist, green, etc.)\n\nTop−income voters voting for right−wing\n\nparties (other parties) −20%\n\n−15%\n\n−10%\n\n−5%\n\n0%\n\n5%\n\n10%\n\n15%\n\n1960−64\n\n1965−69\n\n1970−74\n\n1975−79\n\n1980−84\n\n1985−89\n\n1990−94\n\n1995−99\n\n2000−04\n\n2005−09\n\n2010−14\n\n2015−19\n\n2020−25\n\nDifference (pp): top 10% minus\n\nbottom 90% voting left\n\n(% of top 10% educated voting left) minus (% of bottom 90% educated voting left)\n\n(% of top 10% earners voting left) minus (% of bottom 90% earners voting left)\n\n**Education and income divides in Western democracies, 1960−2025**\n\n**Interpretation.** In the 1960s, both higher−educated and high−income voters were less likely to vote for left−wing (democratic / labor / social−democratic / socialist / green) parties than lower−educated and low−income voters by more than 10 percentage points. The left vote has gradually become associated with higher education voters, giving rising to a multi−elite party system. Figures correspond to five−year averages for Australia, Britain, Canada, Denmark, France, Germany, Italy, the Netherlands, Norway, Sweden, Switzerland, and the U.S. Estimates control for income/education, age, gender, religion, church attendance, rural/urban, region, race/ethnicity, employment status, and marital status (in country−years for which these variables are available). **Sources and series:** Gethin et al. (2021) and _World Political Cleavages and Inequality Database_ (wpid.world).\n\n**Figure 16. Divides between large cities and smaller towns have reached levels unseen in a century**\n\n**Urban voters vote more for left parties**\n\n80% 90%\n\n100% 110% 120% 130% 140% 150% 160% 170% 180%\n\n1840\n\n1860\n\n1880\n\n1900\n\n1920\n\n1940\n\n1960\n\n1980\n\n2000\n\n2020\n\nLeft vote:\n\nurban 50% vs. rural 50%\n\n(% of national average)\n\nEuropean elections 1994–2024\n\nFrench legislative elections 1848–2022\n\n**Interpretation.** This panel shows the ratio of the left−wing vote in urban areas to that in rural areas. It compares the 50% most urban with the 50% most rural (by agglomeration size). In both European elections (1994–2024) and legislative elections (1848–2022), the urban–rural gap widens markedly from the mid−1990s onward, with a sharp rise in the 2024 European election. **Sources and series:** Cagé and Piketty (2025) and unehistoiredunconflitpolitique.fr.\n\n**Ratio of left wing vote beween urban and rural areas, 1848−2022**\n\n24"],"full_content":"Executive Summary\n\n**Figure 15. We need political action but political coalitions are difficult to form**\n\nHigher−educated voters voting for left−wing\n\nparties (democratic, labor, social− democratic, socialist, green, etc.)\n\nTop−income voters voting for right−wing\n\nparties (other parties) −20%\n\n−15%\n\n−10%\n\n−5%\n\n0%\n\n5%\n\n10%\n\n15%\n\n1960−64\n\n1965−69\n\n1970−74\n\n1975−79\n\n1980−84\n\n1985−89\n\n1990−94\n\n1995−99\n\n2000−04\n\n2005−09\n\n2010−14\n\n2015−19\n\n2020−25\n\nDifference (pp): top 10% minus\n\nbottom 90% voting left\n\n(% of top 10% educated voting left) minus (% of bottom 90% educated voting left)\n\n(% of top 10% earners voting left) minus (% of bottom 90% earners voting left)\n\n**Education and income divides in Western democracies, 1960−2025**\n\n**Interpretation.** In the 1960s, both higher−educated and high−income voters were less likely to vote for left−wing (democratic / labor / social−democratic / socialist / green) parties than lower−educated and low−income voters by more than 10 percentage points. The left vote has gradually become associated with higher education voters, giving rising to a multi−elite party system. Figures correspond to five−year averages for Australia, Britain, Canada, Denmark, France, Germany, Italy, the Netherlands, Norway, Sweden, Switzerland, and the U.S. Estimates control for income/education, age, gender, religion, church attendance, rural/urban, region, race/ethnicity, employment status, and marital status (in country−years for which these variables are available). **Sources and series:** Gethin et al. (2021) and _World Political Cleavages and Inequality Database_ (wpid.world).\n\n**Figure 16. Divides between large cities and smaller towns have reached levels unseen in a century**\n\n**Urban voters vote more for left parties**\n\n80% 90%\n\n100% 110% 120% 130% 140% 150% 160% 170% 180%\n\n1840\n\n1860\n\n1880\n\n1900\n\n1920\n\n1940\n\n1960\n\n1980\n\n2000\n\n2020\n\nLeft vote:\n\nurban 50% vs. rural 50%\n\n(% of national average)\n\nEuropean elections 1994–2024\n\nFrench legislative elections 1848–2022\n\n**Interpretation.** This panel shows the ratio of the left−wing vote in urban areas to that in rural areas. It compares the 50% most urban with the 50% most rural (by agglomeration size). In both European elections (1994–2024) and legislative elections (1848–2022), the urban–rural gap widens markedly from the mid−1990s onward, with a sharp rise in the 2024 European election. **Sources and series:** Cagé and Piketty (2025) and unehistoiredunconflitpolitique.fr.\n\n**Ratio of left wing vote beween urban and rural areas, 1848−2022**\n\n24","publish_date":null,"title":"Executive Summary Figure15. Weneedpoliticalactionbutpoliticalcoalitionsare difficulttoform Education and income divides in Western democracies, 1960−2025 Figure16. Dividesbetweenlargecitiesandsmallertownshave reachedlevelsunseeninacentury Ratio of left wing","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p24.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p24.pdf"},"request":{"example_id":"chart/World_Inequality_Report_2026_p24","product_type":"parse","source_file_path":"World_Inequality_Report_2026_p24.pdf"},"started_at":"2026-11-04T19:45:38.695760"} +{"completed_at":"2026-11-04T19:45:41.202166","latency_in_ms":1457,"output":{"example_id":"chart/World_Inequality_Report_2026_p25","grounded_pages":[],"layout_pages":[],"markdown":"Executive Summary\n\n**Figure 17. Without redistribution, political inequality will increase**\n\n0%\n\n10%\n\n20%\n\n30%\n\n40%\n\n50%\n\n60%\n\nP0−P10\n\nP10−P20\n\nP20−P30\n\nP30−P40\n\nP40−P50\n\nP50−P60\n\nP60−P70\n\nP70−P80\n\nP80−P90\n\nP90−P100\n\nPercentage of total political\n\ndonations by income decile\n\nFrance South Korea\n\n**Political donations by income decile, % of total**\n\n**Interpretation.** Average shares of total political donations by income decile in France and South Korea (2013–2021). Donations are highly concentrated at the top, with the richest decile contributing the largest share. **Sources and series:** Cagé (2024).\n\n25","pages":[{"line_numbers":[],"markdown":"Executive Summary\n\n**Figure 17. Without redistribution, political inequality will increase**\n\n0%\n\n10%\n\n20%\n\n30%\n\n40%\n\n50%\n\n60%\n\nP0−P10\n\nP10−P20\n\nP20−P30\n\nP30−P40\n\nP40−P50\n\nP50−P60\n\nP60−P70\n\nP70−P80\n\nP80−P90\n\nP90−P100\n\nPercentage of total political\n\ndonations by income decile\n\nFrance South Korea\n\n**Political donations by income decile, % of total**\n\n**Interpretation.** Average shares of total political donations by income decile in France and South Korea (2013–2021). Donations are highly concentrated at the top, with the richest decile contributing the largest share. **Sources and series:** Cagé (2024).\n\n25","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Executive Summary\n\n**Figure 17. Without redistribution, political inequality will increase**\n\n0%\n\n10%\n\n20%\n\n30%\n\n40%\n\n50%\n\n60%\n\nP0−P10\n\nP10−P20\n\nP20−P30\n\nP30−P40\n\nP40−P50\n\nP50−P60\n\nP60−P70\n\nP70−P80\n\nP80−P90\n\nP90−P100\n\nPercentage of total political\n\ndonations by income decile\n\nFrance South Korea\n\n**Political donations by income decile, % of total**\n\n**Interpretation.** Average shares of total political donations by income decile in France and South Korea (2013–2021). Donations are highly concentrated at the top, with the richest decile contributing the largest share. **Sources and series:** Cagé (2024).\n\n25","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p25.pdf"]},"response":{"errors":[],"extract_id":"extract_2fab8c4b93f15fcbedd2a3885e801bff","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Executive Summary\n\n**Figure 17. Without redistribution, political inequality will increase**\n\n0%\n\n10%\n\n20%\n\n30%\n\n40%\n\n50%\n\n60%\n\nP0−P10\n\nP10−P20\n\nP20−P30\n\nP30−P40\n\nP40−P50\n\nP50−P60\n\nP60−P70\n\nP70−P80\n\nP80−P90\n\nP90−P100\n\nPercentage of total political\n\ndonations by income decile\n\nFrance South Korea\n\n**Political donations by income decile, % of total**\n\n**Interpretation.** Average shares of total political donations by income decile in France and South Korea (2013–2021). Donations are highly concentrated at the top, with the richest decile contributing the largest share. **Sources and series:** Cagé (2024).\n\n25"],"full_content":"Executive Summary\n\n**Figure 17. Without redistribution, political inequality will increase**\n\n0%\n\n10%\n\n20%\n\n30%\n\n40%\n\n50%\n\n60%\n\nP0−P10\n\nP10−P20\n\nP20−P30\n\nP30−P40\n\nP40−P50\n\nP50−P60\n\nP60−P70\n\nP70−P80\n\nP80−P90\n\nP90−P100\n\nPercentage of total political\n\ndonations by income decile\n\nFrance South Korea\n\n**Political donations by income decile, % of total**\n\n**Interpretation.** Average shares of total political donations by income decile in France and South Korea (2013–2021). Donations are highly concentrated at the top, with the richest decile contributing the largest share. **Sources and series:** Cagé (2024).\n\n25","publish_date":null,"title":"Executive Summary Figure17. Withoutredistribution,politicalinequalitywill increase Political donations by income decile, % of total 25","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p25.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p25.pdf"},"request":{"example_id":"chart/World_Inequality_Report_2026_p25","product_type":"parse","source_file_path":"World_Inequality_Report_2026_p25.pdf"},"started_at":"2026-11-04T19:45:39.744546"} +{"completed_at":"2026-11-04T19:45:43.079203","latency_in_ms":1538,"output":{"example_id":"chart/World_Inequality_Report_2026_p36","grounded_pages":[],"layout_pages":[],"markdown":"Chapter 1. Global Economic Inequality\n\n**Figure 1.1. The world is becoming richer**\n\n1,000\n\n2,000\n\n5,000\n\n10,000\n\n1,000\n\n2,000\n\n5,000\n\n10,000\n\n1800 1825 1850 1875 1900 1925 1950 1975 2000 2025\n\nYearly per capita national income\n\n€ (at 2025 PPP), log scale\n\nWorld Population\n\nNational income World population\n\n**Per capita income and population, 1800−2025**\n\n**Interpretation.** World population increased from 1 billion in 1800 to 8 billion in 2025, corresponding to an average annual growth rate of about 0.9% per year. Yearly income per person increased from about €900 in 1800 to about €14,000 in 2025, a multiplication by about 16 (corresponding to average annual growth rate of about 1.2% per year). **Sources and series:** Gómez−Carrera et al. (2025), Nievas and Piketty (2025), and wir2026.wid.world/methodology.\n\nonly 556 million individuals, roughly the size of the combined adult populations of the United States, Pakistan, and Brazil. Looking more closely at the very top, the numbers become even smaller but highly significant. The top 1% includes about 56 million adults, similar to the adult population of the United Kingdom. The top 0.1% (5.6 million adults) is similar in size to the total population of Singapore. The top 0.01% amounts to 556,000 adults, about the total population of Genoa in Italy. The top 0.001%, or 56,000 adults, could all fit inside a football stadium. Going further, the top 0.0001% (around 5,600 adults) would fill a concert arena, the top 0.00001% (560 adults) a theater, and the top 0.000001% (56 adults) a single classroom. These comparisons will help illustrate just how concentrated the very top of the distribution is, and they will serve as a reference throughout the report to help readers grasp the magnitude of global inequality.\n\n**Extreme and rising income inequality**\n\nKeeping these figures in mind, **Figure 1.3** illustrates the extent of global inequality. The top 10% (about 560 million adults) receives 53% of global income, while the bottom 50% (roughly 2.8 billion adults) captures only 8%. The contrast becomes even sharper when we zoom into the right-hand panel: the top 1%, a group of just 56 million people, earns 2.5 times more than the entire bottom half of humanity. Put differently, a population comparable to the United Kingdom’s receives more income than a group as large as the combined populations of China, India, the United States, Indonesia, Nigeria, Brazil, and Russia. If we look beyond these broad groups and zoom in further still, the concentration of income becomes even starker. **Figure 1.4** highlights the concentration of income even more clearly. The third column shows that the top 0.1% earns as much as the entire bottom 50%. This means that a group of people no larger than the population of Singapore takes in the same income as half of the world’s population. At the very extremes, inequality becomes staggering:\n\n36","pages":[{"line_numbers":[],"markdown":"Chapter 1. Global Economic Inequality\n\n**Figure 1.1. The world is becoming richer**\n\n1,000\n\n2,000\n\n5,000\n\n10,000\n\n1,000\n\n2,000\n\n5,000\n\n10,000\n\n1800 1825 1850 1875 1900 1925 1950 1975 2000 2025\n\nYearly per capita national income\n\n€ (at 2025 PPP), log scale\n\nWorld Population\n\nNational income World population\n\n**Per capita income and population, 1800−2025**\n\n**Interpretation.** World population increased from 1 billion in 1800 to 8 billion in 2025, corresponding to an average annual growth rate of about 0.9% per year. Yearly income per person increased from about €900 in 1800 to about €14,000 in 2025, a multiplication by about 16 (corresponding to average annual growth rate of about 1.2% per year). **Sources and series:** Gómez−Carrera et al. (2025), Nievas and Piketty (2025), and wir2026.wid.world/methodology.\n\nonly 556 million individuals, roughly the size of the combined adult populations of the United States, Pakistan, and Brazil. Looking more closely at the very top, the numbers become even smaller but highly significant. The top 1% includes about 56 million adults, similar to the adult population of the United Kingdom. The top 0.1% (5.6 million adults) is similar in size to the total population of Singapore. The top 0.01% amounts to 556,000 adults, about the total population of Genoa in Italy. The top 0.001%, or 56,000 adults, could all fit inside a football stadium. Going further, the top 0.0001% (around 5,600 adults) would fill a concert arena, the top 0.00001% (560 adults) a theater, and the top 0.000001% (56 adults) a single classroom. These comparisons will help illustrate just how concentrated the very top of the distribution is, and they will serve as a reference throughout the report to help readers grasp the magnitude of global inequality.\n\n**Extreme and rising income inequality**\n\nKeeping these figures in mind, **Figure 1.3** illustrates the extent of global inequality. The top 10% (about 560 million adults) receives 53% of global income, while the bottom 50% (roughly 2.8 billion adults) captures only 8%. The contrast becomes even sharper when we zoom into the right-hand panel: the top 1%, a group of just 56 million people, earns 2.5 times more than the entire bottom half of humanity. Put differently, a population comparable to the United Kingdom’s receives more income than a group as large as the combined populations of China, India, the United States, Indonesia, Nigeria, Brazil, and Russia. If we look beyond these broad groups and zoom in further still, the concentration of income becomes even starker. **Figure 1.4** highlights the concentration of income even more clearly. The third column shows that the top 0.1% earns as much as the entire bottom 50%. This means that a group of people no larger than the population of Singapore takes in the same income as half of the world’s population. At the very extremes, inequality becomes staggering:\n\n36","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Chapter 1. Global Economic Inequality\n\n**Figure 1.1. The world is becoming richer**\n\n1,000\n\n2,000\n\n5,000\n\n10,000\n\n1,000\n\n2,000\n\n5,000\n\n10,000\n\n1800 1825 1850 1875 1900 1925 1950 1975 2000 2025\n\nYearly per capita national income\n\n€ (at 2025 PPP), log scale\n\nWorld Population\n\nNational income World population\n\n**Per capita income and population, 1800−2025**\n\n**Interpretation.** World population increased from 1 billion in 1800 to 8 billion in 2025, corresponding to an average annual growth rate of about 0.9% per year. Yearly income per person increased from about €900 in 1800 to about €14,000 in 2025, a multiplication by about 16 (corresponding to average annual growth rate of about 1.2% per year). **Sources and series:** Gómez−Carrera et al. (2025), Nievas and Piketty (2025), and wir2026.wid.world/methodology.\n\nonly 556 million individuals, roughly the size of the combined adult populations of the United States, Pakistan, and Brazil. Looking more closely at the very top, the numbers become even smaller but highly significant. The top 1% includes about 56 million adults, similar to the adult population of the United Kingdom. The top 0.1% (5.6 million adults) is similar in size to the total population of Singapore. The top 0.01% amounts to 556,000 adults, about the total population of Genoa in Italy. The top 0.001%, or 56,000 adults, could all fit inside a football stadium. Going further, the top 0.0001% (around 5,600 adults) would fill a concert arena, the top 0.00001% (560 adults) a theater, and the top 0.000001% (56 adults) a single classroom. These comparisons will help illustrate just how concentrated the very top of the distribution is, and they will serve as a reference throughout the report to help readers grasp the magnitude of global inequality.\n\n**Extreme and rising income inequality**\n\nKeeping these figures in mind, **Figure 1.3** illustrates the extent of global inequality. The top 10% (about 560 million adults) receives 53% of global income, while the bottom 50% (roughly 2.8 billion adults) captures only 8%. The contrast becomes even sharper when we zoom into the right-hand panel: the top 1%, a group of just 56 million people, earns 2.5 times more than the entire bottom half of humanity. Put differently, a population comparable to the United Kingdom’s receives more income than a group as large as the combined populations of China, India, the United States, Indonesia, Nigeria, Brazil, and Russia. If we look beyond these broad groups and zoom in further still, the concentration of income becomes even starker. **Figure 1.4** highlights the concentration of income even more clearly. The third column shows that the top 0.1% earns as much as the entire bottom 50%. This means that a group of people no larger than the population of Singapore takes in the same income as half of the world’s population. At the very extremes, inequality becomes staggering:\n\n36","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p36.pdf"]},"response":{"errors":[],"extract_id":"extract_97321a3f71afaa1cc5d4b5926da6fa49","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Chapter 1. Global Economic Inequality\n\n**Figure 1.1. The world is becoming richer**\n\n1,000\n\n2,000\n\n5,000\n\n10,000\n\n1,000\n\n2,000\n\n5,000\n\n10,000\n\n1800 1825 1850 1875 1900 1925 1950 1975 2000 2025\n\nYearly per capita national income\n\n€ (at 2025 PPP), log scale\n\nWorld Population\n\nNational income World population\n\n**Per capita income and population, 1800−2025**\n\n**Interpretation.** World population increased from 1 billion in 1800 to 8 billion in 2025, corresponding to an average annual growth rate of about 0.9% per year. Yearly income per person increased from about €900 in 1800 to about €14,000 in 2025, a multiplication by about 16 (corresponding to average annual growth rate of about 1.2% per year). **Sources and series:** Gómez−Carrera et al. (2025), Nievas and Piketty (2025), and wir2026.wid.world/methodology.\n\nonly 556 million individuals, roughly the size of the combined adult populations of the United States, Pakistan, and Brazil. Looking more closely at the very top, the numbers become even smaller but highly significant. The top 1% includes about 56 million adults, similar to the adult population of the United Kingdom. The top 0.1% (5.6 million adults) is similar in size to the total population of Singapore. The top 0.01% amounts to 556,000 adults, about the total population of Genoa in Italy. The top 0.001%, or 56,000 adults, could all fit inside a football stadium. Going further, the top 0.0001% (around 5,600 adults) would fill a concert arena, the top 0.00001% (560 adults) a theater, and the top 0.000001% (56 adults) a single classroom. These comparisons will help illustrate just how concentrated the very top of the distribution is, and they will serve as a reference throughout the report to help readers grasp the magnitude of global inequality.\n\n**Extreme and rising income inequality**\n\nKeeping these figures in mind, **Figure 1.3** illustrates the extent of global inequality. The top 10% (about 560 million adults) receives 53% of global income, while the bottom 50% (roughly 2.8 billion adults) captures only 8%. The contrast becomes even sharper when we zoom into the right-hand panel: the top 1%, a group of just 56 million people, earns 2.5 times more than the entire bottom half of humanity. Put differently, a population comparable to the United Kingdom’s receives more income than a group as large as the combined populations of China, India, the United States, Indonesia, Nigeria, Brazil, and Russia. If we look beyond these broad groups and zoom in further still, the concentration of income becomes even starker. **Figure 1.4** highlights the concentration of income even more clearly. The third column shows that the top 0.1% earns as much as the entire bottom 50%. This means that a group of people no larger than the population of Singapore takes in the same income as half of the world’s population. At the very extremes, inequality becomes staggering:\n\n36"],"full_content":"Chapter 1. Global Economic Inequality\n\n**Figure 1.1. The world is becoming richer**\n\n1,000\n\n2,000\n\n5,000\n\n10,000\n\n1,000\n\n2,000\n\n5,000\n\n10,000\n\n1800 1825 1850 1875 1900 1925 1950 1975 2000 2025\n\nYearly per capita national income\n\n€ (at 2025 PPP), log scale\n\nWorld Population\n\nNational income World population\n\n**Per capita income and population, 1800−2025**\n\n**Interpretation.** World population increased from 1 billion in 1800 to 8 billion in 2025, corresponding to an average annual growth rate of about 0.9% per year. Yearly income per person increased from about €900 in 1800 to about €14,000 in 2025, a multiplication by about 16 (corresponding to average annual growth rate of about 1.2% per year). **Sources and series:** Gómez−Carrera et al. (2025), Nievas and Piketty (2025), and wir2026.wid.world/methodology.\n\nonly 556 million individuals, roughly the size of the combined adult populations of the United States, Pakistan, and Brazil. Looking more closely at the very top, the numbers become even smaller but highly significant. The top 1% includes about 56 million adults, similar to the adult population of the United Kingdom. The top 0.1% (5.6 million adults) is similar in size to the total population of Singapore. The top 0.01% amounts to 556,000 adults, about the total population of Genoa in Italy. The top 0.001%, or 56,000 adults, could all fit inside a football stadium. Going further, the top 0.0001% (around 5,600 adults) would fill a concert arena, the top 0.00001% (560 adults) a theater, and the top 0.000001% (56 adults) a single classroom. These comparisons will help illustrate just how concentrated the very top of the distribution is, and they will serve as a reference throughout the report to help readers grasp the magnitude of global inequality.\n\n**Extreme and rising income inequality**\n\nKeeping these figures in mind, **Figure 1.3** illustrates the extent of global inequality. The top 10% (about 560 million adults) receives 53% of global income, while the bottom 50% (roughly 2.8 billion adults) captures only 8%. The contrast becomes even sharper when we zoom into the right-hand panel: the top 1%, a group of just 56 million people, earns 2.5 times more than the entire bottom half of humanity. Put differently, a population comparable to the United Kingdom’s receives more income than a group as large as the combined populations of China, India, the United States, Indonesia, Nigeria, Brazil, and Russia. If we look beyond these broad groups and zoom in further still, the concentration of income becomes even starker. **Figure 1.4** highlights the concentration of income even more clearly. The third column shows that the top 0.1% earns as much as the entire bottom 50%. This means that a group of people no larger than the population of Singapore takes in the same income as half of the world’s population. At the very extremes, inequality becomes staggering:\n\n36","publish_date":null,"title":"Chapter 1. Global Economic Inequality Figure1.1. Theworldisbecomingricher only 556 million individuals, roughly the size of the combined adult populations of","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p36.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p36.pdf"},"request":{"example_id":"chart/World_Inequality_Report_2026_p36","product_type":"parse","source_file_path":"World_Inequality_Report_2026_p36.pdf"},"started_at":"2026-11-04T19:45:41.540323"} +{"completed_at":"2026-11-04T19:45:45.948953","latency_in_ms":2866,"output":{"example_id":"chart/World_Inequality_Report_2026_p38","grounded_pages":[],"layout_pages":[],"markdown":"Chapter 1. Global Economic Inequality\n\n**Figure 1.3. Income and wealth shares are distributed very unequally**\n\n8%\n\n38%\n\n53%\n\n2%\n\n23%\n\n75%\n\n0%\n\n10%\n\n20%\n\n30%\n\n40%\n\n50%\n\n60%\n\n70%\n\n80%\n\n90%\n\n100%\n\nIncome Wealth\n\nShare of global income or wealth\n\nBottom 50% Middle 40% Top 10%\n\n**By broad groups**\n\n8%\n\n38%\n\n33%\n\n20%\n\n2%\n\n23%\n\n38%\n\n37%\n\n0%\n\n10%\n\n20%\n\n30%\n\n40%\n\n50%\n\n60%\n\n70%\n\n80%\n\n90%\n\n100%\n\nIncome Wealth\n\nShare of global income or wealth\n\nNext 9% Top 1%\n\n**More detail on the top 10%**\n\n**Interpretation.** The global bottom 50% capture 8% of total income and own 2% of global wealth (2025 PPP). The top 10% capture 53% of income and own 75% of wealth, while the P90–99 capture 33% of income and own 38% of wealth. Moreover, the top 1% capture 20% of income and hold 37% of wealth. Income is measured after pensions and unemployment benefits are received by individuals and before taxes and transfers. **Sources and series:** Arias−Osorio et al. (2025) and wir2026.wid.world/methodology.\n\n**Global income and wealth inequality, 2025**\n\ncurve. It shows that while the bottom half of the world has enjoyed relatively robust growth since 1980, the middle 40% experienced stagnation, with some groups growing at less than 1% per year. Meanwhile, growth accelerated again at the very top, with the richest 1% and especially the top 0.1% capturing the fastest gains. The result is a polarized pattern: the poor and the rich have seen their incomes rise, while the global middle class has benefited the least over the past four decades. These uneven growth dynamics explain why today’s distribution of income is so skewed: the gains of the past decades have consolidated mainly at the very top. Such polarization also carries political implications: the relative exclusion of large middle-income groups, the stagnation of many poorer groups in rich countries, and the growing influence of the global plutocracy all raise pressing questions for democratic stability and global governance; we will return to this point in **Chapter 8** .\n\n**Wealth inequality is larger, more extreme, and rising faster**\n\nSo far, we have seen that, when examining incomes, inequality is very large. But income inequality only tells part of the story, since it largely reflects labor earnings. Capital income, which is even more concentrated and closely tied to wealth ownership, adds another layer of inequality.\n\n**Figure 1.3** makes this clear: the global top 10% owns three-quarters of all wealth, while the bottom 50% holds just 2%. Zooming further in, the concentration becomes staggering. The top 1% alone, roughly the adult population of the United Kingdom, controls 37% of global wealth. This is more than eighteen times the wealth of the entire bottom half of the world population, a group as large as the combined adult populations of China, India, the United States, Indonesia, Nigeria, Brazil, and Russia.\n\n**Figure 1.6** sheds light on just how extreme inequality becomes at the very top. The top one-in-a-million (about 5,600 adults, enough to fill a concert arena) collectively hold 3% of\n\n38","pages":[{"line_numbers":[],"markdown":"Chapter 1. Global Economic Inequality\n\n**Figure 1.3. Income and wealth shares are distributed very unequally**\n\n8%\n\n38%\n\n53%\n\n2%\n\n23%\n\n75%\n\n0%\n\n10%\n\n20%\n\n30%\n\n40%\n\n50%\n\n60%\n\n70%\n\n80%\n\n90%\n\n100%\n\nIncome Wealth\n\nShare of global income or wealth\n\nBottom 50% Middle 40% Top 10%\n\n**By broad groups**\n\n8%\n\n38%\n\n33%\n\n20%\n\n2%\n\n23%\n\n38%\n\n37%\n\n0%\n\n10%\n\n20%\n\n30%\n\n40%\n\n50%\n\n60%\n\n70%\n\n80%\n\n90%\n\n100%\n\nIncome Wealth\n\nShare of global income or wealth\n\nNext 9% Top 1%\n\n**More detail on the top 10%**\n\n**Interpretation.** The global bottom 50% capture 8% of total income and own 2% of global wealth (2025 PPP). The top 10% capture 53% of income and own 75% of wealth, while the P90–99 capture 33% of income and own 38% of wealth. Moreover, the top 1% capture 20% of income and hold 37% of wealth. Income is measured after pensions and unemployment benefits are received by individuals and before taxes and transfers. **Sources and series:** Arias−Osorio et al. (2025) and wir2026.wid.world/methodology.\n\n**Global income and wealth inequality, 2025**\n\ncurve. It shows that while the bottom half of the world has enjoyed relatively robust growth since 1980, the middle 40% experienced stagnation, with some groups growing at less than 1% per year. Meanwhile, growth accelerated again at the very top, with the richest 1% and especially the top 0.1% capturing the fastest gains. The result is a polarized pattern: the poor and the rich have seen their incomes rise, while the global middle class has benefited the least over the past four decades. These uneven growth dynamics explain why today’s distribution of income is so skewed: the gains of the past decades have consolidated mainly at the very top. Such polarization also carries political implications: the relative exclusion of large middle-income groups, the stagnation of many poorer groups in rich countries, and the growing influence of the global plutocracy all raise pressing questions for democratic stability and global governance; we will return to this point in **Chapter 8** .\n\n**Wealth inequality is larger, more extreme, and rising faster**\n\nSo far, we have seen that, when examining incomes, inequality is very large. But income inequality only tells part of the story, since it largely reflects labor earnings. Capital income, which is even more concentrated and closely tied to wealth ownership, adds another layer of inequality.\n\n**Figure 1.3** makes this clear: the global top 10% owns three-quarters of all wealth, while the bottom 50% holds just 2%. Zooming further in, the concentration becomes staggering. The top 1% alone, roughly the adult population of the United Kingdom, controls 37% of global wealth. This is more than eighteen times the wealth of the entire bottom half of the world population, a group as large as the combined adult populations of China, India, the United States, Indonesia, Nigeria, Brazil, and Russia.\n\n**Figure 1.6** sheds light on just how extreme inequality becomes at the very top. The top one-in-a-million (about 5,600 adults, enough to fill a concert arena) collectively hold 3% of\n\n38","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Chapter 1. Global Economic Inequality\n\n**Figure 1.3. Income and wealth shares are distributed very unequally**\n\n8%\n\n38%\n\n53%\n\n2%\n\n23%\n\n75%\n\n0%\n\n10%\n\n20%\n\n30%\n\n40%\n\n50%\n\n60%\n\n70%\n\n80%\n\n90%\n\n100%\n\nIncome Wealth\n\nShare of global income or wealth\n\nBottom 50% Middle 40% Top 10%\n\n**By broad groups**\n\n8%\n\n38%\n\n33%\n\n20%\n\n2%\n\n23%\n\n38%\n\n37%\n\n0%\n\n10%\n\n20%\n\n30%\n\n40%\n\n50%\n\n60%\n\n70%\n\n80%\n\n90%\n\n100%\n\nIncome Wealth\n\nShare of global income or wealth\n\nNext 9% Top 1%\n\n**More detail on the top 10%**\n\n**Interpretation.** The global bottom 50% capture 8% of total income and own 2% of global wealth (2025 PPP). The top 10% capture 53% of income and own 75% of wealth, while the P90–99 capture 33% of income and own 38% of wealth. Moreover, the top 1% capture 20% of income and hold 37% of wealth. Income is measured after pensions and unemployment benefits are received by individuals and before taxes and transfers. **Sources and series:** Arias−Osorio et al. (2025) and wir2026.wid.world/methodology.\n\n**Global income and wealth inequality, 2025**\n\ncurve. It shows that while the bottom half of the world has enjoyed relatively robust growth since 1980, the middle 40% experienced stagnation, with some groups growing at less than 1% per year. Meanwhile, growth accelerated again at the very top, with the richest 1% and especially the top 0.1% capturing the fastest gains. The result is a polarized pattern: the poor and the rich have seen their incomes rise, while the global middle class has benefited the least over the past four decades. These uneven growth dynamics explain why today’s distribution of income is so skewed: the gains of the past decades have consolidated mainly at the very top. Such polarization also carries political implications: the relative exclusion of large middle-income groups, the stagnation of many poorer groups in rich countries, and the growing influence of the global plutocracy all raise pressing questions for democratic stability and global governance; we will return to this point in **Chapter 8** .\n\n**Wealth inequality is larger, more extreme, and rising faster**\n\nSo far, we have seen that, when examining incomes, inequality is very large. But income inequality only tells part of the story, since it largely reflects labor earnings. Capital income, which is even more concentrated and closely tied to wealth ownership, adds another layer of inequality.\n\n**Figure 1.3** makes this clear: the global top 10% owns three-quarters of all wealth, while the bottom 50% holds just 2%. Zooming further in, the concentration becomes staggering. The top 1% alone, roughly the adult population of the United Kingdom, controls 37% of global wealth. This is more than eighteen times the wealth of the entire bottom half of the world population, a group as large as the combined adult populations of China, India, the United States, Indonesia, Nigeria, Brazil, and Russia.\n\n**Figure 1.6** sheds light on just how extreme inequality becomes at the very top. The top one-in-a-million (about 5,600 adults, enough to fill a concert arena) collectively hold 3% of\n\n38","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p38.pdf"]},"response":{"errors":[],"extract_id":"extract_4664a5e75f7ac9598e27856fc7959a6d","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Chapter 1. Global Economic Inequality\n\n**Figure 1.3. Income and wealth shares are distributed very unequally**\n\n8%\n\n38%\n\n53%\n\n2%\n\n23%\n\n75%\n\n0%\n\n10%\n\n20%\n\n30%\n\n40%\n\n50%\n\n60%\n\n70%\n\n80%\n\n90%\n\n100%\n\nIncome Wealth\n\nShare of global income or wealth\n\nBottom 50% Middle 40% Top 10%\n\n**By broad groups**\n\n8%\n\n38%\n\n33%\n\n20%\n\n2%\n\n23%\n\n38%\n\n37%\n\n0%\n\n10%\n\n20%\n\n30%\n\n40%\n\n50%\n\n60%\n\n70%\n\n80%\n\n90%\n\n100%\n\nIncome Wealth\n\nShare of global income or wealth\n\nNext 9% Top 1%\n\n**More detail on the top 10%**\n\n**Interpretation.** The global bottom 50% capture 8% of total income and own 2% of global wealth (2025 PPP). The top 10% capture 53% of income and own 75% of wealth, while the P90–99 capture 33% of income and own 38% of wealth. Moreover, the top 1% capture 20% of income and hold 37% of wealth. Income is measured after pensions and unemployment benefits are received by individuals and before taxes and transfers. **Sources and series:** Arias−Osorio et al. (2025) and wir2026.wid.world/methodology.\n\n**Global income and wealth inequality, 2025**\n\ncurve. It shows that while the bottom half of the world has enjoyed relatively robust growth since 1980, the middle 40% experienced stagnation, with some groups growing at less than 1% per year. Meanwhile, growth accelerated again at the very top, with the richest 1% and especially the top 0.1% capturing the fastest gains. The result is a polarized pattern: the poor and the rich have seen their incomes rise, while the global middle class has benefited the least over the past four decades. These uneven growth dynamics explain why today’s distribution of income is so skewed: the gains of the past decades have consolidated mainly at the very top. Such polarization also carries political implications: the relative exclusion of large middle-income groups, the stagnation of many poorer groups in rich countries, and the growing influence of the global plutocracy all raise pressing questions for democratic stability and global governance; we will return to this point in **Chapter 8** .\n\n**Wealth inequality is larger, more extreme, and rising faster**\n\nSo far, we have seen that, when examining incomes, inequality is very large. But income inequality only tells part of the story, since it largely reflects labor earnings. Capital income, which is even more concentrated and closely tied to wealth ownership, adds another layer of inequality.\n\n**Figure 1.3** makes this clear: the global top 10% owns three-quarters of all wealth, while the bottom 50% holds just 2%. Zooming further in, the concentration becomes staggering. The top 1% alone, roughly the adult population of the United Kingdom, controls 37% of global wealth. This is more than eighteen times the wealth of the entire bottom half of the world population, a group as large as the combined adult populations of China, India, the United States, Indonesia, Nigeria, Brazil, and Russia.\n\n**Figure 1.6** sheds light on just how extreme inequality becomes at the very top. The top one-in-a-million (about 5,600 adults, enough to fill a concert arena) collectively hold 3% of\n\n38"],"full_content":"Chapter 1. Global Economic Inequality\n\n**Figure 1.3. Income and wealth shares are distributed very unequally**\n\n8%\n\n38%\n\n53%\n\n2%\n\n23%\n\n75%\n\n0%\n\n10%\n\n20%\n\n30%\n\n40%\n\n50%\n\n60%\n\n70%\n\n80%\n\n90%\n\n100%\n\nIncome Wealth\n\nShare of global income or wealth\n\nBottom 50% Middle 40% Top 10%\n\n**By broad groups**\n\n8%\n\n38%\n\n33%\n\n20%\n\n2%\n\n23%\n\n38%\n\n37%\n\n0%\n\n10%\n\n20%\n\n30%\n\n40%\n\n50%\n\n60%\n\n70%\n\n80%\n\n90%\n\n100%\n\nIncome Wealth\n\nShare of global income or wealth\n\nNext 9% Top 1%\n\n**More detail on the top 10%**\n\n**Interpretation.** The global bottom 50% capture 8% of total income and own 2% of global wealth (2025 PPP). The top 10% capture 53% of income and own 75% of wealth, while the P90–99 capture 33% of income and own 38% of wealth. Moreover, the top 1% capture 20% of income and hold 37% of wealth. Income is measured after pensions and unemployment benefits are received by individuals and before taxes and transfers. **Sources and series:** Arias−Osorio et al. (2025) and wir2026.wid.world/methodology.\n\n**Global income and wealth inequality, 2025**\n\ncurve. It shows that while the bottom half of the world has enjoyed relatively robust growth since 1980, the middle 40% experienced stagnation, with some groups growing at less than 1% per year. Meanwhile, growth accelerated again at the very top, with the richest 1% and especially the top 0.1% capturing the fastest gains. The result is a polarized pattern: the poor and the rich have seen their incomes rise, while the global middle class has benefited the least over the past four decades. These uneven growth dynamics explain why today’s distribution of income is so skewed: the gains of the past decades have consolidated mainly at the very top. Such polarization also carries political implications: the relative exclusion of large middle-income groups, the stagnation of many poorer groups in rich countries, and the growing influence of the global plutocracy all raise pressing questions for democratic stability and global governance; we will return to this point in **Chapter 8** .\n\n**Wealth inequality is larger, more extreme, and rising faster**\n\nSo far, we have seen that, when examining incomes, inequality is very large. But income inequality only tells part of the story, since it largely reflects labor earnings. Capital income, which is even more concentrated and closely tied to wealth ownership, adds another layer of inequality.\n\n**Figure 1.3** makes this clear: the global top 10% owns three-quarters of all wealth, while the bottom 50% holds just 2%. Zooming further in, the concentration becomes staggering. The top 1% alone, roughly the adult population of the United Kingdom, controls 37% of global wealth. This is more than eighteen times the wealth of the entire bottom half of the world population, a group as large as the combined adult populations of China, India, the United States, Indonesia, Nigeria, Brazil, and Russia.\n\n**Figure 1.6** sheds light on just how extreme inequality becomes at the very top. The top one-in-a-million (about 5,600 adults, enough to fill a concert arena) collectively hold 3% of\n\n38","publish_date":null,"title":"Chapter 1. Global Economic Inequality Figure1.3. Incomeandwealthsharesaredistributedvery unequally curve. It shows that while the bottom half of the world has","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p38.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p38.pdf"},"request":{"example_id":"chart/World_Inequality_Report_2026_p38","product_type":"parse","source_file_path":"World_Inequality_Report_2026_p38.pdf"},"started_at":"2026-11-04T19:45:43.082184"} +{"completed_at":"2026-11-04T19:45:46.214924","latency_in_ms":1233,"output":{"example_id":"chart/World_Inequality_Report_2026_p43","grounded_pages":[],"layout_pages":[],"markdown":"Chapter 1. Global Economic Inequality\n\n**Figure 1.8. Extreme wealth inequality is persistent and increasing**\n\nThe top 0.001% (about 56,000 adults) have 3 times more wealth than half of the entire world population combined.\n\n0%\n\n1%\n\n2%\n\n3%\n\n4%\n\n5%\n\n6%\n\n7%\n\n1995 2000 2005 2010 2015 2020 2025\n\nShare of global personal wealth (%)\n\nBottom 50% Top 0.001%\n\n**Extreme wealth inequality, 1995−2025**\n\n**Interpretation.** The share of personal wealth held by the richest 0.001% of adults rose from around 3.8% of total wealth in 1995 to nearly 6.1% in 2025. After a very slight increase, the share of wealth owned by the poorest half of the population has stagnated since the early 2000s at around 2%. Net personal wealth is equal to the sum of financial assets (e.g. equity or bonds) and non−financial assets (e.g. housing or land) owned by individuals, net of their debts. **Sources and series:** Arias−Osorio et al. (2025) and wir2026.wid.world/methodology.\n\nit is shaped not only by economics, but also by deep-rooted historical, political, and cultural legacies. It also highlights another key dimension: inequality does not just separate rich and poor individuals in the same context; it also entrenches divides between entire parts of the world. Regional comparisons allow us to see both how far apart regions stand from each other ( **Figure 1.12** ) and how unequal they are internally ( **Figure 1.13** ). The specific country groupings are detailed in **Box 1.1.1** .\n\n**Figure 1.12** contrasts average income and wealth per adult across world regions in 2025, relative to the world average. The patterns are striking. Wealthier regions are also typically higher-income regions, though there are exceptions. East Asia, for instance, has a higher average wealth level than Russia & Central Asia but a lower average income. At the top of the scale, North America & Oceania and Europe stand well above the global mean. In these regions, and in East Asia, wealth levels exceed income levels relative to the world average. In the rest of the world, however, income levels are relatively higher than wealth levels.\n\nAt the bottom of the scale, Sub-Saharan Africa, South & Southeast Asia, and Latin America remain far below the global average on both counts, though Latin America’s average income is somewhat closer to the world mean. The Middle East & North Africa and Russia & Central Asia occupy an intermediate position: their average incomes are closer to East Asia’s and not far from Latin America’s, but their average wealth remains much lower. The contrast is sharp: for instance, in 2025, average wealth in North America & Oceania is 338% of the world average, while in Sub-Saharan Africa it is just 20%. Put differently, the average adult in North America & Oceania owns more than sixteen times the wealth of the average adult in Sub-Saharan Africa. Turning to **Figure 1.13** , the focus shifts from differences between regions to disparities within them, using the top 10%/bottom 50% (T10/B50) ratio. This simple but powerful metric asks: on average, how many times more does the top 10% earn (or own) compared to the poorest half? The results reveal enormous divides. ⁵ First, wealth gaps everywhere are far larger\n\n43","pages":[{"line_numbers":[],"markdown":"Chapter 1. Global Economic Inequality\n\n**Figure 1.8. Extreme wealth inequality is persistent and increasing**\n\nThe top 0.001% (about 56,000 adults) have 3 times more wealth than half of the entire world population combined.\n\n0%\n\n1%\n\n2%\n\n3%\n\n4%\n\n5%\n\n6%\n\n7%\n\n1995 2000 2005 2010 2015 2020 2025\n\nShare of global personal wealth (%)\n\nBottom 50% Top 0.001%\n\n**Extreme wealth inequality, 1995−2025**\n\n**Interpretation.** The share of personal wealth held by the richest 0.001% of adults rose from around 3.8% of total wealth in 1995 to nearly 6.1% in 2025. After a very slight increase, the share of wealth owned by the poorest half of the population has stagnated since the early 2000s at around 2%. Net personal wealth is equal to the sum of financial assets (e.g. equity or bonds) and non−financial assets (e.g. housing or land) owned by individuals, net of their debts. **Sources and series:** Arias−Osorio et al. (2025) and wir2026.wid.world/methodology.\n\nit is shaped not only by economics, but also by deep-rooted historical, political, and cultural legacies. It also highlights another key dimension: inequality does not just separate rich and poor individuals in the same context; it also entrenches divides between entire parts of the world. Regional comparisons allow us to see both how far apart regions stand from each other ( **Figure 1.12** ) and how unequal they are internally ( **Figure 1.13** ). The specific country groupings are detailed in **Box 1.1.1** .\n\n**Figure 1.12** contrasts average income and wealth per adult across world regions in 2025, relative to the world average. The patterns are striking. Wealthier regions are also typically higher-income regions, though there are exceptions. East Asia, for instance, has a higher average wealth level than Russia & Central Asia but a lower average income. At the top of the scale, North America & Oceania and Europe stand well above the global mean. In these regions, and in East Asia, wealth levels exceed income levels relative to the world average. In the rest of the world, however, income levels are relatively higher than wealth levels.\n\nAt the bottom of the scale, Sub-Saharan Africa, South & Southeast Asia, and Latin America remain far below the global average on both counts, though Latin America’s average income is somewhat closer to the world mean. The Middle East & North Africa and Russia & Central Asia occupy an intermediate position: their average incomes are closer to East Asia’s and not far from Latin America’s, but their average wealth remains much lower. The contrast is sharp: for instance, in 2025, average wealth in North America & Oceania is 338% of the world average, while in Sub-Saharan Africa it is just 20%. Put differently, the average adult in North America & Oceania owns more than sixteen times the wealth of the average adult in Sub-Saharan Africa. Turning to **Figure 1.13** , the focus shifts from differences between regions to disparities within them, using the top 10%/bottom 50% (T10/B50) ratio. This simple but powerful metric asks: on average, how many times more does the top 10% earn (or own) compared to the poorest half? The results reveal enormous divides. ⁵ First, wealth gaps everywhere are far larger\n\n43","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Chapter 1. Global Economic Inequality\n\n**Figure 1.8. Extreme wealth inequality is persistent and increasing**\n\nThe top 0.001% (about 56,000 adults) have 3 times more wealth than half of the entire world population combined.\n\n0%\n\n1%\n\n2%\n\n3%\n\n4%\n\n5%\n\n6%\n\n7%\n\n1995 2000 2005 2010 2015 2020 2025\n\nShare of global personal wealth (%)\n\nBottom 50% Top 0.001%\n\n**Extreme wealth inequality, 1995−2025**\n\n**Interpretation.** The share of personal wealth held by the richest 0.001% of adults rose from around 3.8% of total wealth in 1995 to nearly 6.1% in 2025. After a very slight increase, the share of wealth owned by the poorest half of the population has stagnated since the early 2000s at around 2%. Net personal wealth is equal to the sum of financial assets (e.g. equity or bonds) and non−financial assets (e.g. housing or land) owned by individuals, net of their debts. **Sources and series:** Arias−Osorio et al. (2025) and wir2026.wid.world/methodology.\n\nit is shaped not only by economics, but also by deep-rooted historical, political, and cultural legacies. It also highlights another key dimension: inequality does not just separate rich and poor individuals in the same context; it also entrenches divides between entire parts of the world. Regional comparisons allow us to see both how far apart regions stand from each other ( **Figure 1.12** ) and how unequal they are internally ( **Figure 1.13** ). The specific country groupings are detailed in **Box 1.1.1** .\n\n**Figure 1.12** contrasts average income and wealth per adult across world regions in 2025, relative to the world average. The patterns are striking. Wealthier regions are also typically higher-income regions, though there are exceptions. East Asia, for instance, has a higher average wealth level than Russia & Central Asia but a lower average income. At the top of the scale, North America & Oceania and Europe stand well above the global mean. In these regions, and in East Asia, wealth levels exceed income levels relative to the world average. In the rest of the world, however, income levels are relatively higher than wealth levels.\n\nAt the bottom of the scale, Sub-Saharan Africa, South & Southeast Asia, and Latin America remain far below the global average on both counts, though Latin America’s average income is somewhat closer to the world mean. The Middle East & North Africa and Russia & Central Asia occupy an intermediate position: their average incomes are closer to East Asia’s and not far from Latin America’s, but their average wealth remains much lower. The contrast is sharp: for instance, in 2025, average wealth in North America & Oceania is 338% of the world average, while in Sub-Saharan Africa it is just 20%. Put differently, the average adult in North America & Oceania owns more than sixteen times the wealth of the average adult in Sub-Saharan Africa. Turning to **Figure 1.13** , the focus shifts from differences between regions to disparities within them, using the top 10%/bottom 50% (T10/B50) ratio. This simple but powerful metric asks: on average, how many times more does the top 10% earn (or own) compared to the poorest half? The results reveal enormous divides. ⁵ First, wealth gaps everywhere are far larger\n\n43","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p43.pdf"]},"response":{"errors":[],"extract_id":"extract_f63df02ecb391830bee2433d44255ef9","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Chapter 1. Global Economic Inequality\n\n**Figure 1.8. Extreme wealth inequality is persistent and increasing**\n\nThe top 0.001% (about 56,000 adults) have 3 times more wealth than half of the entire world population combined.\n\n0%\n\n1%\n\n2%\n\n3%\n\n4%\n\n5%\n\n6%\n\n7%\n\n1995 2000 2005 2010 2015 2020 2025\n\nShare of global personal wealth (%)\n\nBottom 50% Top 0.001%\n\n**Extreme wealth inequality, 1995−2025**\n\n**Interpretation.** The share of personal wealth held by the richest 0.001% of adults rose from around 3.8% of total wealth in 1995 to nearly 6.1% in 2025. After a very slight increase, the share of wealth owned by the poorest half of the population has stagnated since the early 2000s at around 2%. Net personal wealth is equal to the sum of financial assets (e.g. equity or bonds) and non−financial assets (e.g. housing or land) owned by individuals, net of their debts. **Sources and series:** Arias−Osorio et al. (2025) and wir2026.wid.world/methodology.\n\nit is shaped not only by economics, but also by deep-rooted historical, political, and cultural legacies. It also highlights another key dimension: inequality does not just separate rich and poor individuals in the same context; it also entrenches divides between entire parts of the world. Regional comparisons allow us to see both how far apart regions stand from each other ( **Figure 1.12** ) and how unequal they are internally ( **Figure 1.13** ). The specific country groupings are detailed in **Box 1.1.1** .\n\n**Figure 1.12** contrasts average income and wealth per adult across world regions in 2025, relative to the world average. The patterns are striking. Wealthier regions are also typically higher-income regions, though there are exceptions. East Asia, for instance, has a higher average wealth level than Russia & Central Asia but a lower average income. At the top of the scale, North America & Oceania and Europe stand well above the global mean. In these regions, and in East Asia, wealth levels exceed income levels relative to the world average. In the rest of the world, however, income levels are relatively higher than wealth levels.\n\nAt the bottom of the scale, Sub-Saharan Africa, South & Southeast Asia, and Latin America remain far below the global average on both counts, though Latin America’s average income is somewhat closer to the world mean. The Middle East & North Africa and Russia & Central Asia occupy an intermediate position: their average incomes are closer to East Asia’s and not far from Latin America’s, but their average wealth remains much lower. The contrast is sharp: for instance, in 2025, average wealth in North America & Oceania is 338% of the world average, while in Sub-Saharan Africa it is just 20%. Put differently, the average adult in North America & Oceania owns more than sixteen times the wealth of the average adult in Sub-Saharan Africa. Turning to **Figure 1.13** , the focus shifts from differences between regions to disparities within them, using the top 10%/bottom 50% (T10/B50) ratio. This simple but powerful metric asks: on average, how many times more does the top 10% earn (or own) compared to the poorest half? The results reveal enormous divides. ⁵ First, wealth gaps everywhere are far larger\n\n43"],"full_content":"Chapter 1. Global Economic Inequality\n\n**Figure 1.8. Extreme wealth inequality is persistent and increasing**\n\nThe top 0.001% (about 56,000 adults) have 3 times more wealth than half of the entire world population combined.\n\n0%\n\n1%\n\n2%\n\n3%\n\n4%\n\n5%\n\n6%\n\n7%\n\n1995 2000 2005 2010 2015 2020 2025\n\nShare of global personal wealth (%)\n\nBottom 50% Top 0.001%\n\n**Extreme wealth inequality, 1995−2025**\n\n**Interpretation.** The share of personal wealth held by the richest 0.001% of adults rose from around 3.8% of total wealth in 1995 to nearly 6.1% in 2025. After a very slight increase, the share of wealth owned by the poorest half of the population has stagnated since the early 2000s at around 2%. Net personal wealth is equal to the sum of financial assets (e.g. equity or bonds) and non−financial assets (e.g. housing or land) owned by individuals, net of their debts. **Sources and series:** Arias−Osorio et al. (2025) and wir2026.wid.world/methodology.\n\nit is shaped not only by economics, but also by deep-rooted historical, political, and cultural legacies. It also highlights another key dimension: inequality does not just separate rich and poor individuals in the same context; it also entrenches divides between entire parts of the world. Regional comparisons allow us to see both how far apart regions stand from each other ( **Figure 1.12** ) and how unequal they are internally ( **Figure 1.13** ). The specific country groupings are detailed in **Box 1.1.1** .\n\n**Figure 1.12** contrasts average income and wealth per adult across world regions in 2025, relative to the world average. The patterns are striking. Wealthier regions are also typically higher-income regions, though there are exceptions. East Asia, for instance, has a higher average wealth level than Russia & Central Asia but a lower average income. At the top of the scale, North America & Oceania and Europe stand well above the global mean. In these regions, and in East Asia, wealth levels exceed income levels relative to the world average. In the rest of the world, however, income levels are relatively higher than wealth levels.\n\nAt the bottom of the scale, Sub-Saharan Africa, South & Southeast Asia, and Latin America remain far below the global average on both counts, though Latin America’s average income is somewhat closer to the world mean. The Middle East & North Africa and Russia & Central Asia occupy an intermediate position: their average incomes are closer to East Asia’s and not far from Latin America’s, but their average wealth remains much lower. The contrast is sharp: for instance, in 2025, average wealth in North America & Oceania is 338% of the world average, while in Sub-Saharan Africa it is just 20%. Put differently, the average adult in North America & Oceania owns more than sixteen times the wealth of the average adult in Sub-Saharan Africa. Turning to **Figure 1.13** , the focus shifts from differences between regions to disparities within them, using the top 10%/bottom 50% (T10/B50) ratio. This simple but powerful metric asks: on average, how many times more does the top 10% earn (or own) compared to the poorest half? The results reveal enormous divides. ⁵ First, wealth gaps everywhere are far larger\n\n43","publish_date":null,"title":"Chapter 1. Global Economic Inequality Figure1.8. Extremewealthinequalityispersistentand increasing it is shaped not only by economics, but At the bottom of","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p43.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p43.pdf"},"request":{"example_id":"chart/World_Inequality_Report_2026_p43","product_type":"parse","source_file_path":"World_Inequality_Report_2026_p43.pdf"},"started_at":"2026-11-04T19:45:44.981203"} +{"completed_at":"2026-11-04T19:45:47.091748","latency_in_ms":1141,"output":{"example_id":"chart/World_Inequality_Report_2026_p59","grounded_pages":[],"layout_pages":[],"markdown":"Chapter 2. Regional Income Inequality\n\n**Figure 2.6. Extreme concentration of income at the very top is a defining feature of the global economy**\n\n19%\n\n45%\n\n24%\n\n12%\n\n13%\n\n42%\n\n28%\n\n17%\n\n13%\n\n41%\n\n26%\n\n20%\n\n14%\n\n35%\n\n28%\n\n23%\n\n10%\n\n35%\n\n35%\n\n20%\n\n14%\n\n31%\n\n34%\n\n21%\n\n11%\n\n32%\n\n33%\n\n24%\n\n8%\n\n35%\n\n37%\n\n20%\n\n0%\n\n10% 20% 30% 40% 50% 60% 70% 80% 90%\n\n100%\n\nEurope East Asia North America & Oceania\n\nSub−\n\nSaharan\n\nAfrica\n\nLatin\n\nAmerica\n\nSouth & Southeast\n\nAsia\n\nRussia & Central\n\nAsia\n\nMiddle East\n\n& North\n\nAfrica\n\nShare of national income (%)\n\nTop 1% Next 9% Middle 40% Bottom 50%\n\n**Bottom 50%, middle 40% and top 10% (top 1% and**\n\n**next 9%) income shares across the world, 2025**\n\n**Interpretation.** In Latin America, the top 1% captures 20% of national income, and the next 9% an additional 37%. Together, the top 10% earns 57%, compared to 36% in Europe. Income is measured after pension and unemployment benefits are received by individuals, but before income taxes and other transfers. **Sources and series:** wir2026.wid.world/methodology, Andreescu and Sodano (2024), Bharti and Mo (2024), El Hariri (2024), Flores and Zúñiga−Cordero (2024), Forward and Fisher−Post (2024), and Loubes and Robilliard (2024).\n\nBy 2025, the picture looks markedly different. China’s position has shifted upward: much of its population has moved into the middle 40%, and a growing share has entered the upper-middle segments of the global distribution. This is also true for other Asian countries. India, by contrast, has lost relative ground: in 1980, a larger part of its population was in the middle 40%, but today almost all are in the bottom 50%. Sub-Saharan Africa has also remained in the lower half of the global distribution, though its population is now more evenly spread within the bottom 60% rather than clustered almost entirely below the 30th percentile, as it was in 1980. At the upper end of the distribution, continuity is unmistakable. North America & Oceania continue to dominate the global top 1%, with Europe also maintaining a large share. The composition of the global elite has diversified somewhat, with the Middle East & North Africa and Russia & Central Asia gaining ground, while Latin America’s representation has declined compared to 1980.\n\nThe shifts in **Figure 2.7** reveal a partial reshaping of the global income hierarchy. The rise of China has expanded the global middle class, while the very top remains concentrated in the Global North, and the bottom is heavily populated by South Asia and Sub-Saharan Africa. In short, the geography of inequality has been reshuffled, but not overturned.\n\n**Income inequality within countries in 2025**\n\n**Figure 2.8** to **Figure 2.12** take us inside national income distributions, showing how income is divided between the bottom 50%, the middle 40%, and the top 10% (with a focus also on the top 1%). Together, they illustrate how inequality plays out not only between regions but also within individual countries, and how the balance between these groups varies across the world.\n\n**Figure 2.8** begins with the bottom 50%. The poorest half of the population captures only a small share of national income almost everywhere. In the most unequal countries of Latin America and Sub-Saharan Africa,\n\n59","pages":[{"line_numbers":[],"markdown":"Chapter 2. Regional Income Inequality\n\n**Figure 2.6. Extreme concentration of income at the very top is a defining feature of the global economy**\n\n19%\n\n45%\n\n24%\n\n12%\n\n13%\n\n42%\n\n28%\n\n17%\n\n13%\n\n41%\n\n26%\n\n20%\n\n14%\n\n35%\n\n28%\n\n23%\n\n10%\n\n35%\n\n35%\n\n20%\n\n14%\n\n31%\n\n34%\n\n21%\n\n11%\n\n32%\n\n33%\n\n24%\n\n8%\n\n35%\n\n37%\n\n20%\n\n0%\n\n10% 20% 30% 40% 50% 60% 70% 80% 90%\n\n100%\n\nEurope East Asia North America & Oceania\n\nSub−\n\nSaharan\n\nAfrica\n\nLatin\n\nAmerica\n\nSouth & Southeast\n\nAsia\n\nRussia & Central\n\nAsia\n\nMiddle East\n\n& North\n\nAfrica\n\nShare of national income (%)\n\nTop 1% Next 9% Middle 40% Bottom 50%\n\n**Bottom 50%, middle 40% and top 10% (top 1% and**\n\n**next 9%) income shares across the world, 2025**\n\n**Interpretation.** In Latin America, the top 1% captures 20% of national income, and the next 9% an additional 37%. Together, the top 10% earns 57%, compared to 36% in Europe. Income is measured after pension and unemployment benefits are received by individuals, but before income taxes and other transfers. **Sources and series:** wir2026.wid.world/methodology, Andreescu and Sodano (2024), Bharti and Mo (2024), El Hariri (2024), Flores and Zúñiga−Cordero (2024), Forward and Fisher−Post (2024), and Loubes and Robilliard (2024).\n\nBy 2025, the picture looks markedly different. China’s position has shifted upward: much of its population has moved into the middle 40%, and a growing share has entered the upper-middle segments of the global distribution. This is also true for other Asian countries. India, by contrast, has lost relative ground: in 1980, a larger part of its population was in the middle 40%, but today almost all are in the bottom 50%. Sub-Saharan Africa has also remained in the lower half of the global distribution, though its population is now more evenly spread within the bottom 60% rather than clustered almost entirely below the 30th percentile, as it was in 1980. At the upper end of the distribution, continuity is unmistakable. North America & Oceania continue to dominate the global top 1%, with Europe also maintaining a large share. The composition of the global elite has diversified somewhat, with the Middle East & North Africa and Russia & Central Asia gaining ground, while Latin America’s representation has declined compared to 1980.\n\nThe shifts in **Figure 2.7** reveal a partial reshaping of the global income hierarchy. The rise of China has expanded the global middle class, while the very top remains concentrated in the Global North, and the bottom is heavily populated by South Asia and Sub-Saharan Africa. In short, the geography of inequality has been reshuffled, but not overturned.\n\n**Income inequality within countries in 2025**\n\n**Figure 2.8** to **Figure 2.12** take us inside national income distributions, showing how income is divided between the bottom 50%, the middle 40%, and the top 10% (with a focus also on the top 1%). Together, they illustrate how inequality plays out not only between regions but also within individual countries, and how the balance between these groups varies across the world.\n\n**Figure 2.8** begins with the bottom 50%. The poorest half of the population captures only a small share of national income almost everywhere. In the most unequal countries of Latin America and Sub-Saharan Africa,\n\n59","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Chapter 2. Regional Income Inequality\n\n**Figure 2.6. Extreme concentration of income at the very top is a defining feature of the global economy**\n\n19%\n\n45%\n\n24%\n\n12%\n\n13%\n\n42%\n\n28%\n\n17%\n\n13%\n\n41%\n\n26%\n\n20%\n\n14%\n\n35%\n\n28%\n\n23%\n\n10%\n\n35%\n\n35%\n\n20%\n\n14%\n\n31%\n\n34%\n\n21%\n\n11%\n\n32%\n\n33%\n\n24%\n\n8%\n\n35%\n\n37%\n\n20%\n\n0%\n\n10% 20% 30% 40% 50% 60% 70% 80% 90%\n\n100%\n\nEurope East Asia North America & Oceania\n\nSub−\n\nSaharan\n\nAfrica\n\nLatin\n\nAmerica\n\nSouth & Southeast\n\nAsia\n\nRussia & Central\n\nAsia\n\nMiddle East\n\n& North\n\nAfrica\n\nShare of national income (%)\n\nTop 1% Next 9% Middle 40% Bottom 50%\n\n**Bottom 50%, middle 40% and top 10% (top 1% and**\n\n**next 9%) income shares across the world, 2025**\n\n**Interpretation.** In Latin America, the top 1% captures 20% of national income, and the next 9% an additional 37%. Together, the top 10% earns 57%, compared to 36% in Europe. Income is measured after pension and unemployment benefits are received by individuals, but before income taxes and other transfers. **Sources and series:** wir2026.wid.world/methodology, Andreescu and Sodano (2024), Bharti and Mo (2024), El Hariri (2024), Flores and Zúñiga−Cordero (2024), Forward and Fisher−Post (2024), and Loubes and Robilliard (2024).\n\nBy 2025, the picture looks markedly different. China’s position has shifted upward: much of its population has moved into the middle 40%, and a growing share has entered the upper-middle segments of the global distribution. This is also true for other Asian countries. India, by contrast, has lost relative ground: in 1980, a larger part of its population was in the middle 40%, but today almost all are in the bottom 50%. Sub-Saharan Africa has also remained in the lower half of the global distribution, though its population is now more evenly spread within the bottom 60% rather than clustered almost entirely below the 30th percentile, as it was in 1980. At the upper end of the distribution, continuity is unmistakable. North America & Oceania continue to dominate the global top 1%, with Europe also maintaining a large share. The composition of the global elite has diversified somewhat, with the Middle East & North Africa and Russia & Central Asia gaining ground, while Latin America’s representation has declined compared to 1980.\n\nThe shifts in **Figure 2.7** reveal a partial reshaping of the global income hierarchy. The rise of China has expanded the global middle class, while the very top remains concentrated in the Global North, and the bottom is heavily populated by South Asia and Sub-Saharan Africa. In short, the geography of inequality has been reshuffled, but not overturned.\n\n**Income inequality within countries in 2025**\n\n**Figure 2.8** to **Figure 2.12** take us inside national income distributions, showing how income is divided between the bottom 50%, the middle 40%, and the top 10% (with a focus also on the top 1%). Together, they illustrate how inequality plays out not only between regions but also within individual countries, and how the balance between these groups varies across the world.\n\n**Figure 2.8** begins with the bottom 50%. The poorest half of the population captures only a small share of national income almost everywhere. In the most unequal countries of Latin America and Sub-Saharan Africa,\n\n59","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p59.pdf"]},"response":{"errors":[],"extract_id":"extract_64ae519a7c7906cabc7527b931d0b66b","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Chapter 2. Regional Income Inequality\n\n**Figure 2.6. Extreme concentration of income at the very top is a defining feature of the global economy**\n\n19%\n\n45%\n\n24%\n\n12%\n\n13%\n\n42%\n\n28%\n\n17%\n\n13%\n\n41%\n\n26%\n\n20%\n\n14%\n\n35%\n\n28%\n\n23%\n\n10%\n\n35%\n\n35%\n\n20%\n\n14%\n\n31%\n\n34%\n\n21%\n\n11%\n\n32%\n\n33%\n\n24%\n\n8%\n\n35%\n\n37%\n\n20%\n\n0%\n\n10% 20% 30% 40% 50% 60% 70% 80% 90%\n\n100%\n\nEurope East Asia North America & Oceania\n\nSub−\n\nSaharan\n\nAfrica\n\nLatin\n\nAmerica\n\nSouth & Southeast\n\nAsia\n\nRussia & Central\n\nAsia\n\nMiddle East\n\n& North\n\nAfrica\n\nShare of national income (%)\n\nTop 1% Next 9% Middle 40% Bottom 50%\n\n**Bottom 50%, middle 40% and top 10% (top 1% and**\n\n**next 9%) income shares across the world, 2025**\n\n**Interpretation.** In Latin America, the top 1% captures 20% of national income, and the next 9% an additional 37%. Together, the top 10% earns 57%, compared to 36% in Europe. Income is measured after pension and unemployment benefits are received by individuals, but before income taxes and other transfers. **Sources and series:** wir2026.wid.world/methodology, Andreescu and Sodano (2024), Bharti and Mo (2024), El Hariri (2024), Flores and Zúñiga−Cordero (2024), Forward and Fisher−Post (2024), and Loubes and Robilliard (2024).\n\nBy 2025, the picture looks markedly different. China’s position has shifted upward: much of its population has moved into the middle 40%, and a growing share has entered the upper-middle segments of the global distribution. This is also true for other Asian countries. India, by contrast, has lost relative ground: in 1980, a larger part of its population was in the middle 40%, but today almost all are in the bottom 50%. Sub-Saharan Africa has also remained in the lower half of the global distribution, though its population is now more evenly spread within the bottom 60% rather than clustered almost entirely below the 30th percentile, as it was in 1980. At the upper end of the distribution, continuity is unmistakable. North America & Oceania continue to dominate the global top 1%, with Europe also maintaining a large share. The composition of the global elite has diversified somewhat, with the Middle East & North Africa and Russia & Central Asia gaining ground, while Latin America’s representation has declined compared to 1980.\n\nThe shifts in **Figure 2.7** reveal a partial reshaping of the global income hierarchy. The rise of China has expanded the global middle class, while the very top remains concentrated in the Global North, and the bottom is heavily populated by South Asia and Sub-Saharan Africa. In short, the geography of inequality has been reshuffled, but not overturned.\n\n**Income inequality within countries in 2025**\n\n**Figure 2.8** to **Figure 2.12** take us inside national income distributions, showing how income is divided between the bottom 50%, the middle 40%, and the top 10% (with a focus also on the top 1%). Together, they illustrate how inequality plays out not only between regions but also within individual countries, and how the balance between these groups varies across the world.\n\n**Figure 2.8** begins with the bottom 50%. The poorest half of the population captures only a small share of national income almost everywhere. In the most unequal countries of Latin America and Sub-Saharan Africa,\n\n59"],"full_content":"Chapter 2. Regional Income Inequality\n\n**Figure 2.6. Extreme concentration of income at the very top is a defining feature of the global economy**\n\n19%\n\n45%\n\n24%\n\n12%\n\n13%\n\n42%\n\n28%\n\n17%\n\n13%\n\n41%\n\n26%\n\n20%\n\n14%\n\n35%\n\n28%\n\n23%\n\n10%\n\n35%\n\n35%\n\n20%\n\n14%\n\n31%\n\n34%\n\n21%\n\n11%\n\n32%\n\n33%\n\n24%\n\n8%\n\n35%\n\n37%\n\n20%\n\n0%\n\n10% 20% 30% 40% 50% 60% 70% 80% 90%\n\n100%\n\nEurope East Asia North America & Oceania\n\nSub−\n\nSaharan\n\nAfrica\n\nLatin\n\nAmerica\n\nSouth & Southeast\n\nAsia\n\nRussia & Central\n\nAsia\n\nMiddle East\n\n& North\n\nAfrica\n\nShare of national income (%)\n\nTop 1% Next 9% Middle 40% Bottom 50%\n\n**Bottom 50%, middle 40% and top 10% (top 1% and**\n\n**next 9%) income shares across the world, 2025**\n\n**Interpretation.** In Latin America, the top 1% captures 20% of national income, and the next 9% an additional 37%. Together, the top 10% earns 57%, compared to 36% in Europe. Income is measured after pension and unemployment benefits are received by individuals, but before income taxes and other transfers. **Sources and series:** wir2026.wid.world/methodology, Andreescu and Sodano (2024), Bharti and Mo (2024), El Hariri (2024), Flores and Zúñiga−Cordero (2024), Forward and Fisher−Post (2024), and Loubes and Robilliard (2024).\n\nBy 2025, the picture looks markedly different. China’s position has shifted upward: much of its population has moved into the middle 40%, and a growing share has entered the upper-middle segments of the global distribution. This is also true for other Asian countries. India, by contrast, has lost relative ground: in 1980, a larger part of its population was in the middle 40%, but today almost all are in the bottom 50%. Sub-Saharan Africa has also remained in the lower half of the global distribution, though its population is now more evenly spread within the bottom 60% rather than clustered almost entirely below the 30th percentile, as it was in 1980. At the upper end of the distribution, continuity is unmistakable. North America & Oceania continue to dominate the global top 1%, with Europe also maintaining a large share. The composition of the global elite has diversified somewhat, with the Middle East & North Africa and Russia & Central Asia gaining ground, while Latin America’s representation has declined compared to 1980.\n\nThe shifts in **Figure 2.7** reveal a partial reshaping of the global income hierarchy. The rise of China has expanded the global middle class, while the very top remains concentrated in the Global North, and the bottom is heavily populated by South Asia and Sub-Saharan Africa. In short, the geography of inequality has been reshuffled, but not overturned.\n\n**Income inequality within countries in 2025**\n\n**Figure 2.8** to **Figure 2.12** take us inside national income distributions, showing how income is divided between the bottom 50%, the middle 40%, and the top 10% (with a focus also on the top 1%). Together, they illustrate how inequality plays out not only between regions but also within individual countries, and how the balance between these groups varies across the world.\n\n**Figure 2.8** begins with the bottom 50%. The poorest half of the population captures only a small share of national income almost everywhere. In the most unequal countries of Latin America and Sub-Saharan Africa,\n\n59","publish_date":null,"title":"Chapter 2. Regional Income Inequality Figure2.6. Extremeconcentrationofincomeattheverytopisa definingfeatureoftheglobaleconomy By 2025, the picture looks markedly The shifts in Figure 2.7 reveal","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p59.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p59.pdf"},"request":{"example_id":"chart/World_Inequality_Report_2026_p59","product_type":"parse","source_file_path":"World_Inequality_Report_2026_p59.pdf"},"started_at":"2026-11-04T19:45:45.950124"} +{"completed_at":"2026-11-04T19:45:49.226887","latency_in_ms":1408,"output":{"example_id":"chart/World_Inequality_Report_2026_p66","grounded_pages":[],"layout_pages":[],"markdown":"Chapter 2. Regional Income Inequality\n\n**Figure 2.13. Redistribution decreases inequality within countries but with large variations**\n\n19\n\n10\n\n35\n\n28\n\n35\n\n18\n\n36\n\n33\n\n40 40\n\n52\n\n48\n\n53 50\n\n72\n\n50\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\n70\n\n80\n\nEurope East Asia\n\nNorth\n\nAmerica &\n\nOceania\n\nRussia & Central\n\nAsia\n\nSouth & Southeast\n\nAsia\n\nSub−\n\nSaharan\n\nAfrica\n\nMiddle East\n\n& North\n\nAfrica\n\nLatin\n\nAmerica\n\nTop 10% average to bottom 50%\n\naverage income\n\nBefore redistribution After redistribution\n\n**Top 10/bottom 50 income gaps across regions, 2025**\n\n**Interpretation.** In North America & Oceania, the bottom 50% earns 35 times less than the top 10% before income tax, whereas after income tax and all transfers, the bottom 50% earns 18 times less than the top 10%. Income is measured after pension and unemployment payments and benefits received by individuals but before other taxes they pay and transfers they receive. **Sources and series:** wir2026.wid.world/methodology and Chancel and Piketty (2021).\n\n**Figure 2.14. Transfers account for a larger share of redistribution than taxes**\n\n−20 −10\n\n0\n\n10 20 30 40 50\n\n1980 1985 1990 1995 2000 2005 2010 2015 2020\n\nReduction in\n\ntop 10% to bottom 50%\n\nincome ratio (%)\n\n**Tax progressivity**\n\n−20 −10\n\n0\n\n10 20 30 40 50\n\n1980 1985 1990 1995 2000 2005 2010 2015 2020\n\n**Redistribution**\n\n**considering taxes and transfers**\n\nEast Asia\n\nEurope\n\nLatin America\n\nMENA\n\nNorth America & Oceania\n\nRussia & Central Asia\n\nSouth & Southeast Asia\n\nSub−Saharan Africa\n\n**Intepretation.** The left panel shows the impact of taxes alone on inequality across regions, measured by the reduction in the top 10% to bottom 50% income ratio (a positive value signals inequality reduction). Taxes are more progressive in North America & Oceania; they consistently reduce inequality more than in any other region, while Latin America and Russia & Central Asia often show regressive tax effects. The right panel, which includes both taxes and transfers, reveals much greater redistributive effects in all regions and especially in Europe and North America & Oceania, highlighting the critical role of transfers in reducing global income inequality. **Sources and series:** wir2026.wid.world/methodology and Fisher−Post and Gethin (2025).\n\n**Reduction in top 10/bottom 50 income gaps, 1980−2025**\n\n66","pages":[{"line_numbers":[],"markdown":"Chapter 2. Regional Income Inequality\n\n**Figure 2.13. Redistribution decreases inequality within countries but with large variations**\n\n19\n\n10\n\n35\n\n28\n\n35\n\n18\n\n36\n\n33\n\n40 40\n\n52\n\n48\n\n53 50\n\n72\n\n50\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\n70\n\n80\n\nEurope East Asia\n\nNorth\n\nAmerica &\n\nOceania\n\nRussia & Central\n\nAsia\n\nSouth & Southeast\n\nAsia\n\nSub−\n\nSaharan\n\nAfrica\n\nMiddle East\n\n& North\n\nAfrica\n\nLatin\n\nAmerica\n\nTop 10% average to bottom 50%\n\naverage income\n\nBefore redistribution After redistribution\n\n**Top 10/bottom 50 income gaps across regions, 2025**\n\n**Interpretation.** In North America & Oceania, the bottom 50% earns 35 times less than the top 10% before income tax, whereas after income tax and all transfers, the bottom 50% earns 18 times less than the top 10%. Income is measured after pension and unemployment payments and benefits received by individuals but before other taxes they pay and transfers they receive. **Sources and series:** wir2026.wid.world/methodology and Chancel and Piketty (2021).\n\n**Figure 2.14. Transfers account for a larger share of redistribution than taxes**\n\n−20 −10\n\n0\n\n10 20 30 40 50\n\n1980 1985 1990 1995 2000 2005 2010 2015 2020\n\nReduction in\n\ntop 10% to bottom 50%\n\nincome ratio (%)\n\n**Tax progressivity**\n\n−20 −10\n\n0\n\n10 20 30 40 50\n\n1980 1985 1990 1995 2000 2005 2010 2015 2020\n\n**Redistribution**\n\n**considering taxes and transfers**\n\nEast Asia\n\nEurope\n\nLatin America\n\nMENA\n\nNorth America & Oceania\n\nRussia & Central Asia\n\nSouth & Southeast Asia\n\nSub−Saharan Africa\n\n**Intepretation.** The left panel shows the impact of taxes alone on inequality across regions, measured by the reduction in the top 10% to bottom 50% income ratio (a positive value signals inequality reduction). Taxes are more progressive in North America & Oceania; they consistently reduce inequality more than in any other region, while Latin America and Russia & Central Asia often show regressive tax effects. The right panel, which includes both taxes and transfers, reveals much greater redistributive effects in all regions and especially in Europe and North America & Oceania, highlighting the critical role of transfers in reducing global income inequality. **Sources and series:** wir2026.wid.world/methodology and Fisher−Post and Gethin (2025).\n\n**Reduction in top 10/bottom 50 income gaps, 1980−2025**\n\n66","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Chapter 2. Regional Income Inequality\n\n**Figure 2.13. Redistribution decreases inequality within countries but with large variations**\n\n19\n\n10\n\n35\n\n28\n\n35\n\n18\n\n36\n\n33\n\n40 40\n\n52\n\n48\n\n53 50\n\n72\n\n50\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\n70\n\n80\n\nEurope East Asia\n\nNorth\n\nAmerica &\n\nOceania\n\nRussia & Central\n\nAsia\n\nSouth & Southeast\n\nAsia\n\nSub−\n\nSaharan\n\nAfrica\n\nMiddle East\n\n& North\n\nAfrica\n\nLatin\n\nAmerica\n\nTop 10% average to bottom 50%\n\naverage income\n\nBefore redistribution After redistribution\n\n**Top 10/bottom 50 income gaps across regions, 2025**\n\n**Interpretation.** In North America & Oceania, the bottom 50% earns 35 times less than the top 10% before income tax, whereas after income tax and all transfers, the bottom 50% earns 18 times less than the top 10%. Income is measured after pension and unemployment payments and benefits received by individuals but before other taxes they pay and transfers they receive. **Sources and series:** wir2026.wid.world/methodology and Chancel and Piketty (2021).\n\n**Figure 2.14. Transfers account for a larger share of redistribution than taxes**\n\n−20 −10\n\n0\n\n10 20 30 40 50\n\n1980 1985 1990 1995 2000 2005 2010 2015 2020\n\nReduction in\n\ntop 10% to bottom 50%\n\nincome ratio (%)\n\n**Tax progressivity**\n\n−20 −10\n\n0\n\n10 20 30 40 50\n\n1980 1985 1990 1995 2000 2005 2010 2015 2020\n\n**Redistribution**\n\n**considering taxes and transfers**\n\nEast Asia\n\nEurope\n\nLatin America\n\nMENA\n\nNorth America & Oceania\n\nRussia & Central Asia\n\nSouth & Southeast Asia\n\nSub−Saharan Africa\n\n**Intepretation.** The left panel shows the impact of taxes alone on inequality across regions, measured by the reduction in the top 10% to bottom 50% income ratio (a positive value signals inequality reduction). Taxes are more progressive in North America & Oceania; they consistently reduce inequality more than in any other region, while Latin America and Russia & Central Asia often show regressive tax effects. The right panel, which includes both taxes and transfers, reveals much greater redistributive effects in all regions and especially in Europe and North America & Oceania, highlighting the critical role of transfers in reducing global income inequality. **Sources and series:** wir2026.wid.world/methodology and Fisher−Post and Gethin (2025).\n\n**Reduction in top 10/bottom 50 income gaps, 1980−2025**\n\n66","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p66.pdf"]},"response":{"errors":[],"extract_id":"extract_26de55f1b5decad1ea7c76dd1a9abc48","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Chapter 2. Regional Income Inequality\n\n**Figure 2.13. Redistribution decreases inequality within countries but with large variations**\n\n19\n\n10\n\n35\n\n28\n\n35\n\n18\n\n36\n\n33\n\n40 40\n\n52\n\n48\n\n53 50\n\n72\n\n50\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\n70\n\n80\n\nEurope East Asia\n\nNorth\n\nAmerica &\n\nOceania\n\nRussia & Central\n\nAsia\n\nSouth & Southeast\n\nAsia\n\nSub−\n\nSaharan\n\nAfrica\n\nMiddle East\n\n& North\n\nAfrica\n\nLatin\n\nAmerica\n\nTop 10% average to bottom 50%\n\naverage income\n\nBefore redistribution After redistribution\n\n**Top 10/bottom 50 income gaps across regions, 2025**\n\n**Interpretation.** In North America & Oceania, the bottom 50% earns 35 times less than the top 10% before income tax, whereas after income tax and all transfers, the bottom 50% earns 18 times less than the top 10%. Income is measured after pension and unemployment payments and benefits received by individuals but before other taxes they pay and transfers they receive. **Sources and series:** wir2026.wid.world/methodology and Chancel and Piketty (2021).\n\n**Figure 2.14. Transfers account for a larger share of redistribution than taxes**\n\n−20 −10\n\n0\n\n10 20 30 40 50\n\n1980 1985 1990 1995 2000 2005 2010 2015 2020\n\nReduction in\n\ntop 10% to bottom 50%\n\nincome ratio (%)\n\n**Tax progressivity**\n\n−20 −10\n\n0\n\n10 20 30 40 50\n\n1980 1985 1990 1995 2000 2005 2010 2015 2020\n\n**Redistribution**\n\n**considering taxes and transfers**\n\nEast Asia\n\nEurope\n\nLatin America\n\nMENA\n\nNorth America & Oceania\n\nRussia & Central Asia\n\nSouth & Southeast Asia\n\nSub−Saharan Africa\n\n**Intepretation.** The left panel shows the impact of taxes alone on inequality across regions, measured by the reduction in the top 10% to bottom 50% income ratio (a positive value signals inequality reduction). Taxes are more progressive in North America & Oceania; they consistently reduce inequality more than in any other region, while Latin America and Russia & Central Asia often show regressive tax effects. The right panel, which includes both taxes and transfers, reveals much greater redistributive effects in all regions and especially in Europe and North America & Oceania, highlighting the critical role of transfers in reducing global income inequality. **Sources and series:** wir2026.wid.world/methodology and Fisher−Post and Gethin (2025).\n\n**Reduction in top 10/bottom 50 income gaps, 1980−2025**\n\n66"],"full_content":"Chapter 2. Regional Income Inequality\n\n**Figure 2.13. Redistribution decreases inequality within countries but with large variations**\n\n19\n\n10\n\n35\n\n28\n\n35\n\n18\n\n36\n\n33\n\n40 40\n\n52\n\n48\n\n53 50\n\n72\n\n50\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\n70\n\n80\n\nEurope East Asia\n\nNorth\n\nAmerica &\n\nOceania\n\nRussia & Central\n\nAsia\n\nSouth & Southeast\n\nAsia\n\nSub−\n\nSaharan\n\nAfrica\n\nMiddle East\n\n& North\n\nAfrica\n\nLatin\n\nAmerica\n\nTop 10% average to bottom 50%\n\naverage income\n\nBefore redistribution After redistribution\n\n**Top 10/bottom 50 income gaps across regions, 2025**\n\n**Interpretation.** In North America & Oceania, the bottom 50% earns 35 times less than the top 10% before income tax, whereas after income tax and all transfers, the bottom 50% earns 18 times less than the top 10%. Income is measured after pension and unemployment payments and benefits received by individuals but before other taxes they pay and transfers they receive. **Sources and series:** wir2026.wid.world/methodology and Chancel and Piketty (2021).\n\n**Figure 2.14. Transfers account for a larger share of redistribution than taxes**\n\n−20 −10\n\n0\n\n10 20 30 40 50\n\n1980 1985 1990 1995 2000 2005 2010 2015 2020\n\nReduction in\n\ntop 10% to bottom 50%\n\nincome ratio (%)\n\n**Tax progressivity**\n\n−20 −10\n\n0\n\n10 20 30 40 50\n\n1980 1985 1990 1995 2000 2005 2010 2015 2020\n\n**Redistribution**\n\n**considering taxes and transfers**\n\nEast Asia\n\nEurope\n\nLatin America\n\nMENA\n\nNorth America & Oceania\n\nRussia & Central Asia\n\nSouth & Southeast Asia\n\nSub−Saharan Africa\n\n**Intepretation.** The left panel shows the impact of taxes alone on inequality across regions, measured by the reduction in the top 10% to bottom 50% income ratio (a positive value signals inequality reduction). Taxes are more progressive in North America & Oceania; they consistently reduce inequality more than in any other region, while Latin America and Russia & Central Asia often show regressive tax effects. The right panel, which includes both taxes and transfers, reveals much greater redistributive effects in all regions and especially in Europe and North America & Oceania, highlighting the critical role of transfers in reducing global income inequality. **Sources and series:** wir2026.wid.world/methodology and Fisher−Post and Gethin (2025).\n\n**Reduction in top 10/bottom 50 income gaps, 1980−2025**\n\n66","publish_date":null,"title":"Chapter 2. Regional Income Inequality Figure2.13. Redistributiondecreasesinequalitywithin countriesbutwithlargevariations Figure2.14. Transfersaccountforalargershareofredistribution thantaxes 66","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p66.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p66.pdf"},"request":{"example_id":"chart/World_Inequality_Report_2026_p66","product_type":"parse","source_file_path":"World_Inequality_Report_2026_p66.pdf"},"started_at":"2026-11-04T19:45:47.818413"} +{"completed_at":"2026-11-04T19:45:49.472204","latency_in_ms":1153,"output":{"example_id":"chart/World_Inequality_Report_2026_p71","grounded_pages":[],"layout_pages":[],"markdown":"Chapter 2. Regional Income Inequality\n\n0%\n\n100%\n\n200%\n\n300%\n\n400%\n\n500%\n\n600%\n\n700%\n\n1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 2020 2025\n\nPer capita national income\n\n(% of world average, 2025 PPP €)\n\n0−100k\n\n100k−1m\n\n1m−10m\n\n10m−50m\n\n50m−100m\n\n100m−500m\n\nover 500m\n\n**Figure B2.2.2. Per capita national income by country size, 1970–2025**\n\n**Interpretation.** Ultra−small countries (pop. <100k) have consistently had above−average per capita income, increasing from 290% of the world average in 1970 to 400% in 2025. In contrast, the most populous countries (pop. >500m) remain significantly below average: from 20% in 1970 to 80% in 2025. This size−income gradient has remained persistent across decades. Population groups are defined using 2025 country sizes. **Sources and series:** Gómez−Carrera et al. (2024) and wir2026.wid.world/methodology.\n\n71","pages":[{"line_numbers":[],"markdown":"Chapter 2. Regional Income Inequality\n\n0%\n\n100%\n\n200%\n\n300%\n\n400%\n\n500%\n\n600%\n\n700%\n\n1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 2020 2025\n\nPer capita national income\n\n(% of world average, 2025 PPP €)\n\n0−100k\n\n100k−1m\n\n1m−10m\n\n10m−50m\n\n50m−100m\n\n100m−500m\n\nover 500m\n\n**Figure B2.2.2. Per capita national income by country size, 1970–2025**\n\n**Interpretation.** Ultra−small countries (pop. <100k) have consistently had above−average per capita income, increasing from 290% of the world average in 1970 to 400% in 2025. In contrast, the most populous countries (pop. >500m) remain significantly below average: from 20% in 1970 to 80% in 2025. This size−income gradient has remained persistent across decades. Population groups are defined using 2025 country sizes. **Sources and series:** Gómez−Carrera et al. (2024) and wir2026.wid.world/methodology.\n\n71","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Chapter 2. Regional Income Inequality\n\n0%\n\n100%\n\n200%\n\n300%\n\n400%\n\n500%\n\n600%\n\n700%\n\n1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 2020 2025\n\nPer capita national income\n\n(% of world average, 2025 PPP €)\n\n0−100k\n\n100k−1m\n\n1m−10m\n\n10m−50m\n\n50m−100m\n\n100m−500m\n\nover 500m\n\n**Figure B2.2.2. Per capita national income by country size, 1970–2025**\n\n**Interpretation.** Ultra−small countries (pop. <100k) have consistently had above−average per capita income, increasing from 290% of the world average in 1970 to 400% in 2025. In contrast, the most populous countries (pop. >500m) remain significantly below average: from 20% in 1970 to 80% in 2025. This size−income gradient has remained persistent across decades. Population groups are defined using 2025 country sizes. **Sources and series:** Gómez−Carrera et al. (2024) and wir2026.wid.world/methodology.\n\n71","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p71.pdf"]},"response":{"errors":[],"extract_id":"extract_6d27b07c0a869f8335a0703c6ba28405","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Chapter 2. Regional Income Inequality\n\n0%\n\n100%\n\n200%\n\n300%\n\n400%\n\n500%\n\n600%\n\n700%\n\n1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 2020 2025\n\nPer capita national income\n\n(% of world average, 2025 PPP €)\n\n0−100k\n\n100k−1m\n\n1m−10m\n\n10m−50m\n\n50m−100m\n\n100m−500m\n\nover 500m\n\n**Figure B2.2.2. Per capita national income by country size, 1970–2025**\n\n**Interpretation.** Ultra−small countries (pop. <100k) have consistently had above−average per capita income, increasing from 290% of the world average in 1970 to 400% in 2025. In contrast, the most populous countries (pop. >500m) remain significantly below average: from 20% in 1970 to 80% in 2025. This size−income gradient has remained persistent across decades. Population groups are defined using 2025 country sizes. **Sources and series:** Gómez−Carrera et al. (2024) and wir2026.wid.world/methodology.\n\n71"],"full_content":"Chapter 2. Regional Income Inequality\n\n0%\n\n100%\n\n200%\n\n300%\n\n400%\n\n500%\n\n600%\n\n700%\n\n1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 2020 2025\n\nPer capita national income\n\n(% of world average, 2025 PPP €)\n\n0−100k\n\n100k−1m\n\n1m−10m\n\n10m−50m\n\n50m−100m\n\n100m−500m\n\nover 500m\n\n**Figure B2.2.2. Per capita national income by country size, 1970–2025**\n\n**Interpretation.** Ultra−small countries (pop. <100k) have consistently had above−average per capita income, increasing from 290% of the world average in 1970 to 400% in 2025. In contrast, the most populous countries (pop. >500m) remain significantly below average: from 20% in 1970 to 80% in 2025. This size−income gradient has remained persistent across decades. Population groups are defined using 2025 country sizes. **Sources and series:** Gómez−Carrera et al. (2024) and wir2026.wid.world/methodology.\n\n71","publish_date":null,"title":"Chapter 2. Regional Income Inequality 71","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p71.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p71.pdf"},"request":{"example_id":"chart/World_Inequality_Report_2026_p71","product_type":"parse","source_file_path":"World_Inequality_Report_2026_p71.pdf"},"started_at":"2026-11-04T19:45:48.318574"} +{"completed_at":"2026-11-04T19:45:51.979097","latency_in_ms":2339,"output":{"example_id":"chart/World_Inequality_Report_2026_p77","grounded_pages":[],"layout_pages":[],"markdown":"Chapter 3. Regional Wealth Inequality\n\n**Figure 3.2. From the post war decades onward, most countries experienced renewed rises in wealth ratios**\n\n0%\n\n100%\n\n200%\n\n300%\n\n400%\n\n500%\n\n600%\n\n700%\n\n800%\n\n900%\n\n1 000%\n\n1860 1880 1900 1920 1940 1960 1980 2000 2020\n\nWealth as a % of national income\n\nChina\n\nFrance\n\nGermany\n\nIndia\n\nJapan\n\nSpain\n\nSweden\n\nUnited Kingdom\n\nUnited States\n\nWorld\n\n**Historical net national wealth to net national income ratio, 1860−2025**\n\n**Interpretation.** This graph shows the historical evolution of the ratio of net national wealth to net national income for certain countries. A higher ratio indicates that a country or region holds more wealth relative to its yearly income, reflecting both accumulated savings and capital gains. The ratio of national wealth to national income collapsed across countries during the first half of the 20th century but has rebounded sharply since the 1980s, especially in China. **Sources and series:** Bauluz et al. (2025) and wir2026.wid.world/methodology.\n\nAmerica, and Sub-Saharan Africa generally record negative foreign assets balances, underscoring their reliance on external capital and the fact that part of their domestic wealth is owned by foreigners. In recent years, Europe has begun to resemble the East Asian case, with residents increasingly holding more assets abroad than foreigners own within Europe. Russia & Central Asia, meanwhile, has seen a similar movement toward a positive position. The overall picture is twofold. First, domestic capital remains the foundation of national wealth everywhere. Second, although net foreign assets are relatively small in scale, they determine the highly embedded nature of global financial linkages. East Asia, Europe, and the Middle East & North Africa hold more assets abroad than foreigners own within their borders. By contrast, a significant portion of the wealth in North America, Latin America, and Sub-Saharan Africa is effectively owned outside their borders. Global wealth is not only concentrated but also deeply interconnected across regions.\n\n**Figure 3.4** illustrates the evolution of net foreign wealth across regions, expressed both in relation to regional GDP (left) and global GDP (right). These balances reveal which regions act as global creditors and which as debtors. In the 19th and early 20th centuries, Europe held the largest foreign asset position in history, with net wealth abroad reaching over 70% of its GDP and close to one-third of world GDP before 1914. Much of this was built on colonial extraction and unequal exchange, as is mirrored by the heavily negative positions of South & Southeast Asia and Sub-Saharan Africa (see Nievas and Piketty 2025). The two world wars, revolutions, and decolonization brought this dominance to an abrupt end, wiping out most of Europe’s external holdings. The mid-20th century saw North America & Oceania briefly become the world’s leading creditors, peaking at around 8% of world GDP in the 1950s. But since the 1970s, the region has shifted into the largest net debtor, with balances now at 18% of world GDP. The most dramatic contemporary\n\n77","pages":[{"line_numbers":[],"markdown":"Chapter 3. Regional Wealth Inequality\n\n**Figure 3.2. From the post war decades onward, most countries experienced renewed rises in wealth ratios**\n\n0%\n\n100%\n\n200%\n\n300%\n\n400%\n\n500%\n\n600%\n\n700%\n\n800%\n\n900%\n\n1 000%\n\n1860 1880 1900 1920 1940 1960 1980 2000 2020\n\nWealth as a % of national income\n\nChina\n\nFrance\n\nGermany\n\nIndia\n\nJapan\n\nSpain\n\nSweden\n\nUnited Kingdom\n\nUnited States\n\nWorld\n\n**Historical net national wealth to net national income ratio, 1860−2025**\n\n**Interpretation.** This graph shows the historical evolution of the ratio of net national wealth to net national income for certain countries. A higher ratio indicates that a country or region holds more wealth relative to its yearly income, reflecting both accumulated savings and capital gains. The ratio of national wealth to national income collapsed across countries during the first half of the 20th century but has rebounded sharply since the 1980s, especially in China. **Sources and series:** Bauluz et al. (2025) and wir2026.wid.world/methodology.\n\nAmerica, and Sub-Saharan Africa generally record negative foreign assets balances, underscoring their reliance on external capital and the fact that part of their domestic wealth is owned by foreigners. In recent years, Europe has begun to resemble the East Asian case, with residents increasingly holding more assets abroad than foreigners own within Europe. Russia & Central Asia, meanwhile, has seen a similar movement toward a positive position. The overall picture is twofold. First, domestic capital remains the foundation of national wealth everywhere. Second, although net foreign assets are relatively small in scale, they determine the highly embedded nature of global financial linkages. East Asia, Europe, and the Middle East & North Africa hold more assets abroad than foreigners own within their borders. By contrast, a significant portion of the wealth in North America, Latin America, and Sub-Saharan Africa is effectively owned outside their borders. Global wealth is not only concentrated but also deeply interconnected across regions.\n\n**Figure 3.4** illustrates the evolution of net foreign wealth across regions, expressed both in relation to regional GDP (left) and global GDP (right). These balances reveal which regions act as global creditors and which as debtors. In the 19th and early 20th centuries, Europe held the largest foreign asset position in history, with net wealth abroad reaching over 70% of its GDP and close to one-third of world GDP before 1914. Much of this was built on colonial extraction and unequal exchange, as is mirrored by the heavily negative positions of South & Southeast Asia and Sub-Saharan Africa (see Nievas and Piketty 2025). The two world wars, revolutions, and decolonization brought this dominance to an abrupt end, wiping out most of Europe’s external holdings. The mid-20th century saw North America & Oceania briefly become the world’s leading creditors, peaking at around 8% of world GDP in the 1950s. But since the 1970s, the region has shifted into the largest net debtor, with balances now at 18% of world GDP. The most dramatic contemporary\n\n77","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Chapter 3. Regional Wealth Inequality\n\n**Figure 3.2. From the post war decades onward, most countries experienced renewed rises in wealth ratios**\n\n0%\n\n100%\n\n200%\n\n300%\n\n400%\n\n500%\n\n600%\n\n700%\n\n800%\n\n900%\n\n1 000%\n\n1860 1880 1900 1920 1940 1960 1980 2000 2020\n\nWealth as a % of national income\n\nChina\n\nFrance\n\nGermany\n\nIndia\n\nJapan\n\nSpain\n\nSweden\n\nUnited Kingdom\n\nUnited States\n\nWorld\n\n**Historical net national wealth to net national income ratio, 1860−2025**\n\n**Interpretation.** This graph shows the historical evolution of the ratio of net national wealth to net national income for certain countries. A higher ratio indicates that a country or region holds more wealth relative to its yearly income, reflecting both accumulated savings and capital gains. The ratio of national wealth to national income collapsed across countries during the first half of the 20th century but has rebounded sharply since the 1980s, especially in China. **Sources and series:** Bauluz et al. (2025) and wir2026.wid.world/methodology.\n\nAmerica, and Sub-Saharan Africa generally record negative foreign assets balances, underscoring their reliance on external capital and the fact that part of their domestic wealth is owned by foreigners. In recent years, Europe has begun to resemble the East Asian case, with residents increasingly holding more assets abroad than foreigners own within Europe. Russia & Central Asia, meanwhile, has seen a similar movement toward a positive position. The overall picture is twofold. First, domestic capital remains the foundation of national wealth everywhere. Second, although net foreign assets are relatively small in scale, they determine the highly embedded nature of global financial linkages. East Asia, Europe, and the Middle East & North Africa hold more assets abroad than foreigners own within their borders. By contrast, a significant portion of the wealth in North America, Latin America, and Sub-Saharan Africa is effectively owned outside their borders. Global wealth is not only concentrated but also deeply interconnected across regions.\n\n**Figure 3.4** illustrates the evolution of net foreign wealth across regions, expressed both in relation to regional GDP (left) and global GDP (right). These balances reveal which regions act as global creditors and which as debtors. In the 19th and early 20th centuries, Europe held the largest foreign asset position in history, with net wealth abroad reaching over 70% of its GDP and close to one-third of world GDP before 1914. Much of this was built on colonial extraction and unequal exchange, as is mirrored by the heavily negative positions of South & Southeast Asia and Sub-Saharan Africa (see Nievas and Piketty 2025). The two world wars, revolutions, and decolonization brought this dominance to an abrupt end, wiping out most of Europe’s external holdings. The mid-20th century saw North America & Oceania briefly become the world’s leading creditors, peaking at around 8% of world GDP in the 1950s. But since the 1970s, the region has shifted into the largest net debtor, with balances now at 18% of world GDP. The most dramatic contemporary\n\n77","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p77.pdf"]},"response":{"errors":[],"extract_id":"extract_8eeb1e02a574f92efb029d4d036fcb96","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Chapter 3. Regional Wealth Inequality\n\n**Figure 3.2. From the post war decades onward, most countries experienced renewed rises in wealth ratios**\n\n0%\n\n100%\n\n200%\n\n300%\n\n400%\n\n500%\n\n600%\n\n700%\n\n800%\n\n900%\n\n1 000%\n\n1860 1880 1900 1920 1940 1960 1980 2000 2020\n\nWealth as a % of national income\n\nChina\n\nFrance\n\nGermany\n\nIndia\n\nJapan\n\nSpain\n\nSweden\n\nUnited Kingdom\n\nUnited States\n\nWorld\n\n**Historical net national wealth to net national income ratio, 1860−2025**\n\n**Interpretation.** This graph shows the historical evolution of the ratio of net national wealth to net national income for certain countries. A higher ratio indicates that a country or region holds more wealth relative to its yearly income, reflecting both accumulated savings and capital gains. The ratio of national wealth to national income collapsed across countries during the first half of the 20th century but has rebounded sharply since the 1980s, especially in China. **Sources and series:** Bauluz et al. (2025) and wir2026.wid.world/methodology.\n\nAmerica, and Sub-Saharan Africa generally record negative foreign assets balances, underscoring their reliance on external capital and the fact that part of their domestic wealth is owned by foreigners. In recent years, Europe has begun to resemble the East Asian case, with residents increasingly holding more assets abroad than foreigners own within Europe. Russia & Central Asia, meanwhile, has seen a similar movement toward a positive position. The overall picture is twofold. First, domestic capital remains the foundation of national wealth everywhere. Second, although net foreign assets are relatively small in scale, they determine the highly embedded nature of global financial linkages. East Asia, Europe, and the Middle East & North Africa hold more assets abroad than foreigners own within their borders. By contrast, a significant portion of the wealth in North America, Latin America, and Sub-Saharan Africa is effectively owned outside their borders. Global wealth is not only concentrated but also deeply interconnected across regions.\n\n**Figure 3.4** illustrates the evolution of net foreign wealth across regions, expressed both in relation to regional GDP (left) and global GDP (right). These balances reveal which regions act as global creditors and which as debtors. In the 19th and early 20th centuries, Europe held the largest foreign asset position in history, with net wealth abroad reaching over 70% of its GDP and close to one-third of world GDP before 1914. Much of this was built on colonial extraction and unequal exchange, as is mirrored by the heavily negative positions of South & Southeast Asia and Sub-Saharan Africa (see Nievas and Piketty 2025). The two world wars, revolutions, and decolonization brought this dominance to an abrupt end, wiping out most of Europe’s external holdings. The mid-20th century saw North America & Oceania briefly become the world’s leading creditors, peaking at around 8% of world GDP in the 1950s. But since the 1970s, the region has shifted into the largest net debtor, with balances now at 18% of world GDP. The most dramatic contemporary\n\n77"],"full_content":"Chapter 3. Regional Wealth Inequality\n\n**Figure 3.2. From the post war decades onward, most countries experienced renewed rises in wealth ratios**\n\n0%\n\n100%\n\n200%\n\n300%\n\n400%\n\n500%\n\n600%\n\n700%\n\n800%\n\n900%\n\n1 000%\n\n1860 1880 1900 1920 1940 1960 1980 2000 2020\n\nWealth as a % of national income\n\nChina\n\nFrance\n\nGermany\n\nIndia\n\nJapan\n\nSpain\n\nSweden\n\nUnited Kingdom\n\nUnited States\n\nWorld\n\n**Historical net national wealth to net national income ratio, 1860−2025**\n\n**Interpretation.** This graph shows the historical evolution of the ratio of net national wealth to net national income for certain countries. A higher ratio indicates that a country or region holds more wealth relative to its yearly income, reflecting both accumulated savings and capital gains. The ratio of national wealth to national income collapsed across countries during the first half of the 20th century but has rebounded sharply since the 1980s, especially in China. **Sources and series:** Bauluz et al. (2025) and wir2026.wid.world/methodology.\n\nAmerica, and Sub-Saharan Africa generally record negative foreign assets balances, underscoring their reliance on external capital and the fact that part of their domestic wealth is owned by foreigners. In recent years, Europe has begun to resemble the East Asian case, with residents increasingly holding more assets abroad than foreigners own within Europe. Russia & Central Asia, meanwhile, has seen a similar movement toward a positive position. The overall picture is twofold. First, domestic capital remains the foundation of national wealth everywhere. Second, although net foreign assets are relatively small in scale, they determine the highly embedded nature of global financial linkages. East Asia, Europe, and the Middle East & North Africa hold more assets abroad than foreigners own within their borders. By contrast, a significant portion of the wealth in North America, Latin America, and Sub-Saharan Africa is effectively owned outside their borders. Global wealth is not only concentrated but also deeply interconnected across regions.\n\n**Figure 3.4** illustrates the evolution of net foreign wealth across regions, expressed both in relation to regional GDP (left) and global GDP (right). These balances reveal which regions act as global creditors and which as debtors. In the 19th and early 20th centuries, Europe held the largest foreign asset position in history, with net wealth abroad reaching over 70% of its GDP and close to one-third of world GDP before 1914. Much of this was built on colonial extraction and unequal exchange, as is mirrored by the heavily negative positions of South & Southeast Asia and Sub-Saharan Africa (see Nievas and Piketty 2025). The two world wars, revolutions, and decolonization brought this dominance to an abrupt end, wiping out most of Europe’s external holdings. The mid-20th century saw North America & Oceania briefly become the world’s leading creditors, peaking at around 8% of world GDP in the 1950s. But since the 1970s, the region has shifted into the largest net debtor, with balances now at 18% of world GDP. The most dramatic contemporary\n\n77","publish_date":null,"title":"Chapter 3. Regional Wealth Inequality Figure3.2. Fromthepostwardecadesonward,mostcountries experiencedrenewedrisesinwealthratios America, and Sub-Saharan Africa generally Figure 3.4 illustrates the evolution of net","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p77.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p77.pdf"},"request":{"example_id":"chart/World_Inequality_Report_2026_p77","product_type":"parse","source_file_path":"World_Inequality_Report_2026_p77.pdf"},"started_at":"2026-11-04T19:45:49.639440"} +{"completed_at":"2026-11-04T19:45:54.100724","latency_in_ms":3178,"output":{"example_id":"chart/World_Inequality_Report_2026_p78","grounded_pages":[],"layout_pages":[],"markdown":"Chapter 3. Regional Wealth Inequality\n\n**Figure 3.3. Domestic capital remains the foundation of national wealth everywhere**\n\n**Net national wealth = Domestic capital + Net foreign assets**\n\n1995 2005 2015 2025 1995 2005 2015 2025 1995 2005 2015 2025\n\n0%\n\n100%\n\n200%\n\n0%\n\n250%\n\n500%\n\n750%\n\n0%\n\n250%\n\n500%\n\n750%\n\nWealth as a % of national income\n\nEast Asia\n\nEurope\n\nLatin America\n\nMENA\n\nNorth America & Oceania\n\nRussia & Central Asia\n\nSouth & Southeast Asia\n\nSub−Saharan Africa\n\n**Wealth composition across regions, 1995−2025**\n\n**Interpretation.** This figure shows that domestic capital makes up the bulk of net national wealth across all regions, while net foreign assets play only a minor role. Most regions exhibit steady increases in national wealth as a share of income since 1995, particularly East Asia and North America & Oceania. Sub−Saharan Africa and Latin America, by contrast, show limited growth and continue to hold negligible net foreign assets. **Notes.** Net national wealth = domestical capital + net foreign assets. **Sources and series:** Bauluz et al. (2025) and wir2026.wid.world/methodology.\n\ntransformation has been in East Asia. Since the 1970s, the region has become the largest creditor. Today, it holds about 12% of the world’s GDP. The Middle East & North Africa has also maintained a strong positive balance since the oil boom of the 1970s, while Europe has rebuilt modest surpluses in recent decades. By contrast, Latin America, Sub-Saharan Africa, and South & Southeast Asia remain consistent debtors, with foreigners owning more of their assets than their residents hold abroad. Foreign asset positions show the persistent global asymmetries of the financial system: today, East Asia, the Middle East & North Africa, and Europe finance the rest of the world, while North America & Oceania and most of the Global South run chronic deficits. It is important to note, as Nievas and Piketty (2025) emphasize, that these historical imbalances reflect not just markets but also power relations and unequal exchange, a theme we revisit in **Chapter 5** .\n\n**Private wealth is rising while public wealth stagnates**\n\nOver the past thirty years, global wealth has risen faster than income, growing from just over 400% of world income in 1995 to more than 600% in 2025 ( **Figure 3.2** ). Yet this rise has been almost entirely concentrated in the private sector (see Bauluz, Brassac, et al. (2025)). **Figure 3.5** shows that private wealth increased from about 350% to over 500% of world income, while public wealth stagnated at around 80–90%. In some regions, public wealth even turned negative, meaning that governments’ liabilities exceeded their assets. The slowdown during the COVID-19 pandemic briefly interrupted the upward trajectory, but the long-run trend remains clear: wealth growth has accumulated in private hands.\n\n**Figure 3.5** shows that East Asia and North America & Oceania now report the highest levels of private wealth, each above 600% of income by 2025 (left-hand-side panel). North America & Oceania’s trajectory, however, has been volatile: strong growth up to 2007, a sharp fall during the global\n\n78","pages":[{"line_numbers":[],"markdown":"Chapter 3. Regional Wealth Inequality\n\n**Figure 3.3. Domestic capital remains the foundation of national wealth everywhere**\n\n**Net national wealth = Domestic capital + Net foreign assets**\n\n1995 2005 2015 2025 1995 2005 2015 2025 1995 2005 2015 2025\n\n0%\n\n100%\n\n200%\n\n0%\n\n250%\n\n500%\n\n750%\n\n0%\n\n250%\n\n500%\n\n750%\n\nWealth as a % of national income\n\nEast Asia\n\nEurope\n\nLatin America\n\nMENA\n\nNorth America & Oceania\n\nRussia & Central Asia\n\nSouth & Southeast Asia\n\nSub−Saharan Africa\n\n**Wealth composition across regions, 1995−2025**\n\n**Interpretation.** This figure shows that domestic capital makes up the bulk of net national wealth across all regions, while net foreign assets play only a minor role. Most regions exhibit steady increases in national wealth as a share of income since 1995, particularly East Asia and North America & Oceania. Sub−Saharan Africa and Latin America, by contrast, show limited growth and continue to hold negligible net foreign assets. **Notes.** Net national wealth = domestical capital + net foreign assets. **Sources and series:** Bauluz et al. (2025) and wir2026.wid.world/methodology.\n\ntransformation has been in East Asia. Since the 1970s, the region has become the largest creditor. Today, it holds about 12% of the world’s GDP. The Middle East & North Africa has also maintained a strong positive balance since the oil boom of the 1970s, while Europe has rebuilt modest surpluses in recent decades. By contrast, Latin America, Sub-Saharan Africa, and South & Southeast Asia remain consistent debtors, with foreigners owning more of their assets than their residents hold abroad. Foreign asset positions show the persistent global asymmetries of the financial system: today, East Asia, the Middle East & North Africa, and Europe finance the rest of the world, while North America & Oceania and most of the Global South run chronic deficits. It is important to note, as Nievas and Piketty (2025) emphasize, that these historical imbalances reflect not just markets but also power relations and unequal exchange, a theme we revisit in **Chapter 5** .\n\n**Private wealth is rising while public wealth stagnates**\n\nOver the past thirty years, global wealth has risen faster than income, growing from just over 400% of world income in 1995 to more than 600% in 2025 ( **Figure 3.2** ). Yet this rise has been almost entirely concentrated in the private sector (see Bauluz, Brassac, et al. (2025)). **Figure 3.5** shows that private wealth increased from about 350% to over 500% of world income, while public wealth stagnated at around 80–90%. In some regions, public wealth even turned negative, meaning that governments’ liabilities exceeded their assets. The slowdown during the COVID-19 pandemic briefly interrupted the upward trajectory, but the long-run trend remains clear: wealth growth has accumulated in private hands.\n\n**Figure 3.5** shows that East Asia and North America & Oceania now report the highest levels of private wealth, each above 600% of income by 2025 (left-hand-side panel). North America & Oceania’s trajectory, however, has been volatile: strong growth up to 2007, a sharp fall during the global\n\n78","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Chapter 3. Regional Wealth Inequality\n\n**Figure 3.3. Domestic capital remains the foundation of national wealth everywhere**\n\n**Net national wealth = Domestic capital + Net foreign assets**\n\n1995 2005 2015 2025 1995 2005 2015 2025 1995 2005 2015 2025\n\n0%\n\n100%\n\n200%\n\n0%\n\n250%\n\n500%\n\n750%\n\n0%\n\n250%\n\n500%\n\n750%\n\nWealth as a % of national income\n\nEast Asia\n\nEurope\n\nLatin America\n\nMENA\n\nNorth America & Oceania\n\nRussia & Central Asia\n\nSouth & Southeast Asia\n\nSub−Saharan Africa\n\n**Wealth composition across regions, 1995−2025**\n\n**Interpretation.** This figure shows that domestic capital makes up the bulk of net national wealth across all regions, while net foreign assets play only a minor role. Most regions exhibit steady increases in national wealth as a share of income since 1995, particularly East Asia and North America & Oceania. Sub−Saharan Africa and Latin America, by contrast, show limited growth and continue to hold negligible net foreign assets. **Notes.** Net national wealth = domestical capital + net foreign assets. **Sources and series:** Bauluz et al. (2025) and wir2026.wid.world/methodology.\n\ntransformation has been in East Asia. Since the 1970s, the region has become the largest creditor. Today, it holds about 12% of the world’s GDP. The Middle East & North Africa has also maintained a strong positive balance since the oil boom of the 1970s, while Europe has rebuilt modest surpluses in recent decades. By contrast, Latin America, Sub-Saharan Africa, and South & Southeast Asia remain consistent debtors, with foreigners owning more of their assets than their residents hold abroad. Foreign asset positions show the persistent global asymmetries of the financial system: today, East Asia, the Middle East & North Africa, and Europe finance the rest of the world, while North America & Oceania and most of the Global South run chronic deficits. It is important to note, as Nievas and Piketty (2025) emphasize, that these historical imbalances reflect not just markets but also power relations and unequal exchange, a theme we revisit in **Chapter 5** .\n\n**Private wealth is rising while public wealth stagnates**\n\nOver the past thirty years, global wealth has risen faster than income, growing from just over 400% of world income in 1995 to more than 600% in 2025 ( **Figure 3.2** ). Yet this rise has been almost entirely concentrated in the private sector (see Bauluz, Brassac, et al. (2025)). **Figure 3.5** shows that private wealth increased from about 350% to over 500% of world income, while public wealth stagnated at around 80–90%. In some regions, public wealth even turned negative, meaning that governments’ liabilities exceeded their assets. The slowdown during the COVID-19 pandemic briefly interrupted the upward trajectory, but the long-run trend remains clear: wealth growth has accumulated in private hands.\n\n**Figure 3.5** shows that East Asia and North America & Oceania now report the highest levels of private wealth, each above 600% of income by 2025 (left-hand-side panel). North America & Oceania’s trajectory, however, has been volatile: strong growth up to 2007, a sharp fall during the global\n\n78","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p78.pdf"]},"response":{"errors":[],"extract_id":"extract_90c4afb0dd641b6b2ee76b7feafcb42c","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Chapter 3. Regional Wealth Inequality\n\n**Figure 3.3. Domestic capital remains the foundation of national wealth everywhere**\n\n**Net national wealth = Domestic capital + Net foreign assets**\n\n1995 2005 2015 2025 1995 2005 2015 2025 1995 2005 2015 2025\n\n0%\n\n100%\n\n200%\n\n0%\n\n250%\n\n500%\n\n750%\n\n0%\n\n250%\n\n500%\n\n750%\n\nWealth as a % of national income\n\nEast Asia\n\nEurope\n\nLatin America\n\nMENA\n\nNorth America & Oceania\n\nRussia & Central Asia\n\nSouth & Southeast Asia\n\nSub−Saharan Africa\n\n**Wealth composition across regions, 1995−2025**\n\n**Interpretation.** This figure shows that domestic capital makes up the bulk of net national wealth across all regions, while net foreign assets play only a minor role. Most regions exhibit steady increases in national wealth as a share of income since 1995, particularly East Asia and North America & Oceania. Sub−Saharan Africa and Latin America, by contrast, show limited growth and continue to hold negligible net foreign assets. **Notes.** Net national wealth = domestical capital + net foreign assets. **Sources and series:** Bauluz et al. (2025) and wir2026.wid.world/methodology.\n\ntransformation has been in East Asia. Since the 1970s, the region has become the largest creditor. Today, it holds about 12% of the world’s GDP. The Middle East & North Africa has also maintained a strong positive balance since the oil boom of the 1970s, while Europe has rebuilt modest surpluses in recent decades. By contrast, Latin America, Sub-Saharan Africa, and South & Southeast Asia remain consistent debtors, with foreigners owning more of their assets than their residents hold abroad. Foreign asset positions show the persistent global asymmetries of the financial system: today, East Asia, the Middle East & North Africa, and Europe finance the rest of the world, while North America & Oceania and most of the Global South run chronic deficits. It is important to note, as Nievas and Piketty (2025) emphasize, that these historical imbalances reflect not just markets but also power relations and unequal exchange, a theme we revisit in **Chapter 5** .\n\n**Private wealth is rising while public wealth stagnates**\n\nOver the past thirty years, global wealth has risen faster than income, growing from just over 400% of world income in 1995 to more than 600% in 2025 ( **Figure 3.2** ). Yet this rise has been almost entirely concentrated in the private sector (see Bauluz, Brassac, et al. (2025)). **Figure 3.5** shows that private wealth increased from about 350% to over 500% of world income, while public wealth stagnated at around 80–90%. In some regions, public wealth even turned negative, meaning that governments’ liabilities exceeded their assets. The slowdown during the COVID-19 pandemic briefly interrupted the upward trajectory, but the long-run trend remains clear: wealth growth has accumulated in private hands.\n\n**Figure 3.5** shows that East Asia and North America & Oceania now report the highest levels of private wealth, each above 600% of income by 2025 (left-hand-side panel). North America & Oceania’s trajectory, however, has been volatile: strong growth up to 2007, a sharp fall during the global\n\n78"],"full_content":"Chapter 3. Regional Wealth Inequality\n\n**Figure 3.3. Domestic capital remains the foundation of national wealth everywhere**\n\n**Net national wealth = Domestic capital + Net foreign assets**\n\n1995 2005 2015 2025 1995 2005 2015 2025 1995 2005 2015 2025\n\n0%\n\n100%\n\n200%\n\n0%\n\n250%\n\n500%\n\n750%\n\n0%\n\n250%\n\n500%\n\n750%\n\nWealth as a % of national income\n\nEast Asia\n\nEurope\n\nLatin America\n\nMENA\n\nNorth America & Oceania\n\nRussia & Central Asia\n\nSouth & Southeast Asia\n\nSub−Saharan Africa\n\n**Wealth composition across regions, 1995−2025**\n\n**Interpretation.** This figure shows that domestic capital makes up the bulk of net national wealth across all regions, while net foreign assets play only a minor role. Most regions exhibit steady increases in national wealth as a share of income since 1995, particularly East Asia and North America & Oceania. Sub−Saharan Africa and Latin America, by contrast, show limited growth and continue to hold negligible net foreign assets. **Notes.** Net national wealth = domestical capital + net foreign assets. **Sources and series:** Bauluz et al. (2025) and wir2026.wid.world/methodology.\n\ntransformation has been in East Asia. Since the 1970s, the region has become the largest creditor. Today, it holds about 12% of the world’s GDP. The Middle East & North Africa has also maintained a strong positive balance since the oil boom of the 1970s, while Europe has rebuilt modest surpluses in recent decades. By contrast, Latin America, Sub-Saharan Africa, and South & Southeast Asia remain consistent debtors, with foreigners owning more of their assets than their residents hold abroad. Foreign asset positions show the persistent global asymmetries of the financial system: today, East Asia, the Middle East & North Africa, and Europe finance the rest of the world, while North America & Oceania and most of the Global South run chronic deficits. It is important to note, as Nievas and Piketty (2025) emphasize, that these historical imbalances reflect not just markets but also power relations and unequal exchange, a theme we revisit in **Chapter 5** .\n\n**Private wealth is rising while public wealth stagnates**\n\nOver the past thirty years, global wealth has risen faster than income, growing from just over 400% of world income in 1995 to more than 600% in 2025 ( **Figure 3.2** ). Yet this rise has been almost entirely concentrated in the private sector (see Bauluz, Brassac, et al. (2025)). **Figure 3.5** shows that private wealth increased from about 350% to over 500% of world income, while public wealth stagnated at around 80–90%. In some regions, public wealth even turned negative, meaning that governments’ liabilities exceeded their assets. The slowdown during the COVID-19 pandemic briefly interrupted the upward trajectory, but the long-run trend remains clear: wealth growth has accumulated in private hands.\n\n**Figure 3.5** shows that East Asia and North America & Oceania now report the highest levels of private wealth, each above 600% of income by 2025 (left-hand-side panel). North America & Oceania’s trajectory, however, has been volatile: strong growth up to 2007, a sharp fall during the global\n\n78","publish_date":null,"title":"Chapter 3. Regional Wealth Inequality Figure3.3. Domesticcapitalremainsthefoundationofnational wealtheverywhere transformation has been in East Asia. Since the 1970s, the region has","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p78.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p78.pdf"},"request":{"example_id":"chart/World_Inequality_Report_2026_p78","product_type":"parse","source_file_path":"World_Inequality_Report_2026_p78.pdf"},"started_at":"2026-11-04T19:45:50.922645"} +{"completed_at":"2026-11-04T19:45:56.573526","latency_in_ms":3437,"output":{"example_id":"chart/World_Inequality_Report_2026_p79","grounded_pages":[],"layout_pages":[],"markdown":"Chapter 3. Regional Wealth Inequality\n\n**Figure 3.4. Since the 1970s, North America & Oceania has shifted into the largest net debtor**\n\n**As % of regional GDP As % of world GDP**\n\n1800 1830 1860 1890 1920 1950 1980 2010 1800 1830 1860 1890 1920 1950 1980 2010\n\n−20%\n\n−10%\n\n0%\n\n10%\n\n20%\n\n30%\n\n−100%\n\n−50%\n\n0%\n\n50%\n\nNet foreign assets, MER\n\nEast Asia\n\nEurope\n\nLatin America\n\nMENA\n\nNorth America & Oceania\n\nRussia & Central Asia\n\nSouth & Southeast Asia\n\nSub−Saharan Africa\n\nWorld\n\n**Net foreign wealth across regions, 1800−2025**\n\n**Interpretation.** Between 1800 and 1914, Europe accumulated a rising share of global foreign assets. By 1914, its net foreign wealth reached 71% of its own GDP. These assets largely vanished after World War I. Measured as a share of world GDP, Europe’s foreign wealth in 1914 was about 6 times larger than East Asia’s foreign wealth in 2025 (12%) and about 18 times larger than that of MENA (4%). During the 20th century, North America & Oceania emerged as a major foreign asset holder, peaking in 1950 at 8% of world GDP. Over the same period, East Asia transitioned to one of the world’s largest foreign asset holders. By 2025, its net foreign wealth stood at 12% of world GDP. In contrast, North America & Oceania held −18%, meaning other regions now hold more assets in North America & Oceania than it holds abroad. **Sources and series:** Bauluz et al. (2025), Nievas and Piketty (2025), and wir2026.wid.world/methodology.\n\nfinancial crisis, recovery in the 2010s, and a renewed decline after COVID-19. Europe has followed a steadier path, but its private wealth also fell more sharply than the global average after 2020. South & Southeast Asia has steadily built up private wealth, ranking fourth globally, while the Middle East & North Africa overtook Russia & Central Asia during the 2010s. By contrast, Latin America and Sub-Saharan Africa remain far below the world average, reflecting weaker asset accumulation relative to income. The picture of public wealth is even more striking (right-hand-side panel). East Asia stands out as the only region with substantial and rising public wealth, thanks to sustained public savings and significant state ownership of assets. The Middle East & North Africa, Sub-Saharan Africa, Latin America, and Russia & Central Asia maintain modestly positive levels, but these are well below those in East Asia. Europe and South & Southeast Asia hover close to zero, showing little capacity to build collective wealth. In North America & Oceania, public wealth is negative: governments owe more than they own, with rising public debts\n\noffsetting limited state assets. Overall, the world has become wealthier, but the ownership of this wealth has shifted toward individuals and corporations. Governments, by contrast, have seen their net position weaken, narrowing their fiscal capacity to invest in collective goods or respond to crises. This imbalance between expanding private fortunes and stagnant public reserves is now a defining feature of the global economy. This shift toward private balance sheets raises a second question: how are the flows of income divided between labor and capital? **Figure 3.6**\n\nanswers this.\n\n**Figure 3.6** tracks how the income flow is split between labor and capital since 1980. Globally, labor’s share falls from about 61% in 1980 to 53% in 2025, while capital’s share rises from 39% to 47%. This rebalancing toward capital income mirrors the wealth patterns in **Figure 3.5** (more wealth overall, and most of it private). ⁸\n\nAs for regional patterns, North America & Oceania and Europe retain comparatively higher labor shares (and lower capital\n\n79","pages":[{"line_numbers":[],"markdown":"Chapter 3. Regional Wealth Inequality\n\n**Figure 3.4. Since the 1970s, North America & Oceania has shifted into the largest net debtor**\n\n**As % of regional GDP As % of world GDP**\n\n1800 1830 1860 1890 1920 1950 1980 2010 1800 1830 1860 1890 1920 1950 1980 2010\n\n−20%\n\n−10%\n\n0%\n\n10%\n\n20%\n\n30%\n\n−100%\n\n−50%\n\n0%\n\n50%\n\nNet foreign assets, MER\n\nEast Asia\n\nEurope\n\nLatin America\n\nMENA\n\nNorth America & Oceania\n\nRussia & Central Asia\n\nSouth & Southeast Asia\n\nSub−Saharan Africa\n\nWorld\n\n**Net foreign wealth across regions, 1800−2025**\n\n**Interpretation.** Between 1800 and 1914, Europe accumulated a rising share of global foreign assets. By 1914, its net foreign wealth reached 71% of its own GDP. These assets largely vanished after World War I. Measured as a share of world GDP, Europe’s foreign wealth in 1914 was about 6 times larger than East Asia’s foreign wealth in 2025 (12%) and about 18 times larger than that of MENA (4%). During the 20th century, North America & Oceania emerged as a major foreign asset holder, peaking in 1950 at 8% of world GDP. Over the same period, East Asia transitioned to one of the world’s largest foreign asset holders. By 2025, its net foreign wealth stood at 12% of world GDP. In contrast, North America & Oceania held −18%, meaning other regions now hold more assets in North America & Oceania than it holds abroad. **Sources and series:** Bauluz et al. (2025), Nievas and Piketty (2025), and wir2026.wid.world/methodology.\n\nfinancial crisis, recovery in the 2010s, and a renewed decline after COVID-19. Europe has followed a steadier path, but its private wealth also fell more sharply than the global average after 2020. South & Southeast Asia has steadily built up private wealth, ranking fourth globally, while the Middle East & North Africa overtook Russia & Central Asia during the 2010s. By contrast, Latin America and Sub-Saharan Africa remain far below the world average, reflecting weaker asset accumulation relative to income. The picture of public wealth is even more striking (right-hand-side panel). East Asia stands out as the only region with substantial and rising public wealth, thanks to sustained public savings and significant state ownership of assets. The Middle East & North Africa, Sub-Saharan Africa, Latin America, and Russia & Central Asia maintain modestly positive levels, but these are well below those in East Asia. Europe and South & Southeast Asia hover close to zero, showing little capacity to build collective wealth. In North America & Oceania, public wealth is negative: governments owe more than they own, with rising public debts\n\noffsetting limited state assets. Overall, the world has become wealthier, but the ownership of this wealth has shifted toward individuals and corporations. Governments, by contrast, have seen their net position weaken, narrowing their fiscal capacity to invest in collective goods or respond to crises. This imbalance between expanding private fortunes and stagnant public reserves is now a defining feature of the global economy. This shift toward private balance sheets raises a second question: how are the flows of income divided between labor and capital? **Figure 3.6**\n\nanswers this.\n\n**Figure 3.6** tracks how the income flow is split between labor and capital since 1980. Globally, labor’s share falls from about 61% in 1980 to 53% in 2025, while capital’s share rises from 39% to 47%. This rebalancing toward capital income mirrors the wealth patterns in **Figure 3.5** (more wealth overall, and most of it private). ⁸\n\nAs for regional patterns, North America & Oceania and Europe retain comparatively higher labor shares (and lower capital\n\n79","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Chapter 3. Regional Wealth Inequality\n\n**Figure 3.4. Since the 1970s, North America & Oceania has shifted into the largest net debtor**\n\n**As % of regional GDP As % of world GDP**\n\n1800 1830 1860 1890 1920 1950 1980 2010 1800 1830 1860 1890 1920 1950 1980 2010\n\n−20%\n\n−10%\n\n0%\n\n10%\n\n20%\n\n30%\n\n−100%\n\n−50%\n\n0%\n\n50%\n\nNet foreign assets, MER\n\nEast Asia\n\nEurope\n\nLatin America\n\nMENA\n\nNorth America & Oceania\n\nRussia & Central Asia\n\nSouth & Southeast Asia\n\nSub−Saharan Africa\n\nWorld\n\n**Net foreign wealth across regions, 1800−2025**\n\n**Interpretation.** Between 1800 and 1914, Europe accumulated a rising share of global foreign assets. By 1914, its net foreign wealth reached 71% of its own GDP. These assets largely vanished after World War I. Measured as a share of world GDP, Europe’s foreign wealth in 1914 was about 6 times larger than East Asia’s foreign wealth in 2025 (12%) and about 18 times larger than that of MENA (4%). During the 20th century, North America & Oceania emerged as a major foreign asset holder, peaking in 1950 at 8% of world GDP. Over the same period, East Asia transitioned to one of the world’s largest foreign asset holders. By 2025, its net foreign wealth stood at 12% of world GDP. In contrast, North America & Oceania held −18%, meaning other regions now hold more assets in North America & Oceania than it holds abroad. **Sources and series:** Bauluz et al. (2025), Nievas and Piketty (2025), and wir2026.wid.world/methodology.\n\nfinancial crisis, recovery in the 2010s, and a renewed decline after COVID-19. Europe has followed a steadier path, but its private wealth also fell more sharply than the global average after 2020. South & Southeast Asia has steadily built up private wealth, ranking fourth globally, while the Middle East & North Africa overtook Russia & Central Asia during the 2010s. By contrast, Latin America and Sub-Saharan Africa remain far below the world average, reflecting weaker asset accumulation relative to income. The picture of public wealth is even more striking (right-hand-side panel). East Asia stands out as the only region with substantial and rising public wealth, thanks to sustained public savings and significant state ownership of assets. The Middle East & North Africa, Sub-Saharan Africa, Latin America, and Russia & Central Asia maintain modestly positive levels, but these are well below those in East Asia. Europe and South & Southeast Asia hover close to zero, showing little capacity to build collective wealth. In North America & Oceania, public wealth is negative: governments owe more than they own, with rising public debts\n\noffsetting limited state assets. Overall, the world has become wealthier, but the ownership of this wealth has shifted toward individuals and corporations. Governments, by contrast, have seen their net position weaken, narrowing their fiscal capacity to invest in collective goods or respond to crises. This imbalance between expanding private fortunes and stagnant public reserves is now a defining feature of the global economy. This shift toward private balance sheets raises a second question: how are the flows of income divided between labor and capital? **Figure 3.6**\n\nanswers this.\n\n**Figure 3.6** tracks how the income flow is split between labor and capital since 1980. Globally, labor’s share falls from about 61% in 1980 to 53% in 2025, while capital’s share rises from 39% to 47%. This rebalancing toward capital income mirrors the wealth patterns in **Figure 3.5** (more wealth overall, and most of it private). ⁸\n\nAs for regional patterns, North America & Oceania and Europe retain comparatively higher labor shares (and lower capital\n\n79","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p79.pdf"]},"response":{"errors":[],"extract_id":"extract_40ae1cdda45c4718e7f53ad9c29a1b48","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Chapter 3. Regional Wealth Inequality\n\n**Figure 3.4. Since the 1970s, North America & Oceania has shifted into the largest net debtor**\n\n**As % of regional GDP As % of world GDP**\n\n1800 1830 1860 1890 1920 1950 1980 2010 1800 1830 1860 1890 1920 1950 1980 2010\n\n−20%\n\n−10%\n\n0%\n\n10%\n\n20%\n\n30%\n\n−100%\n\n−50%\n\n0%\n\n50%\n\nNet foreign assets, MER\n\nEast Asia\n\nEurope\n\nLatin America\n\nMENA\n\nNorth America & Oceania\n\nRussia & Central Asia\n\nSouth & Southeast Asia\n\nSub−Saharan Africa\n\nWorld\n\n**Net foreign wealth across regions, 1800−2025**\n\n**Interpretation.** Between 1800 and 1914, Europe accumulated a rising share of global foreign assets. By 1914, its net foreign wealth reached 71% of its own GDP. These assets largely vanished after World War I. Measured as a share of world GDP, Europe’s foreign wealth in 1914 was about 6 times larger than East Asia’s foreign wealth in 2025 (12%) and about 18 times larger than that of MENA (4%). During the 20th century, North America & Oceania emerged as a major foreign asset holder, peaking in 1950 at 8% of world GDP. Over the same period, East Asia transitioned to one of the world’s largest foreign asset holders. By 2025, its net foreign wealth stood at 12% of world GDP. In contrast, North America & Oceania held −18%, meaning other regions now hold more assets in North America & Oceania than it holds abroad. **Sources and series:** Bauluz et al. (2025), Nievas and Piketty (2025), and wir2026.wid.world/methodology.\n\nfinancial crisis, recovery in the 2010s, and a renewed decline after COVID-19. Europe has followed a steadier path, but its private wealth also fell more sharply than the global average after 2020. South & Southeast Asia has steadily built up private wealth, ranking fourth globally, while the Middle East & North Africa overtook Russia & Central Asia during the 2010s. By contrast, Latin America and Sub-Saharan Africa remain far below the world average, reflecting weaker asset accumulation relative to income. The picture of public wealth is even more striking (right-hand-side panel). East Asia stands out as the only region with substantial and rising public wealth, thanks to sustained public savings and significant state ownership of assets. The Middle East & North Africa, Sub-Saharan Africa, Latin America, and Russia & Central Asia maintain modestly positive levels, but these are well below those in East Asia. Europe and South & Southeast Asia hover close to zero, showing little capacity to build collective wealth. In North America & Oceania, public wealth is negative: governments owe more than they own, with rising public debts\n\noffsetting limited state assets. Overall, the world has become wealthier, but the ownership of this wealth has shifted toward individuals and corporations. Governments, by contrast, have seen their net position weaken, narrowing their fiscal capacity to invest in collective goods or respond to crises. This imbalance between expanding private fortunes and stagnant public reserves is now a defining feature of the global economy. This shift toward private balance sheets raises a second question: how are the flows of income divided between labor and capital? **Figure 3.6**\n\nanswers this.\n\n**Figure 3.6** tracks how the income flow is split between labor and capital since 1980. Globally, labor’s share falls from about 61% in 1980 to 53% in 2025, while capital’s share rises from 39% to 47%. This rebalancing toward capital income mirrors the wealth patterns in **Figure 3.5** (more wealth overall, and most of it private). ⁸\n\nAs for regional patterns, North America & Oceania and Europe retain comparatively higher labor shares (and lower capital\n\n79"],"full_content":"Chapter 3. Regional Wealth Inequality\n\n**Figure 3.4. Since the 1970s, North America & Oceania has shifted into the largest net debtor**\n\n**As % of regional GDP As % of world GDP**\n\n1800 1830 1860 1890 1920 1950 1980 2010 1800 1830 1860 1890 1920 1950 1980 2010\n\n−20%\n\n−10%\n\n0%\n\n10%\n\n20%\n\n30%\n\n−100%\n\n−50%\n\n0%\n\n50%\n\nNet foreign assets, MER\n\nEast Asia\n\nEurope\n\nLatin America\n\nMENA\n\nNorth America & Oceania\n\nRussia & Central Asia\n\nSouth & Southeast Asia\n\nSub−Saharan Africa\n\nWorld\n\n**Net foreign wealth across regions, 1800−2025**\n\n**Interpretation.** Between 1800 and 1914, Europe accumulated a rising share of global foreign assets. By 1914, its net foreign wealth reached 71% of its own GDP. These assets largely vanished after World War I. Measured as a share of world GDP, Europe’s foreign wealth in 1914 was about 6 times larger than East Asia’s foreign wealth in 2025 (12%) and about 18 times larger than that of MENA (4%). During the 20th century, North America & Oceania emerged as a major foreign asset holder, peaking in 1950 at 8% of world GDP. Over the same period, East Asia transitioned to one of the world’s largest foreign asset holders. By 2025, its net foreign wealth stood at 12% of world GDP. In contrast, North America & Oceania held −18%, meaning other regions now hold more assets in North America & Oceania than it holds abroad. **Sources and series:** Bauluz et al. (2025), Nievas and Piketty (2025), and wir2026.wid.world/methodology.\n\nfinancial crisis, recovery in the 2010s, and a renewed decline after COVID-19. Europe has followed a steadier path, but its private wealth also fell more sharply than the global average after 2020. South & Southeast Asia has steadily built up private wealth, ranking fourth globally, while the Middle East & North Africa overtook Russia & Central Asia during the 2010s. By contrast, Latin America and Sub-Saharan Africa remain far below the world average, reflecting weaker asset accumulation relative to income. The picture of public wealth is even more striking (right-hand-side panel). East Asia stands out as the only region with substantial and rising public wealth, thanks to sustained public savings and significant state ownership of assets. The Middle East & North Africa, Sub-Saharan Africa, Latin America, and Russia & Central Asia maintain modestly positive levels, but these are well below those in East Asia. Europe and South & Southeast Asia hover close to zero, showing little capacity to build collective wealth. In North America & Oceania, public wealth is negative: governments owe more than they own, with rising public debts\n\noffsetting limited state assets. Overall, the world has become wealthier, but the ownership of this wealth has shifted toward individuals and corporations. Governments, by contrast, have seen their net position weaken, narrowing their fiscal capacity to invest in collective goods or respond to crises. This imbalance between expanding private fortunes and stagnant public reserves is now a defining feature of the global economy. This shift toward private balance sheets raises a second question: how are the flows of income divided between labor and capital? **Figure 3.6**\n\nanswers this.\n\n**Figure 3.6** tracks how the income flow is split between labor and capital since 1980. Globally, labor’s share falls from about 61% in 1980 to 53% in 2025, while capital’s share rises from 39% to 47%. This rebalancing toward capital income mirrors the wealth patterns in **Figure 3.5** (more wealth overall, and most of it private). ⁸\n\nAs for regional patterns, North America & Oceania and Europe retain comparatively higher labor shares (and lower capital\n\n79","publish_date":null,"title":"Chapter 3. Regional Wealth Inequality Figure3.4. Sincethe1970s,NorthAmerica&Oceaniahas shiftedintothelargestnetdebtor financial crisis, recovery in the 2010s, and offsetting limited state assets. a","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p79.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p79.pdf"},"request":{"example_id":"chart/World_Inequality_Report_2026_p79","product_type":"parse","source_file_path":"World_Inequality_Report_2026_p79.pdf"},"started_at":"2026-11-04T19:45:53.136229"} +{"completed_at":"2026-11-04T19:45:59.438944","latency_in_ms":5001,"output":{"example_id":"chart/World_Inequality_Report_2026_p81","grounded_pages":[],"layout_pages":[],"markdown":"Chapter 3. Regional Wealth Inequality\n\n**Figure 3.6. The rising capital share in global income**\n\n**Capital share of income Labor share of income**\n\n1980 1990 2000 2010 2020 1980 1990 2000 2010 2020\n\n30%\n\n40%\n\n50%\n\n60%\n\n70%\n\n30%\n\n40%\n\n50%\n\n60%\n\n70%\n\nShare of total factor−price GDP\n\nEast Asia\n\nEurope\n\nLatin America\n\nMENA\n\nNorth America & Oceania\n\nRussia & Central Asia\n\nSouth & Southeast Asia\n\nSub−Saharan Africa\n\nWorld\n\n**Capital and income as a share of total factor−price GDP, 1980–2025**\n\n**Interpretation.** This figure shows trends in the composition of income between labor and capital. At the global level, the share of income going to labor declined from 61% in 1980 to 53% in 2025. Meanwhile, the capital share increased from 39% to 47% over the same period. This shift reflects a combination of rising returns to capital, growing depreciation (CFC), and stagnating labor compensation in many regions. Capital shares are substantially larger in poorer regions than in richer ones. **Sources and series:** Dietrich et al. (2025).\n\nSub-Saharan Africa is concentrated at the very low end, with minimal representation beyond the lower deciles. Europe and North America & Oceania are overrepresented at the top tail and dominate the highest wealth brackets. Compared with the income distribution shown earlier in **Figure 2.5** , the contrast is clear: wealth inequality is larger than income inequality. Regions that account for a substantial share of middle-income earners, such as East Asia and parts of South & Southeast Asia, remain underrepresented at the very top of the wealth distribution, while Europe and North America & Oceania are disproportionately present there. Put differently, location still shapes an individual’s chances of reaching the top of the global wealth ladder.\n\n**Figure 3.8** disaggregates household net wealth in 2025 into four groups: the bottom 50%, the middle 40%, the next 9%, and the top 1%, for each world region. The picture is stark: in every region, the top 10% owns the majority of wealth, while the bottom half owns almost none. Across regions, the bottom 50% holds between 1% and 5% of total wealth: just 1% in North America\n\n& Oceania, Sub-Saharan Africa, and the Middle East & North Africa; 3% in Europe, Latin America, and Russia & Central Asia; and 5% in South & Southeast Asia and East Asia. Put plainly, in every region, half of the population owns no more than 5% of that region’s wealth. By contrast, the top 10% controls 60–74% of wealth, about 60% in Europe; 65–70% in South & Southeast Asia, East Asia, Latin America, and North America & Oceania; 70% in Sub-Saharan Africa; 73% in the Middle East & North Africa; and 74% in Russia & Central Asia. In every region, one-tenth of the population owns at least 60% of all wealth. Disaggregating the top 10% shows how concentrated the very top is. The top 1% alone holds a quarter of all wealth in Europe (25%), around a third in North America & Oceania (34%), South & Southeast Asia (35%), Sub-Saharan Africa (36%), Latin America (36%), the Middle East & North Africa (37%), and an excessively high 46% in Russia & Central Asia. In several regions, North America & Oceania, Latin America, the Middle East & North Africa, Sub-Saharan Africa, and Russia & Central Asia, the top\n\n81","pages":[{"line_numbers":[],"markdown":"Chapter 3. Regional Wealth Inequality\n\n**Figure 3.6. The rising capital share in global income**\n\n**Capital share of income Labor share of income**\n\n1980 1990 2000 2010 2020 1980 1990 2000 2010 2020\n\n30%\n\n40%\n\n50%\n\n60%\n\n70%\n\n30%\n\n40%\n\n50%\n\n60%\n\n70%\n\nShare of total factor−price GDP\n\nEast Asia\n\nEurope\n\nLatin America\n\nMENA\n\nNorth America & Oceania\n\nRussia & Central Asia\n\nSouth & Southeast Asia\n\nSub−Saharan Africa\n\nWorld\n\n**Capital and income as a share of total factor−price GDP, 1980–2025**\n\n**Interpretation.** This figure shows trends in the composition of income between labor and capital. At the global level, the share of income going to labor declined from 61% in 1980 to 53% in 2025. Meanwhile, the capital share increased from 39% to 47% over the same period. This shift reflects a combination of rising returns to capital, growing depreciation (CFC), and stagnating labor compensation in many regions. Capital shares are substantially larger in poorer regions than in richer ones. **Sources and series:** Dietrich et al. (2025).\n\nSub-Saharan Africa is concentrated at the very low end, with minimal representation beyond the lower deciles. Europe and North America & Oceania are overrepresented at the top tail and dominate the highest wealth brackets. Compared with the income distribution shown earlier in **Figure 2.5** , the contrast is clear: wealth inequality is larger than income inequality. Regions that account for a substantial share of middle-income earners, such as East Asia and parts of South & Southeast Asia, remain underrepresented at the very top of the wealth distribution, while Europe and North America & Oceania are disproportionately present there. Put differently, location still shapes an individual’s chances of reaching the top of the global wealth ladder.\n\n**Figure 3.8** disaggregates household net wealth in 2025 into four groups: the bottom 50%, the middle 40%, the next 9%, and the top 1%, for each world region. The picture is stark: in every region, the top 10% owns the majority of wealth, while the bottom half owns almost none. Across regions, the bottom 50% holds between 1% and 5% of total wealth: just 1% in North America\n\n& Oceania, Sub-Saharan Africa, and the Middle East & North Africa; 3% in Europe, Latin America, and Russia & Central Asia; and 5% in South & Southeast Asia and East Asia. Put plainly, in every region, half of the population owns no more than 5% of that region’s wealth. By contrast, the top 10% controls 60–74% of wealth, about 60% in Europe; 65–70% in South & Southeast Asia, East Asia, Latin America, and North America & Oceania; 70% in Sub-Saharan Africa; 73% in the Middle East & North Africa; and 74% in Russia & Central Asia. In every region, one-tenth of the population owns at least 60% of all wealth. Disaggregating the top 10% shows how concentrated the very top is. The top 1% alone holds a quarter of all wealth in Europe (25%), around a third in North America & Oceania (34%), South & Southeast Asia (35%), Sub-Saharan Africa (36%), Latin America (36%), the Middle East & North Africa (37%), and an excessively high 46% in Russia & Central Asia. In several regions, North America & Oceania, Latin America, the Middle East & North Africa, Sub-Saharan Africa, and Russia & Central Asia, the top\n\n81","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Chapter 3. Regional Wealth Inequality\n\n**Figure 3.6. The rising capital share in global income**\n\n**Capital share of income Labor share of income**\n\n1980 1990 2000 2010 2020 1980 1990 2000 2010 2020\n\n30%\n\n40%\n\n50%\n\n60%\n\n70%\n\n30%\n\n40%\n\n50%\n\n60%\n\n70%\n\nShare of total factor−price GDP\n\nEast Asia\n\nEurope\n\nLatin America\n\nMENA\n\nNorth America & Oceania\n\nRussia & Central Asia\n\nSouth & Southeast Asia\n\nSub−Saharan Africa\n\nWorld\n\n**Capital and income as a share of total factor−price GDP, 1980–2025**\n\n**Interpretation.** This figure shows trends in the composition of income between labor and capital. At the global level, the share of income going to labor declined from 61% in 1980 to 53% in 2025. Meanwhile, the capital share increased from 39% to 47% over the same period. This shift reflects a combination of rising returns to capital, growing depreciation (CFC), and stagnating labor compensation in many regions. Capital shares are substantially larger in poorer regions than in richer ones. **Sources and series:** Dietrich et al. (2025).\n\nSub-Saharan Africa is concentrated at the very low end, with minimal representation beyond the lower deciles. Europe and North America & Oceania are overrepresented at the top tail and dominate the highest wealth brackets. Compared with the income distribution shown earlier in **Figure 2.5** , the contrast is clear: wealth inequality is larger than income inequality. Regions that account for a substantial share of middle-income earners, such as East Asia and parts of South & Southeast Asia, remain underrepresented at the very top of the wealth distribution, while Europe and North America & Oceania are disproportionately present there. Put differently, location still shapes an individual’s chances of reaching the top of the global wealth ladder.\n\n**Figure 3.8** disaggregates household net wealth in 2025 into four groups: the bottom 50%, the middle 40%, the next 9%, and the top 1%, for each world region. The picture is stark: in every region, the top 10% owns the majority of wealth, while the bottom half owns almost none. Across regions, the bottom 50% holds between 1% and 5% of total wealth: just 1% in North America\n\n& Oceania, Sub-Saharan Africa, and the Middle East & North Africa; 3% in Europe, Latin America, and Russia & Central Asia; and 5% in South & Southeast Asia and East Asia. Put plainly, in every region, half of the population owns no more than 5% of that region’s wealth. By contrast, the top 10% controls 60–74% of wealth, about 60% in Europe; 65–70% in South & Southeast Asia, East Asia, Latin America, and North America & Oceania; 70% in Sub-Saharan Africa; 73% in the Middle East & North Africa; and 74% in Russia & Central Asia. In every region, one-tenth of the population owns at least 60% of all wealth. Disaggregating the top 10% shows how concentrated the very top is. The top 1% alone holds a quarter of all wealth in Europe (25%), around a third in North America & Oceania (34%), South & Southeast Asia (35%), Sub-Saharan Africa (36%), Latin America (36%), the Middle East & North Africa (37%), and an excessively high 46% in Russia & Central Asia. In several regions, North America & Oceania, Latin America, the Middle East & North Africa, Sub-Saharan Africa, and Russia & Central Asia, the top\n\n81","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p81.pdf"]},"response":{"errors":[],"extract_id":"extract_53d3adba3d7a620cfbf86f0ab8cdd751","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Chapter 3. Regional Wealth Inequality\n\n**Figure 3.6. The rising capital share in global income**\n\n**Capital share of income Labor share of income**\n\n1980 1990 2000 2010 2020 1980 1990 2000 2010 2020\n\n30%\n\n40%\n\n50%\n\n60%\n\n70%\n\n30%\n\n40%\n\n50%\n\n60%\n\n70%\n\nShare of total factor−price GDP\n\nEast Asia\n\nEurope\n\nLatin America\n\nMENA\n\nNorth America & Oceania\n\nRussia & Central Asia\n\nSouth & Southeast Asia\n\nSub−Saharan Africa\n\nWorld\n\n**Capital and income as a share of total factor−price GDP, 1980–2025**\n\n**Interpretation.** This figure shows trends in the composition of income between labor and capital. At the global level, the share of income going to labor declined from 61% in 1980 to 53% in 2025. Meanwhile, the capital share increased from 39% to 47% over the same period. This shift reflects a combination of rising returns to capital, growing depreciation (CFC), and stagnating labor compensation in many regions. Capital shares are substantially larger in poorer regions than in richer ones. **Sources and series:** Dietrich et al. (2025).\n\nSub-Saharan Africa is concentrated at the very low end, with minimal representation beyond the lower deciles. Europe and North America & Oceania are overrepresented at the top tail and dominate the highest wealth brackets. Compared with the income distribution shown earlier in **Figure 2.5** , the contrast is clear: wealth inequality is larger than income inequality. Regions that account for a substantial share of middle-income earners, such as East Asia and parts of South & Southeast Asia, remain underrepresented at the very top of the wealth distribution, while Europe and North America & Oceania are disproportionately present there. Put differently, location still shapes an individual’s chances of reaching the top of the global wealth ladder.\n\n**Figure 3.8** disaggregates household net wealth in 2025 into four groups: the bottom 50%, the middle 40%, the next 9%, and the top 1%, for each world region. The picture is stark: in every region, the top 10% owns the majority of wealth, while the bottom half owns almost none. Across regions, the bottom 50% holds between 1% and 5% of total wealth: just 1% in North America\n\n& Oceania, Sub-Saharan Africa, and the Middle East & North Africa; 3% in Europe, Latin America, and Russia & Central Asia; and 5% in South & Southeast Asia and East Asia. Put plainly, in every region, half of the population owns no more than 5% of that region’s wealth. By contrast, the top 10% controls 60–74% of wealth, about 60% in Europe; 65–70% in South & Southeast Asia, East Asia, Latin America, and North America & Oceania; 70% in Sub-Saharan Africa; 73% in the Middle East & North Africa; and 74% in Russia & Central Asia. In every region, one-tenth of the population owns at least 60% of all wealth. Disaggregating the top 10% shows how concentrated the very top is. The top 1% alone holds a quarter of all wealth in Europe (25%), around a third in North America & Oceania (34%), South & Southeast Asia (35%), Sub-Saharan Africa (36%), Latin America (36%), the Middle East & North Africa (37%), and an excessively high 46% in Russia & Central Asia. In several regions, North America & Oceania, Latin America, the Middle East & North Africa, Sub-Saharan Africa, and Russia & Central Asia, the top\n\n81"],"full_content":"Chapter 3. Regional Wealth Inequality\n\n**Figure 3.6. The rising capital share in global income**\n\n**Capital share of income Labor share of income**\n\n1980 1990 2000 2010 2020 1980 1990 2000 2010 2020\n\n30%\n\n40%\n\n50%\n\n60%\n\n70%\n\n30%\n\n40%\n\n50%\n\n60%\n\n70%\n\nShare of total factor−price GDP\n\nEast Asia\n\nEurope\n\nLatin America\n\nMENA\n\nNorth America & Oceania\n\nRussia & Central Asia\n\nSouth & Southeast Asia\n\nSub−Saharan Africa\n\nWorld\n\n**Capital and income as a share of total factor−price GDP, 1980–2025**\n\n**Interpretation.** This figure shows trends in the composition of income between labor and capital. At the global level, the share of income going to labor declined from 61% in 1980 to 53% in 2025. Meanwhile, the capital share increased from 39% to 47% over the same period. This shift reflects a combination of rising returns to capital, growing depreciation (CFC), and stagnating labor compensation in many regions. Capital shares are substantially larger in poorer regions than in richer ones. **Sources and series:** Dietrich et al. (2025).\n\nSub-Saharan Africa is concentrated at the very low end, with minimal representation beyond the lower deciles. Europe and North America & Oceania are overrepresented at the top tail and dominate the highest wealth brackets. Compared with the income distribution shown earlier in **Figure 2.5** , the contrast is clear: wealth inequality is larger than income inequality. Regions that account for a substantial share of middle-income earners, such as East Asia and parts of South & Southeast Asia, remain underrepresented at the very top of the wealth distribution, while Europe and North America & Oceania are disproportionately present there. Put differently, location still shapes an individual’s chances of reaching the top of the global wealth ladder.\n\n**Figure 3.8** disaggregates household net wealth in 2025 into four groups: the bottom 50%, the middle 40%, the next 9%, and the top 1%, for each world region. The picture is stark: in every region, the top 10% owns the majority of wealth, while the bottom half owns almost none. Across regions, the bottom 50% holds between 1% and 5% of total wealth: just 1% in North America\n\n& Oceania, Sub-Saharan Africa, and the Middle East & North Africa; 3% in Europe, Latin America, and Russia & Central Asia; and 5% in South & Southeast Asia and East Asia. Put plainly, in every region, half of the population owns no more than 5% of that region’s wealth. By contrast, the top 10% controls 60–74% of wealth, about 60% in Europe; 65–70% in South & Southeast Asia, East Asia, Latin America, and North America & Oceania; 70% in Sub-Saharan Africa; 73% in the Middle East & North Africa; and 74% in Russia & Central Asia. In every region, one-tenth of the population owns at least 60% of all wealth. Disaggregating the top 10% shows how concentrated the very top is. The top 1% alone holds a quarter of all wealth in Europe (25%), around a third in North America & Oceania (34%), South & Southeast Asia (35%), Sub-Saharan Africa (36%), Latin America (36%), the Middle East & North Africa (37%), and an excessively high 46% in Russia & Central Asia. In several regions, North America & Oceania, Latin America, the Middle East & North Africa, Sub-Saharan Africa, and Russia & Central Asia, the top\n\n81","publish_date":null,"title":"Chapter 3. Regional Wealth Inequality Figure3.6. Therisingcapitalshareinglobalincome & Oceania, Sub-Saharan Africa, and the Sub-Saharan Africa is concentrated at the Middle","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p81.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p81.pdf"},"request":{"example_id":"chart/World_Inequality_Report_2026_p81","product_type":"parse","source_file_path":"World_Inequality_Report_2026_p81.pdf"},"started_at":"2026-11-04T19:45:54.437658"} +{"completed_at":"2026-11-04T19:45:57.585896","latency_in_ms":1084,"output":{"example_id":"chart/World_Inequality_Report_2026_p83","grounded_pages":[],"layout_pages":[],"markdown":"Chapter 3. Regional Wealth Inequality\n\n**Figure 3.8. Extreme wealth inequality is high in all regions**\n\n3%\n\n37%\n\n35%\n\n25%\n\n5%\n\n30%\n\n30%\n\n35%\n\n1%\n\n30%\n\n35%\n\n34%\n\n3%\n\n28%\n\n33%\n\n36%\n\n5%\n\n26%\n\n39%\n\n30%\n\n1%\n\n29%\n\n34%\n\n36%\n\n1%\n\n26%\n\n36%\n\n37%\n\n3%\n\n23%\n\n28%\n\n46%\n\n0%\n\n10%\n\n20%\n\n30%\n\n40%\n\n50%\n\n60%\n\n70%\n\n80%\n\n90%\n\n100%\n\nEurope East Asia\n\nNorth America & Oceania\n\nSouth & Southeast\n\nAsia\n\nLatin\n\nAmerica\n\nSub−\n\nSaharan\n\nAfrica\n\nMiddle East\n\n& North\n\nAfrica\n\nRussia & Central\n\nAsia\n\nShare of personal wealth (%)\n\nTop 1% Next 9% Middle 40% Bottom 50%\n\n**Bottom 50%, middle 40% and top 10% (top 1% and next 9%)**\n\n**wealth shares across the world, 2025**\n\n**Interpretation.** In Latin America, the top 1% captures 36% of national wealth, and the next 9% an additional 33%. Together, the top 10% holds 69%, compared to 60% in Europe. Net personal wealth is equal to the sum of financial assets (e.g. equity or bonds) and non−financial assets (e.g., housing or land) owned by individuals, net of their debts. **Sources and series:** Arias−Osorio et al. (2025) and wir2026.wid.world/methodology.\n\n**Country-by-country patterns of wealth concentration**\n\nHaving mapped the global distribution of wealth by region ( **Figure 3.7** – **Figure 3.9** ), we now zoom in on how wealth is split within countries. The four maps in\n\n**Figure 3.10** – **Figure 3.14** mirror the income analysis in **Chapter 2** ( **Figure 2.9** – **Figure 2.13** ), but for net household wealth. ⁹ The picture is consistently starker than for income.\n\n**Figure 3.10** shows the share of household wealth owned by the bottom 50%. Everywhere in the world, the bottom half owns only a sliver of national wealth, at most around 14% and, in many places, just 1%. Large parts of Latin America, Sub-Saharan Africa, and the Middle East & North Africa fall in the lowest band on the map (below roughly 1–3.8%). The same pattern appears in parts of Central Europe and in the United States. By contrast, several Western European countries, Australia and New Zealand, and large Asian economies such as China and India sit one or two bands higher: their bottom halves still own little, but noticeably more than in the most unequal settings.\n\nThe shares of the middle 40% are illustrated in **Figure 3.11** . The highest bands cluster in Europe and in Oceania, where the middle 40% command a sizable portion of national wealth (around two-fifths, broadly in line with their population weight). The United States and Canada stand out with a much thinner middle share, closer to Latin American and African patterns. In Asia, the middle 40% also capture small shares in China and India.\n\n**Figure 3.12** highlights that the top decile owns the majority of wealth in most countries. The darkest colors cover Latin America, southern Africa, the Middle East, Russia, China, and India, as well as the United States, where the top 10% typically control well over 60% of household wealth. Europe and Oceania are lighter on the map: concentration is still high by any standard, but is less than in other parts of the world. Strikingly, nowhere does the top 10% own less than about 45% of total wealth; in some countries, their share approaches 86%, an extraordinary concentration for a group that represents one person in ten. Focusing on the very top sharpens the contrast ( **Figure 3.13** ). The top 1% takes\n\n83","pages":[{"line_numbers":[],"markdown":"Chapter 3. Regional Wealth Inequality\n\n**Figure 3.8. Extreme wealth inequality is high in all regions**\n\n3%\n\n37%\n\n35%\n\n25%\n\n5%\n\n30%\n\n30%\n\n35%\n\n1%\n\n30%\n\n35%\n\n34%\n\n3%\n\n28%\n\n33%\n\n36%\n\n5%\n\n26%\n\n39%\n\n30%\n\n1%\n\n29%\n\n34%\n\n36%\n\n1%\n\n26%\n\n36%\n\n37%\n\n3%\n\n23%\n\n28%\n\n46%\n\n0%\n\n10%\n\n20%\n\n30%\n\n40%\n\n50%\n\n60%\n\n70%\n\n80%\n\n90%\n\n100%\n\nEurope East Asia\n\nNorth America & Oceania\n\nSouth & Southeast\n\nAsia\n\nLatin\n\nAmerica\n\nSub−\n\nSaharan\n\nAfrica\n\nMiddle East\n\n& North\n\nAfrica\n\nRussia & Central\n\nAsia\n\nShare of personal wealth (%)\n\nTop 1% Next 9% Middle 40% Bottom 50%\n\n**Bottom 50%, middle 40% and top 10% (top 1% and next 9%)**\n\n**wealth shares across the world, 2025**\n\n**Interpretation.** In Latin America, the top 1% captures 36% of national wealth, and the next 9% an additional 33%. Together, the top 10% holds 69%, compared to 60% in Europe. Net personal wealth is equal to the sum of financial assets (e.g. equity or bonds) and non−financial assets (e.g., housing or land) owned by individuals, net of their debts. **Sources and series:** Arias−Osorio et al. (2025) and wir2026.wid.world/methodology.\n\n**Country-by-country patterns of wealth concentration**\n\nHaving mapped the global distribution of wealth by region ( **Figure 3.7** – **Figure 3.9** ), we now zoom in on how wealth is split within countries. The four maps in\n\n**Figure 3.10** – **Figure 3.14** mirror the income analysis in **Chapter 2** ( **Figure 2.9** – **Figure 2.13** ), but for net household wealth. ⁹ The picture is consistently starker than for income.\n\n**Figure 3.10** shows the share of household wealth owned by the bottom 50%. Everywhere in the world, the bottom half owns only a sliver of national wealth, at most around 14% and, in many places, just 1%. Large parts of Latin America, Sub-Saharan Africa, and the Middle East & North Africa fall in the lowest band on the map (below roughly 1–3.8%). The same pattern appears in parts of Central Europe and in the United States. By contrast, several Western European countries, Australia and New Zealand, and large Asian economies such as China and India sit one or two bands higher: their bottom halves still own little, but noticeably more than in the most unequal settings.\n\nThe shares of the middle 40% are illustrated in **Figure 3.11** . The highest bands cluster in Europe and in Oceania, where the middle 40% command a sizable portion of national wealth (around two-fifths, broadly in line with their population weight). The United States and Canada stand out with a much thinner middle share, closer to Latin American and African patterns. In Asia, the middle 40% also capture small shares in China and India.\n\n**Figure 3.12** highlights that the top decile owns the majority of wealth in most countries. The darkest colors cover Latin America, southern Africa, the Middle East, Russia, China, and India, as well as the United States, where the top 10% typically control well over 60% of household wealth. Europe and Oceania are lighter on the map: concentration is still high by any standard, but is less than in other parts of the world. Strikingly, nowhere does the top 10% own less than about 45% of total wealth; in some countries, their share approaches 86%, an extraordinary concentration for a group that represents one person in ten. Focusing on the very top sharpens the contrast ( **Figure 3.13** ). The top 1% takes\n\n83","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Chapter 3. Regional Wealth Inequality\n\n**Figure 3.8. Extreme wealth inequality is high in all regions**\n\n3%\n\n37%\n\n35%\n\n25%\n\n5%\n\n30%\n\n30%\n\n35%\n\n1%\n\n30%\n\n35%\n\n34%\n\n3%\n\n28%\n\n33%\n\n36%\n\n5%\n\n26%\n\n39%\n\n30%\n\n1%\n\n29%\n\n34%\n\n36%\n\n1%\n\n26%\n\n36%\n\n37%\n\n3%\n\n23%\n\n28%\n\n46%\n\n0%\n\n10%\n\n20%\n\n30%\n\n40%\n\n50%\n\n60%\n\n70%\n\n80%\n\n90%\n\n100%\n\nEurope East Asia\n\nNorth America & Oceania\n\nSouth & Southeast\n\nAsia\n\nLatin\n\nAmerica\n\nSub−\n\nSaharan\n\nAfrica\n\nMiddle East\n\n& North\n\nAfrica\n\nRussia & Central\n\nAsia\n\nShare of personal wealth (%)\n\nTop 1% Next 9% Middle 40% Bottom 50%\n\n**Bottom 50%, middle 40% and top 10% (top 1% and next 9%)**\n\n**wealth shares across the world, 2025**\n\n**Interpretation.** In Latin America, the top 1% captures 36% of national wealth, and the next 9% an additional 33%. Together, the top 10% holds 69%, compared to 60% in Europe. Net personal wealth is equal to the sum of financial assets (e.g. equity or bonds) and non−financial assets (e.g., housing or land) owned by individuals, net of their debts. **Sources and series:** Arias−Osorio et al. (2025) and wir2026.wid.world/methodology.\n\n**Country-by-country patterns of wealth concentration**\n\nHaving mapped the global distribution of wealth by region ( **Figure 3.7** – **Figure 3.9** ), we now zoom in on how wealth is split within countries. The four maps in\n\n**Figure 3.10** – **Figure 3.14** mirror the income analysis in **Chapter 2** ( **Figure 2.9** – **Figure 2.13** ), but for net household wealth. ⁹ The picture is consistently starker than for income.\n\n**Figure 3.10** shows the share of household wealth owned by the bottom 50%. Everywhere in the world, the bottom half owns only a sliver of national wealth, at most around 14% and, in many places, just 1%. Large parts of Latin America, Sub-Saharan Africa, and the Middle East & North Africa fall in the lowest band on the map (below roughly 1–3.8%). The same pattern appears in parts of Central Europe and in the United States. By contrast, several Western European countries, Australia and New Zealand, and large Asian economies such as China and India sit one or two bands higher: their bottom halves still own little, but noticeably more than in the most unequal settings.\n\nThe shares of the middle 40% are illustrated in **Figure 3.11** . The highest bands cluster in Europe and in Oceania, where the middle 40% command a sizable portion of national wealth (around two-fifths, broadly in line with their population weight). The United States and Canada stand out with a much thinner middle share, closer to Latin American and African patterns. In Asia, the middle 40% also capture small shares in China and India.\n\n**Figure 3.12** highlights that the top decile owns the majority of wealth in most countries. The darkest colors cover Latin America, southern Africa, the Middle East, Russia, China, and India, as well as the United States, where the top 10% typically control well over 60% of household wealth. Europe and Oceania are lighter on the map: concentration is still high by any standard, but is less than in other parts of the world. Strikingly, nowhere does the top 10% own less than about 45% of total wealth; in some countries, their share approaches 86%, an extraordinary concentration for a group that represents one person in ten. Focusing on the very top sharpens the contrast ( **Figure 3.13** ). The top 1% takes\n\n83","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p83.pdf"]},"response":{"errors":[],"extract_id":"extract_a957c4fd4d327538b26be3ad2ddba1ae","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Chapter 3. Regional Wealth Inequality\n\n**Figure 3.8. Extreme wealth inequality is high in all regions**\n\n3%\n\n37%\n\n35%\n\n25%\n\n5%\n\n30%\n\n30%\n\n35%\n\n1%\n\n30%\n\n35%\n\n34%\n\n3%\n\n28%\n\n33%\n\n36%\n\n5%\n\n26%\n\n39%\n\n30%\n\n1%\n\n29%\n\n34%\n\n36%\n\n1%\n\n26%\n\n36%\n\n37%\n\n3%\n\n23%\n\n28%\n\n46%\n\n0%\n\n10%\n\n20%\n\n30%\n\n40%\n\n50%\n\n60%\n\n70%\n\n80%\n\n90%\n\n100%\n\nEurope East Asia\n\nNorth America & Oceania\n\nSouth & Southeast\n\nAsia\n\nLatin\n\nAmerica\n\nSub−\n\nSaharan\n\nAfrica\n\nMiddle East\n\n& North\n\nAfrica\n\nRussia & Central\n\nAsia\n\nShare of personal wealth (%)\n\nTop 1% Next 9% Middle 40% Bottom 50%\n\n**Bottom 50%, middle 40% and top 10% (top 1% and next 9%)**\n\n**wealth shares across the world, 2025**\n\n**Interpretation.** In Latin America, the top 1% captures 36% of national wealth, and the next 9% an additional 33%. Together, the top 10% holds 69%, compared to 60% in Europe. Net personal wealth is equal to the sum of financial assets (e.g. equity or bonds) and non−financial assets (e.g., housing or land) owned by individuals, net of their debts. **Sources and series:** Arias−Osorio et al. (2025) and wir2026.wid.world/methodology.\n\n**Country-by-country patterns of wealth concentration**\n\nHaving mapped the global distribution of wealth by region ( **Figure 3.7** – **Figure 3.9** ), we now zoom in on how wealth is split within countries. The four maps in\n\n**Figure 3.10** – **Figure 3.14** mirror the income analysis in **Chapter 2** ( **Figure 2.9** – **Figure 2.13** ), but for net household wealth. ⁹ The picture is consistently starker than for income.\n\n**Figure 3.10** shows the share of household wealth owned by the bottom 50%. Everywhere in the world, the bottom half owns only a sliver of national wealth, at most around 14% and, in many places, just 1%. Large parts of Latin America, Sub-Saharan Africa, and the Middle East & North Africa fall in the lowest band on the map (below roughly 1–3.8%). The same pattern appears in parts of Central Europe and in the United States. By contrast, several Western European countries, Australia and New Zealand, and large Asian economies such as China and India sit one or two bands higher: their bottom halves still own little, but noticeably more than in the most unequal settings.\n\nThe shares of the middle 40% are illustrated in **Figure 3.11** . The highest bands cluster in Europe and in Oceania, where the middle 40% command a sizable portion of national wealth (around two-fifths, broadly in line with their population weight). The United States and Canada stand out with a much thinner middle share, closer to Latin American and African patterns. In Asia, the middle 40% also capture small shares in China and India.\n\n**Figure 3.12** highlights that the top decile owns the majority of wealth in most countries. The darkest colors cover Latin America, southern Africa, the Middle East, Russia, China, and India, as well as the United States, where the top 10% typically control well over 60% of household wealth. Europe and Oceania are lighter on the map: concentration is still high by any standard, but is less than in other parts of the world. Strikingly, nowhere does the top 10% own less than about 45% of total wealth; in some countries, their share approaches 86%, an extraordinary concentration for a group that represents one person in ten. Focusing on the very top sharpens the contrast ( **Figure 3.13** ). The top 1% takes\n\n83"],"full_content":"Chapter 3. Regional Wealth Inequality\n\n**Figure 3.8. Extreme wealth inequality is high in all regions**\n\n3%\n\n37%\n\n35%\n\n25%\n\n5%\n\n30%\n\n30%\n\n35%\n\n1%\n\n30%\n\n35%\n\n34%\n\n3%\n\n28%\n\n33%\n\n36%\n\n5%\n\n26%\n\n39%\n\n30%\n\n1%\n\n29%\n\n34%\n\n36%\n\n1%\n\n26%\n\n36%\n\n37%\n\n3%\n\n23%\n\n28%\n\n46%\n\n0%\n\n10%\n\n20%\n\n30%\n\n40%\n\n50%\n\n60%\n\n70%\n\n80%\n\n90%\n\n100%\n\nEurope East Asia\n\nNorth America & Oceania\n\nSouth & Southeast\n\nAsia\n\nLatin\n\nAmerica\n\nSub−\n\nSaharan\n\nAfrica\n\nMiddle East\n\n& North\n\nAfrica\n\nRussia & Central\n\nAsia\n\nShare of personal wealth (%)\n\nTop 1% Next 9% Middle 40% Bottom 50%\n\n**Bottom 50%, middle 40% and top 10% (top 1% and next 9%)**\n\n**wealth shares across the world, 2025**\n\n**Interpretation.** In Latin America, the top 1% captures 36% of national wealth, and the next 9% an additional 33%. Together, the top 10% holds 69%, compared to 60% in Europe. Net personal wealth is equal to the sum of financial assets (e.g. equity or bonds) and non−financial assets (e.g., housing or land) owned by individuals, net of their debts. **Sources and series:** Arias−Osorio et al. (2025) and wir2026.wid.world/methodology.\n\n**Country-by-country patterns of wealth concentration**\n\nHaving mapped the global distribution of wealth by region ( **Figure 3.7** – **Figure 3.9** ), we now zoom in on how wealth is split within countries. The four maps in\n\n**Figure 3.10** – **Figure 3.14** mirror the income analysis in **Chapter 2** ( **Figure 2.9** – **Figure 2.13** ), but for net household wealth. ⁹ The picture is consistently starker than for income.\n\n**Figure 3.10** shows the share of household wealth owned by the bottom 50%. Everywhere in the world, the bottom half owns only a sliver of national wealth, at most around 14% and, in many places, just 1%. Large parts of Latin America, Sub-Saharan Africa, and the Middle East & North Africa fall in the lowest band on the map (below roughly 1–3.8%). The same pattern appears in parts of Central Europe and in the United States. By contrast, several Western European countries, Australia and New Zealand, and large Asian economies such as China and India sit one or two bands higher: their bottom halves still own little, but noticeably more than in the most unequal settings.\n\nThe shares of the middle 40% are illustrated in **Figure 3.11** . The highest bands cluster in Europe and in Oceania, where the middle 40% command a sizable portion of national wealth (around two-fifths, broadly in line with their population weight). The United States and Canada stand out with a much thinner middle share, closer to Latin American and African patterns. In Asia, the middle 40% also capture small shares in China and India.\n\n**Figure 3.12** highlights that the top decile owns the majority of wealth in most countries. The darkest colors cover Latin America, southern Africa, the Middle East, Russia, China, and India, as well as the United States, where the top 10% typically control well over 60% of household wealth. Europe and Oceania are lighter on the map: concentration is still high by any standard, but is less than in other parts of the world. Strikingly, nowhere does the top 10% own less than about 45% of total wealth; in some countries, their share approaches 86%, an extraordinary concentration for a group that represents one person in ten. Focusing on the very top sharpens the contrast ( **Figure 3.13** ). The top 1% takes\n\n83","publish_date":null,"title":"Chapter 3. Regional Wealth Inequality Figure3.8. Extremewealthinequalityishighinallregions The shares of the middle 40% are illustrated in Figure 3.11. The highest","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p83.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p83.pdf"},"request":{"example_id":"chart/World_Inequality_Report_2026_p83","product_type":"parse","source_file_path":"World_Inequality_Report_2026_p83.pdf"},"started_at":"2026-11-04T19:45:56.501549"} +{"completed_at":"2026-11-04T19:31:04.385934","latency_in_ms":1387,"output":{"example_id":"chart/World_Inequality_Report_2026_p95","grounded_pages":[],"layout_pages":[],"markdown":"Chapter 4. Gender Inequality\n\n**Figure 4.4. Female average incomes are smaller than males’ everywhere**\n\n40%\n\n36%\n\n16%\n\n40% 37%\n\n20%\n\n28%\n\n34%\n\nGender parity\n\n0%\n\n10%\n\n20%\n\n30%\n\n40%\n\n50%\n\n60%\n\nMiddle East\n\n& North\n\nAfrica\n\nSouth & Southeast\n\nAsia\n\nSub−\n\nSaharan\n\nAfrica\n\nEast Asia\n\nLatin\n\nAmerica\n\nRussia & Central\n\nAsia\n\nEurope North America & Oceania\n\nFemale labor income share (%)\n\n1990 2000 2010 2020 2025\n\n**Female labor income shares, 1990−2025**\n\n**Interpretation.** This figure shows the evolution of the female labor income share between 1990 and 2025 across world regions. In 2025, female workers earn about 16% of total labor income in the Middle East & North Africa, but about 40% in North America & Oceania and Europe. At the global level, women earned 27.8% of labor income in 1990 and 28.2% in 2025. While some progress has been made, gender parity remains distant in all regions. **Sources and series:** Neef and Robilliard (2021), Gabrielli et al. (2024), and wir2026.wid.world/methodology.\n\nlabor force participation experience stronger growth and a more equitable distribution of income. Yet, despite these benefits, progress has been slow and uneven, suggesting that employment inequalities are deeply embedded in economic and social structures.\n\n**Employed women earn less than employed men**\n\nEven when women overcome barriers to employment, they face another persistent challenge: lower pay. **Figure 4.9** highlights the global gender pay gap, showing that employed women consistently earn less than employed men across all regions. This gap exists at every income level, in both high- and low-income regions, with a few gains during recent decades in Latin America, North America & Oceania, Europe, and Russia & Central Asia. The gap is still present despite decades of anti-discrimination laws and advocacy. The magnitude varies by region: the gap is widest in Sub-Saharan Africa and South & Southeast Asia. Employed women earn\n\nabout 75% of what employed men earn in North America & Oceania, Europe, Russia & Central Asia, and East Asia. The persistence of this divide underscores that it is not simply a legacy of the past, but a structural feature of contemporary labor markets. Several factors contribute to the wage gap. Occupational segregation plays a major role: women are overrepresented in sectors that pay less, such as education, healthcare, and domestic services, and underrepresented in higher-paying fields like finance, engineering, and technology. Within firms, women are less likely to occupy senior positions and more likely to be hired in part-time or precarious roles, which reduces average earnings. The economic consequences are far-reaching. Lower pay compounds over time, leading to smaller savings, weaker pensions, and reduced wealth accumulation. Women not only earn less during their working years but can also accumulate lower wealth, reinforcing inequalities across generations. The gender pay gap is, therefore, more than a matter of fairness\n\n95","pages":[{"line_numbers":[],"markdown":"Chapter 4. Gender Inequality\n\n**Figure 4.4. Female average incomes are smaller than males’ everywhere**\n\n40%\n\n36%\n\n16%\n\n40% 37%\n\n20%\n\n28%\n\n34%\n\nGender parity\n\n0%\n\n10%\n\n20%\n\n30%\n\n40%\n\n50%\n\n60%\n\nMiddle East\n\n& North\n\nAfrica\n\nSouth & Southeast\n\nAsia\n\nSub−\n\nSaharan\n\nAfrica\n\nEast Asia\n\nLatin\n\nAmerica\n\nRussia & Central\n\nAsia\n\nEurope North America & Oceania\n\nFemale labor income share (%)\n\n1990 2000 2010 2020 2025\n\n**Female labor income shares, 1990−2025**\n\n**Interpretation.** This figure shows the evolution of the female labor income share between 1990 and 2025 across world regions. In 2025, female workers earn about 16% of total labor income in the Middle East & North Africa, but about 40% in North America & Oceania and Europe. At the global level, women earned 27.8% of labor income in 1990 and 28.2% in 2025. While some progress has been made, gender parity remains distant in all regions. **Sources and series:** Neef and Robilliard (2021), Gabrielli et al. (2024), and wir2026.wid.world/methodology.\n\nlabor force participation experience stronger growth and a more equitable distribution of income. Yet, despite these benefits, progress has been slow and uneven, suggesting that employment inequalities are deeply embedded in economic and social structures.\n\n**Employed women earn less than employed men**\n\nEven when women overcome barriers to employment, they face another persistent challenge: lower pay. **Figure 4.9** highlights the global gender pay gap, showing that employed women consistently earn less than employed men across all regions. This gap exists at every income level, in both high- and low-income regions, with a few gains during recent decades in Latin America, North America & Oceania, Europe, and Russia & Central Asia. The gap is still present despite decades of anti-discrimination laws and advocacy. The magnitude varies by region: the gap is widest in Sub-Saharan Africa and South & Southeast Asia. Employed women earn\n\nabout 75% of what employed men earn in North America & Oceania, Europe, Russia & Central Asia, and East Asia. The persistence of this divide underscores that it is not simply a legacy of the past, but a structural feature of contemporary labor markets. Several factors contribute to the wage gap. Occupational segregation plays a major role: women are overrepresented in sectors that pay less, such as education, healthcare, and domestic services, and underrepresented in higher-paying fields like finance, engineering, and technology. Within firms, women are less likely to occupy senior positions and more likely to be hired in part-time or precarious roles, which reduces average earnings. The economic consequences are far-reaching. Lower pay compounds over time, leading to smaller savings, weaker pensions, and reduced wealth accumulation. Women not only earn less during their working years but can also accumulate lower wealth, reinforcing inequalities across generations. The gender pay gap is, therefore, more than a matter of fairness\n\n95","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Chapter 4. Gender Inequality\n\n**Figure 4.4. Female average incomes are smaller than males’ everywhere**\n\n40%\n\n36%\n\n16%\n\n40% 37%\n\n20%\n\n28%\n\n34%\n\nGender parity\n\n0%\n\n10%\n\n20%\n\n30%\n\n40%\n\n50%\n\n60%\n\nMiddle East\n\n& North\n\nAfrica\n\nSouth & Southeast\n\nAsia\n\nSub−\n\nSaharan\n\nAfrica\n\nEast Asia\n\nLatin\n\nAmerica\n\nRussia & Central\n\nAsia\n\nEurope North America & Oceania\n\nFemale labor income share (%)\n\n1990 2000 2010 2020 2025\n\n**Female labor income shares, 1990−2025**\n\n**Interpretation.** This figure shows the evolution of the female labor income share between 1990 and 2025 across world regions. In 2025, female workers earn about 16% of total labor income in the Middle East & North Africa, but about 40% in North America & Oceania and Europe. At the global level, women earned 27.8% of labor income in 1990 and 28.2% in 2025. While some progress has been made, gender parity remains distant in all regions. **Sources and series:** Neef and Robilliard (2021), Gabrielli et al. (2024), and wir2026.wid.world/methodology.\n\nlabor force participation experience stronger growth and a more equitable distribution of income. Yet, despite these benefits, progress has been slow and uneven, suggesting that employment inequalities are deeply embedded in economic and social structures.\n\n**Employed women earn less than employed men**\n\nEven when women overcome barriers to employment, they face another persistent challenge: lower pay. **Figure 4.9** highlights the global gender pay gap, showing that employed women consistently earn less than employed men across all regions. This gap exists at every income level, in both high- and low-income regions, with a few gains during recent decades in Latin America, North America & Oceania, Europe, and Russia & Central Asia. The gap is still present despite decades of anti-discrimination laws and advocacy. The magnitude varies by region: the gap is widest in Sub-Saharan Africa and South & Southeast Asia. Employed women earn\n\nabout 75% of what employed men earn in North America & Oceania, Europe, Russia & Central Asia, and East Asia. The persistence of this divide underscores that it is not simply a legacy of the past, but a structural feature of contemporary labor markets. Several factors contribute to the wage gap. Occupational segregation plays a major role: women are overrepresented in sectors that pay less, such as education, healthcare, and domestic services, and underrepresented in higher-paying fields like finance, engineering, and technology. Within firms, women are less likely to occupy senior positions and more likely to be hired in part-time or precarious roles, which reduces average earnings. The economic consequences are far-reaching. Lower pay compounds over time, leading to smaller savings, weaker pensions, and reduced wealth accumulation. Women not only earn less during their working years but can also accumulate lower wealth, reinforcing inequalities across generations. The gender pay gap is, therefore, more than a matter of fairness\n\n95","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p95.pdf"]},"response":{"errors":[],"extract_id":"extract_ed9efe1e0b8ea5d1983e67a3d8923851","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Chapter 4. Gender Inequality\n\n**Figure 4.4. Female average incomes are smaller than males’ everywhere**\n\n40%\n\n36%\n\n16%\n\n40% 37%\n\n20%\n\n28%\n\n34%\n\nGender parity\n\n0%\n\n10%\n\n20%\n\n30%\n\n40%\n\n50%\n\n60%\n\nMiddle East\n\n& North\n\nAfrica\n\nSouth & Southeast\n\nAsia\n\nSub−\n\nSaharan\n\nAfrica\n\nEast Asia\n\nLatin\n\nAmerica\n\nRussia & Central\n\nAsia\n\nEurope North America & Oceania\n\nFemale labor income share (%)\n\n1990 2000 2010 2020 2025\n\n**Female labor income shares, 1990−2025**\n\n**Interpretation.** This figure shows the evolution of the female labor income share between 1990 and 2025 across world regions. In 2025, female workers earn about 16% of total labor income in the Middle East & North Africa, but about 40% in North America & Oceania and Europe. At the global level, women earned 27.8% of labor income in 1990 and 28.2% in 2025. While some progress has been made, gender parity remains distant in all regions. **Sources and series:** Neef and Robilliard (2021), Gabrielli et al. (2024), and wir2026.wid.world/methodology.\n\nlabor force participation experience stronger growth and a more equitable distribution of income. Yet, despite these benefits, progress has been slow and uneven, suggesting that employment inequalities are deeply embedded in economic and social structures.\n\n**Employed women earn less than employed men**\n\nEven when women overcome barriers to employment, they face another persistent challenge: lower pay. **Figure 4.9** highlights the global gender pay gap, showing that employed women consistently earn less than employed men across all regions. This gap exists at every income level, in both high- and low-income regions, with a few gains during recent decades in Latin America, North America & Oceania, Europe, and Russia & Central Asia. The gap is still present despite decades of anti-discrimination laws and advocacy. The magnitude varies by region: the gap is widest in Sub-Saharan Africa and South & Southeast Asia. Employed women earn\n\nabout 75% of what employed men earn in North America & Oceania, Europe, Russia & Central Asia, and East Asia. The persistence of this divide underscores that it is not simply a legacy of the past, but a structural feature of contemporary labor markets. Several factors contribute to the wage gap. Occupational segregation plays a major role: women are overrepresented in sectors that pay less, such as education, healthcare, and domestic services, and underrepresented in higher-paying fields like finance, engineering, and technology. Within firms, women are less likely to occupy senior positions and more likely to be hired in part-time or precarious roles, which reduces average earnings. The economic consequences are far-reaching. Lower pay compounds over time, leading to smaller savings, weaker pensions, and reduced wealth accumulation. Women not only earn less during their working years but can also accumulate lower wealth, reinforcing inequalities across generations. The gender pay gap is, therefore, more than a matter of fairness\n\n95"],"full_content":"Chapter 4. Gender Inequality\n\n**Figure 4.4. Female average incomes are smaller than males’ everywhere**\n\n40%\n\n36%\n\n16%\n\n40% 37%\n\n20%\n\n28%\n\n34%\n\nGender parity\n\n0%\n\n10%\n\n20%\n\n30%\n\n40%\n\n50%\n\n60%\n\nMiddle East\n\n& North\n\nAfrica\n\nSouth & Southeast\n\nAsia\n\nSub−\n\nSaharan\n\nAfrica\n\nEast Asia\n\nLatin\n\nAmerica\n\nRussia & Central\n\nAsia\n\nEurope North America & Oceania\n\nFemale labor income share (%)\n\n1990 2000 2010 2020 2025\n\n**Female labor income shares, 1990−2025**\n\n**Interpretation.** This figure shows the evolution of the female labor income share between 1990 and 2025 across world regions. In 2025, female workers earn about 16% of total labor income in the Middle East & North Africa, but about 40% in North America & Oceania and Europe. At the global level, women earned 27.8% of labor income in 1990 and 28.2% in 2025. While some progress has been made, gender parity remains distant in all regions. **Sources and series:** Neef and Robilliard (2021), Gabrielli et al. (2024), and wir2026.wid.world/methodology.\n\nlabor force participation experience stronger growth and a more equitable distribution of income. Yet, despite these benefits, progress has been slow and uneven, suggesting that employment inequalities are deeply embedded in economic and social structures.\n\n**Employed women earn less than employed men**\n\nEven when women overcome barriers to employment, they face another persistent challenge: lower pay. **Figure 4.9** highlights the global gender pay gap, showing that employed women consistently earn less than employed men across all regions. This gap exists at every income level, in both high- and low-income regions, with a few gains during recent decades in Latin America, North America & Oceania, Europe, and Russia & Central Asia. The gap is still present despite decades of anti-discrimination laws and advocacy. The magnitude varies by region: the gap is widest in Sub-Saharan Africa and South & Southeast Asia. Employed women earn\n\nabout 75% of what employed men earn in North America & Oceania, Europe, Russia & Central Asia, and East Asia. The persistence of this divide underscores that it is not simply a legacy of the past, but a structural feature of contemporary labor markets. Several factors contribute to the wage gap. Occupational segregation plays a major role: women are overrepresented in sectors that pay less, such as education, healthcare, and domestic services, and underrepresented in higher-paying fields like finance, engineering, and technology. Within firms, women are less likely to occupy senior positions and more likely to be hired in part-time or precarious roles, which reduces average earnings. The economic consequences are far-reaching. Lower pay compounds over time, leading to smaller savings, weaker pensions, and reduced wealth accumulation. Women not only earn less during their working years but can also accumulate lower wealth, reinforcing inequalities across generations. The gender pay gap is, therefore, more than a matter of fairness\n\n95","publish_date":null,"title":"Chapter 4. Gender Inequality Figure4.4. Femaleaverageincomesaresmallerthanmales’ everywhere labor force participation experience stronger about 75% of what employed men earn in","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p95.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p95.pdf"},"request":{"example_id":"chart/World_Inequality_Report_2026_p95","product_type":"parse","source_file_path":"World_Inequality_Report_2026_p95.pdf"},"started_at":"2026-11-04T19:31:02.998151"} +{"completed_at":"2026-11-04T19:45:59.857139","latency_in_ms":1156,"output":{"example_id":"chart/World_Inequality_Report_2026_p98","grounded_pages":[],"layout_pages":[],"markdown":"Chapter 4. Gender Inequality\n\n**Figure 4.8. Women are less likely than men to hold a job in the labor market**\n\n89%\n\n73%\n\n32%\n\n91% 93%\n\n34%\n\n62%\n\n74%\n\nGender parity\n\n0%\n\n10% 20% 30% 40% 50% 60% 70% 80% 90%\n\n100%\n\nMiddle East\n\n& North\n\nAfrica\n\nSouth & Southeast\n\nAsia\n\nSub−\n\nSaharan\n\nAfrica\n\nLatin\n\nAmerica\n\nEast Asia\n\nEurope North America & Oceania\n\nRussia & Central\n\nAsia\n\nEmployed women relative to\n\nemployed men (%)\n\n1990 2000 2010 2020 2025\n\n**Employed women relative to employed men, 1990−2025**\n\n**Interpretation.** This figure shows the evolution of the gender total employment ratio between 1990 and 2025 across world regions. The indicator measures the share of employed women relative to employed men, regardless of how much they earn. In 2025, female employment remains well below parity in several regions: only 32% of women are employed per 100 employed men in Middle East & North Africa. In contrast, employment ratios are close to gender parity in Russia & Central Asia (93%), North America & Oceania (91%), and Europe (89%). The global gender employment gap (the gap up to 100%) increased slightly (from 67% in 1990 to 60% in 2025), but progress has been made across some regions. **Sources and series:** wir2026.wid.world/methodology and Gabrielli et al. (2024).\n\n**Figure 4.9. Employed women earn less than employed men everywhere**\n\n75% 81% 84%\n\n75% 75%\n\n66% 66%\n\n74%\n\nGender parity\n\n0%\n\n10% 20% 30% 40% 50% 60% 70% 80% 90%\n\n100%\n\nSub−\n\nSaharan\n\nAfrica\n\nSouth & Southeast\n\nAsia\n\nEast Asia\n\nRussia & Central\n\nAsia\n\nEurope North America & Oceania\n\nLatin\n\nAmerica\n\nMiddle East\n\n& North\n\nAfrica\n\nAverage earnings of women to\n\nthose of men (%)\n\n1990 2000 2010 2020 2025\n\n**Gender earnings ratio across the world, 1990−2025**\n\n**Interpretation.** This figure shows the evolution of the earnings gender ratio between 1990 and 2025 across world regions. The indicator measures the average earnings of employed women as a share of the average earnings of employed men. In 2025, the gender earnings gap remains wide in several regions: women earn only 66% of what men earn in both Sub−Saharan Africa and South & Southeast Asia. In contrast, earnings ratios are highest in Middle East & North Africa (84%) and Latin America (81%). Across high−income regions such as North America & Oceania and Europe (and also Russia & Central Asia), the average female−to−male earnings ratio hovers around 75%. At the global level, women earned 69% of men’s average income in 1990 and 71% in 2025. While modest progress has been made over the past 35 years, gender parity in earnings remains out of reach in all regions. **Sources and series:** wir2026.wid.world/methodology and Gabrielli et al. (2024).\n\n98","pages":[{"line_numbers":[],"markdown":"Chapter 4. Gender Inequality\n\n**Figure 4.8. Women are less likely than men to hold a job in the labor market**\n\n89%\n\n73%\n\n32%\n\n91% 93%\n\n34%\n\n62%\n\n74%\n\nGender parity\n\n0%\n\n10% 20% 30% 40% 50% 60% 70% 80% 90%\n\n100%\n\nMiddle East\n\n& North\n\nAfrica\n\nSouth & Southeast\n\nAsia\n\nSub−\n\nSaharan\n\nAfrica\n\nLatin\n\nAmerica\n\nEast Asia\n\nEurope North America & Oceania\n\nRussia & Central\n\nAsia\n\nEmployed women relative to\n\nemployed men (%)\n\n1990 2000 2010 2020 2025\n\n**Employed women relative to employed men, 1990−2025**\n\n**Interpretation.** This figure shows the evolution of the gender total employment ratio between 1990 and 2025 across world regions. The indicator measures the share of employed women relative to employed men, regardless of how much they earn. In 2025, female employment remains well below parity in several regions: only 32% of women are employed per 100 employed men in Middle East & North Africa. In contrast, employment ratios are close to gender parity in Russia & Central Asia (93%), North America & Oceania (91%), and Europe (89%). The global gender employment gap (the gap up to 100%) increased slightly (from 67% in 1990 to 60% in 2025), but progress has been made across some regions. **Sources and series:** wir2026.wid.world/methodology and Gabrielli et al. (2024).\n\n**Figure 4.9. Employed women earn less than employed men everywhere**\n\n75% 81% 84%\n\n75% 75%\n\n66% 66%\n\n74%\n\nGender parity\n\n0%\n\n10% 20% 30% 40% 50% 60% 70% 80% 90%\n\n100%\n\nSub−\n\nSaharan\n\nAfrica\n\nSouth & Southeast\n\nAsia\n\nEast Asia\n\nRussia & Central\n\nAsia\n\nEurope North America & Oceania\n\nLatin\n\nAmerica\n\nMiddle East\n\n& North\n\nAfrica\n\nAverage earnings of women to\n\nthose of men (%)\n\n1990 2000 2010 2020 2025\n\n**Gender earnings ratio across the world, 1990−2025**\n\n**Interpretation.** This figure shows the evolution of the earnings gender ratio between 1990 and 2025 across world regions. The indicator measures the average earnings of employed women as a share of the average earnings of employed men. In 2025, the gender earnings gap remains wide in several regions: women earn only 66% of what men earn in both Sub−Saharan Africa and South & Southeast Asia. In contrast, earnings ratios are highest in Middle East & North Africa (84%) and Latin America (81%). Across high−income regions such as North America & Oceania and Europe (and also Russia & Central Asia), the average female−to−male earnings ratio hovers around 75%. At the global level, women earned 69% of men’s average income in 1990 and 71% in 2025. While modest progress has been made over the past 35 years, gender parity in earnings remains out of reach in all regions. **Sources and series:** wir2026.wid.world/methodology and Gabrielli et al. (2024).\n\n98","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Chapter 4. Gender Inequality\n\n**Figure 4.8. Women are less likely than men to hold a job in the labor market**\n\n89%\n\n73%\n\n32%\n\n91% 93%\n\n34%\n\n62%\n\n74%\n\nGender parity\n\n0%\n\n10% 20% 30% 40% 50% 60% 70% 80% 90%\n\n100%\n\nMiddle East\n\n& North\n\nAfrica\n\nSouth & Southeast\n\nAsia\n\nSub−\n\nSaharan\n\nAfrica\n\nLatin\n\nAmerica\n\nEast Asia\n\nEurope North America & Oceania\n\nRussia & Central\n\nAsia\n\nEmployed women relative to\n\nemployed men (%)\n\n1990 2000 2010 2020 2025\n\n**Employed women relative to employed men, 1990−2025**\n\n**Interpretation.** This figure shows the evolution of the gender total employment ratio between 1990 and 2025 across world regions. The indicator measures the share of employed women relative to employed men, regardless of how much they earn. In 2025, female employment remains well below parity in several regions: only 32% of women are employed per 100 employed men in Middle East & North Africa. In contrast, employment ratios are close to gender parity in Russia & Central Asia (93%), North America & Oceania (91%), and Europe (89%). The global gender employment gap (the gap up to 100%) increased slightly (from 67% in 1990 to 60% in 2025), but progress has been made across some regions. **Sources and series:** wir2026.wid.world/methodology and Gabrielli et al. (2024).\n\n**Figure 4.9. Employed women earn less than employed men everywhere**\n\n75% 81% 84%\n\n75% 75%\n\n66% 66%\n\n74%\n\nGender parity\n\n0%\n\n10% 20% 30% 40% 50% 60% 70% 80% 90%\n\n100%\n\nSub−\n\nSaharan\n\nAfrica\n\nSouth & Southeast\n\nAsia\n\nEast Asia\n\nRussia & Central\n\nAsia\n\nEurope North America & Oceania\n\nLatin\n\nAmerica\n\nMiddle East\n\n& North\n\nAfrica\n\nAverage earnings of women to\n\nthose of men (%)\n\n1990 2000 2010 2020 2025\n\n**Gender earnings ratio across the world, 1990−2025**\n\n**Interpretation.** This figure shows the evolution of the earnings gender ratio between 1990 and 2025 across world regions. The indicator measures the average earnings of employed women as a share of the average earnings of employed men. In 2025, the gender earnings gap remains wide in several regions: women earn only 66% of what men earn in both Sub−Saharan Africa and South & Southeast Asia. In contrast, earnings ratios are highest in Middle East & North Africa (84%) and Latin America (81%). Across high−income regions such as North America & Oceania and Europe (and also Russia & Central Asia), the average female−to−male earnings ratio hovers around 75%. At the global level, women earned 69% of men’s average income in 1990 and 71% in 2025. While modest progress has been made over the past 35 years, gender parity in earnings remains out of reach in all regions. **Sources and series:** wir2026.wid.world/methodology and Gabrielli et al. (2024).\n\n98","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p98.pdf"]},"response":{"errors":[],"extract_id":"extract_7013ad8b200dd19b7d2ad69b09232a21","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Chapter 4. Gender Inequality\n\n**Figure 4.8. Women are less likely than men to hold a job in the labor market**\n\n89%\n\n73%\n\n32%\n\n91% 93%\n\n34%\n\n62%\n\n74%\n\nGender parity\n\n0%\n\n10% 20% 30% 40% 50% 60% 70% 80% 90%\n\n100%\n\nMiddle East\n\n& North\n\nAfrica\n\nSouth & Southeast\n\nAsia\n\nSub−\n\nSaharan\n\nAfrica\n\nLatin\n\nAmerica\n\nEast Asia\n\nEurope North America & Oceania\n\nRussia & Central\n\nAsia\n\nEmployed women relative to\n\nemployed men (%)\n\n1990 2000 2010 2020 2025\n\n**Employed women relative to employed men, 1990−2025**\n\n**Interpretation.** This figure shows the evolution of the gender total employment ratio between 1990 and 2025 across world regions. The indicator measures the share of employed women relative to employed men, regardless of how much they earn. In 2025, female employment remains well below parity in several regions: only 32% of women are employed per 100 employed men in Middle East & North Africa. In contrast, employment ratios are close to gender parity in Russia & Central Asia (93%), North America & Oceania (91%), and Europe (89%). The global gender employment gap (the gap up to 100%) increased slightly (from 67% in 1990 to 60% in 2025), but progress has been made across some regions. **Sources and series:** wir2026.wid.world/methodology and Gabrielli et al. (2024).\n\n**Figure 4.9. Employed women earn less than employed men everywhere**\n\n75% 81% 84%\n\n75% 75%\n\n66% 66%\n\n74%\n\nGender parity\n\n0%\n\n10% 20% 30% 40% 50% 60% 70% 80% 90%\n\n100%\n\nSub−\n\nSaharan\n\nAfrica\n\nSouth & Southeast\n\nAsia\n\nEast Asia\n\nRussia & Central\n\nAsia\n\nEurope North America & Oceania\n\nLatin\n\nAmerica\n\nMiddle East\n\n& North\n\nAfrica\n\nAverage earnings of women to\n\nthose of men (%)\n\n1990 2000 2010 2020 2025\n\n**Gender earnings ratio across the world, 1990−2025**\n\n**Interpretation.** This figure shows the evolution of the earnings gender ratio between 1990 and 2025 across world regions. The indicator measures the average earnings of employed women as a share of the average earnings of employed men. In 2025, the gender earnings gap remains wide in several regions: women earn only 66% of what men earn in both Sub−Saharan Africa and South & Southeast Asia. In contrast, earnings ratios are highest in Middle East & North Africa (84%) and Latin America (81%). Across high−income regions such as North America & Oceania and Europe (and also Russia & Central Asia), the average female−to−male earnings ratio hovers around 75%. At the global level, women earned 69% of men’s average income in 1990 and 71% in 2025. While modest progress has been made over the past 35 years, gender parity in earnings remains out of reach in all regions. **Sources and series:** wir2026.wid.world/methodology and Gabrielli et al. (2024).\n\n98"],"full_content":"Chapter 4. Gender Inequality\n\n**Figure 4.8. Women are less likely than men to hold a job in the labor market**\n\n89%\n\n73%\n\n32%\n\n91% 93%\n\n34%\n\n62%\n\n74%\n\nGender parity\n\n0%\n\n10% 20% 30% 40% 50% 60% 70% 80% 90%\n\n100%\n\nMiddle East\n\n& North\n\nAfrica\n\nSouth & Southeast\n\nAsia\n\nSub−\n\nSaharan\n\nAfrica\n\nLatin\n\nAmerica\n\nEast Asia\n\nEurope North America & Oceania\n\nRussia & Central\n\nAsia\n\nEmployed women relative to\n\nemployed men (%)\n\n1990 2000 2010 2020 2025\n\n**Employed women relative to employed men, 1990−2025**\n\n**Interpretation.** This figure shows the evolution of the gender total employment ratio between 1990 and 2025 across world regions. The indicator measures the share of employed women relative to employed men, regardless of how much they earn. In 2025, female employment remains well below parity in several regions: only 32% of women are employed per 100 employed men in Middle East & North Africa. In contrast, employment ratios are close to gender parity in Russia & Central Asia (93%), North America & Oceania (91%), and Europe (89%). The global gender employment gap (the gap up to 100%) increased slightly (from 67% in 1990 to 60% in 2025), but progress has been made across some regions. **Sources and series:** wir2026.wid.world/methodology and Gabrielli et al. (2024).\n\n**Figure 4.9. Employed women earn less than employed men everywhere**\n\n75% 81% 84%\n\n75% 75%\n\n66% 66%\n\n74%\n\nGender parity\n\n0%\n\n10% 20% 30% 40% 50% 60% 70% 80% 90%\n\n100%\n\nSub−\n\nSaharan\n\nAfrica\n\nSouth & Southeast\n\nAsia\n\nEast Asia\n\nRussia & Central\n\nAsia\n\nEurope North America & Oceania\n\nLatin\n\nAmerica\n\nMiddle East\n\n& North\n\nAfrica\n\nAverage earnings of women to\n\nthose of men (%)\n\n1990 2000 2010 2020 2025\n\n**Gender earnings ratio across the world, 1990−2025**\n\n**Interpretation.** This figure shows the evolution of the earnings gender ratio between 1990 and 2025 across world regions. The indicator measures the average earnings of employed women as a share of the average earnings of employed men. In 2025, the gender earnings gap remains wide in several regions: women earn only 66% of what men earn in both Sub−Saharan Africa and South & Southeast Asia. In contrast, earnings ratios are highest in Middle East & North Africa (84%) and Latin America (81%). Across high−income regions such as North America & Oceania and Europe (and also Russia & Central Asia), the average female−to−male earnings ratio hovers around 75%. At the global level, women earned 69% of men’s average income in 1990 and 71% in 2025. While modest progress has been made over the past 35 years, gender parity in earnings remains out of reach in all regions. **Sources and series:** wir2026.wid.world/methodology and Gabrielli et al. (2024).\n\n98","publish_date":null,"title":"Chapter 4. Gender Inequality Figure4.8. Womenarelesslikelythanmentoholdajobinthe labormarket Figure4.9. Employedwomenearnlessthanemployedmen everywhere 98","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p98.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p98.pdf"},"request":{"example_id":"chart/World_Inequality_Report_2026_p98","product_type":"parse","source_file_path":"World_Inequality_Report_2026_p98.pdf"},"started_at":"2026-11-04T19:45:58.700601"} +{"completed_at":"2026-11-04T19:46:01.314425","latency_in_ms":1060,"output":{"example_id":"chart/World_Inequality_Report_2026_p99","grounded_pages":[],"layout_pages":[],"markdown":"Chapter 4. Gender Inequality\n\n**Figure 4.10. The high school enrollment gender gap has decreased in the last 25 years**\n\nGender parity\n\n80%\n\n85%\n\n90%\n\n95%\n\n100%\n\n105%\n\n2000 2005 2010 2015 2020 2025\n\nFemale high school enrollment\n\n(as % of male high school enrollment)\n\nHigh income Low & middle income World\n\n**Female high school enrollment as a share of male enrollment, 2000−2025**\n\n**Interpretation.** This figure shows the evolution of the gender gap in high school enrollment from 2000 to 2025 across country income groups. The indicator measures female enrollment as a share of male enrollment. At the global level, this share increased from 88% in 2000 to 98% in 2025, indicating near gender parity. In high−income countries, the ratio reached 101%, and in low− & middle−income countries, the ratio reached 98%, reflecting considerable progress over two decades. **Sources and series:** wir2026.wid.world/methodology.\n\n**Figure 4.11. Education alone cannot fully close the gap**\n\n10\\.5%\n\n8\\.3%\n\n12\\.1%\n\n10%\n\n9\\.3%\n\n6\\.5%\n\n11%\n\n8\\.7%\n\n6\\.3%\n\n4\\.2%\n\n9%\n\n6\\.6%\n\n9\\.7%\n\n7\\.8%\n\n9\\.9%\n\n7\\.6%\n\n0% 1% 2% 3% 4% 5% 6% 7% 8% 9%\n\n10% 11% 12% 13% 14%\n\nMENA Other Asia\n\nChina India World Sub−Sah.\n\nAfrica\n\nLatin\n\nAmerica\n\nEurope /\n\nU.S.\n\nReturn to an additional year of schooling (%)\n\nMen Women\n\n**Returns to schooling by gender, 2019**\n\n**Interpretation** . The figure plots returns to a year of schooling by gender and world region in 2019. Estimates correspond to the effect of one additional year of schooling on the log of personal income, estimated separately by gender using modified Mincerian equations that control for an experience quartic. In all world regions, the return to a year of schooling is higher for women than for men. **Sources and series:** Gethin (2024).\n\n99","pages":[{"line_numbers":[],"markdown":"Chapter 4. Gender Inequality\n\n**Figure 4.10. The high school enrollment gender gap has decreased in the last 25 years**\n\nGender parity\n\n80%\n\n85%\n\n90%\n\n95%\n\n100%\n\n105%\n\n2000 2005 2010 2015 2020 2025\n\nFemale high school enrollment\n\n(as % of male high school enrollment)\n\nHigh income Low & middle income World\n\n**Female high school enrollment as a share of male enrollment, 2000−2025**\n\n**Interpretation.** This figure shows the evolution of the gender gap in high school enrollment from 2000 to 2025 across country income groups. The indicator measures female enrollment as a share of male enrollment. At the global level, this share increased from 88% in 2000 to 98% in 2025, indicating near gender parity. In high−income countries, the ratio reached 101%, and in low− & middle−income countries, the ratio reached 98%, reflecting considerable progress over two decades. **Sources and series:** wir2026.wid.world/methodology.\n\n**Figure 4.11. Education alone cannot fully close the gap**\n\n10\\.5%\n\n8\\.3%\n\n12\\.1%\n\n10%\n\n9\\.3%\n\n6\\.5%\n\n11%\n\n8\\.7%\n\n6\\.3%\n\n4\\.2%\n\n9%\n\n6\\.6%\n\n9\\.7%\n\n7\\.8%\n\n9\\.9%\n\n7\\.6%\n\n0% 1% 2% 3% 4% 5% 6% 7% 8% 9%\n\n10% 11% 12% 13% 14%\n\nMENA Other Asia\n\nChina India World Sub−Sah.\n\nAfrica\n\nLatin\n\nAmerica\n\nEurope /\n\nU.S.\n\nReturn to an additional year of schooling (%)\n\nMen Women\n\n**Returns to schooling by gender, 2019**\n\n**Interpretation** . The figure plots returns to a year of schooling by gender and world region in 2019. Estimates correspond to the effect of one additional year of schooling on the log of personal income, estimated separately by gender using modified Mincerian equations that control for an experience quartic. In all world regions, the return to a year of schooling is higher for women than for men. **Sources and series:** Gethin (2024).\n\n99","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Chapter 4. Gender Inequality\n\n**Figure 4.10. The high school enrollment gender gap has decreased in the last 25 years**\n\nGender parity\n\n80%\n\n85%\n\n90%\n\n95%\n\n100%\n\n105%\n\n2000 2005 2010 2015 2020 2025\n\nFemale high school enrollment\n\n(as % of male high school enrollment)\n\nHigh income Low & middle income World\n\n**Female high school enrollment as a share of male enrollment, 2000−2025**\n\n**Interpretation.** This figure shows the evolution of the gender gap in high school enrollment from 2000 to 2025 across country income groups. The indicator measures female enrollment as a share of male enrollment. At the global level, this share increased from 88% in 2000 to 98% in 2025, indicating near gender parity. In high−income countries, the ratio reached 101%, and in low− & middle−income countries, the ratio reached 98%, reflecting considerable progress over two decades. **Sources and series:** wir2026.wid.world/methodology.\n\n**Figure 4.11. Education alone cannot fully close the gap**\n\n10\\.5%\n\n8\\.3%\n\n12\\.1%\n\n10%\n\n9\\.3%\n\n6\\.5%\n\n11%\n\n8\\.7%\n\n6\\.3%\n\n4\\.2%\n\n9%\n\n6\\.6%\n\n9\\.7%\n\n7\\.8%\n\n9\\.9%\n\n7\\.6%\n\n0% 1% 2% 3% 4% 5% 6% 7% 8% 9%\n\n10% 11% 12% 13% 14%\n\nMENA Other Asia\n\nChina India World Sub−Sah.\n\nAfrica\n\nLatin\n\nAmerica\n\nEurope /\n\nU.S.\n\nReturn to an additional year of schooling (%)\n\nMen Women\n\n**Returns to schooling by gender, 2019**\n\n**Interpretation** . The figure plots returns to a year of schooling by gender and world region in 2019. Estimates correspond to the effect of one additional year of schooling on the log of personal income, estimated separately by gender using modified Mincerian equations that control for an experience quartic. In all world regions, the return to a year of schooling is higher for women than for men. **Sources and series:** Gethin (2024).\n\n99","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p99.pdf"]},"response":{"errors":[],"extract_id":"extract_2a1ad85e21ce9f0d36adccb66b9d7a31","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Chapter 4. Gender Inequality\n\n**Figure 4.10. The high school enrollment gender gap has decreased in the last 25 years**\n\nGender parity\n\n80%\n\n85%\n\n90%\n\n95%\n\n100%\n\n105%\n\n2000 2005 2010 2015 2020 2025\n\nFemale high school enrollment\n\n(as % of male high school enrollment)\n\nHigh income Low & middle income World\n\n**Female high school enrollment as a share of male enrollment, 2000−2025**\n\n**Interpretation.** This figure shows the evolution of the gender gap in high school enrollment from 2000 to 2025 across country income groups. The indicator measures female enrollment as a share of male enrollment. At the global level, this share increased from 88% in 2000 to 98% in 2025, indicating near gender parity. In high−income countries, the ratio reached 101%, and in low− & middle−income countries, the ratio reached 98%, reflecting considerable progress over two decades. **Sources and series:** wir2026.wid.world/methodology.\n\n**Figure 4.11. Education alone cannot fully close the gap**\n\n10\\.5%\n\n8\\.3%\n\n12\\.1%\n\n10%\n\n9\\.3%\n\n6\\.5%\n\n11%\n\n8\\.7%\n\n6\\.3%\n\n4\\.2%\n\n9%\n\n6\\.6%\n\n9\\.7%\n\n7\\.8%\n\n9\\.9%\n\n7\\.6%\n\n0% 1% 2% 3% 4% 5% 6% 7% 8% 9%\n\n10% 11% 12% 13% 14%\n\nMENA Other Asia\n\nChina India World Sub−Sah.\n\nAfrica\n\nLatin\n\nAmerica\n\nEurope /\n\nU.S.\n\nReturn to an additional year of schooling (%)\n\nMen Women\n\n**Returns to schooling by gender, 2019**\n\n**Interpretation** . The figure plots returns to a year of schooling by gender and world region in 2019. Estimates correspond to the effect of one additional year of schooling on the log of personal income, estimated separately by gender using modified Mincerian equations that control for an experience quartic. In all world regions, the return to a year of schooling is higher for women than for men. **Sources and series:** Gethin (2024).\n\n99"],"full_content":"Chapter 4. Gender Inequality\n\n**Figure 4.10. The high school enrollment gender gap has decreased in the last 25 years**\n\nGender parity\n\n80%\n\n85%\n\n90%\n\n95%\n\n100%\n\n105%\n\n2000 2005 2010 2015 2020 2025\n\nFemale high school enrollment\n\n(as % of male high school enrollment)\n\nHigh income Low & middle income World\n\n**Female high school enrollment as a share of male enrollment, 2000−2025**\n\n**Interpretation.** This figure shows the evolution of the gender gap in high school enrollment from 2000 to 2025 across country income groups. The indicator measures female enrollment as a share of male enrollment. At the global level, this share increased from 88% in 2000 to 98% in 2025, indicating near gender parity. In high−income countries, the ratio reached 101%, and in low− & middle−income countries, the ratio reached 98%, reflecting considerable progress over two decades. **Sources and series:** wir2026.wid.world/methodology.\n\n**Figure 4.11. Education alone cannot fully close the gap**\n\n10\\.5%\n\n8\\.3%\n\n12\\.1%\n\n10%\n\n9\\.3%\n\n6\\.5%\n\n11%\n\n8\\.7%\n\n6\\.3%\n\n4\\.2%\n\n9%\n\n6\\.6%\n\n9\\.7%\n\n7\\.8%\n\n9\\.9%\n\n7\\.6%\n\n0% 1% 2% 3% 4% 5% 6% 7% 8% 9%\n\n10% 11% 12% 13% 14%\n\nMENA Other Asia\n\nChina India World Sub−Sah.\n\nAfrica\n\nLatin\n\nAmerica\n\nEurope /\n\nU.S.\n\nReturn to an additional year of schooling (%)\n\nMen Women\n\n**Returns to schooling by gender, 2019**\n\n**Interpretation** . The figure plots returns to a year of schooling by gender and world region in 2019. Estimates correspond to the effect of one additional year of schooling on the log of personal income, estimated separately by gender using modified Mincerian equations that control for an experience quartic. In all world regions, the return to a year of schooling is higher for women than for men. **Sources and series:** Gethin (2024).\n\n99","publish_date":null,"title":"Chapter 4. Gender Inequality Figure4.10. Thehighschoolenrollmentgendergaphas decreasedinthelast25years Figure4.11. Educationalonecannotfullyclosethegap 99","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p99.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Inequality_Report_2026_p99.pdf"},"request":{"example_id":"chart/World_Inequality_Report_2026_p99","product_type":"parse","source_file_path":"World_Inequality_Report_2026_p99.pdf"},"started_at":"2026-11-04T19:46:00.254368"} +{"completed_at":"2026-11-04T19:46:03.655600","latency_in_ms":2209,"output":{"example_id":"chart/World_Manufacturing_Production_2024_Q3_p11","grounded_pages":[],"layout_pages":[],"markdown":"# **Findings by industry groups 3**\n\n### Higher-technology\n\nindustries maintained their strong progress, outperforming other industries\n\nThe latest global trends of industrial sectors grouped by technolog- ical intensity are shown in Figure 3.1. The data reveal that higher- technology industries continued to exhibit strong dynamism, even thriving amidst numerous challenges.\n\nIn recent periods, industries classified as medium-high and high- technology (MHT) have generally outperformed other manufacturing sectors. While they experienced a temporary slowdown in the first quarter of 2024, they returned to growth in the next quarter, increasing output by 1.5 per cent. This upward trend continued into the third quarter of 2024 with a 1.0 per cent quarterly increase. Conversely, lower-technology manufacturing sectors remained flat.\n\n90\n\n100\n\n110\n\n120\n\n130\n\n140\n\nQ1 2015 Q3 Q1 2016 Q3 Q1 2017 Q3 Q1 2018 Q3 Q1 2019 Q3 Q1 2020 Q3 Q1 2021 Q3 Q1 2022 Q3 Q1 2023 Q3 Q1 2024 Q3\n\nIndex (2015 = 100)\n\nMedium-high and high tech Other manufacturing industries\n\n**Figure 3.1 |** Global index of manufacturing production, industries by technology level\n\nSignificant signs\n\nof growth of\n\n### pharmaceuticals\n\nin Q3 2024\n\nAs shown in Figure 3.2, the solid performance of higher-technology manufacturing was mostly driven by the production of pharmaceu- ticals (2.6 per cent), computers and electronics (2.1 per cent), and electrical equipment (1.0 per cent). Machinery, on the other hand, showed the worst performance in the current quarter among MHT industries.","pages":[{"line_numbers":[],"markdown":"# **Findings by industry groups 3**\n\n### Higher-technology\n\nindustries maintained their strong progress, outperforming other industries\n\nThe latest global trends of industrial sectors grouped by technolog- ical intensity are shown in Figure 3.1. The data reveal that higher- technology industries continued to exhibit strong dynamism, even thriving amidst numerous challenges.\n\nIn recent periods, industries classified as medium-high and high- technology (MHT) have generally outperformed other manufacturing sectors. While they experienced a temporary slowdown in the first quarter of 2024, they returned to growth in the next quarter, increasing output by 1.5 per cent. This upward trend continued into the third quarter of 2024 with a 1.0 per cent quarterly increase. Conversely, lower-technology manufacturing sectors remained flat.\n\n90\n\n100\n\n110\n\n120\n\n130\n\n140\n\nQ1 2015 Q3 Q1 2016 Q3 Q1 2017 Q3 Q1 2018 Q3 Q1 2019 Q3 Q1 2020 Q3 Q1 2021 Q3 Q1 2022 Q3 Q1 2023 Q3 Q1 2024 Q3\n\nIndex (2015 = 100)\n\nMedium-high and high tech Other manufacturing industries\n\n**Figure 3.1 |** Global index of manufacturing production, industries by technology level\n\nSignificant signs\n\nof growth of\n\n### pharmaceuticals\n\nin Q3 2024\n\nAs shown in Figure 3.2, the solid performance of higher-technology manufacturing was mostly driven by the production of pharmaceu- ticals (2.6 per cent), computers and electronics (2.1 per cent), and electrical equipment (1.0 per cent). Machinery, on the other hand, showed the worst performance in the current quarter among MHT industries.","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"# **Findings by industry groups 3**\n\n### Higher-technology\n\nindustries maintained their strong progress, outperforming other industries\n\nThe latest global trends of industrial sectors grouped by technolog- ical intensity are shown in Figure 3.1. The data reveal that higher- technology industries continued to exhibit strong dynamism, even thriving amidst numerous challenges.\n\nIn recent periods, industries classified as medium-high and high- technology (MHT) have generally outperformed other manufacturing sectors. While they experienced a temporary slowdown in the first quarter of 2024, they returned to growth in the next quarter, increasing output by 1.5 per cent. This upward trend continued into the third quarter of 2024 with a 1.0 per cent quarterly increase. Conversely, lower-technology manufacturing sectors remained flat.\n\n90\n\n100\n\n110\n\n120\n\n130\n\n140\n\nQ1 2015 Q3 Q1 2016 Q3 Q1 2017 Q3 Q1 2018 Q3 Q1 2019 Q3 Q1 2020 Q3 Q1 2021 Q3 Q1 2022 Q3 Q1 2023 Q3 Q1 2024 Q3\n\nIndex (2015 = 100)\n\nMedium-high and high tech Other manufacturing industries\n\n**Figure 3.1 |** Global index of manufacturing production, industries by technology level\n\nSignificant signs\n\nof growth of\n\n### pharmaceuticals\n\nin Q3 2024\n\nAs shown in Figure 3.2, the solid performance of higher-technology manufacturing was mostly driven by the production of pharmaceu- ticals (2.6 per cent), computers and electronics (2.1 per cent), and electrical equipment (1.0 per cent). Machinery, on the other hand, showed the worst performance in the current quarter among MHT industries.","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Manufacturing_Production_2024_Q3_p11.pdf"]},"response":{"errors":[],"extract_id":"extract_f8da2b877c86be385c7b002889444755","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["# **Findings by industry groups 3**\n\n### Higher-technology\n\nindustries maintained their strong progress, outperforming other industries\n\nThe latest global trends of industrial sectors grouped by technolog- ical intensity are shown in Figure 3.1. The data reveal that higher- technology industries continued to exhibit strong dynamism, even thriving amidst numerous challenges.\n\nIn recent periods, industries classified as medium-high and high- technology (MHT) have generally outperformed other manufacturing sectors. While they experienced a temporary slowdown in the first quarter of 2024, they returned to growth in the next quarter, increasing output by 1.5 per cent. This upward trend continued into the third quarter of 2024 with a 1.0 per cent quarterly increase. Conversely, lower-technology manufacturing sectors remained flat.\n\n90\n\n100\n\n110\n\n120\n\n130\n\n140\n\nQ1 2015 Q3 Q1 2016 Q3 Q1 2017 Q3 Q1 2018 Q3 Q1 2019 Q3 Q1 2020 Q3 Q1 2021 Q3 Q1 2022 Q3 Q1 2023 Q3 Q1 2024 Q3\n\nIndex (2015 = 100)\n\nMedium-high and high tech Other manufacturing industries\n\n**Figure 3.1 |** Global index of manufacturing production, industries by technology level\n\nSignificant signs\n\nof growth of\n\n### pharmaceuticals\n\nin Q3 2024\n\nAs shown in Figure 3.2, the solid performance of higher-technology manufacturing was mostly driven by the production of pharmaceu- ticals (2.6 per cent), computers and electronics (2.1 per cent), and electrical equipment (1.0 per cent). Machinery, on the other hand, showed the worst performance in the current quarter among MHT industries."],"full_content":"# **Findings by industry groups 3**\n\n### Higher-technology\n\nindustries maintained their strong progress, outperforming other industries\n\nThe latest global trends of industrial sectors grouped by technolog- ical intensity are shown in Figure 3.1. The data reveal that higher- technology industries continued to exhibit strong dynamism, even thriving amidst numerous challenges.\n\nIn recent periods, industries classified as medium-high and high- technology (MHT) have generally outperformed other manufacturing sectors. While they experienced a temporary slowdown in the first quarter of 2024, they returned to growth in the next quarter, increasing output by 1.5 per cent. This upward trend continued into the third quarter of 2024 with a 1.0 per cent quarterly increase. Conversely, lower-technology manufacturing sectors remained flat.\n\n90\n\n100\n\n110\n\n120\n\n130\n\n140\n\nQ1 2015 Q3 Q1 2016 Q3 Q1 2017 Q3 Q1 2018 Q3 Q1 2019 Q3 Q1 2020 Q3 Q1 2021 Q3 Q1 2022 Q3 Q1 2023 Q3 Q1 2024 Q3\n\nIndex (2015 = 100)\n\nMedium-high and high tech Other manufacturing industries\n\n**Figure 3.1 |** Global index of manufacturing production, industries by technology level\n\nSignificant signs\n\nof growth of\n\n### pharmaceuticals\n\nin Q3 2024\n\nAs shown in Figure 3.2, the solid performance of higher-technology manufacturing was mostly driven by the production of pharmaceu- ticals (2.6 per cent), computers and electronics (2.1 per cent), and electrical equipment (1.0 per cent). Machinery, on the other hand, showed the worst performance in the current quarter among MHT industries.","publish_date":null,"title":"3","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Manufacturing_Production_2024_Q3_p11.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/World_Manufacturing_Production_2024_Q3_p11.pdf"},"request":{"example_id":"chart/World_Manufacturing_Production_2024_Q3_p11","product_type":"parse","source_file_path":"World_Manufacturing_Production_2024_Q3_p11.pdf"},"started_at":"2026-11-04T19:46:01.445782"} +{"completed_at":"2026-11-04T19:46:07.408010","latency_in_ms":3751,"output":{"example_id":"chart/ac8b3538-en_p103","grounded_pages":[],"layout_pages":[],"markdown":" **101**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Figure 2.13. Many low-skilled, green-driven occupations have worse labour market security than other low-skill occupations**\n\nMarginal effect of labour market insecurity indicators on the share of green-driven occupations, by type of occupation and controlling for main individual and job (or last job) characteristics, 2015-19\n\nNote: The chart reports the point estimate (and 95% confidence intervals) of the percentage difference in the incidence of each type of green-driven occupation between unemployed and employed workers. Unemployed workers are classified according to their last occupation. Estimates are obtained from a linear regression with the inverse hyperbolic sine of the share of each type of green-driven occupation within all, high, medium or low-skilled occupations as dependent variable and including an unemployment dummy, the share of 1-digit ISCO occupations, educational attainment (3 classes), gender, age (3 classes), sectors (5 categories), time and country dummies as explanatory variables. Except for the case of all occupations, regressions are estimated separately for high, medium, and low-skill occupations. “Total” indicates the full sample with all occupations. Standard errors are clustered on the dimensions of variability of the dependent variable. The reported point estimates and confidence intervals refer to the estimated coefficient of the unemployment dummy and are expressed in percentage of the untransformed dependent variable. The sample excludes Agriculture and includes Austria, Belgium, Czechia, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Ireland, Italy, Latvia, Lithuania, Poland, Portugal, the Slovak Republic, Spain, Sweden, Switzerland, the United Kingdom and the United States. Unemployed workers with an unemployment spell equal to or longer than 7 years are excluded in all countries except in the United States. Reading: Controlling for demographic characteristics and job characteristics and restricting the sample to high-skill occupations, the percentage share of high-skill, green new and emerging occupations is, on average, 2.8% lower among the employed than among the unemployed (where data for the unemployed refer to the occupation in the last held job), although the difference is not statistically significant. A lower incidence of a given type of occupation among the employed than among the unemployed is indicative of lower labour market security for that type of occupation than for the other occupations included in the regression. Source: Secretariat’s estimates based on version 24.1 of the O\\*NET database and the Current Population Survey, for the United States and the EU Labour Force Surveys, for all other countries.\n\nStatLink 2 https://stat.link/yd270q\n\nOne factor that may explain the divergent patterns of labour market security among different types of green- driven occupations is the different incidence of fixed-term contracts. The evidence indeed suggests that, on average, fixed-term contracts are often associated with lower labour market security, and in particular a higher risk of job loss – see e.g. OECD (2014[15]). ⁵¹ Consistent with this hypothesis, the labour market insecurity pattern concerning low-skill occupations presented in Figure 2.13 is mirrored by similar results as regards the type of contract. Within low-skill occupations, fixed-term contracts are, on average, more frequent among green increased demand and new and emerging occupations (by 24% and 26%, respectively), while the opposite is true for green-enhanced skills occupations (see Figure 2.14). By contrast, the patterns depicted in Figure 2.13 and Figure 2.14 diverge as regards high-skill occupations. Whatever their type, high-skill, green-driven occupations are systematically less frequent among employees with fixed-term contracts but no less frequent among the unemployed than other high-skill occupations, suggesting that contract type is not the main driver of labour market security in these high- skill occupations. In other words, these results suggest that the typical worker in high-skill, green-driven occupations is more likely to hold an open-ended contract than workers in other high-skill occupations, and nevertheless has often no better labour market security. ⁵²\n\n\\-25\n\n\\-15\n\n\\-5\n\n5\n\n15\n\n25\n\n35\n\n45\n\nTotal High-skill Medium-skill Low-skill Total High-skill Medium-skill Low-skill Total High-skill Medium-skill Low-skill\n\nGreen new and emerging occupations Green-enhanced skills occupations Green increased demand occupations\n\n%","pages":[{"line_numbers":[],"markdown":" **101**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Figure 2.13. Many low-skilled, green-driven occupations have worse labour market security than other low-skill occupations**\n\nMarginal effect of labour market insecurity indicators on the share of green-driven occupations, by type of occupation and controlling for main individual and job (or last job) characteristics, 2015-19\n\nNote: The chart reports the point estimate (and 95% confidence intervals) of the percentage difference in the incidence of each type of green-driven occupation between unemployed and employed workers. Unemployed workers are classified according to their last occupation. Estimates are obtained from a linear regression with the inverse hyperbolic sine of the share of each type of green-driven occupation within all, high, medium or low-skilled occupations as dependent variable and including an unemployment dummy, the share of 1-digit ISCO occupations, educational attainment (3 classes), gender, age (3 classes), sectors (5 categories), time and country dummies as explanatory variables. Except for the case of all occupations, regressions are estimated separately for high, medium, and low-skill occupations. “Total” indicates the full sample with all occupations. Standard errors are clustered on the dimensions of variability of the dependent variable. The reported point estimates and confidence intervals refer to the estimated coefficient of the unemployment dummy and are expressed in percentage of the untransformed dependent variable. The sample excludes Agriculture and includes Austria, Belgium, Czechia, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Ireland, Italy, Latvia, Lithuania, Poland, Portugal, the Slovak Republic, Spain, Sweden, Switzerland, the United Kingdom and the United States. Unemployed workers with an unemployment spell equal to or longer than 7 years are excluded in all countries except in the United States. Reading: Controlling for demographic characteristics and job characteristics and restricting the sample to high-skill occupations, the percentage share of high-skill, green new and emerging occupations is, on average, 2.8% lower among the employed than among the unemployed (where data for the unemployed refer to the occupation in the last held job), although the difference is not statistically significant. A lower incidence of a given type of occupation among the employed than among the unemployed is indicative of lower labour market security for that type of occupation than for the other occupations included in the regression. Source: Secretariat’s estimates based on version 24.1 of the O\\*NET database and the Current Population Survey, for the United States and the EU Labour Force Surveys, for all other countries.\n\nStatLink 2 https://stat.link/yd270q\n\nOne factor that may explain the divergent patterns of labour market security among different types of green- driven occupations is the different incidence of fixed-term contracts. The evidence indeed suggests that, on average, fixed-term contracts are often associated with lower labour market security, and in particular a higher risk of job loss – see e.g. OECD (2014[15]). ⁵¹ Consistent with this hypothesis, the labour market insecurity pattern concerning low-skill occupations presented in Figure 2.13 is mirrored by similar results as regards the type of contract. Within low-skill occupations, fixed-term contracts are, on average, more frequent among green increased demand and new and emerging occupations (by 24% and 26%, respectively), while the opposite is true for green-enhanced skills occupations (see Figure 2.14). By contrast, the patterns depicted in Figure 2.13 and Figure 2.14 diverge as regards high-skill occupations. Whatever their type, high-skill, green-driven occupations are systematically less frequent among employees with fixed-term contracts but no less frequent among the unemployed than other high-skill occupations, suggesting that contract type is not the main driver of labour market security in these high- skill occupations. In other words, these results suggest that the typical worker in high-skill, green-driven occupations is more likely to hold an open-ended contract than workers in other high-skill occupations, and nevertheless has often no better labour market security. ⁵²\n\n\\-25\n\n\\-15\n\n\\-5\n\n5\n\n15\n\n25\n\n35\n\n45\n\nTotal High-skill Medium-skill Low-skill Total High-skill Medium-skill Low-skill Total High-skill Medium-skill Low-skill\n\nGreen new and emerging occupations Green-enhanced skills occupations Green increased demand occupations\n\n%","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":" **101**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Figure 2.13. Many low-skilled, green-driven occupations have worse labour market security than other low-skill occupations**\n\nMarginal effect of labour market insecurity indicators on the share of green-driven occupations, by type of occupation and controlling for main individual and job (or last job) characteristics, 2015-19\n\nNote: The chart reports the point estimate (and 95% confidence intervals) of the percentage difference in the incidence of each type of green-driven occupation between unemployed and employed workers. Unemployed workers are classified according to their last occupation. Estimates are obtained from a linear regression with the inverse hyperbolic sine of the share of each type of green-driven occupation within all, high, medium or low-skilled occupations as dependent variable and including an unemployment dummy, the share of 1-digit ISCO occupations, educational attainment (3 classes), gender, age (3 classes), sectors (5 categories), time and country dummies as explanatory variables. Except for the case of all occupations, regressions are estimated separately for high, medium, and low-skill occupations. “Total” indicates the full sample with all occupations. Standard errors are clustered on the dimensions of variability of the dependent variable. The reported point estimates and confidence intervals refer to the estimated coefficient of the unemployment dummy and are expressed in percentage of the untransformed dependent variable. The sample excludes Agriculture and includes Austria, Belgium, Czechia, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Ireland, Italy, Latvia, Lithuania, Poland, Portugal, the Slovak Republic, Spain, Sweden, Switzerland, the United Kingdom and the United States. Unemployed workers with an unemployment spell equal to or longer than 7 years are excluded in all countries except in the United States. Reading: Controlling for demographic characteristics and job characteristics and restricting the sample to high-skill occupations, the percentage share of high-skill, green new and emerging occupations is, on average, 2.8% lower among the employed than among the unemployed (where data for the unemployed refer to the occupation in the last held job), although the difference is not statistically significant. A lower incidence of a given type of occupation among the employed than among the unemployed is indicative of lower labour market security for that type of occupation than for the other occupations included in the regression. Source: Secretariat’s estimates based on version 24.1 of the O\\*NET database and the Current Population Survey, for the United States and the EU Labour Force Surveys, for all other countries.\n\nStatLink 2 https://stat.link/yd270q\n\nOne factor that may explain the divergent patterns of labour market security among different types of green- driven occupations is the different incidence of fixed-term contracts. The evidence indeed suggests that, on average, fixed-term contracts are often associated with lower labour market security, and in particular a higher risk of job loss – see e.g. OECD (2014[15]). ⁵¹ Consistent with this hypothesis, the labour market insecurity pattern concerning low-skill occupations presented in Figure 2.13 is mirrored by similar results as regards the type of contract. Within low-skill occupations, fixed-term contracts are, on average, more frequent among green increased demand and new and emerging occupations (by 24% and 26%, respectively), while the opposite is true for green-enhanced skills occupations (see Figure 2.14). By contrast, the patterns depicted in Figure 2.13 and Figure 2.14 diverge as regards high-skill occupations. Whatever their type, high-skill, green-driven occupations are systematically less frequent among employees with fixed-term contracts but no less frequent among the unemployed than other high-skill occupations, suggesting that contract type is not the main driver of labour market security in these high- skill occupations. In other words, these results suggest that the typical worker in high-skill, green-driven occupations is more likely to hold an open-ended contract than workers in other high-skill occupations, and nevertheless has often no better labour market security. ⁵²\n\n\\-25\n\n\\-15\n\n\\-5\n\n5\n\n15\n\n25\n\n35\n\n45\n\nTotal High-skill Medium-skill Low-skill Total High-skill Medium-skill Low-skill Total High-skill Medium-skill Low-skill\n\nGreen new and emerging occupations Green-enhanced skills occupations Green increased demand occupations\n\n%","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p103.pdf"]},"response":{"errors":[],"extract_id":"extract_946a934f9fd8bc8a341fb5900376ec36","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":[" **101**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Figure 2.13. Many low-skilled, green-driven occupations have worse labour market security than other low-skill occupations**\n\nMarginal effect of labour market insecurity indicators on the share of green-driven occupations, by type of occupation and controlling for main individual and job (or last job) characteristics, 2015-19\n\nNote: The chart reports the point estimate (and 95% confidence intervals) of the percentage difference in the incidence of each type of green-driven occupation between unemployed and employed workers. Unemployed workers are classified according to their last occupation. Estimates are obtained from a linear regression with the inverse hyperbolic sine of the share of each type of green-driven occupation within all, high, medium or low-skilled occupations as dependent variable and including an unemployment dummy, the share of 1-digit ISCO occupations, educational attainment (3 classes), gender, age (3 classes), sectors (5 categories), time and country dummies as explanatory variables. Except for the case of all occupations, regressions are estimated separately for high, medium, and low-skill occupations. “Total” indicates the full sample with all occupations. Standard errors are clustered on the dimensions of variability of the dependent variable. The reported point estimates and confidence intervals refer to the estimated coefficient of the unemployment dummy and are expressed in percentage of the untransformed dependent variable. The sample excludes Agriculture and includes Austria, Belgium, Czechia, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Ireland, Italy, Latvia, Lithuania, Poland, Portugal, the Slovak Republic, Spain, Sweden, Switzerland, the United Kingdom and the United States. Unemployed workers with an unemployment spell equal to or longer than 7 years are excluded in all countries except in the United States. Reading: Controlling for demographic characteristics and job characteristics and restricting the sample to high-skill occupations, the percentage share of high-skill, green new and emerging occupations is, on average, 2.8% lower among the employed than among the unemployed (where data for the unemployed refer to the occupation in the last held job), although the difference is not statistically significant. A lower incidence of a given type of occupation among the employed than among the unemployed is indicative of lower labour market security for that type of occupation than for the other occupations included in the regression. Source: Secretariat’s estimates based on version 24.1 of the O\\*NET database and the Current Population Survey, for the United States and the EU Labour Force Surveys, for all other countries.\n\nStatLink 2 https://stat.link/yd270q\n\nOne factor that may explain the divergent patterns of labour market security among different types of green- driven occupations is the different incidence of fixed-term contracts. The evidence indeed suggests that, on average, fixed-term contracts are often associated with lower labour market security, and in particular a higher risk of job loss – see e.g. OECD (2014[15]). ⁵¹ Consistent with this hypothesis, the labour market insecurity pattern concerning low-skill occupations presented in Figure 2.13 is mirrored by similar results as regards the type of contract. Within low-skill occupations, fixed-term contracts are, on average, more frequent among green increased demand and new and emerging occupations (by 24% and 26%, respectively), while the opposite is true for green-enhanced skills occupations (see Figure 2.14). By contrast, the patterns depicted in Figure 2.13 and Figure 2.14 diverge as regards high-skill occupations. Whatever their type, high-skill, green-driven occupations are systematically less frequent among employees with fixed-term contracts but no less frequent among the unemployed than other high-skill occupations, suggesting that contract type is not the main driver of labour market security in these high- skill occupations. In other words, these results suggest that the typical worker in high-skill, green-driven occupations is more likely to hold an open-ended contract than workers in other high-skill occupations, and nevertheless has often no better labour market security. ⁵²\n\n\\-25\n\n\\-15\n\n\\-5\n\n5\n\n15\n\n25\n\n35\n\n45\n\nTotal High-skill Medium-skill Low-skill Total High-skill Medium-skill Low-skill Total High-skill Medium-skill Low-skill\n\nGreen new and emerging occupations Green-enhanced skills occupations Green increased demand occupations\n\n%"],"full_content":" **101**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Figure 2.13. Many low-skilled, green-driven occupations have worse labour market security than other low-skill occupations**\n\nMarginal effect of labour market insecurity indicators on the share of green-driven occupations, by type of occupation and controlling for main individual and job (or last job) characteristics, 2015-19\n\nNote: The chart reports the point estimate (and 95% confidence intervals) of the percentage difference in the incidence of each type of green-driven occupation between unemployed and employed workers. Unemployed workers are classified according to their last occupation. Estimates are obtained from a linear regression with the inverse hyperbolic sine of the share of each type of green-driven occupation within all, high, medium or low-skilled occupations as dependent variable and including an unemployment dummy, the share of 1-digit ISCO occupations, educational attainment (3 classes), gender, age (3 classes), sectors (5 categories), time and country dummies as explanatory variables. Except for the case of all occupations, regressions are estimated separately for high, medium, and low-skill occupations. “Total” indicates the full sample with all occupations. Standard errors are clustered on the dimensions of variability of the dependent variable. The reported point estimates and confidence intervals refer to the estimated coefficient of the unemployment dummy and are expressed in percentage of the untransformed dependent variable. The sample excludes Agriculture and includes Austria, Belgium, Czechia, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Ireland, Italy, Latvia, Lithuania, Poland, Portugal, the Slovak Republic, Spain, Sweden, Switzerland, the United Kingdom and the United States. Unemployed workers with an unemployment spell equal to or longer than 7 years are excluded in all countries except in the United States. Reading: Controlling for demographic characteristics and job characteristics and restricting the sample to high-skill occupations, the percentage share of high-skill, green new and emerging occupations is, on average, 2.8% lower among the employed than among the unemployed (where data for the unemployed refer to the occupation in the last held job), although the difference is not statistically significant. A lower incidence of a given type of occupation among the employed than among the unemployed is indicative of lower labour market security for that type of occupation than for the other occupations included in the regression. Source: Secretariat’s estimates based on version 24.1 of the O\\*NET database and the Current Population Survey, for the United States and the EU Labour Force Surveys, for all other countries.\n\nStatLink 2 https://stat.link/yd270q\n\nOne factor that may explain the divergent patterns of labour market security among different types of green- driven occupations is the different incidence of fixed-term contracts. The evidence indeed suggests that, on average, fixed-term contracts are often associated with lower labour market security, and in particular a higher risk of job loss – see e.g. OECD (2014[15]). ⁵¹ Consistent with this hypothesis, the labour market insecurity pattern concerning low-skill occupations presented in Figure 2.13 is mirrored by similar results as regards the type of contract. Within low-skill occupations, fixed-term contracts are, on average, more frequent among green increased demand and new and emerging occupations (by 24% and 26%, respectively), while the opposite is true for green-enhanced skills occupations (see Figure 2.14). By contrast, the patterns depicted in Figure 2.13 and Figure 2.14 diverge as regards high-skill occupations. Whatever their type, high-skill, green-driven occupations are systematically less frequent among employees with fixed-term contracts but no less frequent among the unemployed than other high-skill occupations, suggesting that contract type is not the main driver of labour market security in these high- skill occupations. In other words, these results suggest that the typical worker in high-skill, green-driven occupations is more likely to hold an open-ended contract than workers in other high-skill occupations, and nevertheless has often no better labour market security. ⁵²\n\n\\-25\n\n\\-15\n\n\\-5\n\n5\n\n15\n\n25\n\n35\n\n45\n\nTotal High-skill Medium-skill Low-skill Total High-skill Medium-skill Low-skill Total High-skill Medium-skill Low-skill\n\nGreen new and emerging occupations Green-enhanced skills occupations Green increased demand occupations\n\n%","publish_date":null,"title":"Figure 2.13. Many low-skilled, green-driven occupations have worse labour market security than other low-skill occupations","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p103.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p103.pdf"},"request":{"example_id":"chart/ac8b3538-en_p103","product_type":"parse","source_file_path":"ac8b3538-en_p103.pdf"},"started_at":"2026-11-04T19:46:03.656990"} +{"completed_at":"2026-11-04T19:46:08.603828","latency_in_ms":3142,"output":{"example_id":"chart/ac8b3538-en_p104","grounded_pages":[],"layout_pages":[],"markdown":"**102** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Figure 2.14. Among high-skill jobs, green-driven occupations have fewer fixed-term contracts**\n\nMarginal effect of the fixed-term contract indicator on the share of green-driven occupations, by type of occupation and controlling for main individual and job (or last job) characteristics, 2015-19\n\nNote: The chart reports the point estimate (and 95% confidence intervals) of the percentage difference in the incidence of each type of green- driven occupation between employees on fixed-term and open-ended contracts. Estimates are obtained from a linear regression with the inverse hyperbolic sine of the share of each type of green-driven occupation within all, high, medium or low-skilled occupations as dependent variable and including a fixed-term contract dummy, the share of 1-digit ISCO occupations, educational attainment (3 classes), gender, age (3 classes), sectors (5 categories), time and country dummies as explanatory variables. Except for the case of all occupations, regressions are estimated separately for high, medium, and low-skill occupations. “Total” indicates the full sample with all occupations. Standard errors are clustered on the dimensions of variability of the dependent variable. The reported point estimates and confidence intervals refer to the estimated coefficient of the fixed-term contract dummy and are expressed in percentage of the untransformed dependent variable. The sample excludes Agriculture and includes Austria, Belgium, Czechia, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Iceland, Ireland, Italy, Latvia, Lithuania, Luxembourg, the Netherlands, Norway, Poland, Portugal, the Slovak Republic, Slovenia, Spain, Sweden, Switzerland and the United Kingdom. Reading: Controlling for demographic characteristics and restricting the sample to high-skill occupations, the percentage share of high-skill, green new and emerging occupations is, on average, 39% higher among employees on open-ended than fixed-term contract. A higher incidence of a given type of occupation among employees on open-ended contract is indicative that the share of fixed-term contracts within that type of occupation is lower than in the other occupations included in the regression. Source: Secretariat’s estimates based on version 24.1 of the O\\*NET database and the Current Population Survey, for the United States and the EU Labour Force Surveys, for all other countries.\n\nStatLink 2 https://stat.link/xc4g6h\n\nOne possible explanation of these results is that high-skill workers in green-driven occupations tend to have high-level positions in innovative startups or activities characterised by high average growth but also high failure rates and shakeouts, and therefore a relatively high risk of job destruction. Wage premia, however, likely compensate for the lack of labour market security. Moreover, high-skill, green-enhanced skills occupations are likely characterised by high rates of restructuring due to changing competences and work requirements – see Section 2.1.2. Workers unable to adapt to the new requirements may therefore find it difficult to maintain their employability. ⁵³ At the same time, some other (not green-driven) high-skill occupations, such as health professionals or software developers, have a relatively high share of fixed- term contracts but still enjoy high labour market security (low unemployment rates) since they face strong labour demand. By contrast, the situation of workers in low-skill occupations is likely different: like their negative wage gap (see Section 2.3.1), their higher unemployment risk and incidence of fixed-term contracts probably reflect the weakness of their position in the labour market.\n\n\\-50\n\n\\-40\n\n\\-30\n\n\\-20\n\n\\-10\n\n0\n\n10\n\n20\n\n30\n\n40\n\nTotal High-skill Medium-skill Low-skill Total High-skill Medium-skill Low-skill Total High-skill Medium-skill Low-skill\n\nGreen new and emerging occupations Green-enhanced skills occupations Green increased demand occupations\n\n%","pages":[{"line_numbers":[],"markdown":"**102** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Figure 2.14. Among high-skill jobs, green-driven occupations have fewer fixed-term contracts**\n\nMarginal effect of the fixed-term contract indicator on the share of green-driven occupations, by type of occupation and controlling for main individual and job (or last job) characteristics, 2015-19\n\nNote: The chart reports the point estimate (and 95% confidence intervals) of the percentage difference in the incidence of each type of green- driven occupation between employees on fixed-term and open-ended contracts. Estimates are obtained from a linear regression with the inverse hyperbolic sine of the share of each type of green-driven occupation within all, high, medium or low-skilled occupations as dependent variable and including a fixed-term contract dummy, the share of 1-digit ISCO occupations, educational attainment (3 classes), gender, age (3 classes), sectors (5 categories), time and country dummies as explanatory variables. Except for the case of all occupations, regressions are estimated separately for high, medium, and low-skill occupations. “Total” indicates the full sample with all occupations. Standard errors are clustered on the dimensions of variability of the dependent variable. The reported point estimates and confidence intervals refer to the estimated coefficient of the fixed-term contract dummy and are expressed in percentage of the untransformed dependent variable. The sample excludes Agriculture and includes Austria, Belgium, Czechia, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Iceland, Ireland, Italy, Latvia, Lithuania, Luxembourg, the Netherlands, Norway, Poland, Portugal, the Slovak Republic, Slovenia, Spain, Sweden, Switzerland and the United Kingdom. Reading: Controlling for demographic characteristics and restricting the sample to high-skill occupations, the percentage share of high-skill, green new and emerging occupations is, on average, 39% higher among employees on open-ended than fixed-term contract. A higher incidence of a given type of occupation among employees on open-ended contract is indicative that the share of fixed-term contracts within that type of occupation is lower than in the other occupations included in the regression. Source: Secretariat’s estimates based on version 24.1 of the O\\*NET database and the Current Population Survey, for the United States and the EU Labour Force Surveys, for all other countries.\n\nStatLink 2 https://stat.link/xc4g6h\n\nOne possible explanation of these results is that high-skill workers in green-driven occupations tend to have high-level positions in innovative startups or activities characterised by high average growth but also high failure rates and shakeouts, and therefore a relatively high risk of job destruction. Wage premia, however, likely compensate for the lack of labour market security. Moreover, high-skill, green-enhanced skills occupations are likely characterised by high rates of restructuring due to changing competences and work requirements – see Section 2.1.2. Workers unable to adapt to the new requirements may therefore find it difficult to maintain their employability. ⁵³ At the same time, some other (not green-driven) high-skill occupations, such as health professionals or software developers, have a relatively high share of fixed- term contracts but still enjoy high labour market security (low unemployment rates) since they face strong labour demand. By contrast, the situation of workers in low-skill occupations is likely different: like their negative wage gap (see Section 2.3.1), their higher unemployment risk and incidence of fixed-term contracts probably reflect the weakness of their position in the labour market.\n\n\\-50\n\n\\-40\n\n\\-30\n\n\\-20\n\n\\-10\n\n0\n\n10\n\n20\n\n30\n\n40\n\nTotal High-skill Medium-skill Low-skill Total High-skill Medium-skill Low-skill Total High-skill Medium-skill Low-skill\n\nGreen new and emerging occupations Green-enhanced skills occupations Green increased demand occupations\n\n%","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**102** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Figure 2.14. Among high-skill jobs, green-driven occupations have fewer fixed-term contracts**\n\nMarginal effect of the fixed-term contract indicator on the share of green-driven occupations, by type of occupation and controlling for main individual and job (or last job) characteristics, 2015-19\n\nNote: The chart reports the point estimate (and 95% confidence intervals) of the percentage difference in the incidence of each type of green- driven occupation between employees on fixed-term and open-ended contracts. Estimates are obtained from a linear regression with the inverse hyperbolic sine of the share of each type of green-driven occupation within all, high, medium or low-skilled occupations as dependent variable and including a fixed-term contract dummy, the share of 1-digit ISCO occupations, educational attainment (3 classes), gender, age (3 classes), sectors (5 categories), time and country dummies as explanatory variables. Except for the case of all occupations, regressions are estimated separately for high, medium, and low-skill occupations. “Total” indicates the full sample with all occupations. Standard errors are clustered on the dimensions of variability of the dependent variable. The reported point estimates and confidence intervals refer to the estimated coefficient of the fixed-term contract dummy and are expressed in percentage of the untransformed dependent variable. The sample excludes Agriculture and includes Austria, Belgium, Czechia, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Iceland, Ireland, Italy, Latvia, Lithuania, Luxembourg, the Netherlands, Norway, Poland, Portugal, the Slovak Republic, Slovenia, Spain, Sweden, Switzerland and the United Kingdom. Reading: Controlling for demographic characteristics and restricting the sample to high-skill occupations, the percentage share of high-skill, green new and emerging occupations is, on average, 39% higher among employees on open-ended than fixed-term contract. A higher incidence of a given type of occupation among employees on open-ended contract is indicative that the share of fixed-term contracts within that type of occupation is lower than in the other occupations included in the regression. Source: Secretariat’s estimates based on version 24.1 of the O\\*NET database and the Current Population Survey, for the United States and the EU Labour Force Surveys, for all other countries.\n\nStatLink 2 https://stat.link/xc4g6h\n\nOne possible explanation of these results is that high-skill workers in green-driven occupations tend to have high-level positions in innovative startups or activities characterised by high average growth but also high failure rates and shakeouts, and therefore a relatively high risk of job destruction. Wage premia, however, likely compensate for the lack of labour market security. Moreover, high-skill, green-enhanced skills occupations are likely characterised by high rates of restructuring due to changing competences and work requirements – see Section 2.1.2. Workers unable to adapt to the new requirements may therefore find it difficult to maintain their employability. ⁵³ At the same time, some other (not green-driven) high-skill occupations, such as health professionals or software developers, have a relatively high share of fixed- term contracts but still enjoy high labour market security (low unemployment rates) since they face strong labour demand. By contrast, the situation of workers in low-skill occupations is likely different: like their negative wage gap (see Section 2.3.1), their higher unemployment risk and incidence of fixed-term contracts probably reflect the weakness of their position in the labour market.\n\n\\-50\n\n\\-40\n\n\\-30\n\n\\-20\n\n\\-10\n\n0\n\n10\n\n20\n\n30\n\n40\n\nTotal High-skill Medium-skill Low-skill Total High-skill Medium-skill Low-skill Total High-skill Medium-skill Low-skill\n\nGreen new and emerging occupations Green-enhanced skills occupations Green increased demand occupations\n\n%","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p104.pdf"]},"response":{"errors":[],"extract_id":"extract_53ce9bd254aedbb29cd6a255f9e52fd3","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**102** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Figure 2.14. Among high-skill jobs, green-driven occupations have fewer fixed-term contracts**\n\nMarginal effect of the fixed-term contract indicator on the share of green-driven occupations, by type of occupation and controlling for main individual and job (or last job) characteristics, 2015-19\n\nNote: The chart reports the point estimate (and 95% confidence intervals) of the percentage difference in the incidence of each type of green- driven occupation between employees on fixed-term and open-ended contracts. Estimates are obtained from a linear regression with the inverse hyperbolic sine of the share of each type of green-driven occupation within all, high, medium or low-skilled occupations as dependent variable and including a fixed-term contract dummy, the share of 1-digit ISCO occupations, educational attainment (3 classes), gender, age (3 classes), sectors (5 categories), time and country dummies as explanatory variables. Except for the case of all occupations, regressions are estimated separately for high, medium, and low-skill occupations. “Total” indicates the full sample with all occupations. Standard errors are clustered on the dimensions of variability of the dependent variable. The reported point estimates and confidence intervals refer to the estimated coefficient of the fixed-term contract dummy and are expressed in percentage of the untransformed dependent variable. The sample excludes Agriculture and includes Austria, Belgium, Czechia, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Iceland, Ireland, Italy, Latvia, Lithuania, Luxembourg, the Netherlands, Norway, Poland, Portugal, the Slovak Republic, Slovenia, Spain, Sweden, Switzerland and the United Kingdom. Reading: Controlling for demographic characteristics and restricting the sample to high-skill occupations, the percentage share of high-skill, green new and emerging occupations is, on average, 39% higher among employees on open-ended than fixed-term contract. A higher incidence of a given type of occupation among employees on open-ended contract is indicative that the share of fixed-term contracts within that type of occupation is lower than in the other occupations included in the regression. Source: Secretariat’s estimates based on version 24.1 of the O\\*NET database and the Current Population Survey, for the United States and the EU Labour Force Surveys, for all other countries.\n\nStatLink 2 https://stat.link/xc4g6h\n\nOne possible explanation of these results is that high-skill workers in green-driven occupations tend to have high-level positions in innovative startups or activities characterised by high average growth but also high failure rates and shakeouts, and therefore a relatively high risk of job destruction. Wage premia, however, likely compensate for the lack of labour market security. Moreover, high-skill, green-enhanced skills occupations are likely characterised by high rates of restructuring due to changing competences and work requirements – see Section 2.1.2. Workers unable to adapt to the new requirements may therefore find it difficult to maintain their employability. ⁵³ At the same time, some other (not green-driven) high-skill occupations, such as health professionals or software developers, have a relatively high share of fixed- term contracts but still enjoy high labour market security (low unemployment rates) since they face strong labour demand. By contrast, the situation of workers in low-skill occupations is likely different: like their negative wage gap (see Section 2.3.1), their higher unemployment risk and incidence of fixed-term contracts probably reflect the weakness of their position in the labour market.\n\n\\-50\n\n\\-40\n\n\\-30\n\n\\-20\n\n\\-10\n\n0\n\n10\n\n20\n\n30\n\n40\n\nTotal High-skill Medium-skill Low-skill Total High-skill Medium-skill Low-skill Total High-skill Medium-skill Low-skill\n\nGreen new and emerging occupations Green-enhanced skills occupations Green increased demand occupations\n\n%"],"full_content":"**102** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Figure 2.14. Among high-skill jobs, green-driven occupations have fewer fixed-term contracts**\n\nMarginal effect of the fixed-term contract indicator on the share of green-driven occupations, by type of occupation and controlling for main individual and job (or last job) characteristics, 2015-19\n\nNote: The chart reports the point estimate (and 95% confidence intervals) of the percentage difference in the incidence of each type of green- driven occupation between employees on fixed-term and open-ended contracts. Estimates are obtained from a linear regression with the inverse hyperbolic sine of the share of each type of green-driven occupation within all, high, medium or low-skilled occupations as dependent variable and including a fixed-term contract dummy, the share of 1-digit ISCO occupations, educational attainment (3 classes), gender, age (3 classes), sectors (5 categories), time and country dummies as explanatory variables. Except for the case of all occupations, regressions are estimated separately for high, medium, and low-skill occupations. “Total” indicates the full sample with all occupations. Standard errors are clustered on the dimensions of variability of the dependent variable. The reported point estimates and confidence intervals refer to the estimated coefficient of the fixed-term contract dummy and are expressed in percentage of the untransformed dependent variable. The sample excludes Agriculture and includes Austria, Belgium, Czechia, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Iceland, Ireland, Italy, Latvia, Lithuania, Luxembourg, the Netherlands, Norway, Poland, Portugal, the Slovak Republic, Slovenia, Spain, Sweden, Switzerland and the United Kingdom. Reading: Controlling for demographic characteristics and restricting the sample to high-skill occupations, the percentage share of high-skill, green new and emerging occupations is, on average, 39% higher among employees on open-ended than fixed-term contract. A higher incidence of a given type of occupation among employees on open-ended contract is indicative that the share of fixed-term contracts within that type of occupation is lower than in the other occupations included in the regression. Source: Secretariat’s estimates based on version 24.1 of the O\\*NET database and the Current Population Survey, for the United States and the EU Labour Force Surveys, for all other countries.\n\nStatLink 2 https://stat.link/xc4g6h\n\nOne possible explanation of these results is that high-skill workers in green-driven occupations tend to have high-level positions in innovative startups or activities characterised by high average growth but also high failure rates and shakeouts, and therefore a relatively high risk of job destruction. Wage premia, however, likely compensate for the lack of labour market security. Moreover, high-skill, green-enhanced skills occupations are likely characterised by high rates of restructuring due to changing competences and work requirements – see Section 2.1.2. Workers unable to adapt to the new requirements may therefore find it difficult to maintain their employability. ⁵³ At the same time, some other (not green-driven) high-skill occupations, such as health professionals or software developers, have a relatively high share of fixed- term contracts but still enjoy high labour market security (low unemployment rates) since they face strong labour demand. By contrast, the situation of workers in low-skill occupations is likely different: like their negative wage gap (see Section 2.3.1), their higher unemployment risk and incidence of fixed-term contracts probably reflect the weakness of their position in the labour market.\n\n\\-50\n\n\\-40\n\n\\-30\n\n\\-20\n\n\\-10\n\n0\n\n10\n\n20\n\n30\n\n40\n\nTotal High-skill Medium-skill Low-skill Total High-skill Medium-skill Low-skill Total High-skill Medium-skill Low-skill\n\nGreen new and emerging occupations Green-enhanced skills occupations Green increased demand occupations\n\n%","publish_date":null,"title":"Figure 2.14. Among high-skill jobs, green-driven occupations have fewer fixed-term contracts","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p104.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p104.pdf"},"request":{"example_id":"chart/ac8b3538-en_p104","product_type":"parse","source_file_path":"ac8b3538-en_p104.pdf"},"started_at":"2026-11-04T19:46:05.461111"} +{"completed_at":"2026-11-04T19:46:08.403558","latency_in_ms":1441,"output":{"example_id":"chart/ac8b3538-en_p119","grounded_pages":[],"layout_pages":[],"markdown":" **117**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n# **Annex 2.C.** **Additional figures**\n\n### **Annex Figure 2.C.1. One in eight green-driven occupations is concentrated in GHG-intensive industries**\n\nPercentage of total employment, average 2015-19\n\nNote: Data refer to the average for 2015-19, except for Canada: 2017-19 and New Zealand: 2018. OECD: Unweighted average of countries shown. GHG: greenhouse gases. Source: Secretariat’s estimates based on version 24.1 of the O\\*NET database and the following country-specific sources: Australian Labour Force Survey; Canadian Labour Force Survey; Japanese Labour Force Survey; United States: Current Population Survey; All other countries: EU Labour Force Survey.\n\nStatLink 2 https://stat.link/urlw1n\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n% Mixed green-driven and GHG-intensive occupations Residual green-driven occupations","pages":[{"line_numbers":[],"markdown":" **117**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n# **Annex 2.C.** **Additional figures**\n\n### **Annex Figure 2.C.1. One in eight green-driven occupations is concentrated in GHG-intensive industries**\n\nPercentage of total employment, average 2015-19\n\nNote: Data refer to the average for 2015-19, except for Canada: 2017-19 and New Zealand: 2018. OECD: Unweighted average of countries shown. GHG: greenhouse gases. Source: Secretariat’s estimates based on version 24.1 of the O\\*NET database and the following country-specific sources: Australian Labour Force Survey; Canadian Labour Force Survey; Japanese Labour Force Survey; United States: Current Population Survey; All other countries: EU Labour Force Survey.\n\nStatLink 2 https://stat.link/urlw1n\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n% Mixed green-driven and GHG-intensive occupations Residual green-driven occupations","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":" **117**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n# **Annex 2.C.** **Additional figures**\n\n### **Annex Figure 2.C.1. One in eight green-driven occupations is concentrated in GHG-intensive industries**\n\nPercentage of total employment, average 2015-19\n\nNote: Data refer to the average for 2015-19, except for Canada: 2017-19 and New Zealand: 2018. OECD: Unweighted average of countries shown. GHG: greenhouse gases. Source: Secretariat’s estimates based on version 24.1 of the O\\*NET database and the following country-specific sources: Australian Labour Force Survey; Canadian Labour Force Survey; Japanese Labour Force Survey; United States: Current Population Survey; All other countries: EU Labour Force Survey.\n\nStatLink 2 https://stat.link/urlw1n\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n% Mixed green-driven and GHG-intensive occupations Residual green-driven occupations","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p119.pdf"]},"response":{"errors":[],"extract_id":"extract_a62b491d9c08f322e13ae1fc7a6ff00e","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":[" **117**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n# **Annex 2.C.** **Additional figures**\n\n### **Annex Figure 2.C.1. One in eight green-driven occupations is concentrated in GHG-intensive industries**\n\nPercentage of total employment, average 2015-19\n\nNote: Data refer to the average for 2015-19, except for Canada: 2017-19 and New Zealand: 2018. OECD: Unweighted average of countries shown. GHG: greenhouse gases. Source: Secretariat’s estimates based on version 24.1 of the O\\*NET database and the following country-specific sources: Australian Labour Force Survey; Canadian Labour Force Survey; Japanese Labour Force Survey; United States: Current Population Survey; All other countries: EU Labour Force Survey.\n\nStatLink 2 https://stat.link/urlw1n\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n% Mixed green-driven and GHG-intensive occupations Residual green-driven occupations"],"full_content":" **117**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n# **Annex 2.C.** **Additional figures**\n\n### **Annex Figure 2.C.1. One in eight green-driven occupations is concentrated in GHG-intensive industries**\n\nPercentage of total employment, average 2015-19\n\nNote: Data refer to the average for 2015-19, except for Canada: 2017-19 and New Zealand: 2018. OECD: Unweighted average of countries shown. GHG: greenhouse gases. Source: Secretariat’s estimates based on version 24.1 of the O\\*NET database and the following country-specific sources: Australian Labour Force Survey; Canadian Labour Force Survey; Japanese Labour Force Survey; United States: Current Population Survey; All other countries: EU Labour Force Survey.\n\nStatLink 2 https://stat.link/urlw1n\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n% Mixed green-driven and GHG-intensive occupations Residual green-driven occupations","publish_date":null,"title":"Annex 2.C. Additional figures","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p119.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p119.pdf"},"request":{"example_id":"chart/ac8b3538-en_p119","product_type":"parse","source_file_path":"ac8b3538-en_p119.pdf"},"started_at":"2026-11-04T19:46:06.961934"} +{"completed_at":"2026-11-04T19:46:11.433193","latency_in_ms":3028,"output":{"example_id":"chart/ac8b3538-en_p125","grounded_pages":[],"layout_pages":[],"markdown":" **123**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\nSource: Source Secretariat’s estimates based on version 24.1 of the O\\*NET database and the following country-specific sources: United States: Current Population Survey; All other countries: EU Labour Force Survey.\n\nStatLink 2 https://stat.link/ai0tc1\n\n### **Annex Figure 2.C.7. GHG-intensive occupations are more likely low educated while green-driven occupations are more heterogeneous**\n\nPercentages, average 2015-19\n\nNote: Countries are ranked by decreasing gap of the share for high education compared to the total. Data refer to the average for 2015-19. Low education refers to less than upper secondary, medium education to upper secondary, high education to more than upper secondary. OECD: Unweighted average of countries shown. GHG: greenhouse gases. Source: Secretariat’s estimates based on version 24.1 of the O\\*NET database and the following country-specific sources: United States: Current Population Survey; All other countries: EU Labour Force Survey.\n\nStatLink 2 https://stat.link/6nria2\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n35\n\n40\n\n45\n\n50\n\n% **A. Green-driven occupations**\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n% **B. GHG-intensive occupations**\n\nHigh Medium Low Total","pages":[{"line_numbers":[],"markdown":" **123**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\nSource: Source Secretariat’s estimates based on version 24.1 of the O\\*NET database and the following country-specific sources: United States: Current Population Survey; All other countries: EU Labour Force Survey.\n\nStatLink 2 https://stat.link/ai0tc1\n\n### **Annex Figure 2.C.7. GHG-intensive occupations are more likely low educated while green-driven occupations are more heterogeneous**\n\nPercentages, average 2015-19\n\nNote: Countries are ranked by decreasing gap of the share for high education compared to the total. Data refer to the average for 2015-19. Low education refers to less than upper secondary, medium education to upper secondary, high education to more than upper secondary. OECD: Unweighted average of countries shown. GHG: greenhouse gases. Source: Secretariat’s estimates based on version 24.1 of the O\\*NET database and the following country-specific sources: United States: Current Population Survey; All other countries: EU Labour Force Survey.\n\nStatLink 2 https://stat.link/6nria2\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n35\n\n40\n\n45\n\n50\n\n% **A. Green-driven occupations**\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n% **B. GHG-intensive occupations**\n\nHigh Medium Low Total","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":" **123**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\nSource: Source Secretariat’s estimates based on version 24.1 of the O\\*NET database and the following country-specific sources: United States: Current Population Survey; All other countries: EU Labour Force Survey.\n\nStatLink 2 https://stat.link/ai0tc1\n\n### **Annex Figure 2.C.7. GHG-intensive occupations are more likely low educated while green-driven occupations are more heterogeneous**\n\nPercentages, average 2015-19\n\nNote: Countries are ranked by decreasing gap of the share for high education compared to the total. Data refer to the average for 2015-19. Low education refers to less than upper secondary, medium education to upper secondary, high education to more than upper secondary. OECD: Unweighted average of countries shown. GHG: greenhouse gases. Source: Secretariat’s estimates based on version 24.1 of the O\\*NET database and the following country-specific sources: United States: Current Population Survey; All other countries: EU Labour Force Survey.\n\nStatLink 2 https://stat.link/6nria2\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n35\n\n40\n\n45\n\n50\n\n% **A. Green-driven occupations**\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n% **B. GHG-intensive occupations**\n\nHigh Medium Low Total","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p125.pdf"]},"response":{"errors":[],"extract_id":"extract_c5004fa1865ebb09eee52e41aebc875c","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":[" **123**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\nSource: Source Secretariat’s estimates based on version 24.1 of the O\\*NET database and the following country-specific sources: United States: Current Population Survey; All other countries: EU Labour Force Survey.\n\nStatLink 2 https://stat.link/ai0tc1\n\n### **Annex Figure 2.C.7. GHG-intensive occupations are more likely low educated while green-driven occupations are more heterogeneous**\n\nPercentages, average 2015-19\n\nNote: Countries are ranked by decreasing gap of the share for high education compared to the total. Data refer to the average for 2015-19. Low education refers to less than upper secondary, medium education to upper secondary, high education to more than upper secondary. OECD: Unweighted average of countries shown. GHG: greenhouse gases. Source: Secretariat’s estimates based on version 24.1 of the O\\*NET database and the following country-specific sources: United States: Current Population Survey; All other countries: EU Labour Force Survey.\n\nStatLink 2 https://stat.link/6nria2\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n35\n\n40\n\n45\n\n50\n\n% **A. Green-driven occupations**\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n% **B. GHG-intensive occupations**\n\nHigh Medium Low Total"],"full_content":" **123**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\nSource: Source Secretariat’s estimates based on version 24.1 of the O\\*NET database and the following country-specific sources: United States: Current Population Survey; All other countries: EU Labour Force Survey.\n\nStatLink 2 https://stat.link/ai0tc1\n\n### **Annex Figure 2.C.7. GHG-intensive occupations are more likely low educated while green-driven occupations are more heterogeneous**\n\nPercentages, average 2015-19\n\nNote: Countries are ranked by decreasing gap of the share for high education compared to the total. Data refer to the average for 2015-19. Low education refers to less than upper secondary, medium education to upper secondary, high education to more than upper secondary. OECD: Unweighted average of countries shown. GHG: greenhouse gases. Source: Secretariat’s estimates based on version 24.1 of the O\\*NET database and the following country-specific sources: United States: Current Population Survey; All other countries: EU Labour Force Survey.\n\nStatLink 2 https://stat.link/6nria2\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n35\n\n40\n\n45\n\n50\n\n% **A. Green-driven occupations**\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n% **B. GHG-intensive occupations**\n\nHigh Medium Low Total","publish_date":null,"title":"Annex Figure 2.C.7. GHG-intensive occupations are more likely low educated while green-driven occupations are more heterogeneous","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p125.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p125.pdf"},"request":{"example_id":"chart/ac8b3538-en_p125","product_type":"parse","source_file_path":"ac8b3538-en_p125.pdf"},"started_at":"2026-11-04T19:46:08.404907"} +{"completed_at":"2026-11-04T19:46:12.579517","latency_in_ms":1922,"output":{"example_id":"chart/ac8b3538-en_p126","grounded_pages":[],"layout_pages":[],"markdown":"**124** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Annex Figure 2.C.8. The educational composition of different types of green-driven jobs differs significantly**\n\nPercentages, average 2015-19\n\nNote: Countries are ranked by decreasing gap of the share for high education compared to the total. Data refer to the average for 2015-19. Low education refers to less than upper secondary, medium education to upper secondary, high education to more than upper secondary. OECD: Unweighted average of countries shown. Source: Secretariat’s estimates based on version 24.1 of the O\\*NET database and the following country-specific sources: United States: Current Population Survey; All other countries: EU Labour Force Survey.\n\n0\n\n1\n\n2\n\n3\n\n4\n\n5\n\n6\n\n7\n\n8\n\n% **A. Green new and emerging occupations**\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n% **B. Green-enhanced skills occupations**\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\n16\n\n18\n\n20\n\n% **C. Green increased demand occupations**\n\nHigh Medium Low Total","pages":[{"line_numbers":[],"markdown":"**124** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Annex Figure 2.C.8. The educational composition of different types of green-driven jobs differs significantly**\n\nPercentages, average 2015-19\n\nNote: Countries are ranked by decreasing gap of the share for high education compared to the total. Data refer to the average for 2015-19. Low education refers to less than upper secondary, medium education to upper secondary, high education to more than upper secondary. OECD: Unweighted average of countries shown. Source: Secretariat’s estimates based on version 24.1 of the O\\*NET database and the following country-specific sources: United States: Current Population Survey; All other countries: EU Labour Force Survey.\n\n0\n\n1\n\n2\n\n3\n\n4\n\n5\n\n6\n\n7\n\n8\n\n% **A. Green new and emerging occupations**\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n% **B. Green-enhanced skills occupations**\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\n16\n\n18\n\n20\n\n% **C. Green increased demand occupations**\n\nHigh Medium Low Total","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**124** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Annex Figure 2.C.8. The educational composition of different types of green-driven jobs differs significantly**\n\nPercentages, average 2015-19\n\nNote: Countries are ranked by decreasing gap of the share for high education compared to the total. Data refer to the average for 2015-19. Low education refers to less than upper secondary, medium education to upper secondary, high education to more than upper secondary. OECD: Unweighted average of countries shown. Source: Secretariat’s estimates based on version 24.1 of the O\\*NET database and the following country-specific sources: United States: Current Population Survey; All other countries: EU Labour Force Survey.\n\n0\n\n1\n\n2\n\n3\n\n4\n\n5\n\n6\n\n7\n\n8\n\n% **A. Green new and emerging occupations**\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n% **B. Green-enhanced skills occupations**\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\n16\n\n18\n\n20\n\n% **C. Green increased demand occupations**\n\nHigh Medium Low Total","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p126.pdf"]},"response":{"errors":[],"extract_id":"extract_1be343a0221965d44c19ddf3546f0208","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**124** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Annex Figure 2.C.8. The educational composition of different types of green-driven jobs differs significantly**\n\nPercentages, average 2015-19\n\nNote: Countries are ranked by decreasing gap of the share for high education compared to the total. Data refer to the average for 2015-19. Low education refers to less than upper secondary, medium education to upper secondary, high education to more than upper secondary. OECD: Unweighted average of countries shown. Source: Secretariat’s estimates based on version 24.1 of the O\\*NET database and the following country-specific sources: United States: Current Population Survey; All other countries: EU Labour Force Survey.\n\n0\n\n1\n\n2\n\n3\n\n4\n\n5\n\n6\n\n7\n\n8\n\n% **A. Green new and emerging occupations**\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n% **B. Green-enhanced skills occupations**\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\n16\n\n18\n\n20\n\n% **C. Green increased demand occupations**\n\nHigh Medium Low Total"],"full_content":"**124** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Annex Figure 2.C.8. The educational composition of different types of green-driven jobs differs significantly**\n\nPercentages, average 2015-19\n\nNote: Countries are ranked by decreasing gap of the share for high education compared to the total. Data refer to the average for 2015-19. Low education refers to less than upper secondary, medium education to upper secondary, high education to more than upper secondary. OECD: Unweighted average of countries shown. Source: Secretariat’s estimates based on version 24.1 of the O\\*NET database and the following country-specific sources: United States: Current Population Survey; All other countries: EU Labour Force Survey.\n\n0\n\n1\n\n2\n\n3\n\n4\n\n5\n\n6\n\n7\n\n8\n\n% **A. Green new and emerging occupations**\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n% **B. Green-enhanced skills occupations**\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\n16\n\n18\n\n20\n\n% **C. Green increased demand occupations**\n\nHigh Medium Low Total","publish_date":null,"title":"Annex Figure 2.C.8. The educational composition of different types of green-driven jobs differs significantly","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p126.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p126.pdf"},"request":{"example_id":"chart/ac8b3538-en_p126","product_type":"parse","source_file_path":"ac8b3538-en_p126.pdf"},"started_at":"2026-11-04T19:46:10.656981"} +{"completed_at":"2026-11-04T19:46:14.638419","latency_in_ms":2367,"output":{"example_id":"chart/ac8b3538-en_p127","grounded_pages":[],"layout_pages":[],"markdown":" **125**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\nStatLink 2 https://stat.link/k7u2yn\n\n### **Annex Figure 2.C.9. Employees in GHG-intensive occupations typically earn low or middle wages**\n\nPercentage shares of GHG-intensive occupations, by wage category and country, 2018\n\nNote: The figure reports the percentage of GHG-intensive occupations in wage and salary employment, by country and hourly wage category. High (resp. low) wage is defined as hourly wage above one-and-a-half times (resp. below two-thirds of) the median wage. Agriculture is excluded except in Australia, Canada and the United States. Data for Belgium, Denmark, France, Greece, Iceland, Italy, Luxembourg, Portugal, Sweden and the United Kingdom do not include firms with less than 10 employees. Data for Canada and the United States refer to 2019. OECD: Unweighted average of countries shown. GHG: greenhouse gases. Countries are ranked by the difference in the incidence of GHG-intensive occupations between high and low wage workers. Reading: In Hungary, 13.3% (resp. 6.2% and 2.2%) of low-wage (resp. middle-wage and high-wage) employees have a job in a GHG-intensive occupation. A larger incidence of a given type of occupation among high-wage workers than among middle or low-wage workers is indicative of a positive wage gap between that occupation and the others. Source: Secretariat’s estimates based on version 24.1 of the O\\*NET database and the following country-specific sources: Australia: Table Builder of the Australian Bureau of Statistics (Labour Force: Characteristics of Employment); Canadian Labour Force Survey; United States: Current Population Survey; All other countries: EU Structure of Earnings Surveys.\n\nStatLink 2 https://stat.link/62q14c\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\n% Low pay Medium pay High pay Total","pages":[{"line_numbers":[],"markdown":" **125**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\nStatLink 2 https://stat.link/k7u2yn\n\n### **Annex Figure 2.C.9. Employees in GHG-intensive occupations typically earn low or middle wages**\n\nPercentage shares of GHG-intensive occupations, by wage category and country, 2018\n\nNote: The figure reports the percentage of GHG-intensive occupations in wage and salary employment, by country and hourly wage category. High (resp. low) wage is defined as hourly wage above one-and-a-half times (resp. below two-thirds of) the median wage. Agriculture is excluded except in Australia, Canada and the United States. Data for Belgium, Denmark, France, Greece, Iceland, Italy, Luxembourg, Portugal, Sweden and the United Kingdom do not include firms with less than 10 employees. Data for Canada and the United States refer to 2019. OECD: Unweighted average of countries shown. GHG: greenhouse gases. Countries are ranked by the difference in the incidence of GHG-intensive occupations between high and low wage workers. Reading: In Hungary, 13.3% (resp. 6.2% and 2.2%) of low-wage (resp. middle-wage and high-wage) employees have a job in a GHG-intensive occupation. A larger incidence of a given type of occupation among high-wage workers than among middle or low-wage workers is indicative of a positive wage gap between that occupation and the others. Source: Secretariat’s estimates based on version 24.1 of the O\\*NET database and the following country-specific sources: Australia: Table Builder of the Australian Bureau of Statistics (Labour Force: Characteristics of Employment); Canadian Labour Force Survey; United States: Current Population Survey; All other countries: EU Structure of Earnings Surveys.\n\nStatLink 2 https://stat.link/62q14c\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\n% Low pay Medium pay High pay Total","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":" **125**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\nStatLink 2 https://stat.link/k7u2yn\n\n### **Annex Figure 2.C.9. Employees in GHG-intensive occupations typically earn low or middle wages**\n\nPercentage shares of GHG-intensive occupations, by wage category and country, 2018\n\nNote: The figure reports the percentage of GHG-intensive occupations in wage and salary employment, by country and hourly wage category. High (resp. low) wage is defined as hourly wage above one-and-a-half times (resp. below two-thirds of) the median wage. Agriculture is excluded except in Australia, Canada and the United States. Data for Belgium, Denmark, France, Greece, Iceland, Italy, Luxembourg, Portugal, Sweden and the United Kingdom do not include firms with less than 10 employees. Data for Canada and the United States refer to 2019. OECD: Unweighted average of countries shown. GHG: greenhouse gases. Countries are ranked by the difference in the incidence of GHG-intensive occupations between high and low wage workers. Reading: In Hungary, 13.3% (resp. 6.2% and 2.2%) of low-wage (resp. middle-wage and high-wage) employees have a job in a GHG-intensive occupation. A larger incidence of a given type of occupation among high-wage workers than among middle or low-wage workers is indicative of a positive wage gap between that occupation and the others. Source: Secretariat’s estimates based on version 24.1 of the O\\*NET database and the following country-specific sources: Australia: Table Builder of the Australian Bureau of Statistics (Labour Force: Characteristics of Employment); Canadian Labour Force Survey; United States: Current Population Survey; All other countries: EU Structure of Earnings Surveys.\n\nStatLink 2 https://stat.link/62q14c\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\n% Low pay Medium pay High pay Total","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p127.pdf"]},"response":{"errors":[],"extract_id":"extract_84b7557fb9290cf80a02623d814f0c18","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":[" **125**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\nStatLink 2 https://stat.link/k7u2yn\n\n### **Annex Figure 2.C.9. Employees in GHG-intensive occupations typically earn low or middle wages**\n\nPercentage shares of GHG-intensive occupations, by wage category and country, 2018\n\nNote: The figure reports the percentage of GHG-intensive occupations in wage and salary employment, by country and hourly wage category. High (resp. low) wage is defined as hourly wage above one-and-a-half times (resp. below two-thirds of) the median wage. Agriculture is excluded except in Australia, Canada and the United States. Data for Belgium, Denmark, France, Greece, Iceland, Italy, Luxembourg, Portugal, Sweden and the United Kingdom do not include firms with less than 10 employees. Data for Canada and the United States refer to 2019. OECD: Unweighted average of countries shown. GHG: greenhouse gases. Countries are ranked by the difference in the incidence of GHG-intensive occupations between high and low wage workers. Reading: In Hungary, 13.3% (resp. 6.2% and 2.2%) of low-wage (resp. middle-wage and high-wage) employees have a job in a GHG-intensive occupation. A larger incidence of a given type of occupation among high-wage workers than among middle or low-wage workers is indicative of a positive wage gap between that occupation and the others. Source: Secretariat’s estimates based on version 24.1 of the O\\*NET database and the following country-specific sources: Australia: Table Builder of the Australian Bureau of Statistics (Labour Force: Characteristics of Employment); Canadian Labour Force Survey; United States: Current Population Survey; All other countries: EU Structure of Earnings Surveys.\n\nStatLink 2 https://stat.link/62q14c\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\n% Low pay Medium pay High pay Total"],"full_content":" **125**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\nStatLink 2 https://stat.link/k7u2yn\n\n### **Annex Figure 2.C.9. Employees in GHG-intensive occupations typically earn low or middle wages**\n\nPercentage shares of GHG-intensive occupations, by wage category and country, 2018\n\nNote: The figure reports the percentage of GHG-intensive occupations in wage and salary employment, by country and hourly wage category. High (resp. low) wage is defined as hourly wage above one-and-a-half times (resp. below two-thirds of) the median wage. Agriculture is excluded except in Australia, Canada and the United States. Data for Belgium, Denmark, France, Greece, Iceland, Italy, Luxembourg, Portugal, Sweden and the United Kingdom do not include firms with less than 10 employees. Data for Canada and the United States refer to 2019. OECD: Unweighted average of countries shown. GHG: greenhouse gases. Countries are ranked by the difference in the incidence of GHG-intensive occupations between high and low wage workers. Reading: In Hungary, 13.3% (resp. 6.2% and 2.2%) of low-wage (resp. middle-wage and high-wage) employees have a job in a GHG-intensive occupation. A larger incidence of a given type of occupation among high-wage workers than among middle or low-wage workers is indicative of a positive wage gap between that occupation and the others. Source: Secretariat’s estimates based on version 24.1 of the O\\*NET database and the following country-specific sources: Australia: Table Builder of the Australian Bureau of Statistics (Labour Force: Characteristics of Employment); Canadian Labour Force Survey; United States: Current Population Survey; All other countries: EU Structure of Earnings Surveys.\n\nStatLink 2 https://stat.link/62q14c\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\n% Low pay Medium pay High pay Total","publish_date":null,"title":"Annex Figure 2.C.9. Employees in GHG-intensive occupations typically earn low or middle wages","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p127.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p127.pdf"},"request":{"example_id":"chart/ac8b3538-en_p127","product_type":"parse","source_file_path":"ac8b3538-en_p127.pdf"},"started_at":"2026-11-04T19:46:12.270960"} +{"completed_at":"2026-11-04T19:46:16.657159","latency_in_ms":2386,"output":{"example_id":"chart/ac8b3538-en_p128","grounded_pages":[],"layout_pages":[],"markdown":"**126** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Annex Figure 2.C.10. The wage gap in GHG-intensive occupations is less negative for women**\n\nPercentage difference between the share of green-driven and GHG-intensive occupations among high and low- wage workers, 2018\n\nNote: The figure reports the cross-country unweighted average difference in the percentage of green-driven and GHG-intensive occupations in wage and salary employment between high and low wage workers, by gender. High (resp. low) wage is defined as hourly wage above one-and- a-half times (resp. below two-thirds of) the median wage. Included countries are Australia, Belgium, Canada, Czechia, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Iceland, Italy, Latvia, Lithuania, Luxembourg, the Netherlands, Norway, Poland, Portugal, the Slovak Republic, Slovenia, Spain, Sweden, the United Kingdom, the United States. Agriculture is excluded except in Australia, Canada and the United States. Data for Belgium, Denmark, France, Greece, Iceland, Italy, Luxembourg, Portugal, Sweden and the United Kingdom do not include firms with less than 10 employees. Data for Canada and the United States refer to 2019. GHG: greenhouse gases. Countries are ranked by the difference in the incidence of GHG-intensive occupations between high and low wage workers. Reading: On average, the share of green-driven occupation is 1.6% higher among high-wage men than among low-wage men. A larger incidence of a given type of occupation among high-wage workers than among middle or low-wage workers is indicative of a positive wage gap between that occupation and the others. Source: Secretariat’s estimates based on version 24.1 of the O\\*NET database and the following country-specific sources: Australia: Table Builder of the Australian Bureau of Statistics (Labour Force: Characteristics of Employment); Canadian Labour Force Survey; United States: Current Population Survey; All other countries: EU Structure of Earnings Surveys.\n\nStatLink 2 https://stat.link/nrm1j0\n\n\\-5\n\n\\-4\n\n\\-3\n\n\\-2\n\n\\-1\n\n0\n\n1\n\n2\n\nGreen-driven occupations GHG-intensive occupations\n\n%\n\nMen Women","pages":[{"line_numbers":[],"markdown":"**126** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Annex Figure 2.C.10. The wage gap in GHG-intensive occupations is less negative for women**\n\nPercentage difference between the share of green-driven and GHG-intensive occupations among high and low- wage workers, 2018\n\nNote: The figure reports the cross-country unweighted average difference in the percentage of green-driven and GHG-intensive occupations in wage and salary employment between high and low wage workers, by gender. High (resp. low) wage is defined as hourly wage above one-and- a-half times (resp. below two-thirds of) the median wage. Included countries are Australia, Belgium, Canada, Czechia, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Iceland, Italy, Latvia, Lithuania, Luxembourg, the Netherlands, Norway, Poland, Portugal, the Slovak Republic, Slovenia, Spain, Sweden, the United Kingdom, the United States. Agriculture is excluded except in Australia, Canada and the United States. Data for Belgium, Denmark, France, Greece, Iceland, Italy, Luxembourg, Portugal, Sweden and the United Kingdom do not include firms with less than 10 employees. Data for Canada and the United States refer to 2019. GHG: greenhouse gases. Countries are ranked by the difference in the incidence of GHG-intensive occupations between high and low wage workers. Reading: On average, the share of green-driven occupation is 1.6% higher among high-wage men than among low-wage men. A larger incidence of a given type of occupation among high-wage workers than among middle or low-wage workers is indicative of a positive wage gap between that occupation and the others. Source: Secretariat’s estimates based on version 24.1 of the O\\*NET database and the following country-specific sources: Australia: Table Builder of the Australian Bureau of Statistics (Labour Force: Characteristics of Employment); Canadian Labour Force Survey; United States: Current Population Survey; All other countries: EU Structure of Earnings Surveys.\n\nStatLink 2 https://stat.link/nrm1j0\n\n\\-5\n\n\\-4\n\n\\-3\n\n\\-2\n\n\\-1\n\n0\n\n1\n\n2\n\nGreen-driven occupations GHG-intensive occupations\n\n%\n\nMen Women","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**126** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Annex Figure 2.C.10. The wage gap in GHG-intensive occupations is less negative for women**\n\nPercentage difference between the share of green-driven and GHG-intensive occupations among high and low- wage workers, 2018\n\nNote: The figure reports the cross-country unweighted average difference in the percentage of green-driven and GHG-intensive occupations in wage and salary employment between high and low wage workers, by gender. High (resp. low) wage is defined as hourly wage above one-and- a-half times (resp. below two-thirds of) the median wage. Included countries are Australia, Belgium, Canada, Czechia, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Iceland, Italy, Latvia, Lithuania, Luxembourg, the Netherlands, Norway, Poland, Portugal, the Slovak Republic, Slovenia, Spain, Sweden, the United Kingdom, the United States. Agriculture is excluded except in Australia, Canada and the United States. Data for Belgium, Denmark, France, Greece, Iceland, Italy, Luxembourg, Portugal, Sweden and the United Kingdom do not include firms with less than 10 employees. Data for Canada and the United States refer to 2019. GHG: greenhouse gases. Countries are ranked by the difference in the incidence of GHG-intensive occupations between high and low wage workers. Reading: On average, the share of green-driven occupation is 1.6% higher among high-wage men than among low-wage men. A larger incidence of a given type of occupation among high-wage workers than among middle or low-wage workers is indicative of a positive wage gap between that occupation and the others. Source: Secretariat’s estimates based on version 24.1 of the O\\*NET database and the following country-specific sources: Australia: Table Builder of the Australian Bureau of Statistics (Labour Force: Characteristics of Employment); Canadian Labour Force Survey; United States: Current Population Survey; All other countries: EU Structure of Earnings Surveys.\n\nStatLink 2 https://stat.link/nrm1j0\n\n\\-5\n\n\\-4\n\n\\-3\n\n\\-2\n\n\\-1\n\n0\n\n1\n\n2\n\nGreen-driven occupations GHG-intensive occupations\n\n%\n\nMen Women","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p128.pdf"]},"response":{"errors":[],"extract_id":"extract_c0bd6920bb4605cc85fa4d547f8226d5","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**126** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Annex Figure 2.C.10. The wage gap in GHG-intensive occupations is less negative for women**\n\nPercentage difference between the share of green-driven and GHG-intensive occupations among high and low- wage workers, 2018\n\nNote: The figure reports the cross-country unweighted average difference in the percentage of green-driven and GHG-intensive occupations in wage and salary employment between high and low wage workers, by gender. High (resp. low) wage is defined as hourly wage above one-and- a-half times (resp. below two-thirds of) the median wage. Included countries are Australia, Belgium, Canada, Czechia, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Iceland, Italy, Latvia, Lithuania, Luxembourg, the Netherlands, Norway, Poland, Portugal, the Slovak Republic, Slovenia, Spain, Sweden, the United Kingdom, the United States. Agriculture is excluded except in Australia, Canada and the United States. Data for Belgium, Denmark, France, Greece, Iceland, Italy, Luxembourg, Portugal, Sweden and the United Kingdom do not include firms with less than 10 employees. Data for Canada and the United States refer to 2019. GHG: greenhouse gases. Countries are ranked by the difference in the incidence of GHG-intensive occupations between high and low wage workers. Reading: On average, the share of green-driven occupation is 1.6% higher among high-wage men than among low-wage men. A larger incidence of a given type of occupation among high-wage workers than among middle or low-wage workers is indicative of a positive wage gap between that occupation and the others. Source: Secretariat’s estimates based on version 24.1 of the O\\*NET database and the following country-specific sources: Australia: Table Builder of the Australian Bureau of Statistics (Labour Force: Characteristics of Employment); Canadian Labour Force Survey; United States: Current Population Survey; All other countries: EU Structure of Earnings Surveys.\n\nStatLink 2 https://stat.link/nrm1j0\n\n\\-5\n\n\\-4\n\n\\-3\n\n\\-2\n\n\\-1\n\n0\n\n1\n\n2\n\nGreen-driven occupations GHG-intensive occupations\n\n%\n\nMen Women"],"full_content":"**126** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Annex Figure 2.C.10. The wage gap in GHG-intensive occupations is less negative for women**\n\nPercentage difference between the share of green-driven and GHG-intensive occupations among high and low- wage workers, 2018\n\nNote: The figure reports the cross-country unweighted average difference in the percentage of green-driven and GHG-intensive occupations in wage and salary employment between high and low wage workers, by gender. High (resp. low) wage is defined as hourly wage above one-and- a-half times (resp. below two-thirds of) the median wage. Included countries are Australia, Belgium, Canada, Czechia, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Iceland, Italy, Latvia, Lithuania, Luxembourg, the Netherlands, Norway, Poland, Portugal, the Slovak Republic, Slovenia, Spain, Sweden, the United Kingdom, the United States. Agriculture is excluded except in Australia, Canada and the United States. Data for Belgium, Denmark, France, Greece, Iceland, Italy, Luxembourg, Portugal, Sweden and the United Kingdom do not include firms with less than 10 employees. Data for Canada and the United States refer to 2019. GHG: greenhouse gases. Countries are ranked by the difference in the incidence of GHG-intensive occupations between high and low wage workers. Reading: On average, the share of green-driven occupation is 1.6% higher among high-wage men than among low-wage men. A larger incidence of a given type of occupation among high-wage workers than among middle or low-wage workers is indicative of a positive wage gap between that occupation and the others. Source: Secretariat’s estimates based on version 24.1 of the O\\*NET database and the following country-specific sources: Australia: Table Builder of the Australian Bureau of Statistics (Labour Force: Characteristics of Employment); Canadian Labour Force Survey; United States: Current Population Survey; All other countries: EU Structure of Earnings Surveys.\n\nStatLink 2 https://stat.link/nrm1j0\n\n\\-5\n\n\\-4\n\n\\-3\n\n\\-2\n\n\\-1\n\n0\n\n1\n\n2\n\nGreen-driven occupations GHG-intensive occupations\n\n%\n\nMen Women","publish_date":null,"title":"Annex Figure 2.C.10. The wage gap in GHG-intensive occupations is less negative for women","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p128.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p128.pdf"},"request":{"example_id":"chart/ac8b3538-en_p128","product_type":"parse","source_file_path":"ac8b3538-en_p128.pdf"},"started_at":"2026-11-04T19:46:14.270415"} +{"completed_at":"2026-11-04T19:46:18.785738","latency_in_ms":2599,"output":{"example_id":"chart/ac8b3538-en_p130","grounded_pages":[],"layout_pages":[],"markdown":"**128** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\nNote: The figure reports the percentage of each type of green-driven occupations in wage and salary employment, by country and hourly wage category. High (resp. low) wage is defined as hourly wage above one-and-a-half times (resp. below two-thirds of) the median wage. Agriculture is excluded except in Australia, Canada and the United States. Data for Belgium, Denmark, France, Greece, Iceland, Italy, Luxembourg, Portugal, Sweden and the United Kingdom do not include firms with less than 10 employees. Data for Canada and the United States refer to 2019. OECD: Unweighted average of countries shown. GHG: greenhouse gases. Countries are ranked by the difference in the incidence of each type of occupations between high and low wage workers. Reading: In Norway, 7.6% (resp. 2.5% and 0.9%) of high-wage (resp. middle-wage and low-wage) employees have a job in a green new and emerging occupation. A larger incidence of a given type of occupation among high-wage workers than among middle or low-wage workers is indicative of a positive wage gap between that occupation and the others. Source: Secretariat’s estimates based on version 24.1 of the O\\*NET database and the following country-specific sources: Australia: Table Builder of the Australian Bureau of Statistics (Labour Force: Characteristics of Employment); Canadian Labour Force Survey; United States: Current Population Survey; All other countries: EU Structure of Earnings Surveys.\n\nStatLink 2 https://stat.link/dmu0tr\n\n### **Annex Figure 2.C.12. Gap in the share of green-driven or GHG-intensive occupations in non- strained jobs compared with strained jobs**\n\nPercentage points, 2021 for EU-OECD countries and 2015 for the United States\n\nNote: The chart reports the difference in percentage points between the share of green-driven occupations in non-strained jobs compared with the share of green-driven occupations in strained jobs (Panel A) and between the share of GHG-intensive occupations in non-strained jobs compared with the share of GHG-intensive occupations in strained jobs (Panel B). OECD: Unweighted average of countries shown. GHG: greenhouse gases. Reading: In Panel A, in Greece, the share of green-driven occupations in non-strained jobs is 5.5 percentage points lower than in strained jobs, which means that workers in green-driven occupations are more strained than the others. In Italy, on the opposite, the share of green-driven occupations in non-strained jobs is 4.4 percentage points higher than in strained jobs, which means that workers in green-driven occupations are less strained than the others. Source: Secretariat’s estimates based on version 24.1 of the O\\*NET database and Secretariat’s estimates based on the European Working Conditions Telephone Survey 2021 and the American Working Conditions Survey 2015.\n\nStatLink 2 https://stat.link/32xqhj\n\n\\-6\n\n\\-4\n\n\\-2\n\n0\n\n2\n\n4\n\n6 p.p.\n\n**A. Green-driven occupations**\n\n\\-8\n\n\\-7\n\n\\-6\n\n\\-5\n\n\\-4\n\n\\-3\n\n\\-2\n\n\\-1\n\n0\n\n1 p.p. **B. GHG-intensive occupations**","pages":[{"line_numbers":[],"markdown":"**128** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\nNote: The figure reports the percentage of each type of green-driven occupations in wage and salary employment, by country and hourly wage category. High (resp. low) wage is defined as hourly wage above one-and-a-half times (resp. below two-thirds of) the median wage. Agriculture is excluded except in Australia, Canada and the United States. Data for Belgium, Denmark, France, Greece, Iceland, Italy, Luxembourg, Portugal, Sweden and the United Kingdom do not include firms with less than 10 employees. Data for Canada and the United States refer to 2019. OECD: Unweighted average of countries shown. GHG: greenhouse gases. Countries are ranked by the difference in the incidence of each type of occupations between high and low wage workers. Reading: In Norway, 7.6% (resp. 2.5% and 0.9%) of high-wage (resp. middle-wage and low-wage) employees have a job in a green new and emerging occupation. A larger incidence of a given type of occupation among high-wage workers than among middle or low-wage workers is indicative of a positive wage gap between that occupation and the others. Source: Secretariat’s estimates based on version 24.1 of the O\\*NET database and the following country-specific sources: Australia: Table Builder of the Australian Bureau of Statistics (Labour Force: Characteristics of Employment); Canadian Labour Force Survey; United States: Current Population Survey; All other countries: EU Structure of Earnings Surveys.\n\nStatLink 2 https://stat.link/dmu0tr\n\n### **Annex Figure 2.C.12. Gap in the share of green-driven or GHG-intensive occupations in non- strained jobs compared with strained jobs**\n\nPercentage points, 2021 for EU-OECD countries and 2015 for the United States\n\nNote: The chart reports the difference in percentage points between the share of green-driven occupations in non-strained jobs compared with the share of green-driven occupations in strained jobs (Panel A) and between the share of GHG-intensive occupations in non-strained jobs compared with the share of GHG-intensive occupations in strained jobs (Panel B). OECD: Unweighted average of countries shown. GHG: greenhouse gases. Reading: In Panel A, in Greece, the share of green-driven occupations in non-strained jobs is 5.5 percentage points lower than in strained jobs, which means that workers in green-driven occupations are more strained than the others. In Italy, on the opposite, the share of green-driven occupations in non-strained jobs is 4.4 percentage points higher than in strained jobs, which means that workers in green-driven occupations are less strained than the others. Source: Secretariat’s estimates based on version 24.1 of the O\\*NET database and Secretariat’s estimates based on the European Working Conditions Telephone Survey 2021 and the American Working Conditions Survey 2015.\n\nStatLink 2 https://stat.link/32xqhj\n\n\\-6\n\n\\-4\n\n\\-2\n\n0\n\n2\n\n4\n\n6 p.p.\n\n**A. Green-driven occupations**\n\n\\-8\n\n\\-7\n\n\\-6\n\n\\-5\n\n\\-4\n\n\\-3\n\n\\-2\n\n\\-1\n\n0\n\n1 p.p. **B. GHG-intensive occupations**","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**128** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\nNote: The figure reports the percentage of each type of green-driven occupations in wage and salary employment, by country and hourly wage category. High (resp. low) wage is defined as hourly wage above one-and-a-half times (resp. below two-thirds of) the median wage. Agriculture is excluded except in Australia, Canada and the United States. Data for Belgium, Denmark, France, Greece, Iceland, Italy, Luxembourg, Portugal, Sweden and the United Kingdom do not include firms with less than 10 employees. Data for Canada and the United States refer to 2019. OECD: Unweighted average of countries shown. GHG: greenhouse gases. Countries are ranked by the difference in the incidence of each type of occupations between high and low wage workers. Reading: In Norway, 7.6% (resp. 2.5% and 0.9%) of high-wage (resp. middle-wage and low-wage) employees have a job in a green new and emerging occupation. A larger incidence of a given type of occupation among high-wage workers than among middle or low-wage workers is indicative of a positive wage gap between that occupation and the others. Source: Secretariat’s estimates based on version 24.1 of the O\\*NET database and the following country-specific sources: Australia: Table Builder of the Australian Bureau of Statistics (Labour Force: Characteristics of Employment); Canadian Labour Force Survey; United States: Current Population Survey; All other countries: EU Structure of Earnings Surveys.\n\nStatLink 2 https://stat.link/dmu0tr\n\n### **Annex Figure 2.C.12. Gap in the share of green-driven or GHG-intensive occupations in non- strained jobs compared with strained jobs**\n\nPercentage points, 2021 for EU-OECD countries and 2015 for the United States\n\nNote: The chart reports the difference in percentage points between the share of green-driven occupations in non-strained jobs compared with the share of green-driven occupations in strained jobs (Panel A) and between the share of GHG-intensive occupations in non-strained jobs compared with the share of GHG-intensive occupations in strained jobs (Panel B). OECD: Unweighted average of countries shown. GHG: greenhouse gases. Reading: In Panel A, in Greece, the share of green-driven occupations in non-strained jobs is 5.5 percentage points lower than in strained jobs, which means that workers in green-driven occupations are more strained than the others. In Italy, on the opposite, the share of green-driven occupations in non-strained jobs is 4.4 percentage points higher than in strained jobs, which means that workers in green-driven occupations are less strained than the others. Source: Secretariat’s estimates based on version 24.1 of the O\\*NET database and Secretariat’s estimates based on the European Working Conditions Telephone Survey 2021 and the American Working Conditions Survey 2015.\n\nStatLink 2 https://stat.link/32xqhj\n\n\\-6\n\n\\-4\n\n\\-2\n\n0\n\n2\n\n4\n\n6 p.p.\n\n**A. Green-driven occupations**\n\n\\-8\n\n\\-7\n\n\\-6\n\n\\-5\n\n\\-4\n\n\\-3\n\n\\-2\n\n\\-1\n\n0\n\n1 p.p. **B. GHG-intensive occupations**","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p130.pdf"]},"response":{"errors":[],"extract_id":"extract_54046b7ad4d42098775c14b53e9ab12f","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**128** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\nNote: The figure reports the percentage of each type of green-driven occupations in wage and salary employment, by country and hourly wage category. High (resp. low) wage is defined as hourly wage above one-and-a-half times (resp. below two-thirds of) the median wage. Agriculture is excluded except in Australia, Canada and the United States. Data for Belgium, Denmark, France, Greece, Iceland, Italy, Luxembourg, Portugal, Sweden and the United Kingdom do not include firms with less than 10 employees. Data for Canada and the United States refer to 2019. OECD: Unweighted average of countries shown. GHG: greenhouse gases. Countries are ranked by the difference in the incidence of each type of occupations between high and low wage workers. Reading: In Norway, 7.6% (resp. 2.5% and 0.9%) of high-wage (resp. middle-wage and low-wage) employees have a job in a green new and emerging occupation. A larger incidence of a given type of occupation among high-wage workers than among middle or low-wage workers is indicative of a positive wage gap between that occupation and the others. Source: Secretariat’s estimates based on version 24.1 of the O\\*NET database and the following country-specific sources: Australia: Table Builder of the Australian Bureau of Statistics (Labour Force: Characteristics of Employment); Canadian Labour Force Survey; United States: Current Population Survey; All other countries: EU Structure of Earnings Surveys.\n\nStatLink 2 https://stat.link/dmu0tr\n\n### **Annex Figure 2.C.12. Gap in the share of green-driven or GHG-intensive occupations in non- strained jobs compared with strained jobs**\n\nPercentage points, 2021 for EU-OECD countries and 2015 for the United States\n\nNote: The chart reports the difference in percentage points between the share of green-driven occupations in non-strained jobs compared with the share of green-driven occupations in strained jobs (Panel A) and between the share of GHG-intensive occupations in non-strained jobs compared with the share of GHG-intensive occupations in strained jobs (Panel B). OECD: Unweighted average of countries shown. GHG: greenhouse gases. Reading: In Panel A, in Greece, the share of green-driven occupations in non-strained jobs is 5.5 percentage points lower than in strained jobs, which means that workers in green-driven occupations are more strained than the others. In Italy, on the opposite, the share of green-driven occupations in non-strained jobs is 4.4 percentage points higher than in strained jobs, which means that workers in green-driven occupations are less strained than the others. Source: Secretariat’s estimates based on version 24.1 of the O\\*NET database and Secretariat’s estimates based on the European Working Conditions Telephone Survey 2021 and the American Working Conditions Survey 2015.\n\nStatLink 2 https://stat.link/32xqhj\n\n\\-6\n\n\\-4\n\n\\-2\n\n0\n\n2\n\n4\n\n6 p.p.\n\n**A. Green-driven occupations**\n\n\\-8\n\n\\-7\n\n\\-6\n\n\\-5\n\n\\-4\n\n\\-3\n\n\\-2\n\n\\-1\n\n0\n\n1 p.p. **B. GHG-intensive occupations**"],"full_content":"**128** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\nNote: The figure reports the percentage of each type of green-driven occupations in wage and salary employment, by country and hourly wage category. High (resp. low) wage is defined as hourly wage above one-and-a-half times (resp. below two-thirds of) the median wage. Agriculture is excluded except in Australia, Canada and the United States. Data for Belgium, Denmark, France, Greece, Iceland, Italy, Luxembourg, Portugal, Sweden and the United Kingdom do not include firms with less than 10 employees. Data for Canada and the United States refer to 2019. OECD: Unweighted average of countries shown. GHG: greenhouse gases. Countries are ranked by the difference in the incidence of each type of occupations between high and low wage workers. Reading: In Norway, 7.6% (resp. 2.5% and 0.9%) of high-wage (resp. middle-wage and low-wage) employees have a job in a green new and emerging occupation. A larger incidence of a given type of occupation among high-wage workers than among middle or low-wage workers is indicative of a positive wage gap between that occupation and the others. Source: Secretariat’s estimates based on version 24.1 of the O\\*NET database and the following country-specific sources: Australia: Table Builder of the Australian Bureau of Statistics (Labour Force: Characteristics of Employment); Canadian Labour Force Survey; United States: Current Population Survey; All other countries: EU Structure of Earnings Surveys.\n\nStatLink 2 https://stat.link/dmu0tr\n\n### **Annex Figure 2.C.12. Gap in the share of green-driven or GHG-intensive occupations in non- strained jobs compared with strained jobs**\n\nPercentage points, 2021 for EU-OECD countries and 2015 for the United States\n\nNote: The chart reports the difference in percentage points between the share of green-driven occupations in non-strained jobs compared with the share of green-driven occupations in strained jobs (Panel A) and between the share of GHG-intensive occupations in non-strained jobs compared with the share of GHG-intensive occupations in strained jobs (Panel B). OECD: Unweighted average of countries shown. GHG: greenhouse gases. Reading: In Panel A, in Greece, the share of green-driven occupations in non-strained jobs is 5.5 percentage points lower than in strained jobs, which means that workers in green-driven occupations are more strained than the others. In Italy, on the opposite, the share of green-driven occupations in non-strained jobs is 4.4 percentage points higher than in strained jobs, which means that workers in green-driven occupations are less strained than the others. Source: Secretariat’s estimates based on version 24.1 of the O\\*NET database and Secretariat’s estimates based on the European Working Conditions Telephone Survey 2021 and the American Working Conditions Survey 2015.\n\nStatLink 2 https://stat.link/32xqhj\n\n\\-6\n\n\\-4\n\n\\-2\n\n0\n\n2\n\n4\n\n6 p.p.\n\n**A. Green-driven occupations**\n\n\\-8\n\n\\-7\n\n\\-6\n\n\\-5\n\n\\-4\n\n\\-3\n\n\\-2\n\n\\-1\n\n0\n\n1 p.p. **B. GHG-intensive occupations**","publish_date":null,"title":"Annex Figure 2.C.12. Gap in the share of green-driven or GHG-intensive occupations in non- strained jobs compared with strained jobs","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p130.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p130.pdf"},"request":{"example_id":"chart/ac8b3538-en_p130","product_type":"parse","source_file_path":"ac8b3538-en_p130.pdf"},"started_at":"2026-11-04T19:46:16.185788"} +{"completed_at":"2026-11-04T19:46:18.968444","latency_in_ms":1514,"output":{"example_id":"chart/ac8b3538-en_p131","grounded_pages":[],"layout_pages":[],"markdown":" **129**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Annex Figure 2.C.13. The incidence of job strain is lower in new and emerging and green-enhanced skills green occupations even when controlling for the job characteristics**\n\nMarginal effect of the job strain indicator on the share of green-driven occupations, controlling for main individual and job characteristics, 2021 for EU-OECD countries and 2015 for the United States\n\nNote: The chart reports the point estimate (and 95% confidence intervals) of the percentage difference in the incidence of each type of green- driven occupation between job strained and not strained workers. Estimates are obtained from a linear regression with the inverse hyperbolic sine of the share of each type of green-driven occupation within all, high, medium or low-skilled occupations as dependent variable and including a job strain dummy, the share of 1-digit ISCO occupations, educational attainment (3 classes), gender, age (3 classes), sectors (5 categories), and country dummies as explanatory variables. Standard errors are clustered on the dimensions of variability of the dependent variable. The reported point estimates and confidence intervals refer to the estimated coefficient of the job strain dummy and are expressed in percentage of the untransformed dependent variable. The sample excludes Agriculture. Reading: Controlling for demographic and job characteristics, the percentage share of green new and emerging occupations is, on average, 12% lower among strained employees, i.e. employees whose number of job demands exceeds the number of job resources. Source: Secretariat’s estimates based on version 24.1 of the O\\*NET database and the European Working Conditions Telephone Survey 2021 and the American Working Conditions Survey 2015.\n\nStatLink 2 https://stat.link/vgu8as\n\n\\-40\n\n\\-30\n\n\\-20\n\n\\-10\n\n0\n\n10\n\n20\n\n30\n\nGreen-driven occupations Green new and emerging\n\noccupations\n\nGreen-enhanced skills\n\noccupations\n\nGreen increased demand\n\noccupations\n\nGHG-intensive occupations\n\n%","pages":[{"line_numbers":[],"markdown":" **129**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Annex Figure 2.C.13. The incidence of job strain is lower in new and emerging and green-enhanced skills green occupations even when controlling for the job characteristics**\n\nMarginal effect of the job strain indicator on the share of green-driven occupations, controlling for main individual and job characteristics, 2021 for EU-OECD countries and 2015 for the United States\n\nNote: The chart reports the point estimate (and 95% confidence intervals) of the percentage difference in the incidence of each type of green- driven occupation between job strained and not strained workers. Estimates are obtained from a linear regression with the inverse hyperbolic sine of the share of each type of green-driven occupation within all, high, medium or low-skilled occupations as dependent variable and including a job strain dummy, the share of 1-digit ISCO occupations, educational attainment (3 classes), gender, age (3 classes), sectors (5 categories), and country dummies as explanatory variables. Standard errors are clustered on the dimensions of variability of the dependent variable. The reported point estimates and confidence intervals refer to the estimated coefficient of the job strain dummy and are expressed in percentage of the untransformed dependent variable. The sample excludes Agriculture. Reading: Controlling for demographic and job characteristics, the percentage share of green new and emerging occupations is, on average, 12% lower among strained employees, i.e. employees whose number of job demands exceeds the number of job resources. Source: Secretariat’s estimates based on version 24.1 of the O\\*NET database and the European Working Conditions Telephone Survey 2021 and the American Working Conditions Survey 2015.\n\nStatLink 2 https://stat.link/vgu8as\n\n\\-40\n\n\\-30\n\n\\-20\n\n\\-10\n\n0\n\n10\n\n20\n\n30\n\nGreen-driven occupations Green new and emerging\n\noccupations\n\nGreen-enhanced skills\n\noccupations\n\nGreen increased demand\n\noccupations\n\nGHG-intensive occupations\n\n%","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":" **129**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Annex Figure 2.C.13. The incidence of job strain is lower in new and emerging and green-enhanced skills green occupations even when controlling for the job characteristics**\n\nMarginal effect of the job strain indicator on the share of green-driven occupations, controlling for main individual and job characteristics, 2021 for EU-OECD countries and 2015 for the United States\n\nNote: The chart reports the point estimate (and 95% confidence intervals) of the percentage difference in the incidence of each type of green- driven occupation between job strained and not strained workers. Estimates are obtained from a linear regression with the inverse hyperbolic sine of the share of each type of green-driven occupation within all, high, medium or low-skilled occupations as dependent variable and including a job strain dummy, the share of 1-digit ISCO occupations, educational attainment (3 classes), gender, age (3 classes), sectors (5 categories), and country dummies as explanatory variables. Standard errors are clustered on the dimensions of variability of the dependent variable. The reported point estimates and confidence intervals refer to the estimated coefficient of the job strain dummy and are expressed in percentage of the untransformed dependent variable. The sample excludes Agriculture. Reading: Controlling for demographic and job characteristics, the percentage share of green new and emerging occupations is, on average, 12% lower among strained employees, i.e. employees whose number of job demands exceeds the number of job resources. Source: Secretariat’s estimates based on version 24.1 of the O\\*NET database and the European Working Conditions Telephone Survey 2021 and the American Working Conditions Survey 2015.\n\nStatLink 2 https://stat.link/vgu8as\n\n\\-40\n\n\\-30\n\n\\-20\n\n\\-10\n\n0\n\n10\n\n20\n\n30\n\nGreen-driven occupations Green new and emerging\n\noccupations\n\nGreen-enhanced skills\n\noccupations\n\nGreen increased demand\n\noccupations\n\nGHG-intensive occupations\n\n%","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p131.pdf"]},"response":{"errors":[],"extract_id":"extract_3e0e2c94d31117cc75a931363206dd2d","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":[" **129**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Annex Figure 2.C.13. The incidence of job strain is lower in new and emerging and green-enhanced skills green occupations even when controlling for the job characteristics**\n\nMarginal effect of the job strain indicator on the share of green-driven occupations, controlling for main individual and job characteristics, 2021 for EU-OECD countries and 2015 for the United States\n\nNote: The chart reports the point estimate (and 95% confidence intervals) of the percentage difference in the incidence of each type of green- driven occupation between job strained and not strained workers. Estimates are obtained from a linear regression with the inverse hyperbolic sine of the share of each type of green-driven occupation within all, high, medium or low-skilled occupations as dependent variable and including a job strain dummy, the share of 1-digit ISCO occupations, educational attainment (3 classes), gender, age (3 classes), sectors (5 categories), and country dummies as explanatory variables. Standard errors are clustered on the dimensions of variability of the dependent variable. The reported point estimates and confidence intervals refer to the estimated coefficient of the job strain dummy and are expressed in percentage of the untransformed dependent variable. The sample excludes Agriculture. Reading: Controlling for demographic and job characteristics, the percentage share of green new and emerging occupations is, on average, 12% lower among strained employees, i.e. employees whose number of job demands exceeds the number of job resources. Source: Secretariat’s estimates based on version 24.1 of the O\\*NET database and the European Working Conditions Telephone Survey 2021 and the American Working Conditions Survey 2015.\n\nStatLink 2 https://stat.link/vgu8as\n\n\\-40\n\n\\-30\n\n\\-20\n\n\\-10\n\n0\n\n10\n\n20\n\n30\n\nGreen-driven occupations Green new and emerging\n\noccupations\n\nGreen-enhanced skills\n\noccupations\n\nGreen increased demand\n\noccupations\n\nGHG-intensive occupations\n\n%"],"full_content":" **129**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Annex Figure 2.C.13. The incidence of job strain is lower in new and emerging and green-enhanced skills green occupations even when controlling for the job characteristics**\n\nMarginal effect of the job strain indicator on the share of green-driven occupations, controlling for main individual and job characteristics, 2021 for EU-OECD countries and 2015 for the United States\n\nNote: The chart reports the point estimate (and 95% confidence intervals) of the percentage difference in the incidence of each type of green- driven occupation between job strained and not strained workers. Estimates are obtained from a linear regression with the inverse hyperbolic sine of the share of each type of green-driven occupation within all, high, medium or low-skilled occupations as dependent variable and including a job strain dummy, the share of 1-digit ISCO occupations, educational attainment (3 classes), gender, age (3 classes), sectors (5 categories), and country dummies as explanatory variables. Standard errors are clustered on the dimensions of variability of the dependent variable. The reported point estimates and confidence intervals refer to the estimated coefficient of the job strain dummy and are expressed in percentage of the untransformed dependent variable. The sample excludes Agriculture. Reading: Controlling for demographic and job characteristics, the percentage share of green new and emerging occupations is, on average, 12% lower among strained employees, i.e. employees whose number of job demands exceeds the number of job resources. Source: Secretariat’s estimates based on version 24.1 of the O\\*NET database and the European Working Conditions Telephone Survey 2021 and the American Working Conditions Survey 2015.\n\nStatLink 2 https://stat.link/vgu8as\n\n\\-40\n\n\\-30\n\n\\-20\n\n\\-10\n\n0\n\n10\n\n20\n\n30\n\nGreen-driven occupations Green new and emerging\n\noccupations\n\nGreen-enhanced skills\n\noccupations\n\nGreen increased demand\n\noccupations\n\nGHG-intensive occupations\n\n%","publish_date":null,"title":"Annex Figure 2.C.13. The incidence of job strain is lower in new and emerging and green-enhanced skills green occupations even","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p131.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p131.pdf"},"request":{"example_id":"chart/ac8b3538-en_p131","product_type":"parse","source_file_path":"ac8b3538-en_p131.pdf"},"started_at":"2026-11-04T19:46:17.453811"} +{"completed_at":"2026-11-04T19:46:22.495336","latency_in_ms":2572,"output":{"example_id":"chart/ac8b3538-en_p143","grounded_pages":[],"layout_pages":[],"markdown":" **141**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Figure 3.1. Classification of high- and low-emission industries**\n\nGHG intensity within 2-digit NACE rev.2 sectors and frequency in top 2 deciles across 31 OECD countries, 2009-20 average\n\nNote: GHG intensity is expressed as CO2- equivalent emissions in tonnes per million EUR of gross value added. Source: Eurostat – Air emissions accounts by NACE Rev. 2 activity.\n\nStatLink 2 https://stat.link/3xgc2z\n\nThe list of sectors included in the definition encompass energy supply, as well as transport services (water, air and land), mining and manufacturing of energy-intensive products (basic metals, non-metallic mineral products, refined petroleum products, chemical- and paper products). The sewerage and waste-collection sector is omitted from the final list of high-emission industries as it is not expected to decline and indeed is a much needed industry for the net-zero transition (see Borgonovi et al. (2023[1])). The agricultural sector is entirely omitted from the list of industries studied in this chapter for two reasons. First, mass layoffs in agriculture are more likely to capture seasonal rather than structural adjustments in employment. Second, employment projections for many OECD countries do not suggest a decline in employment growth in the agricultural sector, in contrast to most other emission-intensive sectors, as current policy packages tend to shelter agricultural production from the climate mitigation effort (Borgonovi et al., 2023[1]).\n\nThe classification of “high-emission industries” shows not only significant consistency across the countries considered, but also across different measures of emission intensity, such as carbon intensity or fossil fuel energy intensity (see Barreto et al. (forthcoming[11]) for details). A key limitation of the present approach is the classification of industries based 2-digit ISIC rev. 4 sectors. This does not allow disentangling subsectors which may be expected to expand because of the net-zero transition such as renewable energy power generation or electrified land transport from sectors that are expected to contract. However, cross-country emission data, whether in the form of GHG emissions, carbon intensity or energy use, are not available at a more disaggregate level.\n\n0 1 000 2 000 3 000 4 000 5 000 6 000 7 000 8 000 9 000\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n35\n\nCrop and animal production, hunting\n\nElectricity, gas, air conditioning\n\nSewerage; waste; remediation\n\nOth. non-metallic mineral products\n\nAir transport\n\nManuf. basic metals\n\nWater transport\n\nCoke and refined petroleum\n\nManuf. of chemicals\n\nFishing and aquaculture\n\nMining and quarrying\n\nLand transport; pipelines\n\nPaper and paper products\n\nForestry and logging\n\nWater collection, treatment and supply\n\nTextiles, wearing apparel, leather\n\nFood products, beverages and tobacco\n\nPostal and courier activities\n\nMetal products, exc. machinery and equip.\n\nAccommodation and food service\n\nRubber and plastic products\n\nWood and cork, exc. furniture\n\nConstruction\n\nBasic pharmaceutical products\n\nWholesale and retail trade; repair\n\nMachinery and equipment n.e.c.\n\nWarehousing and support for transportation\n\nPrinting and reproduction of recorded media\n\nWholesale trade, exc. motor vehicles\n\nOther service activities\n\nElectrical equipment\n\nRetail trade, exc. of motor vehicles\n\nComputers, electronic and optical products\n\nOther transport equipment\n\nFurniture; other manufacturing\n\nMotor vehicles, trailers and semi-trailers\n\nRepair and installation of machinery and…\n\nAdministrative and support service activities\n\nArts, entertainment and recreation\n\nOther professional, scientific and technical…\n\nPublic admin., defence; compulsory social…\n\nHuman health and social work\n\nArchitectural and engineering\n\nMotion picture, video, sound; braoadcasting\n\nAdvertising and market research\n\nScientific research and development\n\nLegal and accounting.; management…\n\nEducation\n\nPublishing activities\n\nActivities auxiliary to finance and insurance\n\nTelecommunications\n\nInsurance and pension, exc. compulsory…\n\nFinance exc. insurance and pension\n\nComputer programming; information services\n\nReal estate\n\nGHG intensity Frequency of countries in the\n\ntop two deciles\n\nFrequency: emission-intensive Rest of the economy GHG intensity: emission-intensive Rest of the economy","pages":[{"line_numbers":[],"markdown":" **141**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Figure 3.1. Classification of high- and low-emission industries**\n\nGHG intensity within 2-digit NACE rev.2 sectors and frequency in top 2 deciles across 31 OECD countries, 2009-20 average\n\nNote: GHG intensity is expressed as CO2- equivalent emissions in tonnes per million EUR of gross value added. Source: Eurostat – Air emissions accounts by NACE Rev. 2 activity.\n\nStatLink 2 https://stat.link/3xgc2z\n\nThe list of sectors included in the definition encompass energy supply, as well as transport services (water, air and land), mining and manufacturing of energy-intensive products (basic metals, non-metallic mineral products, refined petroleum products, chemical- and paper products). The sewerage and waste-collection sector is omitted from the final list of high-emission industries as it is not expected to decline and indeed is a much needed industry for the net-zero transition (see Borgonovi et al. (2023[1])). The agricultural sector is entirely omitted from the list of industries studied in this chapter for two reasons. First, mass layoffs in agriculture are more likely to capture seasonal rather than structural adjustments in employment. Second, employment projections for many OECD countries do not suggest a decline in employment growth in the agricultural sector, in contrast to most other emission-intensive sectors, as current policy packages tend to shelter agricultural production from the climate mitigation effort (Borgonovi et al., 2023[1]).\n\nThe classification of “high-emission industries” shows not only significant consistency across the countries considered, but also across different measures of emission intensity, such as carbon intensity or fossil fuel energy intensity (see Barreto et al. (forthcoming[11]) for details). A key limitation of the present approach is the classification of industries based 2-digit ISIC rev. 4 sectors. This does not allow disentangling subsectors which may be expected to expand because of the net-zero transition such as renewable energy power generation or electrified land transport from sectors that are expected to contract. However, cross-country emission data, whether in the form of GHG emissions, carbon intensity or energy use, are not available at a more disaggregate level.\n\n0 1 000 2 000 3 000 4 000 5 000 6 000 7 000 8 000 9 000\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n35\n\nCrop and animal production, hunting\n\nElectricity, gas, air conditioning\n\nSewerage; waste; remediation\n\nOth. non-metallic mineral products\n\nAir transport\n\nManuf. basic metals\n\nWater transport\n\nCoke and refined petroleum\n\nManuf. of chemicals\n\nFishing and aquaculture\n\nMining and quarrying\n\nLand transport; pipelines\n\nPaper and paper products\n\nForestry and logging\n\nWater collection, treatment and supply\n\nTextiles, wearing apparel, leather\n\nFood products, beverages and tobacco\n\nPostal and courier activities\n\nMetal products, exc. machinery and equip.\n\nAccommodation and food service\n\nRubber and plastic products\n\nWood and cork, exc. furniture\n\nConstruction\n\nBasic pharmaceutical products\n\nWholesale and retail trade; repair\n\nMachinery and equipment n.e.c.\n\nWarehousing and support for transportation\n\nPrinting and reproduction of recorded media\n\nWholesale trade, exc. motor vehicles\n\nOther service activities\n\nElectrical equipment\n\nRetail trade, exc. of motor vehicles\n\nComputers, electronic and optical products\n\nOther transport equipment\n\nFurniture; other manufacturing\n\nMotor vehicles, trailers and semi-trailers\n\nRepair and installation of machinery and…\n\nAdministrative and support service activities\n\nArts, entertainment and recreation\n\nOther professional, scientific and technical…\n\nPublic admin., defence; compulsory social…\n\nHuman health and social work\n\nArchitectural and engineering\n\nMotion picture, video, sound; braoadcasting\n\nAdvertising and market research\n\nScientific research and development\n\nLegal and accounting.; management…\n\nEducation\n\nPublishing activities\n\nActivities auxiliary to finance and insurance\n\nTelecommunications\n\nInsurance and pension, exc. compulsory…\n\nFinance exc. insurance and pension\n\nComputer programming; information services\n\nReal estate\n\nGHG intensity Frequency of countries in the\n\ntop two deciles\n\nFrequency: emission-intensive Rest of the economy GHG intensity: emission-intensive Rest of the economy","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":" **141**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Figure 3.1. Classification of high- and low-emission industries**\n\nGHG intensity within 2-digit NACE rev.2 sectors and frequency in top 2 deciles across 31 OECD countries, 2009-20 average\n\nNote: GHG intensity is expressed as CO2- equivalent emissions in tonnes per million EUR of gross value added. Source: Eurostat – Air emissions accounts by NACE Rev. 2 activity.\n\nStatLink 2 https://stat.link/3xgc2z\n\nThe list of sectors included in the definition encompass energy supply, as well as transport services (water, air and land), mining and manufacturing of energy-intensive products (basic metals, non-metallic mineral products, refined petroleum products, chemical- and paper products). The sewerage and waste-collection sector is omitted from the final list of high-emission industries as it is not expected to decline and indeed is a much needed industry for the net-zero transition (see Borgonovi et al. (2023[1])). The agricultural sector is entirely omitted from the list of industries studied in this chapter for two reasons. First, mass layoffs in agriculture are more likely to capture seasonal rather than structural adjustments in employment. Second, employment projections for many OECD countries do not suggest a decline in employment growth in the agricultural sector, in contrast to most other emission-intensive sectors, as current policy packages tend to shelter agricultural production from the climate mitigation effort (Borgonovi et al., 2023[1]).\n\nThe classification of “high-emission industries” shows not only significant consistency across the countries considered, but also across different measures of emission intensity, such as carbon intensity or fossil fuel energy intensity (see Barreto et al. (forthcoming[11]) for details). A key limitation of the present approach is the classification of industries based 2-digit ISIC rev. 4 sectors. This does not allow disentangling subsectors which may be expected to expand because of the net-zero transition such as renewable energy power generation or electrified land transport from sectors that are expected to contract. However, cross-country emission data, whether in the form of GHG emissions, carbon intensity or energy use, are not available at a more disaggregate level.\n\n0 1 000 2 000 3 000 4 000 5 000 6 000 7 000 8 000 9 000\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n35\n\nCrop and animal production, hunting\n\nElectricity, gas, air conditioning\n\nSewerage; waste; remediation\n\nOth. non-metallic mineral products\n\nAir transport\n\nManuf. basic metals\n\nWater transport\n\nCoke and refined petroleum\n\nManuf. of chemicals\n\nFishing and aquaculture\n\nMining and quarrying\n\nLand transport; pipelines\n\nPaper and paper products\n\nForestry and logging\n\nWater collection, treatment and supply\n\nTextiles, wearing apparel, leather\n\nFood products, beverages and tobacco\n\nPostal and courier activities\n\nMetal products, exc. machinery and equip.\n\nAccommodation and food service\n\nRubber and plastic products\n\nWood and cork, exc. furniture\n\nConstruction\n\nBasic pharmaceutical products\n\nWholesale and retail trade; repair\n\nMachinery and equipment n.e.c.\n\nWarehousing and support for transportation\n\nPrinting and reproduction of recorded media\n\nWholesale trade, exc. motor vehicles\n\nOther service activities\n\nElectrical equipment\n\nRetail trade, exc. of motor vehicles\n\nComputers, electronic and optical products\n\nOther transport equipment\n\nFurniture; other manufacturing\n\nMotor vehicles, trailers and semi-trailers\n\nRepair and installation of machinery and…\n\nAdministrative and support service activities\n\nArts, entertainment and recreation\n\nOther professional, scientific and technical…\n\nPublic admin., defence; compulsory social…\n\nHuman health and social work\n\nArchitectural and engineering\n\nMotion picture, video, sound; braoadcasting\n\nAdvertising and market research\n\nScientific research and development\n\nLegal and accounting.; management…\n\nEducation\n\nPublishing activities\n\nActivities auxiliary to finance and insurance\n\nTelecommunications\n\nInsurance and pension, exc. compulsory…\n\nFinance exc. insurance and pension\n\nComputer programming; information services\n\nReal estate\n\nGHG intensity Frequency of countries in the\n\ntop two deciles\n\nFrequency: emission-intensive Rest of the economy GHG intensity: emission-intensive Rest of the economy","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p143.pdf"]},"response":{"errors":[],"extract_id":"extract_bd2deda4c136f95b9e560c63623d3441","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":[" **141**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Figure 3.1. Classification of high- and low-emission industries**\n\nGHG intensity within 2-digit NACE rev.2 sectors and frequency in top 2 deciles across 31 OECD countries, 2009-20 average\n\nNote: GHG intensity is expressed as CO2- equivalent emissions in tonnes per million EUR of gross value added. Source: Eurostat – Air emissions accounts by NACE Rev. 2 activity.\n\nStatLink 2 https://stat.link/3xgc2z\n\nThe list of sectors included in the definition encompass energy supply, as well as transport services (water, air and land), mining and manufacturing of energy-intensive products (basic metals, non-metallic mineral products, refined petroleum products, chemical- and paper products). The sewerage and waste-collection sector is omitted from the final list of high-emission industries as it is not expected to decline and indeed is a much needed industry for the net-zero transition (see Borgonovi et al. (2023[1])). The agricultural sector is entirely omitted from the list of industries studied in this chapter for two reasons. First, mass layoffs in agriculture are more likely to capture seasonal rather than structural adjustments in employment. Second, employment projections for many OECD countries do not suggest a decline in employment growth in the agricultural sector, in contrast to most other emission-intensive sectors, as current policy packages tend to shelter agricultural production from the climate mitigation effort (Borgonovi et al., 2023[1]).\n\nThe classification of “high-emission industries” shows not only significant consistency across the countries considered, but also across different measures of emission intensity, such as carbon intensity or fossil fuel energy intensity (see Barreto et al. (forthcoming[11]) for details). A key limitation of the present approach is the classification of industries based 2-digit ISIC rev. 4 sectors. This does not allow disentangling subsectors which may be expected to expand because of the net-zero transition such as renewable energy power generation or electrified land transport from sectors that are expected to contract. However, cross-country emission data, whether in the form of GHG emissions, carbon intensity or energy use, are not available at a more disaggregate level.\n\n0 1 000 2 000 3 000 4 000 5 000 6 000 7 000 8 000 9 000\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n35\n\nCrop and animal production, hunting\n\nElectricity, gas, air conditioning\n\nSewerage; waste; remediation\n\nOth. non-metallic mineral products\n\nAir transport\n\nManuf. basic metals\n\nWater transport\n\nCoke and refined petroleum\n\nManuf. of chemicals\n\nFishing and aquaculture\n\nMining and quarrying\n\nLand transport; pipelines\n\nPaper and paper products\n\nForestry and logging\n\nWater collection, treatment and supply\n\nTextiles, wearing apparel, leather\n\nFood products, beverages and tobacco\n\nPostal and courier activities\n\nMetal products, exc. machinery and equip.\n\nAccommodation and food service\n\nRubber and plastic products\n\nWood and cork, exc. furniture\n\nConstruction\n\nBasic pharmaceutical products\n\nWholesale and retail trade; repair\n\nMachinery and equipment n.e.c.\n\nWarehousing and support for transportation\n\nPrinting and reproduction of recorded media\n\nWholesale trade, exc. motor vehicles\n\nOther service activities\n\nElectrical equipment\n\nRetail trade, exc. of motor vehicles\n\nComputers, electronic and optical products\n\nOther transport equipment\n\nFurniture; other manufacturing\n\nMotor vehicles, trailers and semi-trailers\n\nRepair and installation of machinery and…\n\nAdministrative and support service activities\n\nArts, entertainment and recreation\n\nOther professional, scientific and technical…\n\nPublic admin., defence; compulsory social…\n\nHuman health and social work\n\nArchitectural and engineering\n\nMotion picture, video, sound; braoadcasting\n\nAdvertising and market research\n\nScientific research and development\n\nLegal and accounting.; management…\n\nEducation\n\nPublishing activities\n\nActivities auxiliary to finance and insurance\n\nTelecommunications\n\nInsurance and pension, exc. compulsory…\n\nFinance exc. insurance and pension\n\nComputer programming; information services\n\nReal estate\n\nGHG intensity Frequency of countries in the\n\ntop two deciles\n\nFrequency: emission-intensive Rest of the economy GHG intensity: emission-intensive Rest of the economy"],"full_content":" **141**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Figure 3.1. Classification of high- and low-emission industries**\n\nGHG intensity within 2-digit NACE rev.2 sectors and frequency in top 2 deciles across 31 OECD countries, 2009-20 average\n\nNote: GHG intensity is expressed as CO2- equivalent emissions in tonnes per million EUR of gross value added. Source: Eurostat – Air emissions accounts by NACE Rev. 2 activity.\n\nStatLink 2 https://stat.link/3xgc2z\n\nThe list of sectors included in the definition encompass energy supply, as well as transport services (water, air and land), mining and manufacturing of energy-intensive products (basic metals, non-metallic mineral products, refined petroleum products, chemical- and paper products). The sewerage and waste-collection sector is omitted from the final list of high-emission industries as it is not expected to decline and indeed is a much needed industry for the net-zero transition (see Borgonovi et al. (2023[1])). The agricultural sector is entirely omitted from the list of industries studied in this chapter for two reasons. First, mass layoffs in agriculture are more likely to capture seasonal rather than structural adjustments in employment. Second, employment projections for many OECD countries do not suggest a decline in employment growth in the agricultural sector, in contrast to most other emission-intensive sectors, as current policy packages tend to shelter agricultural production from the climate mitigation effort (Borgonovi et al., 2023[1]).\n\nThe classification of “high-emission industries” shows not only significant consistency across the countries considered, but also across different measures of emission intensity, such as carbon intensity or fossil fuel energy intensity (see Barreto et al. (forthcoming[11]) for details). A key limitation of the present approach is the classification of industries based 2-digit ISIC rev. 4 sectors. This does not allow disentangling subsectors which may be expected to expand because of the net-zero transition such as renewable energy power generation or electrified land transport from sectors that are expected to contract. However, cross-country emission data, whether in the form of GHG emissions, carbon intensity or energy use, are not available at a more disaggregate level.\n\n0 1 000 2 000 3 000 4 000 5 000 6 000 7 000 8 000 9 000\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n35\n\nCrop and animal production, hunting\n\nElectricity, gas, air conditioning\n\nSewerage; waste; remediation\n\nOth. non-metallic mineral products\n\nAir transport\n\nManuf. basic metals\n\nWater transport\n\nCoke and refined petroleum\n\nManuf. of chemicals\n\nFishing and aquaculture\n\nMining and quarrying\n\nLand transport; pipelines\n\nPaper and paper products\n\nForestry and logging\n\nWater collection, treatment and supply\n\nTextiles, wearing apparel, leather\n\nFood products, beverages and tobacco\n\nPostal and courier activities\n\nMetal products, exc. machinery and equip.\n\nAccommodation and food service\n\nRubber and plastic products\n\nWood and cork, exc. furniture\n\nConstruction\n\nBasic pharmaceutical products\n\nWholesale and retail trade; repair\n\nMachinery and equipment n.e.c.\n\nWarehousing and support for transportation\n\nPrinting and reproduction of recorded media\n\nWholesale trade, exc. motor vehicles\n\nOther service activities\n\nElectrical equipment\n\nRetail trade, exc. of motor vehicles\n\nComputers, electronic and optical products\n\nOther transport equipment\n\nFurniture; other manufacturing\n\nMotor vehicles, trailers and semi-trailers\n\nRepair and installation of machinery and…\n\nAdministrative and support service activities\n\nArts, entertainment and recreation\n\nOther professional, scientific and technical…\n\nPublic admin., defence; compulsory social…\n\nHuman health and social work\n\nArchitectural and engineering\n\nMotion picture, video, sound; braoadcasting\n\nAdvertising and market research\n\nScientific research and development\n\nLegal and accounting.; management…\n\nEducation\n\nPublishing activities\n\nActivities auxiliary to finance and insurance\n\nTelecommunications\n\nInsurance and pension, exc. compulsory…\n\nFinance exc. insurance and pension\n\nComputer programming; information services\n\nReal estate\n\nGHG intensity Frequency of countries in the\n\ntop two deciles\n\nFrequency: emission-intensive Rest of the economy GHG intensity: emission-intensive Rest of the economy","publish_date":null,"title":"Figure 3.1. Classification of high- and low-emission industries","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p143.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p143.pdf"},"request":{"example_id":"chart/ac8b3538-en_p143","product_type":"parse","source_file_path":"ac8b3538-en_p143.pdf"},"started_at":"2026-11-04T19:46:19.922907"} +{"completed_at":"2026-11-04T19:46:23.104709","latency_in_ms":1337,"output":{"example_id":"chart/ac8b3538-en_p144","grounded_pages":[],"layout_pages":[],"markdown":"**142** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n**_3\\.1.1. Emissions are highly concentrated in specific sectors accounting for just 7% of employment_**\n\nHigh-emission industries, as defined in this chapter, account for about 80% of total GHG emissions across the OECD. There is some moderate variation across countries due to differences in the national energy mix, the size of high-emission industries and emission regulations. For instance, while high-emission industries in France and the Slovak Republic emit slightly less than 70% of national GHG emissions, the same sectors emit slightly more than 90% in the Czech Republic, Denmark, Greece, Finland, Iceland and Norway (Figure 3.2). Despite their significant contribution to emissions, high-emission industries employ only about 7% of the total workforce on average among OECD countries, suggesting that emissions are highly concentrated (Figure 3.2). Importantly, this fact implies that any future reductions in emissions will be concentrated in a comparatively small segment of the labour market. This is particularly true in Ireland, the Netherlands and the United Kingdom, where the share of the workforce in high-emission sectors is just below 5%, but even in Poland, the country with the highest share, where 79% of emissions is concentrated in industries representing 12% of the workforce.\n\n### **Figure 3.2. High-emission industries are responsible for most emissions but employ only a fraction of the workforce**\n\nShare of GHG emissions and employment in high-emission industries by country, 2019\\*\n\nNote: \\* Data refer to 2016 and to CO2 emissions for Mexico and the United States. Average across 26 OECD countries shown. Agriculture is excluded. Source: OECD National Accounts and Eurostat Air Emissions Accounts.\n\nStatLink 2 https://stat.link/ejxuol\n\n**_3\\.1.2. Employment in high-emission industries has been declining and the decline is expected to accelerate in the years to come_**\n\nReducing GHG emissions can be achieved by reducing economic activity in high-emission sectors as well as by changing the production technology within these sectors – see also OECD (2023[14]). Both processes require restructuring that may involve job losses and arising opportunities. Alongside previous declines in GHG emissions, employment in high-emission industries has indeed been shrinking substantially in EU countries during the period 2000-19, while employment in other sectors has continued to expand (see Figure 3.4) – in some cases also through green investment coupled with the explicit aim of creating quality\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\n70\n\n80\n\n90\n\n100\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\n16\n\n18\n\n20\n\nShare in total emissions, % Share in employment, %\n\nShare in employment Share in emissions (right scale)","pages":[{"line_numbers":[],"markdown":"**142** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n**_3\\.1.1. Emissions are highly concentrated in specific sectors accounting for just 7% of employment_**\n\nHigh-emission industries, as defined in this chapter, account for about 80% of total GHG emissions across the OECD. There is some moderate variation across countries due to differences in the national energy mix, the size of high-emission industries and emission regulations. For instance, while high-emission industries in France and the Slovak Republic emit slightly less than 70% of national GHG emissions, the same sectors emit slightly more than 90% in the Czech Republic, Denmark, Greece, Finland, Iceland and Norway (Figure 3.2). Despite their significant contribution to emissions, high-emission industries employ only about 7% of the total workforce on average among OECD countries, suggesting that emissions are highly concentrated (Figure 3.2). Importantly, this fact implies that any future reductions in emissions will be concentrated in a comparatively small segment of the labour market. This is particularly true in Ireland, the Netherlands and the United Kingdom, where the share of the workforce in high-emission sectors is just below 5%, but even in Poland, the country with the highest share, where 79% of emissions is concentrated in industries representing 12% of the workforce.\n\n### **Figure 3.2. High-emission industries are responsible for most emissions but employ only a fraction of the workforce**\n\nShare of GHG emissions and employment in high-emission industries by country, 2019\\*\n\nNote: \\* Data refer to 2016 and to CO2 emissions for Mexico and the United States. Average across 26 OECD countries shown. Agriculture is excluded. Source: OECD National Accounts and Eurostat Air Emissions Accounts.\n\nStatLink 2 https://stat.link/ejxuol\n\n**_3\\.1.2. Employment in high-emission industries has been declining and the decline is expected to accelerate in the years to come_**\n\nReducing GHG emissions can be achieved by reducing economic activity in high-emission sectors as well as by changing the production technology within these sectors – see also OECD (2023[14]). Both processes require restructuring that may involve job losses and arising opportunities. Alongside previous declines in GHG emissions, employment in high-emission industries has indeed been shrinking substantially in EU countries during the period 2000-19, while employment in other sectors has continued to expand (see Figure 3.4) – in some cases also through green investment coupled with the explicit aim of creating quality\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\n70\n\n80\n\n90\n\n100\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\n16\n\n18\n\n20\n\nShare in total emissions, % Share in employment, %\n\nShare in employment Share in emissions (right scale)","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**142** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n**_3\\.1.1. Emissions are highly concentrated in specific sectors accounting for just 7% of employment_**\n\nHigh-emission industries, as defined in this chapter, account for about 80% of total GHG emissions across the OECD. There is some moderate variation across countries due to differences in the national energy mix, the size of high-emission industries and emission regulations. For instance, while high-emission industries in France and the Slovak Republic emit slightly less than 70% of national GHG emissions, the same sectors emit slightly more than 90% in the Czech Republic, Denmark, Greece, Finland, Iceland and Norway (Figure 3.2). Despite their significant contribution to emissions, high-emission industries employ only about 7% of the total workforce on average among OECD countries, suggesting that emissions are highly concentrated (Figure 3.2). Importantly, this fact implies that any future reductions in emissions will be concentrated in a comparatively small segment of the labour market. This is particularly true in Ireland, the Netherlands and the United Kingdom, where the share of the workforce in high-emission sectors is just below 5%, but even in Poland, the country with the highest share, where 79% of emissions is concentrated in industries representing 12% of the workforce.\n\n### **Figure 3.2. High-emission industries are responsible for most emissions but employ only a fraction of the workforce**\n\nShare of GHG emissions and employment in high-emission industries by country, 2019\\*\n\nNote: \\* Data refer to 2016 and to CO2 emissions for Mexico and the United States. Average across 26 OECD countries shown. Agriculture is excluded. Source: OECD National Accounts and Eurostat Air Emissions Accounts.\n\nStatLink 2 https://stat.link/ejxuol\n\n**_3\\.1.2. Employment in high-emission industries has been declining and the decline is expected to accelerate in the years to come_**\n\nReducing GHG emissions can be achieved by reducing economic activity in high-emission sectors as well as by changing the production technology within these sectors – see also OECD (2023[14]). Both processes require restructuring that may involve job losses and arising opportunities. Alongside previous declines in GHG emissions, employment in high-emission industries has indeed been shrinking substantially in EU countries during the period 2000-19, while employment in other sectors has continued to expand (see Figure 3.4) – in some cases also through green investment coupled with the explicit aim of creating quality\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\n70\n\n80\n\n90\n\n100\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\n16\n\n18\n\n20\n\nShare in total emissions, % Share in employment, %\n\nShare in employment Share in emissions (right scale)","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p144.pdf"]},"response":{"errors":[],"extract_id":"extract_942c6dee6d1a19042ccf4901a761a989","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**142** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n**_3\\.1.1. Emissions are highly concentrated in specific sectors accounting for just 7% of employment_**\n\nHigh-emission industries, as defined in this chapter, account for about 80% of total GHG emissions across the OECD. There is some moderate variation across countries due to differences in the national energy mix, the size of high-emission industries and emission regulations. For instance, while high-emission industries in France and the Slovak Republic emit slightly less than 70% of national GHG emissions, the same sectors emit slightly more than 90% in the Czech Republic, Denmark, Greece, Finland, Iceland and Norway (Figure 3.2). Despite their significant contribution to emissions, high-emission industries employ only about 7% of the total workforce on average among OECD countries, suggesting that emissions are highly concentrated (Figure 3.2). Importantly, this fact implies that any future reductions in emissions will be concentrated in a comparatively small segment of the labour market. This is particularly true in Ireland, the Netherlands and the United Kingdom, where the share of the workforce in high-emission sectors is just below 5%, but even in Poland, the country with the highest share, where 79% of emissions is concentrated in industries representing 12% of the workforce.\n\n### **Figure 3.2. High-emission industries are responsible for most emissions but employ only a fraction of the workforce**\n\nShare of GHG emissions and employment in high-emission industries by country, 2019\\*\n\nNote: \\* Data refer to 2016 and to CO2 emissions for Mexico and the United States. Average across 26 OECD countries shown. Agriculture is excluded. Source: OECD National Accounts and Eurostat Air Emissions Accounts.\n\nStatLink 2 https://stat.link/ejxuol\n\n**_3\\.1.2. Employment in high-emission industries has been declining and the decline is expected to accelerate in the years to come_**\n\nReducing GHG emissions can be achieved by reducing economic activity in high-emission sectors as well as by changing the production technology within these sectors – see also OECD (2023[14]). Both processes require restructuring that may involve job losses and arising opportunities. Alongside previous declines in GHG emissions, employment in high-emission industries has indeed been shrinking substantially in EU countries during the period 2000-19, while employment in other sectors has continued to expand (see Figure 3.4) – in some cases also through green investment coupled with the explicit aim of creating quality\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\n70\n\n80\n\n90\n\n100\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\n16\n\n18\n\n20\n\nShare in total emissions, % Share in employment, %\n\nShare in employment Share in emissions (right scale)"],"full_content":"**142** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n**_3\\.1.1. Emissions are highly concentrated in specific sectors accounting for just 7% of employment_**\n\nHigh-emission industries, as defined in this chapter, account for about 80% of total GHG emissions across the OECD. There is some moderate variation across countries due to differences in the national energy mix, the size of high-emission industries and emission regulations. For instance, while high-emission industries in France and the Slovak Republic emit slightly less than 70% of national GHG emissions, the same sectors emit slightly more than 90% in the Czech Republic, Denmark, Greece, Finland, Iceland and Norway (Figure 3.2). Despite their significant contribution to emissions, high-emission industries employ only about 7% of the total workforce on average among OECD countries, suggesting that emissions are highly concentrated (Figure 3.2). Importantly, this fact implies that any future reductions in emissions will be concentrated in a comparatively small segment of the labour market. This is particularly true in Ireland, the Netherlands and the United Kingdom, where the share of the workforce in high-emission sectors is just below 5%, but even in Poland, the country with the highest share, where 79% of emissions is concentrated in industries representing 12% of the workforce.\n\n### **Figure 3.2. High-emission industries are responsible for most emissions but employ only a fraction of the workforce**\n\nShare of GHG emissions and employment in high-emission industries by country, 2019\\*\n\nNote: \\* Data refer to 2016 and to CO2 emissions for Mexico and the United States. Average across 26 OECD countries shown. Agriculture is excluded. Source: OECD National Accounts and Eurostat Air Emissions Accounts.\n\nStatLink 2 https://stat.link/ejxuol\n\n**_3\\.1.2. Employment in high-emission industries has been declining and the decline is expected to accelerate in the years to come_**\n\nReducing GHG emissions can be achieved by reducing economic activity in high-emission sectors as well as by changing the production technology within these sectors – see also OECD (2023[14]). Both processes require restructuring that may involve job losses and arising opportunities. Alongside previous declines in GHG emissions, employment in high-emission industries has indeed been shrinking substantially in EU countries during the period 2000-19, while employment in other sectors has continued to expand (see Figure 3.4) – in some cases also through green investment coupled with the explicit aim of creating quality\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\n70\n\n80\n\n90\n\n100\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\n16\n\n18\n\n20\n\nShare in total emissions, % Share in employment, %\n\nShare in employment Share in emissions (right scale)","publish_date":null,"title":"Figure 3.2. High-emission industries are responsible for most emissions but employ only a fraction of the workforce","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p144.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p144.pdf"},"request":{"example_id":"chart/ac8b3538-en_p144","product_type":"parse","source_file_path":"ac8b3538-en_p144.pdf"},"started_at":"2026-11-04T19:46:21.767513"} +{"completed_at":"2026-11-04T19:46:26.394090","latency_in_ms":3287,"output":{"example_id":"chart/ac8b3538-en_p147","grounded_pages":[],"layout_pages":[],"markdown":" **145**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\ndeclines are expected to markedly accelerate with respect to the past. In the EU, for example, the projected declines in the number of employed, which are expected to average 2.3% per year, are more than twice as large as the historical employment decline, which averaged 1.1% between 2000 and 2019 (Figure 3.4).\n\nThe projected pace of decline under EU Fit-for-55 also differs across high-emission industries. For example, employment in the manufacturing of chemicals as well as coke and refined petroleum is expected to decrease by 3.9% and 5.4% on average per year, respectively, while employment in manufacturing of basic metals and other non-metallic minerals as well as in mining and quarrying is expected to decrease by 2.5% and 3.3% on average per year. The manufacturing of paper as well as air and water transport see somewhat smaller average annual declines of just under 2% of employment, while land transport is projected to decline by on average 0.3% per year. Notably, the projections point to an increase in employment in electricity, gas, steam and air conditioning supply, which is a result of a strong increase in the use of greener sources of electricity generation in the energy mix of EU countries (which are bundled together with more GHG-intensive energy production in this industry) as evident from the more finely grained projections by Borgonovi et al. (2023[1]) or IEA (2023[26]). For example, employment in fossil-fuel powered electricity generation is projected to decline by about 80%, while employment in renewable and nuclear electricity generation is expected to substitute for it with an increase by 80%.\n\n### **Figure 3.4. The speed of employment declines in high-emission sectors is projected to increase significantly**\n\nPast average annual employment changes between 2000 and 2019 and projected average annual employment changes between 2019 and 2030 under a 55% emission reduction for EU countries relative to 1990 levels (Fit for 55), average across countries, percentage\n\nNote: The figure shows historical changes in employment between 2000 and 2019 as well as projected employment changes between 2019 and 2030, accounting for the Fit for 55 targets to reduce CO2 – equivalent emissions by 55% in 2030 compared to 1990 levels. Agriculture is excluded. More information can be found in Borgonovi et al. (2023[1]). Source: ENV-Linkages model.\n\nStatLink 2 https://stat.link/iadv1g\n\n**_3\\.1.3. Workers in high-emissions sectors are well paid despite comparatively low education, more likely to be male and live in rural areas_**\n\nThe characteristics of workers in high-emission industries differ notably from those of workers in low- emission industries (Figure 3.5, Panel A). For example, workers in high-emission industries are more likely to have low educational attainment (20% versus 15%) and less likely to have high educational attainment (25% versus 40%) than workers in low-emission industries. Furthermore, workers in high-emission\n\n\\-6\n\n\\-5\n\n\\-4\n\n\\-3\n\n\\-2\n\n\\-1\n\n0\n\n1\n\n2\n\nLow\n\nemission\n\nHigh\n\nemission\n\nElectricity, gas, steam\n\nand air\n\nconditioning\n\nsupply\n\nLand\n\ntransport and transport via\n\npipelines\n\nWater\n\ntransport\n\nManufacture\n\nof paper\n\nAir\n\ntransport\n\nManufacture\n\nof\n\nbasic metals\n\nMining and quarrying\n\nManufacture\n\nof other\n\nnon-metallic\n\nmineral products\n\nManufacture\n\nof\n\nchemicals\n\nManufacture\n\nof coke\n\nand\n\nrefined\n\npetroleum\n\n% Past (2000-19) Projections (2019-30)","pages":[{"line_numbers":[],"markdown":" **145**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\ndeclines are expected to markedly accelerate with respect to the past. In the EU, for example, the projected declines in the number of employed, which are expected to average 2.3% per year, are more than twice as large as the historical employment decline, which averaged 1.1% between 2000 and 2019 (Figure 3.4).\n\nThe projected pace of decline under EU Fit-for-55 also differs across high-emission industries. For example, employment in the manufacturing of chemicals as well as coke and refined petroleum is expected to decrease by 3.9% and 5.4% on average per year, respectively, while employment in manufacturing of basic metals and other non-metallic minerals as well as in mining and quarrying is expected to decrease by 2.5% and 3.3% on average per year. The manufacturing of paper as well as air and water transport see somewhat smaller average annual declines of just under 2% of employment, while land transport is projected to decline by on average 0.3% per year. Notably, the projections point to an increase in employment in electricity, gas, steam and air conditioning supply, which is a result of a strong increase in the use of greener sources of electricity generation in the energy mix of EU countries (which are bundled together with more GHG-intensive energy production in this industry) as evident from the more finely grained projections by Borgonovi et al. (2023[1]) or IEA (2023[26]). For example, employment in fossil-fuel powered electricity generation is projected to decline by about 80%, while employment in renewable and nuclear electricity generation is expected to substitute for it with an increase by 80%.\n\n### **Figure 3.4. The speed of employment declines in high-emission sectors is projected to increase significantly**\n\nPast average annual employment changes between 2000 and 2019 and projected average annual employment changes between 2019 and 2030 under a 55% emission reduction for EU countries relative to 1990 levels (Fit for 55), average across countries, percentage\n\nNote: The figure shows historical changes in employment between 2000 and 2019 as well as projected employment changes between 2019 and 2030, accounting for the Fit for 55 targets to reduce CO2 – equivalent emissions by 55% in 2030 compared to 1990 levels. Agriculture is excluded. More information can be found in Borgonovi et al. (2023[1]). Source: ENV-Linkages model.\n\nStatLink 2 https://stat.link/iadv1g\n\n**_3\\.1.3. Workers in high-emissions sectors are well paid despite comparatively low education, more likely to be male and live in rural areas_**\n\nThe characteristics of workers in high-emission industries differ notably from those of workers in low- emission industries (Figure 3.5, Panel A). For example, workers in high-emission industries are more likely to have low educational attainment (20% versus 15%) and less likely to have high educational attainment (25% versus 40%) than workers in low-emission industries. Furthermore, workers in high-emission\n\n\\-6\n\n\\-5\n\n\\-4\n\n\\-3\n\n\\-2\n\n\\-1\n\n0\n\n1\n\n2\n\nLow\n\nemission\n\nHigh\n\nemission\n\nElectricity, gas, steam\n\nand air\n\nconditioning\n\nsupply\n\nLand\n\ntransport and transport via\n\npipelines\n\nWater\n\ntransport\n\nManufacture\n\nof paper\n\nAir\n\ntransport\n\nManufacture\n\nof\n\nbasic metals\n\nMining and quarrying\n\nManufacture\n\nof other\n\nnon-metallic\n\nmineral products\n\nManufacture\n\nof\n\nchemicals\n\nManufacture\n\nof coke\n\nand\n\nrefined\n\npetroleum\n\n% Past (2000-19) Projections (2019-30)","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":" **145**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\ndeclines are expected to markedly accelerate with respect to the past. In the EU, for example, the projected declines in the number of employed, which are expected to average 2.3% per year, are more than twice as large as the historical employment decline, which averaged 1.1% between 2000 and 2019 (Figure 3.4).\n\nThe projected pace of decline under EU Fit-for-55 also differs across high-emission industries. For example, employment in the manufacturing of chemicals as well as coke and refined petroleum is expected to decrease by 3.9% and 5.4% on average per year, respectively, while employment in manufacturing of basic metals and other non-metallic minerals as well as in mining and quarrying is expected to decrease by 2.5% and 3.3% on average per year. The manufacturing of paper as well as air and water transport see somewhat smaller average annual declines of just under 2% of employment, while land transport is projected to decline by on average 0.3% per year. Notably, the projections point to an increase in employment in electricity, gas, steam and air conditioning supply, which is a result of a strong increase in the use of greener sources of electricity generation in the energy mix of EU countries (which are bundled together with more GHG-intensive energy production in this industry) as evident from the more finely grained projections by Borgonovi et al. (2023[1]) or IEA (2023[26]). For example, employment in fossil-fuel powered electricity generation is projected to decline by about 80%, while employment in renewable and nuclear electricity generation is expected to substitute for it with an increase by 80%.\n\n### **Figure 3.4. The speed of employment declines in high-emission sectors is projected to increase significantly**\n\nPast average annual employment changes between 2000 and 2019 and projected average annual employment changes between 2019 and 2030 under a 55% emission reduction for EU countries relative to 1990 levels (Fit for 55), average across countries, percentage\n\nNote: The figure shows historical changes in employment between 2000 and 2019 as well as projected employment changes between 2019 and 2030, accounting for the Fit for 55 targets to reduce CO2 – equivalent emissions by 55% in 2030 compared to 1990 levels. Agriculture is excluded. More information can be found in Borgonovi et al. (2023[1]). Source: ENV-Linkages model.\n\nStatLink 2 https://stat.link/iadv1g\n\n**_3\\.1.3. Workers in high-emissions sectors are well paid despite comparatively low education, more likely to be male and live in rural areas_**\n\nThe characteristics of workers in high-emission industries differ notably from those of workers in low- emission industries (Figure 3.5, Panel A). For example, workers in high-emission industries are more likely to have low educational attainment (20% versus 15%) and less likely to have high educational attainment (25% versus 40%) than workers in low-emission industries. Furthermore, workers in high-emission\n\n\\-6\n\n\\-5\n\n\\-4\n\n\\-3\n\n\\-2\n\n\\-1\n\n0\n\n1\n\n2\n\nLow\n\nemission\n\nHigh\n\nemission\n\nElectricity, gas, steam\n\nand air\n\nconditioning\n\nsupply\n\nLand\n\ntransport and transport via\n\npipelines\n\nWater\n\ntransport\n\nManufacture\n\nof paper\n\nAir\n\ntransport\n\nManufacture\n\nof\n\nbasic metals\n\nMining and quarrying\n\nManufacture\n\nof other\n\nnon-metallic\n\nmineral products\n\nManufacture\n\nof\n\nchemicals\n\nManufacture\n\nof coke\n\nand\n\nrefined\n\npetroleum\n\n% Past (2000-19) Projections (2019-30)","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p147.pdf"]},"response":{"errors":[],"extract_id":"extract_ae1e67b3fd050d578b118162842c396b","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":[" **145**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\ndeclines are expected to markedly accelerate with respect to the past. In the EU, for example, the projected declines in the number of employed, which are expected to average 2.3% per year, are more than twice as large as the historical employment decline, which averaged 1.1% between 2000 and 2019 (Figure 3.4).\n\nThe projected pace of decline under EU Fit-for-55 also differs across high-emission industries. For example, employment in the manufacturing of chemicals as well as coke and refined petroleum is expected to decrease by 3.9% and 5.4% on average per year, respectively, while employment in manufacturing of basic metals and other non-metallic minerals as well as in mining and quarrying is expected to decrease by 2.5% and 3.3% on average per year. The manufacturing of paper as well as air and water transport see somewhat smaller average annual declines of just under 2% of employment, while land transport is projected to decline by on average 0.3% per year. Notably, the projections point to an increase in employment in electricity, gas, steam and air conditioning supply, which is a result of a strong increase in the use of greener sources of electricity generation in the energy mix of EU countries (which are bundled together with more GHG-intensive energy production in this industry) as evident from the more finely grained projections by Borgonovi et al. (2023[1]) or IEA (2023[26]). For example, employment in fossil-fuel powered electricity generation is projected to decline by about 80%, while employment in renewable and nuclear electricity generation is expected to substitute for it with an increase by 80%.\n\n### **Figure 3.4. The speed of employment declines in high-emission sectors is projected to increase significantly**\n\nPast average annual employment changes between 2000 and 2019 and projected average annual employment changes between 2019 and 2030 under a 55% emission reduction for EU countries relative to 1990 levels (Fit for 55), average across countries, percentage\n\nNote: The figure shows historical changes in employment between 2000 and 2019 as well as projected employment changes between 2019 and 2030, accounting for the Fit for 55 targets to reduce CO2 – equivalent emissions by 55% in 2030 compared to 1990 levels. Agriculture is excluded. More information can be found in Borgonovi et al. (2023[1]). Source: ENV-Linkages model.\n\nStatLink 2 https://stat.link/iadv1g\n\n**_3\\.1.3. Workers in high-emissions sectors are well paid despite comparatively low education, more likely to be male and live in rural areas_**\n\nThe characteristics of workers in high-emission industries differ notably from those of workers in low- emission industries (Figure 3.5, Panel A). For example, workers in high-emission industries are more likely to have low educational attainment (20% versus 15%) and less likely to have high educational attainment (25% versus 40%) than workers in low-emission industries. Furthermore, workers in high-emission\n\n\\-6\n\n\\-5\n\n\\-4\n\n\\-3\n\n\\-2\n\n\\-1\n\n0\n\n1\n\n2\n\nLow\n\nemission\n\nHigh\n\nemission\n\nElectricity, gas, steam\n\nand air\n\nconditioning\n\nsupply\n\nLand\n\ntransport and transport via\n\npipelines\n\nWater\n\ntransport\n\nManufacture\n\nof paper\n\nAir\n\ntransport\n\nManufacture\n\nof\n\nbasic metals\n\nMining and quarrying\n\nManufacture\n\nof other\n\nnon-metallic\n\nmineral products\n\nManufacture\n\nof\n\nchemicals\n\nManufacture\n\nof coke\n\nand\n\nrefined\n\npetroleum\n\n% Past (2000-19) Projections (2019-30)"],"full_content":" **145**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\ndeclines are expected to markedly accelerate with respect to the past. In the EU, for example, the projected declines in the number of employed, which are expected to average 2.3% per year, are more than twice as large as the historical employment decline, which averaged 1.1% between 2000 and 2019 (Figure 3.4).\n\nThe projected pace of decline under EU Fit-for-55 also differs across high-emission industries. For example, employment in the manufacturing of chemicals as well as coke and refined petroleum is expected to decrease by 3.9% and 5.4% on average per year, respectively, while employment in manufacturing of basic metals and other non-metallic minerals as well as in mining and quarrying is expected to decrease by 2.5% and 3.3% on average per year. The manufacturing of paper as well as air and water transport see somewhat smaller average annual declines of just under 2% of employment, while land transport is projected to decline by on average 0.3% per year. Notably, the projections point to an increase in employment in electricity, gas, steam and air conditioning supply, which is a result of a strong increase in the use of greener sources of electricity generation in the energy mix of EU countries (which are bundled together with more GHG-intensive energy production in this industry) as evident from the more finely grained projections by Borgonovi et al. (2023[1]) or IEA (2023[26]). For example, employment in fossil-fuel powered electricity generation is projected to decline by about 80%, while employment in renewable and nuclear electricity generation is expected to substitute for it with an increase by 80%.\n\n### **Figure 3.4. The speed of employment declines in high-emission sectors is projected to increase significantly**\n\nPast average annual employment changes between 2000 and 2019 and projected average annual employment changes between 2019 and 2030 under a 55% emission reduction for EU countries relative to 1990 levels (Fit for 55), average across countries, percentage\n\nNote: The figure shows historical changes in employment between 2000 and 2019 as well as projected employment changes between 2019 and 2030, accounting for the Fit for 55 targets to reduce CO2 – equivalent emissions by 55% in 2030 compared to 1990 levels. Agriculture is excluded. More information can be found in Borgonovi et al. (2023[1]). Source: ENV-Linkages model.\n\nStatLink 2 https://stat.link/iadv1g\n\n**_3\\.1.3. Workers in high-emissions sectors are well paid despite comparatively low education, more likely to be male and live in rural areas_**\n\nThe characteristics of workers in high-emission industries differ notably from those of workers in low- emission industries (Figure 3.5, Panel A). For example, workers in high-emission industries are more likely to have low educational attainment (20% versus 15%) and less likely to have high educational attainment (25% versus 40%) than workers in low-emission industries. Furthermore, workers in high-emission\n\n\\-6\n\n\\-5\n\n\\-4\n\n\\-3\n\n\\-2\n\n\\-1\n\n0\n\n1\n\n2\n\nLow\n\nemission\n\nHigh\n\nemission\n\nElectricity, gas, steam\n\nand air\n\nconditioning\n\nsupply\n\nLand\n\ntransport and transport via\n\npipelines\n\nWater\n\ntransport\n\nManufacture\n\nof paper\n\nAir\n\ntransport\n\nManufacture\n\nof\n\nbasic metals\n\nMining and quarrying\n\nManufacture\n\nof other\n\nnon-metallic\n\nmineral products\n\nManufacture\n\nof\n\nchemicals\n\nManufacture\n\nof coke\n\nand\n\nrefined\n\npetroleum\n\n% Past (2000-19) Projections (2019-30)","publish_date":null,"title":"Figure 3.4. The speed of employment declines in high-emission sectors is projected to increase significantly","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p147.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p147.pdf"},"request":{"example_id":"chart/ac8b3538-en_p147","product_type":"parse","source_file_path":"ac8b3538-en_p147.pdf"},"started_at":"2026-11-04T19:46:23.106668"} +{"completed_at":"2026-11-04T19:46:25.789453","latency_in_ms":1327,"output":{"example_id":"chart/ac8b3538-en_p148","grounded_pages":[],"layout_pages":[],"markdown":"**146** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\nindustries also participate less frequently in formal and non-formal education and training programmes (Figure 3.5, Panel A). However, despite their relatively low educational attainment, they are less likely to earn low wages (i.e. a wage in the bottom two deciles of the wage distribution) and slightly more likely to earn high wages (i.e. a wage in the upper two deciles) than workers in other sectors. Indeed, higher wages in high-emission industries reflect higher firm wage premia, i.e. employers paying higher wages irrespective of workforce composition, rather than differences in skills among workers. ⁶ Workers in high- emission industries are further much more likely to be male, somewhat older and more likely to live in rural areas (Figure 3.5, Panel A). The combination of low skills, relatively high wages and living in rural areas, in particular, could have potentially important implications for the costs of job displacement.\n\n### **Figure 3.5. Workers in high-emission industries differ markedly from other workers**\n\nCharacteristics of workers in high-emission and low-emission sectors, EU-28 countries, 2018/19, percentage of each category\n\nNote: Data refer to the share of each category, expressed in percentages, except for age (average number of years). Low education refers to less than upper secondary, high education to more than upper secondary, tertiary. Participation in formal and non-formal education and training refers to participation over the 4 weeks prior to survey response in the EU-LFS. Cities refer to densely populated areas, rural areas refer to thinly populated areas. Towns and suburbs (intermediate density areas) are not shown. \\* High (low) wages refer to the top (bottom) two deciles of the hourly pay from the main job. \\*\\* High (low) wage premia refer to the top (bottom) two deciles of the firm-related wage premia distribution obtained from a regression of hourly pay onto categorical firm identifiers, age groups, gender, and education levels, and exclude Finland, Iceland, Luxembourg and Slovenia. Agriculture is excluded. Source: Eurostat, EU-LFS, 2019; Eurostat, Structure of Earnings Survey, 2018 (deciles).\n\nStatLink 2 https://stat.link/a710ng\n\n### **3\\.2. Analysing the consequences of job loss in high-emission industries**\n\nThis section provides an in-depth examination of the costs of job displacement in high-emission sectors based on harmonised linked employer-employee data from 14 OECD countries and a consistent econometric framework – see Barreto et al. (forthcoming[11]) for more details.\n\n**_3\\.2.1. Methodology and data for analysing job displacement in high-emission industries_**\n\nDetermining the cost of job displacement for a given worker poses some methodological challenges. Ideally, one would like to compare the outcomes of the same worker in two states of the world, one in which the worker loses employment and another in which the worker continues to be employed. As this is\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\nWomen Age Low-\n\neducated\n\nHigh-\n\neducated\n\nIn formal\n\neducation or\n\ntraining\n\nIn non formal\n\neducation and training\n\nRural Cities Low wage\\* High wage\\* Low wage\n\npremia\\*\\*\n\nHigh wage premia\\*\\*\n\n%\n\nHigh-emission industries Rest of the economy","pages":[{"line_numbers":[],"markdown":"**146** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\nindustries also participate less frequently in formal and non-formal education and training programmes (Figure 3.5, Panel A). However, despite their relatively low educational attainment, they are less likely to earn low wages (i.e. a wage in the bottom two deciles of the wage distribution) and slightly more likely to earn high wages (i.e. a wage in the upper two deciles) than workers in other sectors. Indeed, higher wages in high-emission industries reflect higher firm wage premia, i.e. employers paying higher wages irrespective of workforce composition, rather than differences in skills among workers. ⁶ Workers in high- emission industries are further much more likely to be male, somewhat older and more likely to live in rural areas (Figure 3.5, Panel A). The combination of low skills, relatively high wages and living in rural areas, in particular, could have potentially important implications for the costs of job displacement.\n\n### **Figure 3.5. Workers in high-emission industries differ markedly from other workers**\n\nCharacteristics of workers in high-emission and low-emission sectors, EU-28 countries, 2018/19, percentage of each category\n\nNote: Data refer to the share of each category, expressed in percentages, except for age (average number of years). Low education refers to less than upper secondary, high education to more than upper secondary, tertiary. Participation in formal and non-formal education and training refers to participation over the 4 weeks prior to survey response in the EU-LFS. Cities refer to densely populated areas, rural areas refer to thinly populated areas. Towns and suburbs (intermediate density areas) are not shown. \\* High (low) wages refer to the top (bottom) two deciles of the hourly pay from the main job. \\*\\* High (low) wage premia refer to the top (bottom) two deciles of the firm-related wage premia distribution obtained from a regression of hourly pay onto categorical firm identifiers, age groups, gender, and education levels, and exclude Finland, Iceland, Luxembourg and Slovenia. Agriculture is excluded. Source: Eurostat, EU-LFS, 2019; Eurostat, Structure of Earnings Survey, 2018 (deciles).\n\nStatLink 2 https://stat.link/a710ng\n\n### **3\\.2. Analysing the consequences of job loss in high-emission industries**\n\nThis section provides an in-depth examination of the costs of job displacement in high-emission sectors based on harmonised linked employer-employee data from 14 OECD countries and a consistent econometric framework – see Barreto et al. (forthcoming[11]) for more details.\n\n**_3\\.2.1. Methodology and data for analysing job displacement in high-emission industries_**\n\nDetermining the cost of job displacement for a given worker poses some methodological challenges. Ideally, one would like to compare the outcomes of the same worker in two states of the world, one in which the worker loses employment and another in which the worker continues to be employed. As this is\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\nWomen Age Low-\n\neducated\n\nHigh-\n\neducated\n\nIn formal\n\neducation or\n\ntraining\n\nIn non formal\n\neducation and training\n\nRural Cities Low wage\\* High wage\\* Low wage\n\npremia\\*\\*\n\nHigh wage premia\\*\\*\n\n%\n\nHigh-emission industries Rest of the economy","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**146** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\nindustries also participate less frequently in formal and non-formal education and training programmes (Figure 3.5, Panel A). However, despite their relatively low educational attainment, they are less likely to earn low wages (i.e. a wage in the bottom two deciles of the wage distribution) and slightly more likely to earn high wages (i.e. a wage in the upper two deciles) than workers in other sectors. Indeed, higher wages in high-emission industries reflect higher firm wage premia, i.e. employers paying higher wages irrespective of workforce composition, rather than differences in skills among workers. ⁶ Workers in high- emission industries are further much more likely to be male, somewhat older and more likely to live in rural areas (Figure 3.5, Panel A). The combination of low skills, relatively high wages and living in rural areas, in particular, could have potentially important implications for the costs of job displacement.\n\n### **Figure 3.5. Workers in high-emission industries differ markedly from other workers**\n\nCharacteristics of workers in high-emission and low-emission sectors, EU-28 countries, 2018/19, percentage of each category\n\nNote: Data refer to the share of each category, expressed in percentages, except for age (average number of years). Low education refers to less than upper secondary, high education to more than upper secondary, tertiary. Participation in formal and non-formal education and training refers to participation over the 4 weeks prior to survey response in the EU-LFS. Cities refer to densely populated areas, rural areas refer to thinly populated areas. Towns and suburbs (intermediate density areas) are not shown. \\* High (low) wages refer to the top (bottom) two deciles of the hourly pay from the main job. \\*\\* High (low) wage premia refer to the top (bottom) two deciles of the firm-related wage premia distribution obtained from a regression of hourly pay onto categorical firm identifiers, age groups, gender, and education levels, and exclude Finland, Iceland, Luxembourg and Slovenia. Agriculture is excluded. Source: Eurostat, EU-LFS, 2019; Eurostat, Structure of Earnings Survey, 2018 (deciles).\n\nStatLink 2 https://stat.link/a710ng\n\n### **3\\.2. Analysing the consequences of job loss in high-emission industries**\n\nThis section provides an in-depth examination of the costs of job displacement in high-emission sectors based on harmonised linked employer-employee data from 14 OECD countries and a consistent econometric framework – see Barreto et al. (forthcoming[11]) for more details.\n\n**_3\\.2.1. Methodology and data for analysing job displacement in high-emission industries_**\n\nDetermining the cost of job displacement for a given worker poses some methodological challenges. Ideally, one would like to compare the outcomes of the same worker in two states of the world, one in which the worker loses employment and another in which the worker continues to be employed. As this is\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\nWomen Age Low-\n\neducated\n\nHigh-\n\neducated\n\nIn formal\n\neducation or\n\ntraining\n\nIn non formal\n\neducation and training\n\nRural Cities Low wage\\* High wage\\* Low wage\n\npremia\\*\\*\n\nHigh wage premia\\*\\*\n\n%\n\nHigh-emission industries Rest of the economy","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p148.pdf"]},"response":{"errors":[],"extract_id":"extract_4457bece6d662e873087c1e16d4d3ea7","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**146** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\nindustries also participate less frequently in formal and non-formal education and training programmes (Figure 3.5, Panel A). However, despite their relatively low educational attainment, they are less likely to earn low wages (i.e. a wage in the bottom two deciles of the wage distribution) and slightly more likely to earn high wages (i.e. a wage in the upper two deciles) than workers in other sectors. Indeed, higher wages in high-emission industries reflect higher firm wage premia, i.e. employers paying higher wages irrespective of workforce composition, rather than differences in skills among workers. ⁶ Workers in high- emission industries are further much more likely to be male, somewhat older and more likely to live in rural areas (Figure 3.5, Panel A). The combination of low skills, relatively high wages and living in rural areas, in particular, could have potentially important implications for the costs of job displacement.\n\n### **Figure 3.5. Workers in high-emission industries differ markedly from other workers**\n\nCharacteristics of workers in high-emission and low-emission sectors, EU-28 countries, 2018/19, percentage of each category\n\nNote: Data refer to the share of each category, expressed in percentages, except for age (average number of years). Low education refers to less than upper secondary, high education to more than upper secondary, tertiary. Participation in formal and non-formal education and training refers to participation over the 4 weeks prior to survey response in the EU-LFS. Cities refer to densely populated areas, rural areas refer to thinly populated areas. Towns and suburbs (intermediate density areas) are not shown. \\* High (low) wages refer to the top (bottom) two deciles of the hourly pay from the main job. \\*\\* High (low) wage premia refer to the top (bottom) two deciles of the firm-related wage premia distribution obtained from a regression of hourly pay onto categorical firm identifiers, age groups, gender, and education levels, and exclude Finland, Iceland, Luxembourg and Slovenia. Agriculture is excluded. Source: Eurostat, EU-LFS, 2019; Eurostat, Structure of Earnings Survey, 2018 (deciles).\n\nStatLink 2 https://stat.link/a710ng\n\n### **3\\.2. Analysing the consequences of job loss in high-emission industries**\n\nThis section provides an in-depth examination of the costs of job displacement in high-emission sectors based on harmonised linked employer-employee data from 14 OECD countries and a consistent econometric framework – see Barreto et al. (forthcoming[11]) for more details.\n\n**_3\\.2.1. Methodology and data for analysing job displacement in high-emission industries_**\n\nDetermining the cost of job displacement for a given worker poses some methodological challenges. Ideally, one would like to compare the outcomes of the same worker in two states of the world, one in which the worker loses employment and another in which the worker continues to be employed. As this is\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\nWomen Age Low-\n\neducated\n\nHigh-\n\neducated\n\nIn formal\n\neducation or\n\ntraining\n\nIn non formal\n\neducation and training\n\nRural Cities Low wage\\* High wage\\* Low wage\n\npremia\\*\\*\n\nHigh wage premia\\*\\*\n\n%\n\nHigh-emission industries Rest of the economy"],"full_content":"**146** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\nindustries also participate less frequently in formal and non-formal education and training programmes (Figure 3.5, Panel A). However, despite their relatively low educational attainment, they are less likely to earn low wages (i.e. a wage in the bottom two deciles of the wage distribution) and slightly more likely to earn high wages (i.e. a wage in the upper two deciles) than workers in other sectors. Indeed, higher wages in high-emission industries reflect higher firm wage premia, i.e. employers paying higher wages irrespective of workforce composition, rather than differences in skills among workers. ⁶ Workers in high- emission industries are further much more likely to be male, somewhat older and more likely to live in rural areas (Figure 3.5, Panel A). The combination of low skills, relatively high wages and living in rural areas, in particular, could have potentially important implications for the costs of job displacement.\n\n### **Figure 3.5. Workers in high-emission industries differ markedly from other workers**\n\nCharacteristics of workers in high-emission and low-emission sectors, EU-28 countries, 2018/19, percentage of each category\n\nNote: Data refer to the share of each category, expressed in percentages, except for age (average number of years). Low education refers to less than upper secondary, high education to more than upper secondary, tertiary. Participation in formal and non-formal education and training refers to participation over the 4 weeks prior to survey response in the EU-LFS. Cities refer to densely populated areas, rural areas refer to thinly populated areas. Towns and suburbs (intermediate density areas) are not shown. \\* High (low) wages refer to the top (bottom) two deciles of the hourly pay from the main job. \\*\\* High (low) wage premia refer to the top (bottom) two deciles of the firm-related wage premia distribution obtained from a regression of hourly pay onto categorical firm identifiers, age groups, gender, and education levels, and exclude Finland, Iceland, Luxembourg and Slovenia. Agriculture is excluded. Source: Eurostat, EU-LFS, 2019; Eurostat, Structure of Earnings Survey, 2018 (deciles).\n\nStatLink 2 https://stat.link/a710ng\n\n### **3\\.2. Analysing the consequences of job loss in high-emission industries**\n\nThis section provides an in-depth examination of the costs of job displacement in high-emission sectors based on harmonised linked employer-employee data from 14 OECD countries and a consistent econometric framework – see Barreto et al. (forthcoming[11]) for more details.\n\n**_3\\.2.1. Methodology and data for analysing job displacement in high-emission industries_**\n\nDetermining the cost of job displacement for a given worker poses some methodological challenges. Ideally, one would like to compare the outcomes of the same worker in two states of the world, one in which the worker loses employment and another in which the worker continues to be employed. As this is\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\nWomen Age Low-\n\neducated\n\nHigh-\n\neducated\n\nIn formal\n\neducation or\n\ntraining\n\nIn non formal\n\neducation and training\n\nRural Cities Low wage\\* High wage\\* Low wage\n\npremia\\*\\*\n\nHigh wage premia\\*\\*\n\n%\n\nHigh-emission industries Rest of the economy","publish_date":null,"title":"Figure 3.5. Workers in high-emission industries differ markedly from other workers 3.2. Analysing the consequences of job loss in high-emission","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p148.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p148.pdf"},"request":{"example_id":"chart/ac8b3538-en_p148","product_type":"parse","source_file_path":"ac8b3538-en_p148.pdf"},"started_at":"2026-11-04T19:46:24.462270"} +{"completed_at":"2026-11-04T19:46:28.975958","latency_in_ms":2878,"output":{"example_id":"chart/ac8b3538-en_p153","grounded_pages":[],"layout_pages":[],"markdown":" **151**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\ndifferences in job displacement costs for workers in high- and low-emission industries (see Box 3.3 for details on the methodology of decomposing earnings losses). ⁷ Differences in days worked in the first year after displacement are mainly related to the return to employment later in the year (i.e. after 1 January), whereas differences in subsequent years mainly reflect lower job stability upon re-employment.\n\nOn average over the six years following the year of job displacement, the lower likelihood of being in employment is the main factor behind earnings losses in both high- and low-emission industries (Figure 3.7 Panel A and Panel B). However, the 6-year average masks considerable variation in the importance of the earnings components over time. Initially, in the first year following displacement, differences in employment overwhelmingly drive earnings losses as reflected by differences in the probability of being employed and the number of days worked upon re-employment later in the year. By contrast, differences in re-employment wages play a marginal role. Over time, however, wages become more important for explaining earnings losses after displacement in high- and low-emission industries (both in relative and absolute terms). This increasing importance of wages may be explained by the earlier re-employment of workers with the highest skills and earnings potential as well as the gradual adjustment of job search toward lower paying jobs over prolonged unemployment duration – see e.g. Maibom et al. (2023[36]) and Hijzen, Upward and Wright (2010[37]). The days worked component falls strongly after the first year following job loss. From then on, differences in days worked mainly reflect differences in job instability upon re-employment while differences in the timing of job finding following displacement are negligible. Displaced workers may face higher job instability upon re-employment as they are more likely to be employed on temporary contracts and have lower job tenure.\n\n### **Figure 3.7. Differences in earnings losses reflect larger wage losses, fewer days worked as well as longer spells out of work**\n\nContribution of daily wages, days worked and employment to overall earnings losses after job loss in high- and low- emission industries in percentage (Panel A and B) and the difference in earnings losses between high- and low- emission industries in percentage points (Panel C), average across countries\n\nNote: Average across countries of the contribution of daily wages, the probability of being employed, and days worked to the total earnings loss of displaced workers relative to observationally comparable non-displaced workers (see Box 3.3). The bars show the contribution of each margin to the overall earnings six years after job loss and on average over the six years following the year of job loss. The contributions of daily wages and days worked are conditional on being employed in a given year. The contribution of the probability of being employed captures the role of years in which no earnings from dependent employment are recorded in which case zero earnings are imputed. Point estimates from country- level regressions are averaged assigning each country an equal weight. The countries included are: Austria, Denmark, Finland, France, Germany, Hungary, the Netherlands, Portugal, Spain and Sweden. Due to missing daily wage and days worked information, Australia, Canada, Estonia and Norway are excluded. Source: National linked employer employee data, see Annex Table 3.B.1 for details.\n\nStatLink 2 https://stat.link/8izqk2\n\n\\-70\n\n\\-60\n\n\\-50\n\n\\-40\n\n\\-30\n\n\\-20\n\n\\-10\n\n0 %\n\n**A. High-emission industries**\n\n\\-60\n\n\\-50\n\n\\-40\n\n\\-30\n\n\\-20\n\n\\-10\n\n0 %\n\n**B. Low-emission industries**\n\n\\-10 -9 -8 -7 -6 -5 -4 -3 -2 -1 0 %\n\n**C. Difference**\n\nWages Employment Days worked","pages":[{"line_numbers":[],"markdown":" **151**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\ndifferences in job displacement costs for workers in high- and low-emission industries (see Box 3.3 for details on the methodology of decomposing earnings losses). ⁷ Differences in days worked in the first year after displacement are mainly related to the return to employment later in the year (i.e. after 1 January), whereas differences in subsequent years mainly reflect lower job stability upon re-employment.\n\nOn average over the six years following the year of job displacement, the lower likelihood of being in employment is the main factor behind earnings losses in both high- and low-emission industries (Figure 3.7 Panel A and Panel B). However, the 6-year average masks considerable variation in the importance of the earnings components over time. Initially, in the first year following displacement, differences in employment overwhelmingly drive earnings losses as reflected by differences in the probability of being employed and the number of days worked upon re-employment later in the year. By contrast, differences in re-employment wages play a marginal role. Over time, however, wages become more important for explaining earnings losses after displacement in high- and low-emission industries (both in relative and absolute terms). This increasing importance of wages may be explained by the earlier re-employment of workers with the highest skills and earnings potential as well as the gradual adjustment of job search toward lower paying jobs over prolonged unemployment duration – see e.g. Maibom et al. (2023[36]) and Hijzen, Upward and Wright (2010[37]). The days worked component falls strongly after the first year following job loss. From then on, differences in days worked mainly reflect differences in job instability upon re-employment while differences in the timing of job finding following displacement are negligible. Displaced workers may face higher job instability upon re-employment as they are more likely to be employed on temporary contracts and have lower job tenure.\n\n### **Figure 3.7. Differences in earnings losses reflect larger wage losses, fewer days worked as well as longer spells out of work**\n\nContribution of daily wages, days worked and employment to overall earnings losses after job loss in high- and low- emission industries in percentage (Panel A and B) and the difference in earnings losses between high- and low- emission industries in percentage points (Panel C), average across countries\n\nNote: Average across countries of the contribution of daily wages, the probability of being employed, and days worked to the total earnings loss of displaced workers relative to observationally comparable non-displaced workers (see Box 3.3). The bars show the contribution of each margin to the overall earnings six years after job loss and on average over the six years following the year of job loss. The contributions of daily wages and days worked are conditional on being employed in a given year. The contribution of the probability of being employed captures the role of years in which no earnings from dependent employment are recorded in which case zero earnings are imputed. Point estimates from country- level regressions are averaged assigning each country an equal weight. The countries included are: Austria, Denmark, Finland, France, Germany, Hungary, the Netherlands, Portugal, Spain and Sweden. Due to missing daily wage and days worked information, Australia, Canada, Estonia and Norway are excluded. Source: National linked employer employee data, see Annex Table 3.B.1 for details.\n\nStatLink 2 https://stat.link/8izqk2\n\n\\-70\n\n\\-60\n\n\\-50\n\n\\-40\n\n\\-30\n\n\\-20\n\n\\-10\n\n0 %\n\n**A. High-emission industries**\n\n\\-60\n\n\\-50\n\n\\-40\n\n\\-30\n\n\\-20\n\n\\-10\n\n0 %\n\n**B. Low-emission industries**\n\n\\-10 -9 -8 -7 -6 -5 -4 -3 -2 -1 0 %\n\n**C. Difference**\n\nWages Employment Days worked","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":" **151**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\ndifferences in job displacement costs for workers in high- and low-emission industries (see Box 3.3 for details on the methodology of decomposing earnings losses). ⁷ Differences in days worked in the first year after displacement are mainly related to the return to employment later in the year (i.e. after 1 January), whereas differences in subsequent years mainly reflect lower job stability upon re-employment.\n\nOn average over the six years following the year of job displacement, the lower likelihood of being in employment is the main factor behind earnings losses in both high- and low-emission industries (Figure 3.7 Panel A and Panel B). However, the 6-year average masks considerable variation in the importance of the earnings components over time. Initially, in the first year following displacement, differences in employment overwhelmingly drive earnings losses as reflected by differences in the probability of being employed and the number of days worked upon re-employment later in the year. By contrast, differences in re-employment wages play a marginal role. Over time, however, wages become more important for explaining earnings losses after displacement in high- and low-emission industries (both in relative and absolute terms). This increasing importance of wages may be explained by the earlier re-employment of workers with the highest skills and earnings potential as well as the gradual adjustment of job search toward lower paying jobs over prolonged unemployment duration – see e.g. Maibom et al. (2023[36]) and Hijzen, Upward and Wright (2010[37]). The days worked component falls strongly after the first year following job loss. From then on, differences in days worked mainly reflect differences in job instability upon re-employment while differences in the timing of job finding following displacement are negligible. Displaced workers may face higher job instability upon re-employment as they are more likely to be employed on temporary contracts and have lower job tenure.\n\n### **Figure 3.7. Differences in earnings losses reflect larger wage losses, fewer days worked as well as longer spells out of work**\n\nContribution of daily wages, days worked and employment to overall earnings losses after job loss in high- and low- emission industries in percentage (Panel A and B) and the difference in earnings losses between high- and low- emission industries in percentage points (Panel C), average across countries\n\nNote: Average across countries of the contribution of daily wages, the probability of being employed, and days worked to the total earnings loss of displaced workers relative to observationally comparable non-displaced workers (see Box 3.3). The bars show the contribution of each margin to the overall earnings six years after job loss and on average over the six years following the year of job loss. The contributions of daily wages and days worked are conditional on being employed in a given year. The contribution of the probability of being employed captures the role of years in which no earnings from dependent employment are recorded in which case zero earnings are imputed. Point estimates from country- level regressions are averaged assigning each country an equal weight. The countries included are: Austria, Denmark, Finland, France, Germany, Hungary, the Netherlands, Portugal, Spain and Sweden. Due to missing daily wage and days worked information, Australia, Canada, Estonia and Norway are excluded. Source: National linked employer employee data, see Annex Table 3.B.1 for details.\n\nStatLink 2 https://stat.link/8izqk2\n\n\\-70\n\n\\-60\n\n\\-50\n\n\\-40\n\n\\-30\n\n\\-20\n\n\\-10\n\n0 %\n\n**A. High-emission industries**\n\n\\-60\n\n\\-50\n\n\\-40\n\n\\-30\n\n\\-20\n\n\\-10\n\n0 %\n\n**B. Low-emission industries**\n\n\\-10 -9 -8 -7 -6 -5 -4 -3 -2 -1 0 %\n\n**C. Difference**\n\nWages Employment Days worked","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p153.pdf"]},"response":{"errors":[],"extract_id":"extract_9d1361ac1ae977c737fe6b643725122e","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":[" **151**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\ndifferences in job displacement costs for workers in high- and low-emission industries (see Box 3.3 for details on the methodology of decomposing earnings losses). ⁷ Differences in days worked in the first year after displacement are mainly related to the return to employment later in the year (i.e. after 1 January), whereas differences in subsequent years mainly reflect lower job stability upon re-employment.\n\nOn average over the six years following the year of job displacement, the lower likelihood of being in employment is the main factor behind earnings losses in both high- and low-emission industries (Figure 3.7 Panel A and Panel B). However, the 6-year average masks considerable variation in the importance of the earnings components over time. Initially, in the first year following displacement, differences in employment overwhelmingly drive earnings losses as reflected by differences in the probability of being employed and the number of days worked upon re-employment later in the year. By contrast, differences in re-employment wages play a marginal role. Over time, however, wages become more important for explaining earnings losses after displacement in high- and low-emission industries (both in relative and absolute terms). This increasing importance of wages may be explained by the earlier re-employment of workers with the highest skills and earnings potential as well as the gradual adjustment of job search toward lower paying jobs over prolonged unemployment duration – see e.g. Maibom et al. (2023[36]) and Hijzen, Upward and Wright (2010[37]). The days worked component falls strongly after the first year following job loss. From then on, differences in days worked mainly reflect differences in job instability upon re-employment while differences in the timing of job finding following displacement are negligible. Displaced workers may face higher job instability upon re-employment as they are more likely to be employed on temporary contracts and have lower job tenure.\n\n### **Figure 3.7. Differences in earnings losses reflect larger wage losses, fewer days worked as well as longer spells out of work**\n\nContribution of daily wages, days worked and employment to overall earnings losses after job loss in high- and low- emission industries in percentage (Panel A and B) and the difference in earnings losses between high- and low- emission industries in percentage points (Panel C), average across countries\n\nNote: Average across countries of the contribution of daily wages, the probability of being employed, and days worked to the total earnings loss of displaced workers relative to observationally comparable non-displaced workers (see Box 3.3). The bars show the contribution of each margin to the overall earnings six years after job loss and on average over the six years following the year of job loss. The contributions of daily wages and days worked are conditional on being employed in a given year. The contribution of the probability of being employed captures the role of years in which no earnings from dependent employment are recorded in which case zero earnings are imputed. Point estimates from country- level regressions are averaged assigning each country an equal weight. The countries included are: Austria, Denmark, Finland, France, Germany, Hungary, the Netherlands, Portugal, Spain and Sweden. Due to missing daily wage and days worked information, Australia, Canada, Estonia and Norway are excluded. Source: National linked employer employee data, see Annex Table 3.B.1 for details.\n\nStatLink 2 https://stat.link/8izqk2\n\n\\-70\n\n\\-60\n\n\\-50\n\n\\-40\n\n\\-30\n\n\\-20\n\n\\-10\n\n0 %\n\n**A. High-emission industries**\n\n\\-60\n\n\\-50\n\n\\-40\n\n\\-30\n\n\\-20\n\n\\-10\n\n0 %\n\n**B. Low-emission industries**\n\n\\-10 -9 -8 -7 -6 -5 -4 -3 -2 -1 0 %\n\n**C. Difference**\n\nWages Employment Days worked"],"full_content":" **151**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\ndifferences in job displacement costs for workers in high- and low-emission industries (see Box 3.3 for details on the methodology of decomposing earnings losses). ⁷ Differences in days worked in the first year after displacement are mainly related to the return to employment later in the year (i.e. after 1 January), whereas differences in subsequent years mainly reflect lower job stability upon re-employment.\n\nOn average over the six years following the year of job displacement, the lower likelihood of being in employment is the main factor behind earnings losses in both high- and low-emission industries (Figure 3.7 Panel A and Panel B). However, the 6-year average masks considerable variation in the importance of the earnings components over time. Initially, in the first year following displacement, differences in employment overwhelmingly drive earnings losses as reflected by differences in the probability of being employed and the number of days worked upon re-employment later in the year. By contrast, differences in re-employment wages play a marginal role. Over time, however, wages become more important for explaining earnings losses after displacement in high- and low-emission industries (both in relative and absolute terms). This increasing importance of wages may be explained by the earlier re-employment of workers with the highest skills and earnings potential as well as the gradual adjustment of job search toward lower paying jobs over prolonged unemployment duration – see e.g. Maibom et al. (2023[36]) and Hijzen, Upward and Wright (2010[37]). The days worked component falls strongly after the first year following job loss. From then on, differences in days worked mainly reflect differences in job instability upon re-employment while differences in the timing of job finding following displacement are negligible. Displaced workers may face higher job instability upon re-employment as they are more likely to be employed on temporary contracts and have lower job tenure.\n\n### **Figure 3.7. Differences in earnings losses reflect larger wage losses, fewer days worked as well as longer spells out of work**\n\nContribution of daily wages, days worked and employment to overall earnings losses after job loss in high- and low- emission industries in percentage (Panel A and B) and the difference in earnings losses between high- and low- emission industries in percentage points (Panel C), average across countries\n\nNote: Average across countries of the contribution of daily wages, the probability of being employed, and days worked to the total earnings loss of displaced workers relative to observationally comparable non-displaced workers (see Box 3.3). The bars show the contribution of each margin to the overall earnings six years after job loss and on average over the six years following the year of job loss. The contributions of daily wages and days worked are conditional on being employed in a given year. The contribution of the probability of being employed captures the role of years in which no earnings from dependent employment are recorded in which case zero earnings are imputed. Point estimates from country- level regressions are averaged assigning each country an equal weight. The countries included are: Austria, Denmark, Finland, France, Germany, Hungary, the Netherlands, Portugal, Spain and Sweden. Due to missing daily wage and days worked information, Australia, Canada, Estonia and Norway are excluded. Source: National linked employer employee data, see Annex Table 3.B.1 for details.\n\nStatLink 2 https://stat.link/8izqk2\n\n\\-70\n\n\\-60\n\n\\-50\n\n\\-40\n\n\\-30\n\n\\-20\n\n\\-10\n\n0 %\n\n**A. High-emission industries**\n\n\\-60\n\n\\-50\n\n\\-40\n\n\\-30\n\n\\-20\n\n\\-10\n\n0 %\n\n**B. Low-emission industries**\n\n\\-10 -9 -8 -7 -6 -5 -4 -3 -2 -1 0 %\n\n**C. Difference**\n\nWages Employment Days worked","publish_date":null,"title":"Figure 3.7. Differences in earnings losses reflect larger wage losses, fewer days worked as well as longer spells out of","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p153.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p153.pdf"},"request":{"example_id":"chart/ac8b3538-en_p153","product_type":"parse","source_file_path":"ac8b3538-en_p153.pdf"},"started_at":"2026-11-04T19:46:26.097442"} +{"completed_at":"2026-11-04T19:46:28.588315","latency_in_ms":1181,"output":{"example_id":"chart/ac8b3538-en_p155","grounded_pages":[],"layout_pages":[],"markdown":" **153**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Figure 3.8. Workers displaced from high-emission industries change sector and occupation more often**\n\nTransition probability six years after job displacement, average across countries, percentage\n\nNote: Bars show the probability of working in a different occupation, region, or sector than that recorded in the year before displacement (relative to matched non-displaced workers) and the vertical markers reflect the corresponding 90% confidence intervals. For withdrawals from dependent employment, the bar corresponds to the probability of permanent exits, i.e. a worker that is employed at time t but non-employed in all subsequent periods. As we exclude self-employment in all countries, it is possible that this partly captures transitions into self-employment. All differences between sectors are statistically significant at the 10% level. For sectoral changes and withdrawals from dependent employment, all 14 countries are considered. For occupational changes, the figure includes Germany, Denmark, Finland, Hungary, Norway and Sweden. For regional changes, the figure includes Australia, Austria, Germany, Denmark, Finland, Hungary, the Netherlands, Norway, and Sweden. Source: National linked employer employee data, see Annex Table 3.B.1 for details.\n\nStatLink 2 https://stat.link/afr2yx\n\nWorkers in high-emission industries are slightly more likely to withdraw from dependent employment following displacement, but again these differences are not statistically significant. Such exits from dependent employment may reflect transitions into self-employment or exits from the labour force, including early retirement (Figure 3.8). After six years, their likelihood of withdrawing from dependent employment is 8%, compared to 7% for low-emission industries. As the sample of displaced workers focuses on prime-aged workers (until the age of 50), this effect may reflect a higher likelihood of entering self-employment–- a status generally not covered in the administrative records–- or disability (Schaller and Stevens, 2015[3]) rather than early retirement. For displaced workers above the age of 50, the probability of withdrawing from dependent employment is larger for workers displaced in high- and low-emission industries, while the difference between the two also marginally increases (results not shown here). Withdrawal from the labour market for older workers is more likely to reflect early retirement (Chan and Huff Stevens, 2001[45]).\n\n**_3\\.2.3. Cross-country differences in the consequences of job displacement in high- emission industries_**\n\nWith a focus on the average across countries included in the analysis, this chapter has so far shown that the costs of job displacement in high-emission industries are substantial and noticeably higher than those for workers displaced in low-emission industries. However, there may be important variation across countries in the level of job displacement costs for workers displaced in high-emission industries as well as in the difference in job displacement cost for workers in low-emission industries. The following sub- section therefore leverages the cross-country nature of the data to explore variation in job displacement costs within and between countries.\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\nSector\n\n(2-digit NACE)\n\nOccupation (3-digit ISCO)\n\nRegion (NUTS3)\n\nWithdrawal from dependent\n\nemployment\n\n%\n\nHigh-emission Low-emission","pages":[{"line_numbers":[],"markdown":" **153**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Figure 3.8. Workers displaced from high-emission industries change sector and occupation more often**\n\nTransition probability six years after job displacement, average across countries, percentage\n\nNote: Bars show the probability of working in a different occupation, region, or sector than that recorded in the year before displacement (relative to matched non-displaced workers) and the vertical markers reflect the corresponding 90% confidence intervals. For withdrawals from dependent employment, the bar corresponds to the probability of permanent exits, i.e. a worker that is employed at time t but non-employed in all subsequent periods. As we exclude self-employment in all countries, it is possible that this partly captures transitions into self-employment. All differences between sectors are statistically significant at the 10% level. For sectoral changes and withdrawals from dependent employment, all 14 countries are considered. For occupational changes, the figure includes Germany, Denmark, Finland, Hungary, Norway and Sweden. For regional changes, the figure includes Australia, Austria, Germany, Denmark, Finland, Hungary, the Netherlands, Norway, and Sweden. Source: National linked employer employee data, see Annex Table 3.B.1 for details.\n\nStatLink 2 https://stat.link/afr2yx\n\nWorkers in high-emission industries are slightly more likely to withdraw from dependent employment following displacement, but again these differences are not statistically significant. Such exits from dependent employment may reflect transitions into self-employment or exits from the labour force, including early retirement (Figure 3.8). After six years, their likelihood of withdrawing from dependent employment is 8%, compared to 7% for low-emission industries. As the sample of displaced workers focuses on prime-aged workers (until the age of 50), this effect may reflect a higher likelihood of entering self-employment–- a status generally not covered in the administrative records–- or disability (Schaller and Stevens, 2015[3]) rather than early retirement. For displaced workers above the age of 50, the probability of withdrawing from dependent employment is larger for workers displaced in high- and low-emission industries, while the difference between the two also marginally increases (results not shown here). Withdrawal from the labour market for older workers is more likely to reflect early retirement (Chan and Huff Stevens, 2001[45]).\n\n**_3\\.2.3. Cross-country differences in the consequences of job displacement in high- emission industries_**\n\nWith a focus on the average across countries included in the analysis, this chapter has so far shown that the costs of job displacement in high-emission industries are substantial and noticeably higher than those for workers displaced in low-emission industries. However, there may be important variation across countries in the level of job displacement costs for workers displaced in high-emission industries as well as in the difference in job displacement cost for workers in low-emission industries. The following sub- section therefore leverages the cross-country nature of the data to explore variation in job displacement costs within and between countries.\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\nSector\n\n(2-digit NACE)\n\nOccupation (3-digit ISCO)\n\nRegion (NUTS3)\n\nWithdrawal from dependent\n\nemployment\n\n%\n\nHigh-emission Low-emission","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":" **153**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Figure 3.8. Workers displaced from high-emission industries change sector and occupation more often**\n\nTransition probability six years after job displacement, average across countries, percentage\n\nNote: Bars show the probability of working in a different occupation, region, or sector than that recorded in the year before displacement (relative to matched non-displaced workers) and the vertical markers reflect the corresponding 90% confidence intervals. For withdrawals from dependent employment, the bar corresponds to the probability of permanent exits, i.e. a worker that is employed at time t but non-employed in all subsequent periods. As we exclude self-employment in all countries, it is possible that this partly captures transitions into self-employment. All differences between sectors are statistically significant at the 10% level. For sectoral changes and withdrawals from dependent employment, all 14 countries are considered. For occupational changes, the figure includes Germany, Denmark, Finland, Hungary, Norway and Sweden. For regional changes, the figure includes Australia, Austria, Germany, Denmark, Finland, Hungary, the Netherlands, Norway, and Sweden. Source: National linked employer employee data, see Annex Table 3.B.1 for details.\n\nStatLink 2 https://stat.link/afr2yx\n\nWorkers in high-emission industries are slightly more likely to withdraw from dependent employment following displacement, but again these differences are not statistically significant. Such exits from dependent employment may reflect transitions into self-employment or exits from the labour force, including early retirement (Figure 3.8). After six years, their likelihood of withdrawing from dependent employment is 8%, compared to 7% for low-emission industries. As the sample of displaced workers focuses on prime-aged workers (until the age of 50), this effect may reflect a higher likelihood of entering self-employment–- a status generally not covered in the administrative records–- or disability (Schaller and Stevens, 2015[3]) rather than early retirement. For displaced workers above the age of 50, the probability of withdrawing from dependent employment is larger for workers displaced in high- and low-emission industries, while the difference between the two also marginally increases (results not shown here). Withdrawal from the labour market for older workers is more likely to reflect early retirement (Chan and Huff Stevens, 2001[45]).\n\n**_3\\.2.3. Cross-country differences in the consequences of job displacement in high- emission industries_**\n\nWith a focus on the average across countries included in the analysis, this chapter has so far shown that the costs of job displacement in high-emission industries are substantial and noticeably higher than those for workers displaced in low-emission industries. However, there may be important variation across countries in the level of job displacement costs for workers displaced in high-emission industries as well as in the difference in job displacement cost for workers in low-emission industries. The following sub- section therefore leverages the cross-country nature of the data to explore variation in job displacement costs within and between countries.\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\nSector\n\n(2-digit NACE)\n\nOccupation (3-digit ISCO)\n\nRegion (NUTS3)\n\nWithdrawal from dependent\n\nemployment\n\n%\n\nHigh-emission Low-emission","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p155.pdf"]},"response":{"errors":[],"extract_id":"extract_be58a45d960eb9c98a4d05183becdd3a","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":[" **153**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Figure 3.8. Workers displaced from high-emission industries change sector and occupation more often**\n\nTransition probability six years after job displacement, average across countries, percentage\n\nNote: Bars show the probability of working in a different occupation, region, or sector than that recorded in the year before displacement (relative to matched non-displaced workers) and the vertical markers reflect the corresponding 90% confidence intervals. For withdrawals from dependent employment, the bar corresponds to the probability of permanent exits, i.e. a worker that is employed at time t but non-employed in all subsequent periods. As we exclude self-employment in all countries, it is possible that this partly captures transitions into self-employment. All differences between sectors are statistically significant at the 10% level. For sectoral changes and withdrawals from dependent employment, all 14 countries are considered. For occupational changes, the figure includes Germany, Denmark, Finland, Hungary, Norway and Sweden. For regional changes, the figure includes Australia, Austria, Germany, Denmark, Finland, Hungary, the Netherlands, Norway, and Sweden. Source: National linked employer employee data, see Annex Table 3.B.1 for details.\n\nStatLink 2 https://stat.link/afr2yx\n\nWorkers in high-emission industries are slightly more likely to withdraw from dependent employment following displacement, but again these differences are not statistically significant. Such exits from dependent employment may reflect transitions into self-employment or exits from the labour force, including early retirement (Figure 3.8). After six years, their likelihood of withdrawing from dependent employment is 8%, compared to 7% for low-emission industries. As the sample of displaced workers focuses on prime-aged workers (until the age of 50), this effect may reflect a higher likelihood of entering self-employment–- a status generally not covered in the administrative records–- or disability (Schaller and Stevens, 2015[3]) rather than early retirement. For displaced workers above the age of 50, the probability of withdrawing from dependent employment is larger for workers displaced in high- and low-emission industries, while the difference between the two also marginally increases (results not shown here). Withdrawal from the labour market for older workers is more likely to reflect early retirement (Chan and Huff Stevens, 2001[45]).\n\n**_3\\.2.3. Cross-country differences in the consequences of job displacement in high- emission industries_**\n\nWith a focus on the average across countries included in the analysis, this chapter has so far shown that the costs of job displacement in high-emission industries are substantial and noticeably higher than those for workers displaced in low-emission industries. However, there may be important variation across countries in the level of job displacement costs for workers displaced in high-emission industries as well as in the difference in job displacement cost for workers in low-emission industries. The following sub- section therefore leverages the cross-country nature of the data to explore variation in job displacement costs within and between countries.\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\nSector\n\n(2-digit NACE)\n\nOccupation (3-digit ISCO)\n\nRegion (NUTS3)\n\nWithdrawal from dependent\n\nemployment\n\n%\n\nHigh-emission Low-emission"],"full_content":" **153**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Figure 3.8. Workers displaced from high-emission industries change sector and occupation more often**\n\nTransition probability six years after job displacement, average across countries, percentage\n\nNote: Bars show the probability of working in a different occupation, region, or sector than that recorded in the year before displacement (relative to matched non-displaced workers) and the vertical markers reflect the corresponding 90% confidence intervals. For withdrawals from dependent employment, the bar corresponds to the probability of permanent exits, i.e. a worker that is employed at time t but non-employed in all subsequent periods. As we exclude self-employment in all countries, it is possible that this partly captures transitions into self-employment. All differences between sectors are statistically significant at the 10% level. For sectoral changes and withdrawals from dependent employment, all 14 countries are considered. For occupational changes, the figure includes Germany, Denmark, Finland, Hungary, Norway and Sweden. For regional changes, the figure includes Australia, Austria, Germany, Denmark, Finland, Hungary, the Netherlands, Norway, and Sweden. Source: National linked employer employee data, see Annex Table 3.B.1 for details.\n\nStatLink 2 https://stat.link/afr2yx\n\nWorkers in high-emission industries are slightly more likely to withdraw from dependent employment following displacement, but again these differences are not statistically significant. Such exits from dependent employment may reflect transitions into self-employment or exits from the labour force, including early retirement (Figure 3.8). After six years, their likelihood of withdrawing from dependent employment is 8%, compared to 7% for low-emission industries. As the sample of displaced workers focuses on prime-aged workers (until the age of 50), this effect may reflect a higher likelihood of entering self-employment–- a status generally not covered in the administrative records–- or disability (Schaller and Stevens, 2015[3]) rather than early retirement. For displaced workers above the age of 50, the probability of withdrawing from dependent employment is larger for workers displaced in high- and low-emission industries, while the difference between the two also marginally increases (results not shown here). Withdrawal from the labour market for older workers is more likely to reflect early retirement (Chan and Huff Stevens, 2001[45]).\n\n**_3\\.2.3. Cross-country differences in the consequences of job displacement in high- emission industries_**\n\nWith a focus on the average across countries included in the analysis, this chapter has so far shown that the costs of job displacement in high-emission industries are substantial and noticeably higher than those for workers displaced in low-emission industries. However, there may be important variation across countries in the level of job displacement costs for workers displaced in high-emission industries as well as in the difference in job displacement cost for workers in low-emission industries. The following sub- section therefore leverages the cross-country nature of the data to explore variation in job displacement costs within and between countries.\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\nSector\n\n(2-digit NACE)\n\nOccupation (3-digit ISCO)\n\nRegion (NUTS3)\n\nWithdrawal from dependent\n\nemployment\n\n%\n\nHigh-emission Low-emission","publish_date":null,"title":"Figure 3.8. Workers displaced from high-emission industries change sector and occupation more often","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p155.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p155.pdf"},"request":{"example_id":"chart/ac8b3538-en_p155","product_type":"parse","source_file_path":"ac8b3538-en_p155.pdf"},"started_at":"2026-11-04T19:46:27.407217"} +{"completed_at":"2026-11-04T19:46:31.361680","latency_in_ms":2382,"output":{"example_id":"chart/ac8b3538-en_p157","grounded_pages":[],"layout_pages":[],"markdown":" **155**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n_Differences in wage losses tend to be firm-related rather than worker-related_\n\nWage losses upon re-employment may be either worker-related or firm-related. Worker-related wage losses stem from the loss of firm-specific skills and a decline in the quality of the match, whereas firm-related wage losses reflect a decline in the generosity of a firm’s wage policies (Lachowska, Mas and Woodbury, 2020[29]). Decomposing the contribution of wage losses to earnings losses into a worker- and a firm-related component suggests that firm-related wage losses account for most of the difference in wage losses between high- and low-emission industries, echoing the results of Barreto, Grundke and Krill (2023[8]). On average across countries, differences in average wage losses between high- and low-emission industries are entirely firm- related (Figure 3.10). Indeed, in several countries worker-related losses to be smaller in high-emission industries. This is the case in France, Hungary, the Netherlands and Sweden. One possible explanation for this may be that workers in high-emission industries have lower skills and hence are less at risk of losing (firm-specific) human capital following job loss. Overall, these results suggest that displaced workers from high-emission industries undergo larger earnings losses in part because they were displaced from higher paying firms. This may mean that these workers move to less productive firms, firms where workers capture lower rents, or lose compensating wage differentials for physically demanding working conditions (Card et al., 2018[46]; Sorkin, 2018[47]; Hirsch and Mueller, 2020[48]). Evidence for Portugal (not shown) suggests that displaced workers in high-emission industries move to less productive firms, while evidence from France by Brandily, Hémet and Malgouyres (2022[49]) shows that displaced workers may re-allocate to higher productivity firms that pay lower wages because of a loss in bargaining power. ⁸ This loss in bargaining power is in line with the finding in Chapter 2 that workers employed in occupations concentrated in high-emission sectors are more likely to be covered by a collective bargaining agreement than other workers.\n\n### **Figure 3.10. Higher wage losses in high-emission sectors tend to be firm-related rather than worker-related**\n\nContribution of wage losses to the difference in overall earnings losses between displaced workers from high- and low-emission sectors, decomposed into differences in firm- and worker-related losses, average over six years following displacement, percentage points\n\nNote: Bars represent the contribution of wage losses to the overall difference in earnings losses between displaced workers from high-emission and low-emission industries on average over the six years after displacement (see Box 3.3). The bars are decomposed into the contribution of a firm-related component (e.g. firm wage policies) and a worker-related component (i.e. human capital, match quality) following Lachowska et al. (2020[29]). Countries are sorted based on the difference in wage losses between high- and low-emission sectors. The figure excludes Australia, Canada, Estonia and Norway due to missing information on daily wages. Source: National linked employer employee data, see Annex Table 3.B.1 for details.\n\nStatLink 2 https://stat.link/q3tfmg\n\n\\-10 -8 -6 -4 -2 0 2 4 6\n\nSWE\n\nNLD\n\nDEU\n\nPRT\n\nDNK\n\nAverage\n\nFIN\n\nAUT\n\nHUN\n\nFRA\n\nESP\n\np.p.\n\nFirm Worker Total (contribution of wage losses)","pages":[{"line_numbers":[],"markdown":" **155**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n_Differences in wage losses tend to be firm-related rather than worker-related_\n\nWage losses upon re-employment may be either worker-related or firm-related. Worker-related wage losses stem from the loss of firm-specific skills and a decline in the quality of the match, whereas firm-related wage losses reflect a decline in the generosity of a firm’s wage policies (Lachowska, Mas and Woodbury, 2020[29]). Decomposing the contribution of wage losses to earnings losses into a worker- and a firm-related component suggests that firm-related wage losses account for most of the difference in wage losses between high- and low-emission industries, echoing the results of Barreto, Grundke and Krill (2023[8]). On average across countries, differences in average wage losses between high- and low-emission industries are entirely firm- related (Figure 3.10). Indeed, in several countries worker-related losses to be smaller in high-emission industries. This is the case in France, Hungary, the Netherlands and Sweden. One possible explanation for this may be that workers in high-emission industries have lower skills and hence are less at risk of losing (firm-specific) human capital following job loss. Overall, these results suggest that displaced workers from high-emission industries undergo larger earnings losses in part because they were displaced from higher paying firms. This may mean that these workers move to less productive firms, firms where workers capture lower rents, or lose compensating wage differentials for physically demanding working conditions (Card et al., 2018[46]; Sorkin, 2018[47]; Hirsch and Mueller, 2020[48]). Evidence for Portugal (not shown) suggests that displaced workers in high-emission industries move to less productive firms, while evidence from France by Brandily, Hémet and Malgouyres (2022[49]) shows that displaced workers may re-allocate to higher productivity firms that pay lower wages because of a loss in bargaining power. ⁸ This loss in bargaining power is in line with the finding in Chapter 2 that workers employed in occupations concentrated in high-emission sectors are more likely to be covered by a collective bargaining agreement than other workers.\n\n### **Figure 3.10. Higher wage losses in high-emission sectors tend to be firm-related rather than worker-related**\n\nContribution of wage losses to the difference in overall earnings losses between displaced workers from high- and low-emission sectors, decomposed into differences in firm- and worker-related losses, average over six years following displacement, percentage points\n\nNote: Bars represent the contribution of wage losses to the overall difference in earnings losses between displaced workers from high-emission and low-emission industries on average over the six years after displacement (see Box 3.3). The bars are decomposed into the contribution of a firm-related component (e.g. firm wage policies) and a worker-related component (i.e. human capital, match quality) following Lachowska et al. (2020[29]). Countries are sorted based on the difference in wage losses between high- and low-emission sectors. The figure excludes Australia, Canada, Estonia and Norway due to missing information on daily wages. Source: National linked employer employee data, see Annex Table 3.B.1 for details.\n\nStatLink 2 https://stat.link/q3tfmg\n\n\\-10 -8 -6 -4 -2 0 2 4 6\n\nSWE\n\nNLD\n\nDEU\n\nPRT\n\nDNK\n\nAverage\n\nFIN\n\nAUT\n\nHUN\n\nFRA\n\nESP\n\np.p.\n\nFirm Worker Total (contribution of wage losses)","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":" **155**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n_Differences in wage losses tend to be firm-related rather than worker-related_\n\nWage losses upon re-employment may be either worker-related or firm-related. Worker-related wage losses stem from the loss of firm-specific skills and a decline in the quality of the match, whereas firm-related wage losses reflect a decline in the generosity of a firm’s wage policies (Lachowska, Mas and Woodbury, 2020[29]). Decomposing the contribution of wage losses to earnings losses into a worker- and a firm-related component suggests that firm-related wage losses account for most of the difference in wage losses between high- and low-emission industries, echoing the results of Barreto, Grundke and Krill (2023[8]). On average across countries, differences in average wage losses between high- and low-emission industries are entirely firm- related (Figure 3.10). Indeed, in several countries worker-related losses to be smaller in high-emission industries. This is the case in France, Hungary, the Netherlands and Sweden. One possible explanation for this may be that workers in high-emission industries have lower skills and hence are less at risk of losing (firm-specific) human capital following job loss. Overall, these results suggest that displaced workers from high-emission industries undergo larger earnings losses in part because they were displaced from higher paying firms. This may mean that these workers move to less productive firms, firms where workers capture lower rents, or lose compensating wage differentials for physically demanding working conditions (Card et al., 2018[46]; Sorkin, 2018[47]; Hirsch and Mueller, 2020[48]). Evidence for Portugal (not shown) suggests that displaced workers in high-emission industries move to less productive firms, while evidence from France by Brandily, Hémet and Malgouyres (2022[49]) shows that displaced workers may re-allocate to higher productivity firms that pay lower wages because of a loss in bargaining power. ⁸ This loss in bargaining power is in line with the finding in Chapter 2 that workers employed in occupations concentrated in high-emission sectors are more likely to be covered by a collective bargaining agreement than other workers.\n\n### **Figure 3.10. Higher wage losses in high-emission sectors tend to be firm-related rather than worker-related**\n\nContribution of wage losses to the difference in overall earnings losses between displaced workers from high- and low-emission sectors, decomposed into differences in firm- and worker-related losses, average over six years following displacement, percentage points\n\nNote: Bars represent the contribution of wage losses to the overall difference in earnings losses between displaced workers from high-emission and low-emission industries on average over the six years after displacement (see Box 3.3). The bars are decomposed into the contribution of a firm-related component (e.g. firm wage policies) and a worker-related component (i.e. human capital, match quality) following Lachowska et al. (2020[29]). Countries are sorted based on the difference in wage losses between high- and low-emission sectors. The figure excludes Australia, Canada, Estonia and Norway due to missing information on daily wages. Source: National linked employer employee data, see Annex Table 3.B.1 for details.\n\nStatLink 2 https://stat.link/q3tfmg\n\n\\-10 -8 -6 -4 -2 0 2 4 6\n\nSWE\n\nNLD\n\nDEU\n\nPRT\n\nDNK\n\nAverage\n\nFIN\n\nAUT\n\nHUN\n\nFRA\n\nESP\n\np.p.\n\nFirm Worker Total (contribution of wage losses)","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p157.pdf"]},"response":{"errors":[],"extract_id":"extract_cbdff51bb144ae1c7d7c233217601ea5","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":[" **155**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n_Differences in wage losses tend to be firm-related rather than worker-related_\n\nWage losses upon re-employment may be either worker-related or firm-related. Worker-related wage losses stem from the loss of firm-specific skills and a decline in the quality of the match, whereas firm-related wage losses reflect a decline in the generosity of a firm’s wage policies (Lachowska, Mas and Woodbury, 2020[29]). Decomposing the contribution of wage losses to earnings losses into a worker- and a firm-related component suggests that firm-related wage losses account for most of the difference in wage losses between high- and low-emission industries, echoing the results of Barreto, Grundke and Krill (2023[8]). On average across countries, differences in average wage losses between high- and low-emission industries are entirely firm- related (Figure 3.10). Indeed, in several countries worker-related losses to be smaller in high-emission industries. This is the case in France, Hungary, the Netherlands and Sweden. One possible explanation for this may be that workers in high-emission industries have lower skills and hence are less at risk of losing (firm-specific) human capital following job loss. Overall, these results suggest that displaced workers from high-emission industries undergo larger earnings losses in part because they were displaced from higher paying firms. This may mean that these workers move to less productive firms, firms where workers capture lower rents, or lose compensating wage differentials for physically demanding working conditions (Card et al., 2018[46]; Sorkin, 2018[47]; Hirsch and Mueller, 2020[48]). Evidence for Portugal (not shown) suggests that displaced workers in high-emission industries move to less productive firms, while evidence from France by Brandily, Hémet and Malgouyres (2022[49]) shows that displaced workers may re-allocate to higher productivity firms that pay lower wages because of a loss in bargaining power. ⁸ This loss in bargaining power is in line with the finding in Chapter 2 that workers employed in occupations concentrated in high-emission sectors are more likely to be covered by a collective bargaining agreement than other workers.\n\n### **Figure 3.10. Higher wage losses in high-emission sectors tend to be firm-related rather than worker-related**\n\nContribution of wage losses to the difference in overall earnings losses between displaced workers from high- and low-emission sectors, decomposed into differences in firm- and worker-related losses, average over six years following displacement, percentage points\n\nNote: Bars represent the contribution of wage losses to the overall difference in earnings losses between displaced workers from high-emission and low-emission industries on average over the six years after displacement (see Box 3.3). The bars are decomposed into the contribution of a firm-related component (e.g. firm wage policies) and a worker-related component (i.e. human capital, match quality) following Lachowska et al. (2020[29]). Countries are sorted based on the difference in wage losses between high- and low-emission sectors. The figure excludes Australia, Canada, Estonia and Norway due to missing information on daily wages. Source: National linked employer employee data, see Annex Table 3.B.1 for details.\n\nStatLink 2 https://stat.link/q3tfmg\n\n\\-10 -8 -6 -4 -2 0 2 4 6\n\nSWE\n\nNLD\n\nDEU\n\nPRT\n\nDNK\n\nAverage\n\nFIN\n\nAUT\n\nHUN\n\nFRA\n\nESP\n\np.p.\n\nFirm Worker Total (contribution of wage losses)"],"full_content":" **155**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n_Differences in wage losses tend to be firm-related rather than worker-related_\n\nWage losses upon re-employment may be either worker-related or firm-related. Worker-related wage losses stem from the loss of firm-specific skills and a decline in the quality of the match, whereas firm-related wage losses reflect a decline in the generosity of a firm’s wage policies (Lachowska, Mas and Woodbury, 2020[29]). Decomposing the contribution of wage losses to earnings losses into a worker- and a firm-related component suggests that firm-related wage losses account for most of the difference in wage losses between high- and low-emission industries, echoing the results of Barreto, Grundke and Krill (2023[8]). On average across countries, differences in average wage losses between high- and low-emission industries are entirely firm- related (Figure 3.10). Indeed, in several countries worker-related losses to be smaller in high-emission industries. This is the case in France, Hungary, the Netherlands and Sweden. One possible explanation for this may be that workers in high-emission industries have lower skills and hence are less at risk of losing (firm-specific) human capital following job loss. Overall, these results suggest that displaced workers from high-emission industries undergo larger earnings losses in part because they were displaced from higher paying firms. This may mean that these workers move to less productive firms, firms where workers capture lower rents, or lose compensating wage differentials for physically demanding working conditions (Card et al., 2018[46]; Sorkin, 2018[47]; Hirsch and Mueller, 2020[48]). Evidence for Portugal (not shown) suggests that displaced workers in high-emission industries move to less productive firms, while evidence from France by Brandily, Hémet and Malgouyres (2022[49]) shows that displaced workers may re-allocate to higher productivity firms that pay lower wages because of a loss in bargaining power. ⁸ This loss in bargaining power is in line with the finding in Chapter 2 that workers employed in occupations concentrated in high-emission sectors are more likely to be covered by a collective bargaining agreement than other workers.\n\n### **Figure 3.10. Higher wage losses in high-emission sectors tend to be firm-related rather than worker-related**\n\nContribution of wage losses to the difference in overall earnings losses between displaced workers from high- and low-emission sectors, decomposed into differences in firm- and worker-related losses, average over six years following displacement, percentage points\n\nNote: Bars represent the contribution of wage losses to the overall difference in earnings losses between displaced workers from high-emission and low-emission industries on average over the six years after displacement (see Box 3.3). The bars are decomposed into the contribution of a firm-related component (e.g. firm wage policies) and a worker-related component (i.e. human capital, match quality) following Lachowska et al. (2020[29]). Countries are sorted based on the difference in wage losses between high- and low-emission sectors. The figure excludes Australia, Canada, Estonia and Norway due to missing information on daily wages. Source: National linked employer employee data, see Annex Table 3.B.1 for details.\n\nStatLink 2 https://stat.link/q3tfmg\n\n\\-10 -8 -6 -4 -2 0 2 4 6\n\nSWE\n\nNLD\n\nDEU\n\nPRT\n\nDNK\n\nAverage\n\nFIN\n\nAUT\n\nHUN\n\nFRA\n\nESP\n\np.p.\n\nFirm Worker Total (contribution of wage losses)","publish_date":null,"title":"Figure 3.10. Higher wage losses in high-emission sectors tend to be firm-related rather than worker-related","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p157.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p157.pdf"},"request":{"example_id":"chart/ac8b3538-en_p157","product_type":"parse","source_file_path":"ac8b3538-en_p157.pdf"},"started_at":"2026-11-04T19:46:28.979193"} +{"completed_at":"2026-11-04T19:46:33.413536","latency_in_ms":3258,"output":{"example_id":"chart/ac8b3538-en_p158","grounded_pages":[],"layout_pages":[],"markdown":"**156** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\nAfter focusing on the within-country variation in job displacement costs, the remainder of this sub-section considers how differences in the costs of job displacement for workers in high-emission industries vary between countries in levels.\n\n_There are large differences in the cost of job displacement in high-emission industries between countries_\n\nThere are large differences in the cost of job displacement between countries (Bertheau et al., 2023[10]). With less than 30%, average annual earnings losses of displaced workers in high-emission industries are particularly low in Australia, Canada, Germany, the Netherlands, Norway and Sweden, but are especially high in Portugal, where they exceed 60% (Figure 3.11). These differences reflect to an important extent differences in the opportunities of displaced workers to find another job as well as the number of days worked once a job is found. In Portugal, the country with the highest earnings losses, more than 43 percentage points of the earnings losses following displacement in high-emission sectors can be explained by being out of employment for an entire year, with a further 24 percentage points attributable to fewer days worked and about 5 percentage points to lower re-employment wages. In contrast, in Sweden, the country with the third lowest earnings losses, employment and days worked respectively explain 14 percentage points and 9 percentage points. By contrast, the contribution of wage losses upon re-employment varies much less between countries, typically between 6 percentage points and 11 percentage point at most. The main exceptions are Sweden and the Netherlands where wage losses upon re-employment are much less important than in the other countries considered. Note that the differences between countries in the cost of job displacement and the importance of joblessness are similar in both high- and low-emission sectors, which suggests that factors related to the incidence of joblessness affect both sectors in a similar way. ⁹\n\n### **Figure 3.11. Country-differences in the costs of job displacement in high-emission industries mainly reflect differences in the incidence of joblessness**\n\nDecomposition of the average earnings losses over six years since displacement in high-emission industries into the contributions associated with wages, employment and days workers by country, percentage\n\nNote: Bars refer to the average effect of job loss on earnings over six years following displacement for workers in high-emission sectors. Countries are sorted based on the level of earnings losses for high-emission sectors. Due to missing information on daily wages and days worked, the bars for Australia, Canada, Estonia and Norway are not decomposed into components. Source: National linked employer employee data, see Annex Table 3.B.1 for details.\n\nStatLink 2 https://stat.link/lfpsgh\n\n\\-80 -70 -60 -50 -40 -30 -20 -10 0\n\nPRT ESP EST HUN FRA\n\nAverage\n\nAUT FIN DNK AUS CAN DEU NLD SWE NOR\n\n%\n\nWages Employment Days worked Total","pages":[{"line_numbers":[],"markdown":"**156** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\nAfter focusing on the within-country variation in job displacement costs, the remainder of this sub-section considers how differences in the costs of job displacement for workers in high-emission industries vary between countries in levels.\n\n_There are large differences in the cost of job displacement in high-emission industries between countries_\n\nThere are large differences in the cost of job displacement between countries (Bertheau et al., 2023[10]). With less than 30%, average annual earnings losses of displaced workers in high-emission industries are particularly low in Australia, Canada, Germany, the Netherlands, Norway and Sweden, but are especially high in Portugal, where they exceed 60% (Figure 3.11). These differences reflect to an important extent differences in the opportunities of displaced workers to find another job as well as the number of days worked once a job is found. In Portugal, the country with the highest earnings losses, more than 43 percentage points of the earnings losses following displacement in high-emission sectors can be explained by being out of employment for an entire year, with a further 24 percentage points attributable to fewer days worked and about 5 percentage points to lower re-employment wages. In contrast, in Sweden, the country with the third lowest earnings losses, employment and days worked respectively explain 14 percentage points and 9 percentage points. By contrast, the contribution of wage losses upon re-employment varies much less between countries, typically between 6 percentage points and 11 percentage point at most. The main exceptions are Sweden and the Netherlands where wage losses upon re-employment are much less important than in the other countries considered. Note that the differences between countries in the cost of job displacement and the importance of joblessness are similar in both high- and low-emission sectors, which suggests that factors related to the incidence of joblessness affect both sectors in a similar way. ⁹\n\n### **Figure 3.11. Country-differences in the costs of job displacement in high-emission industries mainly reflect differences in the incidence of joblessness**\n\nDecomposition of the average earnings losses over six years since displacement in high-emission industries into the contributions associated with wages, employment and days workers by country, percentage\n\nNote: Bars refer to the average effect of job loss on earnings over six years following displacement for workers in high-emission sectors. Countries are sorted based on the level of earnings losses for high-emission sectors. Due to missing information on daily wages and days worked, the bars for Australia, Canada, Estonia and Norway are not decomposed into components. Source: National linked employer employee data, see Annex Table 3.B.1 for details.\n\nStatLink 2 https://stat.link/lfpsgh\n\n\\-80 -70 -60 -50 -40 -30 -20 -10 0\n\nPRT ESP EST HUN FRA\n\nAverage\n\nAUT FIN DNK AUS CAN DEU NLD SWE NOR\n\n%\n\nWages Employment Days worked Total","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**156** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\nAfter focusing on the within-country variation in job displacement costs, the remainder of this sub-section considers how differences in the costs of job displacement for workers in high-emission industries vary between countries in levels.\n\n_There are large differences in the cost of job displacement in high-emission industries between countries_\n\nThere are large differences in the cost of job displacement between countries (Bertheau et al., 2023[10]). With less than 30%, average annual earnings losses of displaced workers in high-emission industries are particularly low in Australia, Canada, Germany, the Netherlands, Norway and Sweden, but are especially high in Portugal, where they exceed 60% (Figure 3.11). These differences reflect to an important extent differences in the opportunities of displaced workers to find another job as well as the number of days worked once a job is found. In Portugal, the country with the highest earnings losses, more than 43 percentage points of the earnings losses following displacement in high-emission sectors can be explained by being out of employment for an entire year, with a further 24 percentage points attributable to fewer days worked and about 5 percentage points to lower re-employment wages. In contrast, in Sweden, the country with the third lowest earnings losses, employment and days worked respectively explain 14 percentage points and 9 percentage points. By contrast, the contribution of wage losses upon re-employment varies much less between countries, typically between 6 percentage points and 11 percentage point at most. The main exceptions are Sweden and the Netherlands where wage losses upon re-employment are much less important than in the other countries considered. Note that the differences between countries in the cost of job displacement and the importance of joblessness are similar in both high- and low-emission sectors, which suggests that factors related to the incidence of joblessness affect both sectors in a similar way. ⁹\n\n### **Figure 3.11. Country-differences in the costs of job displacement in high-emission industries mainly reflect differences in the incidence of joblessness**\n\nDecomposition of the average earnings losses over six years since displacement in high-emission industries into the contributions associated with wages, employment and days workers by country, percentage\n\nNote: Bars refer to the average effect of job loss on earnings over six years following displacement for workers in high-emission sectors. Countries are sorted based on the level of earnings losses for high-emission sectors. Due to missing information on daily wages and days worked, the bars for Australia, Canada, Estonia and Norway are not decomposed into components. Source: National linked employer employee data, see Annex Table 3.B.1 for details.\n\nStatLink 2 https://stat.link/lfpsgh\n\n\\-80 -70 -60 -50 -40 -30 -20 -10 0\n\nPRT ESP EST HUN FRA\n\nAverage\n\nAUT FIN DNK AUS CAN DEU NLD SWE NOR\n\n%\n\nWages Employment Days worked Total","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p158.pdf"]},"response":{"errors":[],"extract_id":"extract_64007ff64f7118bc6cb2914ec90185dc","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**156** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\nAfter focusing on the within-country variation in job displacement costs, the remainder of this sub-section considers how differences in the costs of job displacement for workers in high-emission industries vary between countries in levels.\n\n_There are large differences in the cost of job displacement in high-emission industries between countries_\n\nThere are large differences in the cost of job displacement between countries (Bertheau et al., 2023[10]). With less than 30%, average annual earnings losses of displaced workers in high-emission industries are particularly low in Australia, Canada, Germany, the Netherlands, Norway and Sweden, but are especially high in Portugal, where they exceed 60% (Figure 3.11). These differences reflect to an important extent differences in the opportunities of displaced workers to find another job as well as the number of days worked once a job is found. In Portugal, the country with the highest earnings losses, more than 43 percentage points of the earnings losses following displacement in high-emission sectors can be explained by being out of employment for an entire year, with a further 24 percentage points attributable to fewer days worked and about 5 percentage points to lower re-employment wages. In contrast, in Sweden, the country with the third lowest earnings losses, employment and days worked respectively explain 14 percentage points and 9 percentage points. By contrast, the contribution of wage losses upon re-employment varies much less between countries, typically between 6 percentage points and 11 percentage point at most. The main exceptions are Sweden and the Netherlands where wage losses upon re-employment are much less important than in the other countries considered. Note that the differences between countries in the cost of job displacement and the importance of joblessness are similar in both high- and low-emission sectors, which suggests that factors related to the incidence of joblessness affect both sectors in a similar way. ⁹\n\n### **Figure 3.11. Country-differences in the costs of job displacement in high-emission industries mainly reflect differences in the incidence of joblessness**\n\nDecomposition of the average earnings losses over six years since displacement in high-emission industries into the contributions associated with wages, employment and days workers by country, percentage\n\nNote: Bars refer to the average effect of job loss on earnings over six years following displacement for workers in high-emission sectors. Countries are sorted based on the level of earnings losses for high-emission sectors. Due to missing information on daily wages and days worked, the bars for Australia, Canada, Estonia and Norway are not decomposed into components. Source: National linked employer employee data, see Annex Table 3.B.1 for details.\n\nStatLink 2 https://stat.link/lfpsgh\n\n\\-80 -70 -60 -50 -40 -30 -20 -10 0\n\nPRT ESP EST HUN FRA\n\nAverage\n\nAUT FIN DNK AUS CAN DEU NLD SWE NOR\n\n%\n\nWages Employment Days worked Total"],"full_content":"**156** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\nAfter focusing on the within-country variation in job displacement costs, the remainder of this sub-section considers how differences in the costs of job displacement for workers in high-emission industries vary between countries in levels.\n\n_There are large differences in the cost of job displacement in high-emission industries between countries_\n\nThere are large differences in the cost of job displacement between countries (Bertheau et al., 2023[10]). With less than 30%, average annual earnings losses of displaced workers in high-emission industries are particularly low in Australia, Canada, Germany, the Netherlands, Norway and Sweden, but are especially high in Portugal, where they exceed 60% (Figure 3.11). These differences reflect to an important extent differences in the opportunities of displaced workers to find another job as well as the number of days worked once a job is found. In Portugal, the country with the highest earnings losses, more than 43 percentage points of the earnings losses following displacement in high-emission sectors can be explained by being out of employment for an entire year, with a further 24 percentage points attributable to fewer days worked and about 5 percentage points to lower re-employment wages. In contrast, in Sweden, the country with the third lowest earnings losses, employment and days worked respectively explain 14 percentage points and 9 percentage points. By contrast, the contribution of wage losses upon re-employment varies much less between countries, typically between 6 percentage points and 11 percentage point at most. The main exceptions are Sweden and the Netherlands where wage losses upon re-employment are much less important than in the other countries considered. Note that the differences between countries in the cost of job displacement and the importance of joblessness are similar in both high- and low-emission sectors, which suggests that factors related to the incidence of joblessness affect both sectors in a similar way. ⁹\n\n### **Figure 3.11. Country-differences in the costs of job displacement in high-emission industries mainly reflect differences in the incidence of joblessness**\n\nDecomposition of the average earnings losses over six years since displacement in high-emission industries into the contributions associated with wages, employment and days workers by country, percentage\n\nNote: Bars refer to the average effect of job loss on earnings over six years following displacement for workers in high-emission sectors. Countries are sorted based on the level of earnings losses for high-emission sectors. Due to missing information on daily wages and days worked, the bars for Australia, Canada, Estonia and Norway are not decomposed into components. Source: National linked employer employee data, see Annex Table 3.B.1 for details.\n\nStatLink 2 https://stat.link/lfpsgh\n\n\\-80 -70 -60 -50 -40 -30 -20 -10 0\n\nPRT ESP EST HUN FRA\n\nAverage\n\nAUT FIN DNK AUS CAN DEU NLD SWE NOR\n\n%\n\nWages Employment Days worked Total","publish_date":null,"title":"Figure 3.11. Country-differences in the costs of job displacement in high-emission industries mainly reflect differences in the incidence of joblessness","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p158.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p158.pdf"},"request":{"example_id":"chart/ac8b3538-en_p158","product_type":"parse","source_file_path":"ac8b3538-en_p158.pdf"},"started_at":"2026-11-04T19:46:30.154589"} +{"completed_at":"2026-11-04T19:46:34.467299","latency_in_ms":2711,"output":{"example_id":"chart/ac8b3538-en_p161","grounded_pages":[],"layout_pages":[],"markdown":" **159**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\nlow-emission industries, and the second component captures the structural effect, i.e. unexplained differences in the cost of job displacement between high- and low-emission industries holding composition constant. A similar decomposition can be used to shed light on the drivers of differences in the cost of job displacement in high-emission industries between countries. The results of the decomposition between industries within countries and within industries between countries are presented in Figure 3.12.\n\n**The role of occupational composition**\n\nWhile the baseline analysis for all countries focuses on the role of firm and worker composition in explaining differences in the cost of job displacement between industries and countries, for a subset of countries with information 3-digit occupations (ISCO), it is possible to go further by analysing the role of differences in the composition of occupations. Occupations are categorised into five groups based on their task content following Autor, Levy and Murnane (2003[50]): non-routine manual, routine-manual, non-routine cognitive analytic, routine-cognitive and non-routine interactive. The analysis reveals that occupational composition plays an important role in explaining differences in the earnings losses of displacement between high- and low-emission industries, on average across countries analysed (Figure 3.13). More specifically, the concentration of routine-manual occupations accounts for 32% of the difference in earnings losses between high- and low-emission industries. This suggests that displaced workers in emission-intensive occupations may have difficulty finding another job because opportunities in the same occupation are limited and they lack the skills to move to other occupations with higher skill requirements.\n\n### **Figure 3.13. The concentration of manual routine occupations in high-emission industries contributes to higher earnings losses from job displacement**\n\nOaxaca-Blinder Decomposition, difference in earnings losses between high- and low-emission sectors including occupational measures\n\nNote: Average across Germany, Finland, Portugal, Hungary and Sweden. Each bar represents the contribution of a given occupational measure to the overall difference in earnings losses between high- and low-emission sectors based on an Oaxaca-Blinder decomposition. The reference category is non-routine interactive. Source: National linked employer employee data, see Annex Table 3.B.1 for details.\n\nStatLink 2 https://stat.link/w1n74g\n\n\\-0.07 -0.06 -0.05 -0.04 -0.03 -0.02 -0.01 0 0.01\n\nRoutine cognitive\n\nNon-routine cognitive analytic\n\nRoutine manual\n\nNon-routine manual\n\nFirm size\n\nFirm wage premia\n\nWorker FE\n\nTenure\n\nGender\n\nAge\n\nUnexplained\n\nExplained\n\nDifference\n\nEarnings relaitve to pre-displacement average, p.p.","pages":[{"line_numbers":[],"markdown":" **159**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\nlow-emission industries, and the second component captures the structural effect, i.e. unexplained differences in the cost of job displacement between high- and low-emission industries holding composition constant. A similar decomposition can be used to shed light on the drivers of differences in the cost of job displacement in high-emission industries between countries. The results of the decomposition between industries within countries and within industries between countries are presented in Figure 3.12.\n\n**The role of occupational composition**\n\nWhile the baseline analysis for all countries focuses on the role of firm and worker composition in explaining differences in the cost of job displacement between industries and countries, for a subset of countries with information 3-digit occupations (ISCO), it is possible to go further by analysing the role of differences in the composition of occupations. Occupations are categorised into five groups based on their task content following Autor, Levy and Murnane (2003[50]): non-routine manual, routine-manual, non-routine cognitive analytic, routine-cognitive and non-routine interactive. The analysis reveals that occupational composition plays an important role in explaining differences in the earnings losses of displacement between high- and low-emission industries, on average across countries analysed (Figure 3.13). More specifically, the concentration of routine-manual occupations accounts for 32% of the difference in earnings losses between high- and low-emission industries. This suggests that displaced workers in emission-intensive occupations may have difficulty finding another job because opportunities in the same occupation are limited and they lack the skills to move to other occupations with higher skill requirements.\n\n### **Figure 3.13. The concentration of manual routine occupations in high-emission industries contributes to higher earnings losses from job displacement**\n\nOaxaca-Blinder Decomposition, difference in earnings losses between high- and low-emission sectors including occupational measures\n\nNote: Average across Germany, Finland, Portugal, Hungary and Sweden. Each bar represents the contribution of a given occupational measure to the overall difference in earnings losses between high- and low-emission sectors based on an Oaxaca-Blinder decomposition. The reference category is non-routine interactive. Source: National linked employer employee data, see Annex Table 3.B.1 for details.\n\nStatLink 2 https://stat.link/w1n74g\n\n\\-0.07 -0.06 -0.05 -0.04 -0.03 -0.02 -0.01 0 0.01\n\nRoutine cognitive\n\nNon-routine cognitive analytic\n\nRoutine manual\n\nNon-routine manual\n\nFirm size\n\nFirm wage premia\n\nWorker FE\n\nTenure\n\nGender\n\nAge\n\nUnexplained\n\nExplained\n\nDifference\n\nEarnings relaitve to pre-displacement average, p.p.","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":" **159**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\nlow-emission industries, and the second component captures the structural effect, i.e. unexplained differences in the cost of job displacement between high- and low-emission industries holding composition constant. A similar decomposition can be used to shed light on the drivers of differences in the cost of job displacement in high-emission industries between countries. The results of the decomposition between industries within countries and within industries between countries are presented in Figure 3.12.\n\n**The role of occupational composition**\n\nWhile the baseline analysis for all countries focuses on the role of firm and worker composition in explaining differences in the cost of job displacement between industries and countries, for a subset of countries with information 3-digit occupations (ISCO), it is possible to go further by analysing the role of differences in the composition of occupations. Occupations are categorised into five groups based on their task content following Autor, Levy and Murnane (2003[50]): non-routine manual, routine-manual, non-routine cognitive analytic, routine-cognitive and non-routine interactive. The analysis reveals that occupational composition plays an important role in explaining differences in the earnings losses of displacement between high- and low-emission industries, on average across countries analysed (Figure 3.13). More specifically, the concentration of routine-manual occupations accounts for 32% of the difference in earnings losses between high- and low-emission industries. This suggests that displaced workers in emission-intensive occupations may have difficulty finding another job because opportunities in the same occupation are limited and they lack the skills to move to other occupations with higher skill requirements.\n\n### **Figure 3.13. The concentration of manual routine occupations in high-emission industries contributes to higher earnings losses from job displacement**\n\nOaxaca-Blinder Decomposition, difference in earnings losses between high- and low-emission sectors including occupational measures\n\nNote: Average across Germany, Finland, Portugal, Hungary and Sweden. Each bar represents the contribution of a given occupational measure to the overall difference in earnings losses between high- and low-emission sectors based on an Oaxaca-Blinder decomposition. The reference category is non-routine interactive. Source: National linked employer employee data, see Annex Table 3.B.1 for details.\n\nStatLink 2 https://stat.link/w1n74g\n\n\\-0.07 -0.06 -0.05 -0.04 -0.03 -0.02 -0.01 0 0.01\n\nRoutine cognitive\n\nNon-routine cognitive analytic\n\nRoutine manual\n\nNon-routine manual\n\nFirm size\n\nFirm wage premia\n\nWorker FE\n\nTenure\n\nGender\n\nAge\n\nUnexplained\n\nExplained\n\nDifference\n\nEarnings relaitve to pre-displacement average, p.p.","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p161.pdf"]},"response":{"errors":[],"extract_id":"extract_70133c4388d1d46fcf3a6f87c609705f","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":[" **159**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\nlow-emission industries, and the second component captures the structural effect, i.e. unexplained differences in the cost of job displacement between high- and low-emission industries holding composition constant. A similar decomposition can be used to shed light on the drivers of differences in the cost of job displacement in high-emission industries between countries. The results of the decomposition between industries within countries and within industries between countries are presented in Figure 3.12.\n\n**The role of occupational composition**\n\nWhile the baseline analysis for all countries focuses on the role of firm and worker composition in explaining differences in the cost of job displacement between industries and countries, for a subset of countries with information 3-digit occupations (ISCO), it is possible to go further by analysing the role of differences in the composition of occupations. Occupations are categorised into five groups based on their task content following Autor, Levy and Murnane (2003[50]): non-routine manual, routine-manual, non-routine cognitive analytic, routine-cognitive and non-routine interactive. The analysis reveals that occupational composition plays an important role in explaining differences in the earnings losses of displacement between high- and low-emission industries, on average across countries analysed (Figure 3.13). More specifically, the concentration of routine-manual occupations accounts for 32% of the difference in earnings losses between high- and low-emission industries. This suggests that displaced workers in emission-intensive occupations may have difficulty finding another job because opportunities in the same occupation are limited and they lack the skills to move to other occupations with higher skill requirements.\n\n### **Figure 3.13. The concentration of manual routine occupations in high-emission industries contributes to higher earnings losses from job displacement**\n\nOaxaca-Blinder Decomposition, difference in earnings losses between high- and low-emission sectors including occupational measures\n\nNote: Average across Germany, Finland, Portugal, Hungary and Sweden. Each bar represents the contribution of a given occupational measure to the overall difference in earnings losses between high- and low-emission sectors based on an Oaxaca-Blinder decomposition. The reference category is non-routine interactive. Source: National linked employer employee data, see Annex Table 3.B.1 for details.\n\nStatLink 2 https://stat.link/w1n74g\n\n\\-0.07 -0.06 -0.05 -0.04 -0.03 -0.02 -0.01 0 0.01\n\nRoutine cognitive\n\nNon-routine cognitive analytic\n\nRoutine manual\n\nNon-routine manual\n\nFirm size\n\nFirm wage premia\n\nWorker FE\n\nTenure\n\nGender\n\nAge\n\nUnexplained\n\nExplained\n\nDifference\n\nEarnings relaitve to pre-displacement average, p.p."],"full_content":" **159**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\nlow-emission industries, and the second component captures the structural effect, i.e. unexplained differences in the cost of job displacement between high- and low-emission industries holding composition constant. A similar decomposition can be used to shed light on the drivers of differences in the cost of job displacement in high-emission industries between countries. The results of the decomposition between industries within countries and within industries between countries are presented in Figure 3.12.\n\n**The role of occupational composition**\n\nWhile the baseline analysis for all countries focuses on the role of firm and worker composition in explaining differences in the cost of job displacement between industries and countries, for a subset of countries with information 3-digit occupations (ISCO), it is possible to go further by analysing the role of differences in the composition of occupations. Occupations are categorised into five groups based on their task content following Autor, Levy and Murnane (2003[50]): non-routine manual, routine-manual, non-routine cognitive analytic, routine-cognitive and non-routine interactive. The analysis reveals that occupational composition plays an important role in explaining differences in the earnings losses of displacement between high- and low-emission industries, on average across countries analysed (Figure 3.13). More specifically, the concentration of routine-manual occupations accounts for 32% of the difference in earnings losses between high- and low-emission industries. This suggests that displaced workers in emission-intensive occupations may have difficulty finding another job because opportunities in the same occupation are limited and they lack the skills to move to other occupations with higher skill requirements.\n\n### **Figure 3.13. The concentration of manual routine occupations in high-emission industries contributes to higher earnings losses from job displacement**\n\nOaxaca-Blinder Decomposition, difference in earnings losses between high- and low-emission sectors including occupational measures\n\nNote: Average across Germany, Finland, Portugal, Hungary and Sweden. Each bar represents the contribution of a given occupational measure to the overall difference in earnings losses between high- and low-emission sectors based on an Oaxaca-Blinder decomposition. The reference category is non-routine interactive. Source: National linked employer employee data, see Annex Table 3.B.1 for details.\n\nStatLink 2 https://stat.link/w1n74g\n\n\\-0.07 -0.06 -0.05 -0.04 -0.03 -0.02 -0.01 0 0.01\n\nRoutine cognitive\n\nNon-routine cognitive analytic\n\nRoutine manual\n\nNon-routine manual\n\nFirm size\n\nFirm wage premia\n\nWorker FE\n\nTenure\n\nGender\n\nAge\n\nUnexplained\n\nExplained\n\nDifference\n\nEarnings relaitve to pre-displacement average, p.p.","publish_date":null,"title":"Figure 3.13. The concentration of manual routine occupations in high-emission industries contributes to higher earnings losses from job displacement","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p161.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p161.pdf"},"request":{"example_id":"chart/ac8b3538-en_p161","product_type":"parse","source_file_path":"ac8b3538-en_p161.pdf"},"started_at":"2026-11-04T19:46:31.755912"} +{"completed_at":"2026-11-04T19:46:34.673119","latency_in_ms":1257,"output":{"example_id":"chart/ac8b3538-en_p182","grounded_pages":[],"layout_pages":[],"markdown":"**180** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n# **Annex 3.A.** **Additional figures**\n\n### **Annex Figure 3.A.1. Displacement costs are slightly larger when including older workers**\n\nDifference in annual earnings between displaced workers and their matched counterparts relative to the time of displacement, average across countries, workers aged 18 to 60, percentage\n\nNote: The figure plots the average coefficients and the corresponding 90% confidence intervals across countries based on Equation (1) for workers aged 16 to 60. The coefficients capture the earnings losses of displaced workers relative to observationally comparable non-displaced workers. The point estimates show the impact of job loss on earnings in event time, where workers are displaced between time 0 and time 1, such that time 1 is the first post-displacement year. Related to this, earnings losses present a drop by construction at time 0, as earnings capture the sum of labour payments over the entire year and consequently already capture part of the displacement effect at time 0. The reference period for earnings losses is k=-2. Point estimates and confidence intervals from country-level regressions are averaged assigning each country an equal weight. The countries included are: Australia, Austria, Canada, Denmark, Estonia, Finland, Germany, Hungary, the Netherlands, Norway, Portugal, Spain, France, Sweden. Source: National linked employer employee data, see Annex Table 3.B.1 for details.\n\nStatLink 2 https://stat.link/yf4gt1\n\n\\-70\n\n\\-60\n\n\\-50\n\n\\-40\n\n\\-30\n\n\\-20\n\n\\-10\n\n0\n\n10\n\n\\-3 -2 -1 0 1 2 3 4 5 6\n\n%\n\nYear relative to job loss\n\nHigh-emission sectors Low-emission sectors","pages":[{"line_numbers":[],"markdown":"**180** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n# **Annex 3.A.** **Additional figures**\n\n### **Annex Figure 3.A.1. Displacement costs are slightly larger when including older workers**\n\nDifference in annual earnings between displaced workers and their matched counterparts relative to the time of displacement, average across countries, workers aged 18 to 60, percentage\n\nNote: The figure plots the average coefficients and the corresponding 90% confidence intervals across countries based on Equation (1) for workers aged 16 to 60. The coefficients capture the earnings losses of displaced workers relative to observationally comparable non-displaced workers. The point estimates show the impact of job loss on earnings in event time, where workers are displaced between time 0 and time 1, such that time 1 is the first post-displacement year. Related to this, earnings losses present a drop by construction at time 0, as earnings capture the sum of labour payments over the entire year and consequently already capture part of the displacement effect at time 0. The reference period for earnings losses is k=-2. Point estimates and confidence intervals from country-level regressions are averaged assigning each country an equal weight. The countries included are: Australia, Austria, Canada, Denmark, Estonia, Finland, Germany, Hungary, the Netherlands, Norway, Portugal, Spain, France, Sweden. Source: National linked employer employee data, see Annex Table 3.B.1 for details.\n\nStatLink 2 https://stat.link/yf4gt1\n\n\\-70\n\n\\-60\n\n\\-50\n\n\\-40\n\n\\-30\n\n\\-20\n\n\\-10\n\n0\n\n10\n\n\\-3 -2 -1 0 1 2 3 4 5 6\n\n%\n\nYear relative to job loss\n\nHigh-emission sectors Low-emission sectors","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**180** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n# **Annex 3.A.** **Additional figures**\n\n### **Annex Figure 3.A.1. Displacement costs are slightly larger when including older workers**\n\nDifference in annual earnings between displaced workers and their matched counterparts relative to the time of displacement, average across countries, workers aged 18 to 60, percentage\n\nNote: The figure plots the average coefficients and the corresponding 90% confidence intervals across countries based on Equation (1) for workers aged 16 to 60. The coefficients capture the earnings losses of displaced workers relative to observationally comparable non-displaced workers. The point estimates show the impact of job loss on earnings in event time, where workers are displaced between time 0 and time 1, such that time 1 is the first post-displacement year. Related to this, earnings losses present a drop by construction at time 0, as earnings capture the sum of labour payments over the entire year and consequently already capture part of the displacement effect at time 0. The reference period for earnings losses is k=-2. Point estimates and confidence intervals from country-level regressions are averaged assigning each country an equal weight. The countries included are: Australia, Austria, Canada, Denmark, Estonia, Finland, Germany, Hungary, the Netherlands, Norway, Portugal, Spain, France, Sweden. Source: National linked employer employee data, see Annex Table 3.B.1 for details.\n\nStatLink 2 https://stat.link/yf4gt1\n\n\\-70\n\n\\-60\n\n\\-50\n\n\\-40\n\n\\-30\n\n\\-20\n\n\\-10\n\n0\n\n10\n\n\\-3 -2 -1 0 1 2 3 4 5 6\n\n%\n\nYear relative to job loss\n\nHigh-emission sectors Low-emission sectors","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p182.pdf"]},"response":{"errors":[],"extract_id":"extract_04dd80d48383cd688b5a2460c68a2446","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**180** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n# **Annex 3.A.** **Additional figures**\n\n### **Annex Figure 3.A.1. Displacement costs are slightly larger when including older workers**\n\nDifference in annual earnings between displaced workers and their matched counterparts relative to the time of displacement, average across countries, workers aged 18 to 60, percentage\n\nNote: The figure plots the average coefficients and the corresponding 90% confidence intervals across countries based on Equation (1) for workers aged 16 to 60. The coefficients capture the earnings losses of displaced workers relative to observationally comparable non-displaced workers. The point estimates show the impact of job loss on earnings in event time, where workers are displaced between time 0 and time 1, such that time 1 is the first post-displacement year. Related to this, earnings losses present a drop by construction at time 0, as earnings capture the sum of labour payments over the entire year and consequently already capture part of the displacement effect at time 0. The reference period for earnings losses is k=-2. Point estimates and confidence intervals from country-level regressions are averaged assigning each country an equal weight. The countries included are: Australia, Austria, Canada, Denmark, Estonia, Finland, Germany, Hungary, the Netherlands, Norway, Portugal, Spain, France, Sweden. Source: National linked employer employee data, see Annex Table 3.B.1 for details.\n\nStatLink 2 https://stat.link/yf4gt1\n\n\\-70\n\n\\-60\n\n\\-50\n\n\\-40\n\n\\-30\n\n\\-20\n\n\\-10\n\n0\n\n10\n\n\\-3 -2 -1 0 1 2 3 4 5 6\n\n%\n\nYear relative to job loss\n\nHigh-emission sectors Low-emission sectors"],"full_content":"**180** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n# **Annex 3.A.** **Additional figures**\n\n### **Annex Figure 3.A.1. Displacement costs are slightly larger when including older workers**\n\nDifference in annual earnings between displaced workers and their matched counterparts relative to the time of displacement, average across countries, workers aged 18 to 60, percentage\n\nNote: The figure plots the average coefficients and the corresponding 90% confidence intervals across countries based on Equation (1) for workers aged 16 to 60. The coefficients capture the earnings losses of displaced workers relative to observationally comparable non-displaced workers. The point estimates show the impact of job loss on earnings in event time, where workers are displaced between time 0 and time 1, such that time 1 is the first post-displacement year. Related to this, earnings losses present a drop by construction at time 0, as earnings capture the sum of labour payments over the entire year and consequently already capture part of the displacement effect at time 0. The reference period for earnings losses is k=-2. Point estimates and confidence intervals from country-level regressions are averaged assigning each country an equal weight. The countries included are: Australia, Austria, Canada, Denmark, Estonia, Finland, Germany, Hungary, the Netherlands, Norway, Portugal, Spain, France, Sweden. Source: National linked employer employee data, see Annex Table 3.B.1 for details.\n\nStatLink 2 https://stat.link/yf4gt1\n\n\\-70\n\n\\-60\n\n\\-50\n\n\\-40\n\n\\-30\n\n\\-20\n\n\\-10\n\n0\n\n10\n\n\\-3 -2 -1 0 1 2 3 4 5 6\n\n%\n\nYear relative to job loss\n\nHigh-emission sectors Low-emission sectors","publish_date":null,"title":"Annex 3.A. Additional figures","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p182.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p182.pdf"},"request":{"example_id":"chart/ac8b3538-en_p182","product_type":"parse","source_file_path":"ac8b3538-en_p182.pdf"},"started_at":"2026-11-04T19:46:33.415128"} +{"completed_at":"2026-11-04T19:46:37.402642","latency_in_ms":2728,"output":{"example_id":"chart/ac8b3538-en_p183","grounded_pages":[],"layout_pages":[],"markdown":" **181**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Annex Figure 3.A.2. Country-rankings in the level of earnings losses following displacement in low- emission industries roughly mirror those for high-emission industries**\n\nDecomposition of the average earnings losses over six years since displacement in low-emission industries into the contributions associated with wages, employment and days workers by country, percentage\n\nNote: Bars refer to the average effect of job loss on earnings over six years following displacement for workers in low-emission sectors. Countries are sorted based on the level of earnings losses for low-emission sectors. Due to missing information on daily wages and days worked, the bars for Australia, Canada, Estonia and Norway are not decomposed into components. Source: National linked employer employee data, see Annex Table 3.B.1 for details.\n\nStatLink 2 https://stat.link/zip64g\n\n\\-70 -65 -60 -55 -50 -45 -40 -35 -30 -25 -20 -15 -10 -5 0 5\n\nPRT EST ESP HUN\n\nAverage\n\nAUT FIN FRA DNK NLD CAN AUS DEU SWE NOR\n\n%\n\nWages Employment Days worked Total","pages":[{"line_numbers":[],"markdown":" **181**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Annex Figure 3.A.2. Country-rankings in the level of earnings losses following displacement in low- emission industries roughly mirror those for high-emission industries**\n\nDecomposition of the average earnings losses over six years since displacement in low-emission industries into the contributions associated with wages, employment and days workers by country, percentage\n\nNote: Bars refer to the average effect of job loss on earnings over six years following displacement for workers in low-emission sectors. Countries are sorted based on the level of earnings losses for low-emission sectors. Due to missing information on daily wages and days worked, the bars for Australia, Canada, Estonia and Norway are not decomposed into components. Source: National linked employer employee data, see Annex Table 3.B.1 for details.\n\nStatLink 2 https://stat.link/zip64g\n\n\\-70 -65 -60 -55 -50 -45 -40 -35 -30 -25 -20 -15 -10 -5 0 5\n\nPRT EST ESP HUN\n\nAverage\n\nAUT FIN FRA DNK NLD CAN AUS DEU SWE NOR\n\n%\n\nWages Employment Days worked Total","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":" **181**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Annex Figure 3.A.2. Country-rankings in the level of earnings losses following displacement in low- emission industries roughly mirror those for high-emission industries**\n\nDecomposition of the average earnings losses over six years since displacement in low-emission industries into the contributions associated with wages, employment and days workers by country, percentage\n\nNote: Bars refer to the average effect of job loss on earnings over six years following displacement for workers in low-emission sectors. Countries are sorted based on the level of earnings losses for low-emission sectors. Due to missing information on daily wages and days worked, the bars for Australia, Canada, Estonia and Norway are not decomposed into components. Source: National linked employer employee data, see Annex Table 3.B.1 for details.\n\nStatLink 2 https://stat.link/zip64g\n\n\\-70 -65 -60 -55 -50 -45 -40 -35 -30 -25 -20 -15 -10 -5 0 5\n\nPRT EST ESP HUN\n\nAverage\n\nAUT FIN FRA DNK NLD CAN AUS DEU SWE NOR\n\n%\n\nWages Employment Days worked Total","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p183.pdf"]},"response":{"errors":[],"extract_id":"extract_ad1e19fffc70aabcc2f5b7d2addf1e7d","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":[" **181**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Annex Figure 3.A.2. Country-rankings in the level of earnings losses following displacement in low- emission industries roughly mirror those for high-emission industries**\n\nDecomposition of the average earnings losses over six years since displacement in low-emission industries into the contributions associated with wages, employment and days workers by country, percentage\n\nNote: Bars refer to the average effect of job loss on earnings over six years following displacement for workers in low-emission sectors. Countries are sorted based on the level of earnings losses for low-emission sectors. Due to missing information on daily wages and days worked, the bars for Australia, Canada, Estonia and Norway are not decomposed into components. Source: National linked employer employee data, see Annex Table 3.B.1 for details.\n\nStatLink 2 https://stat.link/zip64g\n\n\\-70 -65 -60 -55 -50 -45 -40 -35 -30 -25 -20 -15 -10 -5 0 5\n\nPRT EST ESP HUN\n\nAverage\n\nAUT FIN FRA DNK NLD CAN AUS DEU SWE NOR\n\n%\n\nWages Employment Days worked Total"],"full_content":" **181**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Annex Figure 3.A.2. Country-rankings in the level of earnings losses following displacement in low- emission industries roughly mirror those for high-emission industries**\n\nDecomposition of the average earnings losses over six years since displacement in low-emission industries into the contributions associated with wages, employment and days workers by country, percentage\n\nNote: Bars refer to the average effect of job loss on earnings over six years following displacement for workers in low-emission sectors. Countries are sorted based on the level of earnings losses for low-emission sectors. Due to missing information on daily wages and days worked, the bars for Australia, Canada, Estonia and Norway are not decomposed into components. Source: National linked employer employee data, see Annex Table 3.B.1 for details.\n\nStatLink 2 https://stat.link/zip64g\n\n\\-70 -65 -60 -55 -50 -45 -40 -35 -30 -25 -20 -15 -10 -5 0 5\n\nPRT EST ESP HUN\n\nAverage\n\nAUT FIN FRA DNK NLD CAN AUS DEU SWE NOR\n\n%\n\nWages Employment Days worked Total","publish_date":null,"title":"Annex Figure 3.A.2. Country-rankings in the level of earnings losses following displacement in low- emission industries roughly mirror those for","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p183.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p183.pdf"},"request":{"example_id":"chart/ac8b3538-en_p183","product_type":"parse","source_file_path":"ac8b3538-en_p183.pdf"},"started_at":"2026-11-04T19:46:34.674431"} +{"completed_at":"2026-11-04T19:46:38.327013","latency_in_ms":1457,"output":{"example_id":"chart/ac8b3538-en_p19","grounded_pages":[],"layout_pages":[],"markdown":" **17**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Infographic 1. Key facts and figures**\n\nCumulative % change in real wages between Q4 2019 and Q1 2024\n\nThe OECD-wide employment is expected to increase, but growth is projected to slow from 1.7% in 2023 to around 0.7% per year over 2024-25.\n\nIn Q1 2024, yearly real wage growth was positive in 29 of the 35 countries for which data are available. However, real wages were still below their Q4 2019 level in 16 of them.\n\n20% of workers are in green-driven jobs (incl. jobs that support ‘green’ activities). Adding jobs in high-emission industries means that over 25% of all jobs are set to be significantly impacted by the net-zero transition.\n\n**More than a quarter of jobs will be strongly impacted by net-zero transition**\n\n100\n\n110\n\n120\n\n130\n\n2000 2005 2010 2015 2020 2025\n\nProjections\n\n**High emission industries represent a small share of overall employment**\n\n% of total greenhouse gas (GHG) emissions and total employment of high emission industries, 2019\n\n% loss of earnings compared to non-displaced workers Required skill level\n\n% among total employment\n\nHigh emission industries account for almost 80% of GHG emissions, but represent only about 7% of overall employment.\n\nWorkers in high-emission industries who lose their jobs experience an average drop in annual earnings of 36% over six years, compared to 29% in other industries. Skill similarities across jobs make transitions feasible for all workers, although low-skilled adults need more retraining.\n\n**Real wages are growing but remain below pre-COVID levels in many OECD countries Employment growth remains strong but expected to slow**\n\nOECD-wide employment, index base 100 in 2000\n\n**Reskilling is key for moving away from emission-intensive occupations The cost of losing jobs in high-emission industries is substantial**\n\nGreen-driven jobs GHG-intensive jobs\n\nNew green-driven jobs GHG-intensive jobs\n\nLow-skill occupations\n\nHigh-skill occupations\n\n\\-60%\n\n\\-40%\n\n\\-20%\n\n0%\n\n\\-2 -1 0 1 2 3 4 5 6\n\n0 10 20 30 40\n\nPoland\n\nAustralia\n\nOECD\n\nJapan\n\nGermany\n\nUS\n\n\\-6 -4 -2 0 2 4 6 8 10\n\nPoland\n\nPortugal\n\nUK US\n\nGermany\n\nSpain\n\nAustralia\n\nProblem solving\n\nSocial skills\n\nProblem solving\n\nSocial skills\n\nVery low Very high\n\nShare of employment Share of emissions\n\n## 7% 79%\n\nYear relative to job loss\n\nLow-emission industries\n\nHigh-emission industries","pages":[{"line_numbers":[],"markdown":" **17**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Infographic 1. Key facts and figures**\n\nCumulative % change in real wages between Q4 2019 and Q1 2024\n\nThe OECD-wide employment is expected to increase, but growth is projected to slow from 1.7% in 2023 to around 0.7% per year over 2024-25.\n\nIn Q1 2024, yearly real wage growth was positive in 29 of the 35 countries for which data are available. However, real wages were still below their Q4 2019 level in 16 of them.\n\n20% of workers are in green-driven jobs (incl. jobs that support ‘green’ activities). Adding jobs in high-emission industries means that over 25% of all jobs are set to be significantly impacted by the net-zero transition.\n\n**More than a quarter of jobs will be strongly impacted by net-zero transition**\n\n100\n\n110\n\n120\n\n130\n\n2000 2005 2010 2015 2020 2025\n\nProjections\n\n**High emission industries represent a small share of overall employment**\n\n% of total greenhouse gas (GHG) emissions and total employment of high emission industries, 2019\n\n% loss of earnings compared to non-displaced workers Required skill level\n\n% among total employment\n\nHigh emission industries account for almost 80% of GHG emissions, but represent only about 7% of overall employment.\n\nWorkers in high-emission industries who lose their jobs experience an average drop in annual earnings of 36% over six years, compared to 29% in other industries. Skill similarities across jobs make transitions feasible for all workers, although low-skilled adults need more retraining.\n\n**Real wages are growing but remain below pre-COVID levels in many OECD countries Employment growth remains strong but expected to slow**\n\nOECD-wide employment, index base 100 in 2000\n\n**Reskilling is key for moving away from emission-intensive occupations The cost of losing jobs in high-emission industries is substantial**\n\nGreen-driven jobs GHG-intensive jobs\n\nNew green-driven jobs GHG-intensive jobs\n\nLow-skill occupations\n\nHigh-skill occupations\n\n\\-60%\n\n\\-40%\n\n\\-20%\n\n0%\n\n\\-2 -1 0 1 2 3 4 5 6\n\n0 10 20 30 40\n\nPoland\n\nAustralia\n\nOECD\n\nJapan\n\nGermany\n\nUS\n\n\\-6 -4 -2 0 2 4 6 8 10\n\nPoland\n\nPortugal\n\nUK US\n\nGermany\n\nSpain\n\nAustralia\n\nProblem solving\n\nSocial skills\n\nProblem solving\n\nSocial skills\n\nVery low Very high\n\nShare of employment Share of emissions\n\n## 7% 79%\n\nYear relative to job loss\n\nLow-emission industries\n\nHigh-emission industries","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":" **17**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Infographic 1. Key facts and figures**\n\nCumulative % change in real wages between Q4 2019 and Q1 2024\n\nThe OECD-wide employment is expected to increase, but growth is projected to slow from 1.7% in 2023 to around 0.7% per year over 2024-25.\n\nIn Q1 2024, yearly real wage growth was positive in 29 of the 35 countries for which data are available. However, real wages were still below their Q4 2019 level in 16 of them.\n\n20% of workers are in green-driven jobs (incl. jobs that support ‘green’ activities). Adding jobs in high-emission industries means that over 25% of all jobs are set to be significantly impacted by the net-zero transition.\n\n**More than a quarter of jobs will be strongly impacted by net-zero transition**\n\n100\n\n110\n\n120\n\n130\n\n2000 2005 2010 2015 2020 2025\n\nProjections\n\n**High emission industries represent a small share of overall employment**\n\n% of total greenhouse gas (GHG) emissions and total employment of high emission industries, 2019\n\n% loss of earnings compared to non-displaced workers Required skill level\n\n% among total employment\n\nHigh emission industries account for almost 80% of GHG emissions, but represent only about 7% of overall employment.\n\nWorkers in high-emission industries who lose their jobs experience an average drop in annual earnings of 36% over six years, compared to 29% in other industries. Skill similarities across jobs make transitions feasible for all workers, although low-skilled adults need more retraining.\n\n**Real wages are growing but remain below pre-COVID levels in many OECD countries Employment growth remains strong but expected to slow**\n\nOECD-wide employment, index base 100 in 2000\n\n**Reskilling is key for moving away from emission-intensive occupations The cost of losing jobs in high-emission industries is substantial**\n\nGreen-driven jobs GHG-intensive jobs\n\nNew green-driven jobs GHG-intensive jobs\n\nLow-skill occupations\n\nHigh-skill occupations\n\n\\-60%\n\n\\-40%\n\n\\-20%\n\n0%\n\n\\-2 -1 0 1 2 3 4 5 6\n\n0 10 20 30 40\n\nPoland\n\nAustralia\n\nOECD\n\nJapan\n\nGermany\n\nUS\n\n\\-6 -4 -2 0 2 4 6 8 10\n\nPoland\n\nPortugal\n\nUK US\n\nGermany\n\nSpain\n\nAustralia\n\nProblem solving\n\nSocial skills\n\nProblem solving\n\nSocial skills\n\nVery low Very high\n\nShare of employment Share of emissions\n\n## 7% 79%\n\nYear relative to job loss\n\nLow-emission industries\n\nHigh-emission industries","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p19.pdf"]},"response":{"errors":[],"extract_id":"extract_8402b298d4873f24206b4c0e077fc432","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":[" **17**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Infographic 1. Key facts and figures**\n\nCumulative % change in real wages between Q4 2019 and Q1 2024\n\nThe OECD-wide employment is expected to increase, but growth is projected to slow from 1.7% in 2023 to around 0.7% per year over 2024-25.\n\nIn Q1 2024, yearly real wage growth was positive in 29 of the 35 countries for which data are available. However, real wages were still below their Q4 2019 level in 16 of them.\n\n20% of workers are in green-driven jobs (incl. jobs that support ‘green’ activities). Adding jobs in high-emission industries means that over 25% of all jobs are set to be significantly impacted by the net-zero transition.\n\n**More than a quarter of jobs will be strongly impacted by net-zero transition**\n\n100\n\n110\n\n120\n\n130\n\n2000 2005 2010 2015 2020 2025\n\nProjections\n\n**High emission industries represent a small share of overall employment**\n\n% of total greenhouse gas (GHG) emissions and total employment of high emission industries, 2019\n\n% loss of earnings compared to non-displaced workers Required skill level\n\n% among total employment\n\nHigh emission industries account for almost 80% of GHG emissions, but represent only about 7% of overall employment.\n\nWorkers in high-emission industries who lose their jobs experience an average drop in annual earnings of 36% over six years, compared to 29% in other industries. Skill similarities across jobs make transitions feasible for all workers, although low-skilled adults need more retraining.\n\n**Real wages are growing but remain below pre-COVID levels in many OECD countries Employment growth remains strong but expected to slow**\n\nOECD-wide employment, index base 100 in 2000\n\n**Reskilling is key for moving away from emission-intensive occupations The cost of losing jobs in high-emission industries is substantial**\n\nGreen-driven jobs GHG-intensive jobs\n\nNew green-driven jobs GHG-intensive jobs\n\nLow-skill occupations\n\nHigh-skill occupations\n\n\\-60%\n\n\\-40%\n\n\\-20%\n\n0%\n\n\\-2 -1 0 1 2 3 4 5 6\n\n0 10 20 30 40\n\nPoland\n\nAustralia\n\nOECD\n\nJapan\n\nGermany\n\nUS\n\n\\-6 -4 -2 0 2 4 6 8 10\n\nPoland\n\nPortugal\n\nUK US\n\nGermany\n\nSpain\n\nAustralia\n\nProblem solving\n\nSocial skills\n\nProblem solving\n\nSocial skills\n\nVery low Very high\n\nShare of employment Share of emissions\n\n## 7% 79%\n\nYear relative to job loss\n\nLow-emission industries\n\nHigh-emission industries"],"full_content":" **17**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Infographic 1. Key facts and figures**\n\nCumulative % change in real wages between Q4 2019 and Q1 2024\n\nThe OECD-wide employment is expected to increase, but growth is projected to slow from 1.7% in 2023 to around 0.7% per year over 2024-25.\n\nIn Q1 2024, yearly real wage growth was positive in 29 of the 35 countries for which data are available. However, real wages were still below their Q4 2019 level in 16 of them.\n\n20% of workers are in green-driven jobs (incl. jobs that support ‘green’ activities). Adding jobs in high-emission industries means that over 25% of all jobs are set to be significantly impacted by the net-zero transition.\n\n**More than a quarter of jobs will be strongly impacted by net-zero transition**\n\n100\n\n110\n\n120\n\n130\n\n2000 2005 2010 2015 2020 2025\n\nProjections\n\n**High emission industries represent a small share of overall employment**\n\n% of total greenhouse gas (GHG) emissions and total employment of high emission industries, 2019\n\n% loss of earnings compared to non-displaced workers Required skill level\n\n% among total employment\n\nHigh emission industries account for almost 80% of GHG emissions, but represent only about 7% of overall employment.\n\nWorkers in high-emission industries who lose their jobs experience an average drop in annual earnings of 36% over six years, compared to 29% in other industries. Skill similarities across jobs make transitions feasible for all workers, although low-skilled adults need more retraining.\n\n**Real wages are growing but remain below pre-COVID levels in many OECD countries Employment growth remains strong but expected to slow**\n\nOECD-wide employment, index base 100 in 2000\n\n**Reskilling is key for moving away from emission-intensive occupations The cost of losing jobs in high-emission industries is substantial**\n\nGreen-driven jobs GHG-intensive jobs\n\nNew green-driven jobs GHG-intensive jobs\n\nLow-skill occupations\n\nHigh-skill occupations\n\n\\-60%\n\n\\-40%\n\n\\-20%\n\n0%\n\n\\-2 -1 0 1 2 3 4 5 6\n\n0 10 20 30 40\n\nPoland\n\nAustralia\n\nOECD\n\nJapan\n\nGermany\n\nUS\n\n\\-6 -4 -2 0 2 4 6 8 10\n\nPoland\n\nPortugal\n\nUK US\n\nGermany\n\nSpain\n\nAustralia\n\nProblem solving\n\nSocial skills\n\nProblem solving\n\nSocial skills\n\nVery low Very high\n\nShare of employment Share of emissions\n\n## 7% 79%\n\nYear relative to job loss\n\nLow-emission industries\n\nHigh-emission industries","publish_date":null,"title":"7 % 7 9 %","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p19.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p19.pdf"},"request":{"example_id":"chart/ac8b3538-en_p19","product_type":"parse","source_file_path":"ac8b3538-en_p19.pdf"},"started_at":"2026-11-04T19:46:36.869175"} +{"completed_at":"2026-11-04T19:46:40.893863","latency_in_ms":2563,"output":{"example_id":"chart/ac8b3538-en_p210","grounded_pages":[],"layout_pages":[],"markdown":"**208** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Figure 4.8. Workers in green-driven and GHG-intensive occupations train less than average**\n\nPoint estimate of the percentage difference between workers that participate in training and those that do not participate, 2018\n\nNote: Panel A and B report the point estimate (and 95% confidence intervals) of the percentage difference in the average incidence of each type of worker between those workers that undertake training and those that do not. Estimates are obtained from a linear regression with the inverse hyperbolic sine of the share of each type of occupation as a dependent variable and including a training dummy (formal training in Panel A and informal training in Panel B), educational attainment (3 classes), gender, age (3 classes) and country dummies as explanatory variables and standard errors clustered on the dimensions of variability of the dependent variable. The reported point estimates and confidence intervals refer to the estimated coefficient of the training dummy and are expressed in percentage of the untransformed dependent variable. Reading: Controlling for demographic characteristics, the percentage share of green-driven occupations is, on average, 50% lower among employees who undertake formal training and 18% lower among employees who undertake informal training, than among those that do not. The share of workers in GHG-intensive occupations participating is 31% lower for formal training and 12% lower for non-formal training. Source: OECD estimates based on version 24.1 of the O\\*NET database and the following country-specific sources: Australia: Table Builder of the Australian Bureau of Statistics (Labour Force: Characteristics of Employment); United States: Current Population Survey; All other countries: EU Structure of Earnings Surveys.\n\nTraining for the green transition is becoming a priority for many governments. In 2022 the European Council adopted recommendations to stimulate learning for sustainable development, which highlights the need to provide learning in formal, non-formal and informal settings (European Council, 2022[29]). The United Nations Environment Programme has recently produced a framework for education on green jobs – guidelines aimed at the higher education community on how to prepare students for the green economy (United Nations Environment Programme, 2021[30]). The importance of the challenge is not lost on OECD countries, with 13 out of 26 of them reporting they are funding new training and apprenticeship programmes for the green transition (OECD, 2023[31]). ¹⁸\n\nIn Canada, the Energy to Digital Growth Education and Upskilling Project (EDGE UP) was developed by Calgary Economic Development in partnership with the Information and Communication Technology Council (ICTC) with the goal of reskilling displaced mid-career oil and gas professionals and supporting them in starting a career in the technology sector in Calgary. The EDGE UP programme created reskilling pathways based on a skills mapping research study that highlighted the transferrable skills of workers in high-emission occupations and their skills gaps when compared with the needs of the most in-demand\n\n\\-60\n\n\\-50\n\n\\-40\n\n\\-30\n\n\\-20\n\n\\-10\n\n0\n\n10\n\n20\n\nNew green-driven\n\noccupations\n\nEstablished green- driven occupations\n\nGHG-intensive occupations\n\n%\n\n**A. Formal**\n\n\\-60\n\n\\-50\n\n\\-40\n\n\\-30\n\n\\-20\n\n\\-10\n\n0\n\n10\n\n20\n\nNew green-driven\n\noccupations\n\nEstablished green- driven occupations\n\nGHG-intensive occupations\n\n%\n\n**B. Not formal**","pages":[{"line_numbers":[],"markdown":"**208** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Figure 4.8. Workers in green-driven and GHG-intensive occupations train less than average**\n\nPoint estimate of the percentage difference between workers that participate in training and those that do not participate, 2018\n\nNote: Panel A and B report the point estimate (and 95% confidence intervals) of the percentage difference in the average incidence of each type of worker between those workers that undertake training and those that do not. Estimates are obtained from a linear regression with the inverse hyperbolic sine of the share of each type of occupation as a dependent variable and including a training dummy (formal training in Panel A and informal training in Panel B), educational attainment (3 classes), gender, age (3 classes) and country dummies as explanatory variables and standard errors clustered on the dimensions of variability of the dependent variable. The reported point estimates and confidence intervals refer to the estimated coefficient of the training dummy and are expressed in percentage of the untransformed dependent variable. Reading: Controlling for demographic characteristics, the percentage share of green-driven occupations is, on average, 50% lower among employees who undertake formal training and 18% lower among employees who undertake informal training, than among those that do not. The share of workers in GHG-intensive occupations participating is 31% lower for formal training and 12% lower for non-formal training. Source: OECD estimates based on version 24.1 of the O\\*NET database and the following country-specific sources: Australia: Table Builder of the Australian Bureau of Statistics (Labour Force: Characteristics of Employment); United States: Current Population Survey; All other countries: EU Structure of Earnings Surveys.\n\nTraining for the green transition is becoming a priority for many governments. In 2022 the European Council adopted recommendations to stimulate learning for sustainable development, which highlights the need to provide learning in formal, non-formal and informal settings (European Council, 2022[29]). The United Nations Environment Programme has recently produced a framework for education on green jobs – guidelines aimed at the higher education community on how to prepare students for the green economy (United Nations Environment Programme, 2021[30]). The importance of the challenge is not lost on OECD countries, with 13 out of 26 of them reporting they are funding new training and apprenticeship programmes for the green transition (OECD, 2023[31]). ¹⁸\n\nIn Canada, the Energy to Digital Growth Education and Upskilling Project (EDGE UP) was developed by Calgary Economic Development in partnership with the Information and Communication Technology Council (ICTC) with the goal of reskilling displaced mid-career oil and gas professionals and supporting them in starting a career in the technology sector in Calgary. The EDGE UP programme created reskilling pathways based on a skills mapping research study that highlighted the transferrable skills of workers in high-emission occupations and their skills gaps when compared with the needs of the most in-demand\n\n\\-60\n\n\\-50\n\n\\-40\n\n\\-30\n\n\\-20\n\n\\-10\n\n0\n\n10\n\n20\n\nNew green-driven\n\noccupations\n\nEstablished green- driven occupations\n\nGHG-intensive occupations\n\n%\n\n**A. Formal**\n\n\\-60\n\n\\-50\n\n\\-40\n\n\\-30\n\n\\-20\n\n\\-10\n\n0\n\n10\n\n20\n\nNew green-driven\n\noccupations\n\nEstablished green- driven occupations\n\nGHG-intensive occupations\n\n%\n\n**B. Not formal**","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**208** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Figure 4.8. Workers in green-driven and GHG-intensive occupations train less than average**\n\nPoint estimate of the percentage difference between workers that participate in training and those that do not participate, 2018\n\nNote: Panel A and B report the point estimate (and 95% confidence intervals) of the percentage difference in the average incidence of each type of worker between those workers that undertake training and those that do not. Estimates are obtained from a linear regression with the inverse hyperbolic sine of the share of each type of occupation as a dependent variable and including a training dummy (formal training in Panel A and informal training in Panel B), educational attainment (3 classes), gender, age (3 classes) and country dummies as explanatory variables and standard errors clustered on the dimensions of variability of the dependent variable. The reported point estimates and confidence intervals refer to the estimated coefficient of the training dummy and are expressed in percentage of the untransformed dependent variable. Reading: Controlling for demographic characteristics, the percentage share of green-driven occupations is, on average, 50% lower among employees who undertake formal training and 18% lower among employees who undertake informal training, than among those that do not. The share of workers in GHG-intensive occupations participating is 31% lower for formal training and 12% lower for non-formal training. Source: OECD estimates based on version 24.1 of the O\\*NET database and the following country-specific sources: Australia: Table Builder of the Australian Bureau of Statistics (Labour Force: Characteristics of Employment); United States: Current Population Survey; All other countries: EU Structure of Earnings Surveys.\n\nTraining for the green transition is becoming a priority for many governments. In 2022 the European Council adopted recommendations to stimulate learning for sustainable development, which highlights the need to provide learning in formal, non-formal and informal settings (European Council, 2022[29]). The United Nations Environment Programme has recently produced a framework for education on green jobs – guidelines aimed at the higher education community on how to prepare students for the green economy (United Nations Environment Programme, 2021[30]). The importance of the challenge is not lost on OECD countries, with 13 out of 26 of them reporting they are funding new training and apprenticeship programmes for the green transition (OECD, 2023[31]). ¹⁸\n\nIn Canada, the Energy to Digital Growth Education and Upskilling Project (EDGE UP) was developed by Calgary Economic Development in partnership with the Information and Communication Technology Council (ICTC) with the goal of reskilling displaced mid-career oil and gas professionals and supporting them in starting a career in the technology sector in Calgary. The EDGE UP programme created reskilling pathways based on a skills mapping research study that highlighted the transferrable skills of workers in high-emission occupations and their skills gaps when compared with the needs of the most in-demand\n\n\\-60\n\n\\-50\n\n\\-40\n\n\\-30\n\n\\-20\n\n\\-10\n\n0\n\n10\n\n20\n\nNew green-driven\n\noccupations\n\nEstablished green- driven occupations\n\nGHG-intensive occupations\n\n%\n\n**A. Formal**\n\n\\-60\n\n\\-50\n\n\\-40\n\n\\-30\n\n\\-20\n\n\\-10\n\n0\n\n10\n\n20\n\nNew green-driven\n\noccupations\n\nEstablished green- driven occupations\n\nGHG-intensive occupations\n\n%\n\n**B. Not formal**","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p210.pdf"]},"response":{"errors":[],"extract_id":"extract_52fce16d7618ee7212dba2238a890519","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**208** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Figure 4.8. Workers in green-driven and GHG-intensive occupations train less than average**\n\nPoint estimate of the percentage difference between workers that participate in training and those that do not participate, 2018\n\nNote: Panel A and B report the point estimate (and 95% confidence intervals) of the percentage difference in the average incidence of each type of worker between those workers that undertake training and those that do not. Estimates are obtained from a linear regression with the inverse hyperbolic sine of the share of each type of occupation as a dependent variable and including a training dummy (formal training in Panel A and informal training in Panel B), educational attainment (3 classes), gender, age (3 classes) and country dummies as explanatory variables and standard errors clustered on the dimensions of variability of the dependent variable. The reported point estimates and confidence intervals refer to the estimated coefficient of the training dummy and are expressed in percentage of the untransformed dependent variable. Reading: Controlling for demographic characteristics, the percentage share of green-driven occupations is, on average, 50% lower among employees who undertake formal training and 18% lower among employees who undertake informal training, than among those that do not. The share of workers in GHG-intensive occupations participating is 31% lower for formal training and 12% lower for non-formal training. Source: OECD estimates based on version 24.1 of the O\\*NET database and the following country-specific sources: Australia: Table Builder of the Australian Bureau of Statistics (Labour Force: Characteristics of Employment); United States: Current Population Survey; All other countries: EU Structure of Earnings Surveys.\n\nTraining for the green transition is becoming a priority for many governments. In 2022 the European Council adopted recommendations to stimulate learning for sustainable development, which highlights the need to provide learning in formal, non-formal and informal settings (European Council, 2022[29]). The United Nations Environment Programme has recently produced a framework for education on green jobs – guidelines aimed at the higher education community on how to prepare students for the green economy (United Nations Environment Programme, 2021[30]). The importance of the challenge is not lost on OECD countries, with 13 out of 26 of them reporting they are funding new training and apprenticeship programmes for the green transition (OECD, 2023[31]). ¹⁸\n\nIn Canada, the Energy to Digital Growth Education and Upskilling Project (EDGE UP) was developed by Calgary Economic Development in partnership with the Information and Communication Technology Council (ICTC) with the goal of reskilling displaced mid-career oil and gas professionals and supporting them in starting a career in the technology sector in Calgary. The EDGE UP programme created reskilling pathways based on a skills mapping research study that highlighted the transferrable skills of workers in high-emission occupations and their skills gaps when compared with the needs of the most in-demand\n\n\\-60\n\n\\-50\n\n\\-40\n\n\\-30\n\n\\-20\n\n\\-10\n\n0\n\n10\n\n20\n\nNew green-driven\n\noccupations\n\nEstablished green- driven occupations\n\nGHG-intensive occupations\n\n%\n\n**A. Formal**\n\n\\-60\n\n\\-50\n\n\\-40\n\n\\-30\n\n\\-20\n\n\\-10\n\n0\n\n10\n\n20\n\nNew green-driven\n\noccupations\n\nEstablished green- driven occupations\n\nGHG-intensive occupations\n\n%\n\n**B. Not formal**"],"full_content":"**208** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Figure 4.8. Workers in green-driven and GHG-intensive occupations train less than average**\n\nPoint estimate of the percentage difference between workers that participate in training and those that do not participate, 2018\n\nNote: Panel A and B report the point estimate (and 95% confidence intervals) of the percentage difference in the average incidence of each type of worker between those workers that undertake training and those that do not. Estimates are obtained from a linear regression with the inverse hyperbolic sine of the share of each type of occupation as a dependent variable and including a training dummy (formal training in Panel A and informal training in Panel B), educational attainment (3 classes), gender, age (3 classes) and country dummies as explanatory variables and standard errors clustered on the dimensions of variability of the dependent variable. The reported point estimates and confidence intervals refer to the estimated coefficient of the training dummy and are expressed in percentage of the untransformed dependent variable. Reading: Controlling for demographic characteristics, the percentage share of green-driven occupations is, on average, 50% lower among employees who undertake formal training and 18% lower among employees who undertake informal training, than among those that do not. The share of workers in GHG-intensive occupations participating is 31% lower for formal training and 12% lower for non-formal training. Source: OECD estimates based on version 24.1 of the O\\*NET database and the following country-specific sources: Australia: Table Builder of the Australian Bureau of Statistics (Labour Force: Characteristics of Employment); United States: Current Population Survey; All other countries: EU Structure of Earnings Surveys.\n\nTraining for the green transition is becoming a priority for many governments. In 2022 the European Council adopted recommendations to stimulate learning for sustainable development, which highlights the need to provide learning in formal, non-formal and informal settings (European Council, 2022[29]). The United Nations Environment Programme has recently produced a framework for education on green jobs – guidelines aimed at the higher education community on how to prepare students for the green economy (United Nations Environment Programme, 2021[30]). The importance of the challenge is not lost on OECD countries, with 13 out of 26 of them reporting they are funding new training and apprenticeship programmes for the green transition (OECD, 2023[31]). ¹⁸\n\nIn Canada, the Energy to Digital Growth Education and Upskilling Project (EDGE UP) was developed by Calgary Economic Development in partnership with the Information and Communication Technology Council (ICTC) with the goal of reskilling displaced mid-career oil and gas professionals and supporting them in starting a career in the technology sector in Calgary. The EDGE UP programme created reskilling pathways based on a skills mapping research study that highlighted the transferrable skills of workers in high-emission occupations and their skills gaps when compared with the needs of the most in-demand\n\n\\-60\n\n\\-50\n\n\\-40\n\n\\-30\n\n\\-20\n\n\\-10\n\n0\n\n10\n\n20\n\nNew green-driven\n\noccupations\n\nEstablished green- driven occupations\n\nGHG-intensive occupations\n\n%\n\n**A. Formal**\n\n\\-60\n\n\\-50\n\n\\-40\n\n\\-30\n\n\\-20\n\n\\-10\n\n0\n\n10\n\n20\n\nNew green-driven\n\noccupations\n\nEstablished green- driven occupations\n\nGHG-intensive occupations\n\n%\n\n**B. Not formal**","publish_date":null,"title":"Figure 4.8. Workers in green-driven and GHG-intensive occupations train less than average","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p210.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p210.pdf"},"request":{"example_id":"chart/ac8b3538-en_p210","product_type":"parse","source_file_path":"ac8b3538-en_p210.pdf"},"started_at":"2026-11-04T19:46:38.330143"} +{"completed_at":"2026-11-04T19:46:40.963809","latency_in_ms":1370,"output":{"example_id":"chart/ac8b3538-en_p23","grounded_pages":[],"layout_pages":[],"markdown":" **21**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\nNon-wage aspects of jobs also need to be monitored to understand trends in job quality in the wake of the COVID-19 pandemic and the recent cost-of-living crisis. Drawing on the conceptual framework developed by the OECD (Cazes, Hijzen and Saint-Martin, 2015[1]; OECD, 2014[2]) and then adopted by the G20 (G20, 2015[3]), the chapter also provides an update on the three key indicators of job quality across countries – earnings quality, labour market security and quality of the working environment.\n\nThe chapter is organised as follows: Section 1.1 reviews recent labour market developments across the OECD countries; Section 1.2 reports on recent wage developments, including an update on statutory minimum wages and negotiated wages; and finally, Section 1.3 presents the latest\n\nOECD job quality indicators and analyses trends in these indicators since 2015. Section 1.4 concludes with policy recommendations.\n\n### **1\\.1. Labour markets have proven resilient in the wake of adverse shocks**\n\nGlobal GDP growth has been moderate, but relatively resilient in 2023 despite the negative shocks from Russia’s war of aggression against Ukraine and the sharp tightening of monetary policy to tackle high inflation. Growth was particularly strong in the United States and many emerging-market economies but saw a slowdown in most European countries (Figure 1.1). The attacks on ships in the Red Sea that started in Fall 2023, have raised shipping costs sharply and lengthened delivery times, disrupting production schedules and raising price pressures. According to the latest indicators, global GDP growth is projected to continue growing at a modest pace of 3.1% in 2024, the same growth as the 3.1% in 2023, followed by a slight pick-up to 3.2% in 2025 as financial conditions ease (OECD, 2024[4]).\n\n### **Figure 1.1. GDP growth has been moderate with significant divergence between countries**\n\nReal GDP indexed to 100 in Q4 2019 in selected OECD countries, seasonally adjusted data\n\nNote: Euro Area refers to the 20 Eurozone countries. Source: OECD (2024), “Quarterly National Accounts”, OECD National Accounts Statistics (database), https://doi.org/10.1787/data-00017-en (accessed on 12 June 2024).\n\nStatLink 2 https://stat.link/glsrka\n\n85\n\n90\n\n95\n\n100\n\n105\n\n110\n\nOECD Euro Area Australia Japan Korea United States","pages":[{"line_numbers":[],"markdown":" **21**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\nNon-wage aspects of jobs also need to be monitored to understand trends in job quality in the wake of the COVID-19 pandemic and the recent cost-of-living crisis. Drawing on the conceptual framework developed by the OECD (Cazes, Hijzen and Saint-Martin, 2015[1]; OECD, 2014[2]) and then adopted by the G20 (G20, 2015[3]), the chapter also provides an update on the three key indicators of job quality across countries – earnings quality, labour market security and quality of the working environment.\n\nThe chapter is organised as follows: Section 1.1 reviews recent labour market developments across the OECD countries; Section 1.2 reports on recent wage developments, including an update on statutory minimum wages and negotiated wages; and finally, Section 1.3 presents the latest\n\nOECD job quality indicators and analyses trends in these indicators since 2015. Section 1.4 concludes with policy recommendations.\n\n### **1\\.1. Labour markets have proven resilient in the wake of adverse shocks**\n\nGlobal GDP growth has been moderate, but relatively resilient in 2023 despite the negative shocks from Russia’s war of aggression against Ukraine and the sharp tightening of monetary policy to tackle high inflation. Growth was particularly strong in the United States and many emerging-market economies but saw a slowdown in most European countries (Figure 1.1). The attacks on ships in the Red Sea that started in Fall 2023, have raised shipping costs sharply and lengthened delivery times, disrupting production schedules and raising price pressures. According to the latest indicators, global GDP growth is projected to continue growing at a modest pace of 3.1% in 2024, the same growth as the 3.1% in 2023, followed by a slight pick-up to 3.2% in 2025 as financial conditions ease (OECD, 2024[4]).\n\n### **Figure 1.1. GDP growth has been moderate with significant divergence between countries**\n\nReal GDP indexed to 100 in Q4 2019 in selected OECD countries, seasonally adjusted data\n\nNote: Euro Area refers to the 20 Eurozone countries. Source: OECD (2024), “Quarterly National Accounts”, OECD National Accounts Statistics (database), https://doi.org/10.1787/data-00017-en (accessed on 12 June 2024).\n\nStatLink 2 https://stat.link/glsrka\n\n85\n\n90\n\n95\n\n100\n\n105\n\n110\n\nOECD Euro Area Australia Japan Korea United States","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":" **21**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\nNon-wage aspects of jobs also need to be monitored to understand trends in job quality in the wake of the COVID-19 pandemic and the recent cost-of-living crisis. Drawing on the conceptual framework developed by the OECD (Cazes, Hijzen and Saint-Martin, 2015[1]; OECD, 2014[2]) and then adopted by the G20 (G20, 2015[3]), the chapter also provides an update on the three key indicators of job quality across countries – earnings quality, labour market security and quality of the working environment.\n\nThe chapter is organised as follows: Section 1.1 reviews recent labour market developments across the OECD countries; Section 1.2 reports on recent wage developments, including an update on statutory minimum wages and negotiated wages; and finally, Section 1.3 presents the latest\n\nOECD job quality indicators and analyses trends in these indicators since 2015. Section 1.4 concludes with policy recommendations.\n\n### **1\\.1. Labour markets have proven resilient in the wake of adverse shocks**\n\nGlobal GDP growth has been moderate, but relatively resilient in 2023 despite the negative shocks from Russia’s war of aggression against Ukraine and the sharp tightening of monetary policy to tackle high inflation. Growth was particularly strong in the United States and many emerging-market economies but saw a slowdown in most European countries (Figure 1.1). The attacks on ships in the Red Sea that started in Fall 2023, have raised shipping costs sharply and lengthened delivery times, disrupting production schedules and raising price pressures. According to the latest indicators, global GDP growth is projected to continue growing at a modest pace of 3.1% in 2024, the same growth as the 3.1% in 2023, followed by a slight pick-up to 3.2% in 2025 as financial conditions ease (OECD, 2024[4]).\n\n### **Figure 1.1. GDP growth has been moderate with significant divergence between countries**\n\nReal GDP indexed to 100 in Q4 2019 in selected OECD countries, seasonally adjusted data\n\nNote: Euro Area refers to the 20 Eurozone countries. Source: OECD (2024), “Quarterly National Accounts”, OECD National Accounts Statistics (database), https://doi.org/10.1787/data-00017-en (accessed on 12 June 2024).\n\nStatLink 2 https://stat.link/glsrka\n\n85\n\n90\n\n95\n\n100\n\n105\n\n110\n\nOECD Euro Area Australia Japan Korea United States","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p23.pdf"]},"response":{"errors":[],"extract_id":"extract_e2b0da5d600cc1f57d5af56ea4d56049","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":[" **21**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\nNon-wage aspects of jobs also need to be monitored to understand trends in job quality in the wake of the COVID-19 pandemic and the recent cost-of-living crisis. Drawing on the conceptual framework developed by the OECD (Cazes, Hijzen and Saint-Martin, 2015[1]; OECD, 2014[2]) and then adopted by the G20 (G20, 2015[3]), the chapter also provides an update on the three key indicators of job quality across countries – earnings quality, labour market security and quality of the working environment.\n\nThe chapter is organised as follows: Section 1.1 reviews recent labour market developments across the OECD countries; Section 1.2 reports on recent wage developments, including an update on statutory minimum wages and negotiated wages; and finally, Section 1.3 presents the latest\n\nOECD job quality indicators and analyses trends in these indicators since 2015. Section 1.4 concludes with policy recommendations.\n\n### **1\\.1. Labour markets have proven resilient in the wake of adverse shocks**\n\nGlobal GDP growth has been moderate, but relatively resilient in 2023 despite the negative shocks from Russia’s war of aggression against Ukraine and the sharp tightening of monetary policy to tackle high inflation. Growth was particularly strong in the United States and many emerging-market economies but saw a slowdown in most European countries (Figure 1.1). The attacks on ships in the Red Sea that started in Fall 2023, have raised shipping costs sharply and lengthened delivery times, disrupting production schedules and raising price pressures. According to the latest indicators, global GDP growth is projected to continue growing at a modest pace of 3.1% in 2024, the same growth as the 3.1% in 2023, followed by a slight pick-up to 3.2% in 2025 as financial conditions ease (OECD, 2024[4]).\n\n### **Figure 1.1. GDP growth has been moderate with significant divergence between countries**\n\nReal GDP indexed to 100 in Q4 2019 in selected OECD countries, seasonally adjusted data\n\nNote: Euro Area refers to the 20 Eurozone countries. Source: OECD (2024), “Quarterly National Accounts”, OECD National Accounts Statistics (database), https://doi.org/10.1787/data-00017-en (accessed on 12 June 2024).\n\nStatLink 2 https://stat.link/glsrka\n\n85\n\n90\n\n95\n\n100\n\n105\n\n110\n\nOECD Euro Area Australia Japan Korea United States"],"full_content":" **21**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\nNon-wage aspects of jobs also need to be monitored to understand trends in job quality in the wake of the COVID-19 pandemic and the recent cost-of-living crisis. Drawing on the conceptual framework developed by the OECD (Cazes, Hijzen and Saint-Martin, 2015[1]; OECD, 2014[2]) and then adopted by the G20 (G20, 2015[3]), the chapter also provides an update on the three key indicators of job quality across countries – earnings quality, labour market security and quality of the working environment.\n\nThe chapter is organised as follows: Section 1.1 reviews recent labour market developments across the OECD countries; Section 1.2 reports on recent wage developments, including an update on statutory minimum wages and negotiated wages; and finally, Section 1.3 presents the latest\n\nOECD job quality indicators and analyses trends in these indicators since 2015. Section 1.4 concludes with policy recommendations.\n\n### **1\\.1. Labour markets have proven resilient in the wake of adverse shocks**\n\nGlobal GDP growth has been moderate, but relatively resilient in 2023 despite the negative shocks from Russia’s war of aggression against Ukraine and the sharp tightening of monetary policy to tackle high inflation. Growth was particularly strong in the United States and many emerging-market economies but saw a slowdown in most European countries (Figure 1.1). The attacks on ships in the Red Sea that started in Fall 2023, have raised shipping costs sharply and lengthened delivery times, disrupting production schedules and raising price pressures. According to the latest indicators, global GDP growth is projected to continue growing at a modest pace of 3.1% in 2024, the same growth as the 3.1% in 2023, followed by a slight pick-up to 3.2% in 2025 as financial conditions ease (OECD, 2024[4]).\n\n### **Figure 1.1. GDP growth has been moderate with significant divergence between countries**\n\nReal GDP indexed to 100 in Q4 2019 in selected OECD countries, seasonally adjusted data\n\nNote: Euro Area refers to the 20 Eurozone countries. Source: OECD (2024), “Quarterly National Accounts”, OECD National Accounts Statistics (database), https://doi.org/10.1787/data-00017-en (accessed on 12 June 2024).\n\nStatLink 2 https://stat.link/glsrka\n\n85\n\n90\n\n95\n\n100\n\n105\n\n110\n\nOECD Euro Area Australia Japan Korea United States","publish_date":null,"title":"1.1. Labour markets have proven resilient in the wake of adverse shocks Figure 1.1. GDP growth has been moderate with","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p23.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p23.pdf"},"request":{"example_id":"chart/ac8b3538-en_p23","product_type":"parse","source_file_path":"ac8b3538-en_p23.pdf"},"started_at":"2026-11-04T19:46:39.592978"} +{"completed_at":"2026-11-04T19:46:42.194109","latency_in_ms":1227,"output":{"example_id":"chart/ac8b3538-en_p232","grounded_pages":[],"layout_pages":[],"markdown":"**230** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\npoint to a globally weighted implicit price on all emissions of more than EUR 600 per tonne of CO2 (Network for Greening the Finanical System, 2023[10]), when using carbon prices as a proxy for all climate policies. Some studies warn that even carbon prices of this order of magnitude will not suffice to meet net-zero emission targets, without accompanying policies to markedly raise the responsiveness of emissions to carbon pricing, such as regulations concerning certain uses of fossil fuels or support for clean technology (D’Arcangelo et al., 2022[11]). Yet, in 72 OECD and non-OECD countries that together account for 80% of global greenhouse gas (GHG) emissions, less than half of all emissions were covered by some form of carbon pricing measure ⁴ in 2021 (OECD, 2023[8]), and prevailing price levels are mostly much too low to incentivise deep emission cuts. In OECD countries, only a small share of emissions is priced at or above EUR 60 per tonne of CO2. Globally, sizeable fossil-fuel subsidies reduce effective carbon prices, sometimes turning them negative in countries with little or no carbon pricing measures in place (OECD, 2022[12]).\n\n### **Figure 5.1. Support for climate policies hinges on perceived gains and losses**\n\nCorrelation between beliefs (as listed) and support for a carbon tax package with cash transfers\n\nNote: Results of regressions of support on standardised variables measuring respondents’ beliefs and perceptions. Dark green indicates variables measuring expected gains and losses. Country fixed effects, treatment indicators, and individual socio-economic characteristics are included but not displayed. The dependent variable is an indicator variable equal to 1 if the respondent (somewhat or strongly) supports each of the main climate policies. N=40 680, R2=0.378. CC: climate change. Source: Adapted from Dechezleprêtre, A. et al. (2022[7]), “Fighting climate change: International attitudes toward climate policies” _,_ https://doi.org/10.1787/3406f29a-en.\n\nStatLink 2 https://stat.link/fknrvx\n\n∗∗∗\n\n∗∗∗\n\n∗∗∗\n\n\\*\\*\n\n​\n\n​\n\n\\*\\*\n\n∗∗∗\n\n∗∗∗\n\n∗∗∗\n\n∗∗∗\n\n∗∗∗\n\n∗∗∗\n\n∗∗∗\n\n∗∗∗\n\n\\-15 -10 -5 0 5 10 15\n\nOwn household would lose\n\nLow-income earners will lose\n\nUnderstands impacts of CC\n\nNet-zero is technically feasible\n\nWorries about the consequences of CC\n\nUnderstands emission across activities/regions\n\nCC is real and caused by human\n\nWill suffer from climate change\n\nHigh-income earners will lose\n\nKnows which gases cause CC\n\nPolicies entail positive economic effects\n\nPolicies would reduce pollution\n\nTrusts the governement\n\nInequality is an important problem\n\nPolicies would reduce emissions\n\nCorrelation (%)","pages":[{"line_numbers":[],"markdown":"**230** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\npoint to a globally weighted implicit price on all emissions of more than EUR 600 per tonne of CO2 (Network for Greening the Finanical System, 2023[10]), when using carbon prices as a proxy for all climate policies. Some studies warn that even carbon prices of this order of magnitude will not suffice to meet net-zero emission targets, without accompanying policies to markedly raise the responsiveness of emissions to carbon pricing, such as regulations concerning certain uses of fossil fuels or support for clean technology (D’Arcangelo et al., 2022[11]). Yet, in 72 OECD and non-OECD countries that together account for 80% of global greenhouse gas (GHG) emissions, less than half of all emissions were covered by some form of carbon pricing measure ⁴ in 2021 (OECD, 2023[8]), and prevailing price levels are mostly much too low to incentivise deep emission cuts. In OECD countries, only a small share of emissions is priced at or above EUR 60 per tonne of CO2. Globally, sizeable fossil-fuel subsidies reduce effective carbon prices, sometimes turning them negative in countries with little or no carbon pricing measures in place (OECD, 2022[12]).\n\n### **Figure 5.1. Support for climate policies hinges on perceived gains and losses**\n\nCorrelation between beliefs (as listed) and support for a carbon tax package with cash transfers\n\nNote: Results of regressions of support on standardised variables measuring respondents’ beliefs and perceptions. Dark green indicates variables measuring expected gains and losses. Country fixed effects, treatment indicators, and individual socio-economic characteristics are included but not displayed. The dependent variable is an indicator variable equal to 1 if the respondent (somewhat or strongly) supports each of the main climate policies. N=40 680, R2=0.378. CC: climate change. Source: Adapted from Dechezleprêtre, A. et al. (2022[7]), “Fighting climate change: International attitudes toward climate policies” _,_ https://doi.org/10.1787/3406f29a-en.\n\nStatLink 2 https://stat.link/fknrvx\n\n∗∗∗\n\n∗∗∗\n\n∗∗∗\n\n\\*\\*\n\n​\n\n​\n\n\\*\\*\n\n∗∗∗\n\n∗∗∗\n\n∗∗∗\n\n∗∗∗\n\n∗∗∗\n\n∗∗∗\n\n∗∗∗\n\n∗∗∗\n\n\\-15 -10 -5 0 5 10 15\n\nOwn household would lose\n\nLow-income earners will lose\n\nUnderstands impacts of CC\n\nNet-zero is technically feasible\n\nWorries about the consequences of CC\n\nUnderstands emission across activities/regions\n\nCC is real and caused by human\n\nWill suffer from climate change\n\nHigh-income earners will lose\n\nKnows which gases cause CC\n\nPolicies entail positive economic effects\n\nPolicies would reduce pollution\n\nTrusts the governement\n\nInequality is an important problem\n\nPolicies would reduce emissions\n\nCorrelation (%)","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**230** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\npoint to a globally weighted implicit price on all emissions of more than EUR 600 per tonne of CO2 (Network for Greening the Finanical System, 2023[10]), when using carbon prices as a proxy for all climate policies. Some studies warn that even carbon prices of this order of magnitude will not suffice to meet net-zero emission targets, without accompanying policies to markedly raise the responsiveness of emissions to carbon pricing, such as regulations concerning certain uses of fossil fuels or support for clean technology (D’Arcangelo et al., 2022[11]). Yet, in 72 OECD and non-OECD countries that together account for 80% of global greenhouse gas (GHG) emissions, less than half of all emissions were covered by some form of carbon pricing measure ⁴ in 2021 (OECD, 2023[8]), and prevailing price levels are mostly much too low to incentivise deep emission cuts. In OECD countries, only a small share of emissions is priced at or above EUR 60 per tonne of CO2. Globally, sizeable fossil-fuel subsidies reduce effective carbon prices, sometimes turning them negative in countries with little or no carbon pricing measures in place (OECD, 2022[12]).\n\n### **Figure 5.1. Support for climate policies hinges on perceived gains and losses**\n\nCorrelation between beliefs (as listed) and support for a carbon tax package with cash transfers\n\nNote: Results of regressions of support on standardised variables measuring respondents’ beliefs and perceptions. Dark green indicates variables measuring expected gains and losses. Country fixed effects, treatment indicators, and individual socio-economic characteristics are included but not displayed. The dependent variable is an indicator variable equal to 1 if the respondent (somewhat or strongly) supports each of the main climate policies. N=40 680, R2=0.378. CC: climate change. Source: Adapted from Dechezleprêtre, A. et al. (2022[7]), “Fighting climate change: International attitudes toward climate policies” _,_ https://doi.org/10.1787/3406f29a-en.\n\nStatLink 2 https://stat.link/fknrvx\n\n∗∗∗\n\n∗∗∗\n\n∗∗∗\n\n\\*\\*\n\n​\n\n​\n\n\\*\\*\n\n∗∗∗\n\n∗∗∗\n\n∗∗∗\n\n∗∗∗\n\n∗∗∗\n\n∗∗∗\n\n∗∗∗\n\n∗∗∗\n\n\\-15 -10 -5 0 5 10 15\n\nOwn household would lose\n\nLow-income earners will lose\n\nUnderstands impacts of CC\n\nNet-zero is technically feasible\n\nWorries about the consequences of CC\n\nUnderstands emission across activities/regions\n\nCC is real and caused by human\n\nWill suffer from climate change\n\nHigh-income earners will lose\n\nKnows which gases cause CC\n\nPolicies entail positive economic effects\n\nPolicies would reduce pollution\n\nTrusts the governement\n\nInequality is an important problem\n\nPolicies would reduce emissions\n\nCorrelation (%)","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p232.pdf"]},"response":{"errors":[],"extract_id":"extract_977d345759f59239365661ef90b13fb4","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**230** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\npoint to a globally weighted implicit price on all emissions of more than EUR 600 per tonne of CO2 (Network for Greening the Finanical System, 2023[10]), when using carbon prices as a proxy for all climate policies. Some studies warn that even carbon prices of this order of magnitude will not suffice to meet net-zero emission targets, without accompanying policies to markedly raise the responsiveness of emissions to carbon pricing, such as regulations concerning certain uses of fossil fuels or support for clean technology (D’Arcangelo et al., 2022[11]). Yet, in 72 OECD and non-OECD countries that together account for 80% of global greenhouse gas (GHG) emissions, less than half of all emissions were covered by some form of carbon pricing measure ⁴ in 2021 (OECD, 2023[8]), and prevailing price levels are mostly much too low to incentivise deep emission cuts. In OECD countries, only a small share of emissions is priced at or above EUR 60 per tonne of CO2. Globally, sizeable fossil-fuel subsidies reduce effective carbon prices, sometimes turning them negative in countries with little or no carbon pricing measures in place (OECD, 2022[12]).\n\n### **Figure 5.1. Support for climate policies hinges on perceived gains and losses**\n\nCorrelation between beliefs (as listed) and support for a carbon tax package with cash transfers\n\nNote: Results of regressions of support on standardised variables measuring respondents’ beliefs and perceptions. Dark green indicates variables measuring expected gains and losses. Country fixed effects, treatment indicators, and individual socio-economic characteristics are included but not displayed. The dependent variable is an indicator variable equal to 1 if the respondent (somewhat or strongly) supports each of the main climate policies. N=40 680, R2=0.378. CC: climate change. Source: Adapted from Dechezleprêtre, A. et al. (2022[7]), “Fighting climate change: International attitudes toward climate policies” _,_ https://doi.org/10.1787/3406f29a-en.\n\nStatLink 2 https://stat.link/fknrvx\n\n∗∗∗\n\n∗∗∗\n\n∗∗∗\n\n\\*\\*\n\n​\n\n​\n\n\\*\\*\n\n∗∗∗\n\n∗∗∗\n\n∗∗∗\n\n∗∗∗\n\n∗∗∗\n\n∗∗∗\n\n∗∗���\n\n∗∗∗\n\n\\-15 -10 -5 0 5 10 15\n\nOwn household would lose\n\nLow-income earners will lose\n\nUnderstands impacts of CC\n\nNet-zero is technically feasible\n\nWorries about the consequences of CC\n\nUnderstands emission across activities/regions\n\nCC is real and caused by human\n\nWill suffer from climate change\n\nHigh-income earners will lose\n\nKnows which gases cause CC\n\nPolicies entail positive economic effects\n\nPolicies would reduce pollution\n\nTrusts the governement\n\nInequality is an important problem\n\nPolicies would reduce emissions\n\nCorrelation (%)"],"full_content":"**230** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\npoint to a globally weighted implicit price on all emissions of more than EUR 600 per tonne of CO2 (Network for Greening the Finanical System, 2023[10]), when using carbon prices as a proxy for all climate policies. Some studies warn that even carbon prices of this order of magnitude will not suffice to meet net-zero emission targets, without accompanying policies to markedly raise the responsiveness of emissions to carbon pricing, such as regulations concerning certain uses of fossil fuels or support for clean technology (D’Arcangelo et al., 2022[11]). Yet, in 72 OECD and non-OECD countries that together account for 80% of global greenhouse gas (GHG) emissions, less than half of all emissions were covered by some form of carbon pricing measure ⁴ in 2021 (OECD, 2023[8]), and prevailing price levels are mostly much too low to incentivise deep emission cuts. In OECD countries, only a small share of emissions is priced at or above EUR 60 per tonne of CO2. Globally, sizeable fossil-fuel subsidies reduce effective carbon prices, sometimes turning them negative in countries with little or no carbon pricing measures in place (OECD, 2022[12]).\n\n### **Figure 5.1. Support for climate policies hinges on perceived gains and losses**\n\nCorrelation between beliefs (as listed) and support for a carbon tax package with cash transfers\n\nNote: Results of regressions of support on standardised variables measuring respondents’ beliefs and perceptions. Dark green indicates variables measuring expected gains and losses. Country fixed effects, treatment indicators, and individual socio-economic characteristics are included but not displayed. The dependent variable is an indicator variable equal to 1 if the respondent (somewhat or strongly) supports each of the main climate policies. N=40 680, R2=0.378. CC: climate change. Source: Adapted from Dechezleprêtre, A. et al. (2022[7]), “Fighting climate change: International attitudes toward climate policies” _,_ https://doi.org/10.1787/3406f29a-en.\n\nStatLink 2 https://stat.link/fknrvx\n\n∗∗∗\n\n∗∗∗\n\n∗∗∗\n\n\\*\\*\n\n​\n\n​\n\n\\*\\*\n\n∗∗∗\n\n∗∗∗\n\n∗∗∗\n\n∗∗∗\n\n∗∗∗\n\n∗∗∗\n\n∗∗∗\n\n∗∗∗\n\n\\-15 -10 -5 0 5 10 15\n\nOwn household would lose\n\nLow-income earners will lose\n\nUnderstands impacts of CC\n\nNet-zero is technically feasible\n\nWorries about the consequences of CC\n\nUnderstands emission across activities/regions\n\nCC is real and caused by human\n\nWill suffer from climate change\n\nHigh-income earners will lose\n\nKnows which gases cause CC\n\nPolicies entail positive economic effects\n\nPolicies would reduce pollution\n\nTrusts the governement\n\nInequality is an important problem\n\nPolicies would reduce emissions\n\nCorrelation (%)","publish_date":null,"title":"Figure 5.1. Support for climate policies hinges on perceived gains and losses","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p232.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p232.pdf"},"request":{"example_id":"chart/ac8b3538-en_p232","product_type":"parse","source_file_path":"ac8b3538-en_p232.pdf"},"started_at":"2026-11-04T19:46:40.966115"} +{"completed_at":"2026-11-04T19:46:45.390608","latency_in_ms":3195,"output":{"example_id":"chart/ac8b3538-en_p239","grounded_pages":[],"layout_pages":[],"markdown":" **237**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Figure 5.2. Evolution of effective carbon rates in 34 OECD countries, 2012-21**\n\nIn 2021 EUR per tonne of CO2, levels for 2012 and 2021\n\nNote: For each country, the two bars refer to the two years (2012, 2021). Highlighted countries are those included in the empirical analysis of this chapter. The chart presents emissions-weighted average Effective Carbon Rates (ECR) over the whole economy, by pricing instrument, presented in 2021 constant EUR. Effective Carbon Rates are the sum of carbon taxes, permit prices from emissions trading systems (ETS) and fuel excise taxes. The emissions-weighted average effective carbon rates presented here are computed excluding emissions from the combustion of biomass. Reading: In 2012, the average ECR in France was of EUR 74.5 per tonne of CO2, with carbon pricing mostly resulting from fuel excise taxes (with an average of EUR 72 per tonne of CO2 across the economy). In 2021, the average ECR in France was of EUR 110 per tonne of CO2, with carbon taxes and ETS prices increasing the most since 2012: the French economy-wide average carbon price resulting from the EU ETS was of EUR 13/tCO2 and from carbon taxes of EUR 23.6/tCO2. Source: OECD Effective Carbon Rates database.\n\nStatLink 2 https://stat.link/8veb70\n\nIn parallel, per-capita carbon emissions from energy use declined in most OECD countries (Figure 5.3). By contrast, in spite of lower average emissions than OECD countries in 2012, non-OECD G20 countries have not all followed the decreasing trend in the OECD area. In particular, China, India and Indonesia have seen a rise in per capita emissions. In India and Indonesia, they nevertheless remained lower than in most OECD countries in 2021. These differences arise due to many factors, including the different levels of developments between most OECD and non-OECD G20 countries. Less developed countries may rely more on energy-intensive resources to foster their development.\n\n0\n\n30\n\n60\n\n90\n\n120\n\n150\n\nEUR per tonne of CO₂\n\nETS (2012, 2021) Carbon taxes (2012, 2021) Fuel excise taxes (2012, 2021)","pages":[{"line_numbers":[],"markdown":" **237**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Figure 5.2. Evolution of effective carbon rates in 34 OECD countries, 2012-21**\n\nIn 2021 EUR per tonne of CO2, levels for 2012 and 2021\n\nNote: For each country, the two bars refer to the two years (2012, 2021). Highlighted countries are those included in the empirical analysis of this chapter. The chart presents emissions-weighted average Effective Carbon Rates (ECR) over the whole economy, by pricing instrument, presented in 2021 constant EUR. Effective Carbon Rates are the sum of carbon taxes, permit prices from emissions trading systems (ETS) and fuel excise taxes. The emissions-weighted average effective carbon rates presented here are computed excluding emissions from the combustion of biomass. Reading: In 2012, the average ECR in France was of EUR 74.5 per tonne of CO2, with carbon pricing mostly resulting from fuel excise taxes (with an average of EUR 72 per tonne of CO2 across the economy). In 2021, the average ECR in France was of EUR 110 per tonne of CO2, with carbon taxes and ETS prices increasing the most since 2012: the French economy-wide average carbon price resulting from the EU ETS was of EUR 13/tCO2 and from carbon taxes of EUR 23.6/tCO2. Source: OECD Effective Carbon Rates database.\n\nStatLink 2 https://stat.link/8veb70\n\nIn parallel, per-capita carbon emissions from energy use declined in most OECD countries (Figure 5.3). By contrast, in spite of lower average emissions than OECD countries in 2012, non-OECD G20 countries have not all followed the decreasing trend in the OECD area. In particular, China, India and Indonesia have seen a rise in per capita emissions. In India and Indonesia, they nevertheless remained lower than in most OECD countries in 2021. These differences arise due to many factors, including the different levels of developments between most OECD and non-OECD G20 countries. Less developed countries may rely more on energy-intensive resources to foster their development.\n\n0\n\n30\n\n60\n\n90\n\n120\n\n150\n\nEUR per tonne of CO₂\n\nETS (2012, 2021) Carbon taxes (2012, 2021) Fuel excise taxes (2012, 2021)","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":" **237**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Figure 5.2. Evolution of effective carbon rates in 34 OECD countries, 2012-21**\n\nIn 2021 EUR per tonne of CO2, levels for 2012 and 2021\n\nNote: For each country, the two bars refer to the two years (2012, 2021). Highlighted countries are those included in the empirical analysis of this chapter. The chart presents emissions-weighted average Effective Carbon Rates (ECR) over the whole economy, by pricing instrument, presented in 2021 constant EUR. Effective Carbon Rates are the sum of carbon taxes, permit prices from emissions trading systems (ETS) and fuel excise taxes. The emissions-weighted average effective carbon rates presented here are computed excluding emissions from the combustion of biomass. Reading: In 2012, the average ECR in France was of EUR 74.5 per tonne of CO2, with carbon pricing mostly resulting from fuel excise taxes (with an average of EUR 72 per tonne of CO2 across the economy). In 2021, the average ECR in France was of EUR 110 per tonne of CO2, with carbon taxes and ETS prices increasing the most since 2012: the French economy-wide average carbon price resulting from the EU ETS was of EUR 13/tCO2 and from carbon taxes of EUR 23.6/tCO2. Source: OECD Effective Carbon Rates database.\n\nStatLink 2 https://stat.link/8veb70\n\nIn parallel, per-capita carbon emissions from energy use declined in most OECD countries (Figure 5.3). By contrast, in spite of lower average emissions than OECD countries in 2012, non-OECD G20 countries have not all followed the decreasing trend in the OECD area. In particular, China, India and Indonesia have seen a rise in per capita emissions. In India and Indonesia, they nevertheless remained lower than in most OECD countries in 2021. These differences arise due to many factors, including the different levels of developments between most OECD and non-OECD G20 countries. Less developed countries may rely more on energy-intensive resources to foster their development.\n\n0\n\n30\n\n60\n\n90\n\n120\n\n150\n\nEUR per tonne of CO₂\n\nETS (2012, 2021) Carbon taxes (2012, 2021) Fuel excise taxes (2012, 2021)","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p239.pdf"]},"response":{"errors":[],"extract_id":"extract_1dc9e7bea537904270c6f4521c047dd1","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":[" **237**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Figure 5.2. Evolution of effective carbon rates in 34 OECD countries, 2012-21**\n\nIn 2021 EUR per tonne of CO2, levels for 2012 and 2021\n\nNote: For each country, the two bars refer to the two years (2012, 2021). Highlighted countries are those included in the empirical analysis of this chapter. The chart presents emissions-weighted average Effective Carbon Rates (ECR) over the whole economy, by pricing instrument, presented in 2021 constant EUR. Effective Carbon Rates are the sum of carbon taxes, permit prices from emissions trading systems (ETS) and fuel excise taxes. The emissions-weighted average effective carbon rates presented here are computed excluding emissions from the combustion of biomass. Reading: In 2012, the average ECR in France was of EUR 74.5 per tonne of CO2, with carbon pricing mostly resulting from fuel excise taxes (with an average of EUR 72 per tonne of CO2 across the economy). In 2021, the average ECR in France was of EUR 110 per tonne of CO2, with carbon taxes and ETS prices increasing the most since 2012: the French economy-wide average carbon price resulting from the EU ETS was of EUR 13/tCO2 and from carbon taxes of EUR 23.6/tCO2. Source: OECD Effective Carbon Rates database.\n\nStatLink 2 https://stat.link/8veb70\n\nIn parallel, per-capita carbon emissions from energy use declined in most OECD countries (Figure 5.3). By contrast, in spite of lower average emissions than OECD countries in 2012, non-OECD G20 countries have not all followed the decreasing trend in the OECD area. In particular, China, India and Indonesia have seen a rise in per capita emissions. In India and Indonesia, they nevertheless remained lower than in most OECD countries in 2021. These differences arise due to many factors, including the different levels of developments between most OECD and non-OECD G20 countries. Less developed countries may rely more on energy-intensive resources to foster their development.\n\n0\n\n30\n\n60\n\n90\n\n120\n\n150\n\nEUR per tonne of CO₂\n\nETS (2012, 2021) Carbon taxes (2012, 2021) Fuel excise taxes (2012, 2021)"],"full_content":" **237**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Figure 5.2. Evolution of effective carbon rates in 34 OECD countries, 2012-21**\n\nIn 2021 EUR per tonne of CO2, levels for 2012 and 2021\n\nNote: For each country, the two bars refer to the two years (2012, 2021). Highlighted countries are those included in the empirical analysis of this chapter. The chart presents emissions-weighted average Effective Carbon Rates (ECR) over the whole economy, by pricing instrument, presented in 2021 constant EUR. Effective Carbon Rates are the sum of carbon taxes, permit prices from emissions trading systems (ETS) and fuel excise taxes. The emissions-weighted average effective carbon rates presented here are computed excluding emissions from the combustion of biomass. Reading: In 2012, the average ECR in France was of EUR 74.5 per tonne of CO2, with carbon pricing mostly resulting from fuel excise taxes (with an average of EUR 72 per tonne of CO2 across the economy). In 2021, the average ECR in France was of EUR 110 per tonne of CO2, with carbon taxes and ETS prices increasing the most since 2012: the French economy-wide average carbon price resulting from the EU ETS was of EUR 13/tCO2 and from carbon taxes of EUR 23.6/tCO2. Source: OECD Effective Carbon Rates database.\n\nStatLink 2 https://stat.link/8veb70\n\nIn parallel, per-capita carbon emissions from energy use declined in most OECD countries (Figure 5.3). By contrast, in spite of lower average emissions than OECD countries in 2012, non-OECD G20 countries have not all followed the decreasing trend in the OECD area. In particular, China, India and Indonesia have seen a rise in per capita emissions. In India and Indonesia, they nevertheless remained lower than in most OECD countries in 2021. These differences arise due to many factors, including the different levels of developments between most OECD and non-OECD G20 countries. Less developed countries may rely more on energy-intensive resources to foster their development.\n\n0\n\n30\n\n60\n\n90\n\n120\n\n150\n\nEUR per tonne of CO₂\n\nETS (2012, 2021) Carbon taxes (2012, 2021) Fuel excise taxes (2012, 2021)","publish_date":null,"title":"Figure 5.2. Evolution of effective carbon rates in 34 OECD countries, 2012-21","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p239.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p239.pdf"},"request":{"example_id":"chart/ac8b3538-en_p239","product_type":"parse","source_file_path":"ac8b3538-en_p239.pdf"},"started_at":"2026-11-04T19:46:42.195584"} +{"completed_at":"2026-11-04T19:45:41.537447","latency_in_ms":2621,"output":{"example_id":"chart/ac8b3538-en_p245","grounded_pages":[],"layout_pages":[],"markdown":" **243**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n**_5\\.3.1. Energy spending is one key driver of household emissions_**\n\nCarbon prices affect household budgets both directly through households’ own fuel consumption, and indirectly via the consumption of other goods and services that give rise to CO2 emissions during the production process. The direct effect is shaped by the pattern of expenditures on heating and transport (fossil-)fuel (Figure 5.4). The figure also shows spending on electricity, which is a derived good, whose production can be fuel intensive. ¹⁶ Energy-related emissions embodied in the full set of derived goods, including, e.g. food and public transport, form one part of the overall carbon footprints presented in the next sub-section.\n\nPoor households save less than the rich, or they dissave, and total consumption therefore accounts for larger portions of their income than for better-off households. In four of the five countries included in Figure 5.4, such a regressive spending pattern holds also for energy consumption. This is particularly visible in Poland and Türkiye, where low-income households spent more than one fifth of their incomes on energy. ¹⁷ Spending on electricity and heating is also regressive in the other European countries. This makes them particularly vulnerable to energy poverty. ¹⁸ But the share of resources devoted to some categories of energy can in fact increase with income. Spending shares for motor fuel are increasing with income in Mexico and Poland, reflecting both income inequality and unequal car ownership. In Germany, motor fuel spending is mostly flat. Mexico is the only country where richer households devote bigger portions of their income to energy overall, confirming that energy can be a luxury item in middle-income countries. ¹⁹ Average spending shares also vary strongly across countries, with plausible drivers including average incomes, climatic factors, as well as energy taxation and subsidies.\n\n### **Figure 5.4. Poorer households typically spend large parts of their income on energy, but spending is** **“top** **\\-heavy** **” in some middle** **\\-income countries**\n\nHousehold expenditures on fuel and other energy, as a percentage of disposable income, by income decile\n\nNote: Groups 1-10 refer to income deciles (equivalised disposable household income). Heating fuel includes expenditure on gas (natural gas and town gas), liquified hydrocarbons, kerosene and other liquid fuels, coal and other solid fuels. Motor fuel includes expenditure on diesel and petrol for transportation. Source: OECD calculations using household budget surveys (2015 for EU countries, 2016 for Mexico, 2019 for Türkiye) and WIOD input-output data (for electricity).\n\nStatLink 2 https://stat.link/h71u94\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10\n\nGermany France Mexico Poland Türkiye\n\n%\n\nIncome decile\n\nHeating fuels Motor fuels Electricity","pages":[{"line_numbers":[],"markdown":" **243**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n**_5\\.3.1. Energy spending is one key driver of household emissions_**\n\nCarbon prices affect household budgets both directly through households’ own fuel consumption, and indirectly via the consumption of other goods and services that give rise to CO2 emissions during the production process. The direct effect is shaped by the pattern of expenditures on heating and transport (fossil-)fuel (Figure 5.4). The figure also shows spending on electricity, which is a derived good, whose production can be fuel intensive. ¹⁶ Energy-related emissions embodied in the full set of derived goods, including, e.g. food and public transport, form one part of the overall carbon footprints presented in the next sub-section.\n\nPoor households save less than the rich, or they dissave, and total consumption therefore accounts for larger portions of their income than for better-off households. In four of the five countries included in Figure 5.4, such a regressive spending pattern holds also for energy consumption. This is particularly visible in Poland and Türkiye, where low-income households spent more than one fifth of their incomes on energy. ¹⁷ Spending on electricity and heating is also regressive in the other European countries. This makes them particularly vulnerable to energy poverty. ¹⁸ But the share of resources devoted to some categories of energy can in fact increase with income. Spending shares for motor fuel are increasing with income in Mexico and Poland, reflecting both income inequality and unequal car ownership. In Germany, motor fuel spending is mostly flat. Mexico is the only country where richer households devote bigger portions of their income to energy overall, confirming that energy can be a luxury item in middle-income countries. ¹⁹ Average spending shares also vary strongly across countries, with plausible drivers including average incomes, climatic factors, as well as energy taxation and subsidies.\n\n### **Figure 5.4. Poorer households typically spend large parts of their income on energy, but spending is** **“top** **\\-heavy** **” in some middle** **\\-income countries**\n\nHousehold expenditures on fuel and other energy, as a percentage of disposable income, by income decile\n\nNote: Groups 1-10 refer to income deciles (equivalised disposable household income). Heating fuel includes expenditure on gas (natural gas and town gas), liquified hydrocarbons, kerosene and other liquid fuels, coal and other solid fuels. Motor fuel includes expenditure on diesel and petrol for transportation. Source: OECD calculations using household budget surveys (2015 for EU countries, 2016 for Mexico, 2019 for Türkiye) and WIOD input-output data (for electricity).\n\nStatLink 2 https://stat.link/h71u94\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10\n\nGermany France Mexico Poland Türkiye\n\n%\n\nIncome decile\n\nHeating fuels Motor fuels Electricity","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":" **243**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n**_5\\.3.1. Energy spending is one key driver of household emissions_**\n\nCarbon prices affect household budgets both directly through households’ own fuel consumption, and indirectly via the consumption of other goods and services that give rise to CO2 emissions during the production process. The direct effect is shaped by the pattern of expenditures on heating and transport (fossil-)fuel (Figure 5.4). The figure also shows spending on electricity, which is a derived good, whose production can be fuel intensive. ¹⁶ Energy-related emissions embodied in the full set of derived goods, including, e.g. food and public transport, form one part of the overall carbon footprints presented in the next sub-section.\n\nPoor households save less than the rich, or they dissave, and total consumption therefore accounts for larger portions of their income than for better-off households. In four of the five countries included in Figure 5.4, such a regressive spending pattern holds also for energy consumption. This is particularly visible in Poland and Türkiye, where low-income households spent more than one fifth of their incomes on energy. ¹⁷ Spending on electricity and heating is also regressive in the other European countries. This makes them particularly vulnerable to energy poverty. ¹⁸ But the share of resources devoted to some categories of energy can in fact increase with income. Spending shares for motor fuel are increasing with income in Mexico and Poland, reflecting both income inequality and unequal car ownership. In Germany, motor fuel spending is mostly flat. Mexico is the only country where richer households devote bigger portions of their income to energy overall, confirming that energy can be a luxury item in middle-income countries. ¹⁹ Average spending shares also vary strongly across countries, with plausible drivers including average incomes, climatic factors, as well as energy taxation and subsidies.\n\n### **Figure 5.4. Poorer households typically spend large parts of their income on energy, but spending is** **“top** **\\-heavy** **” in some middle** **\\-income countries**\n\nHousehold expenditures on fuel and other energy, as a percentage of disposable income, by income decile\n\nNote: Groups 1-10 refer to income deciles (equivalised disposable household income). Heating fuel includes expenditure on gas (natural gas and town gas), liquified hydrocarbons, kerosene and other liquid fuels, coal and other solid fuels. Motor fuel includes expenditure on diesel and petrol for transportation. Source: OECD calculations using household budget surveys (2015 for EU countries, 2016 for Mexico, 2019 for Türkiye) and WIOD input-output data (for electricity).\n\nStatLink 2 https://stat.link/h71u94\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10\n\nGermany France Mexico Poland Türkiye\n\n%\n\nIncome decile\n\nHeating fuels Motor fuels Electricity","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p245.pdf"]},"response":{"errors":[],"extract_id":"extract_42e05e42867819fbcce512f21b35afd4","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":[" **243**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n**_5\\.3.1. Energy spending is one key driver of household emissions_**\n\nCarbon prices affect household budgets both directly through households’ own fuel consumption, and indirectly via the consumption of other goods and services that give rise to CO2 emissions during the production process. The direct effect is shaped by the pattern of expenditures on heating and transport (fossil-)fuel (Figure 5.4). The figure also shows spending on electricity, which is a derived good, whose production can be fuel intensive. ¹⁶ Energy-related emissions embodied in the full set of derived goods, including, e.g. food and public transport, form one part of the overall carbon footprints presented in the next sub-section.\n\nPoor households save less than the rich, or they dissave, and total consumption therefore accounts for larger portions of their income than for better-off households. In four of the five countries included in Figure 5.4, such a regressive spending pattern holds also for energy consumption. This is particularly visible in Poland and Türkiye, where low-income households spent more than one fifth of their incomes on energy. ¹⁷ Spending on electricity and heating is also regressive in the other European countries. This makes them particularly vulnerable to energy poverty. ¹⁸ But the share of resources devoted to some categories of energy can in fact increase with income. Spending shares for motor fuel are increasing with income in Mexico and Poland, reflecting both income inequality and unequal car ownership. In Germany, motor fuel spending is mostly flat. Mexico is the only country where richer households devote bigger portions of their income to energy overall, confirming that energy can be a luxury item in middle-income countries. ¹⁹ Average spending shares also vary strongly across countries, with plausible drivers including average incomes, climatic factors, as well as energy taxation and subsidies.\n\n### **Figure 5.4. Poorer households typically spend large parts of their income on energy, but spending is** **“top** **\\-heavy** **” in some middle** **\\-income countries**\n\nHousehold expenditures on fuel and other energy, as a percentage of disposable income, by income decile\n\nNote: Groups 1-10 refer to income deciles (equivalised disposable household income). Heating fuel includes expenditure on gas (natural gas and town gas), liquified hydrocarbons, kerosene and other liquid fuels, coal and other solid fuels. Motor fuel includes expenditure on diesel and petrol for transportation. Source: OECD calculations using household budget surveys (2015 for EU countries, 2016 for Mexico, 2019 for Türkiye) and WIOD input-output data (for electricity).\n\nStatLink 2 https://stat.link/h71u94\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10\n\nGermany France Mexico Poland Türkiye\n\n%\n\nIncome decile\n\nHeating fuels Motor fuels Electricity"],"full_content":" **243**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n**_5\\.3.1. Energy spending is one key driver of household emissions_**\n\nCarbon prices affect household budgets both directly through households’ own fuel consumption, and indirectly via the consumption of other goods and services that give rise to CO2 emissions during the production process. The direct effect is shaped by the pattern of expenditures on heating and transport (fossil-)fuel (Figure 5.4). The figure also shows spending on electricity, which is a derived good, whose production can be fuel intensive. ¹⁶ Energy-related emissions embodied in the full set of derived goods, including, e.g. food and public transport, form one part of the overall carbon footprints presented in the next sub-section.\n\nPoor households save less than the rich, or they dissave, and total consumption therefore accounts for larger portions of their income than for better-off households. In four of the five countries included in Figure 5.4, such a regressive spending pattern holds also for energy consumption. This is particularly visible in Poland and Türkiye, where low-income households spent more than one fifth of their incomes on energy. ¹⁷ Spending on electricity and heating is also regressive in the other European countries. This makes them particularly vulnerable to energy poverty. ¹⁸ But the share of resources devoted to some categories of energy can in fact increase with income. Spending shares for motor fuel are increasing with income in Mexico and Poland, reflecting both income inequality and unequal car ownership. In Germany, motor fuel spending is mostly flat. Mexico is the only country where richer households devote bigger portions of their income to energy overall, confirming that energy can be a luxury item in middle-income countries. ¹⁹ Average spending shares also vary strongly across countries, with plausible drivers including average incomes, climatic factors, as well as energy taxation and subsidies.\n\n### **Figure 5.4. Poorer households typically spend large parts of their income on energy, but spending is** **“top** **\\-heavy** **” in some middle** **\\-income countries**\n\nHousehold expenditures on fuel and other energy, as a percentage of disposable income, by income decile\n\nNote: Groups 1-10 refer to income deciles (equivalised disposable household income). Heating fuel includes expenditure on gas (natural gas and town gas), liquified hydrocarbons, kerosene and other liquid fuels, coal and other solid fuels. Motor fuel includes expenditure on diesel and petrol for transportation. Source: OECD calculations using household budget surveys (2015 for EU countries, 2016 for Mexico, 2019 for Türkiye) and WIOD input-output data (for electricity).\n\nStatLink 2 https://stat.link/h71u94\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10\n\nGermany France Mexico Poland Türkiye\n\n%\n\nIncome decile\n\nHeating fuels Motor fuels Electricity","publish_date":null,"title":"Figure 5.4. Poorer households typically spend large parts of their income on energy, but spending is “top-heavy” in some middle-income","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p245.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p245.pdf"},"request":{"example_id":"chart/ac8b3538-en_p245","product_type":"parse","source_file_path":"ac8b3538-en_p245.pdf"},"started_at":"2026-11-04T19:45:38.916077"} +{"completed_at":"2026-11-04T19:45:43.520840","latency_in_ms":2923,"output":{"example_id":"chart/ac8b3538-en_p246","grounded_pages":[],"layout_pages":[],"markdown":"**244** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n**_5\\.3.2. Who pollutes the most? Carbon footprints linked to household consumption_**\n\nEnergy consumption is a key driver of emissions, but it is not the only one. There are several reasons why energy spending is only a partial indicator of households’ carbon footprints. First, emissions from electricity generation vary enormously between countries, by a factor of 15 across the five countries considered here. ²⁰ Relatedly, for each of the main fuel consumption categories, emissions vary by type of fuel (see Annex Figure 5.B.1), as do prices before accounting for carbon pricing. For lower-income or rural households, domestic fuels can include high shares of solid fuels (coal, coke, peat, firewood), which have much higher emission factors than liquid fuel. Emission factors are lower for natural gas, an energy source that can be more common in urban areas. Motor fuels are commonly more expensive than domestic fuels and produce fewer emissions per energy unit than firewood, coal, or heating oil. Per unit of fuel expenditure, emissions – and therefore also the effect of a given carbon price – thus tend to be higher for domestic fuels than for motor fuels.\n\nSecond, non-fuel expenditures account for a large share of household spending. ²¹ Per unit of spending, fuel use creates more emissions than other consumption, but due to the size of non-fuel spending, the production of non-fuel goods and services is a significant driver of carbon footprints. Across the five countries, direct emissions from households’ own use of fossil fuels mostly account for around half of the total emissions linked to consumption (Figure 5.5). These estimates incorporate all household consumption, following the “consumer responsibility” principle (see figure note). Fuel spending was a lesser driver of total emissions in France, partly reflecting the balance of spending on fossil fuels and electricity, but also the types of fuel (coal and other solid fuel, liquid fuel, gas) used by households and in production processes. ²² Emissions linked to imported inputs or final goods (other than fuel) are also significant but comparatively small, accounting for less than 10% of emissions in each of the five countries. (This small share is reassuring in the context of the chapter’s later estimates of carbon price burdens, which do not account for differential carbon price trends in source countries.)\n\n### **Figure 5.5** **. Fuel expenditure is not the only driver of households’ carbon footpr** **ints**\n\nEmissions from fuel (“direct”) and non-fuel (“indirect”) consumption, in percentage of total\n\nNote: “Direct” includes households’ own consumption of fossil fuels, both domestically sourced and imported. “Indirect, domestic” includes households’ own non-fuel, domestically sourced consumption. “Indirect imported” accounts for emissions linked to all other non-domestically sourced inputs and consumption goods. Estimates are based on the “consumer responsibility” principle, accounting for all household consumption. They therefore attribute emissions to the country where the final good is consumed, and this differs from the emissions per capita that are physically released in a given country. It is important to account for the consumption of imported goods in order to determine the carbon price embedded in households’ consumption baskets. Source: OECD calculations using IEA emissions factors for different fuels, World Input-Output Database (WIOD) as well as household budget surveys (2015 for EU countries, 2016 for Mexico, 2019 for Türkiye).\n\nStatLink 2 https://stat.link/hs30kq\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\n70\n\n80\n\n90\n\n100\n\nFrance Germany Poland Mexico Türkiye\n\nShare of total (%)\n\nDirect Indirect, domestic Indirect, imported","pages":[{"line_numbers":[],"markdown":"**244** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n**_5\\.3.2. Who pollutes the most? Carbon footprints linked to household consumption_**\n\nEnergy consumption is a key driver of emissions, but it is not the only one. There are several reasons why energy spending is only a partial indicator of households’ carbon footprints. First, emissions from electricity generation vary enormously between countries, by a factor of 15 across the five countries considered here. ²⁰ Relatedly, for each of the main fuel consumption categories, emissions vary by type of fuel (see Annex Figure 5.B.1), as do prices before accounting for carbon pricing. For lower-income or rural households, domestic fuels can include high shares of solid fuels (coal, coke, peat, firewood), which have much higher emission factors than liquid fuel. Emission factors are lower for natural gas, an energy source that can be more common in urban areas. Motor fuels are commonly more expensive than domestic fuels and produce fewer emissions per energy unit than firewood, coal, or heating oil. Per unit of fuel expenditure, emissions – and therefore also the effect of a given carbon price – thus tend to be higher for domestic fuels than for motor fuels.\n\nSecond, non-fuel expenditures account for a large share of household spending. ²¹ Per unit of spending, fuel use creates more emissions than other consumption, but due to the size of non-fuel spending, the production of non-fuel goods and services is a significant driver of carbon footprints. Across the five countries, direct emissions from households’ own use of fossil fuels mostly account for around half of the total emissions linked to consumption (Figure 5.5). These estimates incorporate all household consumption, following the “consumer responsibility” principle (see figure note). Fuel spending was a lesser driver of total emissions in France, partly reflecting the balance of spending on fossil fuels and electricity, but also the types of fuel (coal and other solid fuel, liquid fuel, gas) used by households and in production processes. ²² Emissions linked to imported inputs or final goods (other than fuel) are also significant but comparatively small, accounting for less than 10% of emissions in each of the five countries. (This small share is reassuring in the context of the chapter’s later estimates of carbon price burdens, which do not account for differential carbon price trends in source countries.)\n\n### **Figure 5.5** **. Fuel expenditure is not the only driver of households’ carbon footpr** **ints**\n\nEmissions from fuel (“direct”) and non-fuel (“indirect”) consumption, in percentage of total\n\nNote: “Direct” includes households’ own consumption of fossil fuels, both domestically sourced and imported. “Indirect, domestic” includes households’ own non-fuel, domestically sourced consumption. “Indirect imported” accounts for emissions linked to all other non-domestically sourced inputs and consumption goods. Estimates are based on the “consumer responsibility” principle, accounting for all household consumption. They therefore attribute emissions to the country where the final good is consumed, and this differs from the emissions per capita that are physically released in a given country. It is important to account for the consumption of imported goods in order to determine the carbon price embedded in households’ consumption baskets. Source: OECD calculations using IEA emissions factors for different fuels, World Input-Output Database (WIOD) as well as household budget surveys (2015 for EU countries, 2016 for Mexico, 2019 for Türkiye).\n\nStatLink 2 https://stat.link/hs30kq\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\n70\n\n80\n\n90\n\n100\n\nFrance Germany Poland Mexico Türkiye\n\nShare of total (%)\n\nDirect Indirect, domestic Indirect, imported","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**244** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n**_5\\.3.2. Who pollutes the most? Carbon footprints linked to household consumption_**\n\nEnergy consumption is a key driver of emissions, but it is not the only one. There are several reasons why energy spending is only a partial indicator of households’ carbon footprints. First, emissions from electricity generation vary enormously between countries, by a factor of 15 across the five countries considered here. ²⁰ Relatedly, for each of the main fuel consumption categories, emissions vary by type of fuel (see Annex Figure 5.B.1), as do prices before accounting for carbon pricing. For lower-income or rural households, domestic fuels can include high shares of solid fuels (coal, coke, peat, firewood), which have much higher emission factors than liquid fuel. Emission factors are lower for natural gas, an energy source that can be more common in urban areas. Motor fuels are commonly more expensive than domestic fuels and produce fewer emissions per energy unit than firewood, coal, or heating oil. Per unit of fuel expenditure, emissions – and therefore also the effect of a given carbon price – thus tend to be higher for domestic fuels than for motor fuels.\n\nSecond, non-fuel expenditures account for a large share of household spending. ²¹ Per unit of spending, fuel use creates more emissions than other consumption, but due to the size of non-fuel spending, the production of non-fuel goods and services is a significant driver of carbon footprints. Across the five countries, direct emissions from households’ own use of fossil fuels mostly account for around half of the total emissions linked to consumption (Figure 5.5). These estimates incorporate all household consumption, following the “consumer responsibility” principle (see figure note). Fuel spending was a lesser driver of total emissions in France, partly reflecting the balance of spending on fossil fuels and electricity, but also the types of fuel (coal and other solid fuel, liquid fuel, gas) used by households and in production processes. ²² Emissions linked to imported inputs or final goods (other than fuel) are also significant but comparatively small, accounting for less than 10% of emissions in each of the five countries. (This small share is reassuring in the context of the chapter’s later estimates of carbon price burdens, which do not account for differential carbon price trends in source countries.)\n\n### **Figure 5.5** **. Fuel expenditure is not the only driver of households’ carbon footpr** **ints**\n\nEmissions from fuel (“direct”) and non-fuel (“indirect”) consumption, in percentage of total\n\nNote: “Direct” includes households’ own consumption of fossil fuels, both domestically sourced and imported. “Indirect, domestic” includes households’ own non-fuel, domestically sourced consumption. “Indirect imported” accounts for emissions linked to all other non-domestically sourced inputs and consumption goods. Estimates are based on the “consumer responsibility” principle, accounting for all household consumption. They therefore attribute emissions to the country where the final good is consumed, and this differs from the emissions per capita that are physically released in a given country. It is important to account for the consumption of imported goods in order to determine the carbon price embedded in households’ consumption baskets. Source: OECD calculations using IEA emissions factors for different fuels, World Input-Output Database (WIOD) as well as household budget surveys (2015 for EU countries, 2016 for Mexico, 2019 for Türkiye).\n\nStatLink 2 https://stat.link/hs30kq\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\n70\n\n80\n\n90\n\n100\n\nFrance Germany Poland Mexico Türkiye\n\nShare of total (%)\n\nDirect Indirect, domestic Indirect, imported","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p246.pdf"]},"response":{"errors":[],"extract_id":"extract_28ff26016897d9800533056520563d71","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**244** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n**_5\\.3.2. Who pollutes the most? Carbon footprints linked to household consumption_**\n\nEnergy consumption is a key driver of emissions, but it is not the only one. There are several reasons why energy spending is only a partial indicator of households’ carbon footprints. First, emissions from electricity generation vary enormously between countries, by a factor of 15 across the five countries considered here. ²⁰ Relatedly, for each of the main fuel consumption categories, emissions vary by type of fuel (see Annex Figure 5.B.1), as do prices before accounting for carbon pricing. For lower-income or rural households, domestic fuels can include high shares of solid fuels (coal, coke, peat, firewood), which have much higher emission factors than liquid fuel. Emission factors are lower for natural gas, an energy source that can be more common in urban areas. Motor fuels are commonly more expensive than domestic fuels and produce fewer emissions per energy unit than firewood, coal, or heating oil. Per unit of fuel expenditure, emissions – and therefore also the effect of a given carbon price – thus tend to be higher for domestic fuels than for motor fuels.\n\nSecond, non-fuel expenditures account for a large share of household spending. ²¹ Per unit of spending, fuel use creates more emissions than other consumption, but due to the size of non-fuel spending, the production of non-fuel goods and services is a significant driver of carbon footprints. Across the five countries, direct emissions from households’ own use of fossil fuels mostly account for around half of the total emissions linked to consumption (Figure 5.5). These estimates incorporate all household consumption, following the “consumer responsibility” principle (see figure note). Fuel spending was a lesser driver of total emissions in France, partly reflecting the balance of spending on fossil fuels and electricity, but also the types of fuel (coal and other solid fuel, liquid fuel, gas) used by households and in production processes. ²² Emissions linked to imported inputs or final goods (other than fuel) are also significant but comparatively small, accounting for less than 10% of emissions in each of the five countries. (This small share is reassuring in the context of the chapter’s later estimates of carbon price burdens, which do not account for differential carbon price trends in source countries.)\n\n### **Figure 5.5** **. Fuel expenditure is not the only driver of households’ carbon footpr** **ints**\n\nEmissions from fuel (“direct”) and non-fuel (“indirect”) consumption, in percentage of total\n\nNote: “Direct” includes households’ own consumption of fossil fuels, both domestically sourced and imported. “Indirect, domestic” includes households’ own non-fuel, domestically sourced consumption. “Indirect imported” accounts for emissions linked to all other non-domestically sourced inputs and consumption goods. Estimates are based on the “consumer responsibility” principle, accounting for all household consumption. They therefore attribute emissions to the country where the final good is consumed, and this differs from the emissions per capita that are physically released in a given country. It is important to account for the consumption of imported goods in order to determine the carbon price embedded in households’ consumption baskets. Source: OECD calculations using IEA emissions factors for different fuels, World Input-Output Database (WIOD) as well as household budget surveys (2015 for EU countries, 2016 for Mexico, 2019 for Türkiye).\n\nStatLink 2 https://stat.link/hs30kq\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\n70\n\n80\n\n90\n\n100\n\nFrance Germany Poland Mexico Türkiye\n\nShare of total (%)\n\nDirect Indirect, domestic Indirect, imported"],"full_content":"**244** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n**_5\\.3.2. Who pollutes the most? Carbon footprints linked to household consumption_**\n\nEnergy consumption is a key driver of emissions, but it is not the only one. There are several reasons why energy spending is only a partial indicator of households’ carbon footprints. First, emissions from electricity generation vary enormously between countries, by a factor of 15 across the five countries considered here. ²⁰ Relatedly, for each of the main fuel consumption categories, emissions vary by type of fuel (see Annex Figure 5.B.1), as do prices before accounting for carbon pricing. For lower-income or rural households, domestic fuels can include high shares of solid fuels (coal, coke, peat, firewood), which have much higher emission factors than liquid fuel. Emission factors are lower for natural gas, an energy source that can be more common in urban areas. Motor fuels are commonly more expensive than domestic fuels and produce fewer emissions per energy unit than firewood, coal, or heating oil. Per unit of fuel expenditure, emissions – and therefore also the effect of a given carbon price – thus tend to be higher for domestic fuels than for motor fuels.\n\nSecond, non-fuel expenditures account for a large share of household spending. ²¹ Per unit of spending, fuel use creates more emissions than other consumption, but due to the size of non-fuel spending, the production of non-fuel goods and services is a significant driver of carbon footprints. Across the five countries, direct emissions from households’ own use of fossil fuels mostly account for around half of the total emissions linked to consumption (Figure 5.5). These estimates incorporate all household consumption, following the “consumer responsibility” principle (see figure note). Fuel spending was a lesser driver of total emissions in France, partly reflecting the balance of spending on fossil fuels and electricity, but also the types of fuel (coal and other solid fuel, liquid fuel, gas) used by households and in production processes. ²² Emissions linked to imported inputs or final goods (other than fuel) are also significant but comparatively small, accounting for less than 10% of emissions in each of the five countries. (This small share is reassuring in the context of the chapter’s later estimates of carbon price burdens, which do not account for differential carbon price trends in source countries.)\n\n### **Figure 5.5** **. Fuel expenditure is not the only driver of households’ carbon footpr** **ints**\n\nEmissions from fuel (“direct”) and non-fuel (“indirect”) consumption, in percentage of total\n\nNote: “Direct” includes households’ own consumption of fossil fuels, both domestically sourced and imported. “Indirect, domestic” includes households’ own non-fuel, domestically sourced consumption. “Indirect imported” accounts for emissions linked to all other non-domestically sourced inputs and consumption goods. Estimates are based on the “consumer responsibility” principle, accounting for all household consumption. They therefore attribute emissions to the country where the final good is consumed, and this differs from the emissions per capita that are physically released in a given country. It is important to account for the consumption of imported goods in order to determine the carbon price embedded in households’ consumption baskets. Source: OECD calculations using IEA emissions factors for different fuels, World Input-Output Database (WIOD) as well as household budget surveys (2015 for EU countries, 2016 for Mexico, 2019 for Türkiye).\n\nStatLink 2 https://stat.link/hs30kq\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\n70\n\n80\n\n90\n\n100\n\nFrance Germany Poland Mexico Türkiye\n\nShare of total (%)\n\nDirect Indirect, domestic Indirect, imported","publish_date":null,"title":"Figure 5.5. Fuel expenditure is not the only driver of households’ carbon footprints","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p246.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p246.pdf"},"request":{"example_id":"chart/ac8b3538-en_p246","product_type":"parse","source_file_path":"ac8b3538-en_p246.pdf"},"started_at":"2026-11-04T19:45:40.597625"} +{"completed_at":"2026-11-04T19:45:42.737705","latency_in_ms":1193,"output":{"example_id":"chart/ac8b3538-en_p247","grounded_pages":[],"layout_pages":[],"markdown":" **245**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\nQuantifying indirect emissions embodied in non-fuel consumption is thus crucial for assessing distributional effects. Embodied emissions are not observed directly. Consumption patterns, and their carbon content, vary in complex ways between countries and households, making the net effect impossible to anticipate without careful modelling. Results in the remainder of this chapter trace emissions from both direct and indirect consumption to individual households, using the modelling approach summarised in Annex 5.A.\n\nAcross countries, differences in carbon footprints are very large, reflecting various factors, including levels of development, population density, consumption patterns, and production technology. Figure 5.6 shows emissions linked to household consumption across countries, at different points of the national emissions distribution (rather than the income distribution). Average emissions (not shown) range from around 1 tonne of CO2 per household and year in Mexico and Türkiye, to 6 tonnes in Poland, and 8 to 9 tonnes in France and Germany. Median emissions of German households were nearly as high as for households at the 8 th decile of the Polish emissions range. The average consumption for the top 10% emitting households in Mexico and Türkiye produced the same amount of emissions as the bottom 3 rd decile in Germany. These country differences in emissions attributed to domestic household consumption can be much bigger than those in terms of per-capita emissions that are physically released in each country (compare with Figure 5.3).\n\n### **Figure 5.6. Emissions from household consumption are very unequal across and within countries**\n\nEmissions from household consumption, tCO2 per household at different points in the national emissions distribution\n\nNote: Average emissions across the national emissions distribution (not income distribution), from lowest-emitting to the highest-emitting households. The ranking variable is emissions linked to household consumption, equivalised to account for household size. Estimates follow the “consumer responsibility” principle, accounting for all household consumption, including both domestically produced and imported goods. They therefore attribute emissions to the country where the final good is consumed, and this differs from the emissions per capita that are physically released in a given country. Source: OECD calculations using IEA emissions factors for different fuels, World Input-Output Database (WIOD) as well as household budget surveys (2015 for EU countries, 2016 for Mexico, 2019 for Türkiye).\n\nStatLink 2 https://stat.link/snwjx0\n\nWithin countries, the key driver of emissions disparities between poor and rich households is total consumption. In the five countries, on average, high-income households (top 10%) spent 4.5 times as much as the poorest 10% (Figure 5.7, top-left panel). Spending inequality was lowest in France and Poland (total spending for the top and bottom 10% differ by a factor of 3) and highest in Mexico (a factor of 9). Emissions associated with consumption therefore also rose strongly with income. On average across countries, profiles of total spending and emissions are very similar, pointing to the importance of total spending levels as a driver of carbon footprints. The findings on the income gradient of consumption-based\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n1 2 3 4 5 6 7 8 9 10\n\nTonne of CO2 per household\n\nEmissions decile\n\nFrance Germany Mexico Poland Türkiye","pages":[{"line_numbers":[],"markdown":" **245**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\nQuantifying indirect emissions embodied in non-fuel consumption is thus crucial for assessing distributional effects. Embodied emissions are not observed directly. Consumption patterns, and their carbon content, vary in complex ways between countries and households, making the net effect impossible to anticipate without careful modelling. Results in the remainder of this chapter trace emissions from both direct and indirect consumption to individual households, using the modelling approach summarised in Annex 5.A.\n\nAcross countries, differences in carbon footprints are very large, reflecting various factors, including levels of development, population density, consumption patterns, and production technology. Figure 5.6 shows emissions linked to household consumption across countries, at different points of the national emissions distribution (rather than the income distribution). Average emissions (not shown) range from around 1 tonne of CO2 per household and year in Mexico and Türkiye, to 6 tonnes in Poland, and 8 to 9 tonnes in France and Germany. Median emissions of German households were nearly as high as for households at the 8 th decile of the Polish emissions range. The average consumption for the top 10% emitting households in Mexico and Türkiye produced the same amount of emissions as the bottom 3 rd decile in Germany. These country differences in emissions attributed to domestic household consumption can be much bigger than those in terms of per-capita emissions that are physically released in each country (compare with Figure 5.3).\n\n### **Figure 5.6. Emissions from household consumption are very unequal across and within countries**\n\nEmissions from household consumption, tCO2 per household at different points in the national emissions distribution\n\nNote: Average emissions across the national emissions distribution (not income distribution), from lowest-emitting to the highest-emitting households. The ranking variable is emissions linked to household consumption, equivalised to account for household size. Estimates follow the “consumer responsibility” principle, accounting for all household consumption, including both domestically produced and imported goods. They therefore attribute emissions to the country where the final good is consumed, and this differs from the emissions per capita that are physically released in a given country. Source: OECD calculations using IEA emissions factors for different fuels, World Input-Output Database (WIOD) as well as household budget surveys (2015 for EU countries, 2016 for Mexico, 2019 for Türkiye).\n\nStatLink 2 https://stat.link/snwjx0\n\nWithin countries, the key driver of emissions disparities between poor and rich households is total consumption. In the five countries, on average, high-income households (top 10%) spent 4.5 times as much as the poorest 10% (Figure 5.7, top-left panel). Spending inequality was lowest in France and Poland (total spending for the top and bottom 10% differ by a factor of 3) and highest in Mexico (a factor of 9). Emissions associated with consumption therefore also rose strongly with income. On average across countries, profiles of total spending and emissions are very similar, pointing to the importance of total spending levels as a driver of carbon footprints. The findings on the income gradient of consumption-based\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n1 2 3 4 5 6 7 8 9 10\n\nTonne of CO2 per household\n\nEmissions decile\n\nFrance Germany Mexico Poland Türkiye","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":" **245**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\nQuantifying indirect emissions embodied in non-fuel consumption is thus crucial for assessing distributional effects. Embodied emissions are not observed directly. Consumption patterns, and their carbon content, vary in complex ways between countries and households, making the net effect impossible to anticipate without careful modelling. Results in the remainder of this chapter trace emissions from both direct and indirect consumption to individual households, using the modelling approach summarised in Annex 5.A.\n\nAcross countries, differences in carbon footprints are very large, reflecting various factors, including levels of development, population density, consumption patterns, and production technology. Figure 5.6 shows emissions linked to household consumption across countries, at different points of the national emissions distribution (rather than the income distribution). Average emissions (not shown) range from around 1 tonne of CO2 per household and year in Mexico and Türkiye, to 6 tonnes in Poland, and 8 to 9 tonnes in France and Germany. Median emissions of German households were nearly as high as for households at the 8 th decile of the Polish emissions range. The average consumption for the top 10% emitting households in Mexico and Türkiye produced the same amount of emissions as the bottom 3 rd decile in Germany. These country differences in emissions attributed to domestic household consumption can be much bigger than those in terms of per-capita emissions that are physically released in each country (compare with Figure 5.3).\n\n### **Figure 5.6. Emissions from household consumption are very unequal across and within countries**\n\nEmissions from household consumption, tCO2 per household at different points in the national emissions distribution\n\nNote: Average emissions across the national emissions distribution (not income distribution), from lowest-emitting to the highest-emitting households. The ranking variable is emissions linked to household consumption, equivalised to account for household size. Estimates follow the “consumer responsibility” principle, accounting for all household consumption, including both domestically produced and imported goods. They therefore attribute emissions to the country where the final good is consumed, and this differs from the emissions per capita that are physically released in a given country. Source: OECD calculations using IEA emissions factors for different fuels, World Input-Output Database (WIOD) as well as household budget surveys (2015 for EU countries, 2016 for Mexico, 2019 for Türkiye).\n\nStatLink 2 https://stat.link/snwjx0\n\nWithin countries, the key driver of emissions disparities between poor and rich households is total consumption. In the five countries, on average, high-income households (top 10%) spent 4.5 times as much as the poorest 10% (Figure 5.7, top-left panel). Spending inequality was lowest in France and Poland (total spending for the top and bottom 10% differ by a factor of 3) and highest in Mexico (a factor of 9). Emissions associated with consumption therefore also rose strongly with income. On average across countries, profiles of total spending and emissions are very similar, pointing to the importance of total spending levels as a driver of carbon footprints. The findings on the income gradient of consumption-based\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n1 2 3 4 5 6 7 8 9 10\n\nTonne of CO2 per household\n\nEmissions decile\n\nFrance Germany Mexico Poland Türkiye","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p247.pdf"]},"response":{"errors":[],"extract_id":"extract_c47149acd5fc4a6078c7f97eafdafd60","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":[" **245**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\nQuantifying indirect emissions embodied in non-fuel consumption is thus crucial for assessing distributional effects. Embodied emissions are not observed directly. Consumption patterns, and their carbon content, vary in complex ways between countries and households, making the net effect impossible to anticipate without careful modelling. Results in the remainder of this chapter trace emissions from both direct and indirect consumption to individual households, using the modelling approach summarised in Annex 5.A.\n\nAcross countries, differences in carbon footprints are very large, reflecting various factors, including levels of development, population density, consumption patterns, and production technology. Figure 5.6 shows emissions linked to household consumption across countries, at different points of the national emissions distribution (rather than the income distribution). Average emissions (not shown) range from around 1 tonne of CO2 per household and year in Mexico and Türkiye, to 6 tonnes in Poland, and 8 to 9 tonnes in France and Germany. Median emissions of German households were nearly as high as for households at the 8 th decile of the Polish emissions range. The average consumption for the top 10% emitting households in Mexico and Türkiye produced the same amount of emissions as the bottom 3 rd decile in Germany. These country differences in emissions attributed to domestic household consumption can be much bigger than those in terms of per-capita emissions that are physically released in each country (compare with Figure 5.3).\n\n### **Figure 5.6. Emissions from household consumption are very unequal across and within countries**\n\nEmissions from household consumption, tCO2 per household at different points in the national emissions distribution\n\nNote: Average emissions across the national emissions distribution (not income distribution), from lowest-emitting to the highest-emitting households. The ranking variable is emissions linked to household consumption, equivalised to account for household size. Estimates follow the “consumer responsibility” principle, accounting for all household consumption, including both domestically produced and imported goods. They therefore attribute emissions to the country where the final good is consumed, and this differs from the emissions per capita that are physically released in a given country. Source: OECD calculations using IEA emissions factors for different fuels, World Input-Output Database (WIOD) as well as household budget surveys (2015 for EU countries, 2016 for Mexico, 2019 for Türkiye).\n\nStatLink 2 https://stat.link/snwjx0\n\nWithin countries, the key driver of emissions disparities between poor and rich households is total consumption. In the five countries, on average, high-income households (top 10%) spent 4.5 times as much as the poorest 10% (Figure 5.7, top-left panel). Spending inequality was lowest in France and Poland (total spending for the top and bottom 10% differ by a factor of 3) and highest in Mexico (a factor of 9). Emissions associated with consumption therefore also rose strongly with income. On average across countries, profiles of total spending and emissions are very similar, pointing to the importance of total spending levels as a driver of carbon footprints. The findings on the income gradient of consumption-based\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n1 2 3 4 5 6 7 8 9 10\n\nTonne of CO2 per household\n\nEmissions decile\n\nFrance Germany Mexico Poland Türkiye"],"full_content":" **245**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\nQuantifying indirect emissions embodied in non-fuel consumption is thus crucial for assessing distributional effects. Embodied emissions are not observed directly. Consumption patterns, and their carbon content, vary in complex ways between countries and households, making the net effect impossible to anticipate without careful modelling. Results in the remainder of this chapter trace emissions from both direct and indirect consumption to individual households, using the modelling approach summarised in Annex 5.A.\n\nAcross countries, differences in carbon footprints are very large, reflecting various factors, including levels of development, population density, consumption patterns, and production technology. Figure 5.6 shows emissions linked to household consumption across countries, at different points of the national emissions distribution (rather than the income distribution). Average emissions (not shown) range from around 1 tonne of CO2 per household and year in Mexico and Türkiye, to 6 tonnes in Poland, and 8 to 9 tonnes in France and Germany. Median emissions of German households were nearly as high as for households at the 8 th decile of the Polish emissions range. The average consumption for the top 10% emitting households in Mexico and Türkiye produced the same amount of emissions as the bottom 3 rd decile in Germany. These country differences in emissions attributed to domestic household consumption can be much bigger than those in terms of per-capita emissions that are physically released in each country (compare with Figure 5.3).\n\n### **Figure 5.6. Emissions from household consumption are very unequal across and within countries**\n\nEmissions from household consumption, tCO2 per household at different points in the national emissions distribution\n\nNote: Average emissions across the national emissions distribution (not income distribution), from lowest-emitting to the highest-emitting households. The ranking variable is emissions linked to household consumption, equivalised to account for household size. Estimates follow the “consumer responsibility” principle, accounting for all household consumption, including both domestically produced and imported goods. They therefore attribute emissions to the country where the final good is consumed, and this differs from the emissions per capita that are physically released in a given country. Source: OECD calculations using IEA emissions factors for different fuels, World Input-Output Database (WIOD) as well as household budget surveys (2015 for EU countries, 2016 for Mexico, 2019 for Türkiye).\n\nStatLink 2 https://stat.link/snwjx0\n\nWithin countries, the key driver of emissions disparities between poor and rich households is total consumption. In the five countries, on average, high-income households (top 10%) spent 4.5 times as much as the poorest 10% (Figure 5.7, top-left panel). Spending inequality was lowest in France and Poland (total spending for the top and bottom 10% differ by a factor of 3) and highest in Mexico (a factor of 9). Emissions associated with consumption therefore also rose strongly with income. On average across countries, profiles of total spending and emissions are very similar, pointing to the importance of total spending levels as a driver of carbon footprints. The findings on the income gradient of consumption-based\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n1 2 3 4 5 6 7 8 9 10\n\nTonne of CO2 per household\n\nEmissions decile\n\nFrance Germany Mexico Poland Türkiye","publish_date":null,"title":"Figure 5.6. Emissions from household consumption are very unequal across and within countries","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p247.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p247.pdf"},"request":{"example_id":"chart/ac8b3538-en_p247","product_type":"parse","source_file_path":"ac8b3538-en_p247.pdf"},"started_at":"2026-11-04T19:45:41.544522"} +{"completed_at":"2026-11-04T19:45:45.912201","latency_in_ms":2675,"output":{"example_id":"chart/ac8b3538-en_p248","grounded_pages":[],"layout_pages":[],"markdown":"**246** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\nemissions echo those found in Chancel, Bothe and Voituriez (2023[112])’s recent study, even though the results presented here are less skewed.\n\n### **Figure 5.7. Carbon footprints are much bigger at high income levels, but in some countries low- income households consume greater shares of high-emission goods**\n\nHousehold expenditure and emissions shares of total, by (disposable) income decile\n\nNote: Average expenditure (resp. emissions) by income group (equivalised disposable household income) as a percentage of overall expenditure (resp. emissions). Estimates follow the “consumer responsibility” principle, accounting for all household consumption, including both domestically produced and imported goods. They therefore attribute emissions to the country where the final good is consumed, and this differs from the emissions per capita that are physically released in a given country. Reading: in Poland, expenditure from the lowest income decile accounts for about 6% of total national expenditure and expenditure from the highest income decile accounts for about 18%. About 5% of emissions accrued to the lowest income decile while 16.5% of emissions accrued to the highest income decile. Source: OECD calculations using IEA emissions factors for different fuels, World Input-Output Database (WIOD) as well as household budget surveys (2015 for EU countries, 2016 for Mexico, 2019 for Türkiye).\n\nStatLink 2 https://stat.link/2jcm09\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n35\n\n1 2 3 4 5 6 7 8 9 10\n\nIncome decile\n\n**France** %\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n35\n\n1 2 3 4 5 6 7 8 9 10\n\nIncome decile\n\n**Germany** %\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n35\n\n1 2 3 4 5 6 7 8 9 10\n\nIncome decile\n\n**Mexico** %\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n35\n\n1 2 3 4 5 6 7 8 9 10\n\nIncome decile\n\n**Poland** %\n\nExpenditure CO2 emissions\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n35\n\n1 2 3 4 5 6 7 8 9 10\n\nIncome decile\n\n**Türkiye** %\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n35\n\n1 2 3 4 5 6 7 8 9 10\n\nIncome decile\n\n**Five-country average** %","pages":[{"line_numbers":[],"markdown":"**246** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\nemissions echo those found in Chancel, Bothe and Voituriez (2023[112])’s recent study, even though the results presented here are less skewed.\n\n### **Figure 5.7. Carbon footprints are much bigger at high income levels, but in some countries low- income households consume greater shares of high-emission goods**\n\nHousehold expenditure and emissions shares of total, by (disposable) income decile\n\nNote: Average expenditure (resp. emissions) by income group (equivalised disposable household income) as a percentage of overall expenditure (resp. emissions). Estimates follow the “consumer responsibility” principle, accounting for all household consumption, including both domestically produced and imported goods. They therefore attribute emissions to the country where the final good is consumed, and this differs from the emissions per capita that are physically released in a given country. Reading: in Poland, expenditure from the lowest income decile accounts for about 6% of total national expenditure and expenditure from the highest income decile accounts for about 18%. About 5% of emissions accrued to the lowest income decile while 16.5% of emissions accrued to the highest income decile. Source: OECD calculations using IEA emissions factors for different fuels, World Input-Output Database (WIOD) as well as household budget surveys (2015 for EU countries, 2016 for Mexico, 2019 for Türkiye).\n\nStatLink 2 https://stat.link/2jcm09\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n35\n\n1 2 3 4 5 6 7 8 9 10\n\nIncome decile\n\n**France** %\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n35\n\n1 2 3 4 5 6 7 8 9 10\n\nIncome decile\n\n**Germany** %\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n35\n\n1 2 3 4 5 6 7 8 9 10\n\nIncome decile\n\n**Mexico** %\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n35\n\n1 2 3 4 5 6 7 8 9 10\n\nIncome decile\n\n**Poland** %\n\nExpenditure CO2 emissions\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n35\n\n1 2 3 4 5 6 7 8 9 10\n\nIncome decile\n\n**Türkiye** %\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n35\n\n1 2 3 4 5 6 7 8 9 10\n\nIncome decile\n\n**Five-country average** %","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**246** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\nemissions echo those found in Chancel, Bothe and Voituriez (2023[112])’s recent study, even though the results presented here are less skewed.\n\n### **Figure 5.7. Carbon footprints are much bigger at high income levels, but in some countries low- income households consume greater shares of high-emission goods**\n\nHousehold expenditure and emissions shares of total, by (disposable) income decile\n\nNote: Average expenditure (resp. emissions) by income group (equivalised disposable household income) as a percentage of overall expenditure (resp. emissions). Estimates follow the “consumer responsibility” principle, accounting for all household consumption, including both domestically produced and imported goods. They therefore attribute emissions to the country where the final good is consumed, and this differs from the emissions per capita that are physically released in a given country. Reading: in Poland, expenditure from the lowest income decile accounts for about 6% of total national expenditure and expenditure from the highest income decile accounts for about 18%. About 5% of emissions accrued to the lowest income decile while 16.5% of emissions accrued to the highest income decile. Source: OECD calculations using IEA emissions factors for different fuels, World Input-Output Database (WIOD) as well as household budget surveys (2015 for EU countries, 2016 for Mexico, 2019 for Türkiye).\n\nStatLink 2 https://stat.link/2jcm09\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n35\n\n1 2 3 4 5 6 7 8 9 10\n\nIncome decile\n\n**France** %\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n35\n\n1 2 3 4 5 6 7 8 9 10\n\nIncome decile\n\n**Germany** %\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n35\n\n1 2 3 4 5 6 7 8 9 10\n\nIncome decile\n\n**Mexico** %\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n35\n\n1 2 3 4 5 6 7 8 9 10\n\nIncome decile\n\n**Poland** %\n\nExpenditure CO2 emissions\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n35\n\n1 2 3 4 5 6 7 8 9 10\n\nIncome decile\n\n**Türkiye** %\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n35\n\n1 2 3 4 5 6 7 8 9 10\n\nIncome decile\n\n**Five-country average** %","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p248.pdf"]},"response":{"errors":[],"extract_id":"extract_189af8b9cd3eaa5207cc78b6523b2eea","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**246** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\nemissions echo those found in Chancel, Bothe and Voituriez (2023[112])’s recent study, even though the results presented here are less skewed.\n\n### **Figure 5.7. Carbon footprints are much bigger at high income levels, but in some countries low- income households consume greater shares of high-emission goods**\n\nHousehold expenditure and emissions shares of total, by (disposable) income decile\n\nNote: Average expenditure (resp. emissions) by income group (equivalised disposable household income) as a percentage of overall expenditure (resp. emissions). Estimates follow the “consumer responsibility” principle, accounting for all household consumption, including both domestically produced and imported goods. They therefore attribute emissions to the country where the final good is consumed, and this differs from the emissions per capita that are physically released in a given country. Reading: in Poland, expenditure from the lowest income decile accounts for about 6% of total national expenditure and expenditure from the highest income decile accounts for about 18%. About 5% of emissions accrued to the lowest income decile while 16.5% of emissions accrued to the highest income decile. Source: OECD calculations using IEA emissions factors for different fuels, World Input-Output Database (WIOD) as well as household budget surveys (2015 for EU countries, 2016 for Mexico, 2019 for Türkiye).\n\nStatLink 2 https://stat.link/2jcm09\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n35\n\n1 2 3 4 5 6 7 8 9 10\n\nIncome decile\n\n**France** %\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n35\n\n1 2 3 4 5 6 7 8 9 10\n\nIncome decile\n\n**Germany** %\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n35\n\n1 2 3 4 5 6 7 8 9 10\n\nIncome decile\n\n**Mexico** %\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n35\n\n1 2 3 4 5 6 7 8 9 10\n\nIncome decile\n\n**Poland** %\n\nExpenditure CO2 emissions\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n35\n\n1 2 3 4 5 6 7 8 9 10\n\nIncome decile\n\n**Türkiye** %\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n35\n\n1 2 3 4 5 6 7 8 9 10\n\nIncome decile\n\n**Five-country average** %"],"full_content":"**246** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\nemissions echo those found in Chancel, Bothe and Voituriez (2023[112])’s recent study, even though the results presented here are less skewed.\n\n### **Figure 5.7. Carbon footprints are much bigger at high income levels, but in some countries low- income households consume greater shares of high-emission goods**\n\nHousehold expenditure and emissions shares of total, by (disposable) income decile\n\nNote: Average expenditure (resp. emissions) by income group (equivalised disposable household income) as a percentage of overall expenditure (resp. emissions). Estimates follow the “consumer responsibility” principle, accounting for all household consumption, including both domestically produced and imported goods. They therefore attribute emissions to the country where the final good is consumed, and this differs from the emissions per capita that are physically released in a given country. Reading: in Poland, expenditure from the lowest income decile accounts for about 6% of total national expenditure and expenditure from the highest income decile accounts for about 18%. About 5% of emissions accrued to the lowest income decile while 16.5% of emissions accrued to the highest income decile. Source: OECD calculations using IEA emissions factors for different fuels, World Input-Output Database (WIOD) as well as household budget surveys (2015 for EU countries, 2016 for Mexico, 2019 for Türkiye).\n\nStatLink 2 https://stat.link/2jcm09\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n35\n\n1 2 3 4 5 6 7 8 9 10\n\nIncome decile\n\n**France** %\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n35\n\n1 2 3 4 5 6 7 8 9 10\n\nIncome decile\n\n**Germany** %\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n35\n\n1 2 3 4 5 6 7 8 9 10\n\nIncome decile\n\n**Mexico** %\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n35\n\n1 2 3 4 5 6 7 8 9 10\n\nIncome decile\n\n**Poland** %\n\nExpenditure CO2 emissions\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n35\n\n1 2 3 4 5 6 7 8 9 10\n\nIncome decile\n\n**Türkiye** %\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n35\n\n1 2 3 4 5 6 7 8 9 10\n\nIncome decile\n\n**Five-country average** %","publish_date":null,"title":"Figure 5.7. Carbon footprints are much bigger at high income levels, but in some countries low- income households consume greater","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p248.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p248.pdf"},"request":{"example_id":"chart/ac8b3538-en_p248","product_type":"parse","source_file_path":"ac8b3538-en_p248.pdf"},"started_at":"2026-11-04T19:45:43.237195"} +{"completed_at":"2026-11-04T19:45:47.926025","latency_in_ms":2751,"output":{"example_id":"chart/ac8b3538-en_p25","grounded_pages":[],"layout_pages":[],"markdown":" **23**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\nAs for employment rates, they progressed more for women than for men in most OECD countries compared to pre-pandemic level, indicating that gender gaps in employment rates are narrowing in many OECD countries. Interestingly, data suggest that the higher the employment gender gap in Q4 2019, the greater was the growth in women’s employment rate between Q4 2019 and Q1 2024 (Annex Figure 1.A.1).\n\nUnemployment rates remain at historically low levels in many OECD countries (Figure 1.3). The unemployment rate for the OECD was already at its pre-COVID-19 level in January 2022 – before Russia’s full-scale invasion of Ukraine. Since then, the unemployment rate declined by a further 0.4 percentage points and stood at 4.9% in May 2024 after a record low of 4.8% in September 2023. Unemployment rates are below their levels of January 2022 in 17 OECD countries, and above that level by more than 0.5 percentage points in 10 countries.\n\nThe most recent data also suggest stable unemployment rates across countries, with only 13 OECD countries having experienced an increase of more than a quarter of a percentage point over the past six months. Gender differences in the changes in unemployment rates between December 2019 and May 2024 are generally small: while not shown here, the gender gap in unemployment rates was rather stable on average for the OECD area except for Colombia, Costa Rica and Greece where it decreased by more than 2 percentage points.\n\n### **Figure 1.3. Unemployment rates remain at historically low levels in many countries**\n\nUnemployment rate (percentage of labour force), seasonally adjusted data\n\nNote: The labour force population includes all those aged 15 or more. Euro Area refers to the 20 Eurozone countries. The labour force population includes all those aged 15 or more. The latest month refers to **Q1 2024** for New Zealand and Switzerland; **February 2024** for Iceland, **March 2024** for the United Kingdom; **April 2024** for Chile, Costa Rica and Türkiye; and **June 2024** for Canada and the United States. Source: OECD (2024), “Unemployment rate” (indicator), https://doi.org/10.1787/52570002-en (accessed on 9 July 2024).\n\nStatLink 2 https://stat.link/1z75tu\n\n**_1\\.1.2. Labour force participation rates continue to increase while average hours worked are slightly below their pre-COVID-19 levels in several countries_**\n\nLabour force participation rates among the working age population have continued to increase in most of the OECD countries over the past year or so ¹ (Figure 1.4, Panel A). In Q1 2024, labour force participation rates were 1.3 percentage points higher than at the end of 2019 on average across the OECD. More than\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\n16\n\n18 %\n\nLatest (May 2024) Pre-crisis (December 2019) January 2022 Increased over the last 6 months","pages":[{"line_numbers":[],"markdown":" **23**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\nAs for employment rates, they progressed more for women than for men in most OECD countries compared to pre-pandemic level, indicating that gender gaps in employment rates are narrowing in many OECD countries. Interestingly, data suggest that the higher the employment gender gap in Q4 2019, the greater was the growth in women’s employment rate between Q4 2019 and Q1 2024 (Annex Figure 1.A.1).\n\nUnemployment rates remain at historically low levels in many OECD countries (Figure 1.3). The unemployment rate for the OECD was already at its pre-COVID-19 level in January 2022 – before Russia’s full-scale invasion of Ukraine. Since then, the unemployment rate declined by a further 0.4 percentage points and stood at 4.9% in May 2024 after a record low of 4.8% in September 2023. Unemployment rates are below their levels of January 2022 in 17 OECD countries, and above that level by more than 0.5 percentage points in 10 countries.\n\nThe most recent data also suggest stable unemployment rates across countries, with only 13 OECD countries having experienced an increase of more than a quarter of a percentage point over the past six months. Gender differences in the changes in unemployment rates between December 2019 and May 2024 are generally small: while not shown here, the gender gap in unemployment rates was rather stable on average for the OECD area except for Colombia, Costa Rica and Greece where it decreased by more than 2 percentage points.\n\n### **Figure 1.3. Unemployment rates remain at historically low levels in many countries**\n\nUnemployment rate (percentage of labour force), seasonally adjusted data\n\nNote: The labour force population includes all those aged 15 or more. Euro Area refers to the 20 Eurozone countries. The labour force population includes all those aged 15 or more. The latest month refers to **Q1 2024** for New Zealand and Switzerland; **February 2024** for Iceland, **March 2024** for the United Kingdom; **April 2024** for Chile, Costa Rica and Türkiye; and **June 2024** for Canada and the United States. Source: OECD (2024), “Unemployment rate” (indicator), https://doi.org/10.1787/52570002-en (accessed on 9 July 2024).\n\nStatLink 2 https://stat.link/1z75tu\n\n**_1\\.1.2. Labour force participation rates continue to increase while average hours worked are slightly below their pre-COVID-19 levels in several countries_**\n\nLabour force participation rates among the working age population have continued to increase in most of the OECD countries over the past year or so ¹ (Figure 1.4, Panel A). In Q1 2024, labour force participation rates were 1.3 percentage points higher than at the end of 2019 on average across the OECD. More than\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\n16\n\n18 %\n\nLatest (May 2024) Pre-crisis (December 2019) January 2022 Increased over the last 6 months","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":" **23**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\nAs for employment rates, they progressed more for women than for men in most OECD countries compared to pre-pandemic level, indicating that gender gaps in employment rates are narrowing in many OECD countries. Interestingly, data suggest that the higher the employment gender gap in Q4 2019, the greater was the growth in women’s employment rate between Q4 2019 and Q1 2024 (Annex Figure 1.A.1).\n\nUnemployment rates remain at historically low levels in many OECD countries (Figure 1.3). The unemployment rate for the OECD was already at its pre-COVID-19 level in January 2022 – before Russia’s full-scale invasion of Ukraine. Since then, the unemployment rate declined by a further 0.4 percentage points and stood at 4.9% in May 2024 after a record low of 4.8% in September 2023. Unemployment rates are below their levels of January 2022 in 17 OECD countries, and above that level by more than 0.5 percentage points in 10 countries.\n\nThe most recent data also suggest stable unemployment rates across countries, with only 13 OECD countries having experienced an increase of more than a quarter of a percentage point over the past six months. Gender differences in the changes in unemployment rates between December 2019 and May 2024 are generally small: while not shown here, the gender gap in unemployment rates was rather stable on average for the OECD area except for Colombia, Costa Rica and Greece where it decreased by more than 2 percentage points.\n\n### **Figure 1.3. Unemployment rates remain at historically low levels in many countries**\n\nUnemployment rate (percentage of labour force), seasonally adjusted data\n\nNote: The labour force population includes all those aged 15 or more. Euro Area refers to the 20 Eurozone countries. The labour force population includes all those aged 15 or more. The latest month refers to **Q1 2024** for New Zealand and Switzerland; **February 2024** for Iceland, **March 2024** for the United Kingdom; **April 2024** for Chile, Costa Rica and Türkiye; and **June 2024** for Canada and the United States. Source: OECD (2024), “Unemployment rate” (indicator), https://doi.org/10.1787/52570002-en (accessed on 9 July 2024).\n\nStatLink 2 https://stat.link/1z75tu\n\n**_1\\.1.2. Labour force participation rates continue to increase while average hours worked are slightly below their pre-COVID-19 levels in several countries_**\n\nLabour force participation rates among the working age population have continued to increase in most of the OECD countries over the past year or so ¹ (Figure 1.4, Panel A). In Q1 2024, labour force participation rates were 1.3 percentage points higher than at the end of 2019 on average across the OECD. More than\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\n16\n\n18 %\n\nLatest (May 2024) Pre-crisis (December 2019) January 2022 Increased over the last 6 months","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p25.pdf"]},"response":{"errors":[],"extract_id":"extract_9acb092ff291266ce672ae07b615cda6","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":[" **23**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\nAs for employment rates, they progressed more for women than for men in most OECD countries compared to pre-pandemic level, indicating that gender gaps in employment rates are narrowing in many OECD countries. Interestingly, data suggest that the higher the employment gender gap in Q4 2019, the greater was the growth in women’s employment rate between Q4 2019 and Q1 2024 (Annex Figure 1.A.1).\n\nUnemployment rates remain at historically low levels in many OECD countries (Figure 1.3). The unemployment rate for the OECD was already at its pre-COVID-19 level in January 2022 – before Russia’s full-scale invasion of Ukraine. Since then, the unemployment rate declined by a further 0.4 percentage points and stood at 4.9% in May 2024 after a record low of 4.8% in September 2023. Unemployment rates are below their levels of January 2022 in 17 OECD countries, and above that level by more than 0.5 percentage points in 10 countries.\n\nThe most recent data also suggest stable unemployment rates across countries, with only 13 OECD countries having experienced an increase of more than a quarter of a percentage point over the past six months. Gender differences in the changes in unemployment rates between December 2019 and May 2024 are generally small: while not shown here, the gender gap in unemployment rates was rather stable on average for the OECD area except for Colombia, Costa Rica and Greece where it decreased by more than 2 percentage points.\n\n### **Figure 1.3. Unemployment rates remain at historically low levels in many countries**\n\nUnemployment rate (percentage of labour force), seasonally adjusted data\n\nNote: The labour force population includes all those aged 15 or more. Euro Area refers to the 20 Eurozone countries. The labour force population includes all those aged 15 or more. The latest month refers to **Q1 2024** for New Zealand and Switzerland; **February 2024** for Iceland, **March 2024** for the United Kingdom; **April 2024** for Chile, Costa Rica and Türkiye; and **June 2024** for Canada and the United States. Source: OECD (2024), “Unemployment rate” (indicator), https://doi.org/10.1787/52570002-en (accessed on 9 July 2024).\n\nStatLink 2 https://stat.link/1z75tu\n\n**_1\\.1.2. Labour force participation rates continue to increase while average hours worked are slightly below their pre-COVID-19 levels in several countries_**\n\nLabour force participation rates among the working age population have continued to increase in most of the OECD countries over the past year or so ¹ (Figure 1.4, Panel A). In Q1 2024, labour force participation rates were 1.3 percentage points higher than at the end of 2019 on average across the OECD. More than\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\n16\n\n18 %\n\nLatest (May 2024) Pre-crisis (December 2019) January 2022 Increased over the last 6 months"],"full_content":" **23**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\nAs for employment rates, they progressed more for women than for men in most OECD countries compared to pre-pandemic level, indicating that gender gaps in employment rates are narrowing in many OECD countries. Interestingly, data suggest that the higher the employment gender gap in Q4 2019, the greater was the growth in women’s employment rate between Q4 2019 and Q1 2024 (Annex Figure 1.A.1).\n\nUnemployment rates remain at historically low levels in many OECD countries (Figure 1.3). The unemployment rate for the OECD was already at its pre-COVID-19 level in January 2022 – before Russia’s full-scale invasion of Ukraine. Since then, the unemployment rate declined by a further 0.4 percentage points and stood at 4.9% in May 2024 after a record low of 4.8% in September 2023. Unemployment rates are below their levels of January 2022 in 17 OECD countries, and above that level by more than 0.5 percentage points in 10 countries.\n\nThe most recent data also suggest stable unemployment rates across countries, with only 13 OECD countries having experienced an increase of more than a quarter of a percentage point over the past six months. Gender differences in the changes in unemployment rates between December 2019 and May 2024 are generally small: while not shown here, the gender gap in unemployment rates was rather stable on average for the OECD area except for Colombia, Costa Rica and Greece where it decreased by more than 2 percentage points.\n\n### **Figure 1.3. Unemployment rates remain at historically low levels in many countries**\n\nUnemployment rate (percentage of labour force), seasonally adjusted data\n\nNote: The labour force population includes all those aged 15 or more. Euro Area refers to the 20 Eurozone countries. The labour force population includes all those aged 15 or more. The latest month refers to **Q1 2024** for New Zealand and Switzerland; **February 2024** for Iceland, **March 2024** for the United Kingdom; **April 2024** for Chile, Costa Rica and Türkiye; and **June 2024** for Canada and the United States. Source: OECD (2024), “Unemployment rate” (indicator), https://doi.org/10.1787/52570002-en (accessed on 9 July 2024).\n\nStatLink 2 https://stat.link/1z75tu\n\n**_1\\.1.2. Labour force participation rates continue to increase while average hours worked are slightly below their pre-COVID-19 levels in several countries_**\n\nLabour force participation rates among the working age population have continued to increase in most of the OECD countries over the past year or so ¹ (Figure 1.4, Panel A). In Q1 2024, labour force participation rates were 1.3 percentage points higher than at the end of 2019 on average across the OECD. More than\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\n16\n\n18 %\n\nLatest (May 2024) Pre-crisis (December 2019) January 2022 Increased over the last 6 months","publish_date":null,"title":"Figure 1.3. Unemployment rates remain at historically low levels in many countries","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p25.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p25.pdf"},"request":{"example_id":"chart/ac8b3538-en_p25","product_type":"parse","source_file_path":"ac8b3538-en_p25.pdf"},"started_at":"2026-11-04T19:45:45.174235"} +{"completed_at":"2026-11-04T19:45:47.947612","latency_in_ms":1728,"output":{"example_id":"chart/ac8b3538-en_p250","grounded_pages":[],"layout_pages":[],"markdown":"**248** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Figure 5.8. Various household characteristics shape carbon footprints**\n\nHousehold characteristics at different emission levels (see figure note)\n\nNote: Ratios compare socio-economic characteristics between high-emitting households (the top 10% of emissions) and low-emitting households (the bottom 10% of emissions) as follows. Number of earners and number of children in the household: Ratio of average numbers of earners and children per household; Other categories: Ratio of number of households in a rural area, or headed by a male, by a person with completed tertiary education, or by a pensioner. Urban/rural indicators are not available from the Mexican and Turkish household budget surveys. Reading, using the example of Germany: The average number of earners in the top 10% emitting households is 4.3 times that in the bottom 10% emitting households (note that the reason for this large relative difference is that almost 80% of the bottom 10% emitting households in Germany had no earner – and the average number of earners in the top 10% emitting households is 1.4). Reading is equivalent for “number of children”. For the other categories, ratios relate to the number of households in the top 10% and bottom 10% emitting households: the number of households located in rural areas, and the number headed by a male, by somebody with tertiary education, and by a pensioner. In other (less technical) terms, in Germany, high-emitting households were three times more likely to live in a rural area than low-emitting households. Source: OECD calculations using IEA emissions factors for different fuels, World Input-Output Database (WIOD) as well as household budget surveys (2015 for EU countries, 2016 for Mexico, 2019 for Türkiye).\n\nStatLink 2 https://stat.link/tfzmcx\n\n**_5\\.3.3. Policy changes: Who paid for recent carbon pricing measures?_**\n\nHouseholds’ carbon footprints are a primary determinant of carbon pricing burdens. But they are not the only one, as real-world carbon pricing measures do not apply uniformly, and not all emissions are therefore priced equally. For instance, excise taxes, carbon taxes and emissions trading systems can, and often do, vary substantially between industry and fuel type. For instance, in the 72 countries covered by the OECD ECR database, the road transport sector faces the highest carbon rates (rates above EUR 60 and EUR 120 per tonne of CO2 mostly occur in that sector), followed by the electricity and off-road transport sectors. In the industry and building sectors 72% and 64%, respectively, of emissions remain unpriced, while almost three-quarters of emissions in the electricity sector face a positive carbon price (OECD, 2023[8]). The correspondence between household emissions and their carbon price burden is thus neither perfect nor straightforward, and depends on the specific design of carbon pricing measures.\n\nThe period between 2012 and 2021 saw considerable policy innovation in the area of carbon pricing. While carbon prices mostly fall far short of those needed for meeting key mitigation commitments, OECD countries often raised them significantly over the period. Initial carbon prices in 2012 and the pace of subsequent changes varied, however, as did policy levers (Section 5.1 and Figure 5.2 above). Elgouacem et al. (forthcoming[14]) gives further details on specific reforms in the five countries. In four of the five countries included in the empirical analysis effective carbon rates increased in (constant) Euro terms – in most of them substantially so. ETS prices and coverage increased in all EU countries, approximately doubling ECRs in\n\n0\n\n1\n\n2\n\n3\n\n4\n\n5\n\n6\n\n7\n\n8\n\n9\n\n10\n\nNumber of earners Number of children Male-headed Head has completed\n\ntertiary education\n\nPensioner-headed Rural area\n\nGermany France Poland Mexico Türkiye\n\nCharacteristics of high-emitting vs low-emitting households 36","pages":[{"line_numbers":[],"markdown":"**248** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Figure 5.8. Various household characteristics shape carbon footprints**\n\nHousehold characteristics at different emission levels (see figure note)\n\nNote: Ratios compare socio-economic characteristics between high-emitting households (the top 10% of emissions) and low-emitting households (the bottom 10% of emissions) as follows. Number of earners and number of children in the household: Ratio of average numbers of earners and children per household; Other categories: Ratio of number of households in a rural area, or headed by a male, by a person with completed tertiary education, or by a pensioner. Urban/rural indicators are not available from the Mexican and Turkish household budget surveys. Reading, using the example of Germany: The average number of earners in the top 10% emitting households is 4.3 times that in the bottom 10% emitting households (note that the reason for this large relative difference is that almost 80% of the bottom 10% emitting households in Germany had no earner – and the average number of earners in the top 10% emitting households is 1.4). Reading is equivalent for “number of children”. For the other categories, ratios relate to the number of households in the top 10% and bottom 10% emitting households: the number of households located in rural areas, and the number headed by a male, by somebody with tertiary education, and by a pensioner. In other (less technical) terms, in Germany, high-emitting households were three times more likely to live in a rural area than low-emitting households. Source: OECD calculations using IEA emissions factors for different fuels, World Input-Output Database (WIOD) as well as household budget surveys (2015 for EU countries, 2016 for Mexico, 2019 for Türkiye).\n\nStatLink 2 https://stat.link/tfzmcx\n\n**_5\\.3.3. Policy changes: Who paid for recent carbon pricing measures?_**\n\nHouseholds’ carbon footprints are a primary determinant of carbon pricing burdens. But they are not the only one, as real-world carbon pricing measures do not apply uniformly, and not all emissions are therefore priced equally. For instance, excise taxes, carbon taxes and emissions trading systems can, and often do, vary substantially between industry and fuel type. For instance, in the 72 countries covered by the OECD ECR database, the road transport sector faces the highest carbon rates (rates above EUR 60 and EUR 120 per tonne of CO2 mostly occur in that sector), followed by the electricity and off-road transport sectors. In the industry and building sectors 72% and 64%, respectively, of emissions remain unpriced, while almost three-quarters of emissions in the electricity sector face a positive carbon price (OECD, 2023[8]). The correspondence between household emissions and their carbon price burden is thus neither perfect nor straightforward, and depends on the specific design of carbon pricing measures.\n\nThe period between 2012 and 2021 saw considerable policy innovation in the area of carbon pricing. While carbon prices mostly fall far short of those needed for meeting key mitigation commitments, OECD countries often raised them significantly over the period. Initial carbon prices in 2012 and the pace of subsequent changes varied, however, as did policy levers (Section 5.1 and Figure 5.2 above). Elgouacem et al. (forthcoming[14]) gives further details on specific reforms in the five countries. In four of the five countries included in the empirical analysis effective carbon rates increased in (constant) Euro terms – in most of them substantially so. ETS prices and coverage increased in all EU countries, approximately doubling ECRs in\n\n0\n\n1\n\n2\n\n3\n\n4\n\n5\n\n6\n\n7\n\n8\n\n9\n\n10\n\nNumber of earners Number of children Male-headed Head has completed\n\ntertiary education\n\nPensioner-headed Rural area\n\nGermany France Poland Mexico Türkiye\n\nCharacteristics of high-emitting vs low-emitting households 36","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**248** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Figure 5.8. Various household characteristics shape carbon footprints**\n\nHousehold characteristics at different emission levels (see figure note)\n\nNote: Ratios compare socio-economic characteristics between high-emitting households (the top 10% of emissions) and low-emitting households (the bottom 10% of emissions) as follows. Number of earners and number of children in the household: Ratio of average numbers of earners and children per household; Other categories: Ratio of number of households in a rural area, or headed by a male, by a person with completed tertiary education, or by a pensioner. Urban/rural indicators are not available from the Mexican and Turkish household budget surveys. Reading, using the example of Germany: The average number of earners in the top 10% emitting households is 4.3 times that in the bottom 10% emitting households (note that the reason for this large relative difference is that almost 80% of the bottom 10% emitting households in Germany had no earner – and the average number of earners in the top 10% emitting households is 1.4). Reading is equivalent for “number of children”. For the other categories, ratios relate to the number of households in the top 10% and bottom 10% emitting households: the number of households located in rural areas, and the number headed by a male, by somebody with tertiary education, and by a pensioner. In other (less technical) terms, in Germany, high-emitting households were three times more likely to live in a rural area than low-emitting households. Source: OECD calculations using IEA emissions factors for different fuels, World Input-Output Database (WIOD) as well as household budget surveys (2015 for EU countries, 2016 for Mexico, 2019 for Türkiye).\n\nStatLink 2 https://stat.link/tfzmcx\n\n**_5\\.3.3. Policy changes: Who paid for recent carbon pricing measures?_**\n\nHouseholds’ carbon footprints are a primary determinant of carbon pricing burdens. But they are not the only one, as real-world carbon pricing measures do not apply uniformly, and not all emissions are therefore priced equally. For instance, excise taxes, carbon taxes and emissions trading systems can, and often do, vary substantially between industry and fuel type. For instance, in the 72 countries covered by the OECD ECR database, the road transport sector faces the highest carbon rates (rates above EUR 60 and EUR 120 per tonne of CO2 mostly occur in that sector), followed by the electricity and off-road transport sectors. In the industry and building sectors 72% and 64%, respectively, of emissions remain unpriced, while almost three-quarters of emissions in the electricity sector face a positive carbon price (OECD, 2023[8]). The correspondence between household emissions and their carbon price burden is thus neither perfect nor straightforward, and depends on the specific design of carbon pricing measures.\n\nThe period between 2012 and 2021 saw considerable policy innovation in the area of carbon pricing. While carbon prices mostly fall far short of those needed for meeting key mitigation commitments, OECD countries often raised them significantly over the period. Initial carbon prices in 2012 and the pace of subsequent changes varied, however, as did policy levers (Section 5.1 and Figure 5.2 above). Elgouacem et al. (forthcoming[14]) gives further details on specific reforms in the five countries. In four of the five countries included in the empirical analysis effective carbon rates increased in (constant) Euro terms – in most of them substantially so. ETS prices and coverage increased in all EU countries, approximately doubling ECRs in\n\n0\n\n1\n\n2\n\n3\n\n4\n\n5\n\n6\n\n7\n\n8\n\n9\n\n10\n\nNumber of earners Number of children Male-headed Head has completed\n\ntertiary education\n\nPensioner-headed Rural area\n\nGermany France Poland Mexico Türkiye\n\nCharacteristics of high-emitting vs low-emitting households 36","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p250.pdf"]},"response":{"errors":[],"extract_id":"extract_62a9dc92c3dafc4ad5c35a4cb29949df","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**248** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Figure 5.8. Various household characteristics shape carbon footprints**\n\nHousehold characteristics at different emission levels (see figure note)\n\nNote: Ratios compare socio-economic characteristics between high-emitting households (the top 10% of emissions) and low-emitting households (the bottom 10% of emissions) as follows. Number of earners and number of children in the household: Ratio of average numbers of earners and children per household; Other categories: Ratio of number of households in a rural area, or headed by a male, by a person with completed tertiary education, or by a pensioner. Urban/rural indicators are not available from the Mexican and Turkish household budget surveys. Reading, using the example of Germany: The average number of earners in the top 10% emitting households is 4.3 times that in the bottom 10% emitting households (note that the reason for this large relative difference is that almost 80% of the bottom 10% emitting households in Germany had no earner – and the average number of earners in the top 10% emitting households is 1.4). Reading is equivalent for “number of children”. For the other categories, ratios relate to the number of households in the top 10% and bottom 10% emitting households: the number of households located in rural areas, and the number headed by a male, by somebody with tertiary education, and by a pensioner. In other (less technical) terms, in Germany, high-emitting households were three times more likely to live in a rural area than low-emitting households. Source: OECD calculations using IEA emissions factors for different fuels, World Input-Output Database (WIOD) as well as household budget surveys (2015 for EU countries, 2016 for Mexico, 2019 for Türkiye).\n\nStatLink 2 https://stat.link/tfzmcx\n\n**_5\\.3.3. Policy changes: Who paid for recent carbon pricing measures?_**\n\nHouseholds’ carbon footprints are a primary determinant of carbon pricing burdens. But they are not the only one, as real-world carbon pricing measures do not apply uniformly, and not all emissions are therefore priced equally. For instance, excise taxes, carbon taxes and emissions trading systems can, and often do, vary substantially between industry and fuel type. For instance, in the 72 countries covered by the OECD ECR database, the road transport sector faces the highest carbon rates (rates above EUR 60 and EUR 120 per tonne of CO2 mostly occur in that sector), followed by the electricity and off-road transport sectors. In the industry and building sectors 72% and 64%, respectively, of emissions remain unpriced, while almost three-quarters of emissions in the electricity sector face a positive carbon price (OECD, 2023[8]). The correspondence between household emissions and their carbon price burden is thus neither perfect nor straightforward, and depends on the specific design of carbon pricing measures.\n\nThe period between 2012 and 2021 saw considerable policy innovation in the area of carbon pricing. While carbon prices mostly fall far short of those needed for meeting key mitigation commitments, OECD countries often raised them significantly over the period. Initial carbon prices in 2012 and the pace of subsequent changes varied, however, as did policy levers (Section 5.1 and Figure 5.2 above). Elgouacem et al. (forthcoming[14]) gives further details on specific reforms in the five countries. In four of the five countries included in the empirical analysis effective carbon rates increased in (constant) Euro terms – in most of them substantially so. ETS prices and coverage increased in all EU countries, approximately doubling ECRs in\n\n0\n\n1\n\n2\n\n3\n\n4\n\n5\n\n6\n\n7\n\n8\n\n9\n\n10\n\nNumber of earners Number of children Male-headed Head has completed\n\ntertiary education\n\nPensioner-headed Rural area\n\nGermany France Poland Mexico Türkiye\n\nCharacteristics of high-emitting vs low-emitting households 36"],"full_content":"**248** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Figure 5.8. Various household characteristics shape carbon footprints**\n\nHousehold characteristics at different emission levels (see figure note)\n\nNote: Ratios compare socio-economic characteristics between high-emitting households (the top 10% of emissions) and low-emitting households (the bottom 10% of emissions) as follows. Number of earners and number of children in the household: Ratio of average numbers of earners and children per household; Other categories: Ratio of number of households in a rural area, or headed by a male, by a person with completed tertiary education, or by a pensioner. Urban/rural indicators are not available from the Mexican and Turkish household budget surveys. Reading, using the example of Germany: The average number of earners in the top 10% emitting households is 4.3 times that in the bottom 10% emitting households (note that the reason for this large relative difference is that almost 80% of the bottom 10% emitting households in Germany had no earner – and the average number of earners in the top 10% emitting households is 1.4). Reading is equivalent for “number of children”. For the other categories, ratios relate to the number of households in the top 10% and bottom 10% emitting households: the number of households located in rural areas, and the number headed by a male, by somebody with tertiary education, and by a pensioner. In other (less technical) terms, in Germany, high-emitting households were three times more likely to live in a rural area than low-emitting households. Source: OECD calculations using IEA emissions factors for different fuels, World Input-Output Database (WIOD) as well as household budget surveys (2015 for EU countries, 2016 for Mexico, 2019 for Türkiye).\n\nStatLink 2 https://stat.link/tfzmcx\n\n**_5\\.3.3. Policy changes: Who paid for recent carbon pricing measures?_**\n\nHouseholds’ carbon footprints are a primary determinant of carbon pricing burdens. But they are not the only one, as real-world carbon pricing measures do not apply uniformly, and not all emissions are therefore priced equally. For instance, excise taxes, carbon taxes and emissions trading systems can, and often do, vary substantially between industry and fuel type. For instance, in the 72 countries covered by the OECD ECR database, the road transport sector faces the highest carbon rates (rates above EUR 60 and EUR 120 per tonne of CO2 mostly occur in that sector), followed by the electricity and off-road transport sectors. In the industry and building sectors 72% and 64%, respectively, of emissions remain unpriced, while almost three-quarters of emissions in the electricity sector face a positive carbon price (OECD, 2023[8]). The correspondence between household emissions and their carbon price burden is thus neither perfect nor straightforward, and depends on the specific design of carbon pricing measures.\n\nThe period between 2012 and 2021 saw considerable policy innovation in the area of carbon pricing. While carbon prices mostly fall far short of those needed for meeting key mitigation commitments, OECD countries often raised them significantly over the period. Initial carbon prices in 2012 and the pace of subsequent changes varied, however, as did policy levers (Section 5.1 and Figure 5.2 above). Elgouacem et al. (forthcoming[14]) gives further details on specific reforms in the five countries. In four of the five countries included in the empirical analysis effective carbon rates increased in (constant) Euro terms – in most of them substantially so. ETS prices and coverage increased in all EU countries, approximately doubling ECRs in\n\n0\n\n1\n\n2\n\n3\n\n4\n\n5\n\n6\n\n7\n\n8\n\n9\n\n10\n\nNumber of earners Number of children Male-headed Head has completed\n\ntertiary education\n\nPensioner-headed Rural area\n\nGermany France Poland Mexico Türkiye\n\nCharacteristics of high-emitting vs low-emitting households 36","publish_date":null,"title":"5.3.3. Policy changes: Who paid for recent carbon pricing measures?","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p250.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p250.pdf"},"request":{"example_id":"chart/ac8b3538-en_p250","product_type":"parse","source_file_path":"ac8b3538-en_p250.pdf"},"started_at":"2026-11-04T19:45:46.218826"} +{"completed_at":"2026-11-04T19:45:49.808923","latency_in_ms":1880,"output":{"example_id":"chart/ac8b3538-en_p252","grounded_pages":[],"layout_pages":[],"markdown":"**250** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\nOverall, household burdens depend strongly on the specific patterns of carbon pricing across sectors. There is a huge spread of prices in practice, both in any given year, and when considering carbon price changes over time (OECD, 2023[8]). This chapter’s mapping from sectors to consumption goods captures this heterogeneity. By contrast, reform assessments based on simple “stylised” carbon pricing scenarios, such as a uniform rate applying to all emissions, are likely to miss key aspects of the distributional impacts of real-world policy trajectories. Moreover, as illustrated in some of the results, concurrent or sequential carbon pricing reforms in different policy areas can produce complex distributional effects, amplifying, or offsetting each other in terms of their impact on households.\n\n### **Figure 5.9. Carbon pricing measures during the past decade: Burdens for households**\n\nPercentage of disposable income, 2012-21\n\nNote: Change in the cost of household-specific consumption baskets, as a share of disposable household incomes, taking consumption baskets (2015 in EU countries, 2016 in Mexico and 2019 in Türkiye) and the fuel mix and carbon intensity of consumption (using the 2016 vintage of the environmentally extended World Input-Output Database – WIOD) as fixed. The changes in carbon pricing burdens are thus only driven by changes in carbon pricing and neither by changes in consumption behaviour nor by changes in the fuel mix and carbon intensity of consumption (see Section 5.2 and Box 5.3 for more details on this assumption). Effective carbon rates for 2012 and 2021 are converted to real terms to match the corresponding household budget survey vintage. Averages by income decile (equivalised disposable household income). For the average household, burdens from carbon pricing reforms were as follows. France: +0.53% of total household income, Germany: +0.50%, Mexico: +1.03%, Poland: +2.34%, Türkiye: +0.09%. ETS: Emission Trading Systems. Changes are computed against the status quo, and do not account for the distributional impacts of inaction. Source: OECD calculations using OECD Effective Carbon Rates data, IEA emissions factors for different fuels, household budget surveys (2015 for EU countries, 2016 for Mexico, 2019 for Türkiye) and World Input-Output Database (WIOD).\n\nStatLink 2 https://stat.link/cx409a\n\nThe welfare impact of carbon price changes depends not only on the carbon price, but also on the growth of labour earnings, which may inform the rate of growth of individual and, hence, household labour income. ²⁷ Indeed, if carbon prices increase at a faster (resp. lower) rate than earnings, then purchasing power (conditional on interactions with other policies such as taxation and benefits) is likely to fall (resp. increase). This can also help inform the impact of carbon prices on the real value of wages. While combining earnings changes and carbon price changes at the individual level is beyond the scope of the analysis conducted in this chapter, Box 5.4 presents some results when comparing the distribution of carbon price changes relative to the mean sectoral labour earnings growth (as proxied by labour cost growth) over the 2012 to 2021 period.\n\n\\-1\n\n0\n\n1\n\n2\n\n3\n\n4\n\n5\n\n1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10\n\nFrance Germany Mexico Poland Türkiye\n\n%\n\nIncome decile\n\nETS Carbon tax Fuel excise taxes Total","pages":[{"line_numbers":[],"markdown":"**250** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\nOverall, household burdens depend strongly on the specific patterns of carbon pricing across sectors. There is a huge spread of prices in practice, both in any given year, and when considering carbon price changes over time (OECD, 2023[8]). This chapter’s mapping from sectors to consumption goods captures this heterogeneity. By contrast, reform assessments based on simple “stylised” carbon pricing scenarios, such as a uniform rate applying to all emissions, are likely to miss key aspects of the distributional impacts of real-world policy trajectories. Moreover, as illustrated in some of the results, concurrent or sequential carbon pricing reforms in different policy areas can produce complex distributional effects, amplifying, or offsetting each other in terms of their impact on households.\n\n### **Figure 5.9. Carbon pricing measures during the past decade: Burdens for households**\n\nPercentage of disposable income, 2012-21\n\nNote: Change in the cost of household-specific consumption baskets, as a share of disposable household incomes, taking consumption baskets (2015 in EU countries, 2016 in Mexico and 2019 in Türkiye) and the fuel mix and carbon intensity of consumption (using the 2016 vintage of the environmentally extended World Input-Output Database – WIOD) as fixed. The changes in carbon pricing burdens are thus only driven by changes in carbon pricing and neither by changes in consumption behaviour nor by changes in the fuel mix and carbon intensity of consumption (see Section 5.2 and Box 5.3 for more details on this assumption). Effective carbon rates for 2012 and 2021 are converted to real terms to match the corresponding household budget survey vintage. Averages by income decile (equivalised disposable household income). For the average household, burdens from carbon pricing reforms were as follows. France: +0.53% of total household income, Germany: +0.50%, Mexico: +1.03%, Poland: +2.34%, Türkiye: +0.09%. ETS: Emission Trading Systems. Changes are computed against the status quo, and do not account for the distributional impacts of inaction. Source: OECD calculations using OECD Effective Carbon Rates data, IEA emissions factors for different fuels, household budget surveys (2015 for EU countries, 2016 for Mexico, 2019 for Türkiye) and World Input-Output Database (WIOD).\n\nStatLink 2 https://stat.link/cx409a\n\nThe welfare impact of carbon price changes depends not only on the carbon price, but also on the growth of labour earnings, which may inform the rate of growth of individual and, hence, household labour income. ²⁷ Indeed, if carbon prices increase at a faster (resp. lower) rate than earnings, then purchasing power (conditional on interactions with other policies such as taxation and benefits) is likely to fall (resp. increase). This can also help inform the impact of carbon prices on the real value of wages. While combining earnings changes and carbon price changes at the individual level is beyond the scope of the analysis conducted in this chapter, Box 5.4 presents some results when comparing the distribution of carbon price changes relative to the mean sectoral labour earnings growth (as proxied by labour cost growth) over the 2012 to 2021 period.\n\n\\-1\n\n0\n\n1\n\n2\n\n3\n\n4\n\n5\n\n1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10\n\nFrance Germany Mexico Poland Türkiye\n\n%\n\nIncome decile\n\nETS Carbon tax Fuel excise taxes Total","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**250** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\nOverall, household burdens depend strongly on the specific patterns of carbon pricing across sectors. There is a huge spread of prices in practice, both in any given year, and when considering carbon price changes over time (OECD, 2023[8]). This chapter’s mapping from sectors to consumption goods captures this heterogeneity. By contrast, reform assessments based on simple “stylised” carbon pricing scenarios, such as a uniform rate applying to all emissions, are likely to miss key aspects of the distributional impacts of real-world policy trajectories. Moreover, as illustrated in some of the results, concurrent or sequential carbon pricing reforms in different policy areas can produce complex distributional effects, amplifying, or offsetting each other in terms of their impact on households.\n\n### **Figure 5.9. Carbon pricing measures during the past decade: Burdens for households**\n\nPercentage of disposable income, 2012-21\n\nNote: Change in the cost of household-specific consumption baskets, as a share of disposable household incomes, taking consumption baskets (2015 in EU countries, 2016 in Mexico and 2019 in Türkiye) and the fuel mix and carbon intensity of consumption (using the 2016 vintage of the environmentally extended World Input-Output Database – WIOD) as fixed. The changes in carbon pricing burdens are thus only driven by changes in carbon pricing and neither by changes in consumption behaviour nor by changes in the fuel mix and carbon intensity of consumption (see Section 5.2 and Box 5.3 for more details on this assumption). Effective carbon rates for 2012 and 2021 are converted to real terms to match the corresponding household budget survey vintage. Averages by income decile (equivalised disposable household income). For the average household, burdens from carbon pricing reforms were as follows. France: +0.53% of total household income, Germany: +0.50%, Mexico: +1.03%, Poland: +2.34%, Türkiye: +0.09%. ETS: Emission Trading Systems. Changes are computed against the status quo, and do not account for the distributional impacts of inaction. Source: OECD calculations using OECD Effective Carbon Rates data, IEA emissions factors for different fuels, household budget surveys (2015 for EU countries, 2016 for Mexico, 2019 for Türkiye) and World Input-Output Database (WIOD).\n\nStatLink 2 https://stat.link/cx409a\n\nThe welfare impact of carbon price changes depends not only on the carbon price, but also on the growth of labour earnings, which may inform the rate of growth of individual and, hence, household labour income. ²⁷ Indeed, if carbon prices increase at a faster (resp. lower) rate than earnings, then purchasing power (conditional on interactions with other policies such as taxation and benefits) is likely to fall (resp. increase). This can also help inform the impact of carbon prices on the real value of wages. While combining earnings changes and carbon price changes at the individual level is beyond the scope of the analysis conducted in this chapter, Box 5.4 presents some results when comparing the distribution of carbon price changes relative to the mean sectoral labour earnings growth (as proxied by labour cost growth) over the 2012 to 2021 period.\n\n\\-1\n\n0\n\n1\n\n2\n\n3\n\n4\n\n5\n\n1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10\n\nFrance Germany Mexico Poland Türkiye\n\n%\n\nIncome decile\n\nETS Carbon tax Fuel excise taxes Total","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p252.pdf"]},"response":{"errors":[],"extract_id":"extract_7818293d44dd5dec4c2e0f34093b37b7","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**250** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\nOverall, household burdens depend strongly on the specific patterns of carbon pricing across sectors. There is a huge spread of prices in practice, both in any given year, and when considering carbon price changes over time (OECD, 2023[8]). This chapter’s mapping from sectors to consumption goods captures this heterogeneity. By contrast, reform assessments based on simple “stylised” carbon pricing scenarios, such as a uniform rate applying to all emissions, are likely to miss key aspects of the distributional impacts of real-world policy trajectories. Moreover, as illustrated in some of the results, concurrent or sequential carbon pricing reforms in different policy areas can produce complex distributional effects, amplifying, or offsetting each other in terms of their impact on households.\n\n### **Figure 5.9. Carbon pricing measures during the past decade: Burdens for households**\n\nPercentage of disposable income, 2012-21\n\nNote: Change in the cost of household-specific consumption baskets, as a share of disposable household incomes, taking consumption baskets (2015 in EU countries, 2016 in Mexico and 2019 in Türkiye) and the fuel mix and carbon intensity of consumption (using the 2016 vintage of the environmentally extended World Input-Output Database – WIOD) as fixed. The changes in carbon pricing burdens are thus only driven by changes in carbon pricing and neither by changes in consumption behaviour nor by changes in the fuel mix and carbon intensity of consumption (see Section 5.2 and Box 5.3 for more details on this assumption). Effective carbon rates for 2012 and 2021 are converted to real terms to match the corresponding household budget survey vintage. Averages by income decile (equivalised disposable household income). For the average household, burdens from carbon pricing reforms were as follows. France: +0.53% of total household income, Germany: +0.50%, Mexico: +1.03%, Poland: +2.34%, Türkiye: +0.09%. ETS: Emission Trading Systems. Changes are computed against the status quo, and do not account for the distributional impacts of inaction. Source: OECD calculations using OECD Effective Carbon Rates data, IEA emissions factors for different fuels, household budget surveys (2015 for EU countries, 2016 for Mexico, 2019 for Türkiye) and World Input-Output Database (WIOD).\n\nStatLink 2 https://stat.link/cx409a\n\nThe welfare impact of carbon price changes depends not only on the carbon price, but also on the growth of labour earnings, which may inform the rate of growth of individual and, hence, household labour income. ²⁷ Indeed, if carbon prices increase at a faster (resp. lower) rate than earnings, then purchasing power (conditional on interactions with other policies such as taxation and benefits) is likely to fall (resp. increase). This can also help inform the impact of carbon prices on the real value of wages. While combining earnings changes and carbon price changes at the individual level is beyond the scope of the analysis conducted in this chapter, Box 5.4 presents some results when comparing the distribution of carbon price changes relative to the mean sectoral labour earnings growth (as proxied by labour cost growth) over the 2012 to 2021 period.\n\n\\-1\n\n0\n\n1\n\n2\n\n3\n\n4\n\n5\n\n1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10\n\nFrance Germany Mexico Poland Türkiye\n\n%\n\nIncome decile\n\nETS Carbon tax Fuel excise taxes Total"],"full_content":"**250** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\nOverall, household burdens depend strongly on the specific patterns of carbon pricing across sectors. There is a huge spread of prices in practice, both in any given year, and when considering carbon price changes over time (OECD, 2023[8]). This chapter’s mapping from sectors to consumption goods captures this heterogeneity. By contrast, reform assessments based on simple “stylised” carbon pricing scenarios, such as a uniform rate applying to all emissions, are likely to miss key aspects of the distributional impacts of real-world policy trajectories. Moreover, as illustrated in some of the results, concurrent or sequential carbon pricing reforms in different policy areas can produce complex distributional effects, amplifying, or offsetting each other in terms of their impact on households.\n\n### **Figure 5.9. Carbon pricing measures during the past decade: Burdens for households**\n\nPercentage of disposable income, 2012-21\n\nNote: Change in the cost of household-specific consumption baskets, as a share of disposable household incomes, taking consumption baskets (2015 in EU countries, 2016 in Mexico and 2019 in Türkiye) and the fuel mix and carbon intensity of consumption (using the 2016 vintage of the environmentally extended World Input-Output Database – WIOD) as fixed. The changes in carbon pricing burdens are thus only driven by changes in carbon pricing and neither by changes in consumption behaviour nor by changes in the fuel mix and carbon intensity of consumption (see Section 5.2 and Box 5.3 for more details on this assumption). Effective carbon rates for 2012 and 2021 are converted to real terms to match the corresponding household budget survey vintage. Averages by income decile (equivalised disposable household income). For the average household, burdens from carbon pricing reforms were as follows. France: +0.53% of total household income, Germany: +0.50%, Mexico: +1.03%, Poland: +2.34%, Türkiye: +0.09%. ETS: Emission Trading Systems. Changes are computed against the status quo, and do not account for the distributional impacts of inaction. Source: OECD calculations using OECD Effective Carbon Rates data, IEA emissions factors for different fuels, household budget surveys (2015 for EU countries, 2016 for Mexico, 2019 for Türkiye) and World Input-Output Database (WIOD).\n\nStatLink 2 https://stat.link/cx409a\n\nThe welfare impact of carbon price changes depends not only on the carbon price, but also on the growth of labour earnings, which may inform the rate of growth of individual and, hence, household labour income. ²⁷ Indeed, if carbon prices increase at a faster (resp. lower) rate than earnings, then purchasing power (conditional on interactions with other policies such as taxation and benefits) is likely to fall (resp. increase). This can also help inform the impact of carbon prices on the real value of wages. While combining earnings changes and carbon price changes at the individual level is beyond the scope of the analysis conducted in this chapter, Box 5.4 presents some results when comparing the distribution of carbon price changes relative to the mean sectoral labour earnings growth (as proxied by labour cost growth) over the 2012 to 2021 period.\n\n\\-1\n\n0\n\n1\n\n2\n\n3\n\n4\n\n5\n\n1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10\n\nFrance Germany Mexico Poland Türkiye\n\n%\n\nIncome decile\n\nETS Carbon tax Fuel excise taxes Total","publish_date":null,"title":"Figure 5.9. Carbon pricing measures during the past decade: Burdens for households","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p252.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p252.pdf"},"request":{"example_id":"chart/ac8b3538-en_p252","product_type":"parse","source_file_path":"ac8b3538-en_p252.pdf"},"started_at":"2026-11-04T19:45:47.928224"} +{"completed_at":"2026-11-04T19:45:50.219752","latency_in_ms":1347,"output":{"example_id":"chart/ac8b3538-en_p253","grounded_pages":[],"layout_pages":[],"markdown":" **251**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Box 5.4. Evolution of labour costs and carbon prices**\n\nFigure 5.10 below reports the growth rate of (nominal) effective carbon rates (ECRs) for households compared with inflation (the growth rate of prices, using the consumer price index – CPI) and labour costs (in the sectors with the highest and lowest growth rates). For almost all countries, carbon prices in households’ budgets grew faster than inflation for all income deciles, indicating the increasing share of purchasing power subject to carbon prices. However, average labour costs also grew over that period, and at a rate faster than inflation. While the data used in the analysis for this chapter prevents the disaggregation of labour cost changes for individual households, Eurostat labour cost data enables the disaggregation of labour costs by (broadly-defined) sector, which enables the calculation of growth rates of labour costs for the sector with the highest increase and the sector with the lowest increase. For Germany, France and Poland, the growth of ECRs in household budgets was higher than labour costs growth in the sector with the highest growth rate. Türkiye experienced the opposite, with the ECR growth rate for households lower than the labour costs growth rate in the highest growth sector. This indicates that in France, Germany and Poland, the welfare effects of increase in carbon prices possibly tended to be negative (i.e. earnings increased less than prices), while welfare effects in Türkiye were possibly positive for individuals in the sectors with the highest earnings growth.\n\n### **Figure 5.10. Growth rates of carbon prices (ECR), of the consumer price index and of labour costs 2012-21, by country**\n\nNote: Mexico does not appear in the graph for lack of comparable labour costs evolution data. This figure reports the average labour cost growth of the lowest growing sector, the highest growing sector, the average CPI growth, and the relative change in nominal ECRs between 2012 and 2021 by household (equivalised disposable) income decile. ECR: Effective Carbon Rates. LC: Labour Cost. CPI: Consumer Price Index. Reading: for instance, for the first decile in Germany, the relative change in nominal ECRs is the difference between the average ECR paid by households in the first decile in Germany in 2021 and in 2012 divided by the average ECR paid by households in Germany in 2012. Source: Eurostat labour costs data, https://ec.europa.eu/eurostat/web/labour-market/information-data/labour-costs and OECD calculations using OECD Effective Carbon Rates data, IEA emissions factors for different fuels, household budget surveys (2015 for EU countries and 2019 for Türkiye) and World Input-Output Database (WIOD)\n\nStatLink 2 https://stat.link/z9e5kr\n\n0\n\n0\\.5\n\n1\n\n1\\.5\n\n2\n\n2\\.5\n\n3\n\n3\\.5\n\n4\n\n1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10\n\nGermany France Poland Türkiye\n\n% change 2012-21\n\nEquivalised disposable income decile\n\nECR(Direct) ECR (Indirect) LC: lowest growing LC: highest growing CPI","pages":[{"line_numbers":[],"markdown":" **251**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Box 5.4. Evolution of labour costs and carbon prices**\n\nFigure 5.10 below reports the growth rate of (nominal) effective carbon rates (ECRs) for households compared with inflation (the growth rate of prices, using the consumer price index – CPI) and labour costs (in the sectors with the highest and lowest growth rates). For almost all countries, carbon prices in households’ budgets grew faster than inflation for all income deciles, indicating the increasing share of purchasing power subject to carbon prices. However, average labour costs also grew over that period, and at a rate faster than inflation. While the data used in the analysis for this chapter prevents the disaggregation of labour cost changes for individual households, Eurostat labour cost data enables the disaggregation of labour costs by (broadly-defined) sector, which enables the calculation of growth rates of labour costs for the sector with the highest increase and the sector with the lowest increase. For Germany, France and Poland, the growth of ECRs in household budgets was higher than labour costs growth in the sector with the highest growth rate. Türkiye experienced the opposite, with the ECR growth rate for households lower than the labour costs growth rate in the highest growth sector. This indicates that in France, Germany and Poland, the welfare effects of increase in carbon prices possibly tended to be negative (i.e. earnings increased less than prices), while welfare effects in Türkiye were possibly positive for individuals in the sectors with the highest earnings growth.\n\n### **Figure 5.10. Growth rates of carbon prices (ECR), of the consumer price index and of labour costs 2012-21, by country**\n\nNote: Mexico does not appear in the graph for lack of comparable labour costs evolution data. This figure reports the average labour cost growth of the lowest growing sector, the highest growing sector, the average CPI growth, and the relative change in nominal ECRs between 2012 and 2021 by household (equivalised disposable) income decile. ECR: Effective Carbon Rates. LC: Labour Cost. CPI: Consumer Price Index. Reading: for instance, for the first decile in Germany, the relative change in nominal ECRs is the difference between the average ECR paid by households in the first decile in Germany in 2021 and in 2012 divided by the average ECR paid by households in Germany in 2012. Source: Eurostat labour costs data, https://ec.europa.eu/eurostat/web/labour-market/information-data/labour-costs and OECD calculations using OECD Effective Carbon Rates data, IEA emissions factors for different fuels, household budget surveys (2015 for EU countries and 2019 for Türkiye) and World Input-Output Database (WIOD)\n\nStatLink 2 https://stat.link/z9e5kr\n\n0\n\n0\\.5\n\n1\n\n1\\.5\n\n2\n\n2\\.5\n\n3\n\n3\\.5\n\n4\n\n1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10\n\nGermany France Poland Türkiye\n\n% change 2012-21\n\nEquivalised disposable income decile\n\nECR(Direct) ECR (Indirect) LC: lowest growing LC: highest growing CPI","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":" **251**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Box 5.4. Evolution of labour costs and carbon prices**\n\nFigure 5.10 below reports the growth rate of (nominal) effective carbon rates (ECRs) for households compared with inflation (the growth rate of prices, using the consumer price index – CPI) and labour costs (in the sectors with the highest and lowest growth rates). For almost all countries, carbon prices in households’ budgets grew faster than inflation for all income deciles, indicating the increasing share of purchasing power subject to carbon prices. However, average labour costs also grew over that period, and at a rate faster than inflation. While the data used in the analysis for this chapter prevents the disaggregation of labour cost changes for individual households, Eurostat labour cost data enables the disaggregation of labour costs by (broadly-defined) sector, which enables the calculation of growth rates of labour costs for the sector with the highest increase and the sector with the lowest increase. For Germany, France and Poland, the growth of ECRs in household budgets was higher than labour costs growth in the sector with the highest growth rate. Türkiye experienced the opposite, with the ECR growth rate for households lower than the labour costs growth rate in the highest growth sector. This indicates that in France, Germany and Poland, the welfare effects of increase in carbon prices possibly tended to be negative (i.e. earnings increased less than prices), while welfare effects in Türkiye were possibly positive for individuals in the sectors with the highest earnings growth.\n\n### **Figure 5.10. Growth rates of carbon prices (ECR), of the consumer price index and of labour costs 2012-21, by country**\n\nNote: Mexico does not appear in the graph for lack of comparable labour costs evolution data. This figure reports the average labour cost growth of the lowest growing sector, the highest growing sector, the average CPI growth, and the relative change in nominal ECRs between 2012 and 2021 by household (equivalised disposable) income decile. ECR: Effective Carbon Rates. LC: Labour Cost. CPI: Consumer Price Index. Reading: for instance, for the first decile in Germany, the relative change in nominal ECRs is the difference between the average ECR paid by households in the first decile in Germany in 2021 and in 2012 divided by the average ECR paid by households in Germany in 2012. Source: Eurostat labour costs data, https://ec.europa.eu/eurostat/web/labour-market/information-data/labour-costs and OECD calculations using OECD Effective Carbon Rates data, IEA emissions factors for different fuels, household budget surveys (2015 for EU countries and 2019 for Türkiye) and World Input-Output Database (WIOD)\n\nStatLink 2 https://stat.link/z9e5kr\n\n0\n\n0\\.5\n\n1\n\n1\\.5\n\n2\n\n2\\.5\n\n3\n\n3\\.5\n\n4\n\n1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10\n\nGermany France Poland Türkiye\n\n% change 2012-21\n\nEquivalised disposable income decile\n\nECR(Direct) ECR (Indirect) LC: lowest growing LC: highest growing CPI","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p253.pdf"]},"response":{"errors":[],"extract_id":"extract_c447f2bf8f87e44e550d045be8454131","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":[" **251**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Box 5.4. Evolution of labour costs and carbon prices**\n\nFigure 5.10 below reports the growth rate of (nominal) effective carbon rates (ECRs) for households compared with inflation (the growth rate of prices, using the consumer price index – CPI) and labour costs (in the sectors with the highest and lowest growth rates). For almost all countries, carbon prices in households’ budgets grew faster than inflation for all income deciles, indicating the increasing share of purchasing power subject to carbon prices. However, average labour costs also grew over that period, and at a rate faster than inflation. While the data used in the analysis for this chapter prevents the disaggregation of labour cost changes for individual households, Eurostat labour cost data enables the disaggregation of labour costs by (broadly-defined) sector, which enables the calculation of growth rates of labour costs for the sector with the highest increase and the sector with the lowest increase. For Germany, France and Poland, the growth of ECRs in household budgets was higher than labour costs growth in the sector with the highest growth rate. Türkiye experienced the opposite, with the ECR growth rate for households lower than the labour costs growth rate in the highest growth sector. This indicates that in France, Germany and Poland, the welfare effects of increase in carbon prices possibly tended to be negative (i.e. earnings increased less than prices), while welfare effects in Türkiye were possibly positive for individuals in the sectors with the highest earnings growth.\n\n### **Figure 5.10. Growth rates of carbon prices (ECR), of the consumer price index and of labour costs 2012-21, by country**\n\nNote: Mexico does not appear in the graph for lack of comparable labour costs evolution data. This figure reports the average labour cost growth of the lowest growing sector, the highest growing sector, the average CPI growth, and the relative change in nominal ECRs between 2012 and 2021 by household (equivalised disposable) income decile. ECR: Effective Carbon Rates. LC: Labour Cost. CPI: Consumer Price Index. Reading: for instance, for the first decile in Germany, the relative change in nominal ECRs is the difference between the average ECR paid by households in the first decile in Germany in 2021 and in 2012 divided by the average ECR paid by households in Germany in 2012. Source: Eurostat labour costs data, https://ec.europa.eu/eurostat/web/labour-market/information-data/labour-costs and OECD calculations using OECD Effective Carbon Rates data, IEA emissions factors for different fuels, household budget surveys (2015 for EU countries and 2019 for Türkiye) and World Input-Output Database (WIOD)\n\nStatLink 2 https://stat.link/z9e5kr\n\n0\n\n0\\.5\n\n1\n\n1\\.5\n\n2\n\n2\\.5\n\n3\n\n3\\.5\n\n4\n\n1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10\n\nGermany France Poland Türkiye\n\n% change 2012-21\n\nEquivalised disposable income decile\n\nECR(Direct) ECR (Indirect) LC: lowest growing LC: highest growing CPI"],"full_content":" **251**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Box 5.4. Evolution of labour costs and carbon prices**\n\nFigure 5.10 below reports the growth rate of (nominal) effective carbon rates (ECRs) for households compared with inflation (the growth rate of prices, using the consumer price index – CPI) and labour costs (in the sectors with the highest and lowest growth rates). For almost all countries, carbon prices in households’ budgets grew faster than inflation for all income deciles, indicating the increasing share of purchasing power subject to carbon prices. However, average labour costs also grew over that period, and at a rate faster than inflation. While the data used in the analysis for this chapter prevents the disaggregation of labour cost changes for individual households, Eurostat labour cost data enables the disaggregation of labour costs by (broadly-defined) sector, which enables the calculation of growth rates of labour costs for the sector with the highest increase and the sector with the lowest increase. For Germany, France and Poland, the growth of ECRs in household budgets was higher than labour costs growth in the sector with the highest growth rate. Türkiye experienced the opposite, with the ECR growth rate for households lower than the labour costs growth rate in the highest growth sector. This indicates that in France, Germany and Poland, the welfare effects of increase in carbon prices possibly tended to be negative (i.e. earnings increased less than prices), while welfare effects in Türkiye were possibly positive for individuals in the sectors with the highest earnings growth.\n\n### **Figure 5.10. Growth rates of carbon prices (ECR), of the consumer price index and of labour costs 2012-21, by country**\n\nNote: Mexico does not appear in the graph for lack of comparable labour costs evolution data. This figure reports the average labour cost growth of the lowest growing sector, the highest growing sector, the average CPI growth, and the relative change in nominal ECRs between 2012 and 2021 by household (equivalised disposable) income decile. ECR: Effective Carbon Rates. LC: Labour Cost. CPI: Consumer Price Index. Reading: for instance, for the first decile in Germany, the relative change in nominal ECRs is the difference between the average ECR paid by households in the first decile in Germany in 2021 and in 2012 divided by the average ECR paid by households in Germany in 2012. Source: Eurostat labour costs data, https://ec.europa.eu/eurostat/web/labour-market/information-data/labour-costs and OECD calculations using OECD Effective Carbon Rates data, IEA emissions factors for different fuels, household budget surveys (2015 for EU countries and 2019 for Türkiye) and World Input-Output Database (WIOD)\n\nStatLink 2 https://stat.link/z9e5kr\n\n0\n\n0\\.5\n\n1\n\n1\\.5\n\n2\n\n2\\.5\n\n3\n\n3\\.5\n\n4\n\n1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10\n\nGermany France Poland Türkiye\n\n% change 2012-21\n\nEquivalised disposable income decile\n\nECR(Direct) ECR (Indirect) LC: lowest growing LC: highest growing CPI","publish_date":null,"title":"Box 5.4. Evolution of labour costs and carbon prices Figure 5.10. Growth rates of carbon prices (ECR), of the consumer","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p253.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p253.pdf"},"request":{"example_id":"chart/ac8b3538-en_p253","product_type":"parse","source_file_path":"ac8b3538-en_p253.pdf"},"started_at":"2026-11-04T19:45:48.872168"} +{"completed_at":"2026-11-04T19:45:51.462035","latency_in_ms":1649,"output":{"example_id":"chart/ac8b3538-en_p27","grounded_pages":[],"layout_pages":[],"markdown":" **25**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\nIn Q1 2024, hours worked per employed person were below their pre-COVID-19 levels in 20 of the 31 countries with recent data available (Figure 1.5). The average decline in hours worked since Q4 2019 across all countries with available data is rather small – just above 1%. ³ In most countries, this decline follows a trend that pre-dates the COVID-19 crisis, though with some notable accelerations in Austria, Finland, Hungary, Ireland, Korea, the Slovak Republic and Spain. Among the five countries where hours worked had been increasing before the COVID-19 crisis, only the Netherlands, Portugal and the United States, saw a decline of more than 1% in the aftermath of the pandemic.\n\nEvidence for Europe indicates that the decline in hours worked over the past 20 years is largely driven by an increase in part-time and a reduction in hours within jobs (as opposed to a compositional shift to jobs typically requiring fewer hours) (Astinova et al., 2024[5]; ECB, 2021[6]). However, the decline in average hours worked since the COVID-19 crisis has not been associated with widespread up-ticks in part-time employment. On the contrary, annual data for 2022 point to a slight decrease in the incidence of part-time in most OECD countries relative to 2019. ⁴\n\nOverall, the cross-country comparison does not lend support to the hypothesis of a generalised post- COVID change in preferences over work-life balance that might have reduced willingness to work, but more evidence is needed to understand patterns observed in specific countries.\n\n### **Figure 1.5. The post-pandemic decline in average hours worked per worker is generally consistent with long-term trends**\n\nPercentage change, seasonally adjusted data\n\nNote: Average hours worked per worker is defined as the total hours worked divided by total employment, except for **Belgium** where it refers to the average hours worked of employees, **Korea,** where it refers to the average usual weekly hours worked per employed, **New Zealand** , where it is defined as the total paid hours divided by filled jobs, and the **United States** where it refers to the average usual weekly hours worked per wage and salary workers. Statistics are not seasonally adjusted for Canada and Mexico and periods reported for those countries refer to Q4 2004-Q4 2019 and Q4 2004-Q4 2023, and Q4 2005-Q4 2019 and Q4 2005-Q4 2023, respectively. The latest quarter available refers to Q3 2023 for Israel, and to Q4 2023 for Belgium and the United Kingdom. OECD is an unweighted average of the 31 OECD countries shown in this Chart (not including Chile, Colombia, Costa Rica, Iceland, Japan, Switzerland and Türkiye). Euro Area refers to the 20 Eurozone countries. Source: OECD (2024), “Quarterly National Accounts”, _OECD National Accounts Statistics_ (database), https://doi.org/10.1787/data-00017-en (accessed on 21 June 2024) for all countries except Australia, Korea, New Zealand, and the United States; Australian Labour Account (Australian Bureau of Statistics) for Australia, Economically Active Population Survey (Statistics Korea) for Korea, Quarterly Employment Survey (QES – Tables QEM034AA and QEM025AA, Stats NZ) for New Zealand, and Current Population Survey (CPS, Bureau of Labor Statistics) for the United States.\n\nStatLink 2 https://stat.link/kudyaz\n\n\\-25\n\n\\-20\n\n\\-15\n\n\\-10\n\n\\-5\n\n0\n\n5 %\n\nQ1 2005-Q4 2019 Q1 2005-Q1 2024","pages":[{"line_numbers":[],"markdown":" **25**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\nIn Q1 2024, hours worked per employed person were below their pre-COVID-19 levels in 20 of the 31 countries with recent data available (Figure 1.5). The average decline in hours worked since Q4 2019 across all countries with available data is rather small – just above 1%. ³ In most countries, this decline follows a trend that pre-dates the COVID-19 crisis, though with some notable accelerations in Austria, Finland, Hungary, Ireland, Korea, the Slovak Republic and Spain. Among the five countries where hours worked had been increasing before the COVID-19 crisis, only the Netherlands, Portugal and the United States, saw a decline of more than 1% in the aftermath of the pandemic.\n\nEvidence for Europe indicates that the decline in hours worked over the past 20 years is largely driven by an increase in part-time and a reduction in hours within jobs (as opposed to a compositional shift to jobs typically requiring fewer hours) (Astinova et al., 2024[5]; ECB, 2021[6]). However, the decline in average hours worked since the COVID-19 crisis has not been associated with widespread up-ticks in part-time employment. On the contrary, annual data for 2022 point to a slight decrease in the incidence of part-time in most OECD countries relative to 2019. ⁴\n\nOverall, the cross-country comparison does not lend support to the hypothesis of a generalised post- COVID change in preferences over work-life balance that might have reduced willingness to work, but more evidence is needed to understand patterns observed in specific countries.\n\n### **Figure 1.5. The post-pandemic decline in average hours worked per worker is generally consistent with long-term trends**\n\nPercentage change, seasonally adjusted data\n\nNote: Average hours worked per worker is defined as the total hours worked divided by total employment, except for **Belgium** where it refers to the average hours worked of employees, **Korea,** where it refers to the average usual weekly hours worked per employed, **New Zealand** , where it is defined as the total paid hours divided by filled jobs, and the **United States** where it refers to the average usual weekly hours worked per wage and salary workers. Statistics are not seasonally adjusted for Canada and Mexico and periods reported for those countries refer to Q4 2004-Q4 2019 and Q4 2004-Q4 2023, and Q4 2005-Q4 2019 and Q4 2005-Q4 2023, respectively. The latest quarter available refers to Q3 2023 for Israel, and to Q4 2023 for Belgium and the United Kingdom. OECD is an unweighted average of the 31 OECD countries shown in this Chart (not including Chile, Colombia, Costa Rica, Iceland, Japan, Switzerland and Türkiye). Euro Area refers to the 20 Eurozone countries. Source: OECD (2024), “Quarterly National Accounts”, _OECD National Accounts Statistics_ (database), https://doi.org/10.1787/data-00017-en (accessed on 21 June 2024) for all countries except Australia, Korea, New Zealand, and the United States; Australian Labour Account (Australian Bureau of Statistics) for Australia, Economically Active Population Survey (Statistics Korea) for Korea, Quarterly Employment Survey (QES – Tables QEM034AA and QEM025AA, Stats NZ) for New Zealand, and Current Population Survey (CPS, Bureau of Labor Statistics) for the United States.\n\nStatLink 2 https://stat.link/kudyaz\n\n\\-25\n\n\\-20\n\n\\-15\n\n\\-10\n\n\\-5\n\n0\n\n5 %\n\nQ1 2005-Q4 2019 Q1 2005-Q1 2024","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":" **25**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\nIn Q1 2024, hours worked per employed person were below their pre-COVID-19 levels in 20 of the 31 countries with recent data available (Figure 1.5). The average decline in hours worked since Q4 2019 across all countries with available data is rather small – just above 1%. ³ In most countries, this decline follows a trend that pre-dates the COVID-19 crisis, though with some notable accelerations in Austria, Finland, Hungary, Ireland, Korea, the Slovak Republic and Spain. Among the five countries where hours worked had been increasing before the COVID-19 crisis, only the Netherlands, Portugal and the United States, saw a decline of more than 1% in the aftermath of the pandemic.\n\nEvidence for Europe indicates that the decline in hours worked over the past 20 years is largely driven by an increase in part-time and a reduction in hours within jobs (as opposed to a compositional shift to jobs typically requiring fewer hours) (Astinova et al., 2024[5]; ECB, 2021[6]). However, the decline in average hours worked since the COVID-19 crisis has not been associated with widespread up-ticks in part-time employment. On the contrary, annual data for 2022 point to a slight decrease in the incidence of part-time in most OECD countries relative to 2019. ⁴\n\nOverall, the cross-country comparison does not lend support to the hypothesis of a generalised post- COVID change in preferences over work-life balance that might have reduced willingness to work, but more evidence is needed to understand patterns observed in specific countries.\n\n### **Figure 1.5. The post-pandemic decline in average hours worked per worker is generally consistent with long-term trends**\n\nPercentage change, seasonally adjusted data\n\nNote: Average hours worked per worker is defined as the total hours worked divided by total employment, except for **Belgium** where it refers to the average hours worked of employees, **Korea,** where it refers to the average usual weekly hours worked per employed, **New Zealand** , where it is defined as the total paid hours divided by filled jobs, and the **United States** where it refers to the average usual weekly hours worked per wage and salary workers. Statistics are not seasonally adjusted for Canada and Mexico and periods reported for those countries refer to Q4 2004-Q4 2019 and Q4 2004-Q4 2023, and Q4 2005-Q4 2019 and Q4 2005-Q4 2023, respectively. The latest quarter available refers to Q3 2023 for Israel, and to Q4 2023 for Belgium and the United Kingdom. OECD is an unweighted average of the 31 OECD countries shown in this Chart (not including Chile, Colombia, Costa Rica, Iceland, Japan, Switzerland and Türkiye). Euro Area refers to the 20 Eurozone countries. Source: OECD (2024), “Quarterly National Accounts”, _OECD National Accounts Statistics_ (database), https://doi.org/10.1787/data-00017-en (accessed on 21 June 2024) for all countries except Australia, Korea, New Zealand, and the United States; Australian Labour Account (Australian Bureau of Statistics) for Australia, Economically Active Population Survey (Statistics Korea) for Korea, Quarterly Employment Survey (QES – Tables QEM034AA and QEM025AA, Stats NZ) for New Zealand, and Current Population Survey (CPS, Bureau of Labor Statistics) for the United States.\n\nStatLink 2 https://stat.link/kudyaz\n\n\\-25\n\n\\-20\n\n\\-15\n\n\\-10\n\n\\-5\n\n0\n\n5 %\n\nQ1 2005-Q4 2019 Q1 2005-Q1 2024","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p27.pdf"]},"response":{"errors":[],"extract_id":"extract_cf5ff595cbb31be18b6732339db1b1e3","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":[" **25**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\nIn Q1 2024, hours worked per employed person were below their pre-COVID-19 levels in 20 of the 31 countries with recent data available (Figure 1.5). The average decline in hours worked since Q4 2019 across all countries with available data is rather small – just above 1%. ³ In most countries, this decline follows a trend that pre-dates the COVID-19 crisis, though with some notable accelerations in Austria, Finland, Hungary, Ireland, Korea, the Slovak Republic and Spain. Among the five countries where hours worked had been increasing before the COVID-19 crisis, only the Netherlands, Portugal and the United States, saw a decline of more than 1% in the aftermath of the pandemic.\n\nEvidence for Europe indicates that the decline in hours worked over the past 20 years is largely driven by an increase in part-time and a reduction in hours within jobs (as opposed to a compositional shift to jobs typically requiring fewer hours) (Astinova et al., 2024[5]; ECB, 2021[6]). However, the decline in average hours worked since the COVID-19 crisis has not been associated with widespread up-ticks in part-time employment. On the contrary, annual data for 2022 point to a slight decrease in the incidence of part-time in most OECD countries relative to 2019. ⁴\n\nOverall, the cross-country comparison does not lend support to the hypothesis of a generalised post- COVID change in preferences over work-life balance that might have reduced willingness to work, but more evidence is needed to understand patterns observed in specific countries.\n\n### **Figure 1.5. The post-pandemic decline in average hours worked per worker is generally consistent with long-term trends**\n\nPercentage change, seasonally adjusted data\n\nNote: Average hours worked per worker is defined as the total hours worked divided by total employment, except for **Belgium** where it refers to the average hours worked of employees, **Korea,** where it refers to the average usual weekly hours worked per employed, **New Zealand** , where it is defined as the total paid hours divided by filled jobs, and the **United States** where it refers to the average usual weekly hours worked per wage and salary workers. Statistics are not seasonally adjusted for Canada and Mexico and periods reported for those countries refer to Q4 2004-Q4 2019 and Q4 2004-Q4 2023, and Q4 2005-Q4 2019 and Q4 2005-Q4 2023, respectively. The latest quarter available refers to Q3 2023 for Israel, and to Q4 2023 for Belgium and the United Kingdom. OECD is an unweighted average of the 31 OECD countries shown in this Chart (not including Chile, Colombia, Costa Rica, Iceland, Japan, Switzerland and Türkiye). Euro Area refers to the 20 Eurozone countries. Source: OECD (2024), “Quarterly National Accounts”, _OECD National Accounts Statistics_ (database), https://doi.org/10.1787/data-00017-en (accessed on 21 June 2024) for all countries except Australia, Korea, New Zealand, and the United States; Australian Labour Account (Australian Bureau of Statistics) for Australia, Economically Active Population Survey (Statistics Korea) for Korea, Quarterly Employment Survey (QES – Tables QEM034AA and QEM025AA, Stats NZ) for New Zealand, and Current Population Survey (CPS, Bureau of Labor Statistics) for the United States.\n\nStatLink 2 https://stat.link/kudyaz\n\n\\-25\n\n\\-20\n\n\\-15\n\n\\-10\n\n\\-5\n\n0\n\n5 %\n\nQ1 2005-Q4 2019 Q1 2005-Q1 2024"],"full_content":" **25**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\nIn Q1 2024, hours worked per employed person were below their pre-COVID-19 levels in 20 of the 31 countries with recent data available (Figure 1.5). The average decline in hours worked since Q4 2019 across all countries with available data is rather small – just above 1%. ³ In most countries, this decline follows a trend that pre-dates the COVID-19 crisis, though with some notable accelerations in Austria, Finland, Hungary, Ireland, Korea, the Slovak Republic and Spain. Among the five countries where hours worked had been increasing before the COVID-19 crisis, only the Netherlands, Portugal and the United States, saw a decline of more than 1% in the aftermath of the pandemic.\n\nEvidence for Europe indicates that the decline in hours worked over the past 20 years is largely driven by an increase in part-time and a reduction in hours within jobs (as opposed to a compositional shift to jobs typically requiring fewer hours) (Astinova et al., 2024[5]; ECB, 2021[6]). However, the decline in average hours worked since the COVID-19 crisis has not been associated with widespread up-ticks in part-time employment. On the contrary, annual data for 2022 point to a slight decrease in the incidence of part-time in most OECD countries relative to 2019. ⁴\n\nOverall, the cross-country comparison does not lend support to the hypothesis of a generalised post- COVID change in preferences over work-life balance that might have reduced willingness to work, but more evidence is needed to understand patterns observed in specific countries.\n\n### **Figure 1.5. The post-pandemic decline in average hours worked per worker is generally consistent with long-term trends**\n\nPercentage change, seasonally adjusted data\n\nNote: Average hours worked per worker is defined as the total hours worked divided by total employment, except for **Belgium** where it refers to the average hours worked of employees, **Korea,** where it refers to the average usual weekly hours worked per employed, **New Zealand** , where it is defined as the total paid hours divided by filled jobs, and the **United States** where it refers to the average usual weekly hours worked per wage and salary workers. Statistics are not seasonally adjusted for Canada and Mexico and periods reported for those countries refer to Q4 2004-Q4 2019 and Q4 2004-Q4 2023, and Q4 2005-Q4 2019 and Q4 2005-Q4 2023, respectively. The latest quarter available refers to Q3 2023 for Israel, and to Q4 2023 for Belgium and the United Kingdom. OECD is an unweighted average of the 31 OECD countries shown in this Chart (not including Chile, Colombia, Costa Rica, Iceland, Japan, Switzerland and Türkiye). Euro Area refers to the 20 Eurozone countries. Source: OECD (2024), “Quarterly National Accounts”, _OECD National Accounts Statistics_ (database), https://doi.org/10.1787/data-00017-en (accessed on 21 June 2024) for all countries except Australia, Korea, New Zealand, and the United States; Australian Labour Account (Australian Bureau of Statistics) for Australia, Economically Active Population Survey (Statistics Korea) for Korea, Quarterly Employment Survey (QES – Tables QEM034AA and QEM025AA, Stats NZ) for New Zealand, and Current Population Survey (CPS, Bureau of Labor Statistics) for the United States.\n\nStatLink 2 https://stat.link/kudyaz\n\n\\-25\n\n\\-20\n\n\\-15\n\n\\-10\n\n\\-5\n\n0\n\n5 %\n\nQ1 2005-Q4 2019 Q1 2005-Q1 2024","publish_date":null,"title":"Figure 1.5. The post-pandemic decline in average hours worked per worker is generally consistent with long-term trends","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p27.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p27.pdf"},"request":{"example_id":"chart/ac8b3538-en_p27","product_type":"parse","source_file_path":"ac8b3538-en_p27.pdf"},"started_at":"2026-11-04T19:45:49.812661"} +{"completed_at":"2026-11-04T19:45:54.210364","latency_in_ms":2745,"output":{"example_id":"chart/ac8b3538-en_p29","grounded_pages":[],"layout_pages":[],"markdown":" **27**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Figure 1.6. Labour markets remain tight in many countries even as pressure is easing**\n\nNote: OECD is the unweighted average of the 23 OECD countries shown in Panel A of this chart (not including Chile, Colombia, Costa Rica, Czechia, Denmark, Estonia, Iceland, Israel, Italy, Japan, Korea, Mexico, New Zealand, Switzerland and Türkiye). The peak (Panel A) refers to the maximum value of the number of vacancies per unemployed reached in Q4 2019-Q4 2023. For Hungary and Latvia, the number of vacancies per unemployed remains always below its Q4 2019 level and consequently there is no peak during the period considered. For Greece, the peak refers to Q4 2023. In Panel A, statistics refer to the number of vacancies (see description below) divided by the number of unemployed (ILO definition). The definition of vacancies is not harmonised across countries. For **European countries** (except Austria, France, Germany, Hungary and Portugal – see below), a vacancy is defined as a paid post that is newly created, unoccupied, or about to become vacant for which the employer is taking active steps and is prepared to take further steps to find a suitable candidate from outside the enterprise concerned; and which the employer intends to fill either immediately or within a specific period. For **Australia** , a vacancy is defined as a job available for immediate filling and for which recruitment action has been taken by the employer. For **Austria** , a vacancy is defined as a job notified by firms to employment agencies which remain unfilled at the end of the month. For **Canada** , a vacancy is defined as a job meeting the following conditions: it is vacant on the reference date (first day of the month) or will become vacant during the month; there are tasks to be carried out during the month for the job in question; and the employer is actively seeking a worker outside the organisation to fill the job. The jobs could be full-time, part-time, permanent, temporary, casual, or seasonal. Jobs reserved for subcontractors, external consultants, or other workers who are not considered employees, are excluded. For **France** , a vacancy is defined as the monthly number of vacancies posted by companies in France Travail. For **Germany** , a vacancy is defined as a job of seven days’ duration or more reported by employers to employment agencies to be filled within 3 months and remaining unfilled at the end of the month. For **Hungary** , a vacancy is defined as the number of vacancies notified to local labour offices as part of the central government territorial administrative unites. For **Portugal** , a vacancy is defined as the number of vacancies reported by employers to be still vacant at the last month of the quarter. For the **United Kingdom** , a vacancy is defined as a position for which employers are actively seeking recruits from outside their business or organisation (excluding agriculture, forestry, and fishing) based on the estimates from the Vacancy Survey. For the **United States** , a vacancy is defined as a job that is not filled on the last business day of the month and a job is considered open if a specific position exists and there is work available for it, the job can be started within 30 days, and there is active recruiting for the position. In Panel B, online job postings on Indeed are indexed to 100 in January 2024. Source: OECD (2024), “Labour: Registered unemployed and job vacancies (Edition 2023)”, Main Economic Indicators (database), https://doi.org/10.1787/cee89a15-en (accessed on 29 May 2024) for Australia, Austria, Germany, Hungary, Portugal and the United Kingdom; Eurostat, Job vacancy statistics by NACE Rev.2 activity (Table jvs\\_q\\_nace2) for Belgium, Finland, Greece, Ireland, Latvia, Lithuania, Luxembourg, the Netherlands, Norway, Poland, the Slovak Republic, Slovenia, Spain and Sweden, Job vacancies, payroll employees, and job vacancy rate (Statistics Canada) for Canada, Les demandeurs d’emploi inscrits à France Travail (DARES) for France; Job Openings and Labor Turnover Survey (Bureau of Labor Statistics, retrieved from FRED) for the United States; Indeed, Online Job Posting Tracker, https://github.com/hiring-lab/job\\_postings\\_tracker (accessed on 12 June 2024).\n\nStatLink 2 https://stat.link/jzo2lw\n\nHence, addressing labour shortages requires unpacking the many and interconnected factors behind them − and whether they differ from pre-COVID labour markets. For instance, the working conditions in some segments of the health sector – such as the long-term care one – have received considerable attention in the wake of the pandemic with a renewed interest in policy solutions to improve the quality of jobs that are already facing significant recruitment difficulties and are expecting further growth in demand as a result of population ageing (OECD, 2023[9]). Box 1.1 presents some possible factors behind labour shortages.\n\n88\n\n90\n\n92\n\n94\n\n96\n\n98\n\n100\n\nAUS GBR USA DEU CAN FRA IRL\n\n**B. Online job postings on Indeed**\n\n**February 2024 = 100, seasonally adjusted**\n\nMar 2024 Apr 2024 May 2024\n\n50\n\n75\n\n100\n\n125\n\n150 175\n\n200\n\n225 250\n\n**A. Vacancies per unemployed person**\n\n**National definitions, seasonally adjusted, base 100 in Q4 2019**\n\nQ4 2023 Peak\n\n446 279","pages":[{"line_numbers":[],"markdown":" **27**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Figure 1.6. Labour markets remain tight in many countries even as pressure is easing**\n\nNote: OECD is the unweighted average of the 23 OECD countries shown in Panel A of this chart (not including Chile, Colombia, Costa Rica, Czechia, Denmark, Estonia, Iceland, Israel, Italy, Japan, Korea, Mexico, New Zealand, Switzerland and Türkiye). The peak (Panel A) refers to the maximum value of the number of vacancies per unemployed reached in Q4 2019-Q4 2023. For Hungary and Latvia, the number of vacancies per unemployed remains always below its Q4 2019 level and consequently there is no peak during the period considered. For Greece, the peak refers to Q4 2023. In Panel A, statistics refer to the number of vacancies (see description below) divided by the number of unemployed (ILO definition). The definition of vacancies is not harmonised across countries. For **European countries** (except Austria, France, Germany, Hungary and Portugal – see below), a vacancy is defined as a paid post that is newly created, unoccupied, or about to become vacant for which the employer is taking active steps and is prepared to take further steps to find a suitable candidate from outside the enterprise concerned; and which the employer intends to fill either immediately or within a specific period. For **Australia** , a vacancy is defined as a job available for immediate filling and for which recruitment action has been taken by the employer. For **Austria** , a vacancy is defined as a job notified by firms to employment agencies which remain unfilled at the end of the month. For **Canada** , a vacancy is defined as a job meeting the following conditions: it is vacant on the reference date (first day of the month) or will become vacant during the month; there are tasks to be carried out during the month for the job in question; and the employer is actively seeking a worker outside the organisation to fill the job. The jobs could be full-time, part-time, permanent, temporary, casual, or seasonal. Jobs reserved for subcontractors, external consultants, or other workers who are not considered employees, are excluded. For **France** , a vacancy is defined as the monthly number of vacancies posted by companies in France Travail. For **Germany** , a vacancy is defined as a job of seven days’ duration or more reported by employers to employment agencies to be filled within 3 months and remaining unfilled at the end of the month. For **Hungary** , a vacancy is defined as the number of vacancies notified to local labour offices as part of the central government territorial administrative unites. For **Portugal** , a vacancy is defined as the number of vacancies reported by employers to be still vacant at the last month of the quarter. For the **United Kingdom** , a vacancy is defined as a position for which employers are actively seeking recruits from outside their business or organisation (excluding agriculture, forestry, and fishing) based on the estimates from the Vacancy Survey. For the **United States** , a vacancy is defined as a job that is not filled on the last business day of the month and a job is considered open if a specific position exists and there is work available for it, the job can be started within 30 days, and there is active recruiting for the position. In Panel B, online job postings on Indeed are indexed to 100 in January 2024. Source: OECD (2024), “Labour: Registered unemployed and job vacancies (Edition 2023)”, Main Economic Indicators (database), https://doi.org/10.1787/cee89a15-en (accessed on 29 May 2024) for Australia, Austria, Germany, Hungary, Portugal and the United Kingdom; Eurostat, Job vacancy statistics by NACE Rev.2 activity (Table jvs\\_q\\_nace2) for Belgium, Finland, Greece, Ireland, Latvia, Lithuania, Luxembourg, the Netherlands, Norway, Poland, the Slovak Republic, Slovenia, Spain and Sweden, Job vacancies, payroll employees, and job vacancy rate (Statistics Canada) for Canada, Les demandeurs d’emploi inscrits à France Travail (DARES) for France; Job Openings and Labor Turnover Survey (Bureau of Labor Statistics, retrieved from FRED) for the United States; Indeed, Online Job Posting Tracker, https://github.com/hiring-lab/job\\_postings\\_tracker (accessed on 12 June 2024).\n\nStatLink 2 https://stat.link/jzo2lw\n\nHence, addressing labour shortages requires unpacking the many and interconnected factors behind them − and whether they differ from pre-COVID labour markets. For instance, the working conditions in some segments of the health sector – such as the long-term care one – have received considerable attention in the wake of the pandemic with a renewed interest in policy solutions to improve the quality of jobs that are already facing significant recruitment difficulties and are expecting further growth in demand as a result of population ageing (OECD, 2023[9]). Box 1.1 presents some possible factors behind labour shortages.\n\n88\n\n90\n\n92\n\n94\n\n96\n\n98\n\n100\n\nAUS GBR USA DEU CAN FRA IRL\n\n**B. Online job postings on Indeed**\n\n**February 2024 = 100, seasonally adjusted**\n\nMar 2024 Apr 2024 May 2024\n\n50\n\n75\n\n100\n\n125\n\n150 175\n\n200\n\n225 250\n\n**A. Vacancies per unemployed person**\n\n**National definitions, seasonally adjusted, base 100 in Q4 2019**\n\nQ4 2023 Peak\n\n446 279","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":" **27**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Figure 1.6. Labour markets remain tight in many countries even as pressure is easing**\n\nNote: OECD is the unweighted average of the 23 OECD countries shown in Panel A of this chart (not including Chile, Colombia, Costa Rica, Czechia, Denmark, Estonia, Iceland, Israel, Italy, Japan, Korea, Mexico, New Zealand, Switzerland and Türkiye). The peak (Panel A) refers to the maximum value of the number of vacancies per unemployed reached in Q4 2019-Q4 2023. For Hungary and Latvia, the number of vacancies per unemployed remains always below its Q4 2019 level and consequently there is no peak during the period considered. For Greece, the peak refers to Q4 2023. In Panel A, statistics refer to the number of vacancies (see description below) divided by the number of unemployed (ILO definition). The definition of vacancies is not harmonised across countries. For **European countries** (except Austria, France, Germany, Hungary and Portugal – see below), a vacancy is defined as a paid post that is newly created, unoccupied, or about to become vacant for which the employer is taking active steps and is prepared to take further steps to find a suitable candidate from outside the enterprise concerned; and which the employer intends to fill either immediately or within a specific period. For **Australia** , a vacancy is defined as a job available for immediate filling and for which recruitment action has been taken by the employer. For **Austria** , a vacancy is defined as a job notified by firms to employment agencies which remain unfilled at the end of the month. For **Canada** , a vacancy is defined as a job meeting the following conditions: it is vacant on the reference date (first day of the month) or will become vacant during the month; there are tasks to be carried out during the month for the job in question; and the employer is actively seeking a worker outside the organisation to fill the job. The jobs could be full-time, part-time, permanent, temporary, casual, or seasonal. Jobs reserved for subcontractors, external consultants, or other workers who are not considered employees, are excluded. For **France** , a vacancy is defined as the monthly number of vacancies posted by companies in France Travail. For **Germany** , a vacancy is defined as a job of seven days’ duration or more reported by employers to employment agencies to be filled within 3 months and remaining unfilled at the end of the month. For **Hungary** , a vacancy is defined as the number of vacancies notified to local labour offices as part of the central government territorial administrative unites. For **Portugal** , a vacancy is defined as the number of vacancies reported by employers to be still vacant at the last month of the quarter. For the **United Kingdom** , a vacancy is defined as a position for which employers are actively seeking recruits from outside their business or organisation (excluding agriculture, forestry, and fishing) based on the estimates from the Vacancy Survey. For the **United States** , a vacancy is defined as a job that is not filled on the last business day of the month and a job is considered open if a specific position exists and there is work available for it, the job can be started within 30 days, and there is active recruiting for the position. In Panel B, online job postings on Indeed are indexed to 100 in January 2024. Source: OECD (2024), “Labour: Registered unemployed and job vacancies (Edition 2023)”, Main Economic Indicators (database), https://doi.org/10.1787/cee89a15-en (accessed on 29 May 2024) for Australia, Austria, Germany, Hungary, Portugal and the United Kingdom; Eurostat, Job vacancy statistics by NACE Rev.2 activity (Table jvs\\_q\\_nace2) for Belgium, Finland, Greece, Ireland, Latvia, Lithuania, Luxembourg, the Netherlands, Norway, Poland, the Slovak Republic, Slovenia, Spain and Sweden, Job vacancies, payroll employees, and job vacancy rate (Statistics Canada) for Canada, Les demandeurs d’emploi inscrits à France Travail (DARES) for France; Job Openings and Labor Turnover Survey (Bureau of Labor Statistics, retrieved from FRED) for the United States; Indeed, Online Job Posting Tracker, https://github.com/hiring-lab/job\\_postings\\_tracker (accessed on 12 June 2024).\n\nStatLink 2 https://stat.link/jzo2lw\n\nHence, addressing labour shortages requires unpacking the many and interconnected factors behind them − and whether they differ from pre-COVID labour markets. For instance, the working conditions in some segments of the health sector – such as the long-term care one – have received considerable attention in the wake of the pandemic with a renewed interest in policy solutions to improve the quality of jobs that are already facing significant recruitment difficulties and are expecting further growth in demand as a result of population ageing (OECD, 2023[9]). Box 1.1 presents some possible factors behind labour shortages.\n\n88\n\n90\n\n92\n\n94\n\n96\n\n98\n\n100\n\nAUS GBR USA DEU CAN FRA IRL\n\n**B. Online job postings on Indeed**\n\n**February 2024 = 100, seasonally adjusted**\n\nMar 2024 Apr 2024 May 2024\n\n50\n\n75\n\n100\n\n125\n\n150 175\n\n200\n\n225 250\n\n**A. Vacancies per unemployed person**\n\n**National definitions, seasonally adjusted, base 100 in Q4 2019**\n\nQ4 2023 Peak\n\n446 279","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p29.pdf"]},"response":{"errors":[],"extract_id":"extract_e138c64f9c7e14e10f4791079ff88159","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":[" **27**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Figure 1.6. Labour markets remain tight in many countries even as pressure is easing**\n\nNote: OECD is the unweighted average of the 23 OECD countries shown in Panel A of this chart (not including Chile, Colombia, Costa Rica, Czechia, Denmark, Estonia, Iceland, Israel, Italy, Japan, Korea, Mexico, New Zealand, Switzerland and Türkiye). The peak (Panel A) refers to the maximum value of the number of vacancies per unemployed reached in Q4 2019-Q4 2023. For Hungary and Latvia, the number of vacancies per unemployed remains always below its Q4 2019 level and consequently there is no peak during the period considered. For Greece, the peak refers to Q4 2023. In Panel A, statistics refer to the number of vacancies (see description below) divided by the number of unemployed (ILO definition). The definition of vacancies is not harmonised across countries. For **European countries** (except Austria, France, Germany, Hungary and Portugal – see below), a vacancy is defined as a paid post that is newly created, unoccupied, or about to become vacant for which the employer is taking active steps and is prepared to take further steps to find a suitable candidate from outside the enterprise concerned; and which the employer intends to fill either immediately or within a specific period. For **Australia** , a vacancy is defined as a job available for immediate filling and for which recruitment action has been taken by the employer. For **Austria** , a vacancy is defined as a job notified by firms to employment agencies which remain unfilled at the end of the month. For **Canada** , a vacancy is defined as a job meeting the following conditions: it is vacant on the reference date (first day of the month) or will become vacant during the month; there are tasks to be carried out during the month for the job in question; and the employer is actively seeking a worker outside the organisation to fill the job. The jobs could be full-time, part-time, permanent, temporary, casual, or seasonal. Jobs reserved for subcontractors, external consultants, or other workers who are not considered employees, are excluded. For **France** , a vacancy is defined as the monthly number of vacancies posted by companies in France Travail. For **Germany** , a vacancy is defined as a job of seven days’ duration or more reported by employers to employment agencies to be filled within 3 months and remaining unfilled at the end of the month. For **Hungary** , a vacancy is defined as the number of vacancies notified to local labour offices as part of the central government territorial administrative unites. For **Portugal** , a vacancy is defined as the number of vacancies reported by employers to be still vacant at the last month of the quarter. For the **United Kingdom** , a vacancy is defined as a position for which employers are actively seeking recruits from outside their business or organisation (excluding agriculture, forestry, and fishing) based on the estimates from the Vacancy Survey. For the **United States** , a vacancy is defined as a job that is not filled on the last business day of the month and a job is considered open if a specific position exists and there is work available for it, the job can be started within 30 days, and there is active recruiting for the position. In Panel B, online job postings on Indeed are indexed to 100 in January 2024. Source: OECD (2024), “Labour: Registered unemployed and job vacancies (Edition 2023)”, Main Economic Indicators (database), https://doi.org/10.1787/cee89a15-en (accessed on 29 May 2024) for Australia, Austria, Germany, Hungary, Portugal and the United Kingdom; Eurostat, Job vacancy statistics by NACE Rev.2 activity (Table jvs\\_q\\_nace2) for Belgium, Finland, Greece, Ireland, Latvia, Lithuania, Luxembourg, the Netherlands, Norway, Poland, the Slovak Republic, Slovenia, Spain and Sweden, Job vacancies, payroll employees, and job vacancy rate (Statistics Canada) for Canada, Les demandeurs d’emploi inscrits à France Travail (DARES) for France; Job Openings and Labor Turnover Survey (Bureau of Labor Statistics, retrieved from FRED) for the United States; Indeed, Online Job Posting Tracker, https://github.com/hiring-lab/job\\_postings\\_tracker (accessed on 12 June 2024).\n\nStatLink 2 https://stat.link/jzo2lw\n\nHence, addressing labour shortages requires unpacking the many and interconnected factors behind them − and whether they differ from pre-COVID labour markets. For instance, the working conditions in some segments of the health sector – such as the long-term care one – have received considerable attention in the wake of the pandemic with a renewed interest in policy solutions to improve the quality of jobs that are already facing significant recruitment difficulties and are expecting further growth in demand as a result of population ageing (OECD, 2023[9]). Box 1.1 presents some possible factors behind labour shortages.\n\n88\n\n90\n\n92\n\n94\n\n96\n\n98\n\n100\n\nAUS GBR USA DEU CAN FRA IRL\n\n**B. Online job postings on Indeed**\n\n**February 2024 = 100, seasonally adjusted**\n\nMar 2024 Apr 2024 May 2024\n\n50\n\n75\n\n100\n\n125\n\n150 175\n\n200\n\n225 250\n\n**A. Vacancies per unemployed person**\n\n**National definitions, seasonally adjusted, base 100 in Q4 2019**\n\nQ4 2023 Peak\n\n446 279"],"full_content":" **27**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Figure 1.6. Labour markets remain tight in many countries even as pressure is easing**\n\nNote: OECD is the unweighted average of the 23 OECD countries shown in Panel A of this chart (not including Chile, Colombia, Costa Rica, Czechia, Denmark, Estonia, Iceland, Israel, Italy, Japan, Korea, Mexico, New Zealand, Switzerland and Türkiye). The peak (Panel A) refers to the maximum value of the number of vacancies per unemployed reached in Q4 2019-Q4 2023. For Hungary and Latvia, the number of vacancies per unemployed remains always below its Q4 2019 level and consequently there is no peak during the period considered. For Greece, the peak refers to Q4 2023. In Panel A, statistics refer to the number of vacancies (see description below) divided by the number of unemployed (ILO definition). The definition of vacancies is not harmonised across countries. For **European countries** (except Austria, France, Germany, Hungary and Portugal – see below), a vacancy is defined as a paid post that is newly created, unoccupied, or about to become vacant for which the employer is taking active steps and is prepared to take further steps to find a suitable candidate from outside the enterprise concerned; and which the employer intends to fill either immediately or within a specific period. For **Australia** , a vacancy is defined as a job available for immediate filling and for which recruitment action has been taken by the employer. For **Austria** , a vacancy is defined as a job notified by firms to employment agencies which remain unfilled at the end of the month. For **Canada** , a vacancy is defined as a job meeting the following conditions: it is vacant on the reference date (first day of the month) or will become vacant during the month; there are tasks to be carried out during the month for the job in question; and the employer is actively seeking a worker outside the organisation to fill the job. The jobs could be full-time, part-time, permanent, temporary, casual, or seasonal. Jobs reserved for subcontractors, external consultants, or other workers who are not considered employees, are excluded. For **France** , a vacancy is defined as the monthly number of vacancies posted by companies in France Travail. For **Germany** , a vacancy is defined as a job of seven days’ duration or more reported by employers to employment agencies to be filled within 3 months and remaining unfilled at the end of the month. For **Hungary** , a vacancy is defined as the number of vacancies notified to local labour offices as part of the central government territorial administrative unites. For **Portugal** , a vacancy is defined as the number of vacancies reported by employers to be still vacant at the last month of the quarter. For the **United Kingdom** , a vacancy is defined as a position for which employers are actively seeking recruits from outside their business or organisation (excluding agriculture, forestry, and fishing) based on the estimates from the Vacancy Survey. For the **United States** , a vacancy is defined as a job that is not filled on the last business day of the month and a job is considered open if a specific position exists and there is work available for it, the job can be started within 30 days, and there is active recruiting for the position. In Panel B, online job postings on Indeed are indexed to 100 in January 2024. Source: OECD (2024), “Labour: Registered unemployed and job vacancies (Edition 2023)”, Main Economic Indicators (database), https://doi.org/10.1787/cee89a15-en (accessed on 29 May 2024) for Australia, Austria, Germany, Hungary, Portugal and the United Kingdom; Eurostat, Job vacancy statistics by NACE Rev.2 activity (Table jvs\\_q\\_nace2) for Belgium, Finland, Greece, Ireland, Latvia, Lithuania, Luxembourg, the Netherlands, Norway, Poland, the Slovak Republic, Slovenia, Spain and Sweden, Job vacancies, payroll employees, and job vacancy rate (Statistics Canada) for Canada, Les demandeurs d’emploi inscrits à France Travail (DARES) for France; Job Openings and Labor Turnover Survey (Bureau of Labor Statistics, retrieved from FRED) for the United States; Indeed, Online Job Posting Tracker, https://github.com/hiring-lab/job\\_postings\\_tracker (accessed on 12 June 2024).\n\nStatLink 2 https://stat.link/jzo2lw\n\nHence, addressing labour shortages requires unpacking the many and interconnected factors behind them − and whether they differ from pre-COVID labour markets. For instance, the working conditions in some segments of the health sector – such as the long-term care one – have received considerable attention in the wake of the pandemic with a renewed interest in policy solutions to improve the quality of jobs that are already facing significant recruitment difficulties and are expecting further growth in demand as a result of population ageing (OECD, 2023[9]). Box 1.1 presents some possible factors behind labour shortages.\n\n88\n\n90\n\n92\n\n94\n\n96\n\n98\n\n100\n\nAUS GBR USA DEU CAN FRA IRL\n\n**B. Online job postings on Indeed**\n\n**February 2024 = 100, seasonally adjusted**\n\nMar 2024 Apr 2024 May 2024\n\n50\n\n75\n\n100\n\n125\n\n150 175\n\n200\n\n225 250\n\n**A. Vacancies per unemployed person**\n\n**National definitions, seasonally adjusted, base 100 in Q4 2019**\n\nQ4 2023 Peak\n\n446 279","publish_date":null,"title":"Figure 1.6. Labour markets remain tight in many countries even as pressure is easing","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p29.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p29.pdf"},"request":{"example_id":"chart/ac8b3538-en_p29","product_type":"parse","source_file_path":"ac8b3538-en_p29.pdf"},"started_at":"2026-11-04T19:45:51.465319"} +{"completed_at":"2026-11-04T19:45:55.910485","latency_in_ms":2669,"output":{"example_id":"chart/ac8b3538-en_p31","grounded_pages":[],"layout_pages":[],"markdown":" **29**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n**_1\\.1.4. Economic growth in the OECD is expected to remain unchanged in 2024 and strengthens modestly in 2025, with marginal increase in unemployment and slowdown of employment growth_**\n\nGlobal growth, which slowed in the second half of 2023, is expected to stabilise and then pick up slightly through 2024-25. In part, this reflects better momentum than expected in the United States and some emerging-market economies. Annual OECD GDP growth is projected to be at 1.7% in 2024 and edge up to 1.8% in 2025 (OECD, 2024[4]).\n\nThe OECD-wide average unemployment rate is projected to rise marginally over 2024-25 to 5% in the fourth quarter of 2025 (Figure 1.7). The OECD-wide employment growth is expected to slow from 1.7% in 2023 to around 0.7% per annum on average over 2024-25, below its 2000-19 trend (Figure 1.7).\n\n### **Figure 1.7. Employment in the OECD is projected to continue to grow in 2024 and 2025, with the unemployment rate also inching up slightly**\n\nNote: (a) Actual value. (p) OECD projection. Euro Area refers to the 17 EU member states using the euro as their currency which are also OECD Member States. The 2000-19 trend refers to the average quarterly employment growth rate prevailing in Q1 2000 to Q4 2019. Source: OECD (2024[4]), _OECD Economic Outlook, Volume 2024 Issue 1: Preliminary version_ , https://doi.org/10.1787/69a0c310-en.\n\nStatLink 2 https://stat.link/dwxjyq\n\nSignificant uncertainty remains, however. Inflation may stay higher for longer, resulting in slower-than- expected reductions in interest rates and leading to further financial vulnerabilities. Growth could be disappointing in China, due to the persistent weakness of property markets or smaller than-anticipated fiscal support over the next two years. Another key downside risk to the outlook relates to the high geopolitical tensions, notably the uncertain course of Russia’s war of aggression against Ukraine, the evolving conflict in the Middle East, and the associated risks of renewed disruptions in global energy and food markets. On the upside, demand growth could prove stronger than expected, if households and firms were to draw more fully on the savings accumulated during COVID-19 (OECD, 2024[4]).\n\n### **1\\.2. Real wages are now growing in a number of countries but often remain below 2019 levels**\n\nHeadline inflation has fallen virtually everywhere primarily because of the partial reversal of the very large rise in energy prices over the previous two years and is expected to further ease. ⁶ After peaking at\n\n100\n\n105\n\n110\n\n115\n\n120\n\n125\n\n130\n\n**B. Trend in employment**\n\n**OECD Area, base 100 in 2000**\n\nActual Projected 2000-19 trend\n\n0 1 2 3 4 5 6 7 8\n\nOECD Euro Area Australia Japan Mexico United States\n\n%\n\n**A. Unemployment rate**\n\n**Percentage of total labour force**\n\nQ4 2019(a) Q4 2023(a) Q4 2024(p) Q4 2025(p)","pages":[{"line_numbers":[],"markdown":" **29**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n**_1\\.1.4. Economic growth in the OECD is expected to remain unchanged in 2024 and strengthens modestly in 2025, with marginal increase in unemployment and slowdown of employment growth_**\n\nGlobal growth, which slowed in the second half of 2023, is expected to stabilise and then pick up slightly through 2024-25. In part, this reflects better momentum than expected in the United States and some emerging-market economies. Annual OECD GDP growth is projected to be at 1.7% in 2024 and edge up to 1.8% in 2025 (OECD, 2024[4]).\n\nThe OECD-wide average unemployment rate is projected to rise marginally over 2024-25 to 5% in the fourth quarter of 2025 (Figure 1.7). The OECD-wide employment growth is expected to slow from 1.7% in 2023 to around 0.7% per annum on average over 2024-25, below its 2000-19 trend (Figure 1.7).\n\n### **Figure 1.7. Employment in the OECD is projected to continue to grow in 2024 and 2025, with the unemployment rate also inching up slightly**\n\nNote: (a) Actual value. (p) OECD projection. Euro Area refers to the 17 EU member states using the euro as their currency which are also OECD Member States. The 2000-19 trend refers to the average quarterly employment growth rate prevailing in Q1 2000 to Q4 2019. Source: OECD (2024[4]), _OECD Economic Outlook, Volume 2024 Issue 1: Preliminary version_ , https://doi.org/10.1787/69a0c310-en.\n\nStatLink 2 https://stat.link/dwxjyq\n\nSignificant uncertainty remains, however. Inflation may stay higher for longer, resulting in slower-than- expected reductions in interest rates and leading to further financial vulnerabilities. Growth could be disappointing in China, due to the persistent weakness of property markets or smaller than-anticipated fiscal support over the next two years. Another key downside risk to the outlook relates to the high geopolitical tensions, notably the uncertain course of Russia’s war of aggression against Ukraine, the evolving conflict in the Middle East, and the associated risks of renewed disruptions in global energy and food markets. On the upside, demand growth could prove stronger than expected, if households and firms were to draw more fully on the savings accumulated during COVID-19 (OECD, 2024[4]).\n\n### **1\\.2. Real wages are now growing in a number of countries but often remain below 2019 levels**\n\nHeadline inflation has fallen virtually everywhere primarily because of the partial reversal of the very large rise in energy prices over the previous two years and is expected to further ease. ⁶ After peaking at\n\n100\n\n105\n\n110\n\n115\n\n120\n\n125\n\n130\n\n**B. Trend in employment**\n\n**OECD Area, base 100 in 2000**\n\nActual Projected 2000-19 trend\n\n0 1 2 3 4 5 6 7 8\n\nOECD Euro Area Australia Japan Mexico United States\n\n%\n\n**A. Unemployment rate**\n\n**Percentage of total labour force**\n\nQ4 2019(a) Q4 2023(a) Q4 2024(p) Q4 2025(p)","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":" **29**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n**_1\\.1.4. Economic growth in the OECD is expected to remain unchanged in 2024 and strengthens modestly in 2025, with marginal increase in unemployment and slowdown of employment growth_**\n\nGlobal growth, which slowed in the second half of 2023, is expected to stabilise and then pick up slightly through 2024-25. In part, this reflects better momentum than expected in the United States and some emerging-market economies. Annual OECD GDP growth is projected to be at 1.7% in 2024 and edge up to 1.8% in 2025 (OECD, 2024[4]).\n\nThe OECD-wide average unemployment rate is projected to rise marginally over 2024-25 to 5% in the fourth quarter of 2025 (Figure 1.7). The OECD-wide employment growth is expected to slow from 1.7% in 2023 to around 0.7% per annum on average over 2024-25, below its 2000-19 trend (Figure 1.7).\n\n### **Figure 1.7. Employment in the OECD is projected to continue to grow in 2024 and 2025, with the unemployment rate also inching up slightly**\n\nNote: (a) Actual value. (p) OECD projection. Euro Area refers to the 17 EU member states using the euro as their currency which are also OECD Member States. The 2000-19 trend refers to the average quarterly employment growth rate prevailing in Q1 2000 to Q4 2019. Source: OECD (2024[4]), _OECD Economic Outlook, Volume 2024 Issue 1: Preliminary version_ , https://doi.org/10.1787/69a0c310-en.\n\nStatLink 2 https://stat.link/dwxjyq\n\nSignificant uncertainty remains, however. Inflation may stay higher for longer, resulting in slower-than- expected reductions in interest rates and leading to further financial vulnerabilities. Growth could be disappointing in China, due to the persistent weakness of property markets or smaller than-anticipated fiscal support over the next two years. Another key downside risk to the outlook relates to the high geopolitical tensions, notably the uncertain course of Russia’s war of aggression against Ukraine, the evolving conflict in the Middle East, and the associated risks of renewed disruptions in global energy and food markets. On the upside, demand growth could prove stronger than expected, if households and firms were to draw more fully on the savings accumulated during COVID-19 (OECD, 2024[4]).\n\n### **1\\.2. Real wages are now growing in a number of countries but often remain below 2019 levels**\n\nHeadline inflation has fallen virtually everywhere primarily because of the partial reversal of the very large rise in energy prices over the previous two years and is expected to further ease. ⁶ After peaking at\n\n100\n\n105\n\n110\n\n115\n\n120\n\n125\n\n130\n\n**B. Trend in employment**\n\n**OECD Area, base 100 in 2000**\n\nActual Projected 2000-19 trend\n\n0 1 2 3 4 5 6 7 8\n\nOECD Euro Area Australia Japan Mexico United States\n\n%\n\n**A. Unemployment rate**\n\n**Percentage of total labour force**\n\nQ4 2019(a) Q4 2023(a) Q4 2024(p) Q4 2025(p)","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p31.pdf"]},"response":{"errors":[],"extract_id":"extract_966b798d0be5564fc14522273f3d0ba6","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":[" **29**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n**_1\\.1.4. Economic growth in the OECD is expected to remain unchanged in 2024 and strengthens modestly in 2025, with marginal increase in unemployment and slowdown of employment growth_**\n\nGlobal growth, which slowed in the second half of 2023, is expected to stabilise and then pick up slightly through 2024-25. In part, this reflects better momentum than expected in the United States and some emerging-market economies. Annual OECD GDP growth is projected to be at 1.7% in 2024 and edge up to 1.8% in 2025 (OECD, 2024[4]).\n\nThe OECD-wide average unemployment rate is projected to rise marginally over 2024-25 to 5% in the fourth quarter of 2025 (Figure 1.7). The OECD-wide employment growth is expected to slow from 1.7% in 2023 to around 0.7% per annum on average over 2024-25, below its 2000-19 trend (Figure 1.7).\n\n### **Figure 1.7. Employment in the OECD is projected to continue to grow in 2024 and 2025, with the unemployment rate also inching up slightly**\n\nNote: (a) Actual value. (p) OECD projection. Euro Area refers to the 17 EU member states using the euro as their currency which are also OECD Member States. The 2000-19 trend refers to the average quarterly employment growth rate prevailing in Q1 2000 to Q4 2019. Source: OECD (2024[4]), _OECD Economic Outlook, Volume 2024 Issue 1: Preliminary version_ , https://doi.org/10.1787/69a0c310-en.\n\nStatLink 2 https://stat.link/dwxjyq\n\nSignificant uncertainty remains, however. Inflation may stay higher for longer, resulting in slower-than- expected reductions in interest rates and leading to further financial vulnerabilities. Growth could be disappointing in China, due to the persistent weakness of property markets or smaller than-anticipated fiscal support over the next two years. Another key downside risk to the outlook relates to the high geopolitical tensions, notably the uncertain course of Russia’s war of aggression against Ukraine, the evolving conflict in the Middle East, and the associated risks of renewed disruptions in global energy and food markets. On the upside, demand growth could prove stronger than expected, if households and firms were to draw more fully on the savings accumulated during COVID-19 (OECD, 2024[4]).\n\n### **1\\.2. Real wages are now growing in a number of countries but often remain below 2019 levels**\n\nHeadline inflation has fallen virtually everywhere primarily because of the partial reversal of the very large rise in energy prices over the previous two years and is expected to further ease. ⁶ After peaking at\n\n100\n\n105\n\n110\n\n115\n\n120\n\n125\n\n130\n\n**B. Trend in employment**\n\n**OECD Area, base 100 in 2000**\n\nActual Projected 2000-19 trend\n\n0 1 2 3 4 5 6 7 8\n\nOECD Euro Area Australia Japan Mexico United States\n\n%\n\n**A. Unemployment rate**\n\n**Percentage of total labour force**\n\nQ4 2019(a) Q4 2023(a) Q4 2024(p) Q4 2025(p)"],"full_content":" **29**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n**_1\\.1.4. Economic growth in the OECD is expected to remain unchanged in 2024 and strengthens modestly in 2025, with marginal increase in unemployment and slowdown of employment growth_**\n\nGlobal growth, which slowed in the second half of 2023, is expected to stabilise and then pick up slightly through 2024-25. In part, this reflects better momentum than expected in the United States and some emerging-market economies. Annual OECD GDP growth is projected to be at 1.7% in 2024 and edge up to 1.8% in 2025 (OECD, 2024[4]).\n\nThe OECD-wide average unemployment rate is projected to rise marginally over 2024-25 to 5% in the fourth quarter of 2025 (Figure 1.7). The OECD-wide employment growth is expected to slow from 1.7% in 2023 to around 0.7% per annum on average over 2024-25, below its 2000-19 trend (Figure 1.7).\n\n### **Figure 1.7. Employment in the OECD is projected to continue to grow in 2024 and 2025, with the unemployment rate also inching up slightly**\n\nNote: (a) Actual value. (p) OECD projection. Euro Area refers to the 17 EU member states using the euro as their currency which are also OECD Member States. The 2000-19 trend refers to the average quarterly employment growth rate prevailing in Q1 2000 to Q4 2019. Source: OECD (2024[4]), _OECD Economic Outlook, Volume 2024 Issue 1: Preliminary version_ , https://doi.org/10.1787/69a0c310-en.\n\nStatLink 2 https://stat.link/dwxjyq\n\nSignificant uncertainty remains, however. Inflation may stay higher for longer, resulting in slower-than- expected reductions in interest rates and leading to further financial vulnerabilities. Growth could be disappointing in China, due to the persistent weakness of property markets or smaller than-anticipated fiscal support over the next two years. Another key downside risk to the outlook relates to the high geopolitical tensions, notably the uncertain course of Russia’s war of aggression against Ukraine, the evolving conflict in the Middle East, and the associated risks of renewed disruptions in global energy and food markets. On the upside, demand growth could prove stronger than expected, if households and firms were to draw more fully on the savings accumulated during COVID-19 (OECD, 2024[4]).\n\n### **1\\.2. Real wages are now growing in a number of countries but often remain below 2019 levels**\n\nHeadline inflation has fallen virtually everywhere primarily because of the partial reversal of the very large rise in energy prices over the previous two years and is expected to further ease. ⁶ After peaking at\n\n100\n\n105\n\n110\n\n115\n\n120\n\n125\n\n130\n\n**B. Trend in employment**\n\n**OECD Area, base 100 in 2000**\n\nActual Projected 2000-19 trend\n\n0 1 2 3 4 5 6 7 8\n\nOECD Euro Area Australia Japan Mexico United States\n\n%\n\n**A. Unemployment rate**\n\n**Percentage of total labour force**\n\nQ4 2019(a) Q4 2023(a) Q4 2024(p) Q4 2025(p)","publish_date":null,"title":"Figure 1.7. Employment in the OECD is projected to continue to grow in 2024 and 2025, with the unemployment rate","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p31.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p31.pdf"},"request":{"example_id":"chart/ac8b3538-en_p31","product_type":"parse","source_file_path":"ac8b3538-en_p31.pdf"},"started_at":"2026-11-04T19:45:53.241100"} +{"completed_at":"2026-11-04T19:45:58.304334","latency_in_ms":3093,"output":{"example_id":"chart/ac8b3538-en_p32","grounded_pages":[],"layout_pages":[],"markdown":"**30** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\nover 10.7% in October 2022, OECD inflation almost halved reaching 5.9% in May 2024. However, inflation remained above the 2% target of central bank for 31 OECD countries – above 8% in Türkiye, and above 4% in five other OECD countries (Figure 1.8).\n\n### **Figure 1.8. Inflation remains high but has declined significantly since the peak of 2022**\n\nInflation defined as annual percentage change in the consumer price index (CPI), May 2024\n\nNote: For Australia and New Zealand, statistics refer to year-on-year changes in Q1 2024. Values on top refer to peaks of inflation above 16%. Euro Area refers to the 20 Eurozone countries. Source: OECD (2024), “Prices: Consumer prices”, Main Economic Indicators (database), https://doi.org/10.1787/0f2e8000-en (accessed on 02 July 2024).\n\nStatLink 2 https://stat.link/59wpm0\n\n**_1\\.2.1. Real wages are now growing year-on-year, but they remain below 2019 levels in many countries_**\n\nYear-on-year real wage growth turned positive in an increasing number of countries over the last year, as inflation declined and nominal wage growth picked up. ⁷ According to the latest data for Q1 2024, yearly real wage growth was positive in 29 of the 35 countries with available data, with an average change across all countries of +3.5%. In Belgium, Canada, France, Japan, New Zealand and Sweden, the annual real wage growth was still negative in Q4 2024 but relatively moderate – with real wages decreasing less than 1% over the year except for Japan ⁸ (Figure 1.9, Panel A).\n\nData on selected countries for different wage indicators and posted wages in online vacancies suggest that real wage growth has continued to improve in the first months of 2024. This is generally the result of moderating inflation while nominal wage growth has remained stable with some indication of a possible deceleration in posted wages (Box 1.2).\n\nSeveral factors contributed to the general improvement in annual real wage growth over the last year, including tight labour markets (Section 1.1.3), increases in statutory minimum wages (see Section 1.2.2 **)** , and the adjustment of negotiated wages (catch-up process and new collective agreements being renegotiated – Box 1.3 **)** .\n\n\\-2\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\n16 %\n\nLatest (May 2024) Peak since January 2021\n\n70 86\n\n26 18 25 18 22 24","pages":[{"line_numbers":[],"markdown":"**30** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\nover 10.7% in October 2022, OECD inflation almost halved reaching 5.9% in May 2024. However, inflation remained above the 2% target of central bank for 31 OECD countries – above 8% in Türkiye, and above 4% in five other OECD countries (Figure 1.8).\n\n### **Figure 1.8. Inflation remains high but has declined significantly since the peak of 2022**\n\nInflation defined as annual percentage change in the consumer price index (CPI), May 2024\n\nNote: For Australia and New Zealand, statistics refer to year-on-year changes in Q1 2024. Values on top refer to peaks of inflation above 16%. Euro Area refers to the 20 Eurozone countries. Source: OECD (2024), “Prices: Consumer prices”, Main Economic Indicators (database), https://doi.org/10.1787/0f2e8000-en (accessed on 02 July 2024).\n\nStatLink 2 https://stat.link/59wpm0\n\n**_1\\.2.1. Real wages are now growing year-on-year, but they remain below 2019 levels in many countries_**\n\nYear-on-year real wage growth turned positive in an increasing number of countries over the last year, as inflation declined and nominal wage growth picked up. ⁷ According to the latest data for Q1 2024, yearly real wage growth was positive in 29 of the 35 countries with available data, with an average change across all countries of +3.5%. In Belgium, Canada, France, Japan, New Zealand and Sweden, the annual real wage growth was still negative in Q4 2024 but relatively moderate – with real wages decreasing less than 1% over the year except for Japan ⁸ (Figure 1.9, Panel A).\n\nData on selected countries for different wage indicators and posted wages in online vacancies suggest that real wage growth has continued to improve in the first months of 2024. This is generally the result of moderating inflation while nominal wage growth has remained stable with some indication of a possible deceleration in posted wages (Box 1.2).\n\nSeveral factors contributed to the general improvement in annual real wage growth over the last year, including tight labour markets (Section 1.1.3), increases in statutory minimum wages (see Section 1.2.2 **)** , and the adjustment of negotiated wages (catch-up process and new collective agreements being renegotiated – Box 1.3 **)** .\n\n\\-2\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\n16 %\n\nLatest (May 2024) Peak since January 2021\n\n70 86\n\n26 18 25 18 22 24","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**30** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\nover 10.7% in October 2022, OECD inflation almost halved reaching 5.9% in May 2024. However, inflation remained above the 2% target of central bank for 31 OECD countries – above 8% in Türkiye, and above 4% in five other OECD countries (Figure 1.8).\n\n### **Figure 1.8. Inflation remains high but has declined significantly since the peak of 2022**\n\nInflation defined as annual percentage change in the consumer price index (CPI), May 2024\n\nNote: For Australia and New Zealand, statistics refer to year-on-year changes in Q1 2024. Values on top refer to peaks of inflation above 16%. Euro Area refers to the 20 Eurozone countries. Source: OECD (2024), “Prices: Consumer prices”, Main Economic Indicators (database), https://doi.org/10.1787/0f2e8000-en (accessed on 02 July 2024).\n\nStatLink 2 https://stat.link/59wpm0\n\n**_1\\.2.1. Real wages are now growing year-on-year, but they remain below 2019 levels in many countries_**\n\nYear-on-year real wage growth turned positive in an increasing number of countries over the last year, as inflation declined and nominal wage growth picked up. ⁷ According to the latest data for Q1 2024, yearly real wage growth was positive in 29 of the 35 countries with available data, with an average change across all countries of +3.5%. In Belgium, Canada, France, Japan, New Zealand and Sweden, the annual real wage growth was still negative in Q4 2024 but relatively moderate – with real wages decreasing less than 1% over the year except for Japan ⁸ (Figure 1.9, Panel A).\n\nData on selected countries for different wage indicators and posted wages in online vacancies suggest that real wage growth has continued to improve in the first months of 2024. This is generally the result of moderating inflation while nominal wage growth has remained stable with some indication of a possible deceleration in posted wages (Box 1.2).\n\nSeveral factors contributed to the general improvement in annual real wage growth over the last year, including tight labour markets (Section 1.1.3), increases in statutory minimum wages (see Section 1.2.2 **)** , and the adjustment of negotiated wages (catch-up process and new collective agreements being renegotiated – Box 1.3 **)** .\n\n\\-2\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\n16 %\n\nLatest (May 2024) Peak since January 2021\n\n70 86\n\n26 18 25 18 22 24","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p32.pdf"]},"response":{"errors":[],"extract_id":"extract_c20e2141b78d804697145727f7c3fa05","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**30** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\nover 10.7% in October 2022, OECD inflation almost halved reaching 5.9% in May 2024. However, inflation remained above the 2% target of central bank for 31 OECD countries – above 8% in Türkiye, and above 4% in five other OECD countries (Figure 1.8).\n\n### **Figure 1.8. Inflation remains high but has declined significantly since the peak of 2022**\n\nInflation defined as annual percentage change in the consumer price index (CPI), May 2024\n\nNote: For Australia and New Zealand, statistics refer to year-on-year changes in Q1 2024. Values on top refer to peaks of inflation above 16%. Euro Area refers to the 20 Eurozone countries. Source: OECD (2024), “Prices: Consumer prices”, Main Economic Indicators (database), https://doi.org/10.1787/0f2e8000-en (accessed on 02 July 2024).\n\nStatLink 2 https://stat.link/59wpm0\n\n**_1\\.2.1. Real wages are now growing year-on-year, but they remain below 2019 levels in many countries_**\n\nYear-on-year real wage growth turned positive in an increasing number of countries over the last year, as inflation declined and nominal wage growth picked up. ⁷ According to the latest data for Q1 2024, yearly real wage growth was positive in 29 of the 35 countries with available data, with an average change across all countries of +3.5%. In Belgium, Canada, France, Japan, New Zealand and Sweden, the annual real wage growth was still negative in Q4 2024 but relatively moderate – with real wages decreasing less than 1% over the year except for Japan ⁸ (Figure 1.9, Panel A).\n\nData on selected countries for different wage indicators and posted wages in online vacancies suggest that real wage growth has continued to improve in the first months of 2024. This is generally the result of moderating inflation while nominal wage growth has remained stable with some indication of a possible deceleration in posted wages (Box 1.2).\n\nSeveral factors contributed to the general improvement in annual real wage growth over the last year, including tight labour markets (Section 1.1.3), increases in statutory minimum wages (see Section 1.2.2 **)** , and the adjustment of negotiated wages (catch-up process and new collective agreements being renegotiated – Box 1.3 **)** .\n\n\\-2\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\n16 %\n\nLatest (May 2024) Peak since January 2021\n\n70 86\n\n26 18 25 18 22 24"],"full_content":"**30** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\nover 10.7% in October 2022, OECD inflation almost halved reaching 5.9% in May 2024. However, inflation remained above the 2% target of central bank for 31 OECD countries – above 8% in Türkiye, and above 4% in five other OECD countries (Figure 1.8).\n\n### **Figure 1.8. Inflation remains high but has declined significantly since the peak of 2022**\n\nInflation defined as annual percentage change in the consumer price index (CPI), May 2024\n\nNote: For Australia and New Zealand, statistics refer to year-on-year changes in Q1 2024. Values on top refer to peaks of inflation above 16%. Euro Area refers to the 20 Eurozone countries. Source: OECD (2024), “Prices: Consumer prices”, Main Economic Indicators (database), https://doi.org/10.1787/0f2e8000-en (accessed on 02 July 2024).\n\nStatLink 2 https://stat.link/59wpm0\n\n**_1\\.2.1. Real wages are now growing year-on-year, but they remain below 2019 levels in many countries_**\n\nYear-on-year real wage growth turned positive in an increasing number of countries over the last year, as inflation declined and nominal wage growth picked up. ⁷ According to the latest data for Q1 2024, yearly real wage growth was positive in 29 of the 35 countries with available data, with an average change across all countries of +3.5%. In Belgium, Canada, France, Japan, New Zealand and Sweden, the annual real wage growth was still negative in Q4 2024 but relatively moderate – with real wages decreasing less than 1% over the year except for Japan ⁸ (Figure 1.9, Panel A).\n\nData on selected countries for different wage indicators and posted wages in online vacancies suggest that real wage growth has continued to improve in the first months of 2024. This is generally the result of moderating inflation while nominal wage growth has remained stable with some indication of a possible deceleration in posted wages (Box 1.2).\n\nSeveral factors contributed to the general improvement in annual real wage growth over the last year, including tight labour markets (Section 1.1.3), increases in statutory minimum wages (see Section 1.2.2 **)** , and the adjustment of negotiated wages (catch-up process and new collective agreements being renegotiated – Box 1.3 **)** .\n\n\\-2\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\n16 %\n\nLatest (May 2024) Peak since January 2021\n\n70 86\n\n26 18 25 18 22 24","publish_date":null,"title":"Figure 1.8. Inflation remains high but has declined significantly since the peak of 2022","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p32.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p32.pdf"},"request":{"example_id":"chart/ac8b3538-en_p32","product_type":"parse","source_file_path":"ac8b3538-en_p32.pdf"},"started_at":"2026-11-04T19:45:55.210681"} +{"completed_at":"2026-11-04T19:45:57.830167","latency_in_ms":1254,"output":{"example_id":"chart/ac8b3538-en_p34","grounded_pages":[],"layout_pages":[],"markdown":"**32** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\nDespite the recent pick-up in their year-on-year growth, real wages remain below their pre-COVID-19 levels in most countries, even though the average change across all 35 countries with available data is positive (Figure 1.9, Panel B). ⁹ By Q1 2024, real wages had recovered at least some of the lost ground in 23 of the 27 countries in which they fell in the aftermath of the COVID-19 crisis – rising above pre-pandemic levels in 11 of them. However, real wages remained well below their pre-pandemic levels in virtually all countries where they fell the most. Overall, in Q1 2024, real wages were still below their Q4 2019 level in 16 of the 35 countries with available data.\n\n### **Box 1.2. Data for selected countries point to continued improvement in real wage growth in recent months generally driven by declining inflation**\n\nFor a limited number of countries, it is possible to gain insights on very recent wage developments using monthly data. This analysis is subject to the caveat that the underlying measures differ between countries (and from those used in the main analysis in Figure 1.10) and are generally not seasonally adjusted.\n\n### **Figure 1.10. Monthly data point to continued improvement in real wage growth**\n\nYear-on-year percentage change\n\nNote: Up to Q4 2023 refers to the average of the monthly observations of the six months ending to December 2023. Latest refer to the average of all monthly observations available after December 2023. The last available data point is April 2024 for Finland, Hungary, Japan, Norway and the United Kingdom; and May 2024 for Iceland and the United States. For Norway, earnings observed in April 2024 corresponds to the preliminary value. Real wage growth is “increasing” (“decreasing”) where the year-on-year percentage change in the average real wage over January to April 2024 is higher (lower) than the year-on-year percentage change in the average real wage over July to December 2023. Source: OECD calculations based on the Wage and salary indices by industry (Statistics Finland) for **Finland** ; the Main earnings data (Central Statistics Office) for **Hungary** ; the Wage indices by sector and month (Statistics Iceland) for **Iceland** ; the Monthly Labour Survey (Ministry of Health, Labour and Welfare) for **Japan** ; the Number of employment and earnings (Statistics Norway) for **Norway** ;; the Monthly Wages and Salaries Survey (Office for National Statistics) for the **United Kingdom** ; and the Current Employment Statistics (Bureau of Labor Statistics, retrieved from FRED) for the **United States** ; OECD (2024), “Prices: Consumer prices”, Main Economic Indicators (database), https://doi.org/10.1787/0f2e8000-en (accessed on 26 June 2024).\n\nStatLink 2 https://stat.link/1kthew\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\n16\n\nHUN NOR GBR JPN ISL USA FIN\n\nIncreasing real wage growth Decreasing real wage growth\n\n%\n\nNominal wage growth Inflation\n\nUp to Q4 2023 Latest Up to Q4 2023 Latest","pages":[{"line_numbers":[],"markdown":"**32** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\nDespite the recent pick-up in their year-on-year growth, real wages remain below their pre-COVID-19 levels in most countries, even though the average change across all 35 countries with available data is positive (Figure 1.9, Panel B). ⁹ By Q1 2024, real wages had recovered at least some of the lost ground in 23 of the 27 countries in which they fell in the aftermath of the COVID-19 crisis – rising above pre-pandemic levels in 11 of them. However, real wages remained well below their pre-pandemic levels in virtually all countries where they fell the most. Overall, in Q1 2024, real wages were still below their Q4 2019 level in 16 of the 35 countries with available data.\n\n### **Box 1.2. Data for selected countries point to continued improvement in real wage growth in recent months generally driven by declining inflation**\n\nFor a limited number of countries, it is possible to gain insights on very recent wage developments using monthly data. This analysis is subject to the caveat that the underlying measures differ between countries (and from those used in the main analysis in Figure 1.10) and are generally not seasonally adjusted.\n\n### **Figure 1.10. Monthly data point to continued improvement in real wage growth**\n\nYear-on-year percentage change\n\nNote: Up to Q4 2023 refers to the average of the monthly observations of the six months ending to December 2023. Latest refer to the average of all monthly observations available after December 2023. The last available data point is April 2024 for Finland, Hungary, Japan, Norway and the United Kingdom; and May 2024 for Iceland and the United States. For Norway, earnings observed in April 2024 corresponds to the preliminary value. Real wage growth is “increasing” (“decreasing”) where the year-on-year percentage change in the average real wage over January to April 2024 is higher (lower) than the year-on-year percentage change in the average real wage over July to December 2023. Source: OECD calculations based on the Wage and salary indices by industry (Statistics Finland) for **Finland** ; the Main earnings data (Central Statistics Office) for **Hungary** ; the Wage indices by sector and month (Statistics Iceland) for **Iceland** ; the Monthly Labour Survey (Ministry of Health, Labour and Welfare) for **Japan** ; the Number of employment and earnings (Statistics Norway) for **Norway** ;; the Monthly Wages and Salaries Survey (Office for National Statistics) for the **United Kingdom** ; and the Current Employment Statistics (Bureau of Labor Statistics, retrieved from FRED) for the **United States** ; OECD (2024), “Prices: Consumer prices”, Main Economic Indicators (database), https://doi.org/10.1787/0f2e8000-en (accessed on 26 June 2024).\n\nStatLink 2 https://stat.link/1kthew\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\n16\n\nHUN NOR GBR JPN ISL USA FIN\n\nIncreasing real wage growth Decreasing real wage growth\n\n%\n\nNominal wage growth Inflation\n\nUp to Q4 2023 Latest Up to Q4 2023 Latest","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**32** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\nDespite the recent pick-up in their year-on-year growth, real wages remain below their pre-COVID-19 levels in most countries, even though the average change across all 35 countries with available data is positive (Figure 1.9, Panel B). ⁹ By Q1 2024, real wages had recovered at least some of the lost ground in 23 of the 27 countries in which they fell in the aftermath of the COVID-19 crisis – rising above pre-pandemic levels in 11 of them. However, real wages remained well below their pre-pandemic levels in virtually all countries where they fell the most. Overall, in Q1 2024, real wages were still below their Q4 2019 level in 16 of the 35 countries with available data.\n\n### **Box 1.2. Data for selected countries point to continued improvement in real wage growth in recent months generally driven by declining inflation**\n\nFor a limited number of countries, it is possible to gain insights on very recent wage developments using monthly data. This analysis is subject to the caveat that the underlying measures differ between countries (and from those used in the main analysis in Figure 1.10) and are generally not seasonally adjusted.\n\n### **Figure 1.10. Monthly data point to continued improvement in real wage growth**\n\nYear-on-year percentage change\n\nNote: Up to Q4 2023 refers to the average of the monthly observations of the six months ending to December 2023. Latest refer to the average of all monthly observations available after December 2023. The last available data point is April 2024 for Finland, Hungary, Japan, Norway and the United Kingdom; and May 2024 for Iceland and the United States. For Norway, earnings observed in April 2024 corresponds to the preliminary value. Real wage growth is “increasing” (“decreasing”) where the year-on-year percentage change in the average real wage over January to April 2024 is higher (lower) than the year-on-year percentage change in the average real wage over July to December 2023. Source: OECD calculations based on the Wage and salary indices by industry (Statistics Finland) for **Finland** ; the Main earnings data (Central Statistics Office) for **Hungary** ; the Wage indices by sector and month (Statistics Iceland) for **Iceland** ; the Monthly Labour Survey (Ministry of Health, Labour and Welfare) for **Japan** ; the Number of employment and earnings (Statistics Norway) for **Norway** ;; the Monthly Wages and Salaries Survey (Office for National Statistics) for the **United Kingdom** ; and the Current Employment Statistics (Bureau of Labor Statistics, retrieved from FRED) for the **United States** ; OECD (2024), “Prices: Consumer prices”, Main Economic Indicators (database), https://doi.org/10.1787/0f2e8000-en (accessed on 26 June 2024).\n\nStatLink 2 https://stat.link/1kthew\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\n16\n\nHUN NOR GBR JPN ISL USA FIN\n\nIncreasing real wage growth Decreasing real wage growth\n\n%\n\nNominal wage growth Inflation\n\nUp to Q4 2023 Latest Up to Q4 2023 Latest","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p34.pdf"]},"response":{"errors":[],"extract_id":"extract_829c3e350c2b563a837e9800bfb5901c","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**32** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\nDespite the recent pick-up in their year-on-year growth, real wages remain below their pre-COVID-19 levels in most countries, even though the average change across all 35 countries with available data is positive (Figure 1.9, Panel B). ⁹ By Q1 2024, real wages had recovered at least some of the lost ground in 23 of the 27 countries in which they fell in the aftermath of the COVID-19 crisis – rising above pre-pandemic levels in 11 of them. However, real wages remained well below their pre-pandemic levels in virtually all countries where they fell the most. Overall, in Q1 2024, real wages were still below their Q4 2019 level in 16 of the 35 countries with available data.\n\n### **Box 1.2. Data for selected countries point to continued improvement in real wage growth in recent months generally driven by declining inflation**\n\nFor a limited number of countries, it is possible to gain insights on very recent wage developments using monthly data. This analysis is subject to the caveat that the underlying measures differ between countries (and from those used in the main analysis in Figure 1.10) and are generally not seasonally adjusted.\n\n### **Figure 1.10. Monthly data point to continued improvement in real wage growth**\n\nYear-on-year percentage change\n\nNote: Up to Q4 2023 refers to the average of the monthly observations of the six months ending to December 2023. Latest refer to the average of all monthly observations available after December 2023. The last available data point is April 2024 for Finland, Hungary, Japan, Norway and the United Kingdom; and May 2024 for Iceland and the United States. For Norway, earnings observed in April 2024 corresponds to the preliminary value. Real wage growth is “increasing” (“decreasing”) where the year-on-year percentage change in the average real wage over January to April 2024 is higher (lower) than the year-on-year percentage change in the average real wage over July to December 2023. Source: OECD calculations based on the Wage and salary indices by industry (Statistics Finland) for **Finland** ; the Main earnings data (Central Statistics Office) for **Hungary** ; the Wage indices by sector and month (Statistics Iceland) for **Iceland** ; the Monthly Labour Survey (Ministry of Health, Labour and Welfare) for **Japan** ; the Number of employment and earnings (Statistics Norway) for **Norway** ;; the Monthly Wages and Salaries Survey (Office for National Statistics) for the **United Kingdom** ; and the Current Employment Statistics (Bureau of Labor Statistics, retrieved from FRED) for the **United States** ; OECD (2024), “Prices: Consumer prices”, Main Economic Indicators (database), https://doi.org/10.1787/0f2e8000-en (accessed on 26 June 2024).\n\nStatLink 2 https://stat.link/1kthew\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\n16\n\nHUN NOR GBR JPN ISL USA FIN\n\nIncreasing real wage growth Decreasing real wage growth\n\n%\n\nNominal wage growth Inflation\n\nUp to Q4 2023 Latest Up to Q4 2023 Latest"],"full_content":"**32** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\nDespite the recent pick-up in their year-on-year growth, real wages remain below their pre-COVID-19 levels in most countries, even though the average change across all 35 countries with available data is positive (Figure 1.9, Panel B). ⁹ By Q1 2024, real wages had recovered at least some of the lost ground in 23 of the 27 countries in which they fell in the aftermath of the COVID-19 crisis – rising above pre-pandemic levels in 11 of them. However, real wages remained well below their pre-pandemic levels in virtually all countries where they fell the most. Overall, in Q1 2024, real wages were still below their Q4 2019 level in 16 of the 35 countries with available data.\n\n### **Box 1.2. Data for selected countries point to continued improvement in real wage growth in recent months generally driven by declining inflation**\n\nFor a limited number of countries, it is possible to gain insights on very recent wage developments using monthly data. This analysis is subject to the caveat that the underlying measures differ between countries (and from those used in the main analysis in Figure 1.10) and are generally not seasonally adjusted.\n\n### **Figure 1.10. Monthly data point to continued improvement in real wage growth**\n\nYear-on-year percentage change\n\nNote: Up to Q4 2023 refers to the average of the monthly observations of the six months ending to December 2023. Latest refer to the average of all monthly observations available after December 2023. The last available data point is April 2024 for Finland, Hungary, Japan, Norway and the United Kingdom; and May 2024 for Iceland and the United States. For Norway, earnings observed in April 2024 corresponds to the preliminary value. Real wage growth is “increasing” (“decreasing”) where the year-on-year percentage change in the average real wage over January to April 2024 is higher (lower) than the year-on-year percentage change in the average real wage over July to December 2023. Source: OECD calculations based on the Wage and salary indices by industry (Statistics Finland) for **Finland** ; the Main earnings data (Central Statistics Office) for **Hungary** ; the Wage indices by sector and month (Statistics Iceland) for **Iceland** ; the Monthly Labour Survey (Ministry of Health, Labour and Welfare) for **Japan** ; the Number of employment and earnings (Statistics Norway) for **Norway** ;; the Monthly Wages and Salaries Survey (Office for National Statistics) for the **United Kingdom** ; and the Current Employment Statistics (Bureau of Labor Statistics, retrieved from FRED) for the **United States** ; OECD (2024), “Prices: Consumer prices”, Main Economic Indicators (database), https://doi.org/10.1787/0f2e8000-en (accessed on 26 June 2024).\n\nStatLink 2 https://stat.link/1kthew\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\n16\n\nHUN NOR GBR JPN ISL USA FIN\n\nIncreasing real wage growth Decreasing real wage growth\n\n%\n\nNominal wage growth Inflation\n\nUp to Q4 2023 Latest Up to Q4 2023 Latest","publish_date":null,"title":"Box 1.2. Data for selected countries point to continued improvement in real wage growth in recent months generally driven by","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p34.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p34.pdf"},"request":{"example_id":"chart/ac8b3538-en_p34","product_type":"parse","source_file_path":"ac8b3538-en_p34.pdf"},"started_at":"2026-11-04T19:45:56.575264"} +{"completed_at":"2026-11-04T19:46:01.068668","latency_in_ms":2819,"output":{"example_id":"chart/ac8b3538-en_p35","grounded_pages":[],"layout_pages":[],"markdown":" **33**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\nThat being said, the data for the months since the end of Q4 2023 point to an improvement in annual real wage growth in four of the seven countries with available data. This is generally driven by a decline in inflation rather than an increase in nominal wage growth.\n\nData from wages advertised in job postings on the online platform _Indeed_ show improving or stable real wage growth in all countries with available data, except Spain and the United States (Figure 1.11). Consistently with the results above, where real wage growth is increasing, this is mainly driven by a fall in inflation rather than a significant up-tick in nominal wage growth. In fact, these data point to a decrease in nominal wage growth in five of the eight countries with available data (Canada, France, Germany, Spain, and the United States).\n\n### **Figure 1.11. Posted wages point to a recent slowdown in nominal wage growth**\n\nYear-on-year percentage change, three-month moving averages, from December 2023 to May 2024\n\nNote: The posted wages are the average year-on-year percentage changes in wages and salaries advertised by job postings on Indeed. Source: Indeed Wage Tracker (https://github.com/hiring-lab/indeed-wage-tracker); OECD (2024), “Prices: Consumer prices”, Main Economic Indicators (database), https://doi.org/10.1787/0f2e8000-en (accessed on 25 June 2024).\n\nStatLink 2 https://stat.link/guxrfj\n\n### **Box 1.3. Year-on-year growth in real negotiated wages has improved and remained negative only in a few countries in early 2024**\n\nReal growth in negotiated wages has improved over the course of 2023 and remained negative only in a few countries (Figure 1.12). In Q1 2024, negotiated wages were increasing in real terms on an annual basis in Austria, Canada, the Euro Area, Germany, Italy, the Netherlands, and the United States but continued to slightly decline in Australia and Sweden and stabilised, after one year of steady growth, in Belgium (Annex Figure 1.C.2). These developments reflect a combination of factors, including the staggered and infrequent nature of collective bargaining, the delay between the date of completion of negotiations and the effective revisions of pay, the infrequent use of automatic indexation to inflation, and the strength of workers’ bargaining power (Araki et al., 2023[11])). Overall, as more rounds of negotiation take place affecting an increasing number of workers, real growth in negotiated wages turns positive in more countries for some time, recovering some of the lost ground.\n\n0 1 2 3 4 5 6 7 8\n\nDec 2023\n\nMay 2024\n\nDec 2023\n\nMay 2024\n\nDec 2023\n\nMay 2024\n\nDec 2023\n\nMay 2024\n\nDec 2023\n\nMay 2024\n\nDec 2023\n\nMay 2024\n\nDec 2023\n\nMay 2024\n\nDec 2023\n\nMay 2024\n\nCanada France Germany Ireland Netherlands Spain United Kingdom\n\nUnited States\n\n%\n\nPosted wage growth Inflation","pages":[{"line_numbers":[],"markdown":" **33**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\nThat being said, the data for the months since the end of Q4 2023 point to an improvement in annual real wage growth in four of the seven countries with available data. This is generally driven by a decline in inflation rather than an increase in nominal wage growth.\n\nData from wages advertised in job postings on the online platform _Indeed_ show improving or stable real wage growth in all countries with available data, except Spain and the United States (Figure 1.11). Consistently with the results above, where real wage growth is increasing, this is mainly driven by a fall in inflation rather than a significant up-tick in nominal wage growth. In fact, these data point to a decrease in nominal wage growth in five of the eight countries with available data (Canada, France, Germany, Spain, and the United States).\n\n### **Figure 1.11. Posted wages point to a recent slowdown in nominal wage growth**\n\nYear-on-year percentage change, three-month moving averages, from December 2023 to May 2024\n\nNote: The posted wages are the average year-on-year percentage changes in wages and salaries advertised by job postings on Indeed. Source: Indeed Wage Tracker (https://github.com/hiring-lab/indeed-wage-tracker); OECD (2024), “Prices: Consumer prices”, Main Economic Indicators (database), https://doi.org/10.1787/0f2e8000-en (accessed on 25 June 2024).\n\nStatLink 2 https://stat.link/guxrfj\n\n### **Box 1.3. Year-on-year growth in real negotiated wages has improved and remained negative only in a few countries in early 2024**\n\nReal growth in negotiated wages has improved over the course of 2023 and remained negative only in a few countries (Figure 1.12). In Q1 2024, negotiated wages were increasing in real terms on an annual basis in Austria, Canada, the Euro Area, Germany, Italy, the Netherlands, and the United States but continued to slightly decline in Australia and Sweden and stabilised, after one year of steady growth, in Belgium (Annex Figure 1.C.2). These developments reflect a combination of factors, including the staggered and infrequent nature of collective bargaining, the delay between the date of completion of negotiations and the effective revisions of pay, the infrequent use of automatic indexation to inflation, and the strength of workers’ bargaining power (Araki et al., 2023[11])). Overall, as more rounds of negotiation take place affecting an increasing number of workers, real growth in negotiated wages turns positive in more countries for some time, recovering some of the lost ground.\n\n0 1 2 3 4 5 6 7 8\n\nDec 2023\n\nMay 2024\n\nDec 2023\n\nMay 2024\n\nDec 2023\n\nMay 2024\n\nDec 2023\n\nMay 2024\n\nDec 2023\n\nMay 2024\n\nDec 2023\n\nMay 2024\n\nDec 2023\n\nMay 2024\n\nDec 2023\n\nMay 2024\n\nCanada France Germany Ireland Netherlands Spain United Kingdom\n\nUnited States\n\n%\n\nPosted wage growth Inflation","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":" **33**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\nThat being said, the data for the months since the end of Q4 2023 point to an improvement in annual real wage growth in four of the seven countries with available data. This is generally driven by a decline in inflation rather than an increase in nominal wage growth.\n\nData from wages advertised in job postings on the online platform _Indeed_ show improving or stable real wage growth in all countries with available data, except Spain and the United States (Figure 1.11). Consistently with the results above, where real wage growth is increasing, this is mainly driven by a fall in inflation rather than a significant up-tick in nominal wage growth. In fact, these data point to a decrease in nominal wage growth in five of the eight countries with available data (Canada, France, Germany, Spain, and the United States).\n\n### **Figure 1.11. Posted wages point to a recent slowdown in nominal wage growth**\n\nYear-on-year percentage change, three-month moving averages, from December 2023 to May 2024\n\nNote: The posted wages are the average year-on-year percentage changes in wages and salaries advertised by job postings on Indeed. Source: Indeed Wage Tracker (https://github.com/hiring-lab/indeed-wage-tracker); OECD (2024), “Prices: Consumer prices”, Main Economic Indicators (database), https://doi.org/10.1787/0f2e8000-en (accessed on 25 June 2024).\n\nStatLink 2 https://stat.link/guxrfj\n\n### **Box 1.3. Year-on-year growth in real negotiated wages has improved and remained negative only in a few countries in early 2024**\n\nReal growth in negotiated wages has improved over the course of 2023 and remained negative only in a few countries (Figure 1.12). In Q1 2024, negotiated wages were increasing in real terms on an annual basis in Austria, Canada, the Euro Area, Germany, Italy, the Netherlands, and the United States but continued to slightly decline in Australia and Sweden and stabilised, after one year of steady growth, in Belgium (Annex Figure 1.C.2). These developments reflect a combination of factors, including the staggered and infrequent nature of collective bargaining, the delay between the date of completion of negotiations and the effective revisions of pay, the infrequent use of automatic indexation to inflation, and the strength of workers’ bargaining power (Araki et al., 2023[11])). Overall, as more rounds of negotiation take place affecting an increasing number of workers, real growth in negotiated wages turns positive in more countries for some time, recovering some of the lost ground.\n\n0 1 2 3 4 5 6 7 8\n\nDec 2023\n\nMay 2024\n\nDec 2023\n\nMay 2024\n\nDec 2023\n\nMay 2024\n\nDec 2023\n\nMay 2024\n\nDec 2023\n\nMay 2024\n\nDec 2023\n\nMay 2024\n\nDec 2023\n\nMay 2024\n\nDec 2023\n\nMay 2024\n\nCanada France Germany Ireland Netherlands Spain United Kingdom\n\nUnited States\n\n%\n\nPosted wage growth Inflation","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p35.pdf"]},"response":{"errors":[],"extract_id":"extract_f025ed03d225b6917c88345d4b0e85cd","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":[" **33**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\nThat being said, the data for the months since the end of Q4 2023 point to an improvement in annual real wage growth in four of the seven countries with available data. This is generally driven by a decline in inflation rather than an increase in nominal wage growth.\n\nData from wages advertised in job postings on the online platform _Indeed_ show improving or stable real wage growth in all countries with available data, except Spain and the United States (Figure 1.11). Consistently with the results above, where real wage growth is increasing, this is mainly driven by a fall in inflation rather than a significant up-tick in nominal wage growth. In fact, these data point to a decrease in nominal wage growth in five of the eight countries with available data (Canada, France, Germany, Spain, and the United States).\n\n### **Figure 1.11. Posted wages point to a recent slowdown in nominal wage growth**\n\nYear-on-year percentage change, three-month moving averages, from December 2023 to May 2024\n\nNote: The posted wages are the average year-on-year percentage changes in wages and salaries advertised by job postings on Indeed. Source: Indeed Wage Tracker (https://github.com/hiring-lab/indeed-wage-tracker); OECD (2024), “Prices: Consumer prices”, Main Economic Indicators (database), https://doi.org/10.1787/0f2e8000-en (accessed on 25 June 2024).\n\nStatLink 2 https://stat.link/guxrfj\n\n### **Box 1.3. Year-on-year growth in real negotiated wages has improved and remained negative only in a few countries in early 2024**\n\nReal growth in negotiated wages has improved over the course of 2023 and remained negative only in a few countries (Figure 1.12). In Q1 2024, negotiated wages were increasing in real terms on an annual basis in Austria, Canada, the Euro Area, Germany, Italy, the Netherlands, and the United States but continued to slightly decline in Australia and Sweden and stabilised, after one year of steady growth, in Belgium (Annex Figure 1.C.2). These developments reflect a combination of factors, including the staggered and infrequent nature of collective bargaining, the delay between the date of completion of negotiations and the effective revisions of pay, the infrequent use of automatic indexation to inflation, and the strength of workers’ bargaining power (Araki et al., 2023[11])). Overall, as more rounds of negotiation take place affecting an increasing number of workers, real growth in negotiated wages turns positive in more countries for some time, recovering some of the lost ground.\n\n0 1 2 3 4 5 6 7 8\n\nDec 2023\n\nMay 2024\n\nDec 2023\n\nMay 2024\n\nDec 2023\n\nMay 2024\n\nDec 2023\n\nMay 2024\n\nDec 2023\n\nMay 2024\n\nDec 2023\n\nMay 2024\n\nDec 2023\n\nMay 2024\n\nDec 2023\n\nMay 2024\n\nCanada France Germany Ireland Netherlands Spain United Kingdom\n\nUnited States\n\n%\n\nPosted wage growth Inflation"],"full_content":" **33**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\nThat being said, the data for the months since the end of Q4 2023 point to an improvement in annual real wage growth in four of the seven countries with available data. This is generally driven by a decline in inflation rather than an increase in nominal wage growth.\n\nData from wages advertised in job postings on the online platform _Indeed_ show improving or stable real wage growth in all countries with available data, except Spain and the United States (Figure 1.11). Consistently with the results above, where real wage growth is increasing, this is mainly driven by a fall in inflation rather than a significant up-tick in nominal wage growth. In fact, these data point to a decrease in nominal wage growth in five of the eight countries with available data (Canada, France, Germany, Spain, and the United States).\n\n### **Figure 1.11. Posted wages point to a recent slowdown in nominal wage growth**\n\nYear-on-year percentage change, three-month moving averages, from December 2023 to May 2024\n\nNote: The posted wages are the average year-on-year percentage changes in wages and salaries advertised by job postings on Indeed. Source: Indeed Wage Tracker (https://github.com/hiring-lab/indeed-wage-tracker); OECD (2024), “Prices: Consumer prices”, Main Economic Indicators (database), https://doi.org/10.1787/0f2e8000-en (accessed on 25 June 2024).\n\nStatLink 2 https://stat.link/guxrfj\n\n### **Box 1.3. Year-on-year growth in real negotiated wages has improved and remained negative only in a few countries in early 2024**\n\nReal growth in negotiated wages has improved over the course of 2023 and remained negative only in a few countries (Figure 1.12). In Q1 2024, negotiated wages were increasing in real terms on an annual basis in Austria, Canada, the Euro Area, Germany, Italy, the Netherlands, and the United States but continued to slightly decline in Australia and Sweden and stabilised, after one year of steady growth, in Belgium (Annex Figure 1.C.2). These developments reflect a combination of factors, including the staggered and infrequent nature of collective bargaining, the delay between the date of completion of negotiations and the effective revisions of pay, the infrequent use of automatic indexation to inflation, and the strength of workers’ bargaining power (Araki et al., 2023[11])). Overall, as more rounds of negotiation take place affecting an increasing number of workers, real growth in negotiated wages turns positive in more countries for some time, recovering some of the lost ground.\n\n0 1 2 3 4 5 6 7 8\n\nDec 2023\n\nMay 2024\n\nDec 2023\n\nMay 2024\n\nDec 2023\n\nMay 2024\n\nDec 2023\n\nMay 2024\n\nDec 2023\n\nMay 2024\n\nDec 2023\n\nMay 2024\n\nDec 2023\n\nMay 2024\n\nDec 2023\n\nMay 2024\n\nCanada France Germany Ireland Netherlands Spain United Kingdom\n\nUnited States\n\n%\n\nPosted wage growth Inflation","publish_date":null,"title":"Figure 1.11. Posted wages point to a recent slowdown in nominal wage growth Box 1.3. Year-on-year growth in real negotiated","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p35.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p35.pdf"},"request":{"example_id":"chart/ac8b3538-en_p35","product_type":"parse","source_file_path":"ac8b3538-en_p35.pdf"},"started_at":"2026-11-04T19:45:58.249440"} +{"completed_at":"2026-11-04T19:46:00.252250","latency_in_ms":1372,"output":{"example_id":"chart/ac8b3538-en_p36","grounded_pages":[],"layout_pages":[],"markdown":"**34** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\nFor Europe, the European Central Bank (ECB) indicator of future wage growth embedded in agreements reached in the latest quarter points to continuing growth in nominal wages, without signs of acceleration (Lane, 2024[12]). In fact, the latest release saw an increase in negotiated wage growth in the first quarter of 2024 to 4.7% – after it slightly moderated from 4.7% in the third quarter 2023 to 4.5% in the fourth quarter of 2023. Further agreements are expected to be renewed in 2024 which might have a significant impact on the dynamics of negotiated wages in the coming quarters.\n\n### **Figure 1.12. Real negotiated wages in selected OECD countries**\n\nYear-on-year percentage change in real negotiated wages (i.e. resulting from collective agreements)\n\nNote: International comparability of data on negotiated wages is affected by differences in definitions and measurement. Statistics are representative of all employees covered by a collective wage agreement for Austria, Belgium, the Euro Area (20), France, Germany, Italy, the Netherlands, Sweden, and the United States. In Canada, statistics refer to collective bargaining settlements of all bargaining units covering 500 or more employees (units of 100 or more employees for the Federal Jurisdiction). For Australia and Canada, statistics refer only to employees affected by an increase of the negotiated wage at date. Wage increases in Austria, Belgium, the Euro Area (20), Germany, Italy, the Netherlands, Sweden, and the United States refers to the average increase in negotiated wages (wages of union workers for the United States) weighted by the employment composition for a reference year (Laspeyres index). The reference year of the employment composition used is 2009 for Sweden, 2010 for Belgium and the Netherlands, January 2015 for the Euro Area (20), 2015 for Germany and Italy, 2016 for Austria, and 2021 for the United States. For Australia, Canada and France, wage increases refer to the average increase in negotiated wages weighted by the number of employees affected in the period considered. In Panel B, private sector for Germany refers to all industries excluding agriculture, public administration, education, health, and other personal services (Sections B to N of the NACE rev. 2). **LS** : wages including lump sums and/or special payments. The **trough** refers to the quarter where the year-on-year percentage change in real negotiated wages was at its lowest value for the indicated country and series (basic wage or wage including lump sums) since Q4 2019. Source: OECD calculations based on national data on negotiated wages, see Annex Table 1.C.3.in (Araki et al., 2023[11]) for further details; and OECD (2024), “Prices: Consumer prices”, Main Economic Indicators (database), https://doi.org/10.1787/0f2e8000-en (accessed on 28 June 2024).\n\nStatLink 2 https://stat.link/g6dcje\n\n**_1\\.2.2. Wages of low pay workers have performed relatively better in many countries_**\n\n_Statutory minimum wages in real terms are above their 2019 level in virtually all countries_\n\nIn May 2024, thanks to significant nominal increases of statutory minimum wages to support the lowest paid during the cost-of-living crisis, the real minimum wage was 12.8% higher than in May 2019 on average across the 30 OECD countries that have a national statutory minimum wage in place. This average figure\n\n\\-12\n\n\\-10\n\n\\-8\n\n\\-6\n\n\\-4\n\n\\-2\n\n0\n\n2\n\n4\n\n6\n\nAUT NLD NLD\n\n(LS)\n\nDEU (LS)\n\nEA20 ITA DEU CAN AUS SWE\n\n%\n\n**A. All sectors**\n\n\\-12\n\n\\-10\n\n\\-8\n\n\\-6\n\n\\-4\n\n\\-2\n\n0\n\n2\n\n4\n\n6\n\nAUT NLD NLD\n\n(LS)\n\nUSA ITA USA (LS)\n\nDEU (LS)\n\nCAN DEU FRA AUS BEL SWE\n\n%\n\n**B. Private sector**\n\nQ1 2024 or latest ( ) Trough since Q4 2019","pages":[{"line_numbers":[],"markdown":"**34** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\nFor Europe, the European Central Bank (ECB) indicator of future wage growth embedded in agreements reached in the latest quarter points to continuing growth in nominal wages, without signs of acceleration (Lane, 2024[12]). In fact, the latest release saw an increase in negotiated wage growth in the first quarter of 2024 to 4.7% – after it slightly moderated from 4.7% in the third quarter 2023 to 4.5% in the fourth quarter of 2023. Further agreements are expected to be renewed in 2024 which might have a significant impact on the dynamics of negotiated wages in the coming quarters.\n\n### **Figure 1.12. Real negotiated wages in selected OECD countries**\n\nYear-on-year percentage change in real negotiated wages (i.e. resulting from collective agreements)\n\nNote: International comparability of data on negotiated wages is affected by differences in definitions and measurement. Statistics are representative of all employees covered by a collective wage agreement for Austria, Belgium, the Euro Area (20), France, Germany, Italy, the Netherlands, Sweden, and the United States. In Canada, statistics refer to collective bargaining settlements of all bargaining units covering 500 or more employees (units of 100 or more employees for the Federal Jurisdiction). For Australia and Canada, statistics refer only to employees affected by an increase of the negotiated wage at date. Wage increases in Austria, Belgium, the Euro Area (20), Germany, Italy, the Netherlands, Sweden, and the United States refers to the average increase in negotiated wages (wages of union workers for the United States) weighted by the employment composition for a reference year (Laspeyres index). The reference year of the employment composition used is 2009 for Sweden, 2010 for Belgium and the Netherlands, January 2015 for the Euro Area (20), 2015 for Germany and Italy, 2016 for Austria, and 2021 for the United States. For Australia, Canada and France, wage increases refer to the average increase in negotiated wages weighted by the number of employees affected in the period considered. In Panel B, private sector for Germany refers to all industries excluding agriculture, public administration, education, health, and other personal services (Sections B to N of the NACE rev. 2). **LS** : wages including lump sums and/or special payments. The **trough** refers to the quarter where the year-on-year percentage change in real negotiated wages was at its lowest value for the indicated country and series (basic wage or wage including lump sums) since Q4 2019. Source: OECD calculations based on national data on negotiated wages, see Annex Table 1.C.3.in (Araki et al., 2023[11]) for further details; and OECD (2024), “Prices: Consumer prices”, Main Economic Indicators (database), https://doi.org/10.1787/0f2e8000-en (accessed on 28 June 2024).\n\nStatLink 2 https://stat.link/g6dcje\n\n**_1\\.2.2. Wages of low pay workers have performed relatively better in many countries_**\n\n_Statutory minimum wages in real terms are above their 2019 level in virtually all countries_\n\nIn May 2024, thanks to significant nominal increases of statutory minimum wages to support the lowest paid during the cost-of-living crisis, the real minimum wage was 12.8% higher than in May 2019 on average across the 30 OECD countries that have a national statutory minimum wage in place. This average figure\n\n\\-12\n\n\\-10\n\n\\-8\n\n\\-6\n\n\\-4\n\n\\-2\n\n0\n\n2\n\n4\n\n6\n\nAUT NLD NLD\n\n(LS)\n\nDEU (LS)\n\nEA20 ITA DEU CAN AUS SWE\n\n%\n\n**A. All sectors**\n\n\\-12\n\n\\-10\n\n\\-8\n\n\\-6\n\n\\-4\n\n\\-2\n\n0\n\n2\n\n4\n\n6\n\nAUT NLD NLD\n\n(LS)\n\nUSA ITA USA (LS)\n\nDEU (LS)\n\nCAN DEU FRA AUS BEL SWE\n\n%\n\n**B. Private sector**\n\nQ1 2024 or latest ( ) Trough since Q4 2019","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**34** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\nFor Europe, the European Central Bank (ECB) indicator of future wage growth embedded in agreements reached in the latest quarter points to continuing growth in nominal wages, without signs of acceleration (Lane, 2024[12]). In fact, the latest release saw an increase in negotiated wage growth in the first quarter of 2024 to 4.7% – after it slightly moderated from 4.7% in the third quarter 2023 to 4.5% in the fourth quarter of 2023. Further agreements are expected to be renewed in 2024 which might have a significant impact on the dynamics of negotiated wages in the coming quarters.\n\n### **Figure 1.12. Real negotiated wages in selected OECD countries**\n\nYear-on-year percentage change in real negotiated wages (i.e. resulting from collective agreements)\n\nNote: International comparability of data on negotiated wages is affected by differences in definitions and measurement. Statistics are representative of all employees covered by a collective wage agreement for Austria, Belgium, the Euro Area (20), France, Germany, Italy, the Netherlands, Sweden, and the United States. In Canada, statistics refer to collective bargaining settlements of all bargaining units covering 500 or more employees (units of 100 or more employees for the Federal Jurisdiction). For Australia and Canada, statistics refer only to employees affected by an increase of the negotiated wage at date. Wage increases in Austria, Belgium, the Euro Area (20), Germany, Italy, the Netherlands, Sweden, and the United States refers to the average increase in negotiated wages (wages of union workers for the United States) weighted by the employment composition for a reference year (Laspeyres index). The reference year of the employment composition used is 2009 for Sweden, 2010 for Belgium and the Netherlands, January 2015 for the Euro Area (20), 2015 for Germany and Italy, 2016 for Austria, and 2021 for the United States. For Australia, Canada and France, wage increases refer to the average increase in negotiated wages weighted by the number of employees affected in the period considered. In Panel B, private sector for Germany refers to all industries excluding agriculture, public administration, education, health, and other personal services (Sections B to N of the NACE rev. 2). **LS** : wages including lump sums and/or special payments. The **trough** refers to the quarter where the year-on-year percentage change in real negotiated wages was at its lowest value for the indicated country and series (basic wage or wage including lump sums) since Q4 2019. Source: OECD calculations based on national data on negotiated wages, see Annex Table 1.C.3.in (Araki et al., 2023[11]) for further details; and OECD (2024), “Prices: Consumer prices”, Main Economic Indicators (database), https://doi.org/10.1787/0f2e8000-en (accessed on 28 June 2024).\n\nStatLink 2 https://stat.link/g6dcje\n\n**_1\\.2.2. Wages of low pay workers have performed relatively better in many countries_**\n\n_Statutory minimum wages in real terms are above their 2019 level in virtually all countries_\n\nIn May 2024, thanks to significant nominal increases of statutory minimum wages to support the lowest paid during the cost-of-living crisis, the real minimum wage was 12.8% higher than in May 2019 on average across the 30 OECD countries that have a national statutory minimum wage in place. This average figure\n\n\\-12\n\n\\-10\n\n\\-8\n\n\\-6\n\n\\-4\n\n\\-2\n\n0\n\n2\n\n4\n\n6\n\nAUT NLD NLD\n\n(LS)\n\nDEU (LS)\n\nEA20 ITA DEU CAN AUS SWE\n\n%\n\n**A. All sectors**\n\n\\-12\n\n\\-10\n\n\\-8\n\n\\-6\n\n\\-4\n\n\\-2\n\n0\n\n2\n\n4\n\n6\n\nAUT NLD NLD\n\n(LS)\n\nUSA ITA USA (LS)\n\nDEU (LS)\n\nCAN DEU FRA AUS BEL SWE\n\n%\n\n**B. Private sector**\n\nQ1 2024 or latest ( ) Trough since Q4 2019","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p36.pdf"]},"response":{"errors":[],"extract_id":"extract_4b01f76fb97663b44bc22216e528130c","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**34** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\nFor Europe, the European Central Bank (ECB) indicator of future wage growth embedded in agreements reached in the latest quarter points to continuing growth in nominal wages, without signs of acceleration (Lane, 2024[12]). In fact, the latest release saw an increase in negotiated wage growth in the first quarter of 2024 to 4.7% – after it slightly moderated from 4.7% in the third quarter 2023 to 4.5% in the fourth quarter of 2023. Further agreements are expected to be renewed in 2024 which might have a significant impact on the dynamics of negotiated wages in the coming quarters.\n\n### **Figure 1.12. Real negotiated wages in selected OECD countries**\n\nYear-on-year percentage change in real negotiated wages (i.e. resulting from collective agreements)\n\nNote: International comparability of data on negotiated wages is affected by differences in definitions and measurement. Statistics are representative of all employees covered by a collective wage agreement for Austria, Belgium, the Euro Area (20), France, Germany, Italy, the Netherlands, Sweden, and the United States. In Canada, statistics refer to collective bargaining settlements of all bargaining units covering 500 or more employees (units of 100 or more employees for the Federal Jurisdiction). For Australia and Canada, statistics refer only to employees affected by an increase of the negotiated wage at date. Wage increases in Austria, Belgium, the Euro Area (20), Germany, Italy, the Netherlands, Sweden, and the United States refers to the average increase in negotiated wages (wages of union workers for the United States) weighted by the employment composition for a reference year (Laspeyres index). The reference year of the employment composition used is 2009 for Sweden, 2010 for Belgium and the Netherlands, January 2015 for the Euro Area (20), 2015 for Germany and Italy, 2016 for Austria, and 2021 for the United States. For Australia, Canada and France, wage increases refer to the average increase in negotiated wages weighted by the number of employees affected in the period considered. In Panel B, private sector for Germany refers to all industries excluding agriculture, public administration, education, health, and other personal services (Sections B to N of the NACE rev. 2). **LS** : wages including lump sums and/or special payments. The **trough** refers to the quarter where the year-on-year percentage change in real negotiated wages was at its lowest value for the indicated country and series (basic wage or wage including lump sums) since Q4 2019. Source: OECD calculations based on national data on negotiated wages, see Annex Table 1.C.3.in (Araki et al., 2023[11]) for further details; and OECD (2024), “Prices: Consumer prices”, Main Economic Indicators (database), https://doi.org/10.1787/0f2e8000-en (accessed on 28 June 2024).\n\nStatLink 2 https://stat.link/g6dcje\n\n**_1\\.2.2. Wages of low pay workers have performed relatively better in many countries_**\n\n_Statutory minimum wages in real terms are above their 2019 level in virtually all countries_\n\nIn May 2024, thanks to significant nominal increases of statutory minimum wages to support the lowest paid during the cost-of-living crisis, the real minimum wage was 12.8% higher than in May 2019 on average across the 30 OECD countries that have a national statutory minimum wage in place. This average figure\n\n\\-12\n\n\\-10\n\n\\-8\n\n\\-6\n\n\\-4\n\n\\-2\n\n0\n\n2\n\n4\n\n6\n\nAUT NLD NLD\n\n(LS)\n\nDEU (LS)\n\nEA20 ITA DEU CAN AUS SWE\n\n%\n\n**A. All sectors**\n\n\\-12\n\n\\-10\n\n\\-8\n\n\\-6\n\n\\-4\n\n\\-2\n\n0\n\n2\n\n4\n\n6\n\nAUT NLD NLD\n\n(LS)\n\nUSA ITA USA (LS)\n\nDEU (LS)\n\nCAN DEU FRA AUS BEL SWE\n\n%\n\n**B. Private sector**\n\nQ1 2024 or latest ( ) Trough since Q4 2019"],"full_content":"**34** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\nFor Europe, the European Central Bank (ECB) indicator of future wage growth embedded in agreements reached in the latest quarter points to continuing growth in nominal wages, without signs of acceleration (Lane, 2024[12]). In fact, the latest release saw an increase in negotiated wage growth in the first quarter of 2024 to 4.7% – after it slightly moderated from 4.7% in the third quarter 2023 to 4.5% in the fourth quarter of 2023. Further agreements are expected to be renewed in 2024 which might have a significant impact on the dynamics of negotiated wages in the coming quarters.\n\n### **Figure 1.12. Real negotiated wages in selected OECD countries**\n\nYear-on-year percentage change in real negotiated wages (i.e. resulting from collective agreements)\n\nNote: International comparability of data on negotiated wages is affected by differences in definitions and measurement. Statistics are representative of all employees covered by a collective wage agreement for Austria, Belgium, the Euro Area (20), France, Germany, Italy, the Netherlands, Sweden, and the United States. In Canada, statistics refer to collective bargaining settlements of all bargaining units covering 500 or more employees (units of 100 or more employees for the Federal Jurisdiction). For Australia and Canada, statistics refer only to employees affected by an increase of the negotiated wage at date. Wage increases in Austria, Belgium, the Euro Area (20), Germany, Italy, the Netherlands, Sweden, and the United States refers to the average increase in negotiated wages (wages of union workers for the United States) weighted by the employment composition for a reference year (Laspeyres index). The reference year of the employment composition used is 2009 for Sweden, 2010 for Belgium and the Netherlands, January 2015 for the Euro Area (20), 2015 for Germany and Italy, 2016 for Austria, and 2021 for the United States. For Australia, Canada and France, wage increases refer to the average increase in negotiated wages weighted by the number of employees affected in the period considered. In Panel B, private sector for Germany refers to all industries excluding agriculture, public administration, education, health, and other personal services (Sections B to N of the NACE rev. 2). **LS** : wages including lump sums and/or special payments. The **trough** refers to the quarter where the year-on-year percentage change in real negotiated wages was at its lowest value for the indicated country and series (basic wage or wage including lump sums) since Q4 2019. Source: OECD calculations based on national data on negotiated wages, see Annex Table 1.C.3.in (Araki et al., 2023[11]) for further details; and OECD (2024), “Prices: Consumer prices”, Main Economic Indicators (database), https://doi.org/10.1787/0f2e8000-en (accessed on 28 June 2024).\n\nStatLink 2 https://stat.link/g6dcje\n\n**_1\\.2.2. Wages of low pay workers have performed relatively better in many countries_**\n\n_Statutory minimum wages in real terms are above their 2019 level in virtually all countries_\n\nIn May 2024, thanks to significant nominal increases of statutory minimum wages to support the lowest paid during the cost-of-living crisis, the real minimum wage was 12.8% higher than in May 2019 on average across the 30 OECD countries that have a national statutory minimum wage in place. This average figure\n\n\\-12\n\n\\-10\n\n\\-8\n\n\\-6\n\n\\-4\n\n\\-2\n\n0\n\n2\n\n4\n\n6\n\nAUT NLD NLD\n\n(LS)\n\nDEU (LS)\n\nEA20 ITA DEU CAN AUS SWE\n\n%\n\n**A. All sectors**\n\n\\-12\n\n\\-10\n\n\\-8\n\n\\-6\n\n\\-4\n\n\\-2\n\n0\n\n2\n\n4\n\n6\n\nAUT NLD NLD\n\n(LS)\n\nUSA ITA USA (LS)\n\nDEU (LS)\n\nCAN DEU FRA AUS BEL SWE\n\n%\n\n**B. Private sector**\n\nQ1 2024 or latest ( ) Trough since Q4 2019","publish_date":null,"title":"Figure 1.12. Real negotiated wages in selected OECD countries","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p36.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p36.pdf"},"request":{"example_id":"chart/ac8b3538-en_p36","product_type":"parse","source_file_path":"ac8b3538-en_p36.pdf"},"started_at":"2026-11-04T19:45:58.879887"} +{"completed_at":"2026-11-04T19:46:02.373219","latency_in_ms":1301,"output":{"example_id":"chart/ac8b3538-en_p39","grounded_pages":[],"layout_pages":[],"markdown":" **37**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Figure 1.14. Real wages in low-pay industries have performed relatively better in most countries**\n\nPercentage change in real hourly wages between Q4 2019 and Q1 2024\n\nNote: Real wages are obtained by deflating nominal wages by consumer price inflation (all items) which are adjusted, for the purpose of this analysis, using the X-13ARIMA-SEATS Seasonal Adjustment Method. Industries are ranked by the median wage in 2019 in the European Structure of Earnings Survey (SES). The ranking of industries is broadly consistent when 2019 data on median wages from the Current Population Survey of the United States are used. **Low-pay** **industries** **(LP)** include Accommodation and food service, Administrative and support service, Arts, entertainment and recreation and Wholesale and retail trade. **Middle-pay** **industries (MP)** include Transportation and storage, Manufacturing, Other service, Real estate activities and Construction. **High-pay** **industries (HP)** include Human health and social work, Education, Professional activities, Information and communication and Finance and insurance. Average employment shares by industry over the four quarters of 2019 are used for aggregation and thus small inconsistencies between changes in wages by industry and changes in average wages are possible. Statistics refer to the percentage change in real hourly wages between Q4 2019 and Q4 2023 for Australia, Canada, Costa Rica, Japan, Korea, Mexico, the Netherlands, New Zealand, the United Kingdom and the United States. OECD is the unweighted average of the 33 OECD countries shown in this chart (not including Chile, Colombia, Israel, Switzerland and Türkiye). **†** : There are missing industries: Arts, entertainment and recreation is not included for the United States; Human health and social work and Education are not included for France. **‡** : Average weekly earnings are used for the United Kingdom. Moreover, wages in the public sector are excluded for Costa Rica, Japan, Korea, Mexico, the United Kingdom and the United States. Source: OECD calculations based on the Wage Price Index (Australian Bureau of Statistics) for **Australia** ; Fixed weighted index of average hourly earnings for all employees (Statistics Canada) for **Canada** ; Encuesta Continua de Empleo (Instituto Nacional de Estadística y Censos, Costa Rica) for **Costa** **Rica** ; the wages and salaries component of labour cost index by NACE Rev. 2 activity (Eurostat) for **European** countries; Monthly Labour Survey (Japanese Ministry of Health, Labour and Welfare) for **Japan** ; Labour Force Survey at Establishments (Korean Ministry of Employment and Labour) for **Korea** ; Nacional de Ocupación y Empleo, Encuesta Telefónica de Ocupación y Empleo, Encuesta Nacional de Ocupación y Empleo Nueva Edición (Instituto Nacional de Estadística y Geografía, Mexico) for **Mexico** ; Labour Cost Index (Statistics New Zealand) for New Zealand; Swiss Wage Index (Swiss Federal Statistical Office) for **Switzerland** , Monthly Wages and Salaries Survey (UK Office for National Statistics) for the **United** **Kingdom** ; and Employment Cost Index (U.S. Bureau of Labor Statistics, retrieved from FRED) for the **United** **States** ; OECD (2024), “Prices: Consumer prices”, Main Economic Indicators (database), https://doi.org/10.1787/0f2e8000-en (accessed on 18 June 2024).\n\nStatLink 2 https://stat.link/ph6ri2\n\nResults by education for the five countries with available data also provide additional support for a general pattern of wage compression, especially at the bottom of the distribution (Figure 1.15). Between 2019 and 2023, real wage growth was stronger for the low- and mid-pay groups by education in four of the five countries (Costa Rica, Mexico, the United Kingdom, and the United States). Canada was the only country among those with available data where real wages grew more for the highest educated.\n\n\\-10\n\n\\-5\n\n0\n\n5\n\n10\n\n15\n\n20\n\nGRC\n\nPOL\n\nKOR\n\nISL\n\nUSA†\n\nOECD\n\nDEU\n\nJPN\n\nFRA†\n\nNLD\n\nNOR\n\nESP\n\nNZL\n\nFIN\n\nCZE\n\nSVN\n\nLUX\n\nBEL\n\nCAN\n\nDNK\n\nITA\n\nLVA\n\nAUT\n\nCRI\n\nMEX\n\nPRT\n\nSVK\n\nGBR‡\n\nIRL\n\nSWE\n\nLTU\n\nHUN\n\nAUS\n\nEST\n\nLP> HP & MP LP> HP only LP> MP only LP < HP & MP\n\n%\n\nHigh-pay (HP) Middle-pay (MP) Low-pay (LP)\n\n21 25 21","pages":[{"line_numbers":[],"markdown":" **37**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Figure 1.14. Real wages in low-pay industries have performed relatively better in most countries**\n\nPercentage change in real hourly wages between Q4 2019 and Q1 2024\n\nNote: Real wages are obtained by deflating nominal wages by consumer price inflation (all items) which are adjusted, for the purpose of this analysis, using the X-13ARIMA-SEATS Seasonal Adjustment Method. Industries are ranked by the median wage in 2019 in the European Structure of Earnings Survey (SES). The ranking of industries is broadly consistent when 2019 data on median wages from the Current Population Survey of the United States are used. **Low-pay** **industries** **(LP)** include Accommodation and food service, Administrative and support service, Arts, entertainment and recreation and Wholesale and retail trade. **Middle-pay** **industries (MP)** include Transportation and storage, Manufacturing, Other service, Real estate activities and Construction. **High-pay** **industries (HP)** include Human health and social work, Education, Professional activities, Information and communication and Finance and insurance. Average employment shares by industry over the four quarters of 2019 are used for aggregation and thus small inconsistencies between changes in wages by industry and changes in average wages are possible. Statistics refer to the percentage change in real hourly wages between Q4 2019 and Q4 2023 for Australia, Canada, Costa Rica, Japan, Korea, Mexico, the Netherlands, New Zealand, the United Kingdom and the United States. OECD is the unweighted average of the 33 OECD countries shown in this chart (not including Chile, Colombia, Israel, Switzerland and Türkiye). **†** : There are missing industries: Arts, entertainment and recreation is not included for the United States; Human health and social work and Education are not included for France. **‡** : Average weekly earnings are used for the United Kingdom. Moreover, wages in the public sector are excluded for Costa Rica, Japan, Korea, Mexico, the United Kingdom and the United States. Source: OECD calculations based on the Wage Price Index (Australian Bureau of Statistics) for **Australia** ; Fixed weighted index of average hourly earnings for all employees (Statistics Canada) for **Canada** ; Encuesta Continua de Empleo (Instituto Nacional de Estadística y Censos, Costa Rica) for **Costa** **Rica** ; the wages and salaries component of labour cost index by NACE Rev. 2 activity (Eurostat) for **European** countries; Monthly Labour Survey (Japanese Ministry of Health, Labour and Welfare) for **Japan** ; Labour Force Survey at Establishments (Korean Ministry of Employment and Labour) for **Korea** ; Nacional de Ocupación y Empleo, Encuesta Telefónica de Ocupación y Empleo, Encuesta Nacional de Ocupación y Empleo Nueva Edición (Instituto Nacional de Estadística y Geografía, Mexico) for **Mexico** ; Labour Cost Index (Statistics New Zealand) for New Zealand; Swiss Wage Index (Swiss Federal Statistical Office) for **Switzerland** , Monthly Wages and Salaries Survey (UK Office for National Statistics) for the **United** **Kingdom** ; and Employment Cost Index (U.S. Bureau of Labor Statistics, retrieved from FRED) for the **United** **States** ; OECD (2024), “Prices: Consumer prices”, Main Economic Indicators (database), https://doi.org/10.1787/0f2e8000-en (accessed on 18 June 2024).\n\nStatLink 2 https://stat.link/ph6ri2\n\nResults by education for the five countries with available data also provide additional support for a general pattern of wage compression, especially at the bottom of the distribution (Figure 1.15). Between 2019 and 2023, real wage growth was stronger for the low- and mid-pay groups by education in four of the five countries (Costa Rica, Mexico, the United Kingdom, and the United States). Canada was the only country among those with available data where real wages grew more for the highest educated.\n\n\\-10\n\n\\-5\n\n0\n\n5\n\n10\n\n15\n\n20\n\nGRC\n\nPOL\n\nKOR\n\nISL\n\nUSA†\n\nOECD\n\nDEU\n\nJPN\n\nFRA†\n\nNLD\n\nNOR\n\nESP\n\nNZL\n\nFIN\n\nCZE\n\nSVN\n\nLUX\n\nBEL\n\nCAN\n\nDNK\n\nITA\n\nLVA\n\nAUT\n\nCRI\n\nMEX\n\nPRT\n\nSVK\n\nGBR‡\n\nIRL\n\nSWE\n\nLTU\n\nHUN\n\nAUS\n\nEST\n\nLP> HP & MP LP> HP only LP> MP only LP < HP & MP\n\n%\n\nHigh-pay (HP) Middle-pay (MP) Low-pay (LP)\n\n21 25 21","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":" **37**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Figure 1.14. Real wages in low-pay industries have performed relatively better in most countries**\n\nPercentage change in real hourly wages between Q4 2019 and Q1 2024\n\nNote: Real wages are obtained by deflating nominal wages by consumer price inflation (all items) which are adjusted, for the purpose of this analysis, using the X-13ARIMA-SEATS Seasonal Adjustment Method. Industries are ranked by the median wage in 2019 in the European Structure of Earnings Survey (SES). The ranking of industries is broadly consistent when 2019 data on median wages from the Current Population Survey of the United States are used. **Low-pay** **industries** **(LP)** include Accommodation and food service, Administrative and support service, Arts, entertainment and recreation and Wholesale and retail trade. **Middle-pay** **industries (MP)** include Transportation and storage, Manufacturing, Other service, Real estate activities and Construction. **High-pay** **industries (HP)** include Human health and social work, Education, Professional activities, Information and communication and Finance and insurance. Average employment shares by industry over the four quarters of 2019 are used for aggregation and thus small inconsistencies between changes in wages by industry and changes in average wages are possible. Statistics refer to the percentage change in real hourly wages between Q4 2019 and Q4 2023 for Australia, Canada, Costa Rica, Japan, Korea, Mexico, the Netherlands, New Zealand, the United Kingdom and the United States. OECD is the unweighted average of the 33 OECD countries shown in this chart (not including Chile, Colombia, Israel, Switzerland and Türkiye). **†** : There are missing industries: Arts, entertainment and recreation is not included for the United States; Human health and social work and Education are not included for France. **‡** : Average weekly earnings are used for the United Kingdom. Moreover, wages in the public sector are excluded for Costa Rica, Japan, Korea, Mexico, the United Kingdom and the United States. Source: OECD calculations based on the Wage Price Index (Australian Bureau of Statistics) for **Australia** ; Fixed weighted index of average hourly earnings for all employees (Statistics Canada) for **Canada** ; Encuesta Continua de Empleo (Instituto Nacional de Estadística y Censos, Costa Rica) for **Costa** **Rica** ; the wages and salaries component of labour cost index by NACE Rev. 2 activity (Eurostat) for **European** countries; Monthly Labour Survey (Japanese Ministry of Health, Labour and Welfare) for **Japan** ; Labour Force Survey at Establishments (Korean Ministry of Employment and Labour) for **Korea** ; Nacional de Ocupación y Empleo, Encuesta Telefónica de Ocupación y Empleo, Encuesta Nacional de Ocupación y Empleo Nueva Edición (Instituto Nacional de Estadística y Geografía, Mexico) for **Mexico** ; Labour Cost Index (Statistics New Zealand) for New Zealand; Swiss Wage Index (Swiss Federal Statistical Office) for **Switzerland** , Monthly Wages and Salaries Survey (UK Office for National Statistics) for the **United** **Kingdom** ; and Employment Cost Index (U.S. Bureau of Labor Statistics, retrieved from FRED) for the **United** **States** ; OECD (2024), “Prices: Consumer prices”, Main Economic Indicators (database), https://doi.org/10.1787/0f2e8000-en (accessed on 18 June 2024).\n\nStatLink 2 https://stat.link/ph6ri2\n\nResults by education for the five countries with available data also provide additional support for a general pattern of wage compression, especially at the bottom of the distribution (Figure 1.15). Between 2019 and 2023, real wage growth was stronger for the low- and mid-pay groups by education in four of the five countries (Costa Rica, Mexico, the United Kingdom, and the United States). Canada was the only country among those with available data where real wages grew more for the highest educated.\n\n\\-10\n\n\\-5\n\n0\n\n5\n\n10\n\n15\n\n20\n\nGRC\n\nPOL\n\nKOR\n\nISL\n\nUSA†\n\nOECD\n\nDEU\n\nJPN\n\nFRA†\n\nNLD\n\nNOR\n\nESP\n\nNZL\n\nFIN\n\nCZE\n\nSVN\n\nLUX\n\nBEL\n\nCAN\n\nDNK\n\nITA\n\nLVA\n\nAUT\n\nCRI\n\nMEX\n\nPRT\n\nSVK\n\nGBR‡\n\nIRL\n\nSWE\n\nLTU\n\nHUN\n\nAUS\n\nEST\n\nLP> HP & MP LP> HP only LP> MP only LP < HP & MP\n\n%\n\nHigh-pay (HP) Middle-pay (MP) Low-pay (LP)\n\n21 25 21","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p39.pdf"]},"response":{"errors":[],"extract_id":"extract_788a14807f536757256af0cf95f818e0","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":[" **37**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Figure 1.14. Real wages in low-pay industries have performed relatively better in most countries**\n\nPercentage change in real hourly wages between Q4 2019 and Q1 2024\n\nNote: Real wages are obtained by deflating nominal wages by consumer price inflation (all items) which are adjusted, for the purpose of this analysis, using the X-13ARIMA-SEATS Seasonal Adjustment Method. Industries are ranked by the median wage in 2019 in the European Structure of Earnings Survey (SES). The ranking of industries is broadly consistent when 2019 data on median wages from the Current Population Survey of the United States are used. **Low-pay** **industries** **(LP)** include Accommodation and food service, Administrative and support service, Arts, entertainment and recreation and Wholesale and retail trade. **Middle-pay** **industries (MP)** include Transportation and storage, Manufacturing, Other service, Real estate activities and Construction. **High-pay** **industries (HP)** include Human health and social work, Education, Professional activities, Information and communication and Finance and insurance. Average employment shares by industry over the four quarters of 2019 are used for aggregation and thus small inconsistencies between changes in wages by industry and changes in average wages are possible. Statistics refer to the percentage change in real hourly wages between Q4 2019 and Q4 2023 for Australia, Canada, Costa Rica, Japan, Korea, Mexico, the Netherlands, New Zealand, the United Kingdom and the United States. OECD is the unweighted average of the 33 OECD countries shown in this chart (not including Chile, Colombia, Israel, Switzerland and Türkiye). **†** : There are missing industries: Arts, entertainment and recreation is not included for the United States; Human health and social work and Education are not included for France. **‡** : Average weekly earnings are used for the United Kingdom. Moreover, wages in the public sector are excluded for Costa Rica, Japan, Korea, Mexico, the United Kingdom and the United States. Source: OECD calculations based on the Wage Price Index (Australian Bureau of Statistics) for **Australia** ; Fixed weighted index of average hourly earnings for all employees (Statistics Canada) for **Canada** ; Encuesta Continua de Empleo (Instituto Nacional de Estadística y Censos, Costa Rica) for **Costa** **Rica** ; the wages and salaries component of labour cost index by NACE Rev. 2 activity (Eurostat) for **European** countries; Monthly Labour Survey (Japanese Ministry of Health, Labour and Welfare) for **Japan** ; Labour Force Survey at Establishments (Korean Ministry of Employment and Labour) for **Korea** ; Nacional de Ocupación y Empleo, Encuesta Telefónica de Ocupación y Empleo, Encuesta Nacional de Ocupación y Empleo Nueva Edición (Instituto Nacional de Estadística y Geografía, Mexico) for **Mexico** ; Labour Cost Index (Statistics New Zealand) for New Zealand; Swiss Wage Index (Swiss Federal Statistical Office) for **Switzerland** , Monthly Wages and Salaries Survey (UK Office for National Statistics) for the **United** **Kingdom** ; and Employment Cost Index (U.S. Bureau of Labor Statistics, retrieved from FRED) for the **United** **States** ; OECD (2024), “Prices: Consumer prices”, Main Economic Indicators (database), https://doi.org/10.1787/0f2e8000-en (accessed on 18 June 2024).\n\nStatLink 2 https://stat.link/ph6ri2\n\nResults by education for the five countries with available data also provide additional support for a general pattern of wage compression, especially at the bottom of the distribution (Figure 1.15). Between 2019 and 2023, real wage growth was stronger for the low- and mid-pay groups by education in four of the five countries (Costa Rica, Mexico, the United Kingdom, and the United States). Canada was the only country among those with available data where real wages grew more for the highest educated.\n\n\\-10\n\n\\-5\n\n0\n\n5\n\n10\n\n15\n\n20\n\nGRC\n\nPOL\n\nKOR\n\nISL\n\nUSA†\n\nOECD\n\nDEU\n\nJPN\n\nFRA†\n\nNLD\n\nNOR\n\nESP\n\nNZL\n\nFIN\n\nCZE\n\nSVN\n\nLUX\n\nBEL\n\nCAN\n\nDNK\n\nITA\n\nLVA\n\nAUT\n\nCRI\n\nMEX\n\nPRT\n\nSVK\n\nGBR‡\n\nIRL\n\nSWE\n\nLTU\n\nHUN\n\nAUS\n\nEST\n\nLP> HP & MP LP> HP only LP> MP only LP < HP & MP\n\n%\n\nHigh-pay (HP) Middle-pay (MP) Low-pay (LP)\n\n21 25 21"],"full_content":" **37**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Figure 1.14. Real wages in low-pay industries have performed relatively better in most countries**\n\nPercentage change in real hourly wages between Q4 2019 and Q1 2024\n\nNote: Real wages are obtained by deflating nominal wages by consumer price inflation (all items) which are adjusted, for the purpose of this analysis, using the X-13ARIMA-SEATS Seasonal Adjustment Method. Industries are ranked by the median wage in 2019 in the European Structure of Earnings Survey (SES). The ranking of industries is broadly consistent when 2019 data on median wages from the Current Population Survey of the United States are used. **Low-pay** **industries** **(LP)** include Accommodation and food service, Administrative and support service, Arts, entertainment and recreation and Wholesale and retail trade. **Middle-pay** **industries (MP)** include Transportation and storage, Manufacturing, Other service, Real estate activities and Construction. **High-pay** **industries (HP)** include Human health and social work, Education, Professional activities, Information and communication and Finance and insurance. Average employment shares by industry over the four quarters of 2019 are used for aggregation and thus small inconsistencies between changes in wages by industry and changes in average wages are possible. Statistics refer to the percentage change in real hourly wages between Q4 2019 and Q4 2023 for Australia, Canada, Costa Rica, Japan, Korea, Mexico, the Netherlands, New Zealand, the United Kingdom and the United States. OECD is the unweighted average of the 33 OECD countries shown in this chart (not including Chile, Colombia, Israel, Switzerland and Türkiye). **†** : There are missing industries: Arts, entertainment and recreation is not included for the United States; Human health and social work and Education are not included for France. **‡** : Average weekly earnings are used for the United Kingdom. Moreover, wages in the public sector are excluded for Costa Rica, Japan, Korea, Mexico, the United Kingdom and the United States. Source: OECD calculations based on the Wage Price Index (Australian Bureau of Statistics) for **Australia** ; Fixed weighted index of average hourly earnings for all employees (Statistics Canada) for **Canada** ; Encuesta Continua de Empleo (Instituto Nacional de Estadística y Censos, Costa Rica) for **Costa** **Rica** ; the wages and salaries component of labour cost index by NACE Rev. 2 activity (Eurostat) for **European** countries; Monthly Labour Survey (Japanese Ministry of Health, Labour and Welfare) for **Japan** ; Labour Force Survey at Establishments (Korean Ministry of Employment and Labour) for **Korea** ; Nacional de Ocupación y Empleo, Encuesta Telefónica de Ocupación y Empleo, Encuesta Nacional de Ocupación y Empleo Nueva Edición (Instituto Nacional de Estadística y Geografía, Mexico) for **Mexico** ; Labour Cost Index (Statistics New Zealand) for New Zealand; Swiss Wage Index (Swiss Federal Statistical Office) for **Switzerland** , Monthly Wages and Salaries Survey (UK Office for National Statistics) for the **United** **Kingdom** ; and Employment Cost Index (U.S. Bureau of Labor Statistics, retrieved from FRED) for the **United** **States** ; OECD (2024), “Prices: Consumer prices”, Main Economic Indicators (database), https://doi.org/10.1787/0f2e8000-en (accessed on 18 June 2024).\n\nStatLink 2 https://stat.link/ph6ri2\n\nResults by education for the five countries with available data also provide additional support for a general pattern of wage compression, especially at the bottom of the distribution (Figure 1.15). Between 2019 and 2023, real wage growth was stronger for the low- and mid-pay groups by education in four of the five countries (Costa Rica, Mexico, the United Kingdom, and the United States). Canada was the only country among those with available data where real wages grew more for the highest educated.\n\n\\-10\n\n\\-5\n\n0\n\n5\n\n10\n\n15\n\n20\n\nGRC\n\nPOL\n\nKOR\n\nISL\n\nUSA†\n\nOECD\n\nDEU\n\nJPN\n\nFRA†\n\nNLD\n\nNOR\n\nESP\n\nNZL\n\nFIN\n\nCZE\n\nSVN\n\nLUX\n\nBEL\n\nCAN\n\nDNK\n\nITA\n\nLVA\n\nAUT\n\nCRI\n\nMEX\n\nPRT\n\nSVK\n\nGBR‡\n\nIRL\n\nSWE\n\nLTU\n\nHUN\n\nAUS\n\nEST\n\nLP> HP & MP LP> HP only LP> MP only LP < HP & MP\n\n%\n\nHigh-pay (HP) Middle-pay (MP) Low-pay (LP)\n\n21 25 21","publish_date":null,"title":"Figure 1.14. Real wages in low-pay industries have performed relatively better in most countries","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p39.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p39.pdf"},"request":{"example_id":"chart/ac8b3538-en_p39","product_type":"parse","source_file_path":"ac8b3538-en_p39.pdf"},"started_at":"2026-11-04T19:46:01.071887"} +{"completed_at":"2026-11-04T19:46:10.767708","latency_in_ms":9268,"output":{"example_id":"chart/ac8b3538-en_p42","grounded_pages":[],"layout_pages":[],"markdown":"**40** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Figure 1.16. Profits are beginning to buffer some of the increase in labour costs**\n\nCumulative percentage change since Q4 2021, seasonally adjusted data\n\nNote: OECD is the unweighted average of the 29 OECD countries shown in this chart (not including Chile, Colombia, Costa Rica, Iceland, Israel, Korea, Mexico, New Zealand and Türkiye). Euro Area represents the 20 Eurozone countries. For Norway, the data are based on mainland Norway. Unit labour costs and unit profits are calculated by dividing compensation of employees and gross operating surplus respectively, by real GDP. For Japan and Norway, gross operating surplus is approximated by deducting compensation of employees from nominal GDP – and hence also include unit net taxes. Source: OECD (2024), “Quarterly National Accounts”, OECD National Accounts Statistics (database), https://doi.org/10.1787/data-00017-en (accessed on 21 June 2024), Cabinet Office, Government of Japan, Economic and Social Research Institute (ESRI) Quarterly Estimates of GDP, www.esri.cao.go.jp/en/sna/sokuhou/sokuhou\\_top.html, for Japan, and Statistics Norway, Quarterly National Accounts, www.ssb.no/en/nasjonalregnskap-og-konjunkturer/nasjonalregnskap, for Norway.\n\nStatLink 2 https://stat.link/c4ghnj\n\nAs a result of the recent changes in the relative dynamics of unit labour costs and unit profits, the contribution of unit profits to domestic price pressures has decreased, but remaining higher than prior to the pandemic in the Euro Area (Figure 1.17) – see also (OECD, 2023[18]). In addition, these results imply a reduction in the profit share of income after its growth between 2019 and 2022 (Araki et al., 2023[11]).\n\nThese developments were largely expected as they reflect the ongoing recovery of purchasing power by wages described above, rather than a warning sign of wage-price spirals (Araki et al., 2023[11]). Indeed, the contribution of unit labour costs to domestic price pressures is likely to remain sustained for some time as this catch-up process continues, unless labour productivity growth picks up. Reassuringly, however, there are currently no signs of further acceleration in nominal wage growth (Box 1.2). Moreover, in many countries, the growth in unit profits over the last three years allows for more buffering against the inflationary pressures stemming from the recovery of real wages (Lane, 2024[12]). ¹¹ In the medium term, however, labour productivity growth is essential to ensure sustainable increases in wages that do not generate increases in unit labour costs and further inflationary pressures.\n\n\\-15 -10 -5 0 5 10 15 20 25 30 35 %\n\nQ1 2023 - Q1 2024 Q1 2023 - Q1 2024 Q1 2022 - Q1 2023 Q1 2022 - Q1 2023 Total Total\n\nUnit labour costs Unit profits","pages":[{"line_numbers":[],"markdown":"**40** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Figure 1.16. Profits are beginning to buffer some of the increase in labour costs**\n\nCumulative percentage change since Q4 2021, seasonally adjusted data\n\nNote: OECD is the unweighted average of the 29 OECD countries shown in this chart (not including Chile, Colombia, Costa Rica, Iceland, Israel, Korea, Mexico, New Zealand and Türkiye). Euro Area represents the 20 Eurozone countries. For Norway, the data are based on mainland Norway. Unit labour costs and unit profits are calculated by dividing compensation of employees and gross operating surplus respectively, by real GDP. For Japan and Norway, gross operating surplus is approximated by deducting compensation of employees from nominal GDP – and hence also include unit net taxes. Source: OECD (2024), “Quarterly National Accounts”, OECD National Accounts Statistics (database), https://doi.org/10.1787/data-00017-en (accessed on 21 June 2024), Cabinet Office, Government of Japan, Economic and Social Research Institute (ESRI) Quarterly Estimates of GDP, www.esri.cao.go.jp/en/sna/sokuhou/sokuhou\\_top.html, for Japan, and Statistics Norway, Quarterly National Accounts, www.ssb.no/en/nasjonalregnskap-og-konjunkturer/nasjonalregnskap, for Norway.\n\nStatLink 2 https://stat.link/c4ghnj\n\nAs a result of the recent changes in the relative dynamics of unit labour costs and unit profits, the contribution of unit profits to domestic price pressures has decreased, but remaining higher than prior to the pandemic in the Euro Area (Figure 1.17) – see also (OECD, 2023[18]). In addition, these results imply a reduction in the profit share of income after its growth between 2019 and 2022 (Araki et al., 2023[11]).\n\nThese developments were largely expected as they reflect the ongoing recovery of purchasing power by wages described above, rather than a warning sign of wage-price spirals (Araki et al., 2023[11]). Indeed, the contribution of unit labour costs to domestic price pressures is likely to remain sustained for some time as this catch-up process continues, unless labour productivity growth picks up. Reassuringly, however, there are currently no signs of further acceleration in nominal wage growth (Box 1.2). Moreover, in many countries, the growth in unit profits over the last three years allows for more buffering against the inflationary pressures stemming from the recovery of real wages (Lane, 2024[12]). ¹¹ In the medium term, however, labour productivity growth is essential to ensure sustainable increases in wages that do not generate increases in unit labour costs and further inflationary pressures.\n\n\\-15 -10 -5 0 5 10 15 20 25 30 35 %\n\nQ1 2023 - Q1 2024 Q1 2023 - Q1 2024 Q1 2022 - Q1 2023 Q1 2022 - Q1 2023 Total Total\n\nUnit labour costs Unit profits","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**40** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Figure 1.16. Profits are beginning to buffer some of the increase in labour costs**\n\nCumulative percentage change since Q4 2021, seasonally adjusted data\n\nNote: OECD is the unweighted average of the 29 OECD countries shown in this chart (not including Chile, Colombia, Costa Rica, Iceland, Israel, Korea, Mexico, New Zealand and Türkiye). Euro Area represents the 20 Eurozone countries. For Norway, the data are based on mainland Norway. Unit labour costs and unit profits are calculated by dividing compensation of employees and gross operating surplus respectively, by real GDP. For Japan and Norway, gross operating surplus is approximated by deducting compensation of employees from nominal GDP – and hence also include unit net taxes. Source: OECD (2024), “Quarterly National Accounts”, OECD National Accounts Statistics (database), https://doi.org/10.1787/data-00017-en (accessed on 21 June 2024), Cabinet Office, Government of Japan, Economic and Social Research Institute (ESRI) Quarterly Estimates of GDP, www.esri.cao.go.jp/en/sna/sokuhou/sokuhou\\_top.html, for Japan, and Statistics Norway, Quarterly National Accounts, www.ssb.no/en/nasjonalregnskap-og-konjunkturer/nasjonalregnskap, for Norway.\n\nStatLink 2 https://stat.link/c4ghnj\n\nAs a result of the recent changes in the relative dynamics of unit labour costs and unit profits, the contribution of unit profits to domestic price pressures has decreased, but remaining higher than prior to the pandemic in the Euro Area (Figure 1.17) – see also (OECD, 2023[18]). In addition, these results imply a reduction in the profit share of income after its growth between 2019 and 2022 (Araki et al., 2023[11]).\n\nThese developments were largely expected as they reflect the ongoing recovery of purchasing power by wages described above, rather than a warning sign of wage-price spirals (Araki et al., 2023[11]). Indeed, the contribution of unit labour costs to domestic price pressures is likely to remain sustained for some time as this catch-up process continues, unless labour productivity growth picks up. Reassuringly, however, there are currently no signs of further acceleration in nominal wage growth (Box 1.2). Moreover, in many countries, the growth in unit profits over the last three years allows for more buffering against the inflationary pressures stemming from the recovery of real wages (Lane, 2024[12]). ¹¹ In the medium term, however, labour productivity growth is essential to ensure sustainable increases in wages that do not generate increases in unit labour costs and further inflationary pressures.\n\n\\-15 -10 -5 0 5 10 15 20 25 30 35 %\n\nQ1 2023 - Q1 2024 Q1 2023 - Q1 2024 Q1 2022 - Q1 2023 Q1 2022 - Q1 2023 Total Total\n\nUnit labour costs Unit profits","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p42.pdf"]},"response":{"errors":[],"extract_id":"extract_3244779dbe777719bfb23e96130918f5","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**40** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Figure 1.16. Profits are beginning to buffer some of the increase in labour costs**\n\nCumulative percentage change since Q4 2021, seasonally adjusted data\n\nNote: OECD is the unweighted average of the 29 OECD countries shown in this chart (not including Chile, Colombia, Costa Rica, Iceland, Israel, Korea, Mexico, New Zealand and Türkiye). Euro Area represents the 20 Eurozone countries. For Norway, the data are based on mainland Norway. Unit labour costs and unit profits are calculated by dividing compensation of employees and gross operating surplus respectively, by real GDP. For Japan and Norway, gross operating surplus is approximated by deducting compensation of employees from nominal GDP – and hence also include unit net taxes. Source: OECD (2024), “Quarterly National Accounts”, OECD National Accounts Statistics (database), https://doi.org/10.1787/data-00017-en (accessed on 21 June 2024), Cabinet Office, Government of Japan, Economic and Social Research Institute (ESRI) Quarterly Estimates of GDP, www.esri.cao.go.jp/en/sna/sokuhou/sokuhou\\_top.html, for Japan, and Statistics Norway, Quarterly National Accounts, www.ssb.no/en/nasjonalregnskap-og-konjunkturer/nasjonalregnskap, for Norway.\n\nStatLink 2 https://stat.link/c4ghnj\n\nAs a result of the recent changes in the relative dynamics of unit labour costs and unit profits, the contribution of unit profits to domestic price pressures has decreased, but remaining higher than prior to the pandemic in the Euro Area (Figure 1.17) – see also (OECD, 2023[18]). In addition, these results imply a reduction in the profit share of income after its growth between 2019 and 2022 (Araki et al., 2023[11]).\n\nThese developments were largely expected as they reflect the ongoing recovery of purchasing power by wages described above, rather than a warning sign of wage-price spirals (Araki et al., 2023[11]). Indeed, the contribution of unit labour costs to domestic price pressures is likely to remain sustained for some time as this catch-up process continues, unless labour productivity growth picks up. Reassuringly, however, there are currently no signs of further acceleration in nominal wage growth (Box 1.2). Moreover, in many countries, the growth in unit profits over the last three years allows for more buffering against the inflationary pressures stemming from the recovery of real wages (Lane, 2024[12]). ¹¹ In the medium term, however, labour productivity growth is essential to ensure sustainable increases in wages that do not generate increases in unit labour costs and further inflationary pressures.\n\n\\-15 -10 -5 0 5 10 15 20 25 30 35 %\n\nQ1 2023 - Q1 2024 Q1 2023 - Q1 2024 Q1 2022 - Q1 2023 Q1 2022 - Q1 2023 Total Total\n\nUnit labour costs Unit profits"],"full_content":"**40** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Figure 1.16. Profits are beginning to buffer some of the increase in labour costs**\n\nCumulative percentage change since Q4 2021, seasonally adjusted data\n\nNote: OECD is the unweighted average of the 29 OECD countries shown in this chart (not including Chile, Colombia, Costa Rica, Iceland, Israel, Korea, Mexico, New Zealand and Türkiye). Euro Area represents the 20 Eurozone countries. For Norway, the data are based on mainland Norway. Unit labour costs and unit profits are calculated by dividing compensation of employees and gross operating surplus respectively, by real GDP. For Japan and Norway, gross operating surplus is approximated by deducting compensation of employees from nominal GDP – and hence also include unit net taxes. Source: OECD (2024), “Quarterly National Accounts”, OECD National Accounts Statistics (database), https://doi.org/10.1787/data-00017-en (accessed on 21 June 2024), Cabinet Office, Government of Japan, Economic and Social Research Institute (ESRI) Quarterly Estimates of GDP, www.esri.cao.go.jp/en/sna/sokuhou/sokuhou\\_top.html, for Japan, and Statistics Norway, Quarterly National Accounts, www.ssb.no/en/nasjonalregnskap-og-konjunkturer/nasjonalregnskap, for Norway.\n\nStatLink 2 https://stat.link/c4ghnj\n\nAs a result of the recent changes in the relative dynamics of unit labour costs and unit profits, the contribution of unit profits to domestic price pressures has decreased, but remaining higher than prior to the pandemic in the Euro Area (Figure 1.17) – see also (OECD, 2023[18]). In addition, these results imply a reduction in the profit share of income after its growth between 2019 and 2022 (Araki et al., 2023[11]).\n\nThese developments were largely expected as they reflect the ongoing recovery of purchasing power by wages described above, rather than a warning sign of wage-price spirals (Araki et al., 2023[11]). Indeed, the contribution of unit labour costs to domestic price pressures is likely to remain sustained for some time as this catch-up process continues, unless labour productivity growth picks up. Reassuringly, however, there are currently no signs of further acceleration in nominal wage growth (Box 1.2). Moreover, in many countries, the growth in unit profits over the last three years allows for more buffering against the inflationary pressures stemming from the recovery of real wages (Lane, 2024[12]). ¹¹ In the medium term, however, labour productivity growth is essential to ensure sustainable increases in wages that do not generate increases in unit labour costs and further inflationary pressures.\n\n\\-15 -10 -5 0 5 10 15 20 25 30 35 %\n\nQ1 2023 - Q1 2024 Q1 2023 - Q1 2024 Q1 2022 - Q1 2023 Q1 2022 - Q1 2023 Total Total\n\nUnit labour costs Unit profits","publish_date":null,"title":"Figure 1.16. Profits are beginning to buffer some of the increase in labour costs","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p42.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p42.pdf"},"request":{"example_id":"chart/ac8b3538-en_p42","product_type":"parse","source_file_path":"ac8b3538-en_p42.pdf"},"started_at":"2026-11-04T19:46:01.499412"} +{"completed_at":"2026-11-04T19:46:06.277034","latency_in_ms":2503,"output":{"example_id":"chart/ac8b3538-en_p48","grounded_pages":[],"layout_pages":[],"markdown":"**46** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Figure 1.20. Job strain in OECD European countries, 2021**\n\nPercentage of employees aged 16-64 in OECD European countries, 2021\n\nNote: Countries are ordered in Panel A by descending order of the incidence of job strain. “Average” is the unweighted average of the 25 OECD European countries shown in this Chart. Figures in Panel B refer to the difference between job demands and job resources, capturing the degree of strain. **Highly strained** : Cases where the number of job demands exceeds by at least two the number of job resources. **Moderately strained** : Cases where the number of job demands exceeds by one the number of job resources. For further details on methodology, see Annex 1.B. Source: OECD calculations based on the European Working Conditions Telephone Survey (EWCTS) 2021 of the European Foundation for the Improvement of Living and Working Conditions (Eurofound).\n\nStatLink 2 https://stat.link/9qidkp\n\nCountry differences in the share of workers’ experiencing job strain are likely to reflect different factors, such as different sectoral and occupational structures, different labour and employment policies, ²⁹ as well as different phases of the pandemic and policy responses to it. Moreover, as data display results for 2021, i.e., a year after most OECD European countries implemented lockdown measures and social distance requirement, results may also be affected by teleworking. For instance, those who were able to work from home fared best, while frontline workers during the COVID-19 pandemic fared poorly on several fronts and reported much higher exposure to physical risks than those working from home (Eurofound, 2022[21]). Controlling for a number of observable factors, such as individual characteristics, occupation structures and teleworking, explains some of the country variation ³⁰ in job strain, notably in the case of Luxembourg, Spain and Switzerland. ³¹ Yet, significant cross-country variation remains unexplained that might be attributed to differences across countries in policy, norms, expectations, or attitudes towards job.\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25 %\n\n**A. Incidence of job strain**\n\nBoth sexes Men Women\n\nFRA\n\nCZE\n\nFIN\n\nBEL\n\nPOL\n\nDEU\n\nAUT\n\nSWE ESP DNK\n\nGRC\n\nNOR\n\nCHE\n\n**Average**\n\nIRL\n\nGBR\n\nITA\n\nSVK\n\nLVA\n\nHUN\n\nLTU\n\nLUX\n\nPRT\n\nNLD\n\nEST SVN\n\n0\n\n1\n\n2\n\n3\n\n4\n\n5\n\n6\n\n7\n\n6 7 8 9 10 11 12 13 14\n\n% of employees highly strained\n\n% of employees moderately strained\n\n**B. Degrees of job strain**","pages":[{"line_numbers":[],"markdown":"**46** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Figure 1.20. Job strain in OECD European countries, 2021**\n\nPercentage of employees aged 16-64 in OECD European countries, 2021\n\nNote: Countries are ordered in Panel A by descending order of the incidence of job strain. “Average” is the unweighted average of the 25 OECD European countries shown in this Chart. Figures in Panel B refer to the difference between job demands and job resources, capturing the degree of strain. **Highly strained** : Cases where the number of job demands exceeds by at least two the number of job resources. **Moderately strained** : Cases where the number of job demands exceeds by one the number of job resources. For further details on methodology, see Annex 1.B. Source: OECD calculations based on the European Working Conditions Telephone Survey (EWCTS) 2021 of the European Foundation for the Improvement of Living and Working Conditions (Eurofound).\n\nStatLink 2 https://stat.link/9qidkp\n\nCountry differences in the share of workers’ experiencing job strain are likely to reflect different factors, such as different sectoral and occupational structures, different labour and employment policies, ²⁹ as well as different phases of the pandemic and policy responses to it. Moreover, as data display results for 2021, i.e., a year after most OECD European countries implemented lockdown measures and social distance requirement, results may also be affected by teleworking. For instance, those who were able to work from home fared best, while frontline workers during the COVID-19 pandemic fared poorly on several fronts and reported much higher exposure to physical risks than those working from home (Eurofound, 2022[21]). Controlling for a number of observable factors, such as individual characteristics, occupation structures and teleworking, explains some of the country variation ³⁰ in job strain, notably in the case of Luxembourg, Spain and Switzerland. ³¹ Yet, significant cross-country variation remains unexplained that might be attributed to differences across countries in policy, norms, expectations, or attitudes towards job.\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25 %\n\n**A. Incidence of job strain**\n\nBoth sexes Men Women\n\nFRA\n\nCZE\n\nFIN\n\nBEL\n\nPOL\n\nDEU\n\nAUT\n\nSWE ESP DNK\n\nGRC\n\nNOR\n\nCHE\n\n**Average**\n\nIRL\n\nGBR\n\nITA\n\nSVK\n\nLVA\n\nHUN\n\nLTU\n\nLUX\n\nPRT\n\nNLD\n\nEST SVN\n\n0\n\n1\n\n2\n\n3\n\n4\n\n5\n\n6\n\n7\n\n6 7 8 9 10 11 12 13 14\n\n% of employees highly strained\n\n% of employees moderately strained\n\n**B. Degrees of job strain**","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**46** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Figure 1.20. Job strain in OECD European countries, 2021**\n\nPercentage of employees aged 16-64 in OECD European countries, 2021\n\nNote: Countries are ordered in Panel A by descending order of the incidence of job strain. “Average” is the unweighted average of the 25 OECD European countries shown in this Chart. Figures in Panel B refer to the difference between job demands and job resources, capturing the degree of strain. **Highly strained** : Cases where the number of job demands exceeds by at least two the number of job resources. **Moderately strained** : Cases where the number of job demands exceeds by one the number of job resources. For further details on methodology, see Annex 1.B. Source: OECD calculations based on the European Working Conditions Telephone Survey (EWCTS) 2021 of the European Foundation for the Improvement of Living and Working Conditions (Eurofound).\n\nStatLink 2 https://stat.link/9qidkp\n\nCountry differences in the share of workers’ experiencing job strain are likely to reflect different factors, such as different sectoral and occupational structures, different labour and employment policies, ²⁹ as well as different phases of the pandemic and policy responses to it. Moreover, as data display results for 2021, i.e., a year after most OECD European countries implemented lockdown measures and social distance requirement, results may also be affected by teleworking. For instance, those who were able to work from home fared best, while frontline workers during the COVID-19 pandemic fared poorly on several fronts and reported much higher exposure to physical risks than those working from home (Eurofound, 2022[21]). Controlling for a number of observable factors, such as individual characteristics, occupation structures and teleworking, explains some of the country variation ³⁰ in job strain, notably in the case of Luxembourg, Spain and Switzerland. ³¹ Yet, significant cross-country variation remains unexplained that might be attributed to differences across countries in policy, norms, expectations, or attitudes towards job.\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25 %\n\n**A. Incidence of job strain**\n\nBoth sexes Men Women\n\nFRA\n\nCZE\n\nFIN\n\nBEL\n\nPOL\n\nDEU\n\nAUT\n\nSWE ESP DNK\n\nGRC\n\nNOR\n\nCHE\n\n**Average**\n\nIRL\n\nGBR\n\nITA\n\nSVK\n\nLVA\n\nHUN\n\nLTU\n\nLUX\n\nPRT\n\nNLD\n\nEST SVN\n\n0\n\n1\n\n2\n\n3\n\n4\n\n5\n\n6\n\n7\n\n6 7 8 9 10 11 12 13 14\n\n% of employees highly strained\n\n% of employees moderately strained\n\n**B. Degrees of job strain**","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p48.pdf"]},"response":{"errors":[],"extract_id":"extract_5fdfd9d4b0fe588173ec7e1c4f0f6117","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**46** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Figure 1.20. Job strain in OECD European countries, 2021**\n\nPercentage of employees aged 16-64 in OECD European countries, 2021\n\nNote: Countries are ordered in Panel A by descending order of the incidence of job strain. “Average” is the unweighted average of the 25 OECD European countries shown in this Chart. Figures in Panel B refer to the difference between job demands and job resources, capturing the degree of strain. **Highly strained** : Cases where the number of job demands exceeds by at least two the number of job resources. **Moderately strained** : Cases where the number of job demands exceeds by one the number of job resources. For further details on methodology, see Annex 1.B. Source: OECD calculations based on the European Working Conditions Telephone Survey (EWCTS) 2021 of the European Foundation for the Improvement of Living and Working Conditions (Eurofound).\n\nStatLink 2 https://stat.link/9qidkp\n\nCountry differences in the share of workers’ experiencing job strain are likely to reflect different factors, such as different sectoral and occupational structures, different labour and employment policies, ²⁹ as well as different phases of the pandemic and policy responses to it. Moreover, as data display results for 2021, i.e., a year after most OECD European countries implemented lockdown measures and social distance requirement, results may also be affected by teleworking. For instance, those who were able to work from home fared best, while frontline workers during the COVID-19 pandemic fared poorly on several fronts and reported much higher exposure to physical risks than those working from home (Eurofound, 2022[21]). Controlling for a number of observable factors, such as individual characteristics, occupation structures and teleworking, explains some of the country variation ³⁰ in job strain, notably in the case of Luxembourg, Spain and Switzerland. ³¹ Yet, significant cross-country variation remains unexplained that might be attributed to differences across countries in policy, norms, expectations, or attitudes towards job.\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25 %\n\n**A. Incidence of job strain**\n\nBoth sexes Men Women\n\nFRA\n\nCZE\n\nFIN\n\nBEL\n\nPOL\n\nDEU\n\nAUT\n\nSWE ESP DNK\n\nGRC\n\nNOR\n\nCHE\n\n**Average**\n\nIRL\n\nGBR\n\nITA\n\nSVK\n\nLVA\n\nHUN\n\nLTU\n\nLUX\n\nPRT\n\nNLD\n\nEST SVN\n\n0\n\n1\n\n2\n\n3\n\n4\n\n5\n\n6\n\n7\n\n6 7 8 9 10 11 12 13 14\n\n% of employees highly strained\n\n% of employees moderately strained\n\n**B. Degrees of job strain**"],"full_content":"**46** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Figure 1.20. Job strain in OECD European countries, 2021**\n\nPercentage of employees aged 16-64 in OECD European countries, 2021\n\nNote: Countries are ordered in Panel A by descending order of the incidence of job strain. “Average” is the unweighted average of the 25 OECD European countries shown in this Chart. Figures in Panel B refer to the difference between job demands and job resources, capturing the degree of strain. **Highly strained** : Cases where the number of job demands exceeds by at least two the number of job resources. **Moderately strained** : Cases where the number of job demands exceeds by one the number of job resources. For further details on methodology, see Annex 1.B. Source: OECD calculations based on the European Working Conditions Telephone Survey (EWCTS) 2021 of the European Foundation for the Improvement of Living and Working Conditions (Eurofound).\n\nStatLink 2 https://stat.link/9qidkp\n\nCountry differences in the share of workers’ experiencing job strain are likely to reflect different factors, such as different sectoral and occupational structures, different labour and employment policies, ²⁹ as well as different phases of the pandemic and policy responses to it. Moreover, as data display results for 2021, i.e., a year after most OECD European countries implemented lockdown measures and social distance requirement, results may also be affected by teleworking. For instance, those who were able to work from home fared best, while frontline workers during the COVID-19 pandemic fared poorly on several fronts and reported much higher exposure to physical risks than those working from home (Eurofound, 2022[21]). Controlling for a number of observable factors, such as individual characteristics, occupation structures and teleworking, explains some of the country variation ³⁰ in job strain, notably in the case of Luxembourg, Spain and Switzerland. ³¹ Yet, significant cross-country variation remains unexplained that might be attributed to differences across countries in policy, norms, expectations, or attitudes towards job.\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25 %\n\n**A. Incidence of job strain**\n\nBoth sexes Men Women\n\nFRA\n\nCZE\n\nFIN\n\nBEL\n\nPOL\n\nDEU\n\nAUT\n\nSWE ESP DNK\n\nGRC\n\nNOR\n\nCHE\n\n**Average**\n\nIRL\n\nGBR\n\nITA\n\nSVK\n\nLVA\n\nHUN\n\nLTU\n\nLUX\n\nPRT\n\nNLD\n\nEST SVN\n\n0\n\n1\n\n2\n\n3\n\n4\n\n5\n\n6\n\n7\n\n6 7 8 9 10 11 12 13 14\n\n% of employees highly strained\n\n% of employees moderately strained\n\n**B. Degrees of job strain**","publish_date":null,"title":"Figure 1.20. Job strain in OECD European countries, 2021","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p48.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p48.pdf"},"request":{"example_id":"chart/ac8b3538-en_p48","product_type":"parse","source_file_path":"ac8b3538-en_p48.pdf"},"started_at":"2026-11-04T19:46:03.773529"} +{"completed_at":"2026-11-04T19:46:07.730563","latency_in_ms":2119,"output":{"example_id":"chart/ac8b3538-en_p53","grounded_pages":[],"layout_pages":[],"markdown":" **51**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Annex Figure 1.A.3. Real wages remain below 2019 levels in several countries**\n\nCumulative percentage change in real hourly wage since Q4 2019\n\nNote: This Chart is a version of Figure 1.9 (Panel B) without any adjustments for seasonality in the CPI series. Otherwise noted, nominal hourly wages refer to a constant-industry-structure “wages and salaries” component of the labour cost index. Statistics refer to the private sector only for Costa Rica, Japan, Korea, Mexico and the United States. Nominal wage series are seasonally adjusted for all countries except for Canada, Costa Rica, Israel, Japan, Korea, Mexico, New Zealand and Switzerland. Nominal hourly wage presents a significant amount of unreported income for Mexico. **†** : Nominal hourly wage refers to the actual wage i.e. without any adjustment for sources of compositional shifts for Costa Rica, Israel, Korea, Mexico and the United Kingdom, and thus comparing these results with the other countries requires caution. Moreover, nominal hourly wage refers to the average monthly wages per employee job for Israel, and to the average weekly earnings for the United Kingdom. **‡** : Nominal hourly wage controls for additional sources of compositional shifts, such as regions for Australia, Canada and New Zealand, job characteristics and workers’ characteristics for Australia and New Zealand, gender for Switzerland, and occupations for the United States. For Switzerland, the quarterly estimates refer to the annual Swiss wage index. Real hourly wage is estimated by deflating the nominal hourly wage by the consumer price index (CPI-all items). Countries are ordered by descending order of the cumulative percentage changes in real hourly wages in Q1 2024 compared to Q4 2019. OECD is the unweighted average of the 35 OECD countries shown in this chart (not including Chile, Colombia and Türkiye). The trough refers to the quarter where real hourly wages were at their lowest value for the indicated country since Q4 2021. The latest date available refers to Q3 2023 for Israel, and Q4 2023 for Canada, Costa Rica, Israel, Japan, Korea, Mexico and New Zealand. Source: OECD calculations based on the Wage Price Index (Australian Bureau of Statistics) for **Australia** ; the Fixed weighted index of average hourly earnings for all employees (Statistics Canada) for **Canada** ; the Encuesta Continua de Empleo (Instituto Nacional de Estadística y Censos) for **Costa** **Rica** ; the Labour cost index by NACE Rev. 2 activity (Eurostat) for the **European** **countries** except the United Kingdom; the Wages and Employment Monthly Statistics (Central Bureau of Statistics) for **Israel** ; the Labour Force Survey at Establishments (Ministry of Employment and Labour) for **Korea** ; the Monthly Labour Survey (Ministry of Health, Labour and Welfare) for **Japan** ; the Encuesta Nacional de Ocupación y Empleo y Encuesta Nacional de Ocupación y Empleo Nueva Edición (Instituto Nacional de Estadística y Geografía) for **Mexico** ; the Labour Cost Index (Stats NZ) for **New Zealand** ; the Swiss Wage Index (Federal Statistical Office) for **Switzerland** ; the Monthly Wages and Salaries Survey (Office for National Statistics) for the **United Kingdom** ; and Employment Cost Index (Bureau of Labor Statistics, retrieved from FRED) for the **United States** . OECD (2024), “Prices: Consumer prices”, Main Economic Indicators (database), https://doi.org/10.1787/0f2e8000-en, (accessed on 21 June 2024).\n\nStatLink 2 https://stat.link/m9kfd5\n\n\\-15\n\n\\-10\n\n\\-5\n\n0\n\n5\n\n10\n\n15\n\n20 %\n\nQ1 2024 or latest Trough since Q4 2021","pages":[{"line_numbers":[],"markdown":" **51**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Annex Figure 1.A.3. Real wages remain below 2019 levels in several countries**\n\nCumulative percentage change in real hourly wage since Q4 2019\n\nNote: This Chart is a version of Figure 1.9 (Panel B) without any adjustments for seasonality in the CPI series. Otherwise noted, nominal hourly wages refer to a constant-industry-structure “wages and salaries” component of the labour cost index. Statistics refer to the private sector only for Costa Rica, Japan, Korea, Mexico and the United States. Nominal wage series are seasonally adjusted for all countries except for Canada, Costa Rica, Israel, Japan, Korea, Mexico, New Zealand and Switzerland. Nominal hourly wage presents a significant amount of unreported income for Mexico. **†** : Nominal hourly wage refers to the actual wage i.e. without any adjustment for sources of compositional shifts for Costa Rica, Israel, Korea, Mexico and the United Kingdom, and thus comparing these results with the other countries requires caution. Moreover, nominal hourly wage refers to the average monthly wages per employee job for Israel, and to the average weekly earnings for the United Kingdom. **‡** : Nominal hourly wage controls for additional sources of compositional shifts, such as regions for Australia, Canada and New Zealand, job characteristics and workers’ characteristics for Australia and New Zealand, gender for Switzerland, and occupations for the United States. For Switzerland, the quarterly estimates refer to the annual Swiss wage index. Real hourly wage is estimated by deflating the nominal hourly wage by the consumer price index (CPI-all items). Countries are ordered by descending order of the cumulative percentage changes in real hourly wages in Q1 2024 compared to Q4 2019. OECD is the unweighted average of the 35 OECD countries shown in this chart (not including Chile, Colombia and Türkiye). The trough refers to the quarter where real hourly wages were at their lowest value for the indicated country since Q4 2021. The latest date available refers to Q3 2023 for Israel, and Q4 2023 for Canada, Costa Rica, Israel, Japan, Korea, Mexico and New Zealand. Source: OECD calculations based on the Wage Price Index (Australian Bureau of Statistics) for **Australia** ; the Fixed weighted index of average hourly earnings for all employees (Statistics Canada) for **Canada** ; the Encuesta Continua de Empleo (Instituto Nacional de Estadística y Censos) for **Costa** **Rica** ; the Labour cost index by NACE Rev. 2 activity (Eurostat) for the **European** **countries** except the United Kingdom; the Wages and Employment Monthly Statistics (Central Bureau of Statistics) for **Israel** ; the Labour Force Survey at Establishments (Ministry of Employment and Labour) for **Korea** ; the Monthly Labour Survey (Ministry of Health, Labour and Welfare) for **Japan** ; the Encuesta Nacional de Ocupación y Empleo y Encuesta Nacional de Ocupación y Empleo Nueva Edición (Instituto Nacional de Estadística y Geografía) for **Mexico** ; the Labour Cost Index (Stats NZ) for **New Zealand** ; the Swiss Wage Index (Federal Statistical Office) for **Switzerland** ; the Monthly Wages and Salaries Survey (Office for National Statistics) for the **United Kingdom** ; and Employment Cost Index (Bureau of Labor Statistics, retrieved from FRED) for the **United States** . OECD (2024), “Prices: Consumer prices”, Main Economic Indicators (database), https://doi.org/10.1787/0f2e8000-en, (accessed on 21 June 2024).\n\nStatLink 2 https://stat.link/m9kfd5\n\n\\-15\n\n\\-10\n\n\\-5\n\n0\n\n5\n\n10\n\n15\n\n20 %\n\nQ1 2024 or latest Trough since Q4 2021","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":" **51**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Annex Figure 1.A.3. Real wages remain below 2019 levels in several countries**\n\nCumulative percentage change in real hourly wage since Q4 2019\n\nNote: This Chart is a version of Figure 1.9 (Panel B) without any adjustments for seasonality in the CPI series. Otherwise noted, nominal hourly wages refer to a constant-industry-structure “wages and salaries” component of the labour cost index. Statistics refer to the private sector only for Costa Rica, Japan, Korea, Mexico and the United States. Nominal wage series are seasonally adjusted for all countries except for Canada, Costa Rica, Israel, Japan, Korea, Mexico, New Zealand and Switzerland. Nominal hourly wage presents a significant amount of unreported income for Mexico. **†** : Nominal hourly wage refers to the actual wage i.e. without any adjustment for sources of compositional shifts for Costa Rica, Israel, Korea, Mexico and the United Kingdom, and thus comparing these results with the other countries requires caution. Moreover, nominal hourly wage refers to the average monthly wages per employee job for Israel, and to the average weekly earnings for the United Kingdom. **‡** : Nominal hourly wage controls for additional sources of compositional shifts, such as regions for Australia, Canada and New Zealand, job characteristics and workers’ characteristics for Australia and New Zealand, gender for Switzerland, and occupations for the United States. For Switzerland, the quarterly estimates refer to the annual Swiss wage index. Real hourly wage is estimated by deflating the nominal hourly wage by the consumer price index (CPI-all items). Countries are ordered by descending order of the cumulative percentage changes in real hourly wages in Q1 2024 compared to Q4 2019. OECD is the unweighted average of the 35 OECD countries shown in this chart (not including Chile, Colombia and Türkiye). The trough refers to the quarter where real hourly wages were at their lowest value for the indicated country since Q4 2021. The latest date available refers to Q3 2023 for Israel, and Q4 2023 for Canada, Costa Rica, Israel, Japan, Korea, Mexico and New Zealand. Source: OECD calculations based on the Wage Price Index (Australian Bureau of Statistics) for **Australia** ; the Fixed weighted index of average hourly earnings for all employees (Statistics Canada) for **Canada** ; the Encuesta Continua de Empleo (Instituto Nacional de Estadística y Censos) for **Costa** **Rica** ; the Labour cost index by NACE Rev. 2 activity (Eurostat) for the **European** **countries** except the United Kingdom; the Wages and Employment Monthly Statistics (Central Bureau of Statistics) for **Israel** ; the Labour Force Survey at Establishments (Ministry of Employment and Labour) for **Korea** ; the Monthly Labour Survey (Ministry of Health, Labour and Welfare) for **Japan** ; the Encuesta Nacional de Ocupación y Empleo y Encuesta Nacional de Ocupación y Empleo Nueva Edición (Instituto Nacional de Estadística y Geografía) for **Mexico** ; the Labour Cost Index (Stats NZ) for **New Zealand** ; the Swiss Wage Index (Federal Statistical Office) for **Switzerland** ; the Monthly Wages and Salaries Survey (Office for National Statistics) for the **United Kingdom** ; and Employment Cost Index (Bureau of Labor Statistics, retrieved from FRED) for the **United States** . OECD (2024), “Prices: Consumer prices”, Main Economic Indicators (database), https://doi.org/10.1787/0f2e8000-en, (accessed on 21 June 2024).\n\nStatLink 2 https://stat.link/m9kfd5\n\n\\-15\n\n\\-10\n\n\\-5\n\n0\n\n5\n\n10\n\n15\n\n20 %\n\nQ1 2024 or latest Trough since Q4 2021","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p53.pdf"]},"response":{"errors":[],"extract_id":"extract_d459ebc53891c32dbb981c47cadccfc1","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":[" **51**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Annex Figure 1.A.3. Real wages remain below 2019 levels in several countries**\n\nCumulative percentage change in real hourly wage since Q4 2019\n\nNote: This Chart is a version of Figure 1.9 (Panel B) without any adjustments for seasonality in the CPI series. Otherwise noted, nominal hourly wages refer to a constant-industry-structure “wages and salaries” component of the labour cost index. Statistics refer to the private sector only for Costa Rica, Japan, Korea, Mexico and the United States. Nominal wage series are seasonally adjusted for all countries except for Canada, Costa Rica, Israel, Japan, Korea, Mexico, New Zealand and Switzerland. Nominal hourly wage presents a significant amount of unreported income for Mexico. **†** : Nominal hourly wage refers to the actual wage i.e. without any adjustment for sources of compositional shifts for Costa Rica, Israel, Korea, Mexico and the United Kingdom, and thus comparing these results with the other countries requires caution. Moreover, nominal hourly wage refers to the average monthly wages per employee job for Israel, and to the average weekly earnings for the United Kingdom. **‡** : Nominal hourly wage controls for additional sources of compositional shifts, such as regions for Australia, Canada and New Zealand, job characteristics and workers’ characteristics for Australia and New Zealand, gender for Switzerland, and occupations for the United States. For Switzerland, the quarterly estimates refer to the annual Swiss wage index. Real hourly wage is estimated by deflating the nominal hourly wage by the consumer price index (CPI-all items). Countries are ordered by descending order of the cumulative percentage changes in real hourly wages in Q1 2024 compared to Q4 2019. OECD is the unweighted average of the 35 OECD countries shown in this chart (not including Chile, Colombia and Türkiye). The trough refers to the quarter where real hourly wages were at their lowest value for the indicated country since Q4 2021. The latest date available refers to Q3 2023 for Israel, and Q4 2023 for Canada, Costa Rica, Israel, Japan, Korea, Mexico and New Zealand. Source: OECD calculations based on the Wage Price Index (Australian Bureau of Statistics) for **Australia** ; the Fixed weighted index of average hourly earnings for all employees (Statistics Canada) for **Canada** ; the Encuesta Continua de Empleo (Instituto Nacional de Estadística y Censos) for **Costa** **Rica** ; the Labour cost index by NACE Rev. 2 activity (Eurostat) for the **European** **countries** except the United Kingdom; the Wages and Employment Monthly Statistics (Central Bureau of Statistics) for **Israel** ; the Labour Force Survey at Establishments (Ministry of Employment and Labour) for **Korea** ; the Monthly Labour Survey (Ministry of Health, Labour and Welfare) for **Japan** ; the Encuesta Nacional de Ocupación y Empleo y Encuesta Nacional de Ocupación y Empleo Nueva Edición (Instituto Nacional de Estadística y Geografía) for **Mexico** ; the Labour Cost Index (Stats NZ) for **New Zealand** ; the Swiss Wage Index (Federal Statistical Office) for **Switzerland** ; the Monthly Wages and Salaries Survey (Office for National Statistics) for the **United Kingdom** ; and Employment Cost Index (Bureau of Labor Statistics, retrieved from FRED) for the **United States** . OECD (2024), “Prices: Consumer prices”, Main Economic Indicators (database), https://doi.org/10.1787/0f2e8000-en, (accessed on 21 June 2024).\n\nStatLink 2 https://stat.link/m9kfd5\n\n\\-15\n\n\\-10\n\n\\-5\n\n0\n\n5\n\n10\n\n15\n\n20 %\n\nQ1 2024 or latest Trough since Q4 2021"],"full_content":" **51**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Annex Figure 1.A.3. Real wages remain below 2019 levels in several countries**\n\nCumulative percentage change in real hourly wage since Q4 2019\n\nNote: This Chart is a version of Figure 1.9 (Panel B) without any adjustments for seasonality in the CPI series. Otherwise noted, nominal hourly wages refer to a constant-industry-structure “wages and salaries” component of the labour cost index. Statistics refer to the private sector only for Costa Rica, Japan, Korea, Mexico and the United States. Nominal wage series are seasonally adjusted for all countries except for Canada, Costa Rica, Israel, Japan, Korea, Mexico, New Zealand and Switzerland. Nominal hourly wage presents a significant amount of unreported income for Mexico. **†** : Nominal hourly wage refers to the actual wage i.e. without any adjustment for sources of compositional shifts for Costa Rica, Israel, Korea, Mexico and the United Kingdom, and thus comparing these results with the other countries requires caution. Moreover, nominal hourly wage refers to the average monthly wages per employee job for Israel, and to the average weekly earnings for the United Kingdom. **‡** : Nominal hourly wage controls for additional sources of compositional shifts, such as regions for Australia, Canada and New Zealand, job characteristics and workers’ characteristics for Australia and New Zealand, gender for Switzerland, and occupations for the United States. For Switzerland, the quarterly estimates refer to the annual Swiss wage index. Real hourly wage is estimated by deflating the nominal hourly wage by the consumer price index (CPI-all items). Countries are ordered by descending order of the cumulative percentage changes in real hourly wages in Q1 2024 compared to Q4 2019. OECD is the unweighted average of the 35 OECD countries shown in this chart (not including Chile, Colombia and Türkiye). The trough refers to the quarter where real hourly wages were at their lowest value for the indicated country since Q4 2021. The latest date available refers to Q3 2023 for Israel, and Q4 2023 for Canada, Costa Rica, Israel, Japan, Korea, Mexico and New Zealand. Source: OECD calculations based on the Wage Price Index (Australian Bureau of Statistics) for **Australia** ; the Fixed weighted index of average hourly earnings for all employees (Statistics Canada) for **Canada** ; the Encuesta Continua de Empleo (Instituto Nacional de Estadística y Censos) for **Costa** **Rica** ; the Labour cost index by NACE Rev. 2 activity (Eurostat) for the **European** **countries** except the United Kingdom; the Wages and Employment Monthly Statistics (Central Bureau of Statistics) for **Israel** ; the Labour Force Survey at Establishments (Ministry of Employment and Labour) for **Korea** ; the Monthly Labour Survey (Ministry of Health, Labour and Welfare) for **Japan** ; the Encuesta Nacional de Ocupación y Empleo y Encuesta Nacional de Ocupación y Empleo Nueva Edición (Instituto Nacional de Estadística y Geografía) for **Mexico** ; the Labour Cost Index (Stats NZ) for **New Zealand** ; the Swiss Wage Index (Federal Statistical Office) for **Switzerland** ; the Monthly Wages and Salaries Survey (Office for National Statistics) for the **United Kingdom** ; and Employment Cost Index (Bureau of Labor Statistics, retrieved from FRED) for the **United States** . OECD (2024), “Prices: Consumer prices”, Main Economic Indicators (database), https://doi.org/10.1787/0f2e8000-en, (accessed on 21 June 2024).\n\nStatLink 2 https://stat.link/m9kfd5\n\n\\-15\n\n\\-10\n\n\\-5\n\n0\n\n5\n\n10\n\n15\n\n20 %\n\nQ1 2024 or latest Trough since Q4 2021","publish_date":null,"title":"Annex Figure 1.A.3. Real wages remain below 2019 levels in several countries","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p53.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p53.pdf"},"request":{"example_id":"chart/ac8b3538-en_p53","product_type":"parse","source_file_path":"ac8b3538-en_p53.pdf"},"started_at":"2026-11-04T19:46:05.611496"} +{"completed_at":"2026-11-04T19:46:09.108379","latency_in_ms":2084,"output":{"example_id":"chart/ac8b3538-en_p55","grounded_pages":[],"layout_pages":[],"markdown":" **53**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Annex Figure 1.B.2. Change in unemployment risk by gender**\n\nPercentage point change in unemployment rate by gender, 2015-22\n\nNote: OECD is the unweighted average of the 38 OECD countries shown in this Chart. Statistics for Canada, Greece, Hungary, Iceland, Israel, Italy, and Slovenia, refer to the percentage point change in 2015-21, to be consistent with the data shown in Figure 1.19. p.p: percentage point. Source: OECD calculations based on the OECD Labour Market Statistics (database), http://dx.doi.org/10.1787/data-00322-en (accessed on 24 April 2024).\n\nStatLink 2 https://stat.link/46w2xu\n\nExisting psychometric scales and indices of job strain provide critical guidance on the type of survey questions that can be used for measuring the various components of total job demands and that of total job resources. Yet, the precise set of questions to be selected among the many included in the EWCTS 2021 inevitably relies on judgment and depends on the purpose of the exercise. Since the approach followed in this chapter gives prominence to objective features of job quality, the questions chosen were those seeking objective and precise information (e.g. whether an individual can choose or change their methods of work), as well as readily interpretable in terms of the quality of the working environment (QWE). Annex Table 1.B.1 reports: i) the set of qualitative variables (i.e. EWCTS 2021 questions) retained to measure the various aspects of QWE; ii) the normalisation procedure used to compare these variables, initially measured on different scales; iii) the way these variables have been aggregated into a reduced number of components, which refer to broad categories of job demands or job resources.\n\n\\-12\n\n\\-10\n\n\\-8\n\n\\-6\n\n\\-4\n\n\\-2\n\n0\n\n2\n\n4\n\n6 p.p\n\nWomen ( ) Men","pages":[{"line_numbers":[],"markdown":" **53**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Annex Figure 1.B.2. Change in unemployment risk by gender**\n\nPercentage point change in unemployment rate by gender, 2015-22\n\nNote: OECD is the unweighted average of the 38 OECD countries shown in this Chart. Statistics for Canada, Greece, Hungary, Iceland, Israel, Italy, and Slovenia, refer to the percentage point change in 2015-21, to be consistent with the data shown in Figure 1.19. p.p: percentage point. Source: OECD calculations based on the OECD Labour Market Statistics (database), http://dx.doi.org/10.1787/data-00322-en (accessed on 24 April 2024).\n\nStatLink 2 https://stat.link/46w2xu\n\nExisting psychometric scales and indices of job strain provide critical guidance on the type of survey questions that can be used for measuring the various components of total job demands and that of total job resources. Yet, the precise set of questions to be selected among the many included in the EWCTS 2021 inevitably relies on judgment and depends on the purpose of the exercise. Since the approach followed in this chapter gives prominence to objective features of job quality, the questions chosen were those seeking objective and precise information (e.g. whether an individual can choose or change their methods of work), as well as readily interpretable in terms of the quality of the working environment (QWE). Annex Table 1.B.1 reports: i) the set of qualitative variables (i.e. EWCTS 2021 questions) retained to measure the various aspects of QWE; ii) the normalisation procedure used to compare these variables, initially measured on different scales; iii) the way these variables have been aggregated into a reduced number of components, which refer to broad categories of job demands or job resources.\n\n\\-12\n\n\\-10\n\n\\-8\n\n\\-6\n\n\\-4\n\n\\-2\n\n0\n\n2\n\n4\n\n6 p.p\n\nWomen ( ) Men","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":" **53**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Annex Figure 1.B.2. Change in unemployment risk by gender**\n\nPercentage point change in unemployment rate by gender, 2015-22\n\nNote: OECD is the unweighted average of the 38 OECD countries shown in this Chart. Statistics for Canada, Greece, Hungary, Iceland, Israel, Italy, and Slovenia, refer to the percentage point change in 2015-21, to be consistent with the data shown in Figure 1.19. p.p: percentage point. Source: OECD calculations based on the OECD Labour Market Statistics (database), http://dx.doi.org/10.1787/data-00322-en (accessed on 24 April 2024).\n\nStatLink 2 https://stat.link/46w2xu\n\nExisting psychometric scales and indices of job strain provide critical guidance on the type of survey questions that can be used for measuring the various components of total job demands and that of total job resources. Yet, the precise set of questions to be selected among the many included in the EWCTS 2021 inevitably relies on judgment and depends on the purpose of the exercise. Since the approach followed in this chapter gives prominence to objective features of job quality, the questions chosen were those seeking objective and precise information (e.g. whether an individual can choose or change their methods of work), as well as readily interpretable in terms of the quality of the working environment (QWE). Annex Table 1.B.1 reports: i) the set of qualitative variables (i.e. EWCTS 2021 questions) retained to measure the various aspects of QWE; ii) the normalisation procedure used to compare these variables, initially measured on different scales; iii) the way these variables have been aggregated into a reduced number of components, which refer to broad categories of job demands or job resources.\n\n\\-12\n\n\\-10\n\n\\-8\n\n\\-6\n\n\\-4\n\n\\-2\n\n0\n\n2\n\n4\n\n6 p.p\n\nWomen ( ) Men","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p55.pdf"]},"response":{"errors":[],"extract_id":"extract_b8439fada2a30f8f854ff2abba5c9c18","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":[" **53**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Annex Figure 1.B.2. Change in unemployment risk by gender**\n\nPercentage point change in unemployment rate by gender, 2015-22\n\nNote: OECD is the unweighted average of the 38 OECD countries shown in this Chart. Statistics for Canada, Greece, Hungary, Iceland, Israel, Italy, and Slovenia, refer to the percentage point change in 2015-21, to be consistent with the data shown in Figure 1.19. p.p: percentage point. Source: OECD calculations based on the OECD Labour Market Statistics (database), http://dx.doi.org/10.1787/data-00322-en (accessed on 24 April 2024).\n\nStatLink 2 https://stat.link/46w2xu\n\nExisting psychometric scales and indices of job strain provide critical guidance on the type of survey questions that can be used for measuring the various components of total job demands and that of total job resources. Yet, the precise set of questions to be selected among the many included in the EWCTS 2021 inevitably relies on judgment and depends on the purpose of the exercise. Since the approach followed in this chapter gives prominence to objective features of job quality, the questions chosen were those seeking objective and precise information (e.g. whether an individual can choose or change their methods of work), as well as readily interpretable in terms of the quality of the working environment (QWE). Annex Table 1.B.1 reports: i) the set of qualitative variables (i.e. EWCTS 2021 questions) retained to measure the various aspects of QWE; ii) the normalisation procedure used to compare these variables, initially measured on different scales; iii) the way these variables have been aggregated into a reduced number of components, which refer to broad categories of job demands or job resources.\n\n\\-12\n\n\\-10\n\n\\-8\n\n\\-6\n\n\\-4\n\n\\-2\n\n0\n\n2\n\n4\n\n6 p.p\n\nWomen ( ) Men"],"full_content":" **53**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Annex Figure 1.B.2. Change in unemployment risk by gender**\n\nPercentage point change in unemployment rate by gender, 2015-22\n\nNote: OECD is the unweighted average of the 38 OECD countries shown in this Chart. Statistics for Canada, Greece, Hungary, Iceland, Israel, Italy, and Slovenia, refer to the percentage point change in 2015-21, to be consistent with the data shown in Figure 1.19. p.p: percentage point. Source: OECD calculations based on the OECD Labour Market Statistics (database), http://dx.doi.org/10.1787/data-00322-en (accessed on 24 April 2024).\n\nStatLink 2 https://stat.link/46w2xu\n\nExisting psychometric scales and indices of job strain provide critical guidance on the type of survey questions that can be used for measuring the various components of total job demands and that of total job resources. Yet, the precise set of questions to be selected among the many included in the EWCTS 2021 inevitably relies on judgment and depends on the purpose of the exercise. Since the approach followed in this chapter gives prominence to objective features of job quality, the questions chosen were those seeking objective and precise information (e.g. whether an individual can choose or change their methods of work), as well as readily interpretable in terms of the quality of the working environment (QWE). Annex Table 1.B.1 reports: i) the set of qualitative variables (i.e. EWCTS 2021 questions) retained to measure the various aspects of QWE; ii) the normalisation procedure used to compare these variables, initially measured on different scales; iii) the way these variables have been aggregated into a reduced number of components, which refer to broad categories of job demands or job resources.\n\n\\-12\n\n\\-10\n\n\\-8\n\n\\-6\n\n\\-4\n\n\\-2\n\n0\n\n2\n\n4\n\n6 p.p\n\nWomen ( ) Men","publish_date":null,"title":"Annex Figure 1.B.2. Change in unemployment risk by gender","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p55.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p55.pdf"},"request":{"example_id":"chart/ac8b3538-en_p55","product_type":"parse","source_file_path":"ac8b3538-en_p55.pdf"},"started_at":"2026-11-04T19:46:07.024222"} +{"completed_at":"2026-11-04T19:46:10.662863","latency_in_ms":2057,"output":{"example_id":"chart/ac8b3538-en_p61","grounded_pages":[],"layout_pages":[],"markdown":" **59**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Annex Figure 1.C.1. Minimum wage evolution, May 2019 to May 2024**\n\nNominal and real minimum wage, cumulative percentage change since May 2019\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Australia**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Belgium**\n\n\\-40\n\n0\n\n40\n\n80\n\n120\n\n160\n\n**Canada**\n\nFederal Jurisiction\n\nNominal minimum wage Real minimum wage\n\n\\-20\n\n0\n\n20\n\n40\n\n60\n\n**Canada** Weighted\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Chile**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Colombia**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Costa Rica**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Czechia**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Estonia**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**France**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Germany**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Greece**\n\n\\-30 -15\n\n0 15 30 45 60 75 90\n\n**Hungary**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Ireland**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Israel**","pages":[{"line_numbers":[],"markdown":" **59**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Annex Figure 1.C.1. Minimum wage evolution, May 2019 to May 2024**\n\nNominal and real minimum wage, cumulative percentage change since May 2019\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Australia**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Belgium**\n\n\\-40\n\n0\n\n40\n\n80\n\n120\n\n160\n\n**Canada**\n\nFederal Jurisiction\n\nNominal minimum wage Real minimum wage\n\n\\-20\n\n0\n\n20\n\n40\n\n60\n\n**Canada** Weighted\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Chile**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Colombia**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Costa Rica**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Czechia**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Estonia**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**France**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Germany**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Greece**\n\n\\-30 -15\n\n0 15 30 45 60 75 90\n\n**Hungary**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Ireland**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Israel**","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":" **59**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Annex Figure 1.C.1. Minimum wage evolution, May 2019 to May 2024**\n\nNominal and real minimum wage, cumulative percentage change since May 2019\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Australia**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Belgium**\n\n\\-40\n\n0\n\n40\n\n80\n\n120\n\n160\n\n**Canada**\n\nFederal Jurisiction\n\nNominal minimum wage Real minimum wage\n\n\\-20\n\n0\n\n20\n\n40\n\n60\n\n**Canada** Weighted\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Chile**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Colombia**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Costa Rica**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Czechia**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Estonia**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**France**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Germany**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Greece**\n\n\\-30 -15\n\n0 15 30 45 60 75 90\n\n**Hungary**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Ireland**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Israel**","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p61.pdf"]},"response":{"errors":[],"extract_id":"extract_389ee41946d434de810fdf720fa38391","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":[" **59**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Annex Figure 1.C.1. Minimum wage evolution, May 2019 to May 2024**\n\nNominal and real minimum wage, cumulative percentage change since May 2019\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Australia**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Belgium**\n\n\\-40\n\n0\n\n40\n\n80\n\n120\n\n160\n\n**Canada**\n\nFederal Jurisiction\n\nNominal minimum wage Real minimum wage\n\n\\-20\n\n0\n\n20\n\n40\n\n60\n\n**Canada** Weighted\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Chile**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Colombia**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Costa Rica**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Czechia**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Estonia**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**France**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Germany**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Greece**\n\n\\-30 -15\n\n0 15 30 45 60 75 90\n\n**Hungary**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Ireland**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Israel**"],"full_content":" **59**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Annex Figure 1.C.1. Minimum wage evolution, May 2019 to May 2024**\n\nNominal and real minimum wage, cumulative percentage change since May 2019\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Australia**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Belgium**\n\n\\-40\n\n0\n\n40\n\n80\n\n120\n\n160\n\n**Canada**\n\nFederal Jurisiction\n\nNominal minimum wage Real minimum wage\n\n\\-20\n\n0\n\n20\n\n40\n\n60\n\n**Canada** Weighted\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Chile**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Colombia**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Costa Rica**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Czechia**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Estonia**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**France**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Germany**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Greece**\n\n\\-30 -15\n\n0 15 30 45 60 75 90\n\n**Hungary**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Ireland**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Israel**","publish_date":null,"title":"Annex Figure 1.C.1. Minimum wage evolution, May 2019 to May 2024","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p61.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p61.pdf"},"request":{"example_id":"chart/ac8b3538-en_p61","product_type":"parse","source_file_path":"ac8b3538-en_p61.pdf"},"started_at":"2026-11-04T19:46:08.605436"} +{"completed_at":"2026-11-04T19:46:13.131277","latency_in_ms":2467,"output":{"example_id":"chart/ac8b3538-en_p62","grounded_pages":[],"layout_pages":[],"markdown":"**60** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Japan**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Korea**\n\n\\-30\n\n\\-15\n\n0\n\n15\n\n30\n\n45\n\n60\n\n75\n\n**Latvia**\n\nNominal minimum wage Real minimum wage\n\n\\-30\n\n\\-15\n\n0\n\n15\n\n30\n\n45\n\n60\n\n75\n\n**Lithuania**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Luxembourg**\n\n\\-60\n\n\\-30\n\n0\n\n30\n\n60\n\n90\n\n120\n\n150\n\n**Mexico**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Netherlands**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**New Zealand**\n\n\\-40\n\n\\-20\n\n0\n\n20\n\n40\n\n60\n\n80\n\n100\n\n**Poland**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Portugal**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Slovak Republic**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Slovenia**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Spain**\n\n\\-200\n\n0\n\n200\n\n400\n\n600\n\n800\n\n**Türkiye**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**United Kingdom**","pages":[{"line_numbers":[],"markdown":"**60** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Japan**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Korea**\n\n\\-30\n\n\\-15\n\n0\n\n15\n\n30\n\n45\n\n60\n\n75\n\n**Latvia**\n\nNominal minimum wage Real minimum wage\n\n\\-30\n\n\\-15\n\n0\n\n15\n\n30\n\n45\n\n60\n\n75\n\n**Lithuania**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Luxembourg**\n\n\\-60\n\n\\-30\n\n0\n\n30\n\n60\n\n90\n\n120\n\n150\n\n**Mexico**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Netherlands**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**New Zealand**\n\n\\-40\n\n\\-20\n\n0\n\n20\n\n40\n\n60\n\n80\n\n100\n\n**Poland**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Portugal**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Slovak Republic**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Slovenia**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Spain**\n\n\\-200\n\n0\n\n200\n\n400\n\n600\n\n800\n\n**Türkiye**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**United Kingdom**","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**60** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Japan**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Korea**\n\n\\-30\n\n\\-15\n\n0\n\n15\n\n30\n\n45\n\n60\n\n75\n\n**Latvia**\n\nNominal minimum wage Real minimum wage\n\n\\-30\n\n\\-15\n\n0\n\n15\n\n30\n\n45\n\n60\n\n75\n\n**Lithuania**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Luxembourg**\n\n\\-60\n\n\\-30\n\n0\n\n30\n\n60\n\n90\n\n120\n\n150\n\n**Mexico**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Netherlands**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**New Zealand**\n\n\\-40\n\n\\-20\n\n0\n\n20\n\n40\n\n60\n\n80\n\n100\n\n**Poland**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Portugal**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Slovak Republic**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Slovenia**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Spain**\n\n\\-200\n\n0\n\n200\n\n400\n\n600\n\n800\n\n**Türkiye**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**United Kingdom**","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p62.pdf"]},"response":{"errors":[],"extract_id":"extract_a6765a92909729a1c12dbf4ab0d99458","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**60** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Japan**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Korea**\n\n\\-30\n\n\\-15\n\n0\n\n15\n\n30\n\n45\n\n60\n\n75\n\n**Latvia**\n\nNominal minimum wage Real minimum wage\n\n\\-30\n\n\\-15\n\n0\n\n15\n\n30\n\n45\n\n60\n\n75\n\n**Lithuania**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Luxembourg**\n\n\\-60\n\n\\-30\n\n0\n\n30\n\n60\n\n90\n\n120\n\n150\n\n**Mexico**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Netherlands**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**New Zealand**\n\n\\-40\n\n\\-20\n\n0\n\n20\n\n40\n\n60\n\n80\n\n100\n\n**Poland**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Portugal**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Slovak Republic**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Slovenia**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Spain**\n\n\\-200\n\n0\n\n200\n\n400\n\n600\n\n800\n\n**Türkiye**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**United Kingdom**"],"full_content":"**60** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Japan**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Korea**\n\n\\-30\n\n\\-15\n\n0\n\n15\n\n30\n\n45\n\n60\n\n75\n\n**Latvia**\n\nNominal minimum wage Real minimum wage\n\n\\-30\n\n\\-15\n\n0\n\n15\n\n30\n\n45\n\n60\n\n75\n\n**Lithuania**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Luxembourg**\n\n\\-60\n\n\\-30\n\n0\n\n30\n\n60\n\n90\n\n120\n\n150\n\n**Mexico**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Netherlands**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**New Zealand**\n\n\\-40\n\n\\-20\n\n0\n\n20\n\n40\n\n60\n\n80\n\n100\n\n**Poland**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Portugal**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Slovak Republic**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Slovenia**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**Spain**\n\n\\-200\n\n0\n\n200\n\n400\n\n600\n\n800\n\n**Türkiye**\n\n\\-20 -10\n\n0 10 20 30 40 50 60\n\n**United Kingdom**","publish_date":null,"title":"60 ","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p62.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p62.pdf"},"request":{"example_id":"chart/ac8b3538-en_p62","product_type":"parse","source_file_path":"ac8b3538-en_p62.pdf"},"started_at":"2026-11-04T19:46:10.664259"} +{"completed_at":"2026-11-04T19:46:15.126279","latency_in_ms":2543,"output":{"example_id":"chart/ac8b3538-en_p64","grounded_pages":[],"layout_pages":[],"markdown":"**62** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Annex Figure 1.C.2. Real wages in selected OECD countries resulting from collective agreements**\n\nYear-on-year percentage change, Q1 2018 to Q1 2024\n\nNote: International comparability of data on negotiated wages is affected by differences in definitions and measurement. Statistics are representative of all employees covered by a collective wage agreement for Austria, Belgium, the Euro Area (20), France, Germany, Italy, the Netherlands, Sweden, and the United States. In Canada, statistics refer to collective bargaining settlements of all bargaining units covering 500 or more employees (units of 100 or more employees for the Federal Jurisdiction). For Australia and Canada, statistics refer only to employees affected by an increase of the negotiated wage at date. Wage increases in Austria, Belgium, the Euro Area (20), Germany, Italy, the Netherlands, Sweden, and the United States refers to the average increase in negotiated wages (wages of union workers for the United States) weighted by the employment composition for a reference year (Laspeyres index). The reference year of the employment composition used is 2009 for Sweden, 2010 for Belgium and the Netherlands, January 2015 for the Euro Area (20), 2015 for Germany and Italy, 2016 for Austria, and 2021 for the United States. For Australia, Canada and France, wage increases refer to the average increase in negotiated wages weighted by the number of employees affected in the period considered. Private sector for Germany refers to all industries excluding agriculture, public administration, education, health, and other personal services (Sections B to N of the NACE rev. 2). Source: OECD calculations based on national data on negotiated wages, see Annex Table 1.C.3.in (Araki et al., 2023[11]) for further details; and OECD (2024), “Prices: Consumer prices”, Main Economic Indicators (database), https://doi.org/10.1787/0f2e8000-en (accessed on 28 June 2024).\n\nStatLink 2 https://stat.link/wz8xun\n\n\\-10 -8 -6 -4 -2 0 2 4 6 8\n\n2018 2019 2020 2021 2022 2023 2024\n\n% **Euro Area**\n\n\\-10 -8 -6 -4 -2 0 2 4 6 8\n\n2018 2019 2020 2021 2022 2023 2024\n\n% **Australia**\n\n\\-10 -8 -6 -4 -2 0 2 4 6 8\n\n2018 2019 2020 2021 2022 2023 2024\n\n% **Austria**\n\nBasic wage (all sectors) Basic wage (private sector)\n\nWage incl. lump sums and special payments (all sectors) Wage incl. lump sums and special payments (private sector)\n\n\\-10 -8 -6 -4 -2 0 2 4 6 8\n\n2018 2019 2020 2021 2022 2023 2024\n\n% **Belgium**\n\n\\-10 -8 -6 -4 -2 0 2 4 6 8\n\n2018 2019 2020 2021 2022 2023 2024\n\n% **Canada**\n\n\\-10 -8 -6 -4 -2 0 2 4 6 8\n\n2018 2019 2020 2021 2022 2023 2024\n\n% **France**\n\n\\-10 -8 -6 -4 -2 0 2 4 6 8\n\n2018 2019 2020 2021 2022 2023 2024\n\n% **Germany**\n\n\\-10 -8 -6 -4 -2 0 2 4 6 8\n\n2018 2019 2020 2021 2022 2023 2024\n\n% **Italy**\n\n\\-10 -8 -6 -4 -2 0 2 4 6 8\n\n2018 2019 2020 2021 2022 2023 2024\n\n% **Netherlands**\n\n\\-10 -8 -6 -4 -2 0 2 4 6 8\n\n2018 2019 2020 2021 2022 2023 2024\n\n% **Sweden**\n\n\\-10 -8 -6 -4 -2 0 2 4 6 8\n\n2018 2019 2020 2021 2022 2023 2024\n\n% **United States**","pages":[{"line_numbers":[],"markdown":"**62** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Annex Figure 1.C.2. Real wages in selected OECD countries resulting from collective agreements**\n\nYear-on-year percentage change, Q1 2018 to Q1 2024\n\nNote: International comparability of data on negotiated wages is affected by differences in definitions and measurement. Statistics are representative of all employees covered by a collective wage agreement for Austria, Belgium, the Euro Area (20), France, Germany, Italy, the Netherlands, Sweden, and the United States. In Canada, statistics refer to collective bargaining settlements of all bargaining units covering 500 or more employees (units of 100 or more employees for the Federal Jurisdiction). For Australia and Canada, statistics refer only to employees affected by an increase of the negotiated wage at date. Wage increases in Austria, Belgium, the Euro Area (20), Germany, Italy, the Netherlands, Sweden, and the United States refers to the average increase in negotiated wages (wages of union workers for the United States) weighted by the employment composition for a reference year (Laspeyres index). The reference year of the employment composition used is 2009 for Sweden, 2010 for Belgium and the Netherlands, January 2015 for the Euro Area (20), 2015 for Germany and Italy, 2016 for Austria, and 2021 for the United States. For Australia, Canada and France, wage increases refer to the average increase in negotiated wages weighted by the number of employees affected in the period considered. Private sector for Germany refers to all industries excluding agriculture, public administration, education, health, and other personal services (Sections B to N of the NACE rev. 2). Source: OECD calculations based on national data on negotiated wages, see Annex Table 1.C.3.in (Araki et al., 2023[11]) for further details; and OECD (2024), “Prices: Consumer prices”, Main Economic Indicators (database), https://doi.org/10.1787/0f2e8000-en (accessed on 28 June 2024).\n\nStatLink 2 https://stat.link/wz8xun\n\n\\-10 -8 -6 -4 -2 0 2 4 6 8\n\n2018 2019 2020 2021 2022 2023 2024\n\n% **Euro Area**\n\n\\-10 -8 -6 -4 -2 0 2 4 6 8\n\n2018 2019 2020 2021 2022 2023 2024\n\n% **Australia**\n\n\\-10 -8 -6 -4 -2 0 2 4 6 8\n\n2018 2019 2020 2021 2022 2023 2024\n\n% **Austria**\n\nBasic wage (all sectors) Basic wage (private sector)\n\nWage incl. lump sums and special payments (all sectors) Wage incl. lump sums and special payments (private sector)\n\n\\-10 -8 -6 -4 -2 0 2 4 6 8\n\n2018 2019 2020 2021 2022 2023 2024\n\n% **Belgium**\n\n\\-10 -8 -6 -4 -2 0 2 4 6 8\n\n2018 2019 2020 2021 2022 2023 2024\n\n% **Canada**\n\n\\-10 -8 -6 -4 -2 0 2 4 6 8\n\n2018 2019 2020 2021 2022 2023 2024\n\n% **France**\n\n\\-10 -8 -6 -4 -2 0 2 4 6 8\n\n2018 2019 2020 2021 2022 2023 2024\n\n% **Germany**\n\n\\-10 -8 -6 -4 -2 0 2 4 6 8\n\n2018 2019 2020 2021 2022 2023 2024\n\n% **Italy**\n\n\\-10 -8 -6 -4 -2 0 2 4 6 8\n\n2018 2019 2020 2021 2022 2023 2024\n\n% **Netherlands**\n\n\\-10 -8 -6 -4 -2 0 2 4 6 8\n\n2018 2019 2020 2021 2022 2023 2024\n\n% **Sweden**\n\n\\-10 -8 -6 -4 -2 0 2 4 6 8\n\n2018 2019 2020 2021 2022 2023 2024\n\n% **United States**","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**62** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Annex Figure 1.C.2. Real wages in selected OECD countries resulting from collective agreements**\n\nYear-on-year percentage change, Q1 2018 to Q1 2024\n\nNote: International comparability of data on negotiated wages is affected by differences in definitions and measurement. Statistics are representative of all employees covered by a collective wage agreement for Austria, Belgium, the Euro Area (20), France, Germany, Italy, the Netherlands, Sweden, and the United States. In Canada, statistics refer to collective bargaining settlements of all bargaining units covering 500 or more employees (units of 100 or more employees for the Federal Jurisdiction). For Australia and Canada, statistics refer only to employees affected by an increase of the negotiated wage at date. Wage increases in Austria, Belgium, the Euro Area (20), Germany, Italy, the Netherlands, Sweden, and the United States refers to the average increase in negotiated wages (wages of union workers for the United States) weighted by the employment composition for a reference year (Laspeyres index). The reference year of the employment composition used is 2009 for Sweden, 2010 for Belgium and the Netherlands, January 2015 for the Euro Area (20), 2015 for Germany and Italy, 2016 for Austria, and 2021 for the United States. For Australia, Canada and France, wage increases refer to the average increase in negotiated wages weighted by the number of employees affected in the period considered. Private sector for Germany refers to all industries excluding agriculture, public administration, education, health, and other personal services (Sections B to N of the NACE rev. 2). Source: OECD calculations based on national data on negotiated wages, see Annex Table 1.C.3.in (Araki et al., 2023[11]) for further details; and OECD (2024), “Prices: Consumer prices”, Main Economic Indicators (database), https://doi.org/10.1787/0f2e8000-en (accessed on 28 June 2024).\n\nStatLink 2 https://stat.link/wz8xun\n\n\\-10 -8 -6 -4 -2 0 2 4 6 8\n\n2018 2019 2020 2021 2022 2023 2024\n\n% **Euro Area**\n\n\\-10 -8 -6 -4 -2 0 2 4 6 8\n\n2018 2019 2020 2021 2022 2023 2024\n\n% **Australia**\n\n\\-10 -8 -6 -4 -2 0 2 4 6 8\n\n2018 2019 2020 2021 2022 2023 2024\n\n% **Austria**\n\nBasic wage (all sectors) Basic wage (private sector)\n\nWage incl. lump sums and special payments (all sectors) Wage incl. lump sums and special payments (private sector)\n\n\\-10 -8 -6 -4 -2 0 2 4 6 8\n\n2018 2019 2020 2021 2022 2023 2024\n\n% **Belgium**\n\n\\-10 -8 -6 -4 -2 0 2 4 6 8\n\n2018 2019 2020 2021 2022 2023 2024\n\n% **Canada**\n\n\\-10 -8 -6 -4 -2 0 2 4 6 8\n\n2018 2019 2020 2021 2022 2023 2024\n\n% **France**\n\n\\-10 -8 -6 -4 -2 0 2 4 6 8\n\n2018 2019 2020 2021 2022 2023 2024\n\n% **Germany**\n\n\\-10 -8 -6 -4 -2 0 2 4 6 8\n\n2018 2019 2020 2021 2022 2023 2024\n\n% **Italy**\n\n\\-10 -8 -6 -4 -2 0 2 4 6 8\n\n2018 2019 2020 2021 2022 2023 2024\n\n% **Netherlands**\n\n\\-10 -8 -6 -4 -2 0 2 4 6 8\n\n2018 2019 2020 2021 2022 2023 2024\n\n% **Sweden**\n\n\\-10 -8 -6 -4 -2 0 2 4 6 8\n\n2018 2019 2020 2021 2022 2023 2024\n\n% **United States**","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p64.pdf"]},"response":{"errors":[],"extract_id":"extract_6e92b0e39bf802ae5363ef1ec1fc940c","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**62** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Annex Figure 1.C.2. Real wages in selected OECD countries resulting from collective agreements**\n\nYear-on-year percentage change, Q1 2018 to Q1 2024\n\nNote: International comparability of data on negotiated wages is affected by differences in definitions and measurement. Statistics are representative of all employees covered by a collective wage agreement for Austria, Belgium, the Euro Area (20), France, Germany, Italy, the Netherlands, Sweden, and the United States. In Canada, statistics refer to collective bargaining settlements of all bargaining units covering 500 or more employees (units of 100 or more employees for the Federal Jurisdiction). For Australia and Canada, statistics refer only to employees affected by an increase of the negotiated wage at date. Wage increases in Austria, Belgium, the Euro Area (20), Germany, Italy, the Netherlands, Sweden, and the United States refers to the average increase in negotiated wages (wages of union workers for the United States) weighted by the employment composition for a reference year (Laspeyres index). The reference year of the employment composition used is 2009 for Sweden, 2010 for Belgium and the Netherlands, January 2015 for the Euro Area (20), 2015 for Germany and Italy, 2016 for Austria, and 2021 for the United States. For Australia, Canada and France, wage increases refer to the average increase in negotiated wages weighted by the number of employees affected in the period considered. Private sector for Germany refers to all industries excluding agriculture, public administration, education, health, and other personal services (Sections B to N of the NACE rev. 2). Source: OECD calculations based on national data on negotiated wages, see Annex Table 1.C.3.in (Araki et al., 2023[11]) for further details; and OECD (2024), “Prices: Consumer prices”, Main Economic Indicators (database), https://doi.org/10.1787/0f2e8000-en (accessed on 28 June 2024).\n\nStatLink 2 https://stat.link/wz8xun\n\n\\-10 -8 -6 -4 -2 0 2 4 6 8\n\n2018 2019 2020 2021 2022 2023 2024\n\n% **Euro Area**\n\n\\-10 -8 -6 -4 -2 0 2 4 6 8\n\n2018 2019 2020 2021 2022 2023 2024\n\n% **Australia**\n\n\\-10 -8 -6 -4 -2 0 2 4 6 8\n\n2018 2019 2020 2021 2022 2023 2024\n\n% **Austria**\n\nBasic wage (all sectors) Basic wage (private sector)\n\nWage incl. lump sums and special payments (all sectors) Wage incl. lump sums and special payments (private sector)\n\n\\-10 -8 -6 -4 -2 0 2 4 6 8\n\n2018 2019 2020 2021 2022 2023 2024\n\n% **Belgium**\n\n\\-10 -8 -6 -4 -2 0 2 4 6 8\n\n2018 2019 2020 2021 2022 2023 2024\n\n% **Canada**\n\n\\-10 -8 -6 -4 -2 0 2 4 6 8\n\n2018 2019 2020 2021 2022 2023 2024\n\n% **France**\n\n\\-10 -8 -6 -4 -2 0 2 4 6 8\n\n2018 2019 2020 2021 2022 2023 2024\n\n% **Germany**\n\n\\-10 -8 -6 -4 -2 0 2 4 6 8\n\n2018 2019 2020 2021 2022 2023 2024\n\n% **Italy**\n\n\\-10 -8 -6 -4 -2 0 2 4 6 8\n\n2018 2019 2020 2021 2022 2023 2024\n\n% **Netherlands**\n\n\\-10 -8 -6 -4 -2 0 2 4 6 8\n\n2018 2019 2020 2021 2022 2023 2024\n\n% **Sweden**\n\n\\-10 -8 -6 -4 -2 0 2 4 6 8\n\n2018 2019 2020 2021 2022 2023 2024\n\n% **United States**"],"full_content":"**62** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Annex Figure 1.C.2. Real wages in selected OECD countries resulting from collective agreements**\n\nYear-on-year percentage change, Q1 2018 to Q1 2024\n\nNote: International comparability of data on negotiated wages is affected by differences in definitions and measurement. Statistics are representative of all employees covered by a collective wage agreement for Austria, Belgium, the Euro Area (20), France, Germany, Italy, the Netherlands, Sweden, and the United States. In Canada, statistics refer to collective bargaining settlements of all bargaining units covering 500 or more employees (units of 100 or more employees for the Federal Jurisdiction). For Australia and Canada, statistics refer only to employees affected by an increase of the negotiated wage at date. Wage increases in Austria, Belgium, the Euro Area (20), Germany, Italy, the Netherlands, Sweden, and the United States refers to the average increase in negotiated wages (wages of union workers for the United States) weighted by the employment composition for a reference year (Laspeyres index). The reference year of the employment composition used is 2009 for Sweden, 2010 for Belgium and the Netherlands, January 2015 for the Euro Area (20), 2015 for Germany and Italy, 2016 for Austria, and 2021 for the United States. For Australia, Canada and France, wage increases refer to the average increase in negotiated wages weighted by the number of employees affected in the period considered. Private sector for Germany refers to all industries excluding agriculture, public administration, education, health, and other personal services (Sections B to N of the NACE rev. 2). Source: OECD calculations based on national data on negotiated wages, see Annex Table 1.C.3.in (Araki et al., 2023[11]) for further details; and OECD (2024), “Prices: Consumer prices”, Main Economic Indicators (database), https://doi.org/10.1787/0f2e8000-en (accessed on 28 June 2024).\n\nStatLink 2 https://stat.link/wz8xun\n\n\\-10 -8 -6 -4 -2 0 2 4 6 8\n\n2018 2019 2020 2021 2022 2023 2024\n\n% **Euro Area**\n\n\\-10 -8 -6 -4 -2 0 2 4 6 8\n\n2018 2019 2020 2021 2022 2023 2024\n\n% **Australia**\n\n\\-10 -8 -6 -4 -2 0 2 4 6 8\n\n2018 2019 2020 2021 2022 2023 2024\n\n% **Austria**\n\nBasic wage (all sectors) Basic wage (private sector)\n\nWage incl. lump sums and special payments (all sectors) Wage incl. lump sums and special payments (private sector)\n\n\\-10 -8 -6 -4 -2 0 2 4 6 8\n\n2018 2019 2020 2021 2022 2023 2024\n\n% **Belgium**\n\n\\-10 -8 -6 -4 -2 0 2 4 6 8\n\n2018 2019 2020 2021 2022 2023 2024\n\n% **Canada**\n\n\\-10 -8 -6 -4 -2 0 2 4 6 8\n\n2018 2019 2020 2021 2022 2023 2024\n\n% **France**\n\n\\-10 -8 -6 -4 -2 0 2 4 6 8\n\n2018 2019 2020 2021 2022 2023 2024\n\n% **Germany**\n\n\\-10 -8 -6 -4 -2 0 2 4 6 8\n\n2018 2019 2020 2021 2022 2023 2024\n\n% **Italy**\n\n\\-10 -8 -6 -4 -2 0 2 4 6 8\n\n2018 2019 2020 2021 2022 2023 2024\n\n% **Netherlands**\n\n\\-10 -8 -6 -4 -2 0 2 4 6 8\n\n2018 2019 2020 2021 2022 2023 2024\n\n% **Sweden**\n\n\\-10 -8 -6 -4 -2 0 2 4 6 8\n\n2018 2019 2020 2021 2022 2023 2024\n\n% **United States**","publish_date":null,"title":"Annex Figure 1.C.2. Real wages in selected OECD countries resulting from collective agreements","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p64.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p64.pdf"},"request":{"example_id":"chart/ac8b3538-en_p64","product_type":"parse","source_file_path":"ac8b3538-en_p64.pdf"},"started_at":"2026-11-04T19:46:12.582883"} +{"completed_at":"2026-11-04T19:46:17.272608","latency_in_ms":2979,"output":{"example_id":"chart/ac8b3538-en_p76","grounded_pages":[],"layout_pages":[],"markdown":"**74** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Figure 2.1. One in seven workers report heat-related discomfort in Europe and the United States**\n\nPercentage of employment who declare to be exposed to high temperature at least half of their working time, 2015\n\nNote: Those who declare to be exposed to high temperature which make them perspire even when not working at least half of their working time are classified as at-risk. The shares are the predictions in May, obtained by running a regression of the exposure to heat at country level on the share of respondents by month of interview. For Panel A, the average is unweighted. For Panel B, averages are unweighted and include Austria, Belgium, Czechia, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Ireland, Italy, Latvia, Lithuania, Luxembourg, the Netherlands, Norway, Poland, Portugal, the Slovak Republic, Slovenia, Spain, Sweden, Switzerland, the United Kingdom and the United States. Youth are those aged from 15 to 24, prime from 25 to 54, and senior from 55 and above. For urbanisation, the United States is not included. Source: Secretariat’s calculations based on data from European Working Conditions Survey (Eurofound) and American Working Conditions Survey (RAND Corporation).\n\nStatLink 2 https://stat.link/852oij\n\nOccupational safety and health (OSH) regulations in OECD countries already cover aspects related to temperature, heat stress and extreme weather at work. However, some countries, often with the direct involvement of social partners, have adapted the legislation to the specific challenges related to climate change (e.g. Lithuania) or developed specific OSH programmes and tools (e.g. Germany, Lithuania,\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\nWomen\n\nMen\n\nYouth\n\nPrime\n\nSenior\n\nLow-skilled\n\nMiddle-skilled\n\nHigh-skilled\n\nLow\n\nMiddle\n\nHigh\n\nRural\n\nUrban\n\nSuburban\n\nGender Age Occupation Education Urbanisation\n\n%\n\n**B. OECD27 average by workers' characteristic**\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30 %\n\n**A.Country average**","pages":[{"line_numbers":[],"markdown":"**74** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Figure 2.1. One in seven workers report heat-related discomfort in Europe and the United States**\n\nPercentage of employment who declare to be exposed to high temperature at least half of their working time, 2015\n\nNote: Those who declare to be exposed to high temperature which make them perspire even when not working at least half of their working time are classified as at-risk. The shares are the predictions in May, obtained by running a regression of the exposure to heat at country level on the share of respondents by month of interview. For Panel A, the average is unweighted. For Panel B, averages are unweighted and include Austria, Belgium, Czechia, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Ireland, Italy, Latvia, Lithuania, Luxembourg, the Netherlands, Norway, Poland, Portugal, the Slovak Republic, Slovenia, Spain, Sweden, Switzerland, the United Kingdom and the United States. Youth are those aged from 15 to 24, prime from 25 to 54, and senior from 55 and above. For urbanisation, the United States is not included. Source: Secretariat’s calculations based on data from European Working Conditions Survey (Eurofound) and American Working Conditions Survey (RAND Corporation).\n\nStatLink 2 https://stat.link/852oij\n\nOccupational safety and health (OSH) regulations in OECD countries already cover aspects related to temperature, heat stress and extreme weather at work. However, some countries, often with the direct involvement of social partners, have adapted the legislation to the specific challenges related to climate change (e.g. Lithuania) or developed specific OSH programmes and tools (e.g. Germany, Lithuania,\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\nWomen\n\nMen\n\nYouth\n\nPrime\n\nSenior\n\nLow-skilled\n\nMiddle-skilled\n\nHigh-skilled\n\nLow\n\nMiddle\n\nHigh\n\nRural\n\nUrban\n\nSuburban\n\nGender Age Occupation Education Urbanisation\n\n%\n\n**B. OECD27 average by workers' characteristic**\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30 %\n\n**A.Country average**","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**74** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Figure 2.1. One in seven workers report heat-related discomfort in Europe and the United States**\n\nPercentage of employment who declare to be exposed to high temperature at least half of their working time, 2015\n\nNote: Those who declare to be exposed to high temperature which make them perspire even when not working at least half of their working time are classified as at-risk. The shares are the predictions in May, obtained by running a regression of the exposure to heat at country level on the share of respondents by month of interview. For Panel A, the average is unweighted. For Panel B, averages are unweighted and include Austria, Belgium, Czechia, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Ireland, Italy, Latvia, Lithuania, Luxembourg, the Netherlands, Norway, Poland, Portugal, the Slovak Republic, Slovenia, Spain, Sweden, Switzerland, the United Kingdom and the United States. Youth are those aged from 15 to 24, prime from 25 to 54, and senior from 55 and above. For urbanisation, the United States is not included. Source: Secretariat’s calculations based on data from European Working Conditions Survey (Eurofound) and American Working Conditions Survey (RAND Corporation).\n\nStatLink 2 https://stat.link/852oij\n\nOccupational safety and health (OSH) regulations in OECD countries already cover aspects related to temperature, heat stress and extreme weather at work. However, some countries, often with the direct involvement of social partners, have adapted the legislation to the specific challenges related to climate change (e.g. Lithuania) or developed specific OSH programmes and tools (e.g. Germany, Lithuania,\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\nWomen\n\nMen\n\nYouth\n\nPrime\n\nSenior\n\nLow-skilled\n\nMiddle-skilled\n\nHigh-skilled\n\nLow\n\nMiddle\n\nHigh\n\nRural\n\nUrban\n\nSuburban\n\nGender Age Occupation Education Urbanisation\n\n%\n\n**B. OECD27 average by workers' characteristic**\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30 %\n\n**A.Country average**","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p76.pdf"]},"response":{"errors":[],"extract_id":"extract_bcd3695e20bac4401e491741f47e9f65","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**74** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Figure 2.1. One in seven workers report heat-related discomfort in Europe and the United States**\n\nPercentage of employment who declare to be exposed to high temperature at least half of their working time, 2015\n\nNote: Those who declare to be exposed to high temperature which make them perspire even when not working at least half of their working time are classified as at-risk. The shares are the predictions in May, obtained by running a regression of the exposure to heat at country level on the share of respondents by month of interview. For Panel A, the average is unweighted. For Panel B, averages are unweighted and include Austria, Belgium, Czechia, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Ireland, Italy, Latvia, Lithuania, Luxembourg, the Netherlands, Norway, Poland, Portugal, the Slovak Republic, Slovenia, Spain, Sweden, Switzerland, the United Kingdom and the United States. Youth are those aged from 15 to 24, prime from 25 to 54, and senior from 55 and above. For urbanisation, the United States is not included. Source: Secretariat’s calculations based on data from European Working Conditions Survey (Eurofound) and American Working Conditions Survey (RAND Corporation).\n\nStatLink 2 https://stat.link/852oij\n\nOccupational safety and health (OSH) regulations in OECD countries already cover aspects related to temperature, heat stress and extreme weather at work. However, some countries, often with the direct involvement of social partners, have adapted the legislation to the specific challenges related to climate change (e.g. Lithuania) or developed specific OSH programmes and tools (e.g. Germany, Lithuania,\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\nWomen\n\nMen\n\nYouth\n\nPrime\n\nSenior\n\nLow-skilled\n\nMiddle-skilled\n\nHigh-skilled\n\nLow\n\nMiddle\n\nHigh\n\nRural\n\nUrban\n\nSuburban\n\nGender Age Occupation Education Urbanisation\n\n%\n\n**B. OECD27 average by workers' characteristic**\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30 %\n\n**A.Country average**"],"full_content":"**74** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Figure 2.1. One in seven workers report heat-related discomfort in Europe and the United States**\n\nPercentage of employment who declare to be exposed to high temperature at least half of their working time, 2015\n\nNote: Those who declare to be exposed to high temperature which make them perspire even when not working at least half of their working time are classified as at-risk. The shares are the predictions in May, obtained by running a regression of the exposure to heat at country level on the share of respondents by month of interview. For Panel A, the average is unweighted. For Panel B, averages are unweighted and include Austria, Belgium, Czechia, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Ireland, Italy, Latvia, Lithuania, Luxembourg, the Netherlands, Norway, Poland, Portugal, the Slovak Republic, Slovenia, Spain, Sweden, Switzerland, the United Kingdom and the United States. Youth are those aged from 15 to 24, prime from 25 to 54, and senior from 55 and above. For urbanisation, the United States is not included. Source: Secretariat’s calculations based on data from European Working Conditions Survey (Eurofound) and American Working Conditions Survey (RAND Corporation).\n\nStatLink 2 https://stat.link/852oij\n\nOccupational safety and health (OSH) regulations in OECD countries already cover aspects related to temperature, heat stress and extreme weather at work. However, some countries, often with the direct involvement of social partners, have adapted the legislation to the specific challenges related to climate change (e.g. Lithuania) or developed specific OSH programmes and tools (e.g. Germany, Lithuania,\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\nWomen\n\nMen\n\nYouth\n\nPrime\n\nSenior\n\nLow-skilled\n\nMiddle-skilled\n\nHigh-skilled\n\nLow\n\nMiddle\n\nHigh\n\nRural\n\nUrban\n\nSuburban\n\nGender Age Occupation Education Urbanisation\n\n%\n\n**B. OECD27 average by workers' characteristic**\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30 %\n\n**A.Country average**","publish_date":null,"title":"Figure 2.1. One in seven workers report heat-related discomfort in Europe and the United States","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p76.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p76.pdf"},"request":{"example_id":"chart/ac8b3538-en_p76","product_type":"parse","source_file_path":"ac8b3538-en_p76.pdf"},"started_at":"2026-11-04T19:46:14.292773"} +{"completed_at":"2026-11-04T19:46:19.921106","latency_in_ms":3688,"output":{"example_id":"chart/ac8b3538-en_p86","grounded_pages":[],"layout_pages":[],"markdown":"**84** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\nincreased demand occupations and green-enhanced skills occupations also increased but at a significantly slower rate. However, if those occupations that are both green-driven and GHG-intensive are excluded, the growth of the remaining, green-driven occupations (labelled in the chart as “green residual”) is more than twice as large (5%). This is because occupations concentrated in high-emission sectors, including those occupations that are both green-driven and GHG-intensive, ²³ experienced a -18% decline over the same period. ²⁴\n\nFigure 2.4 suggests that the shares of green-driven and GHG-intensive occupations are positively correlated across countries, ²⁵ which is, to some extent, linked to countries’ industrial composition. Not the entire economy is equally affected by the net-zero transition. While green-driven occupations can be found in all sectors, Figure 2.5 shows that, compared to the distribution of employment across sectors, green- driven occupations are more likely to be found in manufacturing, utilities and mining, construction and transports. GHG-intensive occupations, partly as a result of the definition itself (see Box 2.4 above), are even more concentrated in these macro-sectors and agriculture (where high emission industries are). Other services, which represent more than two-thirds of total employment, are composed mostly by occupations that are neither green-driven nor GHG-intensive.\n\n### **Figure 2.4. Green new and emerging occupations are growing fast**\n\nPercentage change in the share of green-driven and GHG-intensive occupations in total employment\n\nNote: Europe: unweighted average of Austria, Czechia, Estonia, Finland, France, Hungary, Lithuania, Luxembourg, the Netherlands, Norway, Poland, the Slovak Republic, Slovenia, Sweden and Switzerland. Green residual occupations are green-driven occupations excluding those occupations that can be both green-driven and GHG-intensive. Source: Secretariat’s estimates based on version 24.1 of the O\\*NET database and the following country-specific sources: United States: Current Population Survey; All other countries: EU Labour Force Survey.\n\nStatLink 2 https://stat.link/h0w8yk\n\n\\-25\n\n\\-20\n\n\\-15\n\n\\-10\n\n\\-5\n\n0\n\n5\n\n10\n\n15\n\nGreen-driven occ. Green new and\n\nemerging occ.\n\nGreen residual\n\nocc.\n\nGHG-intensive\n\nocc.\n\n%\n\n**Europe**\n\n\\-10\n\n\\-5\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n35\n\nGreen-driven occ. Green new and\n\nemerging occ.\n\nGreen residual\n\nocc.\n\nGHG-intensive\n\nocc.\n\n%\n\n**United States**\n\n2011-15 2015-19 2019-22 2011-22","pages":[{"line_numbers":[],"markdown":"**84** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\nincreased demand occupations and green-enhanced skills occupations also increased but at a significantly slower rate. However, if those occupations that are both green-driven and GHG-intensive are excluded, the growth of the remaining, green-driven occupations (labelled in the chart as “green residual”) is more than twice as large (5%). This is because occupations concentrated in high-emission sectors, including those occupations that are both green-driven and GHG-intensive, ²³ experienced a -18% decline over the same period. ²⁴\n\nFigure 2.4 suggests that the shares of green-driven and GHG-intensive occupations are positively correlated across countries, ²⁵ which is, to some extent, linked to countries’ industrial composition. Not the entire economy is equally affected by the net-zero transition. While green-driven occupations can be found in all sectors, Figure 2.5 shows that, compared to the distribution of employment across sectors, green- driven occupations are more likely to be found in manufacturing, utilities and mining, construction and transports. GHG-intensive occupations, partly as a result of the definition itself (see Box 2.4 above), are even more concentrated in these macro-sectors and agriculture (where high emission industries are). Other services, which represent more than two-thirds of total employment, are composed mostly by occupations that are neither green-driven nor GHG-intensive.\n\n### **Figure 2.4. Green new and emerging occupations are growing fast**\n\nPercentage change in the share of green-driven and GHG-intensive occupations in total employment\n\nNote: Europe: unweighted average of Austria, Czechia, Estonia, Finland, France, Hungary, Lithuania, Luxembourg, the Netherlands, Norway, Poland, the Slovak Republic, Slovenia, Sweden and Switzerland. Green residual occupations are green-driven occupations excluding those occupations that can be both green-driven and GHG-intensive. Source: Secretariat’s estimates based on version 24.1 of the O\\*NET database and the following country-specific sources: United States: Current Population Survey; All other countries: EU Labour Force Survey.\n\nStatLink 2 https://stat.link/h0w8yk\n\n\\-25\n\n\\-20\n\n\\-15\n\n\\-10\n\n\\-5\n\n0\n\n5\n\n10\n\n15\n\nGreen-driven occ. Green new and\n\nemerging occ.\n\nGreen residual\n\nocc.\n\nGHG-intensive\n\nocc.\n\n%\n\n**Europe**\n\n\\-10\n\n\\-5\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n35\n\nGreen-driven occ. Green new and\n\nemerging occ.\n\nGreen residual\n\nocc.\n\nGHG-intensive\n\nocc.\n\n%\n\n**United States**\n\n2011-15 2015-19 2019-22 2011-22","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**84** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\nincreased demand occupations and green-enhanced skills occupations also increased but at a significantly slower rate. However, if those occupations that are both green-driven and GHG-intensive are excluded, the growth of the remaining, green-driven occupations (labelled in the chart as “green residual”) is more than twice as large (5%). This is because occupations concentrated in high-emission sectors, including those occupations that are both green-driven and GHG-intensive, ²³ experienced a -18% decline over the same period. ²⁴\n\nFigure 2.4 suggests that the shares of green-driven and GHG-intensive occupations are positively correlated across countries, ²⁵ which is, to some extent, linked to countries’ industrial composition. Not the entire economy is equally affected by the net-zero transition. While green-driven occupations can be found in all sectors, Figure 2.5 shows that, compared to the distribution of employment across sectors, green- driven occupations are more likely to be found in manufacturing, utilities and mining, construction and transports. GHG-intensive occupations, partly as a result of the definition itself (see Box 2.4 above), are even more concentrated in these macro-sectors and agriculture (where high emission industries are). Other services, which represent more than two-thirds of total employment, are composed mostly by occupations that are neither green-driven nor GHG-intensive.\n\n### **Figure 2.4. Green new and emerging occupations are growing fast**\n\nPercentage change in the share of green-driven and GHG-intensive occupations in total employment\n\nNote: Europe: unweighted average of Austria, Czechia, Estonia, Finland, France, Hungary, Lithuania, Luxembourg, the Netherlands, Norway, Poland, the Slovak Republic, Slovenia, Sweden and Switzerland. Green residual occupations are green-driven occupations excluding those occupations that can be both green-driven and GHG-intensive. Source: Secretariat’s estimates based on version 24.1 of the O\\*NET database and the following country-specific sources: United States: Current Population Survey; All other countries: EU Labour Force Survey.\n\nStatLink 2 https://stat.link/h0w8yk\n\n\\-25\n\n\\-20\n\n\\-15\n\n\\-10\n\n\\-5\n\n0\n\n5\n\n10\n\n15\n\nGreen-driven occ. Green new and\n\nemerging occ.\n\nGreen residual\n\nocc.\n\nGHG-intensive\n\nocc.\n\n%\n\n**Europe**\n\n\\-10\n\n\\-5\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n35\n\nGreen-driven occ. Green new and\n\nemerging occ.\n\nGreen residual\n\nocc.\n\nGHG-intensive\n\nocc.\n\n%\n\n**United States**\n\n2011-15 2015-19 2019-22 2011-22","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p86.pdf"]},"response":{"errors":[],"extract_id":"extract_d98e4d244e7e5ca78b1398823d2b58da","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**84** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\nincreased demand occupations and green-enhanced skills occupations also increased but at a significantly slower rate. However, if those occupations that are both green-driven and GHG-intensive are excluded, the growth of the remaining, green-driven occupations (labelled in the chart as “green residual”) is more than twice as large (5%). This is because occupations concentrated in high-emission sectors, including those occupations that are both green-driven and GHG-intensive, ²³ experienced a -18% decline over the same period. ²⁴\n\nFigure 2.4 suggests that the shares of green-driven and GHG-intensive occupations are positively correlated across countries, ²⁵ which is, to some extent, linked to countries’ industrial composition. Not the entire economy is equally affected by the net-zero transition. While green-driven occupations can be found in all sectors, Figure 2.5 shows that, compared to the distribution of employment across sectors, green- driven occupations are more likely to be found in manufacturing, utilities and mining, construction and transports. GHG-intensive occupations, partly as a result of the definition itself (see Box 2.4 above), are even more concentrated in these macro-sectors and agriculture (where high emission industries are). Other services, which represent more than two-thirds of total employment, are composed mostly by occupations that are neither green-driven nor GHG-intensive.\n\n### **Figure 2.4. Green new and emerging occupations are growing fast**\n\nPercentage change in the share of green-driven and GHG-intensive occupations in total employment\n\nNote: Europe: unweighted average of Austria, Czechia, Estonia, Finland, France, Hungary, Lithuania, Luxembourg, the Netherlands, Norway, Poland, the Slovak Republic, Slovenia, Sweden and Switzerland. Green residual occupations are green-driven occupations excluding those occupations that can be both green-driven and GHG-intensive. Source: Secretariat’s estimates based on version 24.1 of the O\\*NET database and the following country-specific sources: United States: Current Population Survey; All other countries: EU Labour Force Survey.\n\nStatLink 2 https://stat.link/h0w8yk\n\n\\-25\n\n\\-20\n\n\\-15\n\n\\-10\n\n\\-5\n\n0\n\n5\n\n10\n\n15\n\nGreen-driven occ. Green new and\n\nemerging occ.\n\nGreen residual\n\nocc.\n\nGHG-intensive\n\nocc.\n\n%\n\n**Europe**\n\n\\-10\n\n\\-5\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n35\n\nGreen-driven occ. Green new and\n\nemerging occ.\n\nGreen residual\n\nocc.\n\nGHG-intensive\n\nocc.\n\n%\n\n**United States**\n\n2011-15 2015-19 2019-22 2011-22"],"full_content":"**84** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\nincreased demand occupations and green-enhanced skills occupations also increased but at a significantly slower rate. However, if those occupations that are both green-driven and GHG-intensive are excluded, the growth of the remaining, green-driven occupations (labelled in the chart as “green residual”) is more than twice as large (5%). This is because occupations concentrated in high-emission sectors, including those occupations that are both green-driven and GHG-intensive, ²³ experienced a -18% decline over the same period. ²⁴\n\nFigure 2.4 suggests that the shares of green-driven and GHG-intensive occupations are positively correlated across countries, ²⁵ which is, to some extent, linked to countries’ industrial composition. Not the entire economy is equally affected by the net-zero transition. While green-driven occupations can be found in all sectors, Figure 2.5 shows that, compared to the distribution of employment across sectors, green- driven occupations are more likely to be found in manufacturing, utilities and mining, construction and transports. GHG-intensive occupations, partly as a result of the definition itself (see Box 2.4 above), are even more concentrated in these macro-sectors and agriculture (where high emission industries are). Other services, which represent more than two-thirds of total employment, are composed mostly by occupations that are neither green-driven nor GHG-intensive.\n\n### **Figure 2.4. Green new and emerging occupations are growing fast**\n\nPercentage change in the share of green-driven and GHG-intensive occupations in total employment\n\nNote: Europe: unweighted average of Austria, Czechia, Estonia, Finland, France, Hungary, Lithuania, Luxembourg, the Netherlands, Norway, Poland, the Slovak Republic, Slovenia, Sweden and Switzerland. Green residual occupations are green-driven occupations excluding those occupations that can be both green-driven and GHG-intensive. Source: Secretariat’s estimates based on version 24.1 of the O\\*NET database and the following country-specific sources: United States: Current Population Survey; All other countries: EU Labour Force Survey.\n\nStatLink 2 https://stat.link/h0w8yk\n\n\\-25\n\n\\-20\n\n\\-15\n\n\\-10\n\n\\-5\n\n0\n\n5\n\n10\n\n15\n\nGreen-driven occ. Green new and\n\nemerging occ.\n\nGreen residual\n\nocc.\n\nGHG-intensive\n\nocc.\n\n%\n\n**Europe**\n\n\\-10\n\n\\-5\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n35\n\nGreen-driven occ. Green new and\n\nemerging occ.\n\nGreen residual\n\nocc.\n\nGHG-intensive\n\nocc.\n\n%\n\n**United States**\n\n2011-15 2015-19 2019-22 2011-22","publish_date":null,"title":"Figure 2.4. Green new and emerging occupations are growing fast","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p86.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p86.pdf"},"request":{"example_id":"chart/ac8b3538-en_p86","product_type":"parse","source_file_path":"ac8b3538-en_p86.pdf"},"started_at":"2026-11-04T19:46:16.232625"} +{"completed_at":"2026-11-04T19:46:20.868606","latency_in_ms":3069,"output":{"example_id":"chart/ac8b3538-en_p88","grounded_pages":[],"layout_pages":[],"markdown":"**86** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\nwhich employ, on average, more high skilled and high educated workers, pay better wages and have higher job quality (see Sections 2.2.2 and 2.3) are, in fact, concentrated in urban areas. Without an effective policy strategy to mitigate the disadvantage of those regions most specialised in GHG-intensive activities and least specialised in green-driven occupations (see Box 2.6), there is a significant risk that the net-zero transition exacerbates regional disparities and endangers social cohesion.\n\n### **Figure 2.6. Both green-driven and GHG-intensive occupations are more widespread in rural than in urban areas**\n\nShare of green-driven and GHG-intensive occupations by rural/urban area, average 2015-19\n\nNote: Countries are ranked by decreasing gap of the share in rural areas compared to total. For European countries, the degree of urbanisation is defined on the share of local population living in urban clusters and in urban centres. It classifies areas into three types: thinly populated area (rural area); intermediate density area (towns and suburbs/small urban area), and densely populated area (cities/large urban area). In the United States, rural and urban areas are defined as non-metropolitan and metropolitan areas, respectively. Source: Secretariat’s estimates based on version 24.1 of the O\\*NET database and the following country-specific sources: United States: Current Population Survey; All other countries: EU Labour Force Survey.\n\nStatLink 2 https://stat.link/1g08k9\n\n14\n\n16\n\n18\n\n20\n\n22\n\n24\n\n26\n\n28\n\n30\n\n% **A. Green-driven occupations**\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\n16\n\n18\n\n20\n\n% **B. GHG-intensive occupations**\n\nRural Urban Suburban Total","pages":[{"line_numbers":[],"markdown":"**86** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\nwhich employ, on average, more high skilled and high educated workers, pay better wages and have higher job quality (see Sections 2.2.2 and 2.3) are, in fact, concentrated in urban areas. Without an effective policy strategy to mitigate the disadvantage of those regions most specialised in GHG-intensive activities and least specialised in green-driven occupations (see Box 2.6), there is a significant risk that the net-zero transition exacerbates regional disparities and endangers social cohesion.\n\n### **Figure 2.6. Both green-driven and GHG-intensive occupations are more widespread in rural than in urban areas**\n\nShare of green-driven and GHG-intensive occupations by rural/urban area, average 2015-19\n\nNote: Countries are ranked by decreasing gap of the share in rural areas compared to total. For European countries, the degree of urbanisation is defined on the share of local population living in urban clusters and in urban centres. It classifies areas into three types: thinly populated area (rural area); intermediate density area (towns and suburbs/small urban area), and densely populated area (cities/large urban area). In the United States, rural and urban areas are defined as non-metropolitan and metropolitan areas, respectively. Source: Secretariat’s estimates based on version 24.1 of the O\\*NET database and the following country-specific sources: United States: Current Population Survey; All other countries: EU Labour Force Survey.\n\nStatLink 2 https://stat.link/1g08k9\n\n14\n\n16\n\n18\n\n20\n\n22\n\n24\n\n26\n\n28\n\n30\n\n% **A. Green-driven occupations**\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\n16\n\n18\n\n20\n\n% **B. GHG-intensive occupations**\n\nRural Urban Suburban Total","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**86** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\nwhich employ, on average, more high skilled and high educated workers, pay better wages and have higher job quality (see Sections 2.2.2 and 2.3) are, in fact, concentrated in urban areas. Without an effective policy strategy to mitigate the disadvantage of those regions most specialised in GHG-intensive activities and least specialised in green-driven occupations (see Box 2.6), there is a significant risk that the net-zero transition exacerbates regional disparities and endangers social cohesion.\n\n### **Figure 2.6. Both green-driven and GHG-intensive occupations are more widespread in rural than in urban areas**\n\nShare of green-driven and GHG-intensive occupations by rural/urban area, average 2015-19\n\nNote: Countries are ranked by decreasing gap of the share in rural areas compared to total. For European countries, the degree of urbanisation is defined on the share of local population living in urban clusters and in urban centres. It classifies areas into three types: thinly populated area (rural area); intermediate density area (towns and suburbs/small urban area), and densely populated area (cities/large urban area). In the United States, rural and urban areas are defined as non-metropolitan and metropolitan areas, respectively. Source: Secretariat’s estimates based on version 24.1 of the O\\*NET database and the following country-specific sources: United States: Current Population Survey; All other countries: EU Labour Force Survey.\n\nStatLink 2 https://stat.link/1g08k9\n\n14\n\n16\n\n18\n\n20\n\n22\n\n24\n\n26\n\n28\n\n30\n\n% **A. Green-driven occupations**\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\n16\n\n18\n\n20\n\n% **B. GHG-intensive occupations**\n\nRural Urban Suburban Total","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p88.pdf"]},"response":{"errors":[],"extract_id":"extract_9de3e61f70e4e7f79b245465bebcad1f","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**86** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\nwhich employ, on average, more high skilled and high educated workers, pay better wages and have higher job quality (see Sections 2.2.2 and 2.3) are, in fact, concentrated in urban areas. Without an effective policy strategy to mitigate the disadvantage of those regions most specialised in GHG-intensive activities and least specialised in green-driven occupations (see Box 2.6), there is a significant risk that the net-zero transition exacerbates regional disparities and endangers social cohesion.\n\n### **Figure 2.6. Both green-driven and GHG-intensive occupations are more widespread in rural than in urban areas**\n\nShare of green-driven and GHG-intensive occupations by rural/urban area, average 2015-19\n\nNote: Countries are ranked by decreasing gap of the share in rural areas compared to total. For European countries, the degree of urbanisation is defined on the share of local population living in urban clusters and in urban centres. It classifies areas into three types: thinly populated area (rural area); intermediate density area (towns and suburbs/small urban area), and densely populated area (cities/large urban area). In the United States, rural and urban areas are defined as non-metropolitan and metropolitan areas, respectively. Source: Secretariat’s estimates based on version 24.1 of the O\\*NET database and the following country-specific sources: United States: Current Population Survey; All other countries: EU Labour Force Survey.\n\nStatLink 2 https://stat.link/1g08k9\n\n14\n\n16\n\n18\n\n20\n\n22\n\n24\n\n26\n\n28\n\n30\n\n% **A. Green-driven occupations**\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\n16\n\n18\n\n20\n\n% **B. GHG-intensive occupations**\n\nRural Urban Suburban Total"],"full_content":"**86** \n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\nwhich employ, on average, more high skilled and high educated workers, pay better wages and have higher job quality (see Sections 2.2.2 and 2.3) are, in fact, concentrated in urban areas. Without an effective policy strategy to mitigate the disadvantage of those regions most specialised in GHG-intensive activities and least specialised in green-driven occupations (see Box 2.6), there is a significant risk that the net-zero transition exacerbates regional disparities and endangers social cohesion.\n\n### **Figure 2.6. Both green-driven and GHG-intensive occupations are more widespread in rural than in urban areas**\n\nShare of green-driven and GHG-intensive occupations by rural/urban area, average 2015-19\n\nNote: Countries are ranked by decreasing gap of the share in rural areas compared to total. For European countries, the degree of urbanisation is defined on the share of local population living in urban clusters and in urban centres. It classifies areas into three types: thinly populated area (rural area); intermediate density area (towns and suburbs/small urban area), and densely populated area (cities/large urban area). In the United States, rural and urban areas are defined as non-metropolitan and metropolitan areas, respectively. Source: Secretariat’s estimates based on version 24.1 of the O\\*NET database and the following country-specific sources: United States: Current Population Survey; All other countries: EU Labour Force Survey.\n\nStatLink 2 https://stat.link/1g08k9\n\n14\n\n16\n\n18\n\n20\n\n22\n\n24\n\n26\n\n28\n\n30\n\n% **A. Green-driven occupations**\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\n16\n\n18\n\n20\n\n% **B. GHG-intensive occupations**\n\nRural Urban Suburban Total","publish_date":null,"title":"Figure 2.6. Both green-driven and GHG-intensive occupations are more widespread in rural than in urban areas","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p88.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p88.pdf"},"request":{"example_id":"chart/ac8b3538-en_p88","product_type":"parse","source_file_path":"ac8b3538-en_p88.pdf"},"started_at":"2026-11-04T19:46:17.799048"} +{"completed_at":"2026-11-04T19:46:21.765282","latency_in_ms":1820,"output":{"example_id":"chart/ac8b3538-en_p93","grounded_pages":[],"layout_pages":[],"markdown":" **91**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Figure 2.8. Workers in green-driven and GHG-intensive occupations tend to have similar characteristics**\n\nGap in percentage points between the share of green-driven (respectively GHG-intensive occupations) in each characteristic and the share in the whole economy\n\nNote: The figure reports gaps in percentage points between the share of green-driven (Panel A) and GHG-intensive (Panel B) occupations by characteristics and the average share. Data refer to the average for 2015-19, except for Canada: 2017-19. In terms of age, youth are 15-34 years old, middle 35-54 years old, older 55+ year old. In terms of occupation, low-skill refers to Service and Sales Workers and Elementary Occupations (ISCO-08 5 and 9, medium-skill to Clerical Support Workers, Craft and Related Trades Workers and Plant and Machine Operators, and Assemblers (ISCO-08 4, 7 and 8), high-skill to Managers, Professionals and Technicians and Associate Professionals (ISCO-08 1, 2 and 3). Low education refers to less than upper secondary, medium education to upper secondary, high education to more than upper secondary. Unweighted average of Australia, Austria, Belgium, Canada, Czechia, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Iceland, Ireland, Italy, Japan, Latvia, Lithuania, Luxembourg, the Netherlands, Norway, Poland, Portugal, Slovak Republic, Slovenia, Spain, Sweden, Switzerland, the United Kingdom and the United States. Reading: The share of women employed in green-driven occupations is 9 percentage points lower than the share of women employed on average. Source: Secretariat’s estimates based on version 24.1 of the O\\*NET database and the following country-specific sources: Australian Labour Force Survey (gender, age); Canadian Labour Force Survey (gender, age, education); Japanese Labour Force Survey (gender, age); United States: Current Population Survey; All other countries: EU Labour Force Survey.\n\nStatLink 2 https://stat.link/y5167t\n\nIn conclusion, the implementation of ambitious measures to combat climate change will result in decreased employment levels within GHG-intensive industries to the benefit of renewable energy generation as well as a larger number of industries that will be instrumental to make the transition happen, such as certain segments in construction and transport. This reallocation may also give rise to disparities across workers and regions. The analysis shown in this section suggests that, even if the reallocation process will go well beyond the two relatively small groups of GHG-intensive and green-driven occupations, these two groups have some similarities in common in terms of the characteristics of the workers who hold them that could help to make the transition manageable. And yet, the risk of widening inequalities across demographic groups and geographical areas is real – see Botta (2019[77]); OECD (2021[2]); and Chapters 3 and 4. In particular, the analysis has shown that green-driven occupations are a heterogenous group of jobs: fast-growing green new and emerging occupations are typically high-skill and employ highly educated workers in urban areas while the much larger but less dynamic groups of green-enhanced skills occupations and green increased demand ones are on average relatively more medium- and low-skill and employ many more low-educated workers.\n\nInterestingly, the characteristics of the workers employed in declining GHG-intensive occupations are remarkably similar to those of the workers more exposed to high temperatures, as shown in Box 2.2. This suggests that while climate change mitigation policies may have costs for some groups of workers, there are also labour market costs of inaction for these same groups.\n\n\\-10\n\n\\-5\n\n0\n\n5\n\n10\n\n15\n\nWomen\n\nMen\n\nYouth\n\nMiddle\n\nOlder\n\nLow-skill\n\nMedium-skill\n\nHigh-skill\n\nLow\n\nMedium\n\nHigh\n\nGender Age Occupation Education\n\np.p. **A. Green-driven occupations**\n\n\\-10\n\n\\-5\n\n0\n\n5\n\n10\n\n15\n\nWomen\n\nMen\n\nYouth\n\nMiddle\n\nOlder\n\nLow-skill\n\nMedium-skill\n\nHigh-skill\n\nLow\n\nMedium\n\nHigh\n\nGender Age Occupation Education\n\np.p. **B. GHG-intensive occupations**","pages":[{"line_numbers":[],"markdown":" **91**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Figure 2.8. Workers in green-driven and GHG-intensive occupations tend to have similar characteristics**\n\nGap in percentage points between the share of green-driven (respectively GHG-intensive occupations) in each characteristic and the share in the whole economy\n\nNote: The figure reports gaps in percentage points between the share of green-driven (Panel A) and GHG-intensive (Panel B) occupations by characteristics and the average share. Data refer to the average for 2015-19, except for Canada: 2017-19. In terms of age, youth are 15-34 years old, middle 35-54 years old, older 55+ year old. In terms of occupation, low-skill refers to Service and Sales Workers and Elementary Occupations (ISCO-08 5 and 9, medium-skill to Clerical Support Workers, Craft and Related Trades Workers and Plant and Machine Operators, and Assemblers (ISCO-08 4, 7 and 8), high-skill to Managers, Professionals and Technicians and Associate Professionals (ISCO-08 1, 2 and 3). Low education refers to less than upper secondary, medium education to upper secondary, high education to more than upper secondary. Unweighted average of Australia, Austria, Belgium, Canada, Czechia, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Iceland, Ireland, Italy, Japan, Latvia, Lithuania, Luxembourg, the Netherlands, Norway, Poland, Portugal, Slovak Republic, Slovenia, Spain, Sweden, Switzerland, the United Kingdom and the United States. Reading: The share of women employed in green-driven occupations is 9 percentage points lower than the share of women employed on average. Source: Secretariat’s estimates based on version 24.1 of the O\\*NET database and the following country-specific sources: Australian Labour Force Survey (gender, age); Canadian Labour Force Survey (gender, age, education); Japanese Labour Force Survey (gender, age); United States: Current Population Survey; All other countries: EU Labour Force Survey.\n\nStatLink 2 https://stat.link/y5167t\n\nIn conclusion, the implementation of ambitious measures to combat climate change will result in decreased employment levels within GHG-intensive industries to the benefit of renewable energy generation as well as a larger number of industries that will be instrumental to make the transition happen, such as certain segments in construction and transport. This reallocation may also give rise to disparities across workers and regions. The analysis shown in this section suggests that, even if the reallocation process will go well beyond the two relatively small groups of GHG-intensive and green-driven occupations, these two groups have some similarities in common in terms of the characteristics of the workers who hold them that could help to make the transition manageable. And yet, the risk of widening inequalities across demographic groups and geographical areas is real – see Botta (2019[77]); OECD (2021[2]); and Chapters 3 and 4. In particular, the analysis has shown that green-driven occupations are a heterogenous group of jobs: fast-growing green new and emerging occupations are typically high-skill and employ highly educated workers in urban areas while the much larger but less dynamic groups of green-enhanced skills occupations and green increased demand ones are on average relatively more medium- and low-skill and employ many more low-educated workers.\n\nInterestingly, the characteristics of the workers employed in declining GHG-intensive occupations are remarkably similar to those of the workers more exposed to high temperatures, as shown in Box 2.2. This suggests that while climate change mitigation policies may have costs for some groups of workers, there are also labour market costs of inaction for these same groups.\n\n\\-10\n\n\\-5\n\n0\n\n5\n\n10\n\n15\n\nWomen\n\nMen\n\nYouth\n\nMiddle\n\nOlder\n\nLow-skill\n\nMedium-skill\n\nHigh-skill\n\nLow\n\nMedium\n\nHigh\n\nGender Age Occupation Education\n\np.p. **A. Green-driven occupations**\n\n\\-10\n\n\\-5\n\n0\n\n5\n\n10\n\n15\n\nWomen\n\nMen\n\nYouth\n\nMiddle\n\nOlder\n\nLow-skill\n\nMedium-skill\n\nHigh-skill\n\nLow\n\nMedium\n\nHigh\n\nGender Age Occupation Education\n\np.p. **B. GHG-intensive occupations**","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":" **91**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Figure 2.8. Workers in green-driven and GHG-intensive occupations tend to have similar characteristics**\n\nGap in percentage points between the share of green-driven (respectively GHG-intensive occupations) in each characteristic and the share in the whole economy\n\nNote: The figure reports gaps in percentage points between the share of green-driven (Panel A) and GHG-intensive (Panel B) occupations by characteristics and the average share. Data refer to the average for 2015-19, except for Canada: 2017-19. In terms of age, youth are 15-34 years old, middle 35-54 years old, older 55+ year old. In terms of occupation, low-skill refers to Service and Sales Workers and Elementary Occupations (ISCO-08 5 and 9, medium-skill to Clerical Support Workers, Craft and Related Trades Workers and Plant and Machine Operators, and Assemblers (ISCO-08 4, 7 and 8), high-skill to Managers, Professionals and Technicians and Associate Professionals (ISCO-08 1, 2 and 3). Low education refers to less than upper secondary, medium education to upper secondary, high education to more than upper secondary. Unweighted average of Australia, Austria, Belgium, Canada, Czechia, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Iceland, Ireland, Italy, Japan, Latvia, Lithuania, Luxembourg, the Netherlands, Norway, Poland, Portugal, Slovak Republic, Slovenia, Spain, Sweden, Switzerland, the United Kingdom and the United States. Reading: The share of women employed in green-driven occupations is 9 percentage points lower than the share of women employed on average. Source: Secretariat’s estimates based on version 24.1 of the O\\*NET database and the following country-specific sources: Australian Labour Force Survey (gender, age); Canadian Labour Force Survey (gender, age, education); Japanese Labour Force Survey (gender, age); United States: Current Population Survey; All other countries: EU Labour Force Survey.\n\nStatLink 2 https://stat.link/y5167t\n\nIn conclusion, the implementation of ambitious measures to combat climate change will result in decreased employment levels within GHG-intensive industries to the benefit of renewable energy generation as well as a larger number of industries that will be instrumental to make the transition happen, such as certain segments in construction and transport. This reallocation may also give rise to disparities across workers and regions. The analysis shown in this section suggests that, even if the reallocation process will go well beyond the two relatively small groups of GHG-intensive and green-driven occupations, these two groups have some similarities in common in terms of the characteristics of the workers who hold them that could help to make the transition manageable. And yet, the risk of widening inequalities across demographic groups and geographical areas is real – see Botta (2019[77]); OECD (2021[2]); and Chapters 3 and 4. In particular, the analysis has shown that green-driven occupations are a heterogenous group of jobs: fast-growing green new and emerging occupations are typically high-skill and employ highly educated workers in urban areas while the much larger but less dynamic groups of green-enhanced skills occupations and green increased demand ones are on average relatively more medium- and low-skill and employ many more low-educated workers.\n\nInterestingly, the characteristics of the workers employed in declining GHG-intensive occupations are remarkably similar to those of the workers more exposed to high temperatures, as shown in Box 2.2. This suggests that while climate change mitigation policies may have costs for some groups of workers, there are also labour market costs of inaction for these same groups.\n\n\\-10\n\n\\-5\n\n0\n\n5\n\n10\n\n15\n\nWomen\n\nMen\n\nYouth\n\nMiddle\n\nOlder\n\nLow-skill\n\nMedium-skill\n\nHigh-skill\n\nLow\n\nMedium\n\nHigh\n\nGender Age Occupation Education\n\np.p. **A. Green-driven occupations**\n\n\\-10\n\n\\-5\n\n0\n\n5\n\n10\n\n15\n\nWomen\n\nMen\n\nYouth\n\nMiddle\n\nOlder\n\nLow-skill\n\nMedium-skill\n\nHigh-skill\n\nLow\n\nMedium\n\nHigh\n\nGender Age Occupation Education\n\np.p. **B. GHG-intensive occupations**","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p93.pdf"]},"response":{"errors":[],"extract_id":"extract_d74c4eb6f1c1f2ec3178e5b6ca970579","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":[" **91**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Figure 2.8. Workers in green-driven and GHG-intensive occupations tend to have similar characteristics**\n\nGap in percentage points between the share of green-driven (respectively GHG-intensive occupations) in each characteristic and the share in the whole economy\n\nNote: The figure reports gaps in percentage points between the share of green-driven (Panel A) and GHG-intensive (Panel B) occupations by characteristics and the average share. Data refer to the average for 2015-19, except for Canada: 2017-19. In terms of age, youth are 15-34 years old, middle 35-54 years old, older 55+ year old. In terms of occupation, low-skill refers to Service and Sales Workers and Elementary Occupations (ISCO-08 5 and 9, medium-skill to Clerical Support Workers, Craft and Related Trades Workers and Plant and Machine Operators, and Assemblers (ISCO-08 4, 7 and 8), high-skill to Managers, Professionals and Technicians and Associate Professionals (ISCO-08 1, 2 and 3). Low education refers to less than upper secondary, medium education to upper secondary, high education to more than upper secondary. Unweighted average of Australia, Austria, Belgium, Canada, Czechia, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Iceland, Ireland, Italy, Japan, Latvia, Lithuania, Luxembourg, the Netherlands, Norway, Poland, Portugal, Slovak Republic, Slovenia, Spain, Sweden, Switzerland, the United Kingdom and the United States. Reading: The share of women employed in green-driven occupations is 9 percentage points lower than the share of women employed on average. Source: Secretariat’s estimates based on version 24.1 of the O\\*NET database and the following country-specific sources: Australian Labour Force Survey (gender, age); Canadian Labour Force Survey (gender, age, education); Japanese Labour Force Survey (gender, age); United States: Current Population Survey; All other countries: EU Labour Force Survey.\n\nStatLink 2 https://stat.link/y5167t\n\nIn conclusion, the implementation of ambitious measures to combat climate change will result in decreased employment levels within GHG-intensive industries to the benefit of renewable energy generation as well as a larger number of industries that will be instrumental to make the transition happen, such as certain segments in construction and transport. This reallocation may also give rise to disparities across workers and regions. The analysis shown in this section suggests that, even if the reallocation process will go well beyond the two relatively small groups of GHG-intensive and green-driven occupations, these two groups have some similarities in common in terms of the characteristics of the workers who hold them that could help to make the transition manageable. And yet, the risk of widening inequalities across demographic groups and geographical areas is real – see Botta (2019[77]); OECD (2021[2]); and Chapters 3 and 4. In particular, the analysis has shown that green-driven occupations are a heterogenous group of jobs: fast-growing green new and emerging occupations are typically high-skill and employ highly educated workers in urban areas while the much larger but less dynamic groups of green-enhanced skills occupations and green increased demand ones are on average relatively more medium- and low-skill and employ many more low-educated workers.\n\nInterestingly, the characteristics of the workers employed in declining GHG-intensive occupations are remarkably similar to those of the workers more exposed to high temperatures, as shown in Box 2.2. This suggests that while climate change mitigation policies may have costs for some groups of workers, there are also labour market costs of inaction for these same groups.\n\n\\-10\n\n\\-5\n\n0\n\n5\n\n10\n\n15\n\nWomen\n\nMen\n\nYouth\n\nMiddle\n\nOlder\n\nLow-skill\n\nMedium-skill\n\nHigh-skill\n\nLow\n\nMedium\n\nHigh\n\nGender Age Occupation Education\n\np.p. **A. Green-driven occupations**\n\n\\-10\n\n\\-5\n\n0\n\n5\n\n10\n\n15\n\nWomen\n\nMen\n\nYouth\n\nMiddle\n\nOlder\n\nLow-skill\n\nMedium-skill\n\nHigh-skill\n\nLow\n\nMedium\n\nHigh\n\nGender Age Occupation Education\n\np.p. **B. GHG-intensive occupations**"],"full_content":" **91**\n\nOECD EMPLOYMENT OUTLOOK 2024 © OECD 2024\n\n### **Figure 2.8. Workers in green-driven and GHG-intensive occupations tend to have similar characteristics**\n\nGap in percentage points between the share of green-driven (respectively GHG-intensive occupations) in each characteristic and the share in the whole economy\n\nNote: The figure reports gaps in percentage points between the share of green-driven (Panel A) and GHG-intensive (Panel B) occupations by characteristics and the average share. Data refer to the average for 2015-19, except for Canada: 2017-19. In terms of age, youth are 15-34 years old, middle 35-54 years old, older 55+ year old. In terms of occupation, low-skill refers to Service and Sales Workers and Elementary Occupations (ISCO-08 5 and 9, medium-skill to Clerical Support Workers, Craft and Related Trades Workers and Plant and Machine Operators, and Assemblers (ISCO-08 4, 7 and 8), high-skill to Managers, Professionals and Technicians and Associate Professionals (ISCO-08 1, 2 and 3). Low education refers to less than upper secondary, medium education to upper secondary, high education to more than upper secondary. Unweighted average of Australia, Austria, Belgium, Canada, Czechia, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Iceland, Ireland, Italy, Japan, Latvia, Lithuania, Luxembourg, the Netherlands, Norway, Poland, Portugal, Slovak Republic, Slovenia, Spain, Sweden, Switzerland, the United Kingdom and the United States. Reading: The share of women employed in green-driven occupations is 9 percentage points lower than the share of women employed on average. Source: Secretariat’s estimates based on version 24.1 of the O\\*NET database and the following country-specific sources: Australian Labour Force Survey (gender, age); Canadian Labour Force Survey (gender, age, education); Japanese Labour Force Survey (gender, age); United States: Current Population Survey; All other countries: EU Labour Force Survey.\n\nStatLink 2 https://stat.link/y5167t\n\nIn conclusion, the implementation of ambitious measures to combat climate change will result in decreased employment levels within GHG-intensive industries to the benefit of renewable energy generation as well as a larger number of industries that will be instrumental to make the transition happen, such as certain segments in construction and transport. This reallocation may also give rise to disparities across workers and regions. The analysis shown in this section suggests that, even if the reallocation process will go well beyond the two relatively small groups of GHG-intensive and green-driven occupations, these two groups have some similarities in common in terms of the characteristics of the workers who hold them that could help to make the transition manageable. And yet, the risk of widening inequalities across demographic groups and geographical areas is real – see Botta (2019[77]); OECD (2021[2]); and Chapters 3 and 4. In particular, the analysis has shown that green-driven occupations are a heterogenous group of jobs: fast-growing green new and emerging occupations are typically high-skill and employ highly educated workers in urban areas while the much larger but less dynamic groups of green-enhanced skills occupations and green increased demand ones are on average relatively more medium- and low-skill and employ many more low-educated workers.\n\nInterestingly, the characteristics of the workers employed in declining GHG-intensive occupations are remarkably similar to those of the workers more exposed to high temperatures, as shown in Box 2.2. This suggests that while climate change mitigation policies may have costs for some groups of workers, there are also labour market costs of inaction for these same groups.\n\n\\-10\n\n\\-5\n\n0\n\n5\n\n10\n\n15\n\nWomen\n\nMen\n\nYouth\n\nMiddle\n\nOlder\n\nLow-skill\n\nMedium-skill\n\nHigh-skill\n\nLow\n\nMedium\n\nHigh\n\nGender Age Occupation Education\n\np.p. **A. Green-driven occupations**\n\n\\-10\n\n\\-5\n\n0\n\n5\n\n10\n\n15\n\nWomen\n\nMen\n\nYouth\n\nMiddle\n\nOlder\n\nLow-skill\n\nMedium-skill\n\nHigh-skill\n\nLow\n\nMedium\n\nHigh\n\nGender Age Occupation Education\n\np.p. **B. GHG-intensive occupations**","publish_date":null,"title":"Figure 2.8. Workers in green-driven and GHG-intensive occupations tend to have similar characteristics","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p93.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/ac8b3538-en_p93.pdf"},"request":{"example_id":"chart/ac8b3538-en_p93","product_type":"parse","source_file_path":"ac8b3538-en_p93.pdf"},"started_at":"2026-11-04T19:46:19.944387"} +{"completed_at":"2026-11-04T19:46:24.599728","latency_in_ms":2527,"output":{"example_id":"chart/andritz-financial-report-2024-e-data_p15","grounded_pages":[],"layout_pages":[],"markdown":"A N D R I T Z f i n a n c i a l r e p o r t 2 0 2 4\n\nM a n a g e m e n t r e p o r t\n\n# **14**\n\n**c)** **Financial result and taxes**\n\nThe financial result amounted to -15.4 MEUR in the reporting period (2023: 3.0 MEUR). The net interest result decreased compared to the previous year’s reference figure due to the decline in the interest rate environment and lower gross liquidity (1,434.4 MEUR versus 2023: 1,787.2 MEUR). The lower gross liquidity is attributable to an increase in net working capital as part of the scheduled processing of large orders and to the repayment of loans during the financial year. The significant decrease in the other financial result compared to the previous year is mainly attributable to a one- time effect in the context of the deconsolidation of a company. The other financial result of 2024 includes a one- time effect from the write-off of a loan granted to this company and the result from companies accounted for using the equity method includes a one-time effect from the measurement of the remaining shares at fair value. The tax rate decreased from 26.7% in the previous year to 23.2%. Above all, the capitalization of loss carryforwards through positive effects in the context of mergers, active management of tax groups and the consistent use of research and investment benefits have contributed to the reduction of the tax rate.\n\n**d)** **Earnings and dividend per share** The earnings per share decreased to 5.02 EUR (2023: 5.15 EUR). At the Annual General Meeting on March 27, 2025, the Executive Board will propose a dividend of 2.60 EUR per share (2023: 2.50 EUR) for the 2024 financial year. This is equal to a payout ratio of 51.8% (2023: 48.5%).\n\nDividend for 2024: Proposal to the Annual General Meeting.\n\n**2,08**\n\n**3,28**\n\n**4,14**\n\n**5,15 5,02**\n\n1,00\n\n1,65\n\n2,10\n\n2,50 2,60\n\n48,1%\n\n50,3% 50,7%\n\n48,5% 51,8%\n\n0,00\n\n1,00\n\n2,00\n\n3,00\n\n4,00\n\n5,00\n\n6,00\n\n2020 2021 2022 2023 2024\n\nEarnings per share (EUR) Dividend per share (EUR) Payout ratio\n\n— **More informationin the notes to the consolidated financial statements chapter C) 16.** Income taxes.","pages":[{"line_numbers":[],"markdown":"A N D R I T Z f i n a n c i a l r e p o r t 2 0 2 4\n\nM a n a g e m e n t r e p o r t\n\n# **14**\n\n**c)** **Financial result and taxes**\n\nThe financial result amounted to -15.4 MEUR in the reporting period (2023: 3.0 MEUR). The net interest result decreased compared to the previous year’s reference figure due to the decline in the interest rate environment and lower gross liquidity (1,434.4 MEUR versus 2023: 1,787.2 MEUR). The lower gross liquidity is attributable to an increase in net working capital as part of the scheduled processing of large orders and to the repayment of loans during the financial year. The significant decrease in the other financial result compared to the previous year is mainly attributable to a one- time effect in the context of the deconsolidation of a company. The other financial result of 2024 includes a one- time effect from the write-off of a loan granted to this company and the result from companies accounted for using the equity method includes a one-time effect from the measurement of the remaining shares at fair value. The tax rate decreased from 26.7% in the previous year to 23.2%. Above all, the capitalization of loss carryforwards through positive effects in the context of mergers, active management of tax groups and the consistent use of research and investment benefits have contributed to the reduction of the tax rate.\n\n**d)** **Earnings and dividend per share** The earnings per share decreased to 5.02 EUR (2023: 5.15 EUR). At the Annual General Meeting on March 27, 2025, the Executive Board will propose a dividend of 2.60 EUR per share (2023: 2.50 EUR) for the 2024 financial year. This is equal to a payout ratio of 51.8% (2023: 48.5%).\n\nDividend for 2024: Proposal to the Annual General Meeting.\n\n**2,08**\n\n**3,28**\n\n**4,14**\n\n**5,15 5,02**\n\n1,00\n\n1,65\n\n2,10\n\n2,50 2,60\n\n48,1%\n\n50,3% 50,7%\n\n48,5% 51,8%\n\n0,00\n\n1,00\n\n2,00\n\n3,00\n\n4,00\n\n5,00\n\n6,00\n\n2020 2021 2022 2023 2024\n\nEarnings per share (EUR) Dividend per share (EUR) Payout ratio\n\n— **More informationin the notes to the consolidated financial statements chapter C) 16.** Income taxes.","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"A N D R I T Z f i n a n c i a l r e p o r t 2 0 2 4\n\nM a n a g e m e n t r e p o r t\n\n# **14**\n\n**c)** **Financial result and taxes**\n\nThe financial result amounted to -15.4 MEUR in the reporting period (2023: 3.0 MEUR). The net interest result decreased compared to the previous year’s reference figure due to the decline in the interest rate environment and lower gross liquidity (1,434.4 MEUR versus 2023: 1,787.2 MEUR). The lower gross liquidity is attributable to an increase in net working capital as part of the scheduled processing of large orders and to the repayment of loans during the financial year. The significant decrease in the other financial result compared to the previous year is mainly attributable to a one- time effect in the context of the deconsolidation of a company. The other financial result of 2024 includes a one- time effect from the write-off of a loan granted to this company and the result from companies accounted for using the equity method includes a one-time effect from the measurement of the remaining shares at fair value. The tax rate decreased from 26.7% in the previous year to 23.2%. Above all, the capitalization of loss carryforwards through positive effects in the context of mergers, active management of tax groups and the consistent use of research and investment benefits have contributed to the reduction of the tax rate.\n\n**d)** **Earnings and dividend per share** The earnings per share decreased to 5.02 EUR (2023: 5.15 EUR). At the Annual General Meeting on March 27, 2025, the Executive Board will propose a dividend of 2.60 EUR per share (2023: 2.50 EUR) for the 2024 financial year. This is equal to a payout ratio of 51.8% (2023: 48.5%).\n\nDividend for 2024: Proposal to the Annual General Meeting.\n\n**2,08**\n\n**3,28**\n\n**4,14**\n\n**5,15 5,02**\n\n1,00\n\n1,65\n\n2,10\n\n2,50 2,60\n\n48,1%\n\n50,3% 50,7%\n\n48,5% 51,8%\n\n0,00\n\n1,00\n\n2,00\n\n3,00\n\n4,00\n\n5,00\n\n6,00\n\n2020 2021 2022 2023 2024\n\nEarnings per share (EUR) Dividend per share (EUR) Payout ratio\n\n— **More informationin the notes to the consolidated financial statements chapter C) 16.** Income taxes.","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/andritz-financial-report-2024-e-data_p15.pdf"]},"response":{"errors":[],"extract_id":"extract_e043fc8603733e075d122bc6d10adad1","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["A N D R I T Z f i n a n c i a l r e p o r t 2 0 2 4\n\nM a n a g e m e n t r e p o r t\n\n# **14**\n\n**c)** **Financial result and taxes**\n\nThe financial result amounted to -15.4 MEUR in the reporting period (2023: 3.0 MEUR). The net interest result decreased compared to the previous year’s reference figure due to the decline in the interest rate environment and lower gross liquidity (1,434.4 MEUR versus 2023: 1,787.2 MEUR). The lower gross liquidity is attributable to an increase in net working capital as part of the scheduled processing of large orders and to the repayment of loans during the financial year. The significant decrease in the other financial result compared to the previous year is mainly attributable to a one- time effect in the context of the deconsolidation of a company. The other financial result of 2024 includes a one- time effect from the write-off of a loan granted to this company and the result from companies accounted for using the equity method includes a one-time effect from the measurement of the remaining shares at fair value. The tax rate decreased from 26.7% in the previous year to 23.2%. Above all, the capitalization of loss carryforwards through positive effects in the context of mergers, active management of tax groups and the consistent use of research and investment benefits have contributed to the reduction of the tax rate.\n\n**d)** **Earnings and dividend per share** The earnings per share decreased to 5.02 EUR (2023: 5.15 EUR). At the Annual General Meeting on March 27, 2025, the Executive Board will propose a dividend of 2.60 EUR per share (2023: 2.50 EUR) for the 2024 financial year. This is equal to a payout ratio of 51.8% (2023: 48.5%).\n\nDividend for 2024: Proposal to the Annual General Meeting.\n\n**2,08**\n\n**3,28**\n\n**4,14**\n\n**5,15 5,02**\n\n1,00\n\n1,65\n\n2,10\n\n2,50 2,60\n\n48,1%\n\n50,3% 50,7%\n\n48,5% 51,8%\n\n0,00\n\n1,00\n\n2,00\n\n3,00\n\n4,00\n\n5,00\n\n6,00\n\n2020 2021 2022 2023 2024\n\nEarnings per share (EUR) Dividend per share (EUR) Payout ratio\n\n— **More informationin the notes to the consolidated financial statements chapter C) 16.** Income taxes."],"full_content":"A N D R I T Z f i n a n c i a l r e p o r t 2 0 2 4\n\nM a n a g e m e n t r e p o r t\n\n# **14**\n\n**c)** **Financial result and taxes**\n\nThe financial result amounted to -15.4 MEUR in the reporting period (2023: 3.0 MEUR). The net interest result decreased compared to the previous year’s reference figure due to the decline in the interest rate environment and lower gross liquidity (1,434.4 MEUR versus 2023: 1,787.2 MEUR). The lower gross liquidity is attributable to an increase in net working capital as part of the scheduled processing of large orders and to the repayment of loans during the financial year. The significant decrease in the other financial result compared to the previous year is mainly attributable to a one- time effect in the context of the deconsolidation of a company. The other financial result of 2024 includes a one- time effect from the write-off of a loan granted to this company and the result from companies accounted for using the equity method includes a one-time effect from the measurement of the remaining shares at fair value. The tax rate decreased from 26.7% in the previous year to 23.2%. Above all, the capitalization of loss carryforwards through positive effects in the context of mergers, active management of tax groups and the consistent use of research and investment benefits have contributed to the reduction of the tax rate.\n\n**d)** **Earnings and dividend per share** The earnings per share decreased to 5.02 EUR (2023: 5.15 EUR). At the Annual General Meeting on March 27, 2025, the Executive Board will propose a dividend of 2.60 EUR per share (2023: 2.50 EUR) for the 2024 financial year. This is equal to a payout ratio of 51.8% (2023: 48.5%).\n\nDividend for 2024: Proposal to the Annual General Meeting.\n\n**2,08**\n\n**3,28**\n\n**4,14**\n\n**5,15 5,02**\n\n1,00\n\n1,65\n\n2,10\n\n2,50 2,60\n\n48,1%\n\n50,3% 50,7%\n\n48,5% 51,8%\n\n0,00\n\n1,00\n\n2,00\n\n3,00\n\n4,00\n\n5,00\n\n6,00\n\n2020 2021 2022 2023 2024\n\nEarnings per share (EUR) Dividend per share (EUR) Payout ratio\n\n— **More informationin the notes to the consolidated financial statements chapter C) 16.** Income taxes.","publish_date":null,"title":"d) Earnings and dividend per share","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/andritz-financial-report-2024-e-data_p15.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/andritz-financial-report-2024-e-data_p15.pdf"},"request":{"example_id":"chart/andritz-financial-report-2024-e-data_p15","product_type":"parse","source_file_path":"andritz-financial-report-2024-e-data_p15.pdf"},"started_at":"2026-11-04T19:46:22.072553"} +{"completed_at":"2026-11-04T19:46:24.482918","latency_in_ms":1182,"output":{"example_id":"chart/article_thebeatjun2025_p10","grounded_pages":[],"layout_pages":[],"markdown":"The **BEAT** | June 2025\n\n# Global Bond Yields Are on the Rise – Not Unique to the U.S.\n\n**TOP THEMES**\n\nSource: Bloomberg, MSIM. As of May 20, 2025. For illustrative purposes only. Not a recommendation to buy or sell any security. It is not possible to invest directly in an index. The views and opinions expressed are those of the portfolio management team at the time of writing/of this presentation and are subject to change at any time due to market, economic, or other conditions, and may not necessarily come to pass. Forecasts/estimates are based on current market conditions, subject to change, and may not necessarily come to pass. **Past performance is not indicative of future results.**\n\n### Long duration assets are being repriced globally, not just in the U.S. In fact, U.S. long-dated bond yields have risen less so than in other major regions thus far in 2025.\n\n### **Japanese Bond Yields Have Risen to Historic Levels** _U.S. 30-year yield (LH), Japan 30-year yield (RH)_\n\n10\n\n0\\.0%\n\n0\\.5%\n\n1\\.0%\n\n1\\.5%\n\n2\\.0%\n\n2\\.5%\n\n3\\.0%\n\n3\\.5%\n\n1\\.0%\n\n2\\.0%\n\n3\\.0%\n\n4\\.0%\n\n5\\.0%\n\n6\\.0%\n\n7\\.0%\n\n2000\n\n2002\n\n2004\n\n2006\n\n2008\n\n2010\n\n2012\n\n2014\n\n2016\n\n2018\n\n2020\n\n2022\n\n2024\n\nJapan 30Y Yield\n\nUS 30Y Yield\n\nUS 30Y Yield (LH) Japan 30Y Yield (RH)\n\n**Japanese bond yields have** **risen to record highs…**\n\n\\-60\n\n\\-40\n\n\\-20\n\n0\n\n20\n\n40\n\n60\n\n80\n\n100\n\nDec 2024 Jan 2025 Feb 2025 Mar 2025 Apr 2025\n\nUnited States Japan Germany\n\n### **U.S. Bond Yields Have Risen Least in 2025** _YTD cumulative change (in bps)_","pages":[{"line_numbers":[],"markdown":"The **BEAT** | June 2025\n\n# Global Bond Yields Are on the Rise – Not Unique to the U.S.\n\n**TOP THEMES**\n\nSource: Bloomberg, MSIM. As of May 20, 2025. For illustrative purposes only. Not a recommendation to buy or sell any security. It is not possible to invest directly in an index. The views and opinions expressed are those of the portfolio management team at the time of writing/of this presentation and are subject to change at any time due to market, economic, or other conditions, and may not necessarily come to pass. Forecasts/estimates are based on current market conditions, subject to change, and may not necessarily come to pass. **Past performance is not indicative of future results.**\n\n### Long duration assets are being repriced globally, not just in the U.S. In fact, U.S. long-dated bond yields have risen less so than in other major regions thus far in 2025.\n\n### **Japanese Bond Yields Have Risen to Historic Levels** _U.S. 30-year yield (LH), Japan 30-year yield (RH)_\n\n10\n\n0\\.0%\n\n0\\.5%\n\n1\\.0%\n\n1\\.5%\n\n2\\.0%\n\n2\\.5%\n\n3\\.0%\n\n3\\.5%\n\n1\\.0%\n\n2\\.0%\n\n3\\.0%\n\n4\\.0%\n\n5\\.0%\n\n6\\.0%\n\n7\\.0%\n\n2000\n\n2002\n\n2004\n\n2006\n\n2008\n\n2010\n\n2012\n\n2014\n\n2016\n\n2018\n\n2020\n\n2022\n\n2024\n\nJapan 30Y Yield\n\nUS 30Y Yield\n\nUS 30Y Yield (LH) Japan 30Y Yield (RH)\n\n**Japanese bond yields have** **risen to record highs…**\n\n\\-60\n\n\\-40\n\n\\-20\n\n0\n\n20\n\n40\n\n60\n\n80\n\n100\n\nDec 2024 Jan 2025 Feb 2025 Mar 2025 Apr 2025\n\nUnited States Japan Germany\n\n### **U.S. Bond Yields Have Risen Least in 2025** _YTD cumulative change (in bps)_","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"The **BEAT** | June 2025\n\n# Global Bond Yields Are on the Rise – Not Unique to the U.S.\n\n**TOP THEMES**\n\nSource: Bloomberg, MSIM. As of May 20, 2025. For illustrative purposes only. Not a recommendation to buy or sell any security. It is not possible to invest directly in an index. The views and opinions expressed are those of the portfolio management team at the time of writing/of this presentation and are subject to change at any time due to market, economic, or other conditions, and may not necessarily come to pass. Forecasts/estimates are based on current market conditions, subject to change, and may not necessarily come to pass. **Past performance is not indicative of future results.**\n\n### Long duration assets are being repriced globally, not just in the U.S. In fact, U.S. long-dated bond yields have risen less so than in other major regions thus far in 2025.\n\n### **Japanese Bond Yields Have Risen to Historic Levels** _U.S. 30-year yield (LH), Japan 30-year yield (RH)_\n\n10\n\n0\\.0%\n\n0\\.5%\n\n1\\.0%\n\n1\\.5%\n\n2\\.0%\n\n2\\.5%\n\n3\\.0%\n\n3\\.5%\n\n1\\.0%\n\n2\\.0%\n\n3\\.0%\n\n4\\.0%\n\n5\\.0%\n\n6\\.0%\n\n7\\.0%\n\n2000\n\n2002\n\n2004\n\n2006\n\n2008\n\n2010\n\n2012\n\n2014\n\n2016\n\n2018\n\n2020\n\n2022\n\n2024\n\nJapan 30Y Yield\n\nUS 30Y Yield\n\nUS 30Y Yield (LH) Japan 30Y Yield (RH)\n\n**Japanese bond yields have** **risen to record highs…**\n\n\\-60\n\n\\-40\n\n\\-20\n\n0\n\n20\n\n40\n\n60\n\n80\n\n100\n\nDec 2024 Jan 2025 Feb 2025 Mar 2025 Apr 2025\n\nUnited States Japan Germany\n\n### **U.S. Bond Yields Have Risen Least in 2025** _YTD cumulative change (in bps)_","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/article_thebeatjun2025_p10.pdf"]},"response":{"errors":[],"extract_id":"extract_b7bac3d92a313fa7fcbf90d6d0c46f87","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["The **BEAT** | June 2025\n\n# Global Bond Yields Are on the Rise – Not Unique to the U.S.\n\n**TOP THEMES**\n\nSource: Bloomberg, MSIM. As of May 20, 2025. For illustrative purposes only. Not a recommendation to buy or sell any security. It is not possible to invest directly in an index. The views and opinions expressed are those of the portfolio management team at the time of writing/of this presentation and are subject to change at any time due to market, economic, or other conditions, and may not necessarily come to pass. Forecasts/estimates are based on current market conditions, subject to change, and may not necessarily come to pass. **Past performance is not indicative of future results.**\n\n### Long duration assets are being repriced globally, not just in the U.S. In fact, U.S. long-dated bond yields have risen less so than in other major regions thus far in 2025.\n\n### **Japanese Bond Yields Have Risen to Historic Levels** _U.S. 30-year yield (LH), Japan 30-year yield (RH)_\n\n10\n\n0\\.0%\n\n0\\.5%\n\n1\\.0%\n\n1\\.5%\n\n2\\.0%\n\n2\\.5%\n\n3\\.0%\n\n3\\.5%\n\n1\\.0%\n\n2\\.0%\n\n3\\.0%\n\n4\\.0%\n\n5\\.0%\n\n6\\.0%\n\n7\\.0%\n\n2000\n\n2002\n\n2004\n\n2006\n\n2008\n\n2010\n\n2012\n\n2014\n\n2016\n\n2018\n\n2020\n\n2022\n\n2024\n\nJapan 30Y Yield\n\nUS 30Y Yield\n\nUS 30Y Yield (LH) Japan 30Y Yield (RH)\n\n**Japanese bond yields have** **risen to record highs…**\n\n\\-60\n\n\\-40\n\n\\-20\n\n0\n\n20\n\n40\n\n60\n\n80\n\n100\n\nDec 2024 Jan 2025 Feb 2025 Mar 2025 Apr 2025\n\nUnited States Japan Germany\n\n### **U.S. Bond Yields Have Risen Least in 2025** _YTD cumulative change (in bps)_"],"full_content":"The **BEAT** | June 2025\n\n# Global Bond Yields Are on the Rise – Not Unique to the U.S.\n\n**TOP THEMES**\n\nSource: Bloomberg, MSIM. As of May 20, 2025. For illustrative purposes only. Not a recommendation to buy or sell any security. It is not possible to invest directly in an index. The views and opinions expressed are those of the portfolio management team at the time of writing/of this presentation and are subject to change at any time due to market, economic, or other conditions, and may not necessarily come to pass. Forecasts/estimates are based on current market conditions, subject to change, and may not necessarily come to pass. **Past performance is not indicative of future results.**\n\n### Long duration assets are being repriced globally, not just in the U.S. In fact, U.S. long-dated bond yields have risen less so than in other major regions thus far in 2025.\n\n### **Japanese Bond Yields Have Risen to Historic Levels** _U.S. 30-year yield (LH), Japan 30-year yield (RH)_\n\n10\n\n0\\.0%\n\n0\\.5%\n\n1\\.0%\n\n1\\.5%\n\n2\\.0%\n\n2\\.5%\n\n3\\.0%\n\n3\\.5%\n\n1\\.0%\n\n2\\.0%\n\n3\\.0%\n\n4\\.0%\n\n5\\.0%\n\n6\\.0%\n\n7\\.0%\n\n2000\n\n2002\n\n2004\n\n2006\n\n2008\n\n2010\n\n2012\n\n2014\n\n2016\n\n2018\n\n2020\n\n2022\n\n2024\n\nJapan 30Y Yield\n\nUS 30Y Yield\n\nUS 30Y Yield (LH) Japan 30Y Yield (RH)\n\n**Japanese bond yields have** **risen to record highs…**\n\n\\-60\n\n\\-40\n\n\\-20\n\n0\n\n20\n\n40\n\n60\n\n80\n\n100\n\nDec 2024 Jan 2025 Feb 2025 Mar 2025 Apr 2025\n\nUnited States Japan Germany\n\n### **U.S. Bond Yields Have Risen Least in 2025** _YTD cumulative change (in bps)_","publish_date":null,"title":"Global Bond Yields Are on the Rise – Not Unique to the U.S.","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/article_thebeatjun2025_p10.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/article_thebeatjun2025_p10.pdf"},"request":{"example_id":"chart/article_thebeatjun2025_p10","product_type":"parse","source_file_path":"article_thebeatjun2025_p10.pdf"},"started_at":"2026-11-04T19:46:23.300280"} +{"completed_at":"2026-11-04T19:46:26.526901","latency_in_ms":2042,"output":{"example_id":"chart/article_thebeatjun2025_p11","grounded_pages":[],"layout_pages":[],"markdown":"The **BEAT** | June 2025\n\n# Municipals: Higher Yields and a Steeper Curve\n\n**TOP THEMES**\n\nSource: Bloomberg, MSIM. As of May 21, 2025. For illustrative purposes only. Not a recommendation to buy or sell any security. It is not possible to invest directly in an index. The views and opinions expressed are those of the portfolio management team at the time of writing/of this presentation and are subject to change at any time due to market, economic, or other conditions, and may not necessarily come to pass. Forecasts/estimates are based on current market conditions, subject to change, and may not necessarily come to pass. **Past performance is not indicative of future results.**\n\n### Longer-dated yields in the municipal market moved higher in May, a continuation of the YTD trend. In the face of heavy supply in recent months, the muni curve has (bear) steepened, to a greater degree than U.S. Treasuries (USTs), resulting in attractive Muni/UST ratios out on the curve.\n\n### **Muni/UST Ratios Look Attractive Farther Out on the Curve** _Muni/UST ratios across selected curves, 12/31/24 and present_\n\n11\n\n60%\n\n65%\n\n70%\n\n75%\n\n80%\n\n85%\n\n90%\n\n95%\n\n2Y 5Y 10Y 30Y\n\n12/31/2024 5/21/2025","pages":[{"line_numbers":[],"markdown":"The **BEAT** | June 2025\n\n# Municipals: Higher Yields and a Steeper Curve\n\n**TOP THEMES**\n\nSource: Bloomberg, MSIM. As of May 21, 2025. For illustrative purposes only. Not a recommendation to buy or sell any security. It is not possible to invest directly in an index. The views and opinions expressed are those of the portfolio management team at the time of writing/of this presentation and are subject to change at any time due to market, economic, or other conditions, and may not necessarily come to pass. Forecasts/estimates are based on current market conditions, subject to change, and may not necessarily come to pass. **Past performance is not indicative of future results.**\n\n### Longer-dated yields in the municipal market moved higher in May, a continuation of the YTD trend. In the face of heavy supply in recent months, the muni curve has (bear) steepened, to a greater degree than U.S. Treasuries (USTs), resulting in attractive Muni/UST ratios out on the curve.\n\n### **Muni/UST Ratios Look Attractive Farther Out on the Curve** _Muni/UST ratios across selected curves, 12/31/24 and present_\n\n11\n\n60%\n\n65%\n\n70%\n\n75%\n\n80%\n\n85%\n\n90%\n\n95%\n\n2Y 5Y 10Y 30Y\n\n12/31/2024 5/21/2025","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"The **BEAT** | June 2025\n\n# Municipals: Higher Yields and a Steeper Curve\n\n**TOP THEMES**\n\nSource: Bloomberg, MSIM. As of May 21, 2025. For illustrative purposes only. Not a recommendation to buy or sell any security. It is not possible to invest directly in an index. The views and opinions expressed are those of the portfolio management team at the time of writing/of this presentation and are subject to change at any time due to market, economic, or other conditions, and may not necessarily come to pass. Forecasts/estimates are based on current market conditions, subject to change, and may not necessarily come to pass. **Past performance is not indicative of future results.**\n\n### Longer-dated yields in the municipal market moved higher in May, a continuation of the YTD trend. In the face of heavy supply in recent months, the muni curve has (bear) steepened, to a greater degree than U.S. Treasuries (USTs), resulting in attractive Muni/UST ratios out on the curve.\n\n### **Muni/UST Ratios Look Attractive Farther Out on the Curve** _Muni/UST ratios across selected curves, 12/31/24 and present_\n\n11\n\n60%\n\n65%\n\n70%\n\n75%\n\n80%\n\n85%\n\n90%\n\n95%\n\n2Y 5Y 10Y 30Y\n\n12/31/2024 5/21/2025","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/article_thebeatjun2025_p11.pdf"]},"response":{"errors":[],"extract_id":"extract_65c45339641b0ca728d033e7db12cbbe","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["The **BEAT** | June 2025\n\n# Municipals: Higher Yields and a Steeper Curve\n\n**TOP THEMES**\n\nSource: Bloomberg, MSIM. As of May 21, 2025. For illustrative purposes only. Not a recommendation to buy or sell any security. It is not possible to invest directly in an index. The views and opinions expressed are those of the portfolio management team at the time of writing/of this presentation and are subject to change at any time due to market, economic, or other conditions, and may not necessarily come to pass. Forecasts/estimates are based on current market conditions, subject to change, and may not necessarily come to pass. **Past performance is not indicative of future results.**\n\n### Longer-dated yields in the municipal market moved higher in May, a continuation of the YTD trend. In the face of heavy supply in recent months, the muni curve has (bear) steepened, to a greater degree than U.S. Treasuries (USTs), resulting in attractive Muni/UST ratios out on the curve.\n\n### **Muni/UST Ratios Look Attractive Farther Out on the Curve** _Muni/UST ratios across selected curves, 12/31/24 and present_\n\n11\n\n60%\n\n65%\n\n70%\n\n75%\n\n80%\n\n85%\n\n90%\n\n95%\n\n2Y 5Y 10Y 30Y\n\n12/31/2024 5/21/2025"],"full_content":"The **BEAT** | June 2025\n\n# Municipals: Higher Yields and a Steeper Curve\n\n**TOP THEMES**\n\nSource: Bloomberg, MSIM. As of May 21, 2025. For illustrative purposes only. Not a recommendation to buy or sell any security. It is not possible to invest directly in an index. The views and opinions expressed are those of the portfolio management team at the time of writing/of this presentation and are subject to change at any time due to market, economic, or other conditions, and may not necessarily come to pass. Forecasts/estimates are based on current market conditions, subject to change, and may not necessarily come to pass. **Past performance is not indicative of future results.**\n\n### Longer-dated yields in the municipal market moved higher in May, a continuation of the YTD trend. In the face of heavy supply in recent months, the muni curve has (bear) steepened, to a greater degree than U.S. Treasuries (USTs), resulting in attractive Muni/UST ratios out on the curve.\n\n### **Muni/UST Ratios Look Attractive Farther Out on the Curve** _Muni/UST ratios across selected curves, 12/31/24 and present_\n\n11\n\n60%\n\n65%\n\n70%\n\n75%\n\n80%\n\n85%\n\n90%\n\n95%\n\n2Y 5Y 10Y 30Y\n\n12/31/2024 5/21/2025","publish_date":null,"title":"Municipals: Higher Yields and a Steeper Curve","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/article_thebeatjun2025_p11.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/article_thebeatjun2025_p11.pdf"},"request":{"example_id":"chart/article_thebeatjun2025_p11","product_type":"parse","source_file_path":"article_thebeatjun2025_p11.pdf"},"started_at":"2026-11-04T19:46:24.484073"} +{"completed_at":"2026-11-04T19:46:29.009707","latency_in_ms":2613,"output":{"example_id":"chart/article_thebeatjun2025_p12","grounded_pages":[],"layout_pages":[],"markdown":"The **BEAT** | June 2025\n\n# Overweight European High Yield Credit\n\n**TOP IDEAS**\n\n### The reduction in global growth risks from the recent positive trade negotiation outcomes should reduce default risks, making lower-quality credit more attractive. European HY credit appears attractive especially for USD-based investors, given an additional 2.4% pick up from hedging euro exposure back to USD.\n\n12\n\n### **European HY Credit Offers the Highest Yield on USD-Hedged Basis** _Yield decomposition across major credit asset classes_\n\nSource: Bloomberg, MSIM. As of May 22, 2025. For illustrative purposes only. Not a recommendation to buy or sell any security. It is not possible to invest directly in an index. The views and opinions expressed are those of the portfolio management team at the time of writing/of this presentation and are subject to change at any time due to market, economic, or other conditions, and may not necessarily come to pass. Forecasts/estimates are based on current market conditions, subject to change, and may not necessarily come to pass. **Past performance is not indicative of future results.**\n\n**8\\.1% 7.9% 7.6%**\n\n**5\\.6% 5.4%**\n\n0%\n\n1%\n\n2%\n\n3%\n\n4%\n\n5%\n\n6%\n\n7%\n\n8%\n\n9%\n\nEUR High Yield EM Hard Currency USD High Yield EUR IG Corp USD IG Corp\n\nBase Rate Option-Adjusted Spread (OAS) EURUSD 3m Hedging Income","pages":[{"line_numbers":[],"markdown":"The **BEAT** | June 2025\n\n# Overweight European High Yield Credit\n\n**TOP IDEAS**\n\n### The reduction in global growth risks from the recent positive trade negotiation outcomes should reduce default risks, making lower-quality credit more attractive. European HY credit appears attractive especially for USD-based investors, given an additional 2.4% pick up from hedging euro exposure back to USD.\n\n12\n\n### **European HY Credit Offers the Highest Yield on USD-Hedged Basis** _Yield decomposition across major credit asset classes_\n\nSource: Bloomberg, MSIM. As of May 22, 2025. For illustrative purposes only. Not a recommendation to buy or sell any security. It is not possible to invest directly in an index. The views and opinions expressed are those of the portfolio management team at the time of writing/of this presentation and are subject to change at any time due to market, economic, or other conditions, and may not necessarily come to pass. Forecasts/estimates are based on current market conditions, subject to change, and may not necessarily come to pass. **Past performance is not indicative of future results.**\n\n**8\\.1% 7.9% 7.6%**\n\n**5\\.6% 5.4%**\n\n0%\n\n1%\n\n2%\n\n3%\n\n4%\n\n5%\n\n6%\n\n7%\n\n8%\n\n9%\n\nEUR High Yield EM Hard Currency USD High Yield EUR IG Corp USD IG Corp\n\nBase Rate Option-Adjusted Spread (OAS) EURUSD 3m Hedging Income","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"The **BEAT** | June 2025\n\n# Overweight European High Yield Credit\n\n**TOP IDEAS**\n\n### The reduction in global growth risks from the recent positive trade negotiation outcomes should reduce default risks, making lower-quality credit more attractive. European HY credit appears attractive especially for USD-based investors, given an additional 2.4% pick up from hedging euro exposure back to USD.\n\n12\n\n### **European HY Credit Offers the Highest Yield on USD-Hedged Basis** _Yield decomposition across major credit asset classes_\n\nSource: Bloomberg, MSIM. As of May 22, 2025. For illustrative purposes only. Not a recommendation to buy or sell any security. It is not possible to invest directly in an index. The views and opinions expressed are those of the portfolio management team at the time of writing/of this presentation and are subject to change at any time due to market, economic, or other conditions, and may not necessarily come to pass. Forecasts/estimates are based on current market conditions, subject to change, and may not necessarily come to pass. **Past performance is not indicative of future results.**\n\n**8\\.1% 7.9% 7.6%**\n\n**5\\.6% 5.4%**\n\n0%\n\n1%\n\n2%\n\n3%\n\n4%\n\n5%\n\n6%\n\n7%\n\n8%\n\n9%\n\nEUR High Yield EM Hard Currency USD High Yield EUR IG Corp USD IG Corp\n\nBase Rate Option-Adjusted Spread (OAS) EURUSD 3m Hedging Income","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/article_thebeatjun2025_p12.pdf"]},"response":{"errors":[],"extract_id":"extract_37457213c22e726cfea62d518aec18f5","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["The **BEAT** | June 2025\n\n# Overweight European High Yield Credit\n\n**TOP IDEAS**\n\n### The reduction in global growth risks from the recent positive trade negotiation outcomes should reduce default risks, making lower-quality credit more attractive. European HY credit appears attractive especially for USD-based investors, given an additional 2.4% pick up from hedging euro exposure back to USD.\n\n12\n\n### **European HY Credit Offers the Highest Yield on USD-Hedged Basis** _Yield decomposition across major credit asset classes_\n\nSource: Bloomberg, MSIM. As of May 22, 2025. For illustrative purposes only. Not a recommendation to buy or sell any security. It is not possible to invest directly in an index. The views and opinions expressed are those of the portfolio management team at the time of writing/of this presentation and are subject to change at any time due to market, economic, or other conditions, and may not necessarily come to pass. Forecasts/estimates are based on current market conditions, subject to change, and may not necessarily come to pass. **Past performance is not indicative of future results.**\n\n**8\\.1% 7.9% 7.6%**\n\n**5\\.6% 5.4%**\n\n0%\n\n1%\n\n2%\n\n3%\n\n4%\n\n5%\n\n6%\n\n7%\n\n8%\n\n9%\n\nEUR High Yield EM Hard Currency USD High Yield EUR IG Corp USD IG Corp\n\nBase Rate Option-Adjusted Spread (OAS) EURUSD 3m Hedging Income"],"full_content":"The **BEAT** | June 2025\n\n# Overweight European High Yield Credit\n\n**TOP IDEAS**\n\n### The reduction in global growth risks from the recent positive trade negotiation outcomes should reduce default risks, making lower-quality credit more attractive. European HY credit appears attractive especially for USD-based investors, given an additional 2.4% pick up from hedging euro exposure back to USD.\n\n12\n\n### **European HY Credit Offers the Highest Yield on USD-Hedged Basis** _Yield decomposition across major credit asset classes_\n\nSource: Bloomberg, MSIM. As of May 22, 2025. For illustrative purposes only. Not a recommendation to buy or sell any security. It is not possible to invest directly in an index. The views and opinions expressed are those of the portfolio management team at the time of writing/of this presentation and are subject to change at any time due to market, economic, or other conditions, and may not necessarily come to pass. Forecasts/estimates are based on current market conditions, subject to change, and may not necessarily come to pass. **Past performance is not indicative of future results.**\n\n**8\\.1% 7.9% 7.6%**\n\n**5\\.6% 5.4%**\n\n0%\n\n1%\n\n2%\n\n3%\n\n4%\n\n5%\n\n6%\n\n7%\n\n8%\n\n9%\n\nEUR High Yield EM Hard Currency USD High Yield EUR IG Corp USD IG Corp\n\nBase Rate Option-Adjusted Spread (OAS) EURUSD 3m Hedging Income","publish_date":null,"title":"Overweight European High Yield Credit","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/article_thebeatjun2025_p12.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/article_thebeatjun2025_p12.pdf"},"request":{"example_id":"chart/article_thebeatjun2025_p12","product_type":"parse","source_file_path":"article_thebeatjun2025_p12.pdf"},"started_at":"2026-11-04T19:46:26.396313"} +{"completed_at":"2026-11-04T19:46:29.543002","latency_in_ms":2073,"output":{"example_id":"chart/article_thebeatjun2025_p15","grounded_pages":[],"layout_pages":[],"markdown":"The **BEAT** | June 2025\n\n# Real Estate Fundamentals Continue to Improve\n\n15\n\n### Real estate has been experiencing a repricing over the last few years in response to higher interest rates, cyclical oversupply and, in certain sectors, secular demand destruction. That said, the long-term operating outlook is markedly improving, with future supply materially decreasing and the demand destruction seen in certain sectors stabilizing. Entry pricing is meaningfully lower, providing a margin of safety and an interesting access point to improving fundamentals.\n\n**TOP IDEAS**\n\nSource: CoStar, Bloomberg, Bureau of Labor Statistics. As of Dec 31, 2024. For illustrative purposes only. Not a recommendation to buy or sell any security. It is not possible to invest directly in an index. The views and opinions expressed are those of the portfolio management team at the time of writing/of this presentation and are subject to change at any time due to market, economic, or other conditions, and may not necessarily come to pass. Forecasts/estimates are based on current market conditions, subject to change, and may not necessarily come to pass. **Past performance is not indicative of future results.** .\n\n### **U.S. Cap Rates Have Stabilized…**\n\n_Real estate sector cap rates (%)_\n\n### **…and Asset Valuations Are Below Replacement Cost**\n\n_Residential price index (LH), construction as % of inventory (RH)_\n\n0%\n\n1%\n\n2%\n\n3%\n\n4%\n\n5%\n\n6%\n\n7%\n\n0\n\n50\n\n100\n\n150\n\n200\n\n250\n\n300\n\n2005\n\n2006\n\n2007\n\n2008\n\n2009\n\n2010\n\n2011\n\n2012\n\n2013\n\n2014\n\n2015\n\n2016\n\n2017\n\n2018\n\n2019\n\n2020\n\n2021\n\n2022\n\n2023\n\n2024\n\nMarket Sale Price Index (L) Construction Price Index (L)\n\nConstruction % of Inventory (R)\n\n0\n\n1\n\n2\n\n3\n\n4\n\n5\n\n6\n\n7\n\n8\n\n9\n\n10\n\n2002\n\n2003\n\n2004\n\n2005\n\n2006\n\n2007\n\n2008\n\n2009\n\n2010\n\n2011\n\n2012\n\n2013\n\n2014\n\n2015\n\n2016\n\n2017\n\n2018\n\n2019\n\n2020\n\n2021\n\n2022\n\n2023\n\n2024\n\nOffice Retail Industrial Residential 10y GOV","pages":[{"line_numbers":[],"markdown":"The **BEAT** | June 2025\n\n# Real Estate Fundamentals Continue to Improve\n\n15\n\n### Real estate has been experiencing a repricing over the last few years in response to higher interest rates, cyclical oversupply and, in certain sectors, secular demand destruction. That said, the long-term operating outlook is markedly improving, with future supply materially decreasing and the demand destruction seen in certain sectors stabilizing. Entry pricing is meaningfully lower, providing a margin of safety and an interesting access point to improving fundamentals.\n\n**TOP IDEAS**\n\nSource: CoStar, Bloomberg, Bureau of Labor Statistics. As of Dec 31, 2024. For illustrative purposes only. Not a recommendation to buy or sell any security. It is not possible to invest directly in an index. The views and opinions expressed are those of the portfolio management team at the time of writing/of this presentation and are subject to change at any time due to market, economic, or other conditions, and may not necessarily come to pass. Forecasts/estimates are based on current market conditions, subject to change, and may not necessarily come to pass. **Past performance is not indicative of future results.** .\n\n### **U.S. Cap Rates Have Stabilized…**\n\n_Real estate sector cap rates (%)_\n\n### **…and Asset Valuations Are Below Replacement Cost**\n\n_Residential price index (LH), construction as % of inventory (RH)_\n\n0%\n\n1%\n\n2%\n\n3%\n\n4%\n\n5%\n\n6%\n\n7%\n\n0\n\n50\n\n100\n\n150\n\n200\n\n250\n\n300\n\n2005\n\n2006\n\n2007\n\n2008\n\n2009\n\n2010\n\n2011\n\n2012\n\n2013\n\n2014\n\n2015\n\n2016\n\n2017\n\n2018\n\n2019\n\n2020\n\n2021\n\n2022\n\n2023\n\n2024\n\nMarket Sale Price Index (L) Construction Price Index (L)\n\nConstruction % of Inventory (R)\n\n0\n\n1\n\n2\n\n3\n\n4\n\n5\n\n6\n\n7\n\n8\n\n9\n\n10\n\n2002\n\n2003\n\n2004\n\n2005\n\n2006\n\n2007\n\n2008\n\n2009\n\n2010\n\n2011\n\n2012\n\n2013\n\n2014\n\n2015\n\n2016\n\n2017\n\n2018\n\n2019\n\n2020\n\n2021\n\n2022\n\n2023\n\n2024\n\nOffice Retail Industrial Residential 10y GOV","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"The **BEAT** | June 2025\n\n# Real Estate Fundamentals Continue to Improve\n\n15\n\n### Real estate has been experiencing a repricing over the last few years in response to higher interest rates, cyclical oversupply and, in certain sectors, secular demand destruction. That said, the long-term operating outlook is markedly improving, with future supply materially decreasing and the demand destruction seen in certain sectors stabilizing. Entry pricing is meaningfully lower, providing a margin of safety and an interesting access point to improving fundamentals.\n\n**TOP IDEAS**\n\nSource: CoStar, Bloomberg, Bureau of Labor Statistics. As of Dec 31, 2024. For illustrative purposes only. Not a recommendation to buy or sell any security. It is not possible to invest directly in an index. The views and opinions expressed are those of the portfolio management team at the time of writing/of this presentation and are subject to change at any time due to market, economic, or other conditions, and may not necessarily come to pass. Forecasts/estimates are based on current market conditions, subject to change, and may not necessarily come to pass. **Past performance is not indicative of future results.** .\n\n### **U.S. Cap Rates Have Stabilized…**\n\n_Real estate sector cap rates (%)_\n\n### **…and Asset Valuations Are Below Replacement Cost**\n\n_Residential price index (LH), construction as % of inventory (RH)_\n\n0%\n\n1%\n\n2%\n\n3%\n\n4%\n\n5%\n\n6%\n\n7%\n\n0\n\n50\n\n100\n\n150\n\n200\n\n250\n\n300\n\n2005\n\n2006\n\n2007\n\n2008\n\n2009\n\n2010\n\n2011\n\n2012\n\n2013\n\n2014\n\n2015\n\n2016\n\n2017\n\n2018\n\n2019\n\n2020\n\n2021\n\n2022\n\n2023\n\n2024\n\nMarket Sale Price Index (L) Construction Price Index (L)\n\nConstruction % of Inventory (R)\n\n0\n\n1\n\n2\n\n3\n\n4\n\n5\n\n6\n\n7\n\n8\n\n9\n\n10\n\n2002\n\n2003\n\n2004\n\n2005\n\n2006\n\n2007\n\n2008\n\n2009\n\n2010\n\n2011\n\n2012\n\n2013\n\n2014\n\n2015\n\n2016\n\n2017\n\n2018\n\n2019\n\n2020\n\n2021\n\n2022\n\n2023\n\n2024\n\nOffice Retail Industrial Residential 10y GOV","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/article_thebeatjun2025_p15.pdf"]},"response":{"errors":[],"extract_id":"extract_13b32ede2e09f83ea32ab849e1bbd1d4","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["The **BEAT** | June 2025\n\n# Real Estate Fundamentals Continue to Improve\n\n15\n\n### Real estate has been experiencing a repricing over the last few years in response to higher interest rates, cyclical oversupply and, in certain sectors, secular demand destruction. That said, the long-term operating outlook is markedly improving, with future supply materially decreasing and the demand destruction seen in certain sectors stabilizing. Entry pricing is meaningfully lower, providing a margin of safety and an interesting access point to improving fundamentals.\n\n**TOP IDEAS**\n\nSource: CoStar, Bloomberg, Bureau of Labor Statistics. As of Dec 31, 2024. For illustrative purposes only. Not a recommendation to buy or sell any security. It is not possible to invest directly in an index. The views and opinions expressed are those of the portfolio management team at the time of writing/of this presentation and are subject to change at any time due to market, economic, or other conditions, and may not necessarily come to pass. Forecasts/estimates are based on current market conditions, subject to change, and may not necessarily come to pass. **Past performance is not indicative of future results.** .\n\n### **U.S. Cap Rates Have Stabilized…**\n\n_Real estate sector cap rates (%)_\n\n### **…and Asset Valuations Are Below Replacement Cost**\n\n_Residential price index (LH), construction as % of inventory (RH)_\n\n0%\n\n1%\n\n2%\n\n3%\n\n4%\n\n5%\n\n6%\n\n7%\n\n0\n\n50\n\n100\n\n150\n\n200\n\n250\n\n300\n\n2005\n\n2006\n\n2007\n\n2008\n\n2009\n\n2010\n\n2011\n\n2012\n\n2013\n\n2014\n\n2015\n\n2016\n\n2017\n\n2018\n\n2019\n\n2020\n\n2021\n\n2022\n\n2023\n\n2024\n\nMarket Sale Price Index (L) Construction Price Index (L)\n\nConstruction % of Inventory (R)\n\n0\n\n1\n\n2\n\n3\n\n4\n\n5\n\n6\n\n7\n\n8\n\n9\n\n10\n\n2002\n\n2003\n\n2004\n\n2005\n\n2006\n\n2007\n\n2008\n\n2009\n\n2010\n\n2011\n\n2012\n\n2013\n\n2014\n\n2015\n\n2016\n\n2017\n\n2018\n\n2019\n\n2020\n\n2021\n\n2022\n\n2023\n\n2024\n\nOffice Retail Industrial Residential 10y GOV"],"full_content":"The **BEAT** | June 2025\n\n# Real Estate Fundamentals Continue to Improve\n\n15\n\n### Real estate has been experiencing a repricing over the last few years in response to higher interest rates, cyclical oversupply and, in certain sectors, secular demand destruction. That said, the long-term operating outlook is markedly improving, with future supply materially decreasing and the demand destruction seen in certain sectors stabilizing. Entry pricing is meaningfully lower, providing a margin of safety and an interesting access point to improving fundamentals.\n\n**TOP IDEAS**\n\nSource: CoStar, Bloomberg, Bureau of Labor Statistics. As of Dec 31, 2024. For illustrative purposes only. Not a recommendation to buy or sell any security. It is not possible to invest directly in an index. The views and opinions expressed are those of the portfolio management team at the time of writing/of this presentation and are subject to change at any time due to market, economic, or other conditions, and may not necessarily come to pass. Forecasts/estimates are based on current market conditions, subject to change, and may not necessarily come to pass. **Past performance is not indicative of future results.** .\n\n### **U.S. Cap Rates Have Stabilized…**\n\n_Real estate sector cap rates (%)_\n\n### **…and Asset Valuations Are Below Replacement Cost**\n\n_Residential price index (LH), construction as % of inventory (RH)_\n\n0%\n\n1%\n\n2%\n\n3%\n\n4%\n\n5%\n\n6%\n\n7%\n\n0\n\n50\n\n100\n\n150\n\n200\n\n250\n\n300\n\n2005\n\n2006\n\n2007\n\n2008\n\n2009\n\n2010\n\n2011\n\n2012\n\n2013\n\n2014\n\n2015\n\n2016\n\n2017\n\n2018\n\n2019\n\n2020\n\n2021\n\n2022\n\n2023\n\n2024\n\nMarket Sale Price Index (L) Construction Price Index (L)\n\nConstruction % of Inventory (R)\n\n0\n\n1\n\n2\n\n3\n\n4\n\n5\n\n6\n\n7\n\n8\n\n9\n\n10\n\n2002\n\n2003\n\n2004\n\n2005\n\n2006\n\n2007\n\n2008\n\n2009\n\n2010\n\n2011\n\n2012\n\n2013\n\n2014\n\n2015\n\n2016\n\n2017\n\n2018\n\n2019\n\n2020\n\n2021\n\n2022\n\n2023\n\n2024\n\nOffice Retail Industrial Residential 10y GOV","publish_date":null,"title":"Real Estate Fundamentals Continue to Improve","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/article_thebeatjun2025_p15.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/article_thebeatjun2025_p15.pdf"},"request":{"example_id":"chart/article_thebeatjun2025_p15","product_type":"parse","source_file_path":"article_thebeatjun2025_p15.pdf"},"started_at":"2026-11-04T19:46:27.469069"} +{"completed_at":"2026-11-04T19:46:30.301880","latency_in_ms":1318,"output":{"example_id":"chart/article_thebeatjun2025_p20","grounded_pages":[],"layout_pages":[],"markdown":"The **BEAT** | June 2025\n\n# Sovereign Bond Yields\n\n20\n\n### **Developed Markets**\n\n(10 yr. Yield)\n\n### **Emerging Markets**\n\n(10 yr. Yield)\n\n**BONDS**\n\n**Past performance is no guarantee of future results.** It is not possible to invest directly in an index. Source: Factset as of 5/31/25. Data provided is for informational use only. See end of report for important additional information.\n\n**1 Mo. Ago**\n\n**12 Mo. Ago**\n\n0%\n\n4%\n\n8%\n\n12%\n\n16%\n\n**May ’25**\n\n**1 Mo. Ago**\n\n**12 Mo. Ago**\n\n0%\n\n4%\n\n8%\n\n12%\n\n16%\n\n**May ’25**","pages":[{"line_numbers":[],"markdown":"The **BEAT** | June 2025\n\n# Sovereign Bond Yields\n\n20\n\n### **Developed Markets**\n\n(10 yr. Yield)\n\n### **Emerging Markets**\n\n(10 yr. Yield)\n\n**BONDS**\n\n**Past performance is no guarantee of future results.** It is not possible to invest directly in an index. Source: Factset as of 5/31/25. Data provided is for informational use only. See end of report for important additional information.\n\n**1 Mo. Ago**\n\n**12 Mo. Ago**\n\n0%\n\n4%\n\n8%\n\n12%\n\n16%\n\n**May ’25**\n\n**1 Mo. Ago**\n\n**12 Mo. Ago**\n\n0%\n\n4%\n\n8%\n\n12%\n\n16%\n\n**May ’25**","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"The **BEAT** | June 2025\n\n# Sovereign Bond Yields\n\n20\n\n### **Developed Markets**\n\n(10 yr. Yield)\n\n### **Emerging Markets**\n\n(10 yr. Yield)\n\n**BONDS**\n\n**Past performance is no guarantee of future results.** It is not possible to invest directly in an index. Source: Factset as of 5/31/25. Data provided is for informational use only. See end of report for important additional information.\n\n**1 Mo. Ago**\n\n**12 Mo. Ago**\n\n0%\n\n4%\n\n8%\n\n12%\n\n16%\n\n**May ’25**\n\n**1 Mo. Ago**\n\n**12 Mo. Ago**\n\n0%\n\n4%\n\n8%\n\n12%\n\n16%\n\n**May ’25**","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/article_thebeatjun2025_p20.pdf"]},"response":{"errors":[],"extract_id":"extract_0843cf0f0431b1f8ab721e41c6651f66","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["The **BEAT** | June 2025\n\n# Sovereign Bond Yields\n\n20\n\n### **Developed Markets**\n\n(10 yr. Yield)\n\n### **Emerging Markets**\n\n(10 yr. Yield)\n\n**BONDS**\n\n**Past performance is no guarantee of future results.** It is not possible to invest directly in an index. Source: Factset as of 5/31/25. Data provided is for informational use only. See end of report for important additional information.\n\n**1 Mo. Ago**\n\n**12 Mo. Ago**\n\n0%\n\n4%\n\n8%\n\n12%\n\n16%\n\n**May ’25**\n\n**1 Mo. Ago**\n\n**12 Mo. Ago**\n\n0%\n\n4%\n\n8%\n\n12%\n\n16%\n\n**May ’25**"],"full_content":"The **BEAT** | June 2025\n\n# Sovereign Bond Yields\n\n20\n\n### **Developed Markets**\n\n(10 yr. Yield)\n\n### **Emerging Markets**\n\n(10 yr. Yield)\n\n**BONDS**\n\n**Past performance is no guarantee of future results.** It is not possible to invest directly in an index. Source: Factset as of 5/31/25. Data provided is for informational use only. See end of report for important additional information.\n\n**1 Mo. Ago**\n\n**12 Mo. Ago**\n\n0%\n\n4%\n\n8%\n\n12%\n\n16%\n\n**May ’25**\n\n**1 Mo. Ago**\n\n**12 Mo. Ago**\n\n0%\n\n4%\n\n8%\n\n12%\n\n16%\n\n**May ’25**","publish_date":null,"title":"Sovereign Bond Yields","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/article_thebeatjun2025_p20.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/article_thebeatjun2025_p20.pdf"},"request":{"example_id":"chart/article_thebeatjun2025_p20","product_type":"parse","source_file_path":"article_thebeatjun2025_p20.pdf"},"started_at":"2026-11-04T19:46:28.983001"} +{"completed_at":"2026-11-04T19:46:31.752543","latency_in_ms":1449,"output":{"example_id":"chart/article_thebeatjun2025_p22","grounded_pages":[],"layout_pages":[],"markdown":"The **BEAT** | June 2025\n\n4\\.29%\n\n4\\.06%\n\n3\\.80%\n\n3\\.62% 3.55%\n\n4\\.49%\n\n4\\.23%\n\n3\\.95%\n\n3\\.74%\n\n3\\.61%\n\n0\\.47%\n\n0\\.58%\n\n0\\.74% 0.83% 0.86%\n\n2\\.48%\n\n1\\.95%\n\n1\\.66% 1.72%\n\n1\\.85%\n\n0\\.0%\n\n1\\.0%\n\n2\\.0%\n\n3\\.0%\n\n4\\.0%\n\n5\\.0%\n\n3M 6M 1Y 2Y 3Y\n\n**BOE**\n\n**U.S. Federal Reserve**\n\n**ECB**\n\n**BOJ**\n\n4\\.27%\n\n3\\.98%\n\n3\\.54%\n\n3\\.30%\n\n3\\.50%\n\n4\\.11%\n\n3\\.92%\n\n3\\.68% 3.66% 3.79%\n\n0\\.55%\n\n0\\.66%\n\n0\\.82%\n\n1\\.05%\n\n0\\.99%\n\n1\\.91%\n\n1\\.76% 1.67%\n\n1\\.88%\n\n2\\.17%\n\n0\\.0%\n\n1\\.0%\n\n2\\.0%\n\n3\\.0%\n\n4\\.0%\n\n5\\.0%\n\n3M 6M 1Y 2Y 3Y\n\n**BOE**\n\n**U.S. Federal Reserve**\n\n**ECB**\n\n**BOJ**\n\n(1%)\n\n0%\n\n1%\n\n2%\n\n3%\n\n4%\n\n5%\n\n6%\n\n7%\n\n8%\n\n9%\n\n10%\n\n'98 '00 '02 '04 '06 '08 '10 '12 '14 '16 '18 '20 '22 '24\n\n**BOE U.S. Federal Reserve**\n\n**ECB**\n\n**BOJ**\n\n# Monetary Policy\n\nSource: Bloomberg, Factset as of 5/31/25. Data provided is for informational use only. See end of report for important additional information. Forecasts/estimates are based on current market conditions, subject to change, and may not necessarily come to pass.\n\n**Current 1-Mo. Ago 12-Mo. Ago**\n\n**U.S. Federal Reserve** 4\\.50% 4.50% 5.50%\n\n**BOE** 4\\.25% 4.50% 5.25%\n\n**BOJ** 0\\.50% 0.50% 0.10%\n\n**ECB** 2\\.40% 2.40% 4.50%\n\n### **Market Expectations for Future Central Bank Rates Central Bank Policy Rates**\n\n**BONDS**\n\n22","pages":[{"line_numbers":[],"markdown":"The **BEAT** | June 2025\n\n4\\.29%\n\n4\\.06%\n\n3\\.80%\n\n3\\.62% 3.55%\n\n4\\.49%\n\n4\\.23%\n\n3\\.95%\n\n3\\.74%\n\n3\\.61%\n\n0\\.47%\n\n0\\.58%\n\n0\\.74% 0.83% 0.86%\n\n2\\.48%\n\n1\\.95%\n\n1\\.66% 1.72%\n\n1\\.85%\n\n0\\.0%\n\n1\\.0%\n\n2\\.0%\n\n3\\.0%\n\n4\\.0%\n\n5\\.0%\n\n3M 6M 1Y 2Y 3Y\n\n**BOE**\n\n**U.S. Federal Reserve**\n\n**ECB**\n\n**BOJ**\n\n4\\.27%\n\n3\\.98%\n\n3\\.54%\n\n3\\.30%\n\n3\\.50%\n\n4\\.11%\n\n3\\.92%\n\n3\\.68% 3.66% 3.79%\n\n0\\.55%\n\n0\\.66%\n\n0\\.82%\n\n1\\.05%\n\n0\\.99%\n\n1\\.91%\n\n1\\.76% 1.67%\n\n1\\.88%\n\n2\\.17%\n\n0\\.0%\n\n1\\.0%\n\n2\\.0%\n\n3\\.0%\n\n4\\.0%\n\n5\\.0%\n\n3M 6M 1Y 2Y 3Y\n\n**BOE**\n\n**U.S. Federal Reserve**\n\n**ECB**\n\n**BOJ**\n\n(1%)\n\n0%\n\n1%\n\n2%\n\n3%\n\n4%\n\n5%\n\n6%\n\n7%\n\n8%\n\n9%\n\n10%\n\n'98 '00 '02 '04 '06 '08 '10 '12 '14 '16 '18 '20 '22 '24\n\n**BOE U.S. Federal Reserve**\n\n**ECB**\n\n**BOJ**\n\n# Monetary Policy\n\nSource: Bloomberg, Factset as of 5/31/25. Data provided is for informational use only. See end of report for important additional information. Forecasts/estimates are based on current market conditions, subject to change, and may not necessarily come to pass.\n\n**Current 1-Mo. Ago 12-Mo. Ago**\n\n**U.S. Federal Reserve** 4\\.50% 4.50% 5.50%\n\n**BOE** 4\\.25% 4.50% 5.25%\n\n**BOJ** 0\\.50% 0.50% 0.10%\n\n**ECB** 2\\.40% 2.40% 4.50%\n\n### **Market Expectations for Future Central Bank Rates Central Bank Policy Rates**\n\n**BONDS**\n\n22","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"The **BEAT** | June 2025\n\n4\\.29%\n\n4\\.06%\n\n3\\.80%\n\n3\\.62% 3.55%\n\n4\\.49%\n\n4\\.23%\n\n3\\.95%\n\n3\\.74%\n\n3\\.61%\n\n0\\.47%\n\n0\\.58%\n\n0\\.74% 0.83% 0.86%\n\n2\\.48%\n\n1\\.95%\n\n1\\.66% 1.72%\n\n1\\.85%\n\n0\\.0%\n\n1\\.0%\n\n2\\.0%\n\n3\\.0%\n\n4\\.0%\n\n5\\.0%\n\n3M 6M 1Y 2Y 3Y\n\n**BOE**\n\n**U.S. Federal Reserve**\n\n**ECB**\n\n**BOJ**\n\n4\\.27%\n\n3\\.98%\n\n3\\.54%\n\n3\\.30%\n\n3\\.50%\n\n4\\.11%\n\n3\\.92%\n\n3\\.68% 3.66% 3.79%\n\n0\\.55%\n\n0\\.66%\n\n0\\.82%\n\n1\\.05%\n\n0\\.99%\n\n1\\.91%\n\n1\\.76% 1.67%\n\n1\\.88%\n\n2\\.17%\n\n0\\.0%\n\n1\\.0%\n\n2\\.0%\n\n3\\.0%\n\n4\\.0%\n\n5\\.0%\n\n3M 6M 1Y 2Y 3Y\n\n**BOE**\n\n**U.S. Federal Reserve**\n\n**ECB**\n\n**BOJ**\n\n(1%)\n\n0%\n\n1%\n\n2%\n\n3%\n\n4%\n\n5%\n\n6%\n\n7%\n\n8%\n\n9%\n\n10%\n\n'98 '00 '02 '04 '06 '08 '10 '12 '14 '16 '18 '20 '22 '24\n\n**BOE U.S. Federal Reserve**\n\n**ECB**\n\n**BOJ**\n\n# Monetary Policy\n\nSource: Bloomberg, Factset as of 5/31/25. Data provided is for informational use only. See end of report for important additional information. Forecasts/estimates are based on current market conditions, subject to change, and may not necessarily come to pass.\n\n**Current 1-Mo. Ago 12-Mo. Ago**\n\n**U.S. Federal Reserve** 4\\.50% 4.50% 5.50%\n\n**BOE** 4\\.25% 4.50% 5.25%\n\n**BOJ** 0\\.50% 0.50% 0.10%\n\n**ECB** 2\\.40% 2.40% 4.50%\n\n### **Market Expectations for Future Central Bank Rates Central Bank Policy Rates**\n\n**BONDS**\n\n22","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/article_thebeatjun2025_p22.pdf"]},"response":{"errors":[],"extract_id":"extract_554fba2ef59c56ec90cc92e355f9022d","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["The **BEAT** | June 2025\n\n4\\.29%\n\n4\\.06%\n\n3\\.80%\n\n3\\.62% 3.55%\n\n4\\.49%\n\n4\\.23%\n\n3\\.95%\n\n3\\.74%\n\n3\\.61%\n\n0\\.47%\n\n0\\.58%\n\n0\\.74% 0.83% 0.86%\n\n2\\.48%\n\n1\\.95%\n\n1\\.66% 1.72%\n\n1\\.85%\n\n0\\.0%\n\n1\\.0%\n\n2\\.0%\n\n3\\.0%\n\n4\\.0%\n\n5\\.0%\n\n3M 6M 1Y 2Y 3Y\n\n**BOE**\n\n**U.S. Federal Reserve**\n\n**ECB**\n\n**BOJ**\n\n4\\.27%\n\n3\\.98%\n\n3\\.54%\n\n3\\.30%\n\n3\\.50%\n\n4\\.11%\n\n3\\.92%\n\n3\\.68% 3.66% 3.79%\n\n0\\.55%\n\n0\\.66%\n\n0\\.82%\n\n1\\.05%\n\n0\\.99%\n\n1\\.91%\n\n1\\.76% 1.67%\n\n1\\.88%\n\n2\\.17%\n\n0\\.0%\n\n1\\.0%\n\n2\\.0%\n\n3\\.0%\n\n4\\.0%\n\n5\\.0%\n\n3M 6M 1Y 2Y 3Y\n\n**BOE**\n\n**U.S. Federal Reserve**\n\n**ECB**\n\n**BOJ**\n\n(1%)\n\n0%\n\n1%\n\n2%\n\n3%\n\n4%\n\n5%\n\n6%\n\n7%\n\n8%\n\n9%\n\n10%\n\n'98 '00 '02 '04 '06 '08 '10 '12 '14 '16 '18 '20 '22 '24\n\n**BOE U.S. Federal Reserve**\n\n**ECB**\n\n**BOJ**\n\n# Monetary Policy\n\nSource: Bloomberg, Factset as of 5/31/25. Data provided is for informational use only. See end of report for important additional information. Forecasts/estimates are based on current market conditions, subject to change, and may not necessarily come to pass.\n\n**Current 1-Mo. Ago 12-Mo. Ago**\n\n**U.S. Federal Reserve** 4\\.50% 4.50% 5.50%\n\n**BOE** 4\\.25% 4.50% 5.25%\n\n**BOJ** 0\\.50% 0.50% 0.10%\n\n**ECB** 2\\.40% 2.40% 4.50%\n\n### **Market Expectations for Future Central Bank Rates Central Bank Policy Rates**\n\n**BONDS**\n\n22"],"full_content":"The **BEAT** | June 2025\n\n4\\.29%\n\n4\\.06%\n\n3\\.80%\n\n3\\.62% 3.55%\n\n4\\.49%\n\n4\\.23%\n\n3\\.95%\n\n3\\.74%\n\n3\\.61%\n\n0\\.47%\n\n0\\.58%\n\n0\\.74% 0.83% 0.86%\n\n2\\.48%\n\n1\\.95%\n\n1\\.66% 1.72%\n\n1\\.85%\n\n0\\.0%\n\n1\\.0%\n\n2\\.0%\n\n3\\.0%\n\n4\\.0%\n\n5\\.0%\n\n3M 6M 1Y 2Y 3Y\n\n**BOE**\n\n**U.S. Federal Reserve**\n\n**ECB**\n\n**BOJ**\n\n4\\.27%\n\n3\\.98%\n\n3\\.54%\n\n3\\.30%\n\n3\\.50%\n\n4\\.11%\n\n3\\.92%\n\n3\\.68% 3.66% 3.79%\n\n0\\.55%\n\n0\\.66%\n\n0\\.82%\n\n1\\.05%\n\n0\\.99%\n\n1\\.91%\n\n1\\.76% 1.67%\n\n1\\.88%\n\n2\\.17%\n\n0\\.0%\n\n1\\.0%\n\n2\\.0%\n\n3\\.0%\n\n4\\.0%\n\n5\\.0%\n\n3M 6M 1Y 2Y 3Y\n\n**BOE**\n\n**U.S. Federal Reserve**\n\n**ECB**\n\n**BOJ**\n\n(1%)\n\n0%\n\n1%\n\n2%\n\n3%\n\n4%\n\n5%\n\n6%\n\n7%\n\n8%\n\n9%\n\n10%\n\n'98 '00 '02 '04 '06 '08 '10 '12 '14 '16 '18 '20 '22 '24\n\n**BOE U.S. Federal Reserve**\n\n**ECB**\n\n**BOJ**\n\n# Monetary Policy\n\nSource: Bloomberg, Factset as of 5/31/25. Data provided is for informational use only. See end of report for important additional information. Forecasts/estimates are based on current market conditions, subject to change, and may not necessarily come to pass.\n\n**Current 1-Mo. Ago 12-Mo. Ago**\n\n**U.S. Federal Reserve** 4\\.50% 4.50% 5.50%\n\n**BOE** 4\\.25% 4.50% 5.25%\n\n**BOJ** 0\\.50% 0.50% 0.10%\n\n**ECB** 2\\.40% 2.40% 4.50%\n\n### **Market Expectations for Future Central Bank Rates Central Bank Policy Rates**\n\n**BONDS**\n\n22","publish_date":null,"title":"Central Bank Policy Rates","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/article_thebeatjun2025_p22.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/article_thebeatjun2025_p22.pdf"},"request":{"example_id":"chart/article_thebeatjun2025_p22","product_type":"parse","source_file_path":"article_thebeatjun2025_p22.pdf"},"started_at":"2026-11-04T19:46:30.303064"} +{"completed_at":"2026-11-04T19:46:32.974469","latency_in_ms":1133,"output":{"example_id":"chart/article_thebeatjun2025_p33","grounded_pages":[],"layout_pages":[],"markdown":"The **BEAT** | June 2025\n\n# Dividend Yields and Volatility Analysis\n\n**EQUITIES**\n\n**Past performance is no guarantee of future results.** It is not possible to invest directly in an index. Source: FactSet and Ned Davis Research as of 5/31/25. Correlation of S&P 500 Stocks is measured by the median 63-day rolling correlation of one day returns data provided is for informational use only. See end of report for important additional information.\n\n33\n\n### **Historical Yields Current Yields**\n\n### **Correlation of S&P 500 Stocks CBOE Market Volatility Index (VIX)**\n\n1\\.0%\n\n1\\.5%\n\n2\\.0%\n\n2\\.5%\n\n3\\.0%\n\n3\\.5%\n\n4\\.0%\n\n'20 '21 '22 '23 '24 '25\n\n**S&P 500**\n\n**MSCI EAFE**\n\n**MSCI EM**\n\n**MSCI World**\n\n1\\.25\n\n1\\.78\n\n2\\.65 2.97\n\n4\\.39\n\n0\\.0%\n\n1\\.0%\n\n2\\.0%\n\n3\\.0%\n\n4\\.0%\n\n5\\.0%\n\nS&P 500 MSCI World MSCI EM MSCI EAFE 10 Yr Treasury\n\n0\\.0\n\n10\\.0\n\n20\\.0\n\n30\\.0\n\n40\\.0\n\n'20 '21 '22 '23 '24 '25\n\n**VIX**\n\n**Average**\n\n0\\.0\n\n0\\.2\n\n0\\.4\n\n0\\.6\n\n0\\.8\n\n1\\.0\n\n'02 '04 '06 '08 '10 '12 '14 '16 '18 '20 '22 '24\n\n**Average**","pages":[{"line_numbers":[],"markdown":"The **BEAT** | June 2025\n\n# Dividend Yields and Volatility Analysis\n\n**EQUITIES**\n\n**Past performance is no guarantee of future results.** It is not possible to invest directly in an index. Source: FactSet and Ned Davis Research as of 5/31/25. Correlation of S&P 500 Stocks is measured by the median 63-day rolling correlation of one day returns data provided is for informational use only. See end of report for important additional information.\n\n33\n\n### **Historical Yields Current Yields**\n\n### **Correlation of S&P 500 Stocks CBOE Market Volatility Index (VIX)**\n\n1\\.0%\n\n1\\.5%\n\n2\\.0%\n\n2\\.5%\n\n3\\.0%\n\n3\\.5%\n\n4\\.0%\n\n'20 '21 '22 '23 '24 '25\n\n**S&P 500**\n\n**MSCI EAFE**\n\n**MSCI EM**\n\n**MSCI World**\n\n1\\.25\n\n1\\.78\n\n2\\.65 2.97\n\n4\\.39\n\n0\\.0%\n\n1\\.0%\n\n2\\.0%\n\n3\\.0%\n\n4\\.0%\n\n5\\.0%\n\nS&P 500 MSCI World MSCI EM MSCI EAFE 10 Yr Treasury\n\n0\\.0\n\n10\\.0\n\n20\\.0\n\n30\\.0\n\n40\\.0\n\n'20 '21 '22 '23 '24 '25\n\n**VIX**\n\n**Average**\n\n0\\.0\n\n0\\.2\n\n0\\.4\n\n0\\.6\n\n0\\.8\n\n1\\.0\n\n'02 '04 '06 '08 '10 '12 '14 '16 '18 '20 '22 '24\n\n**Average**","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"The **BEAT** | June 2025\n\n# Dividend Yields and Volatility Analysis\n\n**EQUITIES**\n\n**Past performance is no guarantee of future results.** It is not possible to invest directly in an index. Source: FactSet and Ned Davis Research as of 5/31/25. Correlation of S&P 500 Stocks is measured by the median 63-day rolling correlation of one day returns data provided is for informational use only. See end of report for important additional information.\n\n33\n\n### **Historical Yields Current Yields**\n\n### **Correlation of S&P 500 Stocks CBOE Market Volatility Index (VIX)**\n\n1\\.0%\n\n1\\.5%\n\n2\\.0%\n\n2\\.5%\n\n3\\.0%\n\n3\\.5%\n\n4\\.0%\n\n'20 '21 '22 '23 '24 '25\n\n**S&P 500**\n\n**MSCI EAFE**\n\n**MSCI EM**\n\n**MSCI World**\n\n1\\.25\n\n1\\.78\n\n2\\.65 2.97\n\n4\\.39\n\n0\\.0%\n\n1\\.0%\n\n2\\.0%\n\n3\\.0%\n\n4\\.0%\n\n5\\.0%\n\nS&P 500 MSCI World MSCI EM MSCI EAFE 10 Yr Treasury\n\n0\\.0\n\n10\\.0\n\n20\\.0\n\n30\\.0\n\n40\\.0\n\n'20 '21 '22 '23 '24 '25\n\n**VIX**\n\n**Average**\n\n0\\.0\n\n0\\.2\n\n0\\.4\n\n0\\.6\n\n0\\.8\n\n1\\.0\n\n'02 '04 '06 '08 '10 '12 '14 '16 '18 '20 '22 '24\n\n**Average**","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/article_thebeatjun2025_p33.pdf"]},"response":{"errors":[],"extract_id":"extract_af6aa53c783e0e4ea63d88ac00f72cc3","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["The **BEAT** | June 2025\n\n# Dividend Yields and Volatility Analysis\n\n**EQUITIES**\n\n**Past performance is no guarantee of future results.** It is not possible to invest directly in an index. Source: FactSet and Ned Davis Research as of 5/31/25. Correlation of S&P 500 Stocks is measured by the median 63-day rolling correlation of one day returns data provided is for informational use only. See end of report for important additional information.\n\n33\n\n### **Historical Yields Current Yields**\n\n### **Correlation of S&P 500 Stocks CBOE Market Volatility Index (VIX)**\n\n1\\.0%\n\n1\\.5%\n\n2\\.0%\n\n2\\.5%\n\n3\\.0%\n\n3\\.5%\n\n4\\.0%\n\n'20 '21 '22 '23 '24 '25\n\n**S&P 500**\n\n**MSCI EAFE**\n\n**MSCI EM**\n\n**MSCI World**\n\n1\\.25\n\n1\\.78\n\n2\\.65 2.97\n\n4\\.39\n\n0\\.0%\n\n1\\.0%\n\n2\\.0%\n\n3\\.0%\n\n4\\.0%\n\n5\\.0%\n\nS&P 500 MSCI World MSCI EM MSCI EAFE 10 Yr Treasury\n\n0\\.0\n\n10\\.0\n\n20\\.0\n\n30\\.0\n\n40\\.0\n\n'20 '21 '22 '23 '24 '25\n\n**VIX**\n\n**Average**\n\n0\\.0\n\n0\\.2\n\n0\\.4\n\n0\\.6\n\n0\\.8\n\n1\\.0\n\n'02 '04 '06 '08 '10 '12 '14 '16 '18 '20 '22 '24\n\n**Average**"],"full_content":"The **BEAT** | June 2025\n\n# Dividend Yields and Volatility Analysis\n\n**EQUITIES**\n\n**Past performance is no guarantee of future results.** It is not possible to invest directly in an index. Source: FactSet and Ned Davis Research as of 5/31/25. Correlation of S&P 500 Stocks is measured by the median 63-day rolling correlation of one day returns data provided is for informational use only. See end of report for important additional information.\n\n33\n\n### **Historical Yields Current Yields**\n\n### **Correlation of S&P 500 Stocks CBOE Market Volatility Index (VIX)**\n\n1\\.0%\n\n1\\.5%\n\n2\\.0%\n\n2\\.5%\n\n3\\.0%\n\n3\\.5%\n\n4\\.0%\n\n'20 '21 '22 '23 '24 '25\n\n**S&P 500**\n\n**MSCI EAFE**\n\n**MSCI EM**\n\n**MSCI World**\n\n1\\.25\n\n1\\.78\n\n2\\.65 2.97\n\n4\\.39\n\n0\\.0%\n\n1\\.0%\n\n2\\.0%\n\n3\\.0%\n\n4\\.0%\n\n5\\.0%\n\nS&P 500 MSCI World MSCI EM MSCI EAFE 10 Yr Treasury\n\n0\\.0\n\n10\\.0\n\n20\\.0\n\n30\\.0\n\n40\\.0\n\n'20 '21 '22 '23 '24 '25\n\n**VIX**\n\n**Average**\n\n0\\.0\n\n0\\.2\n\n0\\.4\n\n0\\.6\n\n0\\.8\n\n1\\.0\n\n'02 '04 '06 '08 '10 '12 '14 '16 '18 '20 '22 '24\n\n**Average**","publish_date":null,"title":"Dividend Yields and Volatility Analysis","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/article_thebeatjun2025_p33.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/article_thebeatjun2025_p33.pdf"},"request":{"example_id":"chart/article_thebeatjun2025_p33","product_type":"parse","source_file_path":"article_thebeatjun2025_p33.pdf"},"started_at":"2026-11-04T19:46:31.841164"} +{"completed_at":"2026-11-04T19:46:34.653883","latency_in_ms":1096,"output":{"example_id":"chart/article_thebeatjun2025_p34","grounded_pages":[],"layout_pages":[],"markdown":"The **BEAT** | June 2025\n\n# Valuation Analysis\n\n**EQUITIES**\n\nSource: FactSet as of 5/31/25. NTM P/E is market price per share divided by expected earnings per share over the next twelve months. Data provided is for informational use only. See end of report for important additional information. Forecasts/estimates are based on current market conditions, subject to change, and may not necessarily come to pass.\n\n### **Regions/Styles: Current NTM P/E vs. 10-Year High, Low, Average**\n\n34\n\n**S&P 500 Russell 2000**\n\n**Russell 1000**\n\n**Growth**\n\n**Russell 1000**\n\n**Value MSCI World**\n\n**MSCI World ex USA Small Cap MSCI EAFE MSCI EM MSCI Europe**\n\n**MSCI AC Asia Pac**\n\n**High Date** 8/20 6/20 8/20 3/21 8/20 8/20 8/20 1/21 6/20 1/21\n\n**Low Date** 12/18 9/22 2/16 9/22 12/18 9/22 9/22 10/18 9/22 12/18\n\n**High**\n\n**Current**\n\n**Average**\n\n**Low**\n\n**MSCI World**\n\n**Current P/E 18.4**\n\n**23\\.9 23.5**\n\n**15\\.2 17.0 15.1 14.2 12.1 14.1 13.6**\n\n**22\\.9**\n\n**63\\.5**\n\n**32\\.7**\n\n**17\\.6**\n\n**21\\.2 19.7**\n\n**17\\.6 15.7**\n\n**17\\.4 17.4**\n\n**14\\.4 15.2 16.8**\n\n**12\\.1 13.5 11.3 11.0 10.1 10.5 11.3**\n\n**21\\.3**\n\n**23\\.3**\n\n**27\\.5**\n\n**16\\.6**\n\n**19\\.1**\n\n**13\\.5 14.6**\n\n**12\\.1**\n\n**14\\.4 13.9**\n\n0\\.0\n\n10\\.0\n\n20\\.0\n\n30\\.0\n\n40\\.0\n\n50\\.0\n\n60\\.0\n\n70\\.0","pages":[{"line_numbers":[],"markdown":"The **BEAT** | June 2025\n\n# Valuation Analysis\n\n**EQUITIES**\n\nSource: FactSet as of 5/31/25. NTM P/E is market price per share divided by expected earnings per share over the next twelve months. Data provided is for informational use only. See end of report for important additional information. Forecasts/estimates are based on current market conditions, subject to change, and may not necessarily come to pass.\n\n### **Regions/Styles: Current NTM P/E vs. 10-Year High, Low, Average**\n\n34\n\n**S&P 500 Russell 2000**\n\n**Russell 1000**\n\n**Growth**\n\n**Russell 1000**\n\n**Value MSCI World**\n\n**MSCI World ex USA Small Cap MSCI EAFE MSCI EM MSCI Europe**\n\n**MSCI AC Asia Pac**\n\n**High Date** 8/20 6/20 8/20 3/21 8/20 8/20 8/20 1/21 6/20 1/21\n\n**Low Date** 12/18 9/22 2/16 9/22 12/18 9/22 9/22 10/18 9/22 12/18\n\n**High**\n\n**Current**\n\n**Average**\n\n**Low**\n\n**MSCI World**\n\n**Current P/E 18.4**\n\n**23\\.9 23.5**\n\n**15\\.2 17.0 15.1 14.2 12.1 14.1 13.6**\n\n**22\\.9**\n\n**63\\.5**\n\n**32\\.7**\n\n**17\\.6**\n\n**21\\.2 19.7**\n\n**17\\.6 15.7**\n\n**17\\.4 17.4**\n\n**14\\.4 15.2 16.8**\n\n**12\\.1 13.5 11.3 11.0 10.1 10.5 11.3**\n\n**21\\.3**\n\n**23\\.3**\n\n**27\\.5**\n\n**16\\.6**\n\n**19\\.1**\n\n**13\\.5 14.6**\n\n**12\\.1**\n\n**14\\.4 13.9**\n\n0\\.0\n\n10\\.0\n\n20\\.0\n\n30\\.0\n\n40\\.0\n\n50\\.0\n\n60\\.0\n\n70\\.0","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"The **BEAT** | June 2025\n\n# Valuation Analysis\n\n**EQUITIES**\n\nSource: FactSet as of 5/31/25. NTM P/E is market price per share divided by expected earnings per share over the next twelve months. Data provided is for informational use only. See end of report for important additional information. Forecasts/estimates are based on current market conditions, subject to change, and may not necessarily come to pass.\n\n### **Regions/Styles: Current NTM P/E vs. 10-Year High, Low, Average**\n\n34\n\n**S&P 500 Russell 2000**\n\n**Russell 1000**\n\n**Growth**\n\n**Russell 1000**\n\n**Value MSCI World**\n\n**MSCI World ex USA Small Cap MSCI EAFE MSCI EM MSCI Europe**\n\n**MSCI AC Asia Pac**\n\n**High Date** 8/20 6/20 8/20 3/21 8/20 8/20 8/20 1/21 6/20 1/21\n\n**Low Date** 12/18 9/22 2/16 9/22 12/18 9/22 9/22 10/18 9/22 12/18\n\n**High**\n\n**Current**\n\n**Average**\n\n**Low**\n\n**MSCI World**\n\n**Current P/E 18.4**\n\n**23\\.9 23.5**\n\n**15\\.2 17.0 15.1 14.2 12.1 14.1 13.6**\n\n**22\\.9**\n\n**63\\.5**\n\n**32\\.7**\n\n**17\\.6**\n\n**21\\.2 19.7**\n\n**17\\.6 15.7**\n\n**17\\.4 17.4**\n\n**14\\.4 15.2 16.8**\n\n**12\\.1 13.5 11.3 11.0 10.1 10.5 11.3**\n\n**21\\.3**\n\n**23\\.3**\n\n**27\\.5**\n\n**16\\.6**\n\n**19\\.1**\n\n**13\\.5 14.6**\n\n**12\\.1**\n\n**14\\.4 13.9**\n\n0\\.0\n\n10\\.0\n\n20\\.0\n\n30\\.0\n\n40\\.0\n\n50\\.0\n\n60\\.0\n\n70\\.0","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/article_thebeatjun2025_p34.pdf"]},"response":{"errors":[],"extract_id":"extract_5550020a3d02d67547ca40ab98469192","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["The **BEAT** | June 2025\n\n# Valuation Analysis\n\n**EQUITIES**\n\nSource: FactSet as of 5/31/25. NTM P/E is market price per share divided by expected earnings per share over the next twelve months. Data provided is for informational use only. See end of report for important additional information. Forecasts/estimates are based on current market conditions, subject to change, and may not necessarily come to pass.\n\n### **Regions/Styles: Current NTM P/E vs. 10-Year High, Low, Average**\n\n34\n\n**S&P 500 Russell 2000**\n\n**Russell 1000**\n\n**Growth**\n\n**Russell 1000**\n\n**Value MSCI World**\n\n**MSCI World ex USA Small Cap MSCI EAFE MSCI EM MSCI Europe**\n\n**MSCI AC Asia Pac**\n\n**High Date** 8/20 6/20 8/20 3/21 8/20 8/20 8/20 1/21 6/20 1/21\n\n**Low Date** 12/18 9/22 2/16 9/22 12/18 9/22 9/22 10/18 9/22 12/18\n\n**High**\n\n**Current**\n\n**Average**\n\n**Low**\n\n**MSCI World**\n\n**Current P/E 18.4**\n\n**23\\.9 23.5**\n\n**15\\.2 17.0 15.1 14.2 12.1 14.1 13.6**\n\n**22\\.9**\n\n**63\\.5**\n\n**32\\.7**\n\n**17\\.6**\n\n**21\\.2 19.7**\n\n**17\\.6 15.7**\n\n**17\\.4 17.4**\n\n**14\\.4 15.2 16.8**\n\n**12\\.1 13.5 11.3 11.0 10.1 10.5 11.3**\n\n**21\\.3**\n\n**23\\.3**\n\n**27\\.5**\n\n**16\\.6**\n\n**19\\.1**\n\n**13\\.5 14.6**\n\n**12\\.1**\n\n**14\\.4 13.9**\n\n0\\.0\n\n10\\.0\n\n20\\.0\n\n30\\.0\n\n40\\.0\n\n50\\.0\n\n60\\.0\n\n70\\.0"],"full_content":"The **BEAT** | June 2025\n\n# Valuation Analysis\n\n**EQUITIES**\n\nSource: FactSet as of 5/31/25. NTM P/E is market price per share divided by expected earnings per share over the next twelve months. Data provided is for informational use only. See end of report for important additional information. Forecasts/estimates are based on current market conditions, subject to change, and may not necessarily come to pass.\n\n### **Regions/Styles: Current NTM P/E vs. 10-Year High, Low, Average**\n\n34\n\n**S&P 500 Russell 2000**\n\n**Russell 1000**\n\n**Growth**\n\n**Russell 1000**\n\n**Value MSCI World**\n\n**MSCI World ex USA Small Cap MSCI EAFE MSCI EM MSCI Europe**\n\n**MSCI AC Asia Pac**\n\n**High Date** 8/20 6/20 8/20 3/21 8/20 8/20 8/20 1/21 6/20 1/21\n\n**Low Date** 12/18 9/22 2/16 9/22 12/18 9/22 9/22 10/18 9/22 12/18\n\n**High**\n\n**Current**\n\n**Average**\n\n**Low**\n\n**MSCI World**\n\n**Current P/E 18.4**\n\n**23\\.9 23.5**\n\n**15\\.2 17.0 15.1 14.2 12.1 14.1 13.6**\n\n**22\\.9**\n\n**63\\.5**\n\n**32\\.7**\n\n**17\\.6**\n\n**21\\.2 19.7**\n\n**17\\.6 15.7**\n\n**17\\.4 17.4**\n\n**14\\.4 15.2 16.8**\n\n**12\\.1 13.5 11.3 11.0 10.1 10.5 11.3**\n\n**21\\.3**\n\n**23\\.3**\n\n**27\\.5**\n\n**16\\.6**\n\n**19\\.1**\n\n**13\\.5 14.6**\n\n**12\\.1**\n\n**14\\.4 13.9**\n\n0\\.0\n\n10\\.0\n\n20\\.0\n\n30\\.0\n\n40\\.0\n\n50\\.0\n\n60\\.0\n\n70\\.0","publish_date":null,"title":"Valuation Analysis","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/article_thebeatjun2025_p34.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/article_thebeatjun2025_p34.pdf"},"request":{"example_id":"chart/article_thebeatjun2025_p34","product_type":"parse","source_file_path":"article_thebeatjun2025_p34.pdf"},"started_at":"2026-11-04T19:46:33.557756"} +{"completed_at":"2026-11-04T19:46:37.610754","latency_in_ms":2622,"output":{"example_id":"chart/article_thebeatjun2025_p35","grounded_pages":[],"layout_pages":[],"markdown":"The **BEAT** | June 2025\n\n**16\\.6**\n\n**24\\.2**\n\n**19\\.8**\n\n**15\\.2 13.5 16.3 18.8 21.7 17.3 17.8**\n\n**24\\.0**\n\n**45\\.3**\n\n**22\\.6**\n\n**77\\.2**\n\n**17\\.7 19.9**\n\n**26\\.4**\n\n**30\\.4**\n\n**22\\.1 21.3**\n\n**10\\.0**\n\n**16\\.9 16.4**\n\n**\\-263.8**\n\n**\\*Not to scale**\n\n**9\\.3**\n\n**13\\.8 13.6 14.7 12.2 14.7**\n\n**18\\.9**\n\n**27\\.9**\n\n**22\\.6**\n\n**14\\.5 16.7 16.0**\n\n**23\\.7 27.1**\n\n**20\\.1 18.1**\n\n(30.0)\n\n(10.0)\n\n10\\.0\n\n30\\.0\n\n50\\.0\n\n70\\.0\n\n# Valuation Analysis\n\n**EQUITIES**\n\nSource: FactSet as of 5/31/25. NTM P/E is market price per share divided by expected earnings per share over the next twelve months. The Real Estate sector is excluded from this 10- year chart since the sector was created on August 31, 2016. Data provided is for informational use only. See end of report for important additional information. Forecasts/estimates are based on current market conditions, subject to change, and may not necessarily come to pass.\n\n### **S&P 500 Sectors: Current NTM P/E vs. 10-Year High, Low, Average**\n\n35\n\n**High**\n\n**Current**\n\n**Average**\n\n**Low**\n\n**S&P 500**\n\n**Current P/E**\n\n**Communication**\n\n**Services**\n\n**Consumer**\n\n**Discretionary**\n\n**Consumer**\n\n**Staples Energy Financials Health Care Industrials**\n\n**Information Technology Materials Utilities**\n\n**High Date** 8/20 6/20 5/25 4/16 11/24 8/24 8/20 6/24 7/20 3/22\n\n**Low Date** 5/18 2/16 4/18 7/20 3/20 3/20 12/18 12/18 9/15 6/15","pages":[{"line_numbers":[],"markdown":"The **BEAT** | June 2025\n\n**16\\.6**\n\n**24\\.2**\n\n**19\\.8**\n\n**15\\.2 13.5 16.3 18.8 21.7 17.3 17.8**\n\n**24\\.0**\n\n**45\\.3**\n\n**22\\.6**\n\n**77\\.2**\n\n**17\\.7 19.9**\n\n**26\\.4**\n\n**30\\.4**\n\n**22\\.1 21.3**\n\n**10\\.0**\n\n**16\\.9 16.4**\n\n**\\-263.8**\n\n**\\*Not to scale**\n\n**9\\.3**\n\n**13\\.8 13.6 14.7 12.2 14.7**\n\n**18\\.9**\n\n**27\\.9**\n\n**22\\.6**\n\n**14\\.5 16.7 16.0**\n\n**23\\.7 27.1**\n\n**20\\.1 18.1**\n\n(30.0)\n\n(10.0)\n\n10\\.0\n\n30\\.0\n\n50\\.0\n\n70\\.0\n\n# Valuation Analysis\n\n**EQUITIES**\n\nSource: FactSet as of 5/31/25. NTM P/E is market price per share divided by expected earnings per share over the next twelve months. The Real Estate sector is excluded from this 10- year chart since the sector was created on August 31, 2016. Data provided is for informational use only. See end of report for important additional information. Forecasts/estimates are based on current market conditions, subject to change, and may not necessarily come to pass.\n\n### **S&P 500 Sectors: Current NTM P/E vs. 10-Year High, Low, Average**\n\n35\n\n**High**\n\n**Current**\n\n**Average**\n\n**Low**\n\n**S&P 500**\n\n**Current P/E**\n\n**Communication**\n\n**Services**\n\n**Consumer**\n\n**Discretionary**\n\n**Consumer**\n\n**Staples Energy Financials Health Care Industrials**\n\n**Information Technology Materials Utilities**\n\n**High Date** 8/20 6/20 5/25 4/16 11/24 8/24 8/20 6/24 7/20 3/22\n\n**Low Date** 5/18 2/16 4/18 7/20 3/20 3/20 12/18 12/18 9/15 6/15","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"The **BEAT** | June 2025\n\n**16\\.6**\n\n**24\\.2**\n\n**19\\.8**\n\n**15\\.2 13.5 16.3 18.8 21.7 17.3 17.8**\n\n**24\\.0**\n\n**45\\.3**\n\n**22\\.6**\n\n**77\\.2**\n\n**17\\.7 19.9**\n\n**26\\.4**\n\n**30\\.4**\n\n**22\\.1 21.3**\n\n**10\\.0**\n\n**16\\.9 16.4**\n\n**\\-263.8**\n\n**\\*Not to scale**\n\n**9\\.3**\n\n**13\\.8 13.6 14.7 12.2 14.7**\n\n**18\\.9**\n\n**27\\.9**\n\n**22\\.6**\n\n**14\\.5 16.7 16.0**\n\n**23\\.7 27.1**\n\n**20\\.1 18.1**\n\n(30.0)\n\n(10.0)\n\n10\\.0\n\n30\\.0\n\n50\\.0\n\n70\\.0\n\n# Valuation Analysis\n\n**EQUITIES**\n\nSource: FactSet as of 5/31/25. NTM P/E is market price per share divided by expected earnings per share over the next twelve months. The Real Estate sector is excluded from this 10- year chart since the sector was created on August 31, 2016. Data provided is for informational use only. See end of report for important additional information. Forecasts/estimates are based on current market conditions, subject to change, and may not necessarily come to pass.\n\n### **S&P 500 Sectors: Current NTM P/E vs. 10-Year High, Low, Average**\n\n35\n\n**High**\n\n**Current**\n\n**Average**\n\n**Low**\n\n**S&P 500**\n\n**Current P/E**\n\n**Communication**\n\n**Services**\n\n**Consumer**\n\n**Discretionary**\n\n**Consumer**\n\n**Staples Energy Financials Health Care Industrials**\n\n**Information Technology Materials Utilities**\n\n**High Date** 8/20 6/20 5/25 4/16 11/24 8/24 8/20 6/24 7/20 3/22\n\n**Low Date** 5/18 2/16 4/18 7/20 3/20 3/20 12/18 12/18 9/15 6/15","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/article_thebeatjun2025_p35.pdf"]},"response":{"errors":[],"extract_id":"extract_50223542f1b3b8a09f259b3c8e3d53b3","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["The **BEAT** | June 2025\n\n**16\\.6**\n\n**24\\.2**\n\n**19\\.8**\n\n**15\\.2 13.5 16.3 18.8 21.7 17.3 17.8**\n\n**24\\.0**\n\n**45\\.3**\n\n**22\\.6**\n\n**77\\.2**\n\n**17\\.7 19.9**\n\n**26\\.4**\n\n**30\\.4**\n\n**22\\.1 21.3**\n\n**10\\.0**\n\n**16\\.9 16.4**\n\n**\\-263.8**\n\n**\\*Not to scale**\n\n**9\\.3**\n\n**13\\.8 13.6 14.7 12.2 14.7**\n\n**18\\.9**\n\n**27\\.9**\n\n**22\\.6**\n\n**14\\.5 16.7 16.0**\n\n**23\\.7 27.1**\n\n**20\\.1 18.1**\n\n(30.0)\n\n(10.0)\n\n10\\.0\n\n30\\.0\n\n50\\.0\n\n70\\.0\n\n# Valuation Analysis\n\n**EQUITIES**\n\nSource: FactSet as of 5/31/25. NTM P/E is market price per share divided by expected earnings per share over the next twelve months. The Real Estate sector is excluded from this 10- year chart since the sector was created on August 31, 2016. Data provided is for informational use only. See end of report for important additional information. Forecasts/estimates are based on current market conditions, subject to change, and may not necessarily come to pass.\n\n### **S&P 500 Sectors: Current NTM P/E vs. 10-Year High, Low, Average**\n\n35\n\n**High**\n\n**Current**\n\n**Average**\n\n**Low**\n\n**S&P 500**\n\n**Current P/E**\n\n**Communication**\n\n**Services**\n\n**Consumer**\n\n**Discretionary**\n\n**Consumer**\n\n**Staples Energy Financials Health Care Industrials**\n\n**Information Technology Materials Utilities**\n\n**High Date** 8/20 6/20 5/25 4/16 11/24 8/24 8/20 6/24 7/20 3/22\n\n**Low Date** 5/18 2/16 4/18 7/20 3/20 3/20 12/18 12/18 9/15 6/15"],"full_content":"The **BEAT** | June 2025\n\n**16\\.6**\n\n**24\\.2**\n\n**19\\.8**\n\n**15\\.2 13.5 16.3 18.8 21.7 17.3 17.8**\n\n**24\\.0**\n\n**45\\.3**\n\n**22\\.6**\n\n**77\\.2**\n\n**17\\.7 19.9**\n\n**26\\.4**\n\n**30\\.4**\n\n**22\\.1 21.3**\n\n**10\\.0**\n\n**16\\.9 16.4**\n\n**\\-263.8**\n\n**\\*Not to scale**\n\n**9\\.3**\n\n**13\\.8 13.6 14.7 12.2 14.7**\n\n**18\\.9**\n\n**27\\.9**\n\n**22\\.6**\n\n**14\\.5 16.7 16.0**\n\n**23\\.7 27.1**\n\n**20\\.1 18.1**\n\n(30.0)\n\n(10.0)\n\n10\\.0\n\n30\\.0\n\n50\\.0\n\n70\\.0\n\n# Valuation Analysis\n\n**EQUITIES**\n\nSource: FactSet as of 5/31/25. NTM P/E is market price per share divided by expected earnings per share over the next twelve months. The Real Estate sector is excluded from this 10- year chart since the sector was created on August 31, 2016. Data provided is for informational use only. See end of report for important additional information. Forecasts/estimates are based on current market conditions, subject to change, and may not necessarily come to pass.\n\n### **S&P 500 Sectors: Current NTM P/E vs. 10-Year High, Low, Average**\n\n35\n\n**High**\n\n**Current**\n\n**Average**\n\n**Low**\n\n**S&P 500**\n\n**Current P/E**\n\n**Communication**\n\n**Services**\n\n**Consumer**\n\n**Discretionary**\n\n**Consumer**\n\n**Staples Energy Financials Health Care Industrials**\n\n**Information Technology Materials Utilities**\n\n**High Date** 8/20 6/20 5/25 4/16 11/24 8/24 8/20 6/24 7/20 3/22\n\n**Low Date** 5/18 2/16 4/18 7/20 3/20 3/20 12/18 12/18 9/15 6/15","publish_date":null,"title":"Valuation Analysis","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/article_thebeatjun2025_p35.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/article_thebeatjun2025_p35.pdf"},"request":{"example_id":"chart/article_thebeatjun2025_p35","product_type":"parse","source_file_path":"article_thebeatjun2025_p35.pdf"},"started_at":"2026-11-04T19:46:34.988588"} +{"completed_at":"2026-11-04T19:46:39.590987","latency_in_ms":2632,"output":{"example_id":"chart/article_thebeatjun2025_p36","grounded_pages":[],"layout_pages":[],"markdown":"The **BEAT** | June 2025\n\n# Corporate Earnings Growth\n\n**EQUITIES**\n\nSource: FactSet as of 5/31/25. Expected EPS Growth is defined as the expected % change in the EPS growth from the beginning of the current calendar year though the end of the calendar year. 2024 EPS Growth is defined as the % change in EPS from the beginning of the year through the end of the year. Data provided is for informational use only. See end of report for important additional information. Forecasts/estimates are based on current market conditions, subject to change, and may not necessarily come to pass.\n\n36\n\n### **Regions/Styles**\n\n### **S&P 500 Sectors**\n\n**33\\.37**\n\n**13\\.93 13.18 12.85 9.22 7.53 7.27 5.47 3.57 3.37**\n\n**1\\.31**\n\n**28\\.97**\n\n**7\\.38**\n\n**18\\.22**\n\n**10\\.48 17.31**\n\n**7\\.46 0.15 3.02 0.50**\n\n(10.0)\n\n0\\.0\n\n10\\.0\n\n20\\.0\n\n30\\.0\n\n40\\.0\n\n50\\.0\n\nRussell 2000 Russell 1000\n\nGrowth\n\nMSCI World ex USA Small Cap\n\nMSCI EM S&P 500 MSCI Asia Pac MSCI World Russell 1000\n\nValue\n\nMSCI EAFE MSCI Europe\n\n**Expected EPS Growth**\n\n**2024 EPS Growth**\n\n**17\\.59 15.81 14.52**\n\n**8\\.73 7.82 6.88 6.67 3.64 0.68 0.09**\n\n**\\-13.03**\n\n**18\\.96**\n\n**23\\.85**\n\n**4\\.01 0.04**\n\n**\\-10.24**\n\n**18\\.74**\n\n**12\\.52**\n\n**4\\.13**\n\n**16\\.84**\n\n**2\\.17**\n\n**\\-17.82** (20.0)\n\n(10.0)\n\n0\\.0\n\n10\\.0\n\n20\\.0\n\n30\\.0\n\nInfo. Tech. Communications Health Care Industrials Materials Financials Utilities Real Estate Discretionary Staples Energy\n\n**Expected EPS Growth**\n\n**2024 EPS Growth**","pages":[{"line_numbers":[],"markdown":"The **BEAT** | June 2025\n\n# Corporate Earnings Growth\n\n**EQUITIES**\n\nSource: FactSet as of 5/31/25. Expected EPS Growth is defined as the expected % change in the EPS growth from the beginning of the current calendar year though the end of the calendar year. 2024 EPS Growth is defined as the % change in EPS from the beginning of the year through the end of the year. Data provided is for informational use only. See end of report for important additional information. Forecasts/estimates are based on current market conditions, subject to change, and may not necessarily come to pass.\n\n36\n\n### **Regions/Styles**\n\n### **S&P 500 Sectors**\n\n**33\\.37**\n\n**13\\.93 13.18 12.85 9.22 7.53 7.27 5.47 3.57 3.37**\n\n**1\\.31**\n\n**28\\.97**\n\n**7\\.38**\n\n**18\\.22**\n\n**10\\.48 17.31**\n\n**7\\.46 0.15 3.02 0.50**\n\n(10.0)\n\n0\\.0\n\n10\\.0\n\n20\\.0\n\n30\\.0\n\n40\\.0\n\n50\\.0\n\nRussell 2000 Russell 1000\n\nGrowth\n\nMSCI World ex USA Small Cap\n\nMSCI EM S&P 500 MSCI Asia Pac MSCI World Russell 1000\n\nValue\n\nMSCI EAFE MSCI Europe\n\n**Expected EPS Growth**\n\n**2024 EPS Growth**\n\n**17\\.59 15.81 14.52**\n\n**8\\.73 7.82 6.88 6.67 3.64 0.68 0.09**\n\n**\\-13.03**\n\n**18\\.96**\n\n**23\\.85**\n\n**4\\.01 0.04**\n\n**\\-10.24**\n\n**18\\.74**\n\n**12\\.52**\n\n**4\\.13**\n\n**16\\.84**\n\n**2\\.17**\n\n**\\-17.82** (20.0)\n\n(10.0)\n\n0\\.0\n\n10\\.0\n\n20\\.0\n\n30\\.0\n\nInfo. Tech. Communications Health Care Industrials Materials Financials Utilities Real Estate Discretionary Staples Energy\n\n**Expected EPS Growth**\n\n**2024 EPS Growth**","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"The **BEAT** | June 2025\n\n# Corporate Earnings Growth\n\n**EQUITIES**\n\nSource: FactSet as of 5/31/25. Expected EPS Growth is defined as the expected % change in the EPS growth from the beginning of the current calendar year though the end of the calendar year. 2024 EPS Growth is defined as the % change in EPS from the beginning of the year through the end of the year. Data provided is for informational use only. See end of report for important additional information. Forecasts/estimates are based on current market conditions, subject to change, and may not necessarily come to pass.\n\n36\n\n### **Regions/Styles**\n\n### **S&P 500 Sectors**\n\n**33\\.37**\n\n**13\\.93 13.18 12.85 9.22 7.53 7.27 5.47 3.57 3.37**\n\n**1\\.31**\n\n**28\\.97**\n\n**7\\.38**\n\n**18\\.22**\n\n**10\\.48 17.31**\n\n**7\\.46 0.15 3.02 0.50**\n\n(10.0)\n\n0\\.0\n\n10\\.0\n\n20\\.0\n\n30\\.0\n\n40\\.0\n\n50\\.0\n\nRussell 2000 Russell 1000\n\nGrowth\n\nMSCI World ex USA Small Cap\n\nMSCI EM S&P 500 MSCI Asia Pac MSCI World Russell 1000\n\nValue\n\nMSCI EAFE MSCI Europe\n\n**Expected EPS Growth**\n\n**2024 EPS Growth**\n\n**17\\.59 15.81 14.52**\n\n**8\\.73 7.82 6.88 6.67 3.64 0.68 0.09**\n\n**\\-13.03**\n\n**18\\.96**\n\n**23\\.85**\n\n**4\\.01 0.04**\n\n**\\-10.24**\n\n**18\\.74**\n\n**12\\.52**\n\n**4\\.13**\n\n**16\\.84**\n\n**2\\.17**\n\n**\\-17.82** (20.0)\n\n(10.0)\n\n0\\.0\n\n10\\.0\n\n20\\.0\n\n30\\.0\n\nInfo. Tech. Communications Health Care Industrials Materials Financials Utilities Real Estate Discretionary Staples Energy\n\n**Expected EPS Growth**\n\n**2024 EPS Growth**","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/article_thebeatjun2025_p36.pdf"]},"response":{"errors":[],"extract_id":"extract_0ac08a77d5326adef225842a25d91e88","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["The **BEAT** | June 2025\n\n# Corporate Earnings Growth\n\n**EQUITIES**\n\nSource: FactSet as of 5/31/25. Expected EPS Growth is defined as the expected % change in the EPS growth from the beginning of the current calendar year though the end of the calendar year. 2024 EPS Growth is defined as the % change in EPS from the beginning of the year through the end of the year. Data provided is for informational use only. See end of report for important additional information. Forecasts/estimates are based on current market conditions, subject to change, and may not necessarily come to pass.\n\n36\n\n### **Regions/Styles**\n\n### **S&P 500 Sectors**\n\n**33\\.37**\n\n**13\\.93 13.18 12.85 9.22 7.53 7.27 5.47 3.57 3.37**\n\n**1\\.31**\n\n**28\\.97**\n\n**7\\.38**\n\n**18\\.22**\n\n**10\\.48 17.31**\n\n**7\\.46 0.15 3.02 0.50**\n\n(10.0)\n\n0\\.0\n\n10\\.0\n\n20\\.0\n\n30\\.0\n\n40\\.0\n\n50\\.0\n\nRussell 2000 Russell 1000\n\nGrowth\n\nMSCI World ex USA Small Cap\n\nMSCI EM S&P 500 MSCI Asia Pac MSCI World Russell 1000\n\nValue\n\nMSCI EAFE MSCI Europe\n\n**Expected EPS Growth**\n\n**2024 EPS Growth**\n\n**17\\.59 15.81 14.52**\n\n**8\\.73 7.82 6.88 6.67 3.64 0.68 0.09**\n\n**\\-13.03**\n\n**18\\.96**\n\n**23\\.85**\n\n**4\\.01 0.04**\n\n**\\-10.24**\n\n**18\\.74**\n\n**12\\.52**\n\n**4\\.13**\n\n**16\\.84**\n\n**2\\.17**\n\n**\\-17.82** (20.0)\n\n(10.0)\n\n0\\.0\n\n10\\.0\n\n20\\.0\n\n30\\.0\n\nInfo. Tech. Communications Health Care Industrials Materials Financials Utilities Real Estate Discretionary Staples Energy\n\n**Expected EPS Growth**\n\n**2024 EPS Growth**"],"full_content":"The **BEAT** | June 2025\n\n# Corporate Earnings Growth\n\n**EQUITIES**\n\nSource: FactSet as of 5/31/25. Expected EPS Growth is defined as the expected % change in the EPS growth from the beginning of the current calendar year though the end of the calendar year. 2024 EPS Growth is defined as the % change in EPS from the beginning of the year through the end of the year. Data provided is for informational use only. See end of report for important additional information. Forecasts/estimates are based on current market conditions, subject to change, and may not necessarily come to pass.\n\n36\n\n### **Regions/Styles**\n\n### **S&P 500 Sectors**\n\n**33\\.37**\n\n**13\\.93 13.18 12.85 9.22 7.53 7.27 5.47 3.57 3.37**\n\n**1\\.31**\n\n**28\\.97**\n\n**7\\.38**\n\n**18\\.22**\n\n**10\\.48 17.31**\n\n**7\\.46 0.15 3.02 0.50**\n\n(10.0)\n\n0\\.0\n\n10\\.0\n\n20\\.0\n\n30\\.0\n\n40\\.0\n\n50\\.0\n\nRussell 2000 Russell 1000\n\nGrowth\n\nMSCI World ex USA Small Cap\n\nMSCI EM S&P 500 MSCI Asia Pac MSCI World Russell 1000\n\nValue\n\nMSCI EAFE MSCI Europe\n\n**Expected EPS Growth**\n\n**2024 EPS Growth**\n\n**17\\.59 15.81 14.52**\n\n**8\\.73 7.82 6.88 6.67 3.64 0.68 0.09**\n\n**\\-13.03**\n\n**18\\.96**\n\n**23\\.85**\n\n**4\\.01 0.04**\n\n**\\-10.24**\n\n**18\\.74**\n\n**12\\.52**\n\n**4\\.13**\n\n**16\\.84**\n\n**2\\.17**\n\n**\\-17.82** (20.0)\n\n(10.0)\n\n0\\.0\n\n10\\.0\n\n20\\.0\n\n30\\.0\n\nInfo. Tech. Communications Health Care Industrials Materials Financials Utilities Real Estate Discretionary Staples Energy\n\n**Expected EPS Growth**\n\n**2024 EPS Growth**","publish_date":null,"title":"EQUITIES Corporate Earnings Growth","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/article_thebeatjun2025_p36.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/article_thebeatjun2025_p36.pdf"},"request":{"example_id":"chart/article_thebeatjun2025_p36","product_type":"parse","source_file_path":"article_thebeatjun2025_p36.pdf"},"started_at":"2026-11-04T19:46:36.958667"} +{"completed_at":"2026-11-04T19:46:39.430743","latency_in_ms":1099,"output":{"example_id":"chart/article_thebeatjun2025_p46","grounded_pages":[],"layout_pages":[],"markdown":"The **BEAT** | June 2025\n\n# Local Sovereign Currency Yields\n\n**ALTERNATIVES**\n\n**Past performance is no guarantee of future results.** It is not possible to invest directly in an index. \\*German Rate. Source: Bloomberg as 5/31/25. Data provided is for informational use only. See end of report for important additional information.\n\n46\n\n### **Developed Market Local Interest Rates % (1 Year) Emerging Market Local Interest Rates % (1 Year)**\n\n1\\.05\n\n0\\.36\n\n3\\.29\n\n3\\.42\n\n3\\.34\n\n4\\.56\n\n5\\.02\n\n4\\.29\n\n4\\.78\n\n4\\.33\n\n5\\.11\n\n\\-0.27\n\n0\\.74\n\n1\\.56\n\n1\\.76\n\n1\\.91\n\n2\\.56\n\n3\\.17\n\n3\\.40\n\n3\\.83\n\n4\\.00\n\n4\\.12\n\n\\-5.00 0.00 5.00 10.00 15.00\n\nSwiss Franc (CHF)\n\nJapanese Yen (JPY)\n\nDanish Krone (DKK)\n\nEuro (EUR) \\*\n\nSwedish Krona (SEK)\n\nCanadian Dollar (CAD)\n\nNew Zealand Dollar (NZD)\n\nAustralian Dollar (AUD)\n\nBritish Pound (GBP)\n\nNorwegian Krone (NOK)\n\nU.S. Dollar (USD)\n\nCurrent 1 Yr Prior\n\n2\\.36\n\n3\\.56\n\n5\\.38\n\n4\\.85\n\n4\\.86\n\n6\\.02\n\n6\\.46\n\n6\\.65\n\n9\\.18\n\n11\\.34\n\n9\\.62\n\n10\\.49\n\n20\\.70\n\n40\\.18\n\n1\\.54\n\n2\\.86\n\n3\\.95\n\n4\\.26\n\n5\\.03\n\n5\\.62\n\n6\\.03\n\n6\\.17\n\n7\\.91\n\n8\\.27\n\n8\\.71\n\n14\\.64\n\n22\\.12\n\n42\\.67\n\n0\\.00 10.00 20.00 30.00 40.00 50.00\n\nThai Baht (THB)\n\nMalaysian Ringgit (MYR)\n\nPeruvian New Sol (PEN)\n\nPolish Zloty (PLN)\n\nChilean Peso (CLP)\n\nPhilippine Peso (PHP)\n\nIndonesian Rupiah (IDR)\n\nHungarian Forint (HUF)\n\nSouth African Rand (ZAR)\n\nMexican Peso (MXN)\n\nColombian Peso (COP)\n\nBrazilian Real (BRL)\n\nNigerian Naira (NGN)\n\nTurkish New Lira (TRY)\n\nCurrent 1 Yr Prior","pages":[{"line_numbers":[],"markdown":"The **BEAT** | June 2025\n\n# Local Sovereign Currency Yields\n\n**ALTERNATIVES**\n\n**Past performance is no guarantee of future results.** It is not possible to invest directly in an index. \\*German Rate. Source: Bloomberg as 5/31/25. Data provided is for informational use only. See end of report for important additional information.\n\n46\n\n### **Developed Market Local Interest Rates % (1 Year) Emerging Market Local Interest Rates % (1 Year)**\n\n1\\.05\n\n0\\.36\n\n3\\.29\n\n3\\.42\n\n3\\.34\n\n4\\.56\n\n5\\.02\n\n4\\.29\n\n4\\.78\n\n4\\.33\n\n5\\.11\n\n\\-0.27\n\n0\\.74\n\n1\\.56\n\n1\\.76\n\n1\\.91\n\n2\\.56\n\n3\\.17\n\n3\\.40\n\n3\\.83\n\n4\\.00\n\n4\\.12\n\n\\-5.00 0.00 5.00 10.00 15.00\n\nSwiss Franc (CHF)\n\nJapanese Yen (JPY)\n\nDanish Krone (DKK)\n\nEuro (EUR) \\*\n\nSwedish Krona (SEK)\n\nCanadian Dollar (CAD)\n\nNew Zealand Dollar (NZD)\n\nAustralian Dollar (AUD)\n\nBritish Pound (GBP)\n\nNorwegian Krone (NOK)\n\nU.S. Dollar (USD)\n\nCurrent 1 Yr Prior\n\n2\\.36\n\n3\\.56\n\n5\\.38\n\n4\\.85\n\n4\\.86\n\n6\\.02\n\n6\\.46\n\n6\\.65\n\n9\\.18\n\n11\\.34\n\n9\\.62\n\n10\\.49\n\n20\\.70\n\n40\\.18\n\n1\\.54\n\n2\\.86\n\n3\\.95\n\n4\\.26\n\n5\\.03\n\n5\\.62\n\n6\\.03\n\n6\\.17\n\n7\\.91\n\n8\\.27\n\n8\\.71\n\n14\\.64\n\n22\\.12\n\n42\\.67\n\n0\\.00 10.00 20.00 30.00 40.00 50.00\n\nThai Baht (THB)\n\nMalaysian Ringgit (MYR)\n\nPeruvian New Sol (PEN)\n\nPolish Zloty (PLN)\n\nChilean Peso (CLP)\n\nPhilippine Peso (PHP)\n\nIndonesian Rupiah (IDR)\n\nHungarian Forint (HUF)\n\nSouth African Rand (ZAR)\n\nMexican Peso (MXN)\n\nColombian Peso (COP)\n\nBrazilian Real (BRL)\n\nNigerian Naira (NGN)\n\nTurkish New Lira (TRY)\n\nCurrent 1 Yr Prior","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"The **BEAT** | June 2025\n\n# Local Sovereign Currency Yields\n\n**ALTERNATIVES**\n\n**Past performance is no guarantee of future results.** It is not possible to invest directly in an index. \\*German Rate. Source: Bloomberg as 5/31/25. Data provided is for informational use only. See end of report for important additional information.\n\n46\n\n### **Developed Market Local Interest Rates % (1 Year) Emerging Market Local Interest Rates % (1 Year)**\n\n1\\.05\n\n0\\.36\n\n3\\.29\n\n3\\.42\n\n3\\.34\n\n4\\.56\n\n5\\.02\n\n4\\.29\n\n4\\.78\n\n4\\.33\n\n5\\.11\n\n\\-0.27\n\n0\\.74\n\n1\\.56\n\n1\\.76\n\n1\\.91\n\n2\\.56\n\n3\\.17\n\n3\\.40\n\n3\\.83\n\n4\\.00\n\n4\\.12\n\n\\-5.00 0.00 5.00 10.00 15.00\n\nSwiss Franc (CHF)\n\nJapanese Yen (JPY)\n\nDanish Krone (DKK)\n\nEuro (EUR) \\*\n\nSwedish Krona (SEK)\n\nCanadian Dollar (CAD)\n\nNew Zealand Dollar (NZD)\n\nAustralian Dollar (AUD)\n\nBritish Pound (GBP)\n\nNorwegian Krone (NOK)\n\nU.S. Dollar (USD)\n\nCurrent 1 Yr Prior\n\n2\\.36\n\n3\\.56\n\n5\\.38\n\n4\\.85\n\n4\\.86\n\n6\\.02\n\n6\\.46\n\n6\\.65\n\n9\\.18\n\n11\\.34\n\n9\\.62\n\n10\\.49\n\n20\\.70\n\n40\\.18\n\n1\\.54\n\n2\\.86\n\n3\\.95\n\n4\\.26\n\n5\\.03\n\n5\\.62\n\n6\\.03\n\n6\\.17\n\n7\\.91\n\n8\\.27\n\n8\\.71\n\n14\\.64\n\n22\\.12\n\n42\\.67\n\n0\\.00 10.00 20.00 30.00 40.00 50.00\n\nThai Baht (THB)\n\nMalaysian Ringgit (MYR)\n\nPeruvian New Sol (PEN)\n\nPolish Zloty (PLN)\n\nChilean Peso (CLP)\n\nPhilippine Peso (PHP)\n\nIndonesian Rupiah (IDR)\n\nHungarian Forint (HUF)\n\nSouth African Rand (ZAR)\n\nMexican Peso (MXN)\n\nColombian Peso (COP)\n\nBrazilian Real (BRL)\n\nNigerian Naira (NGN)\n\nTurkish New Lira (TRY)\n\nCurrent 1 Yr Prior","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/article_thebeatjun2025_p46.pdf"]},"response":{"errors":[],"extract_id":"extract_96fdf09135b8a6844a04ef1cf9dd28ec","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["The **BEAT** | June 2025\n\n# Local Sovereign Currency Yields\n\n**ALTERNATIVES**\n\n**Past performance is no guarantee of future results.** It is not possible to invest directly in an index. \\*German Rate. Source: Bloomberg as 5/31/25. Data provided is for informational use only. See end of report for important additional information.\n\n46\n\n### **Developed Market Local Interest Rates % (1 Year) Emerging Market Local Interest Rates % (1 Year)**\n\n1\\.05\n\n0\\.36\n\n3\\.29\n\n3\\.42\n\n3\\.34\n\n4\\.56\n\n5\\.02\n\n4\\.29\n\n4\\.78\n\n4\\.33\n\n5\\.11\n\n\\-0.27\n\n0\\.74\n\n1\\.56\n\n1\\.76\n\n1\\.91\n\n2\\.56\n\n3\\.17\n\n3\\.40\n\n3\\.83\n\n4\\.00\n\n4\\.12\n\n\\-5.00 0.00 5.00 10.00 15.00\n\nSwiss Franc (CHF)\n\nJapanese Yen (JPY)\n\nDanish Krone (DKK)\n\nEuro (EUR) \\*\n\nSwedish Krona (SEK)\n\nCanadian Dollar (CAD)\n\nNew Zealand Dollar (NZD)\n\nAustralian Dollar (AUD)\n\nBritish Pound (GBP)\n\nNorwegian Krone (NOK)\n\nU.S. Dollar (USD)\n\nCurrent 1 Yr Prior\n\n2\\.36\n\n3\\.56\n\n5\\.38\n\n4\\.85\n\n4\\.86\n\n6\\.02\n\n6\\.46\n\n6\\.65\n\n9\\.18\n\n11\\.34\n\n9\\.62\n\n10\\.49\n\n20\\.70\n\n40\\.18\n\n1\\.54\n\n2\\.86\n\n3\\.95\n\n4\\.26\n\n5\\.03\n\n5\\.62\n\n6\\.03\n\n6\\.17\n\n7\\.91\n\n8\\.27\n\n8\\.71\n\n14\\.64\n\n22\\.12\n\n42\\.67\n\n0\\.00 10.00 20.00 30.00 40.00 50.00\n\nThai Baht (THB)\n\nMalaysian Ringgit (MYR)\n\nPeruvian New Sol (PEN)\n\nPolish Zloty (PLN)\n\nChilean Peso (CLP)\n\nPhilippine Peso (PHP)\n\nIndonesian Rupiah (IDR)\n\nHungarian Forint (HUF)\n\nSouth African Rand (ZAR)\n\nMexican Peso (MXN)\n\nColombian Peso (COP)\n\nBrazilian Real (BRL)\n\nNigerian Naira (NGN)\n\nTurkish New Lira (TRY)\n\nCurrent 1 Yr Prior"],"full_content":"The **BEAT** | June 2025\n\n# Local Sovereign Currency Yields\n\n**ALTERNATIVES**\n\n**Past performance is no guarantee of future results.** It is not possible to invest directly in an index. \\*German Rate. Source: Bloomberg as 5/31/25. Data provided is for informational use only. See end of report for important additional information.\n\n46\n\n### **Developed Market Local Interest Rates % (1 Year) Emerging Market Local Interest Rates % (1 Year)**\n\n1\\.05\n\n0\\.36\n\n3\\.29\n\n3\\.42\n\n3\\.34\n\n4\\.56\n\n5\\.02\n\n4\\.29\n\n4\\.78\n\n4\\.33\n\n5\\.11\n\n\\-0.27\n\n0\\.74\n\n1\\.56\n\n1\\.76\n\n1\\.91\n\n2\\.56\n\n3\\.17\n\n3\\.40\n\n3\\.83\n\n4\\.00\n\n4\\.12\n\n\\-5.00 0.00 5.00 10.00 15.00\n\nSwiss Franc (CHF)\n\nJapanese Yen (JPY)\n\nDanish Krone (DKK)\n\nEuro (EUR) \\*\n\nSwedish Krona (SEK)\n\nCanadian Dollar (CAD)\n\nNew Zealand Dollar (NZD)\n\nAustralian Dollar (AUD)\n\nBritish Pound (GBP)\n\nNorwegian Krone (NOK)\n\nU.S. Dollar (USD)\n\nCurrent 1 Yr Prior\n\n2\\.36\n\n3\\.56\n\n5\\.38\n\n4\\.85\n\n4\\.86\n\n6\\.02\n\n6\\.46\n\n6\\.65\n\n9\\.18\n\n11\\.34\n\n9\\.62\n\n10\\.49\n\n20\\.70\n\n40\\.18\n\n1\\.54\n\n2\\.86\n\n3\\.95\n\n4\\.26\n\n5\\.03\n\n5\\.62\n\n6\\.03\n\n6\\.17\n\n7\\.91\n\n8\\.27\n\n8\\.71\n\n14\\.64\n\n22\\.12\n\n42\\.67\n\n0\\.00 10.00 20.00 30.00 40.00 50.00\n\nThai Baht (THB)\n\nMalaysian Ringgit (MYR)\n\nPeruvian New Sol (PEN)\n\nPolish Zloty (PLN)\n\nChilean Peso (CLP)\n\nPhilippine Peso (PHP)\n\nIndonesian Rupiah (IDR)\n\nHungarian Forint (HUF)\n\nSouth African Rand (ZAR)\n\nMexican Peso (MXN)\n\nColombian Peso (COP)\n\nBrazilian Real (BRL)\n\nNigerian Naira (NGN)\n\nTurkish New Lira (TRY)\n\nCurrent 1 Yr Prior","publish_date":null,"title":"Local Sovereign Currency Yields","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/article_thebeatjun2025_p46.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/article_thebeatjun2025_p46.pdf"},"request":{"example_id":"chart/article_thebeatjun2025_p46","product_type":"parse","source_file_path":"article_thebeatjun2025_p46.pdf"},"started_at":"2026-11-04T19:46:38.330992"} +{"completed_at":"2026-11-04T19:46:42.192880","latency_in_ms":2302,"output":{"example_id":"chart/article_thebeatjun2025_p6","grounded_pages":[],"layout_pages":[],"markdown":"The **BEAT** | June 2025\n\n# With Recent Developments, Have We Reached Peak Tariff De-escalation?\n\n**TOP THEMES**\n\nSource: Bloomberg, Yale Budget Lab, MSIM. As of May 20, 2025. For illustrative purposes only. Not a recommendation to buy or sell any security. It is not possible to invest directly in an index. The views and opinions expressed are those of the portfolio management team at the time of writing/of this presentation and are subject to change at any time due to market, economic, or other conditions, and may not necessarily come to pass. Forecasts/estimates are based on current market conditions, subject to change, and may not necessarily come to pass. **Past performance is not indicative of future results.**\n\nWith the May 12 th temporary pause/reduction of tariffs on China, the overall average effective tariff rate (as estimated by the\n\n### Yale Budget Lab) fell by just over 10%, down to 17.8% on a pre-substitution basis. With the recent de-escalation with China proving to be much more significant than market expectations, further reductions seem less likely.\n\n### **Following the Recent Trade Deal With China, Have We Reached Peak Tariff De-escalation?** _Average pre/post-substitution effective tariff rate (%), as estimated by the Yale Budget Lab_\n\n6\n\n10\n\n12\n\n14\n\n16\n\n18\n\n20\n\n22\n\n24\n\n26\n\n28\n\n30\n\n3/15/2025\n\n3/17/2025\n\n3/19/2025\n\n3/21/2025\n\n3/23/2025\n\n3/25/2025\n\n3/27/2025\n\n3/29/2025\n\n3/31/2025\n\n4/2/2025\n\n4/4/2025\n\n4/6/2025\n\n4/8/2025\n\n4/10/2025\n\n4/12/2025\n\n4/14/2025\n\n4/16/2025\n\n4/18/2025\n\n4/20/2025\n\n4/22/2025\n\n4/24/2025\n\n4/26/2025\n\n4/28/2025\n\n4/30/2025\n\n5/2/2025\n\n5/4/2025\n\n5/6/2025\n\n5/8/2025\n\n5/10/2025\n\n5/12/2025\n\n5/14/2025\n\n5/16/2025\n\n5/18/2025\n\n5/20/2025\n\nPre-Substitution Average Effective Tariff Rate Post-Substitution Average Effective Tariff Rate\n\n**Pre-Liberation day**\n\n**April 2** **nd** **Announcement**\n\n**April 9** **th** **Delay, China Escalation**\n\n**US/China De-escalation**","pages":[{"line_numbers":[],"markdown":"The **BEAT** | June 2025\n\n# With Recent Developments, Have We Reached Peak Tariff De-escalation?\n\n**TOP THEMES**\n\nSource: Bloomberg, Yale Budget Lab, MSIM. As of May 20, 2025. For illustrative purposes only. Not a recommendation to buy or sell any security. It is not possible to invest directly in an index. The views and opinions expressed are those of the portfolio management team at the time of writing/of this presentation and are subject to change at any time due to market, economic, or other conditions, and may not necessarily come to pass. Forecasts/estimates are based on current market conditions, subject to change, and may not necessarily come to pass. **Past performance is not indicative of future results.**\n\nWith the May 12 th temporary pause/reduction of tariffs on China, the overall average effective tariff rate (as estimated by the\n\n### Yale Budget Lab) fell by just over 10%, down to 17.8% on a pre-substitution basis. With the recent de-escalation with China proving to be much more significant than market expectations, further reductions seem less likely.\n\n### **Following the Recent Trade Deal With China, Have We Reached Peak Tariff De-escalation?** _Average pre/post-substitution effective tariff rate (%), as estimated by the Yale Budget Lab_\n\n6\n\n10\n\n12\n\n14\n\n16\n\n18\n\n20\n\n22\n\n24\n\n26\n\n28\n\n30\n\n3/15/2025\n\n3/17/2025\n\n3/19/2025\n\n3/21/2025\n\n3/23/2025\n\n3/25/2025\n\n3/27/2025\n\n3/29/2025\n\n3/31/2025\n\n4/2/2025\n\n4/4/2025\n\n4/6/2025\n\n4/8/2025\n\n4/10/2025\n\n4/12/2025\n\n4/14/2025\n\n4/16/2025\n\n4/18/2025\n\n4/20/2025\n\n4/22/2025\n\n4/24/2025\n\n4/26/2025\n\n4/28/2025\n\n4/30/2025\n\n5/2/2025\n\n5/4/2025\n\n5/6/2025\n\n5/8/2025\n\n5/10/2025\n\n5/12/2025\n\n5/14/2025\n\n5/16/2025\n\n5/18/2025\n\n5/20/2025\n\nPre-Substitution Average Effective Tariff Rate Post-Substitution Average Effective Tariff Rate\n\n**Pre-Liberation day**\n\n**April 2** **nd** **Announcement**\n\n**April 9** **th** **Delay, China Escalation**\n\n**US/China De-escalation**","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"The **BEAT** | June 2025\n\n# With Recent Developments, Have We Reached Peak Tariff De-escalation?\n\n**TOP THEMES**\n\nSource: Bloomberg, Yale Budget Lab, MSIM. As of May 20, 2025. For illustrative purposes only. Not a recommendation to buy or sell any security. It is not possible to invest directly in an index. The views and opinions expressed are those of the portfolio management team at the time of writing/of this presentation and are subject to change at any time due to market, economic, or other conditions, and may not necessarily come to pass. Forecasts/estimates are based on current market conditions, subject to change, and may not necessarily come to pass. **Past performance is not indicative of future results.**\n\nWith the May 12 th temporary pause/reduction of tariffs on China, the overall average effective tariff rate (as estimated by the\n\n### Yale Budget Lab) fell by just over 10%, down to 17.8% on a pre-substitution basis. With the recent de-escalation with China proving to be much more significant than market expectations, further reductions seem less likely.\n\n### **Following the Recent Trade Deal With China, Have We Reached Peak Tariff De-escalation?** _Average pre/post-substitution effective tariff rate (%), as estimated by the Yale Budget Lab_\n\n6\n\n10\n\n12\n\n14\n\n16\n\n18\n\n20\n\n22\n\n24\n\n26\n\n28\n\n30\n\n3/15/2025\n\n3/17/2025\n\n3/19/2025\n\n3/21/2025\n\n3/23/2025\n\n3/25/2025\n\n3/27/2025\n\n3/29/2025\n\n3/31/2025\n\n4/2/2025\n\n4/4/2025\n\n4/6/2025\n\n4/8/2025\n\n4/10/2025\n\n4/12/2025\n\n4/14/2025\n\n4/16/2025\n\n4/18/2025\n\n4/20/2025\n\n4/22/2025\n\n4/24/2025\n\n4/26/2025\n\n4/28/2025\n\n4/30/2025\n\n5/2/2025\n\n5/4/2025\n\n5/6/2025\n\n5/8/2025\n\n5/10/2025\n\n5/12/2025\n\n5/14/2025\n\n5/16/2025\n\n5/18/2025\n\n5/20/2025\n\nPre-Substitution Average Effective Tariff Rate Post-Substitution Average Effective Tariff Rate\n\n**Pre-Liberation day**\n\n**April 2** **nd** **Announcement**\n\n**April 9** **th** **Delay, China Escalation**\n\n**US/China De-escalation**","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/article_thebeatjun2025_p6.pdf"]},"response":{"errors":[],"extract_id":"extract_107f3efc66bfe7a4a1d38e2de814ddbe","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["The **BEAT** | June 2025\n\n# With Recent Developments, Have We Reached Peak Tariff De-escalation?\n\n**TOP THEMES**\n\nSource: Bloomberg, Yale Budget Lab, MSIM. As of May 20, 2025. For illustrative purposes only. Not a recommendation to buy or sell any security. It is not possible to invest directly in an index. The views and opinions expressed are those of the portfolio management team at the time of writing/of this presentation and are subject to change at any time due to market, economic, or other conditions, and may not necessarily come to pass. Forecasts/estimates are based on current market conditions, subject to change, and may not necessarily come to pass. **Past performance is not indicative of future results.**\n\nWith the May 12 th temporary pause/reduction of tariffs on China, the overall average effective tariff rate (as estimated by the\n\n### Yale Budget Lab) fell by just over 10%, down to 17.8% on a pre-substitution basis. With the recent de-escalation with China proving to be much more significant than market expectations, further reductions seem less likely.\n\n### **Following the Recent Trade Deal With China, Have We Reached Peak Tariff De-escalation?** _Average pre/post-substitution effective tariff rate (%), as estimated by the Yale Budget Lab_\n\n6\n\n10\n\n12\n\n14\n\n16\n\n18\n\n20\n\n22\n\n24\n\n26\n\n28\n\n30\n\n3/15/2025\n\n3/17/2025\n\n3/19/2025\n\n3/21/2025\n\n3/23/2025\n\n3/25/2025\n\n3/27/2025\n\n3/29/2025\n\n3/31/2025\n\n4/2/2025\n\n4/4/2025\n\n4/6/2025\n\n4/8/2025\n\n4/10/2025\n\n4/12/2025\n\n4/14/2025\n\n4/16/2025\n\n4/18/2025\n\n4/20/2025\n\n4/22/2025\n\n4/24/2025\n\n4/26/2025\n\n4/28/2025\n\n4/30/2025\n\n5/2/2025\n\n5/4/2025\n\n5/6/2025\n\n5/8/2025\n\n5/10/2025\n\n5/12/2025\n\n5/14/2025\n\n5/16/2025\n\n5/18/2025\n\n5/20/2025\n\nPre-Substitution Average Effective Tariff Rate Post-Substitution Average Effective Tariff Rate\n\n**Pre-Liberation day**\n\n**April 2** **nd** **Announcement**\n\n**April 9** **th** **Delay, China Escalation**\n\n**US/China De-escalation**"],"full_content":"The **BEAT** | June 2025\n\n# With Recent Developments, Have We Reached Peak Tariff De-escalation?\n\n**TOP THEMES**\n\nSource: Bloomberg, Yale Budget Lab, MSIM. As of May 20, 2025. For illustrative purposes only. Not a recommendation to buy or sell any security. It is not possible to invest directly in an index. The views and opinions expressed are those of the portfolio management team at the time of writing/of this presentation and are subject to change at any time due to market, economic, or other conditions, and may not necessarily come to pass. Forecasts/estimates are based on current market conditions, subject to change, and may not necessarily come to pass. **Past performance is not indicative of future results.**\n\nWith the May 12 th temporary pause/reduction of tariffs on China, the overall average effective tariff rate (as estimated by the\n\n### Yale Budget Lab) fell by just over 10%, down to 17.8% on a pre-substitution basis. With the recent de-escalation with China proving to be much more significant than market expectations, further reductions seem less likely.\n\n### **Following the Recent Trade Deal With China, Have We Reached Peak Tariff De-escalation?** _Average pre/post-substitution effective tariff rate (%), as estimated by the Yale Budget Lab_\n\n6\n\n10\n\n12\n\n14\n\n16\n\n18\n\n20\n\n22\n\n24\n\n26\n\n28\n\n30\n\n3/15/2025\n\n3/17/2025\n\n3/19/2025\n\n3/21/2025\n\n3/23/2025\n\n3/25/2025\n\n3/27/2025\n\n3/29/2025\n\n3/31/2025\n\n4/2/2025\n\n4/4/2025\n\n4/6/2025\n\n4/8/2025\n\n4/10/2025\n\n4/12/2025\n\n4/14/2025\n\n4/16/2025\n\n4/18/2025\n\n4/20/2025\n\n4/22/2025\n\n4/24/2025\n\n4/26/2025\n\n4/28/2025\n\n4/30/2025\n\n5/2/2025\n\n5/4/2025\n\n5/6/2025\n\n5/8/2025\n\n5/10/2025\n\n5/12/2025\n\n5/14/2025\n\n5/16/2025\n\n5/18/2025\n\n5/20/2025\n\nPre-Substitution Average Effective Tariff Rate Post-Substitution Average Effective Tariff Rate\n\n**Pre-Liberation day**\n\n**April 2** **nd** **Announcement**\n\n**April 9** **th** **Delay, China Escalation**\n\n**US/China De-escalation**","publish_date":null,"title":"With Recent Developments, Have We Reached Peak Tariff De-escalation?","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/article_thebeatjun2025_p6.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/article_thebeatjun2025_p6.pdf"},"request":{"example_id":"chart/article_thebeatjun2025_p6","product_type":"parse","source_file_path":"article_thebeatjun2025_p6.pdf"},"started_at":"2026-11-04T19:46:39.890017"} +{"completed_at":"2026-11-04T19:46:42.434775","latency_in_ms":1215,"output":{"example_id":"chart/article_thebeatjun2025_p7","grounded_pages":[],"layout_pages":[],"markdown":"The **BEAT** | June 2025\n\n# With De-escalation in Focus, Markets Have Exhibited “Risk-On” Trends\n\n**TOP THEMES**\n\nSource: Bloomberg, MSIM. As of May 20, 2025. For illustrative purposes only. Not a recommendation to buy or sell any security. It is not possible to invest directly in an index. The views and opinions expressed are those of the portfolio management team at the time of writing/of this presentation and are subject to change at any time due to market, economic, or other conditions, and may not necessarily come to pass. Forecasts/estimates are based on current market conditions, subject to change, and may not necessarily come to pass. **Past performance is not indicative of future results.**\n\nSince the reciprocal tariffs delay announced on April 9 th , equity market internals have signaled risk-on trends across several\n\n### perspectives: On average, Cyclicals (ex energy) have outperformed Defensives, the high-volatility factor has outperformed low-volatility (sector neutral basis), and our tariff “exposed” basket of stocks has outperformed the tariff “resilient” basket.\n\n### **The “Risk** **\\-** **On” Trade Has Returned Since Early April, with De** **\\-escalation in Trade Tensions a Key Catalyst** _YTD cumulative performance for key pair trades_\n\n7\n\n\\-35%\n\n\\-30%\n\n\\-25%\n\n\\-20%\n\n\\-15%\n\n\\-10%\n\n\\-5%\n\n0%\n\n5%\n\nJan 2025 Feb 2025 Mar 2025 Apr 2025 May 2025\n\nS&P 500 Tariff Exposed vs. Tariff Resilient S&P 500 Cyclicals (ex Energy) vs. Defensives S&P 500 Sector Neutral High Vol vs. Low Vol Long-Short\n\n**De-escalation news flow has** **reversed most of the “risk** **\\-** **off”** **moves in equity factors and sectors**","pages":[{"line_numbers":[],"markdown":"The **BEAT** | June 2025\n\n# With De-escalation in Focus, Markets Have Exhibited “Risk-On” Trends\n\n**TOP THEMES**\n\nSource: Bloomberg, MSIM. As of May 20, 2025. For illustrative purposes only. Not a recommendation to buy or sell any security. It is not possible to invest directly in an index. The views and opinions expressed are those of the portfolio management team at the time of writing/of this presentation and are subject to change at any time due to market, economic, or other conditions, and may not necessarily come to pass. Forecasts/estimates are based on current market conditions, subject to change, and may not necessarily come to pass. **Past performance is not indicative of future results.**\n\nSince the reciprocal tariffs delay announced on April 9 th , equity market internals have signaled risk-on trends across several\n\n### perspectives: On average, Cyclicals (ex energy) have outperformed Defensives, the high-volatility factor has outperformed low-volatility (sector neutral basis), and our tariff “exposed” basket of stocks has outperformed the tariff “resilient” basket.\n\n### **The “Risk** **\\-** **On” Trade Has Returned Since Early April, with De** **\\-escalation in Trade Tensions a Key Catalyst** _YTD cumulative performance for key pair trades_\n\n7\n\n\\-35%\n\n\\-30%\n\n\\-25%\n\n\\-20%\n\n\\-15%\n\n\\-10%\n\n\\-5%\n\n0%\n\n5%\n\nJan 2025 Feb 2025 Mar 2025 Apr 2025 May 2025\n\nS&P 500 Tariff Exposed vs. Tariff Resilient S&P 500 Cyclicals (ex Energy) vs. Defensives S&P 500 Sector Neutral High Vol vs. Low Vol Long-Short\n\n**De-escalation news flow has** **reversed most of the “risk** **\\-** **off”** **moves in equity factors and sectors**","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"The **BEAT** | June 2025\n\n# With De-escalation in Focus, Markets Have Exhibited “Risk-On” Trends\n\n**TOP THEMES**\n\nSource: Bloomberg, MSIM. As of May 20, 2025. For illustrative purposes only. Not a recommendation to buy or sell any security. It is not possible to invest directly in an index. The views and opinions expressed are those of the portfolio management team at the time of writing/of this presentation and are subject to change at any time due to market, economic, or other conditions, and may not necessarily come to pass. Forecasts/estimates are based on current market conditions, subject to change, and may not necessarily come to pass. **Past performance is not indicative of future results.**\n\nSince the reciprocal tariffs delay announced on April 9 th , equity market internals have signaled risk-on trends across several\n\n### perspectives: On average, Cyclicals (ex energy) have outperformed Defensives, the high-volatility factor has outperformed low-volatility (sector neutral basis), and our tariff “exposed” basket of stocks has outperformed the tariff “resilient” basket.\n\n### **The “Risk** **\\-** **On” Trade Has Returned Since Early April, with De** **\\-escalation in Trade Tensions a Key Catalyst** _YTD cumulative performance for key pair trades_\n\n7\n\n\\-35%\n\n\\-30%\n\n\\-25%\n\n\\-20%\n\n\\-15%\n\n\\-10%\n\n\\-5%\n\n0%\n\n5%\n\nJan 2025 Feb 2025 Mar 2025 Apr 2025 May 2025\n\nS&P 500 Tariff Exposed vs. Tariff Resilient S&P 500 Cyclicals (ex Energy) vs. Defensives S&P 500 Sector Neutral High Vol vs. Low Vol Long-Short\n\n**De-escalation news flow has** **reversed most of the “risk** **\\-** **off”** **moves in equity factors and sectors**","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/article_thebeatjun2025_p7.pdf"]},"response":{"errors":[],"extract_id":"extract_90aff09e89708707ca6bd3ec31f0a638","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["The **BEAT** | June 2025\n\n# With De-escalation in Focus, Markets Have Exhibited “Risk-On” Trends\n\n**TOP THEMES**\n\nSource: Bloomberg, MSIM. As of May 20, 2025. For illustrative purposes only. Not a recommendation to buy or sell any security. It is not possible to invest directly in an index. The views and opinions expressed are those of the portfolio management team at the time of writing/of this presentation and are subject to change at any time due to market, economic, or other conditions, and may not necessarily come to pass. Forecasts/estimates are based on current market conditions, subject to change, and may not necessarily come to pass. **Past performance is not indicative of future results.**\n\nSince the reciprocal tariffs delay announced on April 9 th , equity market internals have signaled risk-on trends across several\n\n### perspectives: On average, Cyclicals (ex energy) have outperformed Defensives, the high-volatility factor has outperformed low-volatility (sector neutral basis), and our tariff “exposed” basket of stocks has outperformed the tariff “resilient” basket.\n\n### **The “Risk** **\\-** **On” Trade Has Returned Since Early April, with De** **\\-escalation in Trade Tensions a Key Catalyst** _YTD cumulative performance for key pair trades_\n\n7\n\n\\-35%\n\n\\-30%\n\n\\-25%\n\n\\-20%\n\n\\-15%\n\n\\-10%\n\n\\-5%\n\n0%\n\n5%\n\nJan 2025 Feb 2025 Mar 2025 Apr 2025 May 2025\n\nS&P 500 Tariff Exposed vs. Tariff Resilient S&P 500 Cyclicals (ex Energy) vs. Defensives S&P 500 Sector Neutral High Vol vs. Low Vol Long-Short\n\n**De-escalation news flow has** **reversed most of the “risk** **\\-** **off”** **moves in equity factors and sectors**"],"full_content":"The **BEAT** | June 2025\n\n# With De-escalation in Focus, Markets Have Exhibited “Risk-On” Trends\n\n**TOP THEMES**\n\nSource: Bloomberg, MSIM. As of May 20, 2025. For illustrative purposes only. Not a recommendation to buy or sell any security. It is not possible to invest directly in an index. The views and opinions expressed are those of the portfolio management team at the time of writing/of this presentation and are subject to change at any time due to market, economic, or other conditions, and may not necessarily come to pass. Forecasts/estimates are based on current market conditions, subject to change, and may not necessarily come to pass. **Past performance is not indicative of future results.**\n\nSince the reciprocal tariffs delay announced on April 9 th , equity market internals have signaled risk-on trends across several\n\n### perspectives: On average, Cyclicals (ex energy) have outperformed Defensives, the high-volatility factor has outperformed low-volatility (sector neutral basis), and our tariff “exposed” basket of stocks has outperformed the tariff “resilient” basket.\n\n### **The “Risk** **\\-** **On” Trade Has Returned Since Early April, with De** **\\-escalation in Trade Tensions a Key Catalyst** _YTD cumulative performance for key pair trades_\n\n7\n\n\\-35%\n\n\\-30%\n\n\\-25%\n\n\\-20%\n\n\\-15%\n\n\\-10%\n\n\\-5%\n\n0%\n\n5%\n\nJan 2025 Feb 2025 Mar 2025 Apr 2025 May 2025\n\nS&P 500 Tariff Exposed vs. Tariff Resilient S&P 500 Cyclicals (ex Energy) vs. Defensives S&P 500 Sector Neutral High Vol vs. Low Vol Long-Short\n\n**De-escalation news flow has** **reversed most of the “risk** **\\-** **off”** **moves in equity factors and sectors**","publish_date":null,"title":"With De-escalation in Focus, Markets Have Exhibited “Risk-On” Trends","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/article_thebeatjun2025_p7.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/article_thebeatjun2025_p7.pdf"},"request":{"example_id":"chart/article_thebeatjun2025_p7","product_type":"parse","source_file_path":"article_thebeatjun2025_p7.pdf"},"started_at":"2026-11-04T19:46:41.219210"} +{"completed_at":"2026-11-04T19:46:43.259643","latency_in_ms":1063,"output":{"example_id":"chart/article_thebeatjun2025_p8","grounded_pages":[],"layout_pages":[],"markdown":"The **BEAT** | June 2025\n\n## Over the Past Three Years, Payroll Gains Have Been Driven by the Public Sector\n\n**TOP THEMES**\n\nSource: Bloomberg, MSIM. As of April 30, 2025. For illustrative purposes only. Not a recommendation to buy or sell any security. It is not possible to invest directly in an index. The views and opinions expressed are those of the portfolio management team at the time of writing/of this presentation and are subject to change at any time due to market, economic, or other conditions, and may not necessarily come to pass. Forecasts/estimates are based on current market conditions, subject to change, and may not necessarily come to pass. **Past performance is not indicative of future results.**\n\n### Over recent years, the majority of payroll job gains have been in the public sector (government, education, health services), resulting in a crowding out of private sector employment. It remains to be seen whether the new administration’s policies can lead to a reconvergence in employment trends, which would be positive for labor productivity and inflation.\n\n### **Public Sector Employment Has Exploded Since 2022, Resulting in a Crowding Out of the Private Sector** _Cumulative growth in total payrolls since 12/31/2022_\n\n8\n\n0%\n\n1%\n\n2%\n\n3%\n\n4%\n\n5%\n\n6%\n\n7%\n\n8%\n\nDec 2022 Mar 2023 Jun 2023 Sep 2023 Dec 2023 Mar 2024 Jun 2024 Sep 2024 Dec 2024 Mar 2025\n\nPrivate Sector Total Payrolls (ex Govt, Education, Health Svcs) Public Sector Total Payrolls (Gov't, Education, Health Svcs)\n\n### **... crowding out the private sector**\n\n### **Since 2022, public sector employment has grown substantially...**","pages":[{"line_numbers":[],"markdown":"The **BEAT** | June 2025\n\n## Over the Past Three Years, Payroll Gains Have Been Driven by the Public Sector\n\n**TOP THEMES**\n\nSource: Bloomberg, MSIM. As of April 30, 2025. For illustrative purposes only. Not a recommendation to buy or sell any security. It is not possible to invest directly in an index. The views and opinions expressed are those of the portfolio management team at the time of writing/of this presentation and are subject to change at any time due to market, economic, or other conditions, and may not necessarily come to pass. Forecasts/estimates are based on current market conditions, subject to change, and may not necessarily come to pass. **Past performance is not indicative of future results.**\n\n### Over recent years, the majority of payroll job gains have been in the public sector (government, education, health services), resulting in a crowding out of private sector employment. It remains to be seen whether the new administration’s policies can lead to a reconvergence in employment trends, which would be positive for labor productivity and inflation.\n\n### **Public Sector Employment Has Exploded Since 2022, Resulting in a Crowding Out of the Private Sector** _Cumulative growth in total payrolls since 12/31/2022_\n\n8\n\n0%\n\n1%\n\n2%\n\n3%\n\n4%\n\n5%\n\n6%\n\n7%\n\n8%\n\nDec 2022 Mar 2023 Jun 2023 Sep 2023 Dec 2023 Mar 2024 Jun 2024 Sep 2024 Dec 2024 Mar 2025\n\nPrivate Sector Total Payrolls (ex Govt, Education, Health Svcs) Public Sector Total Payrolls (Gov't, Education, Health Svcs)\n\n### **... crowding out the private sector**\n\n### **Since 2022, public sector employment has grown substantially...**","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"The **BEAT** | June 2025\n\n## Over the Past Three Years, Payroll Gains Have Been Driven by the Public Sector\n\n**TOP THEMES**\n\nSource: Bloomberg, MSIM. As of April 30, 2025. For illustrative purposes only. Not a recommendation to buy or sell any security. It is not possible to invest directly in an index. The views and opinions expressed are those of the portfolio management team at the time of writing/of this presentation and are subject to change at any time due to market, economic, or other conditions, and may not necessarily come to pass. Forecasts/estimates are based on current market conditions, subject to change, and may not necessarily come to pass. **Past performance is not indicative of future results.**\n\n### Over recent years, the majority of payroll job gains have been in the public sector (government, education, health services), resulting in a crowding out of private sector employment. It remains to be seen whether the new administration’s policies can lead to a reconvergence in employment trends, which would be positive for labor productivity and inflation.\n\n### **Public Sector Employment Has Exploded Since 2022, Resulting in a Crowding Out of the Private Sector** _Cumulative growth in total payrolls since 12/31/2022_\n\n8\n\n0%\n\n1%\n\n2%\n\n3%\n\n4%\n\n5%\n\n6%\n\n7%\n\n8%\n\nDec 2022 Mar 2023 Jun 2023 Sep 2023 Dec 2023 Mar 2024 Jun 2024 Sep 2024 Dec 2024 Mar 2025\n\nPrivate Sector Total Payrolls (ex Govt, Education, Health Svcs) Public Sector Total Payrolls (Gov't, Education, Health Svcs)\n\n### **... crowding out the private sector**\n\n### **Since 2022, public sector employment has grown substantially...**","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/article_thebeatjun2025_p8.pdf"]},"response":{"errors":[],"extract_id":"extract_4d93fc905cc2610fd304aa70605effeb","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["The **BEAT** | June 2025\n\n## Over the Past Three Years, Payroll Gains Have Been Driven by the Public Sector\n\n**TOP THEMES**\n\nSource: Bloomberg, MSIM. As of April 30, 2025. For illustrative purposes only. Not a recommendation to buy or sell any security. It is not possible to invest directly in an index. The views and opinions expressed are those of the portfolio management team at the time of writing/of this presentation and are subject to change at any time due to market, economic, or other conditions, and may not necessarily come to pass. Forecasts/estimates are based on current market conditions, subject to change, and may not necessarily come to pass. **Past performance is not indicative of future results.**\n\n### Over recent years, the majority of payroll job gains have been in the public sector (government, education, health services), resulting in a crowding out of private sector employment. It remains to be seen whether the new administration’s policies can lead to a reconvergence in employment trends, which would be positive for labor productivity and inflation.\n\n### **Public Sector Employment Has Exploded Since 2022, Resulting in a Crowding Out of the Private Sector** _Cumulative growth in total payrolls since 12/31/2022_\n\n8\n\n0%\n\n1%\n\n2%\n\n3%\n\n4%\n\n5%\n\n6%\n\n7%\n\n8%\n\nDec 2022 Mar 2023 Jun 2023 Sep 2023 Dec 2023 Mar 2024 Jun 2024 Sep 2024 Dec 2024 Mar 2025\n\nPrivate Sector Total Payrolls (ex Govt, Education, Health Svcs) Public Sector Total Payrolls (Gov't, Education, Health Svcs)\n\n### **... crowding out the private sector**\n\n### **Since 2022, public sector employment has grown substantially...**"],"full_content":"The **BEAT** | June 2025\n\n## Over the Past Three Years, Payroll Gains Have Been Driven by the Public Sector\n\n**TOP THEMES**\n\nSource: Bloomberg, MSIM. As of April 30, 2025. For illustrative purposes only. Not a recommendation to buy or sell any security. It is not possible to invest directly in an index. The views and opinions expressed are those of the portfolio management team at the time of writing/of this presentation and are subject to change at any time due to market, economic, or other conditions, and may not necessarily come to pass. Forecasts/estimates are based on current market conditions, subject to change, and may not necessarily come to pass. **Past performance is not indicative of future results.**\n\n### Over recent years, the majority of payroll job gains have been in the public sector (government, education, health services), resulting in a crowding out of private sector employment. It remains to be seen whether the new administration’s policies can lead to a reconvergence in employment trends, which would be positive for labor productivity and inflation.\n\n### **Public Sector Employment Has Exploded Since 2022, Resulting in a Crowding Out of the Private Sector** _Cumulative growth in total payrolls since 12/31/2022_\n\n8\n\n0%\n\n1%\n\n2%\n\n3%\n\n4%\n\n5%\n\n6%\n\n7%\n\n8%\n\nDec 2022 Mar 2023 Jun 2023 Sep 2023 Dec 2023 Mar 2024 Jun 2024 Sep 2024 Dec 2024 Mar 2025\n\nPrivate Sector Total Payrolls (ex Govt, Education, Health Svcs) Public Sector Total Payrolls (Gov't, Education, Health Svcs)\n\n### **... crowding out the private sector**\n\n### **Since 2022, public sector employment has grown substantially...**","publish_date":null,"title":"Over the Past Three Years, Payroll Gains Have Been Driven by the Public Sector","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/article_thebeatjun2025_p8.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/article_thebeatjun2025_p8.pdf"},"request":{"example_id":"chart/article_thebeatjun2025_p8","product_type":"parse","source_file_path":"article_thebeatjun2025_p8.pdf"},"started_at":"2026-11-04T19:46:42.195923"} +{"completed_at":"2026-11-04T19:45:43.894595","latency_in_ms":4807,"output":{"example_id":"chart/aviva-plc-annual-report-and-accounts-2024_p143","grounded_pages":[],"layout_pages":[],"markdown":"**Illustration of the Policy**\n\nThe charts below illustrate how much EDs could earn under different performance scenarios in one financial year:\n\n• Minimum – basic salary, pension or cash in lieu of pension and benefits, no bonus and no vesting of the LTIP.\n\n• Target - basic salary, pension or cash in lieu of pension, benefits, and: – A bonus of 100% and a LTIP of 350% of basic salary (with notional LTIP vesting at 50% of maximum) for the Group CEO. – A bonus of 100% and a LTIP of 225% of basic salary (with notional LTIP vesting at 50% of maximum) for the Group CFO.\n\n• Maximum – basic salary, pension or cash in lieu of pension, benefits, and: – A bonus of 200% and a LTIP of 350% of basic salary (with notional LTIP vesting at maximum) for the Group CEO. – A bonus of 150% and a LTIP of 225% of basic salary (with notional LTIP vesting at maximum) for the Group CFO.\n\n• Maximum with share price appreciation – indicative maximum remuneration, assuming a notional LTIP vesting at maximum and share price appreciation of 50% on the LTIP.\n\n**Potential earnings by pay element - Amanda Blanc**\n\n**Potential earnings by pay element - Charlotte Jones**\n\nNotes to the charts 1. The charts are illustrative only and the actual value EDs could earn is subject to business performance and share price movement to the date of vesting of the LTIP and of the deferred share element of the annual bonus 2. Fixed pay consists of basic salary, pension as described in table 1, and estimated value of benefits provided under the Policy, excluding any one-offs. Actual figures may vary in future years. 3. The value of the deferred element of the annual bonus assumes a constant share price and does not include additional shares awarded in lieu of dividends that may accrue during the vesting period 4. The value of the LTIP assumes a constant share price (with the exception of the maximum with share price increase scenario) and does not include additional shares awarded in lieu of dividends that may accrue during the vesting period 5. The LTIP is as proposed to be awarded in 2025, which would vest in 2028, subject to the satisfaction of performance conditions. The shares would then be subject to a further two-year holding period.\n\nDirectors’ Remuneration Policy\n\n141\n\n£m\n\n2025 Minimum 2025 Target 2025 Maximum 2025 Maximum with\n\nshare price appreciation\n\n0\\.0\n\n5\\.0\n\n10\\.0\n\n£m\n\n£0.9\n\n£2.5m\n\n£3.7m\n\n£4.5m\n\n2025 Minimum 2025 Target 2025 Maximum 2025 Maximum with\n\nshare price appreciation\n\n0\\.0\n\n2\\.5\n\n5\\.0\n\n£1.5m 25% 30%\n\n44%\n\n62%\n\n24%\n\n14%\n\n52%\n\n30%\n\n18%\n\n£8.2m\n\n£10.4m\n\n100%\n\n56%\n\n25%\n\n19%\n\n46%\n\n31%\n\n24%\n\n30%\n\n35%\n\n34%\n\n£4.9m\n\n**Remuneration elements**\n\n**Fixed pay Annual bonus LTIP**\n\n**Aviva plc** Annual Report and Accounts 2024\n\nStrategic Report **Governance Report** IFRS Financial Statements Other Information\n\n100%","pages":[{"line_numbers":[],"markdown":"**Illustration of the Policy**\n\nThe charts below illustrate how much EDs could earn under different performance scenarios in one financial year:\n\n• Minimum – basic salary, pension or cash in lieu of pension and benefits, no bonus and no vesting of the LTIP.\n\n• Target - basic salary, pension or cash in lieu of pension, benefits, and: – A bonus of 100% and a LTIP of 350% of basic salary (with notional LTIP vesting at 50% of maximum) for the Group CEO. – A bonus of 100% and a LTIP of 225% of basic salary (with notional LTIP vesting at 50% of maximum) for the Group CFO.\n\n• Maximum – basic salary, pension or cash in lieu of pension, benefits, and: – A bonus of 200% and a LTIP of 350% of basic salary (with notional LTIP vesting at maximum) for the Group CEO. – A bonus of 150% and a LTIP of 225% of basic salary (with notional LTIP vesting at maximum) for the Group CFO.\n\n• Maximum with share price appreciation – indicative maximum remuneration, assuming a notional LTIP vesting at maximum and share price appreciation of 50% on the LTIP.\n\n**Potential earnings by pay element - Amanda Blanc**\n\n**Potential earnings by pay element - Charlotte Jones**\n\nNotes to the charts 1. The charts are illustrative only and the actual value EDs could earn is subject to business performance and share price movement to the date of vesting of the LTIP and of the deferred share element of the annual bonus 2. Fixed pay consists of basic salary, pension as described in table 1, and estimated value of benefits provided under the Policy, excluding any one-offs. Actual figures may vary in future years. 3. The value of the deferred element of the annual bonus assumes a constant share price and does not include additional shares awarded in lieu of dividends that may accrue during the vesting period 4. The value of the LTIP assumes a constant share price (with the exception of the maximum with share price increase scenario) and does not include additional shares awarded in lieu of dividends that may accrue during the vesting period 5. The LTIP is as proposed to be awarded in 2025, which would vest in 2028, subject to the satisfaction of performance conditions. The shares would then be subject to a further two-year holding period.\n\nDirectors’ Remuneration Policy\n\n141\n\n£m\n\n2025 Minimum 2025 Target 2025 Maximum 2025 Maximum with\n\nshare price appreciation\n\n0\\.0\n\n5\\.0\n\n10\\.0\n\n£m\n\n£0.9\n\n£2.5m\n\n£3.7m\n\n£4.5m\n\n2025 Minimum 2025 Target 2025 Maximum 2025 Maximum with\n\nshare price appreciation\n\n0\\.0\n\n2\\.5\n\n5\\.0\n\n£1.5m 25% 30%\n\n44%\n\n62%\n\n24%\n\n14%\n\n52%\n\n30%\n\n18%\n\n£8.2m\n\n£10.4m\n\n100%\n\n56%\n\n25%\n\n19%\n\n46%\n\n31%\n\n24%\n\n30%\n\n35%\n\n34%\n\n£4.9m\n\n**Remuneration elements**\n\n**Fixed pay Annual bonus LTIP**\n\n**Aviva plc** Annual Report and Accounts 2024\n\nStrategic Report **Governance Report** IFRS Financial Statements Other Information\n\n100%","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**Illustration of the Policy**\n\nThe charts below illustrate how much EDs could earn under different performance scenarios in one financial year:\n\n• Minimum – basic salary, pension or cash in lieu of pension and benefits, no bonus and no vesting of the LTIP.\n\n• Target - basic salary, pension or cash in lieu of pension, benefits, and: – A bonus of 100% and a LTIP of 350% of basic salary (with notional LTIP vesting at 50% of maximum) for the Group CEO. – A bonus of 100% and a LTIP of 225% of basic salary (with notional LTIP vesting at 50% of maximum) for the Group CFO.\n\n• Maximum – basic salary, pension or cash in lieu of pension, benefits, and: – A bonus of 200% and a LTIP of 350% of basic salary (with notional LTIP vesting at maximum) for the Group CEO. – A bonus of 150% and a LTIP of 225% of basic salary (with notional LTIP vesting at maximum) for the Group CFO.\n\n• Maximum with share price appreciation – indicative maximum remuneration, assuming a notional LTIP vesting at maximum and share price appreciation of 50% on the LTIP.\n\n**Potential earnings by pay element - Amanda Blanc**\n\n**Potential earnings by pay element - Charlotte Jones**\n\nNotes to the charts 1. The charts are illustrative only and the actual value EDs could earn is subject to business performance and share price movement to the date of vesting of the LTIP and of the deferred share element of the annual bonus 2. Fixed pay consists of basic salary, pension as described in table 1, and estimated value of benefits provided under the Policy, excluding any one-offs. Actual figures may vary in future years. 3. The value of the deferred element of the annual bonus assumes a constant share price and does not include additional shares awarded in lieu of dividends that may accrue during the vesting period 4. The value of the LTIP assumes a constant share price (with the exception of the maximum with share price increase scenario) and does not include additional shares awarded in lieu of dividends that may accrue during the vesting period 5. The LTIP is as proposed to be awarded in 2025, which would vest in 2028, subject to the satisfaction of performance conditions. The shares would then be subject to a further two-year holding period.\n\nDirectors’ Remuneration Policy\n\n141\n\n£m\n\n2025 Minimum 2025 Target 2025 Maximum 2025 Maximum with\n\nshare price appreciation\n\n0\\.0\n\n5\\.0\n\n10\\.0\n\n£m\n\n£0.9\n\n£2.5m\n\n£3.7m\n\n£4.5m\n\n2025 Minimum 2025 Target 2025 Maximum 2025 Maximum with\n\nshare price appreciation\n\n0\\.0\n\n2\\.5\n\n5\\.0\n\n£1.5m 25% 30%\n\n44%\n\n62%\n\n24%\n\n14%\n\n52%\n\n30%\n\n18%\n\n£8.2m\n\n£10.4m\n\n100%\n\n56%\n\n25%\n\n19%\n\n46%\n\n31%\n\n24%\n\n30%\n\n35%\n\n34%\n\n£4.9m\n\n**Remuneration elements**\n\n**Fixed pay Annual bonus LTIP**\n\n**Aviva plc** Annual Report and Accounts 2024\n\nStrategic Report **Governance Report** IFRS Financial Statements Other Information\n\n100%","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/aviva-plc-annual-report-and-accounts-2024_p143.pdf"]},"response":{"errors":[],"extract_id":"extract_e3faf83c07352692985fa12ca1acf086","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**Illustration of the Policy**\n\nThe charts below illustrate how much EDs could earn under different performance scenarios in one financial year:\n\n• Minimum – basic salary, pension or cash in lieu of pension and benefits, no bonus and no vesting of the LTIP.\n\n• Target - basic salary, pension or cash in lieu of pension, benefits, and: – A bonus of 100% and a LTIP of 350% of basic salary (with notional LTIP vesting at 50% of maximum) for the Group CEO. – A bonus of 100% and a LTIP of 225% of basic salary (with notional LTIP vesting at 50% of maximum) for the Group CFO.\n\n• Maximum – basic salary, pension or cash in lieu of pension, benefits, and: – A bonus of 200% and a LTIP of 350% of basic salary (with notional LTIP vesting at maximum) for the Group CEO. – A bonus of 150% and a LTIP of 225% of basic salary (with notional LTIP vesting at maximum) for the Group CFO.\n\n• Maximum with share price appreciation – indicative maximum remuneration, assuming a notional LTIP vesting at maximum and share price appreciation of 50% on the LTIP.\n\n**Potential earnings by pay element - Amanda Blanc**\n\n**Potential earnings by pay element - Charlotte Jones**\n\nNotes to the charts 1. The charts are illustrative only and the actual value EDs could earn is subject to business performance and share price movement to the date of vesting of the LTIP and of the deferred share element of the annual bonus 2. Fixed pay consists of basic salary, pension as described in table 1, and estimated value of benefits provided under the Policy, excluding any one-offs. Actual figures may vary in future years. 3. The value of the deferred element of the annual bonus assumes a constant share price and does not include additional shares awarded in lieu of dividends that may accrue during the vesting period 4. The value of the LTIP assumes a constant share price (with the exception of the maximum with share price increase scenario) and does not include additional shares awarded in lieu of dividends that may accrue during the vesting period 5. The LTIP is as proposed to be awarded in 2025, which would vest in 2028, subject to the satisfaction of performance conditions. The shares would then be subject to a further two-year holding period.\n\nDirectors’ Remuneration Policy\n\n141\n\n£m\n\n2025 Minimum 2025 Target 2025 Maximum 2025 Maximum with\n\nshare price appreciation\n\n0\\.0\n\n5\\.0\n\n10\\.0\n\n£m\n\n£0.9\n\n£2.5m\n\n£3.7m\n\n£4.5m\n\n2025 Minimum 2025 Target 2025 Maximum 2025 Maximum with\n\nshare price appreciation\n\n0\\.0\n\n2\\.5\n\n5\\.0\n\n£1.5m 25% 30%\n\n44%\n\n62%\n\n24%\n\n14%\n\n52%\n\n30%\n\n18%\n\n£8.2m\n\n£10.4m\n\n100%\n\n56%\n\n25%\n\n19%\n\n46%\n\n31%\n\n24%\n\n30%\n\n35%\n\n34%\n\n£4.9m\n\n**Remuneration elements**\n\n**Fixed pay Annual bonus LTIP**\n\n**Aviva plc** Annual Report and Accounts 2024\n\nStrategic Report **Governance Report** IFRS Financial Statements Other Information\n\n100%"],"full_content":"**Illustration of the Policy**\n\nThe charts below illustrate how much EDs could earn under different performance scenarios in one financial year:\n\n• Minimum – basic salary, pension or cash in lieu of pension and benefits, no bonus and no vesting of the LTIP.\n\n• Target - basic salary, pension or cash in lieu of pension, benefits, and: – A bonus of 100% and a LTIP of 350% of basic salary (with notional LTIP vesting at 50% of maximum) for the Group CEO. – A bonus of 100% and a LTIP of 225% of basic salary (with notional LTIP vesting at 50% of maximum) for the Group CFO.\n\n• Maximum – basic salary, pension or cash in lieu of pension, benefits, and: – A bonus of 200% and a LTIP of 350% of basic salary (with notional LTIP vesting at maximum) for the Group CEO. – A bonus of 150% and a LTIP of 225% of basic salary (with notional LTIP vesting at maximum) for the Group CFO.\n\n• Maximum with share price appreciation – indicative maximum remuneration, assuming a notional LTIP vesting at maximum and share price appreciation of 50% on the LTIP.\n\n**Potential earnings by pay element - Amanda Blanc**\n\n**Potential earnings by pay element - Charlotte Jones**\n\nNotes to the charts 1. The charts are illustrative only and the actual value EDs could earn is subject to business performance and share price movement to the date of vesting of the LTIP and of the deferred share element of the annual bonus 2. Fixed pay consists of basic salary, pension as described in table 1, and estimated value of benefits provided under the Policy, excluding any one-offs. Actual figures may vary in future years. 3. The value of the deferred element of the annual bonus assumes a constant share price and does not include additional shares awarded in lieu of dividends that may accrue during the vesting period 4. The value of the LTIP assumes a constant share price (with the exception of the maximum with share price increase scenario) and does not include additional shares awarded in lieu of dividends that may accrue during the vesting period 5. The LTIP is as proposed to be awarded in 2025, which would vest in 2028, subject to the satisfaction of performance conditions. The shares would then be subject to a further two-year holding period.\n\nDirectors’ Remuneration Policy\n\n141\n\n£m\n\n2025 Minimum 2025 Target 2025 Maximum 2025 Maximum with\n\nshare price appreciation\n\n0\\.0\n\n5\\.0\n\n10\\.0\n\n£m\n\n£0.9\n\n£2.5m\n\n£3.7m\n\n£4.5m\n\n2025 Minimum 2025 Target 2025 Maximum 2025 Maximum with\n\nshare price appreciation\n\n0\\.0\n\n2\\.5\n\n5\\.0\n\n£1.5m 25% 30%\n\n44%\n\n62%\n\n24%\n\n14%\n\n52%\n\n30%\n\n18%\n\n£8.2m\n\n£10.4m\n\n100%\n\n56%\n\n25%\n\n19%\n\n46%\n\n31%\n\n24%\n\n30%\n\n35%\n\n34%\n\n£4.9m\n\n**Remuneration elements**\n\n**Fixed pay Annual bonus LTIP**\n\n**Aviva plc** Annual Report and Accounts 2024\n\nStrategic Report **Governance Report** IFRS Financial Statements Other Information\n\n100%","publish_date":null,"title":"Potential earnings by pay element - Amanda Blanc","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/aviva-plc-annual-report-and-accounts-2024_p143.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/aviva-plc-annual-report-and-accounts-2024_p143.pdf"},"request":{"example_id":"chart/aviva-plc-annual-report-and-accounts-2024_p143","product_type":"parse","source_file_path":"aviva-plc-annual-report-and-accounts-2024_p143.pdf"},"started_at":"2026-11-04T19:45:39.086634"} +{"completed_at":"2026-11-04T19:45:44.979839","latency_in_ms":4126,"output":{"example_id":"chart/b263dc5d-en_p106","grounded_pages":[],"layout_pages":[],"markdown":"**104** \n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 3.1. Change in average literacy and numeracy proficiency between cycles, before and after accounting for demographic changes**\n\nDifference in mean proficiency scores between cycles, after reweighting Cycle 2 to match Cycle 1's distribution of age, immigrant background and gender (Cycle 2 _minus_ Cycle 1)\n\nNote: Adults aged 16-65; does not include adults who in Cycle 2 were only administered the doorstep interview due to a language barrier, to maximise the comparability across cycles (see Box 1.1 in Chapter 1 and Box 3.2). Unadjusted differences are the differences between the averages in each cycle. Adjusted differences are based on a method analogous to post-stratification that reweights the samples in Cycle 2 so that the demographic characteristics of these samples match those of the samples in Cycle 1 (see Note 1). Demographic characteristics considered are: age (in ten-year age brackets), gender and immigrant background. Adjusted differences represent a hypothetical scenario of how the proficiency of a population matching the demographic profile of the population in Cycle 1 would have changed over time. Darker colours denote differences that are statistically significant at the 5% level. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the ote for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the unadjusted change in mean literacy proficiency scores._ Source: OECD (2018[4]; 2015[5]; 2012[6]), _Survey of Adult Skills (PIAAC) databases_ , http://www.oecd.org/skills/piaac/publicdataandanalysis/ (accessed on 23 September 2024); Tables A.3.1 (L) and A.3.1 (N) in Annex A.\n\nFinland\n\nenmark\n\nSweden\n\norway\n\nFlemish Region ( E\n\nEstonia\n\nGermany\n\nEngland (U\n\nCanada\n\netherlands\n\nSpain\n\nIreland\n\nItaly\n\napan\n\nFrance\n\nC echia\n\nAustria\n\nUnited States\n\nSlo ak Republic\n\norea\n\nPoland\n\nChile\n\nSingapore\n\nIsrael\n\new ealand\n\nithuania\n\nHungary\n\nFinland\n\nenmark\n\nSweden\n\norway\n\nFlemish Region ( E\n\nEstonia\n\nGermany\n\nEngland (U\n\nCanada\n\netherlands\n\nSpain\n\nIreland\n\nItaly\n\napan\n\nFrance\n\nC echia\n\nAustria\n\nUnited States\n\nSlo ak Republic\n\norea\n\nPoland\n\nChile\n\nSingapore\n\nIsrael\n\new ealand\n\nithuania\n\nHungary\n\nAfter accounting for demographic changes efore accounting for demographic changes\n\n30\n\n20\n\n10\n\n0\n\n10\n\n20\n\n30\n\n0\n\n30\n\n20\n\n10\n\n0\n\n10\n\n20\n\n30\n\n0\n\n**. iteracy**\n\n**B. umeracy**\n\nRound, Cycle 1:","pages":[{"line_numbers":[],"markdown":"**104** \n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 3.1. Change in average literacy and numeracy proficiency between cycles, before and after accounting for demographic changes**\n\nDifference in mean proficiency scores between cycles, after reweighting Cycle 2 to match Cycle 1's distribution of age, immigrant background and gender (Cycle 2 _minus_ Cycle 1)\n\nNote: Adults aged 16-65; does not include adults who in Cycle 2 were only administered the doorstep interview due to a language barrier, to maximise the comparability across cycles (see Box 1.1 in Chapter 1 and Box 3.2). Unadjusted differences are the differences between the averages in each cycle. Adjusted differences are based on a method analogous to post-stratification that reweights the samples in Cycle 2 so that the demographic characteristics of these samples match those of the samples in Cycle 1 (see Note 1). Demographic characteristics considered are: age (in ten-year age brackets), gender and immigrant background. Adjusted differences represent a hypothetical scenario of how the proficiency of a population matching the demographic profile of the population in Cycle 1 would have changed over time. Darker colours denote differences that are statistically significant at the 5% level. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the ote for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the unadjusted change in mean literacy proficiency scores._ Source: OECD (2018[4]; 2015[5]; 2012[6]), _Survey of Adult Skills (PIAAC) databases_ , http://www.oecd.org/skills/piaac/publicdataandanalysis/ (accessed on 23 September 2024); Tables A.3.1 (L) and A.3.1 (N) in Annex A.\n\nFinland\n\nenmark\n\nSweden\n\norway\n\nFlemish Region ( E\n\nEstonia\n\nGermany\n\nEngland (U\n\nCanada\n\netherlands\n\nSpain\n\nIreland\n\nItaly\n\napan\n\nFrance\n\nC echia\n\nAustria\n\nUnited States\n\nSlo ak Republic\n\norea\n\nPoland\n\nChile\n\nSingapore\n\nIsrael\n\new ealand\n\nithuania\n\nHungary\n\nFinland\n\nenmark\n\nSweden\n\norway\n\nFlemish Region ( E\n\nEstonia\n\nGermany\n\nEngland (U\n\nCanada\n\netherlands\n\nSpain\n\nIreland\n\nItaly\n\napan\n\nFrance\n\nC echia\n\nAustria\n\nUnited States\n\nSlo ak Republic\n\norea\n\nPoland\n\nChile\n\nSingapore\n\nIsrael\n\new ealand\n\nithuania\n\nHungary\n\nAfter accounting for demographic changes efore accounting for demographic changes\n\n30\n\n20\n\n10\n\n0\n\n10\n\n20\n\n30\n\n0\n\n30\n\n20\n\n10\n\n0\n\n10\n\n20\n\n30\n\n0\n\n**. iteracy**\n\n**B. umeracy**\n\nRound, Cycle 1:","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**104** \n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 3.1. Change in average literacy and numeracy proficiency between cycles, before and after accounting for demographic changes**\n\nDifference in mean proficiency scores between cycles, after reweighting Cycle 2 to match Cycle 1's distribution of age, immigrant background and gender (Cycle 2 _minus_ Cycle 1)\n\nNote: Adults aged 16-65; does not include adults who in Cycle 2 were only administered the doorstep interview due to a language barrier, to maximise the comparability across cycles (see Box 1.1 in Chapter 1 and Box 3.2). Unadjusted differences are the differences between the averages in each cycle. Adjusted differences are based on a method analogous to post-stratification that reweights the samples in Cycle 2 so that the demographic characteristics of these samples match those of the samples in Cycle 1 (see Note 1). Demographic characteristics considered are: age (in ten-year age brackets), gender and immigrant background. Adjusted differences represent a hypothetical scenario of how the proficiency of a population matching the demographic profile of the population in Cycle 1 would have changed over time. Darker colours denote differences that are statistically significant at the 5% level. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the ote for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the unadjusted change in mean literacy proficiency scores._ Source: OECD (2018[4]; 2015[5]; 2012[6]), _Survey of Adult Skills (PIAAC) databases_ , http://www.oecd.org/skills/piaac/publicdataandanalysis/ (accessed on 23 September 2024); Tables A.3.1 (L) and A.3.1 (N) in Annex A.\n\nFinland\n\nenmark\n\nSweden\n\norway\n\nFlemish Region ( E\n\nEstonia\n\nGermany\n\nEngland (U\n\nCanada\n\netherlands\n\nSpain\n\nIreland\n\nItaly\n\napan\n\nFrance\n\nC echia\n\nAustria\n\nUnited States\n\nSlo ak Republic\n\norea\n\nPoland\n\nChile\n\nSingapore\n\nIsrael\n\new ealand\n\nithuania\n\nHungary\n\nFinland\n\nenmark\n\nSweden\n\norway\n\nFlemish Region ( E\n\nEstonia\n\nGermany\n\nEngland (U\n\nCanada\n\netherlands\n\nSpain\n\nIreland\n\nItaly\n\napan\n\nFrance\n\nC echia\n\nAustria\n\nUnited States\n\nSlo ak Republic\n\norea\n\nPoland\n\nChile\n\nSingapore\n\nIsrael\n\new ealand\n\nithuania\n\nHungary\n\nAfter accounting for demographic changes efore accounting for demographic changes\n\n30\n\n20\n\n10\n\n0\n\n10\n\n20\n\n30\n\n0\n\n30\n\n20\n\n10\n\n0\n\n10\n\n20\n\n30\n\n0\n\n**. iteracy**\n\n**B. umeracy**\n\nRound, Cycle 1:","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p106.pdf"]},"response":{"errors":[],"extract_id":"extract_2306222a180cdeb13b2dcd00e44dc789","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**104** \n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 3.1. Change in average literacy and numeracy proficiency between cycles, before and after accounting for demographic changes**\n\nDifference in mean proficiency scores between cycles, after reweighting Cycle 2 to match Cycle 1's distribution of age, immigrant background and gender (Cycle 2 _minus_ Cycle 1)\n\nNote: Adults aged 16-65; does not include adults who in Cycle 2 were only administered the doorstep interview due to a language barrier, to maximise the comparability across cycles (see Box 1.1 in Chapter 1 and Box 3.2). Unadjusted differences are the differences between the averages in each cycle. Adjusted differences are based on a method analogous to post-stratification that reweights the samples in Cycle 2 so that the demographic characteristics of these samples match those of the samples in Cycle 1 (see Note 1). Demographic characteristics considered are: age (in ten-year age brackets), gender and immigrant background. Adjusted differences represent a hypothetical scenario of how the proficiency of a population matching the demographic profile of the population in Cycle 1 would have changed over time. Darker colours denote differences that are statistically significant at the 5% level. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the ote for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the unadjusted change in mean literacy proficiency scores._ Source: OECD (2018[4]; 2015[5]; 2012[6]), _Survey of Adult Skills (PIAAC) databases_ , http://www.oecd.org/skills/piaac/publicdataandanalysis/ (accessed on 23 September 2024); Tables A.3.1 (L) and A.3.1 (N) in Annex A.\n\nFinland\n\nenmark\n\nSweden\n\norway\n\nFlemish Region ( E\n\nEstonia\n\nGermany\n\nEngland (U\n\nCanada\n\netherlands\n\nSpain\n\nIreland\n\nItaly\n\napan\n\nFrance\n\nC echia\n\nAustria\n\nUnited States\n\nSlo ak Republic\n\norea\n\nPoland\n\nChile\n\nSingapore\n\nIsrael\n\new ealand\n\nithuania\n\nHungary\n\nFinland\n\nenmark\n\nSweden\n\norway\n\nFlemish Region ( E\n\nEstonia\n\nGermany\n\nEngland (U\n\nCanada\n\netherlands\n\nSpain\n\nIreland\n\nItaly\n\napan\n\nFrance\n\nC echia\n\nAustria\n\nUnited States\n\nSlo ak Republic\n\norea\n\nPoland\n\nChile\n\nSingapore\n\nIsrael\n\new ealand\n\nithuania\n\nHungary\n\nAfter accounting for demographic changes efore accounting for demographic changes\n\n30\n\n20\n\n10\n\n0\n\n10\n\n20\n\n30\n\n0\n\n30\n\n20\n\n10\n\n0\n\n10\n\n20\n\n30\n\n0\n\n**. iteracy**\n\n**B. umeracy**\n\nRound, Cycle 1:"],"full_content":"**104** \n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 3.1. Change in average literacy and numeracy proficiency between cycles, before and after accounting for demographic changes**\n\nDifference in mean proficiency scores between cycles, after reweighting Cycle 2 to match Cycle 1's distribution of age, immigrant background and gender (Cycle 2 _minus_ Cycle 1)\n\nNote: Adults aged 16-65; does not include adults who in Cycle 2 were only administered the doorstep interview due to a language barrier, to maximise the comparability across cycles (see Box 1.1 in Chapter 1 and Box 3.2). Unadjusted differences are the differences between the averages in each cycle. Adjusted differences are based on a method analogous to post-stratification that reweights the samples in Cycle 2 so that the demographic characteristics of these samples match those of the samples in Cycle 1 (see Note 1). Demographic characteristics considered are: age (in ten-year age brackets), gender and immigrant background. Adjusted differences represent a hypothetical scenario of how the proficiency of a population matching the demographic profile of the population in Cycle 1 would have changed over time. Darker colours denote differences that are statistically significant at the 5% level. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the ote for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the unadjusted change in mean literacy proficiency scores._ Source: OECD (2018[4]; 2015[5]; 2012[6]), _Survey of Adult Skills (PIAAC) databases_ , http://www.oecd.org/skills/piaac/publicdataandanalysis/ (accessed on 23 September 2024); Tables A.3.1 (L) and A.3.1 (N) in Annex A.\n\nFinland\n\nenmark\n\nSweden\n\norway\n\nFlemish Region ( E\n\nEstonia\n\nGermany\n\nEngland (U\n\nCanada\n\netherlands\n\nSpain\n\nIreland\n\nItaly\n\napan\n\nFrance\n\nC echia\n\nAustria\n\nUnited States\n\nSlo ak Republic\n\norea\n\nPoland\n\nChile\n\nSingapore\n\nIsrael\n\new ealand\n\nithuania\n\nHungary\n\nFinland\n\nenmark\n\nSweden\n\norway\n\nFlemish Region ( E\n\nEstonia\n\nGermany\n\nEngland (U\n\nCanada\n\netherlands\n\nSpain\n\nIreland\n\nItaly\n\napan\n\nFrance\n\nC echia\n\nAustria\n\nUnited States\n\nSlo ak Republic\n\norea\n\nPoland\n\nChile\n\nSingapore\n\nIsrael\n\new ealand\n\nithuania\n\nHungary\n\nAfter accounting for demographic changes efore accounting for demographic changes\n\n30\n\n20\n\n10\n\n0\n\n10\n\n20\n\n30\n\n0\n\n30\n\n20\n\n10\n\n0\n\n10\n\n20\n\n30\n\n0\n\n**. iteracy**\n\n**B. umeracy**\n\nRound, Cycle 1:","publish_date":null,"title":"Figure 3.1. Change in average literacy and numeracy proficiency between cycles, before and after accounting for demographic changes","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p106.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p106.pdf"},"request":{"example_id":"chart/b263dc5d-en_p106","product_type":"parse","source_file_path":"b263dc5d-en_p106.pdf"},"started_at":"2026-11-04T19:45:40.852859"} +{"completed_at":"2026-11-04T19:45:43.291884","latency_in_ms":1264,"output":{"example_id":"chart/b263dc5d-en_p116","grounded_pages":[],"layout_pages":[],"markdown":"**114** \n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 3.5. Long-term trends in literacy proficiency**\n\nAverage literacy proficiency scores in IALS and Cycles 1 and 2 of the Survey of Adult Skills\n\nNote: Adults aged 16-65; does not include adults who in Cycle 2 were only administered the doorstep interview due to a language barrier, to maximise the comparability across cycles (see Box 1.1 in Chapter 1 and Box 3.2). \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the ote for Poland in the Reader’s Guide. Source: Statistics Canada (2024[12]), _International Adult Literacy Survey (IALS),_ https://www150.statcan.gc.ca/n1/ (accessed on 23 September 2024); OECD (2018[4]; 2015[5]; 2012[6]), _Survey of Adult Skills (PIAAC) databases_ , http://www.oecd.org/skills/piaac/publicdataandanalysis/ (accessed on 23 September 2024); Table A.3.23 (L) in Annex A.\n\n**mproving egative but flattening improving**\n\n**lat lat decreasing**\n\n310\n\n290\n\n2 0\n\n2 0\n\n230\n\n210\n\n199 1998 2002 200 2010 201 2018 2022\n\nScore\n\near\n\n**ositive flattening**\n\n310\n\n290\n\n2 0\n\n2 0\n\n230\n\n210\n\n199 1998 2002 200 2010 201 2018 2022\n\n1998 2002 200 2010 201 2018 2022 199 1998 2002 200 2010 201 2018 2022 ear 199\n\n199 1998 2002 200 2010 201 2018 2022 ear 199 1998 2002 200 2010 201 2018 2022\n\nScore\n\n310\n\n290\n\n2 0\n\n2 0\n\n230\n\n210\n\nScore\n\n310\n\n290\n\n2 0\n\n2 0\n\n230\n\n210\n\nScore\n\n310\n\n290\n\n2 0\n\n2 0\n\n230\n\n210\n\nScore\n\n310\n\n290\n\n2 0\n\n2 0\n\n230\n\n210\n\nScore\n\n**ump s aped**\n\nFinland\n\nCanada\n\nIreland\n\nC echia\n\netherlands\n\nItaly\n\nEngland (U\n\new ealand\n\nPoland\n\nHungary\n\nFlemish Region ( E\n\nUnited States\n\nChile\n\nenmark\n\norway\n\nGermany\n\nSweden","pages":[{"line_numbers":[],"markdown":"**114** \n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 3.5. Long-term trends in literacy proficiency**\n\nAverage literacy proficiency scores in IALS and Cycles 1 and 2 of the Survey of Adult Skills\n\nNote: Adults aged 16-65; does not include adults who in Cycle 2 were only administered the doorstep interview due to a language barrier, to maximise the comparability across cycles (see Box 1.1 in Chapter 1 and Box 3.2). \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the ote for Poland in the Reader’s Guide. Source: Statistics Canada (2024[12]), _International Adult Literacy Survey (IALS),_ https://www150.statcan.gc.ca/n1/ (accessed on 23 September 2024); OECD (2018[4]; 2015[5]; 2012[6]), _Survey of Adult Skills (PIAAC) databases_ , http://www.oecd.org/skills/piaac/publicdataandanalysis/ (accessed on 23 September 2024); Table A.3.23 (L) in Annex A.\n\n**mproving egative but flattening improving**\n\n**lat lat decreasing**\n\n310\n\n290\n\n2 0\n\n2 0\n\n230\n\n210\n\n199 1998 2002 200 2010 201 2018 2022\n\nScore\n\near\n\n**ositive flattening**\n\n310\n\n290\n\n2 0\n\n2 0\n\n230\n\n210\n\n199 1998 2002 200 2010 201 2018 2022\n\n1998 2002 200 2010 201 2018 2022 199 1998 2002 200 2010 201 2018 2022 ear 199\n\n199 1998 2002 200 2010 201 2018 2022 ear 199 1998 2002 200 2010 201 2018 2022\n\nScore\n\n310\n\n290\n\n2 0\n\n2 0\n\n230\n\n210\n\nScore\n\n310\n\n290\n\n2 0\n\n2 0\n\n230\n\n210\n\nScore\n\n310\n\n290\n\n2 0\n\n2 0\n\n230\n\n210\n\nScore\n\n310\n\n290\n\n2 0\n\n2 0\n\n230\n\n210\n\nScore\n\n**ump s aped**\n\nFinland\n\nCanada\n\nIreland\n\nC echia\n\netherlands\n\nItaly\n\nEngland (U\n\new ealand\n\nPoland\n\nHungary\n\nFlemish Region ( E\n\nUnited States\n\nChile\n\nenmark\n\norway\n\nGermany\n\nSweden","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**114** \n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 3.5. Long-term trends in literacy proficiency**\n\nAverage literacy proficiency scores in IALS and Cycles 1 and 2 of the Survey of Adult Skills\n\nNote: Adults aged 16-65; does not include adults who in Cycle 2 were only administered the doorstep interview due to a language barrier, to maximise the comparability across cycles (see Box 1.1 in Chapter 1 and Box 3.2). \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the ote for Poland in the Reader’s Guide. Source: Statistics Canada (2024[12]), _International Adult Literacy Survey (IALS),_ https://www150.statcan.gc.ca/n1/ (accessed on 23 September 2024); OECD (2018[4]; 2015[5]; 2012[6]), _Survey of Adult Skills (PIAAC) databases_ , http://www.oecd.org/skills/piaac/publicdataandanalysis/ (accessed on 23 September 2024); Table A.3.23 (L) in Annex A.\n\n**mproving egative but flattening improving**\n\n**lat lat decreasing**\n\n310\n\n290\n\n2 0\n\n2 0\n\n230\n\n210\n\n199 1998 2002 200 2010 201 2018 2022\n\nScore\n\near\n\n**ositive flattening**\n\n310\n\n290\n\n2 0\n\n2 0\n\n230\n\n210\n\n199 1998 2002 200 2010 201 2018 2022\n\n1998 2002 200 2010 201 2018 2022 199 1998 2002 200 2010 201 2018 2022 ear 199\n\n199 1998 2002 200 2010 201 2018 2022 ear 199 1998 2002 200 2010 201 2018 2022\n\nScore\n\n310\n\n290\n\n2 0\n\n2 0\n\n230\n\n210\n\nScore\n\n310\n\n290\n\n2 0\n\n2 0\n\n230\n\n210\n\nScore\n\n310\n\n290\n\n2 0\n\n2 0\n\n230\n\n210\n\nScore\n\n310\n\n290\n\n2 0\n\n2 0\n\n230\n\n210\n\nScore\n\n**ump s aped**\n\nFinland\n\nCanada\n\nIreland\n\nC echia\n\netherlands\n\nItaly\n\nEngland (U\n\new ealand\n\nPoland\n\nHungary\n\nFlemish Region ( E\n\nUnited States\n\nChile\n\nenmark\n\norway\n\nGermany\n\nSweden","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p116.pdf"]},"response":{"errors":[],"extract_id":"extract_7ccaaf9b334b76d9a995baafcb0ae507","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**114** \n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 3.5. Long-term trends in literacy proficiency**\n\nAverage literacy proficiency scores in IALS and Cycles 1 and 2 of the Survey of Adult Skills\n\nNote: Adults aged 16-65; does not include adults who in Cycle 2 were only administered the doorstep interview due to a language barrier, to maximise the comparability across cycles (see Box 1.1 in Chapter 1 and Box 3.2). \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the ote for Poland in the Reader’s Guide. Source: Statistics Canada (2024[12]), _International Adult Literacy Survey (IALS),_ https://www150.statcan.gc.ca/n1/ (accessed on 23 September 2024); OECD (2018[4]; 2015[5]; 2012[6]), _Survey of Adult Skills (PIAAC) databases_ , http://www.oecd.org/skills/piaac/publicdataandanalysis/ (accessed on 23 September 2024); Table A.3.23 (L) in Annex A.\n\n**mproving egative but flattening improving**\n\n**lat lat decreasing**\n\n310\n\n290\n\n2 0\n\n2 0\n\n230\n\n210\n\n199 1998 2002 200 2010 201 2018 2022\n\nScore\n\near\n\n**ositive flattening**\n\n310\n\n290\n\n2 0\n\n2 0\n\n230\n\n210\n\n199 1998 2002 200 2010 201 2018 2022\n\n1998 2002 200 2010 201 2018 2022 199 1998 2002 200 2010 201 2018 2022 ear 199\n\n199 1998 2002 200 2010 201 2018 2022 ear 199 1998 2002 200 2010 201 2018 2022\n\nScore\n\n310\n\n290\n\n2 0\n\n2 0\n\n230\n\n210\n\nScore\n\n310\n\n290\n\n2 0\n\n2 0\n\n230\n\n210\n\nScore\n\n310\n\n290\n\n2 0\n\n2 0\n\n230\n\n210\n\nScore\n\n310\n\n290\n\n2 0\n\n2 0\n\n230\n\n210\n\nScore\n\n**ump s aped**\n\nFinland\n\nCanada\n\nIreland\n\nC echia\n\netherlands\n\nItaly\n\nEngland (U\n\new ealand\n\nPoland\n\nHungary\n\nFlemish Region ( E\n\nUnited States\n\nChile\n\nenmark\n\norway\n\nGermany\n\nSweden"],"full_content":"**114** \n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 3.5. Long-term trends in literacy proficiency**\n\nAverage literacy proficiency scores in IALS and Cycles 1 and 2 of the Survey of Adult Skills\n\nNote: Adults aged 16-65; does not include adults who in Cycle 2 were only administered the doorstep interview due to a language barrier, to maximise the comparability across cycles (see Box 1.1 in Chapter 1 and Box 3.2). \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the ote for Poland in the Reader’s Guide. Source: Statistics Canada (2024[12]), _International Adult Literacy Survey (IALS),_ https://www150.statcan.gc.ca/n1/ (accessed on 23 September 2024); OECD (2018[4]; 2015[5]; 2012[6]), _Survey of Adult Skills (PIAAC) databases_ , http://www.oecd.org/skills/piaac/publicdataandanalysis/ (accessed on 23 September 2024); Table A.3.23 (L) in Annex A.\n\n**mproving egative but flattening improving**\n\n**lat lat decreasing**\n\n310\n\n290\n\n2 0\n\n2 0\n\n230\n\n210\n\n199 1998 2002 200 2010 201 2018 2022\n\nScore\n\near\n\n**ositive flattening**\n\n310\n\n290\n\n2 0\n\n2 0\n\n230\n\n210\n\n199 1998 2002 200 2010 201 2018 2022\n\n1998 2002 200 2010 201 2018 2022 199 1998 2002 200 2010 201 2018 2022 ear 199\n\n199 1998 2002 200 2010 201 2018 2022 ear 199 1998 2002 200 2010 201 2018 2022\n\nScore\n\n310\n\n290\n\n2 0\n\n2 0\n\n230\n\n210\n\nScore\n\n310\n\n290\n\n2 0\n\n2 0\n\n230\n\n210\n\nScore\n\n310\n\n290\n\n2 0\n\n2 0\n\n230\n\n210\n\nScore\n\n310\n\n290\n\n2 0\n\n2 0\n\n230\n\n210\n\nScore\n\n**ump s aped**\n\nFinland\n\nCanada\n\nIreland\n\nC echia\n\netherlands\n\nItaly\n\nEngland (U\n\new ealand\n\nPoland\n\nHungary\n\nFlemish Region ( E\n\nUnited States\n\nChile\n\nenmark\n\norway\n\nGermany\n\nSweden","publish_date":null,"title":"Figure 3.5. Long-term trends in literacy proficiency","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p116.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p116.pdf"},"request":{"example_id":"chart/b263dc5d-en_p116","product_type":"parse","source_file_path":"b263dc5d-en_p116.pdf"},"started_at":"2026-11-04T19:45:42.026926"} +{"completed_at":"2026-11-04T19:45:44.774045","latency_in_ms":1480,"output":{"example_id":"chart/b263dc5d-en_p118","grounded_pages":[],"layout_pages":[],"markdown":"**116** \n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 3.6. Change in educational attainment of the adult population between cycles**\n\nDifference in the shares of adults with below upper secondary, upper secondary and tertiary education (Cycle 2 _minus_ Cycle 1); 25-65 year-olds\n\nNote: Does not include adults who in Cycle 2 were only administered the doorstep interview due to a language barrier, to maximise the comparability across cycles (see Box 1.1 in Chapter 1 and Box 3.2). See Note 2 for an explanation of the classification of educational attainment. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the Note for Poland in the Reader’s Guide. _Countries and economies are ranked in ascending order of the change in the share of tertiary-educated adults._ Source: OECD (2018[4]; 2015[5]; 2012[6]), _Survey of Adult Skills (PIAAC) databases_ , http://www.oecd.org/skills/piaac/publicdataandanalysis/ (accessed on 23 September 2024); Table B.3.1 (Trend) in Annex B.\n\n_Changes in skills proficiency by educational attainment_\n\nAcross the countries and economies participating in both cycles of the Survey of Adult Skills, declines in literacy proficiency were most widespread among low-educated adults (Figure 3.7). The average proficiency of adults with below upper secondary education significantly decreased between the two cycles in all countries and economies except in Canada, Denmark, Finland, Israel, Spain and Sweden. These decreases ranged from 12 points in the Flemish Region (Belgium) to 46 points in Lithuania. The proficiency of adults with upper secondary education fell in 19 countries and economies, remained unchanged in 7, and increased in Finland. Among tertiary-educated adults, average proficiency fell in 13 countries and economies, and only increased in Finland.\n\nChanges in numeracy followed a similar pattern, with declines in numeracy proficiency being more common among low-educated adults (see Table A.3.11 (N) in Annex A). Thirteen countries and economies showed a significant decrease in the average numeracy score of this group, while only Singapore recorded an increase. The average numeracy proficiency of tertiary-educated adults increased only in Canada, Estonia and Finland, and declined in Hungary, Israel, Italy, Korea, Lithuania, New Zealand, Poland and the Slovak Republic. Particularly large declines in the numeracy proficiency (over 20 points) of tertiary- educated adults were found in Lithuania, Poland and the Slovak Republic.\n\nThe widespread decline in proficiency among low-educated adults has widened disparities in proficiency by educational background in most of the participating countries and economies. Austria, France, Japan, New Zealand and Norway saw large increases in the proficiency gap between adults with below upper secondary and tertiary education (Figure 3.8) In contrast, in Singapore, which has the largest inequality in literacy by educational attainment, the gap changed less because proficiency declined among both low- educated and highly educated adults. The gap narrowed in Poland due to declining literacy proficiency among tertiary-educated adults (Figure 3.7).\n\norea\n\nIreland\n\nFlemish Region ( E\n\nenmark\n\netherlands\n\nFrance\n\nEngland (U\n\nSpain\n\norway\n\nEstonia\n\nSweden\n\nGermany\n\nUnited States\n\nAustria\n\nSlo ak Republic\n\nFinland\n\nItaly\n\nC echia\n\napan\n\nCanada\n\nPoland\n\nithuania\n\nSingapore\n\nChile\n\nIsrael\n\new ealand\n\nHungary\n\nPercentage point difference 20\n\n10\n\n0\n\n10\n\n20\n\nUpper secondary Tertiary elow upper secondary\n\nRound, Cycle 1:","pages":[{"line_numbers":[],"markdown":"**116** \n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 3.6. Change in educational attainment of the adult population between cycles**\n\nDifference in the shares of adults with below upper secondary, upper secondary and tertiary education (Cycle 2 _minus_ Cycle 1); 25-65 year-olds\n\nNote: Does not include adults who in Cycle 2 were only administered the doorstep interview due to a language barrier, to maximise the comparability across cycles (see Box 1.1 in Chapter 1 and Box 3.2). See Note 2 for an explanation of the classification of educational attainment. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the Note for Poland in the Reader’s Guide. _Countries and economies are ranked in ascending order of the change in the share of tertiary-educated adults._ Source: OECD (2018[4]; 2015[5]; 2012[6]), _Survey of Adult Skills (PIAAC) databases_ , http://www.oecd.org/skills/piaac/publicdataandanalysis/ (accessed on 23 September 2024); Table B.3.1 (Trend) in Annex B.\n\n_Changes in skills proficiency by educational attainment_\n\nAcross the countries and economies participating in both cycles of the Survey of Adult Skills, declines in literacy proficiency were most widespread among low-educated adults (Figure 3.7). The average proficiency of adults with below upper secondary education significantly decreased between the two cycles in all countries and economies except in Canada, Denmark, Finland, Israel, Spain and Sweden. These decreases ranged from 12 points in the Flemish Region (Belgium) to 46 points in Lithuania. The proficiency of adults with upper secondary education fell in 19 countries and economies, remained unchanged in 7, and increased in Finland. Among tertiary-educated adults, average proficiency fell in 13 countries and economies, and only increased in Finland.\n\nChanges in numeracy followed a similar pattern, with declines in numeracy proficiency being more common among low-educated adults (see Table A.3.11 (N) in Annex A). Thirteen countries and economies showed a significant decrease in the average numeracy score of this group, while only Singapore recorded an increase. The average numeracy proficiency of tertiary-educated adults increased only in Canada, Estonia and Finland, and declined in Hungary, Israel, Italy, Korea, Lithuania, New Zealand, Poland and the Slovak Republic. Particularly large declines in the numeracy proficiency (over 20 points) of tertiary- educated adults were found in Lithuania, Poland and the Slovak Republic.\n\nThe widespread decline in proficiency among low-educated adults has widened disparities in proficiency by educational background in most of the participating countries and economies. Austria, France, Japan, New Zealand and Norway saw large increases in the proficiency gap between adults with below upper secondary and tertiary education (Figure 3.8) In contrast, in Singapore, which has the largest inequality in literacy by educational attainment, the gap changed less because proficiency declined among both low- educated and highly educated adults. The gap narrowed in Poland due to declining literacy proficiency among tertiary-educated adults (Figure 3.7).\n\norea\n\nIreland\n\nFlemish Region ( E\n\nenmark\n\netherlands\n\nFrance\n\nEngland (U\n\nSpain\n\norway\n\nEstonia\n\nSweden\n\nGermany\n\nUnited States\n\nAustria\n\nSlo ak Republic\n\nFinland\n\nItaly\n\nC echia\n\napan\n\nCanada\n\nPoland\n\nithuania\n\nSingapore\n\nChile\n\nIsrael\n\new ealand\n\nHungary\n\nPercentage point difference 20\n\n10\n\n0\n\n10\n\n20\n\nUpper secondary Tertiary elow upper secondary\n\nRound, Cycle 1:","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**116** \n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 3.6. Change in educational attainment of the adult population between cycles**\n\nDifference in the shares of adults with below upper secondary, upper secondary and tertiary education (Cycle 2 _minus_ Cycle 1); 25-65 year-olds\n\nNote: Does not include adults who in Cycle 2 were only administered the doorstep interview due to a language barrier, to maximise the comparability across cycles (see Box 1.1 in Chapter 1 and Box 3.2). See Note 2 for an explanation of the classification of educational attainment. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the Note for Poland in the Reader’s Guide. _Countries and economies are ranked in ascending order of the change in the share of tertiary-educated adults._ Source: OECD (2018[4]; 2015[5]; 2012[6]), _Survey of Adult Skills (PIAAC) databases_ , http://www.oecd.org/skills/piaac/publicdataandanalysis/ (accessed on 23 September 2024); Table B.3.1 (Trend) in Annex B.\n\n_Changes in skills proficiency by educational attainment_\n\nAcross the countries and economies participating in both cycles of the Survey of Adult Skills, declines in literacy proficiency were most widespread among low-educated adults (Figure 3.7). The average proficiency of adults with below upper secondary education significantly decreased between the two cycles in all countries and economies except in Canada, Denmark, Finland, Israel, Spain and Sweden. These decreases ranged from 12 points in the Flemish Region (Belgium) to 46 points in Lithuania. The proficiency of adults with upper secondary education fell in 19 countries and economies, remained unchanged in 7, and increased in Finland. Among tertiary-educated adults, average proficiency fell in 13 countries and economies, and only increased in Finland.\n\nChanges in numeracy followed a similar pattern, with declines in numeracy proficiency being more common among low-educated adults (see Table A.3.11 (N) in Annex A). Thirteen countries and economies showed a significant decrease in the average numeracy score of this group, while only Singapore recorded an increase. The average numeracy proficiency of tertiary-educated adults increased only in Canada, Estonia and Finland, and declined in Hungary, Israel, Italy, Korea, Lithuania, New Zealand, Poland and the Slovak Republic. Particularly large declines in the numeracy proficiency (over 20 points) of tertiary- educated adults were found in Lithuania, Poland and the Slovak Republic.\n\nThe widespread decline in proficiency among low-educated adults has widened disparities in proficiency by educational background in most of the participating countries and economies. Austria, France, Japan, New Zealand and Norway saw large increases in the proficiency gap between adults with below upper secondary and tertiary education (Figure 3.8) In contrast, in Singapore, which has the largest inequality in literacy by educational attainment, the gap changed less because proficiency declined among both low- educated and highly educated adults. The gap narrowed in Poland due to declining literacy proficiency among tertiary-educated adults (Figure 3.7).\n\norea\n\nIreland\n\nFlemish Region ( E\n\nenmark\n\netherlands\n\nFrance\n\nEngland (U\n\nSpain\n\norway\n\nEstonia\n\nSweden\n\nGermany\n\nUnited States\n\nAustria\n\nSlo ak Republic\n\nFinland\n\nItaly\n\nC echia\n\napan\n\nCanada\n\nPoland\n\nithuania\n\nSingapore\n\nChile\n\nIsrael\n\new ealand\n\nHungary\n\nPercentage point difference 20\n\n10\n\n0\n\n10\n\n20\n\nUpper secondary Tertiary elow upper secondary\n\nRound, Cycle 1:","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p118.pdf"]},"response":{"errors":[],"extract_id":"extract_9419402f4fb9d537162922a680944f2d","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**116** \n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 3.6. Change in educational attainment of the adult population between cycles**\n\nDifference in the shares of adults with below upper secondary, upper secondary and tertiary education (Cycle 2 _minus_ Cycle 1); 25-65 year-olds\n\nNote: Does not include adults who in Cycle 2 were only administered the doorstep interview due to a language barrier, to maximise the comparability across cycles (see Box 1.1 in Chapter 1 and Box 3.2). See Note 2 for an explanation of the classification of educational attainment. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the Note for Poland in the Reader’s Guide. _Countries and economies are ranked in ascending order of the change in the share of tertiary-educated adults._ Source: OECD (2018[4]; 2015[5]; 2012[6]), _Survey of Adult Skills (PIAAC) databases_ , http://www.oecd.org/skills/piaac/publicdataandanalysis/ (accessed on 23 September 2024); Table B.3.1 (Trend) in Annex B.\n\n_Changes in skills proficiency by educational attainment_\n\nAcross the countries and economies participating in both cycles of the Survey of Adult Skills, declines in literacy proficiency were most widespread among low-educated adults (Figure 3.7). The average proficiency of adults with below upper secondary education significantly decreased between the two cycles in all countries and economies except in Canada, Denmark, Finland, Israel, Spain and Sweden. These decreases ranged from 12 points in the Flemish Region (Belgium) to 46 points in Lithuania. The proficiency of adults with upper secondary education fell in 19 countries and economies, remained unchanged in 7, and increased in Finland. Among tertiary-educated adults, average proficiency fell in 13 countries and economies, and only increased in Finland.\n\nChanges in numeracy followed a similar pattern, with declines in numeracy proficiency being more common among low-educated adults (see Table A.3.11 (N) in Annex A). Thirteen countries and economies showed a significant decrease in the average numeracy score of this group, while only Singapore recorded an increase. The average numeracy proficiency of tertiary-educated adults increased only in Canada, Estonia and Finland, and declined in Hungary, Israel, Italy, Korea, Lithuania, New Zealand, Poland and the Slovak Republic. Particularly large declines in the numeracy proficiency (over 20 points) of tertiary- educated adults were found in Lithuania, Poland and the Slovak Republic.\n\nThe widespread decline in proficiency among low-educated adults has widened disparities in proficiency by educational background in most of the participating countries and economies. Austria, France, Japan, New Zealand and Norway saw large increases in the proficiency gap between adults with below upper secondary and tertiary education (Figure 3.8) In contrast, in Singapore, which has the largest inequality in literacy by educational attainment, the gap changed less because proficiency declined among both low- educated and highly educated adults. The gap narrowed in Poland due to declining literacy proficiency among tertiary-educated adults (Figure 3.7).\n\norea\n\nIreland\n\nFlemish Region ( E\n\nenmark\n\netherlands\n\nFrance\n\nEngland (U\n\nSpain\n\norway\n\nEstonia\n\nSweden\n\nGermany\n\nUnited States\n\nAustria\n\nSlo ak Republic\n\nFinland\n\nItaly\n\nC echia\n\napan\n\nCanada\n\nPoland\n\nithuania\n\nSingapore\n\nChile\n\nIsrael\n\new ealand\n\nHungary\n\nPercentage point difference 20\n\n10\n\n0\n\n10\n\n20\n\nUpper secondary Tertiary elow upper secondary\n\nRound, Cycle 1:"],"full_content":"**116** \n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 3.6. Change in educational attainment of the adult population between cycles**\n\nDifference in the shares of adults with below upper secondary, upper secondary and tertiary education (Cycle 2 _minus_ Cycle 1); 25-65 year-olds\n\nNote: Does not include adults who in Cycle 2 were only administered the doorstep interview due to a language barrier, to maximise the comparability across cycles (see Box 1.1 in Chapter 1 and Box 3.2). See Note 2 for an explanation of the classification of educational attainment. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the Note for Poland in the Reader’s Guide. _Countries and economies are ranked in ascending order of the change in the share of tertiary-educated adults._ Source: OECD (2018[4]; 2015[5]; 2012[6]), _Survey of Adult Skills (PIAAC) databases_ , http://www.oecd.org/skills/piaac/publicdataandanalysis/ (accessed on 23 September 2024); Table B.3.1 (Trend) in Annex B.\n\n_Changes in skills proficiency by educational attainment_\n\nAcross the countries and economies participating in both cycles of the Survey of Adult Skills, declines in literacy proficiency were most widespread among low-educated adults (Figure 3.7). The average proficiency of adults with below upper secondary education significantly decreased between the two cycles in all countries and economies except in Canada, Denmark, Finland, Israel, Spain and Sweden. These decreases ranged from 12 points in the Flemish Region (Belgium) to 46 points in Lithuania. The proficiency of adults with upper secondary education fell in 19 countries and economies, remained unchanged in 7, and increased in Finland. Among tertiary-educated adults, average proficiency fell in 13 countries and economies, and only increased in Finland.\n\nChanges in numeracy followed a similar pattern, with declines in numeracy proficiency being more common among low-educated adults (see Table A.3.11 (N) in Annex A). Thirteen countries and economies showed a significant decrease in the average numeracy score of this group, while only Singapore recorded an increase. The average numeracy proficiency of tertiary-educated adults increased only in Canada, Estonia and Finland, and declined in Hungary, Israel, Italy, Korea, Lithuania, New Zealand, Poland and the Slovak Republic. Particularly large declines in the numeracy proficiency (over 20 points) of tertiary- educated adults were found in Lithuania, Poland and the Slovak Republic.\n\nThe widespread decline in proficiency among low-educated adults has widened disparities in proficiency by educational background in most of the participating countries and economies. Austria, France, Japan, New Zealand and Norway saw large increases in the proficiency gap between adults with below upper secondary and tertiary education (Figure 3.8) In contrast, in Singapore, which has the largest inequality in literacy by educational attainment, the gap changed less because proficiency declined among both low- educated and highly educated adults. The gap narrowed in Poland due to declining literacy proficiency among tertiary-educated adults (Figure 3.7).\n\norea\n\nIreland\n\nFlemish Region ( E\n\nenmark\n\netherlands\n\nFrance\n\nEngland (U\n\nSpain\n\norway\n\nEstonia\n\nSweden\n\nGermany\n\nUnited States\n\nAustria\n\nSlo ak Republic\n\nFinland\n\nItaly\n\nC echia\n\napan\n\nCanada\n\nPoland\n\nithuania\n\nSingapore\n\nChile\n\nIsrael\n\new ealand\n\nHungary\n\nPercentage point difference 20\n\n10\n\n0\n\n10\n\n20\n\nUpper secondary Tertiary elow upper secondary\n\nRound, Cycle 1:","publish_date":null,"title":"Figure 3.6. Change in educational attainment of the adult population between cycles","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p118.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p118.pdf"},"request":{"example_id":"chart/b263dc5d-en_p118","product_type":"parse","source_file_path":"b263dc5d-en_p118.pdf"},"started_at":"2026-11-04T19:45:43.293878"} +{"completed_at":"2026-11-04T19:45:46.800473","latency_in_ms":1529,"output":{"example_id":"chart/b263dc5d-en_p119","grounded_pages":[],"layout_pages":[],"markdown":" **117**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 3.7. Change in literacy proficiency between cycles, by educational attainment**\n\nAdjusted and unadjusted difference in mean literacy scores between cycles (Cycle 2 _minus_ Cycle 1); 25-65 year- olds\n\nNote: Does not include adults who in Cycle 2 were only administered the doorstep interview due to a language barrier, to maximise the comparability across cycles (see Box 1.1 in Chapter 1 and Box 3.2). See Note 2 for an explanation of the classification of educational attainment. Unadjusted differences are the differences between the averages in each cycle. Adjusted differences are based on a method analogous to post- stratification that reweights the samples in Cycle 2 so that education groups have the same demographic characteristics as in Cycle 1 (see Note 1). Demographic characteristics considered are: age (in ten-year age brackets), gender and immigrant background. Darker colours denote differences that are statistically significant at the 5% level. In Chile and the United States, more than 20% of the subsample of adults with below upper secondary education only completed the assessment of reading and numeracy components in the second cycle. As the methodology for estimating the proficiency of adults who only took the components assessment has changed between the cycles, the change in proficiency for this group is not reported. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the Note for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the unadjusted change in literacy proficiency among adults with below upper secondary education._ Source: OECD (2018[4]; 2015[5]; 2012[6]), _Survey of Adult Skills (PIAAC) databases_ , http://www.oecd.org/skills/piaac/publicdataandanalysis/ (accessed on 23 September 2024); Table A.3.11 (L) in Annex A.\n\nSweden\n\nFinland\n\nSpain\n\nenmark\n\nCanada\n\nFlemish Region ( E\n\nEstonia\n\nItaly\n\netherlands\n\nGermany\n\nC echia\n\nSlo ak Republic\n\nIreland\n\nEngland (U\n\norway\n\nPoland\n\nFrance\n\napan\n\norea\n\nAustria\n\nUnited States\n\nIsrael\n\nSingapore\n\new ealand\n\nithuania\n\nChile\n\nHungary\n\nSweden\n\nFinland\n\nSpain\n\nenmark\n\nCanada\n\nFlemish Region ( E\n\nEstonia\n\nItaly\n\netherlands\n\nGermany\n\nC echia\n\nSlo ak Republic\n\nIreland\n\nEngland (U\n\norway\n\nPoland\n\nFrance\n\napan\n\norea\n\nAustria\n\nUnited States\n\nIsrael\n\nSingapore\n\new ealand\n\nithuania\n\nChile\n\nHungary\n\nAd usted Unad u sted\n\n20\n\n10\n\n0\n\n10\n\n20\n\n30\n\n0\n\n0\n\n20\n\n10\n\n0\n\n10\n\n20\n\n30\n\n0\n\n0\n\n20\n\n10\n\n0\n\n10\n\n20\n\n30\n\n0\n\n0\n\n**ertiary**\n\n**pper secondary**\n\n**Belo upper secondary**\n\nRound, Cycle 1:\n\nRound, Cycle 1:","pages":[{"line_numbers":[],"markdown":" **117**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 3.7. Change in literacy proficiency between cycles, by educational attainment**\n\nAdjusted and unadjusted difference in mean literacy scores between cycles (Cycle 2 _minus_ Cycle 1); 25-65 year- olds\n\nNote: Does not include adults who in Cycle 2 were only administered the doorstep interview due to a language barrier, to maximise the comparability across cycles (see Box 1.1 in Chapter 1 and Box 3.2). See Note 2 for an explanation of the classification of educational attainment. Unadjusted differences are the differences between the averages in each cycle. Adjusted differences are based on a method analogous to post- stratification that reweights the samples in Cycle 2 so that education groups have the same demographic characteristics as in Cycle 1 (see Note 1). Demographic characteristics considered are: age (in ten-year age brackets), gender and immigrant background. Darker colours denote differences that are statistically significant at the 5% level. In Chile and the United States, more than 20% of the subsample of adults with below upper secondary education only completed the assessment of reading and numeracy components in the second cycle. As the methodology for estimating the proficiency of adults who only took the components assessment has changed between the cycles, the change in proficiency for this group is not reported. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the Note for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the unadjusted change in literacy proficiency among adults with below upper secondary education._ Source: OECD (2018[4]; 2015[5]; 2012[6]), _Survey of Adult Skills (PIAAC) databases_ , http://www.oecd.org/skills/piaac/publicdataandanalysis/ (accessed on 23 September 2024); Table A.3.11 (L) in Annex A.\n\nSweden\n\nFinland\n\nSpain\n\nenmark\n\nCanada\n\nFlemish Region ( E\n\nEstonia\n\nItaly\n\netherlands\n\nGermany\n\nC echia\n\nSlo ak Republic\n\nIreland\n\nEngland (U\n\norway\n\nPoland\n\nFrance\n\napan\n\norea\n\nAustria\n\nUnited States\n\nIsrael\n\nSingapore\n\new ealand\n\nithuania\n\nChile\n\nHungary\n\nSweden\n\nFinland\n\nSpain\n\nenmark\n\nCanada\n\nFlemish Region ( E\n\nEstonia\n\nItaly\n\netherlands\n\nGermany\n\nC echia\n\nSlo ak Republic\n\nIreland\n\nEngland (U\n\norway\n\nPoland\n\nFrance\n\napan\n\norea\n\nAustria\n\nUnited States\n\nIsrael\n\nSingapore\n\new ealand\n\nithuania\n\nChile\n\nHungary\n\nAd usted Unad u sted\n\n20\n\n10\n\n0\n\n10\n\n20\n\n30\n\n0\n\n0\n\n20\n\n10\n\n0\n\n10\n\n20\n\n30\n\n0\n\n0\n\n20\n\n10\n\n0\n\n10\n\n20\n\n30\n\n0\n\n0\n\n**ertiary**\n\n**pper secondary**\n\n**Belo upper secondary**\n\nRound, Cycle 1:\n\nRound, Cycle 1:","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":" **117**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 3.7. Change in literacy proficiency between cycles, by educational attainment**\n\nAdjusted and unadjusted difference in mean literacy scores between cycles (Cycle 2 _minus_ Cycle 1); 25-65 year- olds\n\nNote: Does not include adults who in Cycle 2 were only administered the doorstep interview due to a language barrier, to maximise the comparability across cycles (see Box 1.1 in Chapter 1 and Box 3.2). See Note 2 for an explanation of the classification of educational attainment. Unadjusted differences are the differences between the averages in each cycle. Adjusted differences are based on a method analogous to post- stratification that reweights the samples in Cycle 2 so that education groups have the same demographic characteristics as in Cycle 1 (see Note 1). Demographic characteristics considered are: age (in ten-year age brackets), gender and immigrant background. Darker colours denote differences that are statistically significant at the 5% level. In Chile and the United States, more than 20% of the subsample of adults with below upper secondary education only completed the assessment of reading and numeracy components in the second cycle. As the methodology for estimating the proficiency of adults who only took the components assessment has changed between the cycles, the change in proficiency for this group is not reported. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the Note for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the unadjusted change in literacy proficiency among adults with below upper secondary education._ Source: OECD (2018[4]; 2015[5]; 2012[6]), _Survey of Adult Skills (PIAAC) databases_ , http://www.oecd.org/skills/piaac/publicdataandanalysis/ (accessed on 23 September 2024); Table A.3.11 (L) in Annex A.\n\nSweden\n\nFinland\n\nSpain\n\nenmark\n\nCanada\n\nFlemish Region ( E\n\nEstonia\n\nItaly\n\netherlands\n\nGermany\n\nC echia\n\nSlo ak Republic\n\nIreland\n\nEngland (U\n\norway\n\nPoland\n\nFrance\n\napan\n\norea\n\nAustria\n\nUnited States\n\nIsrael\n\nSingapore\n\new ealand\n\nithuania\n\nChile\n\nHungary\n\nSweden\n\nFinland\n\nSpain\n\nenmark\n\nCanada\n\nFlemish Region ( E\n\nEstonia\n\nItaly\n\netherlands\n\nGermany\n\nC echia\n\nSlo ak Republic\n\nIreland\n\nEngland (U\n\norway\n\nPoland\n\nFrance\n\napan\n\norea\n\nAustria\n\nUnited States\n\nIsrael\n\nSingapore\n\new ealand\n\nithuania\n\nChile\n\nHungary\n\nAd usted Unad u sted\n\n20\n\n10\n\n0\n\n10\n\n20\n\n30\n\n0\n\n0\n\n20\n\n10\n\n0\n\n10\n\n20\n\n30\n\n0\n\n0\n\n20\n\n10\n\n0\n\n10\n\n20\n\n30\n\n0\n\n0\n\n**ertiary**\n\n**pper secondary**\n\n**Belo upper secondary**\n\nRound, Cycle 1:\n\nRound, Cycle 1:","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p119.pdf"]},"response":{"errors":[],"extract_id":"extract_b0d914520f647da10fece7f073e2e0ee","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":[" **117**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 3.7. Change in literacy proficiency between cycles, by educational attainment**\n\nAdjusted and unadjusted difference in mean literacy scores between cycles (Cycle 2 _minus_ Cycle 1); 25-65 year- olds\n\nNote: Does not include adults who in Cycle 2 were only administered the doorstep interview due to a language barrier, to maximise the comparability across cycles (see Box 1.1 in Chapter 1 and Box 3.2). See Note 2 for an explanation of the classification of educational attainment. Unadjusted differences are the differences between the averages in each cycle. Adjusted differences are based on a method analogous to post- stratification that reweights the samples in Cycle 2 so that education groups have the same demographic characteristics as in Cycle 1 (see Note 1). Demographic characteristics considered are: age (in ten-year age brackets), gender and immigrant background. Darker colours denote differences that are statistically significant at the 5% level. In Chile and the United States, more than 20% of the subsample of adults with below upper secondary education only completed the assessment of reading and numeracy components in the second cycle. As the methodology for estimating the proficiency of adults who only took the components assessment has changed between the cycles, the change in proficiency for this group is not reported. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the Note for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the unadjusted change in literacy proficiency among adults with below upper secondary education._ Source: OECD (2018[4]; 2015[5]; 2012[6]), _Survey of Adult Skills (PIAAC) databases_ , http://www.oecd.org/skills/piaac/publicdataandanalysis/ (accessed on 23 September 2024); Table A.3.11 (L) in Annex A.\n\nSweden\n\nFinland\n\nSpain\n\nenmark\n\nCanada\n\nFlemish Region ( E\n\nEstonia\n\nItaly\n\netherlands\n\nGermany\n\nC echia\n\nSlo ak Republic\n\nIreland\n\nEngland (U\n\norway\n\nPoland\n\nFrance\n\napan\n\norea\n\nAustria\n\nUnited States\n\nIsrael\n\nSingapore\n\new ealand\n\nithuania\n\nChile\n\nHungary\n\nSweden\n\nFinland\n\nSpain\n\nenmark\n\nCanada\n\nFlemish Region ( E\n\nEstonia\n\nItaly\n\netherlands\n\nGermany\n\nC echia\n\nSlo ak Republic\n\nIreland\n\nEngland (U\n\norway\n\nPoland\n\nFrance\n\napan\n\norea\n\nAustria\n\nUnited States\n\nIsrael\n\nSingapore\n\new ealand\n\nithuania\n\nChile\n\nHungary\n\nAd usted Unad u sted\n\n20\n\n10\n\n0\n\n10\n\n20\n\n30\n\n0\n\n0\n\n20\n\n10\n\n0\n\n10\n\n20\n\n30\n\n0\n\n0\n\n20\n\n10\n\n0\n\n10\n\n20\n\n30\n\n0\n\n0\n\n**ertiary**\n\n**pper secondary**\n\n**Belo upper secondary**\n\nRound, Cycle 1:\n\nRound, Cycle 1:"],"full_content":" **117**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 3.7. Change in literacy proficiency between cycles, by educational attainment**\n\nAdjusted and unadjusted difference in mean literacy scores between cycles (Cycle 2 _minus_ Cycle 1); 25-65 year- olds\n\nNote: Does not include adults who in Cycle 2 were only administered the doorstep interview due to a language barrier, to maximise the comparability across cycles (see Box 1.1 in Chapter 1 and Box 3.2). See Note 2 for an explanation of the classification of educational attainment. Unadjusted differences are the differences between the averages in each cycle. Adjusted differences are based on a method analogous to post- stratification that reweights the samples in Cycle 2 so that education groups have the same demographic characteristics as in Cycle 1 (see Note 1). Demographic characteristics considered are: age (in ten-year age brackets), gender and immigrant background. Darker colours denote differences that are statistically significant at the 5% level. In Chile and the United States, more than 20% of the subsample of adults with below upper secondary education only completed the assessment of reading and numeracy components in the second cycle. As the methodology for estimating the proficiency of adults who only took the components assessment has changed between the cycles, the change in proficiency for this group is not reported. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the Note for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the unadjusted change in literacy proficiency among adults with below upper secondary education._ Source: OECD (2018[4]; 2015[5]; 2012[6]), _Survey of Adult Skills (PIAAC) databases_ , http://www.oecd.org/skills/piaac/publicdataandanalysis/ (accessed on 23 September 2024); Table A.3.11 (L) in Annex A.\n\nSweden\n\nFinland\n\nSpain\n\nenmark\n\nCanada\n\nFlemish Region ( E\n\nEstonia\n\nItaly\n\netherlands\n\nGermany\n\nC echia\n\nSlo ak Republic\n\nIreland\n\nEngland (U\n\norway\n\nPoland\n\nFrance\n\napan\n\norea\n\nAustria\n\nUnited States\n\nIsrael\n\nSingapore\n\new ealand\n\nithuania\n\nChile\n\nHungary\n\nSweden\n\nFinland\n\nSpain\n\nenmark\n\nCanada\n\nFlemish Region ( E\n\nEstonia\n\nItaly\n\netherlands\n\nGermany\n\nC echia\n\nSlo ak Republic\n\nIreland\n\nEngland (U\n\norway\n\nPoland\n\nFrance\n\napan\n\norea\n\nAustria\n\nUnited States\n\nIsrael\n\nSingapore\n\new ealand\n\nithuania\n\nChile\n\nHungary\n\nAd usted Unad u sted\n\n20\n\n10\n\n0\n\n10\n\n20\n\n30\n\n0\n\n0\n\n20\n\n10\n\n0\n\n10\n\n20\n\n30\n\n0\n\n0\n\n20\n\n10\n\n0\n\n10\n\n20\n\n30\n\n0\n\n0\n\n**ertiary**\n\n**pper secondary**\n\n**Belo upper secondary**\n\nRound, Cycle 1:\n\nRound, Cycle 1:","publish_date":null,"title":"Figure 3.7. Change in literacy proficiency between cycles, by educational attainment","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p119.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p119.pdf"},"request":{"example_id":"chart/b263dc5d-en_p119","product_type":"parse","source_file_path":"b263dc5d-en_p119.pdf"},"started_at":"2026-11-04T19:45:45.271075"} +{"completed_at":"2026-11-04T19:45:48.118470","latency_in_ms":1316,"output":{"example_id":"chart/b263dc5d-en_p120","grounded_pages":[],"layout_pages":[],"markdown":"**118** \n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 3.8. Change in the gap in literacy proficiency between highly and low-educated adults**\n\nAdjusted and unadjusted change between cycles in the average score difference between adults with tertiary education and adults with below upper secondary education (Cycle 2 _minus_ Cycle 1); 25-65 year-olds\n\nNote: Does not include adults who in Cycle 2 were only administered the doorstep interview due to a language barrier, to maximise the comparability across cycles (see Box 1.1 in Chapter 1 and Box 3.2). See Note 2 for an explanation of the classification of educational attainment. Unadjusted changes in the proficiency gap are the changes in the difference between the two contrast categories across the cycles. Adjusted changes are obtained by subtracting the coefficients of regression models estimated separately for each cycle. These coefficients show the difference between the categories in each cycle, after accounting for: gender, immigrant background, age, parental education and language spoken at home. Darker colours denote differences that are statistically significant at the 5% level. In Chile and the United States, more than 20% of the subsample of adults with below upper secondary education only completed the assessment of reading and numeracy components in the second cycle. As the methodology for estimating the proficiency of adults who only took the components assessment has changed between the cycles, the change in the proficiency gap is not reported for these countries. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the Note for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the unadjusted change in the gap._ Source: OECD (2018[4]; 2015[5]; 2012[6]), _Survey of Adult Skills (PIAAC) databases_ , http://www.oecd.org/skills/piaac/publicdataandanalysis/ (accessed on 23 September 2024); Tables A.3.12 (L) and A.3.13 (L) in Annex A.\n\nThe numeracy proficiency gap between low- and highly educated adults widened significantly in 11 countries and economies, with large increases (over 20 points) in Austria, Japan and New Zealand. In Singapore, the gap in numeracy narrowed due to improvements among the low-educated, while in Israel and the Slovak Republic it narrowed because of proficiency declines among tertiary-educated adults (see Tables A.3.11 (N) and A.3.12 (N) in Annex A).\n\n_Changes in skills proficiency by educational attainment among younger and older adults_\n\nWhen interpreting proficiency trends for different education groups, it is important to consider that the proficiency of younger and older adults with the same educational attainment may have evolved differently due to their different educational experiences. Older adults may have received a very different education from that received by younger people today. The relationship between older adults’ formal education and their proficiency is also presumably weaker due to the length of time they have been out of education. In contrast, trends in the proficiency of younger adults with different attainment levels are more influenced by\n\napan\n\nAustria\n\nFrance\n\norway\n\nEstonia\n\nGermany\n\nEngland (U\n\nFinland\n\nFlemish Region ( E\n\nenmark\n\nIreland\n\nC echia\n\norea\n\netherlands\n\nCanada\n\nItaly\n\nSpain\n\nSlo ak Republic\n\nSweden\n\nPoland\n\new ealand\n\nithuania\n\nSingapore\n\nIsrael\n\nHungary\n\nScore point difference 0\n\n3\n\n30\n\n2\n\n20\n\n1\n\n10\n\n0\n\n10\n\n1\n\n20\n\nAd usted Unad u sted\n\nRound, Cycle 1:","pages":[{"line_numbers":[],"markdown":"**118** \n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 3.8. Change in the gap in literacy proficiency between highly and low-educated adults**\n\nAdjusted and unadjusted change between cycles in the average score difference between adults with tertiary education and adults with below upper secondary education (Cycle 2 _minus_ Cycle 1); 25-65 year-olds\n\nNote: Does not include adults who in Cycle 2 were only administered the doorstep interview due to a language barrier, to maximise the comparability across cycles (see Box 1.1 in Chapter 1 and Box 3.2). See Note 2 for an explanation of the classification of educational attainment. Unadjusted changes in the proficiency gap are the changes in the difference between the two contrast categories across the cycles. Adjusted changes are obtained by subtracting the coefficients of regression models estimated separately for each cycle. These coefficients show the difference between the categories in each cycle, after accounting for: gender, immigrant background, age, parental education and language spoken at home. Darker colours denote differences that are statistically significant at the 5% level. In Chile and the United States, more than 20% of the subsample of adults with below upper secondary education only completed the assessment of reading and numeracy components in the second cycle. As the methodology for estimating the proficiency of adults who only took the components assessment has changed between the cycles, the change in the proficiency gap is not reported for these countries. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the Note for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the unadjusted change in the gap._ Source: OECD (2018[4]; 2015[5]; 2012[6]), _Survey of Adult Skills (PIAAC) databases_ , http://www.oecd.org/skills/piaac/publicdataandanalysis/ (accessed on 23 September 2024); Tables A.3.12 (L) and A.3.13 (L) in Annex A.\n\nThe numeracy proficiency gap between low- and highly educated adults widened significantly in 11 countries and economies, with large increases (over 20 points) in Austria, Japan and New Zealand. In Singapore, the gap in numeracy narrowed due to improvements among the low-educated, while in Israel and the Slovak Republic it narrowed because of proficiency declines among tertiary-educated adults (see Tables A.3.11 (N) and A.3.12 (N) in Annex A).\n\n_Changes in skills proficiency by educational attainment among younger and older adults_\n\nWhen interpreting proficiency trends for different education groups, it is important to consider that the proficiency of younger and older adults with the same educational attainment may have evolved differently due to their different educational experiences. Older adults may have received a very different education from that received by younger people today. The relationship between older adults’ formal education and their proficiency is also presumably weaker due to the length of time they have been out of education. In contrast, trends in the proficiency of younger adults with different attainment levels are more influenced by\n\napan\n\nAustria\n\nFrance\n\norway\n\nEstonia\n\nGermany\n\nEngland (U\n\nFinland\n\nFlemish Region ( E\n\nenmark\n\nIreland\n\nC echia\n\norea\n\netherlands\n\nCanada\n\nItaly\n\nSpain\n\nSlo ak Republic\n\nSweden\n\nPoland\n\new ealand\n\nithuania\n\nSingapore\n\nIsrael\n\nHungary\n\nScore point difference 0\n\n3\n\n30\n\n2\n\n20\n\n1\n\n10\n\n0\n\n10\n\n1\n\n20\n\nAd usted Unad u sted\n\nRound, Cycle 1:","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**118** \n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 3.8. Change in the gap in literacy proficiency between highly and low-educated adults**\n\nAdjusted and unadjusted change between cycles in the average score difference between adults with tertiary education and adults with below upper secondary education (Cycle 2 _minus_ Cycle 1); 25-65 year-olds\n\nNote: Does not include adults who in Cycle 2 were only administered the doorstep interview due to a language barrier, to maximise the comparability across cycles (see Box 1.1 in Chapter 1 and Box 3.2). See Note 2 for an explanation of the classification of educational attainment. Unadjusted changes in the proficiency gap are the changes in the difference between the two contrast categories across the cycles. Adjusted changes are obtained by subtracting the coefficients of regression models estimated separately for each cycle. These coefficients show the difference between the categories in each cycle, after accounting for: gender, immigrant background, age, parental education and language spoken at home. Darker colours denote differences that are statistically significant at the 5% level. In Chile and the United States, more than 20% of the subsample of adults with below upper secondary education only completed the assessment of reading and numeracy components in the second cycle. As the methodology for estimating the proficiency of adults who only took the components assessment has changed between the cycles, the change in the proficiency gap is not reported for these countries. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the Note for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the unadjusted change in the gap._ Source: OECD (2018[4]; 2015[5]; 2012[6]), _Survey of Adult Skills (PIAAC) databases_ , http://www.oecd.org/skills/piaac/publicdataandanalysis/ (accessed on 23 September 2024); Tables A.3.12 (L) and A.3.13 (L) in Annex A.\n\nThe numeracy proficiency gap between low- and highly educated adults widened significantly in 11 countries and economies, with large increases (over 20 points) in Austria, Japan and New Zealand. In Singapore, the gap in numeracy narrowed due to improvements among the low-educated, while in Israel and the Slovak Republic it narrowed because of proficiency declines among tertiary-educated adults (see Tables A.3.11 (N) and A.3.12 (N) in Annex A).\n\n_Changes in skills proficiency by educational attainment among younger and older adults_\n\nWhen interpreting proficiency trends for different education groups, it is important to consider that the proficiency of younger and older adults with the same educational attainment may have evolved differently due to their different educational experiences. Older adults may have received a very different education from that received by younger people today. The relationship between older adults’ formal education and their proficiency is also presumably weaker due to the length of time they have been out of education. In contrast, trends in the proficiency of younger adults with different attainment levels are more influenced by\n\napan\n\nAustria\n\nFrance\n\norway\n\nEstonia\n\nGermany\n\nEngland (U\n\nFinland\n\nFlemish Region ( E\n\nenmark\n\nIreland\n\nC echia\n\norea\n\netherlands\n\nCanada\n\nItaly\n\nSpain\n\nSlo ak Republic\n\nSweden\n\nPoland\n\new ealand\n\nithuania\n\nSingapore\n\nIsrael\n\nHungary\n\nScore point difference 0\n\n3\n\n30\n\n2\n\n20\n\n1\n\n10\n\n0\n\n10\n\n1\n\n20\n\nAd usted Unad u sted\n\nRound, Cycle 1:","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p120.pdf"]},"response":{"errors":[],"extract_id":"extract_d4cf436541c7f5f67fb51f00c08954fe","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**118** \n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 3.8. Change in the gap in literacy proficiency between highly and low-educated adults**\n\nAdjusted and unadjusted change between cycles in the average score difference between adults with tertiary education and adults with below upper secondary education (Cycle 2 _minus_ Cycle 1); 25-65 year-olds\n\nNote: Does not include adults who in Cycle 2 were only administered the doorstep interview due to a language barrier, to maximise the comparability across cycles (see Box 1.1 in Chapter 1 and Box 3.2). See Note 2 for an explanation of the classification of educational attainment. Unadjusted changes in the proficiency gap are the changes in the difference between the two contrast categories across the cycles. Adjusted changes are obtained by subtracting the coefficients of regression models estimated separately for each cycle. These coefficients show the difference between the categories in each cycle, after accounting for: gender, immigrant background, age, parental education and language spoken at home. Darker colours denote differences that are statistically significant at the 5% level. In Chile and the United States, more than 20% of the subsample of adults with below upper secondary education only completed the assessment of reading and numeracy components in the second cycle. As the methodology for estimating the proficiency of adults who only took the components assessment has changed between the cycles, the change in the proficiency gap is not reported for these countries. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the Note for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the unadjusted change in the gap._ Source: OECD (2018[4]; 2015[5]; 2012[6]), _Survey of Adult Skills (PIAAC) databases_ , http://www.oecd.org/skills/piaac/publicdataandanalysis/ (accessed on 23 September 2024); Tables A.3.12 (L) and A.3.13 (L) in Annex A.\n\nThe numeracy proficiency gap between low- and highly educated adults widened significantly in 11 countries and economies, with large increases (over 20 points) in Austria, Japan and New Zealand. In Singapore, the gap in numeracy narrowed due to improvements among the low-educated, while in Israel and the Slovak Republic it narrowed because of proficiency declines among tertiary-educated adults (see Tables A.3.11 (N) and A.3.12 (N) in Annex A).\n\n_Changes in skills proficiency by educational attainment among younger and older adults_\n\nWhen interpreting proficiency trends for different education groups, it is important to consider that the proficiency of younger and older adults with the same educational attainment may have evolved differently due to their different educational experiences. Older adults may have received a very different education from that received by younger people today. The relationship between older adults’ formal education and their proficiency is also presumably weaker due to the length of time they have been out of education. In contrast, trends in the proficiency of younger adults with different attainment levels are more influenced by\n\napan\n\nAustria\n\nFrance\n\norway\n\nEstonia\n\nGermany\n\nEngland (U\n\nFinland\n\nFlemish Region ( E\n\nenmark\n\nIreland\n\nC echia\n\norea\n\netherlands\n\nCanada\n\nItaly\n\nSpain\n\nSlo ak Republic\n\nSweden\n\nPoland\n\new ealand\n\nithuania\n\nSingapore\n\nIsrael\n\nHungary\n\nScore point difference 0\n\n3\n\n30\n\n2\n\n20\n\n1\n\n10\n\n0\n\n10\n\n1\n\n20\n\nAd usted Unad u sted\n\nRound, Cycle 1:"],"full_content":"**118** \n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 3.8. Change in the gap in literacy proficiency between highly and low-educated adults**\n\nAdjusted and unadjusted change between cycles in the average score difference between adults with tertiary education and adults with below upper secondary education (Cycle 2 _minus_ Cycle 1); 25-65 year-olds\n\nNote: Does not include adults who in Cycle 2 were only administered the doorstep interview due to a language barrier, to maximise the comparability across cycles (see Box 1.1 in Chapter 1 and Box 3.2). See Note 2 for an explanation of the classification of educational attainment. Unadjusted changes in the proficiency gap are the changes in the difference between the two contrast categories across the cycles. Adjusted changes are obtained by subtracting the coefficients of regression models estimated separately for each cycle. These coefficients show the difference between the categories in each cycle, after accounting for: gender, immigrant background, age, parental education and language spoken at home. Darker colours denote differences that are statistically significant at the 5% level. In Chile and the United States, more than 20% of the subsample of adults with below upper secondary education only completed the assessment of reading and numeracy components in the second cycle. As the methodology for estimating the proficiency of adults who only took the components assessment has changed between the cycles, the change in the proficiency gap is not reported for these countries. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the Note for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the unadjusted change in the gap._ Source: OECD (2018[4]; 2015[5]; 2012[6]), _Survey of Adult Skills (PIAAC) databases_ , http://www.oecd.org/skills/piaac/publicdataandanalysis/ (accessed on 23 September 2024); Tables A.3.12 (L) and A.3.13 (L) in Annex A.\n\nThe numeracy proficiency gap between low- and highly educated adults widened significantly in 11 countries and economies, with large increases (over 20 points) in Austria, Japan and New Zealand. In Singapore, the gap in numeracy narrowed due to improvements among the low-educated, while in Israel and the Slovak Republic it narrowed because of proficiency declines among tertiary-educated adults (see Tables A.3.11 (N) and A.3.12 (N) in Annex A).\n\n_Changes in skills proficiency by educational attainment among younger and older adults_\n\nWhen interpreting proficiency trends for different education groups, it is important to consider that the proficiency of younger and older adults with the same educational attainment may have evolved differently due to their different educational experiences. Older adults may have received a very different education from that received by younger people today. The relationship between older adults’ formal education and their proficiency is also presumably weaker due to the length of time they have been out of education. In contrast, trends in the proficiency of younger adults with different attainment levels are more influenced by\n\napan\n\nAustria\n\nFrance\n\norway\n\nEstonia\n\nGermany\n\nEngland (U\n\nFinland\n\nFlemish Region ( E\n\nenmark\n\nIreland\n\nC echia\n\norea\n\netherlands\n\nCanada\n\nItaly\n\nSpain\n\nSlo ak Republic\n\nSweden\n\nPoland\n\new ealand\n\nithuania\n\nSingapore\n\nIsrael\n\nHungary\n\nScore point difference 0\n\n3\n\n30\n\n2\n\n20\n\n1\n\n10\n\n0\n\n10\n\n1\n\n20\n\nAd usted Unad u sted\n\nRound, Cycle 1:","publish_date":null,"title":"Figure 3.8. Change in the gap in literacy proficiency between highly and low-educated adults","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p120.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p120.pdf"},"request":{"example_id":"chart/b263dc5d-en_p120","product_type":"parse","source_file_path":"b263dc5d-en_p120.pdf"},"started_at":"2026-11-04T19:45:46.802203"} +{"completed_at":"2026-11-04T19:45:49.229160","latency_in_ms":1279,"output":{"example_id":"chart/b263dc5d-en_p123","grounded_pages":[],"layout_pages":[],"markdown":" **121**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n_Changes in the age structure of the population_\n\nOver recent decades, all OECD countries have seen a steady increase in the share of older adults in the population. This trend is projected to continue, with the proportion of people aged 65 and over expected to increase from 18% in 2023 to 25% by 2050, and that of people aged 50 and over from 37% to 44% (OECD, 2024[17]). In the years between the two cycles of the Survey of Adult Skills, most participating countries and economies have experienced a change in the age structure of their population (Figure 3.10). This shift was most pronounced in Austria, Germany, Italy and Korea, where the share of 55-65 year-olds increased by 5 percentage points or more between 2012 and 2023. Among countries taking part in Round 2 of the first cycle, this share has grown by 5 percentage points in Lithuania and 4 percentage points in Singapore between 2015 and 2023.\n\n### **Figure 3.10. Change in the age composition of the adult population between cycles**\n\nDifference in the relative size of 10-year age groups between cycles (Cycle 2 _minus_ Cycle 1)\n\nNote: Adults aged 16-65; does not include adults who in Cycle 2 were only administered the doorstep interview due to a language barrier, to maximise the comparability across cycles (see Box 1.1 in Chapter 1 and Box 3.2). \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the ote for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the change in the share of 55-65 year-olds._ Source: OECD (2018[4]; 2015[5]; 2012[6]), _Survey of Adult Skills (PIAAC) databases_ , http://www.oecd.org/skills/piaac/publicdataandanalysis/ (accessed on 23 September 2024); Table B.3.6 (Trend) in Annex B.\n\n_Changes in skills proficiency by age_\n\nFigure 3.11 presents differences in literacy proficiency between adults of the same age at the time of the first and second cycle (i.e. cohort effects). It shows how proficiency has changed between assessments within five different age groups across the participating countries and economies. The evolution of proficiency in the youngest age group will reflect, to a larger extent, current and recent changes in education, such as expanding participation or improving education quality due to recent reforms. In contrast, changes in proficiency among the older age groups may be the result of influences that occurred long before the period covered by the assessment – reaching as far back as the 1950s and 1960s, when the cohort aged 55-65 in the first Survey of Adult Skills began compulsory schooling. In addition, differences among older adults will also reflect more recent changes in the world of work that would have provided or limited their opportunities to maintain and develop skills.\n\norea\n\nGermany\n\nAustria\n\nItaly\n\nSpain\n\netherlands\n\nIreland\n\nSlo ak Republic\n\nEngland (U\n\nCanada\n\nenmark\n\norway\n\nFlemish Region ( E\n\nFinland\n\nEstonia\n\nUnited States\n\nFrance\n\nPoland\n\nSweden\n\nC echia\n\napan\n\nithuania\n\nSingapore\n\nChile\n\new ealand\n\nIsrael\n\nHungary\n\nPercentage point difference 10\n\n8\n\n2\n\n0\n\n2\n\n8\n\nyear olds year olds 1 2 year olds 2 3 year olds\n\nRound, Cycle 1:","pages":[{"line_numbers":[],"markdown":" **121**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n_Changes in the age structure of the population_\n\nOver recent decades, all OECD countries have seen a steady increase in the share of older adults in the population. This trend is projected to continue, with the proportion of people aged 65 and over expected to increase from 18% in 2023 to 25% by 2050, and that of people aged 50 and over from 37% to 44% (OECD, 2024[17]). In the years between the two cycles of the Survey of Adult Skills, most participating countries and economies have experienced a change in the age structure of their population (Figure 3.10). This shift was most pronounced in Austria, Germany, Italy and Korea, where the share of 55-65 year-olds increased by 5 percentage points or more between 2012 and 2023. Among countries taking part in Round 2 of the first cycle, this share has grown by 5 percentage points in Lithuania and 4 percentage points in Singapore between 2015 and 2023.\n\n### **Figure 3.10. Change in the age composition of the adult population between cycles**\n\nDifference in the relative size of 10-year age groups between cycles (Cycle 2 _minus_ Cycle 1)\n\nNote: Adults aged 16-65; does not include adults who in Cycle 2 were only administered the doorstep interview due to a language barrier, to maximise the comparability across cycles (see Box 1.1 in Chapter 1 and Box 3.2). \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the ote for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the change in the share of 55-65 year-olds._ Source: OECD (2018[4]; 2015[5]; 2012[6]), _Survey of Adult Skills (PIAAC) databases_ , http://www.oecd.org/skills/piaac/publicdataandanalysis/ (accessed on 23 September 2024); Table B.3.6 (Trend) in Annex B.\n\n_Changes in skills proficiency by age_\n\nFigure 3.11 presents differences in literacy proficiency between adults of the same age at the time of the first and second cycle (i.e. cohort effects). It shows how proficiency has changed between assessments within five different age groups across the participating countries and economies. The evolution of proficiency in the youngest age group will reflect, to a larger extent, current and recent changes in education, such as expanding participation or improving education quality due to recent reforms. In contrast, changes in proficiency among the older age groups may be the result of influences that occurred long before the period covered by the assessment – reaching as far back as the 1950s and 1960s, when the cohort aged 55-65 in the first Survey of Adult Skills began compulsory schooling. In addition, differences among older adults will also reflect more recent changes in the world of work that would have provided or limited their opportunities to maintain and develop skills.\n\norea\n\nGermany\n\nAustria\n\nItaly\n\nSpain\n\netherlands\n\nIreland\n\nSlo ak Republic\n\nEngland (U\n\nCanada\n\nenmark\n\norway\n\nFlemish Region ( E\n\nFinland\n\nEstonia\n\nUnited States\n\nFrance\n\nPoland\n\nSweden\n\nC echia\n\napan\n\nithuania\n\nSingapore\n\nChile\n\new ealand\n\nIsrael\n\nHungary\n\nPercentage point difference 10\n\n8\n\n2\n\n0\n\n2\n\n8\n\nyear olds year olds 1 2 year olds 2 3 year olds\n\nRound, Cycle 1:","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":" **121**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n_Changes in the age structure of the population_\n\nOver recent decades, all OECD countries have seen a steady increase in the share of older adults in the population. This trend is projected to continue, with the proportion of people aged 65 and over expected to increase from 18% in 2023 to 25% by 2050, and that of people aged 50 and over from 37% to 44% (OECD, 2024[17]). In the years between the two cycles of the Survey of Adult Skills, most participating countries and economies have experienced a change in the age structure of their population (Figure 3.10). This shift was most pronounced in Austria, Germany, Italy and Korea, where the share of 55-65 year-olds increased by 5 percentage points or more between 2012 and 2023. Among countries taking part in Round 2 of the first cycle, this share has grown by 5 percentage points in Lithuania and 4 percentage points in Singapore between 2015 and 2023.\n\n### **Figure 3.10. Change in the age composition of the adult population between cycles**\n\nDifference in the relative size of 10-year age groups between cycles (Cycle 2 _minus_ Cycle 1)\n\nNote: Adults aged 16-65; does not include adults who in Cycle 2 were only administered the doorstep interview due to a language barrier, to maximise the comparability across cycles (see Box 1.1 in Chapter 1 and Box 3.2). \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the ote for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the change in the share of 55-65 year-olds._ Source: OECD (2018[4]; 2015[5]; 2012[6]), _Survey of Adult Skills (PIAAC) databases_ , http://www.oecd.org/skills/piaac/publicdataandanalysis/ (accessed on 23 September 2024); Table B.3.6 (Trend) in Annex B.\n\n_Changes in skills proficiency by age_\n\nFigure 3.11 presents differences in literacy proficiency between adults of the same age at the time of the first and second cycle (i.e. cohort effects). It shows how proficiency has changed between assessments within five different age groups across the participating countries and economies. The evolution of proficiency in the youngest age group will reflect, to a larger extent, current and recent changes in education, such as expanding participation or improving education quality due to recent reforms. In contrast, changes in proficiency among the older age groups may be the result of influences that occurred long before the period covered by the assessment – reaching as far back as the 1950s and 1960s, when the cohort aged 55-65 in the first Survey of Adult Skills began compulsory schooling. In addition, differences among older adults will also reflect more recent changes in the world of work that would have provided or limited their opportunities to maintain and develop skills.\n\norea\n\nGermany\n\nAustria\n\nItaly\n\nSpain\n\netherlands\n\nIreland\n\nSlo ak Republic\n\nEngland (U\n\nCanada\n\nenmark\n\norway\n\nFlemish Region ( E\n\nFinland\n\nEstonia\n\nUnited States\n\nFrance\n\nPoland\n\nSweden\n\nC echia\n\napan\n\nithuania\n\nSingapore\n\nChile\n\new ealand\n\nIsrael\n\nHungary\n\nPercentage point difference 10\n\n8\n\n2\n\n0\n\n2\n\n8\n\nyear olds year olds 1 2 year olds 2 3 year olds\n\nRound, Cycle 1:","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p123.pdf"]},"response":{"errors":[],"extract_id":"extract_03d4e34aa5cd1dec47e281f107c585c8","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":[" **121**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n_Changes in the age structure of the population_\n\nOver recent decades, all OECD countries have seen a steady increase in the share of older adults in the population. This trend is projected to continue, with the proportion of people aged 65 and over expected to increase from 18% in 2023 to 25% by 2050, and that of people aged 50 and over from 37% to 44% (OECD, 2024[17]). In the years between the two cycles of the Survey of Adult Skills, most participating countries and economies have experienced a change in the age structure of their population (Figure 3.10). This shift was most pronounced in Austria, Germany, Italy and Korea, where the share of 55-65 year-olds increased by 5 percentage points or more between 2012 and 2023. Among countries taking part in Round 2 of the first cycle, this share has grown by 5 percentage points in Lithuania and 4 percentage points in Singapore between 2015 and 2023.\n\n### **Figure 3.10. Change in the age composition of the adult population between cycles**\n\nDifference in the relative size of 10-year age groups between cycles (Cycle 2 _minus_ Cycle 1)\n\nNote: Adults aged 16-65; does not include adults who in Cycle 2 were only administered the doorstep interview due to a language barrier, to maximise the comparability across cycles (see Box 1.1 in Chapter 1 and Box 3.2). \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the ote for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the change in the share of 55-65 year-olds._ Source: OECD (2018[4]; 2015[5]; 2012[6]), _Survey of Adult Skills (PIAAC) databases_ , http://www.oecd.org/skills/piaac/publicdataandanalysis/ (accessed on 23 September 2024); Table B.3.6 (Trend) in Annex B.\n\n_Changes in skills proficiency by age_\n\nFigure 3.11 presents differences in literacy proficiency between adults of the same age at the time of the first and second cycle (i.e. cohort effects). It shows how proficiency has changed between assessments within five different age groups across the participating countries and economies. The evolution of proficiency in the youngest age group will reflect, to a larger extent, current and recent changes in education, such as expanding participation or improving education quality due to recent reforms. In contrast, changes in proficiency among the older age groups may be the result of influences that occurred long before the period covered by the assessment – reaching as far back as the 1950s and 1960s, when the cohort aged 55-65 in the first Survey of Adult Skills began compulsory schooling. In addition, differences among older adults will also reflect more recent changes in the world of work that would have provided or limited their opportunities to maintain and develop skills.\n\norea\n\nGermany\n\nAustria\n\nItaly\n\nSpain\n\netherlands\n\nIreland\n\nSlo ak Republic\n\nEngland (U\n\nCanada\n\nenmark\n\norway\n\nFlemish Region ( E\n\nFinland\n\nEstonia\n\nUnited States\n\nFrance\n\nPoland\n\nSweden\n\nC echia\n\napan\n\nithuania\n\nSingapore\n\nChile\n\new ealand\n\nIsrael\n\nHungary\n\nPercentage point difference 10\n\n8\n\n2\n\n0\n\n2\n\n8\n\nyear olds year olds 1 2 year olds 2 3 year olds\n\nRound, Cycle 1:"],"full_content":" **121**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n_Changes in the age structure of the population_\n\nOver recent decades, all OECD countries have seen a steady increase in the share of older adults in the population. This trend is projected to continue, with the proportion of people aged 65 and over expected to increase from 18% in 2023 to 25% by 2050, and that of people aged 50 and over from 37% to 44% (OECD, 2024[17]). In the years between the two cycles of the Survey of Adult Skills, most participating countries and economies have experienced a change in the age structure of their population (Figure 3.10). This shift was most pronounced in Austria, Germany, Italy and Korea, where the share of 55-65 year-olds increased by 5 percentage points or more between 2012 and 2023. Among countries taking part in Round 2 of the first cycle, this share has grown by 5 percentage points in Lithuania and 4 percentage points in Singapore between 2015 and 2023.\n\n### **Figure 3.10. Change in the age composition of the adult population between cycles**\n\nDifference in the relative size of 10-year age groups between cycles (Cycle 2 _minus_ Cycle 1)\n\nNote: Adults aged 16-65; does not include adults who in Cycle 2 were only administered the doorstep interview due to a language barrier, to maximise the comparability across cycles (see Box 1.1 in Chapter 1 and Box 3.2). \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the ote for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the change in the share of 55-65 year-olds._ Source: OECD (2018[4]; 2015[5]; 2012[6]), _Survey of Adult Skills (PIAAC) databases_ , http://www.oecd.org/skills/piaac/publicdataandanalysis/ (accessed on 23 September 2024); Table B.3.6 (Trend) in Annex B.\n\n_Changes in skills proficiency by age_\n\nFigure 3.11 presents differences in literacy proficiency between adults of the same age at the time of the first and second cycle (i.e. cohort effects). It shows how proficiency has changed between assessments within five different age groups across the participating countries and economies. The evolution of proficiency in the youngest age group will reflect, to a larger extent, current and recent changes in education, such as expanding participation or improving education quality due to recent reforms. In contrast, changes in proficiency among the older age groups may be the result of influences that occurred long before the period covered by the assessment – reaching as far back as the 1950s and 1960s, when the cohort aged 55-65 in the first Survey of Adult Skills began compulsory schooling. In addition, differences among older adults will also reflect more recent changes in the world of work that would have provided or limited their opportunities to maintain and develop skills.\n\norea\n\nGermany\n\nAustria\n\nItaly\n\nSpain\n\netherlands\n\nIreland\n\nSlo ak Republic\n\nEngland (U\n\nCanada\n\nenmark\n\norway\n\nFlemish Region ( E\n\nFinland\n\nEstonia\n\nUnited States\n\nFrance\n\nPoland\n\nSweden\n\nC echia\n\napan\n\nithuania\n\nSingapore\n\nChile\n\new ealand\n\nIsrael\n\nHungary\n\nPercentage point difference 10\n\n8\n\n2\n\n0\n\n2\n\n8\n\nyear olds year olds 1 2 year olds 2 3 year olds\n\nRound, Cycle 1:","publish_date":null,"title":"Figure 3.10. Change in the age composition of the adult population between cycles","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p123.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p123.pdf"},"request":{"example_id":"chart/b263dc5d-en_p123","product_type":"parse","source_file_path":"b263dc5d-en_p123.pdf"},"started_at":"2026-11-04T19:45:47.949548"} +{"completed_at":"2026-11-04T19:45:50.767963","latency_in_ms":1536,"output":{"example_id":"chart/b263dc5d-en_p128","grounded_pages":[],"layout_pages":[],"markdown":"**126** \n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 3.13. Change in the gap in literacy proficiency between younger and older adults**\n\nAdjusted and unadjusted changes between cycles in the mean score difference between 25-34 year-olds and 55-65 year-olds (Cycle 2 _minus_ Cycle 1)\n\nNote: Does not include adults who in Cycle 2 were only administered the doorstep interview due to a language barrier, to maximise the comparability across cycles (see Box 1.1 in Chapter 1 and Box 3.2). Unadjusted changes in the proficiency gap are the changes in the difference between the two contrast categories across the cycles. Adjusted changes are obtained by subtracting the coefficients of regression models estimated separately for each cycle. These coefficients show the difference between the categories in each cycle, after accounting for: gender, immigrant background, educational attainment, parental education and language spoken at home. Darker colours denote differences that are statistically significant at the 5% level. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the ote for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the unadjusted change in the gap._ Source: OECD (2018[4]; 2015[5]; 2012[6]), _Survey of Adult Skills (PIAAC) databases_ , http://www.oecd.org/skills/piaac/publicdataandanalysis/ (accessed on 23 September 2024); Tables A.3.8 (L) and A.3.9 (L) in Annex A.\n\n**_Changes in skills proficiency related to immigrant background_**\n\nSince the first Survey of Adult Skills, most OECD countries have seen unprecedented numbers of migrants arriving in their territories, many of whom are asylum seekers (OECD, 2024[18]). ³ This has posed challenges to many countries, while also presenting opportunities. Migrants increase and diversify the supply of labour, which can contribute to economic growth and help mitigate the negative impacts of population ageing (OECD, 2016[19]). A precondition for this is that they have the skills required to succeed in daily life and work in the host country (OECD, 2019[20]). While some migrants will already possess these skills, others need to overcome substantial language and cultural barriers to acquire them.\n\nHow immigrants’ skills develop depends on two main factors: the profile of migrants coming in and the institutions and policies that support their integration. Countries often admit migrants with specific qualifications to address labour shortages or support innovative sectors. In contrast, during the refugee crises of the 2010s, many countries, particularly in Europe, admitted large numbers of low-skilled migrants who were unfamiliar with the host country’s language. This can alter the average skills proficiency of the immigrant population and – depending on the size of the group – the adult population as a whole. Once migrants have settled in the host country, there are usually mechanisms in place to support their social and labour-market integration, such as language training and assistance in searching for and applying for work. These policies also impact migrants’ skills levels. Their effectiveness may vary across countries and economies due to differences in quality but also differences in the linguistic proximity between migrants’ countries of origin and destination.\n\nEstonia\n\nFrance\n\nAustria\n\norea\n\nC echia\n\nEngland (U\n\nenmark\n\nCanada\n\nFinland\n\napan\n\nIreland\n\nUnited States\n\norway\n\nFlemish Region ( E\n\netherlands\n\nGermany\n\nItaly\n\nPoland\n\nSlo ak Republic\n\nSpain\n\nSweden\n\nChile\n\nithuania\n\nSingapore\n\nIsrael\n\new ealand\n\nHungary\n\nScore point difference 0\n\n30\n\n20\n\n10\n\n0\n\n10\n\n20\n\nAd usted Unad u sted\n\nRound, Cycle 1:","pages":[{"line_numbers":[],"markdown":"**126** \n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 3.13. Change in the gap in literacy proficiency between younger and older adults**\n\nAdjusted and unadjusted changes between cycles in the mean score difference between 25-34 year-olds and 55-65 year-olds (Cycle 2 _minus_ Cycle 1)\n\nNote: Does not include adults who in Cycle 2 were only administered the doorstep interview due to a language barrier, to maximise the comparability across cycles (see Box 1.1 in Chapter 1 and Box 3.2). Unadjusted changes in the proficiency gap are the changes in the difference between the two contrast categories across the cycles. Adjusted changes are obtained by subtracting the coefficients of regression models estimated separately for each cycle. These coefficients show the difference between the categories in each cycle, after accounting for: gender, immigrant background, educational attainment, parental education and language spoken at home. Darker colours denote differences that are statistically significant at the 5% level. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the ote for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the unadjusted change in the gap._ Source: OECD (2018[4]; 2015[5]; 2012[6]), _Survey of Adult Skills (PIAAC) databases_ , http://www.oecd.org/skills/piaac/publicdataandanalysis/ (accessed on 23 September 2024); Tables A.3.8 (L) and A.3.9 (L) in Annex A.\n\n**_Changes in skills proficiency related to immigrant background_**\n\nSince the first Survey of Adult Skills, most OECD countries have seen unprecedented numbers of migrants arriving in their territories, many of whom are asylum seekers (OECD, 2024[18]). ³ This has posed challenges to many countries, while also presenting opportunities. Migrants increase and diversify the supply of labour, which can contribute to economic growth and help mitigate the negative impacts of population ageing (OECD, 2016[19]). A precondition for this is that they have the skills required to succeed in daily life and work in the host country (OECD, 2019[20]). While some migrants will already possess these skills, others need to overcome substantial language and cultural barriers to acquire them.\n\nHow immigrants’ skills develop depends on two main factors: the profile of migrants coming in and the institutions and policies that support their integration. Countries often admit migrants with specific qualifications to address labour shortages or support innovative sectors. In contrast, during the refugee crises of the 2010s, many countries, particularly in Europe, admitted large numbers of low-skilled migrants who were unfamiliar with the host country’s language. This can alter the average skills proficiency of the immigrant population and – depending on the size of the group – the adult population as a whole. Once migrants have settled in the host country, there are usually mechanisms in place to support their social and labour-market integration, such as language training and assistance in searching for and applying for work. These policies also impact migrants’ skills levels. Their effectiveness may vary across countries and economies due to differences in quality but also differences in the linguistic proximity between migrants’ countries of origin and destination.\n\nEstonia\n\nFrance\n\nAustria\n\norea\n\nC echia\n\nEngland (U\n\nenmark\n\nCanada\n\nFinland\n\napan\n\nIreland\n\nUnited States\n\norway\n\nFlemish Region ( E\n\netherlands\n\nGermany\n\nItaly\n\nPoland\n\nSlo ak Republic\n\nSpain\n\nSweden\n\nChile\n\nithuania\n\nSingapore\n\nIsrael\n\new ealand\n\nHungary\n\nScore point difference 0\n\n30\n\n20\n\n10\n\n0\n\n10\n\n20\n\nAd usted Unad u sted\n\nRound, Cycle 1:","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**126** \n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 3.13. Change in the gap in literacy proficiency between younger and older adults**\n\nAdjusted and unadjusted changes between cycles in the mean score difference between 25-34 year-olds and 55-65 year-olds (Cycle 2 _minus_ Cycle 1)\n\nNote: Does not include adults who in Cycle 2 were only administered the doorstep interview due to a language barrier, to maximise the comparability across cycles (see Box 1.1 in Chapter 1 and Box 3.2). Unadjusted changes in the proficiency gap are the changes in the difference between the two contrast categories across the cycles. Adjusted changes are obtained by subtracting the coefficients of regression models estimated separately for each cycle. These coefficients show the difference between the categories in each cycle, after accounting for: gender, immigrant background, educational attainment, parental education and language spoken at home. Darker colours denote differences that are statistically significant at the 5% level. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the ote for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the unadjusted change in the gap._ Source: OECD (2018[4]; 2015[5]; 2012[6]), _Survey of Adult Skills (PIAAC) databases_ , http://www.oecd.org/skills/piaac/publicdataandanalysis/ (accessed on 23 September 2024); Tables A.3.8 (L) and A.3.9 (L) in Annex A.\n\n**_Changes in skills proficiency related to immigrant background_**\n\nSince the first Survey of Adult Skills, most OECD countries have seen unprecedented numbers of migrants arriving in their territories, many of whom are asylum seekers (OECD, 2024[18]). ³ This has posed challenges to many countries, while also presenting opportunities. Migrants increase and diversify the supply of labour, which can contribute to economic growth and help mitigate the negative impacts of population ageing (OECD, 2016[19]). A precondition for this is that they have the skills required to succeed in daily life and work in the host country (OECD, 2019[20]). While some migrants will already possess these skills, others need to overcome substantial language and cultural barriers to acquire them.\n\nHow immigrants’ skills develop depends on two main factors: the profile of migrants coming in and the institutions and policies that support their integration. Countries often admit migrants with specific qualifications to address labour shortages or support innovative sectors. In contrast, during the refugee crises of the 2010s, many countries, particularly in Europe, admitted large numbers of low-skilled migrants who were unfamiliar with the host country’s language. This can alter the average skills proficiency of the immigrant population and – depending on the size of the group – the adult population as a whole. Once migrants have settled in the host country, there are usually mechanisms in place to support their social and labour-market integration, such as language training and assistance in searching for and applying for work. These policies also impact migrants’ skills levels. Their effectiveness may vary across countries and economies due to differences in quality but also differences in the linguistic proximity between migrants’ countries of origin and destination.\n\nEstonia\n\nFrance\n\nAustria\n\norea\n\nC echia\n\nEngland (U\n\nenmark\n\nCanada\n\nFinland\n\napan\n\nIreland\n\nUnited States\n\norway\n\nFlemish Region ( E\n\netherlands\n\nGermany\n\nItaly\n\nPoland\n\nSlo ak Republic\n\nSpain\n\nSweden\n\nChile\n\nithuania\n\nSingapore\n\nIsrael\n\new ealand\n\nHungary\n\nScore point difference 0\n\n30\n\n20\n\n10\n\n0\n\n10\n\n20\n\nAd usted Unad u sted\n\nRound, Cycle 1:","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p128.pdf"]},"response":{"errors":[],"extract_id":"extract_d8bcbe1a07f05299baa219600c070247","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**126** \n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 3.13. Change in the gap in literacy proficiency between younger and older adults**\n\nAdjusted and unadjusted changes between cycles in the mean score difference between 25-34 year-olds and 55-65 year-olds (Cycle 2 _minus_ Cycle 1)\n\nNote: Does not include adults who in Cycle 2 were only administered the doorstep interview due to a language barrier, to maximise the comparability across cycles (see Box 1.1 in Chapter 1 and Box 3.2). Unadjusted changes in the proficiency gap are the changes in the difference between the two contrast categories across the cycles. Adjusted changes are obtained by subtracting the coefficients of regression models estimated separately for each cycle. These coefficients show the difference between the categories in each cycle, after accounting for: gender, immigrant background, educational attainment, parental education and language spoken at home. Darker colours denote differences that are statistically significant at the 5% level. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the ote for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the unadjusted change in the gap._ Source: OECD (2018[4]; 2015[5]; 2012[6]), _Survey of Adult Skills (PIAAC) databases_ , http://www.oecd.org/skills/piaac/publicdataandanalysis/ (accessed on 23 September 2024); Tables A.3.8 (L) and A.3.9 (L) in Annex A.\n\n**_Changes in skills proficiency related to immigrant background_**\n\nSince the first Survey of Adult Skills, most OECD countries have seen unprecedented numbers of migrants arriving in their territories, many of whom are asylum seekers (OECD, 2024[18]). ³ This has posed challenges to many countries, while also presenting opportunities. Migrants increase and diversify the supply of labour, which can contribute to economic growth and help mitigate the negative impacts of population ageing (OECD, 2016[19]). A precondition for this is that they have the skills required to succeed in daily life and work in the host country (OECD, 2019[20]). While some migrants will already possess these skills, others need to overcome substantial language and cultural barriers to acquire them.\n\nHow immigrants’ skills develop depends on two main factors: the profile of migrants coming in and the institutions and policies that support their integration. Countries often admit migrants with specific qualifications to address labour shortages or support innovative sectors. In contrast, during the refugee crises of the 2010s, many countries, particularly in Europe, admitted large numbers of low-skilled migrants who were unfamiliar with the host country’s language. This can alter the average skills proficiency of the immigrant population and – depending on the size of the group – the adult population as a whole. Once migrants have settled in the host country, there are usually mechanisms in place to support their social and labour-market integration, such as language training and assistance in searching for and applying for work. These policies also impact migrants’ skills levels. Their effectiveness may vary across countries and economies due to differences in quality but also differences in the linguistic proximity between migrants’ countries of origin and destination.\n\nEstonia\n\nFrance\n\nAustria\n\norea\n\nC echia\n\nEngland (U\n\nenmark\n\nCanada\n\nFinland\n\napan\n\nIreland\n\nUnited States\n\norway\n\nFlemish Region ( E\n\netherlands\n\nGermany\n\nItaly\n\nPoland\n\nSlo ak Republic\n\nSpain\n\nSweden\n\nChile\n\nithuania\n\nSingapore\n\nIsrael\n\new ealand\n\nHungary\n\nScore point difference 0\n\n30\n\n20\n\n10\n\n0\n\n10\n\n20\n\nAd usted Unad u sted\n\nRound, Cycle 1:"],"full_content":"**126** \n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 3.13. Change in the gap in literacy proficiency between younger and older adults**\n\nAdjusted and unadjusted changes between cycles in the mean score difference between 25-34 year-olds and 55-65 year-olds (Cycle 2 _minus_ Cycle 1)\n\nNote: Does not include adults who in Cycle 2 were only administered the doorstep interview due to a language barrier, to maximise the comparability across cycles (see Box 1.1 in Chapter 1 and Box 3.2). Unadjusted changes in the proficiency gap are the changes in the difference between the two contrast categories across the cycles. Adjusted changes are obtained by subtracting the coefficients of regression models estimated separately for each cycle. These coefficients show the difference between the categories in each cycle, after accounting for: gender, immigrant background, educational attainment, parental education and language spoken at home. Darker colours denote differences that are statistically significant at the 5% level. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the ote for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the unadjusted change in the gap._ Source: OECD (2018[4]; 2015[5]; 2012[6]), _Survey of Adult Skills (PIAAC) databases_ , http://www.oecd.org/skills/piaac/publicdataandanalysis/ (accessed on 23 September 2024); Tables A.3.8 (L) and A.3.9 (L) in Annex A.\n\n**_Changes in skills proficiency related to immigrant background_**\n\nSince the first Survey of Adult Skills, most OECD countries have seen unprecedented numbers of migrants arriving in their territories, many of whom are asylum seekers (OECD, 2024[18]). ³ This has posed challenges to many countries, while also presenting opportunities. Migrants increase and diversify the supply of labour, which can contribute to economic growth and help mitigate the negative impacts of population ageing (OECD, 2016[19]). A precondition for this is that they have the skills required to succeed in daily life and work in the host country (OECD, 2019[20]). While some migrants will already possess these skills, others need to overcome substantial language and cultural barriers to acquire them.\n\nHow immigrants’ skills develop depends on two main factors: the profile of migrants coming in and the institutions and policies that support their integration. Countries often admit migrants with specific qualifications to address labour shortages or support innovative sectors. In contrast, during the refugee crises of the 2010s, many countries, particularly in Europe, admitted large numbers of low-skilled migrants who were unfamiliar with the host country’s language. This can alter the average skills proficiency of the immigrant population and – depending on the size of the group – the adult population as a whole. Once migrants have settled in the host country, there are usually mechanisms in place to support their social and labour-market integration, such as language training and assistance in searching for and applying for work. These policies also impact migrants’ skills levels. Their effectiveness may vary across countries and economies due to differences in quality but also differences in the linguistic proximity between migrants’ countries of origin and destination.\n\nEstonia\n\nFrance\n\nAustria\n\norea\n\nC echia\n\nEngland (U\n\nenmark\n\nCanada\n\nFinland\n\napan\n\nIreland\n\nUnited States\n\norway\n\nFlemish Region ( E\n\netherlands\n\nGermany\n\nItaly\n\nPoland\n\nSlo ak Republic\n\nSpain\n\nSweden\n\nChile\n\nithuania\n\nSingapore\n\nIsrael\n\new ealand\n\nHungary\n\nScore point difference 0\n\n30\n\n20\n\n10\n\n0\n\n10\n\n20\n\nAd usted Unad u sted\n\nRound, Cycle 1:","publish_date":null,"title":"Figure 3.13. Change in the gap in literacy proficiency between younger and older adults","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p128.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p128.pdf"},"request":{"example_id":"chart/b263dc5d-en_p128","product_type":"parse","source_file_path":"b263dc5d-en_p128.pdf"},"started_at":"2026-11-04T19:45:49.231581"} +{"completed_at":"2026-11-04T19:45:51.720434","latency_in_ms":1792,"output":{"example_id":"chart/b263dc5d-en_p130","grounded_pages":[],"layout_pages":[],"markdown":"**128** \n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 3.14. Change in the share of foreign-born adults in the adult population between cycles**\n\nNote: Adults aged 16-65; does not include adults who in Cycle 2 were only administered the doorstep interview due to a language barrier, to maximise the comparability across cycles (see Box 1.1 in Chapter 1 and Box 3.2). Japan and Poland are excluded due to small numbers of foreign-born adults. _Countries and economies are ranked in descending order of the change in the share of foreign-born adults._ Source: OECD (2018[4]; 2015[5]; 2012[6]), _Survey of Adult Skills (PIAAC) databases_ , http://www.oecd.org/skills/piaac/publicdataandanalysis/ (accessed on 23 September 2024); Table B.3.10 (Trend) in Annex B.\n\nOverall, the patterns of change for the native-born population largely mirror those of the entire adult population, but there are exceptions in some countries. In the Flemish Region (Belgium), Germany and Norway, the literacy proficiency of adults without an immigrant background has improved since the first cycle, and it has remained stable in France. Meanwhile, the proficiency of adults with an immigrant background developed less favourably in these countries and economies.\n\nOther countries have also experienced less favourable trends for immigrants than for native-born adults. In Chile and Singapore, literacy proficiency declined among immigrants, but not among native-born adults. In Austria and New Zealand, the decline in literacy proficiency for foreign-born adults was more pronounced than for native-born adults.\n\nMigrants’ numeracy skills have developed more positively in many countries over the past decade. Canada, Denmark, England (United Kingdom), Estonia, Finland, Norway and Sweden saw improvements in the proficiency of foreign-born adults (see Table A.3.14 (N) in Annex A). In Finland, Norway and Sweden, these gains were particularly large, exceeding 20 points. In contrast, numeracy proficiency among immigrants only declined significantly between cycles in Hungary and New Zealand.\n\nWhen interpreting changes in the literacy proficiency of immigrants it is important to consider how long they have been living in the host country. Learning a new language and integrating into a new society takes time, while policies supporting immigrants’ integration also take time to contribute effectively to skills formation. Recent migrants will have had less of both time and such resources. Their skills proficiency is typically lower and more strongly influenced by the opportunities available in their country of origin as well as their linguistic distance to the host country. Changes in proficiency among recently arrived migrants must therefore be interpreted in the context of the changing profile of incoming migrants, which is shaped by each country’s immigration patterns and policies. The skills profile of long-term migrants may also vary\n\norway\n\nSweden\n\nEngland (U\n\nAustria\n\nSpain\n\nCanada\n\nItaly\n\nIreland\n\nGermany\n\nFlemish Region ( E\n\netherlands\n\nUnited States\n\nFinland\n\nFrance\n\norea\n\nenmark\n\nSlo ak Republic\n\nC echia\n\nEstonia\n\new ealand\n\nChile\n\nSingapore\n\nithuania\n\nIsrael\n\nHungary\n\nPercentage point differences 10\n\n8\n\n2\n\n0\n\n2\n\nRound, Cycle 1:","pages":[{"line_numbers":[],"markdown":"**128** \n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 3.14. Change in the share of foreign-born adults in the adult population between cycles**\n\nNote: Adults aged 16-65; does not include adults who in Cycle 2 were only administered the doorstep interview due to a language barrier, to maximise the comparability across cycles (see Box 1.1 in Chapter 1 and Box 3.2). Japan and Poland are excluded due to small numbers of foreign-born adults. _Countries and economies are ranked in descending order of the change in the share of foreign-born adults._ Source: OECD (2018[4]; 2015[5]; 2012[6]), _Survey of Adult Skills (PIAAC) databases_ , http://www.oecd.org/skills/piaac/publicdataandanalysis/ (accessed on 23 September 2024); Table B.3.10 (Trend) in Annex B.\n\nOverall, the patterns of change for the native-born population largely mirror those of the entire adult population, but there are exceptions in some countries. In the Flemish Region (Belgium), Germany and Norway, the literacy proficiency of adults without an immigrant background has improved since the first cycle, and it has remained stable in France. Meanwhile, the proficiency of adults with an immigrant background developed less favourably in these countries and economies.\n\nOther countries have also experienced less favourable trends for immigrants than for native-born adults. In Chile and Singapore, literacy proficiency declined among immigrants, but not among native-born adults. In Austria and New Zealand, the decline in literacy proficiency for foreign-born adults was more pronounced than for native-born adults.\n\nMigrants’ numeracy skills have developed more positively in many countries over the past decade. Canada, Denmark, England (United Kingdom), Estonia, Finland, Norway and Sweden saw improvements in the proficiency of foreign-born adults (see Table A.3.14 (N) in Annex A). In Finland, Norway and Sweden, these gains were particularly large, exceeding 20 points. In contrast, numeracy proficiency among immigrants only declined significantly between cycles in Hungary and New Zealand.\n\nWhen interpreting changes in the literacy proficiency of immigrants it is important to consider how long they have been living in the host country. Learning a new language and integrating into a new society takes time, while policies supporting immigrants’ integration also take time to contribute effectively to skills formation. Recent migrants will have had less of both time and such resources. Their skills proficiency is typically lower and more strongly influenced by the opportunities available in their country of origin as well as their linguistic distance to the host country. Changes in proficiency among recently arrived migrants must therefore be interpreted in the context of the changing profile of incoming migrants, which is shaped by each country’s immigration patterns and policies. The skills profile of long-term migrants may also vary\n\norway\n\nSweden\n\nEngland (U\n\nAustria\n\nSpain\n\nCanada\n\nItaly\n\nIreland\n\nGermany\n\nFlemish Region ( E\n\netherlands\n\nUnited States\n\nFinland\n\nFrance\n\norea\n\nenmark\n\nSlo ak Republic\n\nC echia\n\nEstonia\n\new ealand\n\nChile\n\nSingapore\n\nithuania\n\nIsrael\n\nHungary\n\nPercentage point differences 10\n\n8\n\n2\n\n0\n\n2\n\nRound, Cycle 1:","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**128** \n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 3.14. Change in the share of foreign-born adults in the adult population between cycles**\n\nNote: Adults aged 16-65; does not include adults who in Cycle 2 were only administered the doorstep interview due to a language barrier, to maximise the comparability across cycles (see Box 1.1 in Chapter 1 and Box 3.2). Japan and Poland are excluded due to small numbers of foreign-born adults. _Countries and economies are ranked in descending order of the change in the share of foreign-born adults._ Source: OECD (2018[4]; 2015[5]; 2012[6]), _Survey of Adult Skills (PIAAC) databases_ , http://www.oecd.org/skills/piaac/publicdataandanalysis/ (accessed on 23 September 2024); Table B.3.10 (Trend) in Annex B.\n\nOverall, the patterns of change for the native-born population largely mirror those of the entire adult population, but there are exceptions in some countries. In the Flemish Region (Belgium), Germany and Norway, the literacy proficiency of adults without an immigrant background has improved since the first cycle, and it has remained stable in France. Meanwhile, the proficiency of adults with an immigrant background developed less favourably in these countries and economies.\n\nOther countries have also experienced less favourable trends for immigrants than for native-born adults. In Chile and Singapore, literacy proficiency declined among immigrants, but not among native-born adults. In Austria and New Zealand, the decline in literacy proficiency for foreign-born adults was more pronounced than for native-born adults.\n\nMigrants’ numeracy skills have developed more positively in many countries over the past decade. Canada, Denmark, England (United Kingdom), Estonia, Finland, Norway and Sweden saw improvements in the proficiency of foreign-born adults (see Table A.3.14 (N) in Annex A). In Finland, Norway and Sweden, these gains were particularly large, exceeding 20 points. In contrast, numeracy proficiency among immigrants only declined significantly between cycles in Hungary and New Zealand.\n\nWhen interpreting changes in the literacy proficiency of immigrants it is important to consider how long they have been living in the host country. Learning a new language and integrating into a new society takes time, while policies supporting immigrants’ integration also take time to contribute effectively to skills formation. Recent migrants will have had less of both time and such resources. Their skills proficiency is typically lower and more strongly influenced by the opportunities available in their country of origin as well as their linguistic distance to the host country. Changes in proficiency among recently arrived migrants must therefore be interpreted in the context of the changing profile of incoming migrants, which is shaped by each country’s immigration patterns and policies. The skills profile of long-term migrants may also vary\n\norway\n\nSweden\n\nEngland (U\n\nAustria\n\nSpain\n\nCanada\n\nItaly\n\nIreland\n\nGermany\n\nFlemish Region ( E\n\netherlands\n\nUnited States\n\nFinland\n\nFrance\n\norea\n\nenmark\n\nSlo ak Republic\n\nC echia\n\nEstonia\n\new ealand\n\nChile\n\nSingapore\n\nithuania\n\nIsrael\n\nHungary\n\nPercentage point differences 10\n\n8\n\n2\n\n0\n\n2\n\nRound, Cycle 1:","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p130.pdf"]},"response":{"errors":[],"extract_id":"extract_93bae17cf5c837e043cfeffb6025fe48","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**128** \n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 3.14. Change in the share of foreign-born adults in the adult population between cycles**\n\nNote: Adults aged 16-65; does not include adults who in Cycle 2 were only administered the doorstep interview due to a language barrier, to maximise the comparability across cycles (see Box 1.1 in Chapter 1 and Box 3.2). Japan and Poland are excluded due to small numbers of foreign-born adults. _Countries and economies are ranked in descending order of the change in the share of foreign-born adults._ Source: OECD (2018[4]; 2015[5]; 2012[6]), _Survey of Adult Skills (PIAAC) databases_ , http://www.oecd.org/skills/piaac/publicdataandanalysis/ (accessed on 23 September 2024); Table B.3.10 (Trend) in Annex B.\n\nOverall, the patterns of change for the native-born population largely mirror those of the entire adult population, but there are exceptions in some countries. In the Flemish Region (Belgium), Germany and Norway, the literacy proficiency of adults without an immigrant background has improved since the first cycle, and it has remained stable in France. Meanwhile, the proficiency of adults with an immigrant background developed less favourably in these countries and economies.\n\nOther countries have also experienced less favourable trends for immigrants than for native-born adults. In Chile and Singapore, literacy proficiency declined among immigrants, but not among native-born adults. In Austria and New Zealand, the decline in literacy proficiency for foreign-born adults was more pronounced than for native-born adults.\n\nMigrants’ numeracy skills have developed more positively in many countries over the past decade. Canada, Denmark, England (United Kingdom), Estonia, Finland, Norway and Sweden saw improvements in the proficiency of foreign-born adults (see Table A.3.14 (N) in Annex A). In Finland, Norway and Sweden, these gains were particularly large, exceeding 20 points. In contrast, numeracy proficiency among immigrants only declined significantly between cycles in Hungary and New Zealand.\n\nWhen interpreting changes in the literacy proficiency of immigrants it is important to consider how long they have been living in the host country. Learning a new language and integrating into a new society takes time, while policies supporting immigrants’ integration also take time to contribute effectively to skills formation. Recent migrants will have had less of both time and such resources. Their skills proficiency is typically lower and more strongly influenced by the opportunities available in their country of origin as well as their linguistic distance to the host country. Changes in proficiency among recently arrived migrants must therefore be interpreted in the context of the changing profile of incoming migrants, which is shaped by each country’s immigration patterns and policies. The skills profile of long-term migrants may also vary\n\norway\n\nSweden\n\nEngland (U\n\nAustria\n\nSpain\n\nCanada\n\nItaly\n\nIreland\n\nGermany\n\nFlemish Region ( E\n\netherlands\n\nUnited States\n\nFinland\n\nFrance\n\norea\n\nenmark\n\nSlo ak Republic\n\nC echia\n\nEstonia\n\new ealand\n\nChile\n\nSingapore\n\nithuania\n\nIsrael\n\nHungary\n\nPercentage point differences 10\n\n8\n\n2\n\n0\n\n2\n\nRound, Cycle 1:"],"full_content":"**128** \n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 3.14. Change in the share of foreign-born adults in the adult population between cycles**\n\nNote: Adults aged 16-65; does not include adults who in Cycle 2 were only administered the doorstep interview due to a language barrier, to maximise the comparability across cycles (see Box 1.1 in Chapter 1 and Box 3.2). Japan and Poland are excluded due to small numbers of foreign-born adults. _Countries and economies are ranked in descending order of the change in the share of foreign-born adults._ Source: OECD (2018[4]; 2015[5]; 2012[6]), _Survey of Adult Skills (PIAAC) databases_ , http://www.oecd.org/skills/piaac/publicdataandanalysis/ (accessed on 23 September 2024); Table B.3.10 (Trend) in Annex B.\n\nOverall, the patterns of change for the native-born population largely mirror those of the entire adult population, but there are exceptions in some countries. In the Flemish Region (Belgium), Germany and Norway, the literacy proficiency of adults without an immigrant background has improved since the first cycle, and it has remained stable in France. Meanwhile, the proficiency of adults with an immigrant background developed less favourably in these countries and economies.\n\nOther countries have also experienced less favourable trends for immigrants than for native-born adults. In Chile and Singapore, literacy proficiency declined among immigrants, but not among native-born adults. In Austria and New Zealand, the decline in literacy proficiency for foreign-born adults was more pronounced than for native-born adults.\n\nMigrants’ numeracy skills have developed more positively in many countries over the past decade. Canada, Denmark, England (United Kingdom), Estonia, Finland, Norway and Sweden saw improvements in the proficiency of foreign-born adults (see Table A.3.14 (N) in Annex A). In Finland, Norway and Sweden, these gains were particularly large, exceeding 20 points. In contrast, numeracy proficiency among immigrants only declined significantly between cycles in Hungary and New Zealand.\n\nWhen interpreting changes in the literacy proficiency of immigrants it is important to consider how long they have been living in the host country. Learning a new language and integrating into a new society takes time, while policies supporting immigrants’ integration also take time to contribute effectively to skills formation. Recent migrants will have had less of both time and such resources. Their skills proficiency is typically lower and more strongly influenced by the opportunities available in their country of origin as well as their linguistic distance to the host country. Changes in proficiency among recently arrived migrants must therefore be interpreted in the context of the changing profile of incoming migrants, which is shaped by each country’s immigration patterns and policies. The skills profile of long-term migrants may also vary\n\norway\n\nSweden\n\nEngland (U\n\nAustria\n\nSpain\n\nCanada\n\nItaly\n\nIreland\n\nGermany\n\nFlemish Region ( E\n\netherlands\n\nUnited States\n\nFinland\n\nFrance\n\norea\n\nenmark\n\nSlo ak Republic\n\nC echia\n\nEstonia\n\new ealand\n\nChile\n\nSingapore\n\nithuania\n\nIsrael\n\nHungary\n\nPercentage point differences 10\n\n8\n\n2\n\n0\n\n2\n\nRound, Cycle 1:","publish_date":null,"title":"Figure 3.14. Change in the share of foreign-born adults in the adult population between cycles","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p130.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p130.pdf"},"request":{"example_id":"chart/b263dc5d-en_p130","product_type":"parse","source_file_path":"b263dc5d-en_p130.pdf"},"started_at":"2026-11-04T19:45:49.928216"} +{"completed_at":"2026-11-04T19:45:53.134230","latency_in_ms":1411,"output":{"example_id":"chart/b263dc5d-en_p131","grounded_pages":[],"layout_pages":[],"markdown":" **129**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\nacross time due to changing patterns of immigration, including return migration in the country of origin. In any event, the skills of long-term migrants are more strongly influenced by conditions in the host country. Changes in their skills over time can therefore offer insights into the effectiveness of the host country’s policies in fostering integration.\n\n### **Figure 3.15. Change in literacy proficiency between cycles, by immigrant background**\n\nAdjusted and unadjusted difference in mean literacy scores between cycles (Cycle 2 _minus_ Cycle 1)\n\nNote: Adults aged 16-65; does not include adults who in Cycle 2 were only administered the doorstep interview due to a language barrier, to maximise the comparability across cycles (see Box 1.1 in Chapter 1 and Box 3.2). Unadjusted differences are the differences between the averages in each cycle. Adjusted differences are based on a method analogous to post-stratification that reweights the samples in Cycle 2 so that the groups by immigrant background have the same demographic characteristics as in Cycle 1 (see Note 1). Demographic characteristics considered are: age and gender. Darker colours denote differences that are statistically significant at the 5% level. Japan and Poland are excluded due to small numbers of foreign-born adults. _Countries and economies are ranked in descending order of the unadjusted change in literacy proficiency among foreign-born adults of foreign- born parents._ Source: OECD (2018[4]; 2015[5]; 2012[6]), _Survey of Adult Skills (PIAAC) databases_ , http://www.oecd.org/skills/piaac/publicdataandanalysis/ (accessed on 23 September 2024); Table A.3.14 (L) in Annex A.\n\nFinland\n\nenmark\n\nSweden\n\nCanada\n\nEstonia\n\nItaly\n\norway\n\nEngland (U\n\nC echia\n\nSpain\n\nIreland\n\netherlands\n\nFlemish Region ( E\n\nFrance\n\nUnited States\n\nGermany\n\nAustria\n\nSlo ak Republic\n\norea\n\nithuania\n\nIsrael\n\nChile\n\nSingapore\n\new ealand\n\nHungary\n\nFinland\n\nenmark\n\nSweden\n\nCanada\n\nEstonia\n\nItaly\n\norway\n\nEngland (U\n\nC echia\n\nSpain\n\nIreland\n\netherlands\n\nFlemish Region ( E\n\nFrance\n\nUnited States\n\nGermany\n\nAustria\n\nSlo ak Republic\n\norea\n\nithuania\n\nIsrael\n\nChile\n\nSingapore\n\new ealand\n\nHungary\n\nAd usted Unad u sted\n\n30\n\n20\n\n10\n\n0\n\n10\n\n20\n\n30\n\n0\n\n30\n\n20\n\n10\n\n0\n\n10\n\n20\n\n30\n\n0\n\nRound, Cycle 1:","pages":[{"line_numbers":[],"markdown":" **129**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\nacross time due to changing patterns of immigration, including return migration in the country of origin. In any event, the skills of long-term migrants are more strongly influenced by conditions in the host country. Changes in their skills over time can therefore offer insights into the effectiveness of the host country’s policies in fostering integration.\n\n### **Figure 3.15. Change in literacy proficiency between cycles, by immigrant background**\n\nAdjusted and unadjusted difference in mean literacy scores between cycles (Cycle 2 _minus_ Cycle 1)\n\nNote: Adults aged 16-65; does not include adults who in Cycle 2 were only administered the doorstep interview due to a language barrier, to maximise the comparability across cycles (see Box 1.1 in Chapter 1 and Box 3.2). Unadjusted differences are the differences between the averages in each cycle. Adjusted differences are based on a method analogous to post-stratification that reweights the samples in Cycle 2 so that the groups by immigrant background have the same demographic characteristics as in Cycle 1 (see Note 1). Demographic characteristics considered are: age and gender. Darker colours denote differences that are statistically significant at the 5% level. Japan and Poland are excluded due to small numbers of foreign-born adults. _Countries and economies are ranked in descending order of the unadjusted change in literacy proficiency among foreign-born adults of foreign- born parents._ Source: OECD (2018[4]; 2015[5]; 2012[6]), _Survey of Adult Skills (PIAAC) databases_ , http://www.oecd.org/skills/piaac/publicdataandanalysis/ (accessed on 23 September 2024); Table A.3.14 (L) in Annex A.\n\nFinland\n\nenmark\n\nSweden\n\nCanada\n\nEstonia\n\nItaly\n\norway\n\nEngland (U\n\nC echia\n\nSpain\n\nIreland\n\netherlands\n\nFlemish Region ( E\n\nFrance\n\nUnited States\n\nGermany\n\nAustria\n\nSlo ak Republic\n\norea\n\nithuania\n\nIsrael\n\nChile\n\nSingapore\n\new ealand\n\nHungary\n\nFinland\n\nenmark\n\nSweden\n\nCanada\n\nEstonia\n\nItaly\n\norway\n\nEngland (U\n\nC echia\n\nSpain\n\nIreland\n\netherlands\n\nFlemish Region ( E\n\nFrance\n\nUnited States\n\nGermany\n\nAustria\n\nSlo ak Republic\n\norea\n\nithuania\n\nIsrael\n\nChile\n\nSingapore\n\new ealand\n\nHungary\n\nAd usted Unad u sted\n\n30\n\n20\n\n10\n\n0\n\n10\n\n20\n\n30\n\n0\n\n30\n\n20\n\n10\n\n0\n\n10\n\n20\n\n30\n\n0\n\nRound, Cycle 1:","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":" **129**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\nacross time due to changing patterns of immigration, including return migration in the country of origin. In any event, the skills of long-term migrants are more strongly influenced by conditions in the host country. Changes in their skills over time can therefore offer insights into the effectiveness of the host country’s policies in fostering integration.\n\n### **Figure 3.15. Change in literacy proficiency between cycles, by immigrant background**\n\nAdjusted and unadjusted difference in mean literacy scores between cycles (Cycle 2 _minus_ Cycle 1)\n\nNote: Adults aged 16-65; does not include adults who in Cycle 2 were only administered the doorstep interview due to a language barrier, to maximise the comparability across cycles (see Box 1.1 in Chapter 1 and Box 3.2). Unadjusted differences are the differences between the averages in each cycle. Adjusted differences are based on a method analogous to post-stratification that reweights the samples in Cycle 2 so that the groups by immigrant background have the same demographic characteristics as in Cycle 1 (see Note 1). Demographic characteristics considered are: age and gender. Darker colours denote differences that are statistically significant at the 5% level. Japan and Poland are excluded due to small numbers of foreign-born adults. _Countries and economies are ranked in descending order of the unadjusted change in literacy proficiency among foreign-born adults of foreign- born parents._ Source: OECD (2018[4]; 2015[5]; 2012[6]), _Survey of Adult Skills (PIAAC) databases_ , http://www.oecd.org/skills/piaac/publicdataandanalysis/ (accessed on 23 September 2024); Table A.3.14 (L) in Annex A.\n\nFinland\n\nenmark\n\nSweden\n\nCanada\n\nEstonia\n\nItaly\n\norway\n\nEngland (U\n\nC echia\n\nSpain\n\nIreland\n\netherlands\n\nFlemish Region ( E\n\nFrance\n\nUnited States\n\nGermany\n\nAustria\n\nSlo ak Republic\n\norea\n\nithuania\n\nIsrael\n\nChile\n\nSingapore\n\new ealand\n\nHungary\n\nFinland\n\nenmark\n\nSweden\n\nCanada\n\nEstonia\n\nItaly\n\norway\n\nEngland (U\n\nC echia\n\nSpain\n\nIreland\n\netherlands\n\nFlemish Region ( E\n\nFrance\n\nUnited States\n\nGermany\n\nAustria\n\nSlo ak Republic\n\norea\n\nithuania\n\nIsrael\n\nChile\n\nSingapore\n\new ealand\n\nHungary\n\nAd usted Unad u sted\n\n30\n\n20\n\n10\n\n0\n\n10\n\n20\n\n30\n\n0\n\n30\n\n20\n\n10\n\n0\n\n10\n\n20\n\n30\n\n0\n\nRound, Cycle 1:","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p131.pdf"]},"response":{"errors":[],"extract_id":"extract_ff290e27f9a696cebc207a8a000a7798","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":[" **129**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\nacross time due to changing patterns of immigration, including return migration in the country of origin. In any event, the skills of long-term migrants are more strongly influenced by conditions in the host country. Changes in their skills over time can therefore offer insights into the effectiveness of the host country’s policies in fostering integration.\n\n### **Figure 3.15. Change in literacy proficiency between cycles, by immigrant background**\n\nAdjusted and unadjusted difference in mean literacy scores between cycles (Cycle 2 _minus_ Cycle 1)\n\nNote: Adults aged 16-65; does not include adults who in Cycle 2 were only administered the doorstep interview due to a language barrier, to maximise the comparability across cycles (see Box 1.1 in Chapter 1 and Box 3.2). Unadjusted differences are the differences between the averages in each cycle. Adjusted differences are based on a method analogous to post-stratification that reweights the samples in Cycle 2 so that the groups by immigrant background have the same demographic characteristics as in Cycle 1 (see Note 1). Demographic characteristics considered are: age and gender. Darker colours denote differences that are statistically significant at the 5% level. Japan and Poland are excluded due to small numbers of foreign-born adults. _Countries and economies are ranked in descending order of the unadjusted change in literacy proficiency among foreign-born adults of foreign- born parents._ Source: OECD (2018[4]; 2015[5]; 2012[6]), _Survey of Adult Skills (PIAAC) databases_ , http://www.oecd.org/skills/piaac/publicdataandanalysis/ (accessed on 23 September 2024); Table A.3.14 (L) in Annex A.\n\nFinland\n\nenmark\n\nSweden\n\nCanada\n\nEstonia\n\nItaly\n\norway\n\nEngland (U\n\nC echia\n\nSpain\n\nIreland\n\netherlands\n\nFlemish Region ( E\n\nFrance\n\nUnited States\n\nGermany\n\nAustria\n\nSlo ak Republic\n\norea\n\nithuania\n\nIsrael\n\nChile\n\nSingapore\n\new ealand\n\nHungary\n\nFinland\n\nenmark\n\nSweden\n\nCanada\n\nEstonia\n\nItaly\n\norway\n\nEngland (U\n\nC echia\n\nSpain\n\nIreland\n\netherlands\n\nFlemish Region ( E\n\nFrance\n\nUnited States\n\nGermany\n\nAustria\n\nSlo ak Republic\n\norea\n\nithuania\n\nIsrael\n\nChile\n\nSingapore\n\new ealand\n\nHungary\n\nAd usted Unad u sted\n\n30\n\n20\n\n10\n\n0\n\n10\n\n20\n\n30\n\n0\n\n30\n\n20\n\n10\n\n0\n\n10\n\n20\n\n30\n\n0\n\nRound, Cycle 1:"],"full_content":" **129**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\nacross time due to changing patterns of immigration, including return migration in the country of origin. In any event, the skills of long-term migrants are more strongly influenced by conditions in the host country. Changes in their skills over time can therefore offer insights into the effectiveness of the host country’s policies in fostering integration.\n\n### **Figure 3.15. Change in literacy proficiency between cycles, by immigrant background**\n\nAdjusted and unadjusted difference in mean literacy scores between cycles (Cycle 2 _minus_ Cycle 1)\n\nNote: Adults aged 16-65; does not include adults who in Cycle 2 were only administered the doorstep interview due to a language barrier, to maximise the comparability across cycles (see Box 1.1 in Chapter 1 and Box 3.2). Unadjusted differences are the differences between the averages in each cycle. Adjusted differences are based on a method analogous to post-stratification that reweights the samples in Cycle 2 so that the groups by immigrant background have the same demographic characteristics as in Cycle 1 (see Note 1). Demographic characteristics considered are: age and gender. Darker colours denote differences that are statistically significant at the 5% level. Japan and Poland are excluded due to small numbers of foreign-born adults. _Countries and economies are ranked in descending order of the unadjusted change in literacy proficiency among foreign-born adults of foreign- born parents._ Source: OECD (2018[4]; 2015[5]; 2012[6]), _Survey of Adult Skills (PIAAC) databases_ , http://www.oecd.org/skills/piaac/publicdataandanalysis/ (accessed on 23 September 2024); Table A.3.14 (L) in Annex A.\n\nFinland\n\nenmark\n\nSweden\n\nCanada\n\nEstonia\n\nItaly\n\norway\n\nEngland (U\n\nC echia\n\nSpain\n\nIreland\n\netherlands\n\nFlemish Region ( E\n\nFrance\n\nUnited States\n\nGermany\n\nAustria\n\nSlo ak Republic\n\norea\n\nithuania\n\nIsrael\n\nChile\n\nSingapore\n\new ealand\n\nHungary\n\nFinland\n\nenmark\n\nSweden\n\nCanada\n\nEstonia\n\nItaly\n\norway\n\nEngland (U\n\nC echia\n\nSpain\n\nIreland\n\netherlands\n\nFlemish Region ( E\n\nFrance\n\nUnited States\n\nGermany\n\nAustria\n\nSlo ak Republic\n\norea\n\nithuania\n\nIsrael\n\nChile\n\nSingapore\n\new ealand\n\nHungary\n\nAd usted Unad u sted\n\n30\n\n20\n\n10\n\n0\n\n10\n\n20\n\n30\n\n0\n\n30\n\n20\n\n10\n\n0\n\n10\n\n20\n\n30\n\n0\n\nRound, Cycle 1:","publish_date":null,"title":"Figure 3.15. Change in literacy proficiency between cycles, by immigrant background","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p131.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p131.pdf"},"request":{"example_id":"chart/b263dc5d-en_p131","product_type":"parse","source_file_path":"b263dc5d-en_p131.pdf"},"started_at":"2026-11-04T19:45:51.722671"} +{"completed_at":"2026-11-04T19:31:07.332794","latency_in_ms":2834,"output":{"example_id":"chart/b263dc5d-en_p132","grounded_pages":[],"layout_pages":[],"markdown":"**130** \n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 3.16. Change in literacy proficiency between cycles, by immigrant background and years spent in the country**\n\nDifference in mean literacy scores between cycles (Cycle 2 _minus_ Cycle 1)\n\nNote: Adults aged 16-65; does not include adults who in Cycle 2 were only administered the doorstep interview due to a language barrier, to maximise the comparability across cycles (see Box 1.1 in Chapter 1 and Box 3.2). Differences are unadjusted. Darker colours denote differences that are statistically significant at the 5% level. Chile, Hungary, Japan, Korea, Lithuania, Poland and the Slovak Republic are excluded due to small numbers of cases in particular subgroups. In the United States, more than 20% of the subsample of foreign-born adults of foreign-born parents who had been living in the country/economy for less than 10 years only completed the assessment of reading and numeracy components in the second cycle. As the methodology for estimating the proficiency of adults who only took the components assessment has changed between the cycles, the change in proficiency for this group is not reported. _Countries and economies are ranked in descending order of the change in literacy proficiency among foreign-born adults who had been living in the country/economy for less than 10 years._ Source: OECD (2018[4]; 2015[5]; 2012[6]), _Survey of Adult Skills (PIAAC) databases_ , http://www.oecd.org/skills/piaac/publicdataandanalysis/ (accessed on 23 September 2024); Table A.3.14 (L) in Annex A.\n\nSome countries have seen literacy proficiency for foreign-born adults evolve differently among those who have lived in the host country for less than ten years and those who have lived there longer (Figure 3.16). In Germany, the skills of recent migrants in 2023 are significantly lower than those of recent migrants in the previous cycle and no changes are observed among long-term migrants. While the average score in 2023 of recent migrants to Germany (203 points) was among the lowest across participating countries and economies, in the first cycle the average score for this subgroup (248 points) was close to the OECD average (240 points; see Table A.3.14 (L) in Annex A). In contrast, in Austria and New Zealand, literacy proficiency declined among both recent and long-term migrants. Estonia, France, Ireland and Singapore saw significant declines among long-term migrants.\n\n_Changes in the proficiency gap between foreign-born and native-born adults_\n\nDeclining literacy proficiency among immigrants leaves them at a significant disadvantage. It means that more immigrants will find it difficult to participate effectively in their host countries’ labour markets or integrate into their communities. Declines in skills among migrants also pose a challenge for society as a\n\nFinland\n\nenmark\n\nEstonia\n\nSweden\n\nEngland (U\n\nIreland\n\nCanada\n\norway\n\nSpain\n\nItaly\n\nFlemish Region ( E\n\nFrance\n\netherlands\n\nAustria\n\nC echia\n\nGermany\n\nSingapore\n\nIsrael\n\new ealand\n\nScore point difference 0\n\n30\n\n20\n\n10\n\n0\n\n10\n\n20\n\n30\n\n0\n\n0\n\nForeign born, more than 10 years in country ati e born\n\nForeign born, less than 10 years in country\n\nRound, Cycle 1:","pages":[{"line_numbers":[],"markdown":"**130** \n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 3.16. Change in literacy proficiency between cycles, by immigrant background and years spent in the country**\n\nDifference in mean literacy scores between cycles (Cycle 2 _minus_ Cycle 1)\n\nNote: Adults aged 16-65; does not include adults who in Cycle 2 were only administered the doorstep interview due to a language barrier, to maximise the comparability across cycles (see Box 1.1 in Chapter 1 and Box 3.2). Differences are unadjusted. Darker colours denote differences that are statistically significant at the 5% level. Chile, Hungary, Japan, Korea, Lithuania, Poland and the Slovak Republic are excluded due to small numbers of cases in particular subgroups. In the United States, more than 20% of the subsample of foreign-born adults of foreign-born parents who had been living in the country/economy for less than 10 years only completed the assessment of reading and numeracy components in the second cycle. As the methodology for estimating the proficiency of adults who only took the components assessment has changed between the cycles, the change in proficiency for this group is not reported. _Countries and economies are ranked in descending order of the change in literacy proficiency among foreign-born adults who had been living in the country/economy for less than 10 years._ Source: OECD (2018[4]; 2015[5]; 2012[6]), _Survey of Adult Skills (PIAAC) databases_ , http://www.oecd.org/skills/piaac/publicdataandanalysis/ (accessed on 23 September 2024); Table A.3.14 (L) in Annex A.\n\nSome countries have seen literacy proficiency for foreign-born adults evolve differently among those who have lived in the host country for less than ten years and those who have lived there longer (Figure 3.16). In Germany, the skills of recent migrants in 2023 are significantly lower than those of recent migrants in the previous cycle and no changes are observed among long-term migrants. While the average score in 2023 of recent migrants to Germany (203 points) was among the lowest across participating countries and economies, in the first cycle the average score for this subgroup (248 points) was close to the OECD average (240 points; see Table A.3.14 (L) in Annex A). In contrast, in Austria and New Zealand, literacy proficiency declined among both recent and long-term migrants. Estonia, France, Ireland and Singapore saw significant declines among long-term migrants.\n\n_Changes in the proficiency gap between foreign-born and native-born adults_\n\nDeclining literacy proficiency among immigrants leaves them at a significant disadvantage. It means that more immigrants will find it difficult to participate effectively in their host countries’ labour markets or integrate into their communities. Declines in skills among migrants also pose a challenge for society as a\n\nFinland\n\nenmark\n\nEstonia\n\nSweden\n\nEngland (U\n\nIreland\n\nCanada\n\norway\n\nSpain\n\nItaly\n\nFlemish Region ( E\n\nFrance\n\netherlands\n\nAustria\n\nC echia\n\nGermany\n\nSingapore\n\nIsrael\n\new ealand\n\nScore point difference 0\n\n30\n\n20\n\n10\n\n0\n\n10\n\n20\n\n30\n\n0\n\n0\n\nForeign born, more than 10 years in country ati e born\n\nForeign born, less than 10 years in country\n\nRound, Cycle 1:","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**130** \n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 3.16. Change in literacy proficiency between cycles, by immigrant background and years spent in the country**\n\nDifference in mean literacy scores between cycles (Cycle 2 _minus_ Cycle 1)\n\nNote: Adults aged 16-65; does not include adults who in Cycle 2 were only administered the doorstep interview due to a language barrier, to maximise the comparability across cycles (see Box 1.1 in Chapter 1 and Box 3.2). Differences are unadjusted. Darker colours denote differences that are statistically significant at the 5% level. Chile, Hungary, Japan, Korea, Lithuania, Poland and the Slovak Republic are excluded due to small numbers of cases in particular subgroups. In the United States, more than 20% of the subsample of foreign-born adults of foreign-born parents who had been living in the country/economy for less than 10 years only completed the assessment of reading and numeracy components in the second cycle. As the methodology for estimating the proficiency of adults who only took the components assessment has changed between the cycles, the change in proficiency for this group is not reported. _Countries and economies are ranked in descending order of the change in literacy proficiency among foreign-born adults who had been living in the country/economy for less than 10 years._ Source: OECD (2018[4]; 2015[5]; 2012[6]), _Survey of Adult Skills (PIAAC) databases_ , http://www.oecd.org/skills/piaac/publicdataandanalysis/ (accessed on 23 September 2024); Table A.3.14 (L) in Annex A.\n\nSome countries have seen literacy proficiency for foreign-born adults evolve differently among those who have lived in the host country for less than ten years and those who have lived there longer (Figure 3.16). In Germany, the skills of recent migrants in 2023 are significantly lower than those of recent migrants in the previous cycle and no changes are observed among long-term migrants. While the average score in 2023 of recent migrants to Germany (203 points) was among the lowest across participating countries and economies, in the first cycle the average score for this subgroup (248 points) was close to the OECD average (240 points; see Table A.3.14 (L) in Annex A). In contrast, in Austria and New Zealand, literacy proficiency declined among both recent and long-term migrants. Estonia, France, Ireland and Singapore saw significant declines among long-term migrants.\n\n_Changes in the proficiency gap between foreign-born and native-born adults_\n\nDeclining literacy proficiency among immigrants leaves them at a significant disadvantage. It means that more immigrants will find it difficult to participate effectively in their host countries’ labour markets or integrate into their communities. Declines in skills among migrants also pose a challenge for society as a\n\nFinland\n\nenmark\n\nEstonia\n\nSweden\n\nEngland (U\n\nIreland\n\nCanada\n\norway\n\nSpain\n\nItaly\n\nFlemish Region ( E\n\nFrance\n\netherlands\n\nAustria\n\nC echia\n\nGermany\n\nSingapore\n\nIsrael\n\new ealand\n\nScore point difference 0\n\n30\n\n20\n\n10\n\n0\n\n10\n\n20\n\n30\n\n0\n\n0\n\nForeign born, more than 10 years in country ati e born\n\nForeign born, less than 10 years in country\n\nRound, Cycle 1:","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p132.pdf"]},"response":{"errors":[],"extract_id":"extract_3c9107da9f27417fdc4fcc5f5602c22f","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**130** \n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 3.16. Change in literacy proficiency between cycles, by immigrant background and years spent in the country**\n\nDifference in mean literacy scores between cycles (Cycle 2 _minus_ Cycle 1)\n\nNote: Adults aged 16-65; does not include adults who in Cycle 2 were only administered the doorstep interview due to a language barrier, to maximise the comparability across cycles (see Box 1.1 in Chapter 1 and Box 3.2). Differences are unadjusted. Darker colours denote differences that are statistically significant at the 5% level. Chile, Hungary, Japan, Korea, Lithuania, Poland and the Slovak Republic are excluded due to small numbers of cases in particular subgroups. In the United States, more than 20% of the subsample of foreign-born adults of foreign-born parents who had been living in the country/economy for less than 10 years only completed the assessment of reading and numeracy components in the second cycle. As the methodology for estimating the proficiency of adults who only took the components assessment has changed between the cycles, the change in proficiency for this group is not reported. _Countries and economies are ranked in descending order of the change in literacy proficiency among foreign-born adults who had been living in the country/economy for less than 10 years._ Source: OECD (2018[4]; 2015[5]; 2012[6]), _Survey of Adult Skills (PIAAC) databases_ , http://www.oecd.org/skills/piaac/publicdataandanalysis/ (accessed on 23 September 2024); Table A.3.14 (L) in Annex A.\n\nSome countries have seen literacy proficiency for foreign-born adults evolve differently among those who have lived in the host country for less than ten years and those who have lived there longer (Figure 3.16). In Germany, the skills of recent migrants in 2023 are significantly lower than those of recent migrants in the previous cycle and no changes are observed among long-term migrants. While the average score in 2023 of recent migrants to Germany (203 points) was among the lowest across participating countries and economies, in the first cycle the average score for this subgroup (248 points) was close to the OECD average (240 points; see Table A.3.14 (L) in Annex A). In contrast, in Austria and New Zealand, literacy proficiency declined among both recent and long-term migrants. Estonia, France, Ireland and Singapore saw significant declines among long-term migrants.\n\n_Changes in the proficiency gap between foreign-born and native-born adults_\n\nDeclining literacy proficiency among immigrants leaves them at a significant disadvantage. It means that more immigrants will find it difficult to participate effectively in their host countries’ labour markets or integrate into their communities. Declines in skills among migrants also pose a challenge for society as a\n\nFinland\n\nenmark\n\nEstonia\n\nSweden\n\nEngland (U\n\nIreland\n\nCanada\n\norway\n\nSpain\n\nItaly\n\nFlemish Region ( E\n\nFrance\n\netherlands\n\nAustria\n\nC echia\n\nGermany\n\nSingapore\n\nIsrael\n\new ealand\n\nScore point difference 0\n\n30\n\n20\n\n10\n\n0\n\n10\n\n20\n\n30\n\n0\n\n0\n\nForeign born, more than 10 years in country ati e born\n\nForeign born, less than 10 years in country\n\nRound, Cycle 1:"],"full_content":"**130** \n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 3.16. Change in literacy proficiency between cycles, by immigrant background and years spent in the country**\n\nDifference in mean literacy scores between cycles (Cycle 2 _minus_ Cycle 1)\n\nNote: Adults aged 16-65; does not include adults who in Cycle 2 were only administered the doorstep interview due to a language barrier, to maximise the comparability across cycles (see Box 1.1 in Chapter 1 and Box 3.2). Differences are unadjusted. Darker colours denote differences that are statistically significant at the 5% level. Chile, Hungary, Japan, Korea, Lithuania, Poland and the Slovak Republic are excluded due to small numbers of cases in particular subgroups. In the United States, more than 20% of the subsample of foreign-born adults of foreign-born parents who had been living in the country/economy for less than 10 years only completed the assessment of reading and numeracy components in the second cycle. As the methodology for estimating the proficiency of adults who only took the components assessment has changed between the cycles, the change in proficiency for this group is not reported. _Countries and economies are ranked in descending order of the change in literacy proficiency among foreign-born adults who had been living in the country/economy for less than 10 years._ Source: OECD (2018[4]; 2015[5]; 2012[6]), _Survey of Adult Skills (PIAAC) databases_ , http://www.oecd.org/skills/piaac/publicdataandanalysis/ (accessed on 23 September 2024); Table A.3.14 (L) in Annex A.\n\nSome countries have seen literacy proficiency for foreign-born adults evolve differently among those who have lived in the host country for less than ten years and those who have lived there longer (Figure 3.16). In Germany, the skills of recent migrants in 2023 are significantly lower than those of recent migrants in the previous cycle and no changes are observed among long-term migrants. While the average score in 2023 of recent migrants to Germany (203 points) was among the lowest across participating countries and economies, in the first cycle the average score for this subgroup (248 points) was close to the OECD average (240 points; see Table A.3.14 (L) in Annex A). In contrast, in Austria and New Zealand, literacy proficiency declined among both recent and long-term migrants. Estonia, France, Ireland and Singapore saw significant declines among long-term migrants.\n\n_Changes in the proficiency gap between foreign-born and native-born adults_\n\nDeclining literacy proficiency among immigrants leaves them at a significant disadvantage. It means that more immigrants will find it difficult to participate effectively in their host countries’ labour markets or integrate into their communities. Declines in skills among migrants also pose a challenge for society as a\n\nFinland\n\nenmark\n\nEstonia\n\nSweden\n\nEngland (U\n\nIreland\n\nCanada\n\norway\n\nSpain\n\nItaly\n\nFlemish Region ( E\n\nFrance\n\netherlands\n\nAustria\n\nC echia\n\nGermany\n\nSingapore\n\nIsrael\n\new ealand\n\nScore point difference 0\n\n30\n\n20\n\n10\n\n0\n\n10\n\n20\n\n30\n\n0\n\n0\n\nForeign born, more than 10 years in country ati e born\n\nForeign born, less than 10 years in country\n\nRound, Cycle 1:","publish_date":null,"title":"Figure 3.16. Change in literacy proficiency between cycles, by immigrant background and years spent in the country","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p132.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p132.pdf"},"request":{"example_id":"chart/b263dc5d-en_p132","product_type":"parse","source_file_path":"b263dc5d-en_p132.pdf"},"started_at":"2026-11-04T19:31:04.498092"} +{"completed_at":"2026-11-04T19:45:58.482716","latency_in_ms":3209,"output":{"example_id":"chart/b263dc5d-en_p133","grounded_pages":[],"layout_pages":[],"markdown":" **131**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\nwhole, as a widening skills gap between migrants and the native-born population risks undermining equity and social cohesion. In eight of the countries and economies with available data, the gap in literacy proficiency between migrants and non-migrants has widened (Figure 3.17). This was typically due to proficiency among foreign-born adults changing less favourably than that of native-born adults (Figure 3.15). Only Norway and the Flemish Region (Belgium) have seen the gap widening because the skills of native-born adults improved while those of foreign-born adults remained stable.\n\n### **Figure 3.17. Change in the gap in literacy proficiency between non-immigrants and immigrants**\n\nAdjusted and unadjusted change between cycles in the mean score difference between native-born adults with native-born parents and foreign-born adults with foreign-born parents (Cycle 2 _minus_ Cycle 1)\n\nNote: Adults aged 16-65; does not include adults who in Cycle 2 were only administered the doorstep interview due to a language barrier, to maximise the comparability across cycles (see Box 1.1 in Chapter 1 and Box 3.2). Unadjusted changes in the proficiency gap are the changes in the difference between the two contrast categories across the cycles. Adjusted changes are obtained by subtracting the coefficients of regression models estimated separately for each cycle. These coefficients show the difference between the categories in each cycle, after accounting for: gender, age, educational attainment, parental education and language spoken at home. Darker colours denote differences that are statistically significant at the 5% level. Japan and Poland are excluded due to small numbers of foreign-born adults. _Countries and economies are ranked in ascending order of the unadjusted change in the gap._ Source: OECD (2018[4]; 2015[5]; 2012[6]), _Survey of Adult Skills (PIAAC) databases_ , http://www.oecd.org/skills/piaac/publicdataandanalysis/ (accessed on 23 September 2024); Tables A.3.15 (L) and A.3.16 (L) in Annex A.\n\nThe widening of the proficiency gap between foreign- and native-born adults was particularly pronounced in Chile, Germany and Singapore (Figure 3.17). As described above, Germany experienced an increase in the relative size of its immigrant population, coupled with low skills among newly arrived migrants (Figure 3.14; see also Table A.3.14 (L) in Annex A). In Singapore, literacy proficiency among long-term migrants has significantly declined (Figure 3.16). Accounting for differences between foreign- and native- born adults in educational attainment, use of the language of the assessment and demographic characteristics accounts for only a fraction of the change in the proficiency gap in these countries. This suggests that additional factors are at play in widening the difference in literacy between these groups.\n\nNumeracy proficiency increased more strongly among migrants than among native-born adults in Canada, England (United Kingdom), Estonia, Finland, Norway and Sweden, narrowing the gap in numeracy between the two groups (see Table A.3.15 (N) in Annex A). In contrast, the gap widened in the\n\nGermany\n\nAustria\n\nFlemish Region ( E\n\nFrance\n\norway\n\nUnited States\n\nSlo ak Republic\n\norea\n\netherlands\n\nEngland (U\n\nSpain\n\nIreland\n\nEstonia\n\nSweden\n\nFinland\n\nenmark\n\nCanada\n\nItaly\n\nC echia\n\nSingapore\n\nChile\n\new ealand\n\nIsrael\n\nithuania\n\nHungary\n\nScore point difference 0\n\n30\n\n20\n\n10\n\n0\n\n10\n\n20\n\nAd usted Unad u sted\n\nRound, Cycle 1:","pages":[{"line_numbers":[],"markdown":" **131**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\nwhole, as a widening skills gap between migrants and the native-born population risks undermining equity and social cohesion. In eight of the countries and economies with available data, the gap in literacy proficiency between migrants and non-migrants has widened (Figure 3.17). This was typically due to proficiency among foreign-born adults changing less favourably than that of native-born adults (Figure 3.15). Only Norway and the Flemish Region (Belgium) have seen the gap widening because the skills of native-born adults improved while those of foreign-born adults remained stable.\n\n### **Figure 3.17. Change in the gap in literacy proficiency between non-immigrants and immigrants**\n\nAdjusted and unadjusted change between cycles in the mean score difference between native-born adults with native-born parents and foreign-born adults with foreign-born parents (Cycle 2 _minus_ Cycle 1)\n\nNote: Adults aged 16-65; does not include adults who in Cycle 2 were only administered the doorstep interview due to a language barrier, to maximise the comparability across cycles (see Box 1.1 in Chapter 1 and Box 3.2). Unadjusted changes in the proficiency gap are the changes in the difference between the two contrast categories across the cycles. Adjusted changes are obtained by subtracting the coefficients of regression models estimated separately for each cycle. These coefficients show the difference between the categories in each cycle, after accounting for: gender, age, educational attainment, parental education and language spoken at home. Darker colours denote differences that are statistically significant at the 5% level. Japan and Poland are excluded due to small numbers of foreign-born adults. _Countries and economies are ranked in ascending order of the unadjusted change in the gap._ Source: OECD (2018[4]; 2015[5]; 2012[6]), _Survey of Adult Skills (PIAAC) databases_ , http://www.oecd.org/skills/piaac/publicdataandanalysis/ (accessed on 23 September 2024); Tables A.3.15 (L) and A.3.16 (L) in Annex A.\n\nThe widening of the proficiency gap between foreign- and native-born adults was particularly pronounced in Chile, Germany and Singapore (Figure 3.17). As described above, Germany experienced an increase in the relative size of its immigrant population, coupled with low skills among newly arrived migrants (Figure 3.14; see also Table A.3.14 (L) in Annex A). In Singapore, literacy proficiency among long-term migrants has significantly declined (Figure 3.16). Accounting for differences between foreign- and native- born adults in educational attainment, use of the language of the assessment and demographic characteristics accounts for only a fraction of the change in the proficiency gap in these countries. This suggests that additional factors are at play in widening the difference in literacy between these groups.\n\nNumeracy proficiency increased more strongly among migrants than among native-born adults in Canada, England (United Kingdom), Estonia, Finland, Norway and Sweden, narrowing the gap in numeracy between the two groups (see Table A.3.15 (N) in Annex A). In contrast, the gap widened in the\n\nGermany\n\nAustria\n\nFlemish Region ( E\n\nFrance\n\norway\n\nUnited States\n\nSlo ak Republic\n\norea\n\netherlands\n\nEngland (U\n\nSpain\n\nIreland\n\nEstonia\n\nSweden\n\nFinland\n\nenmark\n\nCanada\n\nItaly\n\nC echia\n\nSingapore\n\nChile\n\new ealand\n\nIsrael\n\nithuania\n\nHungary\n\nScore point difference 0\n\n30\n\n20\n\n10\n\n0\n\n10\n\n20\n\nAd usted Unad u sted\n\nRound, Cycle 1:","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":" **131**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\nwhole, as a widening skills gap between migrants and the native-born population risks undermining equity and social cohesion. In eight of the countries and economies with available data, the gap in literacy proficiency between migrants and non-migrants has widened (Figure 3.17). This was typically due to proficiency among foreign-born adults changing less favourably than that of native-born adults (Figure 3.15). Only Norway and the Flemish Region (Belgium) have seen the gap widening because the skills of native-born adults improved while those of foreign-born adults remained stable.\n\n### **Figure 3.17. Change in the gap in literacy proficiency between non-immigrants and immigrants**\n\nAdjusted and unadjusted change between cycles in the mean score difference between native-born adults with native-born parents and foreign-born adults with foreign-born parents (Cycle 2 _minus_ Cycle 1)\n\nNote: Adults aged 16-65; does not include adults who in Cycle 2 were only administered the doorstep interview due to a language barrier, to maximise the comparability across cycles (see Box 1.1 in Chapter 1 and Box 3.2). Unadjusted changes in the proficiency gap are the changes in the difference between the two contrast categories across the cycles. Adjusted changes are obtained by subtracting the coefficients of regression models estimated separately for each cycle. These coefficients show the difference between the categories in each cycle, after accounting for: gender, age, educational attainment, parental education and language spoken at home. Darker colours denote differences that are statistically significant at the 5% level. Japan and Poland are excluded due to small numbers of foreign-born adults. _Countries and economies are ranked in ascending order of the unadjusted change in the gap._ Source: OECD (2018[4]; 2015[5]; 2012[6]), _Survey of Adult Skills (PIAAC) databases_ , http://www.oecd.org/skills/piaac/publicdataandanalysis/ (accessed on 23 September 2024); Tables A.3.15 (L) and A.3.16 (L) in Annex A.\n\nThe widening of the proficiency gap between foreign- and native-born adults was particularly pronounced in Chile, Germany and Singapore (Figure 3.17). As described above, Germany experienced an increase in the relative size of its immigrant population, coupled with low skills among newly arrived migrants (Figure 3.14; see also Table A.3.14 (L) in Annex A). In Singapore, literacy proficiency among long-term migrants has significantly declined (Figure 3.16). Accounting for differences between foreign- and native- born adults in educational attainment, use of the language of the assessment and demographic characteristics accounts for only a fraction of the change in the proficiency gap in these countries. This suggests that additional factors are at play in widening the difference in literacy between these groups.\n\nNumeracy proficiency increased more strongly among migrants than among native-born adults in Canada, England (United Kingdom), Estonia, Finland, Norway and Sweden, narrowing the gap in numeracy between the two groups (see Table A.3.15 (N) in Annex A). In contrast, the gap widened in the\n\nGermany\n\nAustria\n\nFlemish Region ( E\n\nFrance\n\norway\n\nUnited States\n\nSlo ak Republic\n\norea\n\netherlands\n\nEngland (U\n\nSpain\n\nIreland\n\nEstonia\n\nSweden\n\nFinland\n\nenmark\n\nCanada\n\nItaly\n\nC echia\n\nSingapore\n\nChile\n\new ealand\n\nIsrael\n\nithuania\n\nHungary\n\nScore point difference 0\n\n30\n\n20\n\n10\n\n0\n\n10\n\n20\n\nAd usted Unad u sted\n\nRound, Cycle 1:","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p133.pdf"]},"response":{"errors":[],"extract_id":"extract_603b9c2665f2860b3edc0df560652414","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":[" **131**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\nwhole, as a widening skills gap between migrants and the native-born population risks undermining equity and social cohesion. In eight of the countries and economies with available data, the gap in literacy proficiency between migrants and non-migrants has widened (Figure 3.17). This was typically due to proficiency among foreign-born adults changing less favourably than that of native-born adults (Figure 3.15). Only Norway and the Flemish Region (Belgium) have seen the gap widening because the skills of native-born adults improved while those of foreign-born adults remained stable.\n\n### **Figure 3.17. Change in the gap in literacy proficiency between non-immigrants and immigrants**\n\nAdjusted and unadjusted change between cycles in the mean score difference between native-born adults with native-born parents and foreign-born adults with foreign-born parents (Cycle 2 _minus_ Cycle 1)\n\nNote: Adults aged 16-65; does not include adults who in Cycle 2 were only administered the doorstep interview due to a language barrier, to maximise the comparability across cycles (see Box 1.1 in Chapter 1 and Box 3.2). Unadjusted changes in the proficiency gap are the changes in the difference between the two contrast categories across the cycles. Adjusted changes are obtained by subtracting the coefficients of regression models estimated separately for each cycle. These coefficients show the difference between the categories in each cycle, after accounting for: gender, age, educational attainment, parental education and language spoken at home. Darker colours denote differences that are statistically significant at the 5% level. Japan and Poland are excluded due to small numbers of foreign-born adults. _Countries and economies are ranked in ascending order of the unadjusted change in the gap._ Source: OECD (2018[4]; 2015[5]; 2012[6]), _Survey of Adult Skills (PIAAC) databases_ , http://www.oecd.org/skills/piaac/publicdataandanalysis/ (accessed on 23 September 2024); Tables A.3.15 (L) and A.3.16 (L) in Annex A.\n\nThe widening of the proficiency gap between foreign- and native-born adults was particularly pronounced in Chile, Germany and Singapore (Figure 3.17). As described above, Germany experienced an increase in the relative size of its immigrant population, coupled with low skills among newly arrived migrants (Figure 3.14; see also Table A.3.14 (L) in Annex A). In Singapore, literacy proficiency among long-term migrants has significantly declined (Figure 3.16). Accounting for differences between foreign- and native- born adults in educational attainment, use of the language of the assessment and demographic characteristics accounts for only a fraction of the change in the proficiency gap in these countries. This suggests that additional factors are at play in widening the difference in literacy between these groups.\n\nNumeracy proficiency increased more strongly among migrants than among native-born adults in Canada, England (United Kingdom), Estonia, Finland, Norway and Sweden, narrowing the gap in numeracy between the two groups (see Table A.3.15 (N) in Annex A). In contrast, the gap widened in the\n\nGermany\n\nAustria\n\nFlemish Region ( E\n\nFrance\n\norway\n\nUnited States\n\nSlo ak Republic\n\norea\n\netherlands\n\nEngland (U\n\nSpain\n\nIreland\n\nEstonia\n\nSweden\n\nFinland\n\nenmark\n\nCanada\n\nItaly\n\nC echia\n\nSingapore\n\nChile\n\new ealand\n\nIsrael\n\nithuania\n\nHungary\n\nScore point difference 0\n\n30\n\n20\n\n10\n\n0\n\n10\n\n20\n\nAd usted Unad u sted\n\nRound, Cycle 1:"],"full_content":" **131**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\nwhole, as a widening skills gap between migrants and the native-born population risks undermining equity and social cohesion. In eight of the countries and economies with available data, the gap in literacy proficiency between migrants and non-migrants has widened (Figure 3.17). This was typically due to proficiency among foreign-born adults changing less favourably than that of native-born adults (Figure 3.15). Only Norway and the Flemish Region (Belgium) have seen the gap widening because the skills of native-born adults improved while those of foreign-born adults remained stable.\n\n### **Figure 3.17. Change in the gap in literacy proficiency between non-immigrants and immigrants**\n\nAdjusted and unadjusted change between cycles in the mean score difference between native-born adults with native-born parents and foreign-born adults with foreign-born parents (Cycle 2 _minus_ Cycle 1)\n\nNote: Adults aged 16-65; does not include adults who in Cycle 2 were only administered the doorstep interview due to a language barrier, to maximise the comparability across cycles (see Box 1.1 in Chapter 1 and Box 3.2). Unadjusted changes in the proficiency gap are the changes in the difference between the two contrast categories across the cycles. Adjusted changes are obtained by subtracting the coefficients of regression models estimated separately for each cycle. These coefficients show the difference between the categories in each cycle, after accounting for: gender, age, educational attainment, parental education and language spoken at home. Darker colours denote differences that are statistically significant at the 5% level. Japan and Poland are excluded due to small numbers of foreign-born adults. _Countries and economies are ranked in ascending order of the unadjusted change in the gap._ Source: OECD (2018[4]; 2015[5]; 2012[6]), _Survey of Adult Skills (PIAAC) databases_ , http://www.oecd.org/skills/piaac/publicdataandanalysis/ (accessed on 23 September 2024); Tables A.3.15 (L) and A.3.16 (L) in Annex A.\n\nThe widening of the proficiency gap between foreign- and native-born adults was particularly pronounced in Chile, Germany and Singapore (Figure 3.17). As described above, Germany experienced an increase in the relative size of its immigrant population, coupled with low skills among newly arrived migrants (Figure 3.14; see also Table A.3.14 (L) in Annex A). In Singapore, literacy proficiency among long-term migrants has significantly declined (Figure 3.16). Accounting for differences between foreign- and native- born adults in educational attainment, use of the language of the assessment and demographic characteristics accounts for only a fraction of the change in the proficiency gap in these countries. This suggests that additional factors are at play in widening the difference in literacy between these groups.\n\nNumeracy proficiency increased more strongly among migrants than among native-born adults in Canada, England (United Kingdom), Estonia, Finland, Norway and Sweden, narrowing the gap in numeracy between the two groups (see Table A.3.15 (N) in Annex A). In contrast, the gap widened in the\n\nGermany\n\nAustria\n\nFlemish Region ( E\n\nFrance\n\norway\n\nUnited States\n\nSlo ak Republic\n\norea\n\netherlands\n\nEngland (U\n\nSpain\n\nIreland\n\nEstonia\n\nSweden\n\nFinland\n\nenmark\n\nCanada\n\nItaly\n\nC echia\n\nSingapore\n\nChile\n\new ealand\n\nIsrael\n\nithuania\n\nHungary\n\nScore point difference 0\n\n30\n\n20\n\n10\n\n0\n\n10\n\n20\n\nAd usted Unad u sted\n\nRound, Cycle 1:","publish_date":null,"title":"Figure 3.17. Change in the gap in literacy proficiency between non-immigrants and immigrants","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p133.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p133.pdf"},"request":{"example_id":"chart/b263dc5d-en_p133","product_type":"parse","source_file_path":"b263dc5d-en_p133.pdf"},"started_at":"2026-11-04T19:45:55.272888"} +{"completed_at":"2026-11-04T19:45:58.247429","latency_in_ms":1386,"output":{"example_id":"chart/b263dc5d-en_p138","grounded_pages":[],"layout_pages":[],"markdown":"**136** \n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 3.19. Change in literacy and numeracy proficiency between cycles, by gender**\n\nDifference in mean literacy and numeracy scores between cycles (Cycle 2 _minus_ Cycle 1)\n\nNote: Adults aged 16-65; does not include adults who in Cycle 2 were only administered the doorstep interview due to a language barrier, to maximise the comparability across cycles (see Box 1.1 in Chapter 1 and Box 3.2). Differences are unadjusted. Darker colours denote differences that are statistically significant at the 5% level. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the ote for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the unadjusted change in literacy proficiency among men._ Source: OECD (2018[4]; 2015[5]; 2012[6]), _Survey of Adult Skills (PIAAC) databases_ , http://www.oecd.org/skills/piaac/publicdataandanalysis/ (accessed on 23 September 2024); Tables A.3.4 (L) and A.3.4 (N) in Annex A.\n\nA similar pattern is observed for numeracy (Figure 3.21, Panel B). Chile, Germany, Korea and the United States saw a narrowing gender gap in numeracy among the oldest age group. In Chile and the United States, men and women aged 45-65 now have equal numeracy scores, on average (see Table A.3.5 (N) in Annex A). Chile, Israel, New Zealand, Poland and Singapore saw gender differences in numeracy narrow among 25-44 year-olds. In Israel, New Zealand and Poland, men and women aged 25- 44 now have the same level of numeracy proficiency, while in Chile and Singapore, men in this age group still have an advantage in numeracy over women (see Table A.3.5 (N) in Annex A).\n\nFinland\n\nenmark\n\nSweden\n\norway\n\nFlemish Region ( E\n\nEngland (U\n\nCanada\n\nEstonia\n\nSpain\n\nGermany\n\nItaly\n\netherlands\n\nIreland\n\napan\n\nFrance\n\nC echia\n\nAustria\n\nUnited States\n\nSlo ak Republic\n\norea\n\nPoland\n\nSingapore\n\nChile\n\nIsrael\n\new ealand\n\nithuania\n\nHungary\n\nFinland\n\nenmark\n\nSweden\n\norway\n\nFlemish Region ( E\n\nEngland (U\n\nCanada\n\nEstonia\n\nSpain\n\nGermany\n\nItaly\n\netherlands\n\nIreland\n\napan\n\nFrance\n\nC echia\n\nAustria\n\nUnited States\n\nSlo ak Republic\n\norea\n\nPoland\n\nSingapore\n\nChile\n\nIsrael\n\new ealand\n\nithuania\n\nHungary\n\nen Women\n\n20\n\n10\n\n0\n\n10\n\n20\n\n30\n\n0\n\n20\n\n10\n\n0\n\n10\n\n20\n\n30\n\n0\n\n**. iteracy**\n\n**B. umeracy**\n\nRound, Cycle 1:","pages":[{"line_numbers":[],"markdown":"**136** \n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 3.19. Change in literacy and numeracy proficiency between cycles, by gender**\n\nDifference in mean literacy and numeracy scores between cycles (Cycle 2 _minus_ Cycle 1)\n\nNote: Adults aged 16-65; does not include adults who in Cycle 2 were only administered the doorstep interview due to a language barrier, to maximise the comparability across cycles (see Box 1.1 in Chapter 1 and Box 3.2). Differences are unadjusted. Darker colours denote differences that are statistically significant at the 5% level. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the ote for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the unadjusted change in literacy proficiency among men._ Source: OECD (2018[4]; 2015[5]; 2012[6]), _Survey of Adult Skills (PIAAC) databases_ , http://www.oecd.org/skills/piaac/publicdataandanalysis/ (accessed on 23 September 2024); Tables A.3.4 (L) and A.3.4 (N) in Annex A.\n\nA similar pattern is observed for numeracy (Figure 3.21, Panel B). Chile, Germany, Korea and the United States saw a narrowing gender gap in numeracy among the oldest age group. In Chile and the United States, men and women aged 45-65 now have equal numeracy scores, on average (see Table A.3.5 (N) in Annex A). Chile, Israel, New Zealand, Poland and Singapore saw gender differences in numeracy narrow among 25-44 year-olds. In Israel, New Zealand and Poland, men and women aged 25- 44 now have the same level of numeracy proficiency, while in Chile and Singapore, men in this age group still have an advantage in numeracy over women (see Table A.3.5 (N) in Annex A).\n\nFinland\n\nenmark\n\nSweden\n\norway\n\nFlemish Region ( E\n\nEngland (U\n\nCanada\n\nEstonia\n\nSpain\n\nGermany\n\nItaly\n\netherlands\n\nIreland\n\napan\n\nFrance\n\nC echia\n\nAustria\n\nUnited States\n\nSlo ak Republic\n\norea\n\nPoland\n\nSingapore\n\nChile\n\nIsrael\n\new ealand\n\nithuania\n\nHungary\n\nFinland\n\nenmark\n\nSweden\n\norway\n\nFlemish Region ( E\n\nEngland (U\n\nCanada\n\nEstonia\n\nSpain\n\nGermany\n\nItaly\n\netherlands\n\nIreland\n\napan\n\nFrance\n\nC echia\n\nAustria\n\nUnited States\n\nSlo ak Republic\n\norea\n\nPoland\n\nSingapore\n\nChile\n\nIsrael\n\new ealand\n\nithuania\n\nHungary\n\nen Women\n\n20\n\n10\n\n0\n\n10\n\n20\n\n30\n\n0\n\n20\n\n10\n\n0\n\n10\n\n20\n\n30\n\n0\n\n**. iteracy**\n\n**B. umeracy**\n\nRound, Cycle 1:","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**136** \n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 3.19. Change in literacy and numeracy proficiency between cycles, by gender**\n\nDifference in mean literacy and numeracy scores between cycles (Cycle 2 _minus_ Cycle 1)\n\nNote: Adults aged 16-65; does not include adults who in Cycle 2 were only administered the doorstep interview due to a language barrier, to maximise the comparability across cycles (see Box 1.1 in Chapter 1 and Box 3.2). Differences are unadjusted. Darker colours denote differences that are statistically significant at the 5% level. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the ote for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the unadjusted change in literacy proficiency among men._ Source: OECD (2018[4]; 2015[5]; 2012[6]), _Survey of Adult Skills (PIAAC) databases_ , http://www.oecd.org/skills/piaac/publicdataandanalysis/ (accessed on 23 September 2024); Tables A.3.4 (L) and A.3.4 (N) in Annex A.\n\nA similar pattern is observed for numeracy (Figure 3.21, Panel B). Chile, Germany, Korea and the United States saw a narrowing gender gap in numeracy among the oldest age group. In Chile and the United States, men and women aged 45-65 now have equal numeracy scores, on average (see Table A.3.5 (N) in Annex A). Chile, Israel, New Zealand, Poland and Singapore saw gender differences in numeracy narrow among 25-44 year-olds. In Israel, New Zealand and Poland, men and women aged 25- 44 now have the same level of numeracy proficiency, while in Chile and Singapore, men in this age group still have an advantage in numeracy over women (see Table A.3.5 (N) in Annex A).\n\nFinland\n\nenmark\n\nSweden\n\norway\n\nFlemish Region ( E\n\nEngland (U\n\nCanada\n\nEstonia\n\nSpain\n\nGermany\n\nItaly\n\netherlands\n\nIreland\n\napan\n\nFrance\n\nC echia\n\nAustria\n\nUnited States\n\nSlo ak Republic\n\norea\n\nPoland\n\nSingapore\n\nChile\n\nIsrael\n\new ealand\n\nithuania\n\nHungary\n\nFinland\n\nenmark\n\nSweden\n\norway\n\nFlemish Region ( E\n\nEngland (U\n\nCanada\n\nEstonia\n\nSpain\n\nGermany\n\nItaly\n\netherlands\n\nIreland\n\napan\n\nFrance\n\nC echia\n\nAustria\n\nUnited States\n\nSlo ak Republic\n\norea\n\nPoland\n\nSingapore\n\nChile\n\nIsrael\n\new ealand\n\nithuania\n\nHungary\n\nen Women\n\n20\n\n10\n\n0\n\n10\n\n20\n\n30\n\n0\n\n20\n\n10\n\n0\n\n10\n\n20\n\n30\n\n0\n\n**. iteracy**\n\n**B. umeracy**\n\nRound, Cycle 1:","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p138.pdf"]},"response":{"errors":[],"extract_id":"extract_6f67a50f8450d2b512d542713a135ca4","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**136** \n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 3.19. Change in literacy and numeracy proficiency between cycles, by gender**\n\nDifference in mean literacy and numeracy scores between cycles (Cycle 2 _minus_ Cycle 1)\n\nNote: Adults aged 16-65; does not include adults who in Cycle 2 were only administered the doorstep interview due to a language barrier, to maximise the comparability across cycles (see Box 1.1 in Chapter 1 and Box 3.2). Differences are unadjusted. Darker colours denote differences that are statistically significant at the 5% level. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the ote for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the unadjusted change in literacy proficiency among men._ Source: OECD (2018[4]; 2015[5]; 2012[6]), _Survey of Adult Skills (PIAAC) databases_ , http://www.oecd.org/skills/piaac/publicdataandanalysis/ (accessed on 23 September 2024); Tables A.3.4 (L) and A.3.4 (N) in Annex A.\n\nA similar pattern is observed for numeracy (Figure 3.21, Panel B). Chile, Germany, Korea and the United States saw a narrowing gender gap in numeracy among the oldest age group. In Chile and the United States, men and women aged 45-65 now have equal numeracy scores, on average (see Table A.3.5 (N) in Annex A). Chile, Israel, New Zealand, Poland and Singapore saw gender differences in numeracy narrow among 25-44 year-olds. In Israel, New Zealand and Poland, men and women aged 25- 44 now have the same level of numeracy proficiency, while in Chile and Singapore, men in this age group still have an advantage in numeracy over women (see Table A.3.5 (N) in Annex A).\n\nFinland\n\nenmark\n\nSweden\n\norway\n\nFlemish Region ( E\n\nEngland (U\n\nCanada\n\nEstonia\n\nSpain\n\nGermany\n\nItaly\n\netherlands\n\nIreland\n\napan\n\nFrance\n\nC echia\n\nAustria\n\nUnited States\n\nSlo ak Republic\n\norea\n\nPoland\n\nSingapore\n\nChile\n\nIsrael\n\new ealand\n\nithuania\n\nHungary\n\nFinland\n\nenmark\n\nSweden\n\norway\n\nFlemish Region ( E\n\nEngland (U\n\nCanada\n\nEstonia\n\nSpain\n\nGermany\n\nItaly\n\netherlands\n\nIreland\n\napan\n\nFrance\n\nC echia\n\nAustria\n\nUnited States\n\nSlo ak Republic\n\norea\n\nPoland\n\nSingapore\n\nChile\n\nIsrael\n\new ealand\n\nithuania\n\nHungary\n\nen Women\n\n20\n\n10\n\n0\n\n10\n\n20\n\n30\n\n0\n\n20\n\n10\n\n0\n\n10\n\n20\n\n30\n\n0\n\n**. iteracy**\n\n**B. umeracy**\n\nRound, Cycle 1:"],"full_content":"**136** \n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 3.19. Change in literacy and numeracy proficiency between cycles, by gender**\n\nDifference in mean literacy and numeracy scores between cycles (Cycle 2 _minus_ Cycle 1)\n\nNote: Adults aged 16-65; does not include adults who in Cycle 2 were only administered the doorstep interview due to a language barrier, to maximise the comparability across cycles (see Box 1.1 in Chapter 1 and Box 3.2). Differences are unadjusted. Darker colours denote differences that are statistically significant at the 5% level. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the ote for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the unadjusted change in literacy proficiency among men._ Source: OECD (2018[4]; 2015[5]; 2012[6]), _Survey of Adult Skills (PIAAC) databases_ , http://www.oecd.org/skills/piaac/publicdataandanalysis/ (accessed on 23 September 2024); Tables A.3.4 (L) and A.3.4 (N) in Annex A.\n\nA similar pattern is observed for numeracy (Figure 3.21, Panel B). Chile, Germany, Korea and the United States saw a narrowing gender gap in numeracy among the oldest age group. In Chile and the United States, men and women aged 45-65 now have equal numeracy scores, on average (see Table A.3.5 (N) in Annex A). Chile, Israel, New Zealand, Poland and Singapore saw gender differences in numeracy narrow among 25-44 year-olds. In Israel, New Zealand and Poland, men and women aged 25- 44 now have the same level of numeracy proficiency, while in Chile and Singapore, men in this age group still have an advantage in numeracy over women (see Table A.3.5 (N) in Annex A).\n\nFinland\n\nenmark\n\nSweden\n\norway\n\nFlemish Region ( E\n\nEngland (U\n\nCanada\n\nEstonia\n\nSpain\n\nGermany\n\nItaly\n\netherlands\n\nIreland\n\napan\n\nFrance\n\nC echia\n\nAustria\n\nUnited States\n\nSlo ak Republic\n\norea\n\nPoland\n\nSingapore\n\nChile\n\nIsrael\n\new ealand\n\nithuania\n\nHungary\n\nFinland\n\nenmark\n\nSweden\n\norway\n\nFlemish Region ( E\n\nEngland (U\n\nCanada\n\nEstonia\n\nSpain\n\nGermany\n\nItaly\n\netherlands\n\nIreland\n\napan\n\nFrance\n\nC echia\n\nAustria\n\nUnited States\n\nSlo ak Republic\n\norea\n\nPoland\n\nSingapore\n\nChile\n\nIsrael\n\new ealand\n\nithuania\n\nHungary\n\nen Women\n\n20\n\n10\n\n0\n\n10\n\n20\n\n30\n\n0\n\n20\n\n10\n\n0\n\n10\n\n20\n\n30\n\n0\n\n**. iteracy**\n\n**B. umeracy**\n\nRound, Cycle 1:","publish_date":null,"title":"Figure 3.19. Change in literacy and numeracy proficiency between cycles, by gender","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p138.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p138.pdf"},"request":{"example_id":"chart/b263dc5d-en_p138","product_type":"parse","source_file_path":"b263dc5d-en_p138.pdf"},"started_at":"2026-11-04T19:45:56.860951"} +{"completed_at":"2026-11-04T19:31:08.381210","latency_in_ms":2975,"output":{"example_id":"chart/b263dc5d-en_p139","grounded_pages":[],"layout_pages":[],"markdown":" **137**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 3.20. Change in the gaps in literacy and numeracy proficiency between men and women**\n\nAdjusted and unadjusted change between cycles in the mean score difference between men and women (Cycle 2 _minus_ Cycle 1)\n\nNote: Adults aged 16-65; does not include adults who in Cycle 2 were only administered the doorstep interview due to a language barrier, to maximise the comparability across cycles (see Box 1.1 in Chapter 1 and Box 3.2). Unadjusted changes in the proficiency gap are the changes in the difference between the two contrast categories across the cycles. Adjusted changes are obtained by subtracting the coefficients of regression models estimated separately for each cycle. These coefficients show the difference between the categories in each cycle after accounting for: immigrant background, age, educational attainment, parental education and language spoken at home. Darker colours denote differences that are statistically significant at the 5% level. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the ote for Poland in the Reader’s Guide. _Countries and economies are ranked in ascending order of the unadjusted change in the gap in literacy._ Source: OECD (2018[4]; 2015[5]; 2012[6]), _Survey of Adult Skills (PIAAC) databases_ , http://www.oecd.org/skills/piaac/publicdataandanalysis/ (accessed on 23 September 2024); Tables A.3.5 (L), A.3.5 (N), A.3.6 (L) and A.3.6 (N) in Annex A.\n\nAd usted Unad u sted\n\nPoland\n\nSlo ak Republic\n\nEngland (U\n\nCanada\n\nenmark\n\nFinland\n\nItaly\n\nC echia\n\nSpain\n\napan\n\nFrance\n\nAustria\n\nUnited States\n\norea\n\nSweden\n\nFlemish Region ( E\n\nIreland\n\norway\n\nEstonia\n\nGermany\n\netherlands\n\nithuania\n\nSingapore\n\nIsrael\n\nChile\n\new ealand\n\nHungary\n\nPoland\n\nSlo ak Republic\n\nEngland (U\n\nCanada\n\nenmark\n\nFinland\n\nItaly\n\nC echia\n\nSpain\n\napan\n\nFrance\n\nAustria\n\nUnited States\n\norea\n\nSweden\n\nFlemish Region ( E\n\nIreland\n\norway\n\nEstonia\n\nGermany\n\netherlands\n\nithuania\n\nSingapore\n\nIsrael\n\nChile\n\new ealand\n\nHungary\n\n10\n\n0\n\n10\n\n1\n\n10\n\n0\n\n10\n\n1\n\n**. iteracy**\n\n**B. umeracy**\n\nRound, Cycle 1:","pages":[{"line_numbers":[],"markdown":" **137**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 3.20. Change in the gaps in literacy and numeracy proficiency between men and women**\n\nAdjusted and unadjusted change between cycles in the mean score difference between men and women (Cycle 2 _minus_ Cycle 1)\n\nNote: Adults aged 16-65; does not include adults who in Cycle 2 were only administered the doorstep interview due to a language barrier, to maximise the comparability across cycles (see Box 1.1 in Chapter 1 and Box 3.2). Unadjusted changes in the proficiency gap are the changes in the difference between the two contrast categories across the cycles. Adjusted changes are obtained by subtracting the coefficients of regression models estimated separately for each cycle. These coefficients show the difference between the categories in each cycle after accounting for: immigrant background, age, educational attainment, parental education and language spoken at home. Darker colours denote differences that are statistically significant at the 5% level. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the ote for Poland in the Reader’s Guide. _Countries and economies are ranked in ascending order of the unadjusted change in the gap in literacy._ Source: OECD (2018[4]; 2015[5]; 2012[6]), _Survey of Adult Skills (PIAAC) databases_ , http://www.oecd.org/skills/piaac/publicdataandanalysis/ (accessed on 23 September 2024); Tables A.3.5 (L), A.3.5 (N), A.3.6 (L) and A.3.6 (N) in Annex A.\n\nAd usted Unad u sted\n\nPoland\n\nSlo ak Republic\n\nEngland (U\n\nCanada\n\nenmark\n\nFinland\n\nItaly\n\nC echia\n\nSpain\n\napan\n\nFrance\n\nAustria\n\nUnited States\n\norea\n\nSweden\n\nFlemish Region ( E\n\nIreland\n\norway\n\nEstonia\n\nGermany\n\netherlands\n\nithuania\n\nSingapore\n\nIsrael\n\nChile\n\new ealand\n\nHungary\n\nPoland\n\nSlo ak Republic\n\nEngland (U\n\nCanada\n\nenmark\n\nFinland\n\nItaly\n\nC echia\n\nSpain\n\napan\n\nFrance\n\nAustria\n\nUnited States\n\norea\n\nSweden\n\nFlemish Region ( E\n\nIreland\n\norway\n\nEstonia\n\nGermany\n\netherlands\n\nithuania\n\nSingapore\n\nIsrael\n\nChile\n\new ealand\n\nHungary\n\n10\n\n0\n\n10\n\n1\n\n10\n\n0\n\n10\n\n1\n\n**. iteracy**\n\n**B. umeracy**\n\nRound, Cycle 1:","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":" **137**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 3.20. Change in the gaps in literacy and numeracy proficiency between men and women**\n\nAdjusted and unadjusted change between cycles in the mean score difference between men and women (Cycle 2 _minus_ Cycle 1)\n\nNote: Adults aged 16-65; does not include adults who in Cycle 2 were only administered the doorstep interview due to a language barrier, to maximise the comparability across cycles (see Box 1.1 in Chapter 1 and Box 3.2). Unadjusted changes in the proficiency gap are the changes in the difference between the two contrast categories across the cycles. Adjusted changes are obtained by subtracting the coefficients of regression models estimated separately for each cycle. These coefficients show the difference between the categories in each cycle after accounting for: immigrant background, age, educational attainment, parental education and language spoken at home. Darker colours denote differences that are statistically significant at the 5% level. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the ote for Poland in the Reader’s Guide. _Countries and economies are ranked in ascending order of the unadjusted change in the gap in literacy._ Source: OECD (2018[4]; 2015[5]; 2012[6]), _Survey of Adult Skills (PIAAC) databases_ , http://www.oecd.org/skills/piaac/publicdataandanalysis/ (accessed on 23 September 2024); Tables A.3.5 (L), A.3.5 (N), A.3.6 (L) and A.3.6 (N) in Annex A.\n\nAd usted Unad u sted\n\nPoland\n\nSlo ak Republic\n\nEngland (U\n\nCanada\n\nenmark\n\nFinland\n\nItaly\n\nC echia\n\nSpain\n\napan\n\nFrance\n\nAustria\n\nUnited States\n\norea\n\nSweden\n\nFlemish Region ( E\n\nIreland\n\norway\n\nEstonia\n\nGermany\n\netherlands\n\nithuania\n\nSingapore\n\nIsrael\n\nChile\n\new ealand\n\nHungary\n\nPoland\n\nSlo ak Republic\n\nEngland (U\n\nCanada\n\nenmark\n\nFinland\n\nItaly\n\nC echia\n\nSpain\n\napan\n\nFrance\n\nAustria\n\nUnited States\n\norea\n\nSweden\n\nFlemish Region ( E\n\nIreland\n\norway\n\nEstonia\n\nGermany\n\netherlands\n\nithuania\n\nSingapore\n\nIsrael\n\nChile\n\new ealand\n\nHungary\n\n10\n\n0\n\n10\n\n1\n\n10\n\n0\n\n10\n\n1\n\n**. iteracy**\n\n**B. umeracy**\n\nRound, Cycle 1:","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p139.pdf"]},"response":{"errors":[],"extract_id":"extract_a70038d15cde327956bb149c23e1ca37","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":[" **137**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 3.20. Change in the gaps in literacy and numeracy proficiency between men and women**\n\nAdjusted and unadjusted change between cycles in the mean score difference between men and women (Cycle 2 _minus_ Cycle 1)\n\nNote: Adults aged 16-65; does not include adults who in Cycle 2 were only administered the doorstep interview due to a language barrier, to maximise the comparability across cycles (see Box 1.1 in Chapter 1 and Box 3.2). Unadjusted changes in the proficiency gap are the changes in the difference between the two contrast categories across the cycles. Adjusted changes are obtained by subtracting the coefficients of regression models estimated separately for each cycle. These coefficients show the difference between the categories in each cycle after accounting for: immigrant background, age, educational attainment, parental education and language spoken at home. Darker colours denote differences that are statistically significant at the 5% level. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the ote for Poland in the Reader’s Guide. _Countries and economies are ranked in ascending order of the unadjusted change in the gap in literacy._ Source: OECD (2018[4]; 2015[5]; 2012[6]), _Survey of Adult Skills (PIAAC) databases_ , http://www.oecd.org/skills/piaac/publicdataandanalysis/ (accessed on 23 September 2024); Tables A.3.5 (L), A.3.5 (N), A.3.6 (L) and A.3.6 (N) in Annex A.\n\nAd usted Unad u sted\n\nPoland\n\nSlo ak Republic\n\nEngland (U\n\nCanada\n\nenmark\n\nFinland\n\nItaly\n\nC echia\n\nSpain\n\napan\n\nFrance\n\nAustria\n\nUnited States\n\norea\n\nSweden\n\nFlemish Region ( E\n\nIreland\n\norway\n\nEstonia\n\nGermany\n\netherlands\n\nithuania\n\nSingapore\n\nIsrael\n\nChile\n\new ealand\n\nHungary\n\nPoland\n\nSlo ak Republic\n\nEngland (U\n\nCanada\n\nenmark\n\nFinland\n\nItaly\n\nC echia\n\nSpain\n\napan\n\nFrance\n\nAustria\n\nUnited States\n\norea\n\nSweden\n\nFlemish Region ( E\n\nIreland\n\norway\n\nEstonia\n\nGermany\n\netherlands\n\nithuania\n\nSingapore\n\nIsrael\n\nChile\n\new ealand\n\nHungary\n\n10\n\n0\n\n10\n\n1\n\n10\n\n0\n\n10\n\n1\n\n**. iteracy**\n\n**B. umeracy**\n\nRound, Cycle 1:"],"full_content":" **137**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 3.20. Change in the gaps in literacy and numeracy proficiency between men and women**\n\nAdjusted and unadjusted change between cycles in the mean score difference between men and women (Cycle 2 _minus_ Cycle 1)\n\nNote: Adults aged 16-65; does not include adults who in Cycle 2 were only administered the doorstep interview due to a language barrier, to maximise the comparability across cycles (see Box 1.1 in Chapter 1 and Box 3.2). Unadjusted changes in the proficiency gap are the changes in the difference between the two contrast categories across the cycles. Adjusted changes are obtained by subtracting the coefficients of regression models estimated separately for each cycle. These coefficients show the difference between the categories in each cycle after accounting for: immigrant background, age, educational attainment, parental education and language spoken at home. Darker colours denote differences that are statistically significant at the 5% level. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the ote for Poland in the Reader’s Guide. _Countries and economies are ranked in ascending order of the unadjusted change in the gap in literacy._ Source: OECD (2018[4]; 2015[5]; 2012[6]), _Survey of Adult Skills (PIAAC) databases_ , http://www.oecd.org/skills/piaac/publicdataandanalysis/ (accessed on 23 September 2024); Tables A.3.5 (L), A.3.5 (N), A.3.6 (L) and A.3.6 (N) in Annex A.\n\nAd usted Unad u sted\n\nPoland\n\nSlo ak Republic\n\nEngland (U\n\nCanada\n\nenmark\n\nFinland\n\nItaly\n\nC echia\n\nSpain\n\napan\n\nFrance\n\nAustria\n\nUnited States\n\norea\n\nSweden\n\nFlemish Region ( E\n\nIreland\n\norway\n\nEstonia\n\nGermany\n\netherlands\n\nithuania\n\nSingapore\n\nIsrael\n\nChile\n\new ealand\n\nHungary\n\nPoland\n\nSlo ak Republic\n\nEngland (U\n\nCanada\n\nenmark\n\nFinland\n\nItaly\n\nC echia\n\nSpain\n\napan\n\nFrance\n\nAustria\n\nUnited States\n\norea\n\nSweden\n\nFlemish Region ( E\n\nIreland\n\norway\n\nEstonia\n\nGermany\n\netherlands\n\nithuania\n\nSingapore\n\nIsrael\n\nChile\n\new ealand\n\nHungary\n\n10\n\n0\n\n10\n\n1\n\n10\n\n0\n\n10\n\n1\n\n**. iteracy**\n\n**B. umeracy**\n\nRound, Cycle 1:","publish_date":null,"title":"Figure 3.20. Change in the gaps in literacy and numeracy proficiency between men and women","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p139.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p139.pdf"},"request":{"example_id":"chart/b263dc5d-en_p139","product_type":"parse","source_file_path":"b263dc5d-en_p139.pdf"},"started_at":"2026-11-04T19:31:05.405882"} +{"completed_at":"2026-11-04T19:46:01.443684","latency_in_ms":2410,"output":{"example_id":"chart/b263dc5d-en_p143","grounded_pages":[],"layout_pages":[],"markdown":" **141**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 3.23. Change in the gap in literacy proficiency between adults with highly educated and low- educated parents**\n\nAdjusted and unadjusted change between cycles in the mean score difference between adults with at least one tertiary-educated parent and adults whose parents have below upper secondary education (Cycle 2 _minus_ Cycle 1)\n\nNote: Adults aged 16-65; does not include adults who in Cycle 2 were only administered the doorstep interview due to a language barrier, to maximise the comparability across cycles (see Box 1.1 in Chapter 1 and Box 3.2). Unadjusted changes are the changes in the difference between the two contrast categories across the cycles. Adjusted changes are obtained by subtracting the coefficients of regression models estimated separately for each cycle. These coefficients show the difference between the categories in each cycle, after accounting for: immigrant background, age, educational attainment, gender and language spoken at home. Darker colours denote differences that are statistically significant at the 5% level. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the ote for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the unadjusted change in the literacy gap._ Source: OECD (2018[4]; 2015[5]; 2012[6]), _Survey of Adult Skills (PIAAC) databases_ , http://www.oecd.org/skills/piaac/publicdataandanalysis/ (accessed on 23 September 2024); Tables A.3.19 (L) and A.3.20 (L) in Annex A.\n\nThe numeracy proficiency of adults with low-educated parents has declined in 11 of the participating countries and economies (see Table A.3.17 (N) in Annex A). Most of these countries saw declines in the numeracy proficiency of adults with medium-educated parents as well. In Singapore, numeracy proficiency among adults with low-educated parents has increased, while in Chile, Finland and Spain this increase was only significant after accounting for demographic changes. In comparison, seven countries saw a decline and four recorded an increase in the numeracy proficiency of adults with highly educated parents. As numeracy generally developed more favourably in this group than among adults with low-educated parents, inequalities in numeracy proficiency increased in many countries (see Table A.3.19 (N) in Annex A).\n\n_Change in skills proficiency related to parental education among younger and older adults_\n\nThis section shows how the proficiency gap associated with socio-economic background (as proxied by parental education) has changed across cycles within three age groups: 16-24 year-olds, 25-44 year-olds and 45-65 year-olds. Due to insufficient sample sizes of some groups, adults with low- and medium- educated parents are considered as a single group (denoted as adults whose parents have at most an upper secondary education) and compared to adults with at least one tertiary-educated parent.\n\nEstonia\n\nAustria\n\nGermany\n\nC echia\n\norea\n\nCanada\n\napan\n\nFlemish Region ( E\n\norway\n\nFinland\n\nFrance\n\nUnited States\n\nIreland\n\nenmark\n\nEngland (U\n\netherlands\n\nSweden\n\nItaly\n\nPoland\n\nSlo ak Republic\n\nSpain\n\nithuania\n\nIsrael\n\new ealand\n\nChile\n\nSingapore\n\nHungary\n\nScore point difference 30\n\n20\n\n10\n\n0\n\n10\n\n20\n\nAd usted Unad u sted\n\nRound, Cycle 1:","pages":[{"line_numbers":[],"markdown":" **141**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 3.23. Change in the gap in literacy proficiency between adults with highly educated and low- educated parents**\n\nAdjusted and unadjusted change between cycles in the mean score difference between adults with at least one tertiary-educated parent and adults whose parents have below upper secondary education (Cycle 2 _minus_ Cycle 1)\n\nNote: Adults aged 16-65; does not include adults who in Cycle 2 were only administered the doorstep interview due to a language barrier, to maximise the comparability across cycles (see Box 1.1 in Chapter 1 and Box 3.2). Unadjusted changes are the changes in the difference between the two contrast categories across the cycles. Adjusted changes are obtained by subtracting the coefficients of regression models estimated separately for each cycle. These coefficients show the difference between the categories in each cycle, after accounting for: immigrant background, age, educational attainment, gender and language spoken at home. Darker colours denote differences that are statistically significant at the 5% level. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the ote for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the unadjusted change in the literacy gap._ Source: OECD (2018[4]; 2015[5]; 2012[6]), _Survey of Adult Skills (PIAAC) databases_ , http://www.oecd.org/skills/piaac/publicdataandanalysis/ (accessed on 23 September 2024); Tables A.3.19 (L) and A.3.20 (L) in Annex A.\n\nThe numeracy proficiency of adults with low-educated parents has declined in 11 of the participating countries and economies (see Table A.3.17 (N) in Annex A). Most of these countries saw declines in the numeracy proficiency of adults with medium-educated parents as well. In Singapore, numeracy proficiency among adults with low-educated parents has increased, while in Chile, Finland and Spain this increase was only significant after accounting for demographic changes. In comparison, seven countries saw a decline and four recorded an increase in the numeracy proficiency of adults with highly educated parents. As numeracy generally developed more favourably in this group than among adults with low-educated parents, inequalities in numeracy proficiency increased in many countries (see Table A.3.19 (N) in Annex A).\n\n_Change in skills proficiency related to parental education among younger and older adults_\n\nThis section shows how the proficiency gap associated with socio-economic background (as proxied by parental education) has changed across cycles within three age groups: 16-24 year-olds, 25-44 year-olds and 45-65 year-olds. Due to insufficient sample sizes of some groups, adults with low- and medium- educated parents are considered as a single group (denoted as adults whose parents have at most an upper secondary education) and compared to adults with at least one tertiary-educated parent.\n\nEstonia\n\nAustria\n\nGermany\n\nC echia\n\norea\n\nCanada\n\napan\n\nFlemish Region ( E\n\norway\n\nFinland\n\nFrance\n\nUnited States\n\nIreland\n\nenmark\n\nEngland (U\n\netherlands\n\nSweden\n\nItaly\n\nPoland\n\nSlo ak Republic\n\nSpain\n\nithuania\n\nIsrael\n\new ealand\n\nChile\n\nSingapore\n\nHungary\n\nScore point difference 30\n\n20\n\n10\n\n0\n\n10\n\n20\n\nAd usted Unad u sted\n\nRound, Cycle 1:","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":" **141**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 3.23. Change in the gap in literacy proficiency between adults with highly educated and low- educated parents**\n\nAdjusted and unadjusted change between cycles in the mean score difference between adults with at least one tertiary-educated parent and adults whose parents have below upper secondary education (Cycle 2 _minus_ Cycle 1)\n\nNote: Adults aged 16-65; does not include adults who in Cycle 2 were only administered the doorstep interview due to a language barrier, to maximise the comparability across cycles (see Box 1.1 in Chapter 1 and Box 3.2). Unadjusted changes are the changes in the difference between the two contrast categories across the cycles. Adjusted changes are obtained by subtracting the coefficients of regression models estimated separately for each cycle. These coefficients show the difference between the categories in each cycle, after accounting for: immigrant background, age, educational attainment, gender and language spoken at home. Darker colours denote differences that are statistically significant at the 5% level. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the ote for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the unadjusted change in the literacy gap._ Source: OECD (2018[4]; 2015[5]; 2012[6]), _Survey of Adult Skills (PIAAC) databases_ , http://www.oecd.org/skills/piaac/publicdataandanalysis/ (accessed on 23 September 2024); Tables A.3.19 (L) and A.3.20 (L) in Annex A.\n\nThe numeracy proficiency of adults with low-educated parents has declined in 11 of the participating countries and economies (see Table A.3.17 (N) in Annex A). Most of these countries saw declines in the numeracy proficiency of adults with medium-educated parents as well. In Singapore, numeracy proficiency among adults with low-educated parents has increased, while in Chile, Finland and Spain this increase was only significant after accounting for demographic changes. In comparison, seven countries saw a decline and four recorded an increase in the numeracy proficiency of adults with highly educated parents. As numeracy generally developed more favourably in this group than among adults with low-educated parents, inequalities in numeracy proficiency increased in many countries (see Table A.3.19 (N) in Annex A).\n\n_Change in skills proficiency related to parental education among younger and older adults_\n\nThis section shows how the proficiency gap associated with socio-economic background (as proxied by parental education) has changed across cycles within three age groups: 16-24 year-olds, 25-44 year-olds and 45-65 year-olds. Due to insufficient sample sizes of some groups, adults with low- and medium- educated parents are considered as a single group (denoted as adults whose parents have at most an upper secondary education) and compared to adults with at least one tertiary-educated parent.\n\nEstonia\n\nAustria\n\nGermany\n\nC echia\n\norea\n\nCanada\n\napan\n\nFlemish Region ( E\n\norway\n\nFinland\n\nFrance\n\nUnited States\n\nIreland\n\nenmark\n\nEngland (U\n\netherlands\n\nSweden\n\nItaly\n\nPoland\n\nSlo ak Republic\n\nSpain\n\nithuania\n\nIsrael\n\new ealand\n\nChile\n\nSingapore\n\nHungary\n\nScore point difference 30\n\n20\n\n10\n\n0\n\n10\n\n20\n\nAd usted Unad u sted\n\nRound, Cycle 1:","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p143.pdf"]},"response":{"errors":[],"extract_id":"extract_60f9b34aae11ef42b2454a953dcc1210","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":[" **141**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 3.23. Change in the gap in literacy proficiency between adults with highly educated and low- educated parents**\n\nAdjusted and unadjusted change between cycles in the mean score difference between adults with at least one tertiary-educated parent and adults whose parents have below upper secondary education (Cycle 2 _minus_ Cycle 1)\n\nNote: Adults aged 16-65; does not include adults who in Cycle 2 were only administered the doorstep interview due to a language barrier, to maximise the comparability across cycles (see Box 1.1 in Chapter 1 and Box 3.2). Unadjusted changes are the changes in the difference between the two contrast categories across the cycles. Adjusted changes are obtained by subtracting the coefficients of regression models estimated separately for each cycle. These coefficients show the difference between the categories in each cycle, after accounting for: immigrant background, age, educational attainment, gender and language spoken at home. Darker colours denote differences that are statistically significant at the 5% level. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the ote for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the unadjusted change in the literacy gap._ Source: OECD (2018[4]; 2015[5]; 2012[6]), _Survey of Adult Skills (PIAAC) databases_ , http://www.oecd.org/skills/piaac/publicdataandanalysis/ (accessed on 23 September 2024); Tables A.3.19 (L) and A.3.20 (L) in Annex A.\n\nThe numeracy proficiency of adults with low-educated parents has declined in 11 of the participating countries and economies (see Table A.3.17 (N) in Annex A). Most of these countries saw declines in the numeracy proficiency of adults with medium-educated parents as well. In Singapore, numeracy proficiency among adults with low-educated parents has increased, while in Chile, Finland and Spain this increase was only significant after accounting for demographic changes. In comparison, seven countries saw a decline and four recorded an increase in the numeracy proficiency of adults with highly educated parents. As numeracy generally developed more favourably in this group than among adults with low-educated parents, inequalities in numeracy proficiency increased in many countries (see Table A.3.19 (N) in Annex A).\n\n_Change in skills proficiency related to parental education among younger and older adults_\n\nThis section shows how the proficiency gap associated with socio-economic background (as proxied by parental education) has changed across cycles within three age groups: 16-24 year-olds, 25-44 year-olds and 45-65 year-olds. Due to insufficient sample sizes of some groups, adults with low- and medium- educated parents are considered as a single group (denoted as adults whose parents have at most an upper secondary education) and compared to adults with at least one tertiary-educated parent.\n\nEstonia\n\nAustria\n\nGermany\n\nC echia\n\norea\n\nCanada\n\napan\n\nFlemish Region ( E\n\norway\n\nFinland\n\nFrance\n\nUnited States\n\nIreland\n\nenmark\n\nEngland (U\n\netherlands\n\nSweden\n\nItaly\n\nPoland\n\nSlo ak Republic\n\nSpain\n\nithuania\n\nIsrael\n\new ealand\n\nChile\n\nSingapore\n\nHungary\n\nScore point difference 30\n\n20\n\n10\n\n0\n\n10\n\n20\n\nAd usted Unad u sted\n\nRound, Cycle 1:"],"full_content":" **141**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 3.23. Change in the gap in literacy proficiency between adults with highly educated and low- educated parents**\n\nAdjusted and unadjusted change between cycles in the mean score difference between adults with at least one tertiary-educated parent and adults whose parents have below upper secondary education (Cycle 2 _minus_ Cycle 1)\n\nNote: Adults aged 16-65; does not include adults who in Cycle 2 were only administered the doorstep interview due to a language barrier, to maximise the comparability across cycles (see Box 1.1 in Chapter 1 and Box 3.2). Unadjusted changes are the changes in the difference between the two contrast categories across the cycles. Adjusted changes are obtained by subtracting the coefficients of regression models estimated separately for each cycle. These coefficients show the difference between the categories in each cycle, after accounting for: immigrant background, age, educational attainment, gender and language spoken at home. Darker colours denote differences that are statistically significant at the 5% level. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the ote for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the unadjusted change in the literacy gap._ Source: OECD (2018[4]; 2015[5]; 2012[6]), _Survey of Adult Skills (PIAAC) databases_ , http://www.oecd.org/skills/piaac/publicdataandanalysis/ (accessed on 23 September 2024); Tables A.3.19 (L) and A.3.20 (L) in Annex A.\n\nThe numeracy proficiency of adults with low-educated parents has declined in 11 of the participating countries and economies (see Table A.3.17 (N) in Annex A). Most of these countries saw declines in the numeracy proficiency of adults with medium-educated parents as well. In Singapore, numeracy proficiency among adults with low-educated parents has increased, while in Chile, Finland and Spain this increase was only significant after accounting for demographic changes. In comparison, seven countries saw a decline and four recorded an increase in the numeracy proficiency of adults with highly educated parents. As numeracy generally developed more favourably in this group than among adults with low-educated parents, inequalities in numeracy proficiency increased in many countries (see Table A.3.19 (N) in Annex A).\n\n_Change in skills proficiency related to parental education among younger and older adults_\n\nThis section shows how the proficiency gap associated with socio-economic background (as proxied by parental education) has changed across cycles within three age groups: 16-24 year-olds, 25-44 year-olds and 45-65 year-olds. Due to insufficient sample sizes of some groups, adults with low- and medium- educated parents are considered as a single group (denoted as adults whose parents have at most an upper secondary education) and compared to adults with at least one tertiary-educated parent.\n\nEstonia\n\nAustria\n\nGermany\n\nC echia\n\norea\n\nCanada\n\napan\n\nFlemish Region ( E\n\norway\n\nFinland\n\nFrance\n\nUnited States\n\nIreland\n\nenmark\n\nEngland (U\n\netherlands\n\nSweden\n\nItaly\n\nPoland\n\nSlo ak Republic\n\nSpain\n\nithuania\n\nIsrael\n\new ealand\n\nChile\n\nSingapore\n\nHungary\n\nScore point difference 30\n\n20\n\n10\n\n0\n\n10\n\n20\n\nAd usted Unad u sted\n\nRound, Cycle 1:","publish_date":null,"title":"Figure 3.23. Change in the gap in literacy proficiency between adults with highly educated and low- educated parents","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p143.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p143.pdf"},"request":{"example_id":"chart/b263dc5d-en_p143","product_type":"parse","source_file_path":"b263dc5d-en_p143.pdf"},"started_at":"2026-11-04T19:45:59.033679"} +{"completed_at":"2026-11-04T19:46:03.483947","latency_in_ms":2260,"output":{"example_id":"chart/b263dc5d-en_p144","grounded_pages":[],"layout_pages":[],"markdown":"**142** \n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 3.24. Change in the gap in literacy proficiency between adults with highly educated and medium-/low-educated parents, by age**\n\nChange between cycles in the mean score difference between adults with at least one tertiary-educated parent and adults with parents with at most an upper secondary education (Cycle 2 _minus_ Cycle 1)\n\nNote: Adults aged 16-65; does not include adults who in Cycle 2 were only administered the doorstep interview due to a language barrier, to maximise the comparability across cycles (see Box 1.1 in Chapter 1 and Box 3.2). Differences are unadjusted. Darker colours denote differences that are statistically significant at the 5% level. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the ote for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the unadjusted change in the literacy gap among 16-24 year-olds._ Source: OECD (2018[4]; 2015[5]; 2012[6]), _Survey of Adult Skills (PIAAC) databases_ , http://www.oecd.org/skills/piaac/publicdataandanalysis/ (accessed on 23 September 2024); Tables A.3.21 (L) and A.3.22 (L) in Annex A.\n\nSkills-related inequalities in literacy have developed differently in different age groups (Figure 3.24). Canada, Norway and Sweden recorded an increase in the socio-economic gap in literacy among 16-24 year-olds, while socio-economic differences among older adults have remained stable or, in the case of Sweden, have decreased. In Hungary and New Zealand, the socio-economic gap in literacy has increased both among 16-24 and 25-44 year-olds. In another group of countries, including Austria, Estonia, Germany, Israel, Japan and Korea, inequalities increased among 45-65 year-olds and/or 25-44 year-olds, but not among the youngest adults. France saw an increase in social inequalities in skills among the oldest age group and a decrease among those aged 25-44. Next to Sweden, proficiency disparities among older adults narrowed in Singapore and Spain.\n\nThe socio-economic gap in numeracy increased especially among young adults (aged 16-24) in Hungary, New Zealand and Sweden, while Germany saw an increase in the gap among those aged 25-44 (see Table A.3.21 (N) in Annex A). In France, Lithuania and Spain, inequalities in numeracy proficiency declined among adults aged 25-44. Singapore and Spain saw a decline in the gap among those aged 45-65, while in Estonia this gap has widened.\n\nTo sum up, inequalities in education and skills remain a key policy challenge in many countries. Policies should aim to provide better access to educational opportunities for disadvantaged adults, particularly disadvantaged youth. Policies should also help socio-economically disadvantaged adults to make the most of these opportunities. Perhaps more important than further widening access to education is ensuring that\n\nSweden\n\norway\n\nCanada\n\napan\n\norea\n\nUnited States\n\nEstonia\n\nFlemish Region ( E\n\nSlo ak Republic\n\nFinland\n\nGermany\n\nC echia\n\nFrance\n\nAustria\n\nenmark\n\nIreland\n\nEngland (U\n\nSpain\n\nPoland\n\netherlands\n\nItaly\n\new ealand\n\nSingapore\n\nIsrael\n\nithuania\n\nChile\n\nHungary\n\nScore point difference 30\n\n20\n\n10\n\n0\n\n10\n\n20\n\n30\n\n2 year olds 1 2 year olds year olds\n\nRound, Cycle 1:","pages":[{"line_numbers":[],"markdown":"**142** \n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 3.24. Change in the gap in literacy proficiency between adults with highly educated and medium-/low-educated parents, by age**\n\nChange between cycles in the mean score difference between adults with at least one tertiary-educated parent and adults with parents with at most an upper secondary education (Cycle 2 _minus_ Cycle 1)\n\nNote: Adults aged 16-65; does not include adults who in Cycle 2 were only administered the doorstep interview due to a language barrier, to maximise the comparability across cycles (see Box 1.1 in Chapter 1 and Box 3.2). Differences are unadjusted. Darker colours denote differences that are statistically significant at the 5% level. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the ote for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the unadjusted change in the literacy gap among 16-24 year-olds._ Source: OECD (2018[4]; 2015[5]; 2012[6]), _Survey of Adult Skills (PIAAC) databases_ , http://www.oecd.org/skills/piaac/publicdataandanalysis/ (accessed on 23 September 2024); Tables A.3.21 (L) and A.3.22 (L) in Annex A.\n\nSkills-related inequalities in literacy have developed differently in different age groups (Figure 3.24). Canada, Norway and Sweden recorded an increase in the socio-economic gap in literacy among 16-24 year-olds, while socio-economic differences among older adults have remained stable or, in the case of Sweden, have decreased. In Hungary and New Zealand, the socio-economic gap in literacy has increased both among 16-24 and 25-44 year-olds. In another group of countries, including Austria, Estonia, Germany, Israel, Japan and Korea, inequalities increased among 45-65 year-olds and/or 25-44 year-olds, but not among the youngest adults. France saw an increase in social inequalities in skills among the oldest age group and a decrease among those aged 25-44. Next to Sweden, proficiency disparities among older adults narrowed in Singapore and Spain.\n\nThe socio-economic gap in numeracy increased especially among young adults (aged 16-24) in Hungary, New Zealand and Sweden, while Germany saw an increase in the gap among those aged 25-44 (see Table A.3.21 (N) in Annex A). In France, Lithuania and Spain, inequalities in numeracy proficiency declined among adults aged 25-44. Singapore and Spain saw a decline in the gap among those aged 45-65, while in Estonia this gap has widened.\n\nTo sum up, inequalities in education and skills remain a key policy challenge in many countries. Policies should aim to provide better access to educational opportunities for disadvantaged adults, particularly disadvantaged youth. Policies should also help socio-economically disadvantaged adults to make the most of these opportunities. Perhaps more important than further widening access to education is ensuring that\n\nSweden\n\norway\n\nCanada\n\napan\n\norea\n\nUnited States\n\nEstonia\n\nFlemish Region ( E\n\nSlo ak Republic\n\nFinland\n\nGermany\n\nC echia\n\nFrance\n\nAustria\n\nenmark\n\nIreland\n\nEngland (U\n\nSpain\n\nPoland\n\netherlands\n\nItaly\n\new ealand\n\nSingapore\n\nIsrael\n\nithuania\n\nChile\n\nHungary\n\nScore point difference 30\n\n20\n\n10\n\n0\n\n10\n\n20\n\n30\n\n2 year olds 1 2 year olds year olds\n\nRound, Cycle 1:","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**142** \n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 3.24. Change in the gap in literacy proficiency between adults with highly educated and medium-/low-educated parents, by age**\n\nChange between cycles in the mean score difference between adults with at least one tertiary-educated parent and adults with parents with at most an upper secondary education (Cycle 2 _minus_ Cycle 1)\n\nNote: Adults aged 16-65; does not include adults who in Cycle 2 were only administered the doorstep interview due to a language barrier, to maximise the comparability across cycles (see Box 1.1 in Chapter 1 and Box 3.2). Differences are unadjusted. Darker colours denote differences that are statistically significant at the 5% level. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the ote for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the unadjusted change in the literacy gap among 16-24 year-olds._ Source: OECD (2018[4]; 2015[5]; 2012[6]), _Survey of Adult Skills (PIAAC) databases_ , http://www.oecd.org/skills/piaac/publicdataandanalysis/ (accessed on 23 September 2024); Tables A.3.21 (L) and A.3.22 (L) in Annex A.\n\nSkills-related inequalities in literacy have developed differently in different age groups (Figure 3.24). Canada, Norway and Sweden recorded an increase in the socio-economic gap in literacy among 16-24 year-olds, while socio-economic differences among older adults have remained stable or, in the case of Sweden, have decreased. In Hungary and New Zealand, the socio-economic gap in literacy has increased both among 16-24 and 25-44 year-olds. In another group of countries, including Austria, Estonia, Germany, Israel, Japan and Korea, inequalities increased among 45-65 year-olds and/or 25-44 year-olds, but not among the youngest adults. France saw an increase in social inequalities in skills among the oldest age group and a decrease among those aged 25-44. Next to Sweden, proficiency disparities among older adults narrowed in Singapore and Spain.\n\nThe socio-economic gap in numeracy increased especially among young adults (aged 16-24) in Hungary, New Zealand and Sweden, while Germany saw an increase in the gap among those aged 25-44 (see Table A.3.21 (N) in Annex A). In France, Lithuania and Spain, inequalities in numeracy proficiency declined among adults aged 25-44. Singapore and Spain saw a decline in the gap among those aged 45-65, while in Estonia this gap has widened.\n\nTo sum up, inequalities in education and skills remain a key policy challenge in many countries. Policies should aim to provide better access to educational opportunities for disadvantaged adults, particularly disadvantaged youth. Policies should also help socio-economically disadvantaged adults to make the most of these opportunities. Perhaps more important than further widening access to education is ensuring that\n\nSweden\n\norway\n\nCanada\n\napan\n\norea\n\nUnited States\n\nEstonia\n\nFlemish Region ( E\n\nSlo ak Republic\n\nFinland\n\nGermany\n\nC echia\n\nFrance\n\nAustria\n\nenmark\n\nIreland\n\nEngland (U\n\nSpain\n\nPoland\n\netherlands\n\nItaly\n\new ealand\n\nSingapore\n\nIsrael\n\nithuania\n\nChile\n\nHungary\n\nScore point difference 30\n\n20\n\n10\n\n0\n\n10\n\n20\n\n30\n\n2 year olds 1 2 year olds year olds\n\nRound, Cycle 1:","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p144.pdf"]},"response":{"errors":[],"extract_id":"extract_015d3fad7a35eb0250eea7a3020d9adc","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**142** \n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 3.24. Change in the gap in literacy proficiency between adults with highly educated and medium-/low-educated parents, by age**\n\nChange between cycles in the mean score difference between adults with at least one tertiary-educated parent and adults with parents with at most an upper secondary education (Cycle 2 _minus_ Cycle 1)\n\nNote: Adults aged 16-65; does not include adults who in Cycle 2 were only administered the doorstep interview due to a language barrier, to maximise the comparability across cycles (see Box 1.1 in Chapter 1 and Box 3.2). Differences are unadjusted. Darker colours denote differences that are statistically significant at the 5% level. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the ote for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the unadjusted change in the literacy gap among 16-24 year-olds._ Source: OECD (2018[4]; 2015[5]; 2012[6]), _Survey of Adult Skills (PIAAC) databases_ , http://www.oecd.org/skills/piaac/publicdataandanalysis/ (accessed on 23 September 2024); Tables A.3.21 (L) and A.3.22 (L) in Annex A.\n\nSkills-related inequalities in literacy have developed differently in different age groups (Figure 3.24). Canada, Norway and Sweden recorded an increase in the socio-economic gap in literacy among 16-24 year-olds, while socio-economic differences among older adults have remained stable or, in the case of Sweden, have decreased. In Hungary and New Zealand, the socio-economic gap in literacy has increased both among 16-24 and 25-44 year-olds. In another group of countries, including Austria, Estonia, Germany, Israel, Japan and Korea, inequalities increased among 45-65 year-olds and/or 25-44 year-olds, but not among the youngest adults. France saw an increase in social inequalities in skills among the oldest age group and a decrease among those aged 25-44. Next to Sweden, proficiency disparities among older adults narrowed in Singapore and Spain.\n\nThe socio-economic gap in numeracy increased especially among young adults (aged 16-24) in Hungary, New Zealand and Sweden, while Germany saw an increase in the gap among those aged 25-44 (see Table A.3.21 (N) in Annex A). In France, Lithuania and Spain, inequalities in numeracy proficiency declined among adults aged 25-44. Singapore and Spain saw a decline in the gap among those aged 45-65, while in Estonia this gap has widened.\n\nTo sum up, inequalities in education and skills remain a key policy challenge in many countries. Policies should aim to provide better access to educational opportunities for disadvantaged adults, particularly disadvantaged youth. Policies should also help socio-economically disadvantaged adults to make the most of these opportunities. Perhaps more important than further widening access to education is ensuring that\n\nSweden\n\norway\n\nCanada\n\napan\n\norea\n\nUnited States\n\nEstonia\n\nFlemish Region ( E\n\nSlo ak Republic\n\nFinland\n\nGermany\n\nC echia\n\nFrance\n\nAustria\n\nenmark\n\nIreland\n\nEngland (U\n\nSpain\n\nPoland\n\netherlands\n\nItaly\n\new ealand\n\nSingapore\n\nIsrael\n\nithuania\n\nChile\n\nHungary\n\nScore point difference 30\n\n20\n\n10\n\n0\n\n10\n\n20\n\n30\n\n2 year olds 1 2 year olds year olds\n\nRound, Cycle 1:"],"full_content":"**142** \n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 3.24. Change in the gap in literacy proficiency between adults with highly educated and medium-/low-educated parents, by age**\n\nChange between cycles in the mean score difference between adults with at least one tertiary-educated parent and adults with parents with at most an upper secondary education (Cycle 2 _minus_ Cycle 1)\n\nNote: Adults aged 16-65; does not include adults who in Cycle 2 were only administered the doorstep interview due to a language barrier, to maximise the comparability across cycles (see Box 1.1 in Chapter 1 and Box 3.2). Differences are unadjusted. Darker colours denote differences that are statistically significant at the 5% level. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the ote for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the unadjusted change in the literacy gap among 16-24 year-olds._ Source: OECD (2018[4]; 2015[5]; 2012[6]), _Survey of Adult Skills (PIAAC) databases_ , http://www.oecd.org/skills/piaac/publicdataandanalysis/ (accessed on 23 September 2024); Tables A.3.21 (L) and A.3.22 (L) in Annex A.\n\nSkills-related inequalities in literacy have developed differently in different age groups (Figure 3.24). Canada, Norway and Sweden recorded an increase in the socio-economic gap in literacy among 16-24 year-olds, while socio-economic differences among older adults have remained stable or, in the case of Sweden, have decreased. In Hungary and New Zealand, the socio-economic gap in literacy has increased both among 16-24 and 25-44 year-olds. In another group of countries, including Austria, Estonia, Germany, Israel, Japan and Korea, inequalities increased among 45-65 year-olds and/or 25-44 year-olds, but not among the youngest adults. France saw an increase in social inequalities in skills among the oldest age group and a decrease among those aged 25-44. Next to Sweden, proficiency disparities among older adults narrowed in Singapore and Spain.\n\nThe socio-economic gap in numeracy increased especially among young adults (aged 16-24) in Hungary, New Zealand and Sweden, while Germany saw an increase in the gap among those aged 25-44 (see Table A.3.21 (N) in Annex A). In France, Lithuania and Spain, inequalities in numeracy proficiency declined among adults aged 25-44. Singapore and Spain saw a decline in the gap among those aged 45-65, while in Estonia this gap has widened.\n\nTo sum up, inequalities in education and skills remain a key policy challenge in many countries. Policies should aim to provide better access to educational opportunities for disadvantaged adults, particularly disadvantaged youth. Policies should also help socio-economically disadvantaged adults to make the most of these opportunities. Perhaps more important than further widening access to education is ensuring that\n\nSweden\n\norway\n\nCanada\n\napan\n\norea\n\nUnited States\n\nEstonia\n\nFlemish Region ( E\n\nSlo ak Republic\n\nFinland\n\nGermany\n\nC echia\n\nFrance\n\nAustria\n\nenmark\n\nIreland\n\nEngland (U\n\nSpain\n\nPoland\n\netherlands\n\nItaly\n\new ealand\n\nSingapore\n\nIsrael\n\nithuania\n\nChile\n\nHungary\n\nScore point difference 30\n\n20\n\n10\n\n0\n\n10\n\n20\n\n30\n\n2 year olds 1 2 year olds year olds\n\nRound, Cycle 1:","publish_date":null,"title":"Figure 3.24. Change in the gap in literacy proficiency between adults with highly educated and medium-/low-educated parents, by age","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p144.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p144.pdf"},"request":{"example_id":"chart/b263dc5d-en_p144","product_type":"parse","source_file_path":"b263dc5d-en_p144.pdf"},"started_at":"2026-11-04T19:46:01.223834"} +{"completed_at":"2026-11-04T19:46:05.236483","latency_in_ms":2950,"output":{"example_id":"chart/b263dc5d-en_p156","grounded_pages":[],"layout_pages":[],"markdown":"**154** \n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 4.2. Relationship between education, numeracy and labour-market outcomes**\n\nChange in likelihood for a one-standard-deviation increase in years of education or numeracy proficiency\n\nNote: Adults aged 25-65 not in formal education; does not include adults who were only administered the doorstep interview due to a language barrier (see Box 1.1 in Chapter 1 and Box 4.1). In addition to numeracy and years of education, estimates account for age, gender, immigrant background, parental education and whether one lives with a partner or has children. Darker colours denote differences that are statistically significant at the 5% level. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the Note for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the percentage effect of a one-standard-deviation increase in numeracy proficiency._ Source: Tables A.4.3 (L, N, A) and A.4.5 (L, N, A) in Annex A.\n\nThe link between skills, education and employment is weaker in 2023 than it was ten years ago, when data from the first cycle of the survey were collected in most of the countries and economies taking part. One possible explanation for this is the tighter conditions prevailing in most labour markets in 2022-23 compared to 2011-12. In a tight labour market, people find work more easily, meaning that even individuals with low skills may succeed in getting hired. Figure 4.3 explores this hypothesis, by plotting the change in the unemployment rate between the two cycles (on the x-axis) against the change in the relationship between numeracy skills and the probability of being employed (on the y-axis). Overall, it shows that a fall in the unemployment rate is accompanied by a corresponding fall in the estimated effect of numeracy skills on\n\numeracy ears of education\n\nUnited States\n\nItaly\n\nEngland (U\n\nC echia\n\netherlands\n\nAustria\n\nIreland\n\norway\n\nGermany\n\nFinland\n\nCanad a\n\nithuan ia\n\new ealand\n\nFrance\n\nEstonia\n\nFlemish Region ( E\n\nPoland\n\n**EC average**\n\nat ia\n\nCroatia\n\nSweden\n\nIsrael\n\nSpain\n\nSwit erland\n\nHungary\n\nenmark\n\nPortugal\n\nSlo ak Republic\n\napan\n\norea\n\nChile\n\nSingapore\n\nItaly\n\nSweden\n\nSpain\n\nEngland (U\n\norway\n\nEstonia\n\nCanad a\n\new ealand\n\nGermany\n\nSwit erland\n\nat ia\n\nithuania\n\nC echia\n\nUnited States\n\n**EC average**\n\nFinland\n\nSingapore\n\nChile\n\norea\n\nSlo ak Republic\n\nCroatia\n\nPoland\n\nIsrael\n\nAustria\n\nPortugal\n\nIreland\n\nHungary\n\napan\n\nenmark\n\nFlemish Region ( E\n\netherlands\n\nFrance\n\n12\n\n10\n\n8\n\n2\n\n0\n\n2\n\nPercentage points **. i eli ood of being active in labour mar et instead of inactive**\n\n2\n\n0\n\n2\n\nPercentage points **B. i eli ood of being employed instead of unemployed**","pages":[{"line_numbers":[],"markdown":"**154** \n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 4.2. Relationship between education, numeracy and labour-market outcomes**\n\nChange in likelihood for a one-standard-deviation increase in years of education or numeracy proficiency\n\nNote: Adults aged 25-65 not in formal education; does not include adults who were only administered the doorstep interview due to a language barrier (see Box 1.1 in Chapter 1 and Box 4.1). In addition to numeracy and years of education, estimates account for age, gender, immigrant background, parental education and whether one lives with a partner or has children. Darker colours denote differences that are statistically significant at the 5% level. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the Note for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the percentage effect of a one-standard-deviation increase in numeracy proficiency._ Source: Tables A.4.3 (L, N, A) and A.4.5 (L, N, A) in Annex A.\n\nThe link between skills, education and employment is weaker in 2023 than it was ten years ago, when data from the first cycle of the survey were collected in most of the countries and economies taking part. One possible explanation for this is the tighter conditions prevailing in most labour markets in 2022-23 compared to 2011-12. In a tight labour market, people find work more easily, meaning that even individuals with low skills may succeed in getting hired. Figure 4.3 explores this hypothesis, by plotting the change in the unemployment rate between the two cycles (on the x-axis) against the change in the relationship between numeracy skills and the probability of being employed (on the y-axis). Overall, it shows that a fall in the unemployment rate is accompanied by a corresponding fall in the estimated effect of numeracy skills on\n\numeracy ears of education\n\nUnited States\n\nItaly\n\nEngland (U\n\nC echia\n\netherlands\n\nAustria\n\nIreland\n\norway\n\nGermany\n\nFinland\n\nCanad a\n\nithuan ia\n\new ealand\n\nFrance\n\nEstonia\n\nFlemish Region ( E\n\nPoland\n\n**EC average**\n\nat ia\n\nCroatia\n\nSweden\n\nIsrael\n\nSpain\n\nSwit erland\n\nHungary\n\nenmark\n\nPortugal\n\nSlo ak Republic\n\napan\n\norea\n\nChile\n\nSingapore\n\nItaly\n\nSweden\n\nSpain\n\nEngland (U\n\norway\n\nEstonia\n\nCanad a\n\new ealand\n\nGermany\n\nSwit erland\n\nat ia\n\nithuania\n\nC echia\n\nUnited States\n\n**EC average**\n\nFinland\n\nSingapore\n\nChile\n\norea\n\nSlo ak Republic\n\nCroatia\n\nPoland\n\nIsrael\n\nAustria\n\nPortugal\n\nIreland\n\nHungary\n\napan\n\nenmark\n\nFlemish Region ( E\n\netherlands\n\nFrance\n\n12\n\n10\n\n8\n\n2\n\n0\n\n2\n\nPercentage points **. i eli ood of being active in labour mar et instead of inactive**\n\n2\n\n0\n\n2\n\nPercentage points **B. i eli ood of being employed instead of unemployed**","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**154** \n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 4.2. Relationship between education, numeracy and labour-market outcomes**\n\nChange in likelihood for a one-standard-deviation increase in years of education or numeracy proficiency\n\nNote: Adults aged 25-65 not in formal education; does not include adults who were only administered the doorstep interview due to a language barrier (see Box 1.1 in Chapter 1 and Box 4.1). In addition to numeracy and years of education, estimates account for age, gender, immigrant background, parental education and whether one lives with a partner or has children. Darker colours denote differences that are statistically significant at the 5% level. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the Note for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the percentage effect of a one-standard-deviation increase in numeracy proficiency._ Source: Tables A.4.3 (L, N, A) and A.4.5 (L, N, A) in Annex A.\n\nThe link between skills, education and employment is weaker in 2023 than it was ten years ago, when data from the first cycle of the survey were collected in most of the countries and economies taking part. One possible explanation for this is the tighter conditions prevailing in most labour markets in 2022-23 compared to 2011-12. In a tight labour market, people find work more easily, meaning that even individuals with low skills may succeed in getting hired. Figure 4.3 explores this hypothesis, by plotting the change in the unemployment rate between the two cycles (on the x-axis) against the change in the relationship between numeracy skills and the probability of being employed (on the y-axis). Overall, it shows that a fall in the unemployment rate is accompanied by a corresponding fall in the estimated effect of numeracy skills on\n\numeracy ears of education\n\nUnited States\n\nItaly\n\nEngland (U\n\nC echia\n\netherlands\n\nAustria\n\nIreland\n\norway\n\nGermany\n\nFinland\n\nCanad a\n\nithuan ia\n\new ealand\n\nFrance\n\nEstonia\n\nFlemish Region ( E\n\nPoland\n\n**EC average**\n\nat ia\n\nCroatia\n\nSweden\n\nIsrael\n\nSpain\n\nSwit erland\n\nHungary\n\nenmark\n\nPortugal\n\nSlo ak Republic\n\napan\n\norea\n\nChile\n\nSingapore\n\nItaly\n\nSweden\n\nSpain\n\nEngland (U\n\norway\n\nEstonia\n\nCanad a\n\new ealand\n\nGermany\n\nSwit erland\n\nat ia\n\nithuania\n\nC echia\n\nUnited States\n\n**EC average**\n\nFinland\n\nSingapore\n\nChile\n\norea\n\nSlo ak Republic\n\nCroatia\n\nPoland\n\nIsrael\n\nAustria\n\nPortugal\n\nIreland\n\nHungary\n\napan\n\nenmark\n\nFlemish Region ( E\n\netherlands\n\nFrance\n\n12\n\n10\n\n8\n\n2\n\n0\n\n2\n\nPercentage points **. i eli ood of being active in labour mar et instead of inactive**\n\n2\n\n0\n\n2\n\nPercentage points **B. i eli ood of being employed instead of unemployed**","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p156.pdf"]},"response":{"errors":[],"extract_id":"extract_bea29aebf0cb2c03d2e8b1ebd3c3f947","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**154** \n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 4.2. Relationship between education, numeracy and labour-market outcomes**\n\nChange in likelihood for a one-standard-deviation increase in years of education or numeracy proficiency\n\nNote: Adults aged 25-65 not in formal education; does not include adults who were only administered the doorstep interview due to a language barrier (see Box 1.1 in Chapter 1 and Box 4.1). In addition to numeracy and years of education, estimates account for age, gender, immigrant background, parental education and whether one lives with a partner or has children. Darker colours denote differences that are statistically significant at the 5% level. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the Note for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the percentage effect of a one-standard-deviation increase in numeracy proficiency._ Source: Tables A.4.3 (L, N, A) and A.4.5 (L, N, A) in Annex A.\n\nThe link between skills, education and employment is weaker in 2023 than it was ten years ago, when data from the first cycle of the survey were collected in most of the countries and economies taking part. One possible explanation for this is the tighter conditions prevailing in most labour markets in 2022-23 compared to 2011-12. In a tight labour market, people find work more easily, meaning that even individuals with low skills may succeed in getting hired. Figure 4.3 explores this hypothesis, by plotting the change in the unemployment rate between the two cycles (on the x-axis) against the change in the relationship between numeracy skills and the probability of being employed (on the y-axis). Overall, it shows that a fall in the unemployment rate is accompanied by a corresponding fall in the estimated effect of numeracy skills on\n\numeracy ears of education\n\nUnited States\n\nItaly\n\nEngland (U\n\nC echia\n\netherlands\n\nAustria\n\nIreland\n\norway\n\nGermany\n\nFinland\n\nCanad a\n\nithuan ia\n\new ealand\n\nFrance\n\nEstonia\n\nFlemish Region ( E\n\nPoland\n\n**EC average**\n\nat ia\n\nCroatia\n\nSweden\n\nIsrael\n\nSpain\n\nSwit erland\n\nHungary\n\nenmark\n\nPortugal\n\nSlo ak Republic\n\napan\n\norea\n\nChile\n\nSingapore\n\nItaly\n\nSweden\n\nSpain\n\nEngland (U\n\norway\n\nEstonia\n\nCanad a\n\new ealand\n\nGermany\n\nSwit erland\n\nat ia\n\nithuania\n\nC echia\n\nUnited States\n\n**EC average**\n\nFinland\n\nSingapore\n\nChile\n\norea\n\nSlo ak Republic\n\nCroatia\n\nPoland\n\nIsrael\n\nAustria\n\nPortugal\n\nIreland\n\nHungary\n\napan\n\nenmark\n\nFlemish Region ( E\n\netherlands\n\nFrance\n\n12\n\n10\n\n8\n\n2\n\n0\n\n2\n\nPercentage points **. i eli ood of being active in labour mar et instead of inactive**\n\n2\n\n0\n\n2\n\nPercentage points **B. i eli ood of being employed instead of unemployed**"],"full_content":"**154** \n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 4.2. Relationship between education, numeracy and labour-market outcomes**\n\nChange in likelihood for a one-standard-deviation increase in years of education or numeracy proficiency\n\nNote: Adults aged 25-65 not in formal education; does not include adults who were only administered the doorstep interview due to a language barrier (see Box 1.1 in Chapter 1 and Box 4.1). In addition to numeracy and years of education, estimates account for age, gender, immigrant background, parental education and whether one lives with a partner or has children. Darker colours denote differences that are statistically significant at the 5% level. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the Note for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the percentage effect of a one-standard-deviation increase in numeracy proficiency._ Source: Tables A.4.3 (L, N, A) and A.4.5 (L, N, A) in Annex A.\n\nThe link between skills, education and employment is weaker in 2023 than it was ten years ago, when data from the first cycle of the survey were collected in most of the countries and economies taking part. One possible explanation for this is the tighter conditions prevailing in most labour markets in 2022-23 compared to 2011-12. In a tight labour market, people find work more easily, meaning that even individuals with low skills may succeed in getting hired. Figure 4.3 explores this hypothesis, by plotting the change in the unemployment rate between the two cycles (on the x-axis) against the change in the relationship between numeracy skills and the probability of being employed (on the y-axis). Overall, it shows that a fall in the unemployment rate is accompanied by a corresponding fall in the estimated effect of numeracy skills on\n\numeracy ears of education\n\nUnited States\n\nItaly\n\nEngland (U\n\nC echia\n\netherlands\n\nAustria\n\nIreland\n\norway\n\nGermany\n\nFinland\n\nCanad a\n\nithuan ia\n\new ealand\n\nFrance\n\nEstonia\n\nFlemish Region ( E\n\nPoland\n\n**EC average**\n\nat ia\n\nCroatia\n\nSweden\n\nIsrael\n\nSpain\n\nSwit erland\n\nHungary\n\nenmark\n\nPortugal\n\nSlo ak Republic\n\napan\n\norea\n\nChile\n\nSingapore\n\nItaly\n\nSweden\n\nSpain\n\nEngland (U\n\norway\n\nEstonia\n\nCanad a\n\new ealand\n\nGermany\n\nSwit erland\n\nat ia\n\nithuania\n\nC echia\n\nUnited States\n\n**EC average**\n\nFinland\n\nSingapore\n\nChile\n\norea\n\nSlo ak Republic\n\nCroatia\n\nPoland\n\nIsrael\n\nAustria\n\nPortugal\n\nIreland\n\nHungary\n\napan\n\nenmark\n\nFlemish Region ( E\n\netherlands\n\nFrance\n\n12\n\n10\n\n8\n\n2\n\n0\n\n2\n\nPercentage points **. i eli ood of being active in labour mar et instead of inactive**\n\n2\n\n0\n\n2\n\nPercentage points **B. i eli ood of being employed instead of unemployed**","publish_date":null,"title":"Figure 4.2. Relationship between education, numeracy and labour-market outcomes","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p156.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p156.pdf"},"request":{"example_id":"chart/b263dc5d-en_p156","product_type":"parse","source_file_path":"b263dc5d-en_p156.pdf"},"started_at":"2026-11-04T19:46:02.286126"} +{"completed_at":"2026-11-04T19:46:05.610178","latency_in_ms":1734,"output":{"example_id":"chart/b263dc5d-en_p159","grounded_pages":[],"layout_pages":[],"markdown":" **157**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 4.4. Median wages, by numeracy proficiency level**\n\nPPP-adjusted 2022 USD\n\nNote: Employed adults aged 25-65 not in formal education; does not include adults who were only administered the doorstep interview due to a language barrier (see Box 1.1 in Chapter 1 and Box 4.1). Includes bonuses and earnings by self-employed individuals. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the Note for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of median gross hourly earnings for Level 4 and above._ Source: Table A.4.7 (N) in Annex A.\n\nNumeracy skills and years of education continue to have statistically significant associations with wages independently of one another, and the magnitude of these effects varies considerably across participating countries and economies (Figure 4.5). The results presented here are adjusted for socio-demographic characteristics. On average, an increase of one standard deviation in an individual’s years of education is associated with a 16% increase in wages. This is broadly in line with existing international evidence (Clark and Abildgaard Nielsen, 2024[38]; Patrinos, 2023[39]). An increase of one standard deviation in numeracy proficiency is associated with a 9% increase in wages. ⁶\n\nIt is important to note that skills are associated with workers’ wages even after accounting for years of education. For 27 out of 31 participating countries and economies, the relationship between numeracy and wages is statistically significant, and in France the estimated effect for numeracy proficiency in fact exceeds that of years of education. The greatest differences between the effects of numeracy and years of education are found in Singapore (22 percentage points). Numeracy proficiency has the largest estimated relationship with wages in Chile, England (United Kingdom) and Germany (over 14%), and the smallest (statistically significant) relationship in the Flemish Region (Belgium), Italy and Poland (under 6%). For education, the largest estimated relationships are in Chile and Singapore (30% or more) and the smallest are in France and Japan (10% or less).\n\ne el 1 and below e el and abo e Total\n\nSwit erland\n\nenmark\n\nUnited States\n\norway\n\nGermany\n\netherlands\n\nIreland\n\nFlemish Region ( E\n\nAustria\n\nSingapore\n\nCanada\n\nEngland (U\n\nFinland\n\new ealand\n\nIsrael\n\n**EC average**\n\nSweden\n\nFrance\n\nItaly\n\norea\n\nSpain\n\nithuania\n\nEstonia\n\nat ia\n\nHungary\n\napan\n\nChile\n\nPortugal\n\nC echia\n\nCroatia\n\nPoland\n\nSlo ak Republic\n\n0\n\n0\n\n0\n\n30\n\n20\n\n10\n\n0\n\nGross hourly earnings","pages":[{"line_numbers":[],"markdown":" **157**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 4.4. Median wages, by numeracy proficiency level**\n\nPPP-adjusted 2022 USD\n\nNote: Employed adults aged 25-65 not in formal education; does not include adults who were only administered the doorstep interview due to a language barrier (see Box 1.1 in Chapter 1 and Box 4.1). Includes bonuses and earnings by self-employed individuals. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the Note for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of median gross hourly earnings for Level 4 and above._ Source: Table A.4.7 (N) in Annex A.\n\nNumeracy skills and years of education continue to have statistically significant associations with wages independently of one another, and the magnitude of these effects varies considerably across participating countries and economies (Figure 4.5). The results presented here are adjusted for socio-demographic characteristics. On average, an increase of one standard deviation in an individual’s years of education is associated with a 16% increase in wages. This is broadly in line with existing international evidence (Clark and Abildgaard Nielsen, 2024[38]; Patrinos, 2023[39]). An increase of one standard deviation in numeracy proficiency is associated with a 9% increase in wages. ⁶\n\nIt is important to note that skills are associated with workers’ wages even after accounting for years of education. For 27 out of 31 participating countries and economies, the relationship between numeracy and wages is statistically significant, and in France the estimated effect for numeracy proficiency in fact exceeds that of years of education. The greatest differences between the effects of numeracy and years of education are found in Singapore (22 percentage points). Numeracy proficiency has the largest estimated relationship with wages in Chile, England (United Kingdom) and Germany (over 14%), and the smallest (statistically significant) relationship in the Flemish Region (Belgium), Italy and Poland (under 6%). For education, the largest estimated relationships are in Chile and Singapore (30% or more) and the smallest are in France and Japan (10% or less).\n\ne el 1 and below e el and abo e Total\n\nSwit erland\n\nenmark\n\nUnited States\n\norway\n\nGermany\n\netherlands\n\nIreland\n\nFlemish Region ( E\n\nAustria\n\nSingapore\n\nCanada\n\nEngland (U\n\nFinland\n\new ealand\n\nIsrael\n\n**EC average**\n\nSweden\n\nFrance\n\nItaly\n\norea\n\nSpain\n\nithuania\n\nEstonia\n\nat ia\n\nHungary\n\napan\n\nChile\n\nPortugal\n\nC echia\n\nCroatia\n\nPoland\n\nSlo ak Republic\n\n0\n\n0\n\n0\n\n30\n\n20\n\n10\n\n0\n\nGross hourly earnings","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":" **157**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 4.4. Median wages, by numeracy proficiency level**\n\nPPP-adjusted 2022 USD\n\nNote: Employed adults aged 25-65 not in formal education; does not include adults who were only administered the doorstep interview due to a language barrier (see Box 1.1 in Chapter 1 and Box 4.1). Includes bonuses and earnings by self-employed individuals. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the Note for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of median gross hourly earnings for Level 4 and above._ Source: Table A.4.7 (N) in Annex A.\n\nNumeracy skills and years of education continue to have statistically significant associations with wages independently of one another, and the magnitude of these effects varies considerably across participating countries and economies (Figure 4.5). The results presented here are adjusted for socio-demographic characteristics. On average, an increase of one standard deviation in an individual’s years of education is associated with a 16% increase in wages. This is broadly in line with existing international evidence (Clark and Abildgaard Nielsen, 2024[38]; Patrinos, 2023[39]). An increase of one standard deviation in numeracy proficiency is associated with a 9% increase in wages. ⁶\n\nIt is important to note that skills are associated with workers’ wages even after accounting for years of education. For 27 out of 31 participating countries and economies, the relationship between numeracy and wages is statistically significant, and in France the estimated effect for numeracy proficiency in fact exceeds that of years of education. The greatest differences between the effects of numeracy and years of education are found in Singapore (22 percentage points). Numeracy proficiency has the largest estimated relationship with wages in Chile, England (United Kingdom) and Germany (over 14%), and the smallest (statistically significant) relationship in the Flemish Region (Belgium), Italy and Poland (under 6%). For education, the largest estimated relationships are in Chile and Singapore (30% or more) and the smallest are in France and Japan (10% or less).\n\ne el 1 and below e el and abo e Total\n\nSwit erland\n\nenmark\n\nUnited States\n\norway\n\nGermany\n\netherlands\n\nIreland\n\nFlemish Region ( E\n\nAustria\n\nSingapore\n\nCanada\n\nEngland (U\n\nFinland\n\new ealand\n\nIsrael\n\n**EC average**\n\nSweden\n\nFrance\n\nItaly\n\norea\n\nSpain\n\nithuania\n\nEstonia\n\nat ia\n\nHungary\n\napan\n\nChile\n\nPortugal\n\nC echia\n\nCroatia\n\nPoland\n\nSlo ak Republic\n\n0\n\n0\n\n0\n\n30\n\n20\n\n10\n\n0\n\nGross hourly earnings","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p159.pdf"]},"response":{"errors":[],"extract_id":"extract_2262cc93930a20edb06ae72ee51c7396","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":[" **157**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 4.4. Median wages, by numeracy proficiency level**\n\nPPP-adjusted 2022 USD\n\nNote: Employed adults aged 25-65 not in formal education; does not include adults who were only administered the doorstep interview due to a language barrier (see Box 1.1 in Chapter 1 and Box 4.1). Includes bonuses and earnings by self-employed individuals. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the Note for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of median gross hourly earnings for Level 4 and above._ Source: Table A.4.7 (N) in Annex A.\n\nNumeracy skills and years of education continue to have statistically significant associations with wages independently of one another, and the magnitude of these effects varies considerably across participating countries and economies (Figure 4.5). The results presented here are adjusted for socio-demographic characteristics. On average, an increase of one standard deviation in an individual’s years of education is associated with a 16% increase in wages. This is broadly in line with existing international evidence (Clark and Abildgaard Nielsen, 2024[38]; Patrinos, 2023[39]). An increase of one standard deviation in numeracy proficiency is associated with a 9% increase in wages. ⁶\n\nIt is important to note that skills are associated with workers’ wages even after accounting for years of education. For 27 out of 31 participating countries and economies, the relationship between numeracy and wages is statistically significant, and in France the estimated effect for numeracy proficiency in fact exceeds that of years of education. The greatest differences between the effects of numeracy and years of education are found in Singapore (22 percentage points). Numeracy proficiency has the largest estimated relationship with wages in Chile, England (United Kingdom) and Germany (over 14%), and the smallest (statistically significant) relationship in the Flemish Region (Belgium), Italy and Poland (under 6%). For education, the largest estimated relationships are in Chile and Singapore (30% or more) and the smallest are in France and Japan (10% or less).\n\ne el 1 and below e el and abo e Total\n\nSwit erland\n\nenmark\n\nUnited States\n\norway\n\nGermany\n\netherlands\n\nIreland\n\nFlemish Region ( E\n\nAustria\n\nSingapore\n\nCanada\n\nEngland (U\n\nFinland\n\new ealand\n\nIsrael\n\n**EC average**\n\nSweden\n\nFrance\n\nItaly\n\norea\n\nSpain\n\nithuania\n\nEstonia\n\nat ia\n\nHungary\n\napan\n\nChile\n\nPortugal\n\nC echia\n\nCroatia\n\nPoland\n\nSlo ak Republic\n\n0\n\n0\n\n0\n\n30\n\n20\n\n10\n\n0\n\nGross hourly earnings"],"full_content":" **157**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 4.4. Median wages, by numeracy proficiency level**\n\nPPP-adjusted 2022 USD\n\nNote: Employed adults aged 25-65 not in formal education; does not include adults who were only administered the doorstep interview due to a language barrier (see Box 1.1 in Chapter 1 and Box 4.1). Includes bonuses and earnings by self-employed individuals. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the Note for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of median gross hourly earnings for Level 4 and above._ Source: Table A.4.7 (N) in Annex A.\n\nNumeracy skills and years of education continue to have statistically significant associations with wages independently of one another, and the magnitude of these effects varies considerably across participating countries and economies (Figure 4.5). The results presented here are adjusted for socio-demographic characteristics. On average, an increase of one standard deviation in an individual’s years of education is associated with a 16% increase in wages. This is broadly in line with existing international evidence (Clark and Abildgaard Nielsen, 2024[38]; Patrinos, 2023[39]). An increase of one standard deviation in numeracy proficiency is associated with a 9% increase in wages. ⁶\n\nIt is important to note that skills are associated with workers’ wages even after accounting for years of education. For 27 out of 31 participating countries and economies, the relationship between numeracy and wages is statistically significant, and in France the estimated effect for numeracy proficiency in fact exceeds that of years of education. The greatest differences between the effects of numeracy and years of education are found in Singapore (22 percentage points). Numeracy proficiency has the largest estimated relationship with wages in Chile, England (United Kingdom) and Germany (over 14%), and the smallest (statistically significant) relationship in the Flemish Region (Belgium), Italy and Poland (under 6%). For education, the largest estimated relationships are in Chile and Singapore (30% or more) and the smallest are in France and Japan (10% or less).\n\ne el 1 and below e el and abo e Total\n\nSwit erland\n\nenmark\n\nUnited States\n\norway\n\nGermany\n\netherlands\n\nIreland\n\nFlemish Region ( E\n\nAustria\n\nSingapore\n\nCanada\n\nEngland (U\n\nFinland\n\new ealand\n\nIsrael\n\n**EC average**\n\nSweden\n\nFrance\n\nItaly\n\norea\n\nSpain\n\nithuania\n\nEstonia\n\nat ia\n\nHungary\n\napan\n\nChile\n\nPortugal\n\nC echia\n\nCroatia\n\nPoland\n\nSlo ak Republic\n\n0\n\n0\n\n0\n\n30\n\n20\n\n10\n\n0\n\nGross hourly earnings","publish_date":null,"title":"Figure 4.4. Median wages, by numeracy proficiency level","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p159.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p159.pdf"},"request":{"example_id":"chart/b263dc5d-en_p159","product_type":"parse","source_file_path":"b263dc5d-en_p159.pdf"},"started_at":"2026-11-04T19:46:03.875334"} +{"completed_at":"2026-11-04T19:46:07.022296","latency_in_ms":1341,"output":{"example_id":"chart/b263dc5d-en_p160","grounded_pages":[],"layout_pages":[],"markdown":"**158** \n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 4.5. Relationship between education, numeracy and wages**\n\nEffect of a one-standard-deviation increase on hourly wages\n\nNote: Employed adults aged 25-65 not in formal education; does not include adults who were only administered the doorstep interview due to a language barrier (see Box 1.1 in Chapter 1 and Box 4.1). In addition to numeracy and years of education, estimates account for work experience, age, gender, immigrant background, and whether one lives with a partner or has children; wages are gross hourly earnings for employed and self-employed individuals, including bonuses, in PPP-adjusted 2022 USD. Darker colours denote differences that are statistically significant at the 5% level. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the Note for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the percentage effect of a one-standard-deviation increase in numeracy proficiency._ Source: Table A.4.8 (L, N, A) in Annex A.\n\n_Decomposition of variation in wages_\n\nThis section considers how much of the observed variation in wages is accounted for (or “explained”) by differences in observable characteristics such as education, numeracy and literacy proficiency, work experience and socio-demographic factors. It decomposes the variance in earnings to understand how important different factors are in explaining the observed differences in wages between individuals. Wage decomposition has been used to explain, for example, gender, race or geographical wage differences (Akee, Jones and Porter, 2019[40]; Blau and Kahn, 2017[41]; He and Jiang, 2023[42]).\n\nOn average, 23% of wage variation can be accounted for by the observable variables of skills proficiency, educational attainment, field of study, job tenure, and individual characteristics – age, gender, immigrant background, and whether one lives with a partner or has children (Figure 4.6). In Croatia, Finland, Germany, Japan, Latvia and Singapore, they account for over 30% of the variation in wages, compared to less than 15% in Ireland, Israel, Korea, Poland and the Slovak Republic. The fact that observable characteristics explain only 23% of the variation in earnings on average means that 77% of the variation is determined by other factors.\n\nTogether, years of education and numeracy and literacy proficiency account for around two-thirds of the total explained variation on average. In Chile and Hungary this share exceeds 80%, whereas for Japan it is less than half. Education consistently accounts for the greatest share of wage variation. It accounts for 9% of the total variation on average and is the greatest single factor in 27 out of 31 participating countries and economies. Proficiency – across both numeracy and literacy domains – accounts for 6% of the variation in wages on average across OECD countries, and at least 10% in England (United Kingdom),\n\n0\n\n3\n\n30\n\n2\n\n20\n\n1\n\n10\n\n0\n\nChile\n\nEngland (U\n\nGermany\n\nSingapore\n\nFrance\n\nCanada\n\nat ia\n\nEstonia\n\nIreland\n\nSwit erland\n\nUnited States\n\nAustria\n\napan\n\nHungary\n\netherlands\n\nC echia\n\nPortugal\n\n**EC average**\n\new ealand\n\nenmark\n\nithuania\n\norway\n\nCroatia\n\nFinland\n\nSweden\n\nFlemish Region ( E\n\nIsrael\n\nItaly\n\nPoland\n\nSpain\n\norea\n\nSlo ak Republic\n\nchange umeracy ears of education","pages":[{"line_numbers":[],"markdown":"**158** \n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 4.5. Relationship between education, numeracy and wages**\n\nEffect of a one-standard-deviation increase on hourly wages\n\nNote: Employed adults aged 25-65 not in formal education; does not include adults who were only administered the doorstep interview due to a language barrier (see Box 1.1 in Chapter 1 and Box 4.1). In addition to numeracy and years of education, estimates account for work experience, age, gender, immigrant background, and whether one lives with a partner or has children; wages are gross hourly earnings for employed and self-employed individuals, including bonuses, in PPP-adjusted 2022 USD. Darker colours denote differences that are statistically significant at the 5% level. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the Note for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the percentage effect of a one-standard-deviation increase in numeracy proficiency._ Source: Table A.4.8 (L, N, A) in Annex A.\n\n_Decomposition of variation in wages_\n\nThis section considers how much of the observed variation in wages is accounted for (or “explained”) by differences in observable characteristics such as education, numeracy and literacy proficiency, work experience and socio-demographic factors. It decomposes the variance in earnings to understand how important different factors are in explaining the observed differences in wages between individuals. Wage decomposition has been used to explain, for example, gender, race or geographical wage differences (Akee, Jones and Porter, 2019[40]; Blau and Kahn, 2017[41]; He and Jiang, 2023[42]).\n\nOn average, 23% of wage variation can be accounted for by the observable variables of skills proficiency, educational attainment, field of study, job tenure, and individual characteristics – age, gender, immigrant background, and whether one lives with a partner or has children (Figure 4.6). In Croatia, Finland, Germany, Japan, Latvia and Singapore, they account for over 30% of the variation in wages, compared to less than 15% in Ireland, Israel, Korea, Poland and the Slovak Republic. The fact that observable characteristics explain only 23% of the variation in earnings on average means that 77% of the variation is determined by other factors.\n\nTogether, years of education and numeracy and literacy proficiency account for around two-thirds of the total explained variation on average. In Chile and Hungary this share exceeds 80%, whereas for Japan it is less than half. Education consistently accounts for the greatest share of wage variation. It accounts for 9% of the total variation on average and is the greatest single factor in 27 out of 31 participating countries and economies. Proficiency – across both numeracy and literacy domains – accounts for 6% of the variation in wages on average across OECD countries, and at least 10% in England (United Kingdom),\n\n0\n\n3\n\n30\n\n2\n\n20\n\n1\n\n10\n\n0\n\nChile\n\nEngland (U\n\nGermany\n\nSingapore\n\nFrance\n\nCanada\n\nat ia\n\nEstonia\n\nIreland\n\nSwit erland\n\nUnited States\n\nAustria\n\napan\n\nHungary\n\netherlands\n\nC echia\n\nPortugal\n\n**EC average**\n\new ealand\n\nenmark\n\nithuania\n\norway\n\nCroatia\n\nFinland\n\nSweden\n\nFlemish Region ( E\n\nIsrael\n\nItaly\n\nPoland\n\nSpain\n\norea\n\nSlo ak Republic\n\nchange umeracy ears of education","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**158** \n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 4.5. Relationship between education, numeracy and wages**\n\nEffect of a one-standard-deviation increase on hourly wages\n\nNote: Employed adults aged 25-65 not in formal education; does not include adults who were only administered the doorstep interview due to a language barrier (see Box 1.1 in Chapter 1 and Box 4.1). In addition to numeracy and years of education, estimates account for work experience, age, gender, immigrant background, and whether one lives with a partner or has children; wages are gross hourly earnings for employed and self-employed individuals, including bonuses, in PPP-adjusted 2022 USD. Darker colours denote differences that are statistically significant at the 5% level. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the Note for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the percentage effect of a one-standard-deviation increase in numeracy proficiency._ Source: Table A.4.8 (L, N, A) in Annex A.\n\n_Decomposition of variation in wages_\n\nThis section considers how much of the observed variation in wages is accounted for (or “explained”) by differences in observable characteristics such as education, numeracy and literacy proficiency, work experience and socio-demographic factors. It decomposes the variance in earnings to understand how important different factors are in explaining the observed differences in wages between individuals. Wage decomposition has been used to explain, for example, gender, race or geographical wage differences (Akee, Jones and Porter, 2019[40]; Blau and Kahn, 2017[41]; He and Jiang, 2023[42]).\n\nOn average, 23% of wage variation can be accounted for by the observable variables of skills proficiency, educational attainment, field of study, job tenure, and individual characteristics – age, gender, immigrant background, and whether one lives with a partner or has children (Figure 4.6). In Croatia, Finland, Germany, Japan, Latvia and Singapore, they account for over 30% of the variation in wages, compared to less than 15% in Ireland, Israel, Korea, Poland and the Slovak Republic. The fact that observable characteristics explain only 23% of the variation in earnings on average means that 77% of the variation is determined by other factors.\n\nTogether, years of education and numeracy and literacy proficiency account for around two-thirds of the total explained variation on average. In Chile and Hungary this share exceeds 80%, whereas for Japan it is less than half. Education consistently accounts for the greatest share of wage variation. It accounts for 9% of the total variation on average and is the greatest single factor in 27 out of 31 participating countries and economies. Proficiency – across both numeracy and literacy domains – accounts for 6% of the variation in wages on average across OECD countries, and at least 10% in England (United Kingdom),\n\n0\n\n3\n\n30\n\n2\n\n20\n\n1\n\n10\n\n0\n\nChile\n\nEngland (U\n\nGermany\n\nSingapore\n\nFrance\n\nCanada\n\nat ia\n\nEstonia\n\nIreland\n\nSwit erland\n\nUnited States\n\nAustria\n\napan\n\nHungary\n\netherlands\n\nC echia\n\nPortugal\n\n**EC average**\n\new ealand\n\nenmark\n\nithuania\n\norway\n\nCroatia\n\nFinland\n\nSweden\n\nFlemish Region ( E\n\nIsrael\n\nItaly\n\nPoland\n\nSpain\n\norea\n\nSlo ak Republic\n\nchange umeracy ears of education","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p160.pdf"]},"response":{"errors":[],"extract_id":"extract_f80935fc4fdcb3889f91e70a6f14e8bc","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**158** \n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 4.5. Relationship between education, numeracy and wages**\n\nEffect of a one-standard-deviation increase on hourly wages\n\nNote: Employed adults aged 25-65 not in formal education; does not include adults who were only administered the doorstep interview due to a language barrier (see Box 1.1 in Chapter 1 and Box 4.1). In addition to numeracy and years of education, estimates account for work experience, age, gender, immigrant background, and whether one lives with a partner or has children; wages are gross hourly earnings for employed and self-employed individuals, including bonuses, in PPP-adjusted 2022 USD. Darker colours denote differences that are statistically significant at the 5% level. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the Note for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the percentage effect of a one-standard-deviation increase in numeracy proficiency._ Source: Table A.4.8 (L, N, A) in Annex A.\n\n_Decomposition of variation in wages_\n\nThis section considers how much of the observed variation in wages is accounted for (or “explained”) by differences in observable characteristics such as education, numeracy and literacy proficiency, work experience and socio-demographic factors. It decomposes the variance in earnings to understand how important different factors are in explaining the observed differences in wages between individuals. Wage decomposition has been used to explain, for example, gender, race or geographical wage differences (Akee, Jones and Porter, 2019[40]; Blau and Kahn, 2017[41]; He and Jiang, 2023[42]).\n\nOn average, 23% of wage variation can be accounted for by the observable variables of skills proficiency, educational attainment, field of study, job tenure, and individual characteristics – age, gender, immigrant background, and whether one lives with a partner or has children (Figure 4.6). In Croatia, Finland, Germany, Japan, Latvia and Singapore, they account for over 30% of the variation in wages, compared to less than 15% in Ireland, Israel, Korea, Poland and the Slovak Republic. The fact that observable characteristics explain only 23% of the variation in earnings on average means that 77% of the variation is determined by other factors.\n\nTogether, years of education and numeracy and literacy proficiency account for around two-thirds of the total explained variation on average. In Chile and Hungary this share exceeds 80%, whereas for Japan it is less than half. Education consistently accounts for the greatest share of wage variation. It accounts for 9% of the total variation on average and is the greatest single factor in 27 out of 31 participating countries and economies. Proficiency – across both numeracy and literacy domains – accounts for 6% of the variation in wages on average across OECD countries, and at least 10% in England (United Kingdom),\n\n0\n\n3\n\n30\n\n2\n\n20\n\n1\n\n10\n\n0\n\nChile\n\nEngland (U\n\nGermany\n\nSingapore\n\nFrance\n\nCanada\n\nat ia\n\nEstonia\n\nIreland\n\nSwit erland\n\nUnited States\n\nAustria\n\napan\n\nHungary\n\netherlands\n\nC echia\n\nPortugal\n\n**EC average**\n\new ealand\n\nenmark\n\nithuania\n\norway\n\nCroatia\n\nFinland\n\nSweden\n\nFlemish Region ( E\n\nIsrael\n\nItaly\n\nPoland\n\nSpain\n\norea\n\nSlo ak Republic\n\nchange umeracy ears of education"],"full_content":"**158** \n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 4.5. Relationship between education, numeracy and wages**\n\nEffect of a one-standard-deviation increase on hourly wages\n\nNote: Employed adults aged 25-65 not in formal education; does not include adults who were only administered the doorstep interview due to a language barrier (see Box 1.1 in Chapter 1 and Box 4.1). In addition to numeracy and years of education, estimates account for work experience, age, gender, immigrant background, and whether one lives with a partner or has children; wages are gross hourly earnings for employed and self-employed individuals, including bonuses, in PPP-adjusted 2022 USD. Darker colours denote differences that are statistically significant at the 5% level. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the Note for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the percentage effect of a one-standard-deviation increase in numeracy proficiency._ Source: Table A.4.8 (L, N, A) in Annex A.\n\n_Decomposition of variation in wages_\n\nThis section considers how much of the observed variation in wages is accounted for (or “explained”) by differences in observable characteristics such as education, numeracy and literacy proficiency, work experience and socio-demographic factors. It decomposes the variance in earnings to understand how important different factors are in explaining the observed differences in wages between individuals. Wage decomposition has been used to explain, for example, gender, race or geographical wage differences (Akee, Jones and Porter, 2019[40]; Blau and Kahn, 2017[41]; He and Jiang, 2023[42]).\n\nOn average, 23% of wage variation can be accounted for by the observable variables of skills proficiency, educational attainment, field of study, job tenure, and individual characteristics – age, gender, immigrant background, and whether one lives with a partner or has children (Figure 4.6). In Croatia, Finland, Germany, Japan, Latvia and Singapore, they account for over 30% of the variation in wages, compared to less than 15% in Ireland, Israel, Korea, Poland and the Slovak Republic. The fact that observable characteristics explain only 23% of the variation in earnings on average means that 77% of the variation is determined by other factors.\n\nTogether, years of education and numeracy and literacy proficiency account for around two-thirds of the total explained variation on average. In Chile and Hungary this share exceeds 80%, whereas for Japan it is less than half. Education consistently accounts for the greatest share of wage variation. It accounts for 9% of the total variation on average and is the greatest single factor in 27 out of 31 participating countries and economies. Proficiency – across both numeracy and literacy domains – accounts for 6% of the variation in wages on average across OECD countries, and at least 10% in England (United Kingdom),\n\n0\n\n3\n\n30\n\n2\n\n20\n\n1\n\n10\n\n0\n\nChile\n\nEngland (U\n\nGermany\n\nSingapore\n\nFrance\n\nCanada\n\nat ia\n\nEstonia\n\nIreland\n\nSwit erland\n\nUnited States\n\nAustria\n\napan\n\nHungary\n\netherlands\n\nC echia\n\nPortugal\n\n**EC average**\n\new ealand\n\nenmark\n\nithuania\n\norway\n\nCroatia\n\nFinland\n\nSweden\n\nFlemish Region ( E\n\nIsrael\n\nItaly\n\nPoland\n\nSpain\n\norea\n\nSlo ak Republic\n\nchange umeracy ears of education","publish_date":null,"title":"Figure 4.5. Relationship between education, numeracy and wages","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p160.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p160.pdf"},"request":{"example_id":"chart/b263dc5d-en_p160","product_type":"parse","source_file_path":"b263dc5d-en_p160.pdf"},"started_at":"2026-11-04T19:46:05.681073"} +{"completed_at":"2026-11-04T19:46:12.266946","latency_in_ms":4857,"output":{"example_id":"chart/b263dc5d-en_p161","grounded_pages":[],"layout_pages":[],"markdown":" **159**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\nFrance and Germany. In Korea and the Slovak Republic, conversely, skills account for less than 1% of overall wage variation.\n\n### **Figure 4.6. Contribution of observable characteristics to variation in wages**\n\nVariation explained by each factor\n\nNote: Employed adults aged 25-65 not in formal education; does not include adults who were only administered the doorstep interview due to a language barrier (see Box 1.1 in Chapter 1 and Box 4.1). Decomposition calculates the percentage contribution to variation in log wages of the following factors: skills proficiency (numeracy and literacy), years of education, job tenure, field of study (nine categories) and individual characteristics (age, gender, immigrant background, and whether one lives with a partner or has children); wages are (log) gross hourly earnings for employed and self-employed individuals, including bonuses, in PPP-adjusted 2022 USD. See Fields (2003[43]) for further details on the decomposition method. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the ote for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the percentage of variation in wages explained by skills proficiency._ Source: Table A.4.9a (L, N) in Annex A.\n\nObservable individual characteristics do not typically account for a large proportion of variation in wages, at just 4% on average across participating OECD countries and economies. However, the share is 6% in Finland and is at least 8% in Estonia, Japan and Latvia. Japan is also an outlier in the explanatory power of years of job tenure, which accounts for 9% of its wage variation, compared to the OECD average of 2%. For 11 countries the share attributable to job tenure is less than 0.5% (Figure 4.6).\n\nThe explanatory power of these observable characteristics differs by age. The total share of wage variation explained increases from 22% on average among 25-34 year-olds to 27% for 45-54 year-olds. This increase is driven in part by changes in the explanatory power of skills, which rises from 5% among 25- 34 year-olds to 7% among 45-54 year-olds. This suggests that numeracy and literacy proficiency matter more for the wages of older workers than for younger ones (see Table A.4.9b (L, N) in Annex A). This increase in the salience of skills proficiency is consistent with the concept of “employer learning”, noted above (OECD, 2014[32]). Even where workers change employers, as their careers progress their formal education recedes further into the past, whereas the concrete skills they develop and demonstrate regularly during their work will remain relevant.\n\nThere are small differences in the patterns of explained wage variation by gender (Figure 4.7). On average, years of education explain a larger share of variation in wages for women than for men (Panel A), whereas numeracy and literacy proficiency accounts for a greater share of variation for men than for women (Panel B). This is consistent with research that finds that women’s wages are more closely linked to formal\n\nProficiency Education Field of study ob tenure Indi idual characteristics\n\n0\n\n3\n\n30\n\n2\n\n20\n\n1\n\n10\n\n0\n\nGermany\n\nFrance\n\nEngland (U\n\nSwit erland\n\nat ia\n\nCanada\n\nC echia\n\nSingapore\n\nEstonia\n\netherlands\n\nChile\n\nHungary\n\nAustria\n\napan\n\new ealand\n\nenmark\n\n**EC average**\n\norway\n\nSweden\n\nPortugal\n\nFinland\n\nCroatia\n\nithuania\n\nFlemish Region ( E\n\nUnited States\n\nIreland\n\nItaly\n\nSpain\n\nIsrael\n\nPoland\n\nSlo ak Republic\n\norea","pages":[{"line_numbers":[],"markdown":" **159**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\nFrance and Germany. In Korea and the Slovak Republic, conversely, skills account for less than 1% of overall wage variation.\n\n### **Figure 4.6. Contribution of observable characteristics to variation in wages**\n\nVariation explained by each factor\n\nNote: Employed adults aged 25-65 not in formal education; does not include adults who were only administered the doorstep interview due to a language barrier (see Box 1.1 in Chapter 1 and Box 4.1). Decomposition calculates the percentage contribution to variation in log wages of the following factors: skills proficiency (numeracy and literacy), years of education, job tenure, field of study (nine categories) and individual characteristics (age, gender, immigrant background, and whether one lives with a partner or has children); wages are (log) gross hourly earnings for employed and self-employed individuals, including bonuses, in PPP-adjusted 2022 USD. See Fields (2003[43]) for further details on the decomposition method. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the ote for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the percentage of variation in wages explained by skills proficiency._ Source: Table A.4.9a (L, N) in Annex A.\n\nObservable individual characteristics do not typically account for a large proportion of variation in wages, at just 4% on average across participating OECD countries and economies. However, the share is 6% in Finland and is at least 8% in Estonia, Japan and Latvia. Japan is also an outlier in the explanatory power of years of job tenure, which accounts for 9% of its wage variation, compared to the OECD average of 2%. For 11 countries the share attributable to job tenure is less than 0.5% (Figure 4.6).\n\nThe explanatory power of these observable characteristics differs by age. The total share of wage variation explained increases from 22% on average among 25-34 year-olds to 27% for 45-54 year-olds. This increase is driven in part by changes in the explanatory power of skills, which rises from 5% among 25- 34 year-olds to 7% among 45-54 year-olds. This suggests that numeracy and literacy proficiency matter more for the wages of older workers than for younger ones (see Table A.4.9b (L, N) in Annex A). This increase in the salience of skills proficiency is consistent with the concept of “employer learning”, noted above (OECD, 2014[32]). Even where workers change employers, as their careers progress their formal education recedes further into the past, whereas the concrete skills they develop and demonstrate regularly during their work will remain relevant.\n\nThere are small differences in the patterns of explained wage variation by gender (Figure 4.7). On average, years of education explain a larger share of variation in wages for women than for men (Panel A), whereas numeracy and literacy proficiency accounts for a greater share of variation for men than for women (Panel B). This is consistent with research that finds that women’s wages are more closely linked to formal\n\nProficiency Education Field of study ob tenure Indi idual characteristics\n\n0\n\n3\n\n30\n\n2\n\n20\n\n1\n\n10\n\n0\n\nGermany\n\nFrance\n\nEngland (U\n\nSwit erland\n\nat ia\n\nCanada\n\nC echia\n\nSingapore\n\nEstonia\n\netherlands\n\nChile\n\nHungary\n\nAustria\n\napan\n\new ealand\n\nenmark\n\n**EC average**\n\norway\n\nSweden\n\nPortugal\n\nFinland\n\nCroatia\n\nithuania\n\nFlemish Region ( E\n\nUnited States\n\nIreland\n\nItaly\n\nSpain\n\nIsrael\n\nPoland\n\nSlo ak Republic\n\norea","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":" **159**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\nFrance and Germany. In Korea and the Slovak Republic, conversely, skills account for less than 1% of overall wage variation.\n\n### **Figure 4.6. Contribution of observable characteristics to variation in wages**\n\nVariation explained by each factor\n\nNote: Employed adults aged 25-65 not in formal education; does not include adults who were only administered the doorstep interview due to a language barrier (see Box 1.1 in Chapter 1 and Box 4.1). Decomposition calculates the percentage contribution to variation in log wages of the following factors: skills proficiency (numeracy and literacy), years of education, job tenure, field of study (nine categories) and individual characteristics (age, gender, immigrant background, and whether one lives with a partner or has children); wages are (log) gross hourly earnings for employed and self-employed individuals, including bonuses, in PPP-adjusted 2022 USD. See Fields (2003[43]) for further details on the decomposition method. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the ote for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the percentage of variation in wages explained by skills proficiency._ Source: Table A.4.9a (L, N) in Annex A.\n\nObservable individual characteristics do not typically account for a large proportion of variation in wages, at just 4% on average across participating OECD countries and economies. However, the share is 6% in Finland and is at least 8% in Estonia, Japan and Latvia. Japan is also an outlier in the explanatory power of years of job tenure, which accounts for 9% of its wage variation, compared to the OECD average of 2%. For 11 countries the share attributable to job tenure is less than 0.5% (Figure 4.6).\n\nThe explanatory power of these observable characteristics differs by age. The total share of wage variation explained increases from 22% on average among 25-34 year-olds to 27% for 45-54 year-olds. This increase is driven in part by changes in the explanatory power of skills, which rises from 5% among 25- 34 year-olds to 7% among 45-54 year-olds. This suggests that numeracy and literacy proficiency matter more for the wages of older workers than for younger ones (see Table A.4.9b (L, N) in Annex A). This increase in the salience of skills proficiency is consistent with the concept of “employer learning”, noted above (OECD, 2014[32]). Even where workers change employers, as their careers progress their formal education recedes further into the past, whereas the concrete skills they develop and demonstrate regularly during their work will remain relevant.\n\nThere are small differences in the patterns of explained wage variation by gender (Figure 4.7). On average, years of education explain a larger share of variation in wages for women than for men (Panel A), whereas numeracy and literacy proficiency accounts for a greater share of variation for men than for women (Panel B). This is consistent with research that finds that women’s wages are more closely linked to formal\n\nProficiency Education Field of study ob tenure Indi idual characteristics\n\n0\n\n3\n\n30\n\n2\n\n20\n\n1\n\n10\n\n0\n\nGermany\n\nFrance\n\nEngland (U\n\nSwit erland\n\nat ia\n\nCanada\n\nC echia\n\nSingapore\n\nEstonia\n\netherlands\n\nChile\n\nHungary\n\nAustria\n\napan\n\new ealand\n\nenmark\n\n**EC average**\n\norway\n\nSweden\n\nPortugal\n\nFinland\n\nCroatia\n\nithuania\n\nFlemish Region ( E\n\nUnited States\n\nIreland\n\nItaly\n\nSpain\n\nIsrael\n\nPoland\n\nSlo ak Republic\n\norea","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p161.pdf"]},"response":{"errors":[],"extract_id":"extract_37d78cafe10bfd76471a6794609a1126","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":[" **159**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\nFrance and Germany. In Korea and the Slovak Republic, conversely, skills account for less than 1% of overall wage variation.\n\n### **Figure 4.6. Contribution of observable characteristics to variation in wages**\n\nVariation explained by each factor\n\nNote: Employed adults aged 25-65 not in formal education; does not include adults who were only administered the doorstep interview due to a language barrier (see Box 1.1 in Chapter 1 and Box 4.1). Decomposition calculates the percentage contribution to variation in log wages of the following factors: skills proficiency (numeracy and literacy), years of education, job tenure, field of study (nine categories) and individual characteristics (age, gender, immigrant background, and whether one lives with a partner or has children); wages are (log) gross hourly earnings for employed and self-employed individuals, including bonuses, in PPP-adjusted 2022 USD. See Fields (2003[43]) for further details on the decomposition method. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the ote for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the percentage of variation in wages explained by skills proficiency._ Source: Table A.4.9a (L, N) in Annex A.\n\nObservable individual characteristics do not typically account for a large proportion of variation in wages, at just 4% on average across participating OECD countries and economies. However, the share is 6% in Finland and is at least 8% in Estonia, Japan and Latvia. Japan is also an outlier in the explanatory power of years of job tenure, which accounts for 9% of its wage variation, compared to the OECD average of 2%. For 11 countries the share attributable to job tenure is less than 0.5% (Figure 4.6).\n\nThe explanatory power of these observable characteristics differs by age. The total share of wage variation explained increases from 22% on average among 25-34 year-olds to 27% for 45-54 year-olds. This increase is driven in part by changes in the explanatory power of skills, which rises from 5% among 25- 34 year-olds to 7% among 45-54 year-olds. This suggests that numeracy and literacy proficiency matter more for the wages of older workers than for younger ones (see Table A.4.9b (L, N) in Annex A). This increase in the salience of skills proficiency is consistent with the concept of “employer learning”, noted above (OECD, 2014[32]). Even where workers change employers, as their careers progress their formal education recedes further into the past, whereas the concrete skills they develop and demonstrate regularly during their work will remain relevant.\n\nThere are small differences in the patterns of explained wage variation by gender (Figure 4.7). On average, years of education explain a larger share of variation in wages for women than for men (Panel A), whereas numeracy and literacy proficiency accounts for a greater share of variation for men than for women (Panel B). This is consistent with research that finds that women’s wages are more closely linked to formal\n\nProficiency Education Field of study ob tenure Indi idual characteristics\n\n0\n\n3\n\n30\n\n2\n\n20\n\n1\n\n10\n\n0\n\nGermany\n\nFrance\n\nEngland (U\n\nSwit erland\n\nat ia\n\nCanada\n\nC echia\n\nSingapore\n\nEstonia\n\netherlands\n\nChile\n\nHungary\n\nAustria\n\napan\n\new ealand\n\nenmark\n\n**EC average**\n\norway\n\nSweden\n\nPortugal\n\nFinland\n\nCroatia\n\nithuania\n\nFlemish Region ( E\n\nUnited States\n\nIreland\n\nItaly\n\nSpain\n\nIsrael\n\nPoland\n\nSlo ak Republic\n\norea"],"full_content":" **159**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\nFrance and Germany. In Korea and the Slovak Republic, conversely, skills account for less than 1% of overall wage variation.\n\n### **Figure 4.6. Contribution of observable characteristics to variation in wages**\n\nVariation explained by each factor\n\nNote: Employed adults aged 25-65 not in formal education; does not include adults who were only administered the doorstep interview due to a language barrier (see Box 1.1 in Chapter 1 and Box 4.1). Decomposition calculates the percentage contribution to variation in log wages of the following factors: skills proficiency (numeracy and literacy), years of education, job tenure, field of study (nine categories) and individual characteristics (age, gender, immigrant background, and whether one lives with a partner or has children); wages are (log) gross hourly earnings for employed and self-employed individuals, including bonuses, in PPP-adjusted 2022 USD. See Fields (2003[43]) for further details on the decomposition method. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the ote for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the percentage of variation in wages explained by skills proficiency._ Source: Table A.4.9a (L, N) in Annex A.\n\nObservable individual characteristics do not typically account for a large proportion of variation in wages, at just 4% on average across participating OECD countries and economies. However, the share is 6% in Finland and is at least 8% in Estonia, Japan and Latvia. Japan is also an outlier in the explanatory power of years of job tenure, which accounts for 9% of its wage variation, compared to the OECD average of 2%. For 11 countries the share attributable to job tenure is less than 0.5% (Figure 4.6).\n\nThe explanatory power of these observable characteristics differs by age. The total share of wage variation explained increases from 22% on average among 25-34 year-olds to 27% for 45-54 year-olds. This increase is driven in part by changes in the explanatory power of skills, which rises from 5% among 25- 34 year-olds to 7% among 45-54 year-olds. This suggests that numeracy and literacy proficiency matter more for the wages of older workers than for younger ones (see Table A.4.9b (L, N) in Annex A). This increase in the salience of skills proficiency is consistent with the concept of “employer learning”, noted above (OECD, 2014[32]). Even where workers change employers, as their careers progress their formal education recedes further into the past, whereas the concrete skills they develop and demonstrate regularly during their work will remain relevant.\n\nThere are small differences in the patterns of explained wage variation by gender (Figure 4.7). On average, years of education explain a larger share of variation in wages for women than for men (Panel A), whereas numeracy and literacy proficiency accounts for a greater share of variation for men than for women (Panel B). This is consistent with research that finds that women’s wages are more closely linked to formal\n\nProficiency Education Field of study ob tenure Indi idual characteristics\n\n0\n\n3\n\n30\n\n2\n\n20\n\n1\n\n10\n\n0\n\nGermany\n\nFrance\n\nEngland (U\n\nSwit erland\n\nat ia\n\nCanada\n\nC echia\n\nSingapore\n\nEstonia\n\netherlands\n\nChile\n\nHungary\n\nAustria\n\napan\n\new ealand\n\nenmark\n\n**EC average**\n\norway\n\nSweden\n\nPortugal\n\nFinland\n\nCroatia\n\nithuania\n\nFlemish Region ( E\n\nUnited States\n\nIreland\n\nItaly\n\nSpain\n\nIsrael\n\nPoland\n\nSlo ak Republic\n\norea","publish_date":null,"title":"Figure 4.6. Contribution of observable characteristics to variation in wages","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p161.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p161.pdf"},"request":{"example_id":"chart/b263dc5d-en_p161","product_type":"parse","source_file_path":"b263dc5d-en_p161.pdf"},"started_at":"2026-11-04T19:46:07.409557"} +{"completed_at":"2026-11-04T19:46:10.650552","latency_in_ms":1881,"output":{"example_id":"chart/b263dc5d-en_p163","grounded_pages":[],"layout_pages":[],"markdown":" **161**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\nThe average total variation in wages accounted for by numeracy and literacy skills and education increased slightly between the two cycles of the Survey of Adult Skills, from 14% to 15%, although this relatively stable picture conceals considerable changes in many countries (Figure 4.8). Looking at the variation in wages accounted for by skill proficiency reveals an increase of 1.3 percentage points, from 4.5% to 5.8%, between cycles. This increase was most pronounced in Czechia (8 percentage points) and France (7 percentage points). On the other hand, there were marked decreases in Singapore (7 percentage points), the United States (6 percentage points) and Israel (6 percentage points). The change is particularly striking in Singapore, where proficiency in literacy and numeracy accounted for the greatest share of variation in wages of any country in the first cycle.\n\n### **Figure 4.8. Trends in the contribution of education and skills to variation in wages**\n\nVariation explained by skills and education, Cycle 1 and Cycle 2\n\nNote: Employed adults aged 25-65 not in formal education; does not include adults who were only administered the doorstep interview due to a language barrier (see Box 1.1 in Chapter 1 and Box 4.1). Decomposition calculates the percentage contribution to variance in log wages of the following factors: skills proficiency (numeracy and literacy), years of education, job tenure, field of study (9 categories) and individual characteristics (age, gender, immigrant background, and whether one lives with a partner or has children); wages are (log) gross hourly earnings for employed and self-employed individuals, including bonuses, in PPP-adjusted 2022 USD. See Fields (2003[43]) for further details on the decomposition method. Cycle 1 data refer to 2012 for all countries except for Chile, Israel, Lithuania, New Zealand, Singapore (all 2015) and Hungary (2017). All Cycle 2 data refer to 2023. Only countries and economies that participated in both cycles of the Survey of Adult Skills are presented. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the Note for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the share of the variation in wages explained by skills and education in Cycle 2._ Source: OECD (2018[35]; 2015[36]; 2012[37]), Survey of Adult Skills (PIAAC) databases, http://www.oecd.org/skills/piaac/publicdataandanalysis/ (accessed on 23 September 2024); Tables A.4.9a (L, N) in Annex A.\n\nAs this analysis shows, skills and education continue to be rewarded in the labour market, particularly in terms of wages, which have a much clearer relationship with numeracy and literacy skills than with employment. As discussed above, one explanation for the smaller relationship with employment is the comparatively tighter labour markets that prevailed during the second cycle (2022-23) compared to the first cycle (2011-17) of the survey; high demand for labour may have led to high employment rates regardless of skill proficiency. Although much of the variation in wages remains unexplained, information- processing skills and, particularly, education do explain a substantial proportion. It is perhaps unsurprising that education remains more important for wages, as it captures a much wider range of the knowledge, skills and attitudes that are rewarded in the labour market (Durlak et al., 2011[17]; Heckman and Kautz,\n\nears of education Cycle 1 Skills Cycle 1 ears of education Cycle 2 Skills Cycle 2\n\n0\n\n3\n\n30\n\n2\n\n20\n\n1\n\n10\n\n0\n\nSingapore\n\nGermany\n\nFinland\n\nHungary\n\nEngland (U\n\netherlands\n\nChile\n\nCanada\n\nC echia\n\nFrance\n\new ealand\n\nSweden\n\n**EC average**\n\nSpain\n\nUnited States\n\nenmark\n\nItaly\n\nEstonia\n\nAustria\n\norway\n\nFlemish Region ( E\n\napan\n\nithuania\n\nIreland\n\nPoland\n\nSlo ak Republic\n\nIsrael\n\norea","pages":[{"line_numbers":[],"markdown":" **161**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\nThe average total variation in wages accounted for by numeracy and literacy skills and education increased slightly between the two cycles of the Survey of Adult Skills, from 14% to 15%, although this relatively stable picture conceals considerable changes in many countries (Figure 4.8). Looking at the variation in wages accounted for by skill proficiency reveals an increase of 1.3 percentage points, from 4.5% to 5.8%, between cycles. This increase was most pronounced in Czechia (8 percentage points) and France (7 percentage points). On the other hand, there were marked decreases in Singapore (7 percentage points), the United States (6 percentage points) and Israel (6 percentage points). The change is particularly striking in Singapore, where proficiency in literacy and numeracy accounted for the greatest share of variation in wages of any country in the first cycle.\n\n### **Figure 4.8. Trends in the contribution of education and skills to variation in wages**\n\nVariation explained by skills and education, Cycle 1 and Cycle 2\n\nNote: Employed adults aged 25-65 not in formal education; does not include adults who were only administered the doorstep interview due to a language barrier (see Box 1.1 in Chapter 1 and Box 4.1). Decomposition calculates the percentage contribution to variance in log wages of the following factors: skills proficiency (numeracy and literacy), years of education, job tenure, field of study (9 categories) and individual characteristics (age, gender, immigrant background, and whether one lives with a partner or has children); wages are (log) gross hourly earnings for employed and self-employed individuals, including bonuses, in PPP-adjusted 2022 USD. See Fields (2003[43]) for further details on the decomposition method. Cycle 1 data refer to 2012 for all countries except for Chile, Israel, Lithuania, New Zealand, Singapore (all 2015) and Hungary (2017). All Cycle 2 data refer to 2023. Only countries and economies that participated in both cycles of the Survey of Adult Skills are presented. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the Note for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the share of the variation in wages explained by skills and education in Cycle 2._ Source: OECD (2018[35]; 2015[36]; 2012[37]), Survey of Adult Skills (PIAAC) databases, http://www.oecd.org/skills/piaac/publicdataandanalysis/ (accessed on 23 September 2024); Tables A.4.9a (L, N) in Annex A.\n\nAs this analysis shows, skills and education continue to be rewarded in the labour market, particularly in terms of wages, which have a much clearer relationship with numeracy and literacy skills than with employment. As discussed above, one explanation for the smaller relationship with employment is the comparatively tighter labour markets that prevailed during the second cycle (2022-23) compared to the first cycle (2011-17) of the survey; high demand for labour may have led to high employment rates regardless of skill proficiency. Although much of the variation in wages remains unexplained, information- processing skills and, particularly, education do explain a substantial proportion. It is perhaps unsurprising that education remains more important for wages, as it captures a much wider range of the knowledge, skills and attitudes that are rewarded in the labour market (Durlak et al., 2011[17]; Heckman and Kautz,\n\nears of education Cycle 1 Skills Cycle 1 ears of education Cycle 2 Skills Cycle 2\n\n0\n\n3\n\n30\n\n2\n\n20\n\n1\n\n10\n\n0\n\nSingapore\n\nGermany\n\nFinland\n\nHungary\n\nEngland (U\n\netherlands\n\nChile\n\nCanada\n\nC echia\n\nFrance\n\new ealand\n\nSweden\n\n**EC average**\n\nSpain\n\nUnited States\n\nenmark\n\nItaly\n\nEstonia\n\nAustria\n\norway\n\nFlemish Region ( E\n\napan\n\nithuania\n\nIreland\n\nPoland\n\nSlo ak Republic\n\nIsrael\n\norea","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":" **161**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\nThe average total variation in wages accounted for by numeracy and literacy skills and education increased slightly between the two cycles of the Survey of Adult Skills, from 14% to 15%, although this relatively stable picture conceals considerable changes in many countries (Figure 4.8). Looking at the variation in wages accounted for by skill proficiency reveals an increase of 1.3 percentage points, from 4.5% to 5.8%, between cycles. This increase was most pronounced in Czechia (8 percentage points) and France (7 percentage points). On the other hand, there were marked decreases in Singapore (7 percentage points), the United States (6 percentage points) and Israel (6 percentage points). The change is particularly striking in Singapore, where proficiency in literacy and numeracy accounted for the greatest share of variation in wages of any country in the first cycle.\n\n### **Figure 4.8. Trends in the contribution of education and skills to variation in wages**\n\nVariation explained by skills and education, Cycle 1 and Cycle 2\n\nNote: Employed adults aged 25-65 not in formal education; does not include adults who were only administered the doorstep interview due to a language barrier (see Box 1.1 in Chapter 1 and Box 4.1). Decomposition calculates the percentage contribution to variance in log wages of the following factors: skills proficiency (numeracy and literacy), years of education, job tenure, field of study (9 categories) and individual characteristics (age, gender, immigrant background, and whether one lives with a partner or has children); wages are (log) gross hourly earnings for employed and self-employed individuals, including bonuses, in PPP-adjusted 2022 USD. See Fields (2003[43]) for further details on the decomposition method. Cycle 1 data refer to 2012 for all countries except for Chile, Israel, Lithuania, New Zealand, Singapore (all 2015) and Hungary (2017). All Cycle 2 data refer to 2023. Only countries and economies that participated in both cycles of the Survey of Adult Skills are presented. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the Note for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the share of the variation in wages explained by skills and education in Cycle 2._ Source: OECD (2018[35]; 2015[36]; 2012[37]), Survey of Adult Skills (PIAAC) databases, http://www.oecd.org/skills/piaac/publicdataandanalysis/ (accessed on 23 September 2024); Tables A.4.9a (L, N) in Annex A.\n\nAs this analysis shows, skills and education continue to be rewarded in the labour market, particularly in terms of wages, which have a much clearer relationship with numeracy and literacy skills than with employment. As discussed above, one explanation for the smaller relationship with employment is the comparatively tighter labour markets that prevailed during the second cycle (2022-23) compared to the first cycle (2011-17) of the survey; high demand for labour may have led to high employment rates regardless of skill proficiency. Although much of the variation in wages remains unexplained, information- processing skills and, particularly, education do explain a substantial proportion. It is perhaps unsurprising that education remains more important for wages, as it captures a much wider range of the knowledge, skills and attitudes that are rewarded in the labour market (Durlak et al., 2011[17]; Heckman and Kautz,\n\nears of education Cycle 1 Skills Cycle 1 ears of education Cycle 2 Skills Cycle 2\n\n0\n\n3\n\n30\n\n2\n\n20\n\n1\n\n10\n\n0\n\nSingapore\n\nGermany\n\nFinland\n\nHungary\n\nEngland (U\n\netherlands\n\nChile\n\nCanada\n\nC echia\n\nFrance\n\new ealand\n\nSweden\n\n**EC average**\n\nSpain\n\nUnited States\n\nenmark\n\nItaly\n\nEstonia\n\nAustria\n\norway\n\nFlemish Region ( E\n\napan\n\nithuania\n\nIreland\n\nPoland\n\nSlo ak Republic\n\nIsrael\n\norea","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p163.pdf"]},"response":{"errors":[],"extract_id":"extract_0442202f145a0ed3bd6afa811013ddd0","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":[" **161**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\nThe average total variation in wages accounted for by numeracy and literacy skills and education increased slightly between the two cycles of the Survey of Adult Skills, from 14% to 15%, although this relatively stable picture conceals considerable changes in many countries (Figure 4.8). Looking at the variation in wages accounted for by skill proficiency reveals an increase of 1.3 percentage points, from 4.5% to 5.8%, between cycles. This increase was most pronounced in Czechia (8 percentage points) and France (7 percentage points). On the other hand, there were marked decreases in Singapore (7 percentage points), the United States (6 percentage points) and Israel (6 percentage points). The change is particularly striking in Singapore, where proficiency in literacy and numeracy accounted for the greatest share of variation in wages of any country in the first cycle.\n\n### **Figure 4.8. Trends in the contribution of education and skills to variation in wages**\n\nVariation explained by skills and education, Cycle 1 and Cycle 2\n\nNote: Employed adults aged 25-65 not in formal education; does not include adults who were only administered the doorstep interview due to a language barrier (see Box 1.1 in Chapter 1 and Box 4.1). Decomposition calculates the percentage contribution to variance in log wages of the following factors: skills proficiency (numeracy and literacy), years of education, job tenure, field of study (9 categories) and individual characteristics (age, gender, immigrant background, and whether one lives with a partner or has children); wages are (log) gross hourly earnings for employed and self-employed individuals, including bonuses, in PPP-adjusted 2022 USD. See Fields (2003[43]) for further details on the decomposition method. Cycle 1 data refer to 2012 for all countries except for Chile, Israel, Lithuania, New Zealand, Singapore (all 2015) and Hungary (2017). All Cycle 2 data refer to 2023. Only countries and economies that participated in both cycles of the Survey of Adult Skills are presented. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the Note for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the share of the variation in wages explained by skills and education in Cycle 2._ Source: OECD (2018[35]; 2015[36]; 2012[37]), Survey of Adult Skills (PIAAC) databases, http://www.oecd.org/skills/piaac/publicdataandanalysis/ (accessed on 23 September 2024); Tables A.4.9a (L, N) in Annex A.\n\nAs this analysis shows, skills and education continue to be rewarded in the labour market, particularly in terms of wages, which have a much clearer relationship with numeracy and literacy skills than with employment. As discussed above, one explanation for the smaller relationship with employment is the comparatively tighter labour markets that prevailed during the second cycle (2022-23) compared to the first cycle (2011-17) of the survey; high demand for labour may have led to high employment rates regardless of skill proficiency. Although much of the variation in wages remains unexplained, information- processing skills and, particularly, education do explain a substantial proportion. It is perhaps unsurprising that education remains more important for wages, as it captures a much wider range of the knowledge, skills and attitudes that are rewarded in the labour market (Durlak et al., 2011[17]; Heckman and Kautz,\n\nears of education Cycle 1 Skills Cycle 1 ears of education Cycle 2 Skills Cycle 2\n\n0\n\n3\n\n30\n\n2\n\n20\n\n1\n\n10\n\n0\n\nSingapore\n\nGermany\n\nFinland\n\nHungary\n\nEngland (U\n\netherlands\n\nChile\n\nCanada\n\nC echia\n\nFrance\n\new ealand\n\nSweden\n\n**EC average**\n\nSpain\n\nUnited States\n\nenmark\n\nItaly\n\nEstonia\n\nAustria\n\norway\n\nFlemish Region ( E\n\napan\n\nithuania\n\nIreland\n\nPoland\n\nSlo ak Republic\n\nIsrael\n\norea"],"full_content":" **161**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\nThe average total variation in wages accounted for by numeracy and literacy skills and education increased slightly between the two cycles of the Survey of Adult Skills, from 14% to 15%, although this relatively stable picture conceals considerable changes in many countries (Figure 4.8). Looking at the variation in wages accounted for by skill proficiency reveals an increase of 1.3 percentage points, from 4.5% to 5.8%, between cycles. This increase was most pronounced in Czechia (8 percentage points) and France (7 percentage points). On the other hand, there were marked decreases in Singapore (7 percentage points), the United States (6 percentage points) and Israel (6 percentage points). The change is particularly striking in Singapore, where proficiency in literacy and numeracy accounted for the greatest share of variation in wages of any country in the first cycle.\n\n### **Figure 4.8. Trends in the contribution of education and skills to variation in wages**\n\nVariation explained by skills and education, Cycle 1 and Cycle 2\n\nNote: Employed adults aged 25-65 not in formal education; does not include adults who were only administered the doorstep interview due to a language barrier (see Box 1.1 in Chapter 1 and Box 4.1). Decomposition calculates the percentage contribution to variance in log wages of the following factors: skills proficiency (numeracy and literacy), years of education, job tenure, field of study (9 categories) and individual characteristics (age, gender, immigrant background, and whether one lives with a partner or has children); wages are (log) gross hourly earnings for employed and self-employed individuals, including bonuses, in PPP-adjusted 2022 USD. See Fields (2003[43]) for further details on the decomposition method. Cycle 1 data refer to 2012 for all countries except for Chile, Israel, Lithuania, New Zealand, Singapore (all 2015) and Hungary (2017). All Cycle 2 data refer to 2023. Only countries and economies that participated in both cycles of the Survey of Adult Skills are presented. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the Note for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the share of the variation in wages explained by skills and education in Cycle 2._ Source: OECD (2018[35]; 2015[36]; 2012[37]), Survey of Adult Skills (PIAAC) databases, http://www.oecd.org/skills/piaac/publicdataandanalysis/ (accessed on 23 September 2024); Tables A.4.9a (L, N) in Annex A.\n\nAs this analysis shows, skills and education continue to be rewarded in the labour market, particularly in terms of wages, which have a much clearer relationship with numeracy and literacy skills than with employment. As discussed above, one explanation for the smaller relationship with employment is the comparatively tighter labour markets that prevailed during the second cycle (2022-23) compared to the first cycle (2011-17) of the survey; high demand for labour may have led to high employment rates regardless of skill proficiency. Although much of the variation in wages remains unexplained, information- processing skills and, particularly, education do explain a substantial proportion. It is perhaps unsurprising that education remains more important for wages, as it captures a much wider range of the knowledge, skills and attitudes that are rewarded in the labour market (Durlak et al., 2011[17]; Heckman and Kautz,\n\nears of education Cycle 1 Skills Cycle 1 ears of education Cycle 2 Skills Cycle 2\n\n0\n\n3\n\n30\n\n2\n\n20\n\n1\n\n10\n\n0\n\nSingapore\n\nGermany\n\nFinland\n\nHungary\n\nEngland (U\n\netherlands\n\nChile\n\nCanada\n\nC echia\n\nFrance\n\new ealand\n\nSweden\n\n**EC average**\n\nSpain\n\nUnited States\n\nenmark\n\nItaly\n\nEstonia\n\nAustria\n\norway\n\nFlemish Region ( E\n\napan\n\nithuania\n\nIreland\n\nPoland\n\nSlo ak Republic\n\nIsrael\n\norea","publish_date":null,"title":"Figure 4.8. Trends in the contribution of education and skills to variation in wages","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p163.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p163.pdf"},"request":{"example_id":"chart/b263dc5d-en_p163","product_type":"parse","source_file_path":"b263dc5d-en_p163.pdf"},"started_at":"2026-11-04T19:46:08.769226"} +{"completed_at":"2026-11-04T19:46:13.300947","latency_in_ms":2531,"output":{"example_id":"chart/b263dc5d-en_p167","grounded_pages":[],"layout_pages":[],"markdown":" **165**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\nThis positive relationship between life satisfaction, health and numeracy holds even after accounting for a number of personal characteristics. Figure 4.10 shows the percentage point difference in reporting high life satisfaction (Panel A) and positive health outcomes (Panel B) between adults with low and high numeracy skills. The regression model underlying these estimates controls for age, gender, years of education, immigrant background, parental educational attainment and whether an individual lives with a partner or has children. As the sample includes people who are not employed, no occupational controls are included.\n\n### **Figure 4.10. Relationship between numeracy and individual well-being**\n\nDifference in likelihood of reporting positive outcomes (high proficiency _minus_ low proficiency)\n\nNote: Adults aged 25-65; does not include adults who were only administered the doorstep interview due to a language barrier (see Box 1.1 in Chapter 1 and Box 4.1). Positive outcomes for life satisfaction and for self-reported health are defined in Box 4.2. High proficiency refers to numeracy proficiency at Level 4 and above and low proficiency to at or below Level 1. In addition to proficiency level, estimates account for years of education, age, gender, immigrant background, parental education and whether one lives with a partner or has children. Darker colours denote differences that are statistically significant at the 5% level. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the ote for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the difference in the likelihood of reporting positive outcomes._ Source: Table A.4.11 (N) in Annex A.\n\nOn average, across OECD countries and economies, individuals with high numeracy skills are 11 percentage points more likely to report positive health outcomes – and 10 percentage points more likely to report high life satisfaction – than individuals with low numeracy skills. Nevertheless, some cases stand out: in Croatia, for instance, individuals with high numeracy skills are 24 percentage points more likely to report high life satisfaction, a larger effect than in any other country, while the greatest gap in health occurs\n\nCroatia\n\napan\n\nPortugal\n\nItaly\n\nEstonia\n\nGermany\n\nCanada\n\norea\n\nSlo ak Republic\n\nC echia\n\nEngland (U\n\nFrance\n\nPoland\n\nat ia\n\norway\n\nSweden\n\n**EC average**\n\nithuania\n\nSwit erland\n\nSingapore\n\nFlemish Region ( E\n\nAustria\n\nUnited States\n\nIreland\n\nChile\n\netherlands\n\nenmark\n\nFinland\n\new ealand\n\nSpain\n\nHungary\n\nIsrael\n\nItaly\n\nCroatia\n\nCanada\n\nFlemish Region ( E\n\nUnited States\n\nEngland (U\n\new ealand\n\nAustria\n\nC echia\n\nGermany\n\nFrance\n\nChile\n\nenmark\n\norway\n\nEstonia\n\netherlands\n\nSingapore\n\nSwit erland\n\n**EC average**\n\nSlo ak Republic\n\nSweden\n\nHungary\n\nIreland\n\napan\n\nat ia\n\nPortugal\n\nIsrael\n\nithuania\n\nSpain\n\nFinland\n\nPoland\n\norea\n\n30\n\n2\n\n20\n\n1\n\n10\n\n0\n\nPercentage points **. ife satisfaction**\n\n30\n\n2\n\n20\n\n1\n\n10\n\n0\n\nPercentage points **B. ealt**","pages":[{"line_numbers":[],"markdown":" **165**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\nThis positive relationship between life satisfaction, health and numeracy holds even after accounting for a number of personal characteristics. Figure 4.10 shows the percentage point difference in reporting high life satisfaction (Panel A) and positive health outcomes (Panel B) between adults with low and high numeracy skills. The regression model underlying these estimates controls for age, gender, years of education, immigrant background, parental educational attainment and whether an individual lives with a partner or has children. As the sample includes people who are not employed, no occupational controls are included.\n\n### **Figure 4.10. Relationship between numeracy and individual well-being**\n\nDifference in likelihood of reporting positive outcomes (high proficiency _minus_ low proficiency)\n\nNote: Adults aged 25-65; does not include adults who were only administered the doorstep interview due to a language barrier (see Box 1.1 in Chapter 1 and Box 4.1). Positive outcomes for life satisfaction and for self-reported health are defined in Box 4.2. High proficiency refers to numeracy proficiency at Level 4 and above and low proficiency to at or below Level 1. In addition to proficiency level, estimates account for years of education, age, gender, immigrant background, parental education and whether one lives with a partner or has children. Darker colours denote differences that are statistically significant at the 5% level. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the ote for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the difference in the likelihood of reporting positive outcomes._ Source: Table A.4.11 (N) in Annex A.\n\nOn average, across OECD countries and economies, individuals with high numeracy skills are 11 percentage points more likely to report positive health outcomes – and 10 percentage points more likely to report high life satisfaction – than individuals with low numeracy skills. Nevertheless, some cases stand out: in Croatia, for instance, individuals with high numeracy skills are 24 percentage points more likely to report high life satisfaction, a larger effect than in any other country, while the greatest gap in health occurs\n\nCroatia\n\napan\n\nPortugal\n\nItaly\n\nEstonia\n\nGermany\n\nCanada\n\norea\n\nSlo ak Republic\n\nC echia\n\nEngland (U\n\nFrance\n\nPoland\n\nat ia\n\norway\n\nSweden\n\n**EC average**\n\nithuania\n\nSwit erland\n\nSingapore\n\nFlemish Region ( E\n\nAustria\n\nUnited States\n\nIreland\n\nChile\n\netherlands\n\nenmark\n\nFinland\n\new ealand\n\nSpain\n\nHungary\n\nIsrael\n\nItaly\n\nCroatia\n\nCanada\n\nFlemish Region ( E\n\nUnited States\n\nEngland (U\n\new ealand\n\nAustria\n\nC echia\n\nGermany\n\nFrance\n\nChile\n\nenmark\n\norway\n\nEstonia\n\netherlands\n\nSingapore\n\nSwit erland\n\n**EC average**\n\nSlo ak Republic\n\nSweden\n\nHungary\n\nIreland\n\napan\n\nat ia\n\nPortugal\n\nIsrael\n\nithuania\n\nSpain\n\nFinland\n\nPoland\n\norea\n\n30\n\n2\n\n20\n\n1\n\n10\n\n0\n\nPercentage points **. ife satisfaction**\n\n30\n\n2\n\n20\n\n1\n\n10\n\n0\n\nPercentage points **B. ealt**","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":" **165**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\nThis positive relationship between life satisfaction, health and numeracy holds even after accounting for a number of personal characteristics. Figure 4.10 shows the percentage point difference in reporting high life satisfaction (Panel A) and positive health outcomes (Panel B) between adults with low and high numeracy skills. The regression model underlying these estimates controls for age, gender, years of education, immigrant background, parental educational attainment and whether an individual lives with a partner or has children. As the sample includes people who are not employed, no occupational controls are included.\n\n### **Figure 4.10. Relationship between numeracy and individual well-being**\n\nDifference in likelihood of reporting positive outcomes (high proficiency _minus_ low proficiency)\n\nNote: Adults aged 25-65; does not include adults who were only administered the doorstep interview due to a language barrier (see Box 1.1 in Chapter 1 and Box 4.1). Positive outcomes for life satisfaction and for self-reported health are defined in Box 4.2. High proficiency refers to numeracy proficiency at Level 4 and above and low proficiency to at or below Level 1. In addition to proficiency level, estimates account for years of education, age, gender, immigrant background, parental education and whether one lives with a partner or has children. Darker colours denote differences that are statistically significant at the 5% level. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the ote for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the difference in the likelihood of reporting positive outcomes._ Source: Table A.4.11 (N) in Annex A.\n\nOn average, across OECD countries and economies, individuals with high numeracy skills are 11 percentage points more likely to report positive health outcomes – and 10 percentage points more likely to report high life satisfaction – than individuals with low numeracy skills. Nevertheless, some cases stand out: in Croatia, for instance, individuals with high numeracy skills are 24 percentage points more likely to report high life satisfaction, a larger effect than in any other country, while the greatest gap in health occurs\n\nCroatia\n\napan\n\nPortugal\n\nItaly\n\nEstonia\n\nGermany\n\nCanada\n\norea\n\nSlo ak Republic\n\nC echia\n\nEngland (U\n\nFrance\n\nPoland\n\nat ia\n\norway\n\nSweden\n\n**EC average**\n\nithuania\n\nSwit erland\n\nSingapore\n\nFlemish Region ( E\n\nAustria\n\nUnited States\n\nIreland\n\nChile\n\netherlands\n\nenmark\n\nFinland\n\new ealand\n\nSpain\n\nHungary\n\nIsrael\n\nItaly\n\nCroatia\n\nCanada\n\nFlemish Region ( E\n\nUnited States\n\nEngland (U\n\new ealand\n\nAustria\n\nC echia\n\nGermany\n\nFrance\n\nChile\n\nenmark\n\norway\n\nEstonia\n\netherlands\n\nSingapore\n\nSwit erland\n\n**EC average**\n\nSlo ak Republic\n\nSweden\n\nHungary\n\nIreland\n\napan\n\nat ia\n\nPortugal\n\nIsrael\n\nithuania\n\nSpain\n\nFinland\n\nPoland\n\norea\n\n30\n\n2\n\n20\n\n1\n\n10\n\n0\n\nPercentage points **. ife satisfaction**\n\n30\n\n2\n\n20\n\n1\n\n10\n\n0\n\nPercentage points **B. ealt**","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p167.pdf"]},"response":{"errors":[],"extract_id":"extract_01c1b8c879231c2e068b2421135681f3","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":[" **165**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\nThis positive relationship between life satisfaction, health and numeracy holds even after accounting for a number of personal characteristics. Figure 4.10 shows the percentage point difference in reporting high life satisfaction (Panel A) and positive health outcomes (Panel B) between adults with low and high numeracy skills. The regression model underlying these estimates controls for age, gender, years of education, immigrant background, parental educational attainment and whether an individual lives with a partner or has children. As the sample includes people who are not employed, no occupational controls are included.\n\n### **Figure 4.10. Relationship between numeracy and individual well-being**\n\nDifference in likelihood of reporting positive outcomes (high proficiency _minus_ low proficiency)\n\nNote: Adults aged 25-65; does not include adults who were only administered the doorstep interview due to a language barrier (see Box 1.1 in Chapter 1 and Box 4.1). Positive outcomes for life satisfaction and for self-reported health are defined in Box 4.2. High proficiency refers to numeracy proficiency at Level 4 and above and low proficiency to at or below Level 1. In addition to proficiency level, estimates account for years of education, age, gender, immigrant background, parental education and whether one lives with a partner or has children. Darker colours denote differences that are statistically significant at the 5% level. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the ote for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the difference in the likelihood of reporting positive outcomes._ Source: Table A.4.11 (N) in Annex A.\n\nOn average, across OECD countries and economies, individuals with high numeracy skills are 11 percentage points more likely to report positive health outcomes – and 10 percentage points more likely to report high life satisfaction – than individuals with low numeracy skills. Nevertheless, some cases stand out: in Croatia, for instance, individuals with high numeracy skills are 24 percentage points more likely to report high life satisfaction, a larger effect than in any other country, while the greatest gap in health occurs\n\nCroatia\n\napan\n\nPortugal\n\nItaly\n\nEstonia\n\nGermany\n\nCanada\n\norea\n\nSlo ak Republic\n\nC echia\n\nEngland (U\n\nFrance\n\nPoland\n\nat ia\n\norway\n\nSweden\n\n**EC average**\n\nithuania\n\nSwit erland\n\nSingapore\n\nFlemish Region ( E\n\nAustria\n\nUnited States\n\nIreland\n\nChile\n\netherlands\n\nenmark\n\nFinland\n\new ealand\n\nSpain\n\nHungary\n\nIsrael\n\nItaly\n\nCroatia\n\nCanada\n\nFlemish Region ( E\n\nUnited States\n\nEngland (U\n\new ealand\n\nAustria\n\nC echia\n\nGermany\n\nFrance\n\nChile\n\nenmark\n\norway\n\nEstonia\n\netherlands\n\nSingapore\n\nSwit erland\n\n**EC average**\n\nSlo ak Republic\n\nSweden\n\nHungary\n\nIreland\n\napan\n\nat ia\n\nPortugal\n\nIsrael\n\nithuania\n\nSpain\n\nFinland\n\nPoland\n\norea\n\n30\n\n2\n\n20\n\n1\n\n10\n\n0\n\nPercentage points **. ife satisfaction**\n\n30\n\n2\n\n20\n\n1\n\n10\n\n0\n\nPercentage points **B. ealt**"],"full_content":" **165**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\nThis positive relationship between life satisfaction, health and numeracy holds even after accounting for a number of personal characteristics. Figure 4.10 shows the percentage point difference in reporting high life satisfaction (Panel A) and positive health outcomes (Panel B) between adults with low and high numeracy skills. The regression model underlying these estimates controls for age, gender, years of education, immigrant background, parental educational attainment and whether an individual lives with a partner or has children. As the sample includes people who are not employed, no occupational controls are included.\n\n### **Figure 4.10. Relationship between numeracy and individual well-being**\n\nDifference in likelihood of reporting positive outcomes (high proficiency _minus_ low proficiency)\n\nNote: Adults aged 25-65; does not include adults who were only administered the doorstep interview due to a language barrier (see Box 1.1 in Chapter 1 and Box 4.1). Positive outcomes for life satisfaction and for self-reported health are defined in Box 4.2. High proficiency refers to numeracy proficiency at Level 4 and above and low proficiency to at or below Level 1. In addition to proficiency level, estimates account for years of education, age, gender, immigrant background, parental education and whether one lives with a partner or has children. Darker colours denote differences that are statistically significant at the 5% level. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the ote for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the difference in the likelihood of reporting positive outcomes._ Source: Table A.4.11 (N) in Annex A.\n\nOn average, across OECD countries and economies, individuals with high numeracy skills are 11 percentage points more likely to report positive health outcomes – and 10 percentage points more likely to report high life satisfaction – than individuals with low numeracy skills. Nevertheless, some cases stand out: in Croatia, for instance, individuals with high numeracy skills are 24 percentage points more likely to report high life satisfaction, a larger effect than in any other country, while the greatest gap in health occurs\n\nCroatia\n\napan\n\nPortugal\n\nItaly\n\nEstonia\n\nGermany\n\nCanada\n\norea\n\nSlo ak Republic\n\nC echia\n\nEngland (U\n\nFrance\n\nPoland\n\nat ia\n\norway\n\nSweden\n\n**EC average**\n\nithuania\n\nSwit erland\n\nSingapore\n\nFlemish Region ( E\n\nAustria\n\nUnited States\n\nIreland\n\nChile\n\netherlands\n\nenmark\n\nFinland\n\new ealand\n\nSpain\n\nHungary\n\nIsrael\n\nItaly\n\nCroatia\n\nCanada\n\nFlemish Region ( E\n\nUnited States\n\nEngland (U\n\new ealand\n\nAustria\n\nC echia\n\nGermany\n\nFrance\n\nChile\n\nenmark\n\norway\n\nEstonia\n\netherlands\n\nSingapore\n\nSwit erland\n\n**EC average**\n\nSlo ak Republic\n\nSweden\n\nHungary\n\nIreland\n\napan\n\nat ia\n\nPortugal\n\nIsrael\n\nithuania\n\nSpain\n\nFinland\n\nPoland\n\norea\n\n30\n\n2\n\n20\n\n1\n\n10\n\n0\n\nPercentage points **. ife satisfaction**\n\n30\n\n2\n\n20\n\n1\n\n10\n\n0\n\nPercentage points **B. ealt**","publish_date":null,"title":"Figure 4.10. Relationship between numeracy and individual well-being","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p167.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p167.pdf"},"request":{"example_id":"chart/b263dc5d-en_p167","product_type":"parse","source_file_path":"b263dc5d-en_p167.pdf"},"started_at":"2026-11-04T19:46:10.769589"} +{"completed_at":"2026-11-04T19:46:16.230955","latency_in_ms":3096,"output":{"example_id":"chart/b263dc5d-en_p170","grounded_pages":[],"layout_pages":[],"markdown":"**168** \n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 4.11. Civic engagement outcomes, by numeracy proficiency level**\n\nAdults aged 25-65 reporting positive outcomes for political efficacy, trust and volunteering\n\nNote: Does not include adults who were only administered the doorstep interview due to a language barrier (see Box 1.1 in Chapter 1 and Box 4.1). Figure plots the unadjusted share of respondents reporting a positive outcome (see Box 4.2 for definitions). \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the Note for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the share of respondents at Level 4 and above reporting positive outcomes._ Source: Table A.4.10 (N) in Annex A.\n\ne el 1 and below e el and abo e Total\n\nSwit erland\n\nC echia\n\napan\n\nSweden\n\nFinland\n\nChile\n\nenmark\n\nGermany\n\norea\n\norway\n\new ealand\n\nCanada\n\nithuania\n\nSingapore\n\n**EC average**\n\nIreland\n\nEstonia\n\nUnited States\n\netherlands\n\nat ia\n\nAustria\n\nEngland (U\n\nPortugal\n\nFrance\n\nFlemish Region ( E\n\nPoland\n\nIsrael\n\nSlo ak Republic\n\nItaly\n\nHungary\n\nCroatia\n\nSpain\n\nenmark\n\norway\n\nFinland\n\netherlands\n\nSweden\n\nSwit erland\n\new ealand\n\nFlemish Region ( E\n\nIreland\n\nAustria\n\nCanada\n\nUnited States\n\nGermany\n\n**EC average**\n\nEngland (U\n\norea\n\nEstonia\n\nSpain\n\nithuania\n\nSingapore\n\nIsrael\n\nItaly\n\nHungary\n\nat ia\n\nPortugal\n\nFrance\n\nC echia\n\nPoland\n\nCroatia\n\napan\n\nSlo ak Republic\n\nChile\n\nUnited States\n\norway\n\new ealand\n\nIreland\n\nIsrael\n\nFinland\n\nenmark\n\nSwit erland\n\nCanada\n\nGermany\n\nAustria\n\netherlands\n\nFlemish Region ( E\n\nHungary\n\n**EC average**\n\nFrance\n\nSlo ak Republic\n\nEngland (U\n\nSweden\n\nat ia\n\nEstonia\n\nSingapore\n\nChile\n\napan\n\nC echia\n\nPortugal\n\nPoland\n\nItaly\n\nSpain\n\nithuania\n\nCroatia\n\norea\n\n80\n\n0\n\n0\n\n0\n\n0\n\n30\n\n20\n\n10\n\n0\n\n**. olitical efficacy**\n\n100\n\n90\n\n80\n\n0\n\n0\n\n0\n\n0\n\n30\n\n20\n\n10\n\n0\n\n80\n\n0\n\n0\n\n0\n\n0\n\n30\n\n20\n\n10\n\n0\n\n**B. rust**\n\n**C. olunteer activities**","pages":[{"line_numbers":[],"markdown":"**168** \n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 4.11. Civic engagement outcomes, by numeracy proficiency level**\n\nAdults aged 25-65 reporting positive outcomes for political efficacy, trust and volunteering\n\nNote: Does not include adults who were only administered the doorstep interview due to a language barrier (see Box 1.1 in Chapter 1 and Box 4.1). Figure plots the unadjusted share of respondents reporting a positive outcome (see Box 4.2 for definitions). \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the Note for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the share of respondents at Level 4 and above reporting positive outcomes._ Source: Table A.4.10 (N) in Annex A.\n\ne el 1 and below e el and abo e Total\n\nSwit erland\n\nC echia\n\napan\n\nSweden\n\nFinland\n\nChile\n\nenmark\n\nGermany\n\norea\n\norway\n\new ealand\n\nCanada\n\nithuania\n\nSingapore\n\n**EC average**\n\nIreland\n\nEstonia\n\nUnited States\n\netherlands\n\nat ia\n\nAustria\n\nEngland (U\n\nPortugal\n\nFrance\n\nFlemish Region ( E\n\nPoland\n\nIsrael\n\nSlo ak Republic\n\nItaly\n\nHungary\n\nCroatia\n\nSpain\n\nenmark\n\norway\n\nFinland\n\netherlands\n\nSweden\n\nSwit erland\n\new ealand\n\nFlemish Region ( E\n\nIreland\n\nAustria\n\nCanada\n\nUnited States\n\nGermany\n\n**EC average**\n\nEngland (U\n\norea\n\nEstonia\n\nSpain\n\nithuania\n\nSingapore\n\nIsrael\n\nItaly\n\nHungary\n\nat ia\n\nPortugal\n\nFrance\n\nC echia\n\nPoland\n\nCroatia\n\napan\n\nSlo ak Republic\n\nChile\n\nUnited States\n\norway\n\new ealand\n\nIreland\n\nIsrael\n\nFinland\n\nenmark\n\nSwit erland\n\nCanada\n\nGermany\n\nAustria\n\netherlands\n\nFlemish Region ( E\n\nHungary\n\n**EC average**\n\nFrance\n\nSlo ak Republic\n\nEngland (U\n\nSweden\n\nat ia\n\nEstonia\n\nSingapore\n\nChile\n\napan\n\nC echia\n\nPortugal\n\nPoland\n\nItaly\n\nSpain\n\nithuania\n\nCroatia\n\norea\n\n80\n\n0\n\n0\n\n0\n\n0\n\n30\n\n20\n\n10\n\n0\n\n**. olitical efficacy**\n\n100\n\n90\n\n80\n\n0\n\n0\n\n0\n\n0\n\n30\n\n20\n\n10\n\n0\n\n80\n\n0\n\n0\n\n0\n\n0\n\n30\n\n20\n\n10\n\n0\n\n**B. rust**\n\n**C. olunteer activities**","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**168** \n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 4.11. Civic engagement outcomes, by numeracy proficiency level**\n\nAdults aged 25-65 reporting positive outcomes for political efficacy, trust and volunteering\n\nNote: Does not include adults who were only administered the doorstep interview due to a language barrier (see Box 1.1 in Chapter 1 and Box 4.1). Figure plots the unadjusted share of respondents reporting a positive outcome (see Box 4.2 for definitions). \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the Note for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the share of respondents at Level 4 and above reporting positive outcomes._ Source: Table A.4.10 (N) in Annex A.\n\ne el 1 and below e el and abo e Total\n\nSwit erland\n\nC echia\n\napan\n\nSweden\n\nFinland\n\nChile\n\nenmark\n\nGermany\n\norea\n\norway\n\new ealand\n\nCanada\n\nithuania\n\nSingapore\n\n**EC average**\n\nIreland\n\nEstonia\n\nUnited States\n\netherlands\n\nat ia\n\nAustria\n\nEngland (U\n\nPortugal\n\nFrance\n\nFlemish Region ( E\n\nPoland\n\nIsrael\n\nSlo ak Republic\n\nItaly\n\nHungary\n\nCroatia\n\nSpain\n\nenmark\n\norway\n\nFinland\n\netherlands\n\nSweden\n\nSwit erland\n\new ealand\n\nFlemish Region ( E\n\nIreland\n\nAustria\n\nCanada\n\nUnited States\n\nGermany\n\n**EC average**\n\nEngland (U\n\norea\n\nEstonia\n\nSpain\n\nithuania\n\nSingapore\n\nIsrael\n\nItaly\n\nHungary\n\nat ia\n\nPortugal\n\nFrance\n\nC echia\n\nPoland\n\nCroatia\n\napan\n\nSlo ak Republic\n\nChile\n\nUnited States\n\norway\n\new ealand\n\nIreland\n\nIsrael\n\nFinland\n\nenmark\n\nSwit erland\n\nCanada\n\nGermany\n\nAustria\n\netherlands\n\nFlemish Region ( E\n\nHungary\n\n**EC average**\n\nFrance\n\nSlo ak Republic\n\nEngland (U\n\nSweden\n\nat ia\n\nEstonia\n\nSingapore\n\nChile\n\napan\n\nC echia\n\nPortugal\n\nPoland\n\nItaly\n\nSpain\n\nithuania\n\nCroatia\n\norea\n\n80\n\n0\n\n0\n\n0\n\n0\n\n30\n\n20\n\n10\n\n0\n\n**. olitical efficacy**\n\n100\n\n90\n\n80\n\n0\n\n0\n\n0\n\n0\n\n30\n\n20\n\n10\n\n0\n\n80\n\n0\n\n0\n\n0\n\n0\n\n30\n\n20\n\n10\n\n0\n\n**B. rust**\n\n**C. olunteer activities**","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p170.pdf"]},"response":{"errors":[],"extract_id":"extract_ed35f47d32557a0dc2471e305ba0d46e","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**168** \n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 4.11. Civic engagement outcomes, by numeracy proficiency level**\n\nAdults aged 25-65 reporting positive outcomes for political efficacy, trust and volunteering\n\nNote: Does not include adults who were only administered the doorstep interview due to a language barrier (see Box 1.1 in Chapter 1 and Box 4.1). Figure plots the unadjusted share of respondents reporting a positive outcome (see Box 4.2 for definitions). \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the Note for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the share of respondents at Level 4 and above reporting positive outcomes._ Source: Table A.4.10 (N) in Annex A.\n\ne el 1 and below e el and abo e Total\n\nSwit erland\n\nC echia\n\napan\n\nSweden\n\nFinland\n\nChile\n\nenmark\n\nGermany\n\norea\n\norway\n\new ealand\n\nCanada\n\nithuania\n\nSingapore\n\n**EC average**\n\nIreland\n\nEstonia\n\nUnited States\n\netherlands\n\nat ia\n\nAustria\n\nEngland (U\n\nPortugal\n\nFrance\n\nFlemish Region ( E\n\nPoland\n\nIsrael\n\nSlo ak Republic\n\nItaly\n\nHungary\n\nCroatia\n\nSpain\n\nenmark\n\norway\n\nFinland\n\netherlands\n\nSweden\n\nSwit erland\n\new ealand\n\nFlemish Region ( E\n\nIreland\n\nAustria\n\nCanada\n\nUnited States\n\nGermany\n\n**EC average**\n\nEngland (U\n\norea\n\nEstonia\n\nSpain\n\nithuania\n\nSingapore\n\nIsrael\n\nItaly\n\nHungary\n\nat ia\n\nPortugal\n\nFrance\n\nC echia\n\nPoland\n\nCroatia\n\napan\n\nSlo ak Republic\n\nChile\n\nUnited States\n\norway\n\new ealand\n\nIreland\n\nIsrael\n\nFinland\n\nenmark\n\nSwit erland\n\nCanada\n\nGermany\n\nAustria\n\netherlands\n\nFlemish Region ( E\n\nHungary\n\n**EC average**\n\nFrance\n\nSlo ak Republic\n\nEngland (U\n\nSweden\n\nat ia\n\nEstonia\n\nSingapore\n\nChile\n\napan\n\nC echia\n\nPortugal\n\nPoland\n\nItaly\n\nSpain\n\nithuania\n\nCroatia\n\norea\n\n80\n\n0\n\n0\n\n0\n\n0\n\n30\n\n20\n\n10\n\n0\n\n**. olitical efficacy**\n\n100\n\n90\n\n80\n\n0\n\n0\n\n0\n\n0\n\n30\n\n20\n\n10\n\n0\n\n80\n\n0\n\n0\n\n0\n\n0\n\n30\n\n20\n\n10\n\n0\n\n**B. rust**\n\n**C. olunteer activities**"],"full_content":"**168** \n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 4.11. Civic engagement outcomes, by numeracy proficiency level**\n\nAdults aged 25-65 reporting positive outcomes for political efficacy, trust and volunteering\n\nNote: Does not include adults who were only administered the doorstep interview due to a language barrier (see Box 1.1 in Chapter 1 and Box 4.1). Figure plots the unadjusted share of respondents reporting a positive outcome (see Box 4.2 for definitions). \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the Note for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the share of respondents at Level 4 and above reporting positive outcomes._ Source: Table A.4.10 (N) in Annex A.\n\ne el 1 and below e el and abo e Total\n\nSwit erland\n\nC echia\n\napan\n\nSweden\n\nFinland\n\nChile\n\nenmark\n\nGermany\n\norea\n\norway\n\new ealand\n\nCanada\n\nithuania\n\nSingapore\n\n**EC average**\n\nIreland\n\nEstonia\n\nUnited States\n\netherlands\n\nat ia\n\nAustria\n\nEngland (U\n\nPortugal\n\nFrance\n\nFlemish Region ( E\n\nPoland\n\nIsrael\n\nSlo ak Republic\n\nItaly\n\nHungary\n\nCroatia\n\nSpain\n\nenmark\n\norway\n\nFinland\n\netherlands\n\nSweden\n\nSwit erland\n\new ealand\n\nFlemish Region ( E\n\nIreland\n\nAustria\n\nCanada\n\nUnited States\n\nGermany\n\n**EC average**\n\nEngland (U\n\norea\n\nEstonia\n\nSpain\n\nithuania\n\nSingapore\n\nIsrael\n\nItaly\n\nHungary\n\nat ia\n\nPortugal\n\nFrance\n\nC echia\n\nPoland\n\nCroatia\n\napan\n\nSlo ak Republic\n\nChile\n\nUnited States\n\norway\n\new ealand\n\nIreland\n\nIsrael\n\nFinland\n\nenmark\n\nSwit erland\n\nCanada\n\nGermany\n\nAustria\n\netherlands\n\nFlemish Region ( E\n\nHungary\n\n**EC average**\n\nFrance\n\nSlo ak Republic\n\nEngland (U\n\nSweden\n\nat ia\n\nEstonia\n\nSingapore\n\nChile\n\napan\n\nC echia\n\nPortugal\n\nPoland\n\nItaly\n\nSpain\n\nithuania\n\nCroatia\n\norea\n\n80\n\n0\n\n0\n\n0\n\n0\n\n30\n\n20\n\n10\n\n0\n\n**. olitical efficacy**\n\n100\n\n90\n\n80\n\n0\n\n0\n\n0\n\n0\n\n30\n\n20\n\n10\n\n0\n\n80\n\n0\n\n0\n\n0\n\n0\n\n30\n\n20\n\n10\n\n0\n\n**B. rust**\n\n**C. olunteer activities**","publish_date":null,"title":"Figure 4.11. Civic engagement outcomes, by numeracy proficiency level","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p170.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p170.pdf"},"request":{"example_id":"chart/b263dc5d-en_p170","product_type":"parse","source_file_path":"b263dc5d-en_p170.pdf"},"started_at":"2026-11-04T19:46:13.134752"} +{"completed_at":"2026-11-04T19:46:16.181202","latency_in_ms":1540,"output":{"example_id":"chart/b263dc5d-en_p175","grounded_pages":[],"layout_pages":[],"markdown":" **173**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 4.13. Mismatches in qualifications, skills and field of study**\n\nEmployed adults aged 25-65 who are not self-employed\n\nNote: Does not include adults who were only administered the doorstep interview due to a language barrier (see Box 1.1 in Chapter 1 and Box 4.1). Mismatch measures are defined in Table 4.2 and Box 4.3. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the ote for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the share of workers who are over-qualified (Panel A), over-skilled (Panel B) and field of study mismatched (Panel C)._ Source: Table A.4.12 in Annex A.\n\ner qualified Under qualified\n\ner skilled Under skilled\n\nEngland (U\n\napan\n\nIsrael\n\new ealand\n\norea\n\nithuania\n\nGermany\n\nCanada\n\nAustria\n\nUnited States\n\nat ia\n\nEstonia\n\nHungary\n\nC echia\n\n**EC average**\n\nIreland\n\nSweden\n\nSpain\n\nenmark\n\nFinland\n\nChile\n\nFrance\n\nSlo ak Republic\n\netherlands\n\norway\n\nItaly\n\nCroatia\n\nSwit erland\n\nPortugal\n\nSingapore\n\nFlemish Region ( E\n\nPoland\n\n0\n\n0\n\n0\n\n30\n\n20\n\n10\n\n0\n\n**. ualification mismatc**\n\nIsrael\n\nUnited States\n\nCanada\n\new ealand\n\nEngland (U\n\nIreland\n\nGermany\n\nenmark\n\nCroatia\n\nSlo ak Republic\n\nSweden\n\nC echia\n\netherlands\n\n**EC average**\n\nHungary\n\nFrance\n\nPortugal\n\nChile\n\nAustria\n\norea\n\nSwit erland\n\nSpain\n\nFlemish Region ( E\n\nSingapore\n\nEstonia\n\norway\n\nat ia\n\nItaly\n\nFinland\n\nPoland\n\nithuania\n\napan\n\n0\n\n0\n\n0\n\n30\n\n20\n\n10\n\n0\n\n**B. ill mismatc**\n\norea\n\napan\n\new ealand\n\nSpain\n\nEstonia\n\nC echia\n\nat ia\n\nPortugal\n\nEngland (U\n\nItaly\n\nFlemish Region ( E\n\nUnited States\n\nCanada\n\nFrance\n\n**EC average**\n\nSwit erland\n\nIreland\n\nIsrael\n\nithuania\n\nSlo ak Republic\n\nSingapore\n\nChile\n\nHungary\n\nSweden\n\nPoland\n\nenmark\n\nGermany\n\norway\n\nAustria\n\nCroatia\n\netherlands\n\nFinland\n\n0\n\n0\n\n0\n\n30\n\n20\n\n10\n\n0\n\n**C. ield of study mismatc**","pages":[{"line_numbers":[],"markdown":" **173**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 4.13. Mismatches in qualifications, skills and field of study**\n\nEmployed adults aged 25-65 who are not self-employed\n\nNote: Does not include adults who were only administered the doorstep interview due to a language barrier (see Box 1.1 in Chapter 1 and Box 4.1). Mismatch measures are defined in Table 4.2 and Box 4.3. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the ote for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the share of workers who are over-qualified (Panel A), over-skilled (Panel B) and field of study mismatched (Panel C)._ Source: Table A.4.12 in Annex A.\n\ner qualified Under qualified\n\ner skilled Under skilled\n\nEngland (U\n\napan\n\nIsrael\n\new ealand\n\norea\n\nithuania\n\nGermany\n\nCanada\n\nAustria\n\nUnited States\n\nat ia\n\nEstonia\n\nHungary\n\nC echia\n\n**EC average**\n\nIreland\n\nSweden\n\nSpain\n\nenmark\n\nFinland\n\nChile\n\nFrance\n\nSlo ak Republic\n\netherlands\n\norway\n\nItaly\n\nCroatia\n\nSwit erland\n\nPortugal\n\nSingapore\n\nFlemish Region ( E\n\nPoland\n\n0\n\n0\n\n0\n\n30\n\n20\n\n10\n\n0\n\n**. ualification mismatc**\n\nIsrael\n\nUnited States\n\nCanada\n\new ealand\n\nEngland (U\n\nIreland\n\nGermany\n\nenmark\n\nCroatia\n\nSlo ak Republic\n\nSweden\n\nC echia\n\netherlands\n\n**EC average**\n\nHungary\n\nFrance\n\nPortugal\n\nChile\n\nAustria\n\norea\n\nSwit erland\n\nSpain\n\nFlemish Region ( E\n\nSingapore\n\nEstonia\n\norway\n\nat ia\n\nItaly\n\nFinland\n\nPoland\n\nithuania\n\napan\n\n0\n\n0\n\n0\n\n30\n\n20\n\n10\n\n0\n\n**B. ill mismatc**\n\norea\n\napan\n\new ealand\n\nSpain\n\nEstonia\n\nC echia\n\nat ia\n\nPortugal\n\nEngland (U\n\nItaly\n\nFlemish Region ( E\n\nUnited States\n\nCanada\n\nFrance\n\n**EC average**\n\nSwit erland\n\nIreland\n\nIsrael\n\nithuania\n\nSlo ak Republic\n\nSingapore\n\nChile\n\nHungary\n\nSweden\n\nPoland\n\nenmark\n\nGermany\n\norway\n\nAustria\n\nCroatia\n\netherlands\n\nFinland\n\n0\n\n0\n\n0\n\n30\n\n20\n\n10\n\n0\n\n**C. ield of study mismatc**","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":" **173**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 4.13. Mismatches in qualifications, skills and field of study**\n\nEmployed adults aged 25-65 who are not self-employed\n\nNote: Does not include adults who were only administered the doorstep interview due to a language barrier (see Box 1.1 in Chapter 1 and Box 4.1). Mismatch measures are defined in Table 4.2 and Box 4.3. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the ote for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the share of workers who are over-qualified (Panel A), over-skilled (Panel B) and field of study mismatched (Panel C)._ Source: Table A.4.12 in Annex A.\n\ner qualified Under qualified\n\ner skilled Under skilled\n\nEngland (U\n\napan\n\nIsrael\n\new ealand\n\norea\n\nithuania\n\nGermany\n\nCanada\n\nAustria\n\nUnited States\n\nat ia\n\nEstonia\n\nHungary\n\nC echia\n\n**EC average**\n\nIreland\n\nSweden\n\nSpain\n\nenmark\n\nFinland\n\nChile\n\nFrance\n\nSlo ak Republic\n\netherlands\n\norway\n\nItaly\n\nCroatia\n\nSwit erland\n\nPortugal\n\nSingapore\n\nFlemish Region ( E\n\nPoland\n\n0\n\n0\n\n0\n\n30\n\n20\n\n10\n\n0\n\n**. ualification mismatc**\n\nIsrael\n\nUnited States\n\nCanada\n\new ealand\n\nEngland (U\n\nIreland\n\nGermany\n\nenmark\n\nCroatia\n\nSlo ak Republic\n\nSweden\n\nC echia\n\netherlands\n\n**EC average**\n\nHungary\n\nFrance\n\nPortugal\n\nChile\n\nAustria\n\norea\n\nSwit erland\n\nSpain\n\nFlemish Region ( E\n\nSingapore\n\nEstonia\n\norway\n\nat ia\n\nItaly\n\nFinland\n\nPoland\n\nithuania\n\napan\n\n0\n\n0\n\n0\n\n30\n\n20\n\n10\n\n0\n\n**B. ill mismatc**\n\norea\n\napan\n\new ealand\n\nSpain\n\nEstonia\n\nC echia\n\nat ia\n\nPortugal\n\nEngland (U\n\nItaly\n\nFlemish Region ( E\n\nUnited States\n\nCanada\n\nFrance\n\n**EC average**\n\nSwit erland\n\nIreland\n\nIsrael\n\nithuania\n\nSlo ak Republic\n\nSingapore\n\nChile\n\nHungary\n\nSweden\n\nPoland\n\nenmark\n\nGermany\n\norway\n\nAustria\n\nCroatia\n\netherlands\n\nFinland\n\n0\n\n0\n\n0\n\n30\n\n20\n\n10\n\n0\n\n**C. ield of study mismatc**","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p175.pdf"]},"response":{"errors":[],"extract_id":"extract_fe3af52bf68de88322db8439660ee39d","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":[" **173**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 4.13. Mismatches in qualifications, skills and field of study**\n\nEmployed adults aged 25-65 who are not self-employed\n\nNote: Does not include adults who were only administered the doorstep interview due to a language barrier (see Box 1.1 in Chapter 1 and Box 4.1). Mismatch measures are defined in Table 4.2 and Box 4.3. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the ote for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the share of workers who are over-qualified (Panel A), over-skilled (Panel B) and field of study mismatched (Panel C)._ Source: Table A.4.12 in Annex A.\n\ner qualified Under qualified\n\ner skilled Under skilled\n\nEngland (U\n\napan\n\nIsrael\n\new ealand\n\norea\n\nithuania\n\nGermany\n\nCanada\n\nAustria\n\nUnited States\n\nat ia\n\nEstonia\n\nHungary\n\nC echia\n\n**EC average**\n\nIreland\n\nSweden\n\nSpain\n\nenmark\n\nFinland\n\nChile\n\nFrance\n\nSlo ak Republic\n\netherlands\n\norway\n\nItaly\n\nCroatia\n\nSwit erland\n\nPortugal\n\nSingapore\n\nFlemish Region ( E\n\nPoland\n\n0\n\n0\n\n0\n\n30\n\n20\n\n10\n\n0\n\n**. ualification mismatc**\n\nIsrael\n\nUnited States\n\nCanada\n\new ealand\n\nEngland (U\n\nIreland\n\nGermany\n\nenmark\n\nCroatia\n\nSlo ak Republic\n\nSweden\n\nC echia\n\netherlands\n\n**EC average**\n\nHungary\n\nFrance\n\nPortugal\n\nChile\n\nAustria\n\norea\n\nSwit erland\n\nSpain\n\nFlemish Region ( E\n\nSingapore\n\nEstonia\n\norway\n\nat ia\n\nItaly\n\nFinland\n\nPoland\n\nithuania\n\napan\n\n0\n\n0\n\n0\n\n30\n\n20\n\n10\n\n0\n\n**B. ill mismatc**\n\norea\n\napan\n\new ealand\n\nSpain\n\nEstonia\n\nC echia\n\nat ia\n\nPortugal\n\nEngland (U\n\nItaly\n\nFlemish Region ( E\n\nUnited States\n\nCanada\n\nFrance\n\n**EC average**\n\nSwit erland\n\nIreland\n\nIsrael\n\nithuania\n\nSlo ak Republic\n\nSingapore\n\nChile\n\nHungary\n\nSweden\n\nPoland\n\nenmark\n\nGermany\n\norway\n\nAustria\n\nCroatia\n\netherlands\n\nFinland\n\n0\n\n0\n\n0\n\n30\n\n20\n\n10\n\n0\n\n**C. ield of study mismatc**"],"full_content":" **173**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 4.13. Mismatches in qualifications, skills and field of study**\n\nEmployed adults aged 25-65 who are not self-employed\n\nNote: Does not include adults who were only administered the doorstep interview due to a language barrier (see Box 1.1 in Chapter 1 and Box 4.1). Mismatch measures are defined in Table 4.2 and Box 4.3. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the ote for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the share of workers who are over-qualified (Panel A), over-skilled (Panel B) and field of study mismatched (Panel C)._ Source: Table A.4.12 in Annex A.\n\ner qualified Under qualified\n\ner skilled Under skilled\n\nEngland (U\n\napan\n\nIsrael\n\new ealand\n\norea\n\nithuania\n\nGermany\n\nCanada\n\nAustria\n\nUnited States\n\nat ia\n\nEstonia\n\nHungary\n\nC echia\n\n**EC average**\n\nIreland\n\nSweden\n\nSpain\n\nenmark\n\nFinland\n\nChile\n\nFrance\n\nSlo ak Republic\n\netherlands\n\norway\n\nItaly\n\nCroatia\n\nSwit erland\n\nPortugal\n\nSingapore\n\nFlemish Region ( E\n\nPoland\n\n0\n\n0\n\n0\n\n30\n\n20\n\n10\n\n0\n\n**. ualification mismatc**\n\nIsrael\n\nUnited States\n\nCanada\n\new ealand\n\nEngland (U\n\nIreland\n\nGermany\n\nenmark\n\nCroatia\n\nSlo ak Republic\n\nSweden\n\nC echia\n\netherlands\n\n**EC average**\n\nHungary\n\nFrance\n\nPortugal\n\nChile\n\nAustria\n\norea\n\nSwit erland\n\nSpain\n\nFlemish Region ( E\n\nSingapore\n\nEstonia\n\norway\n\nat ia\n\nItaly\n\nFinland\n\nPoland\n\nithuania\n\napan\n\n0\n\n0\n\n0\n\n30\n\n20\n\n10\n\n0\n\n**B. ill mismatc**\n\norea\n\napan\n\new ealand\n\nSpain\n\nEstonia\n\nC echia\n\nat ia\n\nPortugal\n\nEngland (U\n\nItaly\n\nFlemish Region ( E\n\nUnited States\n\nCanada\n\nFrance\n\n**EC average**\n\nSwit erland\n\nIreland\n\nIsrael\n\nithuania\n\nSlo ak Republic\n\nSingapore\n\nChile\n\nHungary\n\nSweden\n\nPoland\n\nenmark\n\nGermany\n\norway\n\nAustria\n\nCroatia\n\netherlands\n\nFinland\n\n0\n\n0\n\n0\n\n30\n\n20\n\n10\n\n0\n\n**C. ield of study mismatc**","publish_date":null,"title":"Figure 4.13. Mismatches in qualifications, skills and field of study","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p175.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p175.pdf"},"request":{"example_id":"chart/b263dc5d-en_p175","product_type":"parse","source_file_path":"b263dc5d-en_p175.pdf"},"started_at":"2026-11-04T19:46:14.640213"} +{"completed_at":"2026-11-04T19:46:17.797627","latency_in_ms":1478,"output":{"example_id":"chart/b263dc5d-en_p177","grounded_pages":[],"layout_pages":[],"markdown":" **175**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Box 4.4. Results on skill gaps from the PIAAC Employer Module**\n\nThe PIAAC Employer Module is a unique survey that captures employers’ perspectives on skill gaps. For employers, skill gaps are mismatches between the skills available in a workforce and those required to meet their current and future business needs. They can be understood as the employer perspective on under-skilling, which denotes a situation where workers believe they do not have the necessary skills for their current job (Marcolin and Quintini, 2023[55]). While skill gaps and under-skilling are closely related, they are not identical.\n\nThe 2023 Survey of Adult Skills and the PIAAC Employer Module are conceptually linked, as they were jointly designed to compare employee and employer perspectives on skill gaps. Data for both surveys were collected simultaneously. The first wave of the PIAAC Employer Module took place in five OECD countries – Hungary, Italy, the Netherlands, Portugal and the Slovak Republic.\n\nAccording to the PIAAC Employer Module, a sizeable share of firms report skill gaps in their organisation. Over half of enterprises in the Slovak Republic state that their workforce has some degree of skill gap (54%), the highest of all countries (Figure 4.15). This is followed by Italy (37%), Portugal (32%), the Netherlands (31%) and Hungary (27%). Where enterprises identify skill gaps, these most often concern only a few or some of their employees. Much smaller proportions of enterprises (<5%) report that most or all their employees do not have the skills needed to perform their jobs.\n\nA notable proportion of enterprises report that they did not know whether they had a skill gap, with high shares in the Netherlands (18%) and Hungary (15%). The uncertainty expressed by many firms about the skill gaps in their workforce is likely to reflect a wider problem of inadequate skills assessment and anticipation, particularly in smaller firms (OECD, 2021[56]). Detailed results of the PIAAC Employer Module are available in OECD (2024[57]).\n\n### **Figure 4.15. Extent of skill gaps in countries participating in the PIAAC Employer Module**\n\nShare of all firms reporting skill gaps, by the proportion of their workforce affected\n\nNote: Some data for the Netherlands are censored due to confidentiality constraints. _Countries are ranked in descending order of the share of firms identifying any skill gaps in their workforce_ . Source: PIAAC Employer Module (2022).\n\nFew Some ost All one on t know\n\n100\n\n90\n\n80\n\n0\n\n0\n\n0\n\n0\n\n30\n\n20\n\n10\n\n0\n\nSlo ak Republic Italy Portugal etherlands Hungary","pages":[{"line_numbers":[],"markdown":" **175**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Box 4.4. Results on skill gaps from the PIAAC Employer Module**\n\nThe PIAAC Employer Module is a unique survey that captures employers’ perspectives on skill gaps. For employers, skill gaps are mismatches between the skills available in a workforce and those required to meet their current and future business needs. They can be understood as the employer perspective on under-skilling, which denotes a situation where workers believe they do not have the necessary skills for their current job (Marcolin and Quintini, 2023[55]). While skill gaps and under-skilling are closely related, they are not identical.\n\nThe 2023 Survey of Adult Skills and the PIAAC Employer Module are conceptually linked, as they were jointly designed to compare employee and employer perspectives on skill gaps. Data for both surveys were collected simultaneously. The first wave of the PIAAC Employer Module took place in five OECD countries – Hungary, Italy, the Netherlands, Portugal and the Slovak Republic.\n\nAccording to the PIAAC Employer Module, a sizeable share of firms report skill gaps in their organisation. Over half of enterprises in the Slovak Republic state that their workforce has some degree of skill gap (54%), the highest of all countries (Figure 4.15). This is followed by Italy (37%), Portugal (32%), the Netherlands (31%) and Hungary (27%). Where enterprises identify skill gaps, these most often concern only a few or some of their employees. Much smaller proportions of enterprises (<5%) report that most or all their employees do not have the skills needed to perform their jobs.\n\nA notable proportion of enterprises report that they did not know whether they had a skill gap, with high shares in the Netherlands (18%) and Hungary (15%). The uncertainty expressed by many firms about the skill gaps in their workforce is likely to reflect a wider problem of inadequate skills assessment and anticipation, particularly in smaller firms (OECD, 2021[56]). Detailed results of the PIAAC Employer Module are available in OECD (2024[57]).\n\n### **Figure 4.15. Extent of skill gaps in countries participating in the PIAAC Employer Module**\n\nShare of all firms reporting skill gaps, by the proportion of their workforce affected\n\nNote: Some data for the Netherlands are censored due to confidentiality constraints. _Countries are ranked in descending order of the share of firms identifying any skill gaps in their workforce_ . Source: PIAAC Employer Module (2022).\n\nFew Some ost All one on t know\n\n100\n\n90\n\n80\n\n0\n\n0\n\n0\n\n0\n\n30\n\n20\n\n10\n\n0\n\nSlo ak Republic Italy Portugal etherlands Hungary","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":" **175**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Box 4.4. Results on skill gaps from the PIAAC Employer Module**\n\nThe PIAAC Employer Module is a unique survey that captures employers’ perspectives on skill gaps. For employers, skill gaps are mismatches between the skills available in a workforce and those required to meet their current and future business needs. They can be understood as the employer perspective on under-skilling, which denotes a situation where workers believe they do not have the necessary skills for their current job (Marcolin and Quintini, 2023[55]). While skill gaps and under-skilling are closely related, they are not identical.\n\nThe 2023 Survey of Adult Skills and the PIAAC Employer Module are conceptually linked, as they were jointly designed to compare employee and employer perspectives on skill gaps. Data for both surveys were collected simultaneously. The first wave of the PIAAC Employer Module took place in five OECD countries – Hungary, Italy, the Netherlands, Portugal and the Slovak Republic.\n\nAccording to the PIAAC Employer Module, a sizeable share of firms report skill gaps in their organisation. Over half of enterprises in the Slovak Republic state that their workforce has some degree of skill gap (54%), the highest of all countries (Figure 4.15). This is followed by Italy (37%), Portugal (32%), the Netherlands (31%) and Hungary (27%). Where enterprises identify skill gaps, these most often concern only a few or some of their employees. Much smaller proportions of enterprises (<5%) report that most or all their employees do not have the skills needed to perform their jobs.\n\nA notable proportion of enterprises report that they did not know whether they had a skill gap, with high shares in the Netherlands (18%) and Hungary (15%). The uncertainty expressed by many firms about the skill gaps in their workforce is likely to reflect a wider problem of inadequate skills assessment and anticipation, particularly in smaller firms (OECD, 2021[56]). Detailed results of the PIAAC Employer Module are available in OECD (2024[57]).\n\n### **Figure 4.15. Extent of skill gaps in countries participating in the PIAAC Employer Module**\n\nShare of all firms reporting skill gaps, by the proportion of their workforce affected\n\nNote: Some data for the Netherlands are censored due to confidentiality constraints. _Countries are ranked in descending order of the share of firms identifying any skill gaps in their workforce_ . Source: PIAAC Employer Module (2022).\n\nFew Some ost All one on t know\n\n100\n\n90\n\n80\n\n0\n\n0\n\n0\n\n0\n\n30\n\n20\n\n10\n\n0\n\nSlo ak Republic Italy Portugal etherlands Hungary","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p177.pdf"]},"response":{"errors":[],"extract_id":"extract_bbc34550782f6e8f9f9b0635ff78f960","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":[" **175**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Box 4.4. Results on skill gaps from the PIAAC Employer Module**\n\nThe PIAAC Employer Module is a unique survey that captures employers’ perspectives on skill gaps. For employers, skill gaps are mismatches between the skills available in a workforce and those required to meet their current and future business needs. They can be understood as the employer perspective on under-skilling, which denotes a situation where workers believe they do not have the necessary skills for their current job (Marcolin and Quintini, 2023[55]). While skill gaps and under-skilling are closely related, they are not identical.\n\nThe 2023 Survey of Adult Skills and the PIAAC Employer Module are conceptually linked, as they were jointly designed to compare employee and employer perspectives on skill gaps. Data for both surveys were collected simultaneously. The first wave of the PIAAC Employer Module took place in five OECD countries – Hungary, Italy, the Netherlands, Portugal and the Slovak Republic.\n\nAccording to the PIAAC Employer Module, a sizeable share of firms report skill gaps in their organisation. Over half of enterprises in the Slovak Republic state that their workforce has some degree of skill gap (54%), the highest of all countries (Figure 4.15). This is followed by Italy (37%), Portugal (32%), the Netherlands (31%) and Hungary (27%). Where enterprises identify skill gaps, these most often concern only a few or some of their employees. Much smaller proportions of enterprises (<5%) report that most or all their employees do not have the skills needed to perform their jobs.\n\nA notable proportion of enterprises report that they did not know whether they had a skill gap, with high shares in the Netherlands (18%) and Hungary (15%). The uncertainty expressed by many firms about the skill gaps in their workforce is likely to reflect a wider problem of inadequate skills assessment and anticipation, particularly in smaller firms (OECD, 2021[56]). Detailed results of the PIAAC Employer Module are available in OECD (2024[57]).\n\n### **Figure 4.15. Extent of skill gaps in countries participating in the PIAAC Employer Module**\n\nShare of all firms reporting skill gaps, by the proportion of their workforce affected\n\nNote: Some data for the Netherlands are censored due to confidentiality constraints. _Countries are ranked in descending order of the share of firms identifying any skill gaps in their workforce_ . Source: PIAAC Employer Module (2022).\n\nFew Some ost All one on t know\n\n100\n\n90\n\n80\n\n0\n\n0\n\n0\n\n0\n\n30\n\n20\n\n10\n\n0\n\nSlo ak Republic Italy Portugal etherlands Hungary"],"full_content":" **175**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Box 4.4. Results on skill gaps from the PIAAC Employer Module**\n\nThe PIAAC Employer Module is a unique survey that captures employers’ perspectives on skill gaps. For employers, skill gaps are mismatches between the skills available in a workforce and those required to meet their current and future business needs. They can be understood as the employer perspective on under-skilling, which denotes a situation where workers believe they do not have the necessary skills for their current job (Marcolin and Quintini, 2023[55]). While skill gaps and under-skilling are closely related, they are not identical.\n\nThe 2023 Survey of Adult Skills and the PIAAC Employer Module are conceptually linked, as they were jointly designed to compare employee and employer perspectives on skill gaps. Data for both surveys were collected simultaneously. The first wave of the PIAAC Employer Module took place in five OECD countries – Hungary, Italy, the Netherlands, Portugal and the Slovak Republic.\n\nAccording to the PIAAC Employer Module, a sizeable share of firms report skill gaps in their organisation. Over half of enterprises in the Slovak Republic state that their workforce has some degree of skill gap (54%), the highest of all countries (Figure 4.15). This is followed by Italy (37%), Portugal (32%), the Netherlands (31%) and Hungary (27%). Where enterprises identify skill gaps, these most often concern only a few or some of their employees. Much smaller proportions of enterprises (<5%) report that most or all their employees do not have the skills needed to perform their jobs.\n\nA notable proportion of enterprises report that they did not know whether they had a skill gap, with high shares in the Netherlands (18%) and Hungary (15%). The uncertainty expressed by many firms about the skill gaps in their workforce is likely to reflect a wider problem of inadequate skills assessment and anticipation, particularly in smaller firms (OECD, 2021[56]). Detailed results of the PIAAC Employer Module are available in OECD (2024[57]).\n\n### **Figure 4.15. Extent of skill gaps in countries participating in the PIAAC Employer Module**\n\nShare of all firms reporting skill gaps, by the proportion of their workforce affected\n\nNote: Some data for the Netherlands are censored due to confidentiality constraints. _Countries are ranked in descending order of the share of firms identifying any skill gaps in their workforce_ . Source: PIAAC Employer Module (2022).\n\nFew Some ost All one on t know\n\n100\n\n90\n\n80\n\n0\n\n0\n\n0\n\n0\n\n30\n\n20\n\n10\n\n0\n\nSlo ak Republic Italy Portugal etherlands Hungary","publish_date":null,"title":"Box 4.4. Results on skill gaps from the PIAAC Employer Module Figure 4.15. Extent of skill gaps in countries participating","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p177.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p177.pdf"},"request":{"example_id":"chart/b263dc5d-en_p177","product_type":"parse","source_file_path":"b263dc5d-en_p177.pdf"},"started_at":"2026-11-04T19:46:16.319048"} +{"completed_at":"2026-11-04T19:46:20.133919","latency_in_ms":1346,"output":{"example_id":"chart/b263dc5d-en_p178","grounded_pages":[],"layout_pages":[],"markdown":"**176** \n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n**_Prevalence of under-skilling related to digital skills_**\n\nThe 2023 Survey of Adult Skills included a new question on the types of skills that under-skilled workers feel they would need to improve to be well-matched to their jobs (Box 4.3). On average across participating OECD countries and economies, 42% of under-skilled respondents feel they need to improve their computer and software skills (see Table A.4.13 in Annex A). This is significantly more than for any other type of skill – perhaps an indication of the rapid changes to labour markets being brought about by artificial intelligence and digitalisation, and the urgency with which respondents feel they need to keep abreast with these trends. Other skills such as project management and organisational skills (26%), communication and presentation skills (26%), and teamwork or leadership skills (24%) are also cited by respondents as areas they most need to develop (see Table A.4.13 in Annex A).\n\nHowever, after taking into account the overall share of under-skilled respondents, the share of all workers stating that they lack sufficient computer and software skills is only 4% across OECD countries and economies. Japan (12%), Estonia (12%), Finland (11%) and Norway (10%) have the highest overall proportions of workers who lack the digital skills required for their current jobs (Figure 4.16). This is largely because these countries have higher shares of self-reported under-skilling. Cross-country differences reflect both the supply of skills (i.e. individuals already have strong digital skills) and the demand for such skills in the labour market (i.e. their jobs may not yet require advanced digital skills). Therefore, high levels of self-reported under-skilling may indicate not only a skill gap but also a greater awareness and desire to improve, highlighting the continued need for countries to invest in digital upskilling and training initiatives to keep pace with technological progress.\n\n### **Figure 4.16. Share of workers reporting inadequate computer and software skills for their job**\n\nEmployed adults aged 25-65 who are not self-employed\n\nNote: Does not include adults who were only administered the doorstep interview due to a language barrier (see Box 1.1 in Chapter 1 and Box 4.1). Figure plots the share of respondents who answered, \"Some of my skills are lower than what is required by my job and need to be further developed,\" to the question, \"Overall, which of the following statements best describes your skills in relation to what is required to do your ob?\", and marked “Computer and software skills” in response to the question, “Which skills were you thinking of when you answered this question?”. Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the Note for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the share of all workers reporting inadequate digital skills._ Source: Tables A.4.12 and A.4.13 in Annex A.\n\napan\n\nEstonia\n\nFinland\n\norway\n\nSweden\n\new ealand\n\nSingapore\n\nGermany\n\nenmark\n\nFlemish Region ( E\n\norea\n\netherlands\n\n**EC average**\n\nat ia\n\nithuan ia\n\nUnited States\n\nIreland\n\nCanad a\n\nChile\n\nEngland (U\n\nC echia\n\nPortugal\n\nFrance\n\nSpain\n\nSwit erland\n\nHungary\n\nAustria\n\nSlo ak Republic\n\nItaly\n\nIsrael\n\nCroatia\n\nPoland\n\n1\n\n12\n\n10\n\n8\n\n2\n\n0","pages":[{"line_numbers":[],"markdown":"**176** \n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n**_Prevalence of under-skilling related to digital skills_**\n\nThe 2023 Survey of Adult Skills included a new question on the types of skills that under-skilled workers feel they would need to improve to be well-matched to their jobs (Box 4.3). On average across participating OECD countries and economies, 42% of under-skilled respondents feel they need to improve their computer and software skills (see Table A.4.13 in Annex A). This is significantly more than for any other type of skill – perhaps an indication of the rapid changes to labour markets being brought about by artificial intelligence and digitalisation, and the urgency with which respondents feel they need to keep abreast with these trends. Other skills such as project management and organisational skills (26%), communication and presentation skills (26%), and teamwork or leadership skills (24%) are also cited by respondents as areas they most need to develop (see Table A.4.13 in Annex A).\n\nHowever, after taking into account the overall share of under-skilled respondents, the share of all workers stating that they lack sufficient computer and software skills is only 4% across OECD countries and economies. Japan (12%), Estonia (12%), Finland (11%) and Norway (10%) have the highest overall proportions of workers who lack the digital skills required for their current jobs (Figure 4.16). This is largely because these countries have higher shares of self-reported under-skilling. Cross-country differences reflect both the supply of skills (i.e. individuals already have strong digital skills) and the demand for such skills in the labour market (i.e. their jobs may not yet require advanced digital skills). Therefore, high levels of self-reported under-skilling may indicate not only a skill gap but also a greater awareness and desire to improve, highlighting the continued need for countries to invest in digital upskilling and training initiatives to keep pace with technological progress.\n\n### **Figure 4.16. Share of workers reporting inadequate computer and software skills for their job**\n\nEmployed adults aged 25-65 who are not self-employed\n\nNote: Does not include adults who were only administered the doorstep interview due to a language barrier (see Box 1.1 in Chapter 1 and Box 4.1). Figure plots the share of respondents who answered, \"Some of my skills are lower than what is required by my job and need to be further developed,\" to the question, \"Overall, which of the following statements best describes your skills in relation to what is required to do your ob?\", and marked “Computer and software skills” in response to the question, “Which skills were you thinking of when you answered this question?”. Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the Note for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the share of all workers reporting inadequate digital skills._ Source: Tables A.4.12 and A.4.13 in Annex A.\n\napan\n\nEstonia\n\nFinland\n\norway\n\nSweden\n\new ealand\n\nSingapore\n\nGermany\n\nenmark\n\nFlemish Region ( E\n\norea\n\netherlands\n\n**EC average**\n\nat ia\n\nithuan ia\n\nUnited States\n\nIreland\n\nCanad a\n\nChile\n\nEngland (U\n\nC echia\n\nPortugal\n\nFrance\n\nSpain\n\nSwit erland\n\nHungary\n\nAustria\n\nSlo ak Republic\n\nItaly\n\nIsrael\n\nCroatia\n\nPoland\n\n1\n\n12\n\n10\n\n8\n\n2\n\n0","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**176** \n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n**_Prevalence of under-skilling related to digital skills_**\n\nThe 2023 Survey of Adult Skills included a new question on the types of skills that under-skilled workers feel they would need to improve to be well-matched to their jobs (Box 4.3). On average across participating OECD countries and economies, 42% of under-skilled respondents feel they need to improve their computer and software skills (see Table A.4.13 in Annex A). This is significantly more than for any other type of skill – perhaps an indication of the rapid changes to labour markets being brought about by artificial intelligence and digitalisation, and the urgency with which respondents feel they need to keep abreast with these trends. Other skills such as project management and organisational skills (26%), communication and presentation skills (26%), and teamwork or leadership skills (24%) are also cited by respondents as areas they most need to develop (see Table A.4.13 in Annex A).\n\nHowever, after taking into account the overall share of under-skilled respondents, the share of all workers stating that they lack sufficient computer and software skills is only 4% across OECD countries and economies. Japan (12%), Estonia (12%), Finland (11%) and Norway (10%) have the highest overall proportions of workers who lack the digital skills required for their current jobs (Figure 4.16). This is largely because these countries have higher shares of self-reported under-skilling. Cross-country differences reflect both the supply of skills (i.e. individuals already have strong digital skills) and the demand for such skills in the labour market (i.e. their jobs may not yet require advanced digital skills). Therefore, high levels of self-reported under-skilling may indicate not only a skill gap but also a greater awareness and desire to improve, highlighting the continued need for countries to invest in digital upskilling and training initiatives to keep pace with technological progress.\n\n### **Figure 4.16. Share of workers reporting inadequate computer and software skills for their job**\n\nEmployed adults aged 25-65 who are not self-employed\n\nNote: Does not include adults who were only administered the doorstep interview due to a language barrier (see Box 1.1 in Chapter 1 and Box 4.1). Figure plots the share of respondents who answered, \"Some of my skills are lower than what is required by my job and need to be further developed,\" to the question, \"Overall, which of the following statements best describes your skills in relation to what is required to do your ob?\", and marked “Computer and software skills” in response to the question, “Which skills were you thinking of when you answered this question?”. Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the Note for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the share of all workers reporting inadequate digital skills._ Source: Tables A.4.12 and A.4.13 in Annex A.\n\napan\n\nEstonia\n\nFinland\n\norway\n\nSweden\n\new ealand\n\nSingapore\n\nGermany\n\nenmark\n\nFlemish Region ( E\n\norea\n\netherlands\n\n**EC average**\n\nat ia\n\nithuan ia\n\nUnited States\n\nIreland\n\nCanad a\n\nChile\n\nEngland (U\n\nC echia\n\nPortugal\n\nFrance\n\nSpain\n\nSwit erland\n\nHungary\n\nAustria\n\nSlo ak Republic\n\nItaly\n\nIsrael\n\nCroatia\n\nPoland\n\n1\n\n12\n\n10\n\n8\n\n2\n\n0","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p178.pdf"]},"response":{"errors":[],"extract_id":"extract_01b2626da5275389f3a96b5ee9701ccd","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**176** \n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n**_Prevalence of under-skilling related to digital skills_**\n\nThe 2023 Survey of Adult Skills included a new question on the types of skills that under-skilled workers feel they would need to improve to be well-matched to their jobs (Box 4.3). On average across participating OECD countries and economies, 42% of under-skilled respondents feel they need to improve their computer and software skills (see Table A.4.13 in Annex A). This is significantly more than for any other type of skill – perhaps an indication of the rapid changes to labour markets being brought about by artificial intelligence and digitalisation, and the urgency with which respondents feel they need to keep abreast with these trends. Other skills such as project management and organisational skills (26%), communication and presentation skills (26%), and teamwork or leadership skills (24%) are also cited by respondents as areas they most need to develop (see Table A.4.13 in Annex A).\n\nHowever, after taking into account the overall share of under-skilled respondents, the share of all workers stating that they lack sufficient computer and software skills is only 4% across OECD countries and economies. Japan (12%), Estonia (12%), Finland (11%) and Norway (10%) have the highest overall proportions of workers who lack the digital skills required for their current jobs (Figure 4.16). This is largely because these countries have higher shares of self-reported under-skilling. Cross-country differences reflect both the supply of skills (i.e. individuals already have strong digital skills) and the demand for such skills in the labour market (i.e. their jobs may not yet require advanced digital skills). Therefore, high levels of self-reported under-skilling may indicate not only a skill gap but also a greater awareness and desire to improve, highlighting the continued need for countries to invest in digital upskilling and training initiatives to keep pace with technological progress.\n\n### **Figure 4.16. Share of workers reporting inadequate computer and software skills for their job**\n\nEmployed adults aged 25-65 who are not self-employed\n\nNote: Does not include adults who were only administered the doorstep interview due to a language barrier (see Box 1.1 in Chapter 1 and Box 4.1). Figure plots the share of respondents who answered, \"Some of my skills are lower than what is required by my job and need to be further developed,\" to the question, \"Overall, which of the following statements best describes your skills in relation to what is required to do your ob?\", and marked “Computer and software skills” in response to the question, “Which skills were you thinking of when you answered this question?”. Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the Note for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the share of all workers reporting inadequate digital skills._ Source: Tables A.4.12 and A.4.13 in Annex A.\n\napan\n\nEstonia\n\nFinland\n\norway\n\nSweden\n\new ealand\n\nSingapore\n\nGermany\n\nenmark\n\nFlemish Region ( E\n\norea\n\netherlands\n\n**EC average**\n\nat ia\n\nithuan ia\n\nUnited States\n\nIreland\n\nCanad a\n\nChile\n\nEngland (U\n\nC echia\n\nPortugal\n\nFrance\n\nSpain\n\nSwit erland\n\nHungary\n\nAustria\n\nSlo ak Republic\n\nItaly\n\nIsrael\n\nCroatia\n\nPoland\n\n1\n\n12\n\n10\n\n8\n\n2\n\n0"],"full_content":"**176** \n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n**_Prevalence of under-skilling related to digital skills_**\n\nThe 2023 Survey of Adult Skills included a new question on the types of skills that under-skilled workers feel they would need to improve to be well-matched to their jobs (Box 4.3). On average across participating OECD countries and economies, 42% of under-skilled respondents feel they need to improve their computer and software skills (see Table A.4.13 in Annex A). This is significantly more than for any other type of skill – perhaps an indication of the rapid changes to labour markets being brought about by artificial intelligence and digitalisation, and the urgency with which respondents feel they need to keep abreast with these trends. Other skills such as project management and organisational skills (26%), communication and presentation skills (26%), and teamwork or leadership skills (24%) are also cited by respondents as areas they most need to develop (see Table A.4.13 in Annex A).\n\nHowever, after taking into account the overall share of under-skilled respondents, the share of all workers stating that they lack sufficient computer and software skills is only 4% across OECD countries and economies. Japan (12%), Estonia (12%), Finland (11%) and Norway (10%) have the highest overall proportions of workers who lack the digital skills required for their current jobs (Figure 4.16). This is largely because these countries have higher shares of self-reported under-skilling. Cross-country differences reflect both the supply of skills (i.e. individuals already have strong digital skills) and the demand for such skills in the labour market (i.e. their jobs may not yet require advanced digital skills). Therefore, high levels of self-reported under-skilling may indicate not only a skill gap but also a greater awareness and desire to improve, highlighting the continued need for countries to invest in digital upskilling and training initiatives to keep pace with technological progress.\n\n### **Figure 4.16. Share of workers reporting inadequate computer and software skills for their job**\n\nEmployed adults aged 25-65 who are not self-employed\n\nNote: Does not include adults who were only administered the doorstep interview due to a language barrier (see Box 1.1 in Chapter 1 and Box 4.1). Figure plots the share of respondents who answered, \"Some of my skills are lower than what is required by my job and need to be further developed,\" to the question, \"Overall, which of the following statements best describes your skills in relation to what is required to do your ob?\", and marked “Computer and software skills” in response to the question, “Which skills were you thinking of when you answered this question?”. Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the Note for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the share of all workers reporting inadequate digital skills._ Source: Tables A.4.12 and A.4.13 in Annex A.\n\napan\n\nEstonia\n\nFinland\n\norway\n\nSweden\n\new ealand\n\nSingapore\n\nGermany\n\nenmark\n\nFlemish Region ( E\n\norea\n\netherlands\n\n**EC average**\n\nat ia\n\nithuan ia\n\nUnited States\n\nIreland\n\nCanad a\n\nChile\n\nEngland (U\n\nC echia\n\nPortugal\n\nFrance\n\nSpain\n\nSwit erland\n\nHungary\n\nAustria\n\nSlo ak Republic\n\nItaly\n\nIsrael\n\nCroatia\n\nPoland\n\n1\n\n12\n\n10\n\n8\n\n2\n\n0","publish_date":null,"title":"Figure 4.16. Share of workers reporting inadequate computer and software skills for their job","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p178.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p178.pdf"},"request":{"example_id":"chart/b263dc5d-en_p178","product_type":"parse","source_file_path":"b263dc5d-en_p178.pdf"},"started_at":"2026-11-04T19:46:18.787891"} +{"completed_at":"2026-11-04T19:46:23.094941","latency_in_ms":2958,"output":{"example_id":"chart/b263dc5d-en_p181","grounded_pages":[],"layout_pages":[],"markdown":" **179**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 4.17. Likelihood of over-qualification, by socio-demographic characteristics [1/2]**\n\nChange in likelihood of over-qualification (relative to reference category)\n\nFinland\n\new ealand\n\nSpain\n\norway\n\nGermany\n\nithuania\n\nat ia\n\nenmark\n\nPortugal\n\nItaly\n\nPoland\n\netherlands\n\nSweden\n\nChile\n\nHungary\n\nSlo ak Republic\n\n**EC average**\n\nUnited States\n\napan\n\nCroatia\n\nC echia\n\nSingapore\n\nFrance\n\nFlemish Region ( E\n\nEstonia\n\nIreland\n\nAustria\n\norea\n\nIsrael\n\nSwit erland\n\nCanada\n\nEngland (U\n\n0\n\n8\n\nPercentage points **. umeracy**\n\nItaly\n\nPortugal\n\nChile\n\nC echia\n\nHungary\n\nEstonia\n\nSlo ak Republic\n\norea\n\nIreland\n\netherlands\n\nFrance\n\nSwit erland\n\napan\n\nSpain\n\nFlemish Region ( E\n\n**EC average**\n\nAustria\n\nSweden\n\nPoland\n\nat ia\n\nGermany\n\nCroatia\n\nenmark\n\norway\n\nUnited States\n\nCanada\n\new ealand\n\nithuania\n\nSingapore\n\nFinland\n\nEngland (U\n\nIsrael\n\n20\n\n1\n\n12\n\n8\n\n0\n\nPercentage points **B. ears of education**\n\nithuania\n\nenmark\n\nat ia\n\norea\n\nGermany\n\napan\n\nCroatia\n\netherlands\n\nUnited States\n\nItaly\n\nFlemish Region ( E\n\nSingapore\n\nPoland\n\nEstonia\n\new ealand\n\nSlo ak Republic\n\nFinland\n\nHungary\n\n**EC average**\n\norway\n\nCanada\n\nSpain\n\nChile\n\nIreland\n\nFrance\n\nSwit erland\n\nPortugal\n\nAustria\n\nEngland (U\n\nSweden\n\nC echia\n\nIsrael\n\n8\n\n0\n\n8\n\nPercentage points **C. lder or ers Ref: ounger or ers**","pages":[{"line_numbers":[],"markdown":" **179**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 4.17. Likelihood of over-qualification, by socio-demographic characteristics [1/2]**\n\nChange in likelihood of over-qualification (relative to reference category)\n\nFinland\n\new ealand\n\nSpain\n\norway\n\nGermany\n\nithuania\n\nat ia\n\nenmark\n\nPortugal\n\nItaly\n\nPoland\n\netherlands\n\nSweden\n\nChile\n\nHungary\n\nSlo ak Republic\n\n**EC average**\n\nUnited States\n\napan\n\nCroatia\n\nC echia\n\nSingapore\n\nFrance\n\nFlemish Region ( E\n\nEstonia\n\nIreland\n\nAustria\n\norea\n\nIsrael\n\nSwit erland\n\nCanada\n\nEngland (U\n\n0\n\n8\n\nPercentage points **. umeracy**\n\nItaly\n\nPortugal\n\nChile\n\nC echia\n\nHungary\n\nEstonia\n\nSlo ak Republic\n\norea\n\nIreland\n\netherlands\n\nFrance\n\nSwit erland\n\napan\n\nSpain\n\nFlemish Region ( E\n\n**EC average**\n\nAustria\n\nSweden\n\nPoland\n\nat ia\n\nGermany\n\nCroatia\n\nenmark\n\norway\n\nUnited States\n\nCanada\n\new ealand\n\nithuania\n\nSingapore\n\nFinland\n\nEngland (U\n\nIsrael\n\n20\n\n1\n\n12\n\n8\n\n0\n\nPercentage points **B. ears of education**\n\nithuania\n\nenmark\n\nat ia\n\norea\n\nGermany\n\napan\n\nCroatia\n\netherlands\n\nUnited States\n\nItaly\n\nFlemish Region ( E\n\nSingapore\n\nPoland\n\nEstonia\n\new ealand\n\nSlo ak Republic\n\nFinland\n\nHungary\n\n**EC average**\n\norway\n\nCanada\n\nSpain\n\nChile\n\nIreland\n\nFrance\n\nSwit erland\n\nPortugal\n\nAustria\n\nEngland (U\n\nSweden\n\nC echia\n\nIsrael\n\n8\n\n0\n\n8\n\nPercentage points **C. lder or ers Ref: ounger or ers**","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":" **179**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 4.17. Likelihood of over-qualification, by socio-demographic characteristics [1/2]**\n\nChange in likelihood of over-qualification (relative to reference category)\n\nFinland\n\new ealand\n\nSpain\n\norway\n\nGermany\n\nithuania\n\nat ia\n\nenmark\n\nPortugal\n\nItaly\n\nPoland\n\netherlands\n\nSweden\n\nChile\n\nHungary\n\nSlo ak Republic\n\n**EC average**\n\nUnited States\n\napan\n\nCroatia\n\nC echia\n\nSingapore\n\nFrance\n\nFlemish Region ( E\n\nEstonia\n\nIreland\n\nAustria\n\norea\n\nIsrael\n\nSwit erland\n\nCanada\n\nEngland (U\n\n0\n\n8\n\nPercentage points **. umeracy**\n\nItaly\n\nPortugal\n\nChile\n\nC echia\n\nHungary\n\nEstonia\n\nSlo ak Republic\n\norea\n\nIreland\n\netherlands\n\nFrance\n\nSwit erland\n\napan\n\nSpain\n\nFlemish Region ( E\n\n**EC average**\n\nAustria\n\nSweden\n\nPoland\n\nat ia\n\nGermany\n\nCroatia\n\nenmark\n\norway\n\nUnited States\n\nCanada\n\new ealand\n\nithuania\n\nSingapore\n\nFinland\n\nEngland (U\n\nIsrael\n\n20\n\n1\n\n12\n\n8\n\n0\n\nPercentage points **B. ears of education**\n\nithuania\n\nenmark\n\nat ia\n\norea\n\nGermany\n\napan\n\nCroatia\n\netherlands\n\nUnited States\n\nItaly\n\nFlemish Region ( E\n\nSingapore\n\nPoland\n\nEstonia\n\new ealand\n\nSlo ak Republic\n\nFinland\n\nHungary\n\n**EC average**\n\norway\n\nCanada\n\nSpain\n\nChile\n\nIreland\n\nFrance\n\nSwit erland\n\nPortugal\n\nAustria\n\nEngland (U\n\nSweden\n\nC echia\n\nIsrael\n\n8\n\n0\n\n8\n\nPercentage points **C. lder or ers Ref: ounger or ers**","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p181.pdf"]},"response":{"errors":[],"extract_id":"extract_2ce4608a9b9fa34e518151edd20485f7","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":[" **179**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 4.17. Likelihood of over-qualification, by socio-demographic characteristics [1/2]**\n\nChange in likelihood of over-qualification (relative to reference category)\n\nFinland\n\new ealand\n\nSpain\n\norway\n\nGermany\n\nithuania\n\nat ia\n\nenmark\n\nPortugal\n\nItaly\n\nPoland\n\netherlands\n\nSweden\n\nChile\n\nHungary\n\nSlo ak Republic\n\n**EC average**\n\nUnited States\n\napan\n\nCroatia\n\nC echia\n\nSingapore\n\nFrance\n\nFlemish Region ( E\n\nEstonia\n\nIreland\n\nAustria\n\norea\n\nIsrael\n\nSwit erland\n\nCanada\n\nEngland (U\n\n0\n\n8\n\nPercentage points **. umeracy**\n\nItaly\n\nPortugal\n\nChile\n\nC echia\n\nHungary\n\nEstonia\n\nSlo ak Republic\n\norea\n\nIreland\n\netherlands\n\nFrance\n\nSwit erland\n\napan\n\nSpain\n\nFlemish Region ( E\n\n**EC average**\n\nAustria\n\nSweden\n\nPoland\n\nat ia\n\nGermany\n\nCroatia\n\nenmark\n\norway\n\nUnited States\n\nCanada\n\new ealand\n\nithuania\n\nSingapore\n\nFinland\n\nEngland (U\n\nIsrael\n\n20\n\n1\n\n12\n\n8\n\n0\n\nPercentage points **B. ears of education**\n\nithuania\n\nenmark\n\nat ia\n\norea\n\nGermany\n\napan\n\nCroatia\n\netherlands\n\nUnited States\n\nItaly\n\nFlemish Region ( E\n\nSingapore\n\nPoland\n\nEstonia\n\new ealand\n\nSlo ak Republic\n\nFinland\n\nHungary\n\n**EC average**\n\norway\n\nCanada\n\nSpain\n\nChile\n\nIreland\n\nFrance\n\nSwit erland\n\nPortugal\n\nAustria\n\nEngland (U\n\nSweden\n\nC echia\n\nIsrael\n\n8\n\n0\n\n8\n\nPercentage points **C. lder or ers Ref: ounger or ers**"],"full_content":" **179**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 4.17. Likelihood of over-qualification, by socio-demographic characteristics [1/2]**\n\nChange in likelihood of over-qualification (relative to reference category)\n\nFinland\n\new ealand\n\nSpain\n\norway\n\nGermany\n\nithuania\n\nat ia\n\nenmark\n\nPortugal\n\nItaly\n\nPoland\n\netherlands\n\nSweden\n\nChile\n\nHungary\n\nSlo ak Republic\n\n**EC average**\n\nUnited States\n\napan\n\nCroatia\n\nC echia\n\nSingapore\n\nFrance\n\nFlemish Region ( E\n\nEstonia\n\nIreland\n\nAustria\n\norea\n\nIsrael\n\nSwit erland\n\nCanada\n\nEngland (U\n\n0\n\n8\n\nPercentage points **. umeracy**\n\nItaly\n\nPortugal\n\nChile\n\nC echia\n\nHungary\n\nEstonia\n\nSlo ak Republic\n\norea\n\nIreland\n\netherlands\n\nFrance\n\nSwit erland\n\napan\n\nSpain\n\nFlemish Region ( E\n\n**EC average**\n\nAustria\n\nSweden\n\nPoland\n\nat ia\n\nGermany\n\nCroatia\n\nenmark\n\norway\n\nUnited States\n\nCanada\n\new ealand\n\nithuania\n\nSingapore\n\nFinland\n\nEngland (U\n\nIsrael\n\n20\n\n1\n\n12\n\n8\n\n0\n\nPercentage points **B. ears of education**\n\nithuania\n\nenmark\n\nat ia\n\norea\n\nGermany\n\napan\n\nCroatia\n\netherlands\n\nUnited States\n\nItaly\n\nFlemish Region ( E\n\nSingapore\n\nPoland\n\nEstonia\n\new ealand\n\nSlo ak Republic\n\nFinland\n\nHungary\n\n**EC average**\n\norway\n\nCanada\n\nSpain\n\nChile\n\nIreland\n\nFrance\n\nSwit erland\n\nPortugal\n\nAustria\n\nEngland (U\n\nSweden\n\nC echia\n\nIsrael\n\n8\n\n0\n\n8\n\nPercentage points **C. lder or ers Ref: ounger or ers**","publish_date":null,"title":"Figure 4.17. Likelihood of over-qualification, by socio-demographic characteristics [1/2]","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p181.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p181.pdf"},"request":{"example_id":"chart/b263dc5d-en_p181","product_type":"parse","source_file_path":"b263dc5d-en_p181.pdf"},"started_at":"2026-11-04T19:46:20.136531"} +{"completed_at":"2026-11-04T19:46:24.111631","latency_in_ms":1657,"output":{"example_id":"chart/b263dc5d-en_p185","grounded_pages":[],"layout_pages":[],"markdown":" **183**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 4.19. Relationship between wages and mismatches, by type of mismatch**\n\nChange in gross hourly wages associated with mismatches\n\nNote: Employed adults aged 25-65 who are not self-employed; does not include adults who were only administered the doorstep interview due to a language barrier (see Box 1.1 in Chapter 1 and Box 4.1). Mismatch measures are defined in Table 4.2 and Box 4.3. Estimates account for age, gender, immigrant background, parental education, whether one lives with a partner or has children, work experience, use of numeracy skills at work, industry and occupation. Wages are (log) gross hourly earnings for employed and self-employed individuals, including bonuses, in PPP-adjusted 2022 USD. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the ote for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the effect of mismatches on wages._ Source: Table A.4.16 of Annex A.\n\nItaly\n\nPoland\n\nAustria\n\nSweden\n\norea\n\nCroatia\n\nFinland\n\nC echia\n\napan\n\nat ia\n\nEstonia\n\nFrance\n\netherlands\n\norway\n\nCanada\n\n**EC average**\n\nenmark\n\nGermany\n\nIreland\n\nIsrael\n\nHungary\n\nFlemish Region ( E\n\nSpain\n\nSlo ak Republic\n\nPortugal\n\nithuania\n\nSwit erland\n\new ealand\n\nEngland (U\n\nUnited States\n\nSingapore\n\nChile\n\n0\n\n12\n\n18\n\n2\n\n12\n\n0\n\n12\n\n18\n\n0\n\n12\n\n18\n\n2\n\n**. ver ualification**\n\norea\n\nithuania\n\nPoland\n\nSlo ak Republic\n\nItaly\n\napan\n\nFlemish Region ( E\n\nSingapore\n\nIsrael\n\netherlands\n\nChile\n\nat ia\n\nCroatia\n\n**EC average**\n\nC echia\n\nenmark\n\nSweden\n\nGermany\n\nHungary\n\nPortugal\n\nAustria\n\nEstonia\n\nCanada\n\nSwit erland\n\norway\n\nUnited States\n\nFrance\n\nFinland\n\new ealand\n\nSpain\n\nIreland\n\nEngland (U\n\n**B. ver s illing**\n\nC echia\n\nChile\n\nIreland\n\netherlands\n\nFlemish Region ( E\n\nEstonia\n\norway\n\norea\n\nFinland\n\nSweden\n\new ealand\n\nFrance\n\nCroatia\n\nenmark\n\nAustria\n\n**EC average**\n\nithuania\n\nGermany\n\nSwit erland\n\nSlo ak Republic\n\nPortugal\n\nPoland\n\nSingapore\n\nCanada\n\nHungary\n\napan\n\nItaly\n\nat ia\n\nSpain\n\nEngland (U\n\nUnited States\n\nIsrael\n\n**C. ield of study mismatc**","pages":[{"line_numbers":[],"markdown":" **183**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 4.19. Relationship between wages and mismatches, by type of mismatch**\n\nChange in gross hourly wages associated with mismatches\n\nNote: Employed adults aged 25-65 who are not self-employed; does not include adults who were only administered the doorstep interview due to a language barrier (see Box 1.1 in Chapter 1 and Box 4.1). Mismatch measures are defined in Table 4.2 and Box 4.3. Estimates account for age, gender, immigrant background, parental education, whether one lives with a partner or has children, work experience, use of numeracy skills at work, industry and occupation. Wages are (log) gross hourly earnings for employed and self-employed individuals, including bonuses, in PPP-adjusted 2022 USD. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the ote for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the effect of mismatches on wages._ Source: Table A.4.16 of Annex A.\n\nItaly\n\nPoland\n\nAustria\n\nSweden\n\norea\n\nCroatia\n\nFinland\n\nC echia\n\napan\n\nat ia\n\nEstonia\n\nFrance\n\netherlands\n\norway\n\nCanada\n\n**EC average**\n\nenmark\n\nGermany\n\nIreland\n\nIsrael\n\nHungary\n\nFlemish Region ( E\n\nSpain\n\nSlo ak Republic\n\nPortugal\n\nithuania\n\nSwit erland\n\new ealand\n\nEngland (U\n\nUnited States\n\nSingapore\n\nChile\n\n0\n\n12\n\n18\n\n2\n\n12\n\n0\n\n12\n\n18\n\n0\n\n12\n\n18\n\n2\n\n**. ver ualification**\n\norea\n\nithuania\n\nPoland\n\nSlo ak Republic\n\nItaly\n\napan\n\nFlemish Region ( E\n\nSingapore\n\nIsrael\n\netherlands\n\nChile\n\nat ia\n\nCroatia\n\n**EC average**\n\nC echia\n\nenmark\n\nSweden\n\nGermany\n\nHungary\n\nPortugal\n\nAustria\n\nEstonia\n\nCanada\n\nSwit erland\n\norway\n\nUnited States\n\nFrance\n\nFinland\n\new ealand\n\nSpain\n\nIreland\n\nEngland (U\n\n**B. ver s illing**\n\nC echia\n\nChile\n\nIreland\n\netherlands\n\nFlemish Region ( E\n\nEstonia\n\norway\n\norea\n\nFinland\n\nSweden\n\new ealand\n\nFrance\n\nCroatia\n\nenmark\n\nAustria\n\n**EC average**\n\nithuania\n\nGermany\n\nSwit erland\n\nSlo ak Republic\n\nPortugal\n\nPoland\n\nSingapore\n\nCanada\n\nHungary\n\napan\n\nItaly\n\nat ia\n\nSpain\n\nEngland (U\n\nUnited States\n\nIsrael\n\n**C. ield of study mismatc**","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":" **183**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 4.19. Relationship between wages and mismatches, by type of mismatch**\n\nChange in gross hourly wages associated with mismatches\n\nNote: Employed adults aged 25-65 who are not self-employed; does not include adults who were only administered the doorstep interview due to a language barrier (see Box 1.1 in Chapter 1 and Box 4.1). Mismatch measures are defined in Table 4.2 and Box 4.3. Estimates account for age, gender, immigrant background, parental education, whether one lives with a partner or has children, work experience, use of numeracy skills at work, industry and occupation. Wages are (log) gross hourly earnings for employed and self-employed individuals, including bonuses, in PPP-adjusted 2022 USD. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the ote for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the effect of mismatches on wages._ Source: Table A.4.16 of Annex A.\n\nItaly\n\nPoland\n\nAustria\n\nSweden\n\norea\n\nCroatia\n\nFinland\n\nC echia\n\napan\n\nat ia\n\nEstonia\n\nFrance\n\netherlands\n\norway\n\nCanada\n\n**EC average**\n\nenmark\n\nGermany\n\nIreland\n\nIsrael\n\nHungary\n\nFlemish Region ( E\n\nSpain\n\nSlo ak Republic\n\nPortugal\n\nithuania\n\nSwit erland\n\new ealand\n\nEngland (U\n\nUnited States\n\nSingapore\n\nChile\n\n0\n\n12\n\n18\n\n2\n\n12\n\n0\n\n12\n\n18\n\n0\n\n12\n\n18\n\n2\n\n**. ver ualification**\n\norea\n\nithuania\n\nPoland\n\nSlo ak Republic\n\nItaly\n\napan\n\nFlemish Region ( E\n\nSingapore\n\nIsrael\n\netherlands\n\nChile\n\nat ia\n\nCroatia\n\n**EC average**\n\nC echia\n\nenmark\n\nSweden\n\nGermany\n\nHungary\n\nPortugal\n\nAustria\n\nEstonia\n\nCanada\n\nSwit erland\n\norway\n\nUnited States\n\nFrance\n\nFinland\n\new ealand\n\nSpain\n\nIreland\n\nEngland (U\n\n**B. ver s illing**\n\nC echia\n\nChile\n\nIreland\n\netherlands\n\nFlemish Region ( E\n\nEstonia\n\norway\n\norea\n\nFinland\n\nSweden\n\new ealand\n\nFrance\n\nCroatia\n\nenmark\n\nAustria\n\n**EC average**\n\nithuania\n\nGermany\n\nSwit erland\n\nSlo ak Republic\n\nPortugal\n\nPoland\n\nSingapore\n\nCanada\n\nHungary\n\napan\n\nItaly\n\nat ia\n\nSpain\n\nEngland (U\n\nUnited States\n\nIsrael\n\n**C. ield of study mismatc**","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p185.pdf"]},"response":{"errors":[],"extract_id":"extract_0b31bcba3c1331628cb27314a4adcebc","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":[" **183**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 4.19. Relationship between wages and mismatches, by type of mismatch**\n\nChange in gross hourly wages associated with mismatches\n\nNote: Employed adults aged 25-65 who are not self-employed; does not include adults who were only administered the doorstep interview due to a language barrier (see Box 1.1 in Chapter 1 and Box 4.1). Mismatch measures are defined in Table 4.2 and Box 4.3. Estimates account for age, gender, immigrant background, parental education, whether one lives with a partner or has children, work experience, use of numeracy skills at work, industry and occupation. Wages are (log) gross hourly earnings for employed and self-employed individuals, including bonuses, in PPP-adjusted 2022 USD. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the ote for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the effect of mismatches on wages._ Source: Table A.4.16 of Annex A.\n\nItaly\n\nPoland\n\nAustria\n\nSweden\n\norea\n\nCroatia\n\nFinland\n\nC echia\n\napan\n\nat ia\n\nEstonia\n\nFrance\n\netherlands\n\norway\n\nCanada\n\n**EC average**\n\nenmark\n\nGermany\n\nIreland\n\nIsrael\n\nHungary\n\nFlemish Region ( E\n\nSpain\n\nSlo ak Republic\n\nPortugal\n\nithuania\n\nSwit erland\n\new ealand\n\nEngland (U\n\nUnited States\n\nSingapore\n\nChile\n\n0\n\n12\n\n18\n\n2\n\n12\n\n0\n\n12\n\n18\n\n0\n\n12\n\n18\n\n2\n\n**. ver ualification**\n\norea\n\nithuania\n\nPoland\n\nSlo ak Republic\n\nItaly\n\napan\n\nFlemish Region ( E\n\nSingapore\n\nIsrael\n\netherlands\n\nChile\n\nat ia\n\nCroatia\n\n**EC average**\n\nC echia\n\nenmark\n\nSweden\n\nGermany\n\nHungary\n\nPortugal\n\nAustria\n\nEstonia\n\nCanada\n\nSwit erland\n\norway\n\nUnited States\n\nFrance\n\nFinland\n\new ealand\n\nSpain\n\nIreland\n\nEngland (U\n\n**B. ver s illing**\n\nC echia\n\nChile\n\nIreland\n\netherlands\n\nFlemish Region ( E\n\nEstonia\n\norway\n\norea\n\nFinland\n\nSweden\n\new ealand\n\nFrance\n\nCroatia\n\nenmark\n\nAustria\n\n**EC average**\n\nithuania\n\nGermany\n\nSwit erland\n\nSlo ak Republic\n\nPortugal\n\nPoland\n\nSingapore\n\nCanada\n\nHungary\n\napan\n\nItaly\n\nat ia\n\nSpain\n\nEngland (U\n\nUnited States\n\nIsrael\n\n**C. ield of study mismatc**"],"full_content":" **183**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 4.19. Relationship between wages and mismatches, by type of mismatch**\n\nChange in gross hourly wages associated with mismatches\n\nNote: Employed adults aged 25-65 who are not self-employed; does not include adults who were only administered the doorstep interview due to a language barrier (see Box 1.1 in Chapter 1 and Box 4.1). Mismatch measures are defined in Table 4.2 and Box 4.3. Estimates account for age, gender, immigrant background, parental education, whether one lives with a partner or has children, work experience, use of numeracy skills at work, industry and occupation. Wages are (log) gross hourly earnings for employed and self-employed individuals, including bonuses, in PPP-adjusted 2022 USD. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the ote for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the effect of mismatches on wages._ Source: Table A.4.16 of Annex A.\n\nItaly\n\nPoland\n\nAustria\n\nSweden\n\norea\n\nCroatia\n\nFinland\n\nC echia\n\napan\n\nat ia\n\nEstonia\n\nFrance\n\netherlands\n\norway\n\nCanada\n\n**EC average**\n\nenmark\n\nGermany\n\nIreland\n\nIsrael\n\nHungary\n\nFlemish Region ( E\n\nSpain\n\nSlo ak Republic\n\nPortugal\n\nithuania\n\nSwit erland\n\new ealand\n\nEngland (U\n\nUnited States\n\nSingapore\n\nChile\n\n0\n\n12\n\n18\n\n2\n\n12\n\n0\n\n12\n\n18\n\n0\n\n12\n\n18\n\n2\n\n**. ver ualification**\n\norea\n\nithuania\n\nPoland\n\nSlo ak Republic\n\nItaly\n\napan\n\nFlemish Region ( E\n\nSingapore\n\nIsrael\n\netherlands\n\nChile\n\nat ia\n\nCroatia\n\n**EC average**\n\nC echia\n\nenmark\n\nSweden\n\nGermany\n\nHungary\n\nPortugal\n\nAustria\n\nEstonia\n\nCanada\n\nSwit erland\n\norway\n\nUnited States\n\nFrance\n\nFinland\n\new ealand\n\nSpain\n\nIreland\n\nEngland (U\n\n**B. ver s illing**\n\nC echia\n\nChile\n\nIreland\n\netherlands\n\nFlemish Region ( E\n\nEstonia\n\norway\n\norea\n\nFinland\n\nSweden\n\new ealand\n\nFrance\n\nCroatia\n\nenmark\n\nAustria\n\n**EC average**\n\nithuania\n\nGermany\n\nSwit erland\n\nSlo ak Republic\n\nPortugal\n\nPoland\n\nSingapore\n\nCanada\n\nHungary\n\napan\n\nItaly\n\nat ia\n\nSpain\n\nEngland (U\n\nUnited States\n\nIsrael\n\n**C. ield of study mismatc**","publish_date":null,"title":"Figure 4.19. Relationship between wages and mismatches, by type of mismatch","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p185.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p185.pdf"},"request":{"example_id":"chart/b263dc5d-en_p185","product_type":"parse","source_file_path":"b263dc5d-en_p185.pdf"},"started_at":"2026-11-04T19:46:22.453969"} +{"completed_at":"2026-11-04T19:46:27.467824","latency_in_ms":3921,"output":{"example_id":"chart/b263dc5d-en_p187","grounded_pages":[],"layout_pages":[],"markdown":" **185**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 4.20. Relationship between over-qualification and life satisfaction**\n\nChange in likelihood of over-qualified workers reporting high life satisfaction\n\nNote: Employed adults aged 25-65 who are not self-employed; does not include adults who were only administered the doorstep interview due to a language barrier (see Box 1.1 in Chapter 1 and Box 4.1). Over-qualification is defined in Table 4.2 and Box 4.3. Life satisfaction is defined in Box 4.2. Estimates account for age, gender, immigrant background, whether one lives with a partner or has children and (log) wages. Darker colours denote differences that are statistically significant at the 5% level. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the ote for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the effect of over-qualification on life satisfaction._ Source: Table A.4.17 of Annex A.\n\n2\n\n0\n\n2\n\n8\n\n10\n\n12\n\n1\n\nSlo ak Republic\n\new ealand\n\nItaly\n\nat ia\n\nithuania\n\nSingapore\n\nFinland\n\norway\n\nC echia\n\netherlands\n\nCanada\n\nEstonia\n\nSweden\n\nEngland (U\n\nCroatia\n\nenmark\n\nSpain\n\nIsrael\n\n**EC average**\n\nFrance\n\nPortugal\n\nGermany\n\nPoland\n\nAustria\n\nChile\n\napan\n\nHungary\n\norea\n\nSwit erland\n\nFlemish Region ( E\n\nIreland\n\nUnited States\n\nPercentage points","pages":[{"line_numbers":[],"markdown":" **185**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 4.20. Relationship between over-qualification and life satisfaction**\n\nChange in likelihood of over-qualified workers reporting high life satisfaction\n\nNote: Employed adults aged 25-65 who are not self-employed; does not include adults who were only administered the doorstep interview due to a language barrier (see Box 1.1 in Chapter 1 and Box 4.1). Over-qualification is defined in Table 4.2 and Box 4.3. Life satisfaction is defined in Box 4.2. Estimates account for age, gender, immigrant background, whether one lives with a partner or has children and (log) wages. Darker colours denote differences that are statistically significant at the 5% level. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the ote for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the effect of over-qualification on life satisfaction._ Source: Table A.4.17 of Annex A.\n\n2\n\n0\n\n2\n\n8\n\n10\n\n12\n\n1\n\nSlo ak Republic\n\new ealand\n\nItaly\n\nat ia\n\nithuania\n\nSingapore\n\nFinland\n\norway\n\nC echia\n\netherlands\n\nCanada\n\nEstonia\n\nSweden\n\nEngland (U\n\nCroatia\n\nenmark\n\nSpain\n\nIsrael\n\n**EC average**\n\nFrance\n\nPortugal\n\nGermany\n\nPoland\n\nAustria\n\nChile\n\napan\n\nHungary\n\norea\n\nSwit erland\n\nFlemish Region ( E\n\nIreland\n\nUnited States\n\nPercentage points","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":" **185**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 4.20. Relationship between over-qualification and life satisfaction**\n\nChange in likelihood of over-qualified workers reporting high life satisfaction\n\nNote: Employed adults aged 25-65 who are not self-employed; does not include adults who were only administered the doorstep interview due to a language barrier (see Box 1.1 in Chapter 1 and Box 4.1). Over-qualification is defined in Table 4.2 and Box 4.3. Life satisfaction is defined in Box 4.2. Estimates account for age, gender, immigrant background, whether one lives with a partner or has children and (log) wages. Darker colours denote differences that are statistically significant at the 5% level. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the ote for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the effect of over-qualification on life satisfaction._ Source: Table A.4.17 of Annex A.\n\n2\n\n0\n\n2\n\n8\n\n10\n\n12\n\n1\n\nSlo ak Republic\n\new ealand\n\nItaly\n\nat ia\n\nithuania\n\nSingapore\n\nFinland\n\norway\n\nC echia\n\netherlands\n\nCanada\n\nEstonia\n\nSweden\n\nEngland (U\n\nCroatia\n\nenmark\n\nSpain\n\nIsrael\n\n**EC average**\n\nFrance\n\nPortugal\n\nGermany\n\nPoland\n\nAustria\n\nChile\n\napan\n\nHungary\n\norea\n\nSwit erland\n\nFlemish Region ( E\n\nIreland\n\nUnited States\n\nPercentage points","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p187.pdf"]},"response":{"errors":[],"extract_id":"extract_851e5ca51b3599dfb5baefc56c54beb4","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":[" **185**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 4.20. Relationship between over-qualification and life satisfaction**\n\nChange in likelihood of over-qualified workers reporting high life satisfaction\n\nNote: Employed adults aged 25-65 who are not self-employed; does not include adults who were only administered the doorstep interview due to a language barrier (see Box 1.1 in Chapter 1 and Box 4.1). Over-qualification is defined in Table 4.2 and Box 4.3. Life satisfaction is defined in Box 4.2. Estimates account for age, gender, immigrant background, whether one lives with a partner or has children and (log) wages. Darker colours denote differences that are statistically significant at the 5% level. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the ote for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the effect of over-qualification on life satisfaction._ Source: Table A.4.17 of Annex A.\n\n2\n\n0\n\n2\n\n8\n\n10\n\n12\n\n1\n\nSlo ak Republic\n\new ealand\n\nItaly\n\nat ia\n\nithuania\n\nSingapore\n\nFinland\n\norway\n\nC echia\n\netherlands\n\nCanada\n\nEstonia\n\nSweden\n\nEngland (U\n\nCroatia\n\nenmark\n\nSpain\n\nIsrael\n\n**EC average**\n\nFrance\n\nPortugal\n\nGermany\n\nPoland\n\nAustria\n\nChile\n\napan\n\nHungary\n\norea\n\nSwit erland\n\nFlemish Region ( E\n\nIreland\n\nUnited States\n\nPercentage points"],"full_content":" **185**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 4.20. Relationship between over-qualification and life satisfaction**\n\nChange in likelihood of over-qualified workers reporting high life satisfaction\n\nNote: Employed adults aged 25-65 who are not self-employed; does not include adults who were only administered the doorstep interview due to a language barrier (see Box 1.1 in Chapter 1 and Box 4.1). Over-qualification is defined in Table 4.2 and Box 4.3. Life satisfaction is defined in Box 4.2. Estimates account for age, gender, immigrant background, whether one lives with a partner or has children and (log) wages. Darker colours denote differences that are statistically significant at the 5% level. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the ote for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the effect of over-qualification on life satisfaction._ Source: Table A.4.17 of Annex A.\n\n2\n\n0\n\n2\n\n8\n\n10\n\n12\n\n1\n\nSlo ak Republic\n\new ealand\n\nItaly\n\nat ia\n\nithuania\n\nSingapore\n\nFinland\n\norway\n\nC echia\n\netherlands\n\nCanada\n\nEstonia\n\nSweden\n\nEngland (U\n\nCroatia\n\nenmark\n\nSpain\n\nIsrael\n\n**EC average**\n\nFrance\n\nPortugal\n\nGermany\n\nPoland\n\nAustria\n\nChile\n\napan\n\nHungary\n\norea\n\nSwit erland\n\nFlemish Region ( E\n\nIreland\n\nUnited States\n\nPercentage points","publish_date":null,"title":"Figure 4.20. Relationship between over-qualification and life satisfaction","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p187.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p187.pdf"},"request":{"example_id":"chart/b263dc5d-en_p187","product_type":"parse","source_file_path":"b263dc5d-en_p187.pdf"},"started_at":"2026-11-04T19:46:23.546440"} +{"completed_at":"2026-11-04T19:46:27.671097","latency_in_ms":3186,"output":{"example_id":"chart/b263dc5d-en_p36","grounded_pages":[],"layout_pages":[],"markdown":"**34** \n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\nPublic policies can also create training and insurance policies to help workers at risk of losing their jobs (e.g. due to the introduction of regulations aimed at reducing emissions) in the transition to new occupations and encourage firms to invest in the training of all workers and improved working conditions to keep older workers in the workforce for longer. Public policies can also respond to ageing societies by reforming pension systems and promoting measures that boost birth rates.\n\nIn democratic societies, the responses to these challenges will depend on citizens’ preferences, in particular those expressed in elections and through collective action. The context in which policy preferences are formed, and collective decisions taken, has significantly changed over the past decade, due in particular to the rise of digital media and the increasing socio-economic and cultural diversity of modern societies.\n\n**_The Internet has changed how people access news and information_**\n\nIn the decade that separated the two cycles of the Survey of Adult Skills, OECD countries have achieved near-universal access to the Internet. On average, the share of adults using the Internet increased from 76% in 2012 to 93% in 2023, with 87% of adults reporting using it daily (up from 61% in 2012; Figure 1.3). In 2022, 73% of adults reported accessing the Internet from a mobile device and 71% used it to read or download newspapers or news magazines (OECD, 2024[31]).\n\n### **Figure 1.3. Evolution of Internet usage, 2012-23**\n\nInternet use in the past three months among 16-74 year-olds; OECD average; in per cent\n\nNote: Data on adults using the Internet in mobility are not available for 2023. Source: OECD (2024[31]), “ICT Access and Usage by Households and Individuals”, https://doi.org/10.1787/b9823565-en (accessed on 29 May 2024).\n\nAs a result, the role of traditional media (the printed press, radio and television) has declined, and adults increasingly use social media to get information and discuss and debate political opinions. In the European Union, 64% of adults reported using the Internet for reading online news sites or newspapers in 2023, and 18% used it for civic or political participation (Eurostat, 2024[32]). A recent survey by the Pew Research Center found that about half of adults in the United States got news from social media “sometimes” or “often” in 2022 (Pew Research Center, 2023[33]).\n\nThe rapid diffusion and circulation of information is not without risks. There is a concern that the increase in the sheer amount of information available has been accompanied by a decrease in the quality (OECD, 2023[34]). Internet platforms and social media also make it easier to spread false and misleading information. Some actors may do so deliberately to distort public debates and fuel polarisation. At times this may form part of a hybrid warfare tactic, to erode the social fabric of open societies and weaken\n\nAdults using the Internet Adults using the Internet daily Adults using the Internet in mobility\n\n2012 2013 201 201 201 201 2018 2019 2020 2021 2022 2023\n\n100\n\n90\n\n80\n\n0\n\n0\n\n0\n\n0\n\n30\n\n20\n\n10\n\n0","pages":[{"line_numbers":[],"markdown":"**34** \n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\nPublic policies can also create training and insurance policies to help workers at risk of losing their jobs (e.g. due to the introduction of regulations aimed at reducing emissions) in the transition to new occupations and encourage firms to invest in the training of all workers and improved working conditions to keep older workers in the workforce for longer. Public policies can also respond to ageing societies by reforming pension systems and promoting measures that boost birth rates.\n\nIn democratic societies, the responses to these challenges will depend on citizens’ preferences, in particular those expressed in elections and through collective action. The context in which policy preferences are formed, and collective decisions taken, has significantly changed over the past decade, due in particular to the rise of digital media and the increasing socio-economic and cultural diversity of modern societies.\n\n**_The Internet has changed how people access news and information_**\n\nIn the decade that separated the two cycles of the Survey of Adult Skills, OECD countries have achieved near-universal access to the Internet. On average, the share of adults using the Internet increased from 76% in 2012 to 93% in 2023, with 87% of adults reporting using it daily (up from 61% in 2012; Figure 1.3). In 2022, 73% of adults reported accessing the Internet from a mobile device and 71% used it to read or download newspapers or news magazines (OECD, 2024[31]).\n\n### **Figure 1.3. Evolution of Internet usage, 2012-23**\n\nInternet use in the past three months among 16-74 year-olds; OECD average; in per cent\n\nNote: Data on adults using the Internet in mobility are not available for 2023. Source: OECD (2024[31]), “ICT Access and Usage by Households and Individuals”, https://doi.org/10.1787/b9823565-en (accessed on 29 May 2024).\n\nAs a result, the role of traditional media (the printed press, radio and television) has declined, and adults increasingly use social media to get information and discuss and debate political opinions. In the European Union, 64% of adults reported using the Internet for reading online news sites or newspapers in 2023, and 18% used it for civic or political participation (Eurostat, 2024[32]). A recent survey by the Pew Research Center found that about half of adults in the United States got news from social media “sometimes” or “often” in 2022 (Pew Research Center, 2023[33]).\n\nThe rapid diffusion and circulation of information is not without risks. There is a concern that the increase in the sheer amount of information available has been accompanied by a decrease in the quality (OECD, 2023[34]). Internet platforms and social media also make it easier to spread false and misleading information. Some actors may do so deliberately to distort public debates and fuel polarisation. At times this may form part of a hybrid warfare tactic, to erode the social fabric of open societies and weaken\n\nAdults using the Internet Adults using the Internet daily Adults using the Internet in mobility\n\n2012 2013 201 201 201 201 2018 2019 2020 2021 2022 2023\n\n100\n\n90\n\n80\n\n0\n\n0\n\n0\n\n0\n\n30\n\n20\n\n10\n\n0","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**34** \n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\nPublic policies can also create training and insurance policies to help workers at risk of losing their jobs (e.g. due to the introduction of regulations aimed at reducing emissions) in the transition to new occupations and encourage firms to invest in the training of all workers and improved working conditions to keep older workers in the workforce for longer. Public policies can also respond to ageing societies by reforming pension systems and promoting measures that boost birth rates.\n\nIn democratic societies, the responses to these challenges will depend on citizens’ preferences, in particular those expressed in elections and through collective action. The context in which policy preferences are formed, and collective decisions taken, has significantly changed over the past decade, due in particular to the rise of digital media and the increasing socio-economic and cultural diversity of modern societies.\n\n**_The Internet has changed how people access news and information_**\n\nIn the decade that separated the two cycles of the Survey of Adult Skills, OECD countries have achieved near-universal access to the Internet. On average, the share of adults using the Internet increased from 76% in 2012 to 93% in 2023, with 87% of adults reporting using it daily (up from 61% in 2012; Figure 1.3). In 2022, 73% of adults reported accessing the Internet from a mobile device and 71% used it to read or download newspapers or news magazines (OECD, 2024[31]).\n\n### **Figure 1.3. Evolution of Internet usage, 2012-23**\n\nInternet use in the past three months among 16-74 year-olds; OECD average; in per cent\n\nNote: Data on adults using the Internet in mobility are not available for 2023. Source: OECD (2024[31]), “ICT Access and Usage by Households and Individuals”, https://doi.org/10.1787/b9823565-en (accessed on 29 May 2024).\n\nAs a result, the role of traditional media (the printed press, radio and television) has declined, and adults increasingly use social media to get information and discuss and debate political opinions. In the European Union, 64% of adults reported using the Internet for reading online news sites or newspapers in 2023, and 18% used it for civic or political participation (Eurostat, 2024[32]). A recent survey by the Pew Research Center found that about half of adults in the United States got news from social media “sometimes” or “often” in 2022 (Pew Research Center, 2023[33]).\n\nThe rapid diffusion and circulation of information is not without risks. There is a concern that the increase in the sheer amount of information available has been accompanied by a decrease in the quality (OECD, 2023[34]). Internet platforms and social media also make it easier to spread false and misleading information. Some actors may do so deliberately to distort public debates and fuel polarisation. At times this may form part of a hybrid warfare tactic, to erode the social fabric of open societies and weaken\n\nAdults using the Internet Adults using the Internet daily Adults using the Internet in mobility\n\n2012 2013 201 201 201 201 2018 2019 2020 2021 2022 2023\n\n100\n\n90\n\n80\n\n0\n\n0\n\n0\n\n0\n\n30\n\n20\n\n10\n\n0","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p36.pdf"]},"response":{"errors":[],"extract_id":"extract_3449600f8cbae94eedd63417ed3884a8","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**34** \n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\nPublic policies can also create training and insurance policies to help workers at risk of losing their jobs (e.g. due to the introduction of regulations aimed at reducing emissions) in the transition to new occupations and encourage firms to invest in the training of all workers and improved working conditions to keep older workers in the workforce for longer. Public policies can also respond to ageing societies by reforming pension systems and promoting measures that boost birth rates.\n\nIn democratic societies, the responses to these challenges will depend on citizens’ preferences, in particular those expressed in elections and through collective action. The context in which policy preferences are formed, and collective decisions taken, has significantly changed over the past decade, due in particular to the rise of digital media and the increasing socio-economic and cultural diversity of modern societies.\n\n**_The Internet has changed how people access news and information_**\n\nIn the decade that separated the two cycles of the Survey of Adult Skills, OECD countries have achieved near-universal access to the Internet. On average, the share of adults using the Internet increased from 76% in 2012 to 93% in 2023, with 87% of adults reporting using it daily (up from 61% in 2012; Figure 1.3). In 2022, 73% of adults reported accessing the Internet from a mobile device and 71% used it to read or download newspapers or news magazines (OECD, 2024[31]).\n\n### **Figure 1.3. Evolution of Internet usage, 2012-23**\n\nInternet use in the past three months among 16-74 year-olds; OECD average; in per cent\n\nNote: Data on adults using the Internet in mobility are not available for 2023. Source: OECD (2024[31]), “ICT Access and Usage by Households and Individuals”, https://doi.org/10.1787/b9823565-en (accessed on 29 May 2024).\n\nAs a result, the role of traditional media (the printed press, radio and television) has declined, and adults increasingly use social media to get information and discuss and debate political opinions. In the European Union, 64% of adults reported using the Internet for reading online news sites or newspapers in 2023, and 18% used it for civic or political participation (Eurostat, 2024[32]). A recent survey by the Pew Research Center found that about half of adults in the United States got news from social media “sometimes” or “often” in 2022 (Pew Research Center, 2023[33]).\n\nThe rapid diffusion and circulation of information is not without risks. There is a concern that the increase in the sheer amount of information available has been accompanied by a decrease in the quality (OECD, 2023[34]). Internet platforms and social media also make it easier to spread false and misleading information. Some actors may do so deliberately to distort public debates and fuel polarisation. At times this may form part of a hybrid warfare tactic, to erode the social fabric of open societies and weaken\n\nAdults using the Internet Adults using the Internet daily Adults using the Internet in mobility\n\n2012 2013 201 201 201 201 2018 2019 2020 2021 2022 2023\n\n100\n\n90\n\n80\n\n0\n\n0\n\n0\n\n0\n\n30\n\n20\n\n10\n\n0"],"full_content":"**34** \n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\nPublic policies can also create training and insurance policies to help workers at risk of losing their jobs (e.g. due to the introduction of regulations aimed at reducing emissions) in the transition to new occupations and encourage firms to invest in the training of all workers and improved working conditions to keep older workers in the workforce for longer. Public policies can also respond to ageing societies by reforming pension systems and promoting measures that boost birth rates.\n\nIn democratic societies, the responses to these challenges will depend on citizens’ preferences, in particular those expressed in elections and through collective action. The context in which policy preferences are formed, and collective decisions taken, has significantly changed over the past decade, due in particular to the rise of digital media and the increasing socio-economic and cultural diversity of modern societies.\n\n**_The Internet has changed how people access news and information_**\n\nIn the decade that separated the two cycles of the Survey of Adult Skills, OECD countries have achieved near-universal access to the Internet. On average, the share of adults using the Internet increased from 76% in 2012 to 93% in 2023, with 87% of adults reporting using it daily (up from 61% in 2012; Figure 1.3). In 2022, 73% of adults reported accessing the Internet from a mobile device and 71% used it to read or download newspapers or news magazines (OECD, 2024[31]).\n\n### **Figure 1.3. Evolution of Internet usage, 2012-23**\n\nInternet use in the past three months among 16-74 year-olds; OECD average; in per cent\n\nNote: Data on adults using the Internet in mobility are not available for 2023. Source: OECD (2024[31]), “ICT Access and Usage by Households and Individuals”, https://doi.org/10.1787/b9823565-en (accessed on 29 May 2024).\n\nAs a result, the role of traditional media (the printed press, radio and television) has declined, and adults increasingly use social media to get information and discuss and debate political opinions. In the European Union, 64% of adults reported using the Internet for reading online news sites or newspapers in 2023, and 18% used it for civic or political participation (Eurostat, 2024[32]). A recent survey by the Pew Research Center found that about half of adults in the United States got news from social media “sometimes” or “often” in 2022 (Pew Research Center, 2023[33]).\n\nThe rapid diffusion and circulation of information is not without risks. There is a concern that the increase in the sheer amount of information available has been accompanied by a decrease in the quality (OECD, 2023[34]). Internet platforms and social media also make it easier to spread false and misleading information. Some actors may do so deliberately to distort public debates and fuel polarisation. At times this may form part of a hybrid warfare tactic, to erode the social fabric of open societies and weaken\n\nAdults using the Internet Adults using the Internet daily Adults using the Internet in mobility\n\n2012 2013 201 201 201 201 2018 2019 2020 2021 2022 2023\n\n100\n\n90\n\n80\n\n0\n\n0\n\n0\n\n0\n\n30\n\n20\n\n10\n\n0","publish_date":null,"title":"Figure 1.3. Evolution of Internet usage, 2012-23","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p36.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p36.pdf"},"request":{"example_id":"chart/b263dc5d-en_p36","product_type":"parse","source_file_path":"b263dc5d-en_p36.pdf"},"started_at":"2026-11-04T19:46:24.484431"} +{"completed_at":"2026-11-04T19:46:28.580075","latency_in_ms":2051,"output":{"example_id":"chart/b263dc5d-en_p69","grounded_pages":[],"layout_pages":[],"markdown":" **67**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\nAs proficiency in literacy and numeracy is correlated (OECD, 2016[4]), adults with low proficiency in one domain are also likely to have low proficiency in other domains. Figure 2.5 shows the share of adults with low proficiency in all three domains. This share is highest in Chile (44%) and lowest in Japan (7%). Figure 2.5 also presents the percentage of low-performing adults in literacy and numeracy only. This share ranges between 48% in Chile and 8% in Japan.\n\n### **Figure 2.5. Share of adults who are low performing in more than one domain**\n\n16-65 year-olds scoring at or below Level 1 in more than one domain\n\nNote: Includes adults who were only administered the doorstep interview due to a language barrier (see Box 1.1 in Chapter 1). \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the ote for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the share of adults performing at or below Level 1 in all three domains._ Source: Table A.2.3 in Annex A.\n\n**_Aggregate inequality in the distribution of skills_**\n\nAggregate skills inequalities can be measured by the dispersion of the distribution of skills within each country or economy. One measure of such dispersion is the inter-decile range, i.e. the distance between the 90th percentile of the national skills distribution (the score below which 90% of adults perform) and the 10th percentile of the national skills distribution (the score below which 10% of adults perform).\n\nOn average, the inter-decile range in literacy in participating OECD countries and economies is 140 points (Figure 2.6, Panel A). Inequality is particularly pronounced in Singapore and the United States, where the inter-decile range exceeds 160 points. The distribution is more compressed in Lithuania and the Slovak Republic, where the inter-decile range is below 115 points.\n\nIn numeracy, the inter-decile range in participating OECD countries and economies averages 144 points (Figure 2.6, Panel B). The difference is widest in New Zealand, Portugal, Singapore and the United States, where the inter-decile range exceeds 160 points, while it is more compressed in Japan, Lithuania and the Slovak Republic, where the inter-decile range is below 130 points.\n\nIn adaptive problem solving, the inter-decile range averages 119 points (Figure 2.6, Panel C). Variation is widest in New Zealand and the United States, with the inter-decile range exceeding 140 points. The skills distribution is more compressed in Lithuania and the Slovak Republic, where the inter-decile range is lower than 100 points.\n\nAt or below e el 1 in literacy, numeracy and adapti e problem sol ing At or below e el 1 in literacy and numeracy\n\n0\n\n0\n\n3\n\n30\n\n2\n\n20\n\n1\n\n10\n\n0\n\nChile\n\nPortugal\n\nPoland\n\nItaly\n\nIsrael\n\nithuania\n\nUnited States\n\nHungary\n\nSpain\n\nFrance\n\norea\n\nat ia\n\nCroatia\n\new ealand\n\n**EC average**\n\nAustria\n\nSingapore\n\nC echia\n\nGermany\n\nSwit erland\n\nIreland\n\nSlo ak Republic\n\nEngland (U\n\nCanada\n\nFlemish Region ( E\n\nenmark\n\nEstonia\n\netherlands\n\norway\n\nFinland\n\nSweden\n\napan","pages":[{"line_numbers":[],"markdown":" **67**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\nAs proficiency in literacy and numeracy is correlated (OECD, 2016[4]), adults with low proficiency in one domain are also likely to have low proficiency in other domains. Figure 2.5 shows the share of adults with low proficiency in all three domains. This share is highest in Chile (44%) and lowest in Japan (7%). Figure 2.5 also presents the percentage of low-performing adults in literacy and numeracy only. This share ranges between 48% in Chile and 8% in Japan.\n\n### **Figure 2.5. Share of adults who are low performing in more than one domain**\n\n16-65 year-olds scoring at or below Level 1 in more than one domain\n\nNote: Includes adults who were only administered the doorstep interview due to a language barrier (see Box 1.1 in Chapter 1). \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the ote for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the share of adults performing at or below Level 1 in all three domains._ Source: Table A.2.3 in Annex A.\n\n**_Aggregate inequality in the distribution of skills_**\n\nAggregate skills inequalities can be measured by the dispersion of the distribution of skills within each country or economy. One measure of such dispersion is the inter-decile range, i.e. the distance between the 90th percentile of the national skills distribution (the score below which 90% of adults perform) and the 10th percentile of the national skills distribution (the score below which 10% of adults perform).\n\nOn average, the inter-decile range in literacy in participating OECD countries and economies is 140 points (Figure 2.6, Panel A). Inequality is particularly pronounced in Singapore and the United States, where the inter-decile range exceeds 160 points. The distribution is more compressed in Lithuania and the Slovak Republic, where the inter-decile range is below 115 points.\n\nIn numeracy, the inter-decile range in participating OECD countries and economies averages 144 points (Figure 2.6, Panel B). The difference is widest in New Zealand, Portugal, Singapore and the United States, where the inter-decile range exceeds 160 points, while it is more compressed in Japan, Lithuania and the Slovak Republic, where the inter-decile range is below 130 points.\n\nIn adaptive problem solving, the inter-decile range averages 119 points (Figure 2.6, Panel C). Variation is widest in New Zealand and the United States, with the inter-decile range exceeding 140 points. The skills distribution is more compressed in Lithuania and the Slovak Republic, where the inter-decile range is lower than 100 points.\n\nAt or below e el 1 in literacy, numeracy and adapti e problem sol ing At or below e el 1 in literacy and numeracy\n\n0\n\n0\n\n3\n\n30\n\n2\n\n20\n\n1\n\n10\n\n0\n\nChile\n\nPortugal\n\nPoland\n\nItaly\n\nIsrael\n\nithuania\n\nUnited States\n\nHungary\n\nSpain\n\nFrance\n\norea\n\nat ia\n\nCroatia\n\new ealand\n\n**EC average**\n\nAustria\n\nSingapore\n\nC echia\n\nGermany\n\nSwit erland\n\nIreland\n\nSlo ak Republic\n\nEngland (U\n\nCanada\n\nFlemish Region ( E\n\nenmark\n\nEstonia\n\netherlands\n\norway\n\nFinland\n\nSweden\n\napan","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":" **67**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\nAs proficiency in literacy and numeracy is correlated (OECD, 2016[4]), adults with low proficiency in one domain are also likely to have low proficiency in other domains. Figure 2.5 shows the share of adults with low proficiency in all three domains. This share is highest in Chile (44%) and lowest in Japan (7%). Figure 2.5 also presents the percentage of low-performing adults in literacy and numeracy only. This share ranges between 48% in Chile and 8% in Japan.\n\n### **Figure 2.5. Share of adults who are low performing in more than one domain**\n\n16-65 year-olds scoring at or below Level 1 in more than one domain\n\nNote: Includes adults who were only administered the doorstep interview due to a language barrier (see Box 1.1 in Chapter 1). \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the ote for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the share of adults performing at or below Level 1 in all three domains._ Source: Table A.2.3 in Annex A.\n\n**_Aggregate inequality in the distribution of skills_**\n\nAggregate skills inequalities can be measured by the dispersion of the distribution of skills within each country or economy. One measure of such dispersion is the inter-decile range, i.e. the distance between the 90th percentile of the national skills distribution (the score below which 90% of adults perform) and the 10th percentile of the national skills distribution (the score below which 10% of adults perform).\n\nOn average, the inter-decile range in literacy in participating OECD countries and economies is 140 points (Figure 2.6, Panel A). Inequality is particularly pronounced in Singapore and the United States, where the inter-decile range exceeds 160 points. The distribution is more compressed in Lithuania and the Slovak Republic, where the inter-decile range is below 115 points.\n\nIn numeracy, the inter-decile range in participating OECD countries and economies averages 144 points (Figure 2.6, Panel B). The difference is widest in New Zealand, Portugal, Singapore and the United States, where the inter-decile range exceeds 160 points, while it is more compressed in Japan, Lithuania and the Slovak Republic, where the inter-decile range is below 130 points.\n\nIn adaptive problem solving, the inter-decile range averages 119 points (Figure 2.6, Panel C). Variation is widest in New Zealand and the United States, with the inter-decile range exceeding 140 points. The skills distribution is more compressed in Lithuania and the Slovak Republic, where the inter-decile range is lower than 100 points.\n\nAt or below e el 1 in literacy, numeracy and adapti e problem sol ing At or below e el 1 in literacy and numeracy\n\n0\n\n0\n\n3\n\n30\n\n2\n\n20\n\n1\n\n10\n\n0\n\nChile\n\nPortugal\n\nPoland\n\nItaly\n\nIsrael\n\nithuania\n\nUnited States\n\nHungary\n\nSpain\n\nFrance\n\norea\n\nat ia\n\nCroatia\n\new ealand\n\n**EC average**\n\nAustria\n\nSingapore\n\nC echia\n\nGermany\n\nSwit erland\n\nIreland\n\nSlo ak Republic\n\nEngland (U\n\nCanada\n\nFlemish Region ( E\n\nenmark\n\nEstonia\n\netherlands\n\norway\n\nFinland\n\nSweden\n\napan","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p69.pdf"]},"response":{"errors":[],"extract_id":"extract_8366c03718a20c6827ac7d044d367b0c","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":[" **67**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\nAs proficiency in literacy and numeracy is correlated (OECD, 2016[4]), adults with low proficiency in one domain are also likely to have low proficiency in other domains. Figure 2.5 shows the share of adults with low proficiency in all three domains. This share is highest in Chile (44%) and lowest in Japan (7%). Figure 2.5 also presents the percentage of low-performing adults in literacy and numeracy only. This share ranges between 48% in Chile and 8% in Japan.\n\n### **Figure 2.5. Share of adults who are low performing in more than one domain**\n\n16-65 year-olds scoring at or below Level 1 in more than one domain\n\nNote: Includes adults who were only administered the doorstep interview due to a language barrier (see Box 1.1 in Chapter 1). \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the ote for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the share of adults performing at or below Level 1 in all three domains._ Source: Table A.2.3 in Annex A.\n\n**_Aggregate inequality in the distribution of skills_**\n\nAggregate skills inequalities can be measured by the dispersion of the distribution of skills within each country or economy. One measure of such dispersion is the inter-decile range, i.e. the distance between the 90th percentile of the national skills distribution (the score below which 90% of adults perform) and the 10th percentile of the national skills distribution (the score below which 10% of adults perform).\n\nOn average, the inter-decile range in literacy in participating OECD countries and economies is 140 points (Figure 2.6, Panel A). Inequality is particularly pronounced in Singapore and the United States, where the inter-decile range exceeds 160 points. The distribution is more compressed in Lithuania and the Slovak Republic, where the inter-decile range is below 115 points.\n\nIn numeracy, the inter-decile range in participating OECD countries and economies averages 144 points (Figure 2.6, Panel B). The difference is widest in New Zealand, Portugal, Singapore and the United States, where the inter-decile range exceeds 160 points, while it is more compressed in Japan, Lithuania and the Slovak Republic, where the inter-decile range is below 130 points.\n\nIn adaptive problem solving, the inter-decile range averages 119 points (Figure 2.6, Panel C). Variation is widest in New Zealand and the United States, with the inter-decile range exceeding 140 points. The skills distribution is more compressed in Lithuania and the Slovak Republic, where the inter-decile range is lower than 100 points.\n\nAt or below e el 1 in literacy, numeracy and adapti e problem sol ing At or below e el 1 in literacy and numeracy\n\n0\n\n0\n\n3\n\n30\n\n2\n\n20\n\n1\n\n10\n\n0\n\nChile\n\nPortugal\n\nPoland\n\nItaly\n\nIsrael\n\nithuania\n\nUnited States\n\nHungary\n\nSpain\n\nFrance\n\norea\n\nat ia\n\nCroatia\n\new ealand\n\n**EC average**\n\nAustria\n\nSingapore\n\nC echia\n\nGermany\n\nSwit erland\n\nIreland\n\nSlo ak Republic\n\nEngland (U\n\nCanada\n\nFlemish Region ( E\n\nenmark\n\nEstonia\n\netherlands\n\norway\n\nFinland\n\nSweden\n\napan"],"full_content":" **67**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\nAs proficiency in literacy and numeracy is correlated (OECD, 2016[4]), adults with low proficiency in one domain are also likely to have low proficiency in other domains. Figure 2.5 shows the share of adults with low proficiency in all three domains. This share is highest in Chile (44%) and lowest in Japan (7%). Figure 2.5 also presents the percentage of low-performing adults in literacy and numeracy only. This share ranges between 48% in Chile and 8% in Japan.\n\n### **Figure 2.5. Share of adults who are low performing in more than one domain**\n\n16-65 year-olds scoring at or below Level 1 in more than one domain\n\nNote: Includes adults who were only administered the doorstep interview due to a language barrier (see Box 1.1 in Chapter 1). \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the ote for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the share of adults performing at or below Level 1 in all three domains._ Source: Table A.2.3 in Annex A.\n\n**_Aggregate inequality in the distribution of skills_**\n\nAggregate skills inequalities can be measured by the dispersion of the distribution of skills within each country or economy. One measure of such dispersion is the inter-decile range, i.e. the distance between the 90th percentile of the national skills distribution (the score below which 90% of adults perform) and the 10th percentile of the national skills distribution (the score below which 10% of adults perform).\n\nOn average, the inter-decile range in literacy in participating OECD countries and economies is 140 points (Figure 2.6, Panel A). Inequality is particularly pronounced in Singapore and the United States, where the inter-decile range exceeds 160 points. The distribution is more compressed in Lithuania and the Slovak Republic, where the inter-decile range is below 115 points.\n\nIn numeracy, the inter-decile range in participating OECD countries and economies averages 144 points (Figure 2.6, Panel B). The difference is widest in New Zealand, Portugal, Singapore and the United States, where the inter-decile range exceeds 160 points, while it is more compressed in Japan, Lithuania and the Slovak Republic, where the inter-decile range is below 130 points.\n\nIn adaptive problem solving, the inter-decile range averages 119 points (Figure 2.6, Panel C). Variation is widest in New Zealand and the United States, with the inter-decile range exceeding 140 points. The skills distribution is more compressed in Lithuania and the Slovak Republic, where the inter-decile range is lower than 100 points.\n\nAt or below e el 1 in literacy, numeracy and adapti e problem sol ing At or below e el 1 in literacy and numeracy\n\n0\n\n0\n\n3\n\n30\n\n2\n\n20\n\n1\n\n10\n\n0\n\nChile\n\nPortugal\n\nPoland\n\nItaly\n\nIsrael\n\nithuania\n\nUnited States\n\nHungary\n\nSpain\n\nFrance\n\norea\n\nat ia\n\nCroatia\n\new ealand\n\n**EC average**\n\nAustria\n\nSingapore\n\nC echia\n\nGermany\n\nSwit erland\n\nIreland\n\nSlo ak Republic\n\nEngland (U\n\nCanada\n\nFlemish Region ( E\n\nenmark\n\nEstonia\n\netherlands\n\norway\n\nFinland\n\nSweden\n\napan","publish_date":null,"title":"Figure 2.5. Share of adults who are low performing in more than one domain","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p69.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p69.pdf"},"request":{"example_id":"chart/b263dc5d-en_p69","product_type":"parse","source_file_path":"b263dc5d-en_p69.pdf"},"started_at":"2026-11-04T19:46:26.528906"} +{"completed_at":"2026-11-04T19:46:31.839827","latency_in_ms":4360,"output":{"example_id":"chart/b263dc5d-en_p70","grounded_pages":[],"layout_pages":[],"markdown":"**68** \n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 2.6. Inequality in the distribution of key information-processing skills**\n\nDifference between the 90th and 10th percentile of the national skills distribution for literacy, numeracy and adaptive problem solving (90th percentile _minus_ 10th percentile)\n\nNote: Adults aged 16-65; includes adults who were only administered the doorstep interview due to a language barrier (see Box 1.1 in Chapter 1). \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the Note for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the difference between the 90_ _th_ _and the 10_ _th_ _percentile._ Source: Table A.2.1 (L, N, A) in Annex A.\n\nUnited States\n\nSingapore\n\new ealand\n\nGermany\n\nIsrael\n\nFinland\n\nSwit erland\n\nAustria\n\nFlemish Region ( E\n\nEstonia\n\nFrance\n\nChile\n\nenmark\n\nPortugal\n\netherlands\n\norway\n\n**EC average**\n\nCroatia\n\nC echia\n\nat ia\n\nCanada\n\nHungary\n\nItaly\n\nEngland (U\n\norea\n\nPoland\n\nIreland\n\nSweden\n\napan\n\nSpain\n\nithuania\n\nSlo ak Republic\n\nUnited States\n\nSingapore\n\nPortugal\n\new ealand\n\nGermany\n\nChile\n\nFrance\n\nSwit erland\n\nIsrael\n\nFlemish Region ( E\n\netherlands\n\nAustria\n\nenmark\n\nEngland (U\n\nFinland\n\nHungary\n\n**EC average**\n\nItaly\n\nCanada\n\nEstonia\n\nat ia\n\nC echia\n\nIreland\n\norway\n\nPoland\n\nSpain\n\norea\n\nSweden\n\nCroatia\n\napan\n\nSlo ak Republic\n\nithuania\n\new ealand\n\nUnited States\n\nSwit erland\n\nSingapore\n\nenmark\n\nIsrael\n\nGermany\n\nFrance\n\nFlemish Region ( E\n\nChile\n\nAustria\n\netherlands\n\nFinland\n\nPortugal\n\nEstonia\n\nEngland (U\n\n**EC average**\n\nat ia\n\nHungary\n\nCanada\n\nC echia\n\nCroatia\n\norway\n\nSpain\n\nIreland\n\nItaly\n\nPoland\n\nSweden\n\norea\n\napan\n\nithuania\n\nSlo ak Republic\n\nScore point difference\n\nScore point difference\n\nScore point difference\n\n**. iteracy**\n\n180\n\n1 0\n\n1 0\n\n120\n\n100\n\n80\n\n180\n\n1 0\n\n1 0\n\n120\n\n100\n\n80\n\n180\n\n1 0\n\n1 0\n\n120\n\n100\n\n80\n\n**B. umeracy**\n\n**C. daptive problem solving**","pages":[{"line_numbers":[],"markdown":"**68** \n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 2.6. Inequality in the distribution of key information-processing skills**\n\nDifference between the 90th and 10th percentile of the national skills distribution for literacy, numeracy and adaptive problem solving (90th percentile _minus_ 10th percentile)\n\nNote: Adults aged 16-65; includes adults who were only administered the doorstep interview due to a language barrier (see Box 1.1 in Chapter 1). \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the Note for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the difference between the 90_ _th_ _and the 10_ _th_ _percentile._ Source: Table A.2.1 (L, N, A) in Annex A.\n\nUnited States\n\nSingapore\n\new ealand\n\nGermany\n\nIsrael\n\nFinland\n\nSwit erland\n\nAustria\n\nFlemish Region ( E\n\nEstonia\n\nFrance\n\nChile\n\nenmark\n\nPortugal\n\netherlands\n\norway\n\n**EC average**\n\nCroatia\n\nC echia\n\nat ia\n\nCanada\n\nHungary\n\nItaly\n\nEngland (U\n\norea\n\nPoland\n\nIreland\n\nSweden\n\napan\n\nSpain\n\nithuania\n\nSlo ak Republic\n\nUnited States\n\nSingapore\n\nPortugal\n\new ealand\n\nGermany\n\nChile\n\nFrance\n\nSwit erland\n\nIsrael\n\nFlemish Region ( E\n\netherlands\n\nAustria\n\nenmark\n\nEngland (U\n\nFinland\n\nHungary\n\n**EC average**\n\nItaly\n\nCanada\n\nEstonia\n\nat ia\n\nC echia\n\nIreland\n\norway\n\nPoland\n\nSpain\n\norea\n\nSweden\n\nCroatia\n\napan\n\nSlo ak Republic\n\nithuania\n\new ealand\n\nUnited States\n\nSwit erland\n\nSingapore\n\nenmark\n\nIsrael\n\nGermany\n\nFrance\n\nFlemish Region ( E\n\nChile\n\nAustria\n\netherlands\n\nFinland\n\nPortugal\n\nEstonia\n\nEngland (U\n\n**EC average**\n\nat ia\n\nHungary\n\nCanada\n\nC echia\n\nCroatia\n\norway\n\nSpain\n\nIreland\n\nItaly\n\nPoland\n\nSweden\n\norea\n\napan\n\nithuania\n\nSlo ak Republic\n\nScore point difference\n\nScore point difference\n\nScore point difference\n\n**. iteracy**\n\n180\n\n1 0\n\n1 0\n\n120\n\n100\n\n80\n\n180\n\n1 0\n\n1 0\n\n120\n\n100\n\n80\n\n180\n\n1 0\n\n1 0\n\n120\n\n100\n\n80\n\n**B. umeracy**\n\n**C. daptive problem solving**","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**68** \n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 2.6. Inequality in the distribution of key information-processing skills**\n\nDifference between the 90th and 10th percentile of the national skills distribution for literacy, numeracy and adaptive problem solving (90th percentile _minus_ 10th percentile)\n\nNote: Adults aged 16-65; includes adults who were only administered the doorstep interview due to a language barrier (see Box 1.1 in Chapter 1). \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the Note for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the difference between the 90_ _th_ _and the 10_ _th_ _percentile._ Source: Table A.2.1 (L, N, A) in Annex A.\n\nUnited States\n\nSingapore\n\new ealand\n\nGermany\n\nIsrael\n\nFinland\n\nSwit erland\n\nAustria\n\nFlemish Region ( E\n\nEstonia\n\nFrance\n\nChile\n\nenmark\n\nPortugal\n\netherlands\n\norway\n\n**EC average**\n\nCroatia\n\nC echia\n\nat ia\n\nCanada\n\nHungary\n\nItaly\n\nEngland (U\n\norea\n\nPoland\n\nIreland\n\nSweden\n\napan\n\nSpain\n\nithuania\n\nSlo ak Republic\n\nUnited States\n\nSingapore\n\nPortugal\n\new ealand\n\nGermany\n\nChile\n\nFrance\n\nSwit erland\n\nIsrael\n\nFlemish Region ( E\n\netherlands\n\nAustria\n\nenmark\n\nEngland (U\n\nFinland\n\nHungary\n\n**EC average**\n\nItaly\n\nCanada\n\nEstonia\n\nat ia\n\nC echia\n\nIreland\n\norway\n\nPoland\n\nSpain\n\norea\n\nSweden\n\nCroatia\n\napan\n\nSlo ak Republic\n\nithuania\n\new ealand\n\nUnited States\n\nSwit erland\n\nSingapore\n\nenmark\n\nIsrael\n\nGermany\n\nFrance\n\nFlemish Region ( E\n\nChile\n\nAustria\n\netherlands\n\nFinland\n\nPortugal\n\nEstonia\n\nEngland (U\n\n**EC average**\n\nat ia\n\nHungary\n\nCanada\n\nC echia\n\nCroatia\n\norway\n\nSpain\n\nIreland\n\nItaly\n\nPoland\n\nSweden\n\norea\n\napan\n\nithuania\n\nSlo ak Republic\n\nScore point difference\n\nScore point difference\n\nScore point difference\n\n**. iteracy**\n\n180\n\n1 0\n\n1 0\n\n120\n\n100\n\n80\n\n180\n\n1 0\n\n1 0\n\n120\n\n100\n\n80\n\n180\n\n1 0\n\n1 0\n\n120\n\n100\n\n80\n\n**B. umeracy**\n\n**C. daptive problem solving**","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p70.pdf"]},"response":{"errors":[],"extract_id":"extract_5cf0c7a9cb5d1a7687a7dee54e139f1d","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**68** \n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 2.6. Inequality in the distribution of key information-processing skills**\n\nDifference between the 90th and 10th percentile of the national skills distribution for literacy, numeracy and adaptive problem solving (90th percentile _minus_ 10th percentile)\n\nNote: Adults aged 16-65; includes adults who were only administered the doorstep interview due to a language barrier (see Box 1.1 in Chapter 1). \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the Note for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the difference between the 90_ _th_ _and the 10_ _th_ _percentile._ Source: Table A.2.1 (L, N, A) in Annex A.\n\nUnited States\n\nSingapore\n\new ealand\n\nGermany\n\nIsrael\n\nFinland\n\nSwit erland\n\nAustria\n\nFlemish Region ( E\n\nEstonia\n\nFrance\n\nChile\n\nenmark\n\nPortugal\n\netherlands\n\norway\n\n**EC average**\n\nCroatia\n\nC echia\n\nat ia\n\nCanada\n\nHungary\n\nItaly\n\nEngland (U\n\norea\n\nPoland\n\nIreland\n\nSweden\n\napan\n\nSpain\n\nithuania\n\nSlo ak Republic\n\nUnited States\n\nSingapore\n\nPortugal\n\new ealand\n\nGermany\n\nChile\n\nFrance\n\nSwit erland\n\nIsrael\n\nFlemish Region ( E\n\netherlands\n\nAustria\n\nenmark\n\nEngland (U\n\nFinland\n\nHungary\n\n**EC average**\n\nItaly\n\nCanada\n\nEstonia\n\nat ia\n\nC echia\n\nIreland\n\norway\n\nPoland\n\nSpain\n\norea\n\nSweden\n\nCroatia\n\napan\n\nSlo ak Republic\n\nithuania\n\new ealand\n\nUnited States\n\nSwit erland\n\nSingapore\n\nenmark\n\nIsrael\n\nGermany\n\nFrance\n\nFlemish Region ( E\n\nChile\n\nAustria\n\netherlands\n\nFinland\n\nPortugal\n\nEstonia\n\nEngland (U\n\n**EC average**\n\nat ia\n\nHungary\n\nCanada\n\nC echia\n\nCroatia\n\norway\n\nSpain\n\nIreland\n\nItaly\n\nPoland\n\nSweden\n\norea\n\napan\n\nithuania\n\nSlo ak Republic\n\nScore point difference\n\nScore point difference\n\nScore point difference\n\n**. iteracy**\n\n180\n\n1 0\n\n1 0\n\n120\n\n100\n\n80\n\n180\n\n1 0\n\n1 0\n\n120\n\n100\n\n80\n\n180\n\n1 0\n\n1 0\n\n120\n\n100\n\n80\n\n**B. umeracy**\n\n**C. daptive problem solving**"],"full_content":"**68** \n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 2.6. Inequality in the distribution of key information-processing skills**\n\nDifference between the 90th and 10th percentile of the national skills distribution for literacy, numeracy and adaptive problem solving (90th percentile _minus_ 10th percentile)\n\nNote: Adults aged 16-65; includes adults who were only administered the doorstep interview due to a language barrier (see Box 1.1 in Chapter 1). \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the Note for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the difference between the 90_ _th_ _and the 10_ _th_ _percentile._ Source: Table A.2.1 (L, N, A) in Annex A.\n\nUnited States\n\nSingapore\n\new ealand\n\nGermany\n\nIsrael\n\nFinland\n\nSwit erland\n\nAustria\n\nFlemish Region ( E\n\nEstonia\n\nFrance\n\nChile\n\nenmark\n\nPortugal\n\netherlands\n\norway\n\n**EC average**\n\nCroatia\n\nC echia\n\nat ia\n\nCanada\n\nHungary\n\nItaly\n\nEngland (U\n\norea\n\nPoland\n\nIreland\n\nSweden\n\napan\n\nSpain\n\nithuania\n\nSlo ak Republic\n\nUnited States\n\nSingapore\n\nPortugal\n\new ealand\n\nGermany\n\nChile\n\nFrance\n\nSwit erland\n\nIsrael\n\nFlemish Region ( E\n\netherlands\n\nAustria\n\nenmark\n\nEngland (U\n\nFinland\n\nHungary\n\n**EC average**\n\nItaly\n\nCanada\n\nEstonia\n\nat ia\n\nC echia\n\nIreland\n\norway\n\nPoland\n\nSpain\n\norea\n\nSweden\n\nCroatia\n\napan\n\nSlo ak Republic\n\nithuania\n\new ealand\n\nUnited States\n\nSwit erland\n\nSingapore\n\nenmark\n\nIsrael\n\nGermany\n\nFrance\n\nFlemish Region ( E\n\nChile\n\nAustria\n\netherlands\n\nFinland\n\nPortugal\n\nEstonia\n\nEngland (U\n\n**EC average**\n\nat ia\n\nHungary\n\nCanada\n\nC echia\n\nCroatia\n\norway\n\nSpain\n\nIreland\n\nItaly\n\nPoland\n\nSweden\n\norea\n\napan\n\nithuania\n\nSlo ak Republic\n\nScore point difference\n\nScore point difference\n\nScore point difference\n\n**. iteracy**\n\n180\n\n1 0\n\n1 0\n\n120\n\n100\n\n80\n\n180\n\n1 0\n\n1 0\n\n120\n\n100\n\n80\n\n180\n\n1 0\n\n1 0\n\n120\n\n100\n\n80\n\n**B. umeracy**\n\n**C. daptive problem solving**","publish_date":null,"title":"Figure 2.6. Inequality in the distribution of key information-processing skills","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p70.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p70.pdf"},"request":{"example_id":"chart/b263dc5d-en_p70","product_type":"parse","source_file_path":"b263dc5d-en_p70.pdf"},"started_at":"2026-11-04T19:46:27.479152"} +{"completed_at":"2026-11-04T19:46:30.859815","latency_in_ms":1848,"output":{"example_id":"chart/b263dc5d-en_p78","grounded_pages":[],"layout_pages":[],"markdown":"**76** \n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Box 2.3. Field of study choices and proficiency in numeracy among tertiary-educated adults**\n\nIndividuals can choose between various fields in their tertiary education. ¹ Over the past decades, technological progress has increased the demand for adults with the right mix of skills to work in technology- rich environments. This has raised the economic returns of obtaining degrees in science, technology, engineering and mathematics (STEM) (OECD, 2022[13]). ² Several countries, including Denmark and the United States, have launched initiatives to increase the number of students enrolled in STEM programmes; encourage learners’ interest in such fields; and foster diversity, equity and inclusion in STEM fields (OECD, n.d.[28]; Office of Science and Technology Policy, 2021[29]).\n\nOn average across OECD countries and economies, 27% of tertiary-educated adults have a STEM degree. This share varies from more than 36% in Germany and Singapore to around 21% in the Netherlands (Figure 2.10).\n\nAdults’ field of study is correlated with differences in numeracy proficiency. This can be attributed both to the fact that access and success to STEM programmes require a high level of numeracy skills, and to the fact that STEM programmes (and employment in STEM sectors) help develop such skills. Among tertiary- educated 25-65 year-olds, the average numeracy proficiency of those who studied in a STEM field is 304 points compared to 285 points for those who studied in non-STEM fields (Figure 2.11).\n\nSignificant differences in average numeracy proficiency between adults with a STEM and a non-STEM tertiary education range from 28 points in Japan to 5 points in Croatia (see Table A.2.6 (N) in Annex A). For other domains, adults who studied a STEM field have an average advantage of 6 points in literacy and 10 points in adaptive problem solving (see Table A.2.6 (L) and Table A.2.6 (A) in Annex A).\n\n### **Figure 2.10. Share of tertiary-educated adults who studied STEM fields**\n\nShare of tertiary-educated 25-65 year-olds who obtained their highest qualification in a science, technology, engineering or mathematics (STEM) field\n\nNote: Does not include adults who were only administered the doorstep interview due to a language barrier, as information on field of study was not collected for those respondents (see Box 1.1 in Chapter 1). STEM stands for science, technology, engineering and mathematics. Educational attainment is based on the International Standard Classification of Education (ISCED) 2011, grouped into below upper secondary (ISCED 1, 2 and 3 short), upper secondary (ISCED 3 and 4) and tertiary (ISCED 5, 6, 7 and 8). Where possible, foreign qualifications are included as the closest corresponding level in the respective national education systems. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the Note for Poland in the Reader’s Guide. C _ountries and economies are ranked in descending order based on the percentage of tertiary-educated adults who studied STEM fields_ . Source: Table B.3.2 in Annex B.\n\n0\n\n3\n\n30\n\n2\n\n20\n\n1\n\n10\n\n0\n\nSingapore\n\nGermany\n\nAustria\n\norea\n\nC echia\n\nIsrael\n\nCanada\n\nSwit erland\n\new ealand\n\nSweden\n\nEstonia\n\nenmark\n\nSpain\n\nFinland\n\n**EC average**\n\nItaly\n\nFrance\n\nIreland\n\nUnited States\n\nEngland (U\n\norway\n\nHungary\n\nithuania\n\nChile\n\nPoland\n\nPortugal\n\nat ia\n\nFlemish Region ( E\n\nCroatia\n\nSlo ak Republic\n\napan\n\netherlands","pages":[{"line_numbers":[],"markdown":"**76** \n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Box 2.3. Field of study choices and proficiency in numeracy among tertiary-educated adults**\n\nIndividuals can choose between various fields in their tertiary education. ¹ Over the past decades, technological progress has increased the demand for adults with the right mix of skills to work in technology- rich environments. This has raised the economic returns of obtaining degrees in science, technology, engineering and mathematics (STEM) (OECD, 2022[13]). ² Several countries, including Denmark and the United States, have launched initiatives to increase the number of students enrolled in STEM programmes; encourage learners’ interest in such fields; and foster diversity, equity and inclusion in STEM fields (OECD, n.d.[28]; Office of Science and Technology Policy, 2021[29]).\n\nOn average across OECD countries and economies, 27% of tertiary-educated adults have a STEM degree. This share varies from more than 36% in Germany and Singapore to around 21% in the Netherlands (Figure 2.10).\n\nAdults’ field of study is correlated with differences in numeracy proficiency. This can be attributed both to the fact that access and success to STEM programmes require a high level of numeracy skills, and to the fact that STEM programmes (and employment in STEM sectors) help develop such skills. Among tertiary- educated 25-65 year-olds, the average numeracy proficiency of those who studied in a STEM field is 304 points compared to 285 points for those who studied in non-STEM fields (Figure 2.11).\n\nSignificant differences in average numeracy proficiency between adults with a STEM and a non-STEM tertiary education range from 28 points in Japan to 5 points in Croatia (see Table A.2.6 (N) in Annex A). For other domains, adults who studied a STEM field have an average advantage of 6 points in literacy and 10 points in adaptive problem solving (see Table A.2.6 (L) and Table A.2.6 (A) in Annex A).\n\n### **Figure 2.10. Share of tertiary-educated adults who studied STEM fields**\n\nShare of tertiary-educated 25-65 year-olds who obtained their highest qualification in a science, technology, engineering or mathematics (STEM) field\n\nNote: Does not include adults who were only administered the doorstep interview due to a language barrier, as information on field of study was not collected for those respondents (see Box 1.1 in Chapter 1). STEM stands for science, technology, engineering and mathematics. Educational attainment is based on the International Standard Classification of Education (ISCED) 2011, grouped into below upper secondary (ISCED 1, 2 and 3 short), upper secondary (ISCED 3 and 4) and tertiary (ISCED 5, 6, 7 and 8). Where possible, foreign qualifications are included as the closest corresponding level in the respective national education systems. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the Note for Poland in the Reader’s Guide. C _ountries and economies are ranked in descending order based on the percentage of tertiary-educated adults who studied STEM fields_ . Source: Table B.3.2 in Annex B.\n\n0\n\n3\n\n30\n\n2\n\n20\n\n1\n\n10\n\n0\n\nSingapore\n\nGermany\n\nAustria\n\norea\n\nC echia\n\nIsrael\n\nCanada\n\nSwit erland\n\new ealand\n\nSweden\n\nEstonia\n\nenmark\n\nSpain\n\nFinland\n\n**EC average**\n\nItaly\n\nFrance\n\nIreland\n\nUnited States\n\nEngland (U\n\norway\n\nHungary\n\nithuania\n\nChile\n\nPoland\n\nPortugal\n\nat ia\n\nFlemish Region ( E\n\nCroatia\n\nSlo ak Republic\n\napan\n\netherlands","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**76** \n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Box 2.3. Field of study choices and proficiency in numeracy among tertiary-educated adults**\n\nIndividuals can choose between various fields in their tertiary education. ¹ Over the past decades, technological progress has increased the demand for adults with the right mix of skills to work in technology- rich environments. This has raised the economic returns of obtaining degrees in science, technology, engineering and mathematics (STEM) (OECD, 2022[13]). ² Several countries, including Denmark and the United States, have launched initiatives to increase the number of students enrolled in STEM programmes; encourage learners’ interest in such fields; and foster diversity, equity and inclusion in STEM fields (OECD, n.d.[28]; Office of Science and Technology Policy, 2021[29]).\n\nOn average across OECD countries and economies, 27% of tertiary-educated adults have a STEM degree. This share varies from more than 36% in Germany and Singapore to around 21% in the Netherlands (Figure 2.10).\n\nAdults’ field of study is correlated with differences in numeracy proficiency. This can be attributed both to the fact that access and success to STEM programmes require a high level of numeracy skills, and to the fact that STEM programmes (and employment in STEM sectors) help develop such skills. Among tertiary- educated 25-65 year-olds, the average numeracy proficiency of those who studied in a STEM field is 304 points compared to 285 points for those who studied in non-STEM fields (Figure 2.11).\n\nSignificant differences in average numeracy proficiency between adults with a STEM and a non-STEM tertiary education range from 28 points in Japan to 5 points in Croatia (see Table A.2.6 (N) in Annex A). For other domains, adults who studied a STEM field have an average advantage of 6 points in literacy and 10 points in adaptive problem solving (see Table A.2.6 (L) and Table A.2.6 (A) in Annex A).\n\n### **Figure 2.10. Share of tertiary-educated adults who studied STEM fields**\n\nShare of tertiary-educated 25-65 year-olds who obtained their highest qualification in a science, technology, engineering or mathematics (STEM) field\n\nNote: Does not include adults who were only administered the doorstep interview due to a language barrier, as information on field of study was not collected for those respondents (see Box 1.1 in Chapter 1). STEM stands for science, technology, engineering and mathematics. Educational attainment is based on the International Standard Classification of Education (ISCED) 2011, grouped into below upper secondary (ISCED 1, 2 and 3 short), upper secondary (ISCED 3 and 4) and tertiary (ISCED 5, 6, 7 and 8). Where possible, foreign qualifications are included as the closest corresponding level in the respective national education systems. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the Note for Poland in the Reader’s Guide. C _ountries and economies are ranked in descending order based on the percentage of tertiary-educated adults who studied STEM fields_ . Source: Table B.3.2 in Annex B.\n\n0\n\n3\n\n30\n\n2\n\n20\n\n1\n\n10\n\n0\n\nSingapore\n\nGermany\n\nAustria\n\norea\n\nC echia\n\nIsrael\n\nCanada\n\nSwit erland\n\new ealand\n\nSweden\n\nEstonia\n\nenmark\n\nSpain\n\nFinland\n\n**EC average**\n\nItaly\n\nFrance\n\nIreland\n\nUnited States\n\nEngland (U\n\norway\n\nHungary\n\nithuania\n\nChile\n\nPoland\n\nPortugal\n\nat ia\n\nFlemish Region ( E\n\nCroatia\n\nSlo ak Republic\n\napan\n\netherlands","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p78.pdf"]},"response":{"errors":[],"extract_id":"extract_6149c0b197142394ee6a5e5ff6896667","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**76** \n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Box 2.3. Field of study choices and proficiency in numeracy among tertiary-educated adults**\n\nIndividuals can choose between various fields in their tertiary education. ¹ Over the past decades, technological progress has increased the demand for adults with the right mix of skills to work in technology- rich environments. This has raised the economic returns of obtaining degrees in science, technology, engineering and mathematics (STEM) (OECD, 2022[13]). ² Several countries, including Denmark and the United States, have launched initiatives to increase the number of students enrolled in STEM programmes; encourage learners’ interest in such fields; and foster diversity, equity and inclusion in STEM fields (OECD, n.d.[28]; Office of Science and Technology Policy, 2021[29]).\n\nOn average across OECD countries and economies, 27% of tertiary-educated adults have a STEM degree. This share varies from more than 36% in Germany and Singapore to around 21% in the Netherlands (Figure 2.10).\n\nAdults’ field of study is correlated with differences in numeracy proficiency. This can be attributed both to the fact that access and success to STEM programmes require a high level of numeracy skills, and to the fact that STEM programmes (and employment in STEM sectors) help develop such skills. Among tertiary- educated 25-65 year-olds, the average numeracy proficiency of those who studied in a STEM field is 304 points compared to 285 points for those who studied in non-STEM fields (Figure 2.11).\n\nSignificant differences in average numeracy proficiency between adults with a STEM and a non-STEM tertiary education range from 28 points in Japan to 5 points in Croatia (see Table A.2.6 (N) in Annex A). For other domains, adults who studied a STEM field have an average advantage of 6 points in literacy and 10 points in adaptive problem solving (see Table A.2.6 (L) and Table A.2.6 (A) in Annex A).\n\n### **Figure 2.10. Share of tertiary-educated adults who studied STEM fields**\n\nShare of tertiary-educated 25-65 year-olds who obtained their highest qualification in a science, technology, engineering or mathematics (STEM) field\n\nNote: Does not include adults who were only administered the doorstep interview due to a language barrier, as information on field of study was not collected for those respondents (see Box 1.1 in Chapter 1). STEM stands for science, technology, engineering and mathematics. Educational attainment is based on the International Standard Classification of Education (ISCED) 2011, grouped into below upper secondary (ISCED 1, 2 and 3 short), upper secondary (ISCED 3 and 4) and tertiary (ISCED 5, 6, 7 and 8). Where possible, foreign qualifications are included as the closest corresponding level in the respective national education systems. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the Note for Poland in the Reader’s Guide. C _ountries and economies are ranked in descending order based on the percentage of tertiary-educated adults who studied STEM fields_ . Source: Table B.3.2 in Annex B.\n\n0\n\n3\n\n30\n\n2\n\n20\n\n1\n\n10\n\n0\n\nSingapore\n\nGermany\n\nAustria\n\norea\n\nC echia\n\nIsrael\n\nCanada\n\nSwit erland\n\new ealand\n\nSweden\n\nEstonia\n\nenmark\n\nSpain\n\nFinland\n\n**EC average**\n\nItaly\n\nFrance\n\nIreland\n\nUnited States\n\nEngland (U\n\norway\n\nHungary\n\nithuania\n\nChile\n\nPoland\n\nPortugal\n\nat ia\n\nFlemish Region ( E\n\nCroatia\n\nSlo ak Republic\n\napan\n\netherlands"],"full_content":"**76** \n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Box 2.3. Field of study choices and proficiency in numeracy among tertiary-educated adults**\n\nIndividuals can choose between various fields in their tertiary education. ¹ Over the past decades, technological progress has increased the demand for adults with the right mix of skills to work in technology- rich environments. This has raised the economic returns of obtaining degrees in science, technology, engineering and mathematics (STEM) (OECD, 2022[13]). ² Several countries, including Denmark and the United States, have launched initiatives to increase the number of students enrolled in STEM programmes; encourage learners’ interest in such fields; and foster diversity, equity and inclusion in STEM fields (OECD, n.d.[28]; Office of Science and Technology Policy, 2021[29]).\n\nOn average across OECD countries and economies, 27% of tertiary-educated adults have a STEM degree. This share varies from more than 36% in Germany and Singapore to around 21% in the Netherlands (Figure 2.10).\n\nAdults’ field of study is correlated with differences in numeracy proficiency. This can be attributed both to the fact that access and success to STEM programmes require a high level of numeracy skills, and to the fact that STEM programmes (and employment in STEM sectors) help develop such skills. Among tertiary- educated 25-65 year-olds, the average numeracy proficiency of those who studied in a STEM field is 304 points compared to 285 points for those who studied in non-STEM fields (Figure 2.11).\n\nSignificant differences in average numeracy proficiency between adults with a STEM and a non-STEM tertiary education range from 28 points in Japan to 5 points in Croatia (see Table A.2.6 (N) in Annex A). For other domains, adults who studied a STEM field have an average advantage of 6 points in literacy and 10 points in adaptive problem solving (see Table A.2.6 (L) and Table A.2.6 (A) in Annex A).\n\n### **Figure 2.10. Share of tertiary-educated adults who studied STEM fields**\n\nShare of tertiary-educated 25-65 year-olds who obtained their highest qualification in a science, technology, engineering or mathematics (STEM) field\n\nNote: Does not include adults who were only administered the doorstep interview due to a language barrier, as information on field of study was not collected for those respondents (see Box 1.1 in Chapter 1). STEM stands for science, technology, engineering and mathematics. Educational attainment is based on the International Standard Classification of Education (ISCED) 2011, grouped into below upper secondary (ISCED 1, 2 and 3 short), upper secondary (ISCED 3 and 4) and tertiary (ISCED 5, 6, 7 and 8). Where possible, foreign qualifications are included as the closest corresponding level in the respective national education systems. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the Note for Poland in the Reader’s Guide. C _ountries and economies are ranked in descending order based on the percentage of tertiary-educated adults who studied STEM fields_ . Source: Table B.3.2 in Annex B.\n\n0\n\n3\n\n30\n\n2\n\n20\n\n1\n\n10\n\n0\n\nSingapore\n\nGermany\n\nAustria\n\norea\n\nC echia\n\nIsrael\n\nCanada\n\nSwit erland\n\new ealand\n\nSweden\n\nEstonia\n\nenmark\n\nSpain\n\nFinland\n\n**EC average**\n\nItaly\n\nFrance\n\nIreland\n\nUnited States\n\nEngland (U\n\norway\n\nHungary\n\nithuania\n\nChile\n\nPoland\n\nPortugal\n\nat ia\n\nFlemish Region ( E\n\nCroatia\n\nSlo ak Republic\n\napan\n\netherlands","publish_date":null,"title":"Box 2.3. Field of study choices and proficiency in numeracy among tertiary-educated adults Figure 2.10. Share of tertiary-educated adults who","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p78.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p78.pdf"},"request":{"example_id":"chart/b263dc5d-en_p78","product_type":"parse","source_file_path":"b263dc5d-en_p78.pdf"},"started_at":"2026-11-04T19:46:29.011159"} +{"completed_at":"2026-11-04T19:46:32.578184","latency_in_ms":1716,"output":{"example_id":"chart/b263dc5d-en_p79","grounded_pages":[],"layout_pages":[],"markdown":" **77**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 2.11. Average numeracy proficiency among tertiary-educated adults, by field of study**\n\nNote: Adults aged 25-65; does not include adults who were only administered the doorstep interview due to a language barrier, as information on field of study was not collected for those respondents (see Box 1.1 in Chapter 1). STEM stands for science, technology, engineering and mathematics. Educational attainment is based on the International Standard Classification of Education (ISCED) 2011, grouped into below upper secondary (ISCED 1, 2 and 3 short), upper secondary (ISCED 3 and 4) and tertiary (ISCED 5, 6, 7 and 8). Where possible, foreign qualifications are included as the closest corresponding level in the respective national education systems. Darker colours denote differences that are statistically significant at the 5% level. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the ote for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the proficiency of tertiary-educated adults who studied STEM fields._ Source: Table A.2.6 (N) in Annex A. 1. Within the International Standard Classification of Education (ISCED), programmes and related qualifications can be classified by the field of education and training as well as the level. The 2023 Survey of Adult Skills distinguishes between the following fields of study: economics, business and administration; law; health; welfare; social and behavioural sciences; journalism and information; information and communication technologies (ICT); natural sciences, mathematics and statistics; engineering and manufacturing; construction; agriculture, forestry, fisheries and environmental studies; personal and community services; security and transport; education and teacher training; humanities, languages and arts; and no main area of study or emphasis (if it was a general education programme). 2. In the 2023 Survey of Adult Skills, adults educated in a STEM field are defined as having obtained their highest educational qualification in one of the following fields: ICT; natural sciences, mathematics and statistics; engineering and manufacturing; or construction.\n\nSTE on STE\n\n3 0\n\n320\n\n300\n\n280\n\n2 0\n\n2 0\n\n220\n\nFinland\n\napan\n\nFlemish Region ( E\n\netherlands\n\nSwit erland\n\nenmark\n\nC echia\n\nGermany\n\norway\n\nEstonia\n\nAustria\n\nSweden\n\nHungary\n\nFrance\n\nCanada\n\nEngland (U\n\nat ia\n\n**EC average**\n\nSingapore\n\nPortugal\n\nUnited States\n\nIreland\n\new ealand\n\nSlo ak Republic\n\nSpain\n\nItaly\n\nIsrael\n\nCroatia\n\norea\n\nithuania\n\nPoland\n\nChile\n\nScore points","pages":[{"line_numbers":[],"markdown":" **77**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 2.11. Average numeracy proficiency among tertiary-educated adults, by field of study**\n\nNote: Adults aged 25-65; does not include adults who were only administered the doorstep interview due to a language barrier, as information on field of study was not collected for those respondents (see Box 1.1 in Chapter 1). STEM stands for science, technology, engineering and mathematics. Educational attainment is based on the International Standard Classification of Education (ISCED) 2011, grouped into below upper secondary (ISCED 1, 2 and 3 short), upper secondary (ISCED 3 and 4) and tertiary (ISCED 5, 6, 7 and 8). Where possible, foreign qualifications are included as the closest corresponding level in the respective national education systems. Darker colours denote differences that are statistically significant at the 5% level. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the ote for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the proficiency of tertiary-educated adults who studied STEM fields._ Source: Table A.2.6 (N) in Annex A. 1. Within the International Standard Classification of Education (ISCED), programmes and related qualifications can be classified by the field of education and training as well as the level. The 2023 Survey of Adult Skills distinguishes between the following fields of study: economics, business and administration; law; health; welfare; social and behavioural sciences; journalism and information; information and communication technologies (ICT); natural sciences, mathematics and statistics; engineering and manufacturing; construction; agriculture, forestry, fisheries and environmental studies; personal and community services; security and transport; education and teacher training; humanities, languages and arts; and no main area of study or emphasis (if it was a general education programme). 2. In the 2023 Survey of Adult Skills, adults educated in a STEM field are defined as having obtained their highest educational qualification in one of the following fields: ICT; natural sciences, mathematics and statistics; engineering and manufacturing; or construction.\n\nSTE on STE\n\n3 0\n\n320\n\n300\n\n280\n\n2 0\n\n2 0\n\n220\n\nFinland\n\napan\n\nFlemish Region ( E\n\netherlands\n\nSwit erland\n\nenmark\n\nC echia\n\nGermany\n\norway\n\nEstonia\n\nAustria\n\nSweden\n\nHungary\n\nFrance\n\nCanada\n\nEngland (U\n\nat ia\n\n**EC average**\n\nSingapore\n\nPortugal\n\nUnited States\n\nIreland\n\new ealand\n\nSlo ak Republic\n\nSpain\n\nItaly\n\nIsrael\n\nCroatia\n\norea\n\nithuania\n\nPoland\n\nChile\n\nScore points","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":" **77**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 2.11. Average numeracy proficiency among tertiary-educated adults, by field of study**\n\nNote: Adults aged 25-65; does not include adults who were only administered the doorstep interview due to a language barrier, as information on field of study was not collected for those respondents (see Box 1.1 in Chapter 1). STEM stands for science, technology, engineering and mathematics. Educational attainment is based on the International Standard Classification of Education (ISCED) 2011, grouped into below upper secondary (ISCED 1, 2 and 3 short), upper secondary (ISCED 3 and 4) and tertiary (ISCED 5, 6, 7 and 8). Where possible, foreign qualifications are included as the closest corresponding level in the respective national education systems. Darker colours denote differences that are statistically significant at the 5% level. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the ote for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the proficiency of tertiary-educated adults who studied STEM fields._ Source: Table A.2.6 (N) in Annex A. 1. Within the International Standard Classification of Education (ISCED), programmes and related qualifications can be classified by the field of education and training as well as the level. The 2023 Survey of Adult Skills distinguishes between the following fields of study: economics, business and administration; law; health; welfare; social and behavioural sciences; journalism and information; information and communication technologies (ICT); natural sciences, mathematics and statistics; engineering and manufacturing; construction; agriculture, forestry, fisheries and environmental studies; personal and community services; security and transport; education and teacher training; humanities, languages and arts; and no main area of study or emphasis (if it was a general education programme). 2. In the 2023 Survey of Adult Skills, adults educated in a STEM field are defined as having obtained their highest educational qualification in one of the following fields: ICT; natural sciences, mathematics and statistics; engineering and manufacturing; or construction.\n\nSTE on STE\n\n3 0\n\n320\n\n300\n\n280\n\n2 0\n\n2 0\n\n220\n\nFinland\n\napan\n\nFlemish Region ( E\n\netherlands\n\nSwit erland\n\nenmark\n\nC echia\n\nGermany\n\norway\n\nEstonia\n\nAustria\n\nSweden\n\nHungary\n\nFrance\n\nCanada\n\nEngland (U\n\nat ia\n\n**EC average**\n\nSingapore\n\nPortugal\n\nUnited States\n\nIreland\n\new ealand\n\nSlo ak Republic\n\nSpain\n\nItaly\n\nIsrael\n\nCroatia\n\norea\n\nithuania\n\nPoland\n\nChile\n\nScore points","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p79.pdf"]},"response":{"errors":[],"extract_id":"extract_92daf5a6b00c80b3126285d5f162f077","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":[" **77**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 2.11. Average numeracy proficiency among tertiary-educated adults, by field of study**\n\nNote: Adults aged 25-65; does not include adults who were only administered the doorstep interview due to a language barrier, as information on field of study was not collected for those respondents (see Box 1.1 in Chapter 1). STEM stands for science, technology, engineering and mathematics. Educational attainment is based on the International Standard Classification of Education (ISCED) 2011, grouped into below upper secondary (ISCED 1, 2 and 3 short), upper secondary (ISCED 3 and 4) and tertiary (ISCED 5, 6, 7 and 8). Where possible, foreign qualifications are included as the closest corresponding level in the respective national education systems. Darker colours denote differences that are statistically significant at the 5% level. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the ote for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the proficiency of tertiary-educated adults who studied STEM fields._ Source: Table A.2.6 (N) in Annex A. 1. Within the International Standard Classification of Education (ISCED), programmes and related qualifications can be classified by the field of education and training as well as the level. The 2023 Survey of Adult Skills distinguishes between the following fields of study: economics, business and administration; law; health; welfare; social and behavioural sciences; journalism and information; information and communication technologies (ICT); natural sciences, mathematics and statistics; engineering and manufacturing; construction; agriculture, forestry, fisheries and environmental studies; personal and community services; security and transport; education and teacher training; humanities, languages and arts; and no main area of study or emphasis (if it was a general education programme). 2. In the 2023 Survey of Adult Skills, adults educated in a STEM field are defined as having obtained their highest educational qualification in one of the following fields: ICT; natural sciences, mathematics and statistics; engineering and manufacturing; or construction.\n\nSTE on STE\n\n3 0\n\n320\n\n300\n\n280\n\n2 0\n\n2 0\n\n220\n\nFinland\n\napan\n\nFlemish Region ( E\n\netherlands\n\nSwit erland\n\nenmark\n\nC echia\n\nGermany\n\norway\n\nEstonia\n\nAustria\n\nSweden\n\nHungary\n\nFrance\n\nCanada\n\nEngland (U\n\nat ia\n\n**EC average**\n\nSingapore\n\nPortugal\n\nUnited States\n\nIreland\n\new ealand\n\nSlo ak Republic\n\nSpain\n\nItaly\n\nIsrael\n\nCroatia\n\norea\n\nithuania\n\nPoland\n\nChile\n\nScore points"],"full_content":" **77**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 2.11. Average numeracy proficiency among tertiary-educated adults, by field of study**\n\nNote: Adults aged 25-65; does not include adults who were only administered the doorstep interview due to a language barrier, as information on field of study was not collected for those respondents (see Box 1.1 in Chapter 1). STEM stands for science, technology, engineering and mathematics. Educational attainment is based on the International Standard Classification of Education (ISCED) 2011, grouped into below upper secondary (ISCED 1, 2 and 3 short), upper secondary (ISCED 3 and 4) and tertiary (ISCED 5, 6, 7 and 8). Where possible, foreign qualifications are included as the closest corresponding level in the respective national education systems. Darker colours denote differences that are statistically significant at the 5% level. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the ote for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the proficiency of tertiary-educated adults who studied STEM fields._ Source: Table A.2.6 (N) in Annex A. 1. Within the International Standard Classification of Education (ISCED), programmes and related qualifications can be classified by the field of education and training as well as the level. The 2023 Survey of Adult Skills distinguishes between the following fields of study: economics, business and administration; law; health; welfare; social and behavioural sciences; journalism and information; information and communication technologies (ICT); natural sciences, mathematics and statistics; engineering and manufacturing; construction; agriculture, forestry, fisheries and environmental studies; personal and community services; security and transport; education and teacher training; humanities, languages and arts; and no main area of study or emphasis (if it was a general education programme). 2. In the 2023 Survey of Adult Skills, adults educated in a STEM field are defined as having obtained their highest educational qualification in one of the following fields: ICT; natural sciences, mathematics and statistics; engineering and manufacturing; or construction.\n\nSTE on STE\n\n3 0\n\n320\n\n300\n\n280\n\n2 0\n\n2 0\n\n220\n\nFinland\n\napan\n\nFlemish Region ( E\n\netherlands\n\nSwit erland\n\nenmark\n\nC echia\n\nGermany\n\norway\n\nEstonia\n\nAustria\n\nSweden\n\nHungary\n\nFrance\n\nCanada\n\nEngland (U\n\nat ia\n\n**EC average**\n\nSingapore\n\nPortugal\n\nUnited States\n\nIreland\n\new ealand\n\nSlo ak Republic\n\nSpain\n\nItaly\n\nIsrael\n\nCroatia\n\norea\n\nithuania\n\nPoland\n\nChile\n\nScore points","publish_date":null,"title":"Figure 2.11. Average numeracy proficiency among tertiary-educated adults, by field of study","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p79.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p79.pdf"},"request":{"example_id":"chart/b263dc5d-en_p79","product_type":"parse","source_file_path":"b263dc5d-en_p79.pdf"},"started_at":"2026-11-04T19:46:30.861349"} +{"completed_at":"2026-11-04T19:46:33.556169","latency_in_ms":1654,"output":{"example_id":"chart/b263dc5d-en_p81","grounded_pages":[],"layout_pages":[],"markdown":" **79**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 2.12. Share of women among STEM and non-STEM graduates**\n\nShare of tertiary-educated 25-65 year-olds who graduated from each field of study who are women\n\nNote: Does not include adults who were only administered the doorstep interview due to a language barrier, as information on field of study was not collected for those respondents (see Box 1.1 in Chapter 1). STEM stands for science, technology, engineering and mathematics. Educational attainment is based on the International Standard Classification of Education (ISCED) 2011, grouped into below upper secondary (ISCED 1, 2 and 3 short), upper secondary (ISCED 3 and 4) and tertiary (ISCED 5, 6, 7 and 8). Where possible, foreign qualifications are included as the closest corresponding level in the respective national education systems. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the Note for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the percentage of STEM graduates who are women._ Source: Table B.3.4 in Annex B.\n\nWhen comparing tertiary-educated men and women who have studied STEM fields, data from the 2023 Survey of Adult Skills show that, on average, women score 8 points lower in numeracy (Figure 2.13). This gap is narrower than the gap among all tertiary-educated adults (16 points; see Table A.2.9 (N) in Annex A). Accounting for other relevant characteristics only marginally helps to explain this gap.\n\nAmong tertiary-educated adults who have chosen STEM fields, women have a small and statistically insignificant advantage in literacy over men (1 point); this gap increases to 2 points and becomes statistically significant after accounting for differences in other background characteristics. In contrast, men in this group have a small but statistically significant advantage over women in adaptive problem solving (3 points), which reduces to 2 points after accounting for relevant socio-demographic characteristics (see Table A.2.9 (L, A) in Annex A).\n\nThese results show that the gender gap in numeracy persists even when comparing men and women who made similar educational choices. Factors other than the observable socio-demographic characteristics examined in this report would need to be investigated in order to understand the reasons behind such differences and to design initiatives that aim at reducing this gender gap. The first cycle of the Survey of Adult Skills also found an advantage for men in numeracy (10 points) among adults working in STEM occupations (OECD, 2015[43]).\n\nSTE on STE\n\n80\n\n0\n\n0\n\n0\n\n0\n\n30\n\n20\n\n10\n\n0\n\nPortugal\n\nCroatia\n\nSweden\n\nat ia\n\nIsrael\n\nIreland\n\nSlo ak Republic\n\nC echia\n\nEstonia\n\nItaly\n\nenmark\n\nSingapore\n\nUnited States\n\norway\n\n**EC average**\n\nFrance\n\nFlemish Region ( E\n\nSpain\n\netherlands\n\nCanada\n\new ealand\n\nithuania\n\nFinland\n\nPoland\n\norea\n\nHungary\n\nAustria\n\nSwit erland\n\nGermany\n\nEngland (U\n\nChile\n\napan","pages":[{"line_numbers":[],"markdown":" **79**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 2.12. Share of women among STEM and non-STEM graduates**\n\nShare of tertiary-educated 25-65 year-olds who graduated from each field of study who are women\n\nNote: Does not include adults who were only administered the doorstep interview due to a language barrier, as information on field of study was not collected for those respondents (see Box 1.1 in Chapter 1). STEM stands for science, technology, engineering and mathematics. Educational attainment is based on the International Standard Classification of Education (ISCED) 2011, grouped into below upper secondary (ISCED 1, 2 and 3 short), upper secondary (ISCED 3 and 4) and tertiary (ISCED 5, 6, 7 and 8). Where possible, foreign qualifications are included as the closest corresponding level in the respective national education systems. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the Note for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the percentage of STEM graduates who are women._ Source: Table B.3.4 in Annex B.\n\nWhen comparing tertiary-educated men and women who have studied STEM fields, data from the 2023 Survey of Adult Skills show that, on average, women score 8 points lower in numeracy (Figure 2.13). This gap is narrower than the gap among all tertiary-educated adults (16 points; see Table A.2.9 (N) in Annex A). Accounting for other relevant characteristics only marginally helps to explain this gap.\n\nAmong tertiary-educated adults who have chosen STEM fields, women have a small and statistically insignificant advantage in literacy over men (1 point); this gap increases to 2 points and becomes statistically significant after accounting for differences in other background characteristics. In contrast, men in this group have a small but statistically significant advantage over women in adaptive problem solving (3 points), which reduces to 2 points after accounting for relevant socio-demographic characteristics (see Table A.2.9 (L, A) in Annex A).\n\nThese results show that the gender gap in numeracy persists even when comparing men and women who made similar educational choices. Factors other than the observable socio-demographic characteristics examined in this report would need to be investigated in order to understand the reasons behind such differences and to design initiatives that aim at reducing this gender gap. The first cycle of the Survey of Adult Skills also found an advantage for men in numeracy (10 points) among adults working in STEM occupations (OECD, 2015[43]).\n\nSTE on STE\n\n80\n\n0\n\n0\n\n0\n\n0\n\n30\n\n20\n\n10\n\n0\n\nPortugal\n\nCroatia\n\nSweden\n\nat ia\n\nIsrael\n\nIreland\n\nSlo ak Republic\n\nC echia\n\nEstonia\n\nItaly\n\nenmark\n\nSingapore\n\nUnited States\n\norway\n\n**EC average**\n\nFrance\n\nFlemish Region ( E\n\nSpain\n\netherlands\n\nCanada\n\new ealand\n\nithuania\n\nFinland\n\nPoland\n\norea\n\nHungary\n\nAustria\n\nSwit erland\n\nGermany\n\nEngland (U\n\nChile\n\napan","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":" **79**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 2.12. Share of women among STEM and non-STEM graduates**\n\nShare of tertiary-educated 25-65 year-olds who graduated from each field of study who are women\n\nNote: Does not include adults who were only administered the doorstep interview due to a language barrier, as information on field of study was not collected for those respondents (see Box 1.1 in Chapter 1). STEM stands for science, technology, engineering and mathematics. Educational attainment is based on the International Standard Classification of Education (ISCED) 2011, grouped into below upper secondary (ISCED 1, 2 and 3 short), upper secondary (ISCED 3 and 4) and tertiary (ISCED 5, 6, 7 and 8). Where possible, foreign qualifications are included as the closest corresponding level in the respective national education systems. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the Note for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the percentage of STEM graduates who are women._ Source: Table B.3.4 in Annex B.\n\nWhen comparing tertiary-educated men and women who have studied STEM fields, data from the 2023 Survey of Adult Skills show that, on average, women score 8 points lower in numeracy (Figure 2.13). This gap is narrower than the gap among all tertiary-educated adults (16 points; see Table A.2.9 (N) in Annex A). Accounting for other relevant characteristics only marginally helps to explain this gap.\n\nAmong tertiary-educated adults who have chosen STEM fields, women have a small and statistically insignificant advantage in literacy over men (1 point); this gap increases to 2 points and becomes statistically significant after accounting for differences in other background characteristics. In contrast, men in this group have a small but statistically significant advantage over women in adaptive problem solving (3 points), which reduces to 2 points after accounting for relevant socio-demographic characteristics (see Table A.2.9 (L, A) in Annex A).\n\nThese results show that the gender gap in numeracy persists even when comparing men and women who made similar educational choices. Factors other than the observable socio-demographic characteristics examined in this report would need to be investigated in order to understand the reasons behind such differences and to design initiatives that aim at reducing this gender gap. The first cycle of the Survey of Adult Skills also found an advantage for men in numeracy (10 points) among adults working in STEM occupations (OECD, 2015[43]).\n\nSTE on STE\n\n80\n\n0\n\n0\n\n0\n\n0\n\n30\n\n20\n\n10\n\n0\n\nPortugal\n\nCroatia\n\nSweden\n\nat ia\n\nIsrael\n\nIreland\n\nSlo ak Republic\n\nC echia\n\nEstonia\n\nItaly\n\nenmark\n\nSingapore\n\nUnited States\n\norway\n\n**EC average**\n\nFrance\n\nFlemish Region ( E\n\nSpain\n\netherlands\n\nCanada\n\new ealand\n\nithuania\n\nFinland\n\nPoland\n\norea\n\nHungary\n\nAustria\n\nSwit erland\n\nGermany\n\nEngland (U\n\nChile\n\napan","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p81.pdf"]},"response":{"errors":[],"extract_id":"extract_40f02114a27510e6161fa3bc5036cb9a","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":[" **79**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 2.12. Share of women among STEM and non-STEM graduates**\n\nShare of tertiary-educated 25-65 year-olds who graduated from each field of study who are women\n\nNote: Does not include adults who were only administered the doorstep interview due to a language barrier, as information on field of study was not collected for those respondents (see Box 1.1 in Chapter 1). STEM stands for science, technology, engineering and mathematics. Educational attainment is based on the International Standard Classification of Education (ISCED) 2011, grouped into below upper secondary (ISCED 1, 2 and 3 short), upper secondary (ISCED 3 and 4) and tertiary (ISCED 5, 6, 7 and 8). Where possible, foreign qualifications are included as the closest corresponding level in the respective national education systems. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the Note for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the percentage of STEM graduates who are women._ Source: Table B.3.4 in Annex B.\n\nWhen comparing tertiary-educated men and women who have studied STEM fields, data from the 2023 Survey of Adult Skills show that, on average, women score 8 points lower in numeracy (Figure 2.13). This gap is narrower than the gap among all tertiary-educated adults (16 points; see Table A.2.9 (N) in Annex A). Accounting for other relevant characteristics only marginally helps to explain this gap.\n\nAmong tertiary-educated adults who have chosen STEM fields, women have a small and statistically insignificant advantage in literacy over men (1 point); this gap increases to 2 points and becomes statistically significant after accounting for differences in other background characteristics. In contrast, men in this group have a small but statistically significant advantage over women in adaptive problem solving (3 points), which reduces to 2 points after accounting for relevant socio-demographic characteristics (see Table A.2.9 (L, A) in Annex A).\n\nThese results show that the gender gap in numeracy persists even when comparing men and women who made similar educational choices. Factors other than the observable socio-demographic characteristics examined in this report would need to be investigated in order to understand the reasons behind such differences and to design initiatives that aim at reducing this gender gap. The first cycle of the Survey of Adult Skills also found an advantage for men in numeracy (10 points) among adults working in STEM occupations (OECD, 2015[43]).\n\nSTE on STE\n\n80\n\n0\n\n0\n\n0\n\n0\n\n30\n\n20\n\n10\n\n0\n\nPortugal\n\nCroatia\n\nSweden\n\nat ia\n\nIsrael\n\nIreland\n\nSlo ak Republic\n\nC echia\n\nEstonia\n\nItaly\n\nenmark\n\nSingapore\n\nUnited States\n\norway\n\n**EC average**\n\nFrance\n\nFlemish Region ( E\n\nSpain\n\netherlands\n\nCanada\n\new ealand\n\nithuania\n\nFinland\n\nPoland\n\norea\n\nHungary\n\nAustria\n\nSwit erland\n\nGermany\n\nEngland (U\n\nChile\n\napan"],"full_content":" **79**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 2.12. Share of women among STEM and non-STEM graduates**\n\nShare of tertiary-educated 25-65 year-olds who graduated from each field of study who are women\n\nNote: Does not include adults who were only administered the doorstep interview due to a language barrier, as information on field of study was not collected for those respondents (see Box 1.1 in Chapter 1). STEM stands for science, technology, engineering and mathematics. Educational attainment is based on the International Standard Classification of Education (ISCED) 2011, grouped into below upper secondary (ISCED 1, 2 and 3 short), upper secondary (ISCED 3 and 4) and tertiary (ISCED 5, 6, 7 and 8). Where possible, foreign qualifications are included as the closest corresponding level in the respective national education systems. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the Note for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the percentage of STEM graduates who are women._ Source: Table B.3.4 in Annex B.\n\nWhen comparing tertiary-educated men and women who have studied STEM fields, data from the 2023 Survey of Adult Skills show that, on average, women score 8 points lower in numeracy (Figure 2.13). This gap is narrower than the gap among all tertiary-educated adults (16 points; see Table A.2.9 (N) in Annex A). Accounting for other relevant characteristics only marginally helps to explain this gap.\n\nAmong tertiary-educated adults who have chosen STEM fields, women have a small and statistically insignificant advantage in literacy over men (1 point); this gap increases to 2 points and becomes statistically significant after accounting for differences in other background characteristics. In contrast, men in this group have a small but statistically significant advantage over women in adaptive problem solving (3 points), which reduces to 2 points after accounting for relevant socio-demographic characteristics (see Table A.2.9 (L, A) in Annex A).\n\nThese results show that the gender gap in numeracy persists even when comparing men and women who made similar educational choices. Factors other than the observable socio-demographic characteristics examined in this report would need to be investigated in order to understand the reasons behind such differences and to design initiatives that aim at reducing this gender gap. The first cycle of the Survey of Adult Skills also found an advantage for men in numeracy (10 points) among adults working in STEM occupations (OECD, 2015[43]).\n\nSTE on STE\n\n80\n\n0\n\n0\n\n0\n\n0\n\n30\n\n20\n\n10\n\n0\n\nPortugal\n\nCroatia\n\nSweden\n\nat ia\n\nIsrael\n\nIreland\n\nSlo ak Republic\n\nC echia\n\nEstonia\n\nItaly\n\nenmark\n\nSingapore\n\nUnited States\n\norway\n\n**EC average**\n\nFrance\n\nFlemish Region ( E\n\nSpain\n\netherlands\n\nCanada\n\new ealand\n\nithuania\n\nFinland\n\nPoland\n\norea\n\nHungary\n\nAustria\n\nSwit erland\n\nGermany\n\nEngland (U\n\nChile\n\napan","publish_date":null,"title":"Figure 2.12. Share of women among STEM and non-STEM graduates","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p81.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p81.pdf"},"request":{"example_id":"chart/b263dc5d-en_p81","product_type":"parse","source_file_path":"b263dc5d-en_p81.pdf"},"started_at":"2026-11-04T19:46:31.901348"} +{"completed_at":"2026-11-04T19:46:36.956172","latency_in_ms":3007,"output":{"example_id":"chart/b263dc5d-en_p82","grounded_pages":[],"layout_pages":[],"markdown":"**80** \n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 2.13. Gender differences in numeracy among STEM graduates**\n\nAdjusted and unadjusted differences in average numeracy scores between tertiary-educated men and women who studied STEM fields (men _minus_ women)\n\nNote: Adults aged 25-65. Does not include adults who were only administered the doorstep interview due to a language barrier, as information on field of study was not collected for those respondents (see Box 1.1 in Chapter 1). STEM stands for science, technology, engineering and mathematics. Educational attainment is based on the International Standard Classification of Education (ISCED) 2011, grouped into below upper secondary (ISCED 1, 2 and 3 short), upper secondary (ISCED 3 and 4) and tertiary (ISCED 5, 6, 7 and 8). Where possible, foreign qualifications are included as the closest corresponding level in the respective national education systems. Unadjusted differences are the differences between the two averages for each contrast category. Adjusted differences are based on a regression model that takes into account differences associated with age, immigrant background, language spoken at home and parents’ educational attainment. Darker colours denote differences that are statistically significant at the 5% level. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the ote for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the unadjusted difference between men and women who studied STEM fields._ Source: Table A.2.9 (N) in Annex A.\n\n_Differences in proficiency between men and women_\n\nOn average across participating OECD countries and economies, women scored 3 points higher than men in literacy, while men outscored women by 10 points in numeracy and 2 points in adaptive problem solving (Figure 2.14). After accounting for relevant background characteristics, the adjusted gender gap shrinks to 1 point in favour of women in literacy (which is not statistically significant), but widens to 12 points in numeracy and 4 points in adaptive problem solving in favour of men. One potential explanation for this increase in the size of the adjusted coefficients in numeracy and adaptive problem solving is gender differences in educational attainment. For example, on average, 48% of women are tertiary educated, but only 40% of men are (see Table B.3.3 in Annex B). As higher levels of education are normally associated with higher proficiency, adjusted differences control for the fact that women appear to underperform in the assessment, compared to what one would expect based on their level of education.\n\nIn literacy, women in 13 countries (Croatia, Estonia, Finland, France, Germany, Hungary, Israel, Latvia, Lithuania, the Netherlands, New Zealand, Norway and Poland) scored significantly higher than men on\n\n2\n\n20\n\n1\n\n10\n\n0\n\n10\n\n1\n\nSweden\n\netherlands\n\napan\n\nPoland\n\nUnited States\n\nC echia\n\nFlemish Region ( E\n\nEstonia\n\nIreland\n\nAustria\n\nFrance\n\nSwit erland\n\new ealand\n\norway\n\nSpain\n\n**EC average**\n\nSingapore\n\nat ia\n\nIsrael\n\nHungary\n\nenmark\n\nPortugal\n\nGermany\n\nFinland\n\nItaly\n\nithuania\n\nEngland (U\n\nCanada\n\nChile\n\nCroatia\n\nSlo ak Republic\n\norea\n\nScore point difference Unad u sted Ad usted\n\nen score higher than women\n\nWomen score higher than men","pages":[{"line_numbers":[],"markdown":"**80** \n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 2.13. Gender differences in numeracy among STEM graduates**\n\nAdjusted and unadjusted differences in average numeracy scores between tertiary-educated men and women who studied STEM fields (men _minus_ women)\n\nNote: Adults aged 25-65. Does not include adults who were only administered the doorstep interview due to a language barrier, as information on field of study was not collected for those respondents (see Box 1.1 in Chapter 1). STEM stands for science, technology, engineering and mathematics. Educational attainment is based on the International Standard Classification of Education (ISCED) 2011, grouped into below upper secondary (ISCED 1, 2 and 3 short), upper secondary (ISCED 3 and 4) and tertiary (ISCED 5, 6, 7 and 8). Where possible, foreign qualifications are included as the closest corresponding level in the respective national education systems. Unadjusted differences are the differences between the two averages for each contrast category. Adjusted differences are based on a regression model that takes into account differences associated with age, immigrant background, language spoken at home and parents’ educational attainment. Darker colours denote differences that are statistically significant at the 5% level. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the ote for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the unadjusted difference between men and women who studied STEM fields._ Source: Table A.2.9 (N) in Annex A.\n\n_Differences in proficiency between men and women_\n\nOn average across participating OECD countries and economies, women scored 3 points higher than men in literacy, while men outscored women by 10 points in numeracy and 2 points in adaptive problem solving (Figure 2.14). After accounting for relevant background characteristics, the adjusted gender gap shrinks to 1 point in favour of women in literacy (which is not statistically significant), but widens to 12 points in numeracy and 4 points in adaptive problem solving in favour of men. One potential explanation for this increase in the size of the adjusted coefficients in numeracy and adaptive problem solving is gender differences in educational attainment. For example, on average, 48% of women are tertiary educated, but only 40% of men are (see Table B.3.3 in Annex B). As higher levels of education are normally associated with higher proficiency, adjusted differences control for the fact that women appear to underperform in the assessment, compared to what one would expect based on their level of education.\n\nIn literacy, women in 13 countries (Croatia, Estonia, Finland, France, Germany, Hungary, Israel, Latvia, Lithuania, the Netherlands, New Zealand, Norway and Poland) scored significantly higher than men on\n\n2\n\n20\n\n1\n\n10\n\n0\n\n10\n\n1\n\nSweden\n\netherlands\n\napan\n\nPoland\n\nUnited States\n\nC echia\n\nFlemish Region ( E\n\nEstonia\n\nIreland\n\nAustria\n\nFrance\n\nSwit erland\n\new ealand\n\norway\n\nSpain\n\n**EC average**\n\nSingapore\n\nat ia\n\nIsrael\n\nHungary\n\nenmark\n\nPortugal\n\nGermany\n\nFinland\n\nItaly\n\nithuania\n\nEngland (U\n\nCanada\n\nChile\n\nCroatia\n\nSlo ak Republic\n\norea\n\nScore point difference Unad u sted Ad usted\n\nen score higher than women\n\nWomen score higher than men","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**80** \n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 2.13. Gender differences in numeracy among STEM graduates**\n\nAdjusted and unadjusted differences in average numeracy scores between tertiary-educated men and women who studied STEM fields (men _minus_ women)\n\nNote: Adults aged 25-65. Does not include adults who were only administered the doorstep interview due to a language barrier, as information on field of study was not collected for those respondents (see Box 1.1 in Chapter 1). STEM stands for science, technology, engineering and mathematics. Educational attainment is based on the International Standard Classification of Education (ISCED) 2011, grouped into below upper secondary (ISCED 1, 2 and 3 short), upper secondary (ISCED 3 and 4) and tertiary (ISCED 5, 6, 7 and 8). Where possible, foreign qualifications are included as the closest corresponding level in the respective national education systems. Unadjusted differences are the differences between the two averages for each contrast category. Adjusted differences are based on a regression model that takes into account differences associated with age, immigrant background, language spoken at home and parents’ educational attainment. Darker colours denote differences that are statistically significant at the 5% level. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the ote for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the unadjusted difference between men and women who studied STEM fields._ Source: Table A.2.9 (N) in Annex A.\n\n_Differences in proficiency between men and women_\n\nOn average across participating OECD countries and economies, women scored 3 points higher than men in literacy, while men outscored women by 10 points in numeracy and 2 points in adaptive problem solving (Figure 2.14). After accounting for relevant background characteristics, the adjusted gender gap shrinks to 1 point in favour of women in literacy (which is not statistically significant), but widens to 12 points in numeracy and 4 points in adaptive problem solving in favour of men. One potential explanation for this increase in the size of the adjusted coefficients in numeracy and adaptive problem solving is gender differences in educational attainment. For example, on average, 48% of women are tertiary educated, but only 40% of men are (see Table B.3.3 in Annex B). As higher levels of education are normally associated with higher proficiency, adjusted differences control for the fact that women appear to underperform in the assessment, compared to what one would expect based on their level of education.\n\nIn literacy, women in 13 countries (Croatia, Estonia, Finland, France, Germany, Hungary, Israel, Latvia, Lithuania, the Netherlands, New Zealand, Norway and Poland) scored significantly higher than men on\n\n2\n\n20\n\n1\n\n10\n\n0\n\n10\n\n1\n\nSweden\n\netherlands\n\napan\n\nPoland\n\nUnited States\n\nC echia\n\nFlemish Region ( E\n\nEstonia\n\nIreland\n\nAustria\n\nFrance\n\nSwit erland\n\new ealand\n\norway\n\nSpain\n\n**EC average**\n\nSingapore\n\nat ia\n\nIsrael\n\nHungary\n\nenmark\n\nPortugal\n\nGermany\n\nFinland\n\nItaly\n\nithuania\n\nEngland (U\n\nCanada\n\nChile\n\nCroatia\n\nSlo ak Republic\n\norea\n\nScore point difference Unad u sted Ad usted\n\nen score higher than women\n\nWomen score higher than men","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p82.pdf"]},"response":{"errors":[],"extract_id":"extract_f7fc0120dc944d825fc3a8d25fec3707","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**80** \n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 2.13. Gender differences in numeracy among STEM graduates**\n\nAdjusted and unadjusted differences in average numeracy scores between tertiary-educated men and women who studied STEM fields (men _minus_ women)\n\nNote: Adults aged 25-65. Does not include adults who were only administered the doorstep interview due to a language barrier, as information on field of study was not collected for those respondents (see Box 1.1 in Chapter 1). STEM stands for science, technology, engineering and mathematics. Educational attainment is based on the International Standard Classification of Education (ISCED) 2011, grouped into below upper secondary (ISCED 1, 2 and 3 short), upper secondary (ISCED 3 and 4) and tertiary (ISCED 5, 6, 7 and 8). Where possible, foreign qualifications are included as the closest corresponding level in the respective national education systems. Unadjusted differences are the differences between the two averages for each contrast category. Adjusted differences are based on a regression model that takes into account differences associated with age, immigrant background, language spoken at home and parents’ educational attainment. Darker colours denote differences that are statistically significant at the 5% level. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the ote for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the unadjusted difference between men and women who studied STEM fields._ Source: Table A.2.9 (N) in Annex A.\n\n_Differences in proficiency between men and women_\n\nOn average across participating OECD countries and economies, women scored 3 points higher than men in literacy, while men outscored women by 10 points in numeracy and 2 points in adaptive problem solving (Figure 2.14). After accounting for relevant background characteristics, the adjusted gender gap shrinks to 1 point in favour of women in literacy (which is not statistically significant), but widens to 12 points in numeracy and 4 points in adaptive problem solving in favour of men. One potential explanation for this increase in the size of the adjusted coefficients in numeracy and adaptive problem solving is gender differences in educational attainment. For example, on average, 48% of women are tertiary educated, but only 40% of men are (see Table B.3.3 in Annex B). As higher levels of education are normally associated with higher proficiency, adjusted differences control for the fact that women appear to underperform in the assessment, compared to what one would expect based on their level of education.\n\nIn literacy, women in 13 countries (Croatia, Estonia, Finland, France, Germany, Hungary, Israel, Latvia, Lithuania, the Netherlands, New Zealand, Norway and Poland) scored significantly higher than men on\n\n2\n\n20\n\n1\n\n10\n\n0\n\n10\n\n1\n\nSweden\n\netherlands\n\napan\n\nPoland\n\nUnited States\n\nC echia\n\nFlemish Region ( E\n\nEstonia\n\nIreland\n\nAustria\n\nFrance\n\nSwit erland\n\new ealand\n\norway\n\nSpain\n\n**EC average**\n\nSingapore\n\nat ia\n\nIsrael\n\nHungary\n\nenmark\n\nPortugal\n\nGermany\n\nFinland\n\nItaly\n\nithuania\n\nEngland (U\n\nCanada\n\nChile\n\nCroatia\n\nSlo ak Republic\n\norea\n\nScore point difference Unad u sted Ad usted\n\nen score higher than women\n\nWomen score higher than men"],"full_content":"**80** \n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 2.13. Gender differences in numeracy among STEM graduates**\n\nAdjusted and unadjusted differences in average numeracy scores between tertiary-educated men and women who studied STEM fields (men _minus_ women)\n\nNote: Adults aged 25-65. Does not include adults who were only administered the doorstep interview due to a language barrier, as information on field of study was not collected for those respondents (see Box 1.1 in Chapter 1). STEM stands for science, technology, engineering and mathematics. Educational attainment is based on the International Standard Classification of Education (ISCED) 2011, grouped into below upper secondary (ISCED 1, 2 and 3 short), upper secondary (ISCED 3 and 4) and tertiary (ISCED 5, 6, 7 and 8). Where possible, foreign qualifications are included as the closest corresponding level in the respective national education systems. Unadjusted differences are the differences between the two averages for each contrast category. Adjusted differences are based on a regression model that takes into account differences associated with age, immigrant background, language spoken at home and parents’ educational attainment. Darker colours denote differences that are statistically significant at the 5% level. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the ote for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the unadjusted difference between men and women who studied STEM fields._ Source: Table A.2.9 (N) in Annex A.\n\n_Differences in proficiency between men and women_\n\nOn average across participating OECD countries and economies, women scored 3 points higher than men in literacy, while men outscored women by 10 points in numeracy and 2 points in adaptive problem solving (Figure 2.14). After accounting for relevant background characteristics, the adjusted gender gap shrinks to 1 point in favour of women in literacy (which is not statistically significant), but widens to 12 points in numeracy and 4 points in adaptive problem solving in favour of men. One potential explanation for this increase in the size of the adjusted coefficients in numeracy and adaptive problem solving is gender differences in educational attainment. For example, on average, 48% of women are tertiary educated, but only 40% of men are (see Table B.3.3 in Annex B). As higher levels of education are normally associated with higher proficiency, adjusted differences control for the fact that women appear to underperform in the assessment, compared to what one would expect based on their level of education.\n\nIn literacy, women in 13 countries (Croatia, Estonia, Finland, France, Germany, Hungary, Israel, Latvia, Lithuania, the Netherlands, New Zealand, Norway and Poland) scored significantly higher than men on\n\n2\n\n20\n\n1\n\n10\n\n0\n\n10\n\n1\n\nSweden\n\netherlands\n\napan\n\nPoland\n\nUnited States\n\nC echia\n\nFlemish Region ( E\n\nEstonia\n\nIreland\n\nAustria\n\nFrance\n\nSwit erland\n\new ealand\n\norway\n\nSpain\n\n**EC average**\n\nSingapore\n\nat ia\n\nIsrael\n\nHungary\n\nenmark\n\nPortugal\n\nGermany\n\nFinland\n\nItaly\n\nithuania\n\nEngland (U\n\nCanada\n\nChile\n\nCroatia\n\nSlo ak Republic\n\norea\n\nScore point difference Unad u sted Ad usted\n\nen score higher than women\n\nWomen score higher than men","publish_date":null,"title":"Figure 2.13. Gender differences in numeracy among STEM graduates","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p82.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p82.pdf"},"request":{"example_id":"chart/b263dc5d-en_p82","product_type":"parse","source_file_path":"b263dc5d-en_p82.pdf"},"started_at":"2026-11-04T19:46:33.948803"} +{"completed_at":"2026-11-04T19:46:36.769856","latency_in_ms":1434,"output":{"example_id":"chart/b263dc5d-en_p94","grounded_pages":[],"layout_pages":[],"markdown":"**92** \n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 2.19. Average proficiency in key information-processing skills, by parental education**\n\nLiteracy, numeracy and adaptive problem solving\n\nNote: Adults aged 16-65; does not include adults who were only administered the doorstep interview due to a language barrier, as information on parental education was not collected for those respondents (see Box 1.1 in Chapter 1). Respondents are categorised as having highly educated parents if at least one parent attained tertiary education; as having medium-educated parents if at least one parent attained upper secondary education and none of the parents attained tertiary education; and as having low-educated parents if neither parent attained upper secondary education. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the ote for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the average proficiency of adults with low-educated parents._ Source: Table A.2.12 (L, N, A) in Annex A.\n\nedium educated parents Highly educated parents ow educated parents\n\nFinland\n\nSweden\n\napan\n\netherlands\n\norway\n\nenmark\n\new ealand\n\nFlemish Region ( E\n\nEngland (U\n\nIreland\n\nSpain\n\nSlo ak Republic\n\nCanada\n\nEstonia\n\nCroatia\n\n**EC average**\n\nItaly\n\nFrance\n\nSingapore\n\nC echia\n\norea\n\nSwit erland\n\nPortugal\n\nAustria\n\nHungary\n\nithuania\n\nGermany\n\nUnited States\n\nPoland\n\nat ia\n\nIsrael\n\nChile\n\n330\n\n310\n\n290\n\n2 0\n\n2 0\n\n230\n\n210\n\n190\n\nScore points **. iteracy**\n\nFinland\n\napan\n\nSweden\n\netherlands\n\norway\n\nenmark\n\nFlemish Region ( E\n\nSingapore\n\nEstonia\n\nSpain\n\nIreland\n\nCanada\n\nAustria\n\nEngland (U\n\nSlo ak Republic\n\new ealand\n\n**EC average**\n\nC echia\n\nSwit erland\n\nCroatia\n\norea\n\nFrance\n\nItaly\n\nat ia\n\nPortugal\n\nithuania\n\nGermany\n\nHungary\n\nIsrael\n\nPoland\n\nUnited States\n\nChile\n\n330\n\n310\n\n290\n\n2 0\n\n2 0\n\n230\n\n210\n\n190\n\nScore points **B. umeracy**\n\nFinland\n\napan\n\nSweden\n\netherlands\n\norway\n\nenmark\n\nFlemish Region ( E\n\nSpain\n\new ealand\n\nSlo ak Republic\n\nEngland (U\n\nIreland\n\nSingapore\n\nEstonia\n\nCanada\n\n**EC average**\n\nFrance\n\nAustria\n\nPortugal\n\nSwit erland\n\nC echia\n\norea\n\nItaly\n\nCroatia\n\nGermany\n\nHungary\n\nat ia\n\nUnited States\n\nithuania\n\nIsrael\n\nPoland\n\nChile\n\n330\n\n310\n\n290\n\n2 0\n\n2 0\n\n230\n\n210\n\n190\n\nScore points **C. daptive problem solving**","pages":[{"line_numbers":[],"markdown":"**92** \n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 2.19. Average proficiency in key information-processing skills, by parental education**\n\nLiteracy, numeracy and adaptive problem solving\n\nNote: Adults aged 16-65; does not include adults who were only administered the doorstep interview due to a language barrier, as information on parental education was not collected for those respondents (see Box 1.1 in Chapter 1). Respondents are categorised as having highly educated parents if at least one parent attained tertiary education; as having medium-educated parents if at least one parent attained upper secondary education and none of the parents attained tertiary education; and as having low-educated parents if neither parent attained upper secondary education. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the ote for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the average proficiency of adults with low-educated parents._ Source: Table A.2.12 (L, N, A) in Annex A.\n\nedium educated parents Highly educated parents ow educated parents\n\nFinland\n\nSweden\n\napan\n\netherlands\n\norway\n\nenmark\n\new ealand\n\nFlemish Region ( E\n\nEngland (U\n\nIreland\n\nSpain\n\nSlo ak Republic\n\nCanada\n\nEstonia\n\nCroatia\n\n**EC average**\n\nItaly\n\nFrance\n\nSingapore\n\nC echia\n\norea\n\nSwit erland\n\nPortugal\n\nAustria\n\nHungary\n\nithuania\n\nGermany\n\nUnited States\n\nPoland\n\nat ia\n\nIsrael\n\nChile\n\n330\n\n310\n\n290\n\n2 0\n\n2 0\n\n230\n\n210\n\n190\n\nScore points **. iteracy**\n\nFinland\n\napan\n\nSweden\n\netherlands\n\norway\n\nenmark\n\nFlemish Region ( E\n\nSingapore\n\nEstonia\n\nSpain\n\nIreland\n\nCanada\n\nAustria\n\nEngland (U\n\nSlo ak Republic\n\new ealand\n\n**EC average**\n\nC echia\n\nSwit erland\n\nCroatia\n\norea\n\nFrance\n\nItaly\n\nat ia\n\nPortugal\n\nithuania\n\nGermany\n\nHungary\n\nIsrael\n\nPoland\n\nUnited States\n\nChile\n\n330\n\n310\n\n290\n\n2 0\n\n2 0\n\n230\n\n210\n\n190\n\nScore points **B. umeracy**\n\nFinland\n\napan\n\nSweden\n\netherlands\n\norway\n\nenmark\n\nFlemish Region ( E\n\nSpain\n\new ealand\n\nSlo ak Republic\n\nEngland (U\n\nIreland\n\nSingapore\n\nEstonia\n\nCanada\n\n**EC average**\n\nFrance\n\nAustria\n\nPortugal\n\nSwit erland\n\nC echia\n\norea\n\nItaly\n\nCroatia\n\nGermany\n\nHungary\n\nat ia\n\nUnited States\n\nithuania\n\nIsrael\n\nPoland\n\nChile\n\n330\n\n310\n\n290\n\n2 0\n\n2 0\n\n230\n\n210\n\n190\n\nScore points **C. daptive problem solving**","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**92** \n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 2.19. Average proficiency in key information-processing skills, by parental education**\n\nLiteracy, numeracy and adaptive problem solving\n\nNote: Adults aged 16-65; does not include adults who were only administered the doorstep interview due to a language barrier, as information on parental education was not collected for those respondents (see Box 1.1 in Chapter 1). Respondents are categorised as having highly educated parents if at least one parent attained tertiary education; as having medium-educated parents if at least one parent attained upper secondary education and none of the parents attained tertiary education; and as having low-educated parents if neither parent attained upper secondary education. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the ote for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the average proficiency of adults with low-educated parents._ Source: Table A.2.12 (L, N, A) in Annex A.\n\nedium educated parents Highly educated parents ow educated parents\n\nFinland\n\nSweden\n\napan\n\netherlands\n\norway\n\nenmark\n\new ealand\n\nFlemish Region ( E\n\nEngland (U\n\nIreland\n\nSpain\n\nSlo ak Republic\n\nCanada\n\nEstonia\n\nCroatia\n\n**EC average**\n\nItaly\n\nFrance\n\nSingapore\n\nC echia\n\norea\n\nSwit erland\n\nPortugal\n\nAustria\n\nHungary\n\nithuania\n\nGermany\n\nUnited States\n\nPoland\n\nat ia\n\nIsrael\n\nChile\n\n330\n\n310\n\n290\n\n2 0\n\n2 0\n\n230\n\n210\n\n190\n\nScore points **. iteracy**\n\nFinland\n\napan\n\nSweden\n\netherlands\n\norway\n\nenmark\n\nFlemish Region ( E\n\nSingapore\n\nEstonia\n\nSpain\n\nIreland\n\nCanada\n\nAustria\n\nEngland (U\n\nSlo ak Republic\n\new ealand\n\n**EC average**\n\nC echia\n\nSwit erland\n\nCroatia\n\norea\n\nFrance\n\nItaly\n\nat ia\n\nPortugal\n\nithuania\n\nGermany\n\nHungary\n\nIsrael\n\nPoland\n\nUnited States\n\nChile\n\n330\n\n310\n\n290\n\n2 0\n\n2 0\n\n230\n\n210\n\n190\n\nScore points **B. umeracy**\n\nFinland\n\napan\n\nSweden\n\netherlands\n\norway\n\nenmark\n\nFlemish Region ( E\n\nSpain\n\new ealand\n\nSlo ak Republic\n\nEngland (U\n\nIreland\n\nSingapore\n\nEstonia\n\nCanada\n\n**EC average**\n\nFrance\n\nAustria\n\nPortugal\n\nSwit erland\n\nC echia\n\norea\n\nItaly\n\nCroatia\n\nGermany\n\nHungary\n\nat ia\n\nUnited States\n\nithuania\n\nIsrael\n\nPoland\n\nChile\n\n330\n\n310\n\n290\n\n2 0\n\n2 0\n\n230\n\n210\n\n190\n\nScore points **C. daptive problem solving**","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p94.pdf"]},"response":{"errors":[],"extract_id":"extract_c481c553e3d9bbe63c7b884c6ce93cfb","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**92** \n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 2.19. Average proficiency in key information-processing skills, by parental education**\n\nLiteracy, numeracy and adaptive problem solving\n\nNote: Adults aged 16-65; does not include adults who were only administered the doorstep interview due to a language barrier, as information on parental education was not collected for those respondents (see Box 1.1 in Chapter 1). Respondents are categorised as having highly educated parents if at least one parent attained tertiary education; as having medium-educated parents if at least one parent attained upper secondary education and none of the parents attained tertiary education; and as having low-educated parents if neither parent attained upper secondary education. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the ote for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the average proficiency of adults with low-educated parents._ Source: Table A.2.12 (L, N, A) in Annex A.\n\nedium educated parents Highly educated parents ow educated parents\n\nFinland\n\nSweden\n\napan\n\netherlands\n\norway\n\nenmark\n\new ealand\n\nFlemish Region ( E\n\nEngland (U\n\nIreland\n\nSpain\n\nSlo ak Republic\n\nCanada\n\nEstonia\n\nCroatia\n\n**EC average**\n\nItaly\n\nFrance\n\nSingapore\n\nC echia\n\norea\n\nSwit erland\n\nPortugal\n\nAustria\n\nHungary\n\nithuania\n\nGermany\n\nUnited States\n\nPoland\n\nat ia\n\nIsrael\n\nChile\n\n330\n\n310\n\n290\n\n2 0\n\n2 0\n\n230\n\n210\n\n190\n\nScore points **. iteracy**\n\nFinland\n\napan\n\nSweden\n\netherlands\n\norway\n\nenmark\n\nFlemish Region ( E\n\nSingapore\n\nEstonia\n\nSpain\n\nIreland\n\nCanada\n\nAustria\n\nEngland (U\n\nSlo ak Republic\n\new ealand\n\n**EC average**\n\nC echia\n\nSwit erland\n\nCroatia\n\norea\n\nFrance\n\nItaly\n\nat ia\n\nPortugal\n\nithuania\n\nGermany\n\nHungary\n\nIsrael\n\nPoland\n\nUnited States\n\nChile\n\n330\n\n310\n\n290\n\n2 0\n\n2 0\n\n230\n\n210\n\n190\n\nScore points **B. umeracy**\n\nFinland\n\napan\n\nSweden\n\netherlands\n\norway\n\nenmark\n\nFlemish Region ( E\n\nSpain\n\new ealand\n\nSlo ak Republic\n\nEngland (U\n\nIreland\n\nSingapore\n\nEstonia\n\nCanada\n\n**EC average**\n\nFrance\n\nAustria\n\nPortugal\n\nSwit erland\n\nC echia\n\norea\n\nItaly\n\nCroatia\n\nGermany\n\nHungary\n\nat ia\n\nUnited States\n\nithuania\n\nIsrael\n\nPoland\n\nChile\n\n330\n\n310\n\n290\n\n2 0\n\n2 0\n\n230\n\n210\n\n190\n\nScore points **C. daptive problem solving**"],"full_content":"**92** \n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 2.19. Average proficiency in key information-processing skills, by parental education**\n\nLiteracy, numeracy and adaptive problem solving\n\nNote: Adults aged 16-65; does not include adults who were only administered the doorstep interview due to a language barrier, as information on parental education was not collected for those respondents (see Box 1.1 in Chapter 1). Respondents are categorised as having highly educated parents if at least one parent attained tertiary education; as having medium-educated parents if at least one parent attained upper secondary education and none of the parents attained tertiary education; and as having low-educated parents if neither parent attained upper secondary education. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the ote for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the average proficiency of adults with low-educated parents._ Source: Table A.2.12 (L, N, A) in Annex A.\n\nedium educated parents Highly educated parents ow educated parents\n\nFinland\n\nSweden\n\napan\n\netherlands\n\norway\n\nenmark\n\new ealand\n\nFlemish Region ( E\n\nEngland (U\n\nIreland\n\nSpain\n\nSlo ak Republic\n\nCanada\n\nEstonia\n\nCroatia\n\n**EC average**\n\nItaly\n\nFrance\n\nSingapore\n\nC echia\n\norea\n\nSwit erland\n\nPortugal\n\nAustria\n\nHungary\n\nithuania\n\nGermany\n\nUnited States\n\nPoland\n\nat ia\n\nIsrael\n\nChile\n\n330\n\n310\n\n290\n\n2 0\n\n2 0\n\n230\n\n210\n\n190\n\nScore points **. iteracy**\n\nFinland\n\napan\n\nSweden\n\netherlands\n\norway\n\nenmark\n\nFlemish Region ( E\n\nSingapore\n\nEstonia\n\nSpain\n\nIreland\n\nCanada\n\nAustria\n\nEngland (U\n\nSlo ak Republic\n\new ealand\n\n**EC average**\n\nC echia\n\nSwit erland\n\nCroatia\n\norea\n\nFrance\n\nItaly\n\nat ia\n\nPortugal\n\nithuania\n\nGermany\n\nHungary\n\nIsrael\n\nPoland\n\nUnited States\n\nChile\n\n330\n\n310\n\n290\n\n2 0\n\n2 0\n\n230\n\n210\n\n190\n\nScore points **B. umeracy**\n\nFinland\n\napan\n\nSweden\n\netherlands\n\norway\n\nenmark\n\nFlemish Region ( E\n\nSpain\n\new ealand\n\nSlo ak Republic\n\nEngland (U\n\nIreland\n\nSingapore\n\nEstonia\n\nCanada\n\n**EC average**\n\nFrance\n\nAustria\n\nPortugal\n\nSwit erland\n\nC echia\n\norea\n\nItaly\n\nCroatia\n\nGermany\n\nHungary\n\nat ia\n\nUnited States\n\nithuania\n\nIsrael\n\nPoland\n\nChile\n\n330\n\n310\n\n290\n\n2 0\n\n2 0\n\n230\n\n210\n\n190\n\nScore points **C. daptive problem solving**","publish_date":null,"title":"Figure 2.19. Average proficiency in key information-processing skills, by parental education","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p94.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p94.pdf"},"request":{"example_id":"chart/b263dc5d-en_p94","product_type":"parse","source_file_path":"b263dc5d-en_p94.pdf"},"started_at":"2026-11-04T19:46:35.335209"} +{"completed_at":"2026-11-04T19:46:38.852664","latency_in_ms":1798,"output":{"example_id":"chart/b263dc5d-en_p95","grounded_pages":[],"layout_pages":[],"markdown":" **93**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 2.20. Differences in key information-processing skills, by parental education**\n\nAdjusted and unadjusted differences in average literacy, numeracy and adaptive problem solving scores between adults (highly educated parents minus low-educated parents)\n\nNote: Adults aged 16-65; does not include adults who were only administered the doorstep interview due to a language barrier, as information on parental education was not collected for those respondents (see Box 1.1 in Chapter 1). Respondents are categorised as having highly educated parents if at least one parent attained tertiary education; as having medium-educated parents if at least one parent attained upper secondary education and none of the parents attained tertiary education; and as having low-educated parents if neither parent attained upper secondary education. Unadjusted differences are the differences between the two averages for each contrast category. Adjusted differences are based on a regression model that takes into account differences associated with gender, age, education, immigrant background and language spoken at home. All differences are statistically significant at the 5% level. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the ote for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the unadjusted difference between adults with highly educated and low-educated parents._ Source: Table A.2.12 (L, N, A) in Annex A.\n\nUnad u sted Ad usted\n\nGermany\n\nSwit erland\n\nUnited States\n\nAustria\n\nHungary\n\nat ia\n\nEstonia\n\nIsrael\n\nSingapore\n\nChile\n\nC echia\n\nFrance\n\nPortugal\n\nFlemish Region ( E\n\nCanada\n\n**EC average**\n\nFinland\n\norea\n\nEngland (U\n\nPoland\n\norway\n\napan\n\nenmark\n\nItaly\n\nCroatia\n\nIreland\n\nithuania\n\netherlands\n\new ealand\n\nSweden\n\nSlo ak Republic\n\nSpain\n\n80\n\n0\n\n0\n\n0\n\n0\n\n30\n\n20\n\n10\n\n0\n\n**. iteracy**\n\nGermany\n\nUnited States\n\nSwit erland\n\nHungary\n\nC echia\n\nChile\n\nat ia\n\nEstonia\n\nFrance\n\nAustria\n\nSingapore\n\nPortugal\n\nEngland (U\n\nFlemish Region ( E\n\nIsrael\n\n**EC average**\n\nCanada\n\nPoland\n\nFinland\n\norea\n\new ealand\n\netherlands\n\norway\n\nSlo ak Republic\n\nenmark\n\nCroatia\n\nItaly\n\nIreland\n\napan\n\nithuania\n\nSweden\n\nSpain\n\n80\n\n0\n\n0\n\n0\n\n0\n\n30\n\n20\n\n10\n\n0\n\n**B. umeracy**\n\nGermany\n\nSwit erland\n\nHungary\n\nat ia\n\nUnited States\n\nC echia\n\nAustria\n\nEstonia\n\nFlemish Region ( E\n\nChile\n\nSingapore\n\nFrance\n\nIsrael\n\nPortugal\n\nCanada\n\nFinland\n\n**EC average**\n\nEngland (U\n\nPoland\n\norea\n\norway\n\nenmark\n\new ealand\n\napan\n\netherlands\n\nCroatia\n\nItaly\n\nIreland\n\nithuania\n\nSweden\n\nSlo ak Republic\n\nSpain\n\n80\n\n0\n\n0\n\n0\n\n0\n\n30\n\n20\n\n10\n\n0\n\n**C. daptive problem solving**\n\nScore point difference\n\nScore point difference\n\nScore point difference","pages":[{"line_numbers":[],"markdown":" **93**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 2.20. Differences in key information-processing skills, by parental education**\n\nAdjusted and unadjusted differences in average literacy, numeracy and adaptive problem solving scores between adults (highly educated parents minus low-educated parents)\n\nNote: Adults aged 16-65; does not include adults who were only administered the doorstep interview due to a language barrier, as information on parental education was not collected for those respondents (see Box 1.1 in Chapter 1). Respondents are categorised as having highly educated parents if at least one parent attained tertiary education; as having medium-educated parents if at least one parent attained upper secondary education and none of the parents attained tertiary education; and as having low-educated parents if neither parent attained upper secondary education. Unadjusted differences are the differences between the two averages for each contrast category. Adjusted differences are based on a regression model that takes into account differences associated with gender, age, education, immigrant background and language spoken at home. All differences are statistically significant at the 5% level. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the ote for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the unadjusted difference between adults with highly educated and low-educated parents._ Source: Table A.2.12 (L, N, A) in Annex A.\n\nUnad u sted Ad usted\n\nGermany\n\nSwit erland\n\nUnited States\n\nAustria\n\nHungary\n\nat ia\n\nEstonia\n\nIsrael\n\nSingapore\n\nChile\n\nC echia\n\nFrance\n\nPortugal\n\nFlemish Region ( E\n\nCanada\n\n**EC average**\n\nFinland\n\norea\n\nEngland (U\n\nPoland\n\norway\n\napan\n\nenmark\n\nItaly\n\nCroatia\n\nIreland\n\nithuania\n\netherlands\n\new ealand\n\nSweden\n\nSlo ak Republic\n\nSpain\n\n80\n\n0\n\n0\n\n0\n\n0\n\n30\n\n20\n\n10\n\n0\n\n**. iteracy**\n\nGermany\n\nUnited States\n\nSwit erland\n\nHungary\n\nC echia\n\nChile\n\nat ia\n\nEstonia\n\nFrance\n\nAustria\n\nSingapore\n\nPortugal\n\nEngland (U\n\nFlemish Region ( E\n\nIsrael\n\n**EC average**\n\nCanada\n\nPoland\n\nFinland\n\norea\n\new ealand\n\netherlands\n\norway\n\nSlo ak Republic\n\nenmark\n\nCroatia\n\nItaly\n\nIreland\n\napan\n\nithuania\n\nSweden\n\nSpain\n\n80\n\n0\n\n0\n\n0\n\n0\n\n30\n\n20\n\n10\n\n0\n\n**B. umeracy**\n\nGermany\n\nSwit erland\n\nHungary\n\nat ia\n\nUnited States\n\nC echia\n\nAustria\n\nEstonia\n\nFlemish Region ( E\n\nChile\n\nSingapore\n\nFrance\n\nIsrael\n\nPortugal\n\nCanada\n\nFinland\n\n**EC average**\n\nEngland (U\n\nPoland\n\norea\n\norway\n\nenmark\n\new ealand\n\napan\n\netherlands\n\nCroatia\n\nItaly\n\nIreland\n\nithuania\n\nSweden\n\nSlo ak Republic\n\nSpain\n\n80\n\n0\n\n0\n\n0\n\n0\n\n30\n\n20\n\n10\n\n0\n\n**C. daptive problem solving**\n\nScore point difference\n\nScore point difference\n\nScore point difference","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":" **93**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 2.20. Differences in key information-processing skills, by parental education**\n\nAdjusted and unadjusted differences in average literacy, numeracy and adaptive problem solving scores between adults (highly educated parents minus low-educated parents)\n\nNote: Adults aged 16-65; does not include adults who were only administered the doorstep interview due to a language barrier, as information on parental education was not collected for those respondents (see Box 1.1 in Chapter 1). Respondents are categorised as having highly educated parents if at least one parent attained tertiary education; as having medium-educated parents if at least one parent attained upper secondary education and none of the parents attained tertiary education; and as having low-educated parents if neither parent attained upper secondary education. Unadjusted differences are the differences between the two averages for each contrast category. Adjusted differences are based on a regression model that takes into account differences associated with gender, age, education, immigrant background and language spoken at home. All differences are statistically significant at the 5% level. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the ote for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the unadjusted difference between adults with highly educated and low-educated parents._ Source: Table A.2.12 (L, N, A) in Annex A.\n\nUnad u sted Ad usted\n\nGermany\n\nSwit erland\n\nUnited States\n\nAustria\n\nHungary\n\nat ia\n\nEstonia\n\nIsrael\n\nSingapore\n\nChile\n\nC echia\n\nFrance\n\nPortugal\n\nFlemish Region ( E\n\nCanada\n\n**EC average**\n\nFinland\n\norea\n\nEngland (U\n\nPoland\n\norway\n\napan\n\nenmark\n\nItaly\n\nCroatia\n\nIreland\n\nithuania\n\netherlands\n\new ealand\n\nSweden\n\nSlo ak Republic\n\nSpain\n\n80\n\n0\n\n0\n\n0\n\n0\n\n30\n\n20\n\n10\n\n0\n\n**. iteracy**\n\nGermany\n\nUnited States\n\nSwit erland\n\nHungary\n\nC echia\n\nChile\n\nat ia\n\nEstonia\n\nFrance\n\nAustria\n\nSingapore\n\nPortugal\n\nEngland (U\n\nFlemish Region ( E\n\nIsrael\n\n**EC average**\n\nCanada\n\nPoland\n\nFinland\n\norea\n\new ealand\n\netherlands\n\norway\n\nSlo ak Republic\n\nenmark\n\nCroatia\n\nItaly\n\nIreland\n\napan\n\nithuania\n\nSweden\n\nSpain\n\n80\n\n0\n\n0\n\n0\n\n0\n\n30\n\n20\n\n10\n\n0\n\n**B. umeracy**\n\nGermany\n\nSwit erland\n\nHungary\n\nat ia\n\nUnited States\n\nC echia\n\nAustria\n\nEstonia\n\nFlemish Region ( E\n\nChile\n\nSingapore\n\nFrance\n\nIsrael\n\nPortugal\n\nCanada\n\nFinland\n\n**EC average**\n\nEngland (U\n\nPoland\n\norea\n\norway\n\nenmark\n\new ealand\n\napan\n\netherlands\n\nCroatia\n\nItaly\n\nIreland\n\nithuania\n\nSweden\n\nSlo ak Republic\n\nSpain\n\n80\n\n0\n\n0\n\n0\n\n0\n\n30\n\n20\n\n10\n\n0\n\n**C. daptive problem solving**\n\nScore point difference\n\nScore point difference\n\nScore point difference","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p95.pdf"]},"response":{"errors":[],"extract_id":"extract_ad4ba9b521ee96c24ee3a73411c7d225","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":[" **93**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 2.20. Differences in key information-processing skills, by parental education**\n\nAdjusted and unadjusted differences in average literacy, numeracy and adaptive problem solving scores between adults (highly educated parents minus low-educated parents)\n\nNote: Adults aged 16-65; does not include adults who were only administered the doorstep interview due to a language barrier, as information on parental education was not collected for those respondents (see Box 1.1 in Chapter 1). Respondents are categorised as having highly educated parents if at least one parent attained tertiary education; as having medium-educated parents if at least one parent attained upper secondary education and none of the parents attained tertiary education; and as having low-educated parents if neither parent attained upper secondary education. Unadjusted differences are the differences between the two averages for each contrast category. Adjusted differences are based on a regression model that takes into account differences associated with gender, age, education, immigrant background and language spoken at home. All differences are statistically significant at the 5% level. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the ote for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the unadjusted difference between adults with highly educated and low-educated parents._ Source: Table A.2.12 (L, N, A) in Annex A.\n\nUnad u sted Ad usted\n\nGermany\n\nSwit erland\n\nUnited States\n\nAustria\n\nHungary\n\nat ia\n\nEstonia\n\nIsrael\n\nSingapore\n\nChile\n\nC echia\n\nFrance\n\nPortugal\n\nFlemish Region ( E\n\nCanada\n\n**EC average**\n\nFinland\n\norea\n\nEngland (U\n\nPoland\n\norway\n\napan\n\nenmark\n\nItaly\n\nCroatia\n\nIreland\n\nithuania\n\netherlands\n\new ealand\n\nSweden\n\nSlo ak Republic\n\nSpain\n\n80\n\n0\n\n0\n\n0\n\n0\n\n30\n\n20\n\n10\n\n0\n\n**. iteracy**\n\nGermany\n\nUnited States\n\nSwit erland\n\nHungary\n\nC echia\n\nChile\n\nat ia\n\nEstonia\n\nFrance\n\nAustria\n\nSingapore\n\nPortugal\n\nEngland (U\n\nFlemish Region ( E\n\nIsrael\n\n**EC average**\n\nCanada\n\nPoland\n\nFinland\n\norea\n\new ealand\n\netherlands\n\norway\n\nSlo ak Republic\n\nenmark\n\nCroatia\n\nItaly\n\nIreland\n\napan\n\nithuania\n\nSweden\n\nSpain\n\n80\n\n0\n\n0\n\n0\n\n0\n\n30\n\n20\n\n10\n\n0\n\n**B. umeracy**\n\nGermany\n\nSwit erland\n\nHungary\n\nat ia\n\nUnited States\n\nC echia\n\nAustria\n\nEstonia\n\nFlemish Region ( E\n\nChile\n\nSingapore\n\nFrance\n\nIsrael\n\nPortugal\n\nCanada\n\nFinland\n\n**EC average**\n\nEngland (U\n\nPoland\n\norea\n\norway\n\nenmark\n\new ealand\n\napan\n\netherlands\n\nCroatia\n\nItaly\n\nIreland\n\nithuania\n\nSweden\n\nSlo ak Republic\n\nSpain\n\n80\n\n0\n\n0\n\n0\n\n0\n\n30\n\n20\n\n10\n\n0\n\n**C. daptive problem solving**\n\nScore point difference\n\nScore point difference\n\nScore point difference"],"full_content":" **93**\n\nDO ADULTS HAVE THE SKILLS THEY NEED TO THRIVE IN A CHANGING WORLD? © OECD 2024\n\n### **Figure 2.20. Differences in key information-processing skills, by parental education**\n\nAdjusted and unadjusted differences in average literacy, numeracy and adaptive problem solving scores between adults (highly educated parents minus low-educated parents)\n\nNote: Adults aged 16-65; does not include adults who were only administered the doorstep interview due to a language barrier, as information on parental education was not collected for those respondents (see Box 1.1 in Chapter 1). Respondents are categorised as having highly educated parents if at least one parent attained tertiary education; as having medium-educated parents if at least one parent attained upper secondary education and none of the parents attained tertiary education; and as having low-educated parents if neither parent attained upper secondary education. Unadjusted differences are the differences between the two averages for each contrast category. Adjusted differences are based on a regression model that takes into account differences associated with gender, age, education, immigrant background and language spoken at home. All differences are statistically significant at the 5% level. \\*Caution is required in interpreting results due to the high share of respondents with unusual response patterns. See the ote for Poland in the Reader’s Guide. _Countries and economies are ranked in descending order of the unadjusted difference between adults with highly educated and low-educated parents._ Source: Table A.2.12 (L, N, A) in Annex A.\n\nUnad u sted Ad usted\n\nGermany\n\nSwit erland\n\nUnited States\n\nAustria\n\nHungary\n\nat ia\n\nEstonia\n\nIsrael\n\nSingapore\n\nChile\n\nC echia\n\nFrance\n\nPortugal\n\nFlemish Region ( E\n\nCanada\n\n**EC average**\n\nFinland\n\norea\n\nEngland (U\n\nPoland\n\norway\n\napan\n\nenmark\n\nItaly\n\nCroatia\n\nIreland\n\nithuania\n\netherlands\n\new ealand\n\nSweden\n\nSlo ak Republic\n\nSpain\n\n80\n\n0\n\n0\n\n0\n\n0\n\n30\n\n20\n\n10\n\n0\n\n**. iteracy**\n\nGermany\n\nUnited States\n\nSwit erland\n\nHungary\n\nC echia\n\nChile\n\nat ia\n\nEstonia\n\nFrance\n\nAustria\n\nSingapore\n\nPortugal\n\nEngland (U\n\nFlemish Region ( E\n\nIsrael\n\n**EC average**\n\nCanada\n\nPoland\n\nFinland\n\norea\n\new ealand\n\netherlands\n\norway\n\nSlo ak Republic\n\nenmark\n\nCroatia\n\nItaly\n\nIreland\n\napan\n\nithuania\n\nSweden\n\nSpain\n\n80\n\n0\n\n0\n\n0\n\n0\n\n30\n\n20\n\n10\n\n0\n\n**B. umeracy**\n\nGermany\n\nSwit erland\n\nHungary\n\nat ia\n\nUnited States\n\nC echia\n\nAustria\n\nEstonia\n\nFlemish Region ( E\n\nChile\n\nSingapore\n\nFrance\n\nIsrael\n\nPortugal\n\nCanada\n\nFinland\n\n**EC average**\n\nEngland (U\n\nPoland\n\norea\n\norway\n\nenmark\n\new ealand\n\napan\n\netherlands\n\nCroatia\n\nItaly\n\nIreland\n\nithuania\n\nSweden\n\nSlo ak Republic\n\nSpain\n\n80\n\n0\n\n0\n\n0\n\n0\n\n30\n\n20\n\n10\n\n0\n\n**C. daptive problem solving**\n\nScore point difference\n\nScore point difference\n\nScore point difference","publish_date":null,"title":"Figure 2.20. Differences in key information-processing skills, by parental education","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p95.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b263dc5d-en_p95.pdf"},"request":{"example_id":"chart/b263dc5d-en_p95","product_type":"parse","source_file_path":"b263dc5d-en_p95.pdf"},"started_at":"2026-11-04T19:46:37.053970"} +{"completed_at":"2026-11-04T19:46:41.624289","latency_in_ms":3110,"output":{"example_id":"chart/b3cd580a-3656-44ed-838a-5f2996ff6fc9_p18","grounded_pages":[],"layout_pages":[],"markdown":"18 © The Economist Intelligence Unit Limited 2024\n\nIndustry outlook 2025 Challenges, opportunities and trends to watch in six sectors\n\n# Increased competition but greater EV profitability\n\n% change, year on year\n\n2024 2025\n\nAsia-Pacific\n\nLatin America\n\nEurope\n\nMiddle East and Africa\n\nNorth America\n\nWorld\n\n0 50 100 150 200 250 300\n\n38 34\n\n195 53\n\n26 30\n\n288 58\n\n32 26\n\n24 16\n\nSource: International Energy Agency; EIU forecasts. Copyright © The Economist Intelligence Unit 2024. All rights reserved.\n\n### Growth in new EV registrations will slow further in 2025\n\nThese trends pose a huge challenge for carmakers as supply chains fracture and costs multiply. Foreign carmakers in China, including Germany’s Volkswagen, will continue to see their Chinese market share tumble in 2025 and beyond. Meanwhile, China’s EV-makers will try to increase exports despite rising trade barriers. BYD, now the world’s biggest EV-maker ahead of Tesla (US), aims to sell 1m models outside China next year, helped by new plants in Brazil and Hungary. As competition intensifies, several automakers will miss the hefty targets that they set to increase the share of EVs in their sales. For example Volvo Cars (Sweden-based, but owned by China’s Geely) aimed to have EVs to make up half of its vehicle sales by 2025, but had achieved just 16% in full-year 2023.\n\nDespite these challenges, EVs are on the path to profitability. Tesla has been profitable since 2019, and 2025 may well be the year when it is joined by legacy automakers. Back in 2022 the finance head of BMW (Germany) said that the profitability of the premium carmaker’s EVs would exceed that of fossil-fuel cars in 2025. Although this milestone has now been pushed to 2026, strong growth in global EV sales, combined with lower prices for commodities such as lithium, could mark a turning point; General Motors (US) and Stellantis (Netherlands) say that they are on track to become profitable on EVs by end-2024. Even so, EVs will remain less profitable than internal combustion engine vehicles, with the investment needed for the EV transition burdening many carmakers’ finances in 2025.","pages":[{"line_numbers":[],"markdown":"18 © The Economist Intelligence Unit Limited 2024\n\nIndustry outlook 2025 Challenges, opportunities and trends to watch in six sectors\n\n# Increased competition but greater EV profitability\n\n% change, year on year\n\n2024 2025\n\nAsia-Pacific\n\nLatin America\n\nEurope\n\nMiddle East and Africa\n\nNorth America\n\nWorld\n\n0 50 100 150 200 250 300\n\n38 34\n\n195 53\n\n26 30\n\n288 58\n\n32 26\n\n24 16\n\nSource: International Energy Agency; EIU forecasts. Copyright © The Economist Intelligence Unit 2024. All rights reserved.\n\n### Growth in new EV registrations will slow further in 2025\n\nThese trends pose a huge challenge for carmakers as supply chains fracture and costs multiply. Foreign carmakers in China, including Germany’s Volkswagen, will continue to see their Chinese market share tumble in 2025 and beyond. Meanwhile, China’s EV-makers will try to increase exports despite rising trade barriers. BYD, now the world’s biggest EV-maker ahead of Tesla (US), aims to sell 1m models outside China next year, helped by new plants in Brazil and Hungary. As competition intensifies, several automakers will miss the hefty targets that they set to increase the share of EVs in their sales. For example Volvo Cars (Sweden-based, but owned by China’s Geely) aimed to have EVs to make up half of its vehicle sales by 2025, but had achieved just 16% in full-year 2023.\n\nDespite these challenges, EVs are on the path to profitability. Tesla has been profitable since 2019, and 2025 may well be the year when it is joined by legacy automakers. Back in 2022 the finance head of BMW (Germany) said that the profitability of the premium carmaker’s EVs would exceed that of fossil-fuel cars in 2025. Although this milestone has now been pushed to 2026, strong growth in global EV sales, combined with lower prices for commodities such as lithium, could mark a turning point; General Motors (US) and Stellantis (Netherlands) say that they are on track to become profitable on EVs by end-2024. Even so, EVs will remain less profitable than internal combustion engine vehicles, with the investment needed for the EV transition burdening many carmakers’ finances in 2025.","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"18 © The Economist Intelligence Unit Limited 2024\n\nIndustry outlook 2025 Challenges, opportunities and trends to watch in six sectors\n\n# Increased competition but greater EV profitability\n\n% change, year on year\n\n2024 2025\n\nAsia-Pacific\n\nLatin America\n\nEurope\n\nMiddle East and Africa\n\nNorth America\n\nWorld\n\n0 50 100 150 200 250 300\n\n38 34\n\n195 53\n\n26 30\n\n288 58\n\n32 26\n\n24 16\n\nSource: International Energy Agency; EIU forecasts. Copyright © The Economist Intelligence Unit 2024. All rights reserved.\n\n### Growth in new EV registrations will slow further in 2025\n\nThese trends pose a huge challenge for carmakers as supply chains fracture and costs multiply. Foreign carmakers in China, including Germany’s Volkswagen, will continue to see their Chinese market share tumble in 2025 and beyond. Meanwhile, China’s EV-makers will try to increase exports despite rising trade barriers. BYD, now the world’s biggest EV-maker ahead of Tesla (US), aims to sell 1m models outside China next year, helped by new plants in Brazil and Hungary. As competition intensifies, several automakers will miss the hefty targets that they set to increase the share of EVs in their sales. For example Volvo Cars (Sweden-based, but owned by China’s Geely) aimed to have EVs to make up half of its vehicle sales by 2025, but had achieved just 16% in full-year 2023.\n\nDespite these challenges, EVs are on the path to profitability. Tesla has been profitable since 2019, and 2025 may well be the year when it is joined by legacy automakers. Back in 2022 the finance head of BMW (Germany) said that the profitability of the premium carmaker’s EVs would exceed that of fossil-fuel cars in 2025. Although this milestone has now been pushed to 2026, strong growth in global EV sales, combined with lower prices for commodities such as lithium, could mark a turning point; General Motors (US) and Stellantis (Netherlands) say that they are on track to become profitable on EVs by end-2024. Even so, EVs will remain less profitable than internal combustion engine vehicles, with the investment needed for the EV transition burdening many carmakers’ finances in 2025.","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b3cd580a-3656-44ed-838a-5f2996ff6fc9_p18.pdf"]},"response":{"errors":[],"extract_id":"extract_8db67d9d1881ebf1b1329e0a22b29800","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["18 © The Economist Intelligence Unit Limited 2024\n\nIndustry outlook 2025 Challenges, opportunities and trends to watch in six sectors\n\n# Increased competition but greater EV profitability\n\n% change, year on year\n\n2024 2025\n\nAsia-Pacific\n\nLatin America\n\nEurope\n\nMiddle East and Africa\n\nNorth America\n\nWorld\n\n0 50 100 150 200 250 300\n\n38 34\n\n195 53\n\n26 30\n\n288 58\n\n32 26\n\n24 16\n\nSource: International Energy Agency; EIU forecasts. Copyright © The Economist Intelligence Unit 2024. All rights reserved.\n\n### Growth in new EV registrations will slow further in 2025\n\nThese trends pose a huge challenge for carmakers as supply chains fracture and costs multiply. Foreign carmakers in China, including Germany’s Volkswagen, will continue to see their Chinese market share tumble in 2025 and beyond. Meanwhile, China’s EV-makers will try to increase exports despite rising trade barriers. BYD, now the world’s biggest EV-maker ahead of Tesla (US), aims to sell 1m models outside China next year, helped by new plants in Brazil and Hungary. As competition intensifies, several automakers will miss the hefty targets that they set to increase the share of EVs in their sales. For example Volvo Cars (Sweden-based, but owned by China’s Geely) aimed to have EVs to make up half of its vehicle sales by 2025, but had achieved just 16% in full-year 2023.\n\nDespite these challenges, EVs are on the path to profitability. Tesla has been profitable since 2019, and 2025 may well be the year when it is joined by legacy automakers. Back in 2022 the finance head of BMW (Germany) said that the profitability of the premium carmaker’s EVs would exceed that of fossil-fuel cars in 2025. Although this milestone has now been pushed to 2026, strong growth in global EV sales, combined with lower prices for commodities such as lithium, could mark a turning point; General Motors (US) and Stellantis (Netherlands) say that they are on track to become profitable on EVs by end-2024. Even so, EVs will remain less profitable than internal combustion engine vehicles, with the investment needed for the EV transition burdening many carmakers’ finances in 2025."],"full_content":"18 © The Economist Intelligence Unit Limited 2024\n\nIndustry outlook 2025 Challenges, opportunities and trends to watch in six sectors\n\n# Increased competition but greater EV profitability\n\n% change, year on year\n\n2024 2025\n\nAsia-Pacific\n\nLatin America\n\nEurope\n\nMiddle East and Africa\n\nNorth America\n\nWorld\n\n0 50 100 150 200 250 300\n\n38 34\n\n195 53\n\n26 30\n\n288 58\n\n32 26\n\n24 16\n\nSource: International Energy Agency; EIU forecasts. Copyright © The Economist Intelligence Unit 2024. All rights reserved.\n\n### Growth in new EV registrations will slow further in 2025\n\nThese trends pose a huge challenge for carmakers as supply chains fracture and costs multiply. Foreign carmakers in China, including Germany’s Volkswagen, will continue to see their Chinese market share tumble in 2025 and beyond. Meanwhile, China’s EV-makers will try to increase exports despite rising trade barriers. BYD, now the world’s biggest EV-maker ahead of Tesla (US), aims to sell 1m models outside China next year, helped by new plants in Brazil and Hungary. As competition intensifies, several automakers will miss the hefty targets that they set to increase the share of EVs in their sales. For example Volvo Cars (Sweden-based, but owned by China’s Geely) aimed to have EVs to make up half of its vehicle sales by 2025, but had achieved just 16% in full-year 2023.\n\nDespite these challenges, EVs are on the path to profitability. Tesla has been profitable since 2019, and 2025 may well be the year when it is joined by legacy automakers. Back in 2022 the finance head of BMW (Germany) said that the profitability of the premium carmaker’s EVs would exceed that of fossil-fuel cars in 2025. Although this milestone has now been pushed to 2026, strong growth in global EV sales, combined with lower prices for commodities such as lithium, could mark a turning point; General Motors (US) and Stellantis (Netherlands) say that they are on track to become profitable on EVs by end-2024. Even so, EVs will remain less profitable than internal combustion engine vehicles, with the investment needed for the EV transition burdening many carmakers’ finances in 2025.","publish_date":null,"title":"Increased competition but greater EV profitability","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b3cd580a-3656-44ed-838a-5f2996ff6fc9_p18.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b3cd580a-3656-44ed-838a-5f2996ff6fc9_p18.pdf"},"request":{"example_id":"chart/b3cd580a-3656-44ed-838a-5f2996ff6fc9_p18","product_type":"parse","source_file_path":"b3cd580a-3656-44ed-838a-5f2996ff6fc9_p18.pdf"},"started_at":"2026-11-04T19:46:38.513506"} +{"completed_at":"2026-11-04T19:46:41.217533","latency_in_ms":1038,"output":{"example_id":"chart/b3cd580a-3656-44ed-838a-5f2996ff6fc9_p23","grounded_pages":[],"layout_pages":[],"markdown":"23 © The Economist Intelligence Unit Limited 2024\n\nIndustry outlook 2025 Challenges, opportunities and trends to watch in six sectors\n\nin March 2024 Dollar Tree, a US discount chain, announced plans to cut 1,000 stores in the US in 2024-25, and in the second quarter McDonalds, a US fast-food chain, reported its first decline in global sales since 2020.\n\nThe retail growth landscape will vary significantly according to region. Asian and the Gulf countries are expected to outpace other regions in retail volume growth, driven by large, young consumer bases with rising incomes, urbanisation and a surge in online shopping.\n\n−8\n\n−6\n\n−4\n\n−2\n\n0\n\n2\n\n4\n\nUSMCA Europe Asia-Pacific South America MEA\n\n2024 2025\n\nSource: EIU.\n\nAsia will lead global retail sales growth, and South America will rebound in 2025\n\nRetail sales volume growth rate, %\n\nCopyright © The Economist Intelligence Unit 2024. All rights reserved.\n\n# While Chinese retail slows, India offers a silver lining\n\nReal growth in retail sales in Asia will outpace that of other regions, but the star of the show will not be China. The world’s second-largest consumer market is expected to grow by nearly 4% year on year in real terms in 2025—the slowest rate since 2022, and markedly so compared with the historical trend (averaging 7% in 2015-19). Income shocks from the prolonged lockdowns, a property market downturn, job market weakness and an ageing population have all contributed to lasting pessimism in the market.\n\nMeanwhile, we expect growth to average about 5% year on year in India (down from 5.3% in 2015-19). For businesses, the opportunities will be twofold. At the lower end of the market, fast-moving consumer goods makers such as PepsiCo and P&G (US) are banking on increased spending by the country’s sizeable rural population (65% of the total) to offset some of the slowdown in China. Recent economic data suggest a strong propensity among this population to spend on packaged goods, cosmetics, toiletries and household goods.","pages":[{"line_numbers":[],"markdown":"23 © The Economist Intelligence Unit Limited 2024\n\nIndustry outlook 2025 Challenges, opportunities and trends to watch in six sectors\n\nin March 2024 Dollar Tree, a US discount chain, announced plans to cut 1,000 stores in the US in 2024-25, and in the second quarter McDonalds, a US fast-food chain, reported its first decline in global sales since 2020.\n\nThe retail growth landscape will vary significantly according to region. Asian and the Gulf countries are expected to outpace other regions in retail volume growth, driven by large, young consumer bases with rising incomes, urbanisation and a surge in online shopping.\n\n−8\n\n−6\n\n−4\n\n−2\n\n0\n\n2\n\n4\n\nUSMCA Europe Asia-Pacific South America MEA\n\n2024 2025\n\nSource: EIU.\n\nAsia will lead global retail sales growth, and South America will rebound in 2025\n\nRetail sales volume growth rate, %\n\nCopyright © The Economist Intelligence Unit 2024. All rights reserved.\n\n# While Chinese retail slows, India offers a silver lining\n\nReal growth in retail sales in Asia will outpace that of other regions, but the star of the show will not be China. The world’s second-largest consumer market is expected to grow by nearly 4% year on year in real terms in 2025—the slowest rate since 2022, and markedly so compared with the historical trend (averaging 7% in 2015-19). Income shocks from the prolonged lockdowns, a property market downturn, job market weakness and an ageing population have all contributed to lasting pessimism in the market.\n\nMeanwhile, we expect growth to average about 5% year on year in India (down from 5.3% in 2015-19). For businesses, the opportunities will be twofold. At the lower end of the market, fast-moving consumer goods makers such as PepsiCo and P&G (US) are banking on increased spending by the country’s sizeable rural population (65% of the total) to offset some of the slowdown in China. Recent economic data suggest a strong propensity among this population to spend on packaged goods, cosmetics, toiletries and household goods.","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"23 © The Economist Intelligence Unit Limited 2024\n\nIndustry outlook 2025 Challenges, opportunities and trends to watch in six sectors\n\nin March 2024 Dollar Tree, a US discount chain, announced plans to cut 1,000 stores in the US in 2024-25, and in the second quarter McDonalds, a US fast-food chain, reported its first decline in global sales since 2020.\n\nThe retail growth landscape will vary significantly according to region. Asian and the Gulf countries are expected to outpace other regions in retail volume growth, driven by large, young consumer bases with rising incomes, urbanisation and a surge in online shopping.\n\n−8\n\n−6\n\n−4\n\n−2\n\n0\n\n2\n\n4\n\nUSMCA Europe Asia-Pacific South America MEA\n\n2024 2025\n\nSource: EIU.\n\nAsia will lead global retail sales growth, and South America will rebound in 2025\n\nRetail sales volume growth rate, %\n\nCopyright © The Economist Intelligence Unit 2024. All rights reserved.\n\n# While Chinese retail slows, India offers a silver lining\n\nReal growth in retail sales in Asia will outpace that of other regions, but the star of the show will not be China. The world’s second-largest consumer market is expected to grow by nearly 4% year on year in real terms in 2025—the slowest rate since 2022, and markedly so compared with the historical trend (averaging 7% in 2015-19). Income shocks from the prolonged lockdowns, a property market downturn, job market weakness and an ageing population have all contributed to lasting pessimism in the market.\n\nMeanwhile, we expect growth to average about 5% year on year in India (down from 5.3% in 2015-19). For businesses, the opportunities will be twofold. At the lower end of the market, fast-moving consumer goods makers such as PepsiCo and P&G (US) are banking on increased spending by the country’s sizeable rural population (65% of the total) to offset some of the slowdown in China. Recent economic data suggest a strong propensity among this population to spend on packaged goods, cosmetics, toiletries and household goods.","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b3cd580a-3656-44ed-838a-5f2996ff6fc9_p23.pdf"]},"response":{"errors":[],"extract_id":"extract_7f4af26773887b1baa803ce07d4526e7","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["23 © The Economist Intelligence Unit Limited 2024\n\nIndustry outlook 2025 Challenges, opportunities and trends to watch in six sectors\n\nin March 2024 Dollar Tree, a US discount chain, announced plans to cut 1,000 stores in the US in 2024-25, and in the second quarter McDonalds, a US fast-food chain, reported its first decline in global sales since 2020.\n\nThe retail growth landscape will vary significantly according to region. Asian and the Gulf countries are expected to outpace other regions in retail volume growth, driven by large, young consumer bases with rising incomes, urbanisation and a surge in online shopping.\n\n−8\n\n−6\n\n−4\n\n−2\n\n0\n\n2\n\n4\n\nUSMCA Europe Asia-Pacific South America MEA\n\n2024 2025\n\nSource: EIU.\n\nAsia will lead global retail sales growth, and South America will rebound in 2025\n\nRetail sales volume growth rate, %\n\nCopyright © The Economist Intelligence Unit 2024. All rights reserved.\n\n# While Chinese retail slows, India offers a silver lining\n\nReal growth in retail sales in Asia will outpace that of other regions, but the star of the show will not be China. The world’s second-largest consumer market is expected to grow by nearly 4% year on year in real terms in 2025—the slowest rate since 2022, and markedly so compared with the historical trend (averaging 7% in 2015-19). Income shocks from the prolonged lockdowns, a property market downturn, job market weakness and an ageing population have all contributed to lasting pessimism in the market.\n\nMeanwhile, we expect growth to average about 5% year on year in India (down from 5.3% in 2015-19). For businesses, the opportunities will be twofold. At the lower end of the market, fast-moving consumer goods makers such as PepsiCo and P&G (US) are banking on increased spending by the country’s sizeable rural population (65% of the total) to offset some of the slowdown in China. Recent economic data suggest a strong propensity among this population to spend on packaged goods, cosmetics, toiletries and household goods."],"full_content":"23 © The Economist Intelligence Unit Limited 2024\n\nIndustry outlook 2025 Challenges, opportunities and trends to watch in six sectors\n\nin March 2024 Dollar Tree, a US discount chain, announced plans to cut 1,000 stores in the US in 2024-25, and in the second quarter McDonalds, a US fast-food chain, reported its first decline in global sales since 2020.\n\nThe retail growth landscape will vary significantly according to region. Asian and the Gulf countries are expected to outpace other regions in retail volume growth, driven by large, young consumer bases with rising incomes, urbanisation and a surge in online shopping.\n\n−8\n\n−6\n\n−4\n\n−2\n\n0\n\n2\n\n4\n\nUSMCA Europe Asia-Pacific South America MEA\n\n2024 2025\n\nSource: EIU.\n\nAsia will lead global retail sales growth, and South America will rebound in 2025\n\nRetail sales volume growth rate, %\n\nCopyright © The Economist Intelligence Unit 2024. All rights reserved.\n\n# While Chinese retail slows, India offers a silver lining\n\nReal growth in retail sales in Asia will outpace that of other regions, but the star of the show will not be China. The world’s second-largest consumer market is expected to grow by nearly 4% year on year in real terms in 2025—the slowest rate since 2022, and markedly so compared with the historical trend (averaging 7% in 2015-19). Income shocks from the prolonged lockdowns, a property market downturn, job market weakness and an ageing population have all contributed to lasting pessimism in the market.\n\nMeanwhile, we expect growth to average about 5% year on year in India (down from 5.3% in 2015-19). For businesses, the opportunities will be twofold. At the lower end of the market, fast-moving consumer goods makers such as PepsiCo and P&G (US) are banking on increased spending by the country’s sizeable rural population (65% of the total) to offset some of the slowdown in China. Recent economic data suggest a strong propensity among this population to spend on packaged goods, cosmetics, toiletries and household goods.","publish_date":null,"title":"While Chinese retail slows, India offers a silver lining","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b3cd580a-3656-44ed-838a-5f2996ff6fc9_p23.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b3cd580a-3656-44ed-838a-5f2996ff6fc9_p23.pdf"},"request":{"example_id":"chart/b3cd580a-3656-44ed-838a-5f2996ff6fc9_p23","product_type":"parse","source_file_path":"b3cd580a-3656-44ed-838a-5f2996ff6fc9_p23.pdf"},"started_at":"2026-11-04T19:46:40.179391"} +{"completed_at":"2026-11-04T19:31:07.451465","latency_in_ms":995,"output":{"example_id":"chart/b3cd580a-3656-44ed-838a-5f2996ff6fc9_p26","grounded_pages":[],"layout_pages":[],"markdown":"26 © The Economist Intelligence Unit Limited 2024\n\nIndustry outlook 2025 Challenges, opportunities and trends to watch in six sectors\n\n**Junk food:** Several countries will take steps to reduce junk food consumption in 2025 in an effort to address rising obesity rates. The UK plans to ban television advertisements for junk food before 9 pm, and on online ads for products that have high fat, sugar and salt content from October 1st 2025. Norway is expected to announce a similar ban, which would come into effect in 2025. Other governments will introduce taxes to discourage consumption while also increasing tax revenue. Slovakia will impose a sugar tax from January, and Vietnam is expected to pass a regulation to tax soft drinks that contain more than 5g of sugar per 100 ml.\n\n**EU travel rules:** After much delay, the European Travel Information and Authorisation System (ETIAS) will come into effect from mid-2025. After that date travellers from non- EU countries that do not currently require a visa to enter the EU will have to apply for travel authorisation before entry. The authorisation, which will cost €7 for those who are not exempt, will apply to around 1.4bn tourists a year from countries including the UK, Australia and the US.\n\n**Green trade crackdown:** New regulations will affect global trade. Low carbon technology from China will face higher tariffs and other\n\n# What to watch\n\nWith home sales being a major driver of consumer expenditure on household goods, including furniture and white goods, we expect spending on these items to be subdued in 2025. Inflation-struck consumers are likely to prioritise spending on essentials, as well as\n\nleisure and entertainment activities, over other discretionary goods such as clothing. Spending on consumer services such as travel, recreation and culture activities have fared well in the past year, and will continue to do so into 2025.\n\n2020 2021 2022 2023 2024 2025\n\n100\n\n110\n\n120\n\n130\n\n140\n\n150\n\n160\n\n170\n\n180\n\nSource: EIU.\n\nHotels & restaurants\n\nFood, beverages & tobacco Clothing & footwear\n\nHousehold goods & services\n\nLeisure & education Housing & household fuels\n\nHotels & restaurants\n\nFood, beverages & tobacco Clothing & footwear\n\nHousehold goods & services\n\nLeisure & education Housing & household fuels\n\n### Consumer spending on household goods & furniture will be sluggish\n\nGlobal consumer spending in US$, 2020=100\n\nCopyright © The Economist Intelligence Unit 2024. All rights reserved.","pages":[{"line_numbers":[],"markdown":"26 © The Economist Intelligence Unit Limited 2024\n\nIndustry outlook 2025 Challenges, opportunities and trends to watch in six sectors\n\n**Junk food:** Several countries will take steps to reduce junk food consumption in 2025 in an effort to address rising obesity rates. The UK plans to ban television advertisements for junk food before 9 pm, and on online ads for products that have high fat, sugar and salt content from October 1st 2025. Norway is expected to announce a similar ban, which would come into effect in 2025. Other governments will introduce taxes to discourage consumption while also increasing tax revenue. Slovakia will impose a sugar tax from January, and Vietnam is expected to pass a regulation to tax soft drinks that contain more than 5g of sugar per 100 ml.\n\n**EU travel rules:** After much delay, the European Travel Information and Authorisation System (ETIAS) will come into effect from mid-2025. After that date travellers from non- EU countries that do not currently require a visa to enter the EU will have to apply for travel authorisation before entry. The authorisation, which will cost €7 for those who are not exempt, will apply to around 1.4bn tourists a year from countries including the UK, Australia and the US.\n\n**Green trade crackdown:** New regulations will affect global trade. Low carbon technology from China will face higher tariffs and other\n\n# What to watch\n\nWith home sales being a major driver of consumer expenditure on household goods, including furniture and white goods, we expect spending on these items to be subdued in 2025. Inflation-struck consumers are likely to prioritise spending on essentials, as well as\n\nleisure and entertainment activities, over other discretionary goods such as clothing. Spending on consumer services such as travel, recreation and culture activities have fared well in the past year, and will continue to do so into 2025.\n\n2020 2021 2022 2023 2024 2025\n\n100\n\n110\n\n120\n\n130\n\n140\n\n150\n\n160\n\n170\n\n180\n\nSource: EIU.\n\nHotels & restaurants\n\nFood, beverages & tobacco Clothing & footwear\n\nHousehold goods & services\n\nLeisure & education Housing & household fuels\n\nHotels & restaurants\n\nFood, beverages & tobacco Clothing & footwear\n\nHousehold goods & services\n\nLeisure & education Housing & household fuels\n\n### Consumer spending on household goods & furniture will be sluggish\n\nGlobal consumer spending in US$, 2020=100\n\nCopyright © The Economist Intelligence Unit 2024. All rights reserved.","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"26 © The Economist Intelligence Unit Limited 2024\n\nIndustry outlook 2025 Challenges, opportunities and trends to watch in six sectors\n\n**Junk food:** Several countries will take steps to reduce junk food consumption in 2025 in an effort to address rising obesity rates. The UK plans to ban television advertisements for junk food before 9 pm, and on online ads for products that have high fat, sugar and salt content from October 1st 2025. Norway is expected to announce a similar ban, which would come into effect in 2025. Other governments will introduce taxes to discourage consumption while also increasing tax revenue. Slovakia will impose a sugar tax from January, and Vietnam is expected to pass a regulation to tax soft drinks that contain more than 5g of sugar per 100 ml.\n\n**EU travel rules:** After much delay, the European Travel Information and Authorisation System (ETIAS) will come into effect from mid-2025. After that date travellers from non- EU countries that do not currently require a visa to enter the EU will have to apply for travel authorisation before entry. The authorisation, which will cost €7 for those who are not exempt, will apply to around 1.4bn tourists a year from countries including the UK, Australia and the US.\n\n**Green trade crackdown:** New regulations will affect global trade. Low carbon technology from China will face higher tariffs and other\n\n# What to watch\n\nWith home sales being a major driver of consumer expenditure on household goods, including furniture and white goods, we expect spending on these items to be subdued in 2025. Inflation-struck consumers are likely to prioritise spending on essentials, as well as\n\nleisure and entertainment activities, over other discretionary goods such as clothing. Spending on consumer services such as travel, recreation and culture activities have fared well in the past year, and will continue to do so into 2025.\n\n2020 2021 2022 2023 2024 2025\n\n100\n\n110\n\n120\n\n130\n\n140\n\n150\n\n160\n\n170\n\n180\n\nSource: EIU.\n\nHotels & restaurants\n\nFood, beverages & tobacco Clothing & footwear\n\nHousehold goods & services\n\nLeisure & education Housing & household fuels\n\nHotels & restaurants\n\nFood, beverages & tobacco Clothing & footwear\n\nHousehold goods & services\n\nLeisure & education Housing & household fuels\n\n### Consumer spending on household goods & furniture will be sluggish\n\nGlobal consumer spending in US$, 2020=100\n\nCopyright © The Economist Intelligence Unit 2024. All rights reserved.","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b3cd580a-3656-44ed-838a-5f2996ff6fc9_p26.pdf"]},"response":{"errors":[],"extract_id":"extract_cfacc113af6b4a495fc9152c027562bd","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["26 © The Economist Intelligence Unit Limited 2024\n\nIndustry outlook 2025 Challenges, opportunities and trends to watch in six sectors\n\n**Junk food:** Several countries will take steps to reduce junk food consumption in 2025 in an effort to address rising obesity rates. The UK plans to ban television advertisements for junk food before 9 pm, and on online ads for products that have high fat, sugar and salt content from October 1st 2025. Norway is expected to announce a similar ban, which would come into effect in 2025. Other governments will introduce taxes to discourage consumption while also increasing tax revenue. Slovakia will impose a sugar tax from January, and Vietnam is expected to pass a regulation to tax soft drinks that contain more than 5g of sugar per 100 ml.\n\n**EU travel rules:** After much delay, the European Travel Information and Authorisation System (ETIAS) will come into effect from mid-2025. After that date travellers from non- EU countries that do not currently require a visa to enter the EU will have to apply for travel authorisation before entry. The authorisation, which will cost €7 for those who are not exempt, will apply to around 1.4bn tourists a year from countries including the UK, Australia and the US.\n\n**Green trade crackdown:** New regulations will affect global trade. Low carbon technology from China will face higher tariffs and other\n\n# What to watch\n\nWith home sales being a major driver of consumer expenditure on household goods, including furniture and white goods, we expect spending on these items to be subdued in 2025. Inflation-struck consumers are likely to prioritise spending on essentials, as well as\n\nleisure and entertainment activities, over other discretionary goods such as clothing. Spending on consumer services such as travel, recreation and culture activities have fared well in the past year, and will continue to do so into 2025.\n\n2020 2021 2022 2023 2024 2025\n\n100\n\n110\n\n120\n\n130\n\n140\n\n150\n\n160\n\n170\n\n180\n\nSource: EIU.\n\nHotels & restaurants\n\nFood, beverages & tobacco Clothing & footwear\n\nHousehold goods & services\n\nLeisure & education Housing & household fuels\n\nHotels & restaurants\n\nFood, beverages & tobacco Clothing & footwear\n\nHousehold goods & services\n\nLeisure & education Housing & household fuels\n\n### Consumer spending on household goods & furniture will be sluggish\n\nGlobal consumer spending in US$, 2020=100\n\nCopyright © The Economist Intelligence Unit 2024. All rights reserved."],"full_content":"26 © The Economist Intelligence Unit Limited 2024\n\nIndustry outlook 2025 Challenges, opportunities and trends to watch in six sectors\n\n**Junk food:** Several countries will take steps to reduce junk food consumption in 2025 in an effort to address rising obesity rates. The UK plans to ban television advertisements for junk food before 9 pm, and on online ads for products that have high fat, sugar and salt content from October 1st 2025. Norway is expected to announce a similar ban, which would come into effect in 2025. Other governments will introduce taxes to discourage consumption while also increasing tax revenue. Slovakia will impose a sugar tax from January, and Vietnam is expected to pass a regulation to tax soft drinks that contain more than 5g of sugar per 100 ml.\n\n**EU travel rules:** After much delay, the European Travel Information and Authorisation System (ETIAS) will come into effect from mid-2025. After that date travellers from non- EU countries that do not currently require a visa to enter the EU will have to apply for travel authorisation before entry. The authorisation, which will cost €7 for those who are not exempt, will apply to around 1.4bn tourists a year from countries including the UK, Australia and the US.\n\n**Green trade crackdown:** New regulations will affect global trade. Low carbon technology from China will face higher tariffs and other\n\n# What to watch\n\nWith home sales being a major driver of consumer expenditure on household goods, including furniture and white goods, we expect spending on these items to be subdued in 2025. Inflation-struck consumers are likely to prioritise spending on essentials, as well as\n\nleisure and entertainment activities, over other discretionary goods such as clothing. Spending on consumer services such as travel, recreation and culture activities have fared well in the past year, and will continue to do so into 2025.\n\n2020 2021 2022 2023 2024 2025\n\n100\n\n110\n\n120\n\n130\n\n140\n\n150\n\n160\n\n170\n\n180\n\nSource: EIU.\n\nHotels & restaurants\n\nFood, beverages & tobacco Clothing & footwear\n\nHousehold goods & services\n\nLeisure & education Housing & household fuels\n\nHotels & restaurants\n\nFood, beverages & tobacco Clothing & footwear\n\nHousehold goods & services\n\nLeisure & education Housing & household fuels\n\n### Consumer spending on household goods & furniture will be sluggish\n\nGlobal consumer spending in US$, 2020=100\n\nCopyright © The Economist Intelligence Unit 2024. All rights reserved.","publish_date":null,"title":"What to watch","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b3cd580a-3656-44ed-838a-5f2996ff6fc9_p26.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b3cd580a-3656-44ed-838a-5f2996ff6fc9_p26.pdf"},"request":{"example_id":"chart/b3cd580a-3656-44ed-838a-5f2996ff6fc9_p26","product_type":"parse","source_file_path":"b3cd580a-3656-44ed-838a-5f2996ff6fc9_p26.pdf"},"started_at":"2026-11-04T19:31:06.455731"} +{"completed_at":"2026-11-04T19:46:43.460540","latency_in_ms":1024,"output":{"example_id":"chart/b3cd580a-3656-44ed-838a-5f2996ff6fc9_p32","grounded_pages":[],"layout_pages":[],"markdown":"32 © The Economist Intelligence Unit Limited 2024\n\nIndustry outlook 2025 Challenges, opportunities and trends to watch in six sectors\n\n2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033\n\n0\n\n2,000\n\n4,000\n\n6,000\n\n8,000\n\n10,000\n\n12,000\n\nCoal Gas Solar Wind\n\nHydro Nuclear\n\nSource: EIU.\n\nWind and solar are taking over\n\nPower generation by source; GWh\n\nCopyright © The Economist Intelligence Unit 2024. All rights reserved.\n\n# Moderating oil prices will support demand growth\n\nan impressive figure in historical terms, albeit lower than the record high set in 2023. Most of this capacity will be installed in China, which is a major source, as well as the main destination, of investment into renewables. Investment into solar panel production will be directed to other countries as trade barriers increase.\n\nHowever, not all is bright for renewables, as there are many risks. Inflation, supply-chain bottlenecks and high interest rates are all hitting the renewable energy sector, but the biggest barrier is the lack of grid infrastructure.\n\nPrices of fossil fuels are coming down from their highs in 2022 and 2024, which will boost demand. However, after their recent weakness, we expect crude oil prices to bounce back to average more than US$77/barrel (dated Brent Blend) in 2025. Based on these forecasts, we expect global oil demand growth to accelerate in 2025, but to slow again in 2026.\n\nOil consumption has already peaked in Europe, and it will peak by the end of the decade in North America (or even earlier if Ms Harris becomes the US president). In neither of these regions has demand recovered to 2019 levels. However, oil consumption will continue to grow beyond 2033 (the end of our forecast period) in the Middle East, Africa and South America.","pages":[{"line_numbers":[],"markdown":"32 © The Economist Intelligence Unit Limited 2024\n\nIndustry outlook 2025 Challenges, opportunities and trends to watch in six sectors\n\n2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033\n\n0\n\n2,000\n\n4,000\n\n6,000\n\n8,000\n\n10,000\n\n12,000\n\nCoal Gas Solar Wind\n\nHydro Nuclear\n\nSource: EIU.\n\nWind and solar are taking over\n\nPower generation by source; GWh\n\nCopyright © The Economist Intelligence Unit 2024. All rights reserved.\n\n# Moderating oil prices will support demand growth\n\nan impressive figure in historical terms, albeit lower than the record high set in 2023. Most of this capacity will be installed in China, which is a major source, as well as the main destination, of investment into renewables. Investment into solar panel production will be directed to other countries as trade barriers increase.\n\nHowever, not all is bright for renewables, as there are many risks. Inflation, supply-chain bottlenecks and high interest rates are all hitting the renewable energy sector, but the biggest barrier is the lack of grid infrastructure.\n\nPrices of fossil fuels are coming down from their highs in 2022 and 2024, which will boost demand. However, after their recent weakness, we expect crude oil prices to bounce back to average more than US$77/barrel (dated Brent Blend) in 2025. Based on these forecasts, we expect global oil demand growth to accelerate in 2025, but to slow again in 2026.\n\nOil consumption has already peaked in Europe, and it will peak by the end of the decade in North America (or even earlier if Ms Harris becomes the US president). In neither of these regions has demand recovered to 2019 levels. However, oil consumption will continue to grow beyond 2033 (the end of our forecast period) in the Middle East, Africa and South America.","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"32 © The Economist Intelligence Unit Limited 2024\n\nIndustry outlook 2025 Challenges, opportunities and trends to watch in six sectors\n\n2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033\n\n0\n\n2,000\n\n4,000\n\n6,000\n\n8,000\n\n10,000\n\n12,000\n\nCoal Gas Solar Wind\n\nHydro Nuclear\n\nSource: EIU.\n\nWind and solar are taking over\n\nPower generation by source; GWh\n\nCopyright © The Economist Intelligence Unit 2024. All rights reserved.\n\n# Moderating oil prices will support demand growth\n\nan impressive figure in historical terms, albeit lower than the record high set in 2023. Most of this capacity will be installed in China, which is a major source, as well as the main destination, of investment into renewables. Investment into solar panel production will be directed to other countries as trade barriers increase.\n\nHowever, not all is bright for renewables, as there are many risks. Inflation, supply-chain bottlenecks and high interest rates are all hitting the renewable energy sector, but the biggest barrier is the lack of grid infrastructure.\n\nPrices of fossil fuels are coming down from their highs in 2022 and 2024, which will boost demand. However, after their recent weakness, we expect crude oil prices to bounce back to average more than US$77/barrel (dated Brent Blend) in 2025. Based on these forecasts, we expect global oil demand growth to accelerate in 2025, but to slow again in 2026.\n\nOil consumption has already peaked in Europe, and it will peak by the end of the decade in North America (or even earlier if Ms Harris becomes the US president). In neither of these regions has demand recovered to 2019 levels. However, oil consumption will continue to grow beyond 2033 (the end of our forecast period) in the Middle East, Africa and South America.","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b3cd580a-3656-44ed-838a-5f2996ff6fc9_p32.pdf"]},"response":{"errors":[],"extract_id":"extract_5b78903ae2e38085abe3a7aaef335ec0","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["32 © The Economist Intelligence Unit Limited 2024\n\nIndustry outlook 2025 Challenges, opportunities and trends to watch in six sectors\n\n2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033\n\n0\n\n2,000\n\n4,000\n\n6,000\n\n8,000\n\n10,000\n\n12,000\n\nCoal Gas Solar Wind\n\nHydro Nuclear\n\nSource: EIU.\n\nWind and solar are taking over\n\nPower generation by source; GWh\n\nCopyright © The Economist Intelligence Unit 2024. All rights reserved.\n\n# Moderating oil prices will support demand growth\n\nan impressive figure in historical terms, albeit lower than the record high set in 2023. Most of this capacity will be installed in China, which is a major source, as well as the main destination, of investment into renewables. Investment into solar panel production will be directed to other countries as trade barriers increase.\n\nHowever, not all is bright for renewables, as there are many risks. Inflation, supply-chain bottlenecks and high interest rates are all hitting the renewable energy sector, but the biggest barrier is the lack of grid infrastructure.\n\nPrices of fossil fuels are coming down from their highs in 2022 and 2024, which will boost demand. However, after their recent weakness, we expect crude oil prices to bounce back to average more than US$77/barrel (dated Brent Blend) in 2025. Based on these forecasts, we expect global oil demand growth to accelerate in 2025, but to slow again in 2026.\n\nOil consumption has already peaked in Europe, and it will peak by the end of the decade in North America (or even earlier if Ms Harris becomes the US president). In neither of these regions has demand recovered to 2019 levels. However, oil consumption will continue to grow beyond 2033 (the end of our forecast period) in the Middle East, Africa and South America."],"full_content":"32 © The Economist Intelligence Unit Limited 2024\n\nIndustry outlook 2025 Challenges, opportunities and trends to watch in six sectors\n\n2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033\n\n0\n\n2,000\n\n4,000\n\n6,000\n\n8,000\n\n10,000\n\n12,000\n\nCoal Gas Solar Wind\n\nHydro Nuclear\n\nSource: EIU.\n\nWind and solar are taking over\n\nPower generation by source; GWh\n\nCopyright © The Economist Intelligence Unit 2024. All rights reserved.\n\n# Moderating oil prices will support demand growth\n\nan impressive figure in historical terms, albeit lower than the record high set in 2023. Most of this capacity will be installed in China, which is a major source, as well as the main destination, of investment into renewables. Investment into solar panel production will be directed to other countries as trade barriers increase.\n\nHowever, not all is bright for renewables, as there are many risks. Inflation, supply-chain bottlenecks and high interest rates are all hitting the renewable energy sector, but the biggest barrier is the lack of grid infrastructure.\n\nPrices of fossil fuels are coming down from their highs in 2022 and 2024, which will boost demand. However, after their recent weakness, we expect crude oil prices to bounce back to average more than US$77/barrel (dated Brent Blend) in 2025. Based on these forecasts, we expect global oil demand growth to accelerate in 2025, but to slow again in 2026.\n\nOil consumption has already peaked in Europe, and it will peak by the end of the decade in North America (or even earlier if Ms Harris becomes the US president). In neither of these regions has demand recovered to 2019 levels. However, oil consumption will continue to grow beyond 2033 (the end of our forecast period) in the Middle East, Africa and South America.","publish_date":null,"title":"Moderating oil prices will support demand growth","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b3cd580a-3656-44ed-838a-5f2996ff6fc9_p32.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b3cd580a-3656-44ed-838a-5f2996ff6fc9_p32.pdf"},"request":{"example_id":"chart/b3cd580a-3656-44ed-838a-5f2996ff6fc9_p32","product_type":"parse","source_file_path":"b3cd580a-3656-44ed-838a-5f2996ff6fc9_p32.pdf"},"started_at":"2026-11-04T19:46:42.436493"} +{"completed_at":"2026-11-04T19:44:47.778088","latency_in_ms":1630,"output":{"example_id":"chart/b3cd580a-3656-44ed-838a-5f2996ff6fc9_p33","grounded_pages":[],"layout_pages":[],"markdown":"33 © The Economist Intelligence Unit Limited 2024\n\nIndustry outlook 2025 Challenges, opportunities and trends to watch in six sectors\n\n# Gas prices will continue to decline\n\nChange in demand, mtoe\n\n−100\n\n−50\n\n0\n\n50\n\n100\n\n150\n\n200\n\n2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033\n\nOil Gas Coal\n\nSource: EIU.\n\n### Global coal and oil demand are set to peak within the next ten years, but gas consumption will remain strong\n\nCopyright © The Economist Intelligence Unit 2024. All rights reserved.\n\nOil demand in China will peak by the end of the decade as the country increasingly adopts electric vehicles (EVs). India will then become the main driver of global demand growth, owing to its growing population, higher purchasing power and rapid economic development, which will result in greater penetration of passenger cars, freight transport and air travel. This, in addition to a low penetration of EVs (we forecast that they will make up only 7% of new-car sales by 2028), will lift oil demand.\n\nOn the supply side, we expect global oil production to rise moderately in\n\n2025\\. In June OPEC+ announced that lower 2024 production quotas agreed in mid-2023 would be extended to end-2025. Additional voluntary cuts agreed by a core group of members, led by Saudi Arabia, were recently extended to end-November, to be phased out incrementally between December 2024 and November 2025. However, we estimate that production in the US, Brazil and Guyana will continue surging, putting downward pressure on oil prices.\n\nNatural gas prices in Europe will continue to trend downwards in 2025 amid weaker demand. After recovering in 2024, we expect that\n\ngas demand in Europe will decline again in 2025, mainly owing to surging power generation from renewables. Although new import","pages":[{"line_numbers":[],"markdown":"33 © The Economist Intelligence Unit Limited 2024\n\nIndustry outlook 2025 Challenges, opportunities and trends to watch in six sectors\n\n# Gas prices will continue to decline\n\nChange in demand, mtoe\n\n−100\n\n−50\n\n0\n\n50\n\n100\n\n150\n\n200\n\n2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033\n\nOil Gas Coal\n\nSource: EIU.\n\n### Global coal and oil demand are set to peak within the next ten years, but gas consumption will remain strong\n\nCopyright © The Economist Intelligence Unit 2024. All rights reserved.\n\nOil demand in China will peak by the end of the decade as the country increasingly adopts electric vehicles (EVs). India will then become the main driver of global demand growth, owing to its growing population, higher purchasing power and rapid economic development, which will result in greater penetration of passenger cars, freight transport and air travel. This, in addition to a low penetration of EVs (we forecast that they will make up only 7% of new-car sales by 2028), will lift oil demand.\n\nOn the supply side, we expect global oil production to rise moderately in\n\n2025\\. In June OPEC+ announced that lower 2024 production quotas agreed in mid-2023 would be extended to end-2025. Additional voluntary cuts agreed by a core group of members, led by Saudi Arabia, were recently extended to end-November, to be phased out incrementally between December 2024 and November 2025. However, we estimate that production in the US, Brazil and Guyana will continue surging, putting downward pressure on oil prices.\n\nNatural gas prices in Europe will continue to trend downwards in 2025 amid weaker demand. After recovering in 2024, we expect that\n\ngas demand in Europe will decline again in 2025, mainly owing to surging power generation from renewables. Although new import","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"33 © The Economist Intelligence Unit Limited 2024\n\nIndustry outlook 2025 Challenges, opportunities and trends to watch in six sectors\n\n# Gas prices will continue to decline\n\nChange in demand, mtoe\n\n−100\n\n−50\n\n0\n\n50\n\n100\n\n150\n\n200\n\n2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033\n\nOil Gas Coal\n\nSource: EIU.\n\n### Global coal and oil demand are set to peak within the next ten years, but gas consumption will remain strong\n\nCopyright © The Economist Intelligence Unit 2024. All rights reserved.\n\nOil demand in China will peak by the end of the decade as the country increasingly adopts electric vehicles (EVs). India will then become the main driver of global demand growth, owing to its growing population, higher purchasing power and rapid economic development, which will result in greater penetration of passenger cars, freight transport and air travel. This, in addition to a low penetration of EVs (we forecast that they will make up only 7% of new-car sales by 2028), will lift oil demand.\n\nOn the supply side, we expect global oil production to rise moderately in\n\n2025\\. In June OPEC+ announced that lower 2024 production quotas agreed in mid-2023 would be extended to end-2025. Additional voluntary cuts agreed by a core group of members, led by Saudi Arabia, were recently extended to end-November, to be phased out incrementally between December 2024 and November 2025. However, we estimate that production in the US, Brazil and Guyana will continue surging, putting downward pressure on oil prices.\n\nNatural gas prices in Europe will continue to trend downwards in 2025 amid weaker demand. After recovering in 2024, we expect that\n\ngas demand in Europe will decline again in 2025, mainly owing to surging power generation from renewables. Although new import","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b3cd580a-3656-44ed-838a-5f2996ff6fc9_p33.pdf"]},"response":{"errors":[],"extract_id":"extract_254851f503f950a482944b6dbc052228","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["33 © The Economist Intelligence Unit Limited 2024\n\nIndustry outlook 2025 Challenges, opportunities and trends to watch in six sectors\n\n# Gas prices will continue to decline\n\nChange in demand, mtoe\n\n−100\n\n−50\n\n0\n\n50\n\n100\n\n150\n\n200\n\n2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033\n\nOil Gas Coal\n\nSource: EIU.\n\n### Global coal and oil demand are set to peak within the next ten years, but gas consumption will remain strong\n\nCopyright © The Economist Intelligence Unit 2024. All rights reserved.\n\nOil demand in China will peak by the end of the decade as the country increasingly adopts electric vehicles (EVs). India will then become the main driver of global demand growth, owing to its growing population, higher purchasing power and rapid economic development, which will result in greater penetration of passenger cars, freight transport and air travel. This, in addition to a low penetration of EVs (we forecast that they will make up only 7% of new-car sales by 2028), will lift oil demand.\n\nOn the supply side, we expect global oil production to rise moderately in\n\n2025\\. In June OPEC+ announced that lower 2024 production quotas agreed in mid-2023 would be extended to end-2025. Additional voluntary cuts agreed by a core group of members, led by Saudi Arabia, were recently extended to end-November, to be phased out incrementally between December 2024 and November 2025. However, we estimate that production in the US, Brazil and Guyana will continue surging, putting downward pressure on oil prices.\n\nNatural gas prices in Europe will continue to trend downwards in 2025 amid weaker demand. After recovering in 2024, we expect that\n\ngas demand in Europe will decline again in 2025, mainly owing to surging power generation from renewables. Although new import"],"full_content":"33 © The Economist Intelligence Unit Limited 2024\n\nIndustry outlook 2025 Challenges, opportunities and trends to watch in six sectors\n\n# Gas prices will continue to decline\n\nChange in demand, mtoe\n\n−100\n\n−50\n\n0\n\n50\n\n100\n\n150\n\n200\n\n2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033\n\nOil Gas Coal\n\nSource: EIU.\n\n### Global coal and oil demand are set to peak within the next ten years, but gas consumption will remain strong\n\nCopyright © The Economist Intelligence Unit 2024. All rights reserved.\n\nOil demand in China will peak by the end of the decade as the country increasingly adopts electric vehicles (EVs). India will then become the main driver of global demand growth, owing to its growing population, higher purchasing power and rapid economic development, which will result in greater penetration of passenger cars, freight transport and air travel. This, in addition to a low penetration of EVs (we forecast that they will make up only 7% of new-car sales by 2028), will lift oil demand.\n\nOn the supply side, we expect global oil production to rise moderately in\n\n2025\\. In June OPEC+ announced that lower 2024 production quotas agreed in mid-2023 would be extended to end-2025. Additional voluntary cuts agreed by a core group of members, led by Saudi Arabia, were recently extended to end-November, to be phased out incrementally between December 2024 and November 2025. However, we estimate that production in the US, Brazil and Guyana will continue surging, putting downward pressure on oil prices.\n\nNatural gas prices in Europe will continue to trend downwards in 2025 amid weaker demand. After recovering in 2024, we expect that\n\ngas demand in Europe will decline again in 2025, mainly owing to surging power generation from renewables. Although new import","publish_date":null,"title":"Gas prices will continue to decline","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b3cd580a-3656-44ed-838a-5f2996ff6fc9_p33.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b3cd580a-3656-44ed-838a-5f2996ff6fc9_p33.pdf"},"request":{"example_id":"chart/b3cd580a-3656-44ed-838a-5f2996ff6fc9_p33","product_type":"parse","source_file_path":"b3cd580a-3656-44ed-838a-5f2996ff6fc9_p33.pdf"},"started_at":"2026-11-04T19:44:46.147512"} +{"completed_at":"2026-11-04T19:44:49.219128","latency_in_ms":1217,"output":{"example_id":"chart/b3cd580a-3656-44ed-838a-5f2996ff6fc9_p49","grounded_pages":[],"layout_pages":[],"markdown":"49 © The Economist Intelligence Unit Limited 2024\n\nIndustry outlook 2025 Challenges, opportunities and trends to watch in six sectors\n\nAs a result, progress on the build-out of public healthcare systems will remain slow, as it has been since the pandemic. About a decade ago, when the UN’s sustainable development goals were adopted, 2023 was set as the target year for achieving the so-called triple billion targets, which included extending UHC to 1bn additional people. During the pandemic these targets\n\nwere deferred to 2025 but it now seems that this timeline will also be missed, jeopardising the even more ambitious targets set for 2030. We also expect minimal progress in lowering the unaffordable out-of- pocket health costs for rural and poorer populations across Africa, Asia and Latin America.\n\n# Medical tourism will drive healthcare investment\n\nMedical tourism will be a major driver of global health and pharmaceutical investment in 2025, as high prices in North America and western Europe cause people to travel abroad to access cheaper healthcare. Several countries, including Thailand, Turkey, the UAE\n\nand Mexico, will focus on becoming top destinations for medical tourists by developing specialisms and boosting recruitment. Malaysia aims to become a premium medical tourism destination by 2025, under its Healthcare Travel Industry Blueprint.\n\n% change in nominal US$ terms\n\n% change in real US$ terms Healthcare spending as a % of GDP\n\nSource: World Bank, OECD, EIU.\n\nCopyright © The Economist Intelligence Unit 2024. All rights reserved.\n\n### Global health spending will grow in real terms in 2025\n\nThe world's healthcare expenditure as a % of GDP will be largely unchanged\n\n2019 2020 2021 2022 2023 2024 2025\n\n−8\n\n−6\n\n−4\n\n−2\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\nHealthcare spending fell in Healthcare spending fell in real terms after the covid- real terms after the covid-\n\n19 pandemic 19 pandemic\n\nHealthcare spending fell in real terms after the covid- 19 pandemic\n\n2018","pages":[{"line_numbers":[],"markdown":"49 © The Economist Intelligence Unit Limited 2024\n\nIndustry outlook 2025 Challenges, opportunities and trends to watch in six sectors\n\nAs a result, progress on the build-out of public healthcare systems will remain slow, as it has been since the pandemic. About a decade ago, when the UN’s sustainable development goals were adopted, 2023 was set as the target year for achieving the so-called triple billion targets, which included extending UHC to 1bn additional people. During the pandemic these targets\n\nwere deferred to 2025 but it now seems that this timeline will also be missed, jeopardising the even more ambitious targets set for 2030. We also expect minimal progress in lowering the unaffordable out-of- pocket health costs for rural and poorer populations across Africa, Asia and Latin America.\n\n# Medical tourism will drive healthcare investment\n\nMedical tourism will be a major driver of global health and pharmaceutical investment in 2025, as high prices in North America and western Europe cause people to travel abroad to access cheaper healthcare. Several countries, including Thailand, Turkey, the UAE\n\nand Mexico, will focus on becoming top destinations for medical tourists by developing specialisms and boosting recruitment. Malaysia aims to become a premium medical tourism destination by 2025, under its Healthcare Travel Industry Blueprint.\n\n% change in nominal US$ terms\n\n% change in real US$ terms Healthcare spending as a % of GDP\n\nSource: World Bank, OECD, EIU.\n\nCopyright © The Economist Intelligence Unit 2024. All rights reserved.\n\n### Global health spending will grow in real terms in 2025\n\nThe world's healthcare expenditure as a % of GDP will be largely unchanged\n\n2019 2020 2021 2022 2023 2024 2025\n\n−8\n\n−6\n\n−4\n\n−2\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\nHealthcare spending fell in Healthcare spending fell in real terms after the covid- real terms after the covid-\n\n19 pandemic 19 pandemic\n\nHealthcare spending fell in real terms after the covid- 19 pandemic\n\n2018","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"49 © The Economist Intelligence Unit Limited 2024\n\nIndustry outlook 2025 Challenges, opportunities and trends to watch in six sectors\n\nAs a result, progress on the build-out of public healthcare systems will remain slow, as it has been since the pandemic. About a decade ago, when the UN’s sustainable development goals were adopted, 2023 was set as the target year for achieving the so-called triple billion targets, which included extending UHC to 1bn additional people. During the pandemic these targets\n\nwere deferred to 2025 but it now seems that this timeline will also be missed, jeopardising the even more ambitious targets set for 2030. We also expect minimal progress in lowering the unaffordable out-of- pocket health costs for rural and poorer populations across Africa, Asia and Latin America.\n\n# Medical tourism will drive healthcare investment\n\nMedical tourism will be a major driver of global health and pharmaceutical investment in 2025, as high prices in North America and western Europe cause people to travel abroad to access cheaper healthcare. Several countries, including Thailand, Turkey, the UAE\n\nand Mexico, will focus on becoming top destinations for medical tourists by developing specialisms and boosting recruitment. Malaysia aims to become a premium medical tourism destination by 2025, under its Healthcare Travel Industry Blueprint.\n\n% change in nominal US$ terms\n\n% change in real US$ terms Healthcare spending as a % of GDP\n\nSource: World Bank, OECD, EIU.\n\nCopyright © The Economist Intelligence Unit 2024. All rights reserved.\n\n### Global health spending will grow in real terms in 2025\n\nThe world's healthcare expenditure as a % of GDP will be largely unchanged\n\n2019 2020 2021 2022 2023 2024 2025\n\n−8\n\n−6\n\n−4\n\n−2\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\nHealthcare spending fell in Healthcare spending fell in real terms after the covid- real terms after the covid-\n\n19 pandemic 19 pandemic\n\nHealthcare spending fell in real terms after the covid- 19 pandemic\n\n2018","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b3cd580a-3656-44ed-838a-5f2996ff6fc9_p49.pdf"]},"response":{"errors":[],"extract_id":"extract_4d457d1f1dee47c79f964d610eea7b52","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["49 © The Economist Intelligence Unit Limited 2024\n\nIndustry outlook 2025 Challenges, opportunities and trends to watch in six sectors\n\nAs a result, progress on the build-out of public healthcare systems will remain slow, as it has been since the pandemic. About a decade ago, when the UN’s sustainable development goals were adopted, 2023 was set as the target year for achieving the so-called triple billion targets, which included extending UHC to 1bn additional people. During the pandemic these targets\n\nwere deferred to 2025 but it now seems that this timeline will also be missed, jeopardising the even more ambitious targets set for 2030. We also expect minimal progress in lowering the unaffordable out-of- pocket health costs for rural and poorer populations across Africa, Asia and Latin America.\n\n# Medical tourism will drive healthcare investment\n\nMedical tourism will be a major driver of global health and pharmaceutical investment in 2025, as high prices in North America and western Europe cause people to travel abroad to access cheaper healthcare. Several countries, including Thailand, Turkey, the UAE\n\nand Mexico, will focus on becoming top destinations for medical tourists by developing specialisms and boosting recruitment. Malaysia aims to become a premium medical tourism destination by 2025, under its Healthcare Travel Industry Blueprint.\n\n% change in nominal US$ terms\n\n% change in real US$ terms Healthcare spending as a % of GDP\n\nSource: World Bank, OECD, EIU.\n\nCopyright © The Economist Intelligence Unit 2024. All rights reserved.\n\n### Global health spending will grow in real terms in 2025\n\nThe world's healthcare expenditure as a % of GDP will be largely unchanged\n\n2019 2020 2021 2022 2023 2024 2025\n\n−8\n\n−6\n\n−4\n\n−2\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\nHealthcare spending fell in Healthcare spending fell in real terms after the covid- real terms after the covid-\n\n19 pandemic 19 pandemic\n\nHealthcare spending fell in real terms after the covid- 19 pandemic\n\n2018"],"full_content":"49 © The Economist Intelligence Unit Limited 2024\n\nIndustry outlook 2025 Challenges, opportunities and trends to watch in six sectors\n\nAs a result, progress on the build-out of public healthcare systems will remain slow, as it has been since the pandemic. About a decade ago, when the UN’s sustainable development goals were adopted, 2023 was set as the target year for achieving the so-called triple billion targets, which included extending UHC to 1bn additional people. During the pandemic these targets\n\nwere deferred to 2025 but it now seems that this timeline will also be missed, jeopardising the even more ambitious targets set for 2030. We also expect minimal progress in lowering the unaffordable out-of- pocket health costs for rural and poorer populations across Africa, Asia and Latin America.\n\n# Medical tourism will drive healthcare investment\n\nMedical tourism will be a major driver of global health and pharmaceutical investment in 2025, as high prices in North America and western Europe cause people to travel abroad to access cheaper healthcare. Several countries, including Thailand, Turkey, the UAE\n\nand Mexico, will focus on becoming top destinations for medical tourists by developing specialisms and boosting recruitment. Malaysia aims to become a premium medical tourism destination by 2025, under its Healthcare Travel Industry Blueprint.\n\n% change in nominal US$ terms\n\n% change in real US$ terms Healthcare spending as a % of GDP\n\nSource: World Bank, OECD, EIU.\n\nCopyright © The Economist Intelligence Unit 2024. All rights reserved.\n\n### Global health spending will grow in real terms in 2025\n\nThe world's healthcare expenditure as a % of GDP will be largely unchanged\n\n2019 2020 2021 2022 2023 2024 2025\n\n−8\n\n−6\n\n−4\n\n−2\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\nHealthcare spending fell in Healthcare spending fell in real terms after the covid- real terms after the covid-\n\n19 pandemic 19 pandemic\n\nHealthcare spending fell in real terms after the covid- 19 pandemic\n\n2018","publish_date":null,"title":"Medical tourism will drive","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b3cd580a-3656-44ed-838a-5f2996ff6fc9_p49.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b3cd580a-3656-44ed-838a-5f2996ff6fc9_p49.pdf"},"request":{"example_id":"chart/b3cd580a-3656-44ed-838a-5f2996ff6fc9_p49","product_type":"parse","source_file_path":"b3cd580a-3656-44ed-838a-5f2996ff6fc9_p49.pdf"},"started_at":"2026-11-04T19:44:48.002081"} +{"completed_at":"2026-11-04T19:44:51.144357","latency_in_ms":2137,"output":{"example_id":"chart/b3cd580a-3656-44ed-838a-5f2996ff6fc9_p6","grounded_pages":[],"layout_pages":[],"markdown":"6 © The Economist Intelligence Unit Limited 2024\n\nIndustry outlook 2025 Challenges, opportunities and trends to watch in six sectors\n\nGeopolitical risks and trade barriers will continue to weigh on the outlook. Our baseline forecast assumes that another large-scale war will not break out in Asia or the Middle East. However, the continued threat of geopolitical conflict, in addition to the continuing war in Ukraine, will lead to economic reconfiguration and policy divergence. The EU and the US are already raising barriers against Chinese exports in areas from automotive and technology to healthcare. Chinese retaliation is likely to intensify in 2025 as rival\n\nblocs emerge across the world. These trends will reshape supply chains over 2025, and could upend our forecast of falling commodity prices and inflation.\n\nClimate change and the efforts to mitigate it will play into these geopolitical rifts. At COP30 in November 2025 governments are due to update their national climate pledges (NDCs), but much will depend on US leadership. The debate over environmental, social and governance (ESG) reporting will\n\n# Geopolitical shocks and climate change will remain the biggest risks\n\n% growth year on year\n\n2024\n\n2025\n\nSource: EIU forecasts.\n\n**Real GDP growth will slow in 2025 in most major markets**\n\nCopyright © The Economist Intelligence Unit 2024. All rights reserved.\n\nIndia\n\nChina\n\nBrazil\n\nUK\n\nUS\n\nRussia\n\nJapan\n\nGermany\n\nWorld\n\n0 1 2 3 4 5 6\n\n6\\.9 6.6\n\n4\\.7 4.8\n\n3 2\n\n1\\.1 1.4\n\n2\\.7 1.7\n\n3\\.6 1.2\n\n0\\.2\n\n1\\.1\n\n0\\.1\n\n0\\.9\n\n2\\.6 2.6\n\n7","pages":[{"line_numbers":[],"markdown":"6 © The Economist Intelligence Unit Limited 2024\n\nIndustry outlook 2025 Challenges, opportunities and trends to watch in six sectors\n\nGeopolitical risks and trade barriers will continue to weigh on the outlook. Our baseline forecast assumes that another large-scale war will not break out in Asia or the Middle East. However, the continued threat of geopolitical conflict, in addition to the continuing war in Ukraine, will lead to economic reconfiguration and policy divergence. The EU and the US are already raising barriers against Chinese exports in areas from automotive and technology to healthcare. Chinese retaliation is likely to intensify in 2025 as rival\n\nblocs emerge across the world. These trends will reshape supply chains over 2025, and could upend our forecast of falling commodity prices and inflation.\n\nClimate change and the efforts to mitigate it will play into these geopolitical rifts. At COP30 in November 2025 governments are due to update their national climate pledges (NDCs), but much will depend on US leadership. The debate over environmental, social and governance (ESG) reporting will\n\n# Geopolitical shocks and climate change will remain the biggest risks\n\n% growth year on year\n\n2024\n\n2025\n\nSource: EIU forecasts.\n\n**Real GDP growth will slow in 2025 in most major markets**\n\nCopyright © The Economist Intelligence Unit 2024. All rights reserved.\n\nIndia\n\nChina\n\nBrazil\n\nUK\n\nUS\n\nRussia\n\nJapan\n\nGermany\n\nWorld\n\n0 1 2 3 4 5 6\n\n6\\.9 6.6\n\n4\\.7 4.8\n\n3 2\n\n1\\.1 1.4\n\n2\\.7 1.7\n\n3\\.6 1.2\n\n0\\.2\n\n1\\.1\n\n0\\.1\n\n0\\.9\n\n2\\.6 2.6\n\n7","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"6 © The Economist Intelligence Unit Limited 2024\n\nIndustry outlook 2025 Challenges, opportunities and trends to watch in six sectors\n\nGeopolitical risks and trade barriers will continue to weigh on the outlook. Our baseline forecast assumes that another large-scale war will not break out in Asia or the Middle East. However, the continued threat of geopolitical conflict, in addition to the continuing war in Ukraine, will lead to economic reconfiguration and policy divergence. The EU and the US are already raising barriers against Chinese exports in areas from automotive and technology to healthcare. Chinese retaliation is likely to intensify in 2025 as rival\n\nblocs emerge across the world. These trends will reshape supply chains over 2025, and could upend our forecast of falling commodity prices and inflation.\n\nClimate change and the efforts to mitigate it will play into these geopolitical rifts. At COP30 in November 2025 governments are due to update their national climate pledges (NDCs), but much will depend on US leadership. The debate over environmental, social and governance (ESG) reporting will\n\n# Geopolitical shocks and climate change will remain the biggest risks\n\n% growth year on year\n\n2024\n\n2025\n\nSource: EIU forecasts.\n\n**Real GDP growth will slow in 2025 in most major markets**\n\nCopyright © The Economist Intelligence Unit 2024. All rights reserved.\n\nIndia\n\nChina\n\nBrazil\n\nUK\n\nUS\n\nRussia\n\nJapan\n\nGermany\n\nWorld\n\n0 1 2 3 4 5 6\n\n6\\.9 6.6\n\n4\\.7 4.8\n\n3 2\n\n1\\.1 1.4\n\n2\\.7 1.7\n\n3\\.6 1.2\n\n0\\.2\n\n1\\.1\n\n0\\.1\n\n0\\.9\n\n2\\.6 2.6\n\n7","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b3cd580a-3656-44ed-838a-5f2996ff6fc9_p6.pdf"]},"response":{"errors":[],"extract_id":"extract_ee66a2ea4befe253f2d660b983512d74","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["6 © The Economist Intelligence Unit Limited 2024\n\nIndustry outlook 2025 Challenges, opportunities and trends to watch in six sectors\n\nGeopolitical risks and trade barriers will continue to weigh on the outlook. Our baseline forecast assumes that another large-scale war will not break out in Asia or the Middle East. However, the continued threat of geopolitical conflict, in addition to the continuing war in Ukraine, will lead to economic reconfiguration and policy divergence. The EU and the US are already raising barriers against Chinese exports in areas from automotive and technology to healthcare. Chinese retaliation is likely to intensify in 2025 as rival\n\nblocs emerge across the world. These trends will reshape supply chains over 2025, and could upend our forecast of falling commodity prices and inflation.\n\nClimate change and the efforts to mitigate it will play into these geopolitical rifts. At COP30 in November 2025 governments are due to update their national climate pledges (NDCs), but much will depend on US leadership. The debate over environmental, social and governance (ESG) reporting will\n\n# Geopolitical shocks and climate change will remain the biggest risks\n\n% growth year on year\n\n2024\n\n2025\n\nSource: EIU forecasts.\n\n**Real GDP growth will slow in 2025 in most major markets**\n\nCopyright © The Economist Intelligence Unit 2024. All rights reserved.\n\nIndia\n\nChina\n\nBrazil\n\nUK\n\nUS\n\nRussia\n\nJapan\n\nGermany\n\nWorld\n\n0 1 2 3 4 5 6\n\n6\\.9 6.6\n\n4\\.7 4.8\n\n3 2\n\n1\\.1 1.4\n\n2\\.7 1.7\n\n3\\.6 1.2\n\n0\\.2\n\n1\\.1\n\n0\\.1\n\n0\\.9\n\n2\\.6 2.6\n\n7"],"full_content":"6 © The Economist Intelligence Unit Limited 2024\n\nIndustry outlook 2025 Challenges, opportunities and trends to watch in six sectors\n\nGeopolitical risks and trade barriers will continue to weigh on the outlook. Our baseline forecast assumes that another large-scale war will not break out in Asia or the Middle East. However, the continued threat of geopolitical conflict, in addition to the continuing war in Ukraine, will lead to economic reconfiguration and policy divergence. The EU and the US are already raising barriers against Chinese exports in areas from automotive and technology to healthcare. Chinese retaliation is likely to intensify in 2025 as rival\n\nblocs emerge across the world. These trends will reshape supply chains over 2025, and could upend our forecast of falling commodity prices and inflation.\n\nClimate change and the efforts to mitigate it will play into these geopolitical rifts. At COP30 in November 2025 governments are due to update their national climate pledges (NDCs), but much will depend on US leadership. The debate over environmental, social and governance (ESG) reporting will\n\n# Geopolitical shocks and climate change will remain the biggest risks\n\n% growth year on year\n\n2024\n\n2025\n\nSource: EIU forecasts.\n\n**Real GDP growth will slow in 2025 in most major markets**\n\nCopyright © The Economist Intelligence Unit 2024. All rights reserved.\n\nIndia\n\nChina\n\nBrazil\n\nUK\n\nUS\n\nRussia\n\nJapan\n\nGermany\n\nWorld\n\n0 1 2 3 4 5 6\n\n6\\.9 6.6\n\n4\\.7 4.8\n\n3 2\n\n1\\.1 1.4\n\n2\\.7 1.7\n\n3\\.6 1.2\n\n0\\.2\n\n1\\.1\n\n0\\.1\n\n0\\.9\n\n2\\.6 2.6\n\n7","publish_date":null,"title":"Geopolitical shocks and climate change will remain the biggest risks","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b3cd580a-3656-44ed-838a-5f2996ff6fc9_p6.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/b3cd580a-3656-44ed-838a-5f2996ff6fc9_p6.pdf"},"request":{"example_id":"chart/b3cd580a-3656-44ed-838a-5f2996ff6fc9_p6","product_type":"parse","source_file_path":"b3cd580a-3656-44ed-838a-5f2996ff6fc9_p6.pdf"},"started_at":"2026-11-04T19:44:49.006958"} +{"completed_at":"2026-11-04T19:44:52.853227","latency_in_ms":2490,"output":{"example_id":"chart/cimquarterlychartbook2025q4-updated_p11","grounded_pages":[],"layout_pages":[],"markdown":"_Source: Cetera Investment Management, National Bureau of Economic Research. Data as of 9/30/2025._ **11**\n\n# U.S. Economic Expansions and Recessions\n\n\\*Current Expansion\n\n37\n\n45\n\n39\n\n24\n\n106\n\n36\n\n58\n\n12\n\n92\n\n120\n\n73\n\n128\n\n65\n\n11 10 8 10 11\n\n16\n\n6\n\n16\n\n8 8\n\n18\n\n2\n\n0\n\n20\n\n40\n\n60\n\n80\n\n100\n\n120\n\n\\# of Months\n\nYear Expansion Started\n\nExpansion\n\nRecession","pages":[{"line_numbers":[],"markdown":"_Source: Cetera Investment Management, National Bureau of Economic Research. Data as of 9/30/2025._ **11**\n\n# U.S. Economic Expansions and Recessions\n\n\\*Current Expansion\n\n37\n\n45\n\n39\n\n24\n\n106\n\n36\n\n58\n\n12\n\n92\n\n120\n\n73\n\n128\n\n65\n\n11 10 8 10 11\n\n16\n\n6\n\n16\n\n8 8\n\n18\n\n2\n\n0\n\n20\n\n40\n\n60\n\n80\n\n100\n\n120\n\n\\# of Months\n\nYear Expansion Started\n\nExpansion\n\nRecession","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"_Source: Cetera Investment Management, National Bureau of Economic Research. Data as of 9/30/2025._ **11**\n\n# U.S. Economic Expansions and Recessions\n\n\\*Current Expansion\n\n37\n\n45\n\n39\n\n24\n\n106\n\n36\n\n58\n\n12\n\n92\n\n120\n\n73\n\n128\n\n65\n\n11 10 8 10 11\n\n16\n\n6\n\n16\n\n8 8\n\n18\n\n2\n\n0\n\n20\n\n40\n\n60\n\n80\n\n100\n\n120\n\n\\# of Months\n\nYear Expansion Started\n\nExpansion\n\nRecession","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/cimquarterlychartbook2025q4-updated_p11.pdf"]},"response":{"errors":[],"extract_id":"extract_d89120a50eaf35eb7cae39d5bdd1afcc","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["_Source: Cetera Investment Management, National Bureau of Economic Research. Data as of 9/30/2025._ **11**\n\n# U.S. Economic Expansions and Recessions\n\n\\*Current Expansion\n\n37\n\n45\n\n39\n\n24\n\n106\n\n36\n\n58\n\n12\n\n92\n\n120\n\n73\n\n128\n\n65\n\n11 10 8 10 11\n\n16\n\n6\n\n16\n\n8 8\n\n18\n\n2\n\n0\n\n20\n\n40\n\n60\n\n80\n\n100\n\n120\n\n\\# of Months\n\nYear Expansion Started\n\nExpansion\n\nRecession"],"full_content":"_Source: Cetera Investment Management, National Bureau of Economic Research. Data as of 9/30/2025._ **11**\n\n# U.S. Economic Expansions and Recessions\n\n\\*Current Expansion\n\n37\n\n45\n\n39\n\n24\n\n106\n\n36\n\n58\n\n12\n\n92\n\n120\n\n73\n\n128\n\n65\n\n11 10 8 10 11\n\n16\n\n6\n\n16\n\n8 8\n\n18\n\n2\n\n0\n\n20\n\n40\n\n60\n\n80\n\n100\n\n120\n\n\\# of Months\n\nYear Expansion Started\n\nExpansion\n\nRecession","publish_date":null,"title":"U.S. Economic Expansions and Recessions","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/cimquarterlychartbook2025q4-updated_p11.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/cimquarterlychartbook2025q4-updated_p11.pdf"},"request":{"example_id":"chart/cimquarterlychartbook2025q4-updated_p11","product_type":"parse","source_file_path":"cimquarterlychartbook2025q4-updated_p11.pdf"},"started_at":"2026-11-04T19:44:50.362606"} +{"completed_at":"2026-11-04T19:44:52.996567","latency_in_ms":1239,"output":{"example_id":"chart/cimquarterlychartbook2025q4-updated_p15","grounded_pages":[],"layout_pages":[],"markdown":"_Source: Cetera Investment Management, FactSet, YCharts. The annual total return includes dividends. The annual max decline is the_\n\n_largest price decline in each calendar year. Return and max drawdown data based on closing prices. Data as of 12/31/2024._ **15**\n\n# S&P 500 Annual Total Return and Max Drawdown\n\n**32\\.5%**\n\n**\\-4.9%**\n\n**21\\.5%** **22\\.6%**\n\n**6\\.3%**\n\n**31\\.7%**\n\n**18\\.7%**\n\n**5\\.3%**\n\n**16\\.6%**\n\n**31\\.7%**\n\n**\\-3.1%**\n\n**30\\.5%**\n\n**7\\.6%**\n\n**10\\.1%**\n\n**1\\.3%**\n\n**37\\.6%**\n\n**23\\.0%**\n\n**33\\.4%**\n\n**28\\.6%**\n\n**21\\.0%**\n\n**\\-9.1%**\n\n**\\-11.9%**\n\n**\\-22.1%**\n\n**28\\.7%**\n\n**10\\.9%**\n\n**4\\.9%**\n\n**15\\.8%**\n\n**5\\.5%**\n\n**\\-37.0%**\n\n**26\\.5%**\n\n**15\\.1%**\n\n**2\\.1%**\n\n**16\\.0%**\n\n**32\\.4%**\n\n**13\\.7%**\n\n**1\\.4%**\n\n**12\\.0%**\n\n**21\\.8%**\n\n**\\-4.4%**\n\n**31\\.5%**\n\n**18\\.4%**\n\n**28\\.7%**\n\n**\\-18.1%**\n\n**26\\.3%25.0%**\n\n\\-17.1%\n\n\\-18.4%\n\n\\-16.6%\n\n\\-6.9%\n\n\\-12.7%\n\n\\-7.7%\n\n\\-9.4%\n\n\\-33.5%\n\n\\-7.6% -7.6%\n\n\\-19.9%\n\n\\-5.7%\n\n\\-6.2%\n\n\\-5.0%\n\n\\-8.9%\n\n\\-2.5%\n\n\\-7.6%\n\n\\-10.8%\n\n\\-19.3%\n\n\\-12.1%\n\n\\-17.2%\n\n\\-29.7%\n\n\\-33.8%\n\n\\-14.1%\n\n\\-8.2% \\-7.2%-7.7%\n\n\\-10.1%\n\n\\-48.8%\n\n\\-27.6%\n\n\\-16.0%\n\n\\-19.4%\n\n\\-9.9%\n\n\\-5.8%\n\n\\-7.4%\n\n\\-12.4%\n\n\\-10.5%\n\n\\-2.8%\n\n\\-19.8%\n\n\\-6.8%\n\n\\-33.9%\n\n\\-5.2%\n\n\\-25.4%\n\n\\-10.3%\n\n\\-8.5%\n\n\\-50.0%\n\n\\-40.0%\n\n\\-30.0%\n\n\\-20.0%\n\n\\-10.0%\n\n0\\.0%\n\n10\\.0%\n\n20\\.0%\n\n30\\.0%\n\n40\\.0%\n\n1980 1985 1990 1995 2000 2005 2010 2015 2020\n\nAnnual Total Return Annual Max Drawdown\n\n**1980 - 2024**\n\nAverage Annual Total Return: 13.5% Average Annual Max Drawdown: -14.1%","pages":[{"line_numbers":[],"markdown":"_Source: Cetera Investment Management, FactSet, YCharts. The annual total return includes dividends. The annual max decline is the_\n\n_largest price decline in each calendar year. Return and max drawdown data based on closing prices. Data as of 12/31/2024._ **15**\n\n# S&P 500 Annual Total Return and Max Drawdown\n\n**32\\.5%**\n\n**\\-4.9%**\n\n**21\\.5%** **22\\.6%**\n\n**6\\.3%**\n\n**31\\.7%**\n\n**18\\.7%**\n\n**5\\.3%**\n\n**16\\.6%**\n\n**31\\.7%**\n\n**\\-3.1%**\n\n**30\\.5%**\n\n**7\\.6%**\n\n**10\\.1%**\n\n**1\\.3%**\n\n**37\\.6%**\n\n**23\\.0%**\n\n**33\\.4%**\n\n**28\\.6%**\n\n**21\\.0%**\n\n**\\-9.1%**\n\n**\\-11.9%**\n\n**\\-22.1%**\n\n**28\\.7%**\n\n**10\\.9%**\n\n**4\\.9%**\n\n**15\\.8%**\n\n**5\\.5%**\n\n**\\-37.0%**\n\n**26\\.5%**\n\n**15\\.1%**\n\n**2\\.1%**\n\n**16\\.0%**\n\n**32\\.4%**\n\n**13\\.7%**\n\n**1\\.4%**\n\n**12\\.0%**\n\n**21\\.8%**\n\n**\\-4.4%**\n\n**31\\.5%**\n\n**18\\.4%**\n\n**28\\.7%**\n\n**\\-18.1%**\n\n**26\\.3%25.0%**\n\n\\-17.1%\n\n\\-18.4%\n\n\\-16.6%\n\n\\-6.9%\n\n\\-12.7%\n\n\\-7.7%\n\n\\-9.4%\n\n\\-33.5%\n\n\\-7.6% -7.6%\n\n\\-19.9%\n\n\\-5.7%\n\n\\-6.2%\n\n\\-5.0%\n\n\\-8.9%\n\n\\-2.5%\n\n\\-7.6%\n\n\\-10.8%\n\n\\-19.3%\n\n\\-12.1%\n\n\\-17.2%\n\n\\-29.7%\n\n\\-33.8%\n\n\\-14.1%\n\n\\-8.2% \\-7.2%-7.7%\n\n\\-10.1%\n\n\\-48.8%\n\n\\-27.6%\n\n\\-16.0%\n\n\\-19.4%\n\n\\-9.9%\n\n\\-5.8%\n\n\\-7.4%\n\n\\-12.4%\n\n\\-10.5%\n\n\\-2.8%\n\n\\-19.8%\n\n\\-6.8%\n\n\\-33.9%\n\n\\-5.2%\n\n\\-25.4%\n\n\\-10.3%\n\n\\-8.5%\n\n\\-50.0%\n\n\\-40.0%\n\n\\-30.0%\n\n\\-20.0%\n\n\\-10.0%\n\n0\\.0%\n\n10\\.0%\n\n20\\.0%\n\n30\\.0%\n\n40\\.0%\n\n1980 1985 1990 1995 2000 2005 2010 2015 2020\n\nAnnual Total Return Annual Max Drawdown\n\n**1980 - 2024**\n\nAverage Annual Total Return: 13.5% Average Annual Max Drawdown: -14.1%","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"_Source: Cetera Investment Management, FactSet, YCharts. The annual total return includes dividends. The annual max decline is the_\n\n_largest price decline in each calendar year. Return and max drawdown data based on closing prices. Data as of 12/31/2024._ **15**\n\n# S&P 500 Annual Total Return and Max Drawdown\n\n**32\\.5%**\n\n**\\-4.9%**\n\n**21\\.5%** **22\\.6%**\n\n**6\\.3%**\n\n**31\\.7%**\n\n**18\\.7%**\n\n**5\\.3%**\n\n**16\\.6%**\n\n**31\\.7%**\n\n**\\-3.1%**\n\n**30\\.5%**\n\n**7\\.6%**\n\n**10\\.1%**\n\n**1\\.3%**\n\n**37\\.6%**\n\n**23\\.0%**\n\n**33\\.4%**\n\n**28\\.6%**\n\n**21\\.0%**\n\n**\\-9.1%**\n\n**\\-11.9%**\n\n**\\-22.1%**\n\n**28\\.7%**\n\n**10\\.9%**\n\n**4\\.9%**\n\n**15\\.8%**\n\n**5\\.5%**\n\n**\\-37.0%**\n\n**26\\.5%**\n\n**15\\.1%**\n\n**2\\.1%**\n\n**16\\.0%**\n\n**32\\.4%**\n\n**13\\.7%**\n\n**1\\.4%**\n\n**12\\.0%**\n\n**21\\.8%**\n\n**\\-4.4%**\n\n**31\\.5%**\n\n**18\\.4%**\n\n**28\\.7%**\n\n**\\-18.1%**\n\n**26\\.3%25.0%**\n\n\\-17.1%\n\n\\-18.4%\n\n\\-16.6%\n\n\\-6.9%\n\n\\-12.7%\n\n\\-7.7%\n\n\\-9.4%\n\n\\-33.5%\n\n\\-7.6% -7.6%\n\n\\-19.9%\n\n\\-5.7%\n\n\\-6.2%\n\n\\-5.0%\n\n\\-8.9%\n\n\\-2.5%\n\n\\-7.6%\n\n\\-10.8%\n\n\\-19.3%\n\n\\-12.1%\n\n\\-17.2%\n\n\\-29.7%\n\n\\-33.8%\n\n\\-14.1%\n\n\\-8.2% \\-7.2%-7.7%\n\n\\-10.1%\n\n\\-48.8%\n\n\\-27.6%\n\n\\-16.0%\n\n\\-19.4%\n\n\\-9.9%\n\n\\-5.8%\n\n\\-7.4%\n\n\\-12.4%\n\n\\-10.5%\n\n\\-2.8%\n\n\\-19.8%\n\n\\-6.8%\n\n\\-33.9%\n\n\\-5.2%\n\n\\-25.4%\n\n\\-10.3%\n\n\\-8.5%\n\n\\-50.0%\n\n\\-40.0%\n\n\\-30.0%\n\n\\-20.0%\n\n\\-10.0%\n\n0\\.0%\n\n10\\.0%\n\n20\\.0%\n\n30\\.0%\n\n40\\.0%\n\n1980 1985 1990 1995 2000 2005 2010 2015 2020\n\nAnnual Total Return Annual Max Drawdown\n\n**1980 - 2024**\n\nAverage Annual Total Return: 13.5% Average Annual Max Drawdown: -14.1%","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/cimquarterlychartbook2025q4-updated_p15.pdf"]},"response":{"errors":[],"extract_id":"extract_fd828460cce17692bce9a459f55440d2","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["_Source: Cetera Investment Management, FactSet, YCharts. The annual total return includes dividends. The annual max decline is the_\n\n_largest price decline in each calendar year. Return and max drawdown data based on closing prices. Data as of 12/31/2024._ **15**\n\n# S&P 500 Annual Total Return and Max Drawdown\n\n**32\\.5%**\n\n**\\-4.9%**\n\n**21\\.5%** **22\\.6%**\n\n**6\\.3%**\n\n**31\\.7%**\n\n**18\\.7%**\n\n**5\\.3%**\n\n**16\\.6%**\n\n**31\\.7%**\n\n**\\-3.1%**\n\n**30\\.5%**\n\n**7\\.6%**\n\n**10\\.1%**\n\n**1\\.3%**\n\n**37\\.6%**\n\n**23\\.0%**\n\n**33\\.4%**\n\n**28\\.6%**\n\n**21\\.0%**\n\n**\\-9.1%**\n\n**\\-11.9%**\n\n**\\-22.1%**\n\n**28\\.7%**\n\n**10\\.9%**\n\n**4\\.9%**\n\n**15\\.8%**\n\n**5\\.5%**\n\n**\\-37.0%**\n\n**26\\.5%**\n\n**15\\.1%**\n\n**2\\.1%**\n\n**16\\.0%**\n\n**32\\.4%**\n\n**13\\.7%**\n\n**1\\.4%**\n\n**12\\.0%**\n\n**21\\.8%**\n\n**\\-4.4%**\n\n**31\\.5%**\n\n**18\\.4%**\n\n**28\\.7%**\n\n**\\-18.1%**\n\n**26\\.3%25.0%**\n\n\\-17.1%\n\n\\-18.4%\n\n\\-16.6%\n\n\\-6.9%\n\n\\-12.7%\n\n\\-7.7%\n\n\\-9.4%\n\n\\-33.5%\n\n\\-7.6% -7.6%\n\n\\-19.9%\n\n\\-5.7%\n\n\\-6.2%\n\n\\-5.0%\n\n\\-8.9%\n\n\\-2.5%\n\n\\-7.6%\n\n\\-10.8%\n\n\\-19.3%\n\n\\-12.1%\n\n\\-17.2%\n\n\\-29.7%\n\n\\-33.8%\n\n\\-14.1%\n\n\\-8.2% \\-7.2%-7.7%\n\n\\-10.1%\n\n\\-48.8%\n\n\\-27.6%\n\n\\-16.0%\n\n\\-19.4%\n\n\\-9.9%\n\n\\-5.8%\n\n\\-7.4%\n\n\\-12.4%\n\n\\-10.5%\n\n\\-2.8%\n\n\\-19.8%\n\n\\-6.8%\n\n\\-33.9%\n\n\\-5.2%\n\n\\-25.4%\n\n\\-10.3%\n\n\\-8.5%\n\n\\-50.0%\n\n\\-40.0%\n\n\\-30.0%\n\n\\-20.0%\n\n\\-10.0%\n\n0\\.0%\n\n10\\.0%\n\n20\\.0%\n\n30\\.0%\n\n40\\.0%\n\n1980 1985 1990 1995 2000 2005 2010 2015 2020\n\nAnnual Total Return Annual Max Drawdown\n\n**1980 - 2024**\n\nAverage Annual Total Return: 13.5% Average Annual Max Drawdown: -14.1%"],"full_content":"_Source: Cetera Investment Management, FactSet, YCharts. The annual total return includes dividends. The annual max decline is the_\n\n_largest price decline in each calendar year. Return and max drawdown data based on closing prices. Data as of 12/31/2024._ **15**\n\n# S&P 500 Annual Total Return and Max Drawdown\n\n**32\\.5%**\n\n**\\-4.9%**\n\n**21\\.5%** **22\\.6%**\n\n**6\\.3%**\n\n**31\\.7%**\n\n**18\\.7%**\n\n**5\\.3%**\n\n**16\\.6%**\n\n**31\\.7%**\n\n**\\-3.1%**\n\n**30\\.5%**\n\n**7\\.6%**\n\n**10\\.1%**\n\n**1\\.3%**\n\n**37\\.6%**\n\n**23\\.0%**\n\n**33\\.4%**\n\n**28\\.6%**\n\n**21\\.0%**\n\n**\\-9.1%**\n\n**\\-11.9%**\n\n**\\-22.1%**\n\n**28\\.7%**\n\n**10\\.9%**\n\n**4\\.9%**\n\n**15\\.8%**\n\n**5\\.5%**\n\n**\\-37.0%**\n\n**26\\.5%**\n\n**15\\.1%**\n\n**2\\.1%**\n\n**16\\.0%**\n\n**32\\.4%**\n\n**13\\.7%**\n\n**1\\.4%**\n\n**12\\.0%**\n\n**21\\.8%**\n\n**\\-4.4%**\n\n**31\\.5%**\n\n**18\\.4%**\n\n**28\\.7%**\n\n**\\-18.1%**\n\n**26\\.3%25.0%**\n\n\\-17.1%\n\n\\-18.4%\n\n\\-16.6%\n\n\\-6.9%\n\n\\-12.7%\n\n\\-7.7%\n\n\\-9.4%\n\n\\-33.5%\n\n\\-7.6% -7.6%\n\n\\-19.9%\n\n\\-5.7%\n\n\\-6.2%\n\n\\-5.0%\n\n\\-8.9%\n\n\\-2.5%\n\n\\-7.6%\n\n\\-10.8%\n\n\\-19.3%\n\n\\-12.1%\n\n\\-17.2%\n\n\\-29.7%\n\n\\-33.8%\n\n\\-14.1%\n\n\\-8.2% \\-7.2%-7.7%\n\n\\-10.1%\n\n\\-48.8%\n\n\\-27.6%\n\n\\-16.0%\n\n\\-19.4%\n\n\\-9.9%\n\n\\-5.8%\n\n\\-7.4%\n\n\\-12.4%\n\n\\-10.5%\n\n\\-2.8%\n\n\\-19.8%\n\n\\-6.8%\n\n\\-33.9%\n\n\\-5.2%\n\n\\-25.4%\n\n\\-10.3%\n\n\\-8.5%\n\n\\-50.0%\n\n\\-40.0%\n\n\\-30.0%\n\n\\-20.0%\n\n\\-10.0%\n\n0\\.0%\n\n10\\.0%\n\n20\\.0%\n\n30\\.0%\n\n40\\.0%\n\n1980 1985 1990 1995 2000 2005 2010 2015 2020\n\nAnnual Total Return Annual Max Drawdown\n\n**1980 - 2024**\n\nAverage Annual Total Return: 13.5% Average Annual Max Drawdown: -14.1%","publish_date":null,"title":"S&P 500 Annual Total Return and Max Drawdown","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/cimquarterlychartbook2025q4-updated_p15.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/cimquarterlychartbook2025q4-updated_p15.pdf"},"request":{"example_id":"chart/cimquarterlychartbook2025q4-updated_p15","product_type":"parse","source_file_path":"cimquarterlychartbook2025q4-updated_p15.pdf"},"started_at":"2026-11-04T19:44:51.757405"} +{"completed_at":"2026-11-04T19:44:54.415625","latency_in_ms":1603,"output":{"example_id":"chart/cimquarterlychartbook2025q4-updated_p16","grounded_pages":[],"layout_pages":[],"markdown":"_Source: Cetera Investment Management, FactSet, Standard & Poor’s. Total returns used. Past performance is not indicative of f_ _uture results. Data as of 9/30/2025._ **16**\n\n# Rolling S&P 500 Returns Since 1950\n\n### The chart shows the rolling 5, 10, and 15-\n\n### year annualized returns for the S&P 500\n\n### (including dividend reinvestment) since\n\n### 1950\\. Returns are through the end of\n\n### each quarter. As an example, the most\n\n### recent 5-year rolling return on this chart\n\n### represents the annualized return from\n\n### 10/1/2020 through 9/30/2025. The\n\n### average return for each rolling chart is\n\n### represented by the dashed black line.\n\n### Shorter holding periods have larger\n\n### variability in returns.\n\n## **Rolling S&P 500 Returns Since 1950**","pages":[{"line_numbers":[],"markdown":"_Source: Cetera Investment Management, FactSet, Standard & Poor’s. Total returns used. Past performance is not indicative of f_ _uture results. Data as of 9/30/2025._ **16**\n\n# Rolling S&P 500 Returns Since 1950\n\n### The chart shows the rolling 5, 10, and 15-\n\n### year annualized returns for the S&P 500\n\n### (including dividend reinvestment) since\n\n### 1950\\. Returns are through the end of\n\n### each quarter. As an example, the most\n\n### recent 5-year rolling return on this chart\n\n### represents the annualized return from\n\n### 10/1/2020 through 9/30/2025. The\n\n### average return for each rolling chart is\n\n### represented by the dashed black line.\n\n### Shorter holding periods have larger\n\n### variability in returns.\n\n## **Rolling S&P 500 Returns Since 1950**","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"_Source: Cetera Investment Management, FactSet, Standard & Poor’s. Total returns used. Past performance is not indicative of f_ _uture results. Data as of 9/30/2025._ **16**\n\n# Rolling S&P 500 Returns Since 1950\n\n### The chart shows the rolling 5, 10, and 15-\n\n### year annualized returns for the S&P 500\n\n### (including dividend reinvestment) since\n\n### 1950\\. Returns are through the end of\n\n### each quarter. As an example, the most\n\n### recent 5-year rolling return on this chart\n\n### represents the annualized return from\n\n### 10/1/2020 through 9/30/2025. The\n\n### average return for each rolling chart is\n\n### represented by the dashed black line.\n\n### Shorter holding periods have larger\n\n### variability in returns.\n\n## **Rolling S&P 500 Returns Since 1950**","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/cimquarterlychartbook2025q4-updated_p16.pdf"]},"response":{"errors":[],"extract_id":"extract_aa0f45f527b73a63ec5f0b67b849d636","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["_Source: Cetera Investment Management, FactSet, Standard & Poor’s. Total returns used. Past performance is not indicative of f_ _uture results. Data as of 9/30/2025._ **16**\n\n# Rolling S&P 500 Returns Since 1950\n\n### The chart shows the rolling 5, 10, and 15-\n\n### year annualized returns for the S&P 500\n\n### (including dividend reinvestment) since\n\n### 1950\\. Returns are through the end of\n\n### each quarter. As an example, the most\n\n### recent 5-year rolling return on this chart\n\n### represents the annualized return from\n\n### 10/1/2020 through 9/30/2025. The\n\n### average return for each rolling chart is\n\n### represented by the dashed black line.\n\n### Shorter holding periods have larger\n\n### variability in returns.\n\n## **Rolling S&P 500 Returns Since 1950**"],"full_content":"_Source: Cetera Investment Management, FactSet, Standard & Poor’s. Total returns used. Past performance is not indicative of f_ _uture results. Data as of 9/30/2025._ **16**\n\n# Rolling S&P 500 Returns Since 1950\n\n### The chart shows the rolling 5, 10, and 15-\n\n### year annualized returns for the S&P 500\n\n### (including dividend reinvestment) since\n\n### 1950\\. Returns are through the end of\n\n### each quarter. As an example, the most\n\n### recent 5-year rolling return on this chart\n\n### represents the annualized return from\n\n### 10/1/2020 through 9/30/2025. The\n\n### average return for each rolling chart is\n\n### represented by the dashed black line.\n\n### Shorter holding periods have larger\n\n### variability in returns.\n\n## **Rolling S&P 500 Returns Since 1950**","publish_date":null,"title":"Rolling S&P 500 Returns Since 1950","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/cimquarterlychartbook2025q4-updated_p16.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/cimquarterlychartbook2025q4-updated_p16.pdf"},"request":{"example_id":"chart/cimquarterlychartbook2025q4-updated_p16","product_type":"parse","source_file_path":"cimquarterlychartbook2025q4-updated_p16.pdf"},"started_at":"2026-11-04T19:44:52.812308"} +{"completed_at":"2026-11-04T19:44:56.289703","latency_in_ms":2896,"output":{"example_id":"chart/cimquarterlychartbook2025q4-updated_p17","grounded_pages":[],"layout_pages":[],"markdown":"_Source: Cetera Investment Management, FactSet, Standard & Poor’s. Data as of 10/3/2025._ **17**\n\n# Valuation Gap\n\n### The S&P 500 is trading at elevated valuations, with a\n\n### forward P/E ratio of 22.9 versus its 15-year average of 17.2. In contrast, the S&P 500 Equal Weight Index has a P/E of just 17.2, suggesting more reasonable valuations for the average stock. A lot of the froth in the headline index’s valuations is concentrated in larger stocks that carry more weight, particularly megacap\n\n### growth names tied to AI.","pages":[{"line_numbers":[],"markdown":"_Source: Cetera Investment Management, FactSet, Standard & Poor’s. Data as of 10/3/2025._ **17**\n\n# Valuation Gap\n\n### The S&P 500 is trading at elevated valuations, with a\n\n### forward P/E ratio of 22.9 versus its 15-year average of 17.2. In contrast, the S&P 500 Equal Weight Index has a P/E of just 17.2, suggesting more reasonable valuations for the average stock. A lot of the froth in the headline index’s valuations is concentrated in larger stocks that carry more weight, particularly megacap\n\n### growth names tied to AI.","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"_Source: Cetera Investment Management, FactSet, Standard & Poor’s. Data as of 10/3/2025._ **17**\n\n# Valuation Gap\n\n### The S&P 500 is trading at elevated valuations, with a\n\n### forward P/E ratio of 22.9 versus its 15-year average of 17.2. In contrast, the S&P 500 Equal Weight Index has a P/E of just 17.2, suggesting more reasonable valuations for the average stock. A lot of the froth in the headline index’s valuations is concentrated in larger stocks that carry more weight, particularly megacap\n\n### growth names tied to AI.","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/cimquarterlychartbook2025q4-updated_p17.pdf"]},"response":{"errors":[],"extract_id":"extract_e75c4640e01a25a1f7dda5ee1b10b75a","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["_Source: Cetera Investment Management, FactSet, Standard & Poor’s. Data as of 10/3/2025._ **17**\n\n# Valuation Gap\n\n### The S&P 500 is trading at elevated valuations, with a\n\n### forward P/E ratio of 22.9 versus its 15-year average of 17.2. In contrast, the S&P 500 Equal Weight Index has a P/E of just 17.2, suggesting more reasonable valuations for the average stock. A lot of the froth in the headline index’s valuations is concentrated in larger stocks that carry more weight, particularly megacap\n\n### growth names tied to AI."],"full_content":"_Source: Cetera Investment Management, FactSet, Standard & Poor’s. Data as of 10/3/2025._ **17**\n\n# Valuation Gap\n\n### The S&P 500 is trading at elevated valuations, with a\n\n### forward P/E ratio of 22.9 versus its 15-year average of 17.2. In contrast, the S&P 500 Equal Weight Index has a P/E of just 17.2, suggesting more reasonable valuations for the average stock. A lot of the froth in the headline index’s valuations is concentrated in larger stocks that carry more weight, particularly megacap\n\n### growth names tied to AI.","publish_date":null,"title":"Valuation Gap","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/cimquarterlychartbook2025q4-updated_p17.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/cimquarterlychartbook2025q4-updated_p17.pdf"},"request":{"example_id":"chart/cimquarterlychartbook2025q4-updated_p17","product_type":"parse","source_file_path":"cimquarterlychartbook2025q4-updated_p17.pdf"},"started_at":"2026-11-04T19:44:53.393172"} +{"completed_at":"2026-11-04T19:44:56.129006","latency_in_ms":1264,"output":{"example_id":"chart/deloitte-2025-global-automotive-consumer-study-january-2025_1_p15","grounded_pages":[],"layout_pages":[],"markdown":"**2025 Global Automotive Consumer Study** | Key Findings: Global Focus Markets\n\n**More than half of surveyed consumers in China owned the same brand of vehicle prior to their current car,** **but nearly a third of survey respondents say they are first-time owners, signaling a need to build strong** **customer relationships to help ensure forward brand loyalty.**\n\n**Percentage of consumers whose prior vehicle was from the same brand as current vehicle**\n\nQ6. Was your prior vehicle from the same brand? Sample size: n = 852 [China]; 1,114 [Germany]; 646 [India]; 452 [Japan]; 618 [Republic of Korea]; 3,488 [Southeast Asia]; 1,044 [UK]; 821 [US]\n\n**However, three quarters of surveyed consumers in China say they intend to switch to another brand of vehicle the next time they are in-market, signaling the intensity of the competitive landscape in China. Conversely, consumers surveyed in Japan are generally more brand loyal, but the intended defection rate is rising across many global markets on a year-over-year basis.**\n\n**Percentage of consumers intending to switch to another brand\\* of vehicle**\n\n\\*Includes switching to a different brand from the same parent or a different brand from a different sales parent. Q5. What brand is the vehicle you drive most often? Q26. What brand are you considering most for your next vehicle? [Brand switching percentage is based on a calculation involving these two questions.] Sample size: n = 830 [China]; 1,073 [Germany]; 633 [India]; 398 [Japan]; 589 [Republic of Korea]; 3,807 [Southeast Asia]; 959 [UK]; 786 [US]\n\n**55% 55% 54% 51% 47% 45% 43% 42%**\n\n**15%**\n\n**36% 31% 45% 48%**\n\n**31% 48% 51%**\n\n**31%**\n\n**9% 14%**\n\n**4% 5%**\n\n**25%**\n\n**9% 7%**\n\nChina Japan India Germany US Southeast Asia Rep. of Korea UK\n\nYes No Current car is my first vehicle\n\n**76% 72% 71%**\n\n**63%**\n\n**56% 54%**\n\n**46%**\n\n**39%**\n\n**73% 78%**\n\n**62% 57% 54% 51%**\n\n**41% 35%**\n\nChina India Southeast Asia Rep. of Korea UK US Germany Japan\n\n2025\n\n2024\n\n15","pages":[{"line_numbers":[],"markdown":"**2025 Global Automotive Consumer Study** | Key Findings: Global Focus Markets\n\n**More than half of surveyed consumers in China owned the same brand of vehicle prior to their current car,** **but nearly a third of survey respondents say they are first-time owners, signaling a need to build strong** **customer relationships to help ensure forward brand loyalty.**\n\n**Percentage of consumers whose prior vehicle was from the same brand as current vehicle**\n\nQ6. Was your prior vehicle from the same brand? Sample size: n = 852 [China]; 1,114 [Germany]; 646 [India]; 452 [Japan]; 618 [Republic of Korea]; 3,488 [Southeast Asia]; 1,044 [UK]; 821 [US]\n\n**However, three quarters of surveyed consumers in China say they intend to switch to another brand of vehicle the next time they are in-market, signaling the intensity of the competitive landscape in China. Conversely, consumers surveyed in Japan are generally more brand loyal, but the intended defection rate is rising across many global markets on a year-over-year basis.**\n\n**Percentage of consumers intending to switch to another brand\\* of vehicle**\n\n\\*Includes switching to a different brand from the same parent or a different brand from a different sales parent. Q5. What brand is the vehicle you drive most often? Q26. What brand are you considering most for your next vehicle? [Brand switching percentage is based on a calculation involving these two questions.] Sample size: n = 830 [China]; 1,073 [Germany]; 633 [India]; 398 [Japan]; 589 [Republic of Korea]; 3,807 [Southeast Asia]; 959 [UK]; 786 [US]\n\n**55% 55% 54% 51% 47% 45% 43% 42%**\n\n**15%**\n\n**36% 31% 45% 48%**\n\n**31% 48% 51%**\n\n**31%**\n\n**9% 14%**\n\n**4% 5%**\n\n**25%**\n\n**9% 7%**\n\nChina Japan India Germany US Southeast Asia Rep. of Korea UK\n\nYes No Current car is my first vehicle\n\n**76% 72% 71%**\n\n**63%**\n\n**56% 54%**\n\n**46%**\n\n**39%**\n\n**73% 78%**\n\n**62% 57% 54% 51%**\n\n**41% 35%**\n\nChina India Southeast Asia Rep. of Korea UK US Germany Japan\n\n2025\n\n2024\n\n15","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**2025 Global Automotive Consumer Study** | Key Findings: Global Focus Markets\n\n**More than half of surveyed consumers in China owned the same brand of vehicle prior to their current car,** **but nearly a third of survey respondents say they are first-time owners, signaling a need to build strong** **customer relationships to help ensure forward brand loyalty.**\n\n**Percentage of consumers whose prior vehicle was from the same brand as current vehicle**\n\nQ6. Was your prior vehicle from the same brand? Sample size: n = 852 [China]; 1,114 [Germany]; 646 [India]; 452 [Japan]; 618 [Republic of Korea]; 3,488 [Southeast Asia]; 1,044 [UK]; 821 [US]\n\n**However, three quarters of surveyed consumers in China say they intend to switch to another brand of vehicle the next time they are in-market, signaling the intensity of the competitive landscape in China. Conversely, consumers surveyed in Japan are generally more brand loyal, but the intended defection rate is rising across many global markets on a year-over-year basis.**\n\n**Percentage of consumers intending to switch to another brand\\* of vehicle**\n\n\\*Includes switching to a different brand from the same parent or a different brand from a different sales parent. Q5. What brand is the vehicle you drive most often? Q26. What brand are you considering most for your next vehicle? [Brand switching percentage is based on a calculation involving these two questions.] Sample size: n = 830 [China]; 1,073 [Germany]; 633 [India]; 398 [Japan]; 589 [Republic of Korea]; 3,807 [Southeast Asia]; 959 [UK]; 786 [US]\n\n**55% 55% 54% 51% 47% 45% 43% 42%**\n\n**15%**\n\n**36% 31% 45% 48%**\n\n**31% 48% 51%**\n\n**31%**\n\n**9% 14%**\n\n**4% 5%**\n\n**25%**\n\n**9% 7%**\n\nChina Japan India Germany US Southeast Asia Rep. of Korea UK\n\nYes No Current car is my first vehicle\n\n**76% 72% 71%**\n\n**63%**\n\n**56% 54%**\n\n**46%**\n\n**39%**\n\n**73% 78%**\n\n**62% 57% 54% 51%**\n\n**41% 35%**\n\nChina India Southeast Asia Rep. of Korea UK US Germany Japan\n\n2025\n\n2024\n\n15","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/deloitte-2025-global-automotive-consumer-study-january-2025_1_p15.pdf"]},"response":{"errors":[],"extract_id":"extract_3481dd6bac56382b8820cdb05167a18f","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**2025 Global Automotive Consumer Study** | Key Findings: Global Focus Markets\n\n**More than half of surveyed consumers in China owned the same brand of vehicle prior to their current car,** **but nearly a third of survey respondents say they are first-time owners, signaling a need to build strong** **customer relationships to help ensure forward brand loyalty.**\n\n**Percentage of consumers whose prior vehicle was from the same brand as current vehicle**\n\nQ6. Was your prior vehicle from the same brand? Sample size: n = 852 [China]; 1,114 [Germany]; 646 [India]; 452 [Japan]; 618 [Republic of Korea]; 3,488 [Southeast Asia]; 1,044 [UK]; 821 [US]\n\n**However, three quarters of surveyed consumers in China say they intend to switch to another brand of vehicle the next time they are in-market, signaling the intensity of the competitive landscape in China. Conversely, consumers surveyed in Japan are generally more brand loyal, but the intended defection rate is rising across many global markets on a year-over-year basis.**\n\n**Percentage of consumers intending to switch to another brand\\* of vehicle**\n\n\\*Includes switching to a different brand from the same parent or a different brand from a different sales parent. Q5. What brand is the vehicle you drive most often? Q26. What brand are you considering most for your next vehicle? [Brand switching percentage is based on a calculation involving these two questions.] Sample size: n = 830 [China]; 1,073 [Germany]; 633 [India]; 398 [Japan]; 589 [Republic of Korea]; 3,807 [Southeast Asia]; 959 [UK]; 786 [US]\n\n**55% 55% 54% 51% 47% 45% 43% 42%**\n\n**15%**\n\n**36% 31% 45% 48%**\n\n**31% 48% 51%**\n\n**31%**\n\n**9% 14%**\n\n**4% 5%**\n\n**25%**\n\n**9% 7%**\n\nChina Japan India Germany US Southeast Asia Rep. of Korea UK\n\nYes No Current car is my first vehicle\n\n**76% 72% 71%**\n\n**63%**\n\n**56% 54%**\n\n**46%**\n\n**39%**\n\n**73% 78%**\n\n**62% 57% 54% 51%**\n\n**41% 35%**\n\nChina India Southeast Asia Rep. of Korea UK US Germany Japan\n\n2025\n\n2024\n\n15"],"full_content":"**2025 Global Automotive Consumer Study** | Key Findings: Global Focus Markets\n\n**More than half of surveyed consumers in China owned the same brand of vehicle prior to their current car,** **but nearly a third of survey respondents say they are first-time owners, signaling a need to build strong** **customer relationships to help ensure forward brand loyalty.**\n\n**Percentage of consumers whose prior vehicle was from the same brand as current vehicle**\n\nQ6. Was your prior vehicle from the same brand? Sample size: n = 852 [China]; 1,114 [Germany]; 646 [India]; 452 [Japan]; 618 [Republic of Korea]; 3,488 [Southeast Asia]; 1,044 [UK]; 821 [US]\n\n**However, three quarters of surveyed consumers in China say they intend to switch to another brand of vehicle the next time they are in-market, signaling the intensity of the competitive landscape in China. Conversely, consumers surveyed in Japan are generally more brand loyal, but the intended defection rate is rising across many global markets on a year-over-year basis.**\n\n**Percentage of consumers intending to switch to another brand\\* of vehicle**\n\n\\*Includes switching to a different brand from the same parent or a different brand from a different sales parent. Q5. What brand is the vehicle you drive most often? Q26. What brand are you considering most for your next vehicle? [Brand switching percentage is based on a calculation involving these two questions.] Sample size: n = 830 [China]; 1,073 [Germany]; 633 [India]; 398 [Japan]; 589 [Republic of Korea]; 3,807 [Southeast Asia]; 959 [UK]; 786 [US]\n\n**55% 55% 54% 51% 47% 45% 43% 42%**\n\n**15%**\n\n**36% 31% 45% 48%**\n\n**31% 48% 51%**\n\n**31%**\n\n**9% 14%**\n\n**4% 5%**\n\n**25%**\n\n**9% 7%**\n\nChina Japan India Germany US Southeast Asia Rep. of Korea UK\n\nYes No Current car is my first vehicle\n\n**76% 72% 71%**\n\n**63%**\n\n**56% 54%**\n\n**46%**\n\n**39%**\n\n**73% 78%**\n\n**62% 57% 54% 51%**\n\n**41% 35%**\n\nChina India Southeast Asia Rep. of Korea UK US Germany Japan\n\n2025\n\n2024\n\n15","publish_date":null,"title":"More than half of surveyed consumers in China owned the same brand of vehicle prior to their current car, but","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/deloitte-2025-global-automotive-consumer-study-january-2025_1_p15.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/deloitte-2025-global-automotive-consumer-study-january-2025_1_p15.pdf"},"request":{"example_id":"chart/deloitte-2025-global-automotive-consumer-study-january-2025_1_p15","product_type":"parse","source_file_path":"deloitte-2025-global-automotive-consumer-study-january-2025_1_p15.pdf"},"started_at":"2026-11-04T19:44:54.864171"} +{"completed_at":"2026-11-04T19:44:57.420652","latency_in_ms":1129,"output":{"example_id":"chart/deloitte-2025-global-automotive-consumer-study-january-2025_1_p17","grounded_pages":[],"layout_pages":[],"markdown":"17\n\n**2025 Global Automotive Consumer Study** | Key Findings: Global Focus Markets\n\n**As global trade policies shift, encouraging a more regional approach to vehicle manufacturing footprints** **and supply chains, consumer sentiment differs among respondents when it comes to whether local manu** **\\- facturing is an important criteria for choosing a vehicle.**\n\n**Importance of next vehicle to be manufactured locally**\n\nNote: Percentages may not add to 100 due to rounding. Q40. How would you most prefer to pay for public EV charging? Sample size: n= 371 [China]; 295 [Germany]; 194 [India]; 103 [Japan]; 213 [Republic of Korea]; 1,163 [Southeast Asia]; 112 [US]\n\nMost preferred mode of payment\n\n**As some original equipment manufacturers (OEMs) continue to explore the potential for ‘agency’ retail models, surveyed consumers in India and China are more receptive to the concept of buying a vehicle directly from the manufacturer compared to consumers in Japan. Also, surveyed consumers in some mar- kets are signaling a high level of interest in purchasing vehicle insurance directly from the manufacturer.**\n\nQ61. To what extent are you interested in acquiring your next vehicle directly from the manufacturer (via an online process)? Q60. The next time you acquire a vehicle, how interested would you be in purchasing insurance directly from the vehicle manufacturer? Sample size: n = 939 [China]; 1,306 [Germany]; 882 [India]; 637 [Japan]; 906 [Republic of Korea]; 5.028 [Southeast Asia]; 1,314 [UK]; 937 [US]\n\n**49%**\n\n**36%**\n\n**15%**\n\n**10%**\n\n**28%**\n\n**19%**\n\n**13%**\n\n**10%**\n\n**29%**\n\n**31%**\n\n**45%**\n\n**34%**\n\n**31%**\n\n**27%**\n\n**22%**\n\n**19%**\n\n**14%**\n\n**22%**\n\n**24%**\n\n**39%**\n\n**24%**\n\n**33%**\n\n**39%**\n\n**31%**\n\n**8%**\n\n**11%**\n\n**16%**\n\n**17%**\n\n**17%**\n\n**21%**\n\n**26%**\n\n**41%**\n\nUK\n\nGermany\n\nJapan\n\nRep. of Korea\n\nUS\n\nSoutheast Asia\n\nChina\n\nIndia\n\nNot important Somewhat important Important Very important\n\n**34%**\n\n**38%**\n\n**40%**\n\n**46%**\n\n**46%**\n\n**59%**\n\n**72%**\n\n**76%**\n\nJapan\n\nGermany\n\nUK\n\nUS\n\nRep. of Korea\n\nSoutheast Asia\n\nChina\n\nIndia\n\nPercentage of surveyed consumers who would be interested in acquiring **vehicle directly from the manufacturer** Percentage of surveyed consumers who would be interested in purchasing **insurance directly from the manufacturer**\n\n**32%**\n\n**37%**\n\n**41%**\n\n**42%**\n\n**55%**\n\n**71%**\n\n**76%**\n\n**82%**\n\nGermany\n\nJapan\n\nUS\n\nUK\n\nRep. of Korea\n\nSoutheast Asia\n\nChina\n\nIndia\n\nNote: Percentages may not add up to 100 due to rounding. Q31. To what extent is it important that your next vehicle be locally manufactured (i.e., manufactured in your country or region)? Sample size: n = 939 [China]; 1,306 [Germany]; 882 [India]; 637 [Japan]; 906 [Republic of Korea]; 5,028 [Southeast Asia]; 1,315 [UK]; 937 [US]","pages":[{"line_numbers":[],"markdown":"17\n\n**2025 Global Automotive Consumer Study** | Key Findings: Global Focus Markets\n\n**As global trade policies shift, encouraging a more regional approach to vehicle manufacturing footprints** **and supply chains, consumer sentiment differs among respondents when it comes to whether local manu** **\\- facturing is an important criteria for choosing a vehicle.**\n\n**Importance of next vehicle to be manufactured locally**\n\nNote: Percentages may not add to 100 due to rounding. Q40. How would you most prefer to pay for public EV charging? Sample size: n= 371 [China]; 295 [Germany]; 194 [India]; 103 [Japan]; 213 [Republic of Korea]; 1,163 [Southeast Asia]; 112 [US]\n\nMost preferred mode of payment\n\n**As some original equipment manufacturers (OEMs) continue to explore the potential for ‘agency’ retail models, surveyed consumers in India and China are more receptive to the concept of buying a vehicle directly from the manufacturer compared to consumers in Japan. Also, surveyed consumers in some mar- kets are signaling a high level of interest in purchasing vehicle insurance directly from the manufacturer.**\n\nQ61. To what extent are you interested in acquiring your next vehicle directly from the manufacturer (via an online process)? Q60. The next time you acquire a vehicle, how interested would you be in purchasing insurance directly from the vehicle manufacturer? Sample size: n = 939 [China]; 1,306 [Germany]; 882 [India]; 637 [Japan]; 906 [Republic of Korea]; 5.028 [Southeast Asia]; 1,314 [UK]; 937 [US]\n\n**49%**\n\n**36%**\n\n**15%**\n\n**10%**\n\n**28%**\n\n**19%**\n\n**13%**\n\n**10%**\n\n**29%**\n\n**31%**\n\n**45%**\n\n**34%**\n\n**31%**\n\n**27%**\n\n**22%**\n\n**19%**\n\n**14%**\n\n**22%**\n\n**24%**\n\n**39%**\n\n**24%**\n\n**33%**\n\n**39%**\n\n**31%**\n\n**8%**\n\n**11%**\n\n**16%**\n\n**17%**\n\n**17%**\n\n**21%**\n\n**26%**\n\n**41%**\n\nUK\n\nGermany\n\nJapan\n\nRep. of Korea\n\nUS\n\nSoutheast Asia\n\nChina\n\nIndia\n\nNot important Somewhat important Important Very important\n\n**34%**\n\n**38%**\n\n**40%**\n\n**46%**\n\n**46%**\n\n**59%**\n\n**72%**\n\n**76%**\n\nJapan\n\nGermany\n\nUK\n\nUS\n\nRep. of Korea\n\nSoutheast Asia\n\nChina\n\nIndia\n\nPercentage of surveyed consumers who would be interested in acquiring **vehicle directly from the manufacturer** Percentage of surveyed consumers who would be interested in purchasing **insurance directly from the manufacturer**\n\n**32%**\n\n**37%**\n\n**41%**\n\n**42%**\n\n**55%**\n\n**71%**\n\n**76%**\n\n**82%**\n\nGermany\n\nJapan\n\nUS\n\nUK\n\nRep. of Korea\n\nSoutheast Asia\n\nChina\n\nIndia\n\nNote: Percentages may not add up to 100 due to rounding. Q31. To what extent is it important that your next vehicle be locally manufactured (i.e., manufactured in your country or region)? Sample size: n = 939 [China]; 1,306 [Germany]; 882 [India]; 637 [Japan]; 906 [Republic of Korea]; 5,028 [Southeast Asia]; 1,315 [UK]; 937 [US]","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"17\n\n**2025 Global Automotive Consumer Study** | Key Findings: Global Focus Markets\n\n**As global trade policies shift, encouraging a more regional approach to vehicle manufacturing footprints** **and supply chains, consumer sentiment differs among respondents when it comes to whether local manu** **\\- facturing is an important criteria for choosing a vehicle.**\n\n**Importance of next vehicle to be manufactured locally**\n\nNote: Percentages may not add to 100 due to rounding. Q40. How would you most prefer to pay for public EV charging? Sample size: n= 371 [China]; 295 [Germany]; 194 [India]; 103 [Japan]; 213 [Republic of Korea]; 1,163 [Southeast Asia]; 112 [US]\n\nMost preferred mode of payment\n\n**As some original equipment manufacturers (OEMs) continue to explore the potential for ‘agency’ retail models, surveyed consumers in India and China are more receptive to the concept of buying a vehicle directly from the manufacturer compared to consumers in Japan. Also, surveyed consumers in some mar- kets are signaling a high level of interest in purchasing vehicle insurance directly from the manufacturer.**\n\nQ61. To what extent are you interested in acquiring your next vehicle directly from the manufacturer (via an online process)? Q60. The next time you acquire a vehicle, how interested would you be in purchasing insurance directly from the vehicle manufacturer? Sample size: n = 939 [China]; 1,306 [Germany]; 882 [India]; 637 [Japan]; 906 [Republic of Korea]; 5.028 [Southeast Asia]; 1,314 [UK]; 937 [US]\n\n**49%**\n\n**36%**\n\n**15%**\n\n**10%**\n\n**28%**\n\n**19%**\n\n**13%**\n\n**10%**\n\n**29%**\n\n**31%**\n\n**45%**\n\n**34%**\n\n**31%**\n\n**27%**\n\n**22%**\n\n**19%**\n\n**14%**\n\n**22%**\n\n**24%**\n\n**39%**\n\n**24%**\n\n**33%**\n\n**39%**\n\n**31%**\n\n**8%**\n\n**11%**\n\n**16%**\n\n**17%**\n\n**17%**\n\n**21%**\n\n**26%**\n\n**41%**\n\nUK\n\nGermany\n\nJapan\n\nRep. of Korea\n\nUS\n\nSoutheast Asia\n\nChina\n\nIndia\n\nNot important Somewhat important Important Very important\n\n**34%**\n\n**38%**\n\n**40%**\n\n**46%**\n\n**46%**\n\n**59%**\n\n**72%**\n\n**76%**\n\nJapan\n\nGermany\n\nUK\n\nUS\n\nRep. of Korea\n\nSoutheast Asia\n\nChina\n\nIndia\n\nPercentage of surveyed consumers who would be interested in acquiring **vehicle directly from the manufacturer** Percentage of surveyed consumers who would be interested in purchasing **insurance directly from the manufacturer**\n\n**32%**\n\n**37%**\n\n**41%**\n\n**42%**\n\n**55%**\n\n**71%**\n\n**76%**\n\n**82%**\n\nGermany\n\nJapan\n\nUS\n\nUK\n\nRep. of Korea\n\nSoutheast Asia\n\nChina\n\nIndia\n\nNote: Percentages may not add up to 100 due to rounding. Q31. To what extent is it important that your next vehicle be locally manufactured (i.e., manufactured in your country or region)? Sample size: n = 939 [China]; 1,306 [Germany]; 882 [India]; 637 [Japan]; 906 [Republic of Korea]; 5,028 [Southeast Asia]; 1,315 [UK]; 937 [US]","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/deloitte-2025-global-automotive-consumer-study-january-2025_1_p17.pdf"]},"response":{"errors":[],"extract_id":"extract_c196949a8426dec8baa05e3720585a20","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["17\n\n**2025 Global Automotive Consumer Study** | Key Findings: Global Focus Markets\n\n**As global trade policies shift, encouraging a more regional approach to vehicle manufacturing footprints** **and supply chains, consumer sentiment differs among respondents when it comes to whether local manu** **\\- facturing is an important criteria for choosing a vehicle.**\n\n**Importance of next vehicle to be manufactured locally**\n\nNote: Percentages may not add to 100 due to rounding. Q40. How would you most prefer to pay for public EV charging? Sample size: n= 371 [China]; 295 [Germany]; 194 [India]; 103 [Japan]; 213 [Republic of Korea]; 1,163 [Southeast Asia]; 112 [US]\n\nMost preferred mode of payment\n\n**As some original equipment manufacturers (OEMs) continue to explore the potential for ‘agency’ retail models, surveyed consumers in India and China are more receptive to the concept of buying a vehicle directly from the manufacturer compared to consumers in Japan. Also, surveyed consumers in some mar- kets are signaling a high level of interest in purchasing vehicle insurance directly from the manufacturer.**\n\nQ61. To what extent are you interested in acquiring your next vehicle directly from the manufacturer (via an online process)? Q60. The next time you acquire a vehicle, how interested would you be in purchasing insurance directly from the vehicle manufacturer? Sample size: n = 939 [China]; 1,306 [Germany]; 882 [India]; 637 [Japan]; 906 [Republic of Korea]; 5.028 [Southeast Asia]; 1,314 [UK]; 937 [US]\n\n**49%**\n\n**36%**\n\n**15%**\n\n**10%**\n\n**28%**\n\n**19%**\n\n**13%**\n\n**10%**\n\n**29%**\n\n**31%**\n\n**45%**\n\n**34%**\n\n**31%**\n\n**27%**\n\n**22%**\n\n**19%**\n\n**14%**\n\n**22%**\n\n**24%**\n\n**39%**\n\n**24%**\n\n**33%**\n\n**39%**\n\n**31%**\n\n**8%**\n\n**11%**\n\n**16%**\n\n**17%**\n\n**17%**\n\n**21%**\n\n**26%**\n\n**41%**\n\nUK\n\nGermany\n\nJapan\n\nRep. of Korea\n\nUS\n\nSoutheast Asia\n\nChina\n\nIndia\n\nNot important Somewhat important Important Very important\n\n**34%**\n\n**38%**\n\n**40%**\n\n**46%**\n\n**46%**\n\n**59%**\n\n**72%**\n\n**76%**\n\nJapan\n\nGermany\n\nUK\n\nUS\n\nRep. of Korea\n\nSoutheast Asia\n\nChina\n\nIndia\n\nPercentage of surveyed consumers who would be interested in acquiring **vehicle directly from the manufacturer** Percentage of surveyed consumers who would be interested in purchasing **insurance directly from the manufacturer**\n\n**32%**\n\n**37%**\n\n**41%**\n\n**42%**\n\n**55%**\n\n**71%**\n\n**76%**\n\n**82%**\n\nGermany\n\nJapan\n\nUS\n\nUK\n\nRep. of Korea\n\nSoutheast Asia\n\nChina\n\nIndia\n\nNote: Percentages may not add up to 100 due to rounding. Q31. To what extent is it important that your next vehicle be locally manufactured (i.e., manufactured in your country or region)? Sample size: n = 939 [China]; 1,306 [Germany]; 882 [India]; 637 [Japan]; 906 [Republic of Korea]; 5,028 [Southeast Asia]; 1,315 [UK]; 937 [US]"],"full_content":"17\n\n**2025 Global Automotive Consumer Study** | Key Findings: Global Focus Markets\n\n**As global trade policies shift, encouraging a more regional approach to vehicle manufacturing footprints** **and supply chains, consumer sentiment differs among respondents when it comes to whether local manu** **\\- facturing is an important criteria for choosing a vehicle.**\n\n**Importance of next vehicle to be manufactured locally**\n\nNote: Percentages may not add to 100 due to rounding. Q40. How would you most prefer to pay for public EV charging? Sample size: n= 371 [China]; 295 [Germany]; 194 [India]; 103 [Japan]; 213 [Republic of Korea]; 1,163 [Southeast Asia]; 112 [US]\n\nMost preferred mode of payment\n\n**As some original equipment manufacturers (OEMs) continue to explore the potential for ‘agency’ retail models, surveyed consumers in India and China are more receptive to the concept of buying a vehicle directly from the manufacturer compared to consumers in Japan. Also, surveyed consumers in some mar- kets are signaling a high level of interest in purchasing vehicle insurance directly from the manufacturer.**\n\nQ61. To what extent are you interested in acquiring your next vehicle directly from the manufacturer (via an online process)? Q60. The next time you acquire a vehicle, how interested would you be in purchasing insurance directly from the vehicle manufacturer? Sample size: n = 939 [China]; 1,306 [Germany]; 882 [India]; 637 [Japan]; 906 [Republic of Korea]; 5.028 [Southeast Asia]; 1,314 [UK]; 937 [US]\n\n**49%**\n\n**36%**\n\n**15%**\n\n**10%**\n\n**28%**\n\n**19%**\n\n**13%**\n\n**10%**\n\n**29%**\n\n**31%**\n\n**45%**\n\n**34%**\n\n**31%**\n\n**27%**\n\n**22%**\n\n**19%**\n\n**14%**\n\n**22%**\n\n**24%**\n\n**39%**\n\n**24%**\n\n**33%**\n\n**39%**\n\n**31%**\n\n**8%**\n\n**11%**\n\n**16%**\n\n**17%**\n\n**17%**\n\n**21%**\n\n**26%**\n\n**41%**\n\nUK\n\nGermany\n\nJapan\n\nRep. of Korea\n\nUS\n\nSoutheast Asia\n\nChina\n\nIndia\n\nNot important Somewhat important Important Very important\n\n**34%**\n\n**38%**\n\n**40%**\n\n**46%**\n\n**46%**\n\n**59%**\n\n**72%**\n\n**76%**\n\nJapan\n\nGermany\n\nUK\n\nUS\n\nRep. of Korea\n\nSoutheast Asia\n\nChina\n\nIndia\n\nPercentage of surveyed consumers who would be interested in acquiring **vehicle directly from the manufacturer** Percentage of surveyed consumers who would be interested in purchasing **insurance directly from the manufacturer**\n\n**32%**\n\n**37%**\n\n**41%**\n\n**42%**\n\n**55%**\n\n**71%**\n\n**76%**\n\n**82%**\n\nGermany\n\nJapan\n\nUS\n\nUK\n\nRep. of Korea\n\nSoutheast Asia\n\nChina\n\nIndia\n\nNote: Percentages may not add up to 100 due to rounding. Q31. To what extent is it important that your next vehicle be locally manufactured (i.e., manufactured in your country or region)? Sample size: n = 939 [China]; 1,306 [Germany]; 882 [India]; 637 [Japan]; 906 [Republic of Korea]; 5,028 [Southeast Asia]; 1,315 [UK]; 937 [US]","publish_date":null,"title":"As global trade policies shift, encouraging a more regional approach to vehicle manufacturing footprints and supply chains, consumer sentiment differs","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/deloitte-2025-global-automotive-consumer-study-january-2025_1_p17.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/deloitte-2025-global-automotive-consumer-study-january-2025_1_p17.pdf"},"request":{"example_id":"chart/deloitte-2025-global-automotive-consumer-study-january-2025_1_p17","product_type":"parse","source_file_path":"deloitte-2025-global-automotive-consumer-study-january-2025_1_p17.pdf"},"started_at":"2026-11-04T19:44:56.291467"} +{"completed_at":"2026-11-04T19:44:58.775011","latency_in_ms":1420,"output":{"example_id":"chart/deloitte-2025-global-automotive-consumer-study-january-2025_1_p20","grounded_pages":[],"layout_pages":[],"markdown":"20\n\n**2025 Global Automotive Consumer Study** | Key Findings: Global Focus Markets\n\n**Consumers surveyed in both India and the United Kingdom are more concerned with the idea of an** **autonomous vehicle fleet when compared to surveyed consumers in Germany, Japan, or China.**\n\n**36%**\n\n**41%**\n\n**43%**\n\n**44%**\n\n**48%**\n\n**52%**\n\n**52%**\n\n**63%**\n\nGermany\n\nJapan\n\nChina\n\nRep. of Korea\n\nSoutheast Asia\n\nUS\n\nUK\n\nIndia\n\n**47%**\n\n**51%**\n\n**52%**\n\n**64%**\n\n**64%**\n\n**65%**\n\n**67%**\n\n**74%**\n\nChina\n\nGermany\n\nJapan\n\nUS\n\nSoutheast Asia\n\nRep. of Korea\n\nUK\n\nIndia\n\nQ56. To what extent are you concerned with each of the following scenarios? Sample size: n = 939 [China]; 1,306 [Germany]; 882 [India]; 637 [Japan]; 906 [Republic of Korea]; 5,028 [Southeast Asia]; 1,314 [UK]; 937 [US]\n\n**More than three quarters of surveyed consumers in India and China believe the addition of artificial** **intelligence in vehicle systems is beneficial. On the other hand, a quarter of surveyed consumers in** **Germany, the United States, and the United Kingdom remain skeptical about the technology.**\n\n**Addition of artificial intelligence in vehicle systems**\n\nNote: Percentages may not add up to 100 due to rounding. Q62. To what extent do you think the addition of artificial intelligence in vehicle systems (e.g., voice activated features, autonomous driving) will be beneficial? Sample size: n = 939 [China]; 1,306 [Germany]; 882 [India]; 637 [Japan]; 906 [Republic of Korea]; 5,028 [Southeast Asia]; 1,314 [UK]; 937 [US]\n\n**25%**\n\n**21%**\n\n**26%**\n\n**26%**\n\n**11%**\n\n**6%**\n\n**5%**\n\n**5%**\n\n**34%**\n\n**37%**\n\n**31%**\n\n**29%**\n\n**27%**\n\n**22%**\n\n**18%**\n\n**13%**\n\n**41%**\n\n**42%**\n\n**43%**\n\n**45%**\n\n**62%**\n\n**72%**\n\n**77%**\n\n**82%**\n\nGermany\n\nJapan\n\nUK\n\nUS\n\nRep. of Korea\n\nSoutheast Asia\n\nChina\n\nIndia\n\nNot benefical Neutral Beneficial\n\nPercentage of consumers concerned about **fully autonomous robotaxi services** operating where they live\n\nPercentage of consumers concerned about **commercial vehicles operating in a fully autonomous mode** on the highway","pages":[{"line_numbers":[],"markdown":"20\n\n**2025 Global Automotive Consumer Study** | Key Findings: Global Focus Markets\n\n**Consumers surveyed in both India and the United Kingdom are more concerned with the idea of an** **autonomous vehicle fleet when compared to surveyed consumers in Germany, Japan, or China.**\n\n**36%**\n\n**41%**\n\n**43%**\n\n**44%**\n\n**48%**\n\n**52%**\n\n**52%**\n\n**63%**\n\nGermany\n\nJapan\n\nChina\n\nRep. of Korea\n\nSoutheast Asia\n\nUS\n\nUK\n\nIndia\n\n**47%**\n\n**51%**\n\n**52%**\n\n**64%**\n\n**64%**\n\n**65%**\n\n**67%**\n\n**74%**\n\nChina\n\nGermany\n\nJapan\n\nUS\n\nSoutheast Asia\n\nRep. of Korea\n\nUK\n\nIndia\n\nQ56. To what extent are you concerned with each of the following scenarios? Sample size: n = 939 [China]; 1,306 [Germany]; 882 [India]; 637 [Japan]; 906 [Republic of Korea]; 5,028 [Southeast Asia]; 1,314 [UK]; 937 [US]\n\n**More than three quarters of surveyed consumers in India and China believe the addition of artificial** **intelligence in vehicle systems is beneficial. On the other hand, a quarter of surveyed consumers in** **Germany, the United States, and the United Kingdom remain skeptical about the technology.**\n\n**Addition of artificial intelligence in vehicle systems**\n\nNote: Percentages may not add up to 100 due to rounding. Q62. To what extent do you think the addition of artificial intelligence in vehicle systems (e.g., voice activated features, autonomous driving) will be beneficial? Sample size: n = 939 [China]; 1,306 [Germany]; 882 [India]; 637 [Japan]; 906 [Republic of Korea]; 5,028 [Southeast Asia]; 1,314 [UK]; 937 [US]\n\n**25%**\n\n**21%**\n\n**26%**\n\n**26%**\n\n**11%**\n\n**6%**\n\n**5%**\n\n**5%**\n\n**34%**\n\n**37%**\n\n**31%**\n\n**29%**\n\n**27%**\n\n**22%**\n\n**18%**\n\n**13%**\n\n**41%**\n\n**42%**\n\n**43%**\n\n**45%**\n\n**62%**\n\n**72%**\n\n**77%**\n\n**82%**\n\nGermany\n\nJapan\n\nUK\n\nUS\n\nRep. of Korea\n\nSoutheast Asia\n\nChina\n\nIndia\n\nNot benefical Neutral Beneficial\n\nPercentage of consumers concerned about **fully autonomous robotaxi services** operating where they live\n\nPercentage of consumers concerned about **commercial vehicles operating in a fully autonomous mode** on the highway","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"20\n\n**2025 Global Automotive Consumer Study** | Key Findings: Global Focus Markets\n\n**Consumers surveyed in both India and the United Kingdom are more concerned with the idea of an** **autonomous vehicle fleet when compared to surveyed consumers in Germany, Japan, or China.**\n\n**36%**\n\n**41%**\n\n**43%**\n\n**44%**\n\n**48%**\n\n**52%**\n\n**52%**\n\n**63%**\n\nGermany\n\nJapan\n\nChina\n\nRep. of Korea\n\nSoutheast Asia\n\nUS\n\nUK\n\nIndia\n\n**47%**\n\n**51%**\n\n**52%**\n\n**64%**\n\n**64%**\n\n**65%**\n\n**67%**\n\n**74%**\n\nChina\n\nGermany\n\nJapan\n\nUS\n\nSoutheast Asia\n\nRep. of Korea\n\nUK\n\nIndia\n\nQ56. To what extent are you concerned with each of the following scenarios? Sample size: n = 939 [China]; 1,306 [Germany]; 882 [India]; 637 [Japan]; 906 [Republic of Korea]; 5,028 [Southeast Asia]; 1,314 [UK]; 937 [US]\n\n**More than three quarters of surveyed consumers in India and China believe the addition of artificial** **intelligence in vehicle systems is beneficial. On the other hand, a quarter of surveyed consumers in** **Germany, the United States, and the United Kingdom remain skeptical about the technology.**\n\n**Addition of artificial intelligence in vehicle systems**\n\nNote: Percentages may not add up to 100 due to rounding. Q62. To what extent do you think the addition of artificial intelligence in vehicle systems (e.g., voice activated features, autonomous driving) will be beneficial? Sample size: n = 939 [China]; 1,306 [Germany]; 882 [India]; 637 [Japan]; 906 [Republic of Korea]; 5,028 [Southeast Asia]; 1,314 [UK]; 937 [US]\n\n**25%**\n\n**21%**\n\n**26%**\n\n**26%**\n\n**11%**\n\n**6%**\n\n**5%**\n\n**5%**\n\n**34%**\n\n**37%**\n\n**31%**\n\n**29%**\n\n**27%**\n\n**22%**\n\n**18%**\n\n**13%**\n\n**41%**\n\n**42%**\n\n**43%**\n\n**45%**\n\n**62%**\n\n**72%**\n\n**77%**\n\n**82%**\n\nGermany\n\nJapan\n\nUK\n\nUS\n\nRep. of Korea\n\nSoutheast Asia\n\nChina\n\nIndia\n\nNot benefical Neutral Beneficial\n\nPercentage of consumers concerned about **fully autonomous robotaxi services** operating where they live\n\nPercentage of consumers concerned about **commercial vehicles operating in a fully autonomous mode** on the highway","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/deloitte-2025-global-automotive-consumer-study-january-2025_1_p20.pdf"]},"response":{"errors":[],"extract_id":"extract_c2367e3063750730b1b4a4b641705c48","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["20\n\n**2025 Global Automotive Consumer Study** | Key Findings: Global Focus Markets\n\n**Consumers surveyed in both India and the United Kingdom are more concerned with the idea of an** **autonomous vehicle fleet when compared to surveyed consumers in Germany, Japan, or China.**\n\n**36%**\n\n**41%**\n\n**43%**\n\n**44%**\n\n**48%**\n\n**52%**\n\n**52%**\n\n**63%**\n\nGermany\n\nJapan\n\nChina\n\nRep. of Korea\n\nSoutheast Asia\n\nUS\n\nUK\n\nIndia\n\n**47%**\n\n**51%**\n\n**52%**\n\n**64%**\n\n**64%**\n\n**65%**\n\n**67%**\n\n**74%**\n\nChina\n\nGermany\n\nJapan\n\nUS\n\nSoutheast Asia\n\nRep. of Korea\n\nUK\n\nIndia\n\nQ56. To what extent are you concerned with each of the following scenarios? Sample size: n = 939 [China]; 1,306 [Germany]; 882 [India]; 637 [Japan]; 906 [Republic of Korea]; 5,028 [Southeast Asia]; 1,314 [UK]; 937 [US]\n\n**More than three quarters of surveyed consumers in India and China believe the addition of artificial** **intelligence in vehicle systems is beneficial. On the other hand, a quarter of surveyed consumers in** **Germany, the United States, and the United Kingdom remain skeptical about the technology.**\n\n**Addition of artificial intelligence in vehicle systems**\n\nNote: Percentages may not add up to 100 due to rounding. Q62. To what extent do you think the addition of artificial intelligence in vehicle systems (e.g., voice activated features, autonomous driving) will be beneficial? Sample size: n = 939 [China]; 1,306 [Germany]; 882 [India]; 637 [Japan]; 906 [Republic of Korea]; 5,028 [Southeast Asia]; 1,314 [UK]; 937 [US]\n\n**25%**\n\n**21%**\n\n**26%**\n\n**26%**\n\n**11%**\n\n**6%**\n\n**5%**\n\n**5%**\n\n**34%**\n\n**37%**\n\n**31%**\n\n**29%**\n\n**27%**\n\n**22%**\n\n**18%**\n\n**13%**\n\n**41%**\n\n**42%**\n\n**43%**\n\n**45%**\n\n**62%**\n\n**72%**\n\n**77%**\n\n**82%**\n\nGermany\n\nJapan\n\nUK\n\nUS\n\nRep. of Korea\n\nSoutheast Asia\n\nChina\n\nIndia\n\nNot benefical Neutral Beneficial\n\nPercentage of consumers concerned about **fully autonomous robotaxi services** operating where they live\n\nPercentage of consumers concerned about **commercial vehicles operating in a fully autonomous mode** on the highway"],"full_content":"20\n\n**2025 Global Automotive Consumer Study** | Key Findings: Global Focus Markets\n\n**Consumers surveyed in both India and the United Kingdom are more concerned with the idea of an** **autonomous vehicle fleet when compared to surveyed consumers in Germany, Japan, or China.**\n\n**36%**\n\n**41%**\n\n**43%**\n\n**44%**\n\n**48%**\n\n**52%**\n\n**52%**\n\n**63%**\n\nGermany\n\nJapan\n\nChina\n\nRep. of Korea\n\nSoutheast Asia\n\nUS\n\nUK\n\nIndia\n\n**47%**\n\n**51%**\n\n**52%**\n\n**64%**\n\n**64%**\n\n**65%**\n\n**67%**\n\n**74%**\n\nChina\n\nGermany\n\nJapan\n\nUS\n\nSoutheast Asia\n\nRep. of Korea\n\nUK\n\nIndia\n\nQ56. To what extent are you concerned with each of the following scenarios? Sample size: n = 939 [China]; 1,306 [Germany]; 882 [India]; 637 [Japan]; 906 [Republic of Korea]; 5,028 [Southeast Asia]; 1,314 [UK]; 937 [US]\n\n**More than three quarters of surveyed consumers in India and China believe the addition of artificial** **intelligence in vehicle systems is beneficial. On the other hand, a quarter of surveyed consumers in** **Germany, the United States, and the United Kingdom remain skeptical about the technology.**\n\n**Addition of artificial intelligence in vehicle systems**\n\nNote: Percentages may not add up to 100 due to rounding. Q62. To what extent do you think the addition of artificial intelligence in vehicle systems (e.g., voice activated features, autonomous driving) will be beneficial? Sample size: n = 939 [China]; 1,306 [Germany]; 882 [India]; 637 [Japan]; 906 [Republic of Korea]; 5,028 [Southeast Asia]; 1,314 [UK]; 937 [US]\n\n**25%**\n\n**21%**\n\n**26%**\n\n**26%**\n\n**11%**\n\n**6%**\n\n**5%**\n\n**5%**\n\n**34%**\n\n**37%**\n\n**31%**\n\n**29%**\n\n**27%**\n\n**22%**\n\n**18%**\n\n**13%**\n\n**41%**\n\n**42%**\n\n**43%**\n\n**45%**\n\n**62%**\n\n**72%**\n\n**77%**\n\n**82%**\n\nGermany\n\nJapan\n\nUK\n\nUS\n\nRep. of Korea\n\nSoutheast Asia\n\nChina\n\nIndia\n\nNot benefical Neutral Beneficial\n\nPercentage of consumers concerned about **fully autonomous robotaxi services** operating where they live\n\nPercentage of consumers concerned about **commercial vehicles operating in a fully autonomous mode** on the highway","publish_date":null,"title":"Consumers surveyed in both India and the United Kingdom are more concerned with the idea of an autonomous vehicle fleet","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/deloitte-2025-global-automotive-consumer-study-january-2025_1_p20.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/deloitte-2025-global-automotive-consumer-study-january-2025_1_p20.pdf"},"request":{"example_id":"chart/deloitte-2025-global-automotive-consumer-study-january-2025_1_p20","product_type":"parse","source_file_path":"deloitte-2025-global-automotive-consumer-study-january-2025_1_p20.pdf"},"started_at":"2026-11-04T19:44:57.354733"} +{"completed_at":"2026-11-04T19:44:59.903939","latency_in_ms":1294,"output":{"example_id":"chart/deloitte-2025-global-automotive-consumer-study-january-2025_1_p8","grounded_pages":[],"layout_pages":[],"markdown":"8\n\n**2025 Global Automotive Consumer Study** | Key Findings: Global Focus Markets\n\nHome Work On-street/public charging station Other\n\nUK\n\n**80%**\n\n**10%**\n\n**11%**\n\n**10%**\n\n**11%**\n\nUS\n\n**79%**\n\nJapan\n\n**72%**\n\nGermany India China Southeast Asia Rep. of Korea\n\n**2%**\n\n**13%**\n\n**14%**\n\n**70%**\n\n**18%**\n\n**12%**\n\n**63%**\n\n**22%**\n\n**15%**\n\n**62%**\n\n**28%**\n\n**10%**\n\n**61%**\n\n**21%**\n\n**17%**\n\n**71%**\n\n**20%**\n\n**9%**\n\nYes, dedicated Yes, shared with a neighbor/others No Don’t know\n\nChina\n\n**77%**\n\n**15%**\n\n**7%**\n\nIndia\n\n**41%**\n\n**44%**\n\nUS\n\n**40%**\n\nGermany UK Southeast Asia Japan\n\n**14%**\n\n**58%**\n\n**2%**\n\n**33%**\n\n**62%**\n\n**5%**\n\n**32%**\n\n**58%**\n\n**8%**\n\n**28%**\n\n**65%**\n\n**4%**\n\nRep. of Korea\n\n**19%**\n\n**7%**\n\n**25%**\n\n**37%**\n\n**56%**\n\n**6%**\n\n**1% 1% 1% 2% 2%**\n\n**50%**\n\n**The push to create public charging capacity may be somewhat overstated in many markets as a** **strong majority of EV intenders surveyed plan to charge their vehicle at home.**\n\n**Expecting to charge electrified vehicle most often at…**\n\n**Yet, many EV intenders surveyed in mature markets that plan to charge at home do not currently have access to a charger.**\n\n**EV charger access among consumers who plan to charge their vehicle at home**\n\nNote: Percentages may not add up to 100 due to rounding. Q43. Where do you expect to charge your EV most often? Sample size: n = 413 [China]; 304 [Germany]; 182 [India]; 64 [Japan]; 166 [Republic of Korea]; 1,223 [Southeast Asia]; 245 [UK]; 111 [US]\n\nNote: Percentages may not add up to 100 due to rounding. Q44. Do you already have access to a charger at your residence? Sample size: n = 260 [China]; 215 [Germany]; 128 [India]; 46 [Japan]; 102 [Republic of Korea]; 755 [Southeast Asia]; 195 [UK]; 88 [US]","pages":[{"line_numbers":[],"markdown":"8\n\n**2025 Global Automotive Consumer Study** | Key Findings: Global Focus Markets\n\nHome Work On-street/public charging station Other\n\nUK\n\n**80%**\n\n**10%**\n\n**11%**\n\n**10%**\n\n**11%**\n\nUS\n\n**79%**\n\nJapan\n\n**72%**\n\nGermany India China Southeast Asia Rep. of Korea\n\n**2%**\n\n**13%**\n\n**14%**\n\n**70%**\n\n**18%**\n\n**12%**\n\n**63%**\n\n**22%**\n\n**15%**\n\n**62%**\n\n**28%**\n\n**10%**\n\n**61%**\n\n**21%**\n\n**17%**\n\n**71%**\n\n**20%**\n\n**9%**\n\nYes, dedicated Yes, shared with a neighbor/others No Don’t know\n\nChina\n\n**77%**\n\n**15%**\n\n**7%**\n\nIndia\n\n**41%**\n\n**44%**\n\nUS\n\n**40%**\n\nGermany UK Southeast Asia Japan\n\n**14%**\n\n**58%**\n\n**2%**\n\n**33%**\n\n**62%**\n\n**5%**\n\n**32%**\n\n**58%**\n\n**8%**\n\n**28%**\n\n**65%**\n\n**4%**\n\nRep. of Korea\n\n**19%**\n\n**7%**\n\n**25%**\n\n**37%**\n\n**56%**\n\n**6%**\n\n**1% 1% 1% 2% 2%**\n\n**50%**\n\n**The push to create public charging capacity may be somewhat overstated in many markets as a** **strong majority of EV intenders surveyed plan to charge their vehicle at home.**\n\n**Expecting to charge electrified vehicle most often at…**\n\n**Yet, many EV intenders surveyed in mature markets that plan to charge at home do not currently have access to a charger.**\n\n**EV charger access among consumers who plan to charge their vehicle at home**\n\nNote: Percentages may not add up to 100 due to rounding. Q43. Where do you expect to charge your EV most often? Sample size: n = 413 [China]; 304 [Germany]; 182 [India]; 64 [Japan]; 166 [Republic of Korea]; 1,223 [Southeast Asia]; 245 [UK]; 111 [US]\n\nNote: Percentages may not add up to 100 due to rounding. Q44. Do you already have access to a charger at your residence? Sample size: n = 260 [China]; 215 [Germany]; 128 [India]; 46 [Japan]; 102 [Republic of Korea]; 755 [Southeast Asia]; 195 [UK]; 88 [US]","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"8\n\n**2025 Global Automotive Consumer Study** | Key Findings: Global Focus Markets\n\nHome Work On-street/public charging station Other\n\nUK\n\n**80%**\n\n**10%**\n\n**11%**\n\n**10%**\n\n**11%**\n\nUS\n\n**79%**\n\nJapan\n\n**72%**\n\nGermany India China Southeast Asia Rep. of Korea\n\n**2%**\n\n**13%**\n\n**14%**\n\n**70%**\n\n**18%**\n\n**12%**\n\n**63%**\n\n**22%**\n\n**15%**\n\n**62%**\n\n**28%**\n\n**10%**\n\n**61%**\n\n**21%**\n\n**17%**\n\n**71%**\n\n**20%**\n\n**9%**\n\nYes, dedicated Yes, shared with a neighbor/others No Don’t know\n\nChina\n\n**77%**\n\n**15%**\n\n**7%**\n\nIndia\n\n**41%**\n\n**44%**\n\nUS\n\n**40%**\n\nGermany UK Southeast Asia Japan\n\n**14%**\n\n**58%**\n\n**2%**\n\n**33%**\n\n**62%**\n\n**5%**\n\n**32%**\n\n**58%**\n\n**8%**\n\n**28%**\n\n**65%**\n\n**4%**\n\nRep. of Korea\n\n**19%**\n\n**7%**\n\n**25%**\n\n**37%**\n\n**56%**\n\n**6%**\n\n**1% 1% 1% 2% 2%**\n\n**50%**\n\n**The push to create public charging capacity may be somewhat overstated in many markets as a** **strong majority of EV intenders surveyed plan to charge their vehicle at home.**\n\n**Expecting to charge electrified vehicle most often at…**\n\n**Yet, many EV intenders surveyed in mature markets that plan to charge at home do not currently have access to a charger.**\n\n**EV charger access among consumers who plan to charge their vehicle at home**\n\nNote: Percentages may not add up to 100 due to rounding. Q43. Where do you expect to charge your EV most often? Sample size: n = 413 [China]; 304 [Germany]; 182 [India]; 64 [Japan]; 166 [Republic of Korea]; 1,223 [Southeast Asia]; 245 [UK]; 111 [US]\n\nNote: Percentages may not add up to 100 due to rounding. Q44. Do you already have access to a charger at your residence? Sample size: n = 260 [China]; 215 [Germany]; 128 [India]; 46 [Japan]; 102 [Republic of Korea]; 755 [Southeast Asia]; 195 [UK]; 88 [US]","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/deloitte-2025-global-automotive-consumer-study-january-2025_1_p8.pdf"]},"response":{"errors":[],"extract_id":"extract_e1e413e1ba2274fc4546ccf45104816d","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["8\n\n**2025 Global Automotive Consumer Study** | Key Findings: Global Focus Markets\n\nHome Work On-street/public charging station Other\n\nUK\n\n**80%**\n\n**10%**\n\n**11%**\n\n**10%**\n\n**11%**\n\nUS\n\n**79%**\n\nJapan\n\n**72%**\n\nGermany India China Southeast Asia Rep. of Korea\n\n**2%**\n\n**13%**\n\n**14%**\n\n**70%**\n\n**18%**\n\n**12%**\n\n**63%**\n\n**22%**\n\n**15%**\n\n**62%**\n\n**28%**\n\n**10%**\n\n**61%**\n\n**21%**\n\n**17%**\n\n**71%**\n\n**20%**\n\n**9%**\n\nYes, dedicated Yes, shared with a neighbor/others No Don’t know\n\nChina\n\n**77%**\n\n**15%**\n\n**7%**\n\nIndia\n\n**41%**\n\n**44%**\n\nUS\n\n**40%**\n\nGermany UK Southeast Asia Japan\n\n**14%**\n\n**58%**\n\n**2%**\n\n**33%**\n\n**62%**\n\n**5%**\n\n**32%**\n\n**58%**\n\n**8%**\n\n**28%**\n\n**65%**\n\n**4%**\n\nRep. of Korea\n\n**19%**\n\n**7%**\n\n**25%**\n\n**37%**\n\n**56%**\n\n**6%**\n\n**1% 1% 1% 2% 2%**\n\n**50%**\n\n**The push to create public charging capacity may be somewhat overstated in many markets as a** **strong majority of EV intenders surveyed plan to charge their vehicle at home.**\n\n**Expecting to charge electrified vehicle most often at…**\n\n**Yet, many EV intenders surveyed in mature markets that plan to charge at home do not currently have access to a charger.**\n\n**EV charger access among consumers who plan to charge their vehicle at home**\n\nNote: Percentages may not add up to 100 due to rounding. Q43. Where do you expect to charge your EV most often? Sample size: n = 413 [China]; 304 [Germany]; 182 [India]; 64 [Japan]; 166 [Republic of Korea]; 1,223 [Southeast Asia]; 245 [UK]; 111 [US]\n\nNote: Percentages may not add up to 100 due to rounding. Q44. Do you already have access to a charger at your residence? Sample size: n = 260 [China]; 215 [Germany]; 128 [India]; 46 [Japan]; 102 [Republic of Korea]; 755 [Southeast Asia]; 195 [UK]; 88 [US]"],"full_content":"8\n\n**2025 Global Automotive Consumer Study** | Key Findings: Global Focus Markets\n\nHome Work On-street/public charging station Other\n\nUK\n\n**80%**\n\n**10%**\n\n**11%**\n\n**10%**\n\n**11%**\n\nUS\n\n**79%**\n\nJapan\n\n**72%**\n\nGermany India China Southeast Asia Rep. of Korea\n\n**2%**\n\n**13%**\n\n**14%**\n\n**70%**\n\n**18%**\n\n**12%**\n\n**63%**\n\n**22%**\n\n**15%**\n\n**62%**\n\n**28%**\n\n**10%**\n\n**61%**\n\n**21%**\n\n**17%**\n\n**71%**\n\n**20%**\n\n**9%**\n\nYes, dedicated Yes, shared with a neighbor/others No Don’t know\n\nChina\n\n**77%**\n\n**15%**\n\n**7%**\n\nIndia\n\n**41%**\n\n**44%**\n\nUS\n\n**40%**\n\nGermany UK Southeast Asia Japan\n\n**14%**\n\n**58%**\n\n**2%**\n\n**33%**\n\n**62%**\n\n**5%**\n\n**32%**\n\n**58%**\n\n**8%**\n\n**28%**\n\n**65%**\n\n**4%**\n\nRep. of Korea\n\n**19%**\n\n**7%**\n\n**25%**\n\n**37%**\n\n**56%**\n\n**6%**\n\n**1% 1% 1% 2% 2%**\n\n**50%**\n\n**The push to create public charging capacity may be somewhat overstated in many markets as a** **strong majority of EV intenders surveyed plan to charge their vehicle at home.**\n\n**Expecting to charge electrified vehicle most often at…**\n\n**Yet, many EV intenders surveyed in mature markets that plan to charge at home do not currently have access to a charger.**\n\n**EV charger access among consumers who plan to charge their vehicle at home**\n\nNote: Percentages may not add up to 100 due to rounding. Q43. Where do you expect to charge your EV most often? Sample size: n = 413 [China]; 304 [Germany]; 182 [India]; 64 [Japan]; 166 [Republic of Korea]; 1,223 [Southeast Asia]; 245 [UK]; 111 [US]\n\nNote: Percentages may not add up to 100 due to rounding. Q44. Do you already have access to a charger at your residence? Sample size: n = 260 [China]; 215 [Germany]; 128 [India]; 46 [Japan]; 102 [Republic of Korea]; 755 [Southeast Asia]; 195 [UK]; 88 [US]","publish_date":null,"title":"The push to create public charging capacity may be somewhat overstated in many markets as a strong majority of EV","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/deloitte-2025-global-automotive-consumer-study-january-2025_1_p8.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/deloitte-2025-global-automotive-consumer-study-january-2025_1_p8.pdf"},"request":{"example_id":"chart/deloitte-2025-global-automotive-consumer-study-january-2025_1_p8","product_type":"parse","source_file_path":"deloitte-2025-global-automotive-consumer-study-january-2025_1_p8.pdf"},"started_at":"2026-11-04T19:44:58.609637"} +{"completed_at":"2026-11-04T19:45:03.653946","latency_in_ms":4072,"output":{"example_id":"chart/deloitte-2025-global-automotive-consumer-study-january-2025_p15","grounded_pages":[],"layout_pages":[],"markdown":"**2025 Global Automotive Consumer Study** | Key Findings: Global Focus Markets\n\n**More than half of surveyed consumers in China owned the same brand of vehicle prior to their current car,** **but nearly a third of survey respondents say they are first-time owners, signaling a need to build strong** **customer relationships to help ensure forward brand loyalty.**\n\n**Percentage of consumers whose prior vehicle was from the same brand as current vehicle**\n\nQ6. Was your prior vehicle from the same brand? Sample size: n = 852 [China]; 1,114 [Germany]; 646 [India]; 452 [Japan]; 618 [Republic of Korea]; 3,488 [Southeast Asia]; 1,044 [UK]; 821 [US]\n\n**However, three quarters of surveyed consumers in China say they intend to switch to another brand of vehicle the next time they are in-market, signaling the intensity of the competitive landscape in China. Conversely, consumers surveyed in Japan are generally more brand loyal, but the intended defection rate is rising across many global markets on a year-over-year basis.**\n\n**Percentage of consumers intending to switch to another brand\\* of vehicle**\n\n\\*Includes switching to a different brand from the same parent or a different brand from a different sales parent. Q5. What brand is the vehicle you drive most often? Q26. What brand are you considering most for your next vehicle? [Brand switching percentage is based on a calculation involving these two questions.] Sample size: n = 830 [China]; 1,073 [Germany]; 633 [India]; 398 [Japan]; 589 [Republic of Korea]; 3,807 [Southeast Asia]; 959 [UK]; 786 [US]\n\n**55% 55% 54% 51% 47% 45% 43% 42%**\n\n**15%**\n\n**36% 31% 45% 48%**\n\n**31% 48% 51%**\n\n**31%**\n\n**9% 14%**\n\n**4% 5%**\n\n**25%**\n\n**9% 7%**\n\nChina Japan India Germany US Southeast Asia Rep. of Korea UK\n\nYes No Current car is my first vehicle\n\n**76% 72% 71%**\n\n**63%**\n\n**56% 54%**\n\n**46%**\n\n**39%**\n\n**73% 78%**\n\n**62% 57% 54% 51%**\n\n**41% 35%**\n\nChina India Southeast Asia Rep. of Korea UK US Germany Japan\n\n2025\n\n2024\n\n15","pages":[{"line_numbers":[],"markdown":"**2025 Global Automotive Consumer Study** | Key Findings: Global Focus Markets\n\n**More than half of surveyed consumers in China owned the same brand of vehicle prior to their current car,** **but nearly a third of survey respondents say they are first-time owners, signaling a need to build strong** **customer relationships to help ensure forward brand loyalty.**\n\n**Percentage of consumers whose prior vehicle was from the same brand as current vehicle**\n\nQ6. Was your prior vehicle from the same brand? Sample size: n = 852 [China]; 1,114 [Germany]; 646 [India]; 452 [Japan]; 618 [Republic of Korea]; 3,488 [Southeast Asia]; 1,044 [UK]; 821 [US]\n\n**However, three quarters of surveyed consumers in China say they intend to switch to another brand of vehicle the next time they are in-market, signaling the intensity of the competitive landscape in China. Conversely, consumers surveyed in Japan are generally more brand loyal, but the intended defection rate is rising across many global markets on a year-over-year basis.**\n\n**Percentage of consumers intending to switch to another brand\\* of vehicle**\n\n\\*Includes switching to a different brand from the same parent or a different brand from a different sales parent. Q5. What brand is the vehicle you drive most often? Q26. What brand are you considering most for your next vehicle? [Brand switching percentage is based on a calculation involving these two questions.] Sample size: n = 830 [China]; 1,073 [Germany]; 633 [India]; 398 [Japan]; 589 [Republic of Korea]; 3,807 [Southeast Asia]; 959 [UK]; 786 [US]\n\n**55% 55% 54% 51% 47% 45% 43% 42%**\n\n**15%**\n\n**36% 31% 45% 48%**\n\n**31% 48% 51%**\n\n**31%**\n\n**9% 14%**\n\n**4% 5%**\n\n**25%**\n\n**9% 7%**\n\nChina Japan India Germany US Southeast Asia Rep. of Korea UK\n\nYes No Current car is my first vehicle\n\n**76% 72% 71%**\n\n**63%**\n\n**56% 54%**\n\n**46%**\n\n**39%**\n\n**73% 78%**\n\n**62% 57% 54% 51%**\n\n**41% 35%**\n\nChina India Southeast Asia Rep. of Korea UK US Germany Japan\n\n2025\n\n2024\n\n15","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**2025 Global Automotive Consumer Study** | Key Findings: Global Focus Markets\n\n**More than half of surveyed consumers in China owned the same brand of vehicle prior to their current car,** **but nearly a third of survey respondents say they are first-time owners, signaling a need to build strong** **customer relationships to help ensure forward brand loyalty.**\n\n**Percentage of consumers whose prior vehicle was from the same brand as current vehicle**\n\nQ6. Was your prior vehicle from the same brand? Sample size: n = 852 [China]; 1,114 [Germany]; 646 [India]; 452 [Japan]; 618 [Republic of Korea]; 3,488 [Southeast Asia]; 1,044 [UK]; 821 [US]\n\n**However, three quarters of surveyed consumers in China say they intend to switch to another brand of vehicle the next time they are in-market, signaling the intensity of the competitive landscape in China. Conversely, consumers surveyed in Japan are generally more brand loyal, but the intended defection rate is rising across many global markets on a year-over-year basis.**\n\n**Percentage of consumers intending to switch to another brand\\* of vehicle**\n\n\\*Includes switching to a different brand from the same parent or a different brand from a different sales parent. Q5. What brand is the vehicle you drive most often? Q26. What brand are you considering most for your next vehicle? [Brand switching percentage is based on a calculation involving these two questions.] Sample size: n = 830 [China]; 1,073 [Germany]; 633 [India]; 398 [Japan]; 589 [Republic of Korea]; 3,807 [Southeast Asia]; 959 [UK]; 786 [US]\n\n**55% 55% 54% 51% 47% 45% 43% 42%**\n\n**15%**\n\n**36% 31% 45% 48%**\n\n**31% 48% 51%**\n\n**31%**\n\n**9% 14%**\n\n**4% 5%**\n\n**25%**\n\n**9% 7%**\n\nChina Japan India Germany US Southeast Asia Rep. of Korea UK\n\nYes No Current car is my first vehicle\n\n**76% 72% 71%**\n\n**63%**\n\n**56% 54%**\n\n**46%**\n\n**39%**\n\n**73% 78%**\n\n**62% 57% 54% 51%**\n\n**41% 35%**\n\nChina India Southeast Asia Rep. of Korea UK US Germany Japan\n\n2025\n\n2024\n\n15","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/deloitte-2025-global-automotive-consumer-study-january-2025_p15.pdf"]},"response":{"errors":[],"extract_id":"extract_16679aeed4b61d863806b019f8e2777c","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**2025 Global Automotive Consumer Study** | Key Findings: Global Focus Markets\n\n**More than half of surveyed consumers in China owned the same brand of vehicle prior to their current car,** **but nearly a third of survey respondents say they are first-time owners, signaling a need to build strong** **customer relationships to help ensure forward brand loyalty.**\n\n**Percentage of consumers whose prior vehicle was from the same brand as current vehicle**\n\nQ6. Was your prior vehicle from the same brand? Sample size: n = 852 [China]; 1,114 [Germany]; 646 [India]; 452 [Japan]; 618 [Republic of Korea]; 3,488 [Southeast Asia]; 1,044 [UK]; 821 [US]\n\n**However, three quarters of surveyed consumers in China say they intend to switch to another brand of vehicle the next time they are in-market, signaling the intensity of the competitive landscape in China. Conversely, consumers surveyed in Japan are generally more brand loyal, but the intended defection rate is rising across many global markets on a year-over-year basis.**\n\n**Percentage of consumers intending to switch to another brand\\* of vehicle**\n\n\\*Includes switching to a different brand from the same parent or a different brand from a different sales parent. Q5. What brand is the vehicle you drive most often? Q26. What brand are you considering most for your next vehicle? [Brand switching percentage is based on a calculation involving these two questions.] Sample size: n = 830 [China]; 1,073 [Germany]; 633 [India]; 398 [Japan]; 589 [Republic of Korea]; 3,807 [Southeast Asia]; 959 [UK]; 786 [US]\n\n**55% 55% 54% 51% 47% 45% 43% 42%**\n\n**15%**\n\n**36% 31% 45% 48%**\n\n**31% 48% 51%**\n\n**31%**\n\n**9% 14%**\n\n**4% 5%**\n\n**25%**\n\n**9% 7%**\n\nChina Japan India Germany US Southeast Asia Rep. of Korea UK\n\nYes No Current car is my first vehicle\n\n**76% 72% 71%**\n\n**63%**\n\n**56% 54%**\n\n**46%**\n\n**39%**\n\n**73% 78%**\n\n**62% 57% 54% 51%**\n\n**41% 35%**\n\nChina India Southeast Asia Rep. of Korea UK US Germany Japan\n\n2025\n\n2024\n\n15"],"full_content":"**2025 Global Automotive Consumer Study** | Key Findings: Global Focus Markets\n\n**More than half of surveyed consumers in China owned the same brand of vehicle prior to their current car,** **but nearly a third of survey respondents say they are first-time owners, signaling a need to build strong** **customer relationships to help ensure forward brand loyalty.**\n\n**Percentage of consumers whose prior vehicle was from the same brand as current vehicle**\n\nQ6. Was your prior vehicle from the same brand? Sample size: n = 852 [China]; 1,114 [Germany]; 646 [India]; 452 [Japan]; 618 [Republic of Korea]; 3,488 [Southeast Asia]; 1,044 [UK]; 821 [US]\n\n**However, three quarters of surveyed consumers in China say they intend to switch to another brand of vehicle the next time they are in-market, signaling the intensity of the competitive landscape in China. Conversely, consumers surveyed in Japan are generally more brand loyal, but the intended defection rate is rising across many global markets on a year-over-year basis.**\n\n**Percentage of consumers intending to switch to another brand\\* of vehicle**\n\n\\*Includes switching to a different brand from the same parent or a different brand from a different sales parent. Q5. What brand is the vehicle you drive most often? Q26. What brand are you considering most for your next vehicle? [Brand switching percentage is based on a calculation involving these two questions.] Sample size: n = 830 [China]; 1,073 [Germany]; 633 [India]; 398 [Japan]; 589 [Republic of Korea]; 3,807 [Southeast Asia]; 959 [UK]; 786 [US]\n\n**55% 55% 54% 51% 47% 45% 43% 42%**\n\n**15%**\n\n**36% 31% 45% 48%**\n\n**31% 48% 51%**\n\n**31%**\n\n**9% 14%**\n\n**4% 5%**\n\n**25%**\n\n**9% 7%**\n\nChina Japan India Germany US Southeast Asia Rep. of Korea UK\n\nYes No Current car is my first vehicle\n\n**76% 72% 71%**\n\n**63%**\n\n**56% 54%**\n\n**46%**\n\n**39%**\n\n**73% 78%**\n\n**62% 57% 54% 51%**\n\n**41% 35%**\n\nChina India Southeast Asia Rep. of Korea UK US Germany Japan\n\n2025\n\n2024\n\n15","publish_date":null,"title":"More than half of surveyed consumers in China owned the same brand of vehicle prior to their current car, but","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/deloitte-2025-global-automotive-consumer-study-january-2025_p15.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/deloitte-2025-global-automotive-consumer-study-january-2025_p15.pdf"},"request":{"example_id":"chart/deloitte-2025-global-automotive-consumer-study-january-2025_p15","product_type":"parse","source_file_path":"deloitte-2025-global-automotive-consumer-study-january-2025_p15.pdf"},"started_at":"2026-11-04T19:44:59.581533"} +{"completed_at":"2026-11-04T19:45:03.461654","latency_in_ms":2282,"output":{"example_id":"chart/deloitte-2025-global-automotive-consumer-study-january-2025_p17","grounded_pages":[],"layout_pages":[],"markdown":"17\n\n**2025 Global Automotive Consumer Study** | Key Findings: Global Focus Markets\n\n**As global trade policies shift, encouraging a more regional approach to vehicle manufacturing footprints** **and supply chains, consumer sentiment differs among respondents when it comes to whether local manu** **\\- facturing is an important criteria for choosing a vehicle.**\n\n**Importance of next vehicle to be manufactured locally**\n\nNote: Percentages may not add to 100 due to rounding. Q40. How would you most prefer to pay for public EV charging? Sample size: n= 371 [China]; 295 [Germany]; 194 [India]; 103 [Japan]; 213 [Republic of Korea]; 1,163 [Southeast Asia]; 112 [US]\n\nMost preferred mode of payment\n\n**As some original equipment manufacturers (OEMs) continue to explore the potential for ‘agency’ retail models, surveyed consumers in India and China are more receptive to the concept of buying a vehicle directly from the manufacturer compared to consumers in Japan. Also, surveyed consumers in some mar- kets are signaling a high level of interest in purchasing vehicle insurance directly from the manufacturer.**\n\nQ61. To what extent are you interested in acquiring your next vehicle directly from the manufacturer (via an online process)? Q60. The next time you acquire a vehicle, how interested would you be in purchasing insurance directly from the vehicle manufacturer? Sample size: n = 939 [China]; 1,306 [Germany]; 882 [India]; 637 [Japan]; 906 [Republic of Korea]; 5.028 [Southeast Asia]; 1,314 [UK]; 937 [US]\n\n**49%**\n\n**36%**\n\n**15%**\n\n**10%**\n\n**28%**\n\n**19%**\n\n**13%**\n\n**10%**\n\n**29%**\n\n**31%**\n\n**45%**\n\n**34%**\n\n**31%**\n\n**27%**\n\n**22%**\n\n**19%**\n\n**14%**\n\n**22%**\n\n**24%**\n\n**39%**\n\n**24%**\n\n**33%**\n\n**39%**\n\n**31%**\n\n**8%**\n\n**11%**\n\n**16%**\n\n**17%**\n\n**17%**\n\n**21%**\n\n**26%**\n\n**41%**\n\nUK\n\nGermany\n\nJapan\n\nRep. of Korea\n\nUS\n\nSoutheast Asia\n\nChina\n\nIndia\n\nNot important Somewhat important Important Very important\n\n**34%**\n\n**38%**\n\n**40%**\n\n**46%**\n\n**46%**\n\n**59%**\n\n**72%**\n\n**76%**\n\nJapan\n\nGermany\n\nUK\n\nUS\n\nRep. of Korea\n\nSoutheast Asia\n\nChina\n\nIndia\n\nPercentage of surveyed consumers who would be interested in acquiring **vehicle directly from the manufacturer** Percentage of surveyed consumers who would be interested in purchasing **insurance directly from the manufacturer**\n\n**32%**\n\n**37%**\n\n**41%**\n\n**42%**\n\n**55%**\n\n**71%**\n\n**76%**\n\n**82%**\n\nGermany\n\nJapan\n\nUS\n\nUK\n\nRep. of Korea\n\nSoutheast Asia\n\nChina\n\nIndia\n\nNote: Percentages may not add up to 100 due to rounding. Q31. To what extent is it important that your next vehicle be locally manufactured (i.e., manufactured in your country or region)? Sample size: n = 939 [China]; 1,306 [Germany]; 882 [India]; 637 [Japan]; 906 [Republic of Korea]; 5,028 [Southeast Asia]; 1,315 [UK]; 937 [US]","pages":[{"line_numbers":[],"markdown":"17\n\n**2025 Global Automotive Consumer Study** | Key Findings: Global Focus Markets\n\n**As global trade policies shift, encouraging a more regional approach to vehicle manufacturing footprints** **and supply chains, consumer sentiment differs among respondents when it comes to whether local manu** **\\- facturing is an important criteria for choosing a vehicle.**\n\n**Importance of next vehicle to be manufactured locally**\n\nNote: Percentages may not add to 100 due to rounding. Q40. How would you most prefer to pay for public EV charging? Sample size: n= 371 [China]; 295 [Germany]; 194 [India]; 103 [Japan]; 213 [Republic of Korea]; 1,163 [Southeast Asia]; 112 [US]\n\nMost preferred mode of payment\n\n**As some original equipment manufacturers (OEMs) continue to explore the potential for ‘agency’ retail models, surveyed consumers in India and China are more receptive to the concept of buying a vehicle directly from the manufacturer compared to consumers in Japan. Also, surveyed consumers in some mar- kets are signaling a high level of interest in purchasing vehicle insurance directly from the manufacturer.**\n\nQ61. To what extent are you interested in acquiring your next vehicle directly from the manufacturer (via an online process)? Q60. The next time you acquire a vehicle, how interested would you be in purchasing insurance directly from the vehicle manufacturer? Sample size: n = 939 [China]; 1,306 [Germany]; 882 [India]; 637 [Japan]; 906 [Republic of Korea]; 5.028 [Southeast Asia]; 1,314 [UK]; 937 [US]\n\n**49%**\n\n**36%**\n\n**15%**\n\n**10%**\n\n**28%**\n\n**19%**\n\n**13%**\n\n**10%**\n\n**29%**\n\n**31%**\n\n**45%**\n\n**34%**\n\n**31%**\n\n**27%**\n\n**22%**\n\n**19%**\n\n**14%**\n\n**22%**\n\n**24%**\n\n**39%**\n\n**24%**\n\n**33%**\n\n**39%**\n\n**31%**\n\n**8%**\n\n**11%**\n\n**16%**\n\n**17%**\n\n**17%**\n\n**21%**\n\n**26%**\n\n**41%**\n\nUK\n\nGermany\n\nJapan\n\nRep. of Korea\n\nUS\n\nSoutheast Asia\n\nChina\n\nIndia\n\nNot important Somewhat important Important Very important\n\n**34%**\n\n**38%**\n\n**40%**\n\n**46%**\n\n**46%**\n\n**59%**\n\n**72%**\n\n**76%**\n\nJapan\n\nGermany\n\nUK\n\nUS\n\nRep. of Korea\n\nSoutheast Asia\n\nChina\n\nIndia\n\nPercentage of surveyed consumers who would be interested in acquiring **vehicle directly from the manufacturer** Percentage of surveyed consumers who would be interested in purchasing **insurance directly from the manufacturer**\n\n**32%**\n\n**37%**\n\n**41%**\n\n**42%**\n\n**55%**\n\n**71%**\n\n**76%**\n\n**82%**\n\nGermany\n\nJapan\n\nUS\n\nUK\n\nRep. of Korea\n\nSoutheast Asia\n\nChina\n\nIndia\n\nNote: Percentages may not add up to 100 due to rounding. Q31. To what extent is it important that your next vehicle be locally manufactured (i.e., manufactured in your country or region)? Sample size: n = 939 [China]; 1,306 [Germany]; 882 [India]; 637 [Japan]; 906 [Republic of Korea]; 5,028 [Southeast Asia]; 1,315 [UK]; 937 [US]","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"17\n\n**2025 Global Automotive Consumer Study** | Key Findings: Global Focus Markets\n\n**As global trade policies shift, encouraging a more regional approach to vehicle manufacturing footprints** **and supply chains, consumer sentiment differs among respondents when it comes to whether local manu** **\\- facturing is an important criteria for choosing a vehicle.**\n\n**Importance of next vehicle to be manufactured locally**\n\nNote: Percentages may not add to 100 due to rounding. Q40. How would you most prefer to pay for public EV charging? Sample size: n= 371 [China]; 295 [Germany]; 194 [India]; 103 [Japan]; 213 [Republic of Korea]; 1,163 [Southeast Asia]; 112 [US]\n\nMost preferred mode of payment\n\n**As some original equipment manufacturers (OEMs) continue to explore the potential for ‘agency’ retail models, surveyed consumers in India and China are more receptive to the concept of buying a vehicle directly from the manufacturer compared to consumers in Japan. Also, surveyed consumers in some mar- kets are signaling a high level of interest in purchasing vehicle insurance directly from the manufacturer.**\n\nQ61. To what extent are you interested in acquiring your next vehicle directly from the manufacturer (via an online process)? Q60. The next time you acquire a vehicle, how interested would you be in purchasing insurance directly from the vehicle manufacturer? Sample size: n = 939 [China]; 1,306 [Germany]; 882 [India]; 637 [Japan]; 906 [Republic of Korea]; 5.028 [Southeast Asia]; 1,314 [UK]; 937 [US]\n\n**49%**\n\n**36%**\n\n**15%**\n\n**10%**\n\n**28%**\n\n**19%**\n\n**13%**\n\n**10%**\n\n**29%**\n\n**31%**\n\n**45%**\n\n**34%**\n\n**31%**\n\n**27%**\n\n**22%**\n\n**19%**\n\n**14%**\n\n**22%**\n\n**24%**\n\n**39%**\n\n**24%**\n\n**33%**\n\n**39%**\n\n**31%**\n\n**8%**\n\n**11%**\n\n**16%**\n\n**17%**\n\n**17%**\n\n**21%**\n\n**26%**\n\n**41%**\n\nUK\n\nGermany\n\nJapan\n\nRep. of Korea\n\nUS\n\nSoutheast Asia\n\nChina\n\nIndia\n\nNot important Somewhat important Important Very important\n\n**34%**\n\n**38%**\n\n**40%**\n\n**46%**\n\n**46%**\n\n**59%**\n\n**72%**\n\n**76%**\n\nJapan\n\nGermany\n\nUK\n\nUS\n\nRep. of Korea\n\nSoutheast Asia\n\nChina\n\nIndia\n\nPercentage of surveyed consumers who would be interested in acquiring **vehicle directly from the manufacturer** Percentage of surveyed consumers who would be interested in purchasing **insurance directly from the manufacturer**\n\n**32%**\n\n**37%**\n\n**41%**\n\n**42%**\n\n**55%**\n\n**71%**\n\n**76%**\n\n**82%**\n\nGermany\n\nJapan\n\nUS\n\nUK\n\nRep. of Korea\n\nSoutheast Asia\n\nChina\n\nIndia\n\nNote: Percentages may not add up to 100 due to rounding. Q31. To what extent is it important that your next vehicle be locally manufactured (i.e., manufactured in your country or region)? Sample size: n = 939 [China]; 1,306 [Germany]; 882 [India]; 637 [Japan]; 906 [Republic of Korea]; 5,028 [Southeast Asia]; 1,315 [UK]; 937 [US]","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/deloitte-2025-global-automotive-consumer-study-january-2025_p17.pdf"]},"response":{"errors":[],"extract_id":"extract_a618a36a2698924df92c66c578913611","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["17\n\n**2025 Global Automotive Consumer Study** | Key Findings: Global Focus Markets\n\n**As global trade policies shift, encouraging a more regional approach to vehicle manufacturing footprints** **and supply chains, consumer sentiment differs among respondents when it comes to whether local manu** **\\- facturing is an important criteria for choosing a vehicle.**\n\n**Importance of next vehicle to be manufactured locally**\n\nNote: Percentages may not add to 100 due to rounding. Q40. How would you most prefer to pay for public EV charging? Sample size: n= 371 [China]; 295 [Germany]; 194 [India]; 103 [Japan]; 213 [Republic of Korea]; 1,163 [Southeast Asia]; 112 [US]\n\nMost preferred mode of payment\n\n**As some original equipment manufacturers (OEMs) continue to explore the potential for ‘agency’ retail models, surveyed consumers in India and China are more receptive to the concept of buying a vehicle directly from the manufacturer compared to consumers in Japan. Also, surveyed consumers in some mar- kets are signaling a high level of interest in purchasing vehicle insurance directly from the manufacturer.**\n\nQ61. To what extent are you interested in acquiring your next vehicle directly from the manufacturer (via an online process)? Q60. The next time you acquire a vehicle, how interested would you be in purchasing insurance directly from the vehicle manufacturer? Sample size: n = 939 [China]; 1,306 [Germany]; 882 [India]; 637 [Japan]; 906 [Republic of Korea]; 5.028 [Southeast Asia]; 1,314 [UK]; 937 [US]\n\n**49%**\n\n**36%**\n\n**15%**\n\n**10%**\n\n**28%**\n\n**19%**\n\n**13%**\n\n**10%**\n\n**29%**\n\n**31%**\n\n**45%**\n\n**34%**\n\n**31%**\n\n**27%**\n\n**22%**\n\n**19%**\n\n**14%**\n\n**22%**\n\n**24%**\n\n**39%**\n\n**24%**\n\n**33%**\n\n**39%**\n\n**31%**\n\n**8%**\n\n**11%**\n\n**16%**\n\n**17%**\n\n**17%**\n\n**21%**\n\n**26%**\n\n**41%**\n\nUK\n\nGermany\n\nJapan\n\nRep. of Korea\n\nUS\n\nSoutheast Asia\n\nChina\n\nIndia\n\nNot important Somewhat important Important Very important\n\n**34%**\n\n**38%**\n\n**40%**\n\n**46%**\n\n**46%**\n\n**59%**\n\n**72%**\n\n**76%**\n\nJapan\n\nGermany\n\nUK\n\nUS\n\nRep. of Korea\n\nSoutheast Asia\n\nChina\n\nIndia\n\nPercentage of surveyed consumers who would be interested in acquiring **vehicle directly from the manufacturer** Percentage of surveyed consumers who would be interested in purchasing **insurance directly from the manufacturer**\n\n**32%**\n\n**37%**\n\n**41%**\n\n**42%**\n\n**55%**\n\n**71%**\n\n**76%**\n\n**82%**\n\nGermany\n\nJapan\n\nUS\n\nUK\n\nRep. of Korea\n\nSoutheast Asia\n\nChina\n\nIndia\n\nNote: Percentages may not add up to 100 due to rounding. Q31. To what extent is it important that your next vehicle be locally manufactured (i.e., manufactured in your country or region)? Sample size: n = 939 [China]; 1,306 [Germany]; 882 [India]; 637 [Japan]; 906 [Republic of Korea]; 5,028 [Southeast Asia]; 1,315 [UK]; 937 [US]"],"full_content":"17\n\n**2025 Global Automotive Consumer Study** | Key Findings: Global Focus Markets\n\n**As global trade policies shift, encouraging a more regional approach to vehicle manufacturing footprints** **and supply chains, consumer sentiment differs among respondents when it comes to whether local manu** **\\- facturing is an important criteria for choosing a vehicle.**\n\n**Importance of next vehicle to be manufactured locally**\n\nNote: Percentages may not add to 100 due to rounding. Q40. How would you most prefer to pay for public EV charging? Sample size: n= 371 [China]; 295 [Germany]; 194 [India]; 103 [Japan]; 213 [Republic of Korea]; 1,163 [Southeast Asia]; 112 [US]\n\nMost preferred mode of payment\n\n**As some original equipment manufacturers (OEMs) continue to explore the potential for ‘agency’ retail models, surveyed consumers in India and China are more receptive to the concept of buying a vehicle directly from the manufacturer compared to consumers in Japan. Also, surveyed consumers in some mar- kets are signaling a high level of interest in purchasing vehicle insurance directly from the manufacturer.**\n\nQ61. To what extent are you interested in acquiring your next vehicle directly from the manufacturer (via an online process)? Q60. The next time you acquire a vehicle, how interested would you be in purchasing insurance directly from the vehicle manufacturer? Sample size: n = 939 [China]; 1,306 [Germany]; 882 [India]; 637 [Japan]; 906 [Republic of Korea]; 5.028 [Southeast Asia]; 1,314 [UK]; 937 [US]\n\n**49%**\n\n**36%**\n\n**15%**\n\n**10%**\n\n**28%**\n\n**19%**\n\n**13%**\n\n**10%**\n\n**29%**\n\n**31%**\n\n**45%**\n\n**34%**\n\n**31%**\n\n**27%**\n\n**22%**\n\n**19%**\n\n**14%**\n\n**22%**\n\n**24%**\n\n**39%**\n\n**24%**\n\n**33%**\n\n**39%**\n\n**31%**\n\n**8%**\n\n**11%**\n\n**16%**\n\n**17%**\n\n**17%**\n\n**21%**\n\n**26%**\n\n**41%**\n\nUK\n\nGermany\n\nJapan\n\nRep. of Korea\n\nUS\n\nSoutheast Asia\n\nChina\n\nIndia\n\nNot important Somewhat important Important Very important\n\n**34%**\n\n**38%**\n\n**40%**\n\n**46%**\n\n**46%**\n\n**59%**\n\n**72%**\n\n**76%**\n\nJapan\n\nGermany\n\nUK\n\nUS\n\nRep. of Korea\n\nSoutheast Asia\n\nChina\n\nIndia\n\nPercentage of surveyed consumers who would be interested in acquiring **vehicle directly from the manufacturer** Percentage of surveyed consumers who would be interested in purchasing **insurance directly from the manufacturer**\n\n**32%**\n\n**37%**\n\n**41%**\n\n**42%**\n\n**55%**\n\n**71%**\n\n**76%**\n\n**82%**\n\nGermany\n\nJapan\n\nUS\n\nUK\n\nRep. of Korea\n\nSoutheast Asia\n\nChina\n\nIndia\n\nNote: Percentages may not add up to 100 due to rounding. Q31. To what extent is it important that your next vehicle be locally manufactured (i.e., manufactured in your country or region)? Sample size: n = 939 [China]; 1,306 [Germany]; 882 [India]; 637 [Japan]; 906 [Republic of Korea]; 5,028 [Southeast Asia]; 1,315 [UK]; 937 [US]","publish_date":null,"title":"As global trade policies shift, encouraging a more regional approach to vehicle manufacturing footprints and supply chains, consumer sentiment differs","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/deloitte-2025-global-automotive-consumer-study-january-2025_p17.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/deloitte-2025-global-automotive-consumer-study-january-2025_p17.pdf"},"request":{"example_id":"chart/deloitte-2025-global-automotive-consumer-study-january-2025_p17","product_type":"parse","source_file_path":"deloitte-2025-global-automotive-consumer-study-january-2025_p17.pdf"},"started_at":"2026-11-04T19:45:01.178972"} +{"completed_at":"2026-11-04T19:45:03.319276","latency_in_ms":1100,"output":{"example_id":"chart/deloitte-2025-global-automotive-consumer-study-january-2025_p24","grounded_pages":[],"layout_pages":[],"markdown":"24\n\n**2025 Global Automotive Consumer Study** | Key Findings: Global Focus Markets\n\n**23%**\n\n**25%**\n\n**37%**\n\n**38%**\n\n**40%**\n\n**49%**\n\n**49%**\n\n**51%**\n\n**18%**\n\n**23%**\n\n**26%**\n\n**29%**\n\n**36%**\n\n**27%**\n\n**23%**\n\n**25%**\n\n**18%**\n\n**21%**\n\n**17%**\n\n**13%**\n\n**11%**\n\n**11%**\n\n**11%**\n\n**11%**\n\n**6%**\n\n**10%**\n\n**8%**\n\n**5%**\n\n**6%**\n\n**5%**\n\n**5%**\n\n**6%**\n\n**2%**\n\n**4%**\n\n**3%**\n\n**2%**\n\n**2%**\n\n**3%**\n\n**3%**\n\n**3%**\n\n**32%**\n\n**18%**\n\n**9%**\n\n**12%**\n\n**4%**\n\n**6%**\n\n**8%**\n\n**5%**\n\nJapan\n\nSouth Korea\n\nGermany\n\nUK\n\nChina\n\nUS\n\nSoutheast Asia\n\nIndia\n\nEvery day 3-4 times per week 1-2 times per week 2-3 times per month Once a month Once every few months\n\n**Half of surveyed consumers in India, Southeast Asia, and the United States say they drive their vehi- cle every day—roughly twice the number of consumers in Japan and South Korea who say the same,** **perhaps reflecting a difference in the quality and/or availability of mass transit across markets.**\n\n**Frequency of driving personal vehicle**\n\nNote: Percentages may not add up to 100 due to rounding. Q70. How often do you drive your current vehicle? Sample size: n = 1,001 [China]; 1,507 [Germany]; 1,000 [India]; 1,000 [Japan]; 1,012 [Republic of Korea]; 6,029 [Southeast Asia]; 1,505 [UK]; 1,002 [US]\n\n**Even though surveyed consumers in India drive a lot, many younger consumers there would be willing to give up vehicle ownership in favor of a MaaS solution. In contrast, only 1 in 3 younger consumers in Germany would be willing to do the same.**\n\n**Willingness to give up vehicle ownership in favor of MaaS (% somewhat willing/willing/very willing) – 18- to 34-year-old respondents**\n\n**70%**\n\n**54% 54%**\n\n**44%**\n\n**38% 37% 37% 35%**\n\nIndia China Southeast Asia US Japan Rep. of Korea UK Germany\n\nQ63. To what extent would you be willing to give up vehicle ownership in favor of a fully available mobility-as-a-service (MaaS) solution going forward? Sample size: n = 261 [China]; 344 [Germany]; 404 [India]; 191 [Japan]; 246 [Republic of Korea]; 2,116 [Southeast Asia]; 392 [UK]; 286 [US]","pages":[{"line_numbers":[],"markdown":"24\n\n**2025 Global Automotive Consumer Study** | Key Findings: Global Focus Markets\n\n**23%**\n\n**25%**\n\n**37%**\n\n**38%**\n\n**40%**\n\n**49%**\n\n**49%**\n\n**51%**\n\n**18%**\n\n**23%**\n\n**26%**\n\n**29%**\n\n**36%**\n\n**27%**\n\n**23%**\n\n**25%**\n\n**18%**\n\n**21%**\n\n**17%**\n\n**13%**\n\n**11%**\n\n**11%**\n\n**11%**\n\n**11%**\n\n**6%**\n\n**10%**\n\n**8%**\n\n**5%**\n\n**6%**\n\n**5%**\n\n**5%**\n\n**6%**\n\n**2%**\n\n**4%**\n\n**3%**\n\n**2%**\n\n**2%**\n\n**3%**\n\n**3%**\n\n**3%**\n\n**32%**\n\n**18%**\n\n**9%**\n\n**12%**\n\n**4%**\n\n**6%**\n\n**8%**\n\n**5%**\n\nJapan\n\nSouth Korea\n\nGermany\n\nUK\n\nChina\n\nUS\n\nSoutheast Asia\n\nIndia\n\nEvery day 3-4 times per week 1-2 times per week 2-3 times per month Once a month Once every few months\n\n**Half of surveyed consumers in India, Southeast Asia, and the United States say they drive their vehi- cle every day—roughly twice the number of consumers in Japan and South Korea who say the same,** **perhaps reflecting a difference in the quality and/or availability of mass transit across markets.**\n\n**Frequency of driving personal vehicle**\n\nNote: Percentages may not add up to 100 due to rounding. Q70. How often do you drive your current vehicle? Sample size: n = 1,001 [China]; 1,507 [Germany]; 1,000 [India]; 1,000 [Japan]; 1,012 [Republic of Korea]; 6,029 [Southeast Asia]; 1,505 [UK]; 1,002 [US]\n\n**Even though surveyed consumers in India drive a lot, many younger consumers there would be willing to give up vehicle ownership in favor of a MaaS solution. In contrast, only 1 in 3 younger consumers in Germany would be willing to do the same.**\n\n**Willingness to give up vehicle ownership in favor of MaaS (% somewhat willing/willing/very willing) – 18- to 34-year-old respondents**\n\n**70%**\n\n**54% 54%**\n\n**44%**\n\n**38% 37% 37% 35%**\n\nIndia China Southeast Asia US Japan Rep. of Korea UK Germany\n\nQ63. To what extent would you be willing to give up vehicle ownership in favor of a fully available mobility-as-a-service (MaaS) solution going forward? Sample size: n = 261 [China]; 344 [Germany]; 404 [India]; 191 [Japan]; 246 [Republic of Korea]; 2,116 [Southeast Asia]; 392 [UK]; 286 [US]","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"24\n\n**2025 Global Automotive Consumer Study** | Key Findings: Global Focus Markets\n\n**23%**\n\n**25%**\n\n**37%**\n\n**38%**\n\n**40%**\n\n**49%**\n\n**49%**\n\n**51%**\n\n**18%**\n\n**23%**\n\n**26%**\n\n**29%**\n\n**36%**\n\n**27%**\n\n**23%**\n\n**25%**\n\n**18%**\n\n**21%**\n\n**17%**\n\n**13%**\n\n**11%**\n\n**11%**\n\n**11%**\n\n**11%**\n\n**6%**\n\n**10%**\n\n**8%**\n\n**5%**\n\n**6%**\n\n**5%**\n\n**5%**\n\n**6%**\n\n**2%**\n\n**4%**\n\n**3%**\n\n**2%**\n\n**2%**\n\n**3%**\n\n**3%**\n\n**3%**\n\n**32%**\n\n**18%**\n\n**9%**\n\n**12%**\n\n**4%**\n\n**6%**\n\n**8%**\n\n**5%**\n\nJapan\n\nSouth Korea\n\nGermany\n\nUK\n\nChina\n\nUS\n\nSoutheast Asia\n\nIndia\n\nEvery day 3-4 times per week 1-2 times per week 2-3 times per month Once a month Once every few months\n\n**Half of surveyed consumers in India, Southeast Asia, and the United States say they drive their vehi- cle every day—roughly twice the number of consumers in Japan and South Korea who say the same,** **perhaps reflecting a difference in the quality and/or availability of mass transit across markets.**\n\n**Frequency of driving personal vehicle**\n\nNote: Percentages may not add up to 100 due to rounding. Q70. How often do you drive your current vehicle? Sample size: n = 1,001 [China]; 1,507 [Germany]; 1,000 [India]; 1,000 [Japan]; 1,012 [Republic of Korea]; 6,029 [Southeast Asia]; 1,505 [UK]; 1,002 [US]\n\n**Even though surveyed consumers in India drive a lot, many younger consumers there would be willing to give up vehicle ownership in favor of a MaaS solution. In contrast, only 1 in 3 younger consumers in Germany would be willing to do the same.**\n\n**Willingness to give up vehicle ownership in favor of MaaS (% somewhat willing/willing/very willing) – 18- to 34-year-old respondents**\n\n**70%**\n\n**54% 54%**\n\n**44%**\n\n**38% 37% 37% 35%**\n\nIndia China Southeast Asia US Japan Rep. of Korea UK Germany\n\nQ63. To what extent would you be willing to give up vehicle ownership in favor of a fully available mobility-as-a-service (MaaS) solution going forward? Sample size: n = 261 [China]; 344 [Germany]; 404 [India]; 191 [Japan]; 246 [Republic of Korea]; 2,116 [Southeast Asia]; 392 [UK]; 286 [US]","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/deloitte-2025-global-automotive-consumer-study-january-2025_p24.pdf"]},"response":{"errors":[],"extract_id":"extract_6b3ae2236adf8f1a842338f8a200dc01","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["24\n\n**2025 Global Automotive Consumer Study** | Key Findings: Global Focus Markets\n\n**23%**\n\n**25%**\n\n**37%**\n\n**38%**\n\n**40%**\n\n**49%**\n\n**49%**\n\n**51%**\n\n**18%**\n\n**23%**\n\n**26%**\n\n**29%**\n\n**36%**\n\n**27%**\n\n**23%**\n\n**25%**\n\n**18%**\n\n**21%**\n\n**17%**\n\n**13%**\n\n**11%**\n\n**11%**\n\n**11%**\n\n**11%**\n\n**6%**\n\n**10%**\n\n**8%**\n\n**5%**\n\n**6%**\n\n**5%**\n\n**5%**\n\n**6%**\n\n**2%**\n\n**4%**\n\n**3%**\n\n**2%**\n\n**2%**\n\n**3%**\n\n**3%**\n\n**3%**\n\n**32%**\n\n**18%**\n\n**9%**\n\n**12%**\n\n**4%**\n\n**6%**\n\n**8%**\n\n**5%**\n\nJapan\n\nSouth Korea\n\nGermany\n\nUK\n\nChina\n\nUS\n\nSoutheast Asia\n\nIndia\n\nEvery day 3-4 times per week 1-2 times per week 2-3 times per month Once a month Once every few months\n\n**Half of surveyed consumers in India, Southeast Asia, and the United States say they drive their vehi- cle every day—roughly twice the number of consumers in Japan and South Korea who say the same,** **perhaps reflecting a difference in the quality and/or availability of mass transit across markets.**\n\n**Frequency of driving personal vehicle**\n\nNote: Percentages may not add up to 100 due to rounding. Q70. How often do you drive your current vehicle? Sample size: n = 1,001 [China]; 1,507 [Germany]; 1,000 [India]; 1,000 [Japan]; 1,012 [Republic of Korea]; 6,029 [Southeast Asia]; 1,505 [UK]; 1,002 [US]\n\n**Even though surveyed consumers in India drive a lot, many younger consumers there would be willing to give up vehicle ownership in favor of a MaaS solution. In contrast, only 1 in 3 younger consumers in Germany would be willing to do the same.**\n\n**Willingness to give up vehicle ownership in favor of MaaS (% somewhat willing/willing/very willing) – 18- to 34-year-old respondents**\n\n**70%**\n\n**54% 54%**\n\n**44%**\n\n**38% 37% 37% 35%**\n\nIndia China Southeast Asia US Japan Rep. of Korea UK Germany\n\nQ63. To what extent would you be willing to give up vehicle ownership in favor of a fully available mobility-as-a-service (MaaS) solution going forward? Sample size: n = 261 [China]; 344 [Germany]; 404 [India]; 191 [Japan]; 246 [Republic of Korea]; 2,116 [Southeast Asia]; 392 [UK]; 286 [US]"],"full_content":"24\n\n**2025 Global Automotive Consumer Study** | Key Findings: Global Focus Markets\n\n**23%**\n\n**25%**\n\n**37%**\n\n**38%**\n\n**40%**\n\n**49%**\n\n**49%**\n\n**51%**\n\n**18%**\n\n**23%**\n\n**26%**\n\n**29%**\n\n**36%**\n\n**27%**\n\n**23%**\n\n**25%**\n\n**18%**\n\n**21%**\n\n**17%**\n\n**13%**\n\n**11%**\n\n**11%**\n\n**11%**\n\n**11%**\n\n**6%**\n\n**10%**\n\n**8%**\n\n**5%**\n\n**6%**\n\n**5%**\n\n**5%**\n\n**6%**\n\n**2%**\n\n**4%**\n\n**3%**\n\n**2%**\n\n**2%**\n\n**3%**\n\n**3%**\n\n**3%**\n\n**32%**\n\n**18%**\n\n**9%**\n\n**12%**\n\n**4%**\n\n**6%**\n\n**8%**\n\n**5%**\n\nJapan\n\nSouth Korea\n\nGermany\n\nUK\n\nChina\n\nUS\n\nSoutheast Asia\n\nIndia\n\nEvery day 3-4 times per week 1-2 times per week 2-3 times per month Once a month Once every few months\n\n**Half of surveyed consumers in India, Southeast Asia, and the United States say they drive their vehi- cle every day—roughly twice the number of consumers in Japan and South Korea who say the same,** **perhaps reflecting a difference in the quality and/or availability of mass transit across markets.**\n\n**Frequency of driving personal vehicle**\n\nNote: Percentages may not add up to 100 due to rounding. Q70. How often do you drive your current vehicle? Sample size: n = 1,001 [China]; 1,507 [Germany]; 1,000 [India]; 1,000 [Japan]; 1,012 [Republic of Korea]; 6,029 [Southeast Asia]; 1,505 [UK]; 1,002 [US]\n\n**Even though surveyed consumers in India drive a lot, many younger consumers there would be willing to give up vehicle ownership in favor of a MaaS solution. In contrast, only 1 in 3 younger consumers in Germany would be willing to do the same.**\n\n**Willingness to give up vehicle ownership in favor of MaaS (% somewhat willing/willing/very willing) – 18- to 34-year-old respondents**\n\n**70%**\n\n**54% 54%**\n\n**44%**\n\n**38% 37% 37% 35%**\n\nIndia China Southeast Asia US Japan Rep. of Korea UK Germany\n\nQ63. To what extent would you be willing to give up vehicle ownership in favor of a fully available mobility-as-a-service (MaaS) solution going forward? Sample size: n = 261 [China]; 344 [Germany]; 404 [India]; 191 [Japan]; 246 [Republic of Korea]; 2,116 [Southeast Asia]; 392 [UK]; 286 [US]","publish_date":null,"title":"Half of surveyed consumers in India, Southeast Asia, and the United States say they drive their vehi- cle every day—roughly","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/deloitte-2025-global-automotive-consumer-study-january-2025_p24.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/deloitte-2025-global-automotive-consumer-study-january-2025_p24.pdf"},"request":{"example_id":"chart/deloitte-2025-global-automotive-consumer-study-january-2025_p24","product_type":"parse","source_file_path":"deloitte-2025-global-automotive-consumer-study-january-2025_p24.pdf"},"started_at":"2026-11-04T19:45:02.218319"} +{"completed_at":"2026-11-04T19:45:05.945266","latency_in_ms":2289,"output":{"example_id":"chart/deloitte-2025-global-automotive-consumer-study-january-2025_p8","grounded_pages":[],"layout_pages":[],"markdown":"8\n\n**2025 Global Automotive Consumer Study** | Key Findings: Global Focus Markets\n\nHome Work On-street/public charging station Other\n\nUK\n\n**80%**\n\n**10%**\n\n**11%**\n\n**10%**\n\n**11%**\n\nUS\n\n**79%**\n\nJapan\n\n**72%**\n\nGermany India China Southeast Asia Rep. of Korea\n\n**2%**\n\n**13%**\n\n**14%**\n\n**70%**\n\n**18%**\n\n**12%**\n\n**63%**\n\n**22%**\n\n**15%**\n\n**62%**\n\n**28%**\n\n**10%**\n\n**61%**\n\n**21%**\n\n**17%**\n\n**71%**\n\n**20%**\n\n**9%**\n\nYes, dedicated Yes, shared with a neighbor/others No Don’t know\n\nChina\n\n**77%**\n\n**15%**\n\n**7%**\n\nIndia\n\n**41%**\n\n**44%**\n\nUS\n\n**40%**\n\nGermany UK Southeast Asia Japan\n\n**14%**\n\n**58%**\n\n**2%**\n\n**33%**\n\n**62%**\n\n**5%**\n\n**32%**\n\n**58%**\n\n**8%**\n\n**28%**\n\n**65%**\n\n**4%**\n\nRep. of Korea\n\n**19%**\n\n**7%**\n\n**25%**\n\n**37%**\n\n**56%**\n\n**6%**\n\n**1% 1% 1% 2% 2%**\n\n**50%**\n\n**The push to create public charging capacity may be somewhat overstated in many markets as a** **strong majority of EV intenders surveyed plan to charge their vehicle at home.**\n\n**Expecting to charge electrified vehicle most often at…**\n\n**Yet, many EV intenders surveyed in mature markets that plan to charge at home do not currently have access to a charger.**\n\n**EV charger access among consumers who plan to charge their vehicle at home**\n\nNote: Percentages may not add up to 100 due to rounding. Q43. Where do you expect to charge your EV most often? Sample size: n = 413 [China]; 304 [Germany]; 182 [India]; 64 [Japan]; 166 [Republic of Korea]; 1,223 [Southeast Asia]; 245 [UK]; 111 [US]\n\nNote: Percentages may not add up to 100 due to rounding. Q44. Do you already have access to a charger at your residence? Sample size: n = 260 [China]; 215 [Germany]; 128 [India]; 46 [Japan]; 102 [Republic of Korea]; 755 [Southeast Asia]; 195 [UK]; 88 [US]","pages":[{"line_numbers":[],"markdown":"8\n\n**2025 Global Automotive Consumer Study** | Key Findings: Global Focus Markets\n\nHome Work On-street/public charging station Other\n\nUK\n\n**80%**\n\n**10%**\n\n**11%**\n\n**10%**\n\n**11%**\n\nUS\n\n**79%**\n\nJapan\n\n**72%**\n\nGermany India China Southeast Asia Rep. of Korea\n\n**2%**\n\n**13%**\n\n**14%**\n\n**70%**\n\n**18%**\n\n**12%**\n\n**63%**\n\n**22%**\n\n**15%**\n\n**62%**\n\n**28%**\n\n**10%**\n\n**61%**\n\n**21%**\n\n**17%**\n\n**71%**\n\n**20%**\n\n**9%**\n\nYes, dedicated Yes, shared with a neighbor/others No Don’t know\n\nChina\n\n**77%**\n\n**15%**\n\n**7%**\n\nIndia\n\n**41%**\n\n**44%**\n\nUS\n\n**40%**\n\nGermany UK Southeast Asia Japan\n\n**14%**\n\n**58%**\n\n**2%**\n\n**33%**\n\n**62%**\n\n**5%**\n\n**32%**\n\n**58%**\n\n**8%**\n\n**28%**\n\n**65%**\n\n**4%**\n\nRep. of Korea\n\n**19%**\n\n**7%**\n\n**25%**\n\n**37%**\n\n**56%**\n\n**6%**\n\n**1% 1% 1% 2% 2%**\n\n**50%**\n\n**The push to create public charging capacity may be somewhat overstated in many markets as a** **strong majority of EV intenders surveyed plan to charge their vehicle at home.**\n\n**Expecting to charge electrified vehicle most often at…**\n\n**Yet, many EV intenders surveyed in mature markets that plan to charge at home do not currently have access to a charger.**\n\n**EV charger access among consumers who plan to charge their vehicle at home**\n\nNote: Percentages may not add up to 100 due to rounding. Q43. Where do you expect to charge your EV most often? Sample size: n = 413 [China]; 304 [Germany]; 182 [India]; 64 [Japan]; 166 [Republic of Korea]; 1,223 [Southeast Asia]; 245 [UK]; 111 [US]\n\nNote: Percentages may not add up to 100 due to rounding. Q44. Do you already have access to a charger at your residence? Sample size: n = 260 [China]; 215 [Germany]; 128 [India]; 46 [Japan]; 102 [Republic of Korea]; 755 [Southeast Asia]; 195 [UK]; 88 [US]","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"8\n\n**2025 Global Automotive Consumer Study** | Key Findings: Global Focus Markets\n\nHome Work On-street/public charging station Other\n\nUK\n\n**80%**\n\n**10%**\n\n**11%**\n\n**10%**\n\n**11%**\n\nUS\n\n**79%**\n\nJapan\n\n**72%**\n\nGermany India China Southeast Asia Rep. of Korea\n\n**2%**\n\n**13%**\n\n**14%**\n\n**70%**\n\n**18%**\n\n**12%**\n\n**63%**\n\n**22%**\n\n**15%**\n\n**62%**\n\n**28%**\n\n**10%**\n\n**61%**\n\n**21%**\n\n**17%**\n\n**71%**\n\n**20%**\n\n**9%**\n\nYes, dedicated Yes, shared with a neighbor/others No Don’t know\n\nChina\n\n**77%**\n\n**15%**\n\n**7%**\n\nIndia\n\n**41%**\n\n**44%**\n\nUS\n\n**40%**\n\nGermany UK Southeast Asia Japan\n\n**14%**\n\n**58%**\n\n**2%**\n\n**33%**\n\n**62%**\n\n**5%**\n\n**32%**\n\n**58%**\n\n**8%**\n\n**28%**\n\n**65%**\n\n**4%**\n\nRep. of Korea\n\n**19%**\n\n**7%**\n\n**25%**\n\n**37%**\n\n**56%**\n\n**6%**\n\n**1% 1% 1% 2% 2%**\n\n**50%**\n\n**The push to create public charging capacity may be somewhat overstated in many markets as a** **strong majority of EV intenders surveyed plan to charge their vehicle at home.**\n\n**Expecting to charge electrified vehicle most often at…**\n\n**Yet, many EV intenders surveyed in mature markets that plan to charge at home do not currently have access to a charger.**\n\n**EV charger access among consumers who plan to charge their vehicle at home**\n\nNote: Percentages may not add up to 100 due to rounding. Q43. Where do you expect to charge your EV most often? Sample size: n = 413 [China]; 304 [Germany]; 182 [India]; 64 [Japan]; 166 [Republic of Korea]; 1,223 [Southeast Asia]; 245 [UK]; 111 [US]\n\nNote: Percentages may not add up to 100 due to rounding. Q44. Do you already have access to a charger at your residence? Sample size: n = 260 [China]; 215 [Germany]; 128 [India]; 46 [Japan]; 102 [Republic of Korea]; 755 [Southeast Asia]; 195 [UK]; 88 [US]","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/deloitte-2025-global-automotive-consumer-study-january-2025_p8.pdf"]},"response":{"errors":[],"extract_id":"extract_f4232b01891912cfc702093b884f559c","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["8\n\n**2025 Global Automotive Consumer Study** | Key Findings: Global Focus Markets\n\nHome Work On-street/public charging station Other\n\nUK\n\n**80%**\n\n**10%**\n\n**11%**\n\n**10%**\n\n**11%**\n\nUS\n\n**79%**\n\nJapan\n\n**72%**\n\nGermany India China Southeast Asia Rep. of Korea\n\n**2%**\n\n**13%**\n\n**14%**\n\n**70%**\n\n**18%**\n\n**12%**\n\n**63%**\n\n**22%**\n\n**15%**\n\n**62%**\n\n**28%**\n\n**10%**\n\n**61%**\n\n**21%**\n\n**17%**\n\n**71%**\n\n**20%**\n\n**9%**\n\nYes, dedicated Yes, shared with a neighbor/others No Don’t know\n\nChina\n\n**77%**\n\n**15%**\n\n**7%**\n\nIndia\n\n**41%**\n\n**44%**\n\nUS\n\n**40%**\n\nGermany UK Southeast Asia Japan\n\n**14%**\n\n**58%**\n\n**2%**\n\n**33%**\n\n**62%**\n\n**5%**\n\n**32%**\n\n**58%**\n\n**8%**\n\n**28%**\n\n**65%**\n\n**4%**\n\nRep. of Korea\n\n**19%**\n\n**7%**\n\n**25%**\n\n**37%**\n\n**56%**\n\n**6%**\n\n**1% 1% 1% 2% 2%**\n\n**50%**\n\n**The push to create public charging capacity may be somewhat overstated in many markets as a** **strong majority of EV intenders surveyed plan to charge their vehicle at home.**\n\n**Expecting to charge electrified vehicle most often at…**\n\n**Yet, many EV intenders surveyed in mature markets that plan to charge at home do not currently have access to a charger.**\n\n**EV charger access among consumers who plan to charge their vehicle at home**\n\nNote: Percentages may not add up to 100 due to rounding. Q43. Where do you expect to charge your EV most often? Sample size: n = 413 [China]; 304 [Germany]; 182 [India]; 64 [Japan]; 166 [Republic of Korea]; 1,223 [Southeast Asia]; 245 [UK]; 111 [US]\n\nNote: Percentages may not add up to 100 due to rounding. Q44. Do you already have access to a charger at your residence? Sample size: n = 260 [China]; 215 [Germany]; 128 [India]; 46 [Japan]; 102 [Republic of Korea]; 755 [Southeast Asia]; 195 [UK]; 88 [US]"],"full_content":"8\n\n**2025 Global Automotive Consumer Study** | Key Findings: Global Focus Markets\n\nHome Work On-street/public charging station Other\n\nUK\n\n**80%**\n\n**10%**\n\n**11%**\n\n**10%**\n\n**11%**\n\nUS\n\n**79%**\n\nJapan\n\n**72%**\n\nGermany India China Southeast Asia Rep. of Korea\n\n**2%**\n\n**13%**\n\n**14%**\n\n**70%**\n\n**18%**\n\n**12%**\n\n**63%**\n\n**22%**\n\n**15%**\n\n**62%**\n\n**28%**\n\n**10%**\n\n**61%**\n\n**21%**\n\n**17%**\n\n**71%**\n\n**20%**\n\n**9%**\n\nYes, dedicated Yes, shared with a neighbor/others No Don’t know\n\nChina\n\n**77%**\n\n**15%**\n\n**7%**\n\nIndia\n\n**41%**\n\n**44%**\n\nUS\n\n**40%**\n\nGermany UK Southeast Asia Japan\n\n**14%**\n\n**58%**\n\n**2%**\n\n**33%**\n\n**62%**\n\n**5%**\n\n**32%**\n\n**58%**\n\n**8%**\n\n**28%**\n\n**65%**\n\n**4%**\n\nRep. of Korea\n\n**19%**\n\n**7%**\n\n**25%**\n\n**37%**\n\n**56%**\n\n**6%**\n\n**1% 1% 1% 2% 2%**\n\n**50%**\n\n**The push to create public charging capacity may be somewhat overstated in many markets as a** **strong majority of EV intenders surveyed plan to charge their vehicle at home.**\n\n**Expecting to charge electrified vehicle most often at…**\n\n**Yet, many EV intenders surveyed in mature markets that plan to charge at home do not currently have access to a charger.**\n\n**EV charger access among consumers who plan to charge their vehicle at home**\n\nNote: Percentages may not add up to 100 due to rounding. Q43. Where do you expect to charge your EV most often? Sample size: n = 413 [China]; 304 [Germany]; 182 [India]; 64 [Japan]; 166 [Republic of Korea]; 1,223 [Southeast Asia]; 245 [UK]; 111 [US]\n\nNote: Percentages may not add up to 100 due to rounding. Q44. Do you already have access to a charger at your residence? Sample size: n = 260 [China]; 215 [Germany]; 128 [India]; 46 [Japan]; 102 [Republic of Korea]; 755 [Southeast Asia]; 195 [UK]; 88 [US]","publish_date":null,"title":"The push to create public charging capacity may be somewhat overstated in many markets as a strong majority of EV","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/deloitte-2025-global-automotive-consumer-study-january-2025_p8.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/deloitte-2025-global-automotive-consumer-study-january-2025_p8.pdf"},"request":{"example_id":"chart/deloitte-2025-global-automotive-consumer-study-january-2025_p8","product_type":"parse","source_file_path":"deloitte-2025-global-automotive-consumer-study-january-2025_p8.pdf"},"started_at":"2026-11-04T19:45:03.655820"} +{"completed_at":"2026-11-04T19:45:06.022422","latency_in_ms":1415,"output":{"example_id":"chart/employment_and_social_developments_in_europe_esde_2024_report_p100","grounded_pages":[],"layout_pages":[],"markdown":"Employment and Social Developments in Europe 2024\n\n104\n\n### Simulation results\n\nChart A3.1 Employment increases for regions not receiving ESF+ investment due to spillover effects Absolute change in employment following ESF+ investment in skills\n\n_Source:_ JRC calculations based on RHOMOLO model. Click here to download chart.\n\n\\-20000\n\n\\-10000\n\n0\n\n10000\n\n20000\n\n30000\n\n40000\n\n2021 2023 2025 2027 2029 2031 2033 2035 2037 2039\n\n_Absolute change in employment_\n\nNet contributors Net receivers\n\nBox A3.3: Calculating Income Stabilisation Coefficients\n\nAnalytical approach **Following earlier research, this note uses EUROMOD to study the stabilisation properties of the tax- benefit systems of the EU-27.** **(** **1** **)** It does so by measuring an income stabilisation coefficient (ISC), defined as the percentage of a market income shock absorbed by the country's tax-benefit system.\n\n**The calculation of the ISC involves simulating a reform scenario that considers a 5% reduction in gross income, uniformly applied to any form of market income.** EUROMOD is then used to calculate taxes, benefits and disposable income for both the baseline and reform scenarios. The model’s underlying data come from EU-SILC. The comparison of the two scenarios provides an estimate of the stabilisation capacity of the tax and benefit systems. The household-level ISC is calculated as:\n\n𝐼𝐼𝐼𝐼𝐶𝐶ℎ = 1 − Δ𝑌𝑌ℎ\n\nΔ𝑀𝑀ℎ\n\n= Δ𝑇𝑇ℎ\n\nΔ𝑀𝑀ℎ\n\n− Δ𝐵𝐵ℎ\n\nΔ𝑀𝑀ℎ\n\nwhere 𝑌𝑌ℎ is household h disposable income, 𝑀𝑀ℎ is gross market income, 𝐵𝐵ℎ are social benefits received, 𝑇𝑇ℎ are taxes and social insurance contributions (SICs) paid and the Δ𝑋𝑋ℎ = 𝑋𝑋ℎ𝑅𝑅 − 𝑋𝑋ℎ𝐵𝐵 is the change in variable 𝑋𝑋 due to the drop in 5% market income between the baseline and the reform scenarios (𝑋𝑋 being either gross market income, social benefits received, taxes or social insurance contributions paid and household disposable income).\n\nThe higher the coefficient, the stronger the stabilisation effect. For instance, a coefficient of 30% indicates that 30% of a shock to market income is absorbed by the public budget and only 70% of the shock is transmitted into disposable income. Ultimately, 𝐼𝐼𝐼𝐼𝐶𝐶ℎ is equal to 100% if no change in disposable income is observed following the shock (i.e. fiscal policies fully absorb the shock) and equal to 0% if the change in market income is fully transmitted to disposable income. Given the nature of the analysis and the type of shock considered (income decrease rather than a change in labour market status), the policy instruments that will react to the changes are taxes, SICs, and all benefits except unemployment benefits. To analyse convergence of ISCs among Member States, the contribution of discretionary policies to the stabilisation of incomes is reduced by considering average ISCs across two years. ( ² )\n\n( ¹ ) (Coady et al., 2023); (Dolls, Fuest and Peichl, 2012); (Dolls et al., 2022). ( ² ) Convergence analysis compares ISCs before the COVID-19 pandemic (2014-2015 average) and the latest years for which data\n\nare available (2022-2023 average).","pages":[{"line_numbers":[],"markdown":"Employment and Social Developments in Europe 2024\n\n104\n\n### Simulation results\n\nChart A3.1 Employment increases for regions not receiving ESF+ investment due to spillover effects Absolute change in employment following ESF+ investment in skills\n\n_Source:_ JRC calculations based on RHOMOLO model. Click here to download chart.\n\n\\-20000\n\n\\-10000\n\n0\n\n10000\n\n20000\n\n30000\n\n40000\n\n2021 2023 2025 2027 2029 2031 2033 2035 2037 2039\n\n_Absolute change in employment_\n\nNet contributors Net receivers\n\nBox A3.3: Calculating Income Stabilisation Coefficients\n\nAnalytical approach **Following earlier research, this note uses EUROMOD to study the stabilisation properties of the tax- benefit systems of the EU-27.** **(** **1** **)** It does so by measuring an income stabilisation coefficient (ISC), defined as the percentage of a market income shock absorbed by the country's tax-benefit system.\n\n**The calculation of the ISC involves simulating a reform scenario that considers a 5% reduction in gross income, uniformly applied to any form of market income.** EUROMOD is then used to calculate taxes, benefits and disposable income for both the baseline and reform scenarios. The model’s underlying data come from EU-SILC. The comparison of the two scenarios provides an estimate of the stabilisation capacity of the tax and benefit systems. The household-level ISC is calculated as:\n\n𝐼𝐼𝐼𝐼𝐶𝐶ℎ = 1 − Δ𝑌𝑌ℎ\n\nΔ𝑀𝑀ℎ\n\n= Δ𝑇𝑇ℎ\n\nΔ𝑀𝑀ℎ\n\n− Δ𝐵𝐵ℎ\n\nΔ𝑀𝑀ℎ\n\nwhere 𝑌𝑌ℎ is household h disposable income, 𝑀𝑀ℎ is gross market income, 𝐵𝐵ℎ are social benefits received, 𝑇𝑇ℎ are taxes and social insurance contributions (SICs) paid and the Δ𝑋𝑋ℎ = 𝑋𝑋ℎ𝑅𝑅 − 𝑋𝑋ℎ𝐵𝐵 is the change in variable 𝑋𝑋 due to the drop in 5% market income between the baseline and the reform scenarios (𝑋𝑋 being either gross market income, social benefits received, taxes or social insurance contributions paid and household disposable income).\n\nThe higher the coefficient, the stronger the stabilisation effect. For instance, a coefficient of 30% indicates that 30% of a shock to market income is absorbed by the public budget and only 70% of the shock is transmitted into disposable income. Ultimately, 𝐼𝐼𝐼𝐼𝐶𝐶ℎ is equal to 100% if no change in disposable income is observed following the shock (i.e. fiscal policies fully absorb the shock) and equal to 0% if the change in market income is fully transmitted to disposable income. Given the nature of the analysis and the type of shock considered (income decrease rather than a change in labour market status), the policy instruments that will react to the changes are taxes, SICs, and all benefits except unemployment benefits. To analyse convergence of ISCs among Member States, the contribution of discretionary policies to the stabilisation of incomes is reduced by considering average ISCs across two years. ( ² )\n\n( ¹ ) (Coady et al., 2023); (Dolls, Fuest and Peichl, 2012); (Dolls et al., 2022). ( ² ) Convergence analysis compares ISCs before the COVID-19 pandemic (2014-2015 average) and the latest years for which data\n\nare available (2022-2023 average).","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Employment and Social Developments in Europe 2024\n\n104\n\n### Simulation results\n\nChart A3.1 Employment increases for regions not receiving ESF+ investment due to spillover effects Absolute change in employment following ESF+ investment in skills\n\n_Source:_ JRC calculations based on RHOMOLO model. Click here to download chart.\n\n\\-20000\n\n\\-10000\n\n0\n\n10000\n\n20000\n\n30000\n\n40000\n\n2021 2023 2025 2027 2029 2031 2033 2035 2037 2039\n\n_Absolute change in employment_\n\nNet contributors Net receivers\n\nBox A3.3: Calculating Income Stabilisation Coefficients\n\nAnalytical approach **Following earlier research, this note uses EUROMOD to study the stabilisation properties of the tax- benefit systems of the EU-27.** **(** **1** **)** It does so by measuring an income stabilisation coefficient (ISC), defined as the percentage of a market income shock absorbed by the country's tax-benefit system.\n\n**The calculation of the ISC involves simulating a reform scenario that considers a 5% reduction in gross income, uniformly applied to any form of market income.** EUROMOD is then used to calculate taxes, benefits and disposable income for both the baseline and reform scenarios. The model’s underlying data come from EU-SILC. The comparison of the two scenarios provides an estimate of the stabilisation capacity of the tax and benefit systems. The household-level ISC is calculated as:\n\n𝐼𝐼𝐼𝐼𝐶𝐶ℎ = 1 − Δ𝑌𝑌ℎ\n\nΔ𝑀𝑀ℎ\n\n= Δ𝑇𝑇ℎ\n\nΔ𝑀𝑀ℎ\n\n− Δ𝐵𝐵ℎ\n\nΔ𝑀𝑀ℎ\n\nwhere 𝑌𝑌ℎ is household h disposable income, 𝑀𝑀ℎ is gross market income, 𝐵𝐵ℎ are social benefits received, 𝑇𝑇ℎ are taxes and social insurance contributions (SICs) paid and the Δ𝑋𝑋ℎ = 𝑋𝑋ℎ𝑅𝑅 − 𝑋𝑋ℎ𝐵𝐵 is the change in variable 𝑋𝑋 due to the drop in 5% market income between the baseline and the reform scenarios (𝑋𝑋 being either gross market income, social benefits received, taxes or social insurance contributions paid and household disposable income).\n\nThe higher the coefficient, the stronger the stabilisation effect. For instance, a coefficient of 30% indicates that 30% of a shock to market income is absorbed by the public budget and only 70% of the shock is transmitted into disposable income. Ultimately, 𝐼𝐼𝐼𝐼𝐶𝐶ℎ is equal to 100% if no change in disposable income is observed following the shock (i.e. fiscal policies fully absorb the shock) and equal to 0% if the change in market income is fully transmitted to disposable income. Given the nature of the analysis and the type of shock considered (income decrease rather than a change in labour market status), the policy instruments that will react to the changes are taxes, SICs, and all benefits except unemployment benefits. To analyse convergence of ISCs among Member States, the contribution of discretionary policies to the stabilisation of incomes is reduced by considering average ISCs across two years. ( ² )\n\n( ¹ ) (Coady et al., 2023); (Dolls, Fuest and Peichl, 2012); (Dolls et al., 2022). ( ² ) Convergence analysis compares ISCs before the COVID-19 pandemic (2014-2015 average) and the latest years for which data\n\nare available (2022-2023 average).","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/employment_and_social_developments_in_europe_esde_2024_report_p100.pdf"]},"response":{"errors":[],"extract_id":"extract_8597c7a9451de03dda6e9333106f4e8f","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Employment and Social Developments in Europe 2024\n\n104\n\n### Simulation results\n\nChart A3.1 Employment increases for regions not receiving ESF+ investment due to spillover effects Absolute change in employment following ESF+ investment in skills\n\n_Source:_ JRC calculations based on RHOMOLO model. Click here to download chart.\n\n\\-20000\n\n\\-10000\n\n0\n\n10000\n\n20000\n\n30000\n\n40000\n\n2021 2023 2025 2027 2029 2031 2033 2035 2037 2039\n\n_Absolute change in employment_\n\nNet contributors Net receivers\n\nBox A3.3: Calculating Income Stabilisation Coefficients\n\nAnalytical approach **Following earlier research, this note uses EUROMOD to study the stabilisation properties of the tax- benefit systems of the EU-27.** **(** **1** **)** It does so by measuring an income stabilisation coefficient (ISC), defined as the percentage of a market income shock absorbed by the country's tax-benefit system.\n\n**The calculation of the ISC involves simulating a reform scenario that considers a 5% reduction in gross income, uniformly applied to any form of market income.** EUROMOD is then used to calculate taxes, benefits and disposable income for both the baseline and reform scenarios. The model’s underlying data come from EU-SILC. The comparison of the two scenarios provides an estimate of the stabilisation capacity of the tax and benefit systems. The household-level ISC is calculated as:\n\n𝐼𝐼𝐼𝐼𝐶𝐶ℎ = 1 − Δ𝑌𝑌ℎ\n\nΔ𝑀𝑀ℎ\n\n= Δ𝑇𝑇ℎ\n\nΔ𝑀𝑀ℎ\n\n− Δ𝐵𝐵ℎ\n\nΔ𝑀𝑀ℎ\n\nwhere 𝑌𝑌ℎ is household h disposable income, 𝑀𝑀ℎ is gross market income, 𝐵𝐵ℎ are social benefits received, 𝑇𝑇ℎ are taxes and social insurance contributions (SICs) paid and the Δ𝑋𝑋ℎ = 𝑋𝑋ℎ𝑅𝑅 − 𝑋𝑋ℎ𝐵𝐵 is the change in variable 𝑋𝑋 due to the drop in 5% market income between the baseline and the reform scenarios (𝑋𝑋 being either gross market income, social benefits received, taxes or social insurance contributions paid and household disposable income).\n\nThe higher the coefficient, the stronger the stabilisation effect. For instance, a coefficient of 30% indicates that 30% of a shock to market income is absorbed by the public budget and only 70% of the shock is transmitted into disposable income. Ultimately, 𝐼𝐼𝐼𝐼𝐶𝐶ℎ is equal to 100% if no change in disposable income is observed following the shock (i.e. fiscal policies fully absorb the shock) and equal to 0% if the change in market income is fully transmitted to disposable income. Given the nature of the analysis and the type of shock considered (income decrease rather than a change in labour market status), the policy instruments that will react to the changes are taxes, SICs, and all benefits except unemployment benefits. To analyse convergence of ISCs among Member States, the contribution of discretionary policies to the stabilisation of incomes is reduced by considering average ISCs across two years. ( ² )\n\n( ¹ ) (Coady et al., 2023); (Dolls, Fuest and Peichl, 2012); (Dolls et al., 2022). ( ² ) Convergence analysis compares ISCs before the COVID-19 pandemic (2014-2015 average) and the latest years for which data\n\nare available (2022-2023 average)."],"full_content":"Employment and Social Developments in Europe 2024\n\n104\n\n### Simulation results\n\nChart A3.1 Employment increases for regions not receiving ESF+ investment due to spillover effects Absolute change in employment following ESF+ investment in skills\n\n_Source:_ JRC calculations based on RHOMOLO model. Click here to download chart.\n\n\\-20000\n\n\\-10000\n\n0\n\n10000\n\n20000\n\n30000\n\n40000\n\n2021 2023 2025 2027 2029 2031 2033 2035 2037 2039\n\n_Absolute change in employment_\n\nNet contributors Net receivers\n\nBox A3.3: Calculating Income Stabilisation Coefficients\n\nAnalytical approach **Following earlier research, this note uses EUROMOD to study the stabilisation properties of the tax- benefit systems of the EU-27.** **(** **1** **)** It does so by measuring an income stabilisation coefficient (ISC), defined as the percentage of a market income shock absorbed by the country's tax-benefit system.\n\n**The calculation of the ISC involves simulating a reform scenario that considers a 5% reduction in gross income, uniformly applied to any form of market income.** EUROMOD is then used to calculate taxes, benefits and disposable income for both the baseline and reform scenarios. The model’s underlying data come from EU-SILC. The comparison of the two scenarios provides an estimate of the stabilisation capacity of the tax and benefit systems. The household-level ISC is calculated as:\n\n𝐼𝐼𝐼𝐼𝐶𝐶ℎ = 1 − Δ𝑌𝑌ℎ\n\nΔ𝑀𝑀ℎ\n\n= Δ𝑇𝑇ℎ\n\nΔ𝑀𝑀ℎ\n\n− Δ𝐵𝐵ℎ\n\nΔ𝑀𝑀ℎ\n\nwhere 𝑌𝑌ℎ is household h disposable income, 𝑀𝑀ℎ is gross market income, 𝐵𝐵ℎ are social benefits received, 𝑇𝑇ℎ are taxes and social insurance contributions (SICs) paid and the Δ𝑋𝑋ℎ = 𝑋𝑋ℎ𝑅𝑅 − 𝑋𝑋ℎ𝐵𝐵 is the change in variable 𝑋𝑋 due to the drop in 5% market income between the baseline and the reform scenarios (𝑋𝑋 being either gross market income, social benefits received, taxes or social insurance contributions paid and household disposable income).\n\nThe higher the coefficient, the stronger the stabilisation effect. For instance, a coefficient of 30% indicates that 30% of a shock to market income is absorbed by the public budget and only 70% of the shock is transmitted into disposable income. Ultimately, 𝐼𝐼𝐼𝐼𝐶𝐶ℎ is equal to 100% if no change in disposable income is observed following the shock (i.e. fiscal policies fully absorb the shock) and equal to 0% if the change in market income is fully transmitted to disposable income. Given the nature of the analysis and the type of shock considered (income decrease rather than a change in labour market status), the policy instruments that will react to the changes are taxes, SICs, and all benefits except unemployment benefits. To analyse convergence of ISCs among Member States, the contribution of discretionary policies to the stabilisation of incomes is reduced by considering average ISCs across two years. ( ² )\n\n( ¹ ) (Coady et al., 2023); (Dolls, Fuest and Peichl, 2012); (Dolls et al., 2022). ( ² ) Convergence analysis compares ISCs before the COVID-19 pandemic (2014-2015 average) and the latest years for which data\n\nare available (2022-2023 average).","publish_date":null,"title":"Simulation results","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/employment_and_social_developments_in_europe_esde_2024_report_p100.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/employment_and_social_developments_in_europe_esde_2024_report_p100.pdf"},"request":{"example_id":"chart/employment_and_social_developments_in_europe_esde_2024_report_p100","product_type":"parse","source_file_path":"employment_and_social_developments_in_europe_esde_2024_report_p100.pdf"},"started_at":"2026-11-04T19:45:04.607158"} +{"completed_at":"2026-11-04T19:45:08.552871","latency_in_ms":2419,"output":{"example_id":"chart/employment_and_social_developments_in_europe_esde_2024_report_p20","grounded_pages":[],"layout_pages":[],"markdown":"Employment and Social Developments in Europe 2024\n\n22\n\nand tightening monetary policy affected economic performance in 2023, with important consequences for households (see Section 4.1.).\n\nChart 1.2 Real GDP grew in most Member States in 2023 Real GDP growth, EU, 2022 (dots) and 2023 (bars) (% change on previous year)\n\n_Note:_ Ireland: modified domestic demand. 2022: data provisional for Belgium, Cyprus, Germany, Greece, Spain, France, Croatia, Portugal and Romania. 2023: data provisional for Belgium, Bulgaria, Cyprus, Germany, Greece, Spain, France, Croatia, Hungary, Luxembourg, the Netherlands, Portugal and Romania.\n\n_Source:_ Eurostat [nama\\_10\\_gdp]. Click here to download chart.\n\n**Inflation is projected to remain stable for the rest of 2024 and to decline further in 2025.** Annual inflation in the EU is forecasted to average at 2.7% in 2024 and then to decline to 2.2% in 2025. Food and non- energy industrial goods are the primary drivers behind this slowdown. ( ¹⁶ ). Yet, the expiration of policy measures introduced to address the energy crisis may continue to put pressure on the prices of consumer energy. Wage pressures remain elevated, with service prices having contributed little to disinflation.\n\nChart 1.3 Inflation decelerated in most Member States in 2023, but remained high despite decreasing prices of energy Left chart: All-items Harmonised Index of Consumer Prices (HICP), 2023; Right chart, Annual inflation rates for selected items, EU\n\n_Source:_ Eurostat [prc\\_hicp\\_aind; prc\\_hicp\\_manr].\n\nClick here to download chart.\n\n**Almost all Member States’ economies are projected to grow in 2024, with an average growth at the EU-level of 1.0%** (0.8% in the euro area). After recovering somewhat at the beginning of 2024 (+0.6% in Q1 and +0.8% in Q2, compared to same quarter of the previous year), GDP growth is expected to reach 1.6% in 2025 (1.4% in the euro area). The EU entered 2024 with a weaker economic outlook than initially expected. Nevertheless, the conditions remain in place for a gradual improvement in economic activity. Continued wage and employment growth are expected to sustain growth in households’ disposable income in 2024 and to continue in 2025 (albeit slightly slower), which would also boost consumption. The contribution of investment to real GDP growth is expected to increase more gradually and the contribution of EU's net external demand is predicted to become neutral. ( ¹⁷ ) Overall, domestic risks related to growth and inflation are broadly balanced in the EU.\n\n**Heightened geopolitical tensions and global policy uncertainty expose the EU economic outlook to several risks.** Russia's ongoing war of aggression against Ukraine and the Middle East conflict could add to upward pressure on prices and stress on supply chains, jeopardising production and trade. Energy commodities are also vulnerable to these downward risks. Persistence of inflation in the US may delay its interest rate cuts, resulting in tighter global financial conditions. In addition, uncertainties linked to China and US economic activity developments could weigh on the EU economy. Finally, climate change risks continue to be of primary concern,\n\n( ¹⁶ ) (European Commission, 2024a) ( ¹⁷ ) (European Commission, 2024a)\n\n\\-8\n\n\\-6\n\n\\-4\n\n\\-2\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\nIE\n\nEE\n\nFI\n\nLU\n\nHU\n\nAT\n\nLV\n\nLT\n\nDE\n\nSE\n\nCZ\n\nNL\n\nPL\n\nEU\n\nEA\n\nFR\n\nIT\n\nBE\n\nSK\n\nSI\n\nBG\n\nEL\n\nRO\n\nPT\n\nDK\n\nES\n\nCY\n\nHR\n\nMT\n\n2023 Series2\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\n16\n\n18\n\nHU\n\nCZ\n\nSK\n\nPL\n\nRO\n\nEE\n\nLV\n\nLT\n\nBG\n\nHR\n\nAT\n\nSI\n\nEU\n\nDE\n\nSE\n\nIT\n\nFR\n\nMT\n\nEA\n\nPT\n\nIE\n\nFI\n\nEL\n\nNL\n\nCY\n\nES\n\nDK\n\nLU\n\nBE\n\n%\n\nAverage 2023 Dec. 2023 Jul-24\n\n\\-20\n\n\\-10\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n2020-01\n\n2020-04\n\n2020-07\n\n2020-10\n\n2021-01\n\n2021-04\n\n2021-07\n\n2021-10\n\n2022-01\n\n2022-04\n\n2022-07\n\n2022-10\n\n2023-01\n\n2023-04\n\n2023-07\n\n2023-10\n\n2024-01\n\n2024-04\n\n2024-07\n\n_Annual inflation rate (%)_\n\nAll-items HICP\n\nFood, alcohol, and tobacco\n\nNon-energy industrial goods\n\nEnergy\n\nServices","pages":[{"line_numbers":[],"markdown":"Employment and Social Developments in Europe 2024\n\n22\n\nand tightening monetary policy affected economic performance in 2023, with important consequences for households (see Section 4.1.).\n\nChart 1.2 Real GDP grew in most Member States in 2023 Real GDP growth, EU, 2022 (dots) and 2023 (bars) (% change on previous year)\n\n_Note:_ Ireland: modified domestic demand. 2022: data provisional for Belgium, Cyprus, Germany, Greece, Spain, France, Croatia, Portugal and Romania. 2023: data provisional for Belgium, Bulgaria, Cyprus, Germany, Greece, Spain, France, Croatia, Hungary, Luxembourg, the Netherlands, Portugal and Romania.\n\n_Source:_ Eurostat [nama\\_10\\_gdp]. Click here to download chart.\n\n**Inflation is projected to remain stable for the rest of 2024 and to decline further in 2025.** Annual inflation in the EU is forecasted to average at 2.7% in 2024 and then to decline to 2.2% in 2025. Food and non- energy industrial goods are the primary drivers behind this slowdown. ( ¹⁶ ). Yet, the expiration of policy measures introduced to address the energy crisis may continue to put pressure on the prices of consumer energy. Wage pressures remain elevated, with service prices having contributed little to disinflation.\n\nChart 1.3 Inflation decelerated in most Member States in 2023, but remained high despite decreasing prices of energy Left chart: All-items Harmonised Index of Consumer Prices (HICP), 2023; Right chart, Annual inflation rates for selected items, EU\n\n_Source:_ Eurostat [prc\\_hicp\\_aind; prc\\_hicp\\_manr].\n\nClick here to download chart.\n\n**Almost all Member States’ economies are projected to grow in 2024, with an average growth at the EU-level of 1.0%** (0.8% in the euro area). After recovering somewhat at the beginning of 2024 (+0.6% in Q1 and +0.8% in Q2, compared to same quarter of the previous year), GDP growth is expected to reach 1.6% in 2025 (1.4% in the euro area). The EU entered 2024 with a weaker economic outlook than initially expected. Nevertheless, the conditions remain in place for a gradual improvement in economic activity. Continued wage and employment growth are expected to sustain growth in households’ disposable income in 2024 and to continue in 2025 (albeit slightly slower), which would also boost consumption. The contribution of investment to real GDP growth is expected to increase more gradually and the contribution of EU's net external demand is predicted to become neutral. ( ¹⁷ ) Overall, domestic risks related to growth and inflation are broadly balanced in the EU.\n\n**Heightened geopolitical tensions and global policy uncertainty expose the EU economic outlook to several risks.** Russia's ongoing war of aggression against Ukraine and the Middle East conflict could add to upward pressure on prices and stress on supply chains, jeopardising production and trade. Energy commodities are also vulnerable to these downward risks. Persistence of inflation in the US may delay its interest rate cuts, resulting in tighter global financial conditions. In addition, uncertainties linked to China and US economic activity developments could weigh on the EU economy. Finally, climate change risks continue to be of primary concern,\n\n( ¹⁶ ) (European Commission, 2024a) ( ¹⁷ ) (European Commission, 2024a)\n\n\\-8\n\n\\-6\n\n\\-4\n\n\\-2\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\nIE\n\nEE\n\nFI\n\nLU\n\nHU\n\nAT\n\nLV\n\nLT\n\nDE\n\nSE\n\nCZ\n\nNL\n\nPL\n\nEU\n\nEA\n\nFR\n\nIT\n\nBE\n\nSK\n\nSI\n\nBG\n\nEL\n\nRO\n\nPT\n\nDK\n\nES\n\nCY\n\nHR\n\nMT\n\n2023 Series2\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\n16\n\n18\n\nHU\n\nCZ\n\nSK\n\nPL\n\nRO\n\nEE\n\nLV\n\nLT\n\nBG\n\nHR\n\nAT\n\nSI\n\nEU\n\nDE\n\nSE\n\nIT\n\nFR\n\nMT\n\nEA\n\nPT\n\nIE\n\nFI\n\nEL\n\nNL\n\nCY\n\nES\n\nDK\n\nLU\n\nBE\n\n%\n\nAverage 2023 Dec. 2023 Jul-24\n\n\\-20\n\n\\-10\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n2020-01\n\n2020-04\n\n2020-07\n\n2020-10\n\n2021-01\n\n2021-04\n\n2021-07\n\n2021-10\n\n2022-01\n\n2022-04\n\n2022-07\n\n2022-10\n\n2023-01\n\n2023-04\n\n2023-07\n\n2023-10\n\n2024-01\n\n2024-04\n\n2024-07\n\n_Annual inflation rate (%)_\n\nAll-items HICP\n\nFood, alcohol, and tobacco\n\nNon-energy industrial goods\n\nEnergy\n\nServices","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Employment and Social Developments in Europe 2024\n\n22\n\nand tightening monetary policy affected economic performance in 2023, with important consequences for households (see Section 4.1.).\n\nChart 1.2 Real GDP grew in most Member States in 2023 Real GDP growth, EU, 2022 (dots) and 2023 (bars) (% change on previous year)\n\n_Note:_ Ireland: modified domestic demand. 2022: data provisional for Belgium, Cyprus, Germany, Greece, Spain, France, Croatia, Portugal and Romania. 2023: data provisional for Belgium, Bulgaria, Cyprus, Germany, Greece, Spain, France, Croatia, Hungary, Luxembourg, the Netherlands, Portugal and Romania.\n\n_Source:_ Eurostat [nama\\_10\\_gdp]. Click here to download chart.\n\n**Inflation is projected to remain stable for the rest of 2024 and to decline further in 2025.** Annual inflation in the EU is forecasted to average at 2.7% in 2024 and then to decline to 2.2% in 2025. Food and non- energy industrial goods are the primary drivers behind this slowdown. ( ¹⁶ ). Yet, the expiration of policy measures introduced to address the energy crisis may continue to put pressure on the prices of consumer energy. Wage pressures remain elevated, with service prices having contributed little to disinflation.\n\nChart 1.3 Inflation decelerated in most Member States in 2023, but remained high despite decreasing prices of energy Left chart: All-items Harmonised Index of Consumer Prices (HICP), 2023; Right chart, Annual inflation rates for selected items, EU\n\n_Source:_ Eurostat [prc\\_hicp\\_aind; prc\\_hicp\\_manr].\n\nClick here to download chart.\n\n**Almost all Member States’ economies are projected to grow in 2024, with an average growth at the EU-level of 1.0%** (0.8% in the euro area). After recovering somewhat at the beginning of 2024 (+0.6% in Q1 and +0.8% in Q2, compared to same quarter of the previous year), GDP growth is expected to reach 1.6% in 2025 (1.4% in the euro area). The EU entered 2024 with a weaker economic outlook than initially expected. Nevertheless, the conditions remain in place for a gradual improvement in economic activity. Continued wage and employment growth are expected to sustain growth in households’ disposable income in 2024 and to continue in 2025 (albeit slightly slower), which would also boost consumption. The contribution of investment to real GDP growth is expected to increase more gradually and the contribution of EU's net external demand is predicted to become neutral. ( ¹⁷ ) Overall, domestic risks related to growth and inflation are broadly balanced in the EU.\n\n**Heightened geopolitical tensions and global policy uncertainty expose the EU economic outlook to several risks.** Russia's ongoing war of aggression against Ukraine and the Middle East conflict could add to upward pressure on prices and stress on supply chains, jeopardising production and trade. Energy commodities are also vulnerable to these downward risks. Persistence of inflation in the US may delay its interest rate cuts, resulting in tighter global financial conditions. In addition, uncertainties linked to China and US economic activity developments could weigh on the EU economy. Finally, climate change risks continue to be of primary concern,\n\n( ¹⁶ ) (European Commission, 2024a) ( ¹⁷ ) (European Commission, 2024a)\n\n\\-8\n\n\\-6\n\n\\-4\n\n\\-2\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\nIE\n\nEE\n\nFI\n\nLU\n\nHU\n\nAT\n\nLV\n\nLT\n\nDE\n\nSE\n\nCZ\n\nNL\n\nPL\n\nEU\n\nEA\n\nFR\n\nIT\n\nBE\n\nSK\n\nSI\n\nBG\n\nEL\n\nRO\n\nPT\n\nDK\n\nES\n\nCY\n\nHR\n\nMT\n\n2023 Series2\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\n16\n\n18\n\nHU\n\nCZ\n\nSK\n\nPL\n\nRO\n\nEE\n\nLV\n\nLT\n\nBG\n\nHR\n\nAT\n\nSI\n\nEU\n\nDE\n\nSE\n\nIT\n\nFR\n\nMT\n\nEA\n\nPT\n\nIE\n\nFI\n\nEL\n\nNL\n\nCY\n\nES\n\nDK\n\nLU\n\nBE\n\n%\n\nAverage 2023 Dec. 2023 Jul-24\n\n\\-20\n\n\\-10\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n2020-01\n\n2020-04\n\n2020-07\n\n2020-10\n\n2021-01\n\n2021-04\n\n2021-07\n\n2021-10\n\n2022-01\n\n2022-04\n\n2022-07\n\n2022-10\n\n2023-01\n\n2023-04\n\n2023-07\n\n2023-10\n\n2024-01\n\n2024-04\n\n2024-07\n\n_Annual inflation rate (%)_\n\nAll-items HICP\n\nFood, alcohol, and tobacco\n\nNon-energy industrial goods\n\nEnergy\n\nServices","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/employment_and_social_developments_in_europe_esde_2024_report_p20.pdf"]},"response":{"errors":[],"extract_id":"extract_4496c6725a603dc672e79097997d2fef","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Employment and Social Developments in Europe 2024\n\n22\n\nand tightening monetary policy affected economic performance in 2023, with important consequences for households (see Section 4.1.).\n\nChart 1.2 Real GDP grew in most Member States in 2023 Real GDP growth, EU, 2022 (dots) and 2023 (bars) (% change on previous year)\n\n_Note:_ Ireland: modified domestic demand. 2022: data provisional for Belgium, Cyprus, Germany, Greece, Spain, France, Croatia, Portugal and Romania. 2023: data provisional for Belgium, Bulgaria, Cyprus, Germany, Greece, Spain, France, Croatia, Hungary, Luxembourg, the Netherlands, Portugal and Romania.\n\n_Source:_ Eurostat [nama\\_10\\_gdp]. Click here to download chart.\n\n**Inflation is projected to remain stable for the rest of 2024 and to decline further in 2025.** Annual inflation in the EU is forecasted to average at 2.7% in 2024 and then to decline to 2.2% in 2025. Food and non- energy industrial goods are the primary drivers behind this slowdown. ( ¹⁶ ). Yet, the expiration of policy measures introduced to address the energy crisis may continue to put pressure on the prices of consumer energy. Wage pressures remain elevated, with service prices having contributed little to disinflation.\n\nChart 1.3 Inflation decelerated in most Member States in 2023, but remained high despite decreasing prices of energy Left chart: All-items Harmonised Index of Consumer Prices (HICP), 2023; Right chart, Annual inflation rates for selected items, EU\n\n_Source:_ Eurostat [prc\\_hicp\\_aind; prc\\_hicp\\_manr].\n\nClick here to download chart.\n\n**Almost all Member States’ economies are projected to grow in 2024, with an average growth at the EU-level of 1.0%** (0.8% in the euro area). After recovering somewhat at the beginning of 2024 (+0.6% in Q1 and +0.8% in Q2, compared to same quarter of the previous year), GDP growth is expected to reach 1.6% in 2025 (1.4% in the euro area). The EU entered 2024 with a weaker economic outlook than initially expected. Nevertheless, the conditions remain in place for a gradual improvement in economic activity. Continued wage and employment growth are expected to sustain growth in households’ disposable income in 2024 and to continue in 2025 (albeit slightly slower), which would also boost consumption. The contribution of investment to real GDP growth is expected to increase more gradually and the contribution of EU's net external demand is predicted to become neutral. ( ¹⁷ ) Overall, domestic risks related to growth and inflation are broadly balanced in the EU.\n\n**Heightened geopolitical tensions and global policy uncertainty expose the EU economic outlook to several risks.** Russia's ongoing war of aggression against Ukraine and the Middle East conflict could add to upward pressure on prices and stress on supply chains, jeopardising production and trade. Energy commodities are also vulnerable to these downward risks. Persistence of inflation in the US may delay its interest rate cuts, resulting in tighter global financial conditions. In addition, uncertainties linked to China and US economic activity developments could weigh on the EU economy. Finally, climate change risks continue to be of primary concern,\n\n( ¹⁶ ) (European Commission, 2024a) ( ¹⁷ ) (European Commission, 2024a)\n\n\\-8\n\n\\-6\n\n\\-4\n\n\\-2\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\nIE\n\nEE\n\nFI\n\nLU\n\nHU\n\nAT\n\nLV\n\nLT\n\nDE\n\nSE\n\nCZ\n\nNL\n\nPL\n\nEU\n\nEA\n\nFR\n\nIT\n\nBE\n\nSK\n\nSI\n\nBG\n\nEL\n\nRO\n\nPT\n\nDK\n\nES\n\nCY\n\nHR\n\nMT\n\n2023 Series2\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\n16\n\n18\n\nHU\n\nCZ\n\nSK\n\nPL\n\nRO\n\nEE\n\nLV\n\nLT\n\nBG\n\nHR\n\nAT\n\nSI\n\nEU\n\nDE\n\nSE\n\nIT\n\nFR\n\nMT\n\nEA\n\nPT\n\nIE\n\nFI\n\nEL\n\nNL\n\nCY\n\nES\n\nDK\n\nLU\n\nBE\n\n%\n\nAverage 2023 Dec. 2023 Jul-24\n\n\\-20\n\n\\-10\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n2020-01\n\n2020-04\n\n2020-07\n\n2020-10\n\n2021-01\n\n2021-04\n\n2021-07\n\n2021-10\n\n2022-01\n\n2022-04\n\n2022-07\n\n2022-10\n\n2023-01\n\n2023-04\n\n2023-07\n\n2023-10\n\n2024-01\n\n2024-04\n\n2024-07\n\n_Annual inflation rate (%)_\n\nAll-items HICP\n\nFood, alcohol, and tobacco\n\nNon-energy industrial goods\n\nEnergy\n\nServices"],"full_content":"Employment and Social Developments in Europe 2024\n\n22\n\nand tightening monetary policy affected economic performance in 2023, with important consequences for households (see Section 4.1.).\n\nChart 1.2 Real GDP grew in most Member States in 2023 Real GDP growth, EU, 2022 (dots) and 2023 (bars) (% change on previous year)\n\n_Note:_ Ireland: modified domestic demand. 2022: data provisional for Belgium, Cyprus, Germany, Greece, Spain, France, Croatia, Portugal and Romania. 2023: data provisional for Belgium, Bulgaria, Cyprus, Germany, Greece, Spain, France, Croatia, Hungary, Luxembourg, the Netherlands, Portugal and Romania.\n\n_Source:_ Eurostat [nama\\_10\\_gdp]. Click here to download chart.\n\n**Inflation is projected to remain stable for the rest of 2024 and to decline further in 2025.** Annual inflation in the EU is forecasted to average at 2.7% in 2024 and then to decline to 2.2% in 2025. Food and non- energy industrial goods are the primary drivers behind this slowdown. ( ¹⁶ ). Yet, the expiration of policy measures introduced to address the energy crisis may continue to put pressure on the prices of consumer energy. Wage pressures remain elevated, with service prices having contributed little to disinflation.\n\nChart 1.3 Inflation decelerated in most Member States in 2023, but remained high despite decreasing prices of energy Left chart: All-items Harmonised Index of Consumer Prices (HICP), 2023; Right chart, Annual inflation rates for selected items, EU\n\n_Source:_ Eurostat [prc\\_hicp\\_aind; prc\\_hicp\\_manr].\n\nClick here to download chart.\n\n**Almost all Member States’ economies are projected to grow in 2024, with an average growth at the EU-level of 1.0%** (0.8% in the euro area). After recovering somewhat at the beginning of 2024 (+0.6% in Q1 and +0.8% in Q2, compared to same quarter of the previous year), GDP growth is expected to reach 1.6% in 2025 (1.4% in the euro area). The EU entered 2024 with a weaker economic outlook than initially expected. Nevertheless, the conditions remain in place for a gradual improvement in economic activity. Continued wage and employment growth are expected to sustain growth in households’ disposable income in 2024 and to continue in 2025 (albeit slightly slower), which would also boost consumption. The contribution of investment to real GDP growth is expected to increase more gradually and the contribution of EU's net external demand is predicted to become neutral. ( ¹⁷ ) Overall, domestic risks related to growth and inflation are broadly balanced in the EU.\n\n**Heightened geopolitical tensions and global policy uncertainty expose the EU economic outlook to several risks.** Russia's ongoing war of aggression against Ukraine and the Middle East conflict could add to upward pressure on prices and stress on supply chains, jeopardising production and trade. Energy commodities are also vulnerable to these downward risks. Persistence of inflation in the US may delay its interest rate cuts, resulting in tighter global financial conditions. In addition, uncertainties linked to China and US economic activity developments could weigh on the EU economy. Finally, climate change risks continue to be of primary concern,\n\n( ¹⁶ ) (European Commission, 2024a) ( ¹⁷ ) (European Commission, 2024a)\n\n\\-8\n\n\\-6\n\n\\-4\n\n\\-2\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\nIE\n\nEE\n\nFI\n\nLU\n\nHU\n\nAT\n\nLV\n\nLT\n\nDE\n\nSE\n\nCZ\n\nNL\n\nPL\n\nEU\n\nEA\n\nFR\n\nIT\n\nBE\n\nSK\n\nSI\n\nBG\n\nEL\n\nRO\n\nPT\n\nDK\n\nES\n\nCY\n\nHR\n\nMT\n\n2023 Series2\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\n16\n\n18\n\nHU\n\nCZ\n\nSK\n\nPL\n\nRO\n\nEE\n\nLV\n\nLT\n\nBG\n\nHR\n\nAT\n\nSI\n\nEU\n\nDE\n\nSE\n\nIT\n\nFR\n\nMT\n\nEA\n\nPT\n\nIE\n\nFI\n\nEL\n\nNL\n\nCY\n\nES\n\nDK\n\nLU\n\nBE\n\n%\n\nAverage 2023 Dec. 2023 Jul-24\n\n\\-20\n\n\\-10\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n2020-01\n\n2020-04\n\n2020-07\n\n2020-10\n\n2021-01\n\n2021-04\n\n2021-07\n\n2021-10\n\n2022-01\n\n2022-04\n\n2022-07\n\n2022-10\n\n2023-01\n\n2023-04\n\n2023-07\n\n2023-10\n\n2024-01\n\n2024-04\n\n2024-07\n\n_Annual inflation rate (%)_\n\nAll-items HICP\n\nFood, alcohol, and tobacco\n\nNon-energy industrial goods\n\nEnergy\n\nServices","publish_date":null,"title":"and tightening monetary policy affected economic performance in 2023, with important consequences for households (see Section 4.1.). Real GDP grew","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/employment_and_social_developments_in_europe_esde_2024_report_p20.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/employment_and_social_developments_in_europe_esde_2024_report_p20.pdf"},"request":{"example_id":"chart/employment_and_social_developments_in_europe_esde_2024_report_p20","product_type":"parse","source_file_path":"employment_and_social_developments_in_europe_esde_2024_report_p20.pdf"},"started_at":"2026-11-04T19:45:06.133854"} +{"completed_at":"2026-11-04T19:45:09.157808","latency_in_ms":1502,"output":{"example_id":"chart/employment_and_social_developments_in_europe_esde_2024_report_p21","grounded_pages":[],"layout_pages":[],"markdown":"Chapter 1: Main employment and social developments\n\n23\n\nwith severe potential costs for natural capital, infrastructure and economic activity. ( ¹⁸ ) Financing the digital infrastructure may also prove a challenge.\n\n### 3\\. LABOUR MARKET DEVELOPMENTS\n\n### 3\\.1. Employment trends\n\n**Despite ongoing economic challenges, labour markets in the EU remain robust.** Alongside modest GDP growth of 0.4% in 2023, employment increased by 1.2% in the EU and 1.4% in the euro area, with 216.5 million people employed in the EU and 168.7 million in the euro area. With the exception of Romania (-0.9% compared to 2022), all countries experienced an increase in employment. The countries with the highest percentage increase were Malta (+6.7 %), Ireland (+3.5%), Estonia (+3.2%), Spain (+3.2%) and Luxembourg (+2.2%). All other Member States recorded a percentage growth below 2.0%. Notwithstanding a rise of 1.0% in total hours worked in the EU (+1.3% in the euro area) in 2023 compared to 2022, there was a slight decrease in the number of hours worked per employed person (-0.2% in the EU and -0.1% in the euro area), still below the pre-COVID-19 pandemic levels in 2019. This downward trend, which has been protracted for the last two decades, may reflect the introduction of more efficient and productive technologies, including the acceleration of digitalisation, as well as changes in work attitudes. After recovering in 2021 and 2022, labour productivity decreased again in 2023 compared to the previous year (-0.8% per employed person and -0.6% per hour), reflecting a combination of higher labour supply and fewer hours worked per employed person, ( ¹⁹ ) as well as low total factor productivity growth.\n\n**In 2023, the sectors that contributed most to employment growth were trade, transport, accommodation and public administration, defence, education, human health and social work activities.** The numbers of people employed in these two sectors increased by 0.8 million and 0.7 million, respectively. In relative terms, employment grew most in information and communication (+4.2%), indicating the elevated need for specialised skills. It rose by 0.9% in construction and 0.2% in industry, and declined by 1.1% in agriculture, forestry and fishing. The number of self-employed people increased by 1.0% and employee numbers grew by 1.3%.\n\nChart 1.4 Number of hours worked per person employed stagnated in 2023 despite a resilient labour market Left chart: Number of people employed, and number of hours worked (2012=100); Right chart: Headcount employment (% change on previous year), 2012-2025\n\n_Note:_ EA = euro area. European Commission (DG ECFIN) 2023 Spring Forecast in the shaded area.\n\n_Source:_ Eurostat [nama\\_10\\_a10\\_e, nama\\_10\\_pe, naida\\_10\\_pe], DG ECFIN Forecast. Click here to download chart.\n\n**According to the European Commission Spring 2024 Economic Forecast, employment should expand by 0.6% in the EU in 2024 and by 0.4% in 2025, driven by a positive carry-over effect of gains during 2023.** Such expansion could be limited by the fact that employers are retaining their workers in a context of labour shortages (so-called labour hoarding). ( ²⁰ ) In July 2024, the new indicator of labour hoarding developed by the Commission increased slightly compared to the previous month (+0.4 pp to 10.8%, three-month moving\n\n( ¹⁸ ) (European Commission, 2024a) ( ¹⁹ ) (European Commission, 2024a) ( ²⁰ ) ‘ _Labour hoarding can be defined as “that part of labour input which is not fully utilised by a company during its production process at any given point in time”_ (ECB, 2003). _Typically, labour hoarding, implying under-utilisation of the workforce, occurs in periods of slack or downturn in economic activity. The rationale for companies not to lay off (redundant) employees in such periods is that (i) dismissing workers usually involves costs, e.g. severance payments, and (ii) recruiting workers once economic activity recovers also entails costs (screening the labour market for candidates, training them, etc.)_ ’. When economic activity picks up again, companies may not hire new workers immediately, but rather rely on the underutilised labour already in the company. The European Commission developed an indicator of labour hoarding, which measures the percentage of managers expecting their firm’s output to decrease, but employment to remain stable or increase, based on the Joint Harmonised EU Programme of Business and Consumer Surveys (BCS) (European Commission, 2023c).\n\n90\n\n95\n\n100\n\n105\n\n110\n\n115\n\n2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023\n\nEU - employment EA - employment EU - hours worked per person employed EA - hours worked per person employed\n\n\\-8\n\n\\-6\n\n\\-4\n\n\\-2\n\n0\n\n2\n\n4\n\n6\n\n2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025\n\nEU EA UK US Japan\n\n_Forecast_","pages":[{"line_numbers":[],"markdown":"Chapter 1: Main employment and social developments\n\n23\n\nwith severe potential costs for natural capital, infrastructure and economic activity. ( ¹⁸ ) Financing the digital infrastructure may also prove a challenge.\n\n### 3\\. LABOUR MARKET DEVELOPMENTS\n\n### 3\\.1. Employment trends\n\n**Despite ongoing economic challenges, labour markets in the EU remain robust.** Alongside modest GDP growth of 0.4% in 2023, employment increased by 1.2% in the EU and 1.4% in the euro area, with 216.5 million people employed in the EU and 168.7 million in the euro area. With the exception of Romania (-0.9% compared to 2022), all countries experienced an increase in employment. The countries with the highest percentage increase were Malta (+6.7 %), Ireland (+3.5%), Estonia (+3.2%), Spain (+3.2%) and Luxembourg (+2.2%). All other Member States recorded a percentage growth below 2.0%. Notwithstanding a rise of 1.0% in total hours worked in the EU (+1.3% in the euro area) in 2023 compared to 2022, there was a slight decrease in the number of hours worked per employed person (-0.2% in the EU and -0.1% in the euro area), still below the pre-COVID-19 pandemic levels in 2019. This downward trend, which has been protracted for the last two decades, may reflect the introduction of more efficient and productive technologies, including the acceleration of digitalisation, as well as changes in work attitudes. After recovering in 2021 and 2022, labour productivity decreased again in 2023 compared to the previous year (-0.8% per employed person and -0.6% per hour), reflecting a combination of higher labour supply and fewer hours worked per employed person, ( ¹⁹ ) as well as low total factor productivity growth.\n\n**In 2023, the sectors that contributed most to employment growth were trade, transport, accommodation and public administration, defence, education, human health and social work activities.** The numbers of people employed in these two sectors increased by 0.8 million and 0.7 million, respectively. In relative terms, employment grew most in information and communication (+4.2%), indicating the elevated need for specialised skills. It rose by 0.9% in construction and 0.2% in industry, and declined by 1.1% in agriculture, forestry and fishing. The number of self-employed people increased by 1.0% and employee numbers grew by 1.3%.\n\nChart 1.4 Number of hours worked per person employed stagnated in 2023 despite a resilient labour market Left chart: Number of people employed, and number of hours worked (2012=100); Right chart: Headcount employment (% change on previous year), 2012-2025\n\n_Note:_ EA = euro area. European Commission (DG ECFIN) 2023 Spring Forecast in the shaded area.\n\n_Source:_ Eurostat [nama\\_10\\_a10\\_e, nama\\_10\\_pe, naida\\_10\\_pe], DG ECFIN Forecast. Click here to download chart.\n\n**According to the European Commission Spring 2024 Economic Forecast, employment should expand by 0.6% in the EU in 2024 and by 0.4% in 2025, driven by a positive carry-over effect of gains during 2023.** Such expansion could be limited by the fact that employers are retaining their workers in a context of labour shortages (so-called labour hoarding). ( ²⁰ ) In July 2024, the new indicator of labour hoarding developed by the Commission increased slightly compared to the previous month (+0.4 pp to 10.8%, three-month moving\n\n( ¹⁸ ) (European Commission, 2024a) ( ¹⁹ ) (European Commission, 2024a) ( ²⁰ ) ‘ _Labour hoarding can be defined as “that part of labour input which is not fully utilised by a company during its production process at any given point in time”_ (ECB, 2003). _Typically, labour hoarding, implying under-utilisation of the workforce, occurs in periods of slack or downturn in economic activity. The rationale for companies not to lay off (redundant) employees in such periods is that (i) dismissing workers usually involves costs, e.g. severance payments, and (ii) recruiting workers once economic activity recovers also entails costs (screening the labour market for candidates, training them, etc.)_ ’. When economic activity picks up again, companies may not hire new workers immediately, but rather rely on the underutilised labour already in the company. The European Commission developed an indicator of labour hoarding, which measures the percentage of managers expecting their firm’s output to decrease, but employment to remain stable or increase, based on the Joint Harmonised EU Programme of Business and Consumer Surveys (BCS) (European Commission, 2023c).\n\n90\n\n95\n\n100\n\n105\n\n110\n\n115\n\n2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023\n\nEU - employment EA - employment EU - hours worked per person employed EA - hours worked per person employed\n\n\\-8\n\n\\-6\n\n\\-4\n\n\\-2\n\n0\n\n2\n\n4\n\n6\n\n2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025\n\nEU EA UK US Japan\n\n_Forecast_","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Chapter 1: Main employment and social developments\n\n23\n\nwith severe potential costs for natural capital, infrastructure and economic activity. ( ¹⁸ ) Financing the digital infrastructure may also prove a challenge.\n\n### 3\\. LABOUR MARKET DEVELOPMENTS\n\n### 3\\.1. Employment trends\n\n**Despite ongoing economic challenges, labour markets in the EU remain robust.** Alongside modest GDP growth of 0.4% in 2023, employment increased by 1.2% in the EU and 1.4% in the euro area, with 216.5 million people employed in the EU and 168.7 million in the euro area. With the exception of Romania (-0.9% compared to 2022), all countries experienced an increase in employment. The countries with the highest percentage increase were Malta (+6.7 %), Ireland (+3.5%), Estonia (+3.2%), Spain (+3.2%) and Luxembourg (+2.2%). All other Member States recorded a percentage growth below 2.0%. Notwithstanding a rise of 1.0% in total hours worked in the EU (+1.3% in the euro area) in 2023 compared to 2022, there was a slight decrease in the number of hours worked per employed person (-0.2% in the EU and -0.1% in the euro area), still below the pre-COVID-19 pandemic levels in 2019. This downward trend, which has been protracted for the last two decades, may reflect the introduction of more efficient and productive technologies, including the acceleration of digitalisation, as well as changes in work attitudes. After recovering in 2021 and 2022, labour productivity decreased again in 2023 compared to the previous year (-0.8% per employed person and -0.6% per hour), reflecting a combination of higher labour supply and fewer hours worked per employed person, ( ¹⁹ ) as well as low total factor productivity growth.\n\n**In 2023, the sectors that contributed most to employment growth were trade, transport, accommodation and public administration, defence, education, human health and social work activities.** The numbers of people employed in these two sectors increased by 0.8 million and 0.7 million, respectively. In relative terms, employment grew most in information and communication (+4.2%), indicating the elevated need for specialised skills. It rose by 0.9% in construction and 0.2% in industry, and declined by 1.1% in agriculture, forestry and fishing. The number of self-employed people increased by 1.0% and employee numbers grew by 1.3%.\n\nChart 1.4 Number of hours worked per person employed stagnated in 2023 despite a resilient labour market Left chart: Number of people employed, and number of hours worked (2012=100); Right chart: Headcount employment (% change on previous year), 2012-2025\n\n_Note:_ EA = euro area. European Commission (DG ECFIN) 2023 Spring Forecast in the shaded area.\n\n_Source:_ Eurostat [nama\\_10\\_a10\\_e, nama\\_10\\_pe, naida\\_10\\_pe], DG ECFIN Forecast. Click here to download chart.\n\n**According to the European Commission Spring 2024 Economic Forecast, employment should expand by 0.6% in the EU in 2024 and by 0.4% in 2025, driven by a positive carry-over effect of gains during 2023.** Such expansion could be limited by the fact that employers are retaining their workers in a context of labour shortages (so-called labour hoarding). ( ²⁰ ) In July 2024, the new indicator of labour hoarding developed by the Commission increased slightly compared to the previous month (+0.4 pp to 10.8%, three-month moving\n\n( ¹⁸ ) (European Commission, 2024a) ( ¹⁹ ) (European Commission, 2024a) ( ²⁰ ) ‘ _Labour hoarding can be defined as “that part of labour input which is not fully utilised by a company during its production process at any given point in time”_ (ECB, 2003). _Typically, labour hoarding, implying under-utilisation of the workforce, occurs in periods of slack or downturn in economic activity. The rationale for companies not to lay off (redundant) employees in such periods is that (i) dismissing workers usually involves costs, e.g. severance payments, and (ii) recruiting workers once economic activity recovers also entails costs (screening the labour market for candidates, training them, etc.)_ ’. When economic activity picks up again, companies may not hire new workers immediately, but rather rely on the underutilised labour already in the company. The European Commission developed an indicator of labour hoarding, which measures the percentage of managers expecting their firm’s output to decrease, but employment to remain stable or increase, based on the Joint Harmonised EU Programme of Business and Consumer Surveys (BCS) (European Commission, 2023c).\n\n90\n\n95\n\n100\n\n105\n\n110\n\n115\n\n2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023\n\nEU - employment EA - employment EU - hours worked per person employed EA - hours worked per person employed\n\n\\-8\n\n\\-6\n\n\\-4\n\n\\-2\n\n0\n\n2\n\n4\n\n6\n\n2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025\n\nEU EA UK US Japan\n\n_Forecast_","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/employment_and_social_developments_in_europe_esde_2024_report_p21.pdf"]},"response":{"errors":[],"extract_id":"extract_5b6f5b130b8eede51307c08a69cf9234","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Chapter 1: Main employment and social developments\n\n23\n\nwith severe potential costs for natural capital, infrastructure and economic activity. ( ¹⁸ ) Financing the digital infrastructure may also prove a challenge.\n\n### 3\\. LABOUR MARKET DEVELOPMENTS\n\n### 3\\.1. Employment trends\n\n**Despite ongoing economic challenges, labour markets in the EU remain robust.** Alongside modest GDP growth of 0.4% in 2023, employment increased by 1.2% in the EU and 1.4% in the euro area, with 216.5 million people employed in the EU and 168.7 million in the euro area. With the exception of Romania (-0.9% compared to 2022), all countries experienced an increase in employment. The countries with the highest percentage increase were Malta (+6.7 %), Ireland (+3.5%), Estonia (+3.2%), Spain (+3.2%) and Luxembourg (+2.2%). All other Member States recorded a percentage growth below 2.0%. Notwithstanding a rise of 1.0% in total hours worked in the EU (+1.3% in the euro area) in 2023 compared to 2022, there was a slight decrease in the number of hours worked per employed person (-0.2% in the EU and -0.1% in the euro area), still below the pre-COVID-19 pandemic levels in 2019. This downward trend, which has been protracted for the last two decades, may reflect the introduction of more efficient and productive technologies, including the acceleration of digitalisation, as well as changes in work attitudes. After recovering in 2021 and 2022, labour productivity decreased again in 2023 compared to the previous year (-0.8% per employed person and -0.6% per hour), reflecting a combination of higher labour supply and fewer hours worked per employed person, ( ¹⁹ ) as well as low total factor productivity growth.\n\n**In 2023, the sectors that contributed most to employment growth were trade, transport, accommodation and public administration, defence, education, human health and social work activities.** The numbers of people employed in these two sectors increased by 0.8 million and 0.7 million, respectively. In relative terms, employment grew most in information and communication (+4.2%), indicating the elevated need for specialised skills. It rose by 0.9% in construction and 0.2% in industry, and declined by 1.1% in agriculture, forestry and fishing. The number of self-employed people increased by 1.0% and employee numbers grew by 1.3%.\n\nChart 1.4 Number of hours worked per person employed stagnated in 2023 despite a resilient labour market Left chart: Number of people employed, and number of hours worked (2012=100); Right chart: Headcount employment (% change on previous year), 2012-2025\n\n_Note:_ EA = euro area. European Commission (DG ECFIN) 2023 Spring Forecast in the shaded area.\n\n_Source:_ Eurostat [nama\\_10\\_a10\\_e, nama\\_10\\_pe, naida\\_10\\_pe], DG ECFIN Forecast. Click here to download chart.\n\n**According to the European Commission Spring 2024 Economic Forecast, employment should expand by 0.6% in the EU in 2024 and by 0.4% in 2025, driven by a positive carry-over effect of gains during 2023.** Such expansion could be limited by the fact that employers are retaining their workers in a context of labour shortages (so-called labour hoarding). ( ²⁰ ) In July 2024, the new indicator of labour hoarding developed by the Commission increased slightly compared to the previous month (+0.4 pp to 10.8%, three-month moving\n\n( ¹⁸ ) (European Commission, 2024a) ( ¹⁹ ) (European Commission, 2024a) ( ²⁰ ) ‘ _Labour hoarding can be defined as “that part of labour input which is not fully utilised by a company during its production process at any given point in time”_ (ECB, 2003). _Typically, labour hoarding, implying under-utilisation of the workforce, occurs in periods of slack or downturn in economic activity. The rationale for companies not to lay off (redundant) employees in such periods is that (i) dismissing workers usually involves costs, e.g. severance payments, and (ii) recruiting workers once economic activity recovers also entails costs (screening the labour market for candidates, training them, etc.)_ ’. When economic activity picks up again, companies may not hire new workers immediately, but rather rely on the underutilised labour already in the company. The European Commission developed an indicator of labour hoarding, which measures the percentage of managers expecting their firm’s output to decrease, but employment to remain stable or increase, based on the Joint Harmonised EU Programme of Business and Consumer Surveys (BCS) (European Commission, 2023c).\n\n90\n\n95\n\n100\n\n105\n\n110\n\n115\n\n2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023\n\nEU - employment EA - employment EU - hours worked per person employed EA - hours worked per person employed\n\n\\-8\n\n\\-6\n\n\\-4\n\n\\-2\n\n0\n\n2\n\n4\n\n6\n\n2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025\n\nEU EA UK US Japan\n\n_Forecast_"],"full_content":"Chapter 1: Main employment and social developments\n\n23\n\nwith severe potential costs for natural capital, infrastructure and economic activity. ( ¹⁸ ) Financing the digital infrastructure may also prove a challenge.\n\n### 3\\. LABOUR MARKET DEVELOPMENTS\n\n### 3\\.1. Employment trends\n\n**Despite ongoing economic challenges, labour markets in the EU remain robust.** Alongside modest GDP growth of 0.4% in 2023, employment increased by 1.2% in the EU and 1.4% in the euro area, with 216.5 million people employed in the EU and 168.7 million in the euro area. With the exception of Romania (-0.9% compared to 2022), all countries experienced an increase in employment. The countries with the highest percentage increase were Malta (+6.7 %), Ireland (+3.5%), Estonia (+3.2%), Spain (+3.2%) and Luxembourg (+2.2%). All other Member States recorded a percentage growth below 2.0%. Notwithstanding a rise of 1.0% in total hours worked in the EU (+1.3% in the euro area) in 2023 compared to 2022, there was a slight decrease in the number of hours worked per employed person (-0.2% in the EU and -0.1% in the euro area), still below the pre-COVID-19 pandemic levels in 2019. This downward trend, which has been protracted for the last two decades, may reflect the introduction of more efficient and productive technologies, including the acceleration of digitalisation, as well as changes in work attitudes. After recovering in 2021 and 2022, labour productivity decreased again in 2023 compared to the previous year (-0.8% per employed person and -0.6% per hour), reflecting a combination of higher labour supply and fewer hours worked per employed person, ( ¹⁹ ) as well as low total factor productivity growth.\n\n**In 2023, the sectors that contributed most to employment growth were trade, transport, accommodation and public administration, defence, education, human health and social work activities.** The numbers of people employed in these two sectors increased by 0.8 million and 0.7 million, respectively. In relative terms, employment grew most in information and communication (+4.2%), indicating the elevated need for specialised skills. It rose by 0.9% in construction and 0.2% in industry, and declined by 1.1% in agriculture, forestry and fishing. The number of self-employed people increased by 1.0% and employee numbers grew by 1.3%.\n\nChart 1.4 Number of hours worked per person employed stagnated in 2023 despite a resilient labour market Left chart: Number of people employed, and number of hours worked (2012=100); Right chart: Headcount employment (% change on previous year), 2012-2025\n\n_Note:_ EA = euro area. European Commission (DG ECFIN) 2023 Spring Forecast in the shaded area.\n\n_Source:_ Eurostat [nama\\_10\\_a10\\_e, nama\\_10\\_pe, naida\\_10\\_pe], DG ECFIN Forecast. Click here to download chart.\n\n**According to the European Commission Spring 2024 Economic Forecast, employment should expand by 0.6% in the EU in 2024 and by 0.4% in 2025, driven by a positive carry-over effect of gains during 2023.** Such expansion could be limited by the fact that employers are retaining their workers in a context of labour shortages (so-called labour hoarding). ( ²⁰ ) In July 2024, the new indicator of labour hoarding developed by the Commission increased slightly compared to the previous month (+0.4 pp to 10.8%, three-month moving\n\n( ¹⁸ ) (European Commission, 2024a) ( ¹⁹ ) (European Commission, 2024a) ( ²⁰ ) ‘ _Labour hoarding can be defined as “that part of labour input which is not fully utilised by a company during its production process at any given point in time”_ (ECB, 2003). _Typically, labour hoarding, implying under-utilisation of the workforce, occurs in periods of slack or downturn in economic activity. The rationale for companies not to lay off (redundant) employees in such periods is that (i) dismissing workers usually involves costs, e.g. severance payments, and (ii) recruiting workers once economic activity recovers also entails costs (screening the labour market for candidates, training them, etc.)_ ’. When economic activity picks up again, companies may not hire new workers immediately, but rather rely on the underutilised labour already in the company. The European Commission developed an indicator of labour hoarding, which measures the percentage of managers expecting their firm’s output to decrease, but employment to remain stable or increase, based on the Joint Harmonised EU Programme of Business and Consumer Surveys (BCS) (European Commission, 2023c).\n\n90\n\n95\n\n100\n\n105\n\n110\n\n115\n\n2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023\n\nEU - employment EA - employment EU - hours worked per person employed EA - hours worked per person employed\n\n\\-8\n\n\\-6\n\n\\-4\n\n\\-2\n\n0\n\n2\n\n4\n\n6\n\n2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025\n\nEU EA UK US Japan\n\n_Forecast_","publish_date":null,"title":"3. L","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/employment_and_social_developments_in_europe_esde_2024_report_p21.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/employment_and_social_developments_in_europe_esde_2024_report_p21.pdf"},"request":{"example_id":"chart/employment_and_social_developments_in_europe_esde_2024_report_p21","product_type":"parse","source_file_path":"employment_and_social_developments_in_europe_esde_2024_report_p21.pdf"},"started_at":"2026-11-04T19:45:07.655416"} +{"completed_at":"2026-11-04T19:45:11.715898","latency_in_ms":2520,"output":{"example_id":"chart/employment_and_social_developments_in_europe_esde_2024_report_p22","grounded_pages":[],"layout_pages":[],"markdown":"Employment and Social Developments in Europe 2024\n\n24\n\naverage), slightly above its long-term average (9.7%) and pre-COVID-19 pandemic levels (7.5%, three month moving average in February 2020) (Chart 1.4). The level was higher in retail (15.5%) and construction (15.3%) and, than in industry (9.7%) and services (8.2%). ( ²¹ )\n\n**In 2023, the employment rate for individuals aged 20-64 reached the highest recorded levels in the EU.** At 75.3%, it corresponded to an increase of 0.7 pp in both the EU and euro area (74.7%) compared to 2022 (Chart 1.5). The 2023 EU employment rate is getting closer to the EU Porto target, which aims to achieve at least 78% of people aged 20-64 in employment in the EU by 2030, with national targets specific to each Member State. The largest increases were recorded in Malta (+1.6 pp), Italy (+1.5 pp), Poland and Spain (both +1.2 pp), while the employment rate declined in Lithuania (-0.5 pp), Slovenia (-0.4 pp), Denmark (-0.3 pp), Finland (-0.2 pp) and Austria (-0.1 pp). Five Member States (Estonia, Ireland, the Netherlands, Slovakia, Sweden) are already above their national target (Chart 1.5).\n\nChart 1.5 Employment rates reached historic levels in 2023, but growth is slowing Employment rate (% of people aged 20-64)\n\n_Source:_ Eurostat [lfsi\\_emp\\_a].\n\nClick here to download chart.\n\n**The gender employment gap (** **22** **) continued to decrease in 2023, albeit slowly, while the employment rate for women surpassed 70% for the first time.** Within the EU, the difference in employment between women and men stood at 10.2 pp, a decline of 0.5 pp from 2022 and 1.1 pp from 2019. The employment rate for women rose to 70.2%, while the rate for men rose to 80.4%. Further improvements are necessary to meet the ambition set out in the action plan to implement the European Pillar of Social Rights of at least halving the gender employment gap by 2030 compared to 2019 (11.3 pp).\n\n**The employment rate rose for all age groups in 2023, but at a slower pace than in previous years.** On a yearly basis, the rate grew by 0.4 pp (to 35.2%) for young workers (aged 15-24), by 0.5 pp (to 82.2%) for prime age workers (aged 25-54) and by 1.7 pp (to 63.9%) for older workers (aged 55-64) (Chart 1.6). Despite the COVID-19 pandemic, over the period 2019-2023, the employment rate for younger workers increased by 1.7 pp. Potentially as a result of several national pension reforms and the entry of more active cohorts in these age groups, the employment rate during that same period rose by 5.3 pp for older workers and by 2.0 pp for prime- age workers during the same period. The gender employment gap was wider for older workers (12.0 pp) than for prime-age workers (10.1 pp) and young workers (4.3 pp).\n\n**In 2023, the employment rate rose for people aged 25-54 with lower and medium levels of education.** The increase was most notable among those with lower education (+0.8 pp, to 64.1%) while it stagnated for those with tertiary education. Despite the implied reduction in the gap, rates remained highest for those with medium-level vocational education (+0.4 pp, to 84.9%) and those with tertiary education (89.7%), who are typically more likely to have the skills required in the labour market.\n\n**High employment disparities exist between the general population and people in vulnerable situations, including non-EU citizens or people with disabilities.** The disability employment gap stood at 21.5 pp in 2023. In 2023, the employment rate was 76.2% for national workers aged 20-64 (+0.8 pp compared to 2022), 77.6% for EU mobile workers (+0.6 pp), and 63.0% for non-EU workers (+1.2 pp).\n\n( ²¹ ) See Business and Consumer Survey here. ( ²² ) Difference between the employment rate of women and men aged 20-64.\n\n66\n\n68\n\n70\n\n72\n\n74\n\n76\n\n78\n\n80\n\n2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023\n\nEU Euro area EU 2030 target\n\n40\n\n45\n\n50\n\n55\n\n60\n\n65\n\n70\n\n75\n\n80\n\n85\n\n90\n\nIT\n\nEL\n\nRO\n\nES\n\nHR\n\nBE\n\nFR\n\nLU\n\nEU\n\nBG\n\nAT\n\nSK\n\nLV\n\nSI\n\nPL\n\nPT\n\nFI\n\nLT\n\nCY\n\nIE\n\nDK\n\nHU\n\nDE\n\nMT\n\nCZ\n\nEE\n\nSE\n\nNL\n\n2022 2023 2012 Target","pages":[{"line_numbers":[],"markdown":"Employment and Social Developments in Europe 2024\n\n24\n\naverage), slightly above its long-term average (9.7%) and pre-COVID-19 pandemic levels (7.5%, three month moving average in February 2020) (Chart 1.4). The level was higher in retail (15.5%) and construction (15.3%) and, than in industry (9.7%) and services (8.2%). ( ²¹ )\n\n**In 2023, the employment rate for individuals aged 20-64 reached the highest recorded levels in the EU.** At 75.3%, it corresponded to an increase of 0.7 pp in both the EU and euro area (74.7%) compared to 2022 (Chart 1.5). The 2023 EU employment rate is getting closer to the EU Porto target, which aims to achieve at least 78% of people aged 20-64 in employment in the EU by 2030, with national targets specific to each Member State. The largest increases were recorded in Malta (+1.6 pp), Italy (+1.5 pp), Poland and Spain (both +1.2 pp), while the employment rate declined in Lithuania (-0.5 pp), Slovenia (-0.4 pp), Denmark (-0.3 pp), Finland (-0.2 pp) and Austria (-0.1 pp). Five Member States (Estonia, Ireland, the Netherlands, Slovakia, Sweden) are already above their national target (Chart 1.5).\n\nChart 1.5 Employment rates reached historic levels in 2023, but growth is slowing Employment rate (% of people aged 20-64)\n\n_Source:_ Eurostat [lfsi\\_emp\\_a].\n\nClick here to download chart.\n\n**The gender employment gap (** **22** **) continued to decrease in 2023, albeit slowly, while the employment rate for women surpassed 70% for the first time.** Within the EU, the difference in employment between women and men stood at 10.2 pp, a decline of 0.5 pp from 2022 and 1.1 pp from 2019. The employment rate for women rose to 70.2%, while the rate for men rose to 80.4%. Further improvements are necessary to meet the ambition set out in the action plan to implement the European Pillar of Social Rights of at least halving the gender employment gap by 2030 compared to 2019 (11.3 pp).\n\n**The employment rate rose for all age groups in 2023, but at a slower pace than in previous years.** On a yearly basis, the rate grew by 0.4 pp (to 35.2%) for young workers (aged 15-24), by 0.5 pp (to 82.2%) for prime age workers (aged 25-54) and by 1.7 pp (to 63.9%) for older workers (aged 55-64) (Chart 1.6). Despite the COVID-19 pandemic, over the period 2019-2023, the employment rate for younger workers increased by 1.7 pp. Potentially as a result of several national pension reforms and the entry of more active cohorts in these age groups, the employment rate during that same period rose by 5.3 pp for older workers and by 2.0 pp for prime- age workers during the same period. The gender employment gap was wider for older workers (12.0 pp) than for prime-age workers (10.1 pp) and young workers (4.3 pp).\n\n**In 2023, the employment rate rose for people aged 25-54 with lower and medium levels of education.** The increase was most notable among those with lower education (+0.8 pp, to 64.1%) while it stagnated for those with tertiary education. Despite the implied reduction in the gap, rates remained highest for those with medium-level vocational education (+0.4 pp, to 84.9%) and those with tertiary education (89.7%), who are typically more likely to have the skills required in the labour market.\n\n**High employment disparities exist between the general population and people in vulnerable situations, including non-EU citizens or people with disabilities.** The disability employment gap stood at 21.5 pp in 2023. In 2023, the employment rate was 76.2% for national workers aged 20-64 (+0.8 pp compared to 2022), 77.6% for EU mobile workers (+0.6 pp), and 63.0% for non-EU workers (+1.2 pp).\n\n( ²¹ ) See Business and Consumer Survey here. ( ²² ) Difference between the employment rate of women and men aged 20-64.\n\n66\n\n68\n\n70\n\n72\n\n74\n\n76\n\n78\n\n80\n\n2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023\n\nEU Euro area EU 2030 target\n\n40\n\n45\n\n50\n\n55\n\n60\n\n65\n\n70\n\n75\n\n80\n\n85\n\n90\n\nIT\n\nEL\n\nRO\n\nES\n\nHR\n\nBE\n\nFR\n\nLU\n\nEU\n\nBG\n\nAT\n\nSK\n\nLV\n\nSI\n\nPL\n\nPT\n\nFI\n\nLT\n\nCY\n\nIE\n\nDK\n\nHU\n\nDE\n\nMT\n\nCZ\n\nEE\n\nSE\n\nNL\n\n2022 2023 2012 Target","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Employment and Social Developments in Europe 2024\n\n24\n\naverage), slightly above its long-term average (9.7%) and pre-COVID-19 pandemic levels (7.5%, three month moving average in February 2020) (Chart 1.4). The level was higher in retail (15.5%) and construction (15.3%) and, than in industry (9.7%) and services (8.2%). ( ²¹ )\n\n**In 2023, the employment rate for individuals aged 20-64 reached the highest recorded levels in the EU.** At 75.3%, it corresponded to an increase of 0.7 pp in both the EU and euro area (74.7%) compared to 2022 (Chart 1.5). The 2023 EU employment rate is getting closer to the EU Porto target, which aims to achieve at least 78% of people aged 20-64 in employment in the EU by 2030, with national targets specific to each Member State. The largest increases were recorded in Malta (+1.6 pp), Italy (+1.5 pp), Poland and Spain (both +1.2 pp), while the employment rate declined in Lithuania (-0.5 pp), Slovenia (-0.4 pp), Denmark (-0.3 pp), Finland (-0.2 pp) and Austria (-0.1 pp). Five Member States (Estonia, Ireland, the Netherlands, Slovakia, Sweden) are already above their national target (Chart 1.5).\n\nChart 1.5 Employment rates reached historic levels in 2023, but growth is slowing Employment rate (% of people aged 20-64)\n\n_Source:_ Eurostat [lfsi\\_emp\\_a].\n\nClick here to download chart.\n\n**The gender employment gap (** **22** **) continued to decrease in 2023, albeit slowly, while the employment rate for women surpassed 70% for the first time.** Within the EU, the difference in employment between women and men stood at 10.2 pp, a decline of 0.5 pp from 2022 and 1.1 pp from 2019. The employment rate for women rose to 70.2%, while the rate for men rose to 80.4%. Further improvements are necessary to meet the ambition set out in the action plan to implement the European Pillar of Social Rights of at least halving the gender employment gap by 2030 compared to 2019 (11.3 pp).\n\n**The employment rate rose for all age groups in 2023, but at a slower pace than in previous years.** On a yearly basis, the rate grew by 0.4 pp (to 35.2%) for young workers (aged 15-24), by 0.5 pp (to 82.2%) for prime age workers (aged 25-54) and by 1.7 pp (to 63.9%) for older workers (aged 55-64) (Chart 1.6). Despite the COVID-19 pandemic, over the period 2019-2023, the employment rate for younger workers increased by 1.7 pp. Potentially as a result of several national pension reforms and the entry of more active cohorts in these age groups, the employment rate during that same period rose by 5.3 pp for older workers and by 2.0 pp for prime- age workers during the same period. The gender employment gap was wider for older workers (12.0 pp) than for prime-age workers (10.1 pp) and young workers (4.3 pp).\n\n**In 2023, the employment rate rose for people aged 25-54 with lower and medium levels of education.** The increase was most notable among those with lower education (+0.8 pp, to 64.1%) while it stagnated for those with tertiary education. Despite the implied reduction in the gap, rates remained highest for those with medium-level vocational education (+0.4 pp, to 84.9%) and those with tertiary education (89.7%), who are typically more likely to have the skills required in the labour market.\n\n**High employment disparities exist between the general population and people in vulnerable situations, including non-EU citizens or people with disabilities.** The disability employment gap stood at 21.5 pp in 2023. In 2023, the employment rate was 76.2% for national workers aged 20-64 (+0.8 pp compared to 2022), 77.6% for EU mobile workers (+0.6 pp), and 63.0% for non-EU workers (+1.2 pp).\n\n( ²¹ ) See Business and Consumer Survey here. ( ²² ) Difference between the employment rate of women and men aged 20-64.\n\n66\n\n68\n\n70\n\n72\n\n74\n\n76\n\n78\n\n80\n\n2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023\n\nEU Euro area EU 2030 target\n\n40\n\n45\n\n50\n\n55\n\n60\n\n65\n\n70\n\n75\n\n80\n\n85\n\n90\n\nIT\n\nEL\n\nRO\n\nES\n\nHR\n\nBE\n\nFR\n\nLU\n\nEU\n\nBG\n\nAT\n\nSK\n\nLV\n\nSI\n\nPL\n\nPT\n\nFI\n\nLT\n\nCY\n\nIE\n\nDK\n\nHU\n\nDE\n\nMT\n\nCZ\n\nEE\n\nSE\n\nNL\n\n2022 2023 2012 Target","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/employment_and_social_developments_in_europe_esde_2024_report_p22.pdf"]},"response":{"errors":[],"extract_id":"extract_bf08790275c85cc040d8ffd6f16bcf56","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Employment and Social Developments in Europe 2024\n\n24\n\naverage), slightly above its long-term average (9.7%) and pre-COVID-19 pandemic levels (7.5%, three month moving average in February 2020) (Chart 1.4). The level was higher in retail (15.5%) and construction (15.3%) and, than in industry (9.7%) and services (8.2%). ( ²¹ )\n\n**In 2023, the employment rate for individuals aged 20-64 reached the highest recorded levels in the EU.** At 75.3%, it corresponded to an increase of 0.7 pp in both the EU and euro area (74.7%) compared to 2022 (Chart 1.5). The 2023 EU employment rate is getting closer to the EU Porto target, which aims to achieve at least 78% of people aged 20-64 in employment in the EU by 2030, with national targets specific to each Member State. The largest increases were recorded in Malta (+1.6 pp), Italy (+1.5 pp), Poland and Spain (both +1.2 pp), while the employment rate declined in Lithuania (-0.5 pp), Slovenia (-0.4 pp), Denmark (-0.3 pp), Finland (-0.2 pp) and Austria (-0.1 pp). Five Member States (Estonia, Ireland, the Netherlands, Slovakia, Sweden) are already above their national target (Chart 1.5).\n\nChart 1.5 Employment rates reached historic levels in 2023, but growth is slowing Employment rate (% of people aged 20-64)\n\n_Source:_ Eurostat [lfsi\\_emp\\_a].\n\nClick here to download chart.\n\n**The gender employment gap (** **22** **) continued to decrease in 2023, albeit slowly, while the employment rate for women surpassed 70% for the first time.** Within the EU, the difference in employment between women and men stood at 10.2 pp, a decline of 0.5 pp from 2022 and 1.1 pp from 2019. The employment rate for women rose to 70.2%, while the rate for men rose to 80.4%. Further improvements are necessary to meet the ambition set out in the action plan to implement the European Pillar of Social Rights of at least halving the gender employment gap by 2030 compared to 2019 (11.3 pp).\n\n**The employment rate rose for all age groups in 2023, but at a slower pace than in previous years.** On a yearly basis, the rate grew by 0.4 pp (to 35.2%) for young workers (aged 15-24), by 0.5 pp (to 82.2%) for prime age workers (aged 25-54) and by 1.7 pp (to 63.9%) for older workers (aged 55-64) (Chart 1.6). Despite the COVID-19 pandemic, over the period 2019-2023, the employment rate for younger workers increased by 1.7 pp. Potentially as a result of several national pension reforms and the entry of more active cohorts in these age groups, the employment rate during that same period rose by 5.3 pp for older workers and by 2.0 pp for prime- age workers during the same period. The gender employment gap was wider for older workers (12.0 pp) than for prime-age workers (10.1 pp) and young workers (4.3 pp).\n\n**In 2023, the employment rate rose for people aged 25-54 with lower and medium levels of education.** The increase was most notable among those with lower education (+0.8 pp, to 64.1%) while it stagnated for those with tertiary education. Despite the implied reduction in the gap, rates remained highest for those with medium-level vocational education (+0.4 pp, to 84.9%) and those with tertiary education (89.7%), who are typically more likely to have the skills required in the labour market.\n\n**High employment disparities exist between the general population and people in vulnerable situations, including non-EU citizens or people with disabilities.** The disability employment gap stood at 21.5 pp in 2023. In 2023, the employment rate was 76.2% for national workers aged 20-64 (+0.8 pp compared to 2022), 77.6% for EU mobile workers (+0.6 pp), and 63.0% for non-EU workers (+1.2 pp).\n\n( ²¹ ) See Business and Consumer Survey here. ( ²² ) Difference between the employment rate of women and men aged 20-64.\n\n66\n\n68\n\n70\n\n72\n\n74\n\n76\n\n78\n\n80\n\n2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023\n\nEU Euro area EU 2030 target\n\n40\n\n45\n\n50\n\n55\n\n60\n\n65\n\n70\n\n75\n\n80\n\n85\n\n90\n\nIT\n\nEL\n\nRO\n\nES\n\nHR\n\nBE\n\nFR\n\nLU\n\nEU\n\nBG\n\nAT\n\nSK\n\nLV\n\nSI\n\nPL\n\nPT\n\nFI\n\nLT\n\nCY\n\nIE\n\nDK\n\nHU\n\nDE\n\nMT\n\nCZ\n\nEE\n\nSE\n\nNL\n\n2022 2023 2012 Target"],"full_content":"Employment and Social Developments in Europe 2024\n\n24\n\naverage), slightly above its long-term average (9.7%) and pre-COVID-19 pandemic levels (7.5%, three month moving average in February 2020) (Chart 1.4). The level was higher in retail (15.5%) and construction (15.3%) and, than in industry (9.7%) and services (8.2%). ( ²¹ )\n\n**In 2023, the employment rate for individuals aged 20-64 reached the highest recorded levels in the EU.** At 75.3%, it corresponded to an increase of 0.7 pp in both the EU and euro area (74.7%) compared to 2022 (Chart 1.5). The 2023 EU employment rate is getting closer to the EU Porto target, which aims to achieve at least 78% of people aged 20-64 in employment in the EU by 2030, with national targets specific to each Member State. The largest increases were recorded in Malta (+1.6 pp), Italy (+1.5 pp), Poland and Spain (both +1.2 pp), while the employment rate declined in Lithuania (-0.5 pp), Slovenia (-0.4 pp), Denmark (-0.3 pp), Finland (-0.2 pp) and Austria (-0.1 pp). Five Member States (Estonia, Ireland, the Netherlands, Slovakia, Sweden) are already above their national target (Chart 1.5).\n\nChart 1.5 Employment rates reached historic levels in 2023, but growth is slowing Employment rate (% of people aged 20-64)\n\n_Source:_ Eurostat [lfsi\\_emp\\_a].\n\nClick here to download chart.\n\n**The gender employment gap (** **22** **) continued to decrease in 2023, albeit slowly, while the employment rate for women surpassed 70% for the first time.** Within the EU, the difference in employment between women and men stood at 10.2 pp, a decline of 0.5 pp from 2022 and 1.1 pp from 2019. The employment rate for women rose to 70.2%, while the rate for men rose to 80.4%. Further improvements are necessary to meet the ambition set out in the action plan to implement the European Pillar of Social Rights of at least halving the gender employment gap by 2030 compared to 2019 (11.3 pp).\n\n**The employment rate rose for all age groups in 2023, but at a slower pace than in previous years.** On a yearly basis, the rate grew by 0.4 pp (to 35.2%) for young workers (aged 15-24), by 0.5 pp (to 82.2%) for prime age workers (aged 25-54) and by 1.7 pp (to 63.9%) for older workers (aged 55-64) (Chart 1.6). Despite the COVID-19 pandemic, over the period 2019-2023, the employment rate for younger workers increased by 1.7 pp. Potentially as a result of several national pension reforms and the entry of more active cohorts in these age groups, the employment rate during that same period rose by 5.3 pp for older workers and by 2.0 pp for prime- age workers during the same period. The gender employment gap was wider for older workers (12.0 pp) than for prime-age workers (10.1 pp) and young workers (4.3 pp).\n\n**In 2023, the employment rate rose for people aged 25-54 with lower and medium levels of education.** The increase was most notable among those with lower education (+0.8 pp, to 64.1%) while it stagnated for those with tertiary education. Despite the implied reduction in the gap, rates remained highest for those with medium-level vocational education (+0.4 pp, to 84.9%) and those with tertiary education (89.7%), who are typically more likely to have the skills required in the labour market.\n\n**High employment disparities exist between the general population and people in vulnerable situations, including non-EU citizens or people with disabilities.** The disability employment gap stood at 21.5 pp in 2023. In 2023, the employment rate was 76.2% for national workers aged 20-64 (+0.8 pp compared to 2022), 77.6% for EU mobile workers (+0.6 pp), and 63.0% for non-EU workers (+1.2 pp).\n\n( ²¹ ) See Business and Consumer Survey here. ( ²² ) Difference between the employment rate of women and men aged 20-64.\n\n66\n\n68\n\n70\n\n72\n\n74\n\n76\n\n78\n\n80\n\n2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023\n\nEU Euro area EU 2030 target\n\n40\n\n45\n\n50\n\n55\n\n60\n\n65\n\n70\n\n75\n\n80\n\n85\n\n90\n\nIT\n\nEL\n\nRO\n\nES\n\nHR\n\nBE\n\nFR\n\nLU\n\nEU\n\nBG\n\nAT\n\nSK\n\nLV\n\nSI\n\nPL\n\nPT\n\nFI\n\nLT\n\nCY\n\nIE\n\nDK\n\nHU\n\nDE\n\nMT\n\nCZ\n\nEE\n\nSE\n\nNL\n\n2022 2023 2012 Target","publish_date":null,"title":"average), slightly above its long-term average (9.7%) and pre-COVID-19 pandemic levels (7.5%, three month moving average in February 2020) (Chart","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/employment_and_social_developments_in_europe_esde_2024_report_p22.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/employment_and_social_developments_in_europe_esde_2024_report_p22.pdf"},"request":{"example_id":"chart/employment_and_social_developments_in_europe_esde_2024_report_p22","product_type":"parse","source_file_path":"employment_and_social_developments_in_europe_esde_2024_report_p22.pdf"},"started_at":"2026-11-04T19:45:09.194960"} +{"completed_at":"2026-11-04T19:45:13.088463","latency_in_ms":2157,"output":{"example_id":"chart/employment_and_social_developments_in_europe_esde_2024_report_p23","grounded_pages":[],"layout_pages":[],"markdown":"Chapter 1: Main employment and social developments\n\n25\n\nChart 1.6 Employment rates rose in all population groups Employment rate by sex, age group, educational attainment level and citizenship (% of population of respective group), EU, 2023\n\n_Note:_ % of population aged 20-64 for all groups, except by educational attainment (% of population aged 25-54). No data for International Standard Classification of Education (ISCED) 3-4 GEN and VOC before 2021. ISCED (0-2) less than primary, primary and lower secondary education; ISCED (3-4 GEN) general upper secondary and post-secondary non-tertiary education; ISCED (3-4 VOC) upper secondary and post-secondary non-tertiary education; ISCED (5-8) tertiary education.\n\n_Source:_ Eurostat [lfsi\\_emp\\_a, lfsa\\_ergaedn].\n\nClick here to download chart.\n\n**The employment rate in rural regions continued to increase, surpassing 75% for the first time.** It is close to the employment rate of the total population at the EU level (75.3%), with differences across Member States.\n\n**Increases in the numbers of permanent and full-time workers were the primary reason for the rise in employment.** The proportion of temporary employment among individuals aged 15-64 in the EU dropped significantly during the COVID-19 pandemic in 2020 and has not returned to the same levels (23.1 million or 11.6% in 2023, 1.6 pp below the 2019 rate and 0.5 pp lower than in 2022). Comparison of the growth rates of employment (+1.2%) and temporary employment (-3.8%) shows that more people have permanent contracts. The share of part-time workers aged 15-64 in the EU increased by 0.2 pp in 2023 (to 17.8%, or 35.4 million people), still below the 2019 rate (19.4%). The share of workers in temporary (12.8%) and part-time (28.5%) employment remained significantly higher among women than men (10.5% and 8.4%, respectively). The gender gap remained stable in both temporary employment (at 2.3 pp) and in part-time employment (at 20.1 pp). Care duties remain the main reason for part-time employment (21.2%), followed by no full-time job found (19.4%) and education and training (14.2%). The proportion of involuntary part-time employment as a share of total part- time employment further decreased to 19.4% in 2023 (-1.5 pp) following a consistent trend since 2014.\n\n**Working conditions have improved over the last decade.** The number of workers working long hours decreased by nearly one-quarter compared to 2014, reaching 6.9% in 2023. Similarly, the share of workers with atypical working time was 33.9% in 2023, having decreased by nearly 5 pp over a 10-year period, with only a slight increase in 2021 due to the COVID-19 pandemic. The proportion of people working more than one job remained remarkably stable between 2020 and 2023 (3.9%), only a slight decrease since 2013 (-0.3 pp).\n\n### 3\\.2. Unemployment rates\n\n**In 2023, unemployment levels remained at record lows.** The unemployment rate among people aged 15- 74 declined modestly in both the EU (by 0.1 pp, to 6.1%) and euro area (by 0.2 pp, to 6.6%), reaching the lowest rates ever recorded by Eurostat. Compared to 2022, the reduction was 0.1 pp for both men (to 5.8%) and women (to 6.4%) (Chart 1.7). Among the signs of ongoing labour hoarding (as companies retain their employees despite an expected decrease in output in the short term) the unemployment rate is expected to remain broadly stable in 2024, at 6.1%, before falling slightly in 2025, to 6.0%. ( ²³ ) In 2023, the unemployment rate remained far higher for people with education levels up to lower secondary (11.9%, -0.4 pp from 2022) than for those with upper secondary and post-secondary (non-tertiary) education (5.6%, -0.1 pp from 2022) or with tertiary education (3.8%, stable from 2022).\n\n**In 13 Member States, the unemployment rate declined compared to 2022.** In most cases, the decrease was less than 1.0 pp, with the exception of Greece (-1.4 pp, to 11.1%). It remained stable in Romania (5.6%) and France (7.3%), and increased by between 0.1 pp and 0.9 pp in the remaining 11 Member States. ( ²⁴ )\n\n( ²³ ) (European Commission, 2024a) ( ²⁴ ) Changes in unemployment in 2023 compared to 2022: • Decrease: Belgium (-0.1 pp), Germany (-0.1 pp), Poland (-0.1 pp), Ireland (-0.2 pp), Slovakia (-0.3 pp), Slovenia (-0.3 pp), Italy (-0.4 pp), Latvia (-0.4 pp), Malta (-0.4 pp), Cyprus (-0.7 pp), Croatia (-0.9 pp), Spain (-0.8 pp), Greece (-1.4 pp).\n\n• Stable: France, Romania.\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\n70\n\n80\n\n90\n\n100\n\nWomen\n\nMen\n\n15-24\n\n25-54\n\n55-64\n\nISCED 0-2\n\nISCED 3-4 GEN\n\nISCED 3-4 VOC\n\nISCED 5-8\n\nNationals\n\nMobile workers\n\nNon-EU workers\n\nTotal 20-64\n\n2022 2023 2012","pages":[{"line_numbers":[],"markdown":"Chapter 1: Main employment and social developments\n\n25\n\nChart 1.6 Employment rates rose in all population groups Employment rate by sex, age group, educational attainment level and citizenship (% of population of respective group), EU, 2023\n\n_Note:_ % of population aged 20-64 for all groups, except by educational attainment (% of population aged 25-54). No data for International Standard Classification of Education (ISCED) 3-4 GEN and VOC before 2021. ISCED (0-2) less than primary, primary and lower secondary education; ISCED (3-4 GEN) general upper secondary and post-secondary non-tertiary education; ISCED (3-4 VOC) upper secondary and post-secondary non-tertiary education; ISCED (5-8) tertiary education.\n\n_Source:_ Eurostat [lfsi\\_emp\\_a, lfsa\\_ergaedn].\n\nClick here to download chart.\n\n**The employment rate in rural regions continued to increase, surpassing 75% for the first time.** It is close to the employment rate of the total population at the EU level (75.3%), with differences across Member States.\n\n**Increases in the numbers of permanent and full-time workers were the primary reason for the rise in employment.** The proportion of temporary employment among individuals aged 15-64 in the EU dropped significantly during the COVID-19 pandemic in 2020 and has not returned to the same levels (23.1 million or 11.6% in 2023, 1.6 pp below the 2019 rate and 0.5 pp lower than in 2022). Comparison of the growth rates of employment (+1.2%) and temporary employment (-3.8%) shows that more people have permanent contracts. The share of part-time workers aged 15-64 in the EU increased by 0.2 pp in 2023 (to 17.8%, or 35.4 million people), still below the 2019 rate (19.4%). The share of workers in temporary (12.8%) and part-time (28.5%) employment remained significantly higher among women than men (10.5% and 8.4%, respectively). The gender gap remained stable in both temporary employment (at 2.3 pp) and in part-time employment (at 20.1 pp). Care duties remain the main reason for part-time employment (21.2%), followed by no full-time job found (19.4%) and education and training (14.2%). The proportion of involuntary part-time employment as a share of total part- time employment further decreased to 19.4% in 2023 (-1.5 pp) following a consistent trend since 2014.\n\n**Working conditions have improved over the last decade.** The number of workers working long hours decreased by nearly one-quarter compared to 2014, reaching 6.9% in 2023. Similarly, the share of workers with atypical working time was 33.9% in 2023, having decreased by nearly 5 pp over a 10-year period, with only a slight increase in 2021 due to the COVID-19 pandemic. The proportion of people working more than one job remained remarkably stable between 2020 and 2023 (3.9%), only a slight decrease since 2013 (-0.3 pp).\n\n### 3\\.2. Unemployment rates\n\n**In 2023, unemployment levels remained at record lows.** The unemployment rate among people aged 15- 74 declined modestly in both the EU (by 0.1 pp, to 6.1%) and euro area (by 0.2 pp, to 6.6%), reaching the lowest rates ever recorded by Eurostat. Compared to 2022, the reduction was 0.1 pp for both men (to 5.8%) and women (to 6.4%) (Chart 1.7). Among the signs of ongoing labour hoarding (as companies retain their employees despite an expected decrease in output in the short term) the unemployment rate is expected to remain broadly stable in 2024, at 6.1%, before falling slightly in 2025, to 6.0%. ( ²³ ) In 2023, the unemployment rate remained far higher for people with education levels up to lower secondary (11.9%, -0.4 pp from 2022) than for those with upper secondary and post-secondary (non-tertiary) education (5.6%, -0.1 pp from 2022) or with tertiary education (3.8%, stable from 2022).\n\n**In 13 Member States, the unemployment rate declined compared to 2022.** In most cases, the decrease was less than 1.0 pp, with the exception of Greece (-1.4 pp, to 11.1%). It remained stable in Romania (5.6%) and France (7.3%), and increased by between 0.1 pp and 0.9 pp in the remaining 11 Member States. ( ²⁴ )\n\n( ²³ ) (European Commission, 2024a) ( ²⁴ ) Changes in unemployment in 2023 compared to 2022: • Decrease: Belgium (-0.1 pp), Germany (-0.1 pp), Poland (-0.1 pp), Ireland (-0.2 pp), Slovakia (-0.3 pp), Slovenia (-0.3 pp), Italy (-0.4 pp), Latvia (-0.4 pp), Malta (-0.4 pp), Cyprus (-0.7 pp), Croatia (-0.9 pp), Spain (-0.8 pp), Greece (-1.4 pp).\n\n• Stable: France, Romania.\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\n70\n\n80\n\n90\n\n100\n\nWomen\n\nMen\n\n15-24\n\n25-54\n\n55-64\n\nISCED 0-2\n\nISCED 3-4 GEN\n\nISCED 3-4 VOC\n\nISCED 5-8\n\nNationals\n\nMobile workers\n\nNon-EU workers\n\nTotal 20-64\n\n2022 2023 2012","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Chapter 1: Main employment and social developments\n\n25\n\nChart 1.6 Employment rates rose in all population groups Employment rate by sex, age group, educational attainment level and citizenship (% of population of respective group), EU, 2023\n\n_Note:_ % of population aged 20-64 for all groups, except by educational attainment (% of population aged 25-54). No data for International Standard Classification of Education (ISCED) 3-4 GEN and VOC before 2021. ISCED (0-2) less than primary, primary and lower secondary education; ISCED (3-4 GEN) general upper secondary and post-secondary non-tertiary education; ISCED (3-4 VOC) upper secondary and post-secondary non-tertiary education; ISCED (5-8) tertiary education.\n\n_Source:_ Eurostat [lfsi\\_emp\\_a, lfsa\\_ergaedn].\n\nClick here to download chart.\n\n**The employment rate in rural regions continued to increase, surpassing 75% for the first time.** It is close to the employment rate of the total population at the EU level (75.3%), with differences across Member States.\n\n**Increases in the numbers of permanent and full-time workers were the primary reason for the rise in employment.** The proportion of temporary employment among individuals aged 15-64 in the EU dropped significantly during the COVID-19 pandemic in 2020 and has not returned to the same levels (23.1 million or 11.6% in 2023, 1.6 pp below the 2019 rate and 0.5 pp lower than in 2022). Comparison of the growth rates of employment (+1.2%) and temporary employment (-3.8%) shows that more people have permanent contracts. The share of part-time workers aged 15-64 in the EU increased by 0.2 pp in 2023 (to 17.8%, or 35.4 million people), still below the 2019 rate (19.4%). The share of workers in temporary (12.8%) and part-time (28.5%) employment remained significantly higher among women than men (10.5% and 8.4%, respectively). The gender gap remained stable in both temporary employment (at 2.3 pp) and in part-time employment (at 20.1 pp). Care duties remain the main reason for part-time employment (21.2%), followed by no full-time job found (19.4%) and education and training (14.2%). The proportion of involuntary part-time employment as a share of total part- time employment further decreased to 19.4% in 2023 (-1.5 pp) following a consistent trend since 2014.\n\n**Working conditions have improved over the last decade.** The number of workers working long hours decreased by nearly one-quarter compared to 2014, reaching 6.9% in 2023. Similarly, the share of workers with atypical working time was 33.9% in 2023, having decreased by nearly 5 pp over a 10-year period, with only a slight increase in 2021 due to the COVID-19 pandemic. The proportion of people working more than one job remained remarkably stable between 2020 and 2023 (3.9%), only a slight decrease since 2013 (-0.3 pp).\n\n### 3\\.2. Unemployment rates\n\n**In 2023, unemployment levels remained at record lows.** The unemployment rate among people aged 15- 74 declined modestly in both the EU (by 0.1 pp, to 6.1%) and euro area (by 0.2 pp, to 6.6%), reaching the lowest rates ever recorded by Eurostat. Compared to 2022, the reduction was 0.1 pp for both men (to 5.8%) and women (to 6.4%) (Chart 1.7). Among the signs of ongoing labour hoarding (as companies retain their employees despite an expected decrease in output in the short term) the unemployment rate is expected to remain broadly stable in 2024, at 6.1%, before falling slightly in 2025, to 6.0%. ( ²³ ) In 2023, the unemployment rate remained far higher for people with education levels up to lower secondary (11.9%, -0.4 pp from 2022) than for those with upper secondary and post-secondary (non-tertiary) education (5.6%, -0.1 pp from 2022) or with tertiary education (3.8%, stable from 2022).\n\n**In 13 Member States, the unemployment rate declined compared to 2022.** In most cases, the decrease was less than 1.0 pp, with the exception of Greece (-1.4 pp, to 11.1%). It remained stable in Romania (5.6%) and France (7.3%), and increased by between 0.1 pp and 0.9 pp in the remaining 11 Member States. ( ²⁴ )\n\n( ²³ ) (European Commission, 2024a) ( ²⁴ ) Changes in unemployment in 2023 compared to 2022: • Decrease: Belgium (-0.1 pp), Germany (-0.1 pp), Poland (-0.1 pp), Ireland (-0.2 pp), Slovakia (-0.3 pp), Slovenia (-0.3 pp), Italy (-0.4 pp), Latvia (-0.4 pp), Malta (-0.4 pp), Cyprus (-0.7 pp), Croatia (-0.9 pp), Spain (-0.8 pp), Greece (-1.4 pp).\n\n• Stable: France, Romania.\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\n70\n\n80\n\n90\n\n100\n\nWomen\n\nMen\n\n15-24\n\n25-54\n\n55-64\n\nISCED 0-2\n\nISCED 3-4 GEN\n\nISCED 3-4 VOC\n\nISCED 5-8\n\nNationals\n\nMobile workers\n\nNon-EU workers\n\nTotal 20-64\n\n2022 2023 2012","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/employment_and_social_developments_in_europe_esde_2024_report_p23.pdf"]},"response":{"errors":[],"extract_id":"extract_48f55e183ee3eedf14f1a2fd2c6e6a6e","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Chapter 1: Main employment and social developments\n\n25\n\nChart 1.6 Employment rates rose in all population groups Employment rate by sex, age group, educational attainment level and citizenship (% of population of respective group), EU, 2023\n\n_Note:_ % of population aged 20-64 for all groups, except by educational attainment (% of population aged 25-54). No data for International Standard Classification of Education (ISCED) 3-4 GEN and VOC before 2021. ISCED (0-2) less than primary, primary and lower secondary education; ISCED (3-4 GEN) general upper secondary and post-secondary non-tertiary education; ISCED (3-4 VOC) upper secondary and post-secondary non-tertiary education; ISCED (5-8) tertiary education.\n\n_Source:_ Eurostat [lfsi\\_emp\\_a, lfsa\\_ergaedn].\n\nClick here to download chart.\n\n**The employment rate in rural regions continued to increase, surpassing 75% for the first time.** It is close to the employment rate of the total population at the EU level (75.3%), with differences across Member States.\n\n**Increases in the numbers of permanent and full-time workers were the primary reason for the rise in employment.** The proportion of temporary employment among individuals aged 15-64 in the EU dropped significantly during the COVID-19 pandemic in 2020 and has not returned to the same levels (23.1 million or 11.6% in 2023, 1.6 pp below the 2019 rate and 0.5 pp lower than in 2022). Comparison of the growth rates of employment (+1.2%) and temporary employment (-3.8%) shows that more people have permanent contracts. The share of part-time workers aged 15-64 in the EU increased by 0.2 pp in 2023 (to 17.8%, or 35.4 million people), still below the 2019 rate (19.4%). The share of workers in temporary (12.8%) and part-time (28.5%) employment remained significantly higher among women than men (10.5% and 8.4%, respectively). The gender gap remained stable in both temporary employment (at 2.3 pp) and in part-time employment (at 20.1 pp). Care duties remain the main reason for part-time employment (21.2%), followed by no full-time job found (19.4%) and education and training (14.2%). The proportion of involuntary part-time employment as a share of total part- time employment further decreased to 19.4% in 2023 (-1.5 pp) following a consistent trend since 2014.\n\n**Working conditions have improved over the last decade.** The number of workers working long hours decreased by nearly one-quarter compared to 2014, reaching 6.9% in 2023. Similarly, the share of workers with atypical working time was 33.9% in 2023, having decreased by nearly 5 pp over a 10-year period, with only a slight increase in 2021 due to the COVID-19 pandemic. The proportion of people working more than one job remained remarkably stable between 2020 and 2023 (3.9%), only a slight decrease since 2013 (-0.3 pp).\n\n### 3\\.2. Unemployment rates\n\n**In 2023, unemployment levels remained at record lows.** The unemployment rate among people aged 15- 74 declined modestly in both the EU (by 0.1 pp, to 6.1%) and euro area (by 0.2 pp, to 6.6%), reaching the lowest rates ever recorded by Eurostat. Compared to 2022, the reduction was 0.1 pp for both men (to 5.8%) and women (to 6.4%) (Chart 1.7). Among the signs of ongoing labour hoarding (as companies retain their employees despite an expected decrease in output in the short term) the unemployment rate is expected to remain broadly stable in 2024, at 6.1%, before falling slightly in 2025, to 6.0%. ( ²³ ) In 2023, the unemployment rate remained far higher for people with education levels up to lower secondary (11.9%, -0.4 pp from 2022) than for those with upper secondary and post-secondary (non-tertiary) education (5.6%, -0.1 pp from 2022) or with tertiary education (3.8%, stable from 2022).\n\n**In 13 Member States, the unemployment rate declined compared to 2022.** In most cases, the decrease was less than 1.0 pp, with the exception of Greece (-1.4 pp, to 11.1%). It remained stable in Romania (5.6%) and France (7.3%), and increased by between 0.1 pp and 0.9 pp in the remaining 11 Member States. ( ²⁴ )\n\n( ²³ ) (European Commission, 2024a) ( ²⁴ ) Changes in unemployment in 2023 compared to 2022: • Decrease: Belgium (-0.1 pp), Germany (-0.1 pp), Poland (-0.1 pp), Ireland (-0.2 pp), Slovakia (-0.3 pp), Slovenia (-0.3 pp), Italy (-0.4 pp), Latvia (-0.4 pp), Malta (-0.4 pp), Cyprus (-0.7 pp), Croatia (-0.9 pp), Spain (-0.8 pp), Greece (-1.4 pp).\n\n• Stable: France, Romania.\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\n70\n\n80\n\n90\n\n100\n\nWomen\n\nMen\n\n15-24\n\n25-54\n\n55-64\n\nISCED 0-2\n\nISCED 3-4 GEN\n\nISCED 3-4 VOC\n\nISCED 5-8\n\nNationals\n\nMobile workers\n\nNon-EU workers\n\nTotal 20-64\n\n2022 2023 2012"],"full_content":"Chapter 1: Main employment and social developments\n\n25\n\nChart 1.6 Employment rates rose in all population groups Employment rate by sex, age group, educational attainment level and citizenship (% of population of respective group), EU, 2023\n\n_Note:_ % of population aged 20-64 for all groups, except by educational attainment (% of population aged 25-54). No data for International Standard Classification of Education (ISCED) 3-4 GEN and VOC before 2021. ISCED (0-2) less than primary, primary and lower secondary education; ISCED (3-4 GEN) general upper secondary and post-secondary non-tertiary education; ISCED (3-4 VOC) upper secondary and post-secondary non-tertiary education; ISCED (5-8) tertiary education.\n\n_Source:_ Eurostat [lfsi\\_emp\\_a, lfsa\\_ergaedn].\n\nClick here to download chart.\n\n**The employment rate in rural regions continued to increase, surpassing 75% for the first time.** It is close to the employment rate of the total population at the EU level (75.3%), with differences across Member States.\n\n**Increases in the numbers of permanent and full-time workers were the primary reason for the rise in employment.** The proportion of temporary employment among individuals aged 15-64 in the EU dropped significantly during the COVID-19 pandemic in 2020 and has not returned to the same levels (23.1 million or 11.6% in 2023, 1.6 pp below the 2019 rate and 0.5 pp lower than in 2022). Comparison of the growth rates of employment (+1.2%) and temporary employment (-3.8%) shows that more people have permanent contracts. The share of part-time workers aged 15-64 in the EU increased by 0.2 pp in 2023 (to 17.8%, or 35.4 million people), still below the 2019 rate (19.4%). The share of workers in temporary (12.8%) and part-time (28.5%) employment remained significantly higher among women than men (10.5% and 8.4%, respectively). The gender gap remained stable in both temporary employment (at 2.3 pp) and in part-time employment (at 20.1 pp). Care duties remain the main reason for part-time employment (21.2%), followed by no full-time job found (19.4%) and education and training (14.2%). The proportion of involuntary part-time employment as a share of total part- time employment further decreased to 19.4% in 2023 (-1.5 pp) following a consistent trend since 2014.\n\n**Working conditions have improved over the last decade.** The number of workers working long hours decreased by nearly one-quarter compared to 2014, reaching 6.9% in 2023. Similarly, the share of workers with atypical working time was 33.9% in 2023, having decreased by nearly 5 pp over a 10-year period, with only a slight increase in 2021 due to the COVID-19 pandemic. The proportion of people working more than one job remained remarkably stable between 2020 and 2023 (3.9%), only a slight decrease since 2013 (-0.3 pp).\n\n### 3\\.2. Unemployment rates\n\n**In 2023, unemployment levels remained at record lows.** The unemployment rate among people aged 15- 74 declined modestly in both the EU (by 0.1 pp, to 6.1%) and euro area (by 0.2 pp, to 6.6%), reaching the lowest rates ever recorded by Eurostat. Compared to 2022, the reduction was 0.1 pp for both men (to 5.8%) and women (to 6.4%) (Chart 1.7). Among the signs of ongoing labour hoarding (as companies retain their employees despite an expected decrease in output in the short term) the unemployment rate is expected to remain broadly stable in 2024, at 6.1%, before falling slightly in 2025, to 6.0%. ( ²³ ) In 2023, the unemployment rate remained far higher for people with education levels up to lower secondary (11.9%, -0.4 pp from 2022) than for those with upper secondary and post-secondary (non-tertiary) education (5.6%, -0.1 pp from 2022) or with tertiary education (3.8%, stable from 2022).\n\n**In 13 Member States, the unemployment rate declined compared to 2022.** In most cases, the decrease was less than 1.0 pp, with the exception of Greece (-1.4 pp, to 11.1%). It remained stable in Romania (5.6%) and France (7.3%), and increased by between 0.1 pp and 0.9 pp in the remaining 11 Member States. ( ²⁴ )\n\n( ²³ ) (European Commission, 2024a) ( ²⁴ ) Changes in unemployment in 2023 compared to 2022: • Decrease: Belgium (-0.1 pp), Germany (-0.1 pp), Poland (-0.1 pp), Ireland (-0.2 pp), Slovakia (-0.3 pp), Slovenia (-0.3 pp), Italy (-0.4 pp), Latvia (-0.4 pp), Malta (-0.4 pp), Cyprus (-0.7 pp), Croatia (-0.9 pp), Spain (-0.8 pp), Greece (-1.4 pp).\n\n• Stable: France, Romania.\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\n70\n\n80\n\n90\n\n100\n\nWomen\n\nMen\n\n15-24\n\n25-54\n\n55-64\n\nISCED 0-2\n\nISCED 3-4 GEN\n\nISCED 3-4 VOC\n\nISCED 5-8\n\nNationals\n\nMobile workers\n\nNon-EU workers\n\nTotal 20-64\n\n2022 2023 2012","publish_date":null,"title":"3.2. Unemployment rates","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/employment_and_social_developments_in_europe_esde_2024_report_p23.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/employment_and_social_developments_in_europe_esde_2024_report_p23.pdf"},"request":{"example_id":"chart/employment_and_social_developments_in_europe_esde_2024_report_p23","product_type":"parse","source_file_path":"employment_and_social_developments_in_europe_esde_2024_report_p23.pdf"},"started_at":"2026-11-04T19:45:10.930802"} +{"completed_at":"2026-11-04T19:45:14.225579","latency_in_ms":1561,"output":{"example_id":"chart/employment_and_social_developments_in_europe_esde_2024_report_p24","grounded_pages":[],"layout_pages":[],"markdown":"Employment and Social Developments in Europe 2024\n\n26\n\n**The long-term unemployment rate fell in 2023.** After a brief increase during the onset of the COVID-19 pandemic, the long-term unemployment rate continued to decline steadily, reaching 2.1% in 2023 (-0.3 pp compared to 2022), with long-term unemployment representing 35.0% of all unemployment (3.5 pp lower than in 2022). Very long-term unemployment dropped to 1.2% (0.2 pp lower than 2022), representing just over one- fifth (20.6%) of total unemployment.\n\nChart 1.7 Unemployment rates reached record lows in 2023 and declined in all population groups Left chart: Unemployment rate (% of population aged 15-74), 2023; Right chart: Unemployment rate, by specific groups of population (% of active population aged 15-74), 2012, 2022, 2023, EU\n\n_Note:_ European Commission (DG ECFIN) forecast in shaded area. No data for ISCED 3-4 GEN and VOC before 2021.\n\n_Source:_ Eurostat [une\\_rt\\_a, une\\_rt\\_a, lfsa\\_urgaed, lfsa\\_urgan], DG ECFIN 2024 Spring Forecast. Click here to download chart.\n\n**The youth unemployment rate remained stable in the EU and decreased slightly in the euro area in 2023.** It averaged 14.5% in the EU and 14.5% in the euro area (-0.1 pp). The most substantial annual declines were recorded in Greece (-4.7 pp, to 26.7%) and Latvia (-3.0 pp, to 12.3%), with the sharpest rises recorded in Hungary (+2.2 pp, to 12.8%) and Finland (+2.0 pp, to 16.2%). The unemployment rate was 2.8 times higher among young workers (aged 15-24) than the rest of population (aged 25-74). The unemployment rate of young people is more sensitive to the economic cycle than other age groups, leading to quicker falls in levels during economic recovery, but higher increases when the economy slows down. These young people have a higher probability of obtaining temporary contracts or being dismissed during recession due to shorter tenures, limited work experience, and lower firing costs. ( ²⁵ )\n\nChart 1.8 NEET rates in the EU fell in 2023 15-29-year-olds (% of respective population)\n\n_Source:_ Eurostat [lfsi\\_neet\\_a]. Click here to download chart.\n\n**The share of people aged 15-29 who were not in employment, education or training (NEET) in the EU fell again in 2023.** The overall NEET rate decreased by 0.5 pp (to 11.2%), compared to a decline of 1.4 pp in 2022. The NEET rate remained somewhat higher for women (12.5%) than men (10.1%), increasing most in Lithuania (+2.8 pp) and Luxembourg (+1.7 pp), and falling most markedly in Italy (-2.9 pp), Poland (-1.6 pp) and Croatia (-1.5 pp) (Chart 1.8). Romania is now the Member State with the highest NEET rate (19.3%), while the Netherlands has the lowest rate (4.7%). Despite the recent decline, the NEET rate remains elevated, negatively\n\n• Increase: Bulgaria (+0.1 pp), the Netherlands (+0.1 pp), Sweden (+0.2 pp), Austria (+0.3 pp), Portugal (+0.3 pp), Czechia (+0.4 pp), Finland (+0.4 pp), Hungary (+0.5 pp), Denmark (+0.6 pp), Luxembourg (+0.6 pp), Estonia (+0.8 pp), Lithuania (+0.9 pp).\n\n( ²⁵ ) (European Commission, 2022d); (European Commission, 2023i). (European Commission, 2023b)\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\n2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025\n\nEU EA _Forecast_\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\nWomen\n\nMen\n\n15-24\n\n25-54\n\n55-74\n\nISCED 0-2\n\nISCED 3-4 GEN\n\nISCED 3-4 VOC\n\nISCED 5-8\n\nNationals\n\nMobile workers\n\nNon-EU workers\n\nTotal 15-74\n\n2022 2023\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\nRO\n\nIT\n\nEL\n\nBG\n\nCY\n\nLT\n\nES\n\nFR\n\nHR\n\nEU\n\nSK\n\nHU\n\nCZ\n\nLV\n\nBE\n\nEE\n\nAT\n\nFI\n\nPL\n\nPT\n\nDE\n\nDK\n\nLU\n\nIE\n\nSI\n\nMT\n\nSE\n\nNL\n\n2022 2023 2012","pages":[{"line_numbers":[],"markdown":"Employment and Social Developments in Europe 2024\n\n26\n\n**The long-term unemployment rate fell in 2023.** After a brief increase during the onset of the COVID-19 pandemic, the long-term unemployment rate continued to decline steadily, reaching 2.1% in 2023 (-0.3 pp compared to 2022), with long-term unemployment representing 35.0% of all unemployment (3.5 pp lower than in 2022). Very long-term unemployment dropped to 1.2% (0.2 pp lower than 2022), representing just over one- fifth (20.6%) of total unemployment.\n\nChart 1.7 Unemployment rates reached record lows in 2023 and declined in all population groups Left chart: Unemployment rate (% of population aged 15-74), 2023; Right chart: Unemployment rate, by specific groups of population (% of active population aged 15-74), 2012, 2022, 2023, EU\n\n_Note:_ European Commission (DG ECFIN) forecast in shaded area. No data for ISCED 3-4 GEN and VOC before 2021.\n\n_Source:_ Eurostat [une\\_rt\\_a, une\\_rt\\_a, lfsa\\_urgaed, lfsa\\_urgan], DG ECFIN 2024 Spring Forecast. Click here to download chart.\n\n**The youth unemployment rate remained stable in the EU and decreased slightly in the euro area in 2023.** It averaged 14.5% in the EU and 14.5% in the euro area (-0.1 pp). The most substantial annual declines were recorded in Greece (-4.7 pp, to 26.7%) and Latvia (-3.0 pp, to 12.3%), with the sharpest rises recorded in Hungary (+2.2 pp, to 12.8%) and Finland (+2.0 pp, to 16.2%). The unemployment rate was 2.8 times higher among young workers (aged 15-24) than the rest of population (aged 25-74). The unemployment rate of young people is more sensitive to the economic cycle than other age groups, leading to quicker falls in levels during economic recovery, but higher increases when the economy slows down. These young people have a higher probability of obtaining temporary contracts or being dismissed during recession due to shorter tenures, limited work experience, and lower firing costs. ( ²⁵ )\n\nChart 1.8 NEET rates in the EU fell in 2023 15-29-year-olds (% of respective population)\n\n_Source:_ Eurostat [lfsi\\_neet\\_a]. Click here to download chart.\n\n**The share of people aged 15-29 who were not in employment, education or training (NEET) in the EU fell again in 2023.** The overall NEET rate decreased by 0.5 pp (to 11.2%), compared to a decline of 1.4 pp in 2022. The NEET rate remained somewhat higher for women (12.5%) than men (10.1%), increasing most in Lithuania (+2.8 pp) and Luxembourg (+1.7 pp), and falling most markedly in Italy (-2.9 pp), Poland (-1.6 pp) and Croatia (-1.5 pp) (Chart 1.8). Romania is now the Member State with the highest NEET rate (19.3%), while the Netherlands has the lowest rate (4.7%). Despite the recent decline, the NEET rate remains elevated, negatively\n\n• Increase: Bulgaria (+0.1 pp), the Netherlands (+0.1 pp), Sweden (+0.2 pp), Austria (+0.3 pp), Portugal (+0.3 pp), Czechia (+0.4 pp), Finland (+0.4 pp), Hungary (+0.5 pp), Denmark (+0.6 pp), Luxembourg (+0.6 pp), Estonia (+0.8 pp), Lithuania (+0.9 pp).\n\n( ²⁵ ) (European Commission, 2022d); (European Commission, 2023i). (European Commission, 2023b)\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\n2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025\n\nEU EA _Forecast_\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\nWomen\n\nMen\n\n15-24\n\n25-54\n\n55-74\n\nISCED 0-2\n\nISCED 3-4 GEN\n\nISCED 3-4 VOC\n\nISCED 5-8\n\nNationals\n\nMobile workers\n\nNon-EU workers\n\nTotal 15-74\n\n2022 2023\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\nRO\n\nIT\n\nEL\n\nBG\n\nCY\n\nLT\n\nES\n\nFR\n\nHR\n\nEU\n\nSK\n\nHU\n\nCZ\n\nLV\n\nBE\n\nEE\n\nAT\n\nFI\n\nPL\n\nPT\n\nDE\n\nDK\n\nLU\n\nIE\n\nSI\n\nMT\n\nSE\n\nNL\n\n2022 2023 2012","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Employment and Social Developments in Europe 2024\n\n26\n\n**The long-term unemployment rate fell in 2023.** After a brief increase during the onset of the COVID-19 pandemic, the long-term unemployment rate continued to decline steadily, reaching 2.1% in 2023 (-0.3 pp compared to 2022), with long-term unemployment representing 35.0% of all unemployment (3.5 pp lower than in 2022). Very long-term unemployment dropped to 1.2% (0.2 pp lower than 2022), representing just over one- fifth (20.6%) of total unemployment.\n\nChart 1.7 Unemployment rates reached record lows in 2023 and declined in all population groups Left chart: Unemployment rate (% of population aged 15-74), 2023; Right chart: Unemployment rate, by specific groups of population (% of active population aged 15-74), 2012, 2022, 2023, EU\n\n_Note:_ European Commission (DG ECFIN) forecast in shaded area. No data for ISCED 3-4 GEN and VOC before 2021.\n\n_Source:_ Eurostat [une\\_rt\\_a, une\\_rt\\_a, lfsa\\_urgaed, lfsa\\_urgan], DG ECFIN 2024 Spring Forecast. Click here to download chart.\n\n**The youth unemployment rate remained stable in the EU and decreased slightly in the euro area in 2023.** It averaged 14.5% in the EU and 14.5% in the euro area (-0.1 pp). The most substantial annual declines were recorded in Greece (-4.7 pp, to 26.7%) and Latvia (-3.0 pp, to 12.3%), with the sharpest rises recorded in Hungary (+2.2 pp, to 12.8%) and Finland (+2.0 pp, to 16.2%). The unemployment rate was 2.8 times higher among young workers (aged 15-24) than the rest of population (aged 25-74). The unemployment rate of young people is more sensitive to the economic cycle than other age groups, leading to quicker falls in levels during economic recovery, but higher increases when the economy slows down. These young people have a higher probability of obtaining temporary contracts or being dismissed during recession due to shorter tenures, limited work experience, and lower firing costs. ( ²⁵ )\n\nChart 1.8 NEET rates in the EU fell in 2023 15-29-year-olds (% of respective population)\n\n_Source:_ Eurostat [lfsi\\_neet\\_a]. Click here to download chart.\n\n**The share of people aged 15-29 who were not in employment, education or training (NEET) in the EU fell again in 2023.** The overall NEET rate decreased by 0.5 pp (to 11.2%), compared to a decline of 1.4 pp in 2022. The NEET rate remained somewhat higher for women (12.5%) than men (10.1%), increasing most in Lithuania (+2.8 pp) and Luxembourg (+1.7 pp), and falling most markedly in Italy (-2.9 pp), Poland (-1.6 pp) and Croatia (-1.5 pp) (Chart 1.8). Romania is now the Member State with the highest NEET rate (19.3%), while the Netherlands has the lowest rate (4.7%). Despite the recent decline, the NEET rate remains elevated, negatively\n\n• Increase: Bulgaria (+0.1 pp), the Netherlands (+0.1 pp), Sweden (+0.2 pp), Austria (+0.3 pp), Portugal (+0.3 pp), Czechia (+0.4 pp), Finland (+0.4 pp), Hungary (+0.5 pp), Denmark (+0.6 pp), Luxembourg (+0.6 pp), Estonia (+0.8 pp), Lithuania (+0.9 pp).\n\n( ²⁵ ) (European Commission, 2022d); (European Commission, 2023i). (European Commission, 2023b)\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\n2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025\n\nEU EA _Forecast_\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\nWomen\n\nMen\n\n15-24\n\n25-54\n\n55-74\n\nISCED 0-2\n\nISCED 3-4 GEN\n\nISCED 3-4 VOC\n\nISCED 5-8\n\nNationals\n\nMobile workers\n\nNon-EU workers\n\nTotal 15-74\n\n2022 2023\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\nRO\n\nIT\n\nEL\n\nBG\n\nCY\n\nLT\n\nES\n\nFR\n\nHR\n\nEU\n\nSK\n\nHU\n\nCZ\n\nLV\n\nBE\n\nEE\n\nAT\n\nFI\n\nPL\n\nPT\n\nDE\n\nDK\n\nLU\n\nIE\n\nSI\n\nMT\n\nSE\n\nNL\n\n2022 2023 2012","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/employment_and_social_developments_in_europe_esde_2024_report_p24.pdf"]},"response":{"errors":[],"extract_id":"extract_7b2ff28ad99aab4866b2ed7c76dde707","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Employment and Social Developments in Europe 2024\n\n26\n\n**The long-term unemployment rate fell in 2023.** After a brief increase during the onset of the COVID-19 pandemic, the long-term unemployment rate continued to decline steadily, reaching 2.1% in 2023 (-0.3 pp compared to 2022), with long-term unemployment representing 35.0% of all unemployment (3.5 pp lower than in 2022). Very long-term unemployment dropped to 1.2% (0.2 pp lower than 2022), representing just over one- fifth (20.6%) of total unemployment.\n\nChart 1.7 Unemployment rates reached record lows in 2023 and declined in all population groups Left chart: Unemployment rate (% of population aged 15-74), 2023; Right chart: Unemployment rate, by specific groups of population (% of active population aged 15-74), 2012, 2022, 2023, EU\n\n_Note:_ European Commission (DG ECFIN) forecast in shaded area. No data for ISCED 3-4 GEN and VOC before 2021.\n\n_Source:_ Eurostat [une\\_rt\\_a, une\\_rt\\_a, lfsa\\_urgaed, lfsa\\_urgan], DG ECFIN 2024 Spring Forecast. Click here to download chart.\n\n**The youth unemployment rate remained stable in the EU and decreased slightly in the euro area in 2023.** It averaged 14.5% in the EU and 14.5% in the euro area (-0.1 pp). The most substantial annual declines were recorded in Greece (-4.7 pp, to 26.7%) and Latvia (-3.0 pp, to 12.3%), with the sharpest rises recorded in Hungary (+2.2 pp, to 12.8%) and Finland (+2.0 pp, to 16.2%). The unemployment rate was 2.8 times higher among young workers (aged 15-24) than the rest of population (aged 25-74). The unemployment rate of young people is more sensitive to the economic cycle than other age groups, leading to quicker falls in levels during economic recovery, but higher increases when the economy slows down. These young people have a higher probability of obtaining temporary contracts or being dismissed during recession due to shorter tenures, limited work experience, and lower firing costs. ( ²⁵ )\n\nChart 1.8 NEET rates in the EU fell in 2023 15-29-year-olds (% of respective population)\n\n_Source:_ Eurostat [lfsi\\_neet\\_a]. Click here to download chart.\n\n**The share of people aged 15-29 who were not in employment, education or training (NEET) in the EU fell again in 2023.** The overall NEET rate decreased by 0.5 pp (to 11.2%), compared to a decline of 1.4 pp in 2022. The NEET rate remained somewhat higher for women (12.5%) than men (10.1%), increasing most in Lithuania (+2.8 pp) and Luxembourg (+1.7 pp), and falling most markedly in Italy (-2.9 pp), Poland (-1.6 pp) and Croatia (-1.5 pp) (Chart 1.8). Romania is now the Member State with the highest NEET rate (19.3%), while the Netherlands has the lowest rate (4.7%). Despite the recent decline, the NEET rate remains elevated, negatively\n\n• Increase: Bulgaria (+0.1 pp), the Netherlands (+0.1 pp), Sweden (+0.2 pp), Austria (+0.3 pp), Portugal (+0.3 pp), Czechia (+0.4 pp), Finland (+0.4 pp), Hungary (+0.5 pp), Denmark (+0.6 pp), Luxembourg (+0.6 pp), Estonia (+0.8 pp), Lithuania (+0.9 pp).\n\n( ²⁵ ) (European Commission, 2022d); (European Commission, 2023i). (European Commission, 2023b)\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\n2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025\n\nEU EA _Forecast_\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\nWomen\n\nMen\n\n15-24\n\n25-54\n\n55-74\n\nISCED 0-2\n\nISCED 3-4 GEN\n\nISCED 3-4 VOC\n\nISCED 5-8\n\nNationals\n\nMobile workers\n\nNon-EU workers\n\nTotal 15-74\n\n2022 2023\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\nRO\n\nIT\n\nEL\n\nBG\n\nCY\n\nLT\n\nES\n\nFR\n\nHR\n\nEU\n\nSK\n\nHU\n\nCZ\n\nLV\n\nBE\n\nEE\n\nAT\n\nFI\n\nPL\n\nPT\n\nDE\n\nDK\n\nLU\n\nIE\n\nSI\n\nMT\n\nSE\n\nNL\n\n2022 2023 2012"],"full_content":"Employment and Social Developments in Europe 2024\n\n26\n\n**The long-term unemployment rate fell in 2023.** After a brief increase during the onset of the COVID-19 pandemic, the long-term unemployment rate continued to decline steadily, reaching 2.1% in 2023 (-0.3 pp compared to 2022), with long-term unemployment representing 35.0% of all unemployment (3.5 pp lower than in 2022). Very long-term unemployment dropped to 1.2% (0.2 pp lower than 2022), representing just over one- fifth (20.6%) of total unemployment.\n\nChart 1.7 Unemployment rates reached record lows in 2023 and declined in all population groups Left chart: Unemployment rate (% of population aged 15-74), 2023; Right chart: Unemployment rate, by specific groups of population (% of active population aged 15-74), 2012, 2022, 2023, EU\n\n_Note:_ European Commission (DG ECFIN) forecast in shaded area. No data for ISCED 3-4 GEN and VOC before 2021.\n\n_Source:_ Eurostat [une\\_rt\\_a, une\\_rt\\_a, lfsa\\_urgaed, lfsa\\_urgan], DG ECFIN 2024 Spring Forecast. Click here to download chart.\n\n**The youth unemployment rate remained stable in the EU and decreased slightly in the euro area in 2023.** It averaged 14.5% in the EU and 14.5% in the euro area (-0.1 pp). The most substantial annual declines were recorded in Greece (-4.7 pp, to 26.7%) and Latvia (-3.0 pp, to 12.3%), with the sharpest rises recorded in Hungary (+2.2 pp, to 12.8%) and Finland (+2.0 pp, to 16.2%). The unemployment rate was 2.8 times higher among young workers (aged 15-24) than the rest of population (aged 25-74). The unemployment rate of young people is more sensitive to the economic cycle than other age groups, leading to quicker falls in levels during economic recovery, but higher increases when the economy slows down. These young people have a higher probability of obtaining temporary contracts or being dismissed during recession due to shorter tenures, limited work experience, and lower firing costs. ( ²⁵ )\n\nChart 1.8 NEET rates in the EU fell in 2023 15-29-year-olds (% of respective population)\n\n_Source:_ Eurostat [lfsi\\_neet\\_a]. Click here to download chart.\n\n**The share of people aged 15-29 who were not in employment, education or training (NEET) in the EU fell again in 2023.** The overall NEET rate decreased by 0.5 pp (to 11.2%), compared to a decline of 1.4 pp in 2022. The NEET rate remained somewhat higher for women (12.5%) than men (10.1%), increasing most in Lithuania (+2.8 pp) and Luxembourg (+1.7 pp), and falling most markedly in Italy (-2.9 pp), Poland (-1.6 pp) and Croatia (-1.5 pp) (Chart 1.8). Romania is now the Member State with the highest NEET rate (19.3%), while the Netherlands has the lowest rate (4.7%). Despite the recent decline, the NEET rate remains elevated, negatively\n\n• Increase: Bulgaria (+0.1 pp), the Netherlands (+0.1 pp), Sweden (+0.2 pp), Austria (+0.3 pp), Portugal (+0.3 pp), Czechia (+0.4 pp), Finland (+0.4 pp), Hungary (+0.5 pp), Denmark (+0.6 pp), Luxembourg (+0.6 pp), Estonia (+0.8 pp), Lithuania (+0.9 pp).\n\n( ²⁵ ) (European Commission, 2022d); (European Commission, 2023i). (European Commission, 2023b)\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\n2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025\n\nEU EA _Forecast_\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\nWomen\n\nMen\n\n15-24\n\n25-54\n\n55-74\n\nISCED 0-2\n\nISCED 3-4 GEN\n\nISCED 3-4 VOC\n\nISCED 5-8\n\nNationals\n\nMobile workers\n\nNon-EU workers\n\nTotal 15-74\n\n2022 2023\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\nRO\n\nIT\n\nEL\n\nBG\n\nCY\n\nLT\n\nES\n\nFR\n\nHR\n\nEU\n\nSK\n\nHU\n\nCZ\n\nLV\n\nBE\n\nEE\n\nAT\n\nFI\n\nPL\n\nPT\n\nDE\n\nDK\n\nLU\n\nIE\n\nSI\n\nMT\n\nSE\n\nNL\n\n2022 2023 2012","publish_date":null,"title":"After a brief increase during the onset of the COVID-19 The long-term unemployment rate fell in 2023. pandemic, the long-term","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/employment_and_social_developments_in_europe_esde_2024_report_p24.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/employment_and_social_developments_in_europe_esde_2024_report_p24.pdf"},"request":{"example_id":"chart/employment_and_social_developments_in_europe_esde_2024_report_p24","product_type":"parse","source_file_path":"employment_and_social_developments_in_europe_esde_2024_report_p24.pdf"},"started_at":"2026-11-04T19:45:12.663774"} +{"completed_at":"2026-11-04T19:45:15.372745","latency_in_ms":1632,"output":{"example_id":"chart/employment_and_social_developments_in_europe_esde_2024_report_p27","grounded_pages":[],"layout_pages":[],"markdown":"Chapter 1: Main employment and social developments\n\n29\n\n(9.7%). ( ³² ) In addition, 64.2% of the population aged 25-64 declared that they had improved their knowledge, skills, or competences in informal ways, such as through exchanges with a family member, friend, or colleague (31.1%), alone (on electronic devices (53.2%), using printed material (37.8%)), visiting learning centres (8.4%), or attending guided tours (15.4%).\n\n**Most people did not see the need to undertake training, while those who did cited schedules, family reasons and costs as their main reasons for not participating in education or training.** In 2022, 42.4% of respondents aged 25-64 did not want to participate in education or training because they did not see the need (75.9%). Of those who wanted to participate and could not (10.2%) or those who participated but wanted to participate more (15.7%), 25.3% listed training schedules as the main barrier to their participation, while 16.0% cited family reasons, and 15.2% highlighted costs (Chart 1.12). ( ³³ )\n\n**Progress towards the EU target of at least 60% of adults participating in learning every year by 2030 is limited.** The level of participation of people aged 25- 64 in formal or non-formal learning (excluding guided- on-the-job training for the monitoring of the target) was at 34.4% in 2011, 37.4% in 2016 and 39.5% in 2022. ( ³⁴ ) In a context where labour and skills shortages constitute a challenge for EU competitiveness, investment in skills continues to be of the utmost importance, especially towards a fair green and digital transition. The Commission promotes investment in skills, notably through the new European Skills Agenda, and the action plan for the implementation of the European Pillar of Social Rights.\n\n### 3\\.6. Public expenditure on labour market policies and social protection\n\n**In the years following the COVID-19 pandemic, total public expenditure remained sustained and then slowly returned to pre-pandemic levels.** Overall, in both nominal terms and as a share of GDP, government expenditure increased in 2020 and remained at high, albeit declining, levels in 2021 in response to the economic and social challenges of the pandemic (Chart 1.13, left chart). In 2022, total public expenditure as a share of GDP returned to similar levels to 2012. The areas of public expenditure that experienced the strongest growth during the first year of COVID-19 pandemic were spending on social protection (+2.6pp between 2019 and 2020, to 21.9% of GDP) and economic affairs ( ³⁵ ) (including many COVID-19 pandemic-related measures; +1.7 pp, to 6.1% of GDP), and health (+1.0 pp, to 8.0%).%), which reflects the measures put in place to mitigate the socio- economic effects of the crisis as well as the additional demand for healthcare, medical appliances and medicines. In 2022, public expenditure higher than in 2019 included economic affairs (+1.4pp, to 5.8% of GDP), health (+0.6pp, to 7.6%) and housing (+0.4pp, to 1%), while social protection spending as a share of GDP almost fully reverted to its pre-pandemic level (+0.1pp, to 19.4%), mainly reflecting positive developments in the labour market. ( ³⁶ )\n\n( ³² ) Taking into consideration all objectives, not just those related to jobs, 53.9% of employed people, 29.2% of unemployed people and 24.4% of people outside the labour force participated in formal or non-formal education and training.\n\n( ³³ ) Other personal reasons (9.3%), lack of support from employer or public services (8.4%), no suitable offer of education or training (7.8%), health or age reasons (6.7%), other reason (4.4%), course booked out (2.8%), distance (2.7%), and too few registrations (1.7%).\n\n( ³⁴ ) Definition of the indicator monitoring the target differs from published data on the Eurobase and excludes guided-on-the-job (GOTJ) learning activities. The specially calculated data excluding GOTJ training for EU and all Member States, including disaggregation by sex, age groups and educational attainment, are available from Eurostat here.\n\n( ³⁵ ) It includes, among others, administration of general economic, commercial and labour affairs, formulation and implementation of policies, regulation or supervision of activities, support to programmes or schemes, promotion of policies and programmes, administration of sectoral affairs and services, grants, loans or subsidies supporting sectors, administration and operation of government agencies engaged in applied research, grants, loans or subsidies to support applied research by non-government bodies. More details in the Manual on sources and methods for the compilation of COFOG statistics here.\n\n( ³⁶ ) Based on latest available administrative data, which extends to 2022.\n\nChart 1.12 Schedule, family reasons and costs are the most common barriers to participation in education or training Main reason for not participating (or not participating more) for willing respondents, 2022, EU\n\n_Note:_ ‘Course booked out’ and ‘Too few registrations’ were not proposed in 2011 and 2016. Percentages calculated based on respondents who answered the question (non-response rate was 11.3% in 2011, 5.7% in 2016, 9.8% in 2022).\n\n_Source:_ Eurostat [TRNG\\_AES\\_179], Directorate-General for Employment, Social Affairs and Inclusion (DG EMPL) calculations.\n\nClick here to download chart.\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n2011 2016 2022","pages":[{"line_numbers":[],"markdown":"Chapter 1: Main employment and social developments\n\n29\n\n(9.7%). ( ³² ) In addition, 64.2% of the population aged 25-64 declared that they had improved their knowledge, skills, or competences in informal ways, such as through exchanges with a family member, friend, or colleague (31.1%), alone (on electronic devices (53.2%), using printed material (37.8%)), visiting learning centres (8.4%), or attending guided tours (15.4%).\n\n**Most people did not see the need to undertake training, while those who did cited schedules, family reasons and costs as their main reasons for not participating in education or training.** In 2022, 42.4% of respondents aged 25-64 did not want to participate in education or training because they did not see the need (75.9%). Of those who wanted to participate and could not (10.2%) or those who participated but wanted to participate more (15.7%), 25.3% listed training schedules as the main barrier to their participation, while 16.0% cited family reasons, and 15.2% highlighted costs (Chart 1.12). ( ³³ )\n\n**Progress towards the EU target of at least 60% of adults participating in learning every year by 2030 is limited.** The level of participation of people aged 25- 64 in formal or non-formal learning (excluding guided- on-the-job training for the monitoring of the target) was at 34.4% in 2011, 37.4% in 2016 and 39.5% in 2022. ( ³⁴ ) In a context where labour and skills shortages constitute a challenge for EU competitiveness, investment in skills continues to be of the utmost importance, especially towards a fair green and digital transition. The Commission promotes investment in skills, notably through the new European Skills Agenda, and the action plan for the implementation of the European Pillar of Social Rights.\n\n### 3\\.6. Public expenditure on labour market policies and social protection\n\n**In the years following the COVID-19 pandemic, total public expenditure remained sustained and then slowly returned to pre-pandemic levels.** Overall, in both nominal terms and as a share of GDP, government expenditure increased in 2020 and remained at high, albeit declining, levels in 2021 in response to the economic and social challenges of the pandemic (Chart 1.13, left chart). In 2022, total public expenditure as a share of GDP returned to similar levels to 2012. The areas of public expenditure that experienced the strongest growth during the first year of COVID-19 pandemic were spending on social protection (+2.6pp between 2019 and 2020, to 21.9% of GDP) and economic affairs ( ³⁵ ) (including many COVID-19 pandemic-related measures; +1.7 pp, to 6.1% of GDP), and health (+1.0 pp, to 8.0%).%), which reflects the measures put in place to mitigate the socio- economic effects of the crisis as well as the additional demand for healthcare, medical appliances and medicines. In 2022, public expenditure higher than in 2019 included economic affairs (+1.4pp, to 5.8% of GDP), health (+0.6pp, to 7.6%) and housing (+0.4pp, to 1%), while social protection spending as a share of GDP almost fully reverted to its pre-pandemic level (+0.1pp, to 19.4%), mainly reflecting positive developments in the labour market. ( ³⁶ )\n\n( ³² ) Taking into consideration all objectives, not just those related to jobs, 53.9% of employed people, 29.2% of unemployed people and 24.4% of people outside the labour force participated in formal or non-formal education and training.\n\n( ³³ ) Other personal reasons (9.3%), lack of support from employer or public services (8.4%), no suitable offer of education or training (7.8%), health or age reasons (6.7%), other reason (4.4%), course booked out (2.8%), distance (2.7%), and too few registrations (1.7%).\n\n( ³⁴ ) Definition of the indicator monitoring the target differs from published data on the Eurobase and excludes guided-on-the-job (GOTJ) learning activities. The specially calculated data excluding GOTJ training for EU and all Member States, including disaggregation by sex, age groups and educational attainment, are available from Eurostat here.\n\n( ³⁵ ) It includes, among others, administration of general economic, commercial and labour affairs, formulation and implementation of policies, regulation or supervision of activities, support to programmes or schemes, promotion of policies and programmes, administration of sectoral affairs and services, grants, loans or subsidies supporting sectors, administration and operation of government agencies engaged in applied research, grants, loans or subsidies to support applied research by non-government bodies. More details in the Manual on sources and methods for the compilation of COFOG statistics here.\n\n( ³⁶ ) Based on latest available administrative data, which extends to 2022.\n\nChart 1.12 Schedule, family reasons and costs are the most common barriers to participation in education or training Main reason for not participating (or not participating more) for willing respondents, 2022, EU\n\n_Note:_ ‘Course booked out’ and ‘Too few registrations’ were not proposed in 2011 and 2016. Percentages calculated based on respondents who answered the question (non-response rate was 11.3% in 2011, 5.7% in 2016, 9.8% in 2022).\n\n_Source:_ Eurostat [TRNG\\_AES\\_179], Directorate-General for Employment, Social Affairs and Inclusion (DG EMPL) calculations.\n\nClick here to download chart.\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n2011 2016 2022","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Chapter 1: Main employment and social developments\n\n29\n\n(9.7%). ( ³² ) In addition, 64.2% of the population aged 25-64 declared that they had improved their knowledge, skills, or competences in informal ways, such as through exchanges with a family member, friend, or colleague (31.1%), alone (on electronic devices (53.2%), using printed material (37.8%)), visiting learning centres (8.4%), or attending guided tours (15.4%).\n\n**Most people did not see the need to undertake training, while those who did cited schedules, family reasons and costs as their main reasons for not participating in education or training.** In 2022, 42.4% of respondents aged 25-64 did not want to participate in education or training because they did not see the need (75.9%). Of those who wanted to participate and could not (10.2%) or those who participated but wanted to participate more (15.7%), 25.3% listed training schedules as the main barrier to their participation, while 16.0% cited family reasons, and 15.2% highlighted costs (Chart 1.12). ( ³³ )\n\n**Progress towards the EU target of at least 60% of adults participating in learning every year by 2030 is limited.** The level of participation of people aged 25- 64 in formal or non-formal learning (excluding guided- on-the-job training for the monitoring of the target) was at 34.4% in 2011, 37.4% in 2016 and 39.5% in 2022. ( ³⁴ ) In a context where labour and skills shortages constitute a challenge for EU competitiveness, investment in skills continues to be of the utmost importance, especially towards a fair green and digital transition. The Commission promotes investment in skills, notably through the new European Skills Agenda, and the action plan for the implementation of the European Pillar of Social Rights.\n\n### 3\\.6. Public expenditure on labour market policies and social protection\n\n**In the years following the COVID-19 pandemic, total public expenditure remained sustained and then slowly returned to pre-pandemic levels.** Overall, in both nominal terms and as a share of GDP, government expenditure increased in 2020 and remained at high, albeit declining, levels in 2021 in response to the economic and social challenges of the pandemic (Chart 1.13, left chart). In 2022, total public expenditure as a share of GDP returned to similar levels to 2012. The areas of public expenditure that experienced the strongest growth during the first year of COVID-19 pandemic were spending on social protection (+2.6pp between 2019 and 2020, to 21.9% of GDP) and economic affairs ( ³⁵ ) (including many COVID-19 pandemic-related measures; +1.7 pp, to 6.1% of GDP), and health (+1.0 pp, to 8.0%).%), which reflects the measures put in place to mitigate the socio- economic effects of the crisis as well as the additional demand for healthcare, medical appliances and medicines. In 2022, public expenditure higher than in 2019 included economic affairs (+1.4pp, to 5.8% of GDP), health (+0.6pp, to 7.6%) and housing (+0.4pp, to 1%), while social protection spending as a share of GDP almost fully reverted to its pre-pandemic level (+0.1pp, to 19.4%), mainly reflecting positive developments in the labour market. ( ³⁶ )\n\n( ³² ) Taking into consideration all objectives, not just those related to jobs, 53.9% of employed people, 29.2% of unemployed people and 24.4% of people outside the labour force participated in formal or non-formal education and training.\n\n( ³³ ) Other personal reasons (9.3%), lack of support from employer or public services (8.4%), no suitable offer of education or training (7.8%), health or age reasons (6.7%), other reason (4.4%), course booked out (2.8%), distance (2.7%), and too few registrations (1.7%).\n\n( ³⁴ ) Definition of the indicator monitoring the target differs from published data on the Eurobase and excludes guided-on-the-job (GOTJ) learning activities. The specially calculated data excluding GOTJ training for EU and all Member States, including disaggregation by sex, age groups and educational attainment, are available from Eurostat here.\n\n( ³⁵ ) It includes, among others, administration of general economic, commercial and labour affairs, formulation and implementation of policies, regulation or supervision of activities, support to programmes or schemes, promotion of policies and programmes, administration of sectoral affairs and services, grants, loans or subsidies supporting sectors, administration and operation of government agencies engaged in applied research, grants, loans or subsidies to support applied research by non-government bodies. More details in the Manual on sources and methods for the compilation of COFOG statistics here.\n\n( ³⁶ ) Based on latest available administrative data, which extends to 2022.\n\nChart 1.12 Schedule, family reasons and costs are the most common barriers to participation in education or training Main reason for not participating (or not participating more) for willing respondents, 2022, EU\n\n_Note:_ ‘Course booked out’ and ‘Too few registrations’ were not proposed in 2011 and 2016. Percentages calculated based on respondents who answered the question (non-response rate was 11.3% in 2011, 5.7% in 2016, 9.8% in 2022).\n\n_Source:_ Eurostat [TRNG\\_AES\\_179], Directorate-General for Employment, Social Affairs and Inclusion (DG EMPL) calculations.\n\nClick here to download chart.\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n2011 2016 2022","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/employment_and_social_developments_in_europe_esde_2024_report_p27.pdf"]},"response":{"errors":[],"extract_id":"extract_d114358b4e1122920938f3bc82d00afe","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Chapter 1: Main employment and social developments\n\n29\n\n(9.7%). ( ³² ) In addition, 64.2% of the population aged 25-64 declared that they had improved their knowledge, skills, or competences in informal ways, such as through exchanges with a family member, friend, or colleague (31.1%), alone (on electronic devices (53.2%), using printed material (37.8%)), visiting learning centres (8.4%), or attending guided tours (15.4%).\n\n**Most people did not see the need to undertake training, while those who did cited schedules, family reasons and costs as their main reasons for not participating in education or training.** In 2022, 42.4% of respondents aged 25-64 did not want to participate in education or training because they did not see the need (75.9%). Of those who wanted to participate and could not (10.2%) or those who participated but wanted to participate more (15.7%), 25.3% listed training schedules as the main barrier to their participation, while 16.0% cited family reasons, and 15.2% highlighted costs (Chart 1.12). ( ³³ )\n\n**Progress towards the EU target of at least 60% of adults participating in learning every year by 2030 is limited.** The level of participation of people aged 25- 64 in formal or non-formal learning (excluding guided- on-the-job training for the monitoring of the target) was at 34.4% in 2011, 37.4% in 2016 and 39.5% in 2022. ( ³⁴ ) In a context where labour and skills shortages constitute a challenge for EU competitiveness, investment in skills continues to be of the utmost importance, especially towards a fair green and digital transition. The Commission promotes investment in skills, notably through the new European Skills Agenda, and the action plan for the implementation of the European Pillar of Social Rights.\n\n### 3\\.6. Public expenditure on labour market policies and social protection\n\n**In the years following the COVID-19 pandemic, total public expenditure remained sustained and then slowly returned to pre-pandemic levels.** Overall, in both nominal terms and as a share of GDP, government expenditure increased in 2020 and remained at high, albeit declining, levels in 2021 in response to the economic and social challenges of the pandemic (Chart 1.13, left chart). In 2022, total public expenditure as a share of GDP returned to similar levels to 2012. The areas of public expenditure that experienced the strongest growth during the first year of COVID-19 pandemic were spending on social protection (+2.6pp between 2019 and 2020, to 21.9% of GDP) and economic affairs ( ³⁵ ) (including many COVID-19 pandemic-related measures; +1.7 pp, to 6.1% of GDP), and health (+1.0 pp, to 8.0%).%), which reflects the measures put in place to mitigate the socio- economic effects of the crisis as well as the additional demand for healthcare, medical appliances and medicines. In 2022, public expenditure higher than in 2019 included economic affairs (+1.4pp, to 5.8% of GDP), health (+0.6pp, to 7.6%) and housing (+0.4pp, to 1%), while social protection spending as a share of GDP almost fully reverted to its pre-pandemic level (+0.1pp, to 19.4%), mainly reflecting positive developments in the labour market. ( ³⁶ )\n\n( ³² ) Taking into consideration all objectives, not just those related to jobs, 53.9% of employed people, 29.2% of unemployed people and 24.4% of people outside the labour force participated in formal or non-formal education and training.\n\n( ³³ ) Other personal reasons (9.3%), lack of support from employer or public services (8.4%), no suitable offer of education or training (7.8%), health or age reasons (6.7%), other reason (4.4%), course booked out (2.8%), distance (2.7%), and too few registrations (1.7%).\n\n( ³⁴ ) Definition of the indicator monitoring the target differs from published data on the Eurobase and excludes guided-on-the-job (GOTJ) learning activities. The specially calculated data excluding GOTJ training for EU and all Member States, including disaggregation by sex, age groups and educational attainment, are available from Eurostat here.\n\n( ³⁵ ) It includes, among others, administration of general economic, commercial and labour affairs, formulation and implementation of policies, regulation or supervision of activities, support to programmes or schemes, promotion of policies and programmes, administration of sectoral affairs and services, grants, loans or subsidies supporting sectors, administration and operation of government agencies engaged in applied research, grants, loans or subsidies to support applied research by non-government bodies. More details in the Manual on sources and methods for the compilation of COFOG statistics here.\n\n( ³⁶ ) Based on latest available administrative data, which extends to 2022.\n\nChart 1.12 Schedule, family reasons and costs are the most common barriers to participation in education or training Main reason for not participating (or not participating more) for willing respondents, 2022, EU\n\n_Note:_ ‘Course booked out’ and ‘Too few registrations’ were not proposed in 2011 and 2016. Percentages calculated based on respondents who answered the question (non-response rate was 11.3% in 2011, 5.7% in 2016, 9.8% in 2022).\n\n_Source:_ Eurostat [TRNG\\_AES\\_179], Directorate-General for Employment, Social Affairs and Inclusion (DG EMPL) calculations.\n\nClick here to download chart.\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n2011 2016 2022"],"full_content":"Chapter 1: Main employment and social developments\n\n29\n\n(9.7%). ( ³² ) In addition, 64.2% of the population aged 25-64 declared that they had improved their knowledge, skills, or competences in informal ways, such as through exchanges with a family member, friend, or colleague (31.1%), alone (on electronic devices (53.2%), using printed material (37.8%)), visiting learning centres (8.4%), or attending guided tours (15.4%).\n\n**Most people did not see the need to undertake training, while those who did cited schedules, family reasons and costs as their main reasons for not participating in education or training.** In 2022, 42.4% of respondents aged 25-64 did not want to participate in education or training because they did not see the need (75.9%). Of those who wanted to participate and could not (10.2%) or those who participated but wanted to participate more (15.7%), 25.3% listed training schedules as the main barrier to their participation, while 16.0% cited family reasons, and 15.2% highlighted costs (Chart 1.12). ( ³³ )\n\n**Progress towards the EU target of at least 60% of adults participating in learning every year by 2030 is limited.** The level of participation of people aged 25- 64 in formal or non-formal learning (excluding guided- on-the-job training for the monitoring of the target) was at 34.4% in 2011, 37.4% in 2016 and 39.5% in 2022. ( ³⁴ ) In a context where labour and skills shortages constitute a challenge for EU competitiveness, investment in skills continues to be of the utmost importance, especially towards a fair green and digital transition. The Commission promotes investment in skills, notably through the new European Skills Agenda, and the action plan for the implementation of the European Pillar of Social Rights.\n\n### 3\\.6. Public expenditure on labour market policies and social protection\n\n**In the years following the COVID-19 pandemic, total public expenditure remained sustained and then slowly returned to pre-pandemic levels.** Overall, in both nominal terms and as a share of GDP, government expenditure increased in 2020 and remained at high, albeit declining, levels in 2021 in response to the economic and social challenges of the pandemic (Chart 1.13, left chart). In 2022, total public expenditure as a share of GDP returned to similar levels to 2012. The areas of public expenditure that experienced the strongest growth during the first year of COVID-19 pandemic were spending on social protection (+2.6pp between 2019 and 2020, to 21.9% of GDP) and economic affairs ( ³⁵ ) (including many COVID-19 pandemic-related measures; +1.7 pp, to 6.1% of GDP), and health (+1.0 pp, to 8.0%).%), which reflects the measures put in place to mitigate the socio- economic effects of the crisis as well as the additional demand for healthcare, medical appliances and medicines. In 2022, public expenditure higher than in 2019 included economic affairs (+1.4pp, to 5.8% of GDP), health (+0.6pp, to 7.6%) and housing (+0.4pp, to 1%), while social protection spending as a share of GDP almost fully reverted to its pre-pandemic level (+0.1pp, to 19.4%), mainly reflecting positive developments in the labour market. ( ³⁶ )\n\n( ³² ) Taking into consideration all objectives, not just those related to jobs, 53.9% of employed people, 29.2% of unemployed people and 24.4% of people outside the labour force participated in formal or non-formal education and training.\n\n( ³³ ) Other personal reasons (9.3%), lack of support from employer or public services (8.4%), no suitable offer of education or training (7.8%), health or age reasons (6.7%), other reason (4.4%), course booked out (2.8%), distance (2.7%), and too few registrations (1.7%).\n\n( ³⁴ ) Definition of the indicator monitoring the target differs from published data on the Eurobase and excludes guided-on-the-job (GOTJ) learning activities. The specially calculated data excluding GOTJ training for EU and all Member States, including disaggregation by sex, age groups and educational attainment, are available from Eurostat here.\n\n( ³⁵ ) It includes, among others, administration of general economic, commercial and labour affairs, formulation and implementation of policies, regulation or supervision of activities, support to programmes or schemes, promotion of policies and programmes, administration of sectoral affairs and services, grants, loans or subsidies supporting sectors, administration and operation of government agencies engaged in applied research, grants, loans or subsidies to support applied research by non-government bodies. More details in the Manual on sources and methods for the compilation of COFOG statistics here.\n\n( ³⁶ ) Based on latest available administrative data, which extends to 2022.\n\nChart 1.12 Schedule, family reasons and costs are the most common barriers to participation in education or training Main reason for not participating (or not participating more) for willing respondents, 2022, EU\n\n_Note:_ ‘Course booked out’ and ‘Too few registrations’ were not proposed in 2011 and 2016. Percentages calculated based on respondents who answered the question (non-response rate was 11.3% in 2011, 5.7% in 2016, 9.8% in 2022).\n\n_Source:_ Eurostat [TRNG\\_AES\\_179], Directorate-General for Employment, Social Affairs and Inclusion (DG EMPL) calculations.\n\nClick here to download chart.\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n2011 2016 2022","publish_date":null,"title":"3.6. Public expenditure on labour market policies and social protection","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/employment_and_social_developments_in_europe_esde_2024_report_p27.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/employment_and_social_developments_in_europe_esde_2024_report_p27.pdf"},"request":{"example_id":"chart/employment_and_social_developments_in_europe_esde_2024_report_p27","product_type":"parse","source_file_path":"employment_and_social_developments_in_europe_esde_2024_report_p27.pdf"},"started_at":"2026-11-04T19:45:13.739940"} +{"completed_at":"2026-11-04T19:45:18.773838","latency_in_ms":3852,"output":{"example_id":"chart/employment_and_social_developments_in_europe_esde_2024_report_p29","grounded_pages":[],"layout_pages":[],"markdown":"Chapter 1: Main employment and social developments\n\n31\n\nthe positive effect of these measures. ( ³⁸ ) As a share of GDP, social protection benefit expenditure decreased to 27.2% in 2022, 3 pp lower than in 2020 and 1.5 pp lower than in 2021. Old age and survivors’ benefits represented the highest proportion of social benefit expenditure (46.3%), followed by sickness and healthcare (30.0%), family and children (8.6%), disability (7.0%), unemployment (4.3%), social exclusion (2.6%) and housing (1.3%). At national level, estimated total expenditure on social protection benefits ranged between 11.3% of GDP in Ireland and 32.2% of GDP in France in 2022. ( ³⁹ ) The significance of social protection benefits is also affected by the large shares of people entitled to such benefits but not claiming them. For instance, national case studies estimated that in Spain, 58% of households entitled to minimum income benefits in 2022 did not request them in Italy, the non-take-up rate of minimum income by eligible households in 2021 was 38.5%, and almost half (46.6%) of potentially eligible households in Belgium remained uncovered in recent years. ( ⁴⁰ )\n\nChart 1.15 Old age, unemployment and health-related benefits contributed to the decline in social protection benefit expenditure in 2022 Growth in social protection expenditure (% change on previous year, in real terms) and contribution, by function, to the yearly increase (purchasing power standards, PPS), both on the left axis, EU\n\n_Note:_ Nominal expenditure converted into real expenditure by deflating with HICP.\n\n_Source:_ Eurostat, ESSPROS [spr\\_exp\\_sum] and ESSPROS early estimates (2022), and prices statistics [prc\\_hicp\\_aind]; DG EMPL calculations. Click here to download chart.\n\n### 4\\. LIVING CONDITIONS, POVERTY, AND INCOME DEVELOPMENTS\n\n**This section presents recent trends in income and living conditions of EU households since the start of the cost-of-living crisis.** It documents income trends for the overall population and for different income groups, the role of social transfers in mitigating income inequality, and the multifaceted nature of poverty and social exclusion.( ⁴¹ ) As the official indicators on inequality and risk of poverty are computed with survey data on income (latest available are 2023 data based on 2022 incomes), these figures are complemented with Eurostat flash estimates on income inequality and poverty (2023 incomes).\n\n### 4\\.1. Income and consumptions trends\n\n**Following declines in the last three quarters of 2022, real gross disposable household income (GDHI) began to rise again in 2023, especially in the last quarter.** The impact of high inflation on real wages prompted real GDHI to decline by 0.3% in the fourth quarter of 2022 compared to the previous year. However, as inflationary pressures eased and nominal wage growth gained pace in 2023, real GDHI saw improvements and was 1.8% higher in the fourth quarter of 2023 compared to the same quarter of 2022 (0.7% in the first, 1.2% in the second, 0.6% in the third quarter in year-on-year comparisons). The positive growth of real GDHI in the first two quarters of 2023 was primarily due to the lower negative impact of taxes on income and wealth. In the context of easing inflation, real compensation of employees contributed positively to GDHI in the second half of 2023 (Chart 1.16). The phase-out of the extraordinary measures after the COVID-19 pandemic and the accompanying energy crisis can explain the negative contribution of net social benefits on the growth of real\n\n( ³⁸ ) (European Commission, 2024c). ( ³⁹ ) 2022 figures are preliminary results based on Eurostat early estimates for expenditure on social protection. ( ⁴⁰ ) Based on latest available administrative data, which extends up to 2022. Spain: (Autoridad Independente Responsabilidad Fiscal, 2023); Belgium and Italy: (Almeida, De Poli and Hernández, 2022); for EU-level developments: (Heylen, 2024), (European Commission, 2022f).\n\n( ⁴¹ ) Note that the social transfers analysed here do not include in-kind transfers such as healthcare coverage. Those are nevertheless known to have strong impact on reducing poverty (see the State of Health Synthesis Report 2023 p.17, and ESTAT on theSTIK - Impact of health social transfers in kind on income distribution and inequality - Statistics Explained (europa.eu)).\n\n\\-8\n\n\\-6\n\n\\-4\n\n\\-2\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022\n\nFamily and children Housing and social exclusion n.e.c. Old age and survivors\n\nSickness, healthcare and disability Unemployment Social protection benefits","pages":[{"line_numbers":[],"markdown":"Chapter 1: Main employment and social developments\n\n31\n\nthe positive effect of these measures. ( ³⁸ ) As a share of GDP, social protection benefit expenditure decreased to 27.2% in 2022, 3 pp lower than in 2020 and 1.5 pp lower than in 2021. Old age and survivors’ benefits represented the highest proportion of social benefit expenditure (46.3%), followed by sickness and healthcare (30.0%), family and children (8.6%), disability (7.0%), unemployment (4.3%), social exclusion (2.6%) and housing (1.3%). At national level, estimated total expenditure on social protection benefits ranged between 11.3% of GDP in Ireland and 32.2% of GDP in France in 2022. ( ³⁹ ) The significance of social protection benefits is also affected by the large shares of people entitled to such benefits but not claiming them. For instance, national case studies estimated that in Spain, 58% of households entitled to minimum income benefits in 2022 did not request them in Italy, the non-take-up rate of minimum income by eligible households in 2021 was 38.5%, and almost half (46.6%) of potentially eligible households in Belgium remained uncovered in recent years. ( ⁴⁰ )\n\nChart 1.15 Old age, unemployment and health-related benefits contributed to the decline in social protection benefit expenditure in 2022 Growth in social protection expenditure (% change on previous year, in real terms) and contribution, by function, to the yearly increase (purchasing power standards, PPS), both on the left axis, EU\n\n_Note:_ Nominal expenditure converted into real expenditure by deflating with HICP.\n\n_Source:_ Eurostat, ESSPROS [spr\\_exp\\_sum] and ESSPROS early estimates (2022), and prices statistics [prc\\_hicp\\_aind]; DG EMPL calculations. Click here to download chart.\n\n### 4\\. LIVING CONDITIONS, POVERTY, AND INCOME DEVELOPMENTS\n\n**This section presents recent trends in income and living conditions of EU households since the start of the cost-of-living crisis.** It documents income trends for the overall population and for different income groups, the role of social transfers in mitigating income inequality, and the multifaceted nature of poverty and social exclusion.( ⁴¹ ) As the official indicators on inequality and risk of poverty are computed with survey data on income (latest available are 2023 data based on 2022 incomes), these figures are complemented with Eurostat flash estimates on income inequality and poverty (2023 incomes).\n\n### 4\\.1. Income and consumptions trends\n\n**Following declines in the last three quarters of 2022, real gross disposable household income (GDHI) began to rise again in 2023, especially in the last quarter.** The impact of high inflation on real wages prompted real GDHI to decline by 0.3% in the fourth quarter of 2022 compared to the previous year. However, as inflationary pressures eased and nominal wage growth gained pace in 2023, real GDHI saw improvements and was 1.8% higher in the fourth quarter of 2023 compared to the same quarter of 2022 (0.7% in the first, 1.2% in the second, 0.6% in the third quarter in year-on-year comparisons). The positive growth of real GDHI in the first two quarters of 2023 was primarily due to the lower negative impact of taxes on income and wealth. In the context of easing inflation, real compensation of employees contributed positively to GDHI in the second half of 2023 (Chart 1.16). The phase-out of the extraordinary measures after the COVID-19 pandemic and the accompanying energy crisis can explain the negative contribution of net social benefits on the growth of real\n\n( ³⁸ ) (European Commission, 2024c). ( ³⁹ ) 2022 figures are preliminary results based on Eurostat early estimates for expenditure on social protection. ( ⁴⁰ ) Based on latest available administrative data, which extends up to 2022. Spain: (Autoridad Independente Responsabilidad Fiscal, 2023); Belgium and Italy: (Almeida, De Poli and Hernández, 2022); for EU-level developments: (Heylen, 2024), (European Commission, 2022f).\n\n( ⁴¹ ) Note that the social transfers analysed here do not include in-kind transfers such as healthcare coverage. Those are nevertheless known to have strong impact on reducing poverty (see the State of Health Synthesis Report 2023 p.17, and ESTAT on theSTIK - Impact of health social transfers in kind on income distribution and inequality - Statistics Explained (europa.eu)).\n\n\\-8\n\n\\-6\n\n\\-4\n\n\\-2\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022\n\nFamily and children Housing and social exclusion n.e.c. Old age and survivors\n\nSickness, healthcare and disability Unemployment Social protection benefits","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Chapter 1: Main employment and social developments\n\n31\n\nthe positive effect of these measures. ( ³⁸ ) As a share of GDP, social protection benefit expenditure decreased to 27.2% in 2022, 3 pp lower than in 2020 and 1.5 pp lower than in 2021. Old age and survivors’ benefits represented the highest proportion of social benefit expenditure (46.3%), followed by sickness and healthcare (30.0%), family and children (8.6%), disability (7.0%), unemployment (4.3%), social exclusion (2.6%) and housing (1.3%). At national level, estimated total expenditure on social protection benefits ranged between 11.3% of GDP in Ireland and 32.2% of GDP in France in 2022. ( ³⁹ ) The significance of social protection benefits is also affected by the large shares of people entitled to such benefits but not claiming them. For instance, national case studies estimated that in Spain, 58% of households entitled to minimum income benefits in 2022 did not request them in Italy, the non-take-up rate of minimum income by eligible households in 2021 was 38.5%, and almost half (46.6%) of potentially eligible households in Belgium remained uncovered in recent years. ( ⁴⁰ )\n\nChart 1.15 Old age, unemployment and health-related benefits contributed to the decline in social protection benefit expenditure in 2022 Growth in social protection expenditure (% change on previous year, in real terms) and contribution, by function, to the yearly increase (purchasing power standards, PPS), both on the left axis, EU\n\n_Note:_ Nominal expenditure converted into real expenditure by deflating with HICP.\n\n_Source:_ Eurostat, ESSPROS [spr\\_exp\\_sum] and ESSPROS early estimates (2022), and prices statistics [prc\\_hicp\\_aind]; DG EMPL calculations. Click here to download chart.\n\n### 4\\. LIVING CONDITIONS, POVERTY, AND INCOME DEVELOPMENTS\n\n**This section presents recent trends in income and living conditions of EU households since the start of the cost-of-living crisis.** It documents income trends for the overall population and for different income groups, the role of social transfers in mitigating income inequality, and the multifaceted nature of poverty and social exclusion.( ⁴¹ ) As the official indicators on inequality and risk of poverty are computed with survey data on income (latest available are 2023 data based on 2022 incomes), these figures are complemented with Eurostat flash estimates on income inequality and poverty (2023 incomes).\n\n### 4\\.1. Income and consumptions trends\n\n**Following declines in the last three quarters of 2022, real gross disposable household income (GDHI) began to rise again in 2023, especially in the last quarter.** The impact of high inflation on real wages prompted real GDHI to decline by 0.3% in the fourth quarter of 2022 compared to the previous year. However, as inflationary pressures eased and nominal wage growth gained pace in 2023, real GDHI saw improvements and was 1.8% higher in the fourth quarter of 2023 compared to the same quarter of 2022 (0.7% in the first, 1.2% in the second, 0.6% in the third quarter in year-on-year comparisons). The positive growth of real GDHI in the first two quarters of 2023 was primarily due to the lower negative impact of taxes on income and wealth. In the context of easing inflation, real compensation of employees contributed positively to GDHI in the second half of 2023 (Chart 1.16). The phase-out of the extraordinary measures after the COVID-19 pandemic and the accompanying energy crisis can explain the negative contribution of net social benefits on the growth of real\n\n( ³⁸ ) (European Commission, 2024c). ( ³⁹ ) 2022 figures are preliminary results based on Eurostat early estimates for expenditure on social protection. ( ⁴⁰ ) Based on latest available administrative data, which extends up to 2022. Spain: (Autoridad Independente Responsabilidad Fiscal, 2023); Belgium and Italy: (Almeida, De Poli and Hernández, 2022); for EU-level developments: (Heylen, 2024), (European Commission, 2022f).\n\n( ⁴¹ ) Note that the social transfers analysed here do not include in-kind transfers such as healthcare coverage. Those are nevertheless known to have strong impact on reducing poverty (see the State of Health Synthesis Report 2023 p.17, and ESTAT on theSTIK - Impact of health social transfers in kind on income distribution and inequality - Statistics Explained (europa.eu)).\n\n\\-8\n\n\\-6\n\n\\-4\n\n\\-2\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022\n\nFamily and children Housing and social exclusion n.e.c. Old age and survivors\n\nSickness, healthcare and disability Unemployment Social protection benefits","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/employment_and_social_developments_in_europe_esde_2024_report_p29.pdf"]},"response":{"errors":[],"extract_id":"extract_bf49589364fc9ca7c427b87996a46ad6","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Chapter 1: Main employment and social developments\n\n31\n\nthe positive effect of these measures. ( ³⁸ ) As a share of GDP, social protection benefit expenditure decreased to 27.2% in 2022, 3 pp lower than in 2020 and 1.5 pp lower than in 2021. Old age and survivors’ benefits represented the highest proportion of social benefit expenditure (46.3%), followed by sickness and healthcare (30.0%), family and children (8.6%), disability (7.0%), unemployment (4.3%), social exclusion (2.6%) and housing (1.3%). At national level, estimated total expenditure on social protection benefits ranged between 11.3% of GDP in Ireland and 32.2% of GDP in France in 2022. ( ³⁹ ) The significance of social protection benefits is also affected by the large shares of people entitled to such benefits but not claiming them. For instance, national case studies estimated that in Spain, 58% of households entitled to minimum income benefits in 2022 did not request them in Italy, the non-take-up rate of minimum income by eligible households in 2021 was 38.5%, and almost half (46.6%) of potentially eligible households in Belgium remained uncovered in recent years. ( ⁴⁰ )\n\nChart 1.15 Old age, unemployment and health-related benefits contributed to the decline in social protection benefit expenditure in 2022 Growth in social protection expenditure (% change on previous year, in real terms) and contribution, by function, to the yearly increase (purchasing power standards, PPS), both on the left axis, EU\n\n_Note:_ Nominal expenditure converted into real expenditure by deflating with HICP.\n\n_Source:_ Eurostat, ESSPROS [spr\\_exp\\_sum] and ESSPROS early estimates (2022), and prices statistics [prc\\_hicp\\_aind]; DG EMPL calculations. Click here to download chart.\n\n### 4\\. LIVING CONDITIONS, POVERTY, AND INCOME DEVELOPMENTS\n\n**This section presents recent trends in income and living conditions of EU households since the start of the cost-of-living crisis.** It documents income trends for the overall population and for different income groups, the role of social transfers in mitigating income inequality, and the multifaceted nature of poverty and social exclusion.( ⁴¹ ) As the official indicators on inequality and risk of poverty are computed with survey data on income (latest available are 2023 data based on 2022 incomes), these figures are complemented with Eurostat flash estimates on income inequality and poverty (2023 incomes).\n\n### 4\\.1. Income and consumptions trends\n\n**Following declines in the last three quarters of 2022, real gross disposable household income (GDHI) began to rise again in 2023, especially in the last quarter.** The impact of high inflation on real wages prompted real GDHI to decline by 0.3% in the fourth quarter of 2022 compared to the previous year. However, as inflationary pressures eased and nominal wage growth gained pace in 2023, real GDHI saw improvements and was 1.8% higher in the fourth quarter of 2023 compared to the same quarter of 2022 (0.7% in the first, 1.2% in the second, 0.6% in the third quarter in year-on-year comparisons). The positive growth of real GDHI in the first two quarters of 2023 was primarily due to the lower negative impact of taxes on income and wealth. In the context of easing inflation, real compensation of employees contributed positively to GDHI in the second half of 2023 (Chart 1.16). The phase-out of the extraordinary measures after the COVID-19 pandemic and the accompanying energy crisis can explain the negative contribution of net social benefits on the growth of real\n\n( ³⁸ ) (European Commission, 2024c). ( ³⁹ ) 2022 figures are preliminary results based on Eurostat early estimates for expenditure on social protection. ( ⁴⁰ ) Based on latest available administrative data, which extends up to 2022. Spain: (Autoridad Independente Responsabilidad Fiscal, 2023); Belgium and Italy: (Almeida, De Poli and Hernández, 2022); for EU-level developments: (Heylen, 2024), (European Commission, 2022f).\n\n( ⁴¹ ) Note that the social transfers analysed here do not include in-kind transfers such as healthcare coverage. Those are nevertheless known to have strong impact on reducing poverty (see the State of Health Synthesis Report 2023 p.17, and ESTAT on theSTIK - Impact of health social transfers in kind on income distribution and inequality - Statistics Explained (europa.eu)).\n\n\\-8\n\n\\-6\n\n\\-4\n\n\\-2\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022\n\nFamily and children Housing and social exclusion n.e.c. Old age and survivors\n\nSickness, healthcare and disability Unemployment Social protection benefits"],"full_content":"Chapter 1: Main employment and social developments\n\n31\n\nthe positive effect of these measures. ( ³⁸ ) As a share of GDP, social protection benefit expenditure decreased to 27.2% in 2022, 3 pp lower than in 2020 and 1.5 pp lower than in 2021. Old age and survivors’ benefits represented the highest proportion of social benefit expenditure (46.3%), followed by sickness and healthcare (30.0%), family and children (8.6%), disability (7.0%), unemployment (4.3%), social exclusion (2.6%) and housing (1.3%). At national level, estimated total expenditure on social protection benefits ranged between 11.3% of GDP in Ireland and 32.2% of GDP in France in 2022. ( ³⁹ ) The significance of social protection benefits is also affected by the large shares of people entitled to such benefits but not claiming them. For instance, national case studies estimated that in Spain, 58% of households entitled to minimum income benefits in 2022 did not request them in Italy, the non-take-up rate of minimum income by eligible households in 2021 was 38.5%, and almost half (46.6%) of potentially eligible households in Belgium remained uncovered in recent years. ( ⁴⁰ )\n\nChart 1.15 Old age, unemployment and health-related benefits contributed to the decline in social protection benefit expenditure in 2022 Growth in social protection expenditure (% change on previous year, in real terms) and contribution, by function, to the yearly increase (purchasing power standards, PPS), both on the left axis, EU\n\n_Note:_ Nominal expenditure converted into real expenditure by deflating with HICP.\n\n_Source:_ Eurostat, ESSPROS [spr\\_exp\\_sum] and ESSPROS early estimates (2022), and prices statistics [prc\\_hicp\\_aind]; DG EMPL calculations. Click here to download chart.\n\n### 4\\. LIVING CONDITIONS, POVERTY, AND INCOME DEVELOPMENTS\n\n**This section presents recent trends in income and living conditions of EU households since the start of the cost-of-living crisis.** It documents income trends for the overall population and for different income groups, the role of social transfers in mitigating income inequality, and the multifaceted nature of poverty and social exclusion.( ⁴¹ ) As the official indicators on inequality and risk of poverty are computed with survey data on income (latest available are 2023 data based on 2022 incomes), these figures are complemented with Eurostat flash estimates on income inequality and poverty (2023 incomes).\n\n### 4\\.1. Income and consumptions trends\n\n**Following declines in the last three quarters of 2022, real gross disposable household income (GDHI) began to rise again in 2023, especially in the last quarter.** The impact of high inflation on real wages prompted real GDHI to decline by 0.3% in the fourth quarter of 2022 compared to the previous year. However, as inflationary pressures eased and nominal wage growth gained pace in 2023, real GDHI saw improvements and was 1.8% higher in the fourth quarter of 2023 compared to the same quarter of 2022 (0.7% in the first, 1.2% in the second, 0.6% in the third quarter in year-on-year comparisons). The positive growth of real GDHI in the first two quarters of 2023 was primarily due to the lower negative impact of taxes on income and wealth. In the context of easing inflation, real compensation of employees contributed positively to GDHI in the second half of 2023 (Chart 1.16). The phase-out of the extraordinary measures after the COVID-19 pandemic and the accompanying energy crisis can explain the negative contribution of net social benefits on the growth of real\n\n( ³⁸ ) (European Commission, 2024c). ( ³⁹ ) 2022 figures are preliminary results based on Eurostat early estimates for expenditure on social protection. ( ⁴⁰ ) Based on latest available administrative data, which extends up to 2022. Spain: (Autoridad Independente Responsabilidad Fiscal, 2023); Belgium and Italy: (Almeida, De Poli and Hernández, 2022); for EU-level developments: (Heylen, 2024), (European Commission, 2022f).\n\n( ⁴¹ ) Note that the social transfers analysed here do not include in-kind transfers such as healthcare coverage. Those are nevertheless known to have strong impact on reducing poverty (see the State of Health Synthesis Report 2023 p.17, and ESTAT on theSTIK - Impact of health social transfers in kind on income distribution and inequality - Statistics Explained (europa.eu)).\n\n\\-8\n\n\\-6\n\n\\-4\n\n\\-2\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022\n\nFamily and children Housing and social exclusion n.e.c. Old age and survivors\n\nSickness, healthcare and disability Unemployment Social protection benefits","publish_date":null,"title":"4. L , ,","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/employment_and_social_developments_in_europe_esde_2024_report_p29.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/employment_and_social_developments_in_europe_esde_2024_report_p29.pdf"},"request":{"example_id":"chart/employment_and_social_developments_in_europe_esde_2024_report_p29","product_type":"parse","source_file_path":"employment_and_social_developments_in_europe_esde_2024_report_p29.pdf"},"started_at":"2026-11-04T19:45:14.920845"} +{"completed_at":"2026-11-04T19:45:18.954737","latency_in_ms":2655,"output":{"example_id":"chart/employment_and_social_developments_in_europe_esde_2024_report_p30","grounded_pages":[],"layout_pages":[],"markdown":"Employment and Social Developments in Europe 2024\n\n32\n\nGDHI observed in most quarters of 2021 and 2022. In the first two quarters of 2023, these net social benefits started to positively contribute to growth again, but stagnated in the second two quarters (see Section 3.6.).\n\nChart 1.16 Real GDP growth, real GDHI growth and its main components Real GDHI and real GDP (% change on previous year), and contribution of GDHI components (pp), 2012-2023, EU\n\n_Note:_ Nominal GDHI converted into real GDHI by deflating it with the price index of household final consumption expenditure [prc\\_hicp\\_aind].\n\n_Source:_ DG EMPL calculations based on Eurostat data, national accounts [nasq\\_10\\_nf\\_tr] and [namq\\_10\\_gdp], data non-seasonally adjusted.\n\nClick here to download chart.\n\n**The price of some necessities continued to increase in 2023, affecting the purchasing power of lower-income households, while energy prices remained high.** Despite easing inflationary pressure, some essential items for households, such as food and beverages (12.6%), clothing and footwear (4.4%) and rents (3.0%), continued to increase in 2023. This negatively impacted households, especially those with lower incomes, which spend a large share of their income on essential consumption items. However, prices grew less substantially for electricity, gas and other fuels (1.4%), which likely eased the financial pressure stemming from energy consumption for lower-income and some middle-income households. In this context, the increase in energy poverty in 2023 was less pronounced compared to 2022 (Chart 1.17).\n\nChart 1.17 Prices of items continued to increase in 2023 compared to 2022 and 2021 Price index of household final consumption expenditure (year-on-year change), by COICOP, HICP annual data, EU\n\n_Note:_ Consumption items selected from the classification of individual consumption by purpose (COICOP); HICP gives comparable measures of inflation for the countries and country groups for which it is produced.\n\n_Source:_ Eurostat data [prc\\_hicp\\_aind].\n\nClick here to download chart.\n\n**In the context of easing inflation, the proportion of people reporting financial distress experienced a decline.** After stagnating at high levels, around 17.1%, from May to December 2023, financial distress slightly declined in the first two quarters of 2024, reaching 16.2% in July (-1.0 pp compared to one year before). In the same month, the share of the population declaring that they needed to draw on savings reached 12.0% (-0.8 pp compared to the same month in 2023), while 4.2% stated that they were running into debt (-0.2 pp) (Chart 1.18).\n\n\\-16\n\n\\-12\n\n\\-8\n\n\\-4\n\n0\n\n4\n\n8\n\n12\n\n16\n\nQ1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3\n\n2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023\n\n_% change on previous year_\n\nCompensation of employees\n\nCompensation of self-employed\n\nNet property income\n\nNet other current transfers\n\nNet social benefits\n\nTaxes on income, wealth (negative)\n\nReal GDHI\n\nReal GDP\n\n\\-10\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\nAll-items HICP\n\nFood and non-alcoholic beverages\n\nClothing and footwear\n\nActual rentals paid by tenants\n\nWater supply\n\nElectricity, gas and other fuels\n\nHousehold furniture\n\nHousehold appliances\n\nHealth\n\nPassenger transport by underground\n\nand tram\n\nMotor cars\n\nFuels and lubricants for personal\n\ntransport equipment\n\nTelephone and telefax services\n\nRecreation and culture\n\nEducation\n\nChild care services\n\nRetirement homes for elderly\n\npersons and residences for disabled\n\npersons\n\nInsurance\n\n2021 2022 2023","pages":[{"line_numbers":[],"markdown":"Employment and Social Developments in Europe 2024\n\n32\n\nGDHI observed in most quarters of 2021 and 2022. In the first two quarters of 2023, these net social benefits started to positively contribute to growth again, but stagnated in the second two quarters (see Section 3.6.).\n\nChart 1.16 Real GDP growth, real GDHI growth and its main components Real GDHI and real GDP (% change on previous year), and contribution of GDHI components (pp), 2012-2023, EU\n\n_Note:_ Nominal GDHI converted into real GDHI by deflating it with the price index of household final consumption expenditure [prc\\_hicp\\_aind].\n\n_Source:_ DG EMPL calculations based on Eurostat data, national accounts [nasq\\_10\\_nf\\_tr] and [namq\\_10\\_gdp], data non-seasonally adjusted.\n\nClick here to download chart.\n\n**The price of some necessities continued to increase in 2023, affecting the purchasing power of lower-income households, while energy prices remained high.** Despite easing inflationary pressure, some essential items for households, such as food and beverages (12.6%), clothing and footwear (4.4%) and rents (3.0%), continued to increase in 2023. This negatively impacted households, especially those with lower incomes, which spend a large share of their income on essential consumption items. However, prices grew less substantially for electricity, gas and other fuels (1.4%), which likely eased the financial pressure stemming from energy consumption for lower-income and some middle-income households. In this context, the increase in energy poverty in 2023 was less pronounced compared to 2022 (Chart 1.17).\n\nChart 1.17 Prices of items continued to increase in 2023 compared to 2022 and 2021 Price index of household final consumption expenditure (year-on-year change), by COICOP, HICP annual data, EU\n\n_Note:_ Consumption items selected from the classification of individual consumption by purpose (COICOP); HICP gives comparable measures of inflation for the countries and country groups for which it is produced.\n\n_Source:_ Eurostat data [prc\\_hicp\\_aind].\n\nClick here to download chart.\n\n**In the context of easing inflation, the proportion of people reporting financial distress experienced a decline.** After stagnating at high levels, around 17.1%, from May to December 2023, financial distress slightly declined in the first two quarters of 2024, reaching 16.2% in July (-1.0 pp compared to one year before). In the same month, the share of the population declaring that they needed to draw on savings reached 12.0% (-0.8 pp compared to the same month in 2023), while 4.2% stated that they were running into debt (-0.2 pp) (Chart 1.18).\n\n\\-16\n\n\\-12\n\n\\-8\n\n\\-4\n\n0\n\n4\n\n8\n\n12\n\n16\n\nQ1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3\n\n2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023\n\n_% change on previous year_\n\nCompensation of employees\n\nCompensation of self-employed\n\nNet property income\n\nNet other current transfers\n\nNet social benefits\n\nTaxes on income, wealth (negative)\n\nReal GDHI\n\nReal GDP\n\n\\-10\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\nAll-items HICP\n\nFood and non-alcoholic beverages\n\nClothing and footwear\n\nActual rentals paid by tenants\n\nWater supply\n\nElectricity, gas and other fuels\n\nHousehold furniture\n\nHousehold appliances\n\nHealth\n\nPassenger transport by underground\n\nand tram\n\nMotor cars\n\nFuels and lubricants for personal\n\ntransport equipment\n\nTelephone and telefax services\n\nRecreation and culture\n\nEducation\n\nChild care services\n\nRetirement homes for elderly\n\npersons and residences for disabled\n\npersons\n\nInsurance\n\n2021 2022 2023","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Employment and Social Developments in Europe 2024\n\n32\n\nGDHI observed in most quarters of 2021 and 2022. In the first two quarters of 2023, these net social benefits started to positively contribute to growth again, but stagnated in the second two quarters (see Section 3.6.).\n\nChart 1.16 Real GDP growth, real GDHI growth and its main components Real GDHI and real GDP (% change on previous year), and contribution of GDHI components (pp), 2012-2023, EU\n\n_Note:_ Nominal GDHI converted into real GDHI by deflating it with the price index of household final consumption expenditure [prc\\_hicp\\_aind].\n\n_Source:_ DG EMPL calculations based on Eurostat data, national accounts [nasq\\_10\\_nf\\_tr] and [namq\\_10\\_gdp], data non-seasonally adjusted.\n\nClick here to download chart.\n\n**The price of some necessities continued to increase in 2023, affecting the purchasing power of lower-income households, while energy prices remained high.** Despite easing inflationary pressure, some essential items for households, such as food and beverages (12.6%), clothing and footwear (4.4%) and rents (3.0%), continued to increase in 2023. This negatively impacted households, especially those with lower incomes, which spend a large share of their income on essential consumption items. However, prices grew less substantially for electricity, gas and other fuels (1.4%), which likely eased the financial pressure stemming from energy consumption for lower-income and some middle-income households. In this context, the increase in energy poverty in 2023 was less pronounced compared to 2022 (Chart 1.17).\n\nChart 1.17 Prices of items continued to increase in 2023 compared to 2022 and 2021 Price index of household final consumption expenditure (year-on-year change), by COICOP, HICP annual data, EU\n\n_Note:_ Consumption items selected from the classification of individual consumption by purpose (COICOP); HICP gives comparable measures of inflation for the countries and country groups for which it is produced.\n\n_Source:_ Eurostat data [prc\\_hicp\\_aind].\n\nClick here to download chart.\n\n**In the context of easing inflation, the proportion of people reporting financial distress experienced a decline.** After stagnating at high levels, around 17.1%, from May to December 2023, financial distress slightly declined in the first two quarters of 2024, reaching 16.2% in July (-1.0 pp compared to one year before). In the same month, the share of the population declaring that they needed to draw on savings reached 12.0% (-0.8 pp compared to the same month in 2023), while 4.2% stated that they were running into debt (-0.2 pp) (Chart 1.18).\n\n\\-16\n\n\\-12\n\n\\-8\n\n\\-4\n\n0\n\n4\n\n8\n\n12\n\n16\n\nQ1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3\n\n2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023\n\n_% change on previous year_\n\nCompensation of employees\n\nCompensation of self-employed\n\nNet property income\n\nNet other current transfers\n\nNet social benefits\n\nTaxes on income, wealth (negative)\n\nReal GDHI\n\nReal GDP\n\n\\-10\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\nAll-items HICP\n\nFood and non-alcoholic beverages\n\nClothing and footwear\n\nActual rentals paid by tenants\n\nWater supply\n\nElectricity, gas and other fuels\n\nHousehold furniture\n\nHousehold appliances\n\nHealth\n\nPassenger transport by underground\n\nand tram\n\nMotor cars\n\nFuels and lubricants for personal\n\ntransport equipment\n\nTelephone and telefax services\n\nRecreation and culture\n\nEducation\n\nChild care services\n\nRetirement homes for elderly\n\npersons and residences for disabled\n\npersons\n\nInsurance\n\n2021 2022 2023","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/employment_and_social_developments_in_europe_esde_2024_report_p30.pdf"]},"response":{"errors":[],"extract_id":"extract_fff7a512384888cda21530f5dd7d2af3","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Employment and Social Developments in Europe 2024\n\n32\n\nGDHI observed in most quarters of 2021 and 2022. In the first two quarters of 2023, these net social benefits started to positively contribute to growth again, but stagnated in the second two quarters (see Section 3.6.).\n\nChart 1.16 Real GDP growth, real GDHI growth and its main components Real GDHI and real GDP (% change on previous year), and contribution of GDHI components (pp), 2012-2023, EU\n\n_Note:_ Nominal GDHI converted into real GDHI by deflating it with the price index of household final consumption expenditure [prc\\_hicp\\_aind].\n\n_Source:_ DG EMPL calculations based on Eurostat data, national accounts [nasq\\_10\\_nf\\_tr] and [namq\\_10\\_gdp], data non-seasonally adjusted.\n\nClick here to download chart.\n\n**The price of some necessities continued to increase in 2023, affecting the purchasing power of lower-income households, while energy prices remained high.** Despite easing inflationary pressure, some essential items for households, such as food and beverages (12.6%), clothing and footwear (4.4%) and rents (3.0%), continued to increase in 2023. This negatively impacted households, especially those with lower incomes, which spend a large share of their income on essential consumption items. However, prices grew less substantially for electricity, gas and other fuels (1.4%), which likely eased the financial pressure stemming from energy consumption for lower-income and some middle-income households. In this context, the increase in energy poverty in 2023 was less pronounced compared to 2022 (Chart 1.17).\n\nChart 1.17 Prices of items continued to increase in 2023 compared to 2022 and 2021 Price index of household final consumption expenditure (year-on-year change), by COICOP, HICP annual data, EU\n\n_Note:_ Consumption items selected from the classification of individual consumption by purpose (COICOP); HICP gives comparable measures of inflation for the countries and country groups for which it is produced.\n\n_Source:_ Eurostat data [prc\\_hicp\\_aind].\n\nClick here to download chart.\n\n**In the context of easing inflation, the proportion of people reporting financial distress experienced a decline.** After stagnating at high levels, around 17.1%, from May to December 2023, financial distress slightly declined in the first two quarters of 2024, reaching 16.2% in July (-1.0 pp compared to one year before). In the same month, the share of the population declaring that they needed to draw on savings reached 12.0% (-0.8 pp compared to the same month in 2023), while 4.2% stated that they were running into debt (-0.2 pp) (Chart 1.18).\n\n\\-16\n\n\\-12\n\n\\-8\n\n\\-4\n\n0\n\n4\n\n8\n\n12\n\n16\n\nQ1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3\n\n2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023\n\n_% change on previous year_\n\nCompensation of employees\n\nCompensation of self-employed\n\nNet property income\n\nNet other current transfers\n\nNet social benefits\n\nTaxes on income, wealth (negative)\n\nReal GDHI\n\nReal GDP\n\n\\-10\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\nAll-items HICP\n\nFood and non-alcoholic beverages\n\nClothing and footwear\n\nActual rentals paid by tenants\n\nWater supply\n\nElectricity, gas and other fuels\n\nHousehold furniture\n\nHousehold appliances\n\nHealth\n\nPassenger transport by underground\n\nand tram\n\nMotor cars\n\nFuels and lubricants for personal\n\ntransport equipment\n\nTelephone and telefax services\n\nRecreation and culture\n\nEducation\n\nChild care services\n\nRetirement homes for elderly\n\npersons and residences for disabled\n\npersons\n\nInsurance\n\n2021 2022 2023"],"full_content":"Employment and Social Developments in Europe 2024\n\n32\n\nGDHI observed in most quarters of 2021 and 2022. In the first two quarters of 2023, these net social benefits started to positively contribute to growth again, but stagnated in the second two quarters (see Section 3.6.).\n\nChart 1.16 Real GDP growth, real GDHI growth and its main components Real GDHI and real GDP (% change on previous year), and contribution of GDHI components (pp), 2012-2023, EU\n\n_Note:_ Nominal GDHI converted into real GDHI by deflating it with the price index of household final consumption expenditure [prc\\_hicp\\_aind].\n\n_Source:_ DG EMPL calculations based on Eurostat data, national accounts [nasq\\_10\\_nf\\_tr] and [namq\\_10\\_gdp], data non-seasonally adjusted.\n\nClick here to download chart.\n\n**The price of some necessities continued to increase in 2023, affecting the purchasing power of lower-income households, while energy prices remained high.** Despite easing inflationary pressure, some essential items for households, such as food and beverages (12.6%), clothing and footwear (4.4%) and rents (3.0%), continued to increase in 2023. This negatively impacted households, especially those with lower incomes, which spend a large share of their income on essential consumption items. However, prices grew less substantially for electricity, gas and other fuels (1.4%), which likely eased the financial pressure stemming from energy consumption for lower-income and some middle-income households. In this context, the increase in energy poverty in 2023 was less pronounced compared to 2022 (Chart 1.17).\n\nChart 1.17 Prices of items continued to increase in 2023 compared to 2022 and 2021 Price index of household final consumption expenditure (year-on-year change), by COICOP, HICP annual data, EU\n\n_Note:_ Consumption items selected from the classification of individual consumption by purpose (COICOP); HICP gives comparable measures of inflation for the countries and country groups for which it is produced.\n\n_Source:_ Eurostat data [prc\\_hicp\\_aind].\n\nClick here to download chart.\n\n**In the context of easing inflation, the proportion of people reporting financial distress experienced a decline.** After stagnating at high levels, around 17.1%, from May to December 2023, financial distress slightly declined in the first two quarters of 2024, reaching 16.2% in July (-1.0 pp compared to one year before). In the same month, the share of the population declaring that they needed to draw on savings reached 12.0% (-0.8 pp compared to the same month in 2023), while 4.2% stated that they were running into debt (-0.2 pp) (Chart 1.18).\n\n\\-16\n\n\\-12\n\n\\-8\n\n\\-4\n\n0\n\n4\n\n8\n\n12\n\n16\n\nQ1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3\n\n2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023\n\n_% change on previous year_\n\nCompensation of employees\n\nCompensation of self-employed\n\nNet property income\n\nNet other current transfers\n\nNet social benefits\n\nTaxes on income, wealth (negative)\n\nReal GDHI\n\nReal GDP\n\n\\-10\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\nAll-items HICP\n\nFood and non-alcoholic beverages\n\nClothing and footwear\n\nActual rentals paid by tenants\n\nWater supply\n\nElectricity, gas and other fuels\n\nHousehold furniture\n\nHousehold appliances\n\nHealth\n\nPassenger transport by underground\n\nand tram\n\nMotor cars\n\nFuels and lubricants for personal\n\ntransport equipment\n\nTelephone and telefax services\n\nRecreation and culture\n\nEducation\n\nChild care services\n\nRetirement homes for elderly\n\npersons and residences for disabled\n\npersons\n\nInsurance\n\n2021 2022 2023","publish_date":null,"title":"GDHI observed in most quarters of 2021 and 2022. In the first two quarters of 2023, these net social benefits","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/employment_and_social_developments_in_europe_esde_2024_report_p30.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/employment_and_social_developments_in_europe_esde_2024_report_p30.pdf"},"request":{"example_id":"chart/employment_and_social_developments_in_europe_esde_2024_report_p30","product_type":"parse","source_file_path":"employment_and_social_developments_in_europe_esde_2024_report_p30.pdf"},"started_at":"2026-11-04T19:45:16.298826"} +{"completed_at":"2026-11-04T19:45:19.516529","latency_in_ms":1207,"output":{"example_id":"chart/employment_and_social_developments_in_europe_esde_2024_report_p31","grounded_pages":[],"layout_pages":[],"markdown":"Chapter 1: Main employment and social developments\n\n33\n\nChart 1.18 Persistent financial distress over the last year, especially for lower-income households Reported financial distress by income quartile, 2012-2024, EU\n\n_Note:_ Lines show the long-term averages for financial distress for the population as a whole and for households in the four income quartiles. Overall share of adults reporting having to draw on savings and/or run into debt are in light grey and dark grey, respectively, which together represent total financial distress.\n\n_Source:_ DG EMPL calculations based on Business and Consumer Surveys, unadjusted data, 12-month moving average. Click here to download chart.\n\n**Lower-income households still experienced particularly elevated levels of financial distress.** In July 2024 this indicator reached 27.6% for the lowest income quartile (-0.7 pp on a yearly basis), remaining at almost 10 pp or more above the shares for other income quartiles. It amounted to 17.9% for the second quartile (- 0.8 pp), 12.5% for the third quartile (-0.8 pp), and 7.6% for the wealthiest quartile (-0.7 pp) (Chart 1.18). Persistent financial distress, especially for lower-income households, could be partly due to average price increases for certain consumer items in 2023, notably food, and ongoing elevated energy prices.\n\nChart 1.19 Real median disposable income did not increase in all countries since 2010 Median equivalised disposable income in real terms (Index SILC 2010 = 100)\n\n_Source:_ Eurostat [ILC\\_DI18].\n\nClick here to download chart.\n\n**Over the last decade, real median disposable income increased for the EU average but remained at 2015 (2014 incomes) levels in 2023 (2022 incomes) in a number of Member States.** This includes some major EU economies (Chart 1.19). The real median disposable income at EU level increased from 97% of the 2010 real median income in 2015 (2014 incomes) to 118.5% in 2023 (2022 incomes). It remained quite stable since 2020. The trend varied considerably across the Member States. In France, Italy, Finland, Denmark, Germany, Belgium, Austria, Slovakia and Sweden, the 2023 real median disposable income remained at 2015 levels, while it increased most in Romania, Latvia, Lithuania, Poland, Bulgaria and Hungary. In Greece, the real median disposable income remains well below the pre-financial crisis level. In nominal terms, the flash estimates (2023 income) nowcast an overall increase of 6.0% for 2023 incomes, with positive changes estimated for all countries. In real terms, however, a slight, non-statistically significant increase (0.2%) is expected at EU level, while at\n\n0\n\n3\n\n6\n\n9\n\n12\n\n15\n\n18\n\n21\n\n24\n\n27\n\n30\n\nJan\n\nMay\n\nSep\n\nJan\n\nMay\n\nSep\n\nJan\n\nMay\n\nSep\n\nJan\n\nMay\n\nSep\n\nJan\n\nMay\n\nSep\n\nJan\n\nMay\n\nSep\n\nJan\n\nMay\n\nSep\n\nJan\n\nMay\n\nSep\n\nJan\n\nMay\n\nSep\n\nJan\n\nMay\n\nSep\n\nJan\n\nMay\n\nSep\n\nJan\n\nMay\n\nSep\n\nJan\n\nMay\n\n2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024\n\n_% of respondents in respective group_\n\nlowest quartile\n\nTOTAL\n\nhighest quartile\n\nsecond quartile\n\nthird quartile\n\n_% need to draw on savings_\n\nFinancial distress by income quartile\n\n_% need to run into debt_\n\n0\n\n50\n\n100\n\n150\n\n200\n\n250\n\nEL\n\nCY\n\nFR\n\nIT\n\nES\n\nFI\n\nDK\n\nDE\n\nBE\n\nAT\n\nSK\n\nLU\n\nPT\n\nNL\n\nSE\n\nEU\n\nSI\n\nCZ\n\nIE\n\nHR\n\nMT\n\nHU\n\nBG\n\nEE\n\nPL\n\nLV\n\nLT\n\nRO\n\n2022 2023 2015","pages":[{"line_numbers":[],"markdown":"Chapter 1: Main employment and social developments\n\n33\n\nChart 1.18 Persistent financial distress over the last year, especially for lower-income households Reported financial distress by income quartile, 2012-2024, EU\n\n_Note:_ Lines show the long-term averages for financial distress for the population as a whole and for households in the four income quartiles. Overall share of adults reporting having to draw on savings and/or run into debt are in light grey and dark grey, respectively, which together represent total financial distress.\n\n_Source:_ DG EMPL calculations based on Business and Consumer Surveys, unadjusted data, 12-month moving average. Click here to download chart.\n\n**Lower-income households still experienced particularly elevated levels of financial distress.** In July 2024 this indicator reached 27.6% for the lowest income quartile (-0.7 pp on a yearly basis), remaining at almost 10 pp or more above the shares for other income quartiles. It amounted to 17.9% for the second quartile (- 0.8 pp), 12.5% for the third quartile (-0.8 pp), and 7.6% for the wealthiest quartile (-0.7 pp) (Chart 1.18). Persistent financial distress, especially for lower-income households, could be partly due to average price increases for certain consumer items in 2023, notably food, and ongoing elevated energy prices.\n\nChart 1.19 Real median disposable income did not increase in all countries since 2010 Median equivalised disposable income in real terms (Index SILC 2010 = 100)\n\n_Source:_ Eurostat [ILC\\_DI18].\n\nClick here to download chart.\n\n**Over the last decade, real median disposable income increased for the EU average but remained at 2015 (2014 incomes) levels in 2023 (2022 incomes) in a number of Member States.** This includes some major EU economies (Chart 1.19). The real median disposable income at EU level increased from 97% of the 2010 real median income in 2015 (2014 incomes) to 118.5% in 2023 (2022 incomes). It remained quite stable since 2020. The trend varied considerably across the Member States. In France, Italy, Finland, Denmark, Germany, Belgium, Austria, Slovakia and Sweden, the 2023 real median disposable income remained at 2015 levels, while it increased most in Romania, Latvia, Lithuania, Poland, Bulgaria and Hungary. In Greece, the real median disposable income remains well below the pre-financial crisis level. In nominal terms, the flash estimates (2023 income) nowcast an overall increase of 6.0% for 2023 incomes, with positive changes estimated for all countries. In real terms, however, a slight, non-statistically significant increase (0.2%) is expected at EU level, while at\n\n0\n\n3\n\n6\n\n9\n\n12\n\n15\n\n18\n\n21\n\n24\n\n27\n\n30\n\nJan\n\nMay\n\nSep\n\nJan\n\nMay\n\nSep\n\nJan\n\nMay\n\nSep\n\nJan\n\nMay\n\nSep\n\nJan\n\nMay\n\nSep\n\nJan\n\nMay\n\nSep\n\nJan\n\nMay\n\nSep\n\nJan\n\nMay\n\nSep\n\nJan\n\nMay\n\nSep\n\nJan\n\nMay\n\nSep\n\nJan\n\nMay\n\nSep\n\nJan\n\nMay\n\nSep\n\nJan\n\nMay\n\n2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024\n\n_% of respondents in respective group_\n\nlowest quartile\n\nTOTAL\n\nhighest quartile\n\nsecond quartile\n\nthird quartile\n\n_% need to draw on savings_\n\nFinancial distress by income quartile\n\n_% need to run into debt_\n\n0\n\n50\n\n100\n\n150\n\n200\n\n250\n\nEL\n\nCY\n\nFR\n\nIT\n\nES\n\nFI\n\nDK\n\nDE\n\nBE\n\nAT\n\nSK\n\nLU\n\nPT\n\nNL\n\nSE\n\nEU\n\nSI\n\nCZ\n\nIE\n\nHR\n\nMT\n\nHU\n\nBG\n\nEE\n\nPL\n\nLV\n\nLT\n\nRO\n\n2022 2023 2015","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Chapter 1: Main employment and social developments\n\n33\n\nChart 1.18 Persistent financial distress over the last year, especially for lower-income households Reported financial distress by income quartile, 2012-2024, EU\n\n_Note:_ Lines show the long-term averages for financial distress for the population as a whole and for households in the four income quartiles. Overall share of adults reporting having to draw on savings and/or run into debt are in light grey and dark grey, respectively, which together represent total financial distress.\n\n_Source:_ DG EMPL calculations based on Business and Consumer Surveys, unadjusted data, 12-month moving average. Click here to download chart.\n\n**Lower-income households still experienced particularly elevated levels of financial distress.** In July 2024 this indicator reached 27.6% for the lowest income quartile (-0.7 pp on a yearly basis), remaining at almost 10 pp or more above the shares for other income quartiles. It amounted to 17.9% for the second quartile (- 0.8 pp), 12.5% for the third quartile (-0.8 pp), and 7.6% for the wealthiest quartile (-0.7 pp) (Chart 1.18). Persistent financial distress, especially for lower-income households, could be partly due to average price increases for certain consumer items in 2023, notably food, and ongoing elevated energy prices.\n\nChart 1.19 Real median disposable income did not increase in all countries since 2010 Median equivalised disposable income in real terms (Index SILC 2010 = 100)\n\n_Source:_ Eurostat [ILC\\_DI18].\n\nClick here to download chart.\n\n**Over the last decade, real median disposable income increased for the EU average but remained at 2015 (2014 incomes) levels in 2023 (2022 incomes) in a number of Member States.** This includes some major EU economies (Chart 1.19). The real median disposable income at EU level increased from 97% of the 2010 real median income in 2015 (2014 incomes) to 118.5% in 2023 (2022 incomes). It remained quite stable since 2020. The trend varied considerably across the Member States. In France, Italy, Finland, Denmark, Germany, Belgium, Austria, Slovakia and Sweden, the 2023 real median disposable income remained at 2015 levels, while it increased most in Romania, Latvia, Lithuania, Poland, Bulgaria and Hungary. In Greece, the real median disposable income remains well below the pre-financial crisis level. In nominal terms, the flash estimates (2023 income) nowcast an overall increase of 6.0% for 2023 incomes, with positive changes estimated for all countries. In real terms, however, a slight, non-statistically significant increase (0.2%) is expected at EU level, while at\n\n0\n\n3\n\n6\n\n9\n\n12\n\n15\n\n18\n\n21\n\n24\n\n27\n\n30\n\nJan\n\nMay\n\nSep\n\nJan\n\nMay\n\nSep\n\nJan\n\nMay\n\nSep\n\nJan\n\nMay\n\nSep\n\nJan\n\nMay\n\nSep\n\nJan\n\nMay\n\nSep\n\nJan\n\nMay\n\nSep\n\nJan\n\nMay\n\nSep\n\nJan\n\nMay\n\nSep\n\nJan\n\nMay\n\nSep\n\nJan\n\nMay\n\nSep\n\nJan\n\nMay\n\nSep\n\nJan\n\nMay\n\n2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024\n\n_% of respondents in respective group_\n\nlowest quartile\n\nTOTAL\n\nhighest quartile\n\nsecond quartile\n\nthird quartile\n\n_% need to draw on savings_\n\nFinancial distress by income quartile\n\n_% need to run into debt_\n\n0\n\n50\n\n100\n\n150\n\n200\n\n250\n\nEL\n\nCY\n\nFR\n\nIT\n\nES\n\nFI\n\nDK\n\nDE\n\nBE\n\nAT\n\nSK\n\nLU\n\nPT\n\nNL\n\nSE\n\nEU\n\nSI\n\nCZ\n\nIE\n\nHR\n\nMT\n\nHU\n\nBG\n\nEE\n\nPL\n\nLV\n\nLT\n\nRO\n\n2022 2023 2015","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/employment_and_social_developments_in_europe_esde_2024_report_p31.pdf"]},"response":{"errors":[],"extract_id":"extract_07a8f2f11af7bc8f4207f6403b9fc1c4","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Chapter 1: Main employment and social developments\n\n33\n\nChart 1.18 Persistent financial distress over the last year, especially for lower-income households Reported financial distress by income quartile, 2012-2024, EU\n\n_Note:_ Lines show the long-term averages for financial distress for the population as a whole and for households in the four income quartiles. Overall share of adults reporting having to draw on savings and/or run into debt are in light grey and dark grey, respectively, which together represent total financial distress.\n\n_Source:_ DG EMPL calculations based on Business and Consumer Surveys, unadjusted data, 12-month moving average. Click here to download chart.\n\n**Lower-income households still experienced particularly elevated levels of financial distress.** In July 2024 this indicator reached 27.6% for the lowest income quartile (-0.7 pp on a yearly basis), remaining at almost 10 pp or more above the shares for other income quartiles. It amounted to 17.9% for the second quartile (- 0.8 pp), 12.5% for the third quartile (-0.8 pp), and 7.6% for the wealthiest quartile (-0.7 pp) (Chart 1.18). Persistent financial distress, especially for lower-income households, could be partly due to average price increases for certain consumer items in 2023, notably food, and ongoing elevated energy prices.\n\nChart 1.19 Real median disposable income did not increase in all countries since 2010 Median equivalised disposable income in real terms (Index SILC 2010 = 100)\n\n_Source:_ Eurostat [ILC\\_DI18].\n\nClick here to download chart.\n\n**Over the last decade, real median disposable income increased for the EU average but remained at 2015 (2014 incomes) levels in 2023 (2022 incomes) in a number of Member States.** This includes some major EU economies (Chart 1.19). The real median disposable income at EU level increased from 97% of the 2010 real median income in 2015 (2014 incomes) to 118.5% in 2023 (2022 incomes). It remained quite stable since 2020. The trend varied considerably across the Member States. In France, Italy, Finland, Denmark, Germany, Belgium, Austria, Slovakia and Sweden, the 2023 real median disposable income remained at 2015 levels, while it increased most in Romania, Latvia, Lithuania, Poland, Bulgaria and Hungary. In Greece, the real median disposable income remains well below the pre-financial crisis level. In nominal terms, the flash estimates (2023 income) nowcast an overall increase of 6.0% for 2023 incomes, with positive changes estimated for all countries. In real terms, however, a slight, non-statistically significant increase (0.2%) is expected at EU level, while at\n\n0\n\n3\n\n6\n\n9\n\n12\n\n15\n\n18\n\n21\n\n24\n\n27\n\n30\n\nJan\n\nMay\n\nSep\n\nJan\n\nMay\n\nSep\n\nJan\n\nMay\n\nSep\n\nJan\n\nMay\n\nSep\n\nJan\n\nMay\n\nSep\n\nJan\n\nMay\n\nSep\n\nJan\n\nMay\n\nSep\n\nJan\n\nMay\n\nSep\n\nJan\n\nMay\n\nSep\n\nJan\n\nMay\n\nSep\n\nJan\n\nMay\n\nSep\n\nJan\n\nMay\n\nSep\n\nJan\n\nMay\n\n2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024\n\n_% of respondents in respective group_\n\nlowest quartile\n\nTOTAL\n\nhighest quartile\n\nsecond quartile\n\nthird quartile\n\n_% need to draw on savings_\n\nFinancial distress by income quartile\n\n_% need to run into debt_\n\n0\n\n50\n\n100\n\n150\n\n200\n\n250\n\nEL\n\nCY\n\nFR\n\nIT\n\nES\n\nFI\n\nDK\n\nDE\n\nBE\n\nAT\n\nSK\n\nLU\n\nPT\n\nNL\n\nSE\n\nEU\n\nSI\n\nCZ\n\nIE\n\nHR\n\nMT\n\nHU\n\nBG\n\nEE\n\nPL\n\nLV\n\nLT\n\nRO\n\n2022 2023 2015"],"full_content":"Chapter 1: Main employment and social developments\n\n33\n\nChart 1.18 Persistent financial distress over the last year, especially for lower-income households Reported financial distress by income quartile, 2012-2024, EU\n\n_Note:_ Lines show the long-term averages for financial distress for the population as a whole and for households in the four income quartiles. Overall share of adults reporting having to draw on savings and/or run into debt are in light grey and dark grey, respectively, which together represent total financial distress.\n\n_Source:_ DG EMPL calculations based on Business and Consumer Surveys, unadjusted data, 12-month moving average. Click here to download chart.\n\n**Lower-income households still experienced particularly elevated levels of financial distress.** In July 2024 this indicator reached 27.6% for the lowest income quartile (-0.7 pp on a yearly basis), remaining at almost 10 pp or more above the shares for other income quartiles. It amounted to 17.9% for the second quartile (- 0.8 pp), 12.5% for the third quartile (-0.8 pp), and 7.6% for the wealthiest quartile (-0.7 pp) (Chart 1.18). Persistent financial distress, especially for lower-income households, could be partly due to average price increases for certain consumer items in 2023, notably food, and ongoing elevated energy prices.\n\nChart 1.19 Real median disposable income did not increase in all countries since 2010 Median equivalised disposable income in real terms (Index SILC 2010 = 100)\n\n_Source:_ Eurostat [ILC\\_DI18].\n\nClick here to download chart.\n\n**Over the last decade, real median disposable income increased for the EU average but remained at 2015 (2014 incomes) levels in 2023 (2022 incomes) in a number of Member States.** This includes some major EU economies (Chart 1.19). The real median disposable income at EU level increased from 97% of the 2010 real median income in 2015 (2014 incomes) to 118.5% in 2023 (2022 incomes). It remained quite stable since 2020. The trend varied considerably across the Member States. In France, Italy, Finland, Denmark, Germany, Belgium, Austria, Slovakia and Sweden, the 2023 real median disposable income remained at 2015 levels, while it increased most in Romania, Latvia, Lithuania, Poland, Bulgaria and Hungary. In Greece, the real median disposable income remains well below the pre-financial crisis level. In nominal terms, the flash estimates (2023 income) nowcast an overall increase of 6.0% for 2023 incomes, with positive changes estimated for all countries. In real terms, however, a slight, non-statistically significant increase (0.2%) is expected at EU level, while at\n\n0\n\n3\n\n6\n\n9\n\n12\n\n15\n\n18\n\n21\n\n24\n\n27\n\n30\n\nJan\n\nMay\n\nSep\n\nJan\n\nMay\n\nSep\n\nJan\n\nMay\n\nSep\n\nJan\n\nMay\n\nSep\n\nJan\n\nMay\n\nSep\n\nJan\n\nMay\n\nSep\n\nJan\n\nMay\n\nSep\n\nJan\n\nMay\n\nSep\n\nJan\n\nMay\n\nSep\n\nJan\n\nMay\n\nSep\n\nJan\n\nMay\n\nSep\n\nJan\n\nMay\n\nSep\n\nJan\n\nMay\n\n2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024\n\n_% of respondents in respective group_\n\nlowest quartile\n\nTOTAL\n\nhighest quartile\n\nsecond quartile\n\nthird quartile\n\n_% need to draw on savings_\n\nFinancial distress by income quartile\n\n_% need to run into debt_\n\n0\n\n50\n\n100\n\n150\n\n200\n\n250\n\nEL\n\nCY\n\nFR\n\nIT\n\nES\n\nFI\n\nDK\n\nDE\n\nBE\n\nAT\n\nSK\n\nLU\n\nPT\n\nNL\n\nSE\n\nEU\n\nSI\n\nCZ\n\nIE\n\nHR\n\nMT\n\nHU\n\nBG\n\nEE\n\nPL\n\nLV\n\nLT\n\nRO\n\n2022 2023 2015","publish_date":null,"title":"Persistent financial distress over the last year, especially for lower-income households In July Lower-income households still experienced particularly elevated levels","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/employment_and_social_developments_in_europe_esde_2024_report_p31.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/employment_and_social_developments_in_europe_esde_2024_report_p31.pdf"},"request":{"example_id":"chart/employment_and_social_developments_in_europe_esde_2024_report_p31","product_type":"parse","source_file_path":"employment_and_social_developments_in_europe_esde_2024_report_p31.pdf"},"started_at":"2026-11-04T19:45:18.308542"} +{"completed_at":"2026-11-04T19:45:20.090629","latency_in_ms":1132,"output":{"example_id":"chart/employment_and_social_developments_in_europe_esde_2024_report_p34","grounded_pages":[],"layout_pages":[],"markdown":"Employment and Social Developments in Europe 2024\n\n36\n\nChart 1.23 Work protects against poverty, but still imperfectly; Severe material and social deprivation at high levels and slightly increasing for the lowest income group Left chart: share of population at-risk-of-poverty (AROP rate) for selected characteristics, 2015-2023, EU; Right chart: share of population in severe material and social deprivation (SMSD rate) by income quintile, 2021-2023, EU\n\n_Note:_ Year refers to EU-SILC survey year; income from previous year. AROP rate: income from previous year, SMSD rate: current year. AROP rate between 2015 and 2018 is estimated. AROP rate break in time series in 2020.\n\n_Source:_ Eurostat, EU SILC Eurostat, EU SILC [ILC\\_LI02] [ILC\\_LI04] [ILC\\_IW01] [ILC\\_PNS6] [ILC\\_MDD13].\n\nClick here to download chart.\n\n**Despite relatively favourable social outcomes, the proportion of the population experiencing energy poverty further increased in 2023.** Energy poverty here operationalised and measured by the ‘inability to keep home adequately warm’, increased by 1.3 pp in 2023, reaching 10.6% (compared to 9.3% in 2022 and 6.9% in 2021). There are large variations across Member States, with energy poverty particularly high in Spain (20.8%), Portugal (20.8%), Bulgaria (20.7%), Lithuania (20.0%) and Greece (19.2%). In the context of high energy prices, energy poverty increased more sharply for the population at risk of poverty (by 2.1 pp), to 22.2% in 2023, reaching more than double the EU average (Chart 1.24, left). The proportion of single people with dependent children experiencing energy poverty has surpassed 2012 levels (Chart 1.24, right). The share of households with arrears on utility bills remained stable compared to 2022, while the proportion of households living in a dwelling with a leaking roof, damp walls, floors or foundation, or rot in window frames or floor increased by 0.7 pp to 15.5% in 2023. Household expenditure on housing, water, electricity gas and other fuels have risen steadily since 2010. ( ⁴⁹ )\n\nChart 1.24 Peak of population unable to keep their home adequately warm in 2023 Left chart: share of population below/above the at-risk-of-poverty threshold unable to keep home adequately warm and average annual index (2015=100) for electricity, gas and other fuels, 2012-2023, EU; Right chart: share of population below/above the at-risk-of-poverty threshold unable to keep home adequately warm by household composition, 2021-2023, EU.\n\n_Source:_ Eurostat, EU-SILC [ILC\\_MDES01] [PRC\\_HICP\\_AIND]. Click here to download chart.\n\n### 5\\. CONCLUSIONS\n\n**In 2023, EU growth was weak in the context of decreasing, but elevated, inflation levels.** Inflation declined throughout the year, initially driven primarily by the decrease in energy prices. Real GDP increased by 0.4% in 2023 and is expected to expand by 1.0% in 2024 and by 1.6% in 2025. However, uncertainties and significant risks persist, especially as geopolitical tensions continue.\n\n**In recent years, the labour market remained resilient, with employment at record highs and unemployment at record lows.** The employment rate reached its highest-ever level in 2023, at 75.3%, while unemployment remained at 6.1%. The labour market situations of population groups in vulnerable situations showed improvement, albeit with significant differences to the general population. Employment growth is expected to be limited in 2024 and 2025.\n\n( ⁴⁹ ) Eurostat [hbs\\_exp\\_t121]\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\n2015 2016 2017 2018 2019 2020 2021 2022 2023\n\nChildren Unemployed\n\nIn-work Pensioners\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\n16\n\n18\n\n20\n\n1st quintile\n\n2nd quintile\n\n3rd quintile\n\n4th quintile\n\n5th quintile\n\nTotal\n\n2021 2022 2023\n\n0\n\n50\n\n100\n\n150\n\n200\n\n0\n\n10\n\n20\n\n30\n\n2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023\n\nAv. annual index (2015=100) - Elec., gas and other fuels (rhs) Above 60% (lhs) Below 60% (lhs)\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\nAbove 60% Below 60% Above 60% Below 60% Above 60% Below 60%\n\nSingle person with dependent children\n\nHH with dependent children HH without dependent\n\nchildren\n\n2021 2022 2023 2012","pages":[{"line_numbers":[],"markdown":"Employment and Social Developments in Europe 2024\n\n36\n\nChart 1.23 Work protects against poverty, but still imperfectly; Severe material and social deprivation at high levels and slightly increasing for the lowest income group Left chart: share of population at-risk-of-poverty (AROP rate) for selected characteristics, 2015-2023, EU; Right chart: share of population in severe material and social deprivation (SMSD rate) by income quintile, 2021-2023, EU\n\n_Note:_ Year refers to EU-SILC survey year; income from previous year. AROP rate: income from previous year, SMSD rate: current year. AROP rate between 2015 and 2018 is estimated. AROP rate break in time series in 2020.\n\n_Source:_ Eurostat, EU SILC Eurostat, EU SILC [ILC\\_LI02] [ILC\\_LI04] [ILC\\_IW01] [ILC\\_PNS6] [ILC\\_MDD13].\n\nClick here to download chart.\n\n**Despite relatively favourable social outcomes, the proportion of the population experiencing energy poverty further increased in 2023.** Energy poverty here operationalised and measured by the ‘inability to keep home adequately warm’, increased by 1.3 pp in 2023, reaching 10.6% (compared to 9.3% in 2022 and 6.9% in 2021). There are large variations across Member States, with energy poverty particularly high in Spain (20.8%), Portugal (20.8%), Bulgaria (20.7%), Lithuania (20.0%) and Greece (19.2%). In the context of high energy prices, energy poverty increased more sharply for the population at risk of poverty (by 2.1 pp), to 22.2% in 2023, reaching more than double the EU average (Chart 1.24, left). The proportion of single people with dependent children experiencing energy poverty has surpassed 2012 levels (Chart 1.24, right). The share of households with arrears on utility bills remained stable compared to 2022, while the proportion of households living in a dwelling with a leaking roof, damp walls, floors or foundation, or rot in window frames or floor increased by 0.7 pp to 15.5% in 2023. Household expenditure on housing, water, electricity gas and other fuels have risen steadily since 2010. ( ⁴⁹ )\n\nChart 1.24 Peak of population unable to keep their home adequately warm in 2023 Left chart: share of population below/above the at-risk-of-poverty threshold unable to keep home adequately warm and average annual index (2015=100) for electricity, gas and other fuels, 2012-2023, EU; Right chart: share of population below/above the at-risk-of-poverty threshold unable to keep home adequately warm by household composition, 2021-2023, EU.\n\n_Source:_ Eurostat, EU-SILC [ILC\\_MDES01] [PRC\\_HICP\\_AIND]. Click here to download chart.\n\n### 5\\. CONCLUSIONS\n\n**In 2023, EU growth was weak in the context of decreasing, but elevated, inflation levels.** Inflation declined throughout the year, initially driven primarily by the decrease in energy prices. Real GDP increased by 0.4% in 2023 and is expected to expand by 1.0% in 2024 and by 1.6% in 2025. However, uncertainties and significant risks persist, especially as geopolitical tensions continue.\n\n**In recent years, the labour market remained resilient, with employment at record highs and unemployment at record lows.** The employment rate reached its highest-ever level in 2023, at 75.3%, while unemployment remained at 6.1%. The labour market situations of population groups in vulnerable situations showed improvement, albeit with significant differences to the general population. Employment growth is expected to be limited in 2024 and 2025.\n\n( ⁴⁹ ) Eurostat [hbs\\_exp\\_t121]\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\n2015 2016 2017 2018 2019 2020 2021 2022 2023\n\nChildren Unemployed\n\nIn-work Pensioners\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\n16\n\n18\n\n20\n\n1st quintile\n\n2nd quintile\n\n3rd quintile\n\n4th quintile\n\n5th quintile\n\nTotal\n\n2021 2022 2023\n\n0\n\n50\n\n100\n\n150\n\n200\n\n0\n\n10\n\n20\n\n30\n\n2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023\n\nAv. annual index (2015=100) - Elec., gas and other fuels (rhs) Above 60% (lhs) Below 60% (lhs)\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\nAbove 60% Below 60% Above 60% Below 60% Above 60% Below 60%\n\nSingle person with dependent children\n\nHH with dependent children HH without dependent\n\nchildren\n\n2021 2022 2023 2012","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Employment and Social Developments in Europe 2024\n\n36\n\nChart 1.23 Work protects against poverty, but still imperfectly; Severe material and social deprivation at high levels and slightly increasing for the lowest income group Left chart: share of population at-risk-of-poverty (AROP rate) for selected characteristics, 2015-2023, EU; Right chart: share of population in severe material and social deprivation (SMSD rate) by income quintile, 2021-2023, EU\n\n_Note:_ Year refers to EU-SILC survey year; income from previous year. AROP rate: income from previous year, SMSD rate: current year. AROP rate between 2015 and 2018 is estimated. AROP rate break in time series in 2020.\n\n_Source:_ Eurostat, EU SILC Eurostat, EU SILC [ILC\\_LI02] [ILC\\_LI04] [ILC\\_IW01] [ILC\\_PNS6] [ILC\\_MDD13].\n\nClick here to download chart.\n\n**Despite relatively favourable social outcomes, the proportion of the population experiencing energy poverty further increased in 2023.** Energy poverty here operationalised and measured by the ‘inability to keep home adequately warm’, increased by 1.3 pp in 2023, reaching 10.6% (compared to 9.3% in 2022 and 6.9% in 2021). There are large variations across Member States, with energy poverty particularly high in Spain (20.8%), Portugal (20.8%), Bulgaria (20.7%), Lithuania (20.0%) and Greece (19.2%). In the context of high energy prices, energy poverty increased more sharply for the population at risk of poverty (by 2.1 pp), to 22.2% in 2023, reaching more than double the EU average (Chart 1.24, left). The proportion of single people with dependent children experiencing energy poverty has surpassed 2012 levels (Chart 1.24, right). The share of households with arrears on utility bills remained stable compared to 2022, while the proportion of households living in a dwelling with a leaking roof, damp walls, floors or foundation, or rot in window frames or floor increased by 0.7 pp to 15.5% in 2023. Household expenditure on housing, water, electricity gas and other fuels have risen steadily since 2010. ( ⁴⁹ )\n\nChart 1.24 Peak of population unable to keep their home adequately warm in 2023 Left chart: share of population below/above the at-risk-of-poverty threshold unable to keep home adequately warm and average annual index (2015=100) for electricity, gas and other fuels, 2012-2023, EU; Right chart: share of population below/above the at-risk-of-poverty threshold unable to keep home adequately warm by household composition, 2021-2023, EU.\n\n_Source:_ Eurostat, EU-SILC [ILC\\_MDES01] [PRC\\_HICP\\_AIND]. Click here to download chart.\n\n### 5\\. CONCLUSIONS\n\n**In 2023, EU growth was weak in the context of decreasing, but elevated, inflation levels.** Inflation declined throughout the year, initially driven primarily by the decrease in energy prices. Real GDP increased by 0.4% in 2023 and is expected to expand by 1.0% in 2024 and by 1.6% in 2025. However, uncertainties and significant risks persist, especially as geopolitical tensions continue.\n\n**In recent years, the labour market remained resilient, with employment at record highs and unemployment at record lows.** The employment rate reached its highest-ever level in 2023, at 75.3%, while unemployment remained at 6.1%. The labour market situations of population groups in vulnerable situations showed improvement, albeit with significant differences to the general population. Employment growth is expected to be limited in 2024 and 2025.\n\n( ⁴⁹ ) Eurostat [hbs\\_exp\\_t121]\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\n2015 2016 2017 2018 2019 2020 2021 2022 2023\n\nChildren Unemployed\n\nIn-work Pensioners\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\n16\n\n18\n\n20\n\n1st quintile\n\n2nd quintile\n\n3rd quintile\n\n4th quintile\n\n5th quintile\n\nTotal\n\n2021 2022 2023\n\n0\n\n50\n\n100\n\n150\n\n200\n\n0\n\n10\n\n20\n\n30\n\n2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023\n\nAv. annual index (2015=100) - Elec., gas and other fuels (rhs) Above 60% (lhs) Below 60% (lhs)\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\nAbove 60% Below 60% Above 60% Below 60% Above 60% Below 60%\n\nSingle person with dependent children\n\nHH with dependent children HH without dependent\n\nchildren\n\n2021 2022 2023 2012","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/employment_and_social_developments_in_europe_esde_2024_report_p34.pdf"]},"response":{"errors":[],"extract_id":"extract_2bca09a4ef6968eda50645de35f9c69c","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Employment and Social Developments in Europe 2024\n\n36\n\nChart 1.23 Work protects against poverty, but still imperfectly; Severe material and social deprivation at high levels and slightly increasing for the lowest income group Left chart: share of population at-risk-of-poverty (AROP rate) for selected characteristics, 2015-2023, EU; Right chart: share of population in severe material and social deprivation (SMSD rate) by income quintile, 2021-2023, EU\n\n_Note:_ Year refers to EU-SILC survey year; income from previous year. AROP rate: income from previous year, SMSD rate: current year. AROP rate between 2015 and 2018 is estimated. AROP rate break in time series in 2020.\n\n_Source:_ Eurostat, EU SILC Eurostat, EU SILC [ILC\\_LI02] [ILC\\_LI04] [ILC\\_IW01] [ILC\\_PNS6] [ILC\\_MDD13].\n\nClick here to download chart.\n\n**Despite relatively favourable social outcomes, the proportion of the population experiencing energy poverty further increased in 2023.** Energy poverty here operationalised and measured by the ‘inability to keep home adequately warm’, increased by 1.3 pp in 2023, reaching 10.6% (compared to 9.3% in 2022 and 6.9% in 2021). There are large variations across Member States, with energy poverty particularly high in Spain (20.8%), Portugal (20.8%), Bulgaria (20.7%), Lithuania (20.0%) and Greece (19.2%). In the context of high energy prices, energy poverty increased more sharply for the population at risk of poverty (by 2.1 pp), to 22.2% in 2023, reaching more than double the EU average (Chart 1.24, left). The proportion of single people with dependent children experiencing energy poverty has surpassed 2012 levels (Chart 1.24, right). The share of households with arrears on utility bills remained stable compared to 2022, while the proportion of households living in a dwelling with a leaking roof, damp walls, floors or foundation, or rot in window frames or floor increased by 0.7 pp to 15.5% in 2023. Household expenditure on housing, water, electricity gas and other fuels have risen steadily since 2010. ( ⁴⁹ )\n\nChart 1.24 Peak of population unable to keep their home adequately warm in 2023 Left chart: share of population below/above the at-risk-of-poverty threshold unable to keep home adequately warm and average annual index (2015=100) for electricity, gas and other fuels, 2012-2023, EU; Right chart: share of population below/above the at-risk-of-poverty threshold unable to keep home adequately warm by household composition, 2021-2023, EU.\n\n_Source:_ Eurostat, EU-SILC [ILC\\_MDES01] [PRC\\_HICP\\_AIND]. Click here to download chart.\n\n### 5\\. CONCLUSIONS\n\n**In 2023, EU growth was weak in the context of decreasing, but elevated, inflation levels.** Inflation declined throughout the year, initially driven primarily by the decrease in energy prices. Real GDP increased by 0.4% in 2023 and is expected to expand by 1.0% in 2024 and by 1.6% in 2025. However, uncertainties and significant risks persist, especially as geopolitical tensions continue.\n\n**In recent years, the labour market remained resilient, with employment at record highs and unemployment at record lows.** The employment rate reached its highest-ever level in 2023, at 75.3%, while unemployment remained at 6.1%. The labour market situations of population groups in vulnerable situations showed improvement, albeit with significant differences to the general population. Employment growth is expected to be limited in 2024 and 2025.\n\n( ⁴⁹ ) Eurostat [hbs\\_exp\\_t121]\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\n2015 2016 2017 2018 2019 2020 2021 2022 2023\n\nChildren Unemployed\n\nIn-work Pensioners\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\n16\n\n18\n\n20\n\n1st quintile\n\n2nd quintile\n\n3rd quintile\n\n4th quintile\n\n5th quintile\n\nTotal\n\n2021 2022 2023\n\n0\n\n50\n\n100\n\n150\n\n200\n\n0\n\n10\n\n20\n\n30\n\n2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023\n\nAv. annual index (2015=100) - Elec., gas and other fuels (rhs) Above 60% (lhs) Below 60% (lhs)\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\nAbove 60% Below 60% Above 60% Below 60% Above 60% Below 60%\n\nSingle person with dependent children\n\nHH with dependent children HH without dependent\n\nchildren\n\n2021 2022 2023 2012"],"full_content":"Employment and Social Developments in Europe 2024\n\n36\n\nChart 1.23 Work protects against poverty, but still imperfectly; Severe material and social deprivation at high levels and slightly increasing for the lowest income group Left chart: share of population at-risk-of-poverty (AROP rate) for selected characteristics, 2015-2023, EU; Right chart: share of population in severe material and social deprivation (SMSD rate) by income quintile, 2021-2023, EU\n\n_Note:_ Year refers to EU-SILC survey year; income from previous year. AROP rate: income from previous year, SMSD rate: current year. AROP rate between 2015 and 2018 is estimated. AROP rate break in time series in 2020.\n\n_Source:_ Eurostat, EU SILC Eurostat, EU SILC [ILC\\_LI02] [ILC\\_LI04] [ILC\\_IW01] [ILC\\_PNS6] [ILC\\_MDD13].\n\nClick here to download chart.\n\n**Despite relatively favourable social outcomes, the proportion of the population experiencing energy poverty further increased in 2023.** Energy poverty here operationalised and measured by the ‘inability to keep home adequately warm’, increased by 1.3 pp in 2023, reaching 10.6% (compared to 9.3% in 2022 and 6.9% in 2021). There are large variations across Member States, with energy poverty particularly high in Spain (20.8%), Portugal (20.8%), Bulgaria (20.7%), Lithuania (20.0%) and Greece (19.2%). In the context of high energy prices, energy poverty increased more sharply for the population at risk of poverty (by 2.1 pp), to 22.2% in 2023, reaching more than double the EU average (Chart 1.24, left). The proportion of single people with dependent children experiencing energy poverty has surpassed 2012 levels (Chart 1.24, right). The share of households with arrears on utility bills remained stable compared to 2022, while the proportion of households living in a dwelling with a leaking roof, damp walls, floors or foundation, or rot in window frames or floor increased by 0.7 pp to 15.5% in 2023. Household expenditure on housing, water, electricity gas and other fuels have risen steadily since 2010. ( ⁴⁹ )\n\nChart 1.24 Peak of population unable to keep their home adequately warm in 2023 Left chart: share of population below/above the at-risk-of-poverty threshold unable to keep home adequately warm and average annual index (2015=100) for electricity, gas and other fuels, 2012-2023, EU; Right chart: share of population below/above the at-risk-of-poverty threshold unable to keep home adequately warm by household composition, 2021-2023, EU.\n\n_Source:_ Eurostat, EU-SILC [ILC\\_MDES01] [PRC\\_HICP\\_AIND]. Click here to download chart.\n\n### 5\\. CONCLUSIONS\n\n**In 2023, EU growth was weak in the context of decreasing, but elevated, inflation levels.** Inflation declined throughout the year, initially driven primarily by the decrease in energy prices. Real GDP increased by 0.4% in 2023 and is expected to expand by 1.0% in 2024 and by 1.6% in 2025. However, uncertainties and significant risks persist, especially as geopolitical tensions continue.\n\n**In recent years, the labour market remained resilient, with employment at record highs and unemployment at record lows.** The employment rate reached its highest-ever level in 2023, at 75.3%, while unemployment remained at 6.1%. The labour market situations of population groups in vulnerable situations showed improvement, albeit with significant differences to the general population. Employment growth is expected to be limited in 2024 and 2025.\n\n( ⁴⁹ ) Eurostat [hbs\\_exp\\_t121]\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\n2015 2016 2017 2018 2019 2020 2021 2022 2023\n\nChildren Unemployed\n\nIn-work Pensioners\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\n16\n\n18\n\n20\n\n1st quintile\n\n2nd quintile\n\n3rd quintile\n\n4th quintile\n\n5th quintile\n\nTotal\n\n2021 2022 2023\n\n0\n\n50\n\n100\n\n150\n\n200\n\n0\n\n10\n\n20\n\n30\n\n2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023\n\nAv. annual index (2015=100) - Elec., gas and other fuels (rhs) Above 60% (lhs) Below 60% (lhs)\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\nAbove 60% Below 60% Above 60% Below 60% Above 60% Below 60%\n\nSingle person with dependent children\n\nHH with dependent children HH without dependent\n\nchildren\n\n2021 2022 2023 2012","publish_date":null,"title":"5. C","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/employment_and_social_developments_in_europe_esde_2024_report_p34.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/employment_and_social_developments_in_europe_esde_2024_report_p34.pdf"},"request":{"example_id":"chart/employment_and_social_developments_in_europe_esde_2024_report_p34","product_type":"parse","source_file_path":"employment_and_social_developments_in_europe_esde_2024_report_p34.pdf"},"started_at":"2026-11-04T19:45:18.958604"} +{"completed_at":"2026-11-04T19:45:22.723673","latency_in_ms":2710,"output":{"example_id":"chart/employment_and_social_developments_in_europe_esde_2024_report_p40","grounded_pages":[],"layout_pages":[],"markdown":"Chapter 2: Social convergence in the EU: taking stock\n\n43\n\n_(Continued on the next page)_\n\nBox 2.1: Convergence in national wealth levels and distribution\n\n**Median net wealth was stable for most of the 2010s** **with disparities between countries declining** (Chart 1). Between 2010 and 2017, there were only mild changes in median wealth across the 14 eurozone countries that participated in the Household Finance and Consumption Survey (HFCS), ( ¹ ) accompanied by oscillating variation in wealth levels across countries. Higher saving rates and higher prices of certain assets (notably real estate) triggered throughout the COVID-19 pandemic resulted in an increase in net wealth in every country in the euro area, except for Finland. ( ² ) That increase coincided with a drop in wealth disparity between countries. No post-COVID-19 household- level data on wealth are available to assess the durability of the changes during the pandemic.\n\nChart 1 Median wealth has increased on average in the euro area, with countries converging upwards Average and standard deviation of median net wealth, 2010 – 2021, euro area\n\n_Notes:_ Median household net wealth in 2021 EUR purchasing power standard (PPS). Dispersion measured as standard deviation of log-transformed median wealth by\n\ncountry. Dark blue line includes the 14 countries participating in both wave 1 and wave 4 of the HFCS: Belgium, Germany, crGreece, Spain, France, Italy, Cyprus, Luxembourg, Malta, the Netherlands, Austria, Portugal, Slovakia, Finland. Light green line also includes Estonia, Ireland, Latvia, Hungary, Slovenia. Yellow line, includes Croatia and Lithuania, in addition to the 19 countries already listed above. Standard deviation is a measure of cross-country variation, the higher the standard deviation, the higher the cross-country variation.\n\n_Source:_ OECD calculations based on the Eurosystem Household Finance and Consumption Survey (HFCS), https://www.ecb.europa.eu/stats/ecb\\_surveys/hfcs/html/index.en.html. **Since 2014, net wealth grew for all income quintiles, benefitting mostly top earners.** Increases in household net wealth were relatively modest between 2014 and 2017, with median wealth among the top fifth of earners (top quintile) recording the largest growth (15%). Between 2017 and 2021, median household net wealth grew over 20% for all income quintiles. In absolute terms, the top income quintile still benefitted most from this growth, increasing their net wealth by EUR 50,000 (PPS) between 2017 and 2021 (Chart 2).\n\n**Wealth concentration increased after the financial crisis but decreased during COVID-19.** The distribution of wealth became more unequal between 2010 and 2017 (Chart 3), with an increase in the share of wealth held by the top 10% at the expense of the bottom 50%. Eleven of the 14 countries with data available since 2010 saw the wealth share held by the top 10% increase between 2014 and 2017 (from 48% to 50%). ( ³ ) At the same time, variation in wealth concentration across countries decreased until 2017 (Chart 3), as wealth in less unequal countries became more concentrated. Breaking the previous trend, during the COVID-19 pandemic there was a decrease in the wealth share of the top 10%, accompanied by a rising share of wealth held by the bottom 50%. ( ⁴ )\n\n( ¹ ) HFCS data have some limitations in capturing the concentration of wealth at the very top of the wealth distribution due to\n\ndifficulties in covering the wealthiest individuals in a survey and due to underreporting of wealth at the top.\n\n( ² ) For the purposes of this analysis, “euro area” here refers to the group of countries represented in the HFCS, i.e. including\n\nHungary in wave 2 and Croatia in wave 3 (joined the eurozone in 2023). Figure 1 note contains a full list of countries included in each time series.\n\n( ³ ) The increase in concentration of wealth between 2014 and 2017 is also apparent for the broader group of 19 countries for\n\nwhich data are available since 2014, although the patterns are weaker.\n\n( ⁴ ) In general, this was due to lower debt levels and higher housing prices, both of which favour the bottom and middle of the\n\ndistribution. Financial asset prices were also lower in 2020 than in 2017 for most countries but grew in 2021.\n\n€ 0\n\n€ 20 000\n\n€ 40 000\n\n€ 60 000\n\n€ 80 000\n\n€ 100 000\n\n€ 120 000\n\n€ 140 000\n\n€ 160 000\n\n€ 180 000\n\n2010 2014 2017 2021\n\nPanel A: Median net wealth\n\n14 countries 19 countries\n\n21 countries\n\n0\\.5\n\n0\\.52\n\n0\\.54\n\n0\\.56\n\n0\\.58\n\n0\\.6\n\n0\\.62\n\n0\\.64\n\n0\\.66\n\n2010 2014 2017 2021\n\nPanel B: Dispersion of median\n\nwealth\n\n14 countries 19 countries\n\n21 countries","pages":[{"line_numbers":[],"markdown":"Chapter 2: Social convergence in the EU: taking stock\n\n43\n\n_(Continued on the next page)_\n\nBox 2.1: Convergence in national wealth levels and distribution\n\n**Median net wealth was stable for most of the 2010s** **with disparities between countries declining** (Chart 1). Between 2010 and 2017, there were only mild changes in median wealth across the 14 eurozone countries that participated in the Household Finance and Consumption Survey (HFCS), ( ¹ ) accompanied by oscillating variation in wealth levels across countries. Higher saving rates and higher prices of certain assets (notably real estate) triggered throughout the COVID-19 pandemic resulted in an increase in net wealth in every country in the euro area, except for Finland. ( ² ) That increase coincided with a drop in wealth disparity between countries. No post-COVID-19 household- level data on wealth are available to assess the durability of the changes during the pandemic.\n\nChart 1 Median wealth has increased on average in the euro area, with countries converging upwards Average and standard deviation of median net wealth, 2010 – 2021, euro area\n\n_Notes:_ Median household net wealth in 2021 EUR purchasing power standard (PPS). Dispersion measured as standard deviation of log-transformed median wealth by\n\ncountry. Dark blue line includes the 14 countries participating in both wave 1 and wave 4 of the HFCS: Belgium, Germany, crGreece, Spain, France, Italy, Cyprus, Luxembourg, Malta, the Netherlands, Austria, Portugal, Slovakia, Finland. Light green line also includes Estonia, Ireland, Latvia, Hungary, Slovenia. Yellow line, includes Croatia and Lithuania, in addition to the 19 countries already listed above. Standard deviation is a measure of cross-country variation, the higher the standard deviation, the higher the cross-country variation.\n\n_Source:_ OECD calculations based on the Eurosystem Household Finance and Consumption Survey (HFCS), https://www.ecb.europa.eu/stats/ecb\\_surveys/hfcs/html/index.en.html. **Since 2014, net wealth grew for all income quintiles, benefitting mostly top earners.** Increases in household net wealth were relatively modest between 2014 and 2017, with median wealth among the top fifth of earners (top quintile) recording the largest growth (15%). Between 2017 and 2021, median household net wealth grew over 20% for all income quintiles. In absolute terms, the top income quintile still benefitted most from this growth, increasing their net wealth by EUR 50,000 (PPS) between 2017 and 2021 (Chart 2).\n\n**Wealth concentration increased after the financial crisis but decreased during COVID-19.** The distribution of wealth became more unequal between 2010 and 2017 (Chart 3), with an increase in the share of wealth held by the top 10% at the expense of the bottom 50%. Eleven of the 14 countries with data available since 2010 saw the wealth share held by the top 10% increase between 2014 and 2017 (from 48% to 50%). ( ³ ) At the same time, variation in wealth concentration across countries decreased until 2017 (Chart 3), as wealth in less unequal countries became more concentrated. Breaking the previous trend, during the COVID-19 pandemic there was a decrease in the wealth share of the top 10%, accompanied by a rising share of wealth held by the bottom 50%. ( ⁴ )\n\n( ¹ ) HFCS data have some limitations in capturing the concentration of wealth at the very top of the wealth distribution due to\n\ndifficulties in covering the wealthiest individuals in a survey and due to underreporting of wealth at the top.\n\n( ² ) For the purposes of this analysis, “euro area” here refers to the group of countries represented in the HFCS, i.e. including\n\nHungary in wave 2 and Croatia in wave 3 (joined the eurozone in 2023). Figure 1 note contains a full list of countries included in each time series.\n\n( ³ ) The increase in concentration of wealth between 2014 and 2017 is also apparent for the broader group of 19 countries for\n\nwhich data are available since 2014, although the patterns are weaker.\n\n( ⁴ ) In general, this was due to lower debt levels and higher housing prices, both of which favour the bottom and middle of the\n\ndistribution. Financial asset prices were also lower in 2020 than in 2017 for most countries but grew in 2021.\n\n€ 0\n\n€ 20 000\n\n€ 40 000\n\n€ 60 000\n\n€ 80 000\n\n€ 100 000\n\n€ 120 000\n\n€ 140 000\n\n€ 160 000\n\n€ 180 000\n\n2010 2014 2017 2021\n\nPanel A: Median net wealth\n\n14 countries 19 countries\n\n21 countries\n\n0\\.5\n\n0\\.52\n\n0\\.54\n\n0\\.56\n\n0\\.58\n\n0\\.6\n\n0\\.62\n\n0\\.64\n\n0\\.66\n\n2010 2014 2017 2021\n\nPanel B: Dispersion of median\n\nwealth\n\n14 countries 19 countries\n\n21 countries","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Chapter 2: Social convergence in the EU: taking stock\n\n43\n\n_(Continued on the next page)_\n\nBox 2.1: Convergence in national wealth levels and distribution\n\n**Median net wealth was stable for most of the 2010s** **with disparities between countries declining** (Chart 1). Between 2010 and 2017, there were only mild changes in median wealth across the 14 eurozone countries that participated in the Household Finance and Consumption Survey (HFCS), ( ¹ ) accompanied by oscillating variation in wealth levels across countries. Higher saving rates and higher prices of certain assets (notably real estate) triggered throughout the COVID-19 pandemic resulted in an increase in net wealth in every country in the euro area, except for Finland. ( ² ) That increase coincided with a drop in wealth disparity between countries. No post-COVID-19 household- level data on wealth are available to assess the durability of the changes during the pandemic.\n\nChart 1 Median wealth has increased on average in the euro area, with countries converging upwards Average and standard deviation of median net wealth, 2010 – 2021, euro area\n\n_Notes:_ Median household net wealth in 2021 EUR purchasing power standard (PPS). Dispersion measured as standard deviation of log-transformed median wealth by\n\ncountry. Dark blue line includes the 14 countries participating in both wave 1 and wave 4 of the HFCS: Belgium, Germany, crGreece, Spain, France, Italy, Cyprus, Luxembourg, Malta, the Netherlands, Austria, Portugal, Slovakia, Finland. Light green line also includes Estonia, Ireland, Latvia, Hungary, Slovenia. Yellow line, includes Croatia and Lithuania, in addition to the 19 countries already listed above. Standard deviation is a measure of cross-country variation, the higher the standard deviation, the higher the cross-country variation.\n\n_Source:_ OECD calculations based on the Eurosystem Household Finance and Consumption Survey (HFCS), https://www.ecb.europa.eu/stats/ecb\\_surveys/hfcs/html/index.en.html. **Since 2014, net wealth grew for all income quintiles, benefitting mostly top earners.** Increases in household net wealth were relatively modest between 2014 and 2017, with median wealth among the top fifth of earners (top quintile) recording the largest growth (15%). Between 2017 and 2021, median household net wealth grew over 20% for all income quintiles. In absolute terms, the top income quintile still benefitted most from this growth, increasing their net wealth by EUR 50,000 (PPS) between 2017 and 2021 (Chart 2).\n\n**Wealth concentration increased after the financial crisis but decreased during COVID-19.** The distribution of wealth became more unequal between 2010 and 2017 (Chart 3), with an increase in the share of wealth held by the top 10% at the expense of the bottom 50%. Eleven of the 14 countries with data available since 2010 saw the wealth share held by the top 10% increase between 2014 and 2017 (from 48% to 50%). ( ³ ) At the same time, variation in wealth concentration across countries decreased until 2017 (Chart 3), as wealth in less unequal countries became more concentrated. Breaking the previous trend, during the COVID-19 pandemic there was a decrease in the wealth share of the top 10%, accompanied by a rising share of wealth held by the bottom 50%. ( ⁴ )\n\n( ¹ ) HFCS data have some limitations in capturing the concentration of wealth at the very top of the wealth distribution due to\n\ndifficulties in covering the wealthiest individuals in a survey and due to underreporting of wealth at the top.\n\n( ² ) For the purposes of this analysis, “euro area” here refers to the group of countries represented in the HFCS, i.e. including\n\nHungary in wave 2 and Croatia in wave 3 (joined the eurozone in 2023). Figure 1 note contains a full list of countries included in each time series.\n\n( ³ ) The increase in concentration of wealth between 2014 and 2017 is also apparent for the broader group of 19 countries for\n\nwhich data are available since 2014, although the patterns are weaker.\n\n( ⁴ ) In general, this was due to lower debt levels and higher housing prices, both of which favour the bottom and middle of the\n\ndistribution. Financial asset prices were also lower in 2020 than in 2017 for most countries but grew in 2021.\n\n€ 0\n\n€ 20 000\n\n€ 40 000\n\n€ 60 000\n\n€ 80 000\n\n€ 100 000\n\n€ 120 000\n\n€ 140 000\n\n€ 160 000\n\n€ 180 000\n\n2010 2014 2017 2021\n\nPanel A: Median net wealth\n\n14 countries 19 countries\n\n21 countries\n\n0\\.5\n\n0\\.52\n\n0\\.54\n\n0\\.56\n\n0\\.58\n\n0\\.6\n\n0\\.62\n\n0\\.64\n\n0\\.66\n\n2010 2014 2017 2021\n\nPanel B: Dispersion of median\n\nwealth\n\n14 countries 19 countries\n\n21 countries","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/employment_and_social_developments_in_europe_esde_2024_report_p40.pdf"]},"response":{"errors":[],"extract_id":"extract_86cc40784fd397a63d72d1b7a8bc2bca","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Chapter 2: Social convergence in the EU: taking stock\n\n43\n\n_(Continued on the next page)_\n\nBox 2.1: Convergence in national wealth levels and distribution\n\n**Median net wealth was stable for most of the 2010s** **with disparities between countries declining** (Chart 1). Between 2010 and 2017, there were only mild changes in median wealth across the 14 eurozone countries that participated in the Household Finance and Consumption Survey (HFCS), ( ¹ ) accompanied by oscillating variation in wealth levels across countries. Higher saving rates and higher prices of certain assets (notably real estate) triggered throughout the COVID-19 pandemic resulted in an increase in net wealth in every country in the euro area, except for Finland. ( ² ) That increase coincided with a drop in wealth disparity between countries. No post-COVID-19 household- level data on wealth are available to assess the durability of the changes during the pandemic.\n\nChart 1 Median wealth has increased on average in the euro area, with countries converging upwards Average and standard deviation of median net wealth, 2010 – 2021, euro area\n\n_Notes:_ Median household net wealth in 2021 EUR purchasing power standard (PPS). Dispersion measured as standard deviation of log-transformed median wealth by\n\ncountry. Dark blue line includes the 14 countries participating in both wave 1 and wave 4 of the HFCS: Belgium, Germany, crGreece, Spain, France, Italy, Cyprus, Luxembourg, Malta, the Netherlands, Austria, Portugal, Slovakia, Finland. Light green line also includes Estonia, Ireland, Latvia, Hungary, Slovenia. Yellow line, includes Croatia and Lithuania, in addition to the 19 countries already listed above. Standard deviation is a measure of cross-country variation, the higher the standard deviation, the higher the cross-country variation.\n\n_Source:_ OECD calculations based on the Eurosystem Household Finance and Consumption Survey (HFCS), https://www.ecb.europa.eu/stats/ecb\\_surveys/hfcs/html/index.en.html. **Since 2014, net wealth grew for all income quintiles, benefitting mostly top earners.** Increases in household net wealth were relatively modest between 2014 and 2017, with median wealth among the top fifth of earners (top quintile) recording the largest growth (15%). Between 2017 and 2021, median household net wealth grew over 20% for all income quintiles. In absolute terms, the top income quintile still benefitted most from this growth, increasing their net wealth by EUR 50,000 (PPS) between 2017 and 2021 (Chart 2).\n\n**Wealth concentration increased after the financial crisis but decreased during COVID-19.** The distribution of wealth became more unequal between 2010 and 2017 (Chart 3), with an increase in the share of wealth held by the top 10% at the expense of the bottom 50%. Eleven of the 14 countries with data available since 2010 saw the wealth share held by the top 10% increase between 2014 and 2017 (from 48% to 50%). ( ³ ) At the same time, variation in wealth concentration across countries decreased until 2017 (Chart 3), as wealth in less unequal countries became more concentrated. Breaking the previous trend, during the COVID-19 pandemic there was a decrease in the wealth share of the top 10%, accompanied by a rising share of wealth held by the bottom 50%. ( ⁴ )\n\n( ¹ ) HFCS data have some limitations in capturing the concentration of wealth at the very top of the wealth distribution due to\n\ndifficulties in covering the wealthiest individuals in a survey and due to underreporting of wealth at the top.\n\n( ² ) For the purposes of this analysis, “euro area” here refers to the group of countries represented in the HFCS, i.e. including\n\nHungary in wave 2 and Croatia in wave 3 (joined the eurozone in 2023). Figure 1 note contains a full list of countries included in each time series.\n\n( ³ ) The increase in concentration of wealth between 2014 and 2017 is also apparent for the broader group of 19 countries for\n\nwhich data are available since 2014, although the patterns are weaker.\n\n( ⁴ ) In general, this was due to lower debt levels and higher housing prices, both of which favour the bottom and middle of the\n\ndistribution. Financial asset prices were also lower in 2020 than in 2017 for most countries but grew in 2021.\n\n€ 0\n\n€ 20 000\n\n€ 40 000\n\n€ 60 000\n\n€ 80 000\n\n€ 100 000\n\n€ 120 000\n\n€ 140 000\n\n€ 160 000\n\n€ 180 000\n\n2010 2014 2017 2021\n\nPanel A: Median net wealth\n\n14 countries 19 countries\n\n21 countries\n\n0\\.5\n\n0\\.52\n\n0\\.54\n\n0\\.56\n\n0\\.58\n\n0\\.6\n\n0\\.62\n\n0\\.64\n\n0\\.66\n\n2010 2014 2017 2021\n\nPanel B: Dispersion of median\n\nwealth\n\n14 countries 19 countries\n\n21 countries"],"full_content":"Chapter 2: Social convergence in the EU: taking stock\n\n43\n\n_(Continued on the next page)_\n\nBox 2.1: Convergence in national wealth levels and distribution\n\n**Median net wealth was stable for most of the 2010s** **with disparities between countries declining** (Chart 1). Between 2010 and 2017, there were only mild changes in median wealth across the 14 eurozone countries that participated in the Household Finance and Consumption Survey (HFCS), ( ¹ ) accompanied by oscillating variation in wealth levels across countries. Higher saving rates and higher prices of certain assets (notably real estate) triggered throughout the COVID-19 pandemic resulted in an increase in net wealth in every country in the euro area, except for Finland. ( ² ) That increase coincided with a drop in wealth disparity between countries. No post-COVID-19 household- level data on wealth are available to assess the durability of the changes during the pandemic.\n\nChart 1 Median wealth has increased on average in the euro area, with countries converging upwards Average and standard deviation of median net wealth, 2010 – 2021, euro area\n\n_Notes:_ Median household net wealth in 2021 EUR purchasing power standard (PPS). Dispersion measured as standard deviation of log-transformed median wealth by\n\ncountry. Dark blue line includes the 14 countries participating in both wave 1 and wave 4 of the HFCS: Belgium, Germany, crGreece, Spain, France, Italy, Cyprus, Luxembourg, Malta, the Netherlands, Austria, Portugal, Slovakia, Finland. Light green line also includes Estonia, Ireland, Latvia, Hungary, Slovenia. Yellow line, includes Croatia and Lithuania, in addition to the 19 countries already listed above. Standard deviation is a measure of cross-country variation, the higher the standard deviation, the higher the cross-country variation.\n\n_Source:_ OECD calculations based on the Eurosystem Household Finance and Consumption Survey (HFCS), https://www.ecb.europa.eu/stats/ecb\\_surveys/hfcs/html/index.en.html. **Since 2014, net wealth grew for all income quintiles, benefitting mostly top earners.** Increases in household net wealth were relatively modest between 2014 and 2017, with median wealth among the top fifth of earners (top quintile) recording the largest growth (15%). Between 2017 and 2021, median household net wealth grew over 20% for all income quintiles. In absolute terms, the top income quintile still benefitted most from this growth, increasing their net wealth by EUR 50,000 (PPS) between 2017 and 2021 (Chart 2).\n\n**Wealth concentration increased after the financial crisis but decreased during COVID-19.** The distribution of wealth became more unequal between 2010 and 2017 (Chart 3), with an increase in the share of wealth held by the top 10% at the expense of the bottom 50%. Eleven of the 14 countries with data available since 2010 saw the wealth share held by the top 10% increase between 2014 and 2017 (from 48% to 50%). ( ³ ) At the same time, variation in wealth concentration across countries decreased until 2017 (Chart 3), as wealth in less unequal countries became more concentrated. Breaking the previous trend, during the COVID-19 pandemic there was a decrease in the wealth share of the top 10%, accompanied by a rising share of wealth held by the bottom 50%. ( ⁴ )\n\n( ¹ ) HFCS data have some limitations in capturing the concentration of wealth at the very top of the wealth distribution due to\n\ndifficulties in covering the wealthiest individuals in a survey and due to underreporting of wealth at the top.\n\n( ² ) For the purposes of this analysis, “euro area” here refers to the group of countries represented in the HFCS, i.e. including\n\nHungary in wave 2 and Croatia in wave 3 (joined the eurozone in 2023). Figure 1 note contains a full list of countries included in each time series.\n\n( ³ ) The increase in concentration of wealth between 2014 and 2017 is also apparent for the broader group of 19 countries for\n\nwhich data are available since 2014, although the patterns are weaker.\n\n( ⁴ ) In general, this was due to lower debt levels and higher housing prices, both of which favour the bottom and middle of the\n\ndistribution. Financial asset prices were also lower in 2020 than in 2017 for most countries but grew in 2021.\n\n€ 0\n\n€ 20 000\n\n€ 40 000\n\n€ 60 000\n\n€ 80 000\n\n€ 100 000\n\n€ 120 000\n\n€ 140 000\n\n€ 160 000\n\n€ 180 000\n\n2010 2014 2017 2021\n\nPanel A: Median net wealth\n\n14 countries 19 countries\n\n21 countries\n\n0\\.5\n\n0\\.52\n\n0\\.54\n\n0\\.56\n\n0\\.58\n\n0\\.6\n\n0\\.62\n\n0\\.64\n\n0\\.66\n\n2010 2014 2017 2021\n\nPanel B: Dispersion of median\n\nwealth\n\n14 countries 19 countries\n\n21 countries","publish_date":null,"title":"Box 2.1: Convergence in national wealth levels and distribution","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/employment_and_social_developments_in_europe_esde_2024_report_p40.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/employment_and_social_developments_in_europe_esde_2024_report_p40.pdf"},"request":{"example_id":"chart/employment_and_social_developments_in_europe_esde_2024_report_p40","product_type":"parse","source_file_path":"employment_and_social_developments_in_europe_esde_2024_report_p40.pdf"},"started_at":"2026-11-04T19:45:20.013636"} +{"completed_at":"2026-11-04T19:45:23.496462","latency_in_ms":2421,"output":{"example_id":"chart/employment_and_social_developments_in_europe_esde_2024_report_p41","grounded_pages":[],"layout_pages":[],"markdown":"Employment and Social Developments in Europe 2024\n\n44\n\n_Box (continued)_\n\n_(Continued on the next page)_\n\nChart 2 Median net wealth grew across the income distribution between 2010 and 2021 Median household net wealth by income quintile, euro area average, 2021 USD PPP, 14 EU countries\n\n_Note:_ Median household net wealth in 2021 EUR PPS. Median household net wealth calculated for: people in the bottom fifth of income earners (lowest quintile); people in the middle fifth of income earners (middle quintile); people in the top fifth of income earners (top quintile).\n\n_Source:_ OECD calculations based on the HFCS, https://www.ecb.europa.eu/stats/ecb\\_surveys/hfcs/html/index.en.html\n\nChart 3 Wealth concentration increased after the financial crisis but decreased during COVID-19 Average and standard deviation of wealth share of top 10% and bottom 50%, 2010-2021\n\n_Note:_ Dispersion measured as standard deviation of wealth share by country. See Figure 1 note for more information. Standard deviation is a measure of cross-country variation, the higher the standard deviation, the higher the cross-country variation.\n\n_Source:_ OECD calculations based on the HFCS.\n\n€ 0\n\n€ 50 000\n\n€ 100 000\n\n€ 150 000\n\n€ 200 000\n\n€ 250 000\n\n€ 300 000\n\n€ 350 000\n\n€ 400 000\n\n€ 450 000\n\n2010 2014 2017 2021\n\nLowest quintile Middle quintile Top quintile\n\n46\\.0% 46.5% 47.0% 47.5% 48.0% 48.5% 49.0% 49.5% 50.0% 50.5% 51.0%\n\n2010 2014 2017 2021\n\nPanel A: Wealth share of the top\n\n10%\n\n14 countries 19 countries\n\n21 countries\n\n0\\.05\n\n0\\.055\n\n0\\.06\n\n0\\.065\n\n0\\.07\n\n0\\.075\n\n0\\.08\n\n0\\.085\n\n2010 2014 2017 2021\n\nPanel B: Dispersion - wealth share\n\nof top 10%\n\n14 countries 19 countries\n\n21 countries\n\n6\\.0% 6.5% 7.0% 7.5% 8.0% 8.5% 9.0% 9.5%\n\n10\\.0% 10.5% 11.0%\n\n2010 2014 2017 2021\n\nPanel C: Wealth share of the\n\nbottom 50%\n\n14 countries 19 countries\n\n21 countries\n\n0\\.02\n\n0\\.025\n\n0\\.03\n\n0\\.035\n\n0\\.04\n\n0\\.045\n\n0\\.05\n\n0\\.055\n\n2010 2014 2017 2021\n\nPanel D: Dispersion - wealth share\n\nof the bottom 50%\n\n14 countries 19 countries\n\n21 countries","pages":[{"line_numbers":[],"markdown":"Employment and Social Developments in Europe 2024\n\n44\n\n_Box (continued)_\n\n_(Continued on the next page)_\n\nChart 2 Median net wealth grew across the income distribution between 2010 and 2021 Median household net wealth by income quintile, euro area average, 2021 USD PPP, 14 EU countries\n\n_Note:_ Median household net wealth in 2021 EUR PPS. Median household net wealth calculated for: people in the bottom fifth of income earners (lowest quintile); people in the middle fifth of income earners (middle quintile); people in the top fifth of income earners (top quintile).\n\n_Source:_ OECD calculations based on the HFCS, https://www.ecb.europa.eu/stats/ecb\\_surveys/hfcs/html/index.en.html\n\nChart 3 Wealth concentration increased after the financial crisis but decreased during COVID-19 Average and standard deviation of wealth share of top 10% and bottom 50%, 2010-2021\n\n_Note:_ Dispersion measured as standard deviation of wealth share by country. See Figure 1 note for more information. Standard deviation is a measure of cross-country variation, the higher the standard deviation, the higher the cross-country variation.\n\n_Source:_ OECD calculations based on the HFCS.\n\n€ 0\n\n€ 50 000\n\n€ 100 000\n\n€ 150 000\n\n€ 200 000\n\n€ 250 000\n\n€ 300 000\n\n€ 350 000\n\n€ 400 000\n\n€ 450 000\n\n2010 2014 2017 2021\n\nLowest quintile Middle quintile Top quintile\n\n46\\.0% 46.5% 47.0% 47.5% 48.0% 48.5% 49.0% 49.5% 50.0% 50.5% 51.0%\n\n2010 2014 2017 2021\n\nPanel A: Wealth share of the top\n\n10%\n\n14 countries 19 countries\n\n21 countries\n\n0\\.05\n\n0\\.055\n\n0\\.06\n\n0\\.065\n\n0\\.07\n\n0\\.075\n\n0\\.08\n\n0\\.085\n\n2010 2014 2017 2021\n\nPanel B: Dispersion - wealth share\n\nof top 10%\n\n14 countries 19 countries\n\n21 countries\n\n6\\.0% 6.5% 7.0% 7.5% 8.0% 8.5% 9.0% 9.5%\n\n10\\.0% 10.5% 11.0%\n\n2010 2014 2017 2021\n\nPanel C: Wealth share of the\n\nbottom 50%\n\n14 countries 19 countries\n\n21 countries\n\n0\\.02\n\n0\\.025\n\n0\\.03\n\n0\\.035\n\n0\\.04\n\n0\\.045\n\n0\\.05\n\n0\\.055\n\n2010 2014 2017 2021\n\nPanel D: Dispersion - wealth share\n\nof the bottom 50%\n\n14 countries 19 countries\n\n21 countries","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Employment and Social Developments in Europe 2024\n\n44\n\n_Box (continued)_\n\n_(Continued on the next page)_\n\nChart 2 Median net wealth grew across the income distribution between 2010 and 2021 Median household net wealth by income quintile, euro area average, 2021 USD PPP, 14 EU countries\n\n_Note:_ Median household net wealth in 2021 EUR PPS. Median household net wealth calculated for: people in the bottom fifth of income earners (lowest quintile); people in the middle fifth of income earners (middle quintile); people in the top fifth of income earners (top quintile).\n\n_Source:_ OECD calculations based on the HFCS, https://www.ecb.europa.eu/stats/ecb\\_surveys/hfcs/html/index.en.html\n\nChart 3 Wealth concentration increased after the financial crisis but decreased during COVID-19 Average and standard deviation of wealth share of top 10% and bottom 50%, 2010-2021\n\n_Note:_ Dispersion measured as standard deviation of wealth share by country. See Figure 1 note for more information. Standard deviation is a measure of cross-country variation, the higher the standard deviation, the higher the cross-country variation.\n\n_Source:_ OECD calculations based on the HFCS.\n\n€ 0\n\n€ 50 000\n\n€ 100 000\n\n€ 150 000\n\n€ 200 000\n\n€ 250 000\n\n€ 300 000\n\n€ 350 000\n\n€ 400 000\n\n€ 450 000\n\n2010 2014 2017 2021\n\nLowest quintile Middle quintile Top quintile\n\n46\\.0% 46.5% 47.0% 47.5% 48.0% 48.5% 49.0% 49.5% 50.0% 50.5% 51.0%\n\n2010 2014 2017 2021\n\nPanel A: Wealth share of the top\n\n10%\n\n14 countries 19 countries\n\n21 countries\n\n0\\.05\n\n0\\.055\n\n0\\.06\n\n0\\.065\n\n0\\.07\n\n0\\.075\n\n0\\.08\n\n0\\.085\n\n2010 2014 2017 2021\n\nPanel B: Dispersion - wealth share\n\nof top 10%\n\n14 countries 19 countries\n\n21 countries\n\n6\\.0% 6.5% 7.0% 7.5% 8.0% 8.5% 9.0% 9.5%\n\n10\\.0% 10.5% 11.0%\n\n2010 2014 2017 2021\n\nPanel C: Wealth share of the\n\nbottom 50%\n\n14 countries 19 countries\n\n21 countries\n\n0\\.02\n\n0\\.025\n\n0\\.03\n\n0\\.035\n\n0\\.04\n\n0\\.045\n\n0\\.05\n\n0\\.055\n\n2010 2014 2017 2021\n\nPanel D: Dispersion - wealth share\n\nof the bottom 50%\n\n14 countries 19 countries\n\n21 countries","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/employment_and_social_developments_in_europe_esde_2024_report_p41.pdf"]},"response":{"errors":[],"extract_id":"extract_c5b9a8afa92cc9c94cf8b8be93b884a9","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Employment and Social Developments in Europe 2024\n\n44\n\n_Box (continued)_\n\n_(Continued on the next page)_\n\nChart 2 Median net wealth grew across the income distribution between 2010 and 2021 Median household net wealth by income quintile, euro area average, 2021 USD PPP, 14 EU countries\n\n_Note:_ Median household net wealth in 2021 EUR PPS. Median household net wealth calculated for: people in the bottom fifth of income earners (lowest quintile); people in the middle fifth of income earners (middle quintile); people in the top fifth of income earners (top quintile).\n\n_Source:_ OECD calculations based on the HFCS, https://www.ecb.europa.eu/stats/ecb\\_surveys/hfcs/html/index.en.html\n\nChart 3 Wealth concentration increased after the financial crisis but decreased during COVID-19 Average and standard deviation of wealth share of top 10% and bottom 50%, 2010-2021\n\n_Note:_ Dispersion measured as standard deviation of wealth share by country. See Figure 1 note for more information. Standard deviation is a measure of cross-country variation, the higher the standard deviation, the higher the cross-country variation.\n\n_Source:_ OECD calculations based on the HFCS.\n\n€ 0\n\n€ 50 000\n\n€ 100 000\n\n€ 150 000\n\n€ 200 000\n\n€ 250 000\n\n€ 300 000\n\n€ 350 000\n\n€ 400 000\n\n€ 450 000\n\n2010 2014 2017 2021\n\nLowest quintile Middle quintile Top quintile\n\n46\\.0% 46.5% 47.0% 47.5% 48.0% 48.5% 49.0% 49.5% 50.0% 50.5% 51.0%\n\n2010 2014 2017 2021\n\nPanel A: Wealth share of the top\n\n10%\n\n14 countries 19 countries\n\n21 countries\n\n0\\.05\n\n0\\.055\n\n0\\.06\n\n0\\.065\n\n0\\.07\n\n0\\.075\n\n0\\.08\n\n0\\.085\n\n2010 2014 2017 2021\n\nPanel B: Dispersion - wealth share\n\nof top 10%\n\n14 countries 19 countries\n\n21 countries\n\n6\\.0% 6.5% 7.0% 7.5% 8.0% 8.5% 9.0% 9.5%\n\n10\\.0% 10.5% 11.0%\n\n2010 2014 2017 2021\n\nPanel C: Wealth share of the\n\nbottom 50%\n\n14 countries 19 countries\n\n21 countries\n\n0\\.02\n\n0\\.025\n\n0\\.03\n\n0\\.035\n\n0\\.04\n\n0\\.045\n\n0\\.05\n\n0\\.055\n\n2010 2014 2017 2021\n\nPanel D: Dispersion - wealth share\n\nof the bottom 50%\n\n14 countries 19 countries\n\n21 countries"],"full_content":"Employment and Social Developments in Europe 2024\n\n44\n\n_Box (continued)_\n\n_(Continued on the next page)_\n\nChart 2 Median net wealth grew across the income distribution between 2010 and 2021 Median household net wealth by income quintile, euro area average, 2021 USD PPP, 14 EU countries\n\n_Note:_ Median household net wealth in 2021 EUR PPS. Median household net wealth calculated for: people in the bottom fifth of income earners (lowest quintile); people in the middle fifth of income earners (middle quintile); people in the top fifth of income earners (top quintile).\n\n_Source:_ OECD calculations based on the HFCS, https://www.ecb.europa.eu/stats/ecb\\_surveys/hfcs/html/index.en.html\n\nChart 3 Wealth concentration increased after the financial crisis but decreased during COVID-19 Average and standard deviation of wealth share of top 10% and bottom 50%, 2010-2021\n\n_Note:_ Dispersion measured as standard deviation of wealth share by country. See Figure 1 note for more information. Standard deviation is a measure of cross-country variation, the higher the standard deviation, the higher the cross-country variation.\n\n_Source:_ OECD calculations based on the HFCS.\n\n€ 0\n\n€ 50 000\n\n€ 100 000\n\n€ 150 000\n\n€ 200 000\n\n€ 250 000\n\n€ 300 000\n\n€ 350 000\n\n€ 400 000\n\n€ 450 000\n\n2010 2014 2017 2021\n\nLowest quintile Middle quintile Top quintile\n\n46\\.0% 46.5% 47.0% 47.5% 48.0% 48.5% 49.0% 49.5% 50.0% 50.5% 51.0%\n\n2010 2014 2017 2021\n\nPanel A: Wealth share of the top\n\n10%\n\n14 countries 19 countries\n\n21 countries\n\n0\\.05\n\n0\\.055\n\n0\\.06\n\n0\\.065\n\n0\\.07\n\n0\\.075\n\n0\\.08\n\n0\\.085\n\n2010 2014 2017 2021\n\nPanel B: Dispersion - wealth share\n\nof top 10%\n\n14 countries 19 countries\n\n21 countries\n\n6\\.0% 6.5% 7.0% 7.5% 8.0% 8.5% 9.0% 9.5%\n\n10\\.0% 10.5% 11.0%\n\n2010 2014 2017 2021\n\nPanel C: Wealth share of the\n\nbottom 50%\n\n14 countries 19 countries\n\n21 countries\n\n0\\.02\n\n0\\.025\n\n0\\.03\n\n0\\.035\n\n0\\.04\n\n0\\.045\n\n0\\.05\n\n0\\.055\n\n2010 2014 2017 2021\n\nPanel D: Dispersion - wealth share\n\nof the bottom 50%\n\n14 countries 19 countries\n\n21 countries","publish_date":null,"title":"Panel A: Wealth share of the top Panel B: Dispersion -wealth share 10% of top 10% Panel C: Wealth share","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/employment_and_social_developments_in_europe_esde_2024_report_p41.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/employment_and_social_developments_in_europe_esde_2024_report_p41.pdf"},"request":{"example_id":"chart/employment_and_social_developments_in_europe_esde_2024_report_p41","product_type":"parse","source_file_path":"employment_and_social_developments_in_europe_esde_2024_report_p41.pdf"},"started_at":"2026-11-04T19:45:21.074742"} +{"completed_at":"2026-11-04T19:45:25.516768","latency_in_ms":2896,"output":{"example_id":"chart/employment_and_social_developments_in_europe_esde_2024_report_p46","grounded_pages":[],"layout_pages":[],"markdown":"Chapter 2: Social convergence in the EU: taking stock\n\n49\n\nChart 2.5 Growing disparities in skill supply pose challenges to regional development Tertiary education attainment and adult participation in learning (last 4 weeks) (% of population aged 30-34), and cross-regional variation (measured by standard deviation), 2007- 2023, EU-NUTS2\n\n_Note:_ Standard deviation is a measure of cross-regional variation, the higher the standard deviation, the higher the cross-regional variation.\n\n_Source:_ DG EMPL calculations based on Eurostat datasets edat\\_lfse\\_04, trng\\_lfse\\_04.\n\nClick here to download chart.\n\nFigure 2.2 Regional variation in tertiary education attainment ranged from 28-48% in most NUTS2 regions Tertiary educational attainment, 2023 (% of people aged 30-34), NUTS 2 regions\n\n_Note:_ Tertiary educational attainment (% of people aged 30-34) by NUTS2 regions, 2023.\n\n_Source:_ Eurostat, edat\\_lfse\\_04 IMAGE Interactive map generator.\n\nClick here to download figure.\n\n**While participation in adult learning improved somewhat at the EU level, disparities increased including in relation to the development of basic digital skills.** The indicator for adult learning participation (over the last four weeks) stood at 7.9% in 2007, increasing to 10.1% in 2014, and 12.8% in 2023. ( ⁸² ) However, differences in national (and regional) participation rates widened, ranging from 1.4% to\n\n( ⁸² ) Analysis refers to the European Pillar of Social Rights Social Scoreboard indicator of adult participation in learning in the past 4 weeks (LFS) and not to the adult participation in learning in the last 12 months (Adult Education Survey (AES)) indicator that monitors the EU\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n2007 2010 2013 2016 2019 2022\n\nTertiary education attainment\n\nAverage (left axis)\n\nStandard deviation (right axis)\n\n0\n\n2\n\n4\n\n6\n\n8\n\n0\n\n5\n\n10\n\n15\n\n2007 2009 2011 2013 2015 2017 2019 2021 2023\n\nAdult learning participation\n\nAverage (left axis)\n\nStandard deviation (right axis)","pages":[{"line_numbers":[],"markdown":"Chapter 2: Social convergence in the EU: taking stock\n\n49\n\nChart 2.5 Growing disparities in skill supply pose challenges to regional development Tertiary education attainment and adult participation in learning (last 4 weeks) (% of population aged 30-34), and cross-regional variation (measured by standard deviation), 2007- 2023, EU-NUTS2\n\n_Note:_ Standard deviation is a measure of cross-regional variation, the higher the standard deviation, the higher the cross-regional variation.\n\n_Source:_ DG EMPL calculations based on Eurostat datasets edat\\_lfse\\_04, trng\\_lfse\\_04.\n\nClick here to download chart.\n\nFigure 2.2 Regional variation in tertiary education attainment ranged from 28-48% in most NUTS2 regions Tertiary educational attainment, 2023 (% of people aged 30-34), NUTS 2 regions\n\n_Note:_ Tertiary educational attainment (% of people aged 30-34) by NUTS2 regions, 2023.\n\n_Source:_ Eurostat, edat\\_lfse\\_04 IMAGE Interactive map generator.\n\nClick here to download figure.\n\n**While participation in adult learning improved somewhat at the EU level, disparities increased including in relation to the development of basic digital skills.** The indicator for adult learning participation (over the last four weeks) stood at 7.9% in 2007, increasing to 10.1% in 2014, and 12.8% in 2023. ( ⁸² ) However, differences in national (and regional) participation rates widened, ranging from 1.4% to\n\n( ⁸² ) Analysis refers to the European Pillar of Social Rights Social Scoreboard indicator of adult participation in learning in the past 4 weeks (LFS) and not to the adult participation in learning in the last 12 months (Adult Education Survey (AES)) indicator that monitors the EU\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n2007 2010 2013 2016 2019 2022\n\nTertiary education attainment\n\nAverage (left axis)\n\nStandard deviation (right axis)\n\n0\n\n2\n\n4\n\n6\n\n8\n\n0\n\n5\n\n10\n\n15\n\n2007 2009 2011 2013 2015 2017 2019 2021 2023\n\nAdult learning participation\n\nAverage (left axis)\n\nStandard deviation (right axis)","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Chapter 2: Social convergence in the EU: taking stock\n\n49\n\nChart 2.5 Growing disparities in skill supply pose challenges to regional development Tertiary education attainment and adult participation in learning (last 4 weeks) (% of population aged 30-34), and cross-regional variation (measured by standard deviation), 2007- 2023, EU-NUTS2\n\n_Note:_ Standard deviation is a measure of cross-regional variation, the higher the standard deviation, the higher the cross-regional variation.\n\n_Source:_ DG EMPL calculations based on Eurostat datasets edat\\_lfse\\_04, trng\\_lfse\\_04.\n\nClick here to download chart.\n\nFigure 2.2 Regional variation in tertiary education attainment ranged from 28-48% in most NUTS2 regions Tertiary educational attainment, 2023 (% of people aged 30-34), NUTS 2 regions\n\n_Note:_ Tertiary educational attainment (% of people aged 30-34) by NUTS2 regions, 2023.\n\n_Source:_ Eurostat, edat\\_lfse\\_04 IMAGE Interactive map generator.\n\nClick here to download figure.\n\n**While participation in adult learning improved somewhat at the EU level, disparities increased including in relation to the development of basic digital skills.** The indicator for adult learning participation (over the last four weeks) stood at 7.9% in 2007, increasing to 10.1% in 2014, and 12.8% in 2023. ( ⁸² ) However, differences in national (and regional) participation rates widened, ranging from 1.4% to\n\n( ⁸² ) Analysis refers to the European Pillar of Social Rights Social Scoreboard indicator of adult participation in learning in the past 4 weeks (LFS) and not to the adult participation in learning in the last 12 months (Adult Education Survey (AES)) indicator that monitors the EU\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n2007 2010 2013 2016 2019 2022\n\nTertiary education attainment\n\nAverage (left axis)\n\nStandard deviation (right axis)\n\n0\n\n2\n\n4\n\n6\n\n8\n\n0\n\n5\n\n10\n\n15\n\n2007 2009 2011 2013 2015 2017 2019 2021 2023\n\nAdult learning participation\n\nAverage (left axis)\n\nStandard deviation (right axis)","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/employment_and_social_developments_in_europe_esde_2024_report_p46.pdf"]},"response":{"errors":[],"extract_id":"extract_c771c52339153d708747821c97eb1891","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Chapter 2: Social convergence in the EU: taking stock\n\n49\n\nChart 2.5 Growing disparities in skill supply pose challenges to regional development Tertiary education attainment and adult participation in learning (last 4 weeks) (% of population aged 30-34), and cross-regional variation (measured by standard deviation), 2007- 2023, EU-NUTS2\n\n_Note:_ Standard deviation is a measure of cross-regional variation, the higher the standard deviation, the higher the cross-regional variation.\n\n_Source:_ DG EMPL calculations based on Eurostat datasets edat\\_lfse\\_04, trng\\_lfse\\_04.\n\nClick here to download chart.\n\nFigure 2.2 Regional variation in tertiary education attainment ranged from 28-48% in most NUTS2 regions Tertiary educational attainment, 2023 (% of people aged 30-34), NUTS 2 regions\n\n_Note:_ Tertiary educational attainment (% of people aged 30-34) by NUTS2 regions, 2023.\n\n_Source:_ Eurostat, edat\\_lfse\\_04 IMAGE Interactive map generator.\n\nClick here to download figure.\n\n**While participation in adult learning improved somewhat at the EU level, disparities increased including in relation to the development of basic digital skills.** The indicator for adult learning participation (over the last four weeks) stood at 7.9% in 2007, increasing to 10.1% in 2014, and 12.8% in 2023. ( ⁸² ) However, differences in national (and regional) participation rates widened, ranging from 1.4% to\n\n( ⁸² ) Analysis refers to the European Pillar of Social Rights Social Scoreboard indicator of adult participation in learning in the past 4 weeks (LFS) and not to the adult participation in learning in the last 12 months (Adult Education Survey (AES)) indicator that monitors the EU\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n2007 2010 2013 2016 2019 2022\n\nTertiary education attainment\n\nAverage (left axis)\n\nStandard deviation (right axis)\n\n0\n\n2\n\n4\n\n6\n\n8\n\n0\n\n5\n\n10\n\n15\n\n2007 2009 2011 2013 2015 2017 2019 2021 2023\n\nAdult learning participation\n\nAverage (left axis)\n\nStandard deviation (right axis)"],"full_content":"Chapter 2: Social convergence in the EU: taking stock\n\n49\n\nChart 2.5 Growing disparities in skill supply pose challenges to regional development Tertiary education attainment and adult participation in learning (last 4 weeks) (% of population aged 30-34), and cross-regional variation (measured by standard deviation), 2007- 2023, EU-NUTS2\n\n_Note:_ Standard deviation is a measure of cross-regional variation, the higher the standard deviation, the higher the cross-regional variation.\n\n_Source:_ DG EMPL calculations based on Eurostat datasets edat\\_lfse\\_04, trng\\_lfse\\_04.\n\nClick here to download chart.\n\nFigure 2.2 Regional variation in tertiary education attainment ranged from 28-48% in most NUTS2 regions Tertiary educational attainment, 2023 (% of people aged 30-34), NUTS 2 regions\n\n_Note:_ Tertiary educational attainment (% of people aged 30-34) by NUTS2 regions, 2023.\n\n_Source:_ Eurostat, edat\\_lfse\\_04 IMAGE Interactive map generator.\n\nClick here to download figure.\n\n**While participation in adult learning improved somewhat at the EU level, disparities increased including in relation to the development of basic digital skills.** The indicator for adult learning participation (over the last four weeks) stood at 7.9% in 2007, increasing to 10.1% in 2014, and 12.8% in 2023. ( ⁸² ) However, differences in national (and regional) participation rates widened, ranging from 1.4% to\n\n( ⁸² ) Analysis refers to the European Pillar of Social Rights Social Scoreboard indicator of adult participation in learning in the past 4 weeks (LFS) and not to the adult participation in learning in the last 12 months (Adult Education Survey (AES)) indicator that monitors the EU\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n2007 2010 2013 2016 2019 2022\n\nTertiary education attainment\n\nAverage (left axis)\n\nStandard deviation (right axis)\n\n0\n\n2\n\n4\n\n6\n\n8\n\n0\n\n5\n\n10\n\n15\n\n2007 2009 2011 2013 2015 2017 2019 2021 2023\n\nAdult learning participation\n\nAverage (left axis)\n\nStandard deviation (right axis)","publish_date":null,"title":"Tertiary education attainment Adult learning participation","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/employment_and_social_developments_in_europe_esde_2024_report_p46.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/employment_and_social_developments_in_europe_esde_2024_report_p46.pdf"},"request":{"example_id":"chart/employment_and_social_developments_in_europe_esde_2024_report_p46","product_type":"parse","source_file_path":"employment_and_social_developments_in_europe_esde_2024_report_p46.pdf"},"started_at":"2026-11-04T19:45:22.620631"} +{"completed_at":"2026-11-04T19:45:25.506150","latency_in_ms":1190,"output":{"example_id":"chart/employment_and_social_developments_in_europe_esde_2024_report_p50","grounded_pages":[],"layout_pages":[],"markdown":"Chapter 2: Social convergence in the EU: taking stock\n\n53\n\nChart 2.7 Variations in health outcome measures have remained stable Healthy life years at age 65 for women and men, and cross-country variation (measured by standard deviation), 2007-2022, EU-27\n\n_Note:_ Latest data point is 2022 Standard deviation is a measure of cross-country variation, the higher the standard deviation, the higher the cross-country variation. EU average levels are weighted values.\n\n_Source:_ DG EMPL calculations based on EU-SILC dataset hlth\\_hlye. Click here to download chart.\n\n### 3\\. CONVERGENCE IN SOCIOECONOMIC OUTCOMES WITHIN MEMBER STATES\n\n**Understanding patterns of regional convergence within countries can better clarify trends in socioeconomic outcomes.** Convergence at EU level (across countries or regions as analysed in section 2) does not necessarily imply regional convergence within countries, for example when capital and metropolitan regions strongly outperform rural and peripheral regions within a country. It is important to look beyond convergence at EU level to capture potential challenges to territorial and social cohesion within specific Member States. This section provides regional convergence analysis within each of the 19 Member States with more than four NUTS2 regions. ( ⁸⁶ ) It focuses on the indicators analysed in section 2 for which regional data are available (GDP per capita, (un)employment and NEET rates, tertiary and adult education rates, and AROPE rates), covering 2007 to 2022. It also provides an in-depth look at within-country convergence in median incomes across very small territorial units (NUTS3 regions) based on unique administrative data collected in eight Member States (Box 2.2(European Commission, )).\n\n**Upward divergence in economic performance at EU level was observed within most EU Member States.** Sixteen of the 19 Member States analysed saw their national GDP per capita increase since 2007 amid growing regional disparities (Table 2.2). The growth in regional disparities was particularly strong (more than 50%) in Denmark and several central and eastern European countries (Bulgaria, Czechia, Poland, Romania). In central and eastern Member States, divergence partly resulted from capital regions outperforming other regions. ( ⁸⁷ ) This was not always the case in other countries, e.g.: increases in regional variation in France were linked to particularly sluggish growth in regions with low levels of GDP per capita. Only Portugal experienced upward convergence in GDP per capita, but this was largely due to low growth in some developed, previously dynamic, regions. ( ⁸⁸ )\n\n**Convergence patterns in labour market outcomes varied from country to country.** Employment rates grew in most Member States since 2007, sometimes accompanied by regional convergence and sometimes by divergence (notably in Romania). Similarly, regional NEET rates converged in some countries and diverged in others, while average national rates largely declined. Concerningly, Denmark, France, and Romania all saw strong regional divergence in NEET rates. Within-country developments in unemployment rates are more reassuring, with 10 of the 18 countries experiencing upward convergence, notably Bulgaria, Czechia, Germany, and Portugal. Overall, within-country variation in labour market outcomes was highly sensitive to specific national and regional factors, but a detailed analysis is beyond the scope of this report.\n\n**Most countries experienced growth in tertiary education attainment accompanied by regional divergence.** This was the case for 16 of the 18 countries analysed. In several eastern and central European Member States (Czechia, Hungary, Poland, Romania) and Portugal, regional differences grew particularly strongly. These developments stemmed from sharp increases in tertiary education attainment in capital regions, reflecting a combination of factors including the concentration of universities, high demand for tertiary-educated workers and associated wage premiums. By contrast, a lack of tertiary education opportunities and an outflow of highly\n\n( ⁸⁶ ) Four regions was considered the lowest number reasonable to perform robust convergence analysis. ( ⁸⁷ ) (Eurofound, 2021b); (Alcidi et al., 2018) ( ⁸⁸ ) (European Commission, 2024b)\n\n0\n\n1\n\n2\n\n3\n\n4\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n2007 2009 2011 2013 2015 2017 2019 2021\n\nHealthy life years at age 65: women, years\n\nAverage (left axis) Standard deviation (right axis)\n\n0\n\n1\n\n2\n\n3\n\n4\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n2007 2009 2011 2013 2015 2017 2019 2021\n\nHealthy life years at age 65: men, years\n\nAverage (left axis) Standard deviation (right axis)","pages":[{"line_numbers":[],"markdown":"Chapter 2: Social convergence in the EU: taking stock\n\n53\n\nChart 2.7 Variations in health outcome measures have remained stable Healthy life years at age 65 for women and men, and cross-country variation (measured by standard deviation), 2007-2022, EU-27\n\n_Note:_ Latest data point is 2022 Standard deviation is a measure of cross-country variation, the higher the standard deviation, the higher the cross-country variation. EU average levels are weighted values.\n\n_Source:_ DG EMPL calculations based on EU-SILC dataset hlth\\_hlye. Click here to download chart.\n\n### 3\\. CONVERGENCE IN SOCIOECONOMIC OUTCOMES WITHIN MEMBER STATES\n\n**Understanding patterns of regional convergence within countries can better clarify trends in socioeconomic outcomes.** Convergence at EU level (across countries or regions as analysed in section 2) does not necessarily imply regional convergence within countries, for example when capital and metropolitan regions strongly outperform rural and peripheral regions within a country. It is important to look beyond convergence at EU level to capture potential challenges to territorial and social cohesion within specific Member States. This section provides regional convergence analysis within each of the 19 Member States with more than four NUTS2 regions. ( ⁸⁶ ) It focuses on the indicators analysed in section 2 for which regional data are available (GDP per capita, (un)employment and NEET rates, tertiary and adult education rates, and AROPE rates), covering 2007 to 2022. It also provides an in-depth look at within-country convergence in median incomes across very small territorial units (NUTS3 regions) based on unique administrative data collected in eight Member States (Box 2.2(European Commission, )).\n\n**Upward divergence in economic performance at EU level was observed within most EU Member States.** Sixteen of the 19 Member States analysed saw their national GDP per capita increase since 2007 amid growing regional disparities (Table 2.2). The growth in regional disparities was particularly strong (more than 50%) in Denmark and several central and eastern European countries (Bulgaria, Czechia, Poland, Romania). In central and eastern Member States, divergence partly resulted from capital regions outperforming other regions. ( ⁸⁷ ) This was not always the case in other countries, e.g.: increases in regional variation in France were linked to particularly sluggish growth in regions with low levels of GDP per capita. Only Portugal experienced upward convergence in GDP per capita, but this was largely due to low growth in some developed, previously dynamic, regions. ( ⁸⁸ )\n\n**Convergence patterns in labour market outcomes varied from country to country.** Employment rates grew in most Member States since 2007, sometimes accompanied by regional convergence and sometimes by divergence (notably in Romania). Similarly, regional NEET rates converged in some countries and diverged in others, while average national rates largely declined. Concerningly, Denmark, France, and Romania all saw strong regional divergence in NEET rates. Within-country developments in unemployment rates are more reassuring, with 10 of the 18 countries experiencing upward convergence, notably Bulgaria, Czechia, Germany, and Portugal. Overall, within-country variation in labour market outcomes was highly sensitive to specific national and regional factors, but a detailed analysis is beyond the scope of this report.\n\n**Most countries experienced growth in tertiary education attainment accompanied by regional divergence.** This was the case for 16 of the 18 countries analysed. In several eastern and central European Member States (Czechia, Hungary, Poland, Romania) and Portugal, regional differences grew particularly strongly. These developments stemmed from sharp increases in tertiary education attainment in capital regions, reflecting a combination of factors including the concentration of universities, high demand for tertiary-educated workers and associated wage premiums. By contrast, a lack of tertiary education opportunities and an outflow of highly\n\n( ⁸⁶ ) Four regions was considered the lowest number reasonable to perform robust convergence analysis. ( ⁸⁷ ) (Eurofound, 2021b); (Alcidi et al., 2018) ( ⁸⁸ ) (European Commission, 2024b)\n\n0\n\n1\n\n2\n\n3\n\n4\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n2007 2009 2011 2013 2015 2017 2019 2021\n\nHealthy life years at age 65: women, years\n\nAverage (left axis) Standard deviation (right axis)\n\n0\n\n1\n\n2\n\n3\n\n4\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n2007 2009 2011 2013 2015 2017 2019 2021\n\nHealthy life years at age 65: men, years\n\nAverage (left axis) Standard deviation (right axis)","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Chapter 2: Social convergence in the EU: taking stock\n\n53\n\nChart 2.7 Variations in health outcome measures have remained stable Healthy life years at age 65 for women and men, and cross-country variation (measured by standard deviation), 2007-2022, EU-27\n\n_Note:_ Latest data point is 2022 Standard deviation is a measure of cross-country variation, the higher the standard deviation, the higher the cross-country variation. EU average levels are weighted values.\n\n_Source:_ DG EMPL calculations based on EU-SILC dataset hlth\\_hlye. Click here to download chart.\n\n### 3\\. CONVERGENCE IN SOCIOECONOMIC OUTCOMES WITHIN MEMBER STATES\n\n**Understanding patterns of regional convergence within countries can better clarify trends in socioeconomic outcomes.** Convergence at EU level (across countries or regions as analysed in section 2) does not necessarily imply regional convergence within countries, for example when capital and metropolitan regions strongly outperform rural and peripheral regions within a country. It is important to look beyond convergence at EU level to capture potential challenges to territorial and social cohesion within specific Member States. This section provides regional convergence analysis within each of the 19 Member States with more than four NUTS2 regions. ( ⁸⁶ ) It focuses on the indicators analysed in section 2 for which regional data are available (GDP per capita, (un)employment and NEET rates, tertiary and adult education rates, and AROPE rates), covering 2007 to 2022. It also provides an in-depth look at within-country convergence in median incomes across very small territorial units (NUTS3 regions) based on unique administrative data collected in eight Member States (Box 2.2(European Commission, )).\n\n**Upward divergence in economic performance at EU level was observed within most EU Member States.** Sixteen of the 19 Member States analysed saw their national GDP per capita increase since 2007 amid growing regional disparities (Table 2.2). The growth in regional disparities was particularly strong (more than 50%) in Denmark and several central and eastern European countries (Bulgaria, Czechia, Poland, Romania). In central and eastern Member States, divergence partly resulted from capital regions outperforming other regions. ( ⁸⁷ ) This was not always the case in other countries, e.g.: increases in regional variation in France were linked to particularly sluggish growth in regions with low levels of GDP per capita. Only Portugal experienced upward convergence in GDP per capita, but this was largely due to low growth in some developed, previously dynamic, regions. ( ⁸⁸ )\n\n**Convergence patterns in labour market outcomes varied from country to country.** Employment rates grew in most Member States since 2007, sometimes accompanied by regional convergence and sometimes by divergence (notably in Romania). Similarly, regional NEET rates converged in some countries and diverged in others, while average national rates largely declined. Concerningly, Denmark, France, and Romania all saw strong regional divergence in NEET rates. Within-country developments in unemployment rates are more reassuring, with 10 of the 18 countries experiencing upward convergence, notably Bulgaria, Czechia, Germany, and Portugal. Overall, within-country variation in labour market outcomes was highly sensitive to specific national and regional factors, but a detailed analysis is beyond the scope of this report.\n\n**Most countries experienced growth in tertiary education attainment accompanied by regional divergence.** This was the case for 16 of the 18 countries analysed. In several eastern and central European Member States (Czechia, Hungary, Poland, Romania) and Portugal, regional differences grew particularly strongly. These developments stemmed from sharp increases in tertiary education attainment in capital regions, reflecting a combination of factors including the concentration of universities, high demand for tertiary-educated workers and associated wage premiums. By contrast, a lack of tertiary education opportunities and an outflow of highly\n\n( ⁸⁶ ) Four regions was considered the lowest number reasonable to perform robust convergence analysis. ( ⁸⁷ ) (Eurofound, 2021b); (Alcidi et al., 2018) ( ⁸⁸ ) (European Commission, 2024b)\n\n0\n\n1\n\n2\n\n3\n\n4\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n2007 2009 2011 2013 2015 2017 2019 2021\n\nHealthy life years at age 65: women, years\n\nAverage (left axis) Standard deviation (right axis)\n\n0\n\n1\n\n2\n\n3\n\n4\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n2007 2009 2011 2013 2015 2017 2019 2021\n\nHealthy life years at age 65: men, years\n\nAverage (left axis) Standard deviation (right axis)","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/employment_and_social_developments_in_europe_esde_2024_report_p50.pdf"]},"response":{"errors":[],"extract_id":"extract_eaf67f70cf8d203bab645bb751bd5248","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Chapter 2: Social convergence in the EU: taking stock\n\n53\n\nChart 2.7 Variations in health outcome measures have remained stable Healthy life years at age 65 for women and men, and cross-country variation (measured by standard deviation), 2007-2022, EU-27\n\n_Note:_ Latest data point is 2022 Standard deviation is a measure of cross-country variation, the higher the standard deviation, the higher the cross-country variation. EU average levels are weighted values.\n\n_Source:_ DG EMPL calculations based on EU-SILC dataset hlth\\_hlye. Click here to download chart.\n\n### 3\\. CONVERGENCE IN SOCIOECONOMIC OUTCOMES WITHIN MEMBER STATES\n\n**Understanding patterns of regional convergence within countries can better clarify trends in socioeconomic outcomes.** Convergence at EU level (across countries or regions as analysed in section 2) does not necessarily imply regional convergence within countries, for example when capital and metropolitan regions strongly outperform rural and peripheral regions within a country. It is important to look beyond convergence at EU level to capture potential challenges to territorial and social cohesion within specific Member States. This section provides regional convergence analysis within each of the 19 Member States with more than four NUTS2 regions. ( ⁸⁶ ) It focuses on the indicators analysed in section 2 for which regional data are available (GDP per capita, (un)employment and NEET rates, tertiary and adult education rates, and AROPE rates), covering 2007 to 2022. It also provides an in-depth look at within-country convergence in median incomes across very small territorial units (NUTS3 regions) based on unique administrative data collected in eight Member States (Box 2.2(European Commission, )).\n\n**Upward divergence in economic performance at EU level was observed within most EU Member States.** Sixteen of the 19 Member States analysed saw their national GDP per capita increase since 2007 amid growing regional disparities (Table 2.2). The growth in regional disparities was particularly strong (more than 50%) in Denmark and several central and eastern European countries (Bulgaria, Czechia, Poland, Romania). In central and eastern Member States, divergence partly resulted from capital regions outperforming other regions. ( ⁸⁷ ) This was not always the case in other countries, e.g.: increases in regional variation in France were linked to particularly sluggish growth in regions with low levels of GDP per capita. Only Portugal experienced upward convergence in GDP per capita, but this was largely due to low growth in some developed, previously dynamic, regions. ( ⁸⁸ )\n\n**Convergence patterns in labour market outcomes varied from country to country.** Employment rates grew in most Member States since 2007, sometimes accompanied by regional convergence and sometimes by divergence (notably in Romania). Similarly, regional NEET rates converged in some countries and diverged in others, while average national rates largely declined. Concerningly, Denmark, France, and Romania all saw strong regional divergence in NEET rates. Within-country developments in unemployment rates are more reassuring, with 10 of the 18 countries experiencing upward convergence, notably Bulgaria, Czechia, Germany, and Portugal. Overall, within-country variation in labour market outcomes was highly sensitive to specific national and regional factors, but a detailed analysis is beyond the scope of this report.\n\n**Most countries experienced growth in tertiary education attainment accompanied by regional divergence.** This was the case for 16 of the 18 countries analysed. In several eastern and central European Member States (Czechia, Hungary, Poland, Romania) and Portugal, regional differences grew particularly strongly. These developments stemmed from sharp increases in tertiary education attainment in capital regions, reflecting a combination of factors including the concentration of universities, high demand for tertiary-educated workers and associated wage premiums. By contrast, a lack of tertiary education opportunities and an outflow of highly\n\n( ⁸⁶ ) Four regions was considered the lowest number reasonable to perform robust convergence analysis. ( ⁸⁷ ) (Eurofound, 2021b); (Alcidi et al., 2018) ( ⁸⁸ ) (European Commission, 2024b)\n\n0\n\n1\n\n2\n\n3\n\n4\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n2007 2009 2011 2013 2015 2017 2019 2021\n\nHealthy life years at age 65: women, years\n\nAverage (left axis) Standard deviation (right axis)\n\n0\n\n1\n\n2\n\n3\n\n4\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n2007 2009 2011 2013 2015 2017 2019 2021\n\nHealthy life years at age 65: men, years\n\nAverage (left axis) Standard deviation (right axis)"],"full_content":"Chapter 2: Social convergence in the EU: taking stock\n\n53\n\nChart 2.7 Variations in health outcome measures have remained stable Healthy life years at age 65 for women and men, and cross-country variation (measured by standard deviation), 2007-2022, EU-27\n\n_Note:_ Latest data point is 2022 Standard deviation is a measure of cross-country variation, the higher the standard deviation, the higher the cross-country variation. EU average levels are weighted values.\n\n_Source:_ DG EMPL calculations based on EU-SILC dataset hlth\\_hlye. Click here to download chart.\n\n### 3\\. CONVERGENCE IN SOCIOECONOMIC OUTCOMES WITHIN MEMBER STATES\n\n**Understanding patterns of regional convergence within countries can better clarify trends in socioeconomic outcomes.** Convergence at EU level (across countries or regions as analysed in section 2) does not necessarily imply regional convergence within countries, for example when capital and metropolitan regions strongly outperform rural and peripheral regions within a country. It is important to look beyond convergence at EU level to capture potential challenges to territorial and social cohesion within specific Member States. This section provides regional convergence analysis within each of the 19 Member States with more than four NUTS2 regions. ( ⁸⁶ ) It focuses on the indicators analysed in section 2 for which regional data are available (GDP per capita, (un)employment and NEET rates, tertiary and adult education rates, and AROPE rates), covering 2007 to 2022. It also provides an in-depth look at within-country convergence in median incomes across very small territorial units (NUTS3 regions) based on unique administrative data collected in eight Member States (Box 2.2(European Commission, )).\n\n**Upward divergence in economic performance at EU level was observed within most EU Member States.** Sixteen of the 19 Member States analysed saw their national GDP per capita increase since 2007 amid growing regional disparities (Table 2.2). The growth in regional disparities was particularly strong (more than 50%) in Denmark and several central and eastern European countries (Bulgaria, Czechia, Poland, Romania). In central and eastern Member States, divergence partly resulted from capital regions outperforming other regions. ( ⁸⁷ ) This was not always the case in other countries, e.g.: increases in regional variation in France were linked to particularly sluggish growth in regions with low levels of GDP per capita. Only Portugal experienced upward convergence in GDP per capita, but this was largely due to low growth in some developed, previously dynamic, regions. ( ⁸⁸ )\n\n**Convergence patterns in labour market outcomes varied from country to country.** Employment rates grew in most Member States since 2007, sometimes accompanied by regional convergence and sometimes by divergence (notably in Romania). Similarly, regional NEET rates converged in some countries and diverged in others, while average national rates largely declined. Concerningly, Denmark, France, and Romania all saw strong regional divergence in NEET rates. Within-country developments in unemployment rates are more reassuring, with 10 of the 18 countries experiencing upward convergence, notably Bulgaria, Czechia, Germany, and Portugal. Overall, within-country variation in labour market outcomes was highly sensitive to specific national and regional factors, but a detailed analysis is beyond the scope of this report.\n\n**Most countries experienced growth in tertiary education attainment accompanied by regional divergence.** This was the case for 16 of the 18 countries analysed. In several eastern and central European Member States (Czechia, Hungary, Poland, Romania) and Portugal, regional differences grew particularly strongly. These developments stemmed from sharp increases in tertiary education attainment in capital regions, reflecting a combination of factors including the concentration of universities, high demand for tertiary-educated workers and associated wage premiums. By contrast, a lack of tertiary education opportunities and an outflow of highly\n\n( ⁸⁶ ) Four regions was considered the lowest number reasonable to perform robust convergence analysis. ( ⁸⁷ ) (Eurofound, 2021b); (Alcidi et al., 2018) ( ⁸⁸ ) (European Commission, 2024b)\n\n0\n\n1\n\n2\n\n3\n\n4\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n2007 2009 2011 2013 2015 2017 2019 2021\n\nHealthy life years at age 65: women, years\n\nAverage (left axis) Standard deviation (right axis)\n\n0\n\n1\n\n2\n\n3\n\n4\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n2007 2009 2011 2013 2015 2017 2019 2021\n\nHealthy life years at age 65: men, years\n\nAverage (left axis) Standard deviation (right axis)","publish_date":null,"title":"3. C M S","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/employment_and_social_developments_in_europe_esde_2024_report_p50.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/employment_and_social_developments_in_europe_esde_2024_report_p50.pdf"},"request":{"example_id":"chart/employment_and_social_developments_in_europe_esde_2024_report_p50","product_type":"parse","source_file_path":"employment_and_social_developments_in_europe_esde_2024_report_p50.pdf"},"started_at":"2026-11-04T19:45:24.315451"} +{"completed_at":"2026-11-04T19:45:28.395853","latency_in_ms":2886,"output":{"example_id":"chart/employment_and_social_developments_in_europe_esde_2024_report_p55","grounded_pages":[],"layout_pages":[],"markdown":"Employment and Social Developments in Europe 2024\n\n58\n\nChart 2.8 Recent declines in gender employment and pay gaps were accompanied by some convergence Gender employment and pay gaps and cross-country variation (standard deviation), 2002-2022, EU-27\n\n_Note:_ Gender employment gap calculated for population aged 20-64. Significant breaks in the EU-LFS employment time series due to revisions of the survey over time, notably in 2005 and 2021. For cross-country variation in gender pay gap over the period 2010-2021: Greece and Croatia omitted due to missing data for several years. Information on gender pay gap for Ireland in 2021 missing. Standard deviation is a measure of cross-country variation, the higher the standard deviation, the higher the cross-country variation.\n\n_Source:_ DG EMPL calculations based on EU-LFS datasets lfsa\\_ergan & sdg\\_05\\_20.\n\nClick here to download chart.\n\n**Further progress towards equality will depend in part on addressing two long-standing challenges: gender segregation and gender differences in the career trajectories of parents. (** **95** **)** Evidence suggests that much of the observed gender gaps in employment and earnings emerge with parenthood and are closely linked to unequal division of unpaid work, limited access to formal childcare services for very young children (see Chapter 3), father’s low take-up of family leaves and, in some Member States, high taxation of second earners. ( ⁹⁶ ) Gender segregation of certain sectors of economic activity and occupations are among the key factors contributing to the gender pay gap (women tend to be overrepresented in jobs with lower salaries) ( ⁹⁷ ) and exacerbating labour shortages. ( ⁹⁸ )\n\nChart 2.9 Slow but steady rise in gender segregation in education, health and social work sectors Proportion of employment in education, human health and social work activities by gender, and cross-country variation in gender gap (standard deviation), 2008-2023, EU-27\n\n_Note:_ Break in the EU-LFS employment time series due to survey revisions in 2021. Standard deviation is a measure of cross-country variation, the higher the standard deviation, the higher the cross-country variation.\n\n_Source:_ DG EMPL calculations based on EU-LFS dataset lfsa\\_egan2 Click here to download chart.\n\n**Gender segregation of work by occupation and sector of economic activity has changed very little over time. (** **99** **)** Recent research showed limited change in existing aggregate measures of gender segregation over the last two decades in the EU. ( ¹⁰⁰ ) This can be illustrated by an analysis of the proportion of women and men employed in education, human health and social work activities, an indicator which the EIGE’s Gender Equality Index uses as a proxy for gender segregation of the EU labour market. Around 30% of all employed women in the EU worked in education, human health and social work activities in 2022, compared to only 8% of\n\n( ⁹⁵ ) (Bertrand, 2020) ( ⁹⁶ ) (Bettio, 2017); (Kleven, Landais and Soegard, 2019b); (Kleven et al., 2019a); (Goldin, 2021); (Eurofound, a) ( ⁹⁷ ) (European Commission, 2022c); (European Commission, 2022e); (Goldin, 2014); (EIGE, 2018) ( ⁹⁸ ) (European Commission, 2023b) ( ⁹⁹ ) (EIGE, 2023); (EIGE, 2018); (European Commission, 2023b) ( ¹⁰⁰ ) (European Commission, 2023b); (Eurofound and European Commission Joint Research Centre, 2021)\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\n16\n\n18\n\n20\n\na) Gender employment gap\n\nGap in pp Standard deviation\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\n16\n\n18\n\n20\n\n2010\n\n2011\n\n2012\n\n2013\n\n2014\n\n2015\n\n2016\n\n2017\n\n2018\n\n2019\n\n2020\n\n2021\n\n2022\n\nb) Gender pay gap\n\nGap in % Standard deviation\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n35\n\n2008\n\n2009\n\n2010\n\n2011\n\n2012\n\n2013\n\n2014\n\n2015\n\n2016\n\n2017\n\n2018\n\n2019\n\n2020\n\n2021\n\n2022\n\n2023\n\na) % of workers in education,\n\nhealth and social work by gender\n\nMen (%) Women (%)\n\nGender gap\n\n0\n\n1\n\n2\n\n3\n\n4\n\n5\n\n6\n\n2008\n\n2009\n\n2010\n\n2011\n\n2012\n\n2013\n\n2014\n\n2015\n\n2016\n\n2017\n\n2018\n\n2019\n\n2020\n\n2021\n\n2022\n\n2023\n\nb) Cross-country variation in gender\n\ngap","pages":[{"line_numbers":[],"markdown":"Employment and Social Developments in Europe 2024\n\n58\n\nChart 2.8 Recent declines in gender employment and pay gaps were accompanied by some convergence Gender employment and pay gaps and cross-country variation (standard deviation), 2002-2022, EU-27\n\n_Note:_ Gender employment gap calculated for population aged 20-64. Significant breaks in the EU-LFS employment time series due to revisions of the survey over time, notably in 2005 and 2021. For cross-country variation in gender pay gap over the period 2010-2021: Greece and Croatia omitted due to missing data for several years. Information on gender pay gap for Ireland in 2021 missing. Standard deviation is a measure of cross-country variation, the higher the standard deviation, the higher the cross-country variation.\n\n_Source:_ DG EMPL calculations based on EU-LFS datasets lfsa\\_ergan & sdg\\_05\\_20.\n\nClick here to download chart.\n\n**Further progress towards equality will depend in part on addressing two long-standing challenges: gender segregation and gender differences in the career trajectories of parents. (** **95** **)** Evidence suggests that much of the observed gender gaps in employment and earnings emerge with parenthood and are closely linked to unequal division of unpaid work, limited access to formal childcare services for very young children (see Chapter 3), father’s low take-up of family leaves and, in some Member States, high taxation of second earners. ( ⁹⁶ ) Gender segregation of certain sectors of economic activity and occupations are among the key factors contributing to the gender pay gap (women tend to be overrepresented in jobs with lower salaries) ( ⁹⁷ ) and exacerbating labour shortages. ( ⁹⁸ )\n\nChart 2.9 Slow but steady rise in gender segregation in education, health and social work sectors Proportion of employment in education, human health and social work activities by gender, and cross-country variation in gender gap (standard deviation), 2008-2023, EU-27\n\n_Note:_ Break in the EU-LFS employment time series due to survey revisions in 2021. Standard deviation is a measure of cross-country variation, the higher the standard deviation, the higher the cross-country variation.\n\n_Source:_ DG EMPL calculations based on EU-LFS dataset lfsa\\_egan2 Click here to download chart.\n\n**Gender segregation of work by occupation and sector of economic activity has changed very little over time. (** **99** **)** Recent research showed limited change in existing aggregate measures of gender segregation over the last two decades in the EU. ( ¹⁰⁰ ) This can be illustrated by an analysis of the proportion of women and men employed in education, human health and social work activities, an indicator which the EIGE’s Gender Equality Index uses as a proxy for gender segregation of the EU labour market. Around 30% of all employed women in the EU worked in education, human health and social work activities in 2022, compared to only 8% of\n\n( ⁹⁵ ) (Bertrand, 2020) ( ⁹⁶ ) (Bettio, 2017); (Kleven, Landais and Soegard, 2019b); (Kleven et al., 2019a); (Goldin, 2021); (Eurofound, a) ( ⁹⁷ ) (European Commission, 2022c); (European Commission, 2022e); (Goldin, 2014); (EIGE, 2018) ( ⁹⁸ ) (European Commission, 2023b) ( ⁹⁹ ) (EIGE, 2023); (EIGE, 2018); (European Commission, 2023b) ( ¹⁰⁰ ) (European Commission, 2023b); (Eurofound and European Commission Joint Research Centre, 2021)\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\n16\n\n18\n\n20\n\na) Gender employment gap\n\nGap in pp Standard deviation\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\n16\n\n18\n\n20\n\n2010\n\n2011\n\n2012\n\n2013\n\n2014\n\n2015\n\n2016\n\n2017\n\n2018\n\n2019\n\n2020\n\n2021\n\n2022\n\nb) Gender pay gap\n\nGap in % Standard deviation\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n35\n\n2008\n\n2009\n\n2010\n\n2011\n\n2012\n\n2013\n\n2014\n\n2015\n\n2016\n\n2017\n\n2018\n\n2019\n\n2020\n\n2021\n\n2022\n\n2023\n\na) % of workers in education,\n\nhealth and social work by gender\n\nMen (%) Women (%)\n\nGender gap\n\n0\n\n1\n\n2\n\n3\n\n4\n\n5\n\n6\n\n2008\n\n2009\n\n2010\n\n2011\n\n2012\n\n2013\n\n2014\n\n2015\n\n2016\n\n2017\n\n2018\n\n2019\n\n2020\n\n2021\n\n2022\n\n2023\n\nb) Cross-country variation in gender\n\ngap","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Employment and Social Developments in Europe 2024\n\n58\n\nChart 2.8 Recent declines in gender employment and pay gaps were accompanied by some convergence Gender employment and pay gaps and cross-country variation (standard deviation), 2002-2022, EU-27\n\n_Note:_ Gender employment gap calculated for population aged 20-64. Significant breaks in the EU-LFS employment time series due to revisions of the survey over time, notably in 2005 and 2021. For cross-country variation in gender pay gap over the period 2010-2021: Greece and Croatia omitted due to missing data for several years. Information on gender pay gap for Ireland in 2021 missing. Standard deviation is a measure of cross-country variation, the higher the standard deviation, the higher the cross-country variation.\n\n_Source:_ DG EMPL calculations based on EU-LFS datasets lfsa\\_ergan & sdg\\_05\\_20.\n\nClick here to download chart.\n\n**Further progress towards equality will depend in part on addressing two long-standing challenges: gender segregation and gender differences in the career trajectories of parents. (** **95** **)** Evidence suggests that much of the observed gender gaps in employment and earnings emerge with parenthood and are closely linked to unequal division of unpaid work, limited access to formal childcare services for very young children (see Chapter 3), father’s low take-up of family leaves and, in some Member States, high taxation of second earners. ( ⁹⁶ ) Gender segregation of certain sectors of economic activity and occupations are among the key factors contributing to the gender pay gap (women tend to be overrepresented in jobs with lower salaries) ( ⁹⁷ ) and exacerbating labour shortages. ( ⁹⁸ )\n\nChart 2.9 Slow but steady rise in gender segregation in education, health and social work sectors Proportion of employment in education, human health and social work activities by gender, and cross-country variation in gender gap (standard deviation), 2008-2023, EU-27\n\n_Note:_ Break in the EU-LFS employment time series due to survey revisions in 2021. Standard deviation is a measure of cross-country variation, the higher the standard deviation, the higher the cross-country variation.\n\n_Source:_ DG EMPL calculations based on EU-LFS dataset lfsa\\_egan2 Click here to download chart.\n\n**Gender segregation of work by occupation and sector of economic activity has changed very little over time. (** **99** **)** Recent research showed limited change in existing aggregate measures of gender segregation over the last two decades in the EU. ( ¹⁰⁰ ) This can be illustrated by an analysis of the proportion of women and men employed in education, human health and social work activities, an indicator which the EIGE’s Gender Equality Index uses as a proxy for gender segregation of the EU labour market. Around 30% of all employed women in the EU worked in education, human health and social work activities in 2022, compared to only 8% of\n\n( ⁹⁵ ) (Bertrand, 2020) ( ⁹⁶ ) (Bettio, 2017); (Kleven, Landais and Soegard, 2019b); (Kleven et al., 2019a); (Goldin, 2021); (Eurofound, a) ( ⁹⁷ ) (European Commission, 2022c); (European Commission, 2022e); (Goldin, 2014); (EIGE, 2018) ( ⁹⁸ ) (European Commission, 2023b) ( ⁹⁹ ) (EIGE, 2023); (EIGE, 2018); (European Commission, 2023b) ( ¹⁰⁰ ) (European Commission, 2023b); (Eurofound and European Commission Joint Research Centre, 2021)\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\n16\n\n18\n\n20\n\na) Gender employment gap\n\nGap in pp Standard deviation\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\n16\n\n18\n\n20\n\n2010\n\n2011\n\n2012\n\n2013\n\n2014\n\n2015\n\n2016\n\n2017\n\n2018\n\n2019\n\n2020\n\n2021\n\n2022\n\nb) Gender pay gap\n\nGap in % Standard deviation\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n35\n\n2008\n\n2009\n\n2010\n\n2011\n\n2012\n\n2013\n\n2014\n\n2015\n\n2016\n\n2017\n\n2018\n\n2019\n\n2020\n\n2021\n\n2022\n\n2023\n\na) % of workers in education,\n\nhealth and social work by gender\n\nMen (%) Women (%)\n\nGender gap\n\n0\n\n1\n\n2\n\n3\n\n4\n\n5\n\n6\n\n2008\n\n2009\n\n2010\n\n2011\n\n2012\n\n2013\n\n2014\n\n2015\n\n2016\n\n2017\n\n2018\n\n2019\n\n2020\n\n2021\n\n2022\n\n2023\n\nb) Cross-country variation in gender\n\ngap","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/employment_and_social_developments_in_europe_esde_2024_report_p55.pdf"]},"response":{"errors":[],"extract_id":"extract_f3d77c33aa69383dce0162b82d2b7bf5","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Employment and Social Developments in Europe 2024\n\n58\n\nChart 2.8 Recent declines in gender employment and pay gaps were accompanied by some convergence Gender employment and pay gaps and cross-country variation (standard deviation), 2002-2022, EU-27\n\n_Note:_ Gender employment gap calculated for population aged 20-64. Significant breaks in the EU-LFS employment time series due to revisions of the survey over time, notably in 2005 and 2021. For cross-country variation in gender pay gap over the period 2010-2021: Greece and Croatia omitted due to missing data for several years. Information on gender pay gap for Ireland in 2021 missing. Standard deviation is a measure of cross-country variation, the higher the standard deviation, the higher the cross-country variation.\n\n_Source:_ DG EMPL calculations based on EU-LFS datasets lfsa\\_ergan & sdg\\_05\\_20.\n\nClick here to download chart.\n\n**Further progress towards equality will depend in part on addressing two long-standing challenges: gender segregation and gender differences in the career trajectories of parents. (** **95** **)** Evidence suggests that much of the observed gender gaps in employment and earnings emerge with parenthood and are closely linked to unequal division of unpaid work, limited access to formal childcare services for very young children (see Chapter 3), father’s low take-up of family leaves and, in some Member States, high taxation of second earners. ( ⁹⁶ ) Gender segregation of certain sectors of economic activity and occupations are among the key factors contributing to the gender pay gap (women tend to be overrepresented in jobs with lower salaries) ( ⁹⁷ ) and exacerbating labour shortages. ( ⁹⁸ )\n\nChart 2.9 Slow but steady rise in gender segregation in education, health and social work sectors Proportion of employment in education, human health and social work activities by gender, and cross-country variation in gender gap (standard deviation), 2008-2023, EU-27\n\n_Note:_ Break in the EU-LFS employment time series due to survey revisions in 2021. Standard deviation is a measure of cross-country variation, the higher the standard deviation, the higher the cross-country variation.\n\n_Source:_ DG EMPL calculations based on EU-LFS dataset lfsa\\_egan2 Click here to download chart.\n\n**Gender segregation of work by occupation and sector of economic activity has changed very little over time. (** **99** **)** Recent research showed limited change in existing aggregate measures of gender segregation over the last two decades in the EU. ( ¹⁰⁰ ) This can be illustrated by an analysis of the proportion of women and men employed in education, human health and social work activities, an indicator which the EIGE’s Gender Equality Index uses as a proxy for gender segregation of the EU labour market. Around 30% of all employed women in the EU worked in education, human health and social work activities in 2022, compared to only 8% of\n\n( ⁹⁵ ) (Bertrand, 2020) ( ⁹⁶ ) (Bettio, 2017); (Kleven, Landais and Soegard, 2019b); (Kleven et al., 2019a); (Goldin, 2021); (Eurofound, a) ( ⁹⁷ ) (European Commission, 2022c); (European Commission, 2022e); (Goldin, 2014); (EIGE, 2018) ( ⁹⁸ ) (European Commission, 2023b) ( ⁹⁹ ) (EIGE, 2023); (EIGE, 2018); (European Commission, 2023b) ( ¹⁰⁰ ) (European Commission, 2023b); (Eurofound and European Commission Joint Research Centre, 2021)\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\n16\n\n18\n\n20\n\na) Gender employment gap\n\nGap in pp Standard deviation\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\n16\n\n18\n\n20\n\n2010\n\n2011\n\n2012\n\n2013\n\n2014\n\n2015\n\n2016\n\n2017\n\n2018\n\n2019\n\n2020\n\n2021\n\n2022\n\nb) Gender pay gap\n\nGap in % Standard deviation\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n35\n\n2008\n\n2009\n\n2010\n\n2011\n\n2012\n\n2013\n\n2014\n\n2015\n\n2016\n\n2017\n\n2018\n\n2019\n\n2020\n\n2021\n\n2022\n\n2023\n\na) % of workers in education,\n\nhealth and social work by gender\n\nMen (%) Women (%)\n\nGender gap\n\n0\n\n1\n\n2\n\n3\n\n4\n\n5\n\n6\n\n2008\n\n2009\n\n2010\n\n2011\n\n2012\n\n2013\n\n2014\n\n2015\n\n2016\n\n2017\n\n2018\n\n2019\n\n2020\n\n2021\n\n2022\n\n2023\n\nb) Cross-country variation in gender\n\ngap"],"full_content":"Employment and Social Developments in Europe 2024\n\n58\n\nChart 2.8 Recent declines in gender employment and pay gaps were accompanied by some convergence Gender employment and pay gaps and cross-country variation (standard deviation), 2002-2022, EU-27\n\n_Note:_ Gender employment gap calculated for population aged 20-64. Significant breaks in the EU-LFS employment time series due to revisions of the survey over time, notably in 2005 and 2021. For cross-country variation in gender pay gap over the period 2010-2021: Greece and Croatia omitted due to missing data for several years. Information on gender pay gap for Ireland in 2021 missing. Standard deviation is a measure of cross-country variation, the higher the standard deviation, the higher the cross-country variation.\n\n_Source:_ DG EMPL calculations based on EU-LFS datasets lfsa\\_ergan & sdg\\_05\\_20.\n\nClick here to download chart.\n\n**Further progress towards equality will depend in part on addressing two long-standing challenges: gender segregation and gender differences in the career trajectories of parents. (** **95** **)** Evidence suggests that much of the observed gender gaps in employment and earnings emerge with parenthood and are closely linked to unequal division of unpaid work, limited access to formal childcare services for very young children (see Chapter 3), father’s low take-up of family leaves and, in some Member States, high taxation of second earners. ( ⁹⁶ ) Gender segregation of certain sectors of economic activity and occupations are among the key factors contributing to the gender pay gap (women tend to be overrepresented in jobs with lower salaries) ( ⁹⁷ ) and exacerbating labour shortages. ( ⁹⁸ )\n\nChart 2.9 Slow but steady rise in gender segregation in education, health and social work sectors Proportion of employment in education, human health and social work activities by gender, and cross-country variation in gender gap (standard deviation), 2008-2023, EU-27\n\n_Note:_ Break in the EU-LFS employment time series due to survey revisions in 2021. Standard deviation is a measure of cross-country variation, the higher the standard deviation, the higher the cross-country variation.\n\n_Source:_ DG EMPL calculations based on EU-LFS dataset lfsa\\_egan2 Click here to download chart.\n\n**Gender segregation of work by occupation and sector of economic activity has changed very little over time. (** **99** **)** Recent research showed limited change in existing aggregate measures of gender segregation over the last two decades in the EU. ( ¹⁰⁰ ) This can be illustrated by an analysis of the proportion of women and men employed in education, human health and social work activities, an indicator which the EIGE’s Gender Equality Index uses as a proxy for gender segregation of the EU labour market. Around 30% of all employed women in the EU worked in education, human health and social work activities in 2022, compared to only 8% of\n\n( ⁹⁵ ) (Bertrand, 2020) ( ⁹⁶ ) (Bettio, 2017); (Kleven, Landais and Soegard, 2019b); (Kleven et al., 2019a); (Goldin, 2021); (Eurofound, a) ( ⁹⁷ ) (European Commission, 2022c); (European Commission, 2022e); (Goldin, 2014); (EIGE, 2018) ( ⁹⁸ ) (European Commission, 2023b) ( ⁹⁹ ) (EIGE, 2023); (EIGE, 2018); (European Commission, 2023b) ( ¹⁰⁰ ) (European Commission, 2023b); (Eurofound and European Commission Joint Research Centre, 2021)\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\n16\n\n18\n\n20\n\na) Gender employment gap\n\nGap in pp Standard deviation\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\n16\n\n18\n\n20\n\n2010\n\n2011\n\n2012\n\n2013\n\n2014\n\n2015\n\n2016\n\n2017\n\n2018\n\n2019\n\n2020\n\n2021\n\n2022\n\nb) Gender pay gap\n\nGap in % Standard deviation\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n35\n\n2008\n\n2009\n\n2010\n\n2011\n\n2012\n\n2013\n\n2014\n\n2015\n\n2016\n\n2017\n\n2018\n\n2019\n\n2020\n\n2021\n\n2022\n\n2023\n\na) % of workers in education,\n\nhealth and social work by gender\n\nMen (%) Women (%)\n\nGender gap\n\n0\n\n1\n\n2\n\n3\n\n4\n\n5\n\n6\n\n2008\n\n2009\n\n2010\n\n2011\n\n2012\n\n2013\n\n2014\n\n2015\n\n2016\n\n2017\n\n2018\n\n2019\n\n2020\n\n2021\n\n2022\n\n2023\n\nb) Cross-country variation in gender\n\ngap","publish_date":null,"title":"Recent declines in gender employment and pay gaps were accompanied by some convergence Further progress towards equality will depend in","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/employment_and_social_developments_in_europe_esde_2024_report_p55.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/employment_and_social_developments_in_europe_esde_2024_report_p55.pdf"},"request":{"example_id":"chart/employment_and_social_developments_in_europe_esde_2024_report_p55","product_type":"parse","source_file_path":"employment_and_social_developments_in_europe_esde_2024_report_p55.pdf"},"started_at":"2026-11-04T19:45:25.509514"} +{"completed_at":"2026-11-04T19:45:27.470363","latency_in_ms":1161,"output":{"example_id":"chart/employment_and_social_developments_in_europe_esde_2024_report_p56","grounded_pages":[],"layout_pages":[],"markdown":"Chapter 2: Social convergence in the EU: taking stock\n\n59\n\nall men, resulting in a gender gap of about 22 pp (Chart 2.9). This gap increased since 2008, when it stood at about 19 pp. Over the same period, the variation in gender segregation in education and health activities across countries decreased (Chart 2.9). In other words, the EU Member States converged, but towards a somewhat more unequal outcome.\n\n**Involvement of women in unpaid care has dropped somewhat since 2007, but women remain considerably more engaged in unpaid care than men** . In 2022, one in four men in the EU were involved daily in unpaid care, and around one in three in housework (Chart 2.10). For women, the corresponding shares were 34% and 63%, respectively. The proportion of men involved in unpaid care remained relatively stable since 2007. Female involvement in unpaid care dropped slightly (by about 6 pp), reflecting the increased availability of formal childcare (see Chapter 3(Cedefop, n.d.)). The share of women doing housework on a daily basis dropped by 16 pp over the same period (notably since 2016), while the involvement of men increased by 10 pp.\n\n**The evidence of convergence in gender gaps in unpaid work is mixed over time and varies considerably between Member States.** In several countries, the share of women involved daily in unpaid care is at least 15 pp higher than men, while in others there is little difference. Although the magnitude of cross- country differences has remained similar since 2007 (Chart 2.10), some evidence points towards countries with larger gaps catching-up with those with smaller gaps. The gender gap in housework varies between 12 pp and 44 pp across countries. The gap has converged since 2016, but there is no evidence of catching-up by countries with the largest gaps.\n\nChart 2.10 Limited upward convergence in gender gaps in unpaid work Daily involvement in unpaid care and housework by gender, and cross-country variation in gender gaps (standard deviation), 2007-2022, EU-27\n\n_Note:_ Involvement in unpaid care defined as ‘People caring for and educating their children or grandchildren, elderly or people with disabilities, every day (%, 18-74 population)’. Involvement in housework defined as ‘People doing cooking and/or housework, every day (%, 18-74 population)’. Standard deviation is a measure of cross-country variation, the higher the standard deviation, the higher the cross-country variation.\n\n_Source:_ DG EMPL calculations based on EIGE’s Gender Statistics Database\n\nClick here to download chart.\n\n**Gender segregation in the labour market and unequal division of unpaid work are firmly rooted in persistent stereotypes about gender-specific skills and household roles. (** **101** **)** Challenging such stereotypes is crucial to promote gender equality in the labour market. This is a key objective of the gender equality policy outlined in the EU Gender Equality Strategy 2020-2025. ( ¹⁰² ) The remainder of this section explores trends and convergence in attitudes towards the involvement of women and men in paid and unpaid work based on data collected in the European Values Survey between 1990 and 2017 for the 20 Member States, for which data are consistently available over such a long-time span. ( ¹⁰³ )\n\n### 4\\.2. Attitudes to women’s work and sharing unpaid work\n\n**Increasingly positive attitudes to women’s involvement in paid work are consistent with long-term reductions in the gender employment gap.** The share of people who do not believe that children suffer when mothers are in paid work increased from less than 40% in 1990 to more than 60% in 2017, with similar proportions among women and men. A comparable increase applies to the share of people who do not consider women to be primarily interested in home and children rather than paid work (Chart 2.11). The proportion of\n\n( ¹⁰¹ ) (Bertrand, 2020); (EIGE, 2023); (EIGE, 2018) ( ¹⁰² ) See EU Gender Equality Strategy 2020-2025 here. ( ¹⁰³ ) The analysis builds on previous research on attitudes in this area, covering earlier developments in the EU context ( (Knight and Brinton, 2017); (Grunow, Begall and Buchler, 2018); (Brinton and Lee, 2016); (Scarborough, Sin and Risman, 2019))\n\n0\n\n10 20 30 40 50 60 70 80 90\n\n2007 2012 2016 2022\n\na) Gap at EU level\n\nUnpaid care (% Men) Unpaid care (% Women)\n\nHousework (% Men) Housework (% Women)\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\n2007 2012 2016 2022\n\nb) Cross-country variation\n\nUnpaid care Housework","pages":[{"line_numbers":[],"markdown":"Chapter 2: Social convergence in the EU: taking stock\n\n59\n\nall men, resulting in a gender gap of about 22 pp (Chart 2.9). This gap increased since 2008, when it stood at about 19 pp. Over the same period, the variation in gender segregation in education and health activities across countries decreased (Chart 2.9). In other words, the EU Member States converged, but towards a somewhat more unequal outcome.\n\n**Involvement of women in unpaid care has dropped somewhat since 2007, but women remain considerably more engaged in unpaid care than men** . In 2022, one in four men in the EU were involved daily in unpaid care, and around one in three in housework (Chart 2.10). For women, the corresponding shares were 34% and 63%, respectively. The proportion of men involved in unpaid care remained relatively stable since 2007. Female involvement in unpaid care dropped slightly (by about 6 pp), reflecting the increased availability of formal childcare (see Chapter 3(Cedefop, n.d.)). The share of women doing housework on a daily basis dropped by 16 pp over the same period (notably since 2016), while the involvement of men increased by 10 pp.\n\n**The evidence of convergence in gender gaps in unpaid work is mixed over time and varies considerably between Member States.** In several countries, the share of women involved daily in unpaid care is at least 15 pp higher than men, while in others there is little difference. Although the magnitude of cross- country differences has remained similar since 2007 (Chart 2.10), some evidence points towards countries with larger gaps catching-up with those with smaller gaps. The gender gap in housework varies between 12 pp and 44 pp across countries. The gap has converged since 2016, but there is no evidence of catching-up by countries with the largest gaps.\n\nChart 2.10 Limited upward convergence in gender gaps in unpaid work Daily involvement in unpaid care and housework by gender, and cross-country variation in gender gaps (standard deviation), 2007-2022, EU-27\n\n_Note:_ Involvement in unpaid care defined as ‘People caring for and educating their children or grandchildren, elderly or people with disabilities, every day (%, 18-74 population)’. Involvement in housework defined as ‘People doing cooking and/or housework, every day (%, 18-74 population)’. Standard deviation is a measure of cross-country variation, the higher the standard deviation, the higher the cross-country variation.\n\n_Source:_ DG EMPL calculations based on EIGE’s Gender Statistics Database\n\nClick here to download chart.\n\n**Gender segregation in the labour market and unequal division of unpaid work are firmly rooted in persistent stereotypes about gender-specific skills and household roles. (** **101** **)** Challenging such stereotypes is crucial to promote gender equality in the labour market. This is a key objective of the gender equality policy outlined in the EU Gender Equality Strategy 2020-2025. ( ¹⁰² ) The remainder of this section explores trends and convergence in attitudes towards the involvement of women and men in paid and unpaid work based on data collected in the European Values Survey between 1990 and 2017 for the 20 Member States, for which data are consistently available over such a long-time span. ( ¹⁰³ )\n\n### 4\\.2. Attitudes to women’s work and sharing unpaid work\n\n**Increasingly positive attitudes to women’s involvement in paid work are consistent with long-term reductions in the gender employment gap.** The share of people who do not believe that children suffer when mothers are in paid work increased from less than 40% in 1990 to more than 60% in 2017, with similar proportions among women and men. A comparable increase applies to the share of people who do not consider women to be primarily interested in home and children rather than paid work (Chart 2.11). The proportion of\n\n( ¹⁰¹ ) (Bertrand, 2020); (EIGE, 2023); (EIGE, 2018) ( ¹⁰² ) See EU Gender Equality Strategy 2020-2025 here. ( ¹⁰³ ) The analysis builds on previous research on attitudes in this area, covering earlier developments in the EU context ( (Knight and Brinton, 2017); (Grunow, Begall and Buchler, 2018); (Brinton and Lee, 2016); (Scarborough, Sin and Risman, 2019))\n\n0\n\n10 20 30 40 50 60 70 80 90\n\n2007 2012 2016 2022\n\na) Gap at EU level\n\nUnpaid care (% Men) Unpaid care (% Women)\n\nHousework (% Men) Housework (% Women)\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\n2007 2012 2016 2022\n\nb) Cross-country variation\n\nUnpaid care Housework","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Chapter 2: Social convergence in the EU: taking stock\n\n59\n\nall men, resulting in a gender gap of about 22 pp (Chart 2.9). This gap increased since 2008, when it stood at about 19 pp. Over the same period, the variation in gender segregation in education and health activities across countries decreased (Chart 2.9). In other words, the EU Member States converged, but towards a somewhat more unequal outcome.\n\n**Involvement of women in unpaid care has dropped somewhat since 2007, but women remain considerably more engaged in unpaid care than men** . In 2022, one in four men in the EU were involved daily in unpaid care, and around one in three in housework (Chart 2.10). For women, the corresponding shares were 34% and 63%, respectively. The proportion of men involved in unpaid care remained relatively stable since 2007. Female involvement in unpaid care dropped slightly (by about 6 pp), reflecting the increased availability of formal childcare (see Chapter 3(Cedefop, n.d.)). The share of women doing housework on a daily basis dropped by 16 pp over the same period (notably since 2016), while the involvement of men increased by 10 pp.\n\n**The evidence of convergence in gender gaps in unpaid work is mixed over time and varies considerably between Member States.** In several countries, the share of women involved daily in unpaid care is at least 15 pp higher than men, while in others there is little difference. Although the magnitude of cross- country differences has remained similar since 2007 (Chart 2.10), some evidence points towards countries with larger gaps catching-up with those with smaller gaps. The gender gap in housework varies between 12 pp and 44 pp across countries. The gap has converged since 2016, but there is no evidence of catching-up by countries with the largest gaps.\n\nChart 2.10 Limited upward convergence in gender gaps in unpaid work Daily involvement in unpaid care and housework by gender, and cross-country variation in gender gaps (standard deviation), 2007-2022, EU-27\n\n_Note:_ Involvement in unpaid care defined as ‘People caring for and educating their children or grandchildren, elderly or people with disabilities, every day (%, 18-74 population)’. Involvement in housework defined as ‘People doing cooking and/or housework, every day (%, 18-74 population)’. Standard deviation is a measure of cross-country variation, the higher the standard deviation, the higher the cross-country variation.\n\n_Source:_ DG EMPL calculations based on EIGE’s Gender Statistics Database\n\nClick here to download chart.\n\n**Gender segregation in the labour market and unequal division of unpaid work are firmly rooted in persistent stereotypes about gender-specific skills and household roles. (** **101** **)** Challenging such stereotypes is crucial to promote gender equality in the labour market. This is a key objective of the gender equality policy outlined in the EU Gender Equality Strategy 2020-2025. ( ¹⁰² ) The remainder of this section explores trends and convergence in attitudes towards the involvement of women and men in paid and unpaid work based on data collected in the European Values Survey between 1990 and 2017 for the 20 Member States, for which data are consistently available over such a long-time span. ( ¹⁰³ )\n\n### 4\\.2. Attitudes to women’s work and sharing unpaid work\n\n**Increasingly positive attitudes to women’s involvement in paid work are consistent with long-term reductions in the gender employment gap.** The share of people who do not believe that children suffer when mothers are in paid work increased from less than 40% in 1990 to more than 60% in 2017, with similar proportions among women and men. A comparable increase applies to the share of people who do not consider women to be primarily interested in home and children rather than paid work (Chart 2.11). The proportion of\n\n( ¹⁰¹ ) (Bertrand, 2020); (EIGE, 2023); (EIGE, 2018) ( ¹⁰² ) See EU Gender Equality Strategy 2020-2025 here. ( ¹⁰³ ) The analysis builds on previous research on attitudes in this area, covering earlier developments in the EU context ( (Knight and Brinton, 2017); (Grunow, Begall and Buchler, 2018); (Brinton and Lee, 2016); (Scarborough, Sin and Risman, 2019))\n\n0\n\n10 20 30 40 50 60 70 80 90\n\n2007 2012 2016 2022\n\na) Gap at EU level\n\nUnpaid care (% Men) Unpaid care (% Women)\n\nHousework (% Men) Housework (% Women)\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\n2007 2012 2016 2022\n\nb) Cross-country variation\n\nUnpaid care Housework","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/employment_and_social_developments_in_europe_esde_2024_report_p56.pdf"]},"response":{"errors":[],"extract_id":"extract_0e655d3bacf3ba342de3c1b4a9ea99e4","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Chapter 2: Social convergence in the EU: taking stock\n\n59\n\nall men, resulting in a gender gap of about 22 pp (Chart 2.9). This gap increased since 2008, when it stood at about 19 pp. Over the same period, the variation in gender segregation in education and health activities across countries decreased (Chart 2.9). In other words, the EU Member States converged, but towards a somewhat more unequal outcome.\n\n**Involvement of women in unpaid care has dropped somewhat since 2007, but women remain considerably more engaged in unpaid care than men** . In 2022, one in four men in the EU were involved daily in unpaid care, and around one in three in housework (Chart 2.10). For women, the corresponding shares were 34% and 63%, respectively. The proportion of men involved in unpaid care remained relatively stable since 2007. Female involvement in unpaid care dropped slightly (by about 6 pp), reflecting the increased availability of formal childcare (see Chapter 3(Cedefop, n.d.)). The share of women doing housework on a daily basis dropped by 16 pp over the same period (notably since 2016), while the involvement of men increased by 10 pp.\n\n**The evidence of convergence in gender gaps in unpaid work is mixed over time and varies considerably between Member States.** In several countries, the share of women involved daily in unpaid care is at least 15 pp higher than men, while in others there is little difference. Although the magnitude of cross- country differences has remained similar since 2007 (Chart 2.10), some evidence points towards countries with larger gaps catching-up with those with smaller gaps. The gender gap in housework varies between 12 pp and 44 pp across countries. The gap has converged since 2016, but there is no evidence of catching-up by countries with the largest gaps.\n\nChart 2.10 Limited upward convergence in gender gaps in unpaid work Daily involvement in unpaid care and housework by gender, and cross-country variation in gender gaps (standard deviation), 2007-2022, EU-27\n\n_Note:_ Involvement in unpaid care defined as ‘People caring for and educating their children or grandchildren, elderly or people with disabilities, every day (%, 18-74 population)’. Involvement in housework defined as ‘People doing cooking and/or housework, every day (%, 18-74 population)’. Standard deviation is a measure of cross-country variation, the higher the standard deviation, the higher the cross-country variation.\n\n_Source:_ DG EMPL calculations based on EIGE’s Gender Statistics Database\n\nClick here to download chart.\n\n**Gender segregation in the labour market and unequal division of unpaid work are firmly rooted in persistent stereotypes about gender-specific skills and household roles. (** **101** **)** Challenging such stereotypes is crucial to promote gender equality in the labour market. This is a key objective of the gender equality policy outlined in the EU Gender Equality Strategy 2020-2025. ( ¹⁰² ) The remainder of this section explores trends and convergence in attitudes towards the involvement of women and men in paid and unpaid work based on data collected in the European Values Survey between 1990 and 2017 for the 20 Member States, for which data are consistently available over such a long-time span. ( ¹⁰³ )\n\n### 4\\.2. Attitudes to women’s work and sharing unpaid work\n\n**Increasingly positive attitudes to women’s involvement in paid work are consistent with long-term reductions in the gender employment gap.** The share of people who do not believe that children suffer when mothers are in paid work increased from less than 40% in 1990 to more than 60% in 2017, with similar proportions among women and men. A comparable increase applies to the share of people who do not consider women to be primarily interested in home and children rather than paid work (Chart 2.11). The proportion of\n\n( ¹⁰¹ ) (Bertrand, 2020); (EIGE, 2023); (EIGE, 2018) ( ¹⁰² ) See EU Gender Equality Strategy 2020-2025 here. ( ¹⁰³ ) The analysis builds on previous research on attitudes in this area, covering earlier developments in the EU context ( (Knight and Brinton, 2017); (Grunow, Begall and Buchler, 2018); (Brinton and Lee, 2016); (Scarborough, Sin and Risman, 2019))\n\n0\n\n10 20 30 40 50 60 70 80 90\n\n2007 2012 2016 2022\n\na) Gap at EU level\n\nUnpaid care (% Men) Unpaid care (% Women)\n\nHousework (% Men) Housework (% Women)\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\n2007 2012 2016 2022\n\nb) Cross-country variation\n\nUnpaid care Housework"],"full_content":"Chapter 2: Social convergence in the EU: taking stock\n\n59\n\nall men, resulting in a gender gap of about 22 pp (Chart 2.9). This gap increased since 2008, when it stood at about 19 pp. Over the same period, the variation in gender segregation in education and health activities across countries decreased (Chart 2.9). In other words, the EU Member States converged, but towards a somewhat more unequal outcome.\n\n**Involvement of women in unpaid care has dropped somewhat since 2007, but women remain considerably more engaged in unpaid care than men** . In 2022, one in four men in the EU were involved daily in unpaid care, and around one in three in housework (Chart 2.10). For women, the corresponding shares were 34% and 63%, respectively. The proportion of men involved in unpaid care remained relatively stable since 2007. Female involvement in unpaid care dropped slightly (by about 6 pp), reflecting the increased availability of formal childcare (see Chapter 3(Cedefop, n.d.)). The share of women doing housework on a daily basis dropped by 16 pp over the same period (notably since 2016), while the involvement of men increased by 10 pp.\n\n**The evidence of convergence in gender gaps in unpaid work is mixed over time and varies considerably between Member States.** In several countries, the share of women involved daily in unpaid care is at least 15 pp higher than men, while in others there is little difference. Although the magnitude of cross- country differences has remained similar since 2007 (Chart 2.10), some evidence points towards countries with larger gaps catching-up with those with smaller gaps. The gender gap in housework varies between 12 pp and 44 pp across countries. The gap has converged since 2016, but there is no evidence of catching-up by countries with the largest gaps.\n\nChart 2.10 Limited upward convergence in gender gaps in unpaid work Daily involvement in unpaid care and housework by gender, and cross-country variation in gender gaps (standard deviation), 2007-2022, EU-27\n\n_Note:_ Involvement in unpaid care defined as ‘People caring for and educating their children or grandchildren, elderly or people with disabilities, every day (%, 18-74 population)’. Involvement in housework defined as ‘People doing cooking and/or housework, every day (%, 18-74 population)’. Standard deviation is a measure of cross-country variation, the higher the standard deviation, the higher the cross-country variation.\n\n_Source:_ DG EMPL calculations based on EIGE’s Gender Statistics Database\n\nClick here to download chart.\n\n**Gender segregation in the labour market and unequal division of unpaid work are firmly rooted in persistent stereotypes about gender-specific skills and household roles. (** **101** **)** Challenging such stereotypes is crucial to promote gender equality in the labour market. This is a key objective of the gender equality policy outlined in the EU Gender Equality Strategy 2020-2025. ( ¹⁰² ) The remainder of this section explores trends and convergence in attitudes towards the involvement of women and men in paid and unpaid work based on data collected in the European Values Survey between 1990 and 2017 for the 20 Member States, for which data are consistently available over such a long-time span. ( ¹⁰³ )\n\n### 4\\.2. Attitudes to women’s work and sharing unpaid work\n\n**Increasingly positive attitudes to women’s involvement in paid work are consistent with long-term reductions in the gender employment gap.** The share of people who do not believe that children suffer when mothers are in paid work increased from less than 40% in 1990 to more than 60% in 2017, with similar proportions among women and men. A comparable increase applies to the share of people who do not consider women to be primarily interested in home and children rather than paid work (Chart 2.11). The proportion of\n\n( ¹⁰¹ ) (Bertrand, 2020); (EIGE, 2023); (EIGE, 2018) ( ¹⁰² ) See EU Gender Equality Strategy 2020-2025 here. ( ¹⁰³ ) The analysis builds on previous research on attitudes in this area, covering earlier developments in the EU context ( (Knight and Brinton, 2017); (Grunow, Begall and Buchler, 2018); (Brinton and Lee, 2016); (Scarborough, Sin and Risman, 2019))\n\n0\n\n10 20 30 40 50 60 70 80 90\n\n2007 2012 2016 2022\n\na) Gap at EU level\n\nUnpaid care (% Men) Unpaid care (% Women)\n\nHousework (% Men) Housework (% Women)\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\n2007 2012 2016 2022\n\nb) Cross-country variation\n\nUnpaid care Housework","publish_date":null,"title":"4.2. Attitudes to women’s work and sharing unpaid work","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/employment_and_social_developments_in_europe_esde_2024_report_p56.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/employment_and_social_developments_in_europe_esde_2024_report_p56.pdf"},"request":{"example_id":"chart/employment_and_social_developments_in_europe_esde_2024_report_p56","product_type":"parse","source_file_path":"employment_and_social_developments_in_europe_esde_2024_report_p56.pdf"},"started_at":"2026-11-04T19:45:26.308496"} +{"completed_at":"2026-11-04T19:45:30.124360","latency_in_ms":2652,"output":{"example_id":"chart/employment_and_social_developments_in_europe_esde_2024_report_p58","grounded_pages":[],"layout_pages":[],"markdown":"Chapter 2: Social convergence in the EU: taking stock\n\n61\n\nChart 2.12 Considerable and persistent variation in attitudes across the EU Proportion of population holding attitudes supporting gender equality (%) by country clusters, 1990-2017\n\n_Note:_ Chart shows the following country clusters: west (Austria, Germany, France, the Netherlands); north (Denmark, Estonia, Finland, Sweden); south (Spain, Italy, Portugal); central (Czechia, Hungary, Poland, Slovenia, Slovakia); and east (Bulgaria, Lithuania, Latvia, Romania). Data on ‘children don’t suffer’ missing for Austria in 1999, and data for ‘women don’t prefer home’ missing for Austria in 1999 and Sweden in 1990.\n\n_Source:_ DG EMPL calculations based on European Values Survey data\n\nClick here to download chart.\n\n**Women and people with higher educational attainment (or incomes) are more likely to support gender equality in the labour market.** Over the 1990-2017 period, women in the EU were around 8 pp more likely than men to reject the stereotypes that working motherhood harms children, that women prefer family over paid work and that men have a preferential right to scarce jobs (Chart 2.13). These views were also more common among people with higher educational attainment or (to a lesser extent) higher incomes. For example, those holding tertiary qualifications were around 13-17 pp more likely to report these attitudes than those without upper secondary or tertiary qualifications. At the same time, people who are aged 65+, are religious, or have more traditional voting preferences are more likely to think that working motherhood harms children, and that paid work is valued differently by women and men (Chart 2.13). The differences in attitudes across population groups showed little change over the 1990-2017 period.\n\n1990 1999 2008 2017\n\n**Share household**\n\n**chores**\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\n70\n\n80\n\n90\n\n100\n\n1990 1999 2008 2017\n\n**Children do not suffer**\n\n1990 1999 2008 2017\n\n**Women do not prefer**\n\n**home**\n\n1990 1999 2008 2017\n\n**Scarce jobs accessible**\n\nWest North South Central East","pages":[{"line_numbers":[],"markdown":"Chapter 2: Social convergence in the EU: taking stock\n\n61\n\nChart 2.12 Considerable and persistent variation in attitudes across the EU Proportion of population holding attitudes supporting gender equality (%) by country clusters, 1990-2017\n\n_Note:_ Chart shows the following country clusters: west (Austria, Germany, France, the Netherlands); north (Denmark, Estonia, Finland, Sweden); south (Spain, Italy, Portugal); central (Czechia, Hungary, Poland, Slovenia, Slovakia); and east (Bulgaria, Lithuania, Latvia, Romania). Data on ‘children don’t suffer’ missing for Austria in 1999, and data for ‘women don’t prefer home’ missing for Austria in 1999 and Sweden in 1990.\n\n_Source:_ DG EMPL calculations based on European Values Survey data\n\nClick here to download chart.\n\n**Women and people with higher educational attainment (or incomes) are more likely to support gender equality in the labour market.** Over the 1990-2017 period, women in the EU were around 8 pp more likely than men to reject the stereotypes that working motherhood harms children, that women prefer family over paid work and that men have a preferential right to scarce jobs (Chart 2.13). These views were also more common among people with higher educational attainment or (to a lesser extent) higher incomes. For example, those holding tertiary qualifications were around 13-17 pp more likely to report these attitudes than those without upper secondary or tertiary qualifications. At the same time, people who are aged 65+, are religious, or have more traditional voting preferences are more likely to think that working motherhood harms children, and that paid work is valued differently by women and men (Chart 2.13). The differences in attitudes across population groups showed little change over the 1990-2017 period.\n\n1990 1999 2008 2017\n\n**Share household**\n\n**chores**\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\n70\n\n80\n\n90\n\n100\n\n1990 1999 2008 2017\n\n**Children do not suffer**\n\n1990 1999 2008 2017\n\n**Women do not prefer**\n\n**home**\n\n1990 1999 2008 2017\n\n**Scarce jobs accessible**\n\nWest North South Central East","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Chapter 2: Social convergence in the EU: taking stock\n\n61\n\nChart 2.12 Considerable and persistent variation in attitudes across the EU Proportion of population holding attitudes supporting gender equality (%) by country clusters, 1990-2017\n\n_Note:_ Chart shows the following country clusters: west (Austria, Germany, France, the Netherlands); north (Denmark, Estonia, Finland, Sweden); south (Spain, Italy, Portugal); central (Czechia, Hungary, Poland, Slovenia, Slovakia); and east (Bulgaria, Lithuania, Latvia, Romania). Data on ‘children don’t suffer’ missing for Austria in 1999, and data for ‘women don’t prefer home’ missing for Austria in 1999 and Sweden in 1990.\n\n_Source:_ DG EMPL calculations based on European Values Survey data\n\nClick here to download chart.\n\n**Women and people with higher educational attainment (or incomes) are more likely to support gender equality in the labour market.** Over the 1990-2017 period, women in the EU were around 8 pp more likely than men to reject the stereotypes that working motherhood harms children, that women prefer family over paid work and that men have a preferential right to scarce jobs (Chart 2.13). These views were also more common among people with higher educational attainment or (to a lesser extent) higher incomes. For example, those holding tertiary qualifications were around 13-17 pp more likely to report these attitudes than those without upper secondary or tertiary qualifications. At the same time, people who are aged 65+, are religious, or have more traditional voting preferences are more likely to think that working motherhood harms children, and that paid work is valued differently by women and men (Chart 2.13). The differences in attitudes across population groups showed little change over the 1990-2017 period.\n\n1990 1999 2008 2017\n\n**Share household**\n\n**chores**\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\n70\n\n80\n\n90\n\n100\n\n1990 1999 2008 2017\n\n**Children do not suffer**\n\n1990 1999 2008 2017\n\n**Women do not prefer**\n\n**home**\n\n1990 1999 2008 2017\n\n**Scarce jobs accessible**\n\nWest North South Central East","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/employment_and_social_developments_in_europe_esde_2024_report_p58.pdf"]},"response":{"errors":[],"extract_id":"extract_cc542d767615671d44f216c470a1076d","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Chapter 2: Social convergence in the EU: taking stock\n\n61\n\nChart 2.12 Considerable and persistent variation in attitudes across the EU Proportion of population holding attitudes supporting gender equality (%) by country clusters, 1990-2017\n\n_Note:_ Chart shows the following country clusters: west (Austria, Germany, France, the Netherlands); north (Denmark, Estonia, Finland, Sweden); south (Spain, Italy, Portugal); central (Czechia, Hungary, Poland, Slovenia, Slovakia); and east (Bulgaria, Lithuania, Latvia, Romania). Data on ‘children don’t suffer’ missing for Austria in 1999, and data for ‘women don’t prefer home’ missing for Austria in 1999 and Sweden in 1990.\n\n_Source:_ DG EMPL calculations based on European Values Survey data\n\nClick here to download chart.\n\n**Women and people with higher educational attainment (or incomes) are more likely to support gender equality in the labour market.** Over the 1990-2017 period, women in the EU were around 8 pp more likely than men to reject the stereotypes that working motherhood harms children, that women prefer family over paid work and that men have a preferential right to scarce jobs (Chart 2.13). These views were also more common among people with higher educational attainment or (to a lesser extent) higher incomes. For example, those holding tertiary qualifications were around 13-17 pp more likely to report these attitudes than those without upper secondary or tertiary qualifications. At the same time, people who are aged 65+, are religious, or have more traditional voting preferences are more likely to think that working motherhood harms children, and that paid work is valued differently by women and men (Chart 2.13). The differences in attitudes across population groups showed little change over the 1990-2017 period.\n\n1990 1999 2008 2017\n\n**Share household**\n\n**chores**\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\n70\n\n80\n\n90\n\n100\n\n1990 1999 2008 2017\n\n**Children do not suffer**\n\n1990 1999 2008 2017\n\n**Women do not prefer**\n\n**home**\n\n1990 1999 2008 2017\n\n**Scarce jobs accessible**\n\nWest North South Central East"],"full_content":"Chapter 2: Social convergence in the EU: taking stock\n\n61\n\nChart 2.12 Considerable and persistent variation in attitudes across the EU Proportion of population holding attitudes supporting gender equality (%) by country clusters, 1990-2017\n\n_Note:_ Chart shows the following country clusters: west (Austria, Germany, France, the Netherlands); north (Denmark, Estonia, Finland, Sweden); south (Spain, Italy, Portugal); central (Czechia, Hungary, Poland, Slovenia, Slovakia); and east (Bulgaria, Lithuania, Latvia, Romania). Data on ‘children don’t suffer’ missing for Austria in 1999, and data for ‘women don’t prefer home’ missing for Austria in 1999 and Sweden in 1990.\n\n_Source:_ DG EMPL calculations based on European Values Survey data\n\nClick here to download chart.\n\n**Women and people with higher educational attainment (or incomes) are more likely to support gender equality in the labour market.** Over the 1990-2017 period, women in the EU were around 8 pp more likely than men to reject the stereotypes that working motherhood harms children, that women prefer family over paid work and that men have a preferential right to scarce jobs (Chart 2.13). These views were also more common among people with higher educational attainment or (to a lesser extent) higher incomes. For example, those holding tertiary qualifications were around 13-17 pp more likely to report these attitudes than those without upper secondary or tertiary qualifications. At the same time, people who are aged 65+, are religious, or have more traditional voting preferences are more likely to think that working motherhood harms children, and that paid work is valued differently by women and men (Chart 2.13). The differences in attitudes across population groups showed little change over the 1990-2017 period.\n\n1990 1999 2008 2017\n\n**Share household**\n\n**chores**\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\n70\n\n80\n\n90\n\n100\n\n1990 1999 2008 2017\n\n**Children do not suffer**\n\n1990 1999 2008 2017\n\n**Women do not prefer**\n\n**home**\n\n1990 1999 2008 2017\n\n**Scarce jobs accessible**\n\nWest North South Central East","publish_date":null,"title":"Considerable and persistent variation in attitudes across the EU Children do not suffer Women do not prefer Share household Scarce","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/employment_and_social_developments_in_europe_esde_2024_report_p58.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/employment_and_social_developments_in_europe_esde_2024_report_p58.pdf"},"request":{"example_id":"chart/employment_and_social_developments_in_europe_esde_2024_report_p58","product_type":"parse","source_file_path":"employment_and_social_developments_in_europe_esde_2024_report_p58.pdf"},"started_at":"2026-11-04T19:45:27.471772"} +{"completed_at":"2026-11-04T19:45:30.064386","latency_in_ms":1183,"output":{"example_id":"chart/employment_and_social_developments_in_europe_esde_2024_report_p59","grounded_pages":[],"layout_pages":[],"markdown":"Employment and Social Developments in Europe 2024\n\n62\n\nChart 2.13 Attitudes towards women’s position in the labour market vary between population groups Marginal change in probability of holding a given attitude (pp) by population group, 1990-2017, EU\n\n_Note:_ Probability differences reported against the following comparison groups: gender (men); age (20-39); educational attainment (finished education by 15 years of age); income (lowest income tercile); children (having at least one child); employment status (not in employment); town size (town of 500 000 inhabitants or more); religion (non-religious); political preferences (left-wing). Only statistically significant results (p-value<0.1) reported. Data on children don’t suffer missing for Austria in 1999; data for women don’t prefer home missing for Austria in 1999 and Sweden in 1990.\n\n_Source:_ DG EMPL calculations based on European Values Survey data.\n\nClick here to download chart.\n\n### 5\\. EU INITIATIVES TO SUPPORT SOCIAL CONVERGENCE\n\n**The EU supports social convergence through its cohesion policy.** Since the 1950s, the main vehicle to foster territorial, economic and social convergence among Member States is European cohesion policy, delivered and implemented through the cohesion policy funds, including the European Development Fund, the Cohesion Fund, the European Social Fund+ and Just Transition Fund. It aims to correct imbalances between countries and regions while delivering on the Union's political priorities, especially the green and digital transitions. Although the funds support all Member States and regions, a large share is concentrated on less developed countries and regions to foster their catching-up and reduce economic, social and territorial disparities in the EU, encouraging territorial cooperation and addressing needs of outermost regions. More recent funding instruments, such as the Recovery and Resilience Facility (RRF) which provides EUR 648 billion to Member States (in 2022 prices) for implementing reforms and investments to make their economies and societies more sustainable, resilient and prepared for the green and digital transitions, also contribute to promoting social convergence. ( ¹⁰⁶ )\n\n**Several initiatives launched within the European Pillar of Social Rights aim to reduce disparities among Member States and regions.** Proclaimed in Gothenburg in 2017, the European Pillar of Social Rights outlines 20 key principles necessary to progress towards a strong social Europe. To strengthen the implementation of the Pillar, the European Pillar of Social Rights Action Plan was adopted in 2021, containing more than 70 legislative and non-legislative actions to promote convergence of socio-economic outcomes in the Member States and proposing concrete targets. ( ¹⁰⁷ ) Key policy initiatives supporting social convergence include\n\n( ¹⁰⁶ ) Please see Recovery and Resilience Facility (2020-2024) – mid-term evaluation (europa.eu). ( ¹⁰⁷ ) An employment rate of 78% among the population aged 20-64, a 60% target for adults in training every year, and a reduction of at least 15 million in the population at risk of poverty or social exclusion (including at least five million children).\n\n\\-20.0 -10.0 0.0 10.0 20.0\n\nPolitics right\n\nPolitics centrist\n\nReligious\n\nTown size <20,000\n\nEmployed\n\nNo children\n\nHigh income\n\nMedium income\n\nHigher education\n\nSecondary education\n\nAged 65+\n\nAged 40 to 64\n\nWomen\n\nChildren don't suffer Women don't prefer home\n\nScarce jobs accessible Share household chores","pages":[{"line_numbers":[],"markdown":"Employment and Social Developments in Europe 2024\n\n62\n\nChart 2.13 Attitudes towards women’s position in the labour market vary between population groups Marginal change in probability of holding a given attitude (pp) by population group, 1990-2017, EU\n\n_Note:_ Probability differences reported against the following comparison groups: gender (men); age (20-39); educational attainment (finished education by 15 years of age); income (lowest income tercile); children (having at least one child); employment status (not in employment); town size (town of 500 000 inhabitants or more); religion (non-religious); political preferences (left-wing). Only statistically significant results (p-value<0.1) reported. Data on children don’t suffer missing for Austria in 1999; data for women don’t prefer home missing for Austria in 1999 and Sweden in 1990.\n\n_Source:_ DG EMPL calculations based on European Values Survey data.\n\nClick here to download chart.\n\n### 5\\. EU INITIATIVES TO SUPPORT SOCIAL CONVERGENCE\n\n**The EU supports social convergence through its cohesion policy.** Since the 1950s, the main vehicle to foster territorial, economic and social convergence among Member States is European cohesion policy, delivered and implemented through the cohesion policy funds, including the European Development Fund, the Cohesion Fund, the European Social Fund+ and Just Transition Fund. It aims to correct imbalances between countries and regions while delivering on the Union's political priorities, especially the green and digital transitions. Although the funds support all Member States and regions, a large share is concentrated on less developed countries and regions to foster their catching-up and reduce economic, social and territorial disparities in the EU, encouraging territorial cooperation and addressing needs of outermost regions. More recent funding instruments, such as the Recovery and Resilience Facility (RRF) which provides EUR 648 billion to Member States (in 2022 prices) for implementing reforms and investments to make their economies and societies more sustainable, resilient and prepared for the green and digital transitions, also contribute to promoting social convergence. ( ¹⁰⁶ )\n\n**Several initiatives launched within the European Pillar of Social Rights aim to reduce disparities among Member States and regions.** Proclaimed in Gothenburg in 2017, the European Pillar of Social Rights outlines 20 key principles necessary to progress towards a strong social Europe. To strengthen the implementation of the Pillar, the European Pillar of Social Rights Action Plan was adopted in 2021, containing more than 70 legislative and non-legislative actions to promote convergence of socio-economic outcomes in the Member States and proposing concrete targets. ( ¹⁰⁷ ) Key policy initiatives supporting social convergence include\n\n( ¹⁰⁶ ) Please see Recovery and Resilience Facility (2020-2024) – mid-term evaluation (europa.eu). ( ¹⁰⁷ ) An employment rate of 78% among the population aged 20-64, a 60% target for adults in training every year, and a reduction of at least 15 million in the population at risk of poverty or social exclusion (including at least five million children).\n\n\\-20.0 -10.0 0.0 10.0 20.0\n\nPolitics right\n\nPolitics centrist\n\nReligious\n\nTown size <20,000\n\nEmployed\n\nNo children\n\nHigh income\n\nMedium income\n\nHigher education\n\nSecondary education\n\nAged 65+\n\nAged 40 to 64\n\nWomen\n\nChildren don't suffer Women don't prefer home\n\nScarce jobs accessible Share household chores","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Employment and Social Developments in Europe 2024\n\n62\n\nChart 2.13 Attitudes towards women’s position in the labour market vary between population groups Marginal change in probability of holding a given attitude (pp) by population group, 1990-2017, EU\n\n_Note:_ Probability differences reported against the following comparison groups: gender (men); age (20-39); educational attainment (finished education by 15 years of age); income (lowest income tercile); children (having at least one child); employment status (not in employment); town size (town of 500 000 inhabitants or more); religion (non-religious); political preferences (left-wing). Only statistically significant results (p-value<0.1) reported. Data on children don’t suffer missing for Austria in 1999; data for women don’t prefer home missing for Austria in 1999 and Sweden in 1990.\n\n_Source:_ DG EMPL calculations based on European Values Survey data.\n\nClick here to download chart.\n\n### 5\\. EU INITIATIVES TO SUPPORT SOCIAL CONVERGENCE\n\n**The EU supports social convergence through its cohesion policy.** Since the 1950s, the main vehicle to foster territorial, economic and social convergence among Member States is European cohesion policy, delivered and implemented through the cohesion policy funds, including the European Development Fund, the Cohesion Fund, the European Social Fund+ and Just Transition Fund. It aims to correct imbalances between countries and regions while delivering on the Union's political priorities, especially the green and digital transitions. Although the funds support all Member States and regions, a large share is concentrated on less developed countries and regions to foster their catching-up and reduce economic, social and territorial disparities in the EU, encouraging territorial cooperation and addressing needs of outermost regions. More recent funding instruments, such as the Recovery and Resilience Facility (RRF) which provides EUR 648 billion to Member States (in 2022 prices) for implementing reforms and investments to make their economies and societies more sustainable, resilient and prepared for the green and digital transitions, also contribute to promoting social convergence. ( ¹⁰⁶ )\n\n**Several initiatives launched within the European Pillar of Social Rights aim to reduce disparities among Member States and regions.** Proclaimed in Gothenburg in 2017, the European Pillar of Social Rights outlines 20 key principles necessary to progress towards a strong social Europe. To strengthen the implementation of the Pillar, the European Pillar of Social Rights Action Plan was adopted in 2021, containing more than 70 legislative and non-legislative actions to promote convergence of socio-economic outcomes in the Member States and proposing concrete targets. ( ¹⁰⁷ ) Key policy initiatives supporting social convergence include\n\n( ¹⁰⁶ ) Please see Recovery and Resilience Facility (2020-2024) – mid-term evaluation (europa.eu). ( ¹⁰⁷ ) An employment rate of 78% among the population aged 20-64, a 60% target for adults in training every year, and a reduction of at least 15 million in the population at risk of poverty or social exclusion (including at least five million children).\n\n\\-20.0 -10.0 0.0 10.0 20.0\n\nPolitics right\n\nPolitics centrist\n\nReligious\n\nTown size <20,000\n\nEmployed\n\nNo children\n\nHigh income\n\nMedium income\n\nHigher education\n\nSecondary education\n\nAged 65+\n\nAged 40 to 64\n\nWomen\n\nChildren don't suffer Women don't prefer home\n\nScarce jobs accessible Share household chores","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/employment_and_social_developments_in_europe_esde_2024_report_p59.pdf"]},"response":{"errors":[],"extract_id":"extract_dc70edb6082c649f1fac87ea26324502","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Employment and Social Developments in Europe 2024\n\n62\n\nChart 2.13 Attitudes towards women’s position in the labour market vary between population groups Marginal change in probability of holding a given attitude (pp) by population group, 1990-2017, EU\n\n_Note:_ Probability differences reported against the following comparison groups: gender (men); age (20-39); educational attainment (finished education by 15 years of age); income (lowest income tercile); children (having at least one child); employment status (not in employment); town size (town of 500 000 inhabitants or more); religion (non-religious); political preferences (left-wing). Only statistically significant results (p-value<0.1) reported. Data on children don’t suffer missing for Austria in 1999; data for women don’t prefer home missing for Austria in 1999 and Sweden in 1990.\n\n_Source:_ DG EMPL calculations based on European Values Survey data.\n\nClick here to download chart.\n\n### 5\\. EU INITIATIVES TO SUPPORT SOCIAL CONVERGENCE\n\n**The EU supports social convergence through its cohesion policy.** Since the 1950s, the main vehicle to foster territorial, economic and social convergence among Member States is European cohesion policy, delivered and implemented through the cohesion policy funds, including the European Development Fund, the Cohesion Fund, the European Social Fund+ and Just Transition Fund. It aims to correct imbalances between countries and regions while delivering on the Union's political priorities, especially the green and digital transitions. Although the funds support all Member States and regions, a large share is concentrated on less developed countries and regions to foster their catching-up and reduce economic, social and territorial disparities in the EU, encouraging territorial cooperation and addressing needs of outermost regions. More recent funding instruments, such as the Recovery and Resilience Facility (RRF) which provides EUR 648 billion to Member States (in 2022 prices) for implementing reforms and investments to make their economies and societies more sustainable, resilient and prepared for the green and digital transitions, also contribute to promoting social convergence. ( ¹⁰⁶ )\n\n**Several initiatives launched within the European Pillar of Social Rights aim to reduce disparities among Member States and regions.** Proclaimed in Gothenburg in 2017, the European Pillar of Social Rights outlines 20 key principles necessary to progress towards a strong social Europe. To strengthen the implementation of the Pillar, the European Pillar of Social Rights Action Plan was adopted in 2021, containing more than 70 legislative and non-legislative actions to promote convergence of socio-economic outcomes in the Member States and proposing concrete targets. ( ¹⁰⁷ ) Key policy initiatives supporting social convergence include\n\n( ¹⁰⁶ ) Please see Recovery and Resilience Facility (2020-2024) – mid-term evaluation (europa.eu). ( ¹⁰⁷ ) An employment rate of 78% among the population aged 20-64, a 60% target for adults in training every year, and a reduction of at least 15 million in the population at risk of poverty or social exclusion (including at least five million children).\n\n\\-20.0 -10.0 0.0 10.0 20.0\n\nPolitics right\n\nPolitics centrist\n\nReligious\n\nTown size <20,000\n\nEmployed\n\nNo children\n\nHigh income\n\nMedium income\n\nHigher education\n\nSecondary education\n\nAged 65+\n\nAged 40 to 64\n\nWomen\n\nChildren don't suffer Women don't prefer home\n\nScarce jobs accessible Share household chores"],"full_content":"Employment and Social Developments in Europe 2024\n\n62\n\nChart 2.13 Attitudes towards women’s position in the labour market vary between population groups Marginal change in probability of holding a given attitude (pp) by population group, 1990-2017, EU\n\n_Note:_ Probability differences reported against the following comparison groups: gender (men); age (20-39); educational attainment (finished education by 15 years of age); income (lowest income tercile); children (having at least one child); employment status (not in employment); town size (town of 500 000 inhabitants or more); religion (non-religious); political preferences (left-wing). Only statistically significant results (p-value<0.1) reported. Data on children don’t suffer missing for Austria in 1999; data for women don’t prefer home missing for Austria in 1999 and Sweden in 1990.\n\n_Source:_ DG EMPL calculations based on European Values Survey data.\n\nClick here to download chart.\n\n### 5\\. EU INITIATIVES TO SUPPORT SOCIAL CONVERGENCE\n\n**The EU supports social convergence through its cohesion policy.** Since the 1950s, the main vehicle to foster territorial, economic and social convergence among Member States is European cohesion policy, delivered and implemented through the cohesion policy funds, including the European Development Fund, the Cohesion Fund, the European Social Fund+ and Just Transition Fund. It aims to correct imbalances between countries and regions while delivering on the Union's political priorities, especially the green and digital transitions. Although the funds support all Member States and regions, a large share is concentrated on less developed countries and regions to foster their catching-up and reduce economic, social and territorial disparities in the EU, encouraging territorial cooperation and addressing needs of outermost regions. More recent funding instruments, such as the Recovery and Resilience Facility (RRF) which provides EUR 648 billion to Member States (in 2022 prices) for implementing reforms and investments to make their economies and societies more sustainable, resilient and prepared for the green and digital transitions, also contribute to promoting social convergence. ( ¹⁰⁶ )\n\n**Several initiatives launched within the European Pillar of Social Rights aim to reduce disparities among Member States and regions.** Proclaimed in Gothenburg in 2017, the European Pillar of Social Rights outlines 20 key principles necessary to progress towards a strong social Europe. To strengthen the implementation of the Pillar, the European Pillar of Social Rights Action Plan was adopted in 2021, containing more than 70 legislative and non-legislative actions to promote convergence of socio-economic outcomes in the Member States and proposing concrete targets. ( ¹⁰⁷ ) Key policy initiatives supporting social convergence include\n\n( ¹⁰⁶ ) Please see Recovery and Resilience Facility (2020-2024) – mid-term evaluation (europa.eu). ( ¹⁰⁷ ) An employment rate of 78% among the population aged 20-64, a 60% target for adults in training every year, and a reduction of at least 15 million in the population at risk of poverty or social exclusion (including at least five million children).\n\n\\-20.0 -10.0 0.0 10.0 20.0\n\nPolitics right\n\nPolitics centrist\n\nReligious\n\nTown size <20,000\n\nEmployed\n\nNo children\n\nHigh income\n\nMedium income\n\nHigher education\n\nSecondary education\n\nAged 65+\n\nAged 40 to 64\n\nWomen\n\nChildren don't suffer Women don't prefer home\n\nScarce jobs accessible Share household chores","publish_date":null,"title":"5. EU","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/employment_and_social_developments_in_europe_esde_2024_report_p59.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/employment_and_social_developments_in_europe_esde_2024_report_p59.pdf"},"request":{"example_id":"chart/employment_and_social_developments_in_europe_esde_2024_report_p59","product_type":"parse","source_file_path":"employment_and_social_developments_in_europe_esde_2024_report_p59.pdf"},"started_at":"2026-11-04T19:45:28.881155"} +{"completed_at":"2026-11-04T19:45:33.629244","latency_in_ms":3953,"output":{"example_id":"chart/employment_and_social_developments_in_europe_esde_2024_report_p65","grounded_pages":[],"layout_pages":[],"markdown":"Employment and Social Developments in Europe 2024\n\n68\n\nCoefficient of variation adjusts standard deviation by developments in outcome mean over time and is therefore insensitive to level changes (e.g., a 50% increase in GDP per capita in all countries considered would not change the value of coefficient of variation). It measures variation relative to the mean rather than in absolute terms.\n\n### Catching-up of worse performing Member States or regions (Beta convergence)\n\nBeta convergence assesses whether countries or regions that start with worse outcomes improve faster over time than those starting with better outcomes. Where this is the case, Member States or regions are considered converging. Conversely, if countries or regions with better initial outcomes improve faster than others over time, this signifies divergence.\n\nBeta convergence is estimated via a logarithmic regression model, which regresses the growth of an indicator over a certain period of time on its initial value. The specification of this model is as follows,\n\nln ∆𝑦𝑦𝑖𝑖,𝑡𝑡 = 𝛼𝛼 + 𝛽𝛽 ln 𝑦𝑦𝑖𝑖,𝑡𝑡−1 + 𝜀𝜀𝑖𝑖,𝑡𝑡\n\nwhere _yi,t_ is the indicator value in country _i_ at a time _t_ , _Δyi,t_ is the growth rate of this indicator at time _t_ , _α_ and _β_ are coefficients to be estimated and _εi,t_ is the error term.\n\nBeta convergence exists if the estimated regression coefficient _β_ is statistically significant and negative, i.e. if countries that performed worse at the start of the period of interest improved faster than those that started in better positions. In this report, whenever beta convergence is analysed, it is referred to as catching-up.\n\n### Direction of convergence\n\nThe report also considers the overall direction of convergence patterns: where convergence is accompanied by improvements in mean EU outcomes (increase in the mean for a maximisation indicator, such as employment rate, or decrease for a minimisation indicator, such as AROPE), this is defined as upward convergence. Where mean EU outcome deteriorates (decrease in the mean for a maximisation indicator and increase for a minimisation indicator), this is considered downward convergence. From an EU policy-making perspective, it is important to achieve upward convergence, i.e., improvement in average outcomes at EU level accompanied by reduced geographical disparities:\n\n### A2.2. REGIONAL CONVERGENCE\n\nChart A2.1 Convergence patterns of economic gains differed across NUTS2 regions GDP per capita in EUR and their cross-regional variation (measured by standard deviation and coefficient of variation, hence adjusted by average EU GDP), EU-NUTS2\n\n_Note:_ Standard deviation is a measure of cross-regional variation, the higher the standard deviation, the higher the cross-regional variation. The EU average levels are unweighted values representing an average NUTS2 region\n\n_Source:_ DG EMPL calculations based on Eurostat dataset nama\\_10r\\_2gdp\n\nClick here to download chart.\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\n70\n\n80\n\n0\n\n5000\n\n10000\n\n15000\n\n20000\n\n25000\n\n30000\n\n35000\n\n40000\n\n2007 2009 2011 2013 2015 2017 2019 2021\n\nAverage (left axis) Standard deviation (left axis)\n\nCoeff of variation (right axis)","pages":[{"line_numbers":[],"markdown":"Employment and Social Developments in Europe 2024\n\n68\n\nCoefficient of variation adjusts standard deviation by developments in outcome mean over time and is therefore insensitive to level changes (e.g., a 50% increase in GDP per capita in all countries considered would not change the value of coefficient of variation). It measures variation relative to the mean rather than in absolute terms.\n\n### Catching-up of worse performing Member States or regions (Beta convergence)\n\nBeta convergence assesses whether countries or regions that start with worse outcomes improve faster over time than those starting with better outcomes. Where this is the case, Member States or regions are considered converging. Conversely, if countries or regions with better initial outcomes improve faster than others over time, this signifies divergence.\n\nBeta convergence is estimated via a logarithmic regression model, which regresses the growth of an indicator over a certain period of time on its initial value. The specification of this model is as follows,\n\nln ∆𝑦𝑦𝑖𝑖,𝑡𝑡 = 𝛼𝛼 + 𝛽𝛽 ln 𝑦𝑦𝑖𝑖,𝑡𝑡−1 + 𝜀𝜀𝑖𝑖,𝑡𝑡\n\nwhere _yi,t_ is the indicator value in country _i_ at a time _t_ , _Δyi,t_ is the growth rate of this indicator at time _t_ , _α_ and _β_ are coefficients to be estimated and _εi,t_ is the error term.\n\nBeta convergence exists if the estimated regression coefficient _β_ is statistically significant and negative, i.e. if countries that performed worse at the start of the period of interest improved faster than those that started in better positions. In this report, whenever beta convergence is analysed, it is referred to as catching-up.\n\n### Direction of convergence\n\nThe report also considers the overall direction of convergence patterns: where convergence is accompanied by improvements in mean EU outcomes (increase in the mean for a maximisation indicator, such as employment rate, or decrease for a minimisation indicator, such as AROPE), this is defined as upward convergence. Where mean EU outcome deteriorates (decrease in the mean for a maximisation indicator and increase for a minimisation indicator), this is considered downward convergence. From an EU policy-making perspective, it is important to achieve upward convergence, i.e., improvement in average outcomes at EU level accompanied by reduced geographical disparities:\n\n### A2.2. REGIONAL CONVERGENCE\n\nChart A2.1 Convergence patterns of economic gains differed across NUTS2 regions GDP per capita in EUR and their cross-regional variation (measured by standard deviation and coefficient of variation, hence adjusted by average EU GDP), EU-NUTS2\n\n_Note:_ Standard deviation is a measure of cross-regional variation, the higher the standard deviation, the higher the cross-regional variation. The EU average levels are unweighted values representing an average NUTS2 region\n\n_Source:_ DG EMPL calculations based on Eurostat dataset nama\\_10r\\_2gdp\n\nClick here to download chart.\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\n70\n\n80\n\n0\n\n5000\n\n10000\n\n15000\n\n20000\n\n25000\n\n30000\n\n35000\n\n40000\n\n2007 2009 2011 2013 2015 2017 2019 2021\n\nAverage (left axis) Standard deviation (left axis)\n\nCoeff of variation (right axis)","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Employment and Social Developments in Europe 2024\n\n68\n\nCoefficient of variation adjusts standard deviation by developments in outcome mean over time and is therefore insensitive to level changes (e.g., a 50% increase in GDP per capita in all countries considered would not change the value of coefficient of variation). It measures variation relative to the mean rather than in absolute terms.\n\n### Catching-up of worse performing Member States or regions (Beta convergence)\n\nBeta convergence assesses whether countries or regions that start with worse outcomes improve faster over time than those starting with better outcomes. Where this is the case, Member States or regions are considered converging. Conversely, if countries or regions with better initial outcomes improve faster than others over time, this signifies divergence.\n\nBeta convergence is estimated via a logarithmic regression model, which regresses the growth of an indicator over a certain period of time on its initial value. The specification of this model is as follows,\n\nln ∆𝑦𝑦𝑖𝑖,𝑡𝑡 = 𝛼𝛼 + 𝛽𝛽 ln 𝑦𝑦𝑖𝑖,𝑡𝑡−1 + 𝜀𝜀𝑖𝑖,𝑡𝑡\n\nwhere _yi,t_ is the indicator value in country _i_ at a time _t_ , _Δyi,t_ is the growth rate of this indicator at time _t_ , _α_ and _β_ are coefficients to be estimated and _εi,t_ is the error term.\n\nBeta convergence exists if the estimated regression coefficient _β_ is statistically significant and negative, i.e. if countries that performed worse at the start of the period of interest improved faster than those that started in better positions. In this report, whenever beta convergence is analysed, it is referred to as catching-up.\n\n### Direction of convergence\n\nThe report also considers the overall direction of convergence patterns: where convergence is accompanied by improvements in mean EU outcomes (increase in the mean for a maximisation indicator, such as employment rate, or decrease for a minimisation indicator, such as AROPE), this is defined as upward convergence. Where mean EU outcome deteriorates (decrease in the mean for a maximisation indicator and increase for a minimisation indicator), this is considered downward convergence. From an EU policy-making perspective, it is important to achieve upward convergence, i.e., improvement in average outcomes at EU level accompanied by reduced geographical disparities:\n\n### A2.2. REGIONAL CONVERGENCE\n\nChart A2.1 Convergence patterns of economic gains differed across NUTS2 regions GDP per capita in EUR and their cross-regional variation (measured by standard deviation and coefficient of variation, hence adjusted by average EU GDP), EU-NUTS2\n\n_Note:_ Standard deviation is a measure of cross-regional variation, the higher the standard deviation, the higher the cross-regional variation. The EU average levels are unweighted values representing an average NUTS2 region\n\n_Source:_ DG EMPL calculations based on Eurostat dataset nama\\_10r\\_2gdp\n\nClick here to download chart.\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\n70\n\n80\n\n0\n\n5000\n\n10000\n\n15000\n\n20000\n\n25000\n\n30000\n\n35000\n\n40000\n\n2007 2009 2011 2013 2015 2017 2019 2021\n\nAverage (left axis) Standard deviation (left axis)\n\nCoeff of variation (right axis)","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/employment_and_social_developments_in_europe_esde_2024_report_p65.pdf"]},"response":{"errors":[],"extract_id":"extract_88ffbc0a88b2a0196b1f035b608f96ba","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Employment and Social Developments in Europe 2024\n\n68\n\nCoefficient of variation adjusts standard deviation by developments in outcome mean over time and is therefore insensitive to level changes (e.g., a 50% increase in GDP per capita in all countries considered would not change the value of coefficient of variation). It measures variation relative to the mean rather than in absolute terms.\n\n### Catching-up of worse performing Member States or regions (Beta convergence)\n\nBeta convergence assesses whether countries or regions that start with worse outcomes improve faster over time than those starting with better outcomes. Where this is the case, Member States or regions are considered converging. Conversely, if countries or regions with better initial outcomes improve faster than others over time, this signifies divergence.\n\nBeta convergence is estimated via a logarithmic regression model, which regresses the growth of an indicator over a certain period of time on its initial value. The specification of this model is as follows,\n\nln ∆𝑦𝑦𝑖𝑖,𝑡𝑡 = 𝛼𝛼 + 𝛽𝛽 ln 𝑦𝑦𝑖𝑖,𝑡𝑡−1 + 𝜀𝜀𝑖𝑖,𝑡𝑡\n\nwhere _yi,t_ is the indicator value in country _i_ at a time _t_ , _Δyi,t_ is the growth rate of this indicator at time _t_ , _α_ and _β_ are coefficients to be estimated and _εi,t_ is the error term.\n\nBeta convergence exists if the estimated regression coefficient _β_ is statistically significant and negative, i.e. if countries that performed worse at the start of the period of interest improved faster than those that started in better positions. In this report, whenever beta convergence is analysed, it is referred to as catching-up.\n\n### Direction of convergence\n\nThe report also considers the overall direction of convergence patterns: where convergence is accompanied by improvements in mean EU outcomes (increase in the mean for a maximisation indicator, such as employment rate, or decrease for a minimisation indicator, such as AROPE), this is defined as upward convergence. Where mean EU outcome deteriorates (decrease in the mean for a maximisation indicator and increase for a minimisation indicator), this is considered downward convergence. From an EU policy-making perspective, it is important to achieve upward convergence, i.e., improvement in average outcomes at EU level accompanied by reduced geographical disparities:\n\n### A2.2. REGIONAL CONVERGENCE\n\nChart A2.1 Convergence patterns of economic gains differed across NUTS2 regions GDP per capita in EUR and their cross-regional variation (measured by standard deviation and coefficient of variation, hence adjusted by average EU GDP), EU-NUTS2\n\n_Note:_ Standard deviation is a measure of cross-regional variation, the higher the standard deviation, the higher the cross-regional variation. The EU average levels are unweighted values representing an average NUTS2 region\n\n_Source:_ DG EMPL calculations based on Eurostat dataset nama\\_10r\\_2gdp\n\nClick here to download chart.\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\n70\n\n80\n\n0\n\n5000\n\n10000\n\n15000\n\n20000\n\n25000\n\n30000\n\n35000\n\n40000\n\n2007 2009 2011 2013 2015 2017 2019 2021\n\nAverage (left axis) Standard deviation (left axis)\n\nCoeff of variation (right axis)"],"full_content":"Employment and Social Developments in Europe 2024\n\n68\n\nCoefficient of variation adjusts standard deviation by developments in outcome mean over time and is therefore insensitive to level changes (e.g., a 50% increase in GDP per capita in all countries considered would not change the value of coefficient of variation). It measures variation relative to the mean rather than in absolute terms.\n\n### Catching-up of worse performing Member States or regions (Beta convergence)\n\nBeta convergence assesses whether countries or regions that start with worse outcomes improve faster over time than those starting with better outcomes. Where this is the case, Member States or regions are considered converging. Conversely, if countries or regions with better initial outcomes improve faster than others over time, this signifies divergence.\n\nBeta convergence is estimated via a logarithmic regression model, which regresses the growth of an indicator over a certain period of time on its initial value. The specification of this model is as follows,\n\nln ∆𝑦𝑦𝑖𝑖,𝑡𝑡 = 𝛼𝛼 + 𝛽𝛽 ln 𝑦𝑦𝑖𝑖,𝑡𝑡−1 + 𝜀𝜀𝑖𝑖,𝑡𝑡\n\nwhere _yi,t_ is the indicator value in country _i_ at a time _t_ , _Δyi,t_ is the growth rate of this indicator at time _t_ , _α_ and _β_ are coefficients to be estimated and _εi,t_ is the error term.\n\nBeta convergence exists if the estimated regression coefficient _β_ is statistically significant and negative, i.e. if countries that performed worse at the start of the period of interest improved faster than those that started in better positions. In this report, whenever beta convergence is analysed, it is referred to as catching-up.\n\n### Direction of convergence\n\nThe report also considers the overall direction of convergence patterns: where convergence is accompanied by improvements in mean EU outcomes (increase in the mean for a maximisation indicator, such as employment rate, or decrease for a minimisation indicator, such as AROPE), this is defined as upward convergence. Where mean EU outcome deteriorates (decrease in the mean for a maximisation indicator and increase for a minimisation indicator), this is considered downward convergence. From an EU policy-making perspective, it is important to achieve upward convergence, i.e., improvement in average outcomes at EU level accompanied by reduced geographical disparities:\n\n### A2.2. REGIONAL CONVERGENCE\n\nChart A2.1 Convergence patterns of economic gains differed across NUTS2 regions GDP per capita in EUR and their cross-regional variation (measured by standard deviation and coefficient of variation, hence adjusted by average EU GDP), EU-NUTS2\n\n_Note:_ Standard deviation is a measure of cross-regional variation, the higher the standard deviation, the higher the cross-regional variation. The EU average levels are unweighted values representing an average NUTS2 region\n\n_Source:_ DG EMPL calculations based on Eurostat dataset nama\\_10r\\_2gdp\n\nClick here to download chart.\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\n70\n\n80\n\n0\n\n5000\n\n10000\n\n15000\n\n20000\n\n25000\n\n30000\n\n35000\n\n40000\n\n2007 2009 2011 2013 2015 2017 2019 2021\n\nAverage (left axis) Standard deviation (left axis)\n\nCoeff of variation (right axis)","publish_date":null,"title":"A2.2. R","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/employment_and_social_developments_in_europe_esde_2024_report_p65.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/employment_and_social_developments_in_europe_esde_2024_report_p65.pdf"},"request":{"example_id":"chart/employment_and_social_developments_in_europe_esde_2024_report_p65","product_type":"parse","source_file_path":"employment_and_social_developments_in_europe_esde_2024_report_p65.pdf"},"started_at":"2026-11-04T19:45:29.675625"} +{"completed_at":"2026-11-04T19:45:31.875348","latency_in_ms":1042,"output":{"example_id":"chart/employment_and_social_developments_in_europe_esde_2024_report_p73","grounded_pages":[],"layout_pages":[],"markdown":"Chapter 3: The role of social investment\n\n77\n\n**A large majority of preschool age children participate in ECEC in the EU, with participation rates converging across Member States. (** **140** **)** The share of children aged three to compulsory school age (CSA) participating in ECEC broadly stagnated, going from about 87% in 2014 to 89% in 2023 (Chart 3.2). The COVID- 19 pandemic resulted in a temporary decline in participation and a spike in cross-country differences in 2020. In 2023, national participation rates still varied considerably. For example, participation was above the Barcelona target of 96% ( ¹⁴¹ ) in seven Member States (Belgium, Denmark, Spain, France, Hungary, the Netherlands, Sweden), but was less than 80% in three countries (Poland, Romania, Slovakia). Cross-country differences declined between 2014 and 2023, as countries with lower participation rates caught up with those with higher participation. ( ¹⁴² )\n\n**While the average ECEC participation rate of the youngest children (0-2 years of age) rose over the last decade, disparities across Member States also increased.** Over the last ten years, the share of children participating rose from about 28% in 2014 to 37% in 2023, with a short lapse during the 2020 closures due to the COVID-19 pandemic and an increase by 1.5 pp in 2023 compared to the previous year (Chart 3.2). To\n\n( ¹⁴⁰ ) (Eurofound, a). ( ¹⁴¹ ) Note that the Barcelona target for this age group (3+) is based on UOE data (educ\\_uoe\\_enra21) and not on data from EU-SILC survey. ( ¹⁴² ) Standard deviation dropped from 5.3 to 4.7. Beta coefficient is negative (-0.07) and statistically significant at 1% significance level.\n\nBox 3.2: Future investment needs in Early Childhood Education and Care\n\n**Further investment is needed to increase children’s participation in ECEC to the Barcelona target levels at EU level** , notably for the under-3s. Such investment would help to address several barriers to children’s access to ECEC, including the lack of available places and high access costs. For example, by creating new ECEC places, training, hiring and paying new staff, or refurbishing existing/building new facilities, while ensuring accessibility and affordability.\n\n**Additional yearly spending on ECEC needed to reach the EU Barcelona Targets by 2030 would amount to 0.085% of EU GDP** (more than EUR 11 billion) according to recent estimates. ( ¹ ) This amount varies by Member State, ranging from 0% (in Member States that have already reached the target) to nearly 0.3% in Greece and around 0.2% in Germany, Poland and Romania (Chart 1).\n\nChart 1 Additional investment needs in ECEC vary by Member State Yearly additional investment in ECEC needed to meet the 2030 Barcelona targets (% of GDP), by Barcelona targets\n\n_Source:_ DG EMPL calculations based on Eurostat data.\n\n( ¹ ) The additional investment need is estimated as follows. First, based on current participation rates (Eurostat ilc\\_caindformal\n\ndataset), the number of additional children who should be enrolled to meet the Barcelona targets is estimated. This number is then multiplied by the yearly average public expenditure on ECEC per participating child (Eurostat educ\\_uoe\\_fine09 dataset) to obtain the additional yearly spending needs to reach the Barcelona targets. It is assumed that the level of spending per participating child remains constant over the years.\n\n0\\.0%\n\n0\\.1%\n\n0\\.2%\n\n0\\.3%\n\nEU BE BG CZ DK DE EE IE EL ES FR HR IT CY LV LT LU HU MT NL AT PL PT RO SI SK FI SE\n\nAdditional spending needed to reach the 2030 Barcelona target for children from 3 to CSA\n\nAdditional spending needed to reach the 2030 country specific Barcelona target for children below 3","pages":[{"line_numbers":[],"markdown":"Chapter 3: The role of social investment\n\n77\n\n**A large majority of preschool age children participate in ECEC in the EU, with participation rates converging across Member States. (** **140** **)** The share of children aged three to compulsory school age (CSA) participating in ECEC broadly stagnated, going from about 87% in 2014 to 89% in 2023 (Chart 3.2). The COVID- 19 pandemic resulted in a temporary decline in participation and a spike in cross-country differences in 2020. In 2023, national participation rates still varied considerably. For example, participation was above the Barcelona target of 96% ( ¹⁴¹ ) in seven Member States (Belgium, Denmark, Spain, France, Hungary, the Netherlands, Sweden), but was less than 80% in three countries (Poland, Romania, Slovakia). Cross-country differences declined between 2014 and 2023, as countries with lower participation rates caught up with those with higher participation. ( ¹⁴² )\n\n**While the average ECEC participation rate of the youngest children (0-2 years of age) rose over the last decade, disparities across Member States also increased.** Over the last ten years, the share of children participating rose from about 28% in 2014 to 37% in 2023, with a short lapse during the 2020 closures due to the COVID-19 pandemic and an increase by 1.5 pp in 2023 compared to the previous year (Chart 3.2). To\n\n( ¹⁴⁰ ) (Eurofound, a). ( ¹⁴¹ ) Note that the Barcelona target for this age group (3+) is based on UOE data (educ\\_uoe\\_enra21) and not on data from EU-SILC survey. ( ¹⁴² ) Standard deviation dropped from 5.3 to 4.7. Beta coefficient is negative (-0.07) and statistically significant at 1% significance level.\n\nBox 3.2: Future investment needs in Early Childhood Education and Care\n\n**Further investment is needed to increase children’s participation in ECEC to the Barcelona target levels at EU level** , notably for the under-3s. Such investment would help to address several barriers to children’s access to ECEC, including the lack of available places and high access costs. For example, by creating new ECEC places, training, hiring and paying new staff, or refurbishing existing/building new facilities, while ensuring accessibility and affordability.\n\n**Additional yearly spending on ECEC needed to reach the EU Barcelona Targets by 2030 would amount to 0.085% of EU GDP** (more than EUR 11 billion) according to recent estimates. ( ¹ ) This amount varies by Member State, ranging from 0% (in Member States that have already reached the target) to nearly 0.3% in Greece and around 0.2% in Germany, Poland and Romania (Chart 1).\n\nChart 1 Additional investment needs in ECEC vary by Member State Yearly additional investment in ECEC needed to meet the 2030 Barcelona targets (% of GDP), by Barcelona targets\n\n_Source:_ DG EMPL calculations based on Eurostat data.\n\n( ¹ ) The additional investment need is estimated as follows. First, based on current participation rates (Eurostat ilc\\_caindformal\n\ndataset), the number of additional children who should be enrolled to meet the Barcelona targets is estimated. This number is then multiplied by the yearly average public expenditure on ECEC per participating child (Eurostat educ\\_uoe\\_fine09 dataset) to obtain the additional yearly spending needs to reach the Barcelona targets. It is assumed that the level of spending per participating child remains constant over the years.\n\n0\\.0%\n\n0\\.1%\n\n0\\.2%\n\n0\\.3%\n\nEU BE BG CZ DK DE EE IE EL ES FR HR IT CY LV LT LU HU MT NL AT PL PT RO SI SK FI SE\n\nAdditional spending needed to reach the 2030 Barcelona target for children from 3 to CSA\n\nAdditional spending needed to reach the 2030 country specific Barcelona target for children below 3","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Chapter 3: The role of social investment\n\n77\n\n**A large majority of preschool age children participate in ECEC in the EU, with participation rates converging across Member States. (** **140** **)** The share of children aged three to compulsory school age (CSA) participating in ECEC broadly stagnated, going from about 87% in 2014 to 89% in 2023 (Chart 3.2). The COVID- 19 pandemic resulted in a temporary decline in participation and a spike in cross-country differences in 2020. In 2023, national participation rates still varied considerably. For example, participation was above the Barcelona target of 96% ( ¹⁴¹ ) in seven Member States (Belgium, Denmark, Spain, France, Hungary, the Netherlands, Sweden), but was less than 80% in three countries (Poland, Romania, Slovakia). Cross-country differences declined between 2014 and 2023, as countries with lower participation rates caught up with those with higher participation. ( ¹⁴² )\n\n**While the average ECEC participation rate of the youngest children (0-2 years of age) rose over the last decade, disparities across Member States also increased.** Over the last ten years, the share of children participating rose from about 28% in 2014 to 37% in 2023, with a short lapse during the 2020 closures due to the COVID-19 pandemic and an increase by 1.5 pp in 2023 compared to the previous year (Chart 3.2). To\n\n( ¹⁴⁰ ) (Eurofound, a). ( ¹⁴¹ ) Note that the Barcelona target for this age group (3+) is based on UOE data (educ\\_uoe\\_enra21) and not on data from EU-SILC survey. ( ¹⁴² ) Standard deviation dropped from 5.3 to 4.7. Beta coefficient is negative (-0.07) and statistically significant at 1% significance level.\n\nBox 3.2: Future investment needs in Early Childhood Education and Care\n\n**Further investment is needed to increase children’s participation in ECEC to the Barcelona target levels at EU level** , notably for the under-3s. Such investment would help to address several barriers to children’s access to ECEC, including the lack of available places and high access costs. For example, by creating new ECEC places, training, hiring and paying new staff, or refurbishing existing/building new facilities, while ensuring accessibility and affordability.\n\n**Additional yearly spending on ECEC needed to reach the EU Barcelona Targets by 2030 would amount to 0.085% of EU GDP** (more than EUR 11 billion) according to recent estimates. ( ¹ ) This amount varies by Member State, ranging from 0% (in Member States that have already reached the target) to nearly 0.3% in Greece and around 0.2% in Germany, Poland and Romania (Chart 1).\n\nChart 1 Additional investment needs in ECEC vary by Member State Yearly additional investment in ECEC needed to meet the 2030 Barcelona targets (% of GDP), by Barcelona targets\n\n_Source:_ DG EMPL calculations based on Eurostat data.\n\n( ¹ ) The additional investment need is estimated as follows. First, based on current participation rates (Eurostat ilc\\_caindformal\n\ndataset), the number of additional children who should be enrolled to meet the Barcelona targets is estimated. This number is then multiplied by the yearly average public expenditure on ECEC per participating child (Eurostat educ\\_uoe\\_fine09 dataset) to obtain the additional yearly spending needs to reach the Barcelona targets. It is assumed that the level of spending per participating child remains constant over the years.\n\n0\\.0%\n\n0\\.1%\n\n0\\.2%\n\n0\\.3%\n\nEU BE BG CZ DK DE EE IE EL ES FR HR IT CY LV LT LU HU MT NL AT PL PT RO SI SK FI SE\n\nAdditional spending needed to reach the 2030 Barcelona target for children from 3 to CSA\n\nAdditional spending needed to reach the 2030 country specific Barcelona target for children below 3","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/employment_and_social_developments_in_europe_esde_2024_report_p73.pdf"]},"response":{"errors":[],"extract_id":"extract_ba5a3685c4d5b751e6f6c54739365821","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Chapter 3: The role of social investment\n\n77\n\n**A large majority of preschool age children participate in ECEC in the EU, with participation rates converging across Member States. (** **140** **)** The share of children aged three to compulsory school age (CSA) participating in ECEC broadly stagnated, going from about 87% in 2014 to 89% in 2023 (Chart 3.2). The COVID- 19 pandemic resulted in a temporary decline in participation and a spike in cross-country differences in 2020. In 2023, national participation rates still varied considerably. For example, participation was above the Barcelona target of 96% ( ¹⁴¹ ) in seven Member States (Belgium, Denmark, Spain, France, Hungary, the Netherlands, Sweden), but was less than 80% in three countries (Poland, Romania, Slovakia). Cross-country differences declined between 2014 and 2023, as countries with lower participation rates caught up with those with higher participation. ( ¹⁴² )\n\n**While the average ECEC participation rate of the youngest children (0-2 years of age) rose over the last decade, disparities across Member States also increased.** Over the last ten years, the share of children participating rose from about 28% in 2014 to 37% in 2023, with a short lapse during the 2020 closures due to the COVID-19 pandemic and an increase by 1.5 pp in 2023 compared to the previous year (Chart 3.2). To\n\n( ¹⁴⁰ ) (Eurofound, a). ( ¹⁴¹ ) Note that the Barcelona target for this age group (3+) is based on UOE data (educ\\_uoe\\_enra21) and not on data from EU-SILC survey. ( ¹⁴² ) Standard deviation dropped from 5.3 to 4.7. Beta coefficient is negative (-0.07) and statistically significant at 1% significance level.\n\nBox 3.2: Future investment needs in Early Childhood Education and Care\n\n**Further investment is needed to increase children’s participation in ECEC to the Barcelona target levels at EU level** , notably for the under-3s. Such investment would help to address several barriers to children’s access to ECEC, including the lack of available places and high access costs. For example, by creating new ECEC places, training, hiring and paying new staff, or refurbishing existing/building new facilities, while ensuring accessibility and affordability.\n\n**Additional yearly spending on ECEC needed to reach the EU Barcelona Targets by 2030 would amount to 0.085% of EU GDP** (more than EUR 11 billion) according to recent estimates. ( ¹ ) This amount varies by Member State, ranging from 0% (in Member States that have already reached the target) to nearly 0.3% in Greece and around 0.2% in Germany, Poland and Romania (Chart 1).\n\nChart 1 Additional investment needs in ECEC vary by Member State Yearly additional investment in ECEC needed to meet the 2030 Barcelona targets (% of GDP), by Barcelona targets\n\n_Source:_ DG EMPL calculations based on Eurostat data.\n\n( ¹ ) The additional investment need is estimated as follows. First, based on current participation rates (Eurostat ilc\\_caindformal\n\ndataset), the number of additional children who should be enrolled to meet the Barcelona targets is estimated. This number is then multiplied by the yearly average public expenditure on ECEC per participating child (Eurostat educ\\_uoe\\_fine09 dataset) to obtain the additional yearly spending needs to reach the Barcelona targets. It is assumed that the level of spending per participating child remains constant over the years.\n\n0\\.0%\n\n0\\.1%\n\n0\\.2%\n\n0\\.3%\n\nEU BE BG CZ DK DE EE IE EL ES FR HR IT CY LV LT LU HU MT NL AT PL PT RO SI SK FI SE\n\nAdditional spending needed to reach the 2030 Barcelona target for children from 3 to CSA\n\nAdditional spending needed to reach the 2030 country specific Barcelona target for children below 3"],"full_content":"Chapter 3: The role of social investment\n\n77\n\n**A large majority of preschool age children participate in ECEC in the EU, with participation rates converging across Member States. (** **140** **)** The share of children aged three to compulsory school age (CSA) participating in ECEC broadly stagnated, going from about 87% in 2014 to 89% in 2023 (Chart 3.2). The COVID- 19 pandemic resulted in a temporary decline in participation and a spike in cross-country differences in 2020. In 2023, national participation rates still varied considerably. For example, participation was above the Barcelona target of 96% ( ¹⁴¹ ) in seven Member States (Belgium, Denmark, Spain, France, Hungary, the Netherlands, Sweden), but was less than 80% in three countries (Poland, Romania, Slovakia). Cross-country differences declined between 2014 and 2023, as countries with lower participation rates caught up with those with higher participation. ( ¹⁴² )\n\n**While the average ECEC participation rate of the youngest children (0-2 years of age) rose over the last decade, disparities across Member States also increased.** Over the last ten years, the share of children participating rose from about 28% in 2014 to 37% in 2023, with a short lapse during the 2020 closures due to the COVID-19 pandemic and an increase by 1.5 pp in 2023 compared to the previous year (Chart 3.2). To\n\n( ¹⁴⁰ ) (Eurofound, a). ( ¹⁴¹ ) Note that the Barcelona target for this age group (3+) is based on UOE data (educ\\_uoe\\_enra21) and not on data from EU-SILC survey. ( ¹⁴² ) Standard deviation dropped from 5.3 to 4.7. Beta coefficient is negative (-0.07) and statistically significant at 1% significance level.\n\nBox 3.2: Future investment needs in Early Childhood Education and Care\n\n**Further investment is needed to increase children’s participation in ECEC to the Barcelona target levels at EU level** , notably for the under-3s. Such investment would help to address several barriers to children’s access to ECEC, including the lack of available places and high access costs. For example, by creating new ECEC places, training, hiring and paying new staff, or refurbishing existing/building new facilities, while ensuring accessibility and affordability.\n\n**Additional yearly spending on ECEC needed to reach the EU Barcelona Targets by 2030 would amount to 0.085% of EU GDP** (more than EUR 11 billion) according to recent estimates. ( ¹ ) This amount varies by Member State, ranging from 0% (in Member States that have already reached the target) to nearly 0.3% in Greece and around 0.2% in Germany, Poland and Romania (Chart 1).\n\nChart 1 Additional investment needs in ECEC vary by Member State Yearly additional investment in ECEC needed to meet the 2030 Barcelona targets (% of GDP), by Barcelona targets\n\n_Source:_ DG EMPL calculations based on Eurostat data.\n\n( ¹ ) The additional investment need is estimated as follows. First, based on current participation rates (Eurostat ilc\\_caindformal\n\ndataset), the number of additional children who should be enrolled to meet the Barcelona targets is estimated. This number is then multiplied by the yearly average public expenditure on ECEC per participating child (Eurostat educ\\_uoe\\_fine09 dataset) to obtain the additional yearly spending needs to reach the Barcelona targets. It is assumed that the level of spending per participating child remains constant over the years.\n\n0\\.0%\n\n0\\.1%\n\n0\\.2%\n\n0\\.3%\n\nEU BE BG CZ DK DE EE IE EL ES FR HR IT CY LV LT LU HU MT NL AT PL PT RO SI SK FI SE\n\nAdditional spending needed to reach the 2030 Barcelona target for children from 3 to CSA\n\nAdditional spending needed to reach the 2030 country specific Barcelona target for children below 3","publish_date":null,"title":"Box 3.2: Future investment needs in Early Childhood Education and Care","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/employment_and_social_developments_in_europe_esde_2024_report_p73.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/employment_and_social_developments_in_europe_esde_2024_report_p73.pdf"},"request":{"example_id":"chart/employment_and_social_developments_in_europe_esde_2024_report_p73","product_type":"parse","source_file_path":"employment_and_social_developments_in_europe_esde_2024_report_p73.pdf"},"started_at":"2026-11-04T19:45:30.832657"} +{"completed_at":"2026-11-04T19:45:35.239619","latency_in_ms":3362,"output":{"example_id":"chart/employment_and_social_developments_in_europe_esde_2024_report_p74","grounded_pages":[],"layout_pages":[],"markdown":"Employment and Social Developments in Europe 2024\n\n78\n\ndate, ten Member States ( ¹⁴³ ) have reached the revised Barcelona participation target of 45% for this age group set at EU level, ( ¹⁴⁴ ) while six countries recorded participation rates lower than 20% in 2022. ( ¹⁴⁵ ) In the remaining Member States, between 20% and 45% of children participated in ECEC. The overall level of disparity between countries grew somewhat since 2014, despite evidence that some Member States with low participation rates caught up with others. ( ¹⁴⁶ ) Low levels of participation in ECEC for some Member States call for higher investment. Box 3.2 provides an estimate of future investment needs in the EU assuming constant level of spending per participating child.\n\nChart 3.2 Growing ECEC participation accompanied by convergence in attending preschool education Proportion of children participating in ECEC, by age group and variation across countries, 2014-2022, EU-27\n\n_Note:_ Data missing for Italy in 2020.\n\n_Source:_ DG EMPL calculations based on EU-SILC data [datasets ilc\\_caindformal (for 2014) and ilc\\_caindform25 (from 2015 onwards)].\n\nClick here to download chart.\n\n**Young children from disadvantaged backgrounds are less likely to participate in ECEC, especially in some Member States. (** **147** **)** In 2022, children facing different types of disadvantage (monetary poverty, material deprivation, low work intensity) had participation rates at least 10 pp lower than those from non- disadvantaged backgrounds, regardless of the type of disadvantage (Chart 3.3). ( ¹⁴⁸ ) However, this gap varied between Member States, with children at risk of poverty or social exclusion participating considerably less often (by more than 20 pp) in some Member States, ( ¹⁴⁹ ) and at a similar rate or more often than those not at risk in others. ( ¹⁵⁰ ) On average across all Member States, children who can benefit most from attending ECEC tend to participate least. ( ¹⁵¹ )\n\n( ¹⁴³ ) Belgium, Denmark, Spain, France, Luxembourg, Malta, the Netherlands, Portugal, Slovenia, Sweden. ( ¹⁴⁴ ) Specific targets apply to Member States that have not reached the 2002 Barcelona target of 33% participation for this age group. ( ¹⁴⁵ ) Bulgaria, Czechia, Lithuania, Poland, Romania, Slovakia. ( ¹⁴⁶ ) Beta coefficient is negative (-0.03) and statistically significant at 10% significance level. ( ¹⁴⁷ ) (Employment Committee and Social Protection Committee, 2023). ( ¹⁴⁸ ) Data for 2023 was not available at the time of writing. For analysis of earlier periods, see (European Commission, 2019b). ( ¹⁴⁹ ) Belgium, France, Italy, the Netherlands, Spain. ( ¹⁵⁰ ) Bulgaria, Denmark, Croatia, Hungary, Austria, Romania. ( ¹⁵¹ ) Convergence analysis cannot be carried out due to data limitations.\n\n25\n\n30\n\n35\n\n40\n\n45\n\na) Participation (0 to 2 years)\n\n80\n\n85\n\n90\n\n95\n\n100\n\nb) Participation (3 to CSA)\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\nc) Cross-country variation\n\n0 to 2 years 3 to CSA","pages":[{"line_numbers":[],"markdown":"Employment and Social Developments in Europe 2024\n\n78\n\ndate, ten Member States ( ¹⁴³ ) have reached the revised Barcelona participation target of 45% for this age group set at EU level, ( ¹⁴⁴ ) while six countries recorded participation rates lower than 20% in 2022. ( ¹⁴⁵ ) In the remaining Member States, between 20% and 45% of children participated in ECEC. The overall level of disparity between countries grew somewhat since 2014, despite evidence that some Member States with low participation rates caught up with others. ( ¹⁴⁶ ) Low levels of participation in ECEC for some Member States call for higher investment. Box 3.2 provides an estimate of future investment needs in the EU assuming constant level of spending per participating child.\n\nChart 3.2 Growing ECEC participation accompanied by convergence in attending preschool education Proportion of children participating in ECEC, by age group and variation across countries, 2014-2022, EU-27\n\n_Note:_ Data missing for Italy in 2020.\n\n_Source:_ DG EMPL calculations based on EU-SILC data [datasets ilc\\_caindformal (for 2014) and ilc\\_caindform25 (from 2015 onwards)].\n\nClick here to download chart.\n\n**Young children from disadvantaged backgrounds are less likely to participate in ECEC, especially in some Member States. (** **147** **)** In 2022, children facing different types of disadvantage (monetary poverty, material deprivation, low work intensity) had participation rates at least 10 pp lower than those from non- disadvantaged backgrounds, regardless of the type of disadvantage (Chart 3.3). ( ¹⁴⁸ ) However, this gap varied between Member States, with children at risk of poverty or social exclusion participating considerably less often (by more than 20 pp) in some Member States, ( ¹⁴⁹ ) and at a similar rate or more often than those not at risk in others. ( ¹⁵⁰ ) On average across all Member States, children who can benefit most from attending ECEC tend to participate least. ( ¹⁵¹ )\n\n( ¹⁴³ ) Belgium, Denmark, Spain, France, Luxembourg, Malta, the Netherlands, Portugal, Slovenia, Sweden. ( ¹⁴⁴ ) Specific targets apply to Member States that have not reached the 2002 Barcelona target of 33% participation for this age group. ( ¹⁴⁵ ) Bulgaria, Czechia, Lithuania, Poland, Romania, Slovakia. ( ¹⁴⁶ ) Beta coefficient is negative (-0.03) and statistically significant at 10% significance level. ( ¹⁴⁷ ) (Employment Committee and Social Protection Committee, 2023). ( ¹⁴⁸ ) Data for 2023 was not available at the time of writing. For analysis of earlier periods, see (European Commission, 2019b). ( ¹⁴⁹ ) Belgium, France, Italy, the Netherlands, Spain. ( ¹⁵⁰ ) Bulgaria, Denmark, Croatia, Hungary, Austria, Romania. ( ¹⁵¹ ) Convergence analysis cannot be carried out due to data limitations.\n\n25\n\n30\n\n35\n\n40\n\n45\n\na) Participation (0 to 2 years)\n\n80\n\n85\n\n90\n\n95\n\n100\n\nb) Participation (3 to CSA)\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\nc) Cross-country variation\n\n0 to 2 years 3 to CSA","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Employment and Social Developments in Europe 2024\n\n78\n\ndate, ten Member States ( ¹⁴³ ) have reached the revised Barcelona participation target of 45% for this age group set at EU level, ( ¹⁴⁴ ) while six countries recorded participation rates lower than 20% in 2022. ( ¹⁴⁵ ) In the remaining Member States, between 20% and 45% of children participated in ECEC. The overall level of disparity between countries grew somewhat since 2014, despite evidence that some Member States with low participation rates caught up with others. ( ¹⁴⁶ ) Low levels of participation in ECEC for some Member States call for higher investment. Box 3.2 provides an estimate of future investment needs in the EU assuming constant level of spending per participating child.\n\nChart 3.2 Growing ECEC participation accompanied by convergence in attending preschool education Proportion of children participating in ECEC, by age group and variation across countries, 2014-2022, EU-27\n\n_Note:_ Data missing for Italy in 2020.\n\n_Source:_ DG EMPL calculations based on EU-SILC data [datasets ilc\\_caindformal (for 2014) and ilc\\_caindform25 (from 2015 onwards)].\n\nClick here to download chart.\n\n**Young children from disadvantaged backgrounds are less likely to participate in ECEC, especially in some Member States. (** **147** **)** In 2022, children facing different types of disadvantage (monetary poverty, material deprivation, low work intensity) had participation rates at least 10 pp lower than those from non- disadvantaged backgrounds, regardless of the type of disadvantage (Chart 3.3). ( ¹⁴⁸ ) However, this gap varied between Member States, with children at risk of poverty or social exclusion participating considerably less often (by more than 20 pp) in some Member States, ( ¹⁴⁹ ) and at a similar rate or more often than those not at risk in others. ( ¹⁵⁰ ) On average across all Member States, children who can benefit most from attending ECEC tend to participate least. ( ¹⁵¹ )\n\n( ¹⁴³ ) Belgium, Denmark, Spain, France, Luxembourg, Malta, the Netherlands, Portugal, Slovenia, Sweden. ( ¹⁴⁴ ) Specific targets apply to Member States that have not reached the 2002 Barcelona target of 33% participation for this age group. ( ¹⁴⁵ ) Bulgaria, Czechia, Lithuania, Poland, Romania, Slovakia. ( ¹⁴⁶ ) Beta coefficient is negative (-0.03) and statistically significant at 10% significance level. ( ¹⁴⁷ ) (Employment Committee and Social Protection Committee, 2023). ( ¹⁴⁸ ) Data for 2023 was not available at the time of writing. For analysis of earlier periods, see (European Commission, 2019b). ( ¹⁴⁹ ) Belgium, France, Italy, the Netherlands, Spain. ( ¹⁵⁰ ) Bulgaria, Denmark, Croatia, Hungary, Austria, Romania. ( ¹⁵¹ ) Convergence analysis cannot be carried out due to data limitations.\n\n25\n\n30\n\n35\n\n40\n\n45\n\na) Participation (0 to 2 years)\n\n80\n\n85\n\n90\n\n95\n\n100\n\nb) Participation (3 to CSA)\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\nc) Cross-country variation\n\n0 to 2 years 3 to CSA","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/employment_and_social_developments_in_europe_esde_2024_report_p74.pdf"]},"response":{"errors":[],"extract_id":"extract_ca7989a328e40e328f5edf1e46ff6eca","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Employment and Social Developments in Europe 2024\n\n78\n\ndate, ten Member States ( ¹⁴³ ) have reached the revised Barcelona participation target of 45% for this age group set at EU level, ( ¹⁴⁴ ) while six countries recorded participation rates lower than 20% in 2022. ( ¹⁴⁵ ) In the remaining Member States, between 20% and 45% of children participated in ECEC. The overall level of disparity between countries grew somewhat since 2014, despite evidence that some Member States with low participation rates caught up with others. ( ¹⁴⁶ ) Low levels of participation in ECEC for some Member States call for higher investment. Box 3.2 provides an estimate of future investment needs in the EU assuming constant level of spending per participating child.\n\nChart 3.2 Growing ECEC participation accompanied by convergence in attending preschool education Proportion of children participating in ECEC, by age group and variation across countries, 2014-2022, EU-27\n\n_Note:_ Data missing for Italy in 2020.\n\n_Source:_ DG EMPL calculations based on EU-SILC data [datasets ilc\\_caindformal (for 2014) and ilc\\_caindform25 (from 2015 onwards)].\n\nClick here to download chart.\n\n**Young children from disadvantaged backgrounds are less likely to participate in ECEC, especially in some Member States. (** **147** **)** In 2022, children facing different types of disadvantage (monetary poverty, material deprivation, low work intensity) had participation rates at least 10 pp lower than those from non- disadvantaged backgrounds, regardless of the type of disadvantage (Chart 3.3). ( ¹⁴⁸ ) However, this gap varied between Member States, with children at risk of poverty or social exclusion participating considerably less often (by more than 20 pp) in some Member States, ( ¹⁴⁹ ) and at a similar rate or more often than those not at risk in others. ( ¹⁵⁰ ) On average across all Member States, children who can benefit most from attending ECEC tend to participate least. ( ¹⁵¹ )\n\n( ¹⁴³ ) Belgium, Denmark, Spain, France, Luxembourg, Malta, the Netherlands, Portugal, Slovenia, Sweden. ( ¹⁴⁴ ) Specific targets apply to Member States that have not reached the 2002 Barcelona target of 33% participation for this age group. ( ¹⁴⁵ ) Bulgaria, Czechia, Lithuania, Poland, Romania, Slovakia. ( ¹⁴⁶ ) Beta coefficient is negative (-0.03) and statistically significant at 10% significance level. ( ¹⁴⁷ ) (Employment Committee and Social Protection Committee, 2023). ( ¹⁴⁸ ) Data for 2023 was not available at the time of writing. For analysis of earlier periods, see (European Commission, 2019b). ( ¹⁴⁹ ) Belgium, France, Italy, the Netherlands, Spain. ( ¹⁵⁰ ) Bulgaria, Denmark, Croatia, Hungary, Austria, Romania. ( ¹⁵¹ ) Convergence analysis cannot be carried out due to data limitations.\n\n25\n\n30\n\n35\n\n40\n\n45\n\na) Participation (0 to 2 years)\n\n80\n\n85\n\n90\n\n95\n\n100\n\nb) Participation (3 to CSA)\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\nc) Cross-country variation\n\n0 to 2 years 3 to CSA"],"full_content":"Employment and Social Developments in Europe 2024\n\n78\n\ndate, ten Member States ( ¹⁴³ ) have reached the revised Barcelona participation target of 45% for this age group set at EU level, ( ¹⁴⁴ ) while six countries recorded participation rates lower than 20% in 2022. ( ¹⁴⁵ ) In the remaining Member States, between 20% and 45% of children participated in ECEC. The overall level of disparity between countries grew somewhat since 2014, despite evidence that some Member States with low participation rates caught up with others. ( ¹⁴⁶ ) Low levels of participation in ECEC for some Member States call for higher investment. Box 3.2 provides an estimate of future investment needs in the EU assuming constant level of spending per participating child.\n\nChart 3.2 Growing ECEC participation accompanied by convergence in attending preschool education Proportion of children participating in ECEC, by age group and variation across countries, 2014-2022, EU-27\n\n_Note:_ Data missing for Italy in 2020.\n\n_Source:_ DG EMPL calculations based on EU-SILC data [datasets ilc\\_caindformal (for 2014) and ilc\\_caindform25 (from 2015 onwards)].\n\nClick here to download chart.\n\n**Young children from disadvantaged backgrounds are less likely to participate in ECEC, especially in some Member States. (** **147** **)** In 2022, children facing different types of disadvantage (monetary poverty, material deprivation, low work intensity) had participation rates at least 10 pp lower than those from non- disadvantaged backgrounds, regardless of the type of disadvantage (Chart 3.3). ( ¹⁴⁸ ) However, this gap varied between Member States, with children at risk of poverty or social exclusion participating considerably less often (by more than 20 pp) in some Member States, ( ¹⁴⁹ ) and at a similar rate or more often than those not at risk in others. ( ¹⁵⁰ ) On average across all Member States, children who can benefit most from attending ECEC tend to participate least. ( ¹⁵¹ )\n\n( ¹⁴³ ) Belgium, Denmark, Spain, France, Luxembourg, Malta, the Netherlands, Portugal, Slovenia, Sweden. ( ¹⁴⁴ ) Specific targets apply to Member States that have not reached the 2002 Barcelona target of 33% participation for this age group. ( ¹⁴⁵ ) Bulgaria, Czechia, Lithuania, Poland, Romania, Slovakia. ( ¹⁴⁶ ) Beta coefficient is negative (-0.03) and statistically significant at 10% significance level. ( ¹⁴⁷ ) (Employment Committee and Social Protection Committee, 2023). ( ¹⁴⁸ ) Data for 2023 was not available at the time of writing. For analysis of earlier periods, see (European Commission, 2019b). ( ¹⁴⁹ ) Belgium, France, Italy, the Netherlands, Spain. ( ¹⁵⁰ ) Bulgaria, Denmark, Croatia, Hungary, Austria, Romania. ( ¹⁵¹ ) Convergence analysis cannot be carried out due to data limitations.\n\n25\n\n30\n\n35\n\n40\n\n45\n\na) Participation (0 to 2 years)\n\n80\n\n85\n\n90\n\n95\n\n100\n\nb) Participation (3 to CSA)\n\n0\n\n5\n\n10\n\n15\n\n20\n\n25\n\nc) Cross-country variation\n\n0 to 2 years 3 to CSA","publish_date":null,"title":"date, ten Member States ( ) have reached the revised Barcelona participation target of 45% for this age group )","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/employment_and_social_developments_in_europe_esde_2024_report_p74.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/employment_and_social_developments_in_europe_esde_2024_report_p74.pdf"},"request":{"example_id":"chart/employment_and_social_developments_in_europe_esde_2024_report_p74","product_type":"parse","source_file_path":"employment_and_social_developments_in_europe_esde_2024_report_p74.pdf"},"started_at":"2026-11-04T19:45:31.876681"} +{"completed_at":"2026-11-04T19:45:33.944445","latency_in_ms":1320,"output":{"example_id":"chart/employment_and_social_developments_in_europe_esde_2024_report_p76","grounded_pages":[],"layout_pages":[],"markdown":"Employment and Social Developments in Europe 2024\n\n80\n\nChart 3.4 School education receives the largest share of public expenditure on education Weighted average EU share of government expenditure on education, by education level, 2012-2022\n\n_Note:_ School education includes (pre-)primary and secondary levels. Further categories of public expenditure on education not included in the chart are post-secondary non-tertiary education, education not definable by level, subsidiary services to education, research and development (R&D) education, and education not elsewhere classified.\n\n_Source:_ Eurostat [gov\\_10a\\_exp]. Click here to download chart.\n\n**PISA outcomes point to worsening basic skills in the EU over time, with potential negative impacts on EU competitiveness and wage developments in the longer term.** The average PISA score across reading, mathematics and science subjects decreased by 10 points since the last wave in 2018, and by 18 points over the last 10 years in the EU (Chart 3.5). ( ¹⁵⁷ ) While this declining pattern was also observed in other industrial countries (with the average PISA score dropping by 9 points since 2018 and 19 points since 2012 in OECD countries on average), the EU performed quite poorly compared to other advanced economies such as the US, the UK, Canada or Japan. In 2022, for example, it had the highest percentage of underachievers across all three subjects, and one of the lowest shares of top performers among those countries. Although PISA outcomes had already started to decline prior to the COVID-19 pandemic, learning losses during lockdowns might have intensified this downward trend among other possible factors. Box 3.3 presents a case study on Italy, showing the persistent impacts of COVID-19 pandemic school disruptions on student achievement, with learning losses still significant in 2021 and 2022.\n\nChart 3.5 Basic skills and efficiency of cumulative expenditure on education per student decreased post-COVID-19 pandemic Average EU PISA score across reading, mathematics and science, and PISA EU efficiency score, 2012-2022\n\n_Note:_ PISA efficiency score measures the number of PISA points produced by USD 1 000 (expressed in purchasing power parity (PPP)) of cumulative expenditure on education per student, following the methodology in European Commission (2023d). It is calculated as 1 000 x (average PISA score over reading, mathematics and science)/cumulative expenditure per student from age 6-15 (in USD, PPP), and can be interpreted as an approximate measure of the pay-off from additional expenditure for a school education system. Yearly averages for Member States for which data are available.\n\n_Source:_ OECD PISA data for 2012, 2015, 2018, 2022.\n\nClick here to download chart.\n\n( ¹⁵⁷ ) To approximate the average annual pace of learning among 15-year-olds, OECD (2023) used 20 score points in PISA as a common benchmark for all countries. However, this benchmark should not be used to convert the differences across countries and changes over time in PISA scores as years-of-schooling equivalents, given differences in the pace of learning at a given age, organisation of schooling, resources invested in education, and quality of education across countries, as well as possible changes in the pace of learning over time.\n\n0%\n\n20%\n\n40%\n\n60%\n\n80%\n\n100%\n\n2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022\n\nPre-primary and primary Secondary Tertiary\n\n0\n\n3\n\n6\n\n9\n\n470\n\n480\n\n490\n\n500\n\n2012 2015 2018 2022\n\nPISA average score (left axis)\n\nPISA efficiency score (right axis)","pages":[{"line_numbers":[],"markdown":"Employment and Social Developments in Europe 2024\n\n80\n\nChart 3.4 School education receives the largest share of public expenditure on education Weighted average EU share of government expenditure on education, by education level, 2012-2022\n\n_Note:_ School education includes (pre-)primary and secondary levels. Further categories of public expenditure on education not included in the chart are post-secondary non-tertiary education, education not definable by level, subsidiary services to education, research and development (R&D) education, and education not elsewhere classified.\n\n_Source:_ Eurostat [gov\\_10a\\_exp]. Click here to download chart.\n\n**PISA outcomes point to worsening basic skills in the EU over time, with potential negative impacts on EU competitiveness and wage developments in the longer term.** The average PISA score across reading, mathematics and science subjects decreased by 10 points since the last wave in 2018, and by 18 points over the last 10 years in the EU (Chart 3.5). ( ¹⁵⁷ ) While this declining pattern was also observed in other industrial countries (with the average PISA score dropping by 9 points since 2018 and 19 points since 2012 in OECD countries on average), the EU performed quite poorly compared to other advanced economies such as the US, the UK, Canada or Japan. In 2022, for example, it had the highest percentage of underachievers across all three subjects, and one of the lowest shares of top performers among those countries. Although PISA outcomes had already started to decline prior to the COVID-19 pandemic, learning losses during lockdowns might have intensified this downward trend among other possible factors. Box 3.3 presents a case study on Italy, showing the persistent impacts of COVID-19 pandemic school disruptions on student achievement, with learning losses still significant in 2021 and 2022.\n\nChart 3.5 Basic skills and efficiency of cumulative expenditure on education per student decreased post-COVID-19 pandemic Average EU PISA score across reading, mathematics and science, and PISA EU efficiency score, 2012-2022\n\n_Note:_ PISA efficiency score measures the number of PISA points produced by USD 1 000 (expressed in purchasing power parity (PPP)) of cumulative expenditure on education per student, following the methodology in European Commission (2023d). It is calculated as 1 000 x (average PISA score over reading, mathematics and science)/cumulative expenditure per student from age 6-15 (in USD, PPP), and can be interpreted as an approximate measure of the pay-off from additional expenditure for a school education system. Yearly averages for Member States for which data are available.\n\n_Source:_ OECD PISA data for 2012, 2015, 2018, 2022.\n\nClick here to download chart.\n\n( ¹⁵⁷ ) To approximate the average annual pace of learning among 15-year-olds, OECD (2023) used 20 score points in PISA as a common benchmark for all countries. However, this benchmark should not be used to convert the differences across countries and changes over time in PISA scores as years-of-schooling equivalents, given differences in the pace of learning at a given age, organisation of schooling, resources invested in education, and quality of education across countries, as well as possible changes in the pace of learning over time.\n\n0%\n\n20%\n\n40%\n\n60%\n\n80%\n\n100%\n\n2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022\n\nPre-primary and primary Secondary Tertiary\n\n0\n\n3\n\n6\n\n9\n\n470\n\n480\n\n490\n\n500\n\n2012 2015 2018 2022\n\nPISA average score (left axis)\n\nPISA efficiency score (right axis)","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Employment and Social Developments in Europe 2024\n\n80\n\nChart 3.4 School education receives the largest share of public expenditure on education Weighted average EU share of government expenditure on education, by education level, 2012-2022\n\n_Note:_ School education includes (pre-)primary and secondary levels. Further categories of public expenditure on education not included in the chart are post-secondary non-tertiary education, education not definable by level, subsidiary services to education, research and development (R&D) education, and education not elsewhere classified.\n\n_Source:_ Eurostat [gov\\_10a\\_exp]. Click here to download chart.\n\n**PISA outcomes point to worsening basic skills in the EU over time, with potential negative impacts on EU competitiveness and wage developments in the longer term.** The average PISA score across reading, mathematics and science subjects decreased by 10 points since the last wave in 2018, and by 18 points over the last 10 years in the EU (Chart 3.5). ( ¹⁵⁷ ) While this declining pattern was also observed in other industrial countries (with the average PISA score dropping by 9 points since 2018 and 19 points since 2012 in OECD countries on average), the EU performed quite poorly compared to other advanced economies such as the US, the UK, Canada or Japan. In 2022, for example, it had the highest percentage of underachievers across all three subjects, and one of the lowest shares of top performers among those countries. Although PISA outcomes had already started to decline prior to the COVID-19 pandemic, learning losses during lockdowns might have intensified this downward trend among other possible factors. Box 3.3 presents a case study on Italy, showing the persistent impacts of COVID-19 pandemic school disruptions on student achievement, with learning losses still significant in 2021 and 2022.\n\nChart 3.5 Basic skills and efficiency of cumulative expenditure on education per student decreased post-COVID-19 pandemic Average EU PISA score across reading, mathematics and science, and PISA EU efficiency score, 2012-2022\n\n_Note:_ PISA efficiency score measures the number of PISA points produced by USD 1 000 (expressed in purchasing power parity (PPP)) of cumulative expenditure on education per student, following the methodology in European Commission (2023d). It is calculated as 1 000 x (average PISA score over reading, mathematics and science)/cumulative expenditure per student from age 6-15 (in USD, PPP), and can be interpreted as an approximate measure of the pay-off from additional expenditure for a school education system. Yearly averages for Member States for which data are available.\n\n_Source:_ OECD PISA data for 2012, 2015, 2018, 2022.\n\nClick here to download chart.\n\n( ¹⁵⁷ ) To approximate the average annual pace of learning among 15-year-olds, OECD (2023) used 20 score points in PISA as a common benchmark for all countries. However, this benchmark should not be used to convert the differences across countries and changes over time in PISA scores as years-of-schooling equivalents, given differences in the pace of learning at a given age, organisation of schooling, resources invested in education, and quality of education across countries, as well as possible changes in the pace of learning over time.\n\n0%\n\n20%\n\n40%\n\n60%\n\n80%\n\n100%\n\n2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022\n\nPre-primary and primary Secondary Tertiary\n\n0\n\n3\n\n6\n\n9\n\n470\n\n480\n\n490\n\n500\n\n2012 2015 2018 2022\n\nPISA average score (left axis)\n\nPISA efficiency score (right axis)","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/employment_and_social_developments_in_europe_esde_2024_report_p76.pdf"]},"response":{"errors":[],"extract_id":"extract_6c066644314b1fe5f20511400f6b3f68","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Employment and Social Developments in Europe 2024\n\n80\n\nChart 3.4 School education receives the largest share of public expenditure on education Weighted average EU share of government expenditure on education, by education level, 2012-2022\n\n_Note:_ School education includes (pre-)primary and secondary levels. Further categories of public expenditure on education not included in the chart are post-secondary non-tertiary education, education not definable by level, subsidiary services to education, research and development (R&D) education, and education not elsewhere classified.\n\n_Source:_ Eurostat [gov\\_10a\\_exp]. Click here to download chart.\n\n**PISA outcomes point to worsening basic skills in the EU over time, with potential negative impacts on EU competitiveness and wage developments in the longer term.** The average PISA score across reading, mathematics and science subjects decreased by 10 points since the last wave in 2018, and by 18 points over the last 10 years in the EU (Chart 3.5). ( ¹⁵⁷ ) While this declining pattern was also observed in other industrial countries (with the average PISA score dropping by 9 points since 2018 and 19 points since 2012 in OECD countries on average), the EU performed quite poorly compared to other advanced economies such as the US, the UK, Canada or Japan. In 2022, for example, it had the highest percentage of underachievers across all three subjects, and one of the lowest shares of top performers among those countries. Although PISA outcomes had already started to decline prior to the COVID-19 pandemic, learning losses during lockdowns might have intensified this downward trend among other possible factors. Box 3.3 presents a case study on Italy, showing the persistent impacts of COVID-19 pandemic school disruptions on student achievement, with learning losses still significant in 2021 and 2022.\n\nChart 3.5 Basic skills and efficiency of cumulative expenditure on education per student decreased post-COVID-19 pandemic Average EU PISA score across reading, mathematics and science, and PISA EU efficiency score, 2012-2022\n\n_Note:_ PISA efficiency score measures the number of PISA points produced by USD 1 000 (expressed in purchasing power parity (PPP)) of cumulative expenditure on education per student, following the methodology in European Commission (2023d). It is calculated as 1 000 x (average PISA score over reading, mathematics and science)/cumulative expenditure per student from age 6-15 (in USD, PPP), and can be interpreted as an approximate measure of the pay-off from additional expenditure for a school education system. Yearly averages for Member States for which data are available.\n\n_Source:_ OECD PISA data for 2012, 2015, 2018, 2022.\n\nClick here to download chart.\n\n( ¹⁵⁷ ) To approximate the average annual pace of learning among 15-year-olds, OECD (2023) used 20 score points in PISA as a common benchmark for all countries. However, this benchmark should not be used to convert the differences across countries and changes over time in PISA scores as years-of-schooling equivalents, given differences in the pace of learning at a given age, organisation of schooling, resources invested in education, and quality of education across countries, as well as possible changes in the pace of learning over time.\n\n0%\n\n20%\n\n40%\n\n60%\n\n80%\n\n100%\n\n2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022\n\nPre-primary and primary Secondary Tertiary\n\n0\n\n3\n\n6\n\n9\n\n470\n\n480\n\n490\n\n500\n\n2012 2015 2018 2022\n\nPISA average score (left axis)\n\nPISA efficiency score (right axis)"],"full_content":"Employment and Social Developments in Europe 2024\n\n80\n\nChart 3.4 School education receives the largest share of public expenditure on education Weighted average EU share of government expenditure on education, by education level, 2012-2022\n\n_Note:_ School education includes (pre-)primary and secondary levels. Further categories of public expenditure on education not included in the chart are post-secondary non-tertiary education, education not definable by level, subsidiary services to education, research and development (R&D) education, and education not elsewhere classified.\n\n_Source:_ Eurostat [gov\\_10a\\_exp]. Click here to download chart.\n\n**PISA outcomes point to worsening basic skills in the EU over time, with potential negative impacts on EU competitiveness and wage developments in the longer term.** The average PISA score across reading, mathematics and science subjects decreased by 10 points since the last wave in 2018, and by 18 points over the last 10 years in the EU (Chart 3.5). ( ¹⁵⁷ ) While this declining pattern was also observed in other industrial countries (with the average PISA score dropping by 9 points since 2018 and 19 points since 2012 in OECD countries on average), the EU performed quite poorly compared to other advanced economies such as the US, the UK, Canada or Japan. In 2022, for example, it had the highest percentage of underachievers across all three subjects, and one of the lowest shares of top performers among those countries. Although PISA outcomes had already started to decline prior to the COVID-19 pandemic, learning losses during lockdowns might have intensified this downward trend among other possible factors. Box 3.3 presents a case study on Italy, showing the persistent impacts of COVID-19 pandemic school disruptions on student achievement, with learning losses still significant in 2021 and 2022.\n\nChart 3.5 Basic skills and efficiency of cumulative expenditure on education per student decreased post-COVID-19 pandemic Average EU PISA score across reading, mathematics and science, and PISA EU efficiency score, 2012-2022\n\n_Note:_ PISA efficiency score measures the number of PISA points produced by USD 1 000 (expressed in purchasing power parity (PPP)) of cumulative expenditure on education per student, following the methodology in European Commission (2023d). It is calculated as 1 000 x (average PISA score over reading, mathematics and science)/cumulative expenditure per student from age 6-15 (in USD, PPP), and can be interpreted as an approximate measure of the pay-off from additional expenditure for a school education system. Yearly averages for Member States for which data are available.\n\n_Source:_ OECD PISA data for 2012, 2015, 2018, 2022.\n\nClick here to download chart.\n\n( ¹⁵⁷ ) To approximate the average annual pace of learning among 15-year-olds, OECD (2023) used 20 score points in PISA as a common benchmark for all countries. However, this benchmark should not be used to convert the differences across countries and changes over time in PISA scores as years-of-schooling equivalents, given differences in the pace of learning at a given age, organisation of schooling, resources invested in education, and quality of education across countries, as well as possible changes in the pace of learning over time.\n\n0%\n\n20%\n\n40%\n\n60%\n\n80%\n\n100%\n\n2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022\n\nPre-primary and primary Secondary Tertiary\n\n0\n\n3\n\n6\n\n9\n\n470\n\n480\n\n490\n\n500\n\n2012 2015 2018 2022\n\nPISA average score (left axis)\n\nPISA efficiency score (right axis)","publish_date":null,"title":"School education receives the largest share of public expenditure on education 100% 80% 60% 40% 20% 0% 2012 2013 2014","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/employment_and_social_developments_in_europe_esde_2024_report_p76.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/employment_and_social_developments_in_europe_esde_2024_report_p76.pdf"},"request":{"example_id":"chart/employment_and_social_developments_in_europe_esde_2024_report_p76","product_type":"parse","source_file_path":"employment_and_social_developments_in_europe_esde_2024_report_p76.pdf"},"started_at":"2026-11-04T19:45:32.623458"} +{"completed_at":"2026-11-04T19:45:35.047317","latency_in_ms":1099,"output":{"example_id":"chart/employment_and_social_developments_in_europe_esde_2024_report_p80","grounded_pages":[],"layout_pages":[],"markdown":"Employment and Social Developments in Europe 2024\n\n84\n\nworkers with additional capital. As a result, GDP is projected to increase by 0.18% in the long term, compared to a scenario without any additional investment in young unemployed people (Chart 3.7).\n\n### Investments in skills enhance productivity\n\n**Regional investment in skills in the context of the European Social Fund Plus (ESF+) can lead to significant increases in economic activity in the long term.** The European Commission’s RHOMOLO model is used to simulate the potential regional macroeconomic impact of ESF+ investments in skills over the 2021- 2027 funding period, which are assumed to increase labour productivity (Box A3.1 in the Annex). ( ¹⁷¹ ) The simulations illustrate that ESF+ funding for skills, in addition to the direct employment and social benefits expected in the areas and sectors targeted by the projects financed, could increase EU GDP by up to 0.039% at its peak in 2036 relative to baseline GDP (Chart 3.8). GDP is expected to remain above its baseline level even after expenditure on ESF+ terminates, as the structural effects of increased labour productivity and corresponding adjustments by firms and households materialise. Overall, the initial cost of the measure, at less than 0.035% of GDP per year, is more than offset by the long-term economic returns, which are significantly larger than the original investment.\n\nChart 3.8 Investment in skills can lead to long-term GDP gains Expenditure on skills-related ESF+ programmes over 2021-2027 programming period (% over baseline GDP) and expected impact of the investment on GDP (% deviation from baseline GDP)\n\n_Note:_ 'Baseline’ constitutes a scenario with no additional investment.\n\n_Source:_ JRC calculations based on RHOMOLO model.\n\nClick here to download chart.\n\n**Targeted investments in skills can lead to employment gains throughout the funding period and in the long-term.** At their peak in 2026, EU-level employment gains are projected to be 0.024% compared to the baseline scenario of no investment, with the largest employment gains expected for the lowest income quintiles (Chart 3.9). ( ¹⁷² ) This impact is larger in regions with more significant investments. ( ¹⁷³ ) When the demand stimulus related to the ESF+ funds ends, employment decreases and temporarily dips below its initial level, ( ¹⁷⁴ ) before recovering and rising above the initial level in the long term. ( ¹⁷⁵ ) Developments in the unemployment rate mirror these results, with decreases in unemployment rates projected in the short and long term, as vacancies for upskilled individuals who benefitted from the programmes are filled (Chart 3.9).\n\n( ¹⁷¹ ) (Christou et al., 2024). ( ¹⁷² ) The lower income quintiles also experience larger increases in real wages compared to the richest income quintile, as their labour productivity increases relatively more.\n\n( ¹⁷³ ) For instance, in the Bulgarian region BG31, Severozapaden, and the Portuguese region PT20, Região Autónoma dos Açores, impacts at the peak are +0.195% and +0.185%, respectively.\n\n( ¹⁷⁴ ) The temporary decrease in employment below initial levels stems from modelling specificities. In the model, agents are myopic and make decisions based only on past and current economic conditions. Thus, they cannot anticipate the end of the monetary injection (at the end of the implementation period), nor the full positive impact of the supply-side effects of the policy intervention that will materialise in the future. The combination of positive shocks (increased government spending and increased productivity) and negative shocks (contributions needed to finance the policy) together with the asymmetric nature of the shocks across regions and agents being myopic can lead to a temporary negative effect at the end of the implementation period.\n\n( ¹⁷⁵ ) Despite ESF+ spending of less than 0.1% of EU GDP for each year of the programming period, employment would still be 0.013% higher than the baseline 20 years after the start of the programme.\n\n0\\.00\n\n0\\.01\n\n0\\.02\n\n0\\.03\n\n0\\.04\n\n0\\.05\n\n2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2036 2037 2038 2039 2040\n\n% deviation from baseline\n\nCohesion Expenditure Changes in GDP compared to baseline","pages":[{"line_numbers":[],"markdown":"Employment and Social Developments in Europe 2024\n\n84\n\nworkers with additional capital. As a result, GDP is projected to increase by 0.18% in the long term, compared to a scenario without any additional investment in young unemployed people (Chart 3.7).\n\n### Investments in skills enhance productivity\n\n**Regional investment in skills in the context of the European Social Fund Plus (ESF+) can lead to significant increases in economic activity in the long term.** The European Commission’s RHOMOLO model is used to simulate the potential regional macroeconomic impact of ESF+ investments in skills over the 2021- 2027 funding period, which are assumed to increase labour productivity (Box A3.1 in the Annex). ( ¹⁷¹ ) The simulations illustrate that ESF+ funding for skills, in addition to the direct employment and social benefits expected in the areas and sectors targeted by the projects financed, could increase EU GDP by up to 0.039% at its peak in 2036 relative to baseline GDP (Chart 3.8). GDP is expected to remain above its baseline level even after expenditure on ESF+ terminates, as the structural effects of increased labour productivity and corresponding adjustments by firms and households materialise. Overall, the initial cost of the measure, at less than 0.035% of GDP per year, is more than offset by the long-term economic returns, which are significantly larger than the original investment.\n\nChart 3.8 Investment in skills can lead to long-term GDP gains Expenditure on skills-related ESF+ programmes over 2021-2027 programming period (% over baseline GDP) and expected impact of the investment on GDP (% deviation from baseline GDP)\n\n_Note:_ 'Baseline’ constitutes a scenario with no additional investment.\n\n_Source:_ JRC calculations based on RHOMOLO model.\n\nClick here to download chart.\n\n**Targeted investments in skills can lead to employment gains throughout the funding period and in the long-term.** At their peak in 2026, EU-level employment gains are projected to be 0.024% compared to the baseline scenario of no investment, with the largest employment gains expected for the lowest income quintiles (Chart 3.9). ( ¹⁷² ) This impact is larger in regions with more significant investments. ( ¹⁷³ ) When the demand stimulus related to the ESF+ funds ends, employment decreases and temporarily dips below its initial level, ( ¹⁷⁴ ) before recovering and rising above the initial level in the long term. ( ¹⁷⁵ ) Developments in the unemployment rate mirror these results, with decreases in unemployment rates projected in the short and long term, as vacancies for upskilled individuals who benefitted from the programmes are filled (Chart 3.9).\n\n( ¹⁷¹ ) (Christou et al., 2024). ( ¹⁷² ) The lower income quintiles also experience larger increases in real wages compared to the richest income quintile, as their labour productivity increases relatively more.\n\n( ¹⁷³ ) For instance, in the Bulgarian region BG31, Severozapaden, and the Portuguese region PT20, Região Autónoma dos Açores, impacts at the peak are +0.195% and +0.185%, respectively.\n\n( ¹⁷⁴ ) The temporary decrease in employment below initial levels stems from modelling specificities. In the model, agents are myopic and make decisions based only on past and current economic conditions. Thus, they cannot anticipate the end of the monetary injection (at the end of the implementation period), nor the full positive impact of the supply-side effects of the policy intervention that will materialise in the future. The combination of positive shocks (increased government spending and increased productivity) and negative shocks (contributions needed to finance the policy) together with the asymmetric nature of the shocks across regions and agents being myopic can lead to a temporary negative effect at the end of the implementation period.\n\n( ¹⁷⁵ ) Despite ESF+ spending of less than 0.1% of EU GDP for each year of the programming period, employment would still be 0.013% higher than the baseline 20 years after the start of the programme.\n\n0\\.00\n\n0\\.01\n\n0\\.02\n\n0\\.03\n\n0\\.04\n\n0\\.05\n\n2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2036 2037 2038 2039 2040\n\n% deviation from baseline\n\nCohesion Expenditure Changes in GDP compared to baseline","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Employment and Social Developments in Europe 2024\n\n84\n\nworkers with additional capital. As a result, GDP is projected to increase by 0.18% in the long term, compared to a scenario without any additional investment in young unemployed people (Chart 3.7).\n\n### Investments in skills enhance productivity\n\n**Regional investment in skills in the context of the European Social Fund Plus (ESF+) can lead to significant increases in economic activity in the long term.** The European Commission’s RHOMOLO model is used to simulate the potential regional macroeconomic impact of ESF+ investments in skills over the 2021- 2027 funding period, which are assumed to increase labour productivity (Box A3.1 in the Annex). ( ¹⁷¹ ) The simulations illustrate that ESF+ funding for skills, in addition to the direct employment and social benefits expected in the areas and sectors targeted by the projects financed, could increase EU GDP by up to 0.039% at its peak in 2036 relative to baseline GDP (Chart 3.8). GDP is expected to remain above its baseline level even after expenditure on ESF+ terminates, as the structural effects of increased labour productivity and corresponding adjustments by firms and households materialise. Overall, the initial cost of the measure, at less than 0.035% of GDP per year, is more than offset by the long-term economic returns, which are significantly larger than the original investment.\n\nChart 3.8 Investment in skills can lead to long-term GDP gains Expenditure on skills-related ESF+ programmes over 2021-2027 programming period (% over baseline GDP) and expected impact of the investment on GDP (% deviation from baseline GDP)\n\n_Note:_ 'Baseline’ constitutes a scenario with no additional investment.\n\n_Source:_ JRC calculations based on RHOMOLO model.\n\nClick here to download chart.\n\n**Targeted investments in skills can lead to employment gains throughout the funding period and in the long-term.** At their peak in 2026, EU-level employment gains are projected to be 0.024% compared to the baseline scenario of no investment, with the largest employment gains expected for the lowest income quintiles (Chart 3.9). ( ¹⁷² ) This impact is larger in regions with more significant investments. ( ¹⁷³ ) When the demand stimulus related to the ESF+ funds ends, employment decreases and temporarily dips below its initial level, ( ¹⁷⁴ ) before recovering and rising above the initial level in the long term. ( ¹⁷⁵ ) Developments in the unemployment rate mirror these results, with decreases in unemployment rates projected in the short and long term, as vacancies for upskilled individuals who benefitted from the programmes are filled (Chart 3.9).\n\n( ¹⁷¹ ) (Christou et al., 2024). ( ¹⁷² ) The lower income quintiles also experience larger increases in real wages compared to the richest income quintile, as their labour productivity increases relatively more.\n\n( ¹⁷³ ) For instance, in the Bulgarian region BG31, Severozapaden, and the Portuguese region PT20, Região Autónoma dos Açores, impacts at the peak are +0.195% and +0.185%, respectively.\n\n( ¹⁷⁴ ) The temporary decrease in employment below initial levels stems from modelling specificities. In the model, agents are myopic and make decisions based only on past and current economic conditions. Thus, they cannot anticipate the end of the monetary injection (at the end of the implementation period), nor the full positive impact of the supply-side effects of the policy intervention that will materialise in the future. The combination of positive shocks (increased government spending and increased productivity) and negative shocks (contributions needed to finance the policy) together with the asymmetric nature of the shocks across regions and agents being myopic can lead to a temporary negative effect at the end of the implementation period.\n\n( ¹⁷⁵ ) Despite ESF+ spending of less than 0.1% of EU GDP for each year of the programming period, employment would still be 0.013% higher than the baseline 20 years after the start of the programme.\n\n0\\.00\n\n0\\.01\n\n0\\.02\n\n0\\.03\n\n0\\.04\n\n0\\.05\n\n2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2036 2037 2038 2039 2040\n\n% deviation from baseline\n\nCohesion Expenditure Changes in GDP compared to baseline","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/employment_and_social_developments_in_europe_esde_2024_report_p80.pdf"]},"response":{"errors":[],"extract_id":"extract_3fcf6ba072ed131ead0c1535ae9a98af","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Employment and Social Developments in Europe 2024\n\n84\n\nworkers with additional capital. As a result, GDP is projected to increase by 0.18% in the long term, compared to a scenario without any additional investment in young unemployed people (Chart 3.7).\n\n### Investments in skills enhance productivity\n\n**Regional investment in skills in the context of the European Social Fund Plus (ESF+) can lead to significant increases in economic activity in the long term.** The European Commission’s RHOMOLO model is used to simulate the potential regional macroeconomic impact of ESF+ investments in skills over the 2021- 2027 funding period, which are assumed to increase labour productivity (Box A3.1 in the Annex). ( ¹⁷¹ ) The simulations illustrate that ESF+ funding for skills, in addition to the direct employment and social benefits expected in the areas and sectors targeted by the projects financed, could increase EU GDP by up to 0.039% at its peak in 2036 relative to baseline GDP (Chart 3.8). GDP is expected to remain above its baseline level even after expenditure on ESF+ terminates, as the structural effects of increased labour productivity and corresponding adjustments by firms and households materialise. Overall, the initial cost of the measure, at less than 0.035% of GDP per year, is more than offset by the long-term economic returns, which are significantly larger than the original investment.\n\nChart 3.8 Investment in skills can lead to long-term GDP gains Expenditure on skills-related ESF+ programmes over 2021-2027 programming period (% over baseline GDP) and expected impact of the investment on GDP (% deviation from baseline GDP)\n\n_Note:_ 'Baseline’ constitutes a scenario with no additional investment.\n\n_Source:_ JRC calculations based on RHOMOLO model.\n\nClick here to download chart.\n\n**Targeted investments in skills can lead to employment gains throughout the funding period and in the long-term.** At their peak in 2026, EU-level employment gains are projected to be 0.024% compared to the baseline scenario of no investment, with the largest employment gains expected for the lowest income quintiles (Chart 3.9). ( ¹⁷² ) This impact is larger in regions with more significant investments. ( ¹⁷³ ) When the demand stimulus related to the ESF+ funds ends, employment decreases and temporarily dips below its initial level, ( ¹⁷⁴ ) before recovering and rising above the initial level in the long term. ( ¹⁷⁵ ) Developments in the unemployment rate mirror these results, with decreases in unemployment rates projected in the short and long term, as vacancies for upskilled individuals who benefitted from the programmes are filled (Chart 3.9).\n\n( ¹⁷¹ ) (Christou et al., 2024). ( ¹⁷² ) The lower income quintiles also experience larger increases in real wages compared to the richest income quintile, as their labour productivity increases relatively more.\n\n( ¹⁷³ ) For instance, in the Bulgarian region BG31, Severozapaden, and the Portuguese region PT20, Região Autónoma dos Açores, impacts at the peak are +0.195% and +0.185%, respectively.\n\n( ¹⁷⁴ ) The temporary decrease in employment below initial levels stems from modelling specificities. In the model, agents are myopic and make decisions based only on past and current economic conditions. Thus, they cannot anticipate the end of the monetary injection (at the end of the implementation period), nor the full positive impact of the supply-side effects of the policy intervention that will materialise in the future. The combination of positive shocks (increased government spending and increased productivity) and negative shocks (contributions needed to finance the policy) together with the asymmetric nature of the shocks across regions and agents being myopic can lead to a temporary negative effect at the end of the implementation period.\n\n( ¹⁷⁵ ) Despite ESF+ spending of less than 0.1% of EU GDP for each year of the programming period, employment would still be 0.013% higher than the baseline 20 years after the start of the programme.\n\n0\\.00\n\n0\\.01\n\n0\\.02\n\n0\\.03\n\n0\\.04\n\n0\\.05\n\n2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2036 2037 2038 2039 2040\n\n% deviation from baseline\n\nCohesion Expenditure Changes in GDP compared to baseline"],"full_content":"Employment and Social Developments in Europe 2024\n\n84\n\nworkers with additional capital. As a result, GDP is projected to increase by 0.18% in the long term, compared to a scenario without any additional investment in young unemployed people (Chart 3.7).\n\n### Investments in skills enhance productivity\n\n**Regional investment in skills in the context of the European Social Fund Plus (ESF+) can lead to significant increases in economic activity in the long term.** The European Commission’s RHOMOLO model is used to simulate the potential regional macroeconomic impact of ESF+ investments in skills over the 2021- 2027 funding period, which are assumed to increase labour productivity (Box A3.1 in the Annex). ( ¹⁷¹ ) The simulations illustrate that ESF+ funding for skills, in addition to the direct employment and social benefits expected in the areas and sectors targeted by the projects financed, could increase EU GDP by up to 0.039% at its peak in 2036 relative to baseline GDP (Chart 3.8). GDP is expected to remain above its baseline level even after expenditure on ESF+ terminates, as the structural effects of increased labour productivity and corresponding adjustments by firms and households materialise. Overall, the initial cost of the measure, at less than 0.035% of GDP per year, is more than offset by the long-term economic returns, which are significantly larger than the original investment.\n\nChart 3.8 Investment in skills can lead to long-term GDP gains Expenditure on skills-related ESF+ programmes over 2021-2027 programming period (% over baseline GDP) and expected impact of the investment on GDP (% deviation from baseline GDP)\n\n_Note:_ 'Baseline’ constitutes a scenario with no additional investment.\n\n_Source:_ JRC calculations based on RHOMOLO model.\n\nClick here to download chart.\n\n**Targeted investments in skills can lead to employment gains throughout the funding period and in the long-term.** At their peak in 2026, EU-level employment gains are projected to be 0.024% compared to the baseline scenario of no investment, with the largest employment gains expected for the lowest income quintiles (Chart 3.9). ( ¹⁷² ) This impact is larger in regions with more significant investments. ( ¹⁷³ ) When the demand stimulus related to the ESF+ funds ends, employment decreases and temporarily dips below its initial level, ( ¹⁷⁴ ) before recovering and rising above the initial level in the long term. ( ¹⁷⁵ ) Developments in the unemployment rate mirror these results, with decreases in unemployment rates projected in the short and long term, as vacancies for upskilled individuals who benefitted from the programmes are filled (Chart 3.9).\n\n( ¹⁷¹ ) (Christou et al., 2024). ( ¹⁷² ) The lower income quintiles also experience larger increases in real wages compared to the richest income quintile, as their labour productivity increases relatively more.\n\n( ¹⁷³ ) For instance, in the Bulgarian region BG31, Severozapaden, and the Portuguese region PT20, Região Autónoma dos Açores, impacts at the peak are +0.195% and +0.185%, respectively.\n\n( ¹⁷⁴ ) The temporary decrease in employment below initial levels stems from modelling specificities. In the model, agents are myopic and make decisions based only on past and current economic conditions. Thus, they cannot anticipate the end of the monetary injection (at the end of the implementation period), nor the full positive impact of the supply-side effects of the policy intervention that will materialise in the future. The combination of positive shocks (increased government spending and increased productivity) and negative shocks (contributions needed to finance the policy) together with the asymmetric nature of the shocks across regions and agents being myopic can lead to a temporary negative effect at the end of the implementation period.\n\n( ¹⁷⁵ ) Despite ESF+ spending of less than 0.1% of EU GDP for each year of the programming period, employment would still be 0.013% higher than the baseline 20 years after the start of the programme.\n\n0\\.00\n\n0\\.01\n\n0\\.02\n\n0\\.03\n\n0\\.04\n\n0\\.05\n\n2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2036 2037 2038 2039 2040\n\n% deviation from baseline\n\nCohesion Expenditure Changes in GDP compared to baseline","publish_date":null,"title":"Investments in skills enhance productivity","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/employment_and_social_developments_in_europe_esde_2024_report_p80.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/employment_and_social_developments_in_europe_esde_2024_report_p80.pdf"},"request":{"example_id":"chart/employment_and_social_developments_in_europe_esde_2024_report_p80","product_type":"parse","source_file_path":"employment_and_social_developments_in_europe_esde_2024_report_p80.pdf"},"started_at":"2026-11-04T19:45:33.947865"} +{"completed_at":"2026-11-04T19:45:36.280129","latency_in_ms":1231,"output":{"example_id":"chart/employment_and_social_developments_in_europe_esde_2024_report_p84","grounded_pages":[],"layout_pages":[],"markdown":"Employment and Social Developments in Europe 2024\n\n88\n\n### 3\\.4. Investing in Active Labour Market Policies\n\n**Well-functioning and effective active labour market policies (** **182** **) can provide economic and social returns in the medium to long term.** Research suggests that investment in ALMPs is associated with economic growth and positively impacts employment by helping to reduce labour mismatches and shortages, equipping workers with relevant skills, supporting job transitions, and strengthening labour market participation of underrepresented groups. ( ¹⁸³ ) Evidence shows that job search assistance programmes are more effective for disadvantaged participants, while employment subsidies and training work better for long-term unemployed people. A review of studies on evaluations of ALMPs found that the average impact of ALMPs increases in the medium to long-term, that women and long-term unemployed people tend to benefit more than other demographic groups, and that ALMPs have more positive effects during recessions. Design and implementation of programmes also play a role in ALMPs’ efficacy, which may depend on the availability of other social policies, including access to ECEC. ( ¹⁸⁴ ) In light of rapid technological advances and changing skills needs, reinforcing investment in ALMPs through lifelong learning policies is key to maximise returns. ( ¹⁸⁵ )\n\n**According to impact evaluation studies, spending on ALMPs can yield positive returns in the longer term, including longer working careers, with corresponding tax revenue, social security contributions and better pensions for workers. (** **186** **)** A recent evaluation identified positive fiscal effects for a direct job creation scheme in Austria in the long run, focused on disadvantaged groups. ( ¹⁸⁷ ) For social outcomes, ALMPs have poverty-reducing effects, particularly for those with lower levels of skills. ( ¹⁸⁸ ) Recent policy impact evaluations in Germany also found positive effects of ALMP participation on well-being. ( ¹⁸⁹ ) However, more evidence is needed from microeconomic evaluations of the social outcomes of ALMPs to better understand their social investment returns for health, well-being, and education-related outcomes, as well as to compare the efficiency of spending on different types of ALMPs.\n\nChart 3.11 Investment in ALMPs is projected to increase employment during and after the investment period Expected impact on employment of investment in labour supply increasing intervention fields of the ESF+, 2021-2027 programming period, by income quintile (% deviation from baseline)\n\n_Note:_ Income quintile 5 indicates the richest quintile, and quintile 1 the poorest.\n\n_Source:_ JRC calculations based on RHOMOLO model. Click here to download chart.\n\n**Investment in ALMPs can have positive effects on employment and GDP well beyond the timeframe of the original investment.** Simulations using the RHOMOLO model (Box A3.2 in the Annex) estimate the long- term effects of increasing the labour supply through ALMP interventions funded through ESF+ in the 2021-2027 programming period. Regional ESF+ investments in ALMPs are expected to improve employment outcomes in the 20 years following the original investment, reaching a peak of +0.11% in 2027 (Chart 3.11). The investments are\n\n( ¹⁸² ) ‘Active labour market policies (ALMPs) are publicly financed interventions intended to improve the functioning of the labour market by inducing changes in labour demand and labour supply, as well as their matching process. Specifically, these policies aim to preserve existing jobs and create new employment opportunities, encouraging labour market attachment and the reintegration of long-term unemployed and inactive individuals, and facilitating the job-search and job-matching process. In practice, they target labour market outsiders – all unemployed and inactive individuals’ (Ernst et al., 2022).\n\n( ¹⁸³ ) (Sakamoto, 2020) ; (Card, Kluve and Weber, 2018) ; (Levy-Yeyati, Montané and Sartorio, 2019). ( ¹⁸⁴ ) (Kluve et al., 2019); (Plavgo and Hemerijck, 2021); (Nieuwenhuis, 2022). ( ¹⁸⁵ ) (European Commission, 2016). ( ¹⁸⁶ ) (Brown and Koetti, 2012). ( ¹⁸⁷ ) (Walch and Dorofeenko, 2020). ( ¹⁸⁸ ) (Rovny, 2014); (Cammeraat, 2020); (European Commission, 2016). ( ¹⁸⁹ ) (Tübbicke and Schiele, 2023).\n\n0\\.00\n\n0\\.02\n\n0\\.04\n\n0\\.06\n\n0\\.08\n\n0\\.10\n\n0\\.12\n\n0\\.14\n\n0\\.16\n\n2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2036 2037 2038 2039 2040\n\n_% deviation from baseline_\n\nQuintile 1 Quintile 2 Quintile 3 Quntile 4 Quintile 5","pages":[{"line_numbers":[],"markdown":"Employment and Social Developments in Europe 2024\n\n88\n\n### 3\\.4. Investing in Active Labour Market Policies\n\n**Well-functioning and effective active labour market policies (** **182** **) can provide economic and social returns in the medium to long term.** Research suggests that investment in ALMPs is associated with economic growth and positively impacts employment by helping to reduce labour mismatches and shortages, equipping workers with relevant skills, supporting job transitions, and strengthening labour market participation of underrepresented groups. ( ¹⁸³ ) Evidence shows that job search assistance programmes are more effective for disadvantaged participants, while employment subsidies and training work better for long-term unemployed people. A review of studies on evaluations of ALMPs found that the average impact of ALMPs increases in the medium to long-term, that women and long-term unemployed people tend to benefit more than other demographic groups, and that ALMPs have more positive effects during recessions. Design and implementation of programmes also play a role in ALMPs’ efficacy, which may depend on the availability of other social policies, including access to ECEC. ( ¹⁸⁴ ) In light of rapid technological advances and changing skills needs, reinforcing investment in ALMPs through lifelong learning policies is key to maximise returns. ( ¹⁸⁵ )\n\n**According to impact evaluation studies, spending on ALMPs can yield positive returns in the longer term, including longer working careers, with corresponding tax revenue, social security contributions and better pensions for workers. (** **186** **)** A recent evaluation identified positive fiscal effects for a direct job creation scheme in Austria in the long run, focused on disadvantaged groups. ( ¹⁸⁷ ) For social outcomes, ALMPs have poverty-reducing effects, particularly for those with lower levels of skills. ( ¹⁸⁸ ) Recent policy impact evaluations in Germany also found positive effects of ALMP participation on well-being. ( ¹⁸⁹ ) However, more evidence is needed from microeconomic evaluations of the social outcomes of ALMPs to better understand their social investment returns for health, well-being, and education-related outcomes, as well as to compare the efficiency of spending on different types of ALMPs.\n\nChart 3.11 Investment in ALMPs is projected to increase employment during and after the investment period Expected impact on employment of investment in labour supply increasing intervention fields of the ESF+, 2021-2027 programming period, by income quintile (% deviation from baseline)\n\n_Note:_ Income quintile 5 indicates the richest quintile, and quintile 1 the poorest.\n\n_Source:_ JRC calculations based on RHOMOLO model. Click here to download chart.\n\n**Investment in ALMPs can have positive effects on employment and GDP well beyond the timeframe of the original investment.** Simulations using the RHOMOLO model (Box A3.2 in the Annex) estimate the long- term effects of increasing the labour supply through ALMP interventions funded through ESF+ in the 2021-2027 programming period. Regional ESF+ investments in ALMPs are expected to improve employment outcomes in the 20 years following the original investment, reaching a peak of +0.11% in 2027 (Chart 3.11). The investments are\n\n( ¹⁸² ) ‘Active labour market policies (ALMPs) are publicly financed interventions intended to improve the functioning of the labour market by inducing changes in labour demand and labour supply, as well as their matching process. Specifically, these policies aim to preserve existing jobs and create new employment opportunities, encouraging labour market attachment and the reintegration of long-term unemployed and inactive individuals, and facilitating the job-search and job-matching process. In practice, they target labour market outsiders – all unemployed and inactive individuals’ (Ernst et al., 2022).\n\n( ¹⁸³ ) (Sakamoto, 2020) ; (Card, Kluve and Weber, 2018) ; (Levy-Yeyati, Montané and Sartorio, 2019). ( ¹⁸⁴ ) (Kluve et al., 2019); (Plavgo and Hemerijck, 2021); (Nieuwenhuis, 2022). ( ¹⁸⁵ ) (European Commission, 2016). ( ¹⁸⁶ ) (Brown and Koetti, 2012). ( ¹⁸⁷ ) (Walch and Dorofeenko, 2020). ( ¹⁸⁸ ) (Rovny, 2014); (Cammeraat, 2020); (European Commission, 2016). ( ¹⁸⁹ ) (Tübbicke and Schiele, 2023).\n\n0\\.00\n\n0\\.02\n\n0\\.04\n\n0\\.06\n\n0\\.08\n\n0\\.10\n\n0\\.12\n\n0\\.14\n\n0\\.16\n\n2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2036 2037 2038 2039 2040\n\n_% deviation from baseline_\n\nQuintile 1 Quintile 2 Quintile 3 Quntile 4 Quintile 5","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Employment and Social Developments in Europe 2024\n\n88\n\n### 3\\.4. Investing in Active Labour Market Policies\n\n**Well-functioning and effective active labour market policies (** **182** **) can provide economic and social returns in the medium to long term.** Research suggests that investment in ALMPs is associated with economic growth and positively impacts employment by helping to reduce labour mismatches and shortages, equipping workers with relevant skills, supporting job transitions, and strengthening labour market participation of underrepresented groups. ( ¹⁸³ ) Evidence shows that job search assistance programmes are more effective for disadvantaged participants, while employment subsidies and training work better for long-term unemployed people. A review of studies on evaluations of ALMPs found that the average impact of ALMPs increases in the medium to long-term, that women and long-term unemployed people tend to benefit more than other demographic groups, and that ALMPs have more positive effects during recessions. Design and implementation of programmes also play a role in ALMPs’ efficacy, which may depend on the availability of other social policies, including access to ECEC. ( ¹⁸⁴ ) In light of rapid technological advances and changing skills needs, reinforcing investment in ALMPs through lifelong learning policies is key to maximise returns. ( ¹⁸⁵ )\n\n**According to impact evaluation studies, spending on ALMPs can yield positive returns in the longer term, including longer working careers, with corresponding tax revenue, social security contributions and better pensions for workers. (** **186** **)** A recent evaluation identified positive fiscal effects for a direct job creation scheme in Austria in the long run, focused on disadvantaged groups. ( ¹⁸⁷ ) For social outcomes, ALMPs have poverty-reducing effects, particularly for those with lower levels of skills. ( ¹⁸⁸ ) Recent policy impact evaluations in Germany also found positive effects of ALMP participation on well-being. ( ¹⁸⁹ ) However, more evidence is needed from microeconomic evaluations of the social outcomes of ALMPs to better understand their social investment returns for health, well-being, and education-related outcomes, as well as to compare the efficiency of spending on different types of ALMPs.\n\nChart 3.11 Investment in ALMPs is projected to increase employment during and after the investment period Expected impact on employment of investment in labour supply increasing intervention fields of the ESF+, 2021-2027 programming period, by income quintile (% deviation from baseline)\n\n_Note:_ Income quintile 5 indicates the richest quintile, and quintile 1 the poorest.\n\n_Source:_ JRC calculations based on RHOMOLO model. Click here to download chart.\n\n**Investment in ALMPs can have positive effects on employment and GDP well beyond the timeframe of the original investment.** Simulations using the RHOMOLO model (Box A3.2 in the Annex) estimate the long- term effects of increasing the labour supply through ALMP interventions funded through ESF+ in the 2021-2027 programming period. Regional ESF+ investments in ALMPs are expected to improve employment outcomes in the 20 years following the original investment, reaching a peak of +0.11% in 2027 (Chart 3.11). The investments are\n\n( ¹⁸² ) ‘Active labour market policies (ALMPs) are publicly financed interventions intended to improve the functioning of the labour market by inducing changes in labour demand and labour supply, as well as their matching process. Specifically, these policies aim to preserve existing jobs and create new employment opportunities, encouraging labour market attachment and the reintegration of long-term unemployed and inactive individuals, and facilitating the job-search and job-matching process. In practice, they target labour market outsiders – all unemployed and inactive individuals’ (Ernst et al., 2022).\n\n( ¹⁸³ ) (Sakamoto, 2020) ; (Card, Kluve and Weber, 2018) ; (Levy-Yeyati, Montané and Sartorio, 2019). ( ¹⁸⁴ ) (Kluve et al., 2019); (Plavgo and Hemerijck, 2021); (Nieuwenhuis, 2022). ( ¹⁸⁵ ) (European Commission, 2016). ( ¹⁸⁶ ) (Brown and Koetti, 2012). ( ¹⁸⁷ ) (Walch and Dorofeenko, 2020). ( ¹⁸⁸ ) (Rovny, 2014); (Cammeraat, 2020); (European Commission, 2016). ( ¹⁸⁹ ) (Tübbicke and Schiele, 2023).\n\n0\\.00\n\n0\\.02\n\n0\\.04\n\n0\\.06\n\n0\\.08\n\n0\\.10\n\n0\\.12\n\n0\\.14\n\n0\\.16\n\n2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2036 2037 2038 2039 2040\n\n_% deviation from baseline_\n\nQuintile 1 Quintile 2 Quintile 3 Quntile 4 Quintile 5","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/employment_and_social_developments_in_europe_esde_2024_report_p84.pdf"]},"response":{"errors":[],"extract_id":"extract_531daf077f6a3c9b4e1b36ef9748d9d0","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Employment and Social Developments in Europe 2024\n\n88\n\n### 3\\.4. Investing in Active Labour Market Policies\n\n**Well-functioning and effective active labour market policies (** **182** **) can provide economic and social returns in the medium to long term.** Research suggests that investment in ALMPs is associated with economic growth and positively impacts employment by helping to reduce labour mismatches and shortages, equipping workers with relevant skills, supporting job transitions, and strengthening labour market participation of underrepresented groups. ( ¹⁸³ ) Evidence shows that job search assistance programmes are more effective for disadvantaged participants, while employment subsidies and training work better for long-term unemployed people. A review of studies on evaluations of ALMPs found that the average impact of ALMPs increases in the medium to long-term, that women and long-term unemployed people tend to benefit more than other demographic groups, and that ALMPs have more positive effects during recessions. Design and implementation of programmes also play a role in ALMPs’ efficacy, which may depend on the availability of other social policies, including access to ECEC. ( ¹⁸⁴ ) In light of rapid technological advances and changing skills needs, reinforcing investment in ALMPs through lifelong learning policies is key to maximise returns. ( ¹⁸⁵ )\n\n**According to impact evaluation studies, spending on ALMPs can yield positive returns in the longer term, including longer working careers, with corresponding tax revenue, social security contributions and better pensions for workers. (** **186** **)** A recent evaluation identified positive fiscal effects for a direct job creation scheme in Austria in the long run, focused on disadvantaged groups. ( ¹⁸⁷ ) For social outcomes, ALMPs have poverty-reducing effects, particularly for those with lower levels of skills. ( ¹⁸⁸ ) Recent policy impact evaluations in Germany also found positive effects of ALMP participation on well-being. ( ¹⁸⁹ ) However, more evidence is needed from microeconomic evaluations of the social outcomes of ALMPs to better understand their social investment returns for health, well-being, and education-related outcomes, as well as to compare the efficiency of spending on different types of ALMPs.\n\nChart 3.11 Investment in ALMPs is projected to increase employment during and after the investment period Expected impact on employment of investment in labour supply increasing intervention fields of the ESF+, 2021-2027 programming period, by income quintile (% deviation from baseline)\n\n_Note:_ Income quintile 5 indicates the richest quintile, and quintile 1 the poorest.\n\n_Source:_ JRC calculations based on RHOMOLO model. Click here to download chart.\n\n**Investment in ALMPs can have positive effects on employment and GDP well beyond the timeframe of the original investment.** Simulations using the RHOMOLO model (Box A3.2 in the Annex) estimate the long- term effects of increasing the labour supply through ALMP interventions funded through ESF+ in the 2021-2027 programming period. Regional ESF+ investments in ALMPs are expected to improve employment outcomes in the 20 years following the original investment, reaching a peak of +0.11% in 2027 (Chart 3.11). The investments are\n\n( ¹⁸² ) ‘Active labour market policies (ALMPs) are publicly financed interventions intended to improve the functioning of the labour market by inducing changes in labour demand and labour supply, as well as their matching process. Specifically, these policies aim to preserve existing jobs and create new employment opportunities, encouraging labour market attachment and the reintegration of long-term unemployed and inactive individuals, and facilitating the job-search and job-matching process. In practice, they target labour market outsiders – all unemployed and inactive individuals’ (Ernst et al., 2022).\n\n( ¹⁸³ ) (Sakamoto, 2020) ; (Card, Kluve and Weber, 2018) ; (Levy-Yeyati, Montané and Sartorio, 2019). ( ¹⁸⁴ ) (Kluve et al., 2019); (Plavgo and Hemerijck, 2021); (Nieuwenhuis, 2022). ( ¹⁸⁵ ) (European Commission, 2016). ( ¹⁸⁶ ) (Brown and Koetti, 2012). ( ¹⁸⁷ ) (Walch and Dorofeenko, 2020). ( ¹⁸⁸ ) (Rovny, 2014); (Cammeraat, 2020); (European Commission, 2016). ( ¹⁸⁹ ) (Tübbicke and Schiele, 2023).\n\n0\\.00\n\n0\\.02\n\n0\\.04\n\n0\\.06\n\n0\\.08\n\n0\\.10\n\n0\\.12\n\n0\\.14\n\n0\\.16\n\n2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2036 2037 2038 2039 2040\n\n_% deviation from baseline_\n\nQuintile 1 Quintile 2 Quintile 3 Quntile 4 Quintile 5"],"full_content":"Employment and Social Developments in Europe 2024\n\n88\n\n### 3\\.4. Investing in Active Labour Market Policies\n\n**Well-functioning and effective active labour market policies (** **182** **) can provide economic and social returns in the medium to long term.** Research suggests that investment in ALMPs is associated with economic growth and positively impacts employment by helping to reduce labour mismatches and shortages, equipping workers with relevant skills, supporting job transitions, and strengthening labour market participation of underrepresented groups. ( ¹⁸³ ) Evidence shows that job search assistance programmes are more effective for disadvantaged participants, while employment subsidies and training work better for long-term unemployed people. A review of studies on evaluations of ALMPs found that the average impact of ALMPs increases in the medium to long-term, that women and long-term unemployed people tend to benefit more than other demographic groups, and that ALMPs have more positive effects during recessions. Design and implementation of programmes also play a role in ALMPs’ efficacy, which may depend on the availability of other social policies, including access to ECEC. ( ¹⁸⁴ ) In light of rapid technological advances and changing skills needs, reinforcing investment in ALMPs through lifelong learning policies is key to maximise returns. ( ¹⁸⁵ )\n\n**According to impact evaluation studies, spending on ALMPs can yield positive returns in the longer term, including longer working careers, with corresponding tax revenue, social security contributions and better pensions for workers. (** **186** **)** A recent evaluation identified positive fiscal effects for a direct job creation scheme in Austria in the long run, focused on disadvantaged groups. ( ¹⁸⁷ ) For social outcomes, ALMPs have poverty-reducing effects, particularly for those with lower levels of skills. ( ¹⁸⁸ ) Recent policy impact evaluations in Germany also found positive effects of ALMP participation on well-being. ( ¹⁸⁹ ) However, more evidence is needed from microeconomic evaluations of the social outcomes of ALMPs to better understand their social investment returns for health, well-being, and education-related outcomes, as well as to compare the efficiency of spending on different types of ALMPs.\n\nChart 3.11 Investment in ALMPs is projected to increase employment during and after the investment period Expected impact on employment of investment in labour supply increasing intervention fields of the ESF+, 2021-2027 programming period, by income quintile (% deviation from baseline)\n\n_Note:_ Income quintile 5 indicates the richest quintile, and quintile 1 the poorest.\n\n_Source:_ JRC calculations based on RHOMOLO model. Click here to download chart.\n\n**Investment in ALMPs can have positive effects on employment and GDP well beyond the timeframe of the original investment.** Simulations using the RHOMOLO model (Box A3.2 in the Annex) estimate the long- term effects of increasing the labour supply through ALMP interventions funded through ESF+ in the 2021-2027 programming period. Regional ESF+ investments in ALMPs are expected to improve employment outcomes in the 20 years following the original investment, reaching a peak of +0.11% in 2027 (Chart 3.11). The investments are\n\n( ¹⁸² ) ‘Active labour market policies (ALMPs) are publicly financed interventions intended to improve the functioning of the labour market by inducing changes in labour demand and labour supply, as well as their matching process. Specifically, these policies aim to preserve existing jobs and create new employment opportunities, encouraging labour market attachment and the reintegration of long-term unemployed and inactive individuals, and facilitating the job-search and job-matching process. In practice, they target labour market outsiders – all unemployed and inactive individuals’ (Ernst et al., 2022).\n\n( ¹⁸³ ) (Sakamoto, 2020) ; (Card, Kluve and Weber, 2018) ; (Levy-Yeyati, Montané and Sartorio, 2019). ( ¹⁸⁴ ) (Kluve et al., 2019); (Plavgo and Hemerijck, 2021); (Nieuwenhuis, 2022). ( ¹⁸⁵ ) (European Commission, 2016). ( ¹⁸⁶ ) (Brown and Koetti, 2012). ( ¹⁸⁷ ) (Walch and Dorofeenko, 2020). ( ¹⁸⁸ ) (Rovny, 2014); (Cammeraat, 2020); (European Commission, 2016). ( ¹⁸⁹ ) (Tübbicke and Schiele, 2023).\n\n0\\.00\n\n0\\.02\n\n0\\.04\n\n0\\.06\n\n0\\.08\n\n0\\.10\n\n0\\.12\n\n0\\.14\n\n0\\.16\n\n2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2036 2037 2038 2039 2040\n\n_% deviation from baseline_\n\nQuintile 1 Quintile 2 Quintile 3 Quntile 4 Quintile 5","publish_date":null,"title":"3.4. Investing in Active Labour Market Policies","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/employment_and_social_developments_in_europe_esde_2024_report_p84.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/employment_and_social_developments_in_europe_esde_2024_report_p84.pdf"},"request":{"example_id":"chart/employment_and_social_developments_in_europe_esde_2024_report_p84","product_type":"parse","source_file_path":"employment_and_social_developments_in_europe_esde_2024_report_p84.pdf"},"started_at":"2026-11-04T19:45:35.048796"} +{"completed_at":"2026-11-04T19:45:39.427258","latency_in_ms":3145,"output":{"example_id":"chart/employment_and_social_developments_in_europe_esde_2024_report_p87","grounded_pages":[],"layout_pages":[],"markdown":"Chapter 3: The role of social investment\n\n91\n\nBox 3.6: Evaluation of job creation schemes in Ireland\n\n**In Ireland, two direct job creation schemes were found to have positive effects on earning outcomes for (mostly long-term unemployed) participants.** The research found that, after initial lock-in effects, the Community Employment (CE) scheme, ( ¹ ) which primarily targets people who have been unemployed for more than 12 months, has a positive impact on total annual earnings, raising the probability of any earnings in a given year and annual weeks of employment in the regular labour market. More specifically, CE participants earn around EUR 2 000 per year more than matched non-participants and work an average of four weeks longer per year. The Tús (Start) scheme, a community work placement scheme offering short-term work opportunities for long-term unemployed people, was also found to have a positive, albeit modest, impact on the number of jobseekers exiting the unemployment register, as well as a lasting positive effect on earnings (Chart 1). ( ² )\n\nChart 1 Women and younger jobseekers experience a greater boost in earnings following participation in Ireland’s Tús scheme Impact of Tús on earnings, by sex, unemployment duration and age\n\n_Source:_ (OECD, 2024a) calculations based on Department of Social Protection administrative data. **The two schemes have positive impacts on social outcomes, with different effects on specific groups of long-term unemployed people.** Participation in CE reduces future reliance on disability allowance and enhances take-up of education subsidies, indicating that the scheme has possible broader health and engagement in education effects. Six years after starting CE, former participants are 6 pp less likely to receive disability allowance, compared to similar participants who did not enter the scheme. This impact is even stronger for older jobseekers. Younger CE participants seem to benefit more in respect of labour market outcomes. For the Tús scheme, women, and jobseekers under 30 experienced a slightly greater increase in their earnings (Chart 1).\n\n( ¹ ) CE is a public work programme and is one of the most widely used ALMPs for long-term unemployed people in Ireland. Its main\n\nobjectives are to connect jobseekers with the labour market to increase employment levels and to promote social inclusion, as it seeks to reduce social isolation and social barriers for jobseekers.\n\n( ² ) (OECD, 2024a)\n\n\\-€500\n\n€0\n\n€500\n\n€1 000\n\n€1 500\n\n€2 000\n\n€2 500\n\n\\-3 -2 -1 0 1 2 3 4 5 6 7 8\n\nOutcome year\n\nA. By sex\n\nTotal Women Men\n\n\\-€500\n\n€0\n\n€500\n\n€1 000\n\n€1 500\n\n€2 000\n\n€2 500\n\n\\-3 -2 -1 0 1 2 3 4 5 6 7 8\n\nOutcome year\n\nC. By age\n\nUnder 30 30 to 50\n\nOver 50 Total\n\n\\-€500\n\n€0\n\n€500\n\n€1 000\n\n€1 500\n\n€2 000\n\n€2 500\n\n\\-3 -2 -1 0 1 2 3 4 5 6 7 8\n\nOutcome year\n\nB. By unemployment\n\nduration One to three years\n\nMore than three years","pages":[{"line_numbers":[],"markdown":"Chapter 3: The role of social investment\n\n91\n\nBox 3.6: Evaluation of job creation schemes in Ireland\n\n**In Ireland, two direct job creation schemes were found to have positive effects on earning outcomes for (mostly long-term unemployed) participants.** The research found that, after initial lock-in effects, the Community Employment (CE) scheme, ( ¹ ) which primarily targets people who have been unemployed for more than 12 months, has a positive impact on total annual earnings, raising the probability of any earnings in a given year and annual weeks of employment in the regular labour market. More specifically, CE participants earn around EUR 2 000 per year more than matched non-participants and work an average of four weeks longer per year. The Tús (Start) scheme, a community work placement scheme offering short-term work opportunities for long-term unemployed people, was also found to have a positive, albeit modest, impact on the number of jobseekers exiting the unemployment register, as well as a lasting positive effect on earnings (Chart 1). ( ² )\n\nChart 1 Women and younger jobseekers experience a greater boost in earnings following participation in Ireland’s Tús scheme Impact of Tús on earnings, by sex, unemployment duration and age\n\n_Source:_ (OECD, 2024a) calculations based on Department of Social Protection administrative data. **The two schemes have positive impacts on social outcomes, with different effects on specific groups of long-term unemployed people.** Participation in CE reduces future reliance on disability allowance and enhances take-up of education subsidies, indicating that the scheme has possible broader health and engagement in education effects. Six years after starting CE, former participants are 6 pp less likely to receive disability allowance, compared to similar participants who did not enter the scheme. This impact is even stronger for older jobseekers. Younger CE participants seem to benefit more in respect of labour market outcomes. For the Tús scheme, women, and jobseekers under 30 experienced a slightly greater increase in their earnings (Chart 1).\n\n( ¹ ) CE is a public work programme and is one of the most widely used ALMPs for long-term unemployed people in Ireland. Its main\n\nobjectives are to connect jobseekers with the labour market to increase employment levels and to promote social inclusion, as it seeks to reduce social isolation and social barriers for jobseekers.\n\n( ² ) (OECD, 2024a)\n\n\\-€500\n\n€0\n\n€500\n\n€1 000\n\n€1 500\n\n€2 000\n\n€2 500\n\n\\-3 -2 -1 0 1 2 3 4 5 6 7 8\n\nOutcome year\n\nA. By sex\n\nTotal Women Men\n\n\\-€500\n\n€0\n\n€500\n\n€1 000\n\n€1 500\n\n€2 000\n\n€2 500\n\n\\-3 -2 -1 0 1 2 3 4 5 6 7 8\n\nOutcome year\n\nC. By age\n\nUnder 30 30 to 50\n\nOver 50 Total\n\n\\-€500\n\n€0\n\n€500\n\n€1 000\n\n€1 500\n\n��2 000\n\n€2 500\n\n\\-3 -2 -1 0 1 2 3 4 5 6 7 8\n\nOutcome year\n\nB. By unemployment\n\nduration One to three years\n\nMore than three years","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Chapter 3: The role of social investment\n\n91\n\nBox 3.6: Evaluation of job creation schemes in Ireland\n\n**In Ireland, two direct job creation schemes were found to have positive effects on earning outcomes for (mostly long-term unemployed) participants.** The research found that, after initial lock-in effects, the Community Employment (CE) scheme, ( ¹ ) which primarily targets people who have been unemployed for more than 12 months, has a positive impact on total annual earnings, raising the probability of any earnings in a given year and annual weeks of employment in the regular labour market. More specifically, CE participants earn around EUR 2 000 per year more than matched non-participants and work an average of four weeks longer per year. The Tús (Start) scheme, a community work placement scheme offering short-term work opportunities for long-term unemployed people, was also found to have a positive, albeit modest, impact on the number of jobseekers exiting the unemployment register, as well as a lasting positive effect on earnings (Chart 1). ( ² )\n\nChart 1 Women and younger jobseekers experience a greater boost in earnings following participation in Ireland’s Tús scheme Impact of Tús on earnings, by sex, unemployment duration and age\n\n_Source:_ (OECD, 2024a) calculations based on Department of Social Protection administrative data. **The two schemes have positive impacts on social outcomes, with different effects on specific groups of long-term unemployed people.** Participation in CE reduces future reliance on disability allowance and enhances take-up of education subsidies, indicating that the scheme has possible broader health and engagement in education effects. Six years after starting CE, former participants are 6 pp less likely to receive disability allowance, compared to similar participants who did not enter the scheme. This impact is even stronger for older jobseekers. Younger CE participants seem to benefit more in respect of labour market outcomes. For the Tús scheme, women, and jobseekers under 30 experienced a slightly greater increase in their earnings (Chart 1).\n\n( ¹ ) CE is a public work programme and is one of the most widely used ALMPs for long-term unemployed people in Ireland. Its main\n\nobjectives are to connect jobseekers with the labour market to increase employment levels and to promote social inclusion, as it seeks to reduce social isolation and social barriers for jobseekers.\n\n( ² ) (OECD, 2024a)\n\n\\-€500\n\n€0\n\n€500\n\n€1 000\n\n€1 500\n\n€2 000\n\n€2 500\n\n\\-3 -2 -1 0 1 2 3 4 5 6 7 8\n\nOutcome year\n\nA. By sex\n\nTotal Women Men\n\n\\-€500\n\n€0\n\n€500\n\n€1 000\n\n€1 500\n\n€2 000\n\n€2 500\n\n\\-3 -2 -1 0 1 2 3 4 5 6 7 8\n\nOutcome year\n\nC. By age\n\nUnder 30 30 to 50\n\nOver 50 Total\n\n\\-€500\n\n€0\n\n€500\n\n€1 000\n\n€1 500\n\n€2 000\n\n€2 500\n\n\\-3 -2 -1 0 1 2 3 4 5 6 7 8\n\nOutcome year\n\nB. By unemployment\n\nduration One to three years\n\nMore than three years","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/employment_and_social_developments_in_europe_esde_2024_report_p87.pdf"]},"response":{"errors":[],"extract_id":"extract_ba023e0ed12b6226fb3d9187a1fb7fc2","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Chapter 3: The role of social investment\n\n91\n\nBox 3.6: Evaluation of job creation schemes in Ireland\n\n**In Ireland, two direct job creation schemes were found to have positive effects on earning outcomes for (mostly long-term unemployed) participants.** The research found that, after initial lock-in effects, the Community Employment (CE) scheme, ( ¹ ) which primarily targets people who have been unemployed for more than 12 months, has a positive impact on total annual earnings, raising the probability of any earnings in a given year and annual weeks of employment in the regular labour market. More specifically, CE participants earn around EUR 2 000 per year more than matched non-participants and work an average of four weeks longer per year. The Tús (Start) scheme, a community work placement scheme offering short-term work opportunities for long-term unemployed people, was also found to have a positive, albeit modest, impact on the number of jobseekers exiting the unemployment register, as well as a lasting positive effect on earnings (Chart 1). ( ² )\n\nChart 1 Women and younger jobseekers experience a greater boost in earnings following participation in Ireland’s Tús scheme Impact of Tús on earnings, by sex, unemployment duration and age\n\n_Source:_ (OECD, 2024a) calculations based on Department of Social Protection administrative data. **The two schemes have positive impacts on social outcomes, with different effects on specific groups of long-term unemployed people.** Participation in CE reduces future reliance on disability allowance and enhances take-up of education subsidies, indicating that the scheme has possible broader health and engagement in education effects. Six years after starting CE, former participants are 6 pp less likely to receive disability allowance, compared to similar participants who did not enter the scheme. This impact is even stronger for older jobseekers. Younger CE participants seem to benefit more in respect of labour market outcomes. For the Tús scheme, women, and jobseekers under 30 experienced a slightly greater increase in their earnings (Chart 1).\n\n( ¹ ) CE is a public work programme and is one of the most widely used ALMPs for long-term unemployed people in Ireland. Its main\n\nobjectives are to connect jobseekers with the labour market to increase employment levels and to promote social inclusion, as it seeks to reduce social isolation and social barriers for jobseekers.\n\n( ² ) (OECD, 2024a)\n\n\\-€500\n\n€0\n\n€500\n\n€1 000\n\n€1 500\n\n€2 000\n\n€2 500\n\n\\-3 -2 -1 0 1 2 3 4 5 6 7 8\n\nOutcome year\n\nA. By sex\n\nTotal Women Men\n\n\\-€500\n\n€0\n\n€500\n\n€1 000\n\n€1 500\n\n€2 000\n\n€2 500\n\n\\-3 -2 -1 0 1 2 3 4 5 6 7 8\n\nOutcome year\n\nC. By age\n\nUnder 30 30 to 50\n\nOver 50 Total\n\n\\-€500\n\n€0\n\n€500\n\n€1 000\n\n€1 500\n\n€2 000\n\n€2 500\n\n\\-3 -2 -1 0 1 2 3 4 5 6 7 8\n\nOutcome year\n\nB. By unemployment\n\nduration One to three years\n\nMore than three years"],"full_content":"Chapter 3: The role of social investment\n\n91\n\nBox 3.6: Evaluation of job creation schemes in Ireland\n\n**In Ireland, two direct job creation schemes were found to have positive effects on earning outcomes for (mostly long-term unemployed) participants.** The research found that, after initial lock-in effects, the Community Employment (CE) scheme, ( ¹ ) which primarily targets people who have been unemployed for more than 12 months, has a positive impact on total annual earnings, raising the probability of any earnings in a given year and annual weeks of employment in the regular labour market. More specifically, CE participants earn around EUR 2 000 per year more than matched non-participants and work an average of four weeks longer per year. The Tús (Start) scheme, a community work placement scheme offering short-term work opportunities for long-term unemployed people, was also found to have a positive, albeit modest, impact on the number of jobseekers exiting the unemployment register, as well as a lasting positive effect on earnings (Chart 1). ( ² )\n\nChart 1 Women and younger jobseekers experience a greater boost in earnings following participation in Ireland’s Tús scheme Impact of Tús on earnings, by sex, unemployment duration and age\n\n_Source:_ (OECD, 2024a) calculations based on Department of Social Protection administrative data. **The two schemes have positive impacts on social outcomes, with different effects on specific groups of long-term unemployed people.** Participation in CE reduces future reliance on disability allowance and enhances take-up of education subsidies, indicating that the scheme has possible broader health and engagement in education effects. Six years after starting CE, former participants are 6 pp less likely to receive disability allowance, compared to similar participants who did not enter the scheme. This impact is even stronger for older jobseekers. Younger CE participants seem to benefit more in respect of labour market outcomes. For the Tús scheme, women, and jobseekers under 30 experienced a slightly greater increase in their earnings (Chart 1).\n\n( ¹ ) CE is a public work programme and is one of the most widely used ALMPs for long-term unemployed people in Ireland. Its main\n\nobjectives are to connect jobseekers with the labour market to increase employment levels and to promote social inclusion, as it seeks to reduce social isolation and social barriers for jobseekers.\n\n( ² ) (OECD, 2024a)\n\n\\-€500\n\n€0\n\n€500\n\n€1 000\n\n€1 500\n\n€2 000\n\n€2 500\n\n\\-3 -2 -1 0 1 2 3 4 5 6 7 8\n\nOutcome year\n\nA. By sex\n\nTotal Women Men\n\n\\-€500\n\n€0\n\n€500\n\n€1 000\n\n€1 500\n\n€2 000\n\n€2 500\n\n\\-3 -2 -1 0 1 2 3 4 5 6 7 8\n\nOutcome year\n\nC. By age\n\nUnder 30 30 to 50\n\nOver 50 Total\n\n\\-€500\n\n€0\n\n€500\n\n€1 000\n\n€1 500\n\n€2 000\n\n€2 500\n\n\\-3 -2 -1 0 1 2 3 4 5 6 7 8\n\nOutcome year\n\nB. By unemployment\n\nduration One to three years\n\nMore than three years","publish_date":null,"title":"Box 3.6: Evaluation of job creation schemes in Ireland","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/employment_and_social_developments_in_europe_esde_2024_report_p87.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/employment_and_social_developments_in_europe_esde_2024_report_p87.pdf"},"request":{"example_id":"chart/employment_and_social_developments_in_europe_esde_2024_report_p87","product_type":"parse","source_file_path":"employment_and_social_developments_in_europe_esde_2024_report_p87.pdf"},"started_at":"2026-11-04T19:45:36.281814"} +{"completed_at":"2026-11-04T19:45:39.084670","latency_in_ms":1429,"output":{"example_id":"chart/employment_and_social_developments_in_europe_esde_2024_report_p89","grounded_pages":[],"layout_pages":[],"markdown":"Chapter 3: The role of social investment\n\n93\n\nChart 3.14 Share of housing costs in household disposable income declined but house prices increased sharply between 2014 and 2023 Weighted EU average share of housing costs in household disposable income and variation across countries (in standard deviation, left chart), and house price index and rent index (right chart), 2014-2023\n\n_Note:_ In the left chart, 2021 data exclude France, which did not report housing costs in 2021. Standard deviation is a measure of cross-country variation, the higher the standard deviation, the higher the cross country variation.\n\n_Source:_ Eurostat [ilc\\_mded01], [prc\\_hpi\\_a] and [prc\\_hicp\\_aind]. Click here to download chart.\n\n**Housing policies can help improve housing affordability.** At national level, housing policies can include the provision of social housing, housing allowances, tax deduction for housing costs, and utility subsidies, as well as rent regulation or subsidies for improving energy efficiency. ( ²⁰⁰ ) These policies often target specific population groups (e.g. low-income households, tenants vs homeowners) and can have other objectives, such as improving the quality of housing or increasing homeownership. Some housing policies, in particular housing allowances, need to be targeted to help contain the fiscal cost and ensure they are not translated into higher housing or rental prices. The impact of housing policies on housing cost reduction depends on many factors. These include their generosity, coverage and take-up, regulatory environment, stock of housing supply and requirements for renovation, as well as employment, environmental, urban and spatial policies **.** The analysis below examines how two housing policies, - housing allowances and social housing - contribute to poverty reduction. These two policies have been selected on the grounds of data availability. Housing allowances are intended to compensate for housing costs based on a means-test and can be granted to both tenants and owner-occupiers. ( ²⁰¹ ) As a proxy for social housing, this report uses social rent subsidies for tenants, thereby capturing the difference between social rent and estimated market rent for the dwelling. ( ²⁰² )\n\n**Housing policies are relevant for the labour market outcomes of individuals.** The lack of affordable housing may affect the accessibility to good quality education, training and job opportunities, and discourage labour mobility, increasing the probability of higher unemployment rates and contributing to labour and skills shortages. ( ²⁰³ ) Living in poor quality accommodation can negatively affect health and well-being, likely reducing productivity. In turn, worse labour market outcomes (e.g. being unemployed, frequently moving in and out of employment, working on a fixed contract, receiving a lower wage) might prevent housing conditions from improving in the absence of buffers. For example, they might reduce the ability to pay rents and the chances of obtaining a mortgage, especially in more expensive areas with better job opportunities, limiting mobility, or not providing sufficient means to improve housing quality. ( ²⁰⁴ ) The design of housing policies and their complementarities to other policies also matter. For instance, in the presence of high transaction costs, housing policies supporting home ownership might reduce labour mobility, increase the acceptance of lower wages or increase demand.\n\n**Increasing the supply of housing, including by investing in social housing, is very often essential to improve housing affordability, especially for households with lower incomes.** Social dwellings are usually publicly owned (with countries investing in the stock of social housing, enabling them to benefit from an economic rent accruing to owners of property) or semi-privately owned (by housing corporations) but can also be privately owned. ( ²⁰⁵ ) Social housing usually takes the form of rental accommodation provided at below-market prices. In recent decades, social housing models in Europe have increasingly focused on lower-income\n\n( ²⁰⁰ ) See (Eurofound, 2023b) for a more detailed mapping of housing policies in the Member States. ( ²⁰¹ ) Housing allowances compensate for housing costs, such as rent, gas, electricity, heating, water or utility bills. They can be granted to both tenants and owner-occupiers, and do not include social housing policy organised through the tax-benefit system and capital transfers (notably investment grants).\n\n( ²⁰² ) Social rent subsidies (as a proxy for social housing) are estimated using the imputed rents variable (HY030G) for social tenants (those who indicated paying below-the-market rent) from EU-SILC. The imputed value is the equivalent market rent that would be paid for a similar dwelling to that occupied, reduced by any rent actually paid.\n\n( ²⁰³ ) (Borg and Brandén, 2018). ( ²⁰⁴ ) (Arundel and John, 2017). ( ²⁰⁵ ) (Eurofound, 2023b) ; (European Commission, 2023f).\n\n90\n\n110\n\n130\n\n150\n\n170\n\n2014 2015 2016 2017 2018 2019 2020 2021 2022 2023\n\nIndex (2015=100)\n\nHouse prices and rents\n\nHouse prices Rents\n\n0\n\n2\n\n4\n\n6\n\n8\n\n0%\n\n10%\n\n20%\n\n30%\n\n40%\n\n2014 2015 2016 2017 2018 2019 2020 2021 2022 2023\n\nStandard deviation\n\nShare of housing costs\n\nEU average (left axis) Variation across countries (right axis)","pages":[{"line_numbers":[],"markdown":"Chapter 3: The role of social investment\n\n93\n\nChart 3.14 Share of housing costs in household disposable income declined but house prices increased sharply between 2014 and 2023 Weighted EU average share of housing costs in household disposable income and variation across countries (in standard deviation, left chart), and house price index and rent index (right chart), 2014-2023\n\n_Note:_ In the left chart, 2021 data exclude France, which did not report housing costs in 2021. Standard deviation is a measure of cross-country variation, the higher the standard deviation, the higher the cross country variation.\n\n_Source:_ Eurostat [ilc\\_mded01], [prc\\_hpi\\_a] and [prc\\_hicp\\_aind]. Click here to download chart.\n\n**Housing policies can help improve housing affordability.** At national level, housing policies can include the provision of social housing, housing allowances, tax deduction for housing costs, and utility subsidies, as well as rent regulation or subsidies for improving energy efficiency. ( ²⁰⁰ ) These policies often target specific population groups (e.g. low-income households, tenants vs homeowners) and can have other objectives, such as improving the quality of housing or increasing homeownership. Some housing policies, in particular housing allowances, need to be targeted to help contain the fiscal cost and ensure they are not translated into higher housing or rental prices. The impact of housing policies on housing cost reduction depends on many factors. These include their generosity, coverage and take-up, regulatory environment, stock of housing supply and requirements for renovation, as well as employment, environmental, urban and spatial policies **.** The analysis below examines how two housing policies, - housing allowances and social housing - contribute to poverty reduction. These two policies have been selected on the grounds of data availability. Housing allowances are intended to compensate for housing costs based on a means-test and can be granted to both tenants and owner-occupiers. ( ²⁰¹ ) As a proxy for social housing, this report uses social rent subsidies for tenants, thereby capturing the difference between social rent and estimated market rent for the dwelling. ( ²⁰² )\n\n**Housing policies are relevant for the labour market outcomes of individuals.** The lack of affordable housing may affect the accessibility to good quality education, training and job opportunities, and discourage labour mobility, increasing the probability of higher unemployment rates and contributing to labour and skills shortages. ( ²⁰³ ) Living in poor quality accommodation can negatively affect health and well-being, likely reducing productivity. In turn, worse labour market outcomes (e.g. being unemployed, frequently moving in and out of employment, working on a fixed contract, receiving a lower wage) might prevent housing conditions from improving in the absence of buffers. For example, they might reduce the ability to pay rents and the chances of obtaining a mortgage, especially in more expensive areas with better job opportunities, limiting mobility, or not providing sufficient means to improve housing quality. ( ²⁰⁴ ) The design of housing policies and their complementarities to other policies also matter. For instance, in the presence of high transaction costs, housing policies supporting home ownership might reduce labour mobility, increase the acceptance of lower wages or increase demand.\n\n**Increasing the supply of housing, including by investing in social housing, is very often essential to improve housing affordability, especially for households with lower incomes.** Social dwellings are usually publicly owned (with countries investing in the stock of social housing, enabling them to benefit from an economic rent accruing to owners of property) or semi-privately owned (by housing corporations) but can also be privately owned. ( ²⁰⁵ ) Social housing usually takes the form of rental accommodation provided at below-market prices. In recent decades, social housing models in Europe have increasingly focused on lower-income\n\n( ²⁰⁰ ) See (Eurofound, 2023b) for a more detailed mapping of housing policies in the Member States. ( ²⁰¹ ) Housing allowances compensate for housing costs, such as rent, gas, electricity, heating, water or utility bills. They can be granted to both tenants and owner-occupiers, and do not include social housing policy organised through the tax-benefit system and capital transfers (notably investment grants).\n\n( ²⁰² ) Social rent subsidies (as a proxy for social housing) are estimated using the imputed rents variable (HY030G) for social tenants (those who indicated paying below-the-market rent) from EU-SILC. The imputed value is the equivalent market rent that would be paid for a similar dwelling to that occupied, reduced by any rent actually paid.\n\n( ²⁰³ ) (Borg and Brandén, 2018). ( ²⁰⁴ ) (Arundel and John, 2017). ( ²⁰⁵ ) (Eurofound, 2023b) ; (European Commission, 2023f).\n\n90\n\n110\n\n130\n\n150\n\n170\n\n2014 2015 2016 2017 2018 2019 2020 2021 2022 2023\n\nIndex (2015=100)\n\nHouse prices and rents\n\nHouse prices Rents\n\n0\n\n2\n\n4\n\n6\n\n8\n\n0%\n\n10%\n\n20%\n\n30%\n\n40%\n\n2014 2015 2016 2017 2018 2019 2020 2021 2022 2023\n\nStandard deviation\n\nShare of housing costs\n\nEU average (left axis) Variation across countries (right axis)","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Chapter 3: The role of social investment\n\n93\n\nChart 3.14 Share of housing costs in household disposable income declined but house prices increased sharply between 2014 and 2023 Weighted EU average share of housing costs in household disposable income and variation across countries (in standard deviation, left chart), and house price index and rent index (right chart), 2014-2023\n\n_Note:_ In the left chart, 2021 data exclude France, which did not report housing costs in 2021. Standard deviation is a measure of cross-country variation, the higher the standard deviation, the higher the cross country variation.\n\n_Source:_ Eurostat [ilc\\_mded01], [prc\\_hpi\\_a] and [prc\\_hicp\\_aind]. Click here to download chart.\n\n**Housing policies can help improve housing affordability.** At national level, housing policies can include the provision of social housing, housing allowances, tax deduction for housing costs, and utility subsidies, as well as rent regulation or subsidies for improving energy efficiency. ( ²⁰⁰ ) These policies often target specific population groups (e.g. low-income households, tenants vs homeowners) and can have other objectives, such as improving the quality of housing or increasing homeownership. Some housing policies, in particular housing allowances, need to be targeted to help contain the fiscal cost and ensure they are not translated into higher housing or rental prices. The impact of housing policies on housing cost reduction depends on many factors. These include their generosity, coverage and take-up, regulatory environment, stock of housing supply and requirements for renovation, as well as employment, environmental, urban and spatial policies **.** The analysis below examines how two housing policies, - housing allowances and social housing - contribute to poverty reduction. These two policies have been selected on the grounds of data availability. Housing allowances are intended to compensate for housing costs based on a means-test and can be granted to both tenants and owner-occupiers. ( ²⁰¹ ) As a proxy for social housing, this report uses social rent subsidies for tenants, thereby capturing the difference between social rent and estimated market rent for the dwelling. ( ²⁰² )\n\n**Housing policies are relevant for the labour market outcomes of individuals.** The lack of affordable housing may affect the accessibility to good quality education, training and job opportunities, and discourage labour mobility, increasing the probability of higher unemployment rates and contributing to labour and skills shortages. ( ²⁰³ ) Living in poor quality accommodation can negatively affect health and well-being, likely reducing productivity. In turn, worse labour market outcomes (e.g. being unemployed, frequently moving in and out of employment, working on a fixed contract, receiving a lower wage) might prevent housing conditions from improving in the absence of buffers. For example, they might reduce the ability to pay rents and the chances of obtaining a mortgage, especially in more expensive areas with better job opportunities, limiting mobility, or not providing sufficient means to improve housing quality. ( ²⁰⁴ ) The design of housing policies and their complementarities to other policies also matter. For instance, in the presence of high transaction costs, housing policies supporting home ownership might reduce labour mobility, increase the acceptance of lower wages or increase demand.\n\n**Increasing the supply of housing, including by investing in social housing, is very often essential to improve housing affordability, especially for households with lower incomes.** Social dwellings are usually publicly owned (with countries investing in the stock of social housing, enabling them to benefit from an economic rent accruing to owners of property) or semi-privately owned (by housing corporations) but can also be privately owned. ( ²⁰⁵ ) Social housing usually takes the form of rental accommodation provided at below-market prices. In recent decades, social housing models in Europe have increasingly focused on lower-income\n\n( ²⁰⁰ ) See (Eurofound, 2023b) for a more detailed mapping of housing policies in the Member States. ( ²⁰¹ ) Housing allowances compensate for housing costs, such as rent, gas, electricity, heating, water or utility bills. They can be granted to both tenants and owner-occupiers, and do not include social housing policy organised through the tax-benefit system and capital transfers (notably investment grants).\n\n( ²⁰² ) Social rent subsidies (as a proxy for social housing) are estimated using the imputed rents variable (HY030G) for social tenants (those who indicated paying below-the-market rent) from EU-SILC. The imputed value is the equivalent market rent that would be paid for a similar dwelling to that occupied, reduced by any rent actually paid.\n\n( ²⁰³ ) (Borg and Brandén, 2018). ( ²⁰⁴ ) (Arundel and John, 2017). ( ²⁰⁵ ) (Eurofound, 2023b) ; (European Commission, 2023f).\n\n90\n\n110\n\n130\n\n150\n\n170\n\n2014 2015 2016 2017 2018 2019 2020 2021 2022 2023\n\nIndex (2015=100)\n\nHouse prices and rents\n\nHouse prices Rents\n\n0\n\n2\n\n4\n\n6\n\n8\n\n0%\n\n10%\n\n20%\n\n30%\n\n40%\n\n2014 2015 2016 2017 2018 2019 2020 2021 2022 2023\n\nStandard deviation\n\nShare of housing costs\n\nEU average (left axis) Variation across countries (right axis)","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/employment_and_social_developments_in_europe_esde_2024_report_p89.pdf"]},"response":{"errors":[],"extract_id":"extract_c51359288d22e21bf25fc5d9b089ccae","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Chapter 3: The role of social investment\n\n93\n\nChart 3.14 Share of housing costs in household disposable income declined but house prices increased sharply between 2014 and 2023 Weighted EU average share of housing costs in household disposable income and variation across countries (in standard deviation, left chart), and house price index and rent index (right chart), 2014-2023\n\n_Note:_ In the left chart, 2021 data exclude France, which did not report housing costs in 2021. Standard deviation is a measure of cross-country variation, the higher the standard deviation, the higher the cross country variation.\n\n_Source:_ Eurostat [ilc\\_mded01], [prc\\_hpi\\_a] and [prc\\_hicp\\_aind]. Click here to download chart.\n\n**Housing policies can help improve housing affordability.** At national level, housing policies can include the provision of social housing, housing allowances, tax deduction for housing costs, and utility subsidies, as well as rent regulation or subsidies for improving energy efficiency. ( ²⁰⁰ ) These policies often target specific population groups (e.g. low-income households, tenants vs homeowners) and can have other objectives, such as improving the quality of housing or increasing homeownership. Some housing policies, in particular housing allowances, need to be targeted to help contain the fiscal cost and ensure they are not translated into higher housing or rental prices. The impact of housing policies on housing cost reduction depends on many factors. These include their generosity, coverage and take-up, regulatory environment, stock of housing supply and requirements for renovation, as well as employment, environmental, urban and spatial policies **.** The analysis below examines how two housing policies, - housing allowances and social housing - contribute to poverty reduction. These two policies have been selected on the grounds of data availability. Housing allowances are intended to compensate for housing costs based on a means-test and can be granted to both tenants and owner-occupiers. ( ²⁰¹ ) As a proxy for social housing, this report uses social rent subsidies for tenants, thereby capturing the difference between social rent and estimated market rent for the dwelling. ( ²⁰² )\n\n**Housing policies are relevant for the labour market outcomes of individuals.** The lack of affordable housing may affect the accessibility to good quality education, training and job opportunities, and discourage labour mobility, increasing the probability of higher unemployment rates and contributing to labour and skills shortages. ( ²⁰³ ) Living in poor quality accommodation can negatively affect health and well-being, likely reducing productivity. In turn, worse labour market outcomes (e.g. being unemployed, frequently moving in and out of employment, working on a fixed contract, receiving a lower wage) might prevent housing conditions from improving in the absence of buffers. For example, they might reduce the ability to pay rents and the chances of obtaining a mortgage, especially in more expensive areas with better job opportunities, limiting mobility, or not providing sufficient means to improve housing quality. ( ²⁰⁴ ) The design of housing policies and their complementarities to other policies also matter. For instance, in the presence of high transaction costs, housing policies supporting home ownership might reduce labour mobility, increase the acceptance of lower wages or increase demand.\n\n**Increasing the supply of housing, including by investing in social housing, is very often essential to improve housing affordability, especially for households with lower incomes.** Social dwellings are usually publicly owned (with countries investing in the stock of social housing, enabling them to benefit from an economic rent accruing to owners of property) or semi-privately owned (by housing corporations) but can also be privately owned. ( ²⁰⁵ ) Social housing usually takes the form of rental accommodation provided at below-market prices. In recent decades, social housing models in Europe have increasingly focused on lower-income\n\n( ²⁰⁰ ) See (Eurofound, 2023b) for a more detailed mapping of housing policies in the Member States. ( ²⁰¹ ) Housing allowances compensate for housing costs, such as rent, gas, electricity, heating, water or utility bills. They can be granted to both tenants and owner-occupiers, and do not include social housing policy organised through the tax-benefit system and capital transfers (notably investment grants).\n\n( ²⁰² ) Social rent subsidies (as a proxy for social housing) are estimated using the imputed rents variable (HY030G) for social tenants (those who indicated paying below-the-market rent) from EU-SILC. The imputed value is the equivalent market rent that would be paid for a similar dwelling to that occupied, reduced by any rent actually paid.\n\n( ²⁰³ ) (Borg and Brandén, 2018). ( ²⁰⁴ ) (Arundel and John, 2017). ( ²⁰⁵ ) (Eurofound, 2023b) ; (European Commission, 2023f).\n\n90\n\n110\n\n130\n\n150\n\n170\n\n2014 2015 2016 2017 2018 2019 2020 2021 2022 2023\n\nIndex (2015=100)\n\nHouse prices and rents\n\nHouse prices Rents\n\n0\n\n2\n\n4\n\n6\n\n8\n\n0%\n\n10%\n\n20%\n\n30%\n\n40%\n\n2014 2015 2016 2017 2018 2019 2020 2021 2022 2023\n\nStandard deviation\n\nShare of housing costs\n\nEU average (left axis) Variation across countries (right axis)"],"full_content":"Chapter 3: The role of social investment\n\n93\n\nChart 3.14 Share of housing costs in household disposable income declined but house prices increased sharply between 2014 and 2023 Weighted EU average share of housing costs in household disposable income and variation across countries (in standard deviation, left chart), and house price index and rent index (right chart), 2014-2023\n\n_Note:_ In the left chart, 2021 data exclude France, which did not report housing costs in 2021. Standard deviation is a measure of cross-country variation, the higher the standard deviation, the higher the cross country variation.\n\n_Source:_ Eurostat [ilc\\_mded01], [prc\\_hpi\\_a] and [prc\\_hicp\\_aind]. Click here to download chart.\n\n**Housing policies can help improve housing affordability.** At national level, housing policies can include the provision of social housing, housing allowances, tax deduction for housing costs, and utility subsidies, as well as rent regulation or subsidies for improving energy efficiency. ( ²⁰⁰ ) These policies often target specific population groups (e.g. low-income households, tenants vs homeowners) and can have other objectives, such as improving the quality of housing or increasing homeownership. Some housing policies, in particular housing allowances, need to be targeted to help contain the fiscal cost and ensure they are not translated into higher housing or rental prices. The impact of housing policies on housing cost reduction depends on many factors. These include their generosity, coverage and take-up, regulatory environment, stock of housing supply and requirements for renovation, as well as employment, environmental, urban and spatial policies **.** The analysis below examines how two housing policies, - housing allowances and social housing - contribute to poverty reduction. These two policies have been selected on the grounds of data availability. Housing allowances are intended to compensate for housing costs based on a means-test and can be granted to both tenants and owner-occupiers. ( ²⁰¹ ) As a proxy for social housing, this report uses social rent subsidies for tenants, thereby capturing the difference between social rent and estimated market rent for the dwelling. ( ²⁰² )\n\n**Housing policies are relevant for the labour market outcomes of individuals.** The lack of affordable housing may affect the accessibility to good quality education, training and job opportunities, and discourage labour mobility, increasing the probability of higher unemployment rates and contributing to labour and skills shortages. ( ²⁰³ ) Living in poor quality accommodation can negatively affect health and well-being, likely reducing productivity. In turn, worse labour market outcomes (e.g. being unemployed, frequently moving in and out of employment, working on a fixed contract, receiving a lower wage) might prevent housing conditions from improving in the absence of buffers. For example, they might reduce the ability to pay rents and the chances of obtaining a mortgage, especially in more expensive areas with better job opportunities, limiting mobility, or not providing sufficient means to improve housing quality. ( ²⁰⁴ ) The design of housing policies and their complementarities to other policies also matter. For instance, in the presence of high transaction costs, housing policies supporting home ownership might reduce labour mobility, increase the acceptance of lower wages or increase demand.\n\n**Increasing the supply of housing, including by investing in social housing, is very often essential to improve housing affordability, especially for households with lower incomes.** Social dwellings are usually publicly owned (with countries investing in the stock of social housing, enabling them to benefit from an economic rent accruing to owners of property) or semi-privately owned (by housing corporations) but can also be privately owned. ( ²⁰⁵ ) Social housing usually takes the form of rental accommodation provided at below-market prices. In recent decades, social housing models in Europe have increasingly focused on lower-income\n\n( ²⁰⁰ ) See (Eurofound, 2023b) for a more detailed mapping of housing policies in the Member States. ( ²⁰¹ ) Housing allowances compensate for housing costs, such as rent, gas, electricity, heating, water or utility bills. They can be granted to both tenants and owner-occupiers, and do not include social housing policy organised through the tax-benefit system and capital transfers (notably investment grants).\n\n( ²⁰² ) Social rent subsidies (as a proxy for social housing) are estimated using the imputed rents variable (HY030G) for social tenants (those who indicated paying below-the-market rent) from EU-SILC. The imputed value is the equivalent market rent that would be paid for a similar dwelling to that occupied, reduced by any rent actually paid.\n\n( ²⁰³ ) (Borg and Brandén, 2018). ( ²⁰⁴ ) (Arundel and John, 2017). ( ²⁰⁵ ) (Eurofound, 2023b) ; (European Commission, 2023f).\n\n90\n\n110\n\n130\n\n150\n\n170\n\n2014 2015 2016 2017 2018 2019 2020 2021 2022 2023\n\nIndex (2015=100)\n\nHouse prices and rents\n\nHouse prices Rents\n\n0\n\n2\n\n4\n\n6\n\n8\n\n0%\n\n10%\n\n20%\n\n30%\n\n40%\n\n2014 2015 2016 2017 2018 2019 2020 2021 2022 2023\n\nStandard deviation\n\nShare of housing costs\n\nEU average (left axis) Variation across countries (right axis)","publish_date":null,"title":"Share of housing costs in household disposable income declined but house prices increased sharply between 2014 and 2023 At national","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/employment_and_social_developments_in_europe_esde_2024_report_p89.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/employment_and_social_developments_in_europe_esde_2024_report_p89.pdf"},"request":{"example_id":"chart/employment_and_social_developments_in_europe_esde_2024_report_p89","product_type":"parse","source_file_path":"employment_and_social_developments_in_europe_esde_2024_report_p89.pdf"},"started_at":"2026-11-04T19:45:37.655621"} +{"completed_at":"2026-11-04T19:44:48.111113","latency_in_ms":1963,"output":{"example_id":"chart/employment_and_social_developments_in_europe_esde_2024_report_p90","grounded_pages":[],"layout_pages":[],"markdown":"Employment and Social Developments in Europe 2024\n\n94\n\nhouseholds, primarily targeting older people and single-parents, while other vulnerable groups, such as single- person households, households without children, and migrants, are often low on the priority list or not eligible for social housing. ( ²⁰⁶ ) In addition to improving housing affordability, social housing can support macroeconomic objectives by easing supply bottlenecks, thereby mitigating house price pressures. This can in turn help to address labour shortages. ( ²⁰⁷ ) Social housing also plays a key role in housing first-type programmes for homeless people. In this respect, social and affordable housing policies underpin the EU social economy strategy. ( ²⁰⁸ ) More broadly, increasing the supply of housing can require a coherent mix of investments and reforms to address diverse issues such as the supply of basic infrastructure, spatial planning and other regulatory restrictions, or an insufficient supply of qualified labour.\n\n**The stock of social housing is quite low in the EU, with big investment gaps.** The lack of affordable, good quality housing has several structural drivers, including low incentives, bottlenecks in construction and investment gaps regarding in particular energy efficiency and social housing production. The construction of social housing has decreased across the EU in recent decades, and some countries, particularly in Eastern Europe, have privatised large parts of the social housing stock. ( ²⁰⁹ ) Currently, social housing as a proportion of the overall housing stock exceeds 20% in only three Member States: the Netherlands (34.1% in 2021), Austria (23.6% in 2019) and Denmark (21.3% in 2022). It is between 10% and 20% in France (14% in 2018), Ireland (12.7% in 2016) and Finland (10.9% in 2021). ( ²¹⁰ ) Long waiting lists are common in Member States with both larger and small social housing stocks, reducing the effectiveness of social housing in improving housing affordability for the most vulnerable households. ( ²¹¹ ) This issue is intensified by tenants who continue to occupy social housing after their income improves. While this chapter focuses on the effects of housing policies on tenants and owner- occupiers, supply-side policies such as investment in the construction and innovation of social housing form a key part of the broad range of housing policies that affect the availability and affordability of housing.\n\nChart 3.15 Public spending on housing increased but dispersion widened between 2014 and 2021 Weighted EU average government expenditure on housing and variation across countries (in standard deviation), 2014-2021\n\n_Note:_ Social housing is part of rent benefits and is included in the green line.\n\n_Source:_ Eurostat [spr\\_exp\\_fho].\n\nClick here to download chart.\n\n**Between 2014 and 2021, government expenditure on housing benefits per inhabitant, aimed at improving access to housing, increased in the EU.** Taking into account differences in the cost of living between countries, it went from 109 to 120 PPS per inhabitant ( ²¹² ) on average, ranging from less than 1 PPS per inhabitant in Bulgaria, Croatia and Portugal to over 300 PPS per inhabitant in Ireland and Finland in 2021. It was strongly driven by expenditure on rent benefits, mostly comprising housing allowances (86.2% of rent benefits in 2021), with the exception of Belgium, Estonia, Ireland, Lithuania, the Netherlands, Romania and Slovenia, where social housing expenditure constituted a major share of rent benefits. Countries with the lowest initial levels of public spending on housing caught up more strongly between 2014 and 2021, ( ²¹³ ) but the overall dispersion\n\n( ²⁰⁶ ) (Eurofound, 2023b); (Scanlon, Fernández Arrigoitia and Whitehead, 2015); (Heylen, 2024). ( ²⁰⁷ ) (Whitehead, 2017); (European Commission, 2023f). ( ²⁰⁸ ) (European Commission, 2021c). ( ²⁰⁹ ) (Scanlon, Fernández Arrigoitia and Whitehead, 2015) ; (Whitehead, 2017) ; (Eurofound, 2023b). ( ²¹⁰ ) See OECD Affordable Housing Database here. ( ²¹¹ ) (Eurofound, 2023b); (Pestel-Institut, 2024). ( ²¹² ) PPS is the artificial currency unit that eliminates price level differences between countries, i.e. one PPS can buy the same volume of goods and services in all countries. Government expenditure on housing in PPS is derived by dividing the government expenditure on housing of a country in national currency by the respective PPPs, which are obtained by comparing price levels for a basket of comparable goods and services representative of consumption patterns in the various countries.\n\n( ²¹³ ) Beta coefficient is negative (-0.03) and statistically significant at 5% significance level, primarily driven by the convergence in rent benefits (beta coefficient for benefits to owner occupiers is not statistically significant).\n\n0\n\n25\n\n50\n\n75\n\n100\n\n125\n\n150\n\n2014 2015 2016 2017 2018 2019 2020 2021\n\nPurchasing power standard (PPS)\n\nper inhabitant\n\nEU average\n\nTotal expenditure on housing Rent benefits Social housing (included in rent benefits) Benefits to owner occupiers\n\n0\n\n25\n\n50\n\n75\n\n100\n\n125\n\n2014 2015 2016 2017 2018 2019 2020 2021\n\nStandard deviation\n\nVariation across countries\n\nTotal expenditure on housing Rent benefits Social housing (included in rent benefits) Benefits to owner occupiers","pages":[{"line_numbers":[],"markdown":"Employment and Social Developments in Europe 2024\n\n94\n\nhouseholds, primarily targeting older people and single-parents, while other vulnerable groups, such as single- person households, households without children, and migrants, are often low on the priority list or not eligible for social housing. ( ²⁰⁶ ) In addition to improving housing affordability, social housing can support macroeconomic objectives by easing supply bottlenecks, thereby mitigating house price pressures. This can in turn help to address labour shortages. ( ²⁰⁷ ) Social housing also plays a key role in housing first-type programmes for homeless people. In this respect, social and affordable housing policies underpin the EU social economy strategy. ( ²⁰⁸ ) More broadly, increasing the supply of housing can require a coherent mix of investments and reforms to address diverse issues such as the supply of basic infrastructure, spatial planning and other regulatory restrictions, or an insufficient supply of qualified labour.\n\n**The stock of social housing is quite low in the EU, with big investment gaps.** The lack of affordable, good quality housing has several structural drivers, including low incentives, bottlenecks in construction and investment gaps regarding in particular energy efficiency and social housing production. The construction of social housing has decreased across the EU in recent decades, and some countries, particularly in Eastern Europe, have privatised large parts of the social housing stock. ( ²⁰⁹ ) Currently, social housing as a proportion of the overall housing stock exceeds 20% in only three Member States: the Netherlands (34.1% in 2021), Austria (23.6% in 2019) and Denmark (21.3% in 2022). It is between 10% and 20% in France (14% in 2018), Ireland (12.7% in 2016) and Finland (10.9% in 2021). ( ²¹⁰ ) Long waiting lists are common in Member States with both larger and small social housing stocks, reducing the effectiveness of social housing in improving housing affordability for the most vulnerable households. ( ²¹¹ ) This issue is intensified by tenants who continue to occupy social housing after their income improves. While this chapter focuses on the effects of housing policies on tenants and owner- occupiers, supply-side policies such as investment in the construction and innovation of social housing form a key part of the broad range of housing policies that affect the availability and affordability of housing.\n\nChart 3.15 Public spending on housing increased but dispersion widened between 2014 and 2021 Weighted EU average government expenditure on housing and variation across countries (in standard deviation), 2014-2021\n\n_Note:_ Social housing is part of rent benefits and is included in the green line.\n\n_Source:_ Eurostat [spr\\_exp\\_fho].\n\nClick here to download chart.\n\n**Between 2014 and 2021, government expenditure on housing benefits per inhabitant, aimed at improving access to housing, increased in the EU.** Taking into account differences in the cost of living between countries, it went from 109 to 120 PPS per inhabitant ( ²¹² ) on average, ranging from less than 1 PPS per inhabitant in Bulgaria, Croatia and Portugal to over 300 PPS per inhabitant in Ireland and Finland in 2021. It was strongly driven by expenditure on rent benefits, mostly comprising housing allowances (86.2% of rent benefits in 2021), with the exception of Belgium, Estonia, Ireland, Lithuania, the Netherlands, Romania and Slovenia, where social housing expenditure constituted a major share of rent benefits. Countries with the lowest initial levels of public spending on housing caught up more strongly between 2014 and 2021, ( ²¹³ ) but the overall dispersion\n\n( ²⁰⁶ ) (Eurofound, 2023b); (Scanlon, Fernández Arrigoitia and Whitehead, 2015); (Heylen, 2024). ( ²⁰⁷ ) (Whitehead, 2017); (European Commission, 2023f). ( ²⁰⁸ ) (European Commission, 2021c). ( ²⁰⁹ ) (Scanlon, Fernández Arrigoitia and Whitehead, 2015) ; (Whitehead, 2017) ; (Eurofound, 2023b). ( ²¹⁰ ) See OECD Affordable Housing Database here. ( ²¹¹ ) (Eurofound, 2023b); (Pestel-Institut, 2024). ( ²¹² ) PPS is the artificial currency unit that eliminates price level differences between countries, i.e. one PPS can buy the same volume of goods and services in all countries. Government expenditure on housing in PPS is derived by dividing the government expenditure on housing of a country in national currency by the respective PPPs, which are obtained by comparing price levels for a basket of comparable goods and services representative of consumption patterns in the various countries.\n\n( ²¹³ ) Beta coefficient is negative (-0.03) and statistically significant at 5% significance level, primarily driven by the convergence in rent benefits (beta coefficient for benefits to owner occupiers is not statistically significant).\n\n0\n\n25\n\n50\n\n75\n\n100\n\n125\n\n150\n\n2014 2015 2016 2017 2018 2019 2020 2021\n\nPurchasing power standard (PPS)\n\nper inhabitant\n\nEU average\n\nTotal expenditure on housing Rent benefits Social housing (included in rent benefits) Benefits to owner occupiers\n\n0\n\n25\n\n50\n\n75\n\n100\n\n125\n\n2014 2015 2016 2017 2018 2019 2020 2021\n\nStandard deviation\n\nVariation across countries\n\nTotal expenditure on housing Rent benefits Social housing (included in rent benefits) Benefits to owner occupiers","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Employment and Social Developments in Europe 2024\n\n94\n\nhouseholds, primarily targeting older people and single-parents, while other vulnerable groups, such as single- person households, households without children, and migrants, are often low on the priority list or not eligible for social housing. ( ²⁰⁶ ) In addition to improving housing affordability, social housing can support macroeconomic objectives by easing supply bottlenecks, thereby mitigating house price pressures. This can in turn help to address labour shortages. ( ²⁰⁷ ) Social housing also plays a key role in housing first-type programmes for homeless people. In this respect, social and affordable housing policies underpin the EU social economy strategy. ( ²⁰⁸ ) More broadly, increasing the supply of housing can require a coherent mix of investments and reforms to address diverse issues such as the supply of basic infrastructure, spatial planning and other regulatory restrictions, or an insufficient supply of qualified labour.\n\n**The stock of social housing is quite low in the EU, with big investment gaps.** The lack of affordable, good quality housing has several structural drivers, including low incentives, bottlenecks in construction and investment gaps regarding in particular energy efficiency and social housing production. The construction of social housing has decreased across the EU in recent decades, and some countries, particularly in Eastern Europe, have privatised large parts of the social housing stock. ( ²⁰⁹ ) Currently, social housing as a proportion of the overall housing stock exceeds 20% in only three Member States: the Netherlands (34.1% in 2021), Austria (23.6% in 2019) and Denmark (21.3% in 2022). It is between 10% and 20% in France (14% in 2018), Ireland (12.7% in 2016) and Finland (10.9% in 2021). ( ²¹⁰ ) Long waiting lists are common in Member States with both larger and small social housing stocks, reducing the effectiveness of social housing in improving housing affordability for the most vulnerable households. ( ²¹¹ ) This issue is intensified by tenants who continue to occupy social housing after their income improves. While this chapter focuses on the effects of housing policies on tenants and owner- occupiers, supply-side policies such as investment in the construction and innovation of social housing form a key part of the broad range of housing policies that affect the availability and affordability of housing.\n\nChart 3.15 Public spending on housing increased but dispersion widened between 2014 and 2021 Weighted EU average government expenditure on housing and variation across countries (in standard deviation), 2014-2021\n\n_Note:_ Social housing is part of rent benefits and is included in the green line.\n\n_Source:_ Eurostat [spr\\_exp\\_fho].\n\nClick here to download chart.\n\n**Between 2014 and 2021, government expenditure on housing benefits per inhabitant, aimed at improving access to housing, increased in the EU.** Taking into account differences in the cost of living between countries, it went from 109 to 120 PPS per inhabitant ( ²¹² ) on average, ranging from less than 1 PPS per inhabitant in Bulgaria, Croatia and Portugal to over 300 PPS per inhabitant in Ireland and Finland in 2021. It was strongly driven by expenditure on rent benefits, mostly comprising housing allowances (86.2% of rent benefits in 2021), with the exception of Belgium, Estonia, Ireland, Lithuania, the Netherlands, Romania and Slovenia, where social housing expenditure constituted a major share of rent benefits. Countries with the lowest initial levels of public spending on housing caught up more strongly between 2014 and 2021, ( ²¹³ ) but the overall dispersion\n\n( ²⁰⁶ ) (Eurofound, 2023b); (Scanlon, Fernández Arrigoitia and Whitehead, 2015); (Heylen, 2024). ( ²⁰⁷ ) (Whitehead, 2017); (European Commission, 2023f). ( ²⁰⁸ ) (European Commission, 2021c). ( ²⁰⁹ ) (Scanlon, Fernández Arrigoitia and Whitehead, 2015) ; (Whitehead, 2017) ; (Eurofound, 2023b). ( ²¹⁰ ) See OECD Affordable Housing Database here. ( ²¹¹ ) (Eurofound, 2023b); (Pestel-Institut, 2024). ( ²¹² ) PPS is the artificial currency unit that eliminates price level differences between countries, i.e. one PPS can buy the same volume of goods and services in all countries. Government expenditure on housing in PPS is derived by dividing the government expenditure on housing of a country in national currency by the respective PPPs, which are obtained by comparing price levels for a basket of comparable goods and services representative of consumption patterns in the various countries.\n\n( ²¹³ ) Beta coefficient is negative (-0.03) and statistically significant at 5% significance level, primarily driven by the convergence in rent benefits (beta coefficient for benefits to owner occupiers is not statistically significant).\n\n0\n\n25\n\n50\n\n75\n\n100\n\n125\n\n150\n\n2014 2015 2016 2017 2018 2019 2020 2021\n\nPurchasing power standard (PPS)\n\nper inhabitant\n\nEU average\n\nTotal expenditure on housing Rent benefits Social housing (included in rent benefits) Benefits to owner occupiers\n\n0\n\n25\n\n50\n\n75\n\n100\n\n125\n\n2014 2015 2016 2017 2018 2019 2020 2021\n\nStandard deviation\n\nVariation across countries\n\nTotal expenditure on housing Rent benefits Social housing (included in rent benefits) Benefits to owner occupiers","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/employment_and_social_developments_in_europe_esde_2024_report_p90.pdf"]},"response":{"errors":[],"extract_id":"extract_31caab777feb036b15b2d1aba21a29cb","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Employment and Social Developments in Europe 2024\n\n94\n\nhouseholds, primarily targeting older people and single-parents, while other vulnerable groups, such as single- person households, households without children, and migrants, are often low on the priority list or not eligible for social housing. ( ²⁰⁶ ) In addition to improving housing affordability, social housing can support macroeconomic objectives by easing supply bottlenecks, thereby mitigating house price pressures. This can in turn help to address labour shortages. ( ²⁰⁷ ) Social housing also plays a key role in housing first-type programmes for homeless people. In this respect, social and affordable housing policies underpin the EU social economy strategy. ( ²⁰⁸ ) More broadly, increasing the supply of housing can require a coherent mix of investments and reforms to address diverse issues such as the supply of basic infrastructure, spatial planning and other regulatory restrictions, or an insufficient supply of qualified labour.\n\n**The stock of social housing is quite low in the EU, with big investment gaps.** The lack of affordable, good quality housing has several structural drivers, including low incentives, bottlenecks in construction and investment gaps regarding in particular energy efficiency and social housing production. The construction of social housing has decreased across the EU in recent decades, and some countries, particularly in Eastern Europe, have privatised large parts of the social housing stock. ( ²⁰⁹ ) Currently, social housing as a proportion of the overall housing stock exceeds 20% in only three Member States: the Netherlands (34.1% in 2021), Austria (23.6% in 2019) and Denmark (21.3% in 2022). It is between 10% and 20% in France (14% in 2018), Ireland (12.7% in 2016) and Finland (10.9% in 2021). ( ²¹⁰ ) Long waiting lists are common in Member States with both larger and small social housing stocks, reducing the effectiveness of social housing in improving housing affordability for the most vulnerable households. ( ²¹¹ ) This issue is intensified by tenants who continue to occupy social housing after their income improves. While this chapter focuses on the effects of housing policies on tenants and owner- occupiers, supply-side policies such as investment in the construction and innovation of social housing form a key part of the broad range of housing policies that affect the availability and affordability of housing.\n\nChart 3.15 Public spending on housing increased but dispersion widened between 2014 and 2021 Weighted EU average government expenditure on housing and variation across countries (in standard deviation), 2014-2021\n\n_Note:_ Social housing is part of rent benefits and is included in the green line.\n\n_Source:_ Eurostat [spr\\_exp\\_fho].\n\nClick here to download chart.\n\n**Between 2014 and 2021, government expenditure on housing benefits per inhabitant, aimed at improving access to housing, increased in the EU.** Taking into account differences in the cost of living between countries, it went from 109 to 120 PPS per inhabitant ( ²¹² ) on average, ranging from less than 1 PPS per inhabitant in Bulgaria, Croatia and Portugal to over 300 PPS per inhabitant in Ireland and Finland in 2021. It was strongly driven by expenditure on rent benefits, mostly comprising housing allowances (86.2% of rent benefits in 2021), with the exception of Belgium, Estonia, Ireland, Lithuania, the Netherlands, Romania and Slovenia, where social housing expenditure constituted a major share of rent benefits. Countries with the lowest initial levels of public spending on housing caught up more strongly between 2014 and 2021, ( ²¹³ ) but the overall dispersion\n\n( ²⁰⁶ ) (Eurofound, 2023b); (Scanlon, Fernández Arrigoitia and Whitehead, 2015); (Heylen, 2024). ( ²⁰⁷ ) (Whitehead, 2017); (European Commission, 2023f). ( ²⁰⁸ ) (European Commission, 2021c). ( ²⁰⁹ ) (Scanlon, Fernández Arrigoitia and Whitehead, 2015) ; (Whitehead, 2017) ; (Eurofound, 2023b). ( ²¹⁰ ) See OECD Affordable Housing Database here. ( ²¹¹ ) (Eurofound, 2023b); (Pestel-Institut, 2024). ( ²¹² ) PPS is the artificial currency unit that eliminates price level differences between countries, i.e. one PPS can buy the same volume of goods and services in all countries. Government expenditure on housing in PPS is derived by dividing the government expenditure on housing of a country in national currency by the respective PPPs, which are obtained by comparing price levels for a basket of comparable goods and services representative of consumption patterns in the various countries.\n\n( ²¹³ ) Beta coefficient is negative (-0.03) and statistically significant at 5% significance level, primarily driven by the convergence in rent benefits (beta coefficient for benefits to owner occupiers is not statistically significant).\n\n0\n\n25\n\n50\n\n75\n\n100\n\n125\n\n150\n\n2014 2015 2016 2017 2018 2019 2020 2021\n\nPurchasing power standard (PPS)\n\nper inhabitant\n\nEU average\n\nTotal expenditure on housing Rent benefits Social housing (included in rent benefits) Benefits to owner occupiers\n\n0\n\n25\n\n50\n\n75\n\n100\n\n125\n\n2014 2015 2016 2017 2018 2019 2020 2021\n\nStandard deviation\n\nVariation across countries\n\nTotal expenditure on housing Rent benefits Social housing (included in rent benefits) Benefits to owner occupiers"],"full_content":"Employment and Social Developments in Europe 2024\n\n94\n\nhouseholds, primarily targeting older people and single-parents, while other vulnerable groups, such as single- person households, households without children, and migrants, are often low on the priority list or not eligible for social housing. ( ²⁰⁶ ) In addition to improving housing affordability, social housing can support macroeconomic objectives by easing supply bottlenecks, thereby mitigating house price pressures. This can in turn help to address labour shortages. ( ²⁰⁷ ) Social housing also plays a key role in housing first-type programmes for homeless people. In this respect, social and affordable housing policies underpin the EU social economy strategy. ( ²⁰⁸ ) More broadly, increasing the supply of housing can require a coherent mix of investments and reforms to address diverse issues such as the supply of basic infrastructure, spatial planning and other regulatory restrictions, or an insufficient supply of qualified labour.\n\n**The stock of social housing is quite low in the EU, with big investment gaps.** The lack of affordable, good quality housing has several structural drivers, including low incentives, bottlenecks in construction and investment gaps regarding in particular energy efficiency and social housing production. The construction of social housing has decreased across the EU in recent decades, and some countries, particularly in Eastern Europe, have privatised large parts of the social housing stock. ( ²⁰⁹ ) Currently, social housing as a proportion of the overall housing stock exceeds 20% in only three Member States: the Netherlands (34.1% in 2021), Austria (23.6% in 2019) and Denmark (21.3% in 2022). It is between 10% and 20% in France (14% in 2018), Ireland (12.7% in 2016) and Finland (10.9% in 2021). ( ²¹⁰ ) Long waiting lists are common in Member States with both larger and small social housing stocks, reducing the effectiveness of social housing in improving housing affordability for the most vulnerable households. ( ²¹¹ ) This issue is intensified by tenants who continue to occupy social housing after their income improves. While this chapter focuses on the effects of housing policies on tenants and owner- occupiers, supply-side policies such as investment in the construction and innovation of social housing form a key part of the broad range of housing policies that affect the availability and affordability of housing.\n\nChart 3.15 Public spending on housing increased but dispersion widened between 2014 and 2021 Weighted EU average government expenditure on housing and variation across countries (in standard deviation), 2014-2021\n\n_Note:_ Social housing is part of rent benefits and is included in the green line.\n\n_Source:_ Eurostat [spr\\_exp\\_fho].\n\nClick here to download chart.\n\n**Between 2014 and 2021, government expenditure on housing benefits per inhabitant, aimed at improving access to housing, increased in the EU.** Taking into account differences in the cost of living between countries, it went from 109 to 120 PPS per inhabitant ( ²¹² ) on average, ranging from less than 1 PPS per inhabitant in Bulgaria, Croatia and Portugal to over 300 PPS per inhabitant in Ireland and Finland in 2021. It was strongly driven by expenditure on rent benefits, mostly comprising housing allowances (86.2% of rent benefits in 2021), with the exception of Belgium, Estonia, Ireland, Lithuania, the Netherlands, Romania and Slovenia, where social housing expenditure constituted a major share of rent benefits. Countries with the lowest initial levels of public spending on housing caught up more strongly between 2014 and 2021, ( ²¹³ ) but the overall dispersion\n\n( ²⁰⁶ ) (Eurofound, 2023b); (Scanlon, Fernández Arrigoitia and Whitehead, 2015); (Heylen, 2024). ( ²⁰⁷ ) (Whitehead, 2017); (European Commission, 2023f). ( ²⁰⁸ ) (European Commission, 2021c). ( ²⁰⁹ ) (Scanlon, Fernández Arrigoitia and Whitehead, 2015) ; (Whitehead, 2017) ; (Eurofound, 2023b). ( ²¹⁰ ) See OECD Affordable Housing Database here. ( ²¹¹ ) (Eurofound, 2023b); (Pestel-Institut, 2024). ( ²¹² ) PPS is the artificial currency unit that eliminates price level differences between countries, i.e. one PPS can buy the same volume of goods and services in all countries. Government expenditure on housing in PPS is derived by dividing the government expenditure on housing of a country in national currency by the respective PPPs, which are obtained by comparing price levels for a basket of comparable goods and services representative of consumption patterns in the various countries.\n\n( ²¹³ ) Beta coefficient is negative (-0.03) and statistically significant at 5% significance level, primarily driven by the convergence in rent benefits (beta coefficient for benefits to owner occupiers is not statistically significant).\n\n0\n\n25\n\n50\n\n75\n\n100\n\n125\n\n150\n\n2014 2015 2016 2017 2018 2019 2020 2021\n\nPurchasing power standard (PPS)\n\nper inhabitant\n\nEU average\n\nTotal expenditure on housing Rent benefits Social housing (included in rent benefits) Benefits to owner occupiers\n\n0\n\n25\n\n50\n\n75\n\n100\n\n125\n\n2014 2015 2016 2017 2018 2019 2020 2021\n\nStandard deviation\n\nVariation across countries\n\nTotal expenditure on housing Rent benefits Social housing (included in rent benefits) Benefits to owner occupiers","publish_date":null,"title":"households, primarily targeting older people and single-parents, while other vulnerable groups, such as single- person households, households without children, and","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/employment_and_social_developments_in_europe_esde_2024_report_p90.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/employment_and_social_developments_in_europe_esde_2024_report_p90.pdf"},"request":{"example_id":"chart/employment_and_social_developments_in_europe_esde_2024_report_p90","product_type":"parse","source_file_path":"employment_and_social_developments_in_europe_esde_2024_report_p90.pdf"},"started_at":"2026-11-04T19:44:46.148100"} +{"completed_at":"2026-11-04T19:44:50.640337","latency_in_ms":2636,"output":{"example_id":"chart/employment_and_social_developments_in_europe_esde_2024_report_p91","grounded_pages":[],"layout_pages":[],"markdown":"Chapter 3: The role of social investment\n\n95\n\nacross countries widened (Chart 3.15). ( ²¹⁴ ) In light of recent trends, many Europeans responding to a recent survey believe that addressing the high cost of living (48%) and the lack of social housing and homelessness (23%) should be a main priority in their country. Around 6 in 10 respondents think that their governments should increase their spending on housing, with variation across countries ranging from below 30% in Finland and Denmark to over 80% in Cyprus, Ireland and Greece. ( ²¹⁵ )\n\nChart 3.16 Housing allowances reduce the AROP rate Differences in AROP and AROP rates before housing allowances, 2022\n\n_Note:_ AROP rate before housing allowances is calculated by excluding housing allowances from equivalised household disposable income and by calculating the share of people with an equivalised adjusted disposable income below 60% of the national median. AROP before housing allowances rate is missing for Romania, which does not collect data on housing allowances.\n\n_Source:_ EU-SILC scientific use files, 2022.\n\nClick here to download chart.\n\n**Social housing and housing allowances reduce the AROP rate in the short term, without accounting for the possible longer-term effects on housing prices and housing affordability.** Comparing the adjusted AROP rate, which excludes social rent subsidies and/or housing allowances from equivalised household disposable income, ( ²¹⁶ ) to the standard AROP rate provides a good measure of the impact of both policy instruments in reducing poverty. It does not however capture other relevant aspects of housing affordability such as supply and demand for affordable housing, underlying drivers or broader societal impacts. In 2022, housing allowances decreased the AROP rate by an estimated 1.4 pp (from 17.7% based on AROP rate before housing allowances to 16.3% of AROP rate) in the EU on average, with the biggest decreases in Finland (4.3 pp), Ireland (3.5 pp), France (3.1 pp) and Germany (3.0 pp) (Chart 3.16). ( ²¹⁷ ) Similarly, in 2019, the provision of social housing proxied through social rent subsidies decreased the AROP rate in the EU by an estimated 0.4 pp on average (from 18.1% in AROP rate before housing allowances and social rent subsidies to 17.7% in AROP rate before housing allowances). ( ²¹⁸ ) The highest decreases were observed in Ireland (3.7 pp) and Belgium (2.4 pp). The higher average impact of housing allowances could be due to relatively low public spending on social housing (Chart 3.15). While this analysis suggests that social housing and housing allowances reduce poverty, it does not account for the impact of other related policies or institutional setups and bottlenecks relevant for the functioning of rental and housing markets, thus likely reflecting the upper bound of the impact. For instance, the potential shift of (part of) housing allowances to landlords through higher rents might reduce the effectiveness\n\n( ²¹⁴ ) While countries with the lowest initial levels catch up relatively to those with the highest initial level of public spending on housing, changes in the public spending of other countries contributes to increased variance thus increasing overall dispersion.\n\n( ²¹⁵ ) 2024 Eurobarometer on Social Europe; 2022 Eurobarometer on Fairness, Inequality and Intergenerational Mobility. ( ²¹⁶ ) Share of people with an equivalised adjusted disposable income below 60% of the national median, and further referred to as AROP before housing allowances and social rent subsidies, and AROP before housing allowances, respectively.\n\n( ²¹⁷ ) Romania is excluded from both averages, as it does not collect data on housing allowances (variable HY070G in EU-SILC). The reported results assume that no housing allowances are shifted towards landlords through higher rents. The rather small impact of housing allowances in reducing the AROP rate in some countries is largely due to low spending on these policies. However, part of the differences in impact on poverty reduction between countries might also be driven by the design of housing allowances, with some countries targeting households at risk of poverty more effectively.\n\n( ²¹⁸ ) Social rent subsidies (as a proxy for social housing) are estimated using the imputed rents variable (HY030G) for social tenants (those who indicated paying below-the-market rent) from EU-SILC. Since 2021, this variable is no longer part of the annual EU-SILC but will be provided on a three-yearly basis as part of the module “labour and housing conditions” (starting in 2023). Some countries (Denmark, Germany, Malta, the Netherlands, Slovakia, Sweden) did not collect data on imputed rents for social tenants for most years considered and are excluded from the analysis on social housing, in addition to Romania (where housing allowances do not exist). While data for social rent subsidies (imputed rents) are available until 2020, data for 2019 were used to avoid estimating the impact of temporary measures adopted during the COVID-19 pandemic.\n\n5%\n\n10%\n\n15%\n\n20%\n\n25%\n\nAT BE BG CY CZ DE DK EE EL ES FI FR HR HU IE IT LT LU LV MT NL PL PT SE SI SK EU\n\n2022\n\nAROP AROP before housing allowances","pages":[{"line_numbers":[],"markdown":"Chapter 3: The role of social investment\n\n95\n\nacross countries widened (Chart 3.15). ( ²¹⁴ ) In light of recent trends, many Europeans responding to a recent survey believe that addressing the high cost of living (48%) and the lack of social housing and homelessness (23%) should be a main priority in their country. Around 6 in 10 respondents think that their governments should increase their spending on housing, with variation across countries ranging from below 30% in Finland and Denmark to over 80% in Cyprus, Ireland and Greece. ( ²¹⁵ )\n\nChart 3.16 Housing allowances reduce the AROP rate Differences in AROP and AROP rates before housing allowances, 2022\n\n_Note:_ AROP rate before housing allowances is calculated by excluding housing allowances from equivalised household disposable income and by calculating the share of people with an equivalised adjusted disposable income below 60% of the national median. AROP before housing allowances rate is missing for Romania, which does not collect data on housing allowances.\n\n_Source:_ EU-SILC scientific use files, 2022.\n\nClick here to download chart.\n\n**Social housing and housing allowances reduce the AROP rate in the short term, without accounting for the possible longer-term effects on housing prices and housing affordability.** Comparing the adjusted AROP rate, which excludes social rent subsidies and/or housing allowances from equivalised household disposable income, ( ²¹⁶ ) to the standard AROP rate provides a good measure of the impact of both policy instruments in reducing poverty. It does not however capture other relevant aspects of housing affordability such as supply and demand for affordable housing, underlying drivers or broader societal impacts. In 2022, housing allowances decreased the AROP rate by an estimated 1.4 pp (from 17.7% based on AROP rate before housing allowances to 16.3% of AROP rate) in the EU on average, with the biggest decreases in Finland (4.3 pp), Ireland (3.5 pp), France (3.1 pp) and Germany (3.0 pp) (Chart 3.16). ( ²¹⁷ ) Similarly, in 2019, the provision of social housing proxied through social rent subsidies decreased the AROP rate in the EU by an estimated 0.4 pp on average (from 18.1% in AROP rate before housing allowances and social rent subsidies to 17.7% in AROP rate before housing allowances). ( ²¹⁸ ) The highest decreases were observed in Ireland (3.7 pp) and Belgium (2.4 pp). The higher average impact of housing allowances could be due to relatively low public spending on social housing (Chart 3.15). While this analysis suggests that social housing and housing allowances reduce poverty, it does not account for the impact of other related policies or institutional setups and bottlenecks relevant for the functioning of rental and housing markets, thus likely reflecting the upper bound of the impact. For instance, the potential shift of (part of) housing allowances to landlords through higher rents might reduce the effectiveness\n\n( ²¹⁴ ) While countries with the lowest initial levels catch up relatively to those with the highest initial level of public spending on housing, changes in the public spending of other countries contributes to increased variance thus increasing overall dispersion.\n\n( ²¹⁵ ) 2024 Eurobarometer on Social Europe; 2022 Eurobarometer on Fairness, Inequality and Intergenerational Mobility. ( ²¹⁶ ) Share of people with an equivalised adjusted disposable income below 60% of the national median, and further referred to as AROP before housing allowances and social rent subsidies, and AROP before housing allowances, respectively.\n\n( ²¹⁷ ) Romania is excluded from both averages, as it does not collect data on housing allowances (variable HY070G in EU-SILC). The reported results assume that no housing allowances are shifted towards landlords through higher rents. The rather small impact of housing allowances in reducing the AROP rate in some countries is largely due to low spending on these policies. However, part of the differences in impact on poverty reduction between countries might also be driven by the design of housing allowances, with some countries targeting households at risk of poverty more effectively.\n\n( ²¹⁸ ) Social rent subsidies (as a proxy for social housing) are estimated using the imputed rents variable (HY030G) for social tenants (those who indicated paying below-the-market rent) from EU-SILC. Since 2021, this variable is no longer part of the annual EU-SILC but will be provided on a three-yearly basis as part of the module “labour and housing conditions” (starting in 2023). Some countries (Denmark, Germany, Malta, the Netherlands, Slovakia, Sweden) did not collect data on imputed rents for social tenants for most years considered and are excluded from the analysis on social housing, in addition to Romania (where housing allowances do not exist). While data for social rent subsidies (imputed rents) are available until 2020, data for 2019 were used to avoid estimating the impact of temporary measures adopted during the COVID-19 pandemic.\n\n5%\n\n10%\n\n15%\n\n20%\n\n25%\n\nAT BE BG CY CZ DE DK EE EL ES FI FR HR HU IE IT LT LU LV MT NL PL PT SE SI SK EU\n\n2022\n\nAROP AROP before housing allowances","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Chapter 3: The role of social investment\n\n95\n\nacross countries widened (Chart 3.15). ( ²¹⁴ ) In light of recent trends, many Europeans responding to a recent survey believe that addressing the high cost of living (48%) and the lack of social housing and homelessness (23%) should be a main priority in their country. Around 6 in 10 respondents think that their governments should increase their spending on housing, with variation across countries ranging from below 30% in Finland and Denmark to over 80% in Cyprus, Ireland and Greece. ( ²¹⁵ )\n\nChart 3.16 Housing allowances reduce the AROP rate Differences in AROP and AROP rates before housing allowances, 2022\n\n_Note:_ AROP rate before housing allowances is calculated by excluding housing allowances from equivalised household disposable income and by calculating the share of people with an equivalised adjusted disposable income below 60% of the national median. AROP before housing allowances rate is missing for Romania, which does not collect data on housing allowances.\n\n_Source:_ EU-SILC scientific use files, 2022.\n\nClick here to download chart.\n\n**Social housing and housing allowances reduce the AROP rate in the short term, without accounting for the possible longer-term effects on housing prices and housing affordability.** Comparing the adjusted AROP rate, which excludes social rent subsidies and/or housing allowances from equivalised household disposable income, ( ²¹⁶ ) to the standard AROP rate provides a good measure of the impact of both policy instruments in reducing poverty. It does not however capture other relevant aspects of housing affordability such as supply and demand for affordable housing, underlying drivers or broader societal impacts. In 2022, housing allowances decreased the AROP rate by an estimated 1.4 pp (from 17.7% based on AROP rate before housing allowances to 16.3% of AROP rate) in the EU on average, with the biggest decreases in Finland (4.3 pp), Ireland (3.5 pp), France (3.1 pp) and Germany (3.0 pp) (Chart 3.16). ( ²¹⁷ ) Similarly, in 2019, the provision of social housing proxied through social rent subsidies decreased the AROP rate in the EU by an estimated 0.4 pp on average (from 18.1% in AROP rate before housing allowances and social rent subsidies to 17.7% in AROP rate before housing allowances). ( ²¹⁸ ) The highest decreases were observed in Ireland (3.7 pp) and Belgium (2.4 pp). The higher average impact of housing allowances could be due to relatively low public spending on social housing (Chart 3.15). While this analysis suggests that social housing and housing allowances reduce poverty, it does not account for the impact of other related policies or institutional setups and bottlenecks relevant for the functioning of rental and housing markets, thus likely reflecting the upper bound of the impact. For instance, the potential shift of (part of) housing allowances to landlords through higher rents might reduce the effectiveness\n\n( ²¹⁴ ) While countries with the lowest initial levels catch up relatively to those with the highest initial level of public spending on housing, changes in the public spending of other countries contributes to increased variance thus increasing overall dispersion.\n\n( ²¹⁵ ) 2024 Eurobarometer on Social Europe; 2022 Eurobarometer on Fairness, Inequality and Intergenerational Mobility. ( ²¹⁶ ) Share of people with an equivalised adjusted disposable income below 60% of the national median, and further referred to as AROP before housing allowances and social rent subsidies, and AROP before housing allowances, respectively.\n\n( ²¹⁷ ) Romania is excluded from both averages, as it does not collect data on housing allowances (variable HY070G in EU-SILC). The reported results assume that no housing allowances are shifted towards landlords through higher rents. The rather small impact of housing allowances in reducing the AROP rate in some countries is largely due to low spending on these policies. However, part of the differences in impact on poverty reduction between countries might also be driven by the design of housing allowances, with some countries targeting households at risk of poverty more effectively.\n\n( ²¹⁸ ) Social rent subsidies (as a proxy for social housing) are estimated using the imputed rents variable (HY030G) for social tenants (those who indicated paying below-the-market rent) from EU-SILC. Since 2021, this variable is no longer part of the annual EU-SILC but will be provided on a three-yearly basis as part of the module “labour and housing conditions” (starting in 2023). Some countries (Denmark, Germany, Malta, the Netherlands, Slovakia, Sweden) did not collect data on imputed rents for social tenants for most years considered and are excluded from the analysis on social housing, in addition to Romania (where housing allowances do not exist). While data for social rent subsidies (imputed rents) are available until 2020, data for 2019 were used to avoid estimating the impact of temporary measures adopted during the COVID-19 pandemic.\n\n5%\n\n10%\n\n15%\n\n20%\n\n25%\n\nAT BE BG CY CZ DE DK EE EL ES FI FR HR HU IE IT LT LU LV MT NL PL PT SE SI SK EU\n\n2022\n\nAROP AROP before housing allowances","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/employment_and_social_developments_in_europe_esde_2024_report_p91.pdf"]},"response":{"errors":[],"extract_id":"extract_a7a8d69b23e008700c1e60813e3ebe6e","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Chapter 3: The role of social investment\n\n95\n\nacross countries widened (Chart 3.15). ( ²¹⁴ ) In light of recent trends, many Europeans responding to a recent survey believe that addressing the high cost of living (48%) and the lack of social housing and homelessness (23%) should be a main priority in their country. Around 6 in 10 respondents think that their governments should increase their spending on housing, with variation across countries ranging from below 30% in Finland and Denmark to over 80% in Cyprus, Ireland and Greece. ( ²¹⁵ )\n\nChart 3.16 Housing allowances reduce the AROP rate Differences in AROP and AROP rates before housing allowances, 2022\n\n_Note:_ AROP rate before housing allowances is calculated by excluding housing allowances from equivalised household disposable income and by calculating the share of people with an equivalised adjusted disposable income below 60% of the national median. AROP before housing allowances rate is missing for Romania, which does not collect data on housing allowances.\n\n_Source:_ EU-SILC scientific use files, 2022.\n\nClick here to download chart.\n\n**Social housing and housing allowances reduce the AROP rate in the short term, without accounting for the possible longer-term effects on housing prices and housing affordability.** Comparing the adjusted AROP rate, which excludes social rent subsidies and/or housing allowances from equivalised household disposable income, ( ²¹��� ) to the standard AROP rate provides a good measure of the impact of both policy instruments in reducing poverty. It does not however capture other relevant aspects of housing affordability such as supply and demand for affordable housing, underlying drivers or broader societal impacts. In 2022, housing allowances decreased the AROP rate by an estimated 1.4 pp (from 17.7% based on AROP rate before housing allowances to 16.3% of AROP rate) in the EU on average, with the biggest decreases in Finland (4.3 pp), Ireland (3.5 pp), France (3.1 pp) and Germany (3.0 pp) (Chart 3.16). ( ²¹⁷ ) Similarly, in 2019, the provision of social housing proxied through social rent subsidies decreased the AROP rate in the EU by an estimated 0.4 pp on average (from 18.1% in AROP rate before housing allowances and social rent subsidies to 17.7% in AROP rate before housing allowances). ( ²¹⁸ ) The highest decreases were observed in Ireland (3.7 pp) and Belgium (2.4 pp). The higher average impact of housing allowances could be due to relatively low public spending on social housing (Chart 3.15). While this analysis suggests that social housing and housing allowances reduce poverty, it does not account for the impact of other related policies or institutional setups and bottlenecks relevant for the functioning of rental and housing markets, thus likely reflecting the upper bound of the impact. For instance, the potential shift of (part of) housing allowances to landlords through higher rents might reduce the effectiveness\n\n( ²¹⁴ ) While countries with the lowest initial levels catch up relatively to those with the highest initial level of public spending on housing, changes in the public spending of other countries contributes to increased variance thus increasing overall dispersion.\n\n( ²¹⁵ ) 2024 Eurobarometer on Social Europe; 2022 Eurobarometer on Fairness, Inequality and Intergenerational Mobility. ( ²¹⁶ ) Share of people with an equivalised adjusted disposable income below 60% of the national median, and further referred to as AROP before housing allowances and social rent subsidies, and AROP before housing allowances, respectively.\n\n( ²¹⁷ ) Romania is excluded from both averages, as it does not collect data on housing allowances (variable HY070G in EU-SILC). The reported results assume that no housing allowances are shifted towards landlords through higher rents. The rather small impact of housing allowances in reducing the AROP rate in some countries is largely due to low spending on these policies. However, part of the differences in impact on poverty reduction between countries might also be driven by the design of housing allowances, with some countries targeting households at risk of poverty more effectively.\n\n( ²¹⁸ ) Social rent subsidies (as a proxy for social housing) are estimated using the imputed rents variable (HY030G) for social tenants (those who indicated paying below-the-market rent) from EU-SILC. Since 2021, this variable is no longer part of the annual EU-SILC but will be provided on a three-yearly basis as part of the module “labour and housing conditions” (starting in 2023). Some countries (Denmark, Germany, Malta, the Netherlands, Slovakia, Sweden) did not collect data on imputed rents for social tenants for most years considered and are excluded from the analysis on social housing, in addition to Romania (where housing allowances do not exist). While data for social rent subsidies (imputed rents) are available until 2020, data for 2019 were used to avoid estimating the impact of temporary measures adopted during the COVID-19 pandemic.\n\n5%\n\n10%\n\n15%\n\n20%\n\n25%\n\nAT BE BG CY CZ DE DK EE EL ES FI FR HR HU IE IT LT LU LV MT NL PL PT SE SI SK EU\n\n2022\n\nAROP AROP before housing allowances"],"full_content":"Chapter 3: The role of social investment\n\n95\n\nacross countries widened (Chart 3.15). ( ²¹⁴ ) In light of recent trends, many Europeans responding to a recent survey believe that addressing the high cost of living (48%) and the lack of social housing and homelessness (23%) should be a main priority in their country. Around 6 in 10 respondents think that their governments should increase their spending on housing, with variation across countries ranging from below 30% in Finland and Denmark to over 80% in Cyprus, Ireland and Greece. ( ²¹⁵ )\n\nChart 3.16 Housing allowances reduce the AROP rate Differences in AROP and AROP rates before housing allowances, 2022\n\n_Note:_ AROP rate before housing allowances is calculated by excluding housing allowances from equivalised household disposable income and by calculating the share of people with an equivalised adjusted disposable income below 60% of the national median. AROP before housing allowances rate is missing for Romania, which does not collect data on housing allowances.\n\n_Source:_ EU-SILC scientific use files, 2022.\n\nClick here to download chart.\n\n**Social housing and housing allowances reduce the AROP rate in the short term, without accounting for the possible longer-term effects on housing prices and housing affordability.** Comparing the adjusted AROP rate, which excludes social rent subsidies and/or housing allowances from equivalised household disposable income, ( ²¹⁶ ) to the standard AROP rate provides a good measure of the impact of both policy instruments in reducing poverty. It does not however capture other relevant aspects of housing affordability such as supply and demand for affordable housing, underlying drivers or broader societal impacts. In 2022, housing allowances decreased the AROP rate by an estimated 1.4 pp (from 17.7% based on AROP rate before housing allowances to 16.3% of AROP rate) in the EU on average, with the biggest decreases in Finland (4.3 pp), Ireland (3.5 pp), France (3.1 pp) and Germany (3.0 pp) (Chart 3.16). ( ²¹⁷ ) Similarly, in 2019, the provision of social housing proxied through social rent subsidies decreased the AROP rate in the EU by an estimated 0.4 pp on average (from 18.1% in AROP rate before housing allowances and social rent subsidies to 17.7% in AROP rate before housing allowances). ( ²¹⁸ ) The highest decreases were observed in Ireland (3.7 pp) and Belgium (2.4 pp). The higher average impact of housing allowances could be due to relatively low public spending on social housing (Chart 3.15). While this analysis suggests that social housing and housing allowances reduce poverty, it does not account for the impact of other related policies or institutional setups and bottlenecks relevant for the functioning of rental and housing markets, thus likely reflecting the upper bound of the impact. For instance, the potential shift of (part of) housing allowances to landlords through higher rents might reduce the effectiveness\n\n( ²¹⁴ ) While countries with the lowest initial levels catch up relatively to those with the highest initial level of public spending on housing, changes in the public spending of other countries contributes to increased variance thus increasing overall dispersion.\n\n( ²¹⁵ ) 2024 Eurobarometer on Social Europe; 2022 Eurobarometer on Fairness, Inequality and Intergenerational Mobility. ( ²¹⁶ ) Share of people with an equivalised adjusted disposable income below 60% of the national median, and further referred to as AROP before housing allowances and social rent subsidies, and AROP before housing allowances, respectively.\n\n( ²¹⁷ ) Romania is excluded from both averages, as it does not collect data on housing allowances (variable HY070G in EU-SILC). The reported results assume that no housing allowances are shifted towards landlords through higher rents. The rather small impact of housing allowances in reducing the AROP rate in some countries is largely due to low spending on these policies. However, part of the differences in impact on poverty reduction between countries might also be driven by the design of housing allowances, with some countries targeting households at risk of poverty more effectively.\n\n( ²¹⁸ ) Social rent subsidies (as a proxy for social housing) are estimated using the imputed rents variable (HY030G) for social tenants (those who indicated paying below-the-market rent) from EU-SILC. Since 2021, this variable is no longer part of the annual EU-SILC but will be provided on a three-yearly basis as part of the module “labour and housing conditions” (starting in 2023). Some countries (Denmark, Germany, Malta, the Netherlands, Slovakia, Sweden) did not collect data on imputed rents for social tenants for most years considered and are excluded from the analysis on social housing, in addition to Romania (where housing allowances do not exist). While data for social rent subsidies (imputed rents) are available until 2020, data for 2019 were used to avoid estimating the impact of temporary measures adopted during the COVID-19 pandemic.\n\n5%\n\n10%\n\n15%\n\n20%\n\n25%\n\nAT BE BG CY CZ DE DK EE EL ES FI FR HR HU IE IT LT LU LV MT NL PL PT SE SI SK EU\n\n2022\n\nAROP AROP before housing allowances","publish_date":null,"title":"across countries widened (Chart 3.15). ( ) In light of recent trends, many Europeans responding to a recent survey believe","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/employment_and_social_developments_in_europe_esde_2024_report_p91.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/employment_and_social_developments_in_europe_esde_2024_report_p91.pdf"},"request":{"example_id":"chart/employment_and_social_developments_in_europe_esde_2024_report_p91","product_type":"parse","source_file_path":"employment_and_social_developments_in_europe_esde_2024_report_p91.pdf"},"started_at":"2026-11-04T19:44:48.003699"} +{"completed_at":"2026-11-04T19:44:51.855826","latency_in_ms":2813,"output":{"example_id":"chart/employment_and_social_developments_in_europe_esde_2024_report_p93","grounded_pages":[],"layout_pages":[],"markdown":"Chapter 3: The role of social investment\n\n97\n\n_(Continued on the next page)_\n\nBox 3.7: Promoting upward social convergence and poverty reduction through social protection\n\n**Well-designed, inclusive social protection systems and social investment policies complement and reinforce one another as necessary components of modern welfare states.** Social protection systems provide a safety net that protects individuals from economic and social risks by helping them maintain at least a basic level of income and thus contributing to macroeconomic stabilisation. They can mitigate the impacts on poverty and economic growth by increasing income and consumption, particularly through countercyclical spending during economic downturns. Evidence also points to complementarities between social protection and social investment, with spending on ALMPs and ECEC tending to have more positive effects where total social protection expenditure is more generous. ( ¹ )\n\n**Income stabilisers have the potential to contain poverty in times of economic downturn.** The shock- absorption properties of effective tax-benefit systems can prevent diverging trends in inequality and social outcomes during recession and strengthen the resilience of a country. The analysis below ( ² ) uses EUROMOD to quantify how the overall tax-benefit system mitigates the transmission of a market income shock to disposable household incomes. It simulates a 5% hypothetical reduction in gross market income ( ³ ) and estimates the share of income loss absorbed by a country’s tax-benefit system. As no changes in labour market status or prices are considered, ( ⁴ ) income stabilisation is driven by increases in means-tested benefits and reductions in taxes ( ⁵ ) and lower social insurance contributions (SICs). Stabilisation properties are measured through an income stabilisation coefficient (ISC), which varies between 0% and 100%, with higher values pointing to stronger stabilisation (Box A3.3 in Annex).\n\nChart 1 Substantial variation in the extent and composition of income stabilisation after a 5% market income shock across Member States Country-level ISC, by policy instrument (%), 2022-2023 average\n\n_Source:_ JRC calculations based on EUROMOD, version I6.0+. **Tax-benefit systems can stabilise incomes, particularly through direct taxes yet with substantial heterogeneity across Member States.** On average, tax-benefit systems in the EU in 2022-2023 would have absorbed almost half of the simulated 5% market income shock, with the ISC averaging 41.4% across the Member States. ( ⁶ ) Direct taxes would have absorbed 29.4% of a market income shock in 2022-2023, followed by SICs (9.8%) and means-tested benefits (<0.5%). However, the variation between countries is substantial, ranging from 17% in Bulgaria to 58.9% in Belgium (Chart 1). Within the different tax-benefit components, direct taxes would have absorbed 29.4% of a market income shock in 2022-2023, followed by SICs (9.8%) and means-tested benefits (<0.5%). Also for the various components, considerable differences emerge between Member States. In five Member States, SICs (rather\n\n( ¹ ) (European Commission, 2016) ( ² ) While social protection addresses both the risk of economic downturn and risks that occur in the context of sickness or\n\nretirement, this section will focus on the role of social protection in stabilising incomes.\n\n( ³ ) Market income includes employment and self-employment incomes, investment and property incomes, pensions from individual\n\nprivate plans, and regular net inter-household transfers, all reported in gross terms.\n\n( ⁴ ) Transitions into unemployment or changes in indexation of benefits following an increase in prices would require additional\n\nassumptions and estimations and are outside the scope of this analysis. See Box A3.3 in the Annex.\n\n( ⁵ ) Due to the progressive nature of personal income taxation. ( ⁶ ) EU-level ISC computed by aggregating changes in market income and disposable incomes across countries. The result is an EU-\n\nlevel weighted average, with the shares of country-specific market income shock out of total market income shock as the weighting factor. EU-level results in this analysis are thus more influenced by larger countries.\n\n0%\n\n10%\n\n20%\n\n30%\n\n40%\n\n50%\n\n60%\n\n70%\n\nBG EE PL CZ CY HR MT HU LV SK EL ES FR RO SE SI IT EU27PT LT DE AT NL FI IE DK LU BE\n\nIncome Stabilisation coefficients by MS\n\nTaxes SICs Benefits","pages":[{"line_numbers":[],"markdown":"Chapter 3: The role of social investment\n\n97\n\n_(Continued on the next page)_\n\nBox 3.7: Promoting upward social convergence and poverty reduction through social protection\n\n**Well-designed, inclusive social protection systems and social investment policies complement and reinforce one another as necessary components of modern welfare states.** Social protection systems provide a safety net that protects individuals from economic and social risks by helping them maintain at least a basic level of income and thus contributing to macroeconomic stabilisation. They can mitigate the impacts on poverty and economic growth by increasing income and consumption, particularly through countercyclical spending during economic downturns. Evidence also points to complementarities between social protection and social investment, with spending on ALMPs and ECEC tending to have more positive effects where total social protection expenditure is more generous. ( ¹ )\n\n**Income stabilisers have the potential to contain poverty in times of economic downturn.** The shock- absorption properties of effective tax-benefit systems can prevent diverging trends in inequality and social outcomes during recession and strengthen the resilience of a country. The analysis below ( ² ) uses EUROMOD to quantify how the overall tax-benefit system mitigates the transmission of a market income shock to disposable household incomes. It simulates a 5% hypothetical reduction in gross market income ( ³ ) and estimates the share of income loss absorbed by a country’s tax-benefit system. As no changes in labour market status or prices are considered, ( ⁴ ) income stabilisation is driven by increases in means-tested benefits and reductions in taxes ( ⁵ ) and lower social insurance contributions (SICs). Stabilisation properties are measured through an income stabilisation coefficient (ISC), which varies between 0% and 100%, with higher values pointing to stronger stabilisation (Box A3.3 in Annex).\n\nChart 1 Substantial variation in the extent and composition of income stabilisation after a 5% market income shock across Member States Country-level ISC, by policy instrument (%), 2022-2023 average\n\n_Source:_ JRC calculations based on EUROMOD, version I6.0+. **Tax-benefit systems can stabilise incomes, particularly through direct taxes yet with substantial heterogeneity across Member States.** On average, tax-benefit systems in the EU in 2022-2023 would have absorbed almost half of the simulated 5% market income shock, with the ISC averaging 41.4% across the Member States. ( ⁶ ) Direct taxes would have absorbed 29.4% of a market income shock in 2022-2023, followed by SICs (9.8%) and means-tested benefits (<0.5%). However, the variation between countries is substantial, ranging from 17% in Bulgaria to 58.9% in Belgium (Chart 1). Within the different tax-benefit components, direct taxes would have absorbed 29.4% of a market income shock in 2022-2023, followed by SICs (9.8%) and means-tested benefits (<0.5%). Also for the various components, considerable differences emerge between Member States. In five Member States, SICs (rather\n\n( ¹ ) (European Commission, 2016) ( ² ) While social protection addresses both the risk of economic downturn and risks that occur in the context of sickness or\n\nretirement, this section will focus on the role of social protection in stabilising incomes.\n\n( ³ ) Market income includes employment and self-employment incomes, investment and property incomes, pensions from individual\n\nprivate plans, and regular net inter-household transfers, all reported in gross terms.\n\n( ⁴ ) Transitions into unemployment or changes in indexation of benefits following an increase in prices would require additional\n\nassumptions and estimations and are outside the scope of this analysis. See Box A3.3 in the Annex.\n\n( ⁵ ) Due to the progressive nature of personal income taxation. ( ⁶ ) EU-level ISC computed by aggregating changes in market income and disposable incomes across countries. The result is an EU-\n\nlevel weighted average, with the shares of country-specific market income shock out of total market income shock as the weighting factor. EU-level results in this analysis are thus more influenced by larger countries.\n\n0%\n\n10%\n\n20%\n\n30%\n\n40%\n\n50%\n\n60%\n\n70%\n\nBG EE PL CZ CY HR MT HU LV SK EL ES FR RO SE SI IT EU27PT LT DE AT NL FI IE DK LU BE\n\nIncome Stabilisation coefficients by MS\n\nTaxes SICs Benefits","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Chapter 3: The role of social investment\n\n97\n\n_(Continued on the next page)_\n\nBox 3.7: Promoting upward social convergence and poverty reduction through social protection\n\n**Well-designed, inclusive social protection systems and social investment policies complement and reinforce one another as necessary components of modern welfare states.** Social protection systems provide a safety net that protects individuals from economic and social risks by helping them maintain at least a basic level of income and thus contributing to macroeconomic stabilisation. They can mitigate the impacts on poverty and economic growth by increasing income and consumption, particularly through countercyclical spending during economic downturns. Evidence also points to complementarities between social protection and social investment, with spending on ALMPs and ECEC tending to have more positive effects where total social protection expenditure is more generous. ( ¹ )\n\n**Income stabilisers have the potential to contain poverty in times of economic downturn.** The shock- absorption properties of effective tax-benefit systems can prevent diverging trends in inequality and social outcomes during recession and strengthen the resilience of a country. The analysis below ( ² ) uses EUROMOD to quantify how the overall tax-benefit system mitigates the transmission of a market income shock to disposable household incomes. It simulates a 5% hypothetical reduction in gross market income ( ³ ) and estimates the share of income loss absorbed by a country’s tax-benefit system. As no changes in labour market status or prices are considered, ( ⁴ ) income stabilisation is driven by increases in means-tested benefits and reductions in taxes ( ⁵ ) and lower social insurance contributions (SICs). Stabilisation properties are measured through an income stabilisation coefficient (ISC), which varies between 0% and 100%, with higher values pointing to stronger stabilisation (Box A3.3 in Annex).\n\nChart 1 Substantial variation in the extent and composition of income stabilisation after a 5% market income shock across Member States Country-level ISC, by policy instrument (%), 2022-2023 average\n\n_Source:_ JRC calculations based on EUROMOD, version I6.0+. **Tax-benefit systems can stabilise incomes, particularly through direct taxes yet with substantial heterogeneity across Member States.** On average, tax-benefit systems in the EU in 2022-2023 would have absorbed almost half of the simulated 5% market income shock, with the ISC averaging 41.4% across the Member States. ( ⁶ ) Direct taxes would have absorbed 29.4% of a market income shock in 2022-2023, followed by SICs (9.8%) and means-tested benefits (<0.5%). However, the variation between countries is substantial, ranging from 17% in Bulgaria to 58.9% in Belgium (Chart 1). Within the different tax-benefit components, direct taxes would have absorbed 29.4% of a market income shock in 2022-2023, followed by SICs (9.8%) and means-tested benefits (<0.5%). Also for the various components, considerable differences emerge between Member States. In five Member States, SICs (rather\n\n( ¹ ) (European Commission, 2016) ( ² ) While social protection addresses both the risk of economic downturn and risks that occur in the context of sickness or\n\nretirement, this section will focus on the role of social protection in stabilising incomes.\n\n( ³ ) Market income includes employment and self-employment incomes, investment and property incomes, pensions from individual\n\nprivate plans, and regular net inter-household transfers, all reported in gross terms.\n\n( ⁴ ) Transitions into unemployment or changes in indexation of benefits following an increase in prices would require additional\n\nassumptions and estimations and are outside the scope of this analysis. See Box A3.3 in the Annex.\n\n( ⁵ ) Due to the progressive nature of personal income taxation. ( ⁶ ) EU-level ISC computed by aggregating changes in market income and disposable incomes across countries. The result is an EU-\n\nlevel weighted average, with the shares of country-specific market income shock out of total market income shock as the weighting factor. EU-level results in this analysis are thus more influenced by larger countries.\n\n0%\n\n10%\n\n20%\n\n30%\n\n40%\n\n50%\n\n60%\n\n70%\n\nBG EE PL CZ CY HR MT HU LV SK EL ES FR RO SE SI IT EU27PT LT DE AT NL FI IE DK LU BE\n\nIncome Stabilisation coefficients by MS\n\nTaxes SICs Benefits","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/employment_and_social_developments_in_europe_esde_2024_report_p93.pdf"]},"response":{"errors":[],"extract_id":"extract_c6014e8e99b50c0f6b680bee1c71372c","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Chapter 3: The role of social investment\n\n97\n\n_(Continued on the next page)_\n\nBox 3.7: Promoting upward social convergence and poverty reduction through social protection\n\n**Well-designed, inclusive social protection systems and social investment policies complement and reinforce one another as necessary components of modern welfare states.** Social protection systems provide a safety net that protects individuals from economic and social risks by helping them maintain at least a basic level of income and thus contributing to macroeconomic stabilisation. They can mitigate the impacts on poverty and economic growth by increasing income and consumption, particularly through countercyclical spending during economic downturns. Evidence also points to complementarities between social protection and social investment, with spending on ALMPs and ECEC tending to have more positive effects where total social protection expenditure is more generous. ( ¹ )\n\n**Income stabilisers have the potential to contain poverty in times of economic downturn.** The shock- absorption properties of effective tax-benefit systems can prevent diverging trends in inequality and social outcomes during recession and strengthen the resilience of a country. The analysis below ( ² ) uses EUROMOD to quantify how the overall tax-benefit system mitigates the transmission of a market income shock to disposable household incomes. It simulates a 5% hypothetical reduction in gross market income ( ³ ) and estimates the share of income loss absorbed by a country’s tax-benefit system. As no changes in labour market status or prices are considered, ( ⁴ ) income stabilisation is driven by increases in means-tested benefits and reductions in taxes ( ⁵ ) and lower social insurance contributions (SICs). Stabilisation properties are measured through an income stabilisation coefficient (ISC), which varies between 0% and 100%, with higher values pointing to stronger stabilisation (Box A3.3 in Annex).\n\nChart 1 Substantial variation in the extent and composition of income stabilisation after a 5% market income shock across Member States Country-level ISC, by policy instrument (%), 2022-2023 average\n\n_Source:_ JRC calculations based on EUROMOD, version I6.0+. **Tax-benefit systems can stabilise incomes, particularly through direct taxes yet with substantial heterogeneity across Member States.** On average, tax-benefit systems in the EU in 2022-2023 would have absorbed almost half of the simulated 5% market income shock, with the ISC averaging 41.4% across the Member States. ( ⁶ ) Direct taxes would have absorbed 29.4% of a market income shock in 2022-2023, followed by SICs (9.8%) and means-tested benefits (<0.5%). However, the variation between countries is substantial, ranging from 17% in Bulgaria to 58.9% in Belgium (Chart 1). Within the different tax-benefit components, direct taxes would have absorbed 29.4% of a market income shock in 2022-2023, followed by SICs (9.8%) and means-tested benefits (<0.5%). Also for the various components, considerable differences emerge between Member States. In five Member States, SICs (rather\n\n( ¹ ) (European Commission, 2016) ( ² ) While social protection addresses both the risk of economic downturn and risks that occur in the context of sickness or\n\nretirement, this section will focus on the role of social protection in stabilising incomes.\n\n( ³ ) Market income includes employment and self-employment incomes, investment and property incomes, pensions from individual\n\nprivate plans, and regular net inter-household transfers, all reported in gross terms.\n\n( ⁴ ) Transitions into unemployment or changes in indexation of benefits following an increase in prices would require additional\n\nassumptions and estimations and are outside the scope of this analysis. See Box A3.3 in the Annex.\n\n( ⁵ ) Due to the progressive nature of personal income taxation. ( ⁶ ) EU-level ISC computed by aggregating changes in market income and disposable incomes across countries. The result is an EU-\n\nlevel weighted average, with the shares of country-specific market income shock out of total market income shock as the weighting factor. EU-level results in this analysis are thus more influenced by larger countries.\n\n0%\n\n10%\n\n20%\n\n30%\n\n40%\n\n50%\n\n60%\n\n70%\n\nBG EE PL CZ CY HR MT HU LV SK EL ES FR RO SE SI IT EU27PT LT DE AT NL FI IE DK LU BE\n\nIncome Stabilisation coefficients by MS\n\nTaxes SICs Benefits"],"full_content":"Chapter 3: The role of social investment\n\n97\n\n_(Continued on the next page)_\n\nBox 3.7: Promoting upward social convergence and poverty reduction through social protection\n\n**Well-designed, inclusive social protection systems and social investment policies complement and reinforce one another as necessary components of modern welfare states.** Social protection systems provide a safety net that protects individuals from economic and social risks by helping them maintain at least a basic level of income and thus contributing to macroeconomic stabilisation. They can mitigate the impacts on poverty and economic growth by increasing income and consumption, particularly through countercyclical spending during economic downturns. Evidence also points to complementarities between social protection and social investment, with spending on ALMPs and ECEC tending to have more positive effects where total social protection expenditure is more generous. ( ¹ )\n\n**Income stabilisers have the potential to contain poverty in times of economic downturn.** The shock- absorption properties of effective tax-benefit systems can prevent diverging trends in inequality and social outcomes during recession and strengthen the resilience of a country. The analysis below ( ² ) uses EUROMOD to quantify how the overall tax-benefit system mitigates the transmission of a market income shock to disposable household incomes. It simulates a 5% hypothetical reduction in gross market income ( ³ ) and estimates the share of income loss absorbed by a country’s tax-benefit system. As no changes in labour market status or prices are considered, ( ⁴ ) income stabilisation is driven by increases in means-tested benefits and reductions in taxes ( ⁵ ) and lower social insurance contributions (SICs). Stabilisation properties are measured through an income stabilisation coefficient (ISC), which varies between 0% and 100%, with higher values pointing to stronger stabilisation (Box A3.3 in Annex).\n\nChart 1 Substantial variation in the extent and composition of income stabilisation after a 5% market income shock across Member States Country-level ISC, by policy instrument (%), 2022-2023 average\n\n_Source:_ JRC calculations based on EUROMOD, version I6.0+. **Tax-benefit systems can stabilise incomes, particularly through direct taxes yet with substantial heterogeneity across Member States.** On average, tax-benefit systems in the EU in 2022-2023 would have absorbed almost half of the simulated 5% market income shock, with the ISC averaging 41.4% across the Member States. ( ⁶ ) Direct taxes would have absorbed 29.4% of a market income shock in 2022-2023, followed by SICs (9.8%) and means-tested benefits (<0.5%). However, the variation between countries is substantial, ranging from 17% in Bulgaria to 58.9% in Belgium (Chart 1). Within the different tax-benefit components, direct taxes would have absorbed 29.4% of a market income shock in 2022-2023, followed by SICs (9.8%) and means-tested benefits (<0.5%). Also for the various components, considerable differences emerge between Member States. In five Member States, SICs (rather\n\n( ¹ ) (European Commission, 2016) ( ² ) While social protection addresses both the risk of economic downturn and risks that occur in the context of sickness or\n\nretirement, this section will focus on the role of social protection in stabilising incomes.\n\n( ³ ) Market income includes employment and self-employment incomes, investment and property incomes, pensions from individual\n\nprivate plans, and regular net inter-household transfers, all reported in gross terms.\n\n( ⁴ ) Transitions into unemployment or changes in indexation of benefits following an increase in prices would require additional\n\nassumptions and estimations and are outside the scope of this analysis. See Box A3.3 in the Annex.\n\n( ⁵ ) Due to the progressive nature of personal income taxation. ( ⁶ ) EU-level ISC computed by aggregating changes in market income and disposable incomes across countries. The result is an EU-\n\nlevel weighted average, with the shares of country-specific market income shock out of total market income shock as the weighting factor. EU-level results in this analysis are thus more influenced by larger countries.\n\n0%\n\n10%\n\n20%\n\n30%\n\n40%\n\n50%\n\n60%\n\n70%\n\nBG EE PL CZ CY HR MT HU LV SK EL ES FR RO SE SI IT EU27PT LT DE AT NL FI IE DK LU BE\n\nIncome Stabilisation coefficients by MS\n\nTaxes SICs Benefits","publish_date":null,"title":"Box 3.7: Promoting upward social convergence and poverty reduction through social protection","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/employment_and_social_developments_in_europe_esde_2024_report_p93.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/employment_and_social_developments_in_europe_esde_2024_report_p93.pdf"},"request":{"example_id":"chart/employment_and_social_developments_in_europe_esde_2024_report_p93","product_type":"parse","source_file_path":"employment_and_social_developments_in_europe_esde_2024_report_p93.pdf"},"started_at":"2026-11-04T19:44:49.042475"} +{"completed_at":"2026-11-04T19:44:51.856904","latency_in_ms":1214,"output":{"example_id":"chart/employment_and_social_developments_in_europe_esde_2024_report_p94","grounded_pages":[],"layout_pages":[],"markdown":"Employment and Social Developments in Europe 2024\n\n98\n\n_Box (continued)_\n\n### 5\\. CONCLUSIONS\n\n**Well-designed social investment policies can enhance productivity and competitiveness and have positive impacts on economic growth** **and fiscal sustainability, as well as employment, poverty reduction and social inclusion.** They contribute to more inclusive and environmentally sustainable economies and societies, underpinning upward social convergence, helping to advance the fair green and digital transitions, and adapting to demographic change in the EU. Simulations also highlight the potential for social investment to positively impact fiscal sustainability, with the resulting long-term GDP growth more than offsetting the initial cost of the measure analysed.\n\n**Measuring the exact returns on social investment policies is challenging.** Returns might only materialise in the medium to long term, or are not always easily captured in monetary terms. In addition, social investments might reinforce one another, including over individuals' life courses, and might also depend on the effectiveness of other policies, such as social protection. Changing skills needs, including in the context of the green and digital transitions, might necessitate additional social investment to fully reap the benefits of previous investment. While many methods are available to estimate the returns on social investment, they are often complex, require good quality (longitudinal) data and cannot account for all relevant elements at once. The efficiency of spending on social investment is also crucial. In this respect, ensuring that social investment is of high quality, well-designed and evidence-informed is key.\n\nthan direct taxes) represent the largest component of income stabilisation.( ⁷ ) The degree of income stabilisation provided by each component also varies by Member State, ranging from 6.3% in Romania to 45.6% in Denmark for direct taxes; from 0% in Denmark ( ⁸ ) to 32.4% in Romania for SICs; and from less than 0.5% in Hungary, Poland, Estonia and Latvia to 5.3% in France for means-tested benefits (Chart 1).\n\n**The degree and composition of income stabilisation after an income shock varies between income quintiles.** On average, incomes at the bottom (43.5%) and top (42.6%) of the income distribution are more cushioned against a market income shock than those in the middle (39.7%) (Chart 2). While increased benefits lead to a larger shock-absorption among households in the bottom income quintile (providing a stabilisation of 18.6%), the reduction in taxes due to the income loss among households in the top income quintile leads to an almost equally high degree of income stabilisation as for the bottom quintile, despite almost no increase in benefits (0.2% stabilisation). ( ⁹ ) Similarly, personal income taxes absorb an increasingly higher share of the market income shock for the higher the income quintile, increasing from 12.3% for households in the first quintile to as much as 35.3% for households in the fifth quintile.\n\nChart 2 Relative significance of taxes and benefits in absorbing a 5% market income shock varies substantially across the income distribution EU-level ISC, by income quintile and policy instrument (%) (2022/2023 average)\n\n_Source:_ JRC calculations based on EUROMOD, version I6.0+.\n\n( ⁷ ) Bulgaria, Croatia, Hungary, Romania, Slovakia. ( ⁸ ) Result for Denmark follows from the classification of labour market contributions in EUROMOD as taxes rather than SIC and\n\nfrom the fact that unemployment benefit contributions and supplementary labour market contributions do not depend directly on earnings.\n\n( ⁹ ) Chart A3.3 in the Annex presents the relative significance of taxes and benefits in absorbing a 5% market income shock, by\n\nincome quintile and Member State.\n\n0\\.0\n\n10\\.0\n\n20\\.0\n\n30\\.0\n\n40\\.0\n\n50\\.0\n\n1 2 3 4 5 Total\n\nEU - ISC 2022/2023\n\nTaxes Social insurance contributions Benefits","pages":[{"line_numbers":[],"markdown":"Employment and Social Developments in Europe 2024\n\n98\n\n_Box (continued)_\n\n### 5\\. CONCLUSIONS\n\n**Well-designed social investment policies can enhance productivity and competitiveness and have positive impacts on economic growth** **and fiscal sustainability, as well as employment, poverty reduction and social inclusion.** They contribute to more inclusive and environmentally sustainable economies and societies, underpinning upward social convergence, helping to advance the fair green and digital transitions, and adapting to demographic change in the EU. Simulations also highlight the potential for social investment to positively impact fiscal sustainability, with the resulting long-term GDP growth more than offsetting the initial cost of the measure analysed.\n\n**Measuring the exact returns on social investment policies is challenging.** Returns might only materialise in the medium to long term, or are not always easily captured in monetary terms. In addition, social investments might reinforce one another, including over individuals' life courses, and might also depend on the effectiveness of other policies, such as social protection. Changing skills needs, including in the context of the green and digital transitions, might necessitate additional social investment to fully reap the benefits of previous investment. While many methods are available to estimate the returns on social investment, they are often complex, require good quality (longitudinal) data and cannot account for all relevant elements at once. The efficiency of spending on social investment is also crucial. In this respect, ensuring that social investment is of high quality, well-designed and evidence-informed is key.\n\nthan direct taxes) represent the largest component of income stabilisation.( ⁷ ) The degree of income stabilisation provided by each component also varies by Member State, ranging from 6.3% in Romania to 45.6% in Denmark for direct taxes; from 0% in Denmark ( ⁸ ) to 32.4% in Romania for SICs; and from less than 0.5% in Hungary, Poland, Estonia and Latvia to 5.3% in France for means-tested benefits (Chart 1).\n\n**The degree and composition of income stabilisation after an income shock varies between income quintiles.** On average, incomes at the bottom (43.5%) and top (42.6%) of the income distribution are more cushioned against a market income shock than those in the middle (39.7%) (Chart 2). While increased benefits lead to a larger shock-absorption among households in the bottom income quintile (providing a stabilisation of 18.6%), the reduction in taxes due to the income loss among households in the top income quintile leads to an almost equally high degree of income stabilisation as for the bottom quintile, despite almost no increase in benefits (0.2% stabilisation). ( ⁹ ) Similarly, personal income taxes absorb an increasingly higher share of the market income shock for the higher the income quintile, increasing from 12.3% for households in the first quintile to as much as 35.3% for households in the fifth quintile.\n\nChart 2 Relative significance of taxes and benefits in absorbing a 5% market income shock varies substantially across the income distribution EU-level ISC, by income quintile and policy instrument (%) (2022/2023 average)\n\n_Source:_ JRC calculations based on EUROMOD, version I6.0+.\n\n( ⁷ ) Bulgaria, Croatia, Hungary, Romania, Slovakia. ( ⁸ ) Result for Denmark follows from the classification of labour market contributions in EUROMOD as taxes rather than SIC and\n\nfrom the fact that unemployment benefit contributions and supplementary labour market contributions do not depend directly on earnings.\n\n( ⁹ ) Chart A3.3 in the Annex presents the relative significance of taxes and benefits in absorbing a 5% market income shock, by\n\nincome quintile and Member State.\n\n0\\.0\n\n10\\.0\n\n20\\.0\n\n30\\.0\n\n40\\.0\n\n50\\.0\n\n1 2 3 4 5 Total\n\nEU - ISC 2022/2023\n\nTaxes Social insurance contributions Benefits","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Employment and Social Developments in Europe 2024\n\n98\n\n_Box (continued)_\n\n### 5\\. CONCLUSIONS\n\n**Well-designed social investment policies can enhance productivity and competitiveness and have positive impacts on economic growth** **and fiscal sustainability, as well as employment, poverty reduction and social inclusion.** They contribute to more inclusive and environmentally sustainable economies and societies, underpinning upward social convergence, helping to advance the fair green and digital transitions, and adapting to demographic change in the EU. Simulations also highlight the potential for social investment to positively impact fiscal sustainability, with the resulting long-term GDP growth more than offsetting the initial cost of the measure analysed.\n\n**Measuring the exact returns on social investment policies is challenging.** Returns might only materialise in the medium to long term, or are not always easily captured in monetary terms. In addition, social investments might reinforce one another, including over individuals' life courses, and might also depend on the effectiveness of other policies, such as social protection. Changing skills needs, including in the context of the green and digital transitions, might necessitate additional social investment to fully reap the benefits of previous investment. While many methods are available to estimate the returns on social investment, they are often complex, require good quality (longitudinal) data and cannot account for all relevant elements at once. The efficiency of spending on social investment is also crucial. In this respect, ensuring that social investment is of high quality, well-designed and evidence-informed is key.\n\nthan direct taxes) represent the largest component of income stabilisation.( ⁷ ) The degree of income stabilisation provided by each component also varies by Member State, ranging from 6.3% in Romania to 45.6% in Denmark for direct taxes; from 0% in Denmark ( ⁸ ) to 32.4% in Romania for SICs; and from less than 0.5% in Hungary, Poland, Estonia and Latvia to 5.3% in France for means-tested benefits (Chart 1).\n\n**The degree and composition of income stabilisation after an income shock varies between income quintiles.** On average, incomes at the bottom (43.5%) and top (42.6%) of the income distribution are more cushioned against a market income shock than those in the middle (39.7%) (Chart 2). While increased benefits lead to a larger shock-absorption among households in the bottom income quintile (providing a stabilisation of 18.6%), the reduction in taxes due to the income loss among households in the top income quintile leads to an almost equally high degree of income stabilisation as for the bottom quintile, despite almost no increase in benefits (0.2% stabilisation). ( ⁹ ) Similarly, personal income taxes absorb an increasingly higher share of the market income shock for the higher the income quintile, increasing from 12.3% for households in the first quintile to as much as 35.3% for households in the fifth quintile.\n\nChart 2 Relative significance of taxes and benefits in absorbing a 5% market income shock varies substantially across the income distribution EU-level ISC, by income quintile and policy instrument (%) (2022/2023 average)\n\n_Source:_ JRC calculations based on EUROMOD, version I6.0+.\n\n( ⁷ ) Bulgaria, Croatia, Hungary, Romania, Slovakia. ( ⁸ ) Result for Denmark follows from the classification of labour market contributions in EUROMOD as taxes rather than SIC and\n\nfrom the fact that unemployment benefit contributions and supplementary labour market contributions do not depend directly on earnings.\n\n( ⁹ ) Chart A3.3 in the Annex presents the relative significance of taxes and benefits in absorbing a 5% market income shock, by\n\nincome quintile and Member State.\n\n0\\.0\n\n10\\.0\n\n20\\.0\n\n30\\.0\n\n40\\.0\n\n50\\.0\n\n1 2 3 4 5 Total\n\nEU - ISC 2022/2023\n\nTaxes Social insurance contributions Benefits","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/employment_and_social_developments_in_europe_esde_2024_report_p94.pdf"]},"response":{"errors":[],"extract_id":"extract_fb5d3b51b83424516248c28e24bdfe13","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Employment and Social Developments in Europe 2024\n\n98\n\n_Box (continued)_\n\n### 5\\. CONCLUSIONS\n\n**Well-designed social investment policies can enhance productivity and competitiveness and have positive impacts on economic growth** **and fiscal sustainability, as well as employment, poverty reduction and social inclusion.** They contribute to more inclusive and environmentally sustainable economies and societies, underpinning upward social convergence, helping to advance the fair green and digital transitions, and adapting to demographic change in the EU. Simulations also highlight the potential for social investment to positively impact fiscal sustainability, with the resulting long-term GDP growth more than offsetting the initial cost of the measure analysed.\n\n**Measuring the exact returns on social investment policies is challenging.** Returns might only materialise in the medium to long term, or are not always easily captured in monetary terms. In addition, social investments might reinforce one another, including over individuals' life courses, and might also depend on the effectiveness of other policies, such as social protection. Changing skills needs, including in the context of the green and digital transitions, might necessitate additional social investment to fully reap the benefits of previous investment. While many methods are available to estimate the returns on social investment, they are often complex, require good quality (longitudinal) data and cannot account for all relevant elements at once. The efficiency of spending on social investment is also crucial. In this respect, ensuring that social investment is of high quality, well-designed and evidence-informed is key.\n\nthan direct taxes) represent the largest component of income stabilisation.( ⁷ ) The degree of income stabilisation provided by each component also varies by Member State, ranging from 6.3% in Romania to 45.6% in Denmark for direct taxes; from 0% in Denmark ( ⁸ ) to 32.4% in Romania for SICs; and from less than 0.5% in Hungary, Poland, Estonia and Latvia to 5.3% in France for means-tested benefits (Chart 1).\n\n**The degree and composition of income stabilisation after an income shock varies between income quintiles.** On average, incomes at the bottom (43.5%) and top (42.6%) of the income distribution are more cushioned against a market income shock than those in the middle (39.7%) (Chart 2). While increased benefits lead to a larger shock-absorption among households in the bottom income quintile (providing a stabilisation of 18.6%), the reduction in taxes due to the income loss among households in the top income quintile leads to an almost equally high degree of income stabilisation as for the bottom quintile, despite almost no increase in benefits (0.2% stabilisation). ( ⁹ ) Similarly, personal income taxes absorb an increasingly higher share of the market income shock for the higher the income quintile, increasing from 12.3% for households in the first quintile to as much as 35.3% for households in the fifth quintile.\n\nChart 2 Relative significance of taxes and benefits in absorbing a 5% market income shock varies substantially across the income distribution EU-level ISC, by income quintile and policy instrument (%) (2022/2023 average)\n\n_Source:_ JRC calculations based on EUROMOD, version I6.0+.\n\n( ⁷ ) Bulgaria, Croatia, Hungary, Romania, Slovakia. ( ⁸ ) Result for Denmark follows from the classification of labour market contributions in EUROMOD as taxes rather than SIC and\n\nfrom the fact that unemployment benefit contributions and supplementary labour market contributions do not depend directly on earnings.\n\n( ⁹ ) Chart A3.3 in the Annex presents the relative significance of taxes and benefits in absorbing a 5% market income shock, by\n\nincome quintile and Member State.\n\n0\\.0\n\n10\\.0\n\n20\\.0\n\n30\\.0\n\n40\\.0\n\n50\\.0\n\n1 2 3 4 5 Total\n\nEU - ISC 2022/2023\n\nTaxes Social insurance contributions Benefits"],"full_content":"Employment and Social Developments in Europe 2024\n\n98\n\n_Box (continued)_\n\n### 5\\. CONCLUSIONS\n\n**Well-designed social investment policies can enhance productivity and competitiveness and have positive impacts on economic growth** **and fiscal sustainability, as well as employment, poverty reduction and social inclusion.** They contribute to more inclusive and environmentally sustainable economies and societies, underpinning upward social convergence, helping to advance the fair green and digital transitions, and adapting to demographic change in the EU. Simulations also highlight the potential for social investment to positively impact fiscal sustainability, with the resulting long-term GDP growth more than offsetting the initial cost of the measure analysed.\n\n**Measuring the exact returns on social investment policies is challenging.** Returns might only materialise in the medium to long term, or are not always easily captured in monetary terms. In addition, social investments might reinforce one another, including over individuals' life courses, and might also depend on the effectiveness of other policies, such as social protection. Changing skills needs, including in the context of the green and digital transitions, might necessitate additional social investment to fully reap the benefits of previous investment. While many methods are available to estimate the returns on social investment, they are often complex, require good quality (longitudinal) data and cannot account for all relevant elements at once. The efficiency of spending on social investment is also crucial. In this respect, ensuring that social investment is of high quality, well-designed and evidence-informed is key.\n\nthan direct taxes) represent the largest component of income stabilisation.( ⁷ ) The degree of income stabilisation provided by each component also varies by Member State, ranging from 6.3% in Romania to 45.6% in Denmark for direct taxes; from 0% in Denmark ( ⁸ ) to 32.4% in Romania for SICs; and from less than 0.5% in Hungary, Poland, Estonia and Latvia to 5.3% in France for means-tested benefits (Chart 1).\n\n**The degree and composition of income stabilisation after an income shock varies between income quintiles.** On average, incomes at the bottom (43.5%) and top (42.6%) of the income distribution are more cushioned against a market income shock than those in the middle (39.7%) (Chart 2). While increased benefits lead to a larger shock-absorption among households in the bottom income quintile (providing a stabilisation of 18.6%), the reduction in taxes due to the income loss among households in the top income quintile leads to an almost equally high degree of income stabilisation as for the bottom quintile, despite almost no increase in benefits (0.2% stabilisation). ( ⁹ ) Similarly, personal income taxes absorb an increasingly higher share of the market income shock for the higher the income quintile, increasing from 12.3% for households in the first quintile to as much as 35.3% for households in the fifth quintile.\n\nChart 2 Relative significance of taxes and benefits in absorbing a 5% market income shock varies substantially across the income distribution EU-level ISC, by income quintile and policy instrument (%) (2022/2023 average)\n\n_Source:_ JRC calculations based on EUROMOD, version I6.0+.\n\n( ⁷ ) Bulgaria, Croatia, Hungary, Romania, Slovakia. ( ⁸ ) Result for Denmark follows from the classification of labour market contributions in EUROMOD as taxes rather than SIC and\n\nfrom the fact that unemployment benefit contributions and supplementary labour market contributions do not depend directly on earnings.\n\n( ⁹ ) Chart A3.3 in the Annex presents the relative significance of taxes and benefits in absorbing a 5% market income shock, by\n\nincome quintile and Member State.\n\n0\\.0\n\n10\\.0\n\n20\\.0\n\n30\\.0\n\n40\\.0\n\n50\\.0\n\n1 2 3 4 5 Total\n\nEU - ISC 2022/2023\n\nTaxes Social insurance contributions Benefits","publish_date":null,"title":"5. C","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/employment_and_social_developments_in_europe_esde_2024_report_p94.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/employment_and_social_developments_in_europe_esde_2024_report_p94.pdf"},"request":{"example_id":"chart/employment_and_social_developments_in_europe_esde_2024_report_p94","product_type":"parse","source_file_path":"employment_and_social_developments_in_europe_esde_2024_report_p94.pdf"},"started_at":"2026-11-04T19:44:50.641972"} +{"completed_at":"2026-11-04T19:44:53.351454","latency_in_ms":1593,"output":{"example_id":"chart/investing-in-education-2025-NC0125093ENN-1_p10","grounded_pages":[],"layout_pages":[],"markdown":"**Investing in Education 2025**\n\n10 Part 1: Early signs of a moderate rebound in investment in education\n\n**The first part of this report looks at the latest data** **1** **on education investment (2023)** **2** **and long-term trends.** 2022 marked the exit from any Covid-19-related measures for EU education systems (UNESCO, 2022). In some EU countries, learning outcomes started to recover in 2023 compared to the period right after COVID-19, thanks to compensatory measures in schools (e.g. targeted additional support for underachieving students or more focused curricula; see Schnepf et al., 2024). However, research at national level suggests that pandemic-related learning losses are likely to persist over the medium-to-long term without effective remedial policy action at system level (Di Pietro, 2023; Maldonado et al., 2024). Moreover, learning outcomes were already declining in the EU before the pandemic (European Commission, 2024b), pointing to structural challenges that education systems have to address, such as the impact of socio-economic background, teacher shortages, declining parental involvement and digital distraction (European Commission, 2025a).\n\n**Public expenditure on education in the EU amounted to 9.6% of total public expenditure and 4.7% of GDP in 2023** . ³ At the country level, the former varied from 14.5% in Estonia and Sweden to 7.3% in Italy; the latter ranged from 7.2% in Sweden to 2.8% in Ireland (Figure 1). ⁴ The correlation between these two indicators is strongly positive, but not perfect (Figure 2). This depends on the different size of total public expenditure in each economy, ⁵ reflecting the variety of long-term approaches to welfare in EU countries.\n\n**Figure 1. Public expenditure on education (2023)**\n\n_Notes:_ Provisional data for BE, DE, ES, FR, PT, SK. See footnote 4 for IE.\n\n_Source:_ Eurostat COFOG data. Online data code: [gov\\_10a\\_exp].\n\n1 Eurostat releases data on government expenditure by function (COFOG) for year t in late February/early March of year t+2. The most recent data available in 2025 refers to 2023.\n\n2 This report will use ‘investment in education’ interchangeably with ‘public expenditure on education’.\n\n3 In monetary terms, this amounted to €806 billion.\n\n4 The low value of this indicator for Ireland (2.8%) is explained by the specific structure of the Irish economy. In most countries, GDP and Gross National Product (GNP) are very close in value, but in Ireland GDP is larger than GNP because of negative net factor income: income outflows are much larger than income inflows due to the presence of many foreign-owned multinational firms, which pay their profits back to their owners abroad (Central Statistics Office, 2025). Consequently, public expenditure on education as a share of GNP would be higher than as a share of GDP.\n\n5 The mathematical relationship between the two indicators is expressed as: (expenditure on education/GDP) = (expenditure on education/total expenditure) \\* (total expenditure/GDP).","pages":[{"line_numbers":[],"markdown":"**Investing in Education 2025**\n\n10 Part 1: Early signs of a moderate rebound in investment in education\n\n**The first part of this report looks at the latest data** **1** **on education investment (2023)** **2** **and long-term trends.** 2022 marked the exit from any Covid-19-related measures for EU education systems (UNESCO, 2022). In some EU countries, learning outcomes started to recover in 2023 compared to the period right after COVID-19, thanks to compensatory measures in schools (e.g. targeted additional support for underachieving students or more focused curricula; see Schnepf et al., 2024). However, research at national level suggests that pandemic-related learning losses are likely to persist over the medium-to-long term without effective remedial policy action at system level (Di Pietro, 2023; Maldonado et al., 2024). Moreover, learning outcomes were already declining in the EU before the pandemic (European Commission, 2024b), pointing to structural challenges that education systems have to address, such as the impact of socio-economic background, teacher shortages, declining parental involvement and digital distraction (European Commission, 2025a).\n\n**Public expenditure on education in the EU amounted to 9.6% of total public expenditure and 4.7% of GDP in 2023** . ³ At the country level, the former varied from 14.5% in Estonia and Sweden to 7.3% in Italy; the latter ranged from 7.2% in Sweden to 2.8% in Ireland (Figure 1). ⁴ The correlation between these two indicators is strongly positive, but not perfect (Figure 2). This depends on the different size of total public expenditure in each economy, ⁵ reflecting the variety of long-term approaches to welfare in EU countries.\n\n**Figure 1. Public expenditure on education (2023)**\n\n_Notes:_ Provisional data for BE, DE, ES, FR, PT, SK. See footnote 4 for IE.\n\n_Source:_ Eurostat COFOG data. Online data code: [gov\\_10a\\_exp].\n\n1 Eurostat releases data on government expenditure by function (COFOG) for year t in late February/early March of year t+2. The most recent data available in 2025 refers to 2023.\n\n2 This report will use ‘investment in education’ interchangeably with ‘public expenditure on education’.\n\n3 In monetary terms, this amounted to €806 billion.\n\n4 The low value of this indicator for Ireland (2.8%) is explained by the specific structure of the Irish economy. In most countries, GDP and Gross National Product (GNP) are very close in value, but in Ireland GDP is larger than GNP because of negative net factor income: income outflows are much larger than income inflows due to the presence of many foreign-owned multinational firms, which pay their profits back to their owners abroad (Central Statistics Office, 2025). Consequently, public expenditure on education as a share of GNP would be higher than as a share of GDP.\n\n5 The mathematical relationship between the two indicators is expressed as: (expenditure on education/GDP) = (expenditure on education/total expenditure) \\* (total expenditure/GDP).","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**Investing in Education 2025**\n\n10 Part 1: Early signs of a moderate rebound in investment in education\n\n**The first part of this report looks at the latest data** **1** **on education investment (2023)** **2** **and long-term trends.** 2022 marked the exit from any Covid-19-related measures for EU education systems (UNESCO, 2022). In some EU countries, learning outcomes started to recover in 2023 compared to the period right after COVID-19, thanks to compensatory measures in schools (e.g. targeted additional support for underachieving students or more focused curricula; see Schnepf et al., 2024). However, research at national level suggests that pandemic-related learning losses are likely to persist over the medium-to-long term without effective remedial policy action at system level (Di Pietro, 2023; Maldonado et al., 2024). Moreover, learning outcomes were already declining in the EU before the pandemic (European Commission, 2024b), pointing to structural challenges that education systems have to address, such as the impact of socio-economic background, teacher shortages, declining parental involvement and digital distraction (European Commission, 2025a).\n\n**Public expenditure on education in the EU amounted to 9.6% of total public expenditure and 4.7% of GDP in 2023** . ³ At the country level, the former varied from 14.5% in Estonia and Sweden to 7.3% in Italy; the latter ranged from 7.2% in Sweden to 2.8% in Ireland (Figure 1). ⁴ The correlation between these two indicators is strongly positive, but not perfect (Figure 2). This depends on the different size of total public expenditure in each economy, ⁵ reflecting the variety of long-term approaches to welfare in EU countries.\n\n**Figure 1. Public expenditure on education (2023)**\n\n_Notes:_ Provisional data for BE, DE, ES, FR, PT, SK. See footnote 4 for IE.\n\n_Source:_ Eurostat COFOG data. Online data code: [gov\\_10a\\_exp].\n\n1 Eurostat releases data on government expenditure by function (COFOG) for year t in late February/early March of year t+2. The most recent data available in 2025 refers to 2023.\n\n2 This report will use ‘investment in education’ interchangeably with ‘public expenditure on education’.\n\n3 In monetary terms, this amounted to €806 billion.\n\n4 The low value of this indicator for Ireland (2.8%) is explained by the specific structure of the Irish economy. In most countries, GDP and Gross National Product (GNP) are very close in value, but in Ireland GDP is larger than GNP because of negative net factor income: income outflows are much larger than income inflows due to the presence of many foreign-owned multinational firms, which pay their profits back to their owners abroad (Central Statistics Office, 2025). Consequently, public expenditure on education as a share of GNP would be higher than as a share of GDP.\n\n5 The mathematical relationship between the two indicators is expressed as: (expenditure on education/GDP) = (expenditure on education/total expenditure) \\* (total expenditure/GDP).","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/investing-in-education-2025-NC0125093ENN-1_p10.pdf"]},"response":{"errors":[],"extract_id":"extract_4b37c5678d155829bf3f253db1911e92","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**Investing in Education 2025**\n\n10 Part 1: Early signs of a moderate rebound in investment in education\n\n**The first part of this report looks at the latest data** **1** **on education investment (2023)** **2** **and long-term trends.** 2022 marked the exit from any Covid-19-related measures for EU education systems (UNESCO, 2022). In some EU countries, learning outcomes started to recover in 2023 compared to the period right after COVID-19, thanks to compensatory measures in schools (e.g. targeted additional support for underachieving students or more focused curricula; see Schnepf et al., 2024). However, research at national level suggests that pandemic-related learning losses are likely to persist over the medium-to-long term without effective remedial policy action at system level (Di Pietro, 2023; Maldonado et al., 2024). Moreover, learning outcomes were already declining in the EU before the pandemic (European Commission, 2024b), pointing to structural challenges that education systems have to address, such as the impact of socio-economic background, teacher shortages, declining parental involvement and digital distraction (European Commission, 2025a).\n\n**Public expenditure on education in the EU amounted to 9.6% of total public expenditure and 4.7% of GDP in 2023** . ³ At the country level, the former varied from 14.5% in Estonia and Sweden to 7.3% in Italy; the latter ranged from 7.2% in Sweden to 2.8% in Ireland (Figure 1). ⁴ The correlation between these two indicators is strongly positive, but not perfect (Figure 2). This depends on the different size of total public expenditure in each economy, ⁵ reflecting the variety of long-term approaches to welfare in EU countries.\n\n**Figure 1. Public expenditure on education (2023)**\n\n_Notes:_ Provisional data for BE, DE, ES, FR, PT, SK. See footnote 4 for IE.\n\n_Source:_ Eurostat COFOG data. Online data code: [gov\\_10a\\_exp].\n\n1 Eurostat releases data on government expenditure by function (COFOG) for year t in late February/early March of year t+2. The most recent data available in 2025 refers to 2023.\n\n2 This report will use ‘investment in education’ interchangeably with ‘public expenditure on education’.\n\n3 In monetary terms, this amounted to €806 billion.\n\n4 The low value of this indicator for Ireland (2.8%) is explained by the specific structure of the Irish economy. In most countries, GDP and Gross National Product (GNP) are very close in value, but in Ireland GDP is larger than GNP because of negative net factor income: income outflows are much larger than income inflows due to the presence of many foreign-owned multinational firms, which pay their profits back to their owners abroad (Central Statistics Office, 2025). Consequently, public expenditure on education as a share of GNP would be higher than as a share of GDP.\n\n5 The mathematical relationship between the two indicators is expressed as: (expenditure on education/GDP) = (expenditure on education/total expenditure) \\* (total expenditure/GDP)."],"full_content":"**Investing in Education 2025**\n\n10 Part 1: Early signs of a moderate rebound in investment in education\n\n**The first part of this report looks at the latest data** **1** **on education investment (2023)** **2** **and long-term trends.** 2022 marked the exit from any Covid-19-related measures for EU education systems (UNESCO, 2022). In some EU countries, learning outcomes started to recover in 2023 compared to the period right after COVID-19, thanks to compensatory measures in schools (e.g. targeted additional support for underachieving students or more focused curricula; see Schnepf et al., 2024). However, research at national level suggests that pandemic-related learning losses are likely to persist over the medium-to-long term without effective remedial policy action at system level (Di Pietro, 2023; Maldonado et al., 2024). Moreover, learning outcomes were already declining in the EU before the pandemic (European Commission, 2024b), pointing to structural challenges that education systems have to address, such as the impact of socio-economic background, teacher shortages, declining parental involvement and digital distraction (European Commission, 2025a).\n\n**Public expenditure on education in the EU amounted to 9.6% of total public expenditure and 4.7% of GDP in 2023** . ³ At the country level, the former varied from 14.5% in Estonia and Sweden to 7.3% in Italy; the latter ranged from 7.2% in Sweden to 2.8% in Ireland (Figure 1). ⁴ The correlation between these two indicators is strongly positive, but not perfect (Figure 2). This depends on the different size of total public expenditure in each economy, ⁵ reflecting the variety of long-term approaches to welfare in EU countries.\n\n**Figure 1. Public expenditure on education (2023)**\n\n_Notes:_ Provisional data for BE, DE, ES, FR, PT, SK. See footnote 4 for IE.\n\n_Source:_ Eurostat COFOG data. Online data code: [gov\\_10a\\_exp].\n\n1 Eurostat releases data on government expenditure by function (COFOG) for year t in late February/early March of year t+2. The most recent data available in 2025 refers to 2023.\n\n2 This report will use ‘investment in education’ interchangeably with ‘public expenditure on education’.\n\n3 In monetary terms, this amounted to €806 billion.\n\n4 The low value of this indicator for Ireland (2.8%) is explained by the specific structure of the Irish economy. In most countries, GDP and Gross National Product (GNP) are very close in value, but in Ireland GDP is larger than GNP because of negative net factor income: income outflows are much larger than income inflows due to the presence of many foreign-owned multinational firms, which pay their profits back to their owners abroad (Central Statistics Office, 2025). Consequently, public expenditure on education as a share of GNP would be higher than as a share of GDP.\n\n5 The mathematical relationship between the two indicators is expressed as: (expenditure on education/GDP) = (expenditure on education/total expenditure) \\* (total expenditure/GDP).","publish_date":null,"title":"Investing in Education 2025 10 Part 1: Early signs of a moderate rebound in investment in education on education investment","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/investing-in-education-2025-NC0125093ENN-1_p10.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/investing-in-education-2025-NC0125093ENN-1_p10.pdf"},"request":{"example_id":"chart/investing-in-education-2025-NC0125093ENN-1_p10","product_type":"parse","source_file_path":"investing-in-education-2025-NC0125093ENN-1_p10.pdf"},"started_at":"2026-11-04T19:44:51.757751"} +{"completed_at":"2026-11-04T19:44:53.991719","latency_in_ms":1137,"output":{"example_id":"chart/investing-in-education-2025-NC0125093ENN-1_p11","grounded_pages":[],"layout_pages":[],"markdown":"11\n\n**Figure 2. Association between public expenditure on education as a share of total public expenditure and as a share of GDP (2023)**\n\n_Notes:_ Provisional data for BE, DE, ES, FR, PT, SK. See footnote 4 for IE. Pearson correlation coefficient (r=0.63) statistically significant at 1% level.\n\n_Source:_ Eurostat COFOG data. Online data code: [gov\\_10a\\_exp].\n\n**The bulk of public expenditure on education goes to schools** . In 2023, school education received more than 70% of public expenditure on education at EU level. This is roughly equally split between, on the one hand, pre-primary and primary levels (35%) and, on the other hand, secondary level (37%), while tertiary education accounted for 16% of public expenditure. Those shares remained largely stable between 2019 and 2023 (European Commission, 2022; European Commission, 2024a). The EU averages mask large differences among Member States (Figure 3). Those differences can be explained by many factors:\n\n• level of involvement of the general government in the education system; • enrolment levels; • the duration of compulsory education; • relative wages in the education sector; • class size and student-teacher ratios; • instruction time; and • the cost of teaching materials and facilities.\n\nAt tertiary level, tuition fees and support for students are also determining factors.\n\n**Staff costs account for almost two-thirds of public expenditure** . 62% of public expenditure on education at EU level went to compensation of employees (i.e. wages and non-wage costs such as employers’ social contributions), ranging from 81% in Belgium to 43% in Sweden. Intermediate consumption (i.e. purchases of non-durable goods, such as teaching materials, and services needed to provide education, such as heating, electricity, cleaning and maintenance services) received 14% of expenditure. Gross capital formation (i.e. investment in acquiring fixed assets and durable goods, such as computers and buildings, and also including the depreciation of fixed assets) accounted for 8% of expenditure (Figure 4). Those shares remained largely stable between 2019 and 2023 (European Commission, 2022; European Commission, 2024a).","pages":[{"line_numbers":[],"markdown":"11\n\n**Figure 2. Association between public expenditure on education as a share of total public expenditure and as a share of GDP (2023)**\n\n_Notes:_ Provisional data for BE, DE, ES, FR, PT, SK. See footnote 4 for IE. Pearson correlation coefficient (r=0.63) statistically significant at 1% level.\n\n_Source:_ Eurostat COFOG data. Online data code: [gov\\_10a\\_exp].\n\n**The bulk of public expenditure on education goes to schools** . In 2023, school education received more than 70% of public expenditure on education at EU level. This is roughly equally split between, on the one hand, pre-primary and primary levels (35%) and, on the other hand, secondary level (37%), while tertiary education accounted for 16% of public expenditure. Those shares remained largely stable between 2019 and 2023 (European Commission, 2022; European Commission, 2024a). The EU averages mask large differences among Member States (Figure 3). Those differences can be explained by many factors:\n\n• level of involvement of the general government in the education system; • enrolment levels; • the duration of compulsory education; • relative wages in the education sector; • class size and student-teacher ratios; • instruction time; and • the cost of teaching materials and facilities.\n\nAt tertiary level, tuition fees and support for students are also determining factors.\n\n**Staff costs account for almost two-thirds of public expenditure** . 62% of public expenditure on education at EU level went to compensation of employees (i.e. wages and non-wage costs such as employers’ social contributions), ranging from 81% in Belgium to 43% in Sweden. Intermediate consumption (i.e. purchases of non-durable goods, such as teaching materials, and services needed to provide education, such as heating, electricity, cleaning and maintenance services) received 14% of expenditure. Gross capital formation (i.e. investment in acquiring fixed assets and durable goods, such as computers and buildings, and also including the depreciation of fixed assets) accounted for 8% of expenditure (Figure 4). Those shares remained largely stable between 2019 and 2023 (European Commission, 2022; European Commission, 2024a).","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"11\n\n**Figure 2. Association between public expenditure on education as a share of total public expenditure and as a share of GDP (2023)**\n\n_Notes:_ Provisional data for BE, DE, ES, FR, PT, SK. See footnote 4 for IE. Pearson correlation coefficient (r=0.63) statistically significant at 1% level.\n\n_Source:_ Eurostat COFOG data. Online data code: [gov\\_10a\\_exp].\n\n**The bulk of public expenditure on education goes to schools** . In 2023, school education received more than 70% of public expenditure on education at EU level. This is roughly equally split between, on the one hand, pre-primary and primary levels (35%) and, on the other hand, secondary level (37%), while tertiary education accounted for 16% of public expenditure. Those shares remained largely stable between 2019 and 2023 (European Commission, 2022; European Commission, 2024a). The EU averages mask large differences among Member States (Figure 3). Those differences can be explained by many factors:\n\n• level of involvement of the general government in the education system; • enrolment levels; • the duration of compulsory education; • relative wages in the education sector; • class size and student-teacher ratios; • instruction time; and • the cost of teaching materials and facilities.\n\nAt tertiary level, tuition fees and support for students are also determining factors.\n\n**Staff costs account for almost two-thirds of public expenditure** . 62% of public expenditure on education at EU level went to compensation of employees (i.e. wages and non-wage costs such as employers’ social contributions), ranging from 81% in Belgium to 43% in Sweden. Intermediate consumption (i.e. purchases of non-durable goods, such as teaching materials, and services needed to provide education, such as heating, electricity, cleaning and maintenance services) received 14% of expenditure. Gross capital formation (i.e. investment in acquiring fixed assets and durable goods, such as computers and buildings, and also including the depreciation of fixed assets) accounted for 8% of expenditure (Figure 4). Those shares remained largely stable between 2019 and 2023 (European Commission, 2022; European Commission, 2024a).","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/investing-in-education-2025-NC0125093ENN-1_p11.pdf"]},"response":{"errors":[],"extract_id":"extract_7d692a7ff6bfbf243d342cb90dae9d9a","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["11\n\n**Figure 2. Association between public expenditure on education as a share of total public expenditure and as a share of GDP (2023)**\n\n_Notes:_ Provisional data for BE, DE, ES, FR, PT, SK. See footnote 4 for IE. Pearson correlation coefficient (r=0.63) statistically significant at 1% level.\n\n_Source:_ Eurostat COFOG data. Online data code: [gov\\_10a\\_exp].\n\n**The bulk of public expenditure on education goes to schools** . In 2023, school education received more than 70% of public expenditure on education at EU level. This is roughly equally split between, on the one hand, pre-primary and primary levels (35%) and, on the other hand, secondary level (37%), while tertiary education accounted for 16% of public expenditure. Those shares remained largely stable between 2019 and 2023 (European Commission, 2022; European Commission, 2024a). The EU averages mask large differences among Member States (Figure 3). Those differences can be explained by many factors:\n\n• level of involvement of the general government in the education system; • enrolment levels; • the duration of compulsory education; • relative wages in the education sector; • class size and student-teacher ratios; • instruction time; and • the cost of teaching materials and facilities.\n\nAt tertiary level, tuition fees and support for students are also determining factors.\n\n**Staff costs account for almost two-thirds of public expenditure** . 62% of public expenditure on education at EU level went to compensation of employees (i.e. wages and non-wage costs such as employers’ social contributions), ranging from 81% in Belgium to 43% in Sweden. Intermediate consumption (i.e. purchases of non-durable goods, such as teaching materials, and services needed to provide education, such as heating, electricity, cleaning and maintenance services) received 14% of expenditure. Gross capital formation (i.e. investment in acquiring fixed assets and durable goods, such as computers and buildings, and also including the depreciation of fixed assets) accounted for 8% of expenditure (Figure 4). Those shares remained largely stable between 2019 and 2023 (European Commission, 2022; European Commission, 2024a)."],"full_content":"11\n\n**Figure 2. Association between public expenditure on education as a share of total public expenditure and as a share of GDP (2023)**\n\n_Notes:_ Provisional data for BE, DE, ES, FR, PT, SK. See footnote 4 for IE. Pearson correlation coefficient (r=0.63) statistically significant at 1% level.\n\n_Source:_ Eurostat COFOG data. Online data code: [gov\\_10a\\_exp].\n\n**The bulk of public expenditure on education goes to schools** . In 2023, school education received more than 70% of public expenditure on education at EU level. This is roughly equally split between, on the one hand, pre-primary and primary levels (35%) and, on the other hand, secondary level (37%), while tertiary education accounted for 16% of public expenditure. Those shares remained largely stable between 2019 and 2023 (European Commission, 2022; European Commission, 2024a). The EU averages mask large differences among Member States (Figure 3). Those differences can be explained by many factors:\n\n• level of involvement of the general government in the education system; • enrolment levels; • the duration of compulsory education; • relative wages in the education sector; • class size and student-teacher ratios; • instruction time; and • the cost of teaching materials and facilities.\n\nAt tertiary level, tuition fees and support for students are also determining factors.\n\n**Staff costs account for almost two-thirds of public expenditure** . 62% of public expenditure on education at EU level went to compensation of employees (i.e. wages and non-wage costs such as employers’ social contributions), ranging from 81% in Belgium to 43% in Sweden. Intermediate consumption (i.e. purchases of non-durable goods, such as teaching materials, and services needed to provide education, such as heating, electricity, cleaning and maintenance services) received 14% of expenditure. Gross capital formation (i.e. investment in acquiring fixed assets and durable goods, such as computers and buildings, and also including the depreciation of fixed assets) accounted for 8% of expenditure (Figure 4). Those shares remained largely stable between 2019 and 2023 (European Commission, 2022; European Commission, 2024a).","publish_date":null,"title":"11 Figure 2. Association between public expenditure on education as a share of total public expenditure and as a share","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/investing-in-education-2025-NC0125093ENN-1_p11.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/investing-in-education-2025-NC0125093ENN-1_p11.pdf"},"request":{"example_id":"chart/investing-in-education-2025-NC0125093ENN-1_p11","product_type":"parse","source_file_path":"investing-in-education-2025-NC0125093ENN-1_p11.pdf"},"started_at":"2026-11-04T19:44:52.854398"} +{"completed_at":"2026-11-04T19:44:56.394787","latency_in_ms":2604,"output":{"example_id":"chart/investing-in-education-2025-NC0125093ENN-1_p12","grounded_pages":[],"layout_pages":[],"markdown":"**Investing in Education 2025**\n\n12 Part 1: Early signs of a moderate rebound in investment in education\n\n**Figure 3. Distribution of public expenditure on education by educational level (2023)**\n\n_Notes:_ Provisional data for BE, DE, ES, FR, PT, SK. ‘Secondary education’ also includes expenditure on post-secondary non-tertiary education’. ‘Other’ is the sum of the following items: education not definable by level, subsidiary services to education, R&D education and education not elsewhere classified.\n\n_Source:_ European Commission services’ calculations based on Eurostat COFOG data. Online data code: [gov\\_10a\\_exp].\n\n**Figure 4. Distribution of public expenditure on education by category (2023)**\n\n_Notes:_ Provisional data for BE, DE, ES, FR, PT, SK. ‘Other’ is the sum of the following items: subsidies, other taxes on production, property income, social benefits, other current transfers, capital transfers.\n\n_Source:_ European Commission services’ calculations based on Eurostat COFOG data. Online data code: [gov\\_10a\\_exp].\n\n**Public expenditure on education in the EU is slowly recovering, especially as a share of total public expenditure** . It had declined from 10% in 2019 to 9.3% in 2020 during the Covid induced- recession and remained roughly constant in 2021 and 2022 at historically low levels, before slightly increasing to 9.6% in 2023. As a share of GDP, it spiked to 4.9% in 2020 due to a strong GDP contraction (European Commission, 2022), then it reverted to its pre-pandemic trend, with a minor rise between 2022 and 2023, from 4.6% to 4.7% (Figure 5). Variations at country level between 2022 and 2023 were usually rather small (see Annex A.1).","pages":[{"line_numbers":[],"markdown":"**Investing in Education 2025**\n\n12 Part 1: Early signs of a moderate rebound in investment in education\n\n**Figure 3. Distribution of public expenditure on education by educational level (2023)**\n\n_Notes:_ Provisional data for BE, DE, ES, FR, PT, SK. ‘Secondary education’ also includes expenditure on post-secondary non-tertiary education’. ‘Other’ is the sum of the following items: education not definable by level, subsidiary services to education, R&D education and education not elsewhere classified.\n\n_Source:_ European Commission services’ calculations based on Eurostat COFOG data. Online data code: [gov\\_10a\\_exp].\n\n**Figure 4. Distribution of public expenditure on education by category (2023)**\n\n_Notes:_ Provisional data for BE, DE, ES, FR, PT, SK. ‘Other’ is the sum of the following items: subsidies, other taxes on production, property income, social benefits, other current transfers, capital transfers.\n\n_Source:_ European Commission services’ calculations based on Eurostat COFOG data. Online data code: [gov\\_10a\\_exp].\n\n**Public expenditure on education in the EU is slowly recovering, especially as a share of total public expenditure** . It had declined from 10% in 2019 to 9.3% in 2020 during the Covid induced- recession and remained roughly constant in 2021 and 2022 at historically low levels, before slightly increasing to 9.6% in 2023. As a share of GDP, it spiked to 4.9% in 2020 due to a strong GDP contraction (European Commission, 2022), then it reverted to its pre-pandemic trend, with a minor rise between 2022 and 2023, from 4.6% to 4.7% (Figure 5). Variations at country level between 2022 and 2023 were usually rather small (see Annex A.1).","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**Investing in Education 2025**\n\n12 Part 1: Early signs of a moderate rebound in investment in education\n\n**Figure 3. Distribution of public expenditure on education by educational level (2023)**\n\n_Notes:_ Provisional data for BE, DE, ES, FR, PT, SK. ‘Secondary education’ also includes expenditure on post-secondary non-tertiary education’. ‘Other’ is the sum of the following items: education not definable by level, subsidiary services to education, R&D education and education not elsewhere classified.\n\n_Source:_ European Commission services’ calculations based on Eurostat COFOG data. Online data code: [gov\\_10a\\_exp].\n\n**Figure 4. Distribution of public expenditure on education by category (2023)**\n\n_Notes:_ Provisional data for BE, DE, ES, FR, PT, SK. ‘Other’ is the sum of the following items: subsidies, other taxes on production, property income, social benefits, other current transfers, capital transfers.\n\n_Source:_ European Commission services’ calculations based on Eurostat COFOG data. Online data code: [gov\\_10a\\_exp].\n\n**Public expenditure on education in the EU is slowly recovering, especially as a share of total public expenditure** . It had declined from 10% in 2019 to 9.3% in 2020 during the Covid induced- recession and remained roughly constant in 2021 and 2022 at historically low levels, before slightly increasing to 9.6% in 2023. As a share of GDP, it spiked to 4.9% in 2020 due to a strong GDP contraction (European Commission, 2022), then it reverted to its pre-pandemic trend, with a minor rise between 2022 and 2023, from 4.6% to 4.7% (Figure 5). Variations at country level between 2022 and 2023 were usually rather small (see Annex A.1).","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/investing-in-education-2025-NC0125093ENN-1_p12.pdf"]},"response":{"errors":[],"extract_id":"extract_cca673919960943f8f602281b7cd5760","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**Investing in Education 2025**\n\n12 Part 1: Early signs of a moderate rebound in investment in education\n\n**Figure 3. Distribution of public expenditure on education by educational level (2023)**\n\n_Notes:_ Provisional data for BE, DE, ES, FR, PT, SK. ‘Secondary education’ also includes expenditure on post-secondary non-tertiary education’. ‘Other’ is the sum of the following items: education not definable by level, subsidiary services to education, R&D education and education not elsewhere classified.\n\n_Source:_ European Commission services’ calculations based on Eurostat COFOG data. Online data code: [gov\\_10a\\_exp].\n\n**Figure 4. Distribution of public expenditure on education by category (2023)**\n\n_Notes:_ Provisional data for BE, DE, ES, FR, PT, SK. ‘Other’ is the sum of the following items: subsidies, other taxes on production, property income, social benefits, other current transfers, capital transfers.\n\n_Source:_ European Commission services’ calculations based on Eurostat COFOG data. Online data code: [gov\\_10a\\_exp].\n\n**Public expenditure on education in the EU is slowly recovering, especially as a share of total public expenditure** . It had declined from 10% in 2019 to 9.3% in 2020 during the Covid induced- recession and remained roughly constant in 2021 and 2022 at historically low levels, before slightly increasing to 9.6% in 2023. As a share of GDP, it spiked to 4.9% in 2020 due to a strong GDP contraction (European Commission, 2022), then it reverted to its pre-pandemic trend, with a minor rise between 2022 and 2023, from 4.6% to 4.7% (Figure 5). Variations at country level between 2022 and 2023 were usually rather small (see Annex A.1)."],"full_content":"**Investing in Education 2025**\n\n12 Part 1: Early signs of a moderate rebound in investment in education\n\n**Figure 3. Distribution of public expenditure on education by educational level (2023)**\n\n_Notes:_ Provisional data for BE, DE, ES, FR, PT, SK. ‘Secondary education’ also includes expenditure on post-secondary non-tertiary education’. ‘Other’ is the sum of the following items: education not definable by level, subsidiary services to education, R&D education and education not elsewhere classified.\n\n_Source:_ European Commission services’ calculations based on Eurostat COFOG data. Online data code: [gov\\_10a\\_exp].\n\n**Figure 4. Distribution of public expenditure on education by category (2023)**\n\n_Notes:_ Provisional data for BE, DE, ES, FR, PT, SK. ‘Other’ is the sum of the following items: subsidies, other taxes on production, property income, social benefits, other current transfers, capital transfers.\n\n_Source:_ European Commission services’ calculations based on Eurostat COFOG data. Online data code: [gov\\_10a\\_exp].\n\n**Public expenditure on education in the EU is slowly recovering, especially as a share of total public expenditure** . It had declined from 10% in 2019 to 9.3% in 2020 during the Covid induced- recession and remained roughly constant in 2021 and 2022 at historically low levels, before slightly increasing to 9.6% in 2023. As a share of GDP, it spiked to 4.9% in 2020 due to a strong GDP contraction (European Commission, 2022), then it reverted to its pre-pandemic trend, with a minor rise between 2022 and 2023, from 4.6% to 4.7% (Figure 5). Variations at country level between 2022 and 2023 were usually rather small (see Annex A.1).","publish_date":null,"title":"Investing in Education 2025 12 Part 1: Early signs of a moderate rebound in investment in education Figure 3. Distribution","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/investing-in-education-2025-NC0125093ENN-1_p12.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/investing-in-education-2025-NC0125093ENN-1_p12.pdf"},"request":{"example_id":"chart/investing-in-education-2025-NC0125093ENN-1_p12","product_type":"parse","source_file_path":"investing-in-education-2025-NC0125093ENN-1_p12.pdf"},"started_at":"2026-11-04T19:44:53.789897"} +{"completed_at":"2026-11-04T19:44:56.301700","latency_in_ms":1412,"output":{"example_id":"chart/investing-in-education-2025-NC0125093ENN-1_p13","grounded_pages":[],"layout_pages":[],"markdown":"13\n\n**Figure 5. Public expenditure on education as a share of total public expenditure and of GDP in the EU-27 (2007-2023)**\n\n_Source:_ Eurostat COFOG data. Online data code: [gov\\_10a\\_exp].\n\n**Since 2020, public expenditure on education increased considerably in the EU in nominal terms, but not after accounting for increases in price levels.** Figure 6 shows that between 2007 (the last year before the Great Recession) and 2023, nominal (i.e. at current prices) public expenditure on education in the EU rose by almost 60%, recording an increase in all years but 2012. Around half of this rise took place after 2020, mostly due to increasing price levels. Real (i.e. at constant prices) ⁶ expenditure grew by around 15% over 2007- 2023, although with year-on-year declines in 2011, 2012, 2020 and 2022. ⁷\n\n**Education still takes up a smaller proportion of total public expenditure than it did before the Covid-19 pandemic.** Figure 7 explains how the structure of total public expenditure changed between 2019 and 2023. ‘Economic affairs’ jumped by 1.7 percentage points. The government measures to mitigate the economic impact of the Covid-19 pandemic (e.g. subsidies to support various productive sectors affected by Covid-related restrictions) caused the strong increases observed in 2020 and 2021. In 2022 and 2023, the financial impact of those measures was significantly reduced, but new measures to mitigate the impact of rising energy prices on productive sectors partly compensated that decrease (European Commission, 2025c). Among other larger expenditure functions, ⁸ only ‘Health’ increased its share during the pandemic, because of the response to new public health needs, but then it reverted to pre-crisis values (European Commission, 2025d). Most other spending areas saw a decrease, and education is still 0.4 percentage points below its 2019 level.\n\n6 This report uses the price deflator for collective consumption expenditure of general government to convert public expenditure on education from current prices to 2015 constant prices.\n\n7 The average yearly increase in public expenditure on education in the EU between 2007 and 2023 was 2.9% in nominal terms and 0.9% in real terms.\n\n8 The increase in ‘Housing and community amenities’ from 1.2% of total public expenditure in 2019 to 2.4% in 2023 mostly depends on capital transfers recorded for payable tax credits in Italy (European Commission, 2025e).","pages":[{"line_numbers":[],"markdown":"13\n\n**Figure 5. Public expenditure on education as a share of total public expenditure and of GDP in the EU-27 (2007-2023)**\n\n_Source:_ Eurostat COFOG data. Online data code: [gov\\_10a\\_exp].\n\n**Since 2020, public expenditure on education increased considerably in the EU in nominal terms, but not after accounting for increases in price levels.** Figure 6 shows that between 2007 (the last year before the Great Recession) and 2023, nominal (i.e. at current prices) public expenditure on education in the EU rose by almost 60%, recording an increase in all years but 2012. Around half of this rise took place after 2020, mostly due to increasing price levels. Real (i.e. at constant prices) ⁶ expenditure grew by around 15% over 2007- 2023, although with year-on-year declines in 2011, 2012, 2020 and 2022. ⁷\n\n**Education still takes up a smaller proportion of total public expenditure than it did before the Covid-19 pandemic.** Figure 7 explains how the structure of total public expenditure changed between 2019 and 2023. ‘Economic affairs’ jumped by 1.7 percentage points. The government measures to mitigate the economic impact of the Covid-19 pandemic (e.g. subsidies to support various productive sectors affected by Covid-related restrictions) caused the strong increases observed in 2020 and 2021. In 2022 and 2023, the financial impact of those measures was significantly reduced, but new measures to mitigate the impact of rising energy prices on productive sectors partly compensated that decrease (European Commission, 2025c). Among other larger expenditure functions, ⁸ only ‘Health’ increased its share during the pandemic, because of the response to new public health needs, but then it reverted to pre-crisis values (European Commission, 2025d). Most other spending areas saw a decrease, and education is still 0.4 percentage points below its 2019 level.\n\n6 This report uses the price deflator for collective consumption expenditure of general government to convert public expenditure on education from current prices to 2015 constant prices.\n\n7 The average yearly increase in public expenditure on education in the EU between 2007 and 2023 was 2.9% in nominal terms and 0.9% in real terms.\n\n8 The increase in ‘Housing and community amenities’ from 1.2% of total public expenditure in 2019 to 2.4% in 2023 mostly depends on capital transfers recorded for payable tax credits in Italy (European Commission, 2025e).","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"13\n\n**Figure 5. Public expenditure on education as a share of total public expenditure and of GDP in the EU-27 (2007-2023)**\n\n_Source:_ Eurostat COFOG data. Online data code: [gov\\_10a\\_exp].\n\n**Since 2020, public expenditure on education increased considerably in the EU in nominal terms, but not after accounting for increases in price levels.** Figure 6 shows that between 2007 (the last year before the Great Recession) and 2023, nominal (i.e. at current prices) public expenditure on education in the EU rose by almost 60%, recording an increase in all years but 2012. Around half of this rise took place after 2020, mostly due to increasing price levels. Real (i.e. at constant prices) ⁶ expenditure grew by around 15% over 2007- 2023, although with year-on-year declines in 2011, 2012, 2020 and 2022. ⁷\n\n**Education still takes up a smaller proportion of total public expenditure than it did before the Covid-19 pandemic.** Figure 7 explains how the structure of total public expenditure changed between 2019 and 2023. ‘Economic affairs’ jumped by 1.7 percentage points. The government measures to mitigate the economic impact of the Covid-19 pandemic (e.g. subsidies to support various productive sectors affected by Covid-related restrictions) caused the strong increases observed in 2020 and 2021. In 2022 and 2023, the financial impact of those measures was significantly reduced, but new measures to mitigate the impact of rising energy prices on productive sectors partly compensated that decrease (European Commission, 2025c). Among other larger expenditure functions, ⁸ only ‘Health’ increased its share during the pandemic, because of the response to new public health needs, but then it reverted to pre-crisis values (European Commission, 2025d). Most other spending areas saw a decrease, and education is still 0.4 percentage points below its 2019 level.\n\n6 This report uses the price deflator for collective consumption expenditure of general government to convert public expenditure on education from current prices to 2015 constant prices.\n\n7 The average yearly increase in public expenditure on education in the EU between 2007 and 2023 was 2.9% in nominal terms and 0.9% in real terms.\n\n8 The increase in ‘Housing and community amenities’ from 1.2% of total public expenditure in 2019 to 2.4% in 2023 mostly depends on capital transfers recorded for payable tax credits in Italy (European Commission, 2025e).","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/investing-in-education-2025-NC0125093ENN-1_p13.pdf"]},"response":{"errors":[],"extract_id":"extract_b36cc4bb2197fbad6de9a6398001ed1c","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["13\n\n**Figure 5. Public expenditure on education as a share of total public expenditure and of GDP in the EU-27 (2007-2023)**\n\n_Source:_ Eurostat COFOG data. Online data code: [gov\\_10a\\_exp].\n\n**Since 2020, public expenditure on education increased considerably in the EU in nominal terms, but not after accounting for increases in price levels.** Figure 6 shows that between 2007 (the last year before the Great Recession) and 2023, nominal (i.e. at current prices) public expenditure on education in the EU rose by almost 60%, recording an increase in all years but 2012. Around half of this rise took place after 2020, mostly due to increasing price levels. Real (i.e. at constant prices) ⁶ expenditure grew by around 15% over 2007- 2023, although with year-on-year declines in 2011, 2012, 2020 and 2022. ⁷\n\n**Education still takes up a smaller proportion of total public expenditure than it did before the Covid-19 pandemic.** Figure 7 explains how the structure of total public expenditure changed between 2019 and 2023. ‘Economic affairs’ jumped by 1.7 percentage points. The government measures to mitigate the economic impact of the Covid-19 pandemic (e.g. subsidies to support various productive sectors affected by Covid-related restrictions) caused the strong increases observed in 2020 and 2021. In 2022 and 2023, the financial impact of those measures was significantly reduced, but new measures to mitigate the impact of rising energy prices on productive sectors partly compensated that decrease (European Commission, 2025c). Among other larger expenditure functions, ⁸ only ‘Health’ increased its share during the pandemic, because of the response to new public health needs, but then it reverted to pre-crisis values (European Commission, 2025d). Most other spending areas saw a decrease, and education is still 0.4 percentage points below its 2019 level.\n\n6 This report uses the price deflator for collective consumption expenditure of general government to convert public expenditure on education from current prices to 2015 constant prices.\n\n7 The average yearly increase in public expenditure on education in the EU between 2007 and 2023 was 2.9% in nominal terms and 0.9% in real terms.\n\n8 The increase in ‘Housing and community amenities’ from 1.2% of total public expenditure in 2019 to 2.4% in 2023 mostly depends on capital transfers recorded for payable tax credits in Italy (European Commission, 2025e)."],"full_content":"13\n\n**Figure 5. Public expenditure on education as a share of total public expenditure and of GDP in the EU-27 (2007-2023)**\n\n_Source:_ Eurostat COFOG data. Online data code: [gov\\_10a\\_exp].\n\n**Since 2020, public expenditure on education increased considerably in the EU in nominal terms, but not after accounting for increases in price levels.** Figure 6 shows that between 2007 (the last year before the Great Recession) and 2023, nominal (i.e. at current prices) public expenditure on education in the EU rose by almost 60%, recording an increase in all years but 2012. Around half of this rise took place after 2020, mostly due to increasing price levels. Real (i.e. at constant prices) ⁶ expenditure grew by around 15% over 2007- 2023, although with year-on-year declines in 2011, 2012, 2020 and 2022. ⁷\n\n**Education still takes up a smaller proportion of total public expenditure than it did before the Covid-19 pandemic.** Figure 7 explains how the structure of total public expenditure changed between 2019 and 2023. ‘Economic affairs’ jumped by 1.7 percentage points. The government measures to mitigate the economic impact of the Covid-19 pandemic (e.g. subsidies to support various productive sectors affected by Covid-related restrictions) caused the strong increases observed in 2020 and 2021. In 2022 and 2023, the financial impact of those measures was significantly reduced, but new measures to mitigate the impact of rising energy prices on productive sectors partly compensated that decrease (European Commission, 2025c). Among other larger expenditure functions, ⁸ only ‘Health’ increased its share during the pandemic, because of the response to new public health needs, but then it reverted to pre-crisis values (European Commission, 2025d). Most other spending areas saw a decrease, and education is still 0.4 percentage points below its 2019 level.\n\n6 This report uses the price deflator for collective consumption expenditure of general government to convert public expenditure on education from current prices to 2015 constant prices.\n\n7 The average yearly increase in public expenditure on education in the EU between 2007 and 2023 was 2.9% in nominal terms and 0.9% in real terms.\n\n8 The increase in ‘Housing and community amenities’ from 1.2% of total public expenditure in 2019 to 2.4% in 2023 mostly depends on capital transfers recorded for payable tax credits in Italy (European Commission, 2025e).","publish_date":null,"title":"13 Figure 5. Public expenditure on education as a share of total public expenditure and of GDP in the EU-27","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/investing-in-education-2025-NC0125093ENN-1_p13.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/investing-in-education-2025-NC0125093ENN-1_p13.pdf"},"request":{"example_id":"chart/investing-in-education-2025-NC0125093ENN-1_p13","product_type":"parse","source_file_path":"investing-in-education-2025-NC0125093ENN-1_p13.pdf"},"started_at":"2026-11-04T19:44:54.888820"} +{"completed_at":"2026-11-04T19:44:58.607437","latency_in_ms":2315,"output":{"example_id":"chart/investing-in-education-2025-NC0125093ENN-1_p14","grounded_pages":[],"layout_pages":[],"markdown":"**Investing in Education 2025**\n\n14 Part 1: Early signs of a moderate rebound in investment in education\n\n**Figure 6. Evolution of nominal and real public expenditure on education in the EU-27 (index 2007=100)**\n\n_Notes:_ Real values are expressed at 2015 constant prices by using the price deflator for collective consumption expenditure of general government.\n\n_Source:_ European Commission services’ calculations based on Eurostat COFOG data and AMECO data\n\n**Figure 7. Public expenditure by function (2019-2023)**\n\n_Source:_ Eurostat COFOG data. Online data code: [gov\\_10a\\_exp].\n\n**Overall, early signs of a moderate rebound in investment in education are emerging against the backdrop of increased competition for public funding.** 2023 data shows that investment in education started to bounce back from its post-pandemic low, especially as a share of total public expenditure. However, education faces a stronger competition from other public functions, and consequently continues to get a lower share of total public expenditure than in the 2010s. Although part of the measures taken in 2020-2021 to respond to the Covid-19 pandemic in the areas of economic affairs and health were discontinued over 2022-2023, new measures were introduced to face the energy crisis and support Ukraine after Russia’s invasion. A clearer picture of whether a new composition of public expenditure in the EU has emerged out of the recent crises and the new economic and geopolitical priorities will likely appear only over the next few years.\n\n**2019 2020 2021 2022 2023 2019-2023** % of total % of total % of total % of total % of total percentage point change\n\nEconomic affairs 10.1 12.3 12.7 12.1 11.8 1.7 Housing and community amenities 1.2 1.2 1.7 2.0 2.4 1.2 Defence 2.6 2.5 2.4 2.5 2.7 0.1 Environmental protection 1.7 1.6 1.5 1.6 1.7 0.0 Health 14.9 15.0 15.7 15.4 14.8 -0.1 Public order and safety 3.6 3.4 3.3 3.4 3.5 -0.1 Recreation, culture and religion 2.5 2.3 2.3 2.3 2.4 -0.1 General public services 12.3 11.4 11.4 11.9 12.0 -0.3 Education 10.0 9.3 9.3 9.4 9.6 -0.4 Social protection 41.2 41.1 39.6 39.3 39.3 -1.9","pages":[{"line_numbers":[],"markdown":"**Investing in Education 2025**\n\n14 Part 1: Early signs of a moderate rebound in investment in education\n\n**Figure 6. Evolution of nominal and real public expenditure on education in the EU-27 (index 2007=100)**\n\n_Notes:_ Real values are expressed at 2015 constant prices by using the price deflator for collective consumption expenditure of general government.\n\n_Source:_ European Commission services’ calculations based on Eurostat COFOG data and AMECO data\n\n**Figure 7. Public expenditure by function (2019-2023)**\n\n_Source:_ Eurostat COFOG data. Online data code: [gov\\_10a\\_exp].\n\n**Overall, early signs of a moderate rebound in investment in education are emerging against the backdrop of increased competition for public funding.** 2023 data shows that investment in education started to bounce back from its post-pandemic low, especially as a share of total public expenditure. However, education faces a stronger competition from other public functions, and consequently continues to get a lower share of total public expenditure than in the 2010s. Although part of the measures taken in 2020-2021 to respond to the Covid-19 pandemic in the areas of economic affairs and health were discontinued over 2022-2023, new measures were introduced to face the energy crisis and support Ukraine after Russia’s invasion. A clearer picture of whether a new composition of public expenditure in the EU has emerged out of the recent crises and the new economic and geopolitical priorities will likely appear only over the next few years.\n\n**2019 2020 2021 2022 2023 2019-2023** % of total % of total % of total % of total % of total percentage point change\n\nEconomic affairs 10.1 12.3 12.7 12.1 11.8 1.7 Housing and community amenities 1.2 1.2 1.7 2.0 2.4 1.2 Defence 2.6 2.5 2.4 2.5 2.7 0.1 Environmental protection 1.7 1.6 1.5 1.6 1.7 0.0 Health 14.9 15.0 15.7 15.4 14.8 -0.1 Public order and safety 3.6 3.4 3.3 3.4 3.5 -0.1 Recreation, culture and religion 2.5 2.3 2.3 2.3 2.4 -0.1 General public services 12.3 11.4 11.4 11.9 12.0 -0.3 Education 10.0 9.3 9.3 9.4 9.6 -0.4 Social protection 41.2 41.1 39.6 39.3 39.3 -1.9","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**Investing in Education 2025**\n\n14 Part 1: Early signs of a moderate rebound in investment in education\n\n**Figure 6. Evolution of nominal and real public expenditure on education in the EU-27 (index 2007=100)**\n\n_Notes:_ Real values are expressed at 2015 constant prices by using the price deflator for collective consumption expenditure of general government.\n\n_Source:_ European Commission services’ calculations based on Eurostat COFOG data and AMECO data\n\n**Figure 7. Public expenditure by function (2019-2023)**\n\n_Source:_ Eurostat COFOG data. Online data code: [gov\\_10a\\_exp].\n\n**Overall, early signs of a moderate rebound in investment in education are emerging against the backdrop of increased competition for public funding.** 2023 data shows that investment in education started to bounce back from its post-pandemic low, especially as a share of total public expenditure. However, education faces a stronger competition from other public functions, and consequently continues to get a lower share of total public expenditure than in the 2010s. Although part of the measures taken in 2020-2021 to respond to the Covid-19 pandemic in the areas of economic affairs and health were discontinued over 2022-2023, new measures were introduced to face the energy crisis and support Ukraine after Russia’s invasion. A clearer picture of whether a new composition of public expenditure in the EU has emerged out of the recent crises and the new economic and geopolitical priorities will likely appear only over the next few years.\n\n**2019 2020 2021 2022 2023 2019-2023** % of total % of total % of total % of total % of total percentage point change\n\nEconomic affairs 10.1 12.3 12.7 12.1 11.8 1.7 Housing and community amenities 1.2 1.2 1.7 2.0 2.4 1.2 Defence 2.6 2.5 2.4 2.5 2.7 0.1 Environmental protection 1.7 1.6 1.5 1.6 1.7 0.0 Health 14.9 15.0 15.7 15.4 14.8 -0.1 Public order and safety 3.6 3.4 3.3 3.4 3.5 -0.1 Recreation, culture and religion 2.5 2.3 2.3 2.3 2.4 -0.1 General public services 12.3 11.4 11.4 11.9 12.0 -0.3 Education 10.0 9.3 9.3 9.4 9.6 -0.4 Social protection 41.2 41.1 39.6 39.3 39.3 -1.9","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/investing-in-education-2025-NC0125093ENN-1_p14.pdf"]},"response":{"errors":[],"extract_id":"extract_99bbf320099e18ddd5df4635a69bde43","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**Investing in Education 2025**\n\n14 Part 1: Early signs of a moderate rebound in investment in education\n\n**Figure 6. Evolution of nominal and real public expenditure on education in the EU-27 (index 2007=100)**\n\n_Notes:_ Real values are expressed at 2015 constant prices by using the price deflator for collective consumption expenditure of general government.\n\n_Source:_ European Commission services’ calculations based on Eurostat COFOG data and AMECO data\n\n**Figure 7. Public expenditure by function (2019-2023)**\n\n_Source:_ Eurostat COFOG data. Online data code: [gov\\_10a\\_exp].\n\n**Overall, early signs of a moderate rebound in investment in education are emerging against the backdrop of increased competition for public funding.** 2023 data shows that investment in education started to bounce back from its post-pandemic low, especially as a share of total public expenditure. However, education faces a stronger competition from other public functions, and consequently continues to get a lower share of total public expenditure than in the 2010s. Although part of the measures taken in 2020-2021 to respond to the Covid-19 pandemic in the areas of economic affairs and health were discontinued over 2022-2023, new measures were introduced to face the energy crisis and support Ukraine after Russia’s invasion. A clearer picture of whether a new composition of public expenditure in the EU has emerged out of the recent crises and the new economic and geopolitical priorities will likely appear only over the next few years.\n\n**2019 2020 2021 2022 2023 2019-2023** % of total % of total % of total % of total % of total percentage point change\n\nEconomic affairs 10.1 12.3 12.7 12.1 11.8 1.7 Housing and community amenities 1.2 1.2 1.7 2.0 2.4 1.2 Defence 2.6 2.5 2.4 2.5 2.7 0.1 Environmental protection 1.7 1.6 1.5 1.6 1.7 0.0 Health 14.9 15.0 15.7 15.4 14.8 -0.1 Public order and safety 3.6 3.4 3.3 3.4 3.5 -0.1 Recreation, culture and religion 2.5 2.3 2.3 2.3 2.4 -0.1 General public services 12.3 11.4 11.4 11.9 12.0 -0.3 Education 10.0 9.3 9.3 9.4 9.6 -0.4 Social protection 41.2 41.1 39.6 39.3 39.3 -1.9"],"full_content":"**Investing in Education 2025**\n\n14 Part 1: Early signs of a moderate rebound in investment in education\n\n**Figure 6. Evolution of nominal and real public expenditure on education in the EU-27 (index 2007=100)**\n\n_Notes:_ Real values are expressed at 2015 constant prices by using the price deflator for collective consumption expenditure of general government.\n\n_Source:_ European Commission services’ calculations based on Eurostat COFOG data and AMECO data\n\n**Figure 7. Public expenditure by function (2019-2023)**\n\n_Source:_ Eurostat COFOG data. Online data code: [gov\\_10a\\_exp].\n\n**Overall, early signs of a moderate rebound in investment in education are emerging against the backdrop of increased competition for public funding.** 2023 data shows that investment in education started to bounce back from its post-pandemic low, especially as a share of total public expenditure. However, education faces a stronger competition from other public functions, and consequently continues to get a lower share of total public expenditure than in the 2010s. Although part of the measures taken in 2020-2021 to respond to the Covid-19 pandemic in the areas of economic affairs and health were discontinued over 2022-2023, new measures were introduced to face the energy crisis and support Ukraine after Russia’s invasion. A clearer picture of whether a new composition of public expenditure in the EU has emerged out of the recent crises and the new economic and geopolitical priorities will likely appear only over the next few years.\n\n**2019 2020 2021 2022 2023 2019-2023** % of total % of total % of total % of total % of total percentage point change\n\nEconomic affairs 10.1 12.3 12.7 12.1 11.8 1.7 Housing and community amenities 1.2 1.2 1.7 2.0 2.4 1.2 Defence 2.6 2.5 2.4 2.5 2.7 0.1 Environmental protection 1.7 1.6 1.5 1.6 1.7 0.0 Health 14.9 15.0 15.7 15.4 14.8 -0.1 Public order and safety 3.6 3.4 3.3 3.4 3.5 -0.1 Recreation, culture and religion 2.5 2.3 2.3 2.3 2.4 -0.1 General public services 12.3 11.4 11.4 11.9 12.0 -0.3 Education 10.0 9.3 9.3 9.4 9.6 -0.4 Social protection 41.2 41.1 39.6 39.3 39.3 -1.9","publish_date":null,"title":"Investing in Education 2025 14 Part 1: Early signs of a moderate rebound in investment in education Figure 6. Evolution","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/investing-in-education-2025-NC0125093ENN-1_p14.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/investing-in-education-2025-NC0125093ENN-1_p14.pdf"},"request":{"example_id":"chart/investing-in-education-2025-NC0125093ENN-1_p14","product_type":"parse","source_file_path":"investing-in-education-2025-NC0125093ENN-1_p14.pdf"},"started_at":"2026-11-04T19:44:56.291799"} +{"completed_at":"2026-11-04T19:44:58.532449","latency_in_ms":1123,"output":{"example_id":"chart/investing-in-education-2025-NC0125093ENN-1_p20","grounded_pages":[],"layout_pages":[],"markdown":"**Investing in Education 2025**\n\n20 Part 2: Prioritising investment in education: economic returns and demographic challenges\n\n**Adverse demographic trends will affect European education systems.** Figure 9 looks at how demographic change could decrease the population of pre-school and school age in the EU between now and 2030. While several countries make their own forecast for the future number of pupils in their national education systems, there are no internationally comparable projections. To compare trends across Member States, one needs to rely on Eurostat demographic projections and use the population aged 3-18 as a proxy for the number of pupils in pre-primary, primary or secondary education. According to the baseline scenario, the number of people aged 3-18 would decrease by 3.5% in the EU-27 by 2030 compared with 2022. In absolute numbers, this would mean around 2.5 million fewer people. At country level, 19 Member States would experience a demographic decline; Italy (-13%), Greece (-12%) and Spain (-11%) would be the most severely affected countries. By contrast, the population aged 3-18 would increase by more than 5% in Malta (13%), Luxembourg (12%), Germany (9%) and Cyprus (6%). As the time horizon is relatively short, these projections are rather insensitive to possible changes in fertility rates. Different assumptions about migration flows could in principle modify the picture. Eurostat also presents two other scenarios, one assuming larger inflows of migrants (“higher migration”) and one assuming smaller inflows (“lower migration”). In the first scenario the EU-27 population aged 3-18 would shrink by 2.2%, while in the second scenario the decrease would reach 4.8%. At country level, the two scenarios would either exacerbate or mitigate the magnitude of the baseline trends, but they would not radically change them.\n\n**Figure 9. Demographic projections for the 3-18 age group (% change between 2022 and 2030)**\n\n_Notes:_ Higher migration: the scenario assumes that the net migration is higher due to a 33% increase applied to the non-EU immigration flows, in each year of the 2023-2030 time horizon. Lower migration: the scenario assumes that the net migration is lower due to a 33% decrease applied to the non-EU immigration flows, in each year of the 2023-2030 time horizon. For further details, please consult Eurostat metadata.\n\n_Source:_ European Commission services’ calculations based on Eurostat population projection data. Online data code: [proj\\_23np].\n\n**An aging population places financial strain on public budgets, as rising pension and healthcare costs may limit government funding for education.** With fewer children enrolling in schools, there is a risk that governments may come under pressure to deprioritise education funding, leading to lower investment in school infrastructure, educational resources, and overall quality.","pages":[{"line_numbers":[],"markdown":"**Investing in Education 2025**\n\n20 Part 2: Prioritising investment in education: economic returns and demographic challenges\n\n**Adverse demographic trends will affect European education systems.** Figure 9 looks at how demographic change could decrease the population of pre-school and school age in the EU between now and 2030. While several countries make their own forecast for the future number of pupils in their national education systems, there are no internationally comparable projections. To compare trends across Member States, one needs to rely on Eurostat demographic projections and use the population aged 3-18 as a proxy for the number of pupils in pre-primary, primary or secondary education. According to the baseline scenario, the number of people aged 3-18 would decrease by 3.5% in the EU-27 by 2030 compared with 2022. In absolute numbers, this would mean around 2.5 million fewer people. At country level, 19 Member States would experience a demographic decline; Italy (-13%), Greece (-12%) and Spain (-11%) would be the most severely affected countries. By contrast, the population aged 3-18 would increase by more than 5% in Malta (13%), Luxembourg (12%), Germany (9%) and Cyprus (6%). As the time horizon is relatively short, these projections are rather insensitive to possible changes in fertility rates. Different assumptions about migration flows could in principle modify the picture. Eurostat also presents two other scenarios, one assuming larger inflows of migrants (“higher migration”) and one assuming smaller inflows (“lower migration”). In the first scenario the EU-27 population aged 3-18 would shrink by 2.2%, while in the second scenario the decrease would reach 4.8%. At country level, the two scenarios would either exacerbate or mitigate the magnitude of the baseline trends, but they would not radically change them.\n\n**Figure 9. Demographic projections for the 3-18 age group (% change between 2022 and 2030)**\n\n_Notes:_ Higher migration: the scenario assumes that the net migration is higher due to a 33% increase applied to the non-EU immigration flows, in each year of the 2023-2030 time horizon. Lower migration: the scenario assumes that the net migration is lower due to a 33% decrease applied to the non-EU immigration flows, in each year of the 2023-2030 time horizon. For further details, please consult Eurostat metadata.\n\n_Source:_ European Commission services’ calculations based on Eurostat population projection data. Online data code: [proj\\_23np].\n\n**An aging population places financial strain on public budgets, as rising pension and healthcare costs may limit government funding for education.** With fewer children enrolling in schools, there is a risk that governments may come under pressure to deprioritise education funding, leading to lower investment in school infrastructure, educational resources, and overall quality.","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**Investing in Education 2025**\n\n20 Part 2: Prioritising investment in education: economic returns and demographic challenges\n\n**Adverse demographic trends will affect European education systems.** Figure 9 looks at how demographic change could decrease the population of pre-school and school age in the EU between now and 2030. While several countries make their own forecast for the future number of pupils in their national education systems, there are no internationally comparable projections. To compare trends across Member States, one needs to rely on Eurostat demographic projections and use the population aged 3-18 as a proxy for the number of pupils in pre-primary, primary or secondary education. According to the baseline scenario, the number of people aged 3-18 would decrease by 3.5% in the EU-27 by 2030 compared with 2022. In absolute numbers, this would mean around 2.5 million fewer people. At country level, 19 Member States would experience a demographic decline; Italy (-13%), Greece (-12%) and Spain (-11%) would be the most severely affected countries. By contrast, the population aged 3-18 would increase by more than 5% in Malta (13%), Luxembourg (12%), Germany (9%) and Cyprus (6%). As the time horizon is relatively short, these projections are rather insensitive to possible changes in fertility rates. Different assumptions about migration flows could in principle modify the picture. Eurostat also presents two other scenarios, one assuming larger inflows of migrants (“higher migration”) and one assuming smaller inflows (“lower migration”). In the first scenario the EU-27 population aged 3-18 would shrink by 2.2%, while in the second scenario the decrease would reach 4.8%. At country level, the two scenarios would either exacerbate or mitigate the magnitude of the baseline trends, but they would not radically change them.\n\n**Figure 9. Demographic projections for the 3-18 age group (% change between 2022 and 2030)**\n\n_Notes:_ Higher migration: the scenario assumes that the net migration is higher due to a 33% increase applied to the non-EU immigration flows, in each year of the 2023-2030 time horizon. Lower migration: the scenario assumes that the net migration is lower due to a 33% decrease applied to the non-EU immigration flows, in each year of the 2023-2030 time horizon. For further details, please consult Eurostat metadata.\n\n_Source:_ European Commission services’ calculations based on Eurostat population projection data. Online data code: [proj\\_23np].\n\n**An aging population places financial strain on public budgets, as rising pension and healthcare costs may limit government funding for education.** With fewer children enrolling in schools, there is a risk that governments may come under pressure to deprioritise education funding, leading to lower investment in school infrastructure, educational resources, and overall quality.","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/investing-in-education-2025-NC0125093ENN-1_p20.pdf"]},"response":{"errors":[],"extract_id":"extract_3141bfe1c0d7ffed42e42c94b6c91353","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**Investing in Education 2025**\n\n20 Part 2: Prioritising investment in education: economic returns and demographic challenges\n\n**Adverse demographic trends will affect European education systems.** Figure 9 looks at how demographic change could decrease the population of pre-school and school age in the EU between now and 2030. While several countries make their own forecast for the future number of pupils in their national education systems, there are no internationally comparable projections. To compare trends across Member States, one needs to rely on Eurostat demographic projections and use the population aged 3-18 as a proxy for the number of pupils in pre-primary, primary or secondary education. According to the baseline scenario, the number of people aged 3-18 would decrease by 3.5% in the EU-27 by 2030 compared with 2022. In absolute numbers, this would mean around 2.5 million fewer people. At country level, 19 Member States would experience a demographic decline; Italy (-13%), Greece (-12%) and Spain (-11%) would be the most severely affected countries. By contrast, the population aged 3-18 would increase by more than 5% in Malta (13%), Luxembourg (12%), Germany (9%) and Cyprus (6%). As the time horizon is relatively short, these projections are rather insensitive to possible changes in fertility rates. Different assumptions about migration flows could in principle modify the picture. Eurostat also presents two other scenarios, one assuming larger inflows of migrants (“higher migration”) and one assuming smaller inflows (“lower migration”). In the first scenario the EU-27 population aged 3-18 would shrink by 2.2%, while in the second scenario the decrease would reach 4.8%. At country level, the two scenarios would either exacerbate or mitigate the magnitude of the baseline trends, but they would not radically change them.\n\n**Figure 9. Demographic projections for the 3-18 age group (% change between 2022 and 2030)**\n\n_Notes:_ Higher migration: the scenario assumes that the net migration is higher due to a 33% increase applied to the non-EU immigration flows, in each year of the 2023-2030 time horizon. Lower migration: the scenario assumes that the net migration is lower due to a 33% decrease applied to the non-EU immigration flows, in each year of the 2023-2030 time horizon. For further details, please consult Eurostat metadata.\n\n_Source:_ European Commission services’ calculations based on Eurostat population projection data. Online data code: [proj\\_23np].\n\n**An aging population places financial strain on public budgets, as rising pension and healthcare costs may limit government funding for education.** With fewer children enrolling in schools, there is a risk that governments may come under pressure to deprioritise education funding, leading to lower investment in school infrastructure, educational resources, and overall quality."],"full_content":"**Investing in Education 2025**\n\n20 Part 2: Prioritising investment in education: economic returns and demographic challenges\n\n**Adverse demographic trends will affect European education systems.** Figure 9 looks at how demographic change could decrease the population of pre-school and school age in the EU between now and 2030. While several countries make their own forecast for the future number of pupils in their national education systems, there are no internationally comparable projections. To compare trends across Member States, one needs to rely on Eurostat demographic projections and use the population aged 3-18 as a proxy for the number of pupils in pre-primary, primary or secondary education. According to the baseline scenario, the number of people aged 3-18 would decrease by 3.5% in the EU-27 by 2030 compared with 2022. In absolute numbers, this would mean around 2.5 million fewer people. At country level, 19 Member States would experience a demographic decline; Italy (-13%), Greece (-12%) and Spain (-11%) would be the most severely affected countries. By contrast, the population aged 3-18 would increase by more than 5% in Malta (13%), Luxembourg (12%), Germany (9%) and Cyprus (6%). As the time horizon is relatively short, these projections are rather insensitive to possible changes in fertility rates. Different assumptions about migration flows could in principle modify the picture. Eurostat also presents two other scenarios, one assuming larger inflows of migrants (“higher migration”) and one assuming smaller inflows (“lower migration”). In the first scenario the EU-27 population aged 3-18 would shrink by 2.2%, while in the second scenario the decrease would reach 4.8%. At country level, the two scenarios would either exacerbate or mitigate the magnitude of the baseline trends, but they would not radically change them.\n\n**Figure 9. Demographic projections for the 3-18 age group (% change between 2022 and 2030)**\n\n_Notes:_ Higher migration: the scenario assumes that the net migration is higher due to a 33% increase applied to the non-EU immigration flows, in each year of the 2023-2030 time horizon. Lower migration: the scenario assumes that the net migration is lower due to a 33% decrease applied to the non-EU immigration flows, in each year of the 2023-2030 time horizon. For further details, please consult Eurostat metadata.\n\n_Source:_ European Commission services’ calculations based on Eurostat population projection data. Online data code: [proj\\_23np].\n\n**An aging population places financial strain on public budgets, as rising pension and healthcare costs may limit government funding for education.** With fewer children enrolling in schools, there is a risk that governments may come under pressure to deprioritise education funding, leading to lower investment in school infrastructure, educational resources, and overall quality.","publish_date":null,"title":"Investing in Education 2025 20 Part 2: Prioritising investment in education: economic returns and demographic challenges Figure 9 looks at","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/investing-in-education-2025-NC0125093ENN-1_p20.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/investing-in-education-2025-NC0125093ENN-1_p20.pdf"},"request":{"example_id":"chart/investing-in-education-2025-NC0125093ENN-1_p20","product_type":"parse","source_file_path":"investing-in-education-2025-NC0125093ENN-1_p20.pdf"},"started_at":"2026-11-04T19:44:57.408540"} +{"completed_at":"2026-11-04T19:45:00.691074","latency_in_ms":1974,"output":{"example_id":"chart/m-trends-2025-en_p21","grounded_pages":[],"layout_pages":[],"markdown":"**Mandiant M-Trends 2025 Report** **21**\n\n# **Most Frequently Seen Malware Families**\n\nFor the fifth consecutive year, BEACON was identified as the most frequently observed malware family in Mandiant investigations globally and was identified in 5.4% of all intrusions. BEACON usage has decreased dramatically since 2021, when it was observed in 28% of Mandiant investigations.\n\n30\n\n20\n\n10\n\n0\n\n2020 2021 2022 2023 2024\n\n### **24** % **28** %\n\n### **15** % **10** % **5** %\n\n**BEACON Usage, 2020-2024**\n\n**Most Frequently Seen Malware Families, 2024**\n\n6\n\n4\n\n2\n\n0\n\nBEACON GOOTLOADER WIREFIRE SYSTEMBC REDBIKE (Akira)\n\nRANSOMHUB LOCKBIT BASTA\n\n### **5\\.4** %\n\n### **2\\.5** % **2\\.5** % **1\\.8** % **1\\.8** % **1\\.8** % **1\\.6** % **1\\.6** %\n\n**Most Frequently Seen Malware Families, 2024**\n\nOf note, in July of 2024, Europol ¹² provided an update on Operation MORPHEUS, a global action against the illicit use of the unlicensed versions of the Cobalt Strike red teaming tool. This operation, conducted with law enforcement and private sector partners, successfully disrupted infrastructure linked to cyber criminal activities. The initiative, which began in 2021, involved flagging 690 IP addresses, 593 of which were taken down by online service providers. Fortra, ¹³ the maintainers of the Cobalt Strike framework, also announced the number of unauthorized copies of Cobalt Strike observed in the wild has decreased by 80% over the past two years as a result of their participation in Operation MORPHEUS. Observed declines in percentages of inves- tigations where Mandiant identified BEACON since 2021 may reflect the success of this effort.","pages":[{"line_numbers":[],"markdown":"**Mandiant M-Trends 2025 Report** **21**\n\n# **Most Frequently Seen Malware Families**\n\nFor the fifth consecutive year, BEACON was identified as the most frequently observed malware family in Mandiant investigations globally and was identified in 5.4% of all intrusions. BEACON usage has decreased dramatically since 2021, when it was observed in 28% of Mandiant investigations.\n\n30\n\n20\n\n10\n\n0\n\n2020 2021 2022 2023 2024\n\n### **24** % **28** %\n\n### **15** % **10** % **5** %\n\n**BEACON Usage, 2020-2024**\n\n**Most Frequently Seen Malware Families, 2024**\n\n6\n\n4\n\n2\n\n0\n\nBEACON GOOTLOADER WIREFIRE SYSTEMBC REDBIKE (Akira)\n\nRANSOMHUB LOCKBIT BASTA\n\n### **5\\.4** %\n\n### **2\\.5** % **2\\.5** % **1\\.8** % **1\\.8** % **1\\.8** % **1\\.6** % **1\\.6** %\n\n**Most Frequently Seen Malware Families, 2024**\n\nOf note, in July of 2024, Europol ¹² provided an update on Operation MORPHEUS, a global action against the illicit use of the unlicensed versions of the Cobalt Strike red teaming tool. This operation, conducted with law enforcement and private sector partners, successfully disrupted infrastructure linked to cyber criminal activities. The initiative, which began in 2021, involved flagging 690 IP addresses, 593 of which were taken down by online service providers. Fortra, ¹³ the maintainers of the Cobalt Strike framework, also announced the number of unauthorized copies of Cobalt Strike observed in the wild has decreased by 80% over the past two years as a result of their participation in Operation MORPHEUS. Observed declines in percentages of inves- tigations where Mandiant identified BEACON since 2021 may reflect the success of this effort.","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**Mandiant M-Trends 2025 Report** **21**\n\n# **Most Frequently Seen Malware Families**\n\nFor the fifth consecutive year, BEACON was identified as the most frequently observed malware family in Mandiant investigations globally and was identified in 5.4% of all intrusions. BEACON usage has decreased dramatically since 2021, when it was observed in 28% of Mandiant investigations.\n\n30\n\n20\n\n10\n\n0\n\n2020 2021 2022 2023 2024\n\n### **24** % **28** %\n\n### **15** % **10** % **5** %\n\n**BEACON Usage, 2020-2024**\n\n**Most Frequently Seen Malware Families, 2024**\n\n6\n\n4\n\n2\n\n0\n\nBEACON GOOTLOADER WIREFIRE SYSTEMBC REDBIKE (Akira)\n\nRANSOMHUB LOCKBIT BASTA\n\n### **5\\.4** %\n\n### **2\\.5** % **2\\.5** % **1\\.8** % **1\\.8** % **1\\.8** % **1\\.6** % **1\\.6** %\n\n**Most Frequently Seen Malware Families, 2024**\n\nOf note, in July of 2024, Europol ¹² provided an update on Operation MORPHEUS, a global action against the illicit use of the unlicensed versions of the Cobalt Strike red teaming tool. This operation, conducted with law enforcement and private sector partners, successfully disrupted infrastructure linked to cyber criminal activities. The initiative, which began in 2021, involved flagging 690 IP addresses, 593 of which were taken down by online service providers. Fortra, ¹³ the maintainers of the Cobalt Strike framework, also announced the number of unauthorized copies of Cobalt Strike observed in the wild has decreased by 80% over the past two years as a result of their participation in Operation MORPHEUS. Observed declines in percentages of inves- tigations where Mandiant identified BEACON since 2021 may reflect the success of this effort.","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/m-trends-2025-en_p21.pdf"]},"response":{"errors":[],"extract_id":"extract_c525d0aaf389a903ad394e76e95f9302","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**Mandiant M-Trends 2025 Report** **21**\n\n# **Most Frequently Seen Malware Families**\n\nFor the fifth consecutive year, BEACON was identified as the most frequently observed malware family in Mandiant investigations globally and was identified in 5.4% of all intrusions. BEACON usage has decreased dramatically since 2021, when it was observed in 28% of Mandiant investigations.\n\n30\n\n20\n\n10\n\n0\n\n2020 2021 2022 2023 2024\n\n### **24** % **28** %\n\n### **15** % **10** % **5** %\n\n**BEACON Usage, 2020-2024**\n\n**Most Frequently Seen Malware Families, 2024**\n\n6\n\n4\n\n2\n\n0\n\nBEACON GOOTLOADER WIREFIRE SYSTEMBC REDBIKE (Akira)\n\nRANSOMHUB LOCKBIT BASTA\n\n### **5\\.4** %\n\n### **2\\.5** % **2\\.5** % **1\\.8** % **1\\.8** % **1\\.8** % **1\\.6** % **1\\.6** %\n\n**Most Frequently Seen Malware Families, 2024**\n\nOf note, in July of 2024, Europol ¹² provided an update on Operation MORPHEUS, a global action against the illicit use of the unlicensed versions of the Cobalt Strike red teaming tool. This operation, conducted with law enforcement and private sector partners, successfully disrupted infrastructure linked to cyber criminal activities. The initiative, which began in 2021, involved flagging 690 IP addresses, 593 of which were taken down by online service providers. Fortra, ¹³ the maintainers of the Cobalt Strike framework, also announced the number of unauthorized copies of Cobalt Strike observed in the wild has decreased by 80% over the past two years as a result of their participation in Operation MORPHEUS. Observed declines in percentages of inves- tigations where Mandiant identified BEACON since 2021 may reflect the success of this effort."],"full_content":"**Mandiant M-Trends 2025 Report** **21**\n\n# **Most Frequently Seen Malware Families**\n\nFor the fifth consecutive year, BEACON was identified as the most frequently observed malware family in Mandiant investigations globally and was identified in 5.4% of all intrusions. BEACON usage has decreased dramatically since 2021, when it was observed in 28% of Mandiant investigations.\n\n30\n\n20\n\n10\n\n0\n\n2020 2021 2022 2023 2024\n\n### **24** % **28** %\n\n### **15** % **10** % **5** %\n\n**BEACON Usage, 2020-2024**\n\n**Most Frequently Seen Malware Families, 2024**\n\n6\n\n4\n\n2\n\n0\n\nBEACON GOOTLOADER WIREFIRE SYSTEMBC REDBIKE (Akira)\n\nRANSOMHUB LOCKBIT BASTA\n\n### **5\\.4** %\n\n### **2\\.5** % **2\\.5** % **1\\.8** % **1\\.8** % **1\\.8** % **1\\.6** % **1\\.6** %\n\n**Most Frequently Seen Malware Families, 2024**\n\nOf note, in July of 2024, Europol ¹² provided an update on Operation MORPHEUS, a global action against the illicit use of the unlicensed versions of the Cobalt Strike red teaming tool. This operation, conducted with law enforcement and private sector partners, successfully disrupted infrastructure linked to cyber criminal activities. The initiative, which began in 2021, involved flagging 690 IP addresses, 593 of which were taken down by online service providers. Fortra, ¹³ the maintainers of the Cobalt Strike framework, also announced the number of unauthorized copies of Cobalt Strike observed in the wild has decreased by 80% over the past two years as a result of their participation in Operation MORPHEUS. Observed declines in percentages of inves- tigations where Mandiant identified BEACON since 2021 may reflect the success of this effort.","publish_date":null,"title":"Most Frequently Seen Malware Families","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/m-trends-2025-en_p21.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/m-trends-2025-en_p21.pdf"},"request":{"example_id":"chart/m-trends-2025-en_p21","product_type":"parse","source_file_path":"m-trends-2025-en_p21.pdf"},"started_at":"2026-11-04T19:44:58.716597"} +{"completed_at":"2026-11-04T19:45:00.886323","latency_in_ms":1181,"output":{"example_id":"chart/m-trends-2025-en_p29","grounded_pages":[],"layout_pages":[],"markdown":"**Mandiant M-Trends 2025 Report** **29**\n\n# **Malware**\n\nUnsurprisingly, the top malware category observed in 2024 ransomware intrusions was ransom- ware, which made up 34% of the malware data set. The next most prevalent categories are in line with the overall malware landscape observed in 2024. Credential stealers made up 12% of malware observed in ransomware-related intrusions, followed by backdoors (10%), utilities (10%), tunnelers (7%), downloaders (6%), and droppers (5%). The other 16% of malware families had other primary purposes such as keyloggers, launchers, installers, and uploaders.\n\nCompared to the overall metrics, ransomware- related intrusions saw a higher percentage of BEACON usage (15%). However, that may be attributable to the bias of the smaller dataset of ransom- ware-related intrusions rather than a true increase in the rate of BEACON usage when compared to all investigations. The next four most frequently observed malware families were ransomware varieties: RANSOMHUB (10%), REDBIKE (aka Akira) (10%), BASTA (9%), and LOCKBIT (9%). The tunneler SYSTEMBC (7%) was the sixth most commonly observed malware in ransomware-related intrusions, though it was the fourth most commonly observed family in all investigations. Several of these also appear in the overall most frequently seen malware families: BEACON, RANSOMHUB, REDBIKE, BASTA, LOCKBIT, and SYSTEMBC. The overlap of most frequently seen families for both overall and ransomware-related intrusions highlights how pervasive and prolific ransomware-related intrusions are.\n\nCompared to global metrics, ransomware- related intrusions saw more malware category variation; however, the ransomware-related malware dataset contains a much smaller proportion of backdoors. Within ransomware operations, this likely coincides with threat actors continuing to rely on remote control and administration tools. Credential stealers also make up double the percentage of the dataset in ransomware-related intrusions in 2024 compared to the overall dataset. Threat actors using ransomware are more likely to rely on publicly available and legitimate tools, such as credential extraction tools (credential stealers and remote administration tools), to accom- plish their objectives.\n\n### **Most Frequently Seen Malware, 2024**\n\nRansomware-Related\n\n% of Intrusions\n\n15\n\n10\n\n5\n\n0 BEACON\n\n**15** %\n\nRANSOMHUB\n\n**10** %\n\nREDBIKE (Akira)\n\n**10** %\n\nBASTA\n\n**9** %\n\nLOCKBIT\n\n**9** %\n\nSYSTEMBC\n\n**7** %\n\n**Malware Family**\n\n**Most Frequently Seen Malware, 2024 Ransomware-Related**\n\n### **Ransomware**\n\n# **34**\n\n%\n\n### **Other**\n\n# **16**\n\n# % **12**\n\n%\n\n**Credential Stealer**\n\n**Backdoor**\n\n# **10**\n\n%\n\n**Utility**\n\n# **10**\n\n%\n\n**Tunneler**\n\n# **7**\n\n%\n\n**Downloader**\n\n# **6**\n\n%\n\n**Dropper**\n\n# **5**\n\n%\n\n**Observed Malware Families by Category, 2024 Ransomware-Related**","pages":[{"line_numbers":[],"markdown":"**Mandiant M-Trends 2025 Report** **29**\n\n# **Malware**\n\nUnsurprisingly, the top malware category observed in 2024 ransomware intrusions was ransom- ware, which made up 34% of the malware data set. The next most prevalent categories are in line with the overall malware landscape observed in 2024. Credential stealers made up 12% of malware observed in ransomware-related intrusions, followed by backdoors (10%), utilities (10%), tunnelers (7%), downloaders (6%), and droppers (5%). The other 16% of malware families had other primary purposes such as keyloggers, launchers, installers, and uploaders.\n\nCompared to the overall metrics, ransomware- related intrusions saw a higher percentage of BEACON usage (15%). However, that may be attributable to the bias of the smaller dataset of ransom- ware-related intrusions rather than a true increase in the rate of BEACON usage when compared to all investigations. The next four most frequently observed malware families were ransomware varieties: RANSOMHUB (10%), REDBIKE (aka Akira) (10%), BASTA (9%), and LOCKBIT (9%). The tunneler SYSTEMBC (7%) was the sixth most commonly observed malware in ransomware-related intrusions, though it was the fourth most commonly observed family in all investigations. Several of these also appear in the overall most frequently seen malware families: BEACON, RANSOMHUB, REDBIKE, BASTA, LOCKBIT, and SYSTEMBC. The overlap of most frequently seen families for both overall and ransomware-related intrusions highlights how pervasive and prolific ransomware-related intrusions are.\n\nCompared to global metrics, ransomware- related intrusions saw more malware category variation; however, the ransomware-related malware dataset contains a much smaller proportion of backdoors. Within ransomware operations, this likely coincides with threat actors continuing to rely on remote control and administration tools. Credential stealers also make up double the percentage of the dataset in ransomware-related intrusions in 2024 compared to the overall dataset. Threat actors using ransomware are more likely to rely on publicly available and legitimate tools, such as credential extraction tools (credential stealers and remote administration tools), to accom- plish their objectives.\n\n### **Most Frequently Seen Malware, 2024**\n\nRansomware-Related\n\n% of Intrusions\n\n15\n\n10\n\n5\n\n0 BEACON\n\n**15** %\n\nRANSOMHUB\n\n**10** %\n\nREDBIKE (Akira)\n\n**10** %\n\nBASTA\n\n**9** %\n\nLOCKBIT\n\n**9** %\n\nSYSTEMBC\n\n**7** %\n\n**Malware Family**\n\n**Most Frequently Seen Malware, 2024 Ransomware-Related**\n\n### **Ransomware**\n\n# **34**\n\n%\n\n### **Other**\n\n# **16**\n\n# % **12**\n\n%\n\n**Credential Stealer**\n\n**Backdoor**\n\n# **10**\n\n%\n\n**Utility**\n\n# **10**\n\n%\n\n**Tunneler**\n\n# **7**\n\n%\n\n**Downloader**\n\n# **6**\n\n%\n\n**Dropper**\n\n# **5**\n\n%\n\n**Observed Malware Families by Category, 2024 Ransomware-Related**","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**Mandiant M-Trends 2025 Report** **29**\n\n# **Malware**\n\nUnsurprisingly, the top malware category observed in 2024 ransomware intrusions was ransom- ware, which made up 34% of the malware data set. The next most prevalent categories are in line with the overall malware landscape observed in 2024. Credential stealers made up 12% of malware observed in ransomware-related intrusions, followed by backdoors (10%), utilities (10%), tunnelers (7%), downloaders (6%), and droppers (5%). The other 16% of malware families had other primary purposes such as keyloggers, launchers, installers, and uploaders.\n\nCompared to the overall metrics, ransomware- related intrusions saw a higher percentage of BEACON usage (15%). However, that may be attributable to the bias of the smaller dataset of ransom- ware-related intrusions rather than a true increase in the rate of BEACON usage when compared to all investigations. The next four most frequently observed malware families were ransomware varieties: RANSOMHUB (10%), REDBIKE (aka Akira) (10%), BASTA (9%), and LOCKBIT (9%). The tunneler SYSTEMBC (7%) was the sixth most commonly observed malware in ransomware-related intrusions, though it was the fourth most commonly observed family in all investigations. Several of these also appear in the overall most frequently seen malware families: BEACON, RANSOMHUB, REDBIKE, BASTA, LOCKBIT, and SYSTEMBC. The overlap of most frequently seen families for both overall and ransomware-related intrusions highlights how pervasive and prolific ransomware-related intrusions are.\n\nCompared to global metrics, ransomware- related intrusions saw more malware category variation; however, the ransomware-related malware dataset contains a much smaller proportion of backdoors. Within ransomware operations, this likely coincides with threat actors continuing to rely on remote control and administration tools. Credential stealers also make up double the percentage of the dataset in ransomware-related intrusions in 2024 compared to the overall dataset. Threat actors using ransomware are more likely to rely on publicly available and legitimate tools, such as credential extraction tools (credential stealers and remote administration tools), to accom- plish their objectives.\n\n### **Most Frequently Seen Malware, 2024**\n\nRansomware-Related\n\n% of Intrusions\n\n15\n\n10\n\n5\n\n0 BEACON\n\n**15** %\n\nRANSOMHUB\n\n**10** %\n\nREDBIKE (Akira)\n\n**10** %\n\nBASTA\n\n**9** %\n\nLOCKBIT\n\n**9** %\n\nSYSTEMBC\n\n**7** %\n\n**Malware Family**\n\n**Most Frequently Seen Malware, 2024 Ransomware-Related**\n\n### **Ransomware**\n\n# **34**\n\n%\n\n### **Other**\n\n# **16**\n\n# % **12**\n\n%\n\n**Credential Stealer**\n\n**Backdoor**\n\n# **10**\n\n%\n\n**Utility**\n\n# **10**\n\n%\n\n**Tunneler**\n\n# **7**\n\n%\n\n**Downloader**\n\n# **6**\n\n%\n\n**Dropper**\n\n# **5**\n\n%\n\n**Observed Malware Families by Category, 2024 Ransomware-Related**","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/m-trends-2025-en_p29.pdf"]},"response":{"errors":[],"extract_id":"extract_8067655ecb8c52c51480415893f6250b","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**Mandiant M-Trends 2025 Report** **29**\n\n# **Malware**\n\nUnsurprisingly, the top malware category observed in 2024 ransomware intrusions was ransom- ware, which made up 34% of the malware data set. The next most prevalent categories are in line with the overall malware landscape observed in 2024. Credential stealers made up 12% of malware observed in ransomware-related intrusions, followed by backdoors (10%), utilities (10%), tunnelers (7%), downloaders (6%), and droppers (5%). The other 16% of malware families had other primary purposes such as keyloggers, launchers, installers, and uploaders.\n\nCompared to the overall metrics, ransomware- related intrusions saw a higher percentage of BEACON usage (15%). However, that may be attributable to the bias of the smaller dataset of ransom- ware-related intrusions rather than a true increase in the rate of BEACON usage when compared to all investigations. The next four most frequently observed malware families were ransomware varieties: RANSOMHUB (10%), REDBIKE (aka Akira) (10%), BASTA (9%), and LOCKBIT (9%). The tunneler SYSTEMBC (7%) was the sixth most commonly observed malware in ransomware-related intrusions, though it was the fourth most commonly observed family in all investigations. Several of these also appear in the overall most frequently seen malware families: BEACON, RANSOMHUB, REDBIKE, BASTA, LOCKBIT, and SYSTEMBC. The overlap of most frequently seen families for both overall and ransomware-related intrusions highlights how pervasive and prolific ransomware-related intrusions are.\n\nCompared to global metrics, ransomware- related intrusions saw more malware category variation; however, the ransomware-related malware dataset contains a much smaller proportion of backdoors. Within ransomware operations, this likely coincides with threat actors continuing to rely on remote control and administration tools. Credential stealers also make up double the percentage of the dataset in ransomware-related intrusions in 2024 compared to the overall dataset. Threat actors using ransomware are more likely to rely on publicly available and legitimate tools, such as credential extraction tools (credential stealers and remote administration tools), to accom- plish their objectives.\n\n### **Most Frequently Seen Malware, 2024**\n\nRansomware-Related\n\n% of Intrusions\n\n15\n\n10\n\n5\n\n0 BEACON\n\n**15** %\n\nRANSOMHUB\n\n**10** %\n\nREDBIKE (Akira)\n\n**10** %\n\nBASTA\n\n**9** %\n\nLOCKBIT\n\n**9** %\n\nSYSTEMBC\n\n**7** %\n\n**Malware Family**\n\n**Most Frequently Seen Malware, 2024 Ransomware-Related**\n\n### **Ransomware**\n\n# **34**\n\n%\n\n### **Other**\n\n# **16**\n\n# % **12**\n\n%\n\n**Credential Stealer**\n\n**Backdoor**\n\n# **10**\n\n%\n\n**Utility**\n\n# **10**\n\n%\n\n**Tunneler**\n\n# **7**\n\n%\n\n**Downloader**\n\n# **6**\n\n%\n\n**Dropper**\n\n# **5**\n\n%\n\n**Observed Malware Families by Category, 2024 Ransomware-Related**"],"full_content":"**Mandiant M-Trends 2025 Report** **29**\n\n# **Malware**\n\nUnsurprisingly, the top malware category observed in 2024 ransomware intrusions was ransom- ware, which made up 34% of the malware data set. The next most prevalent categories are in line with the overall malware landscape observed in 2024. Credential stealers made up 12% of malware observed in ransomware-related intrusions, followed by backdoors (10%), utilities (10%), tunnelers (7%), downloaders (6%), and droppers (5%). The other 16% of malware families had other primary purposes such as keyloggers, launchers, installers, and uploaders.\n\nCompared to the overall metrics, ransomware- related intrusions saw a higher percentage of BEACON usage (15%). However, that may be attributable to the bias of the smaller dataset of ransom- ware-related intrusions rather than a true increase in the rate of BEACON usage when compared to all investigations. The next four most frequently observed malware families were ransomware varieties: RANSOMHUB (10%), REDBIKE (aka Akira) (10%), BASTA (9%), and LOCKBIT (9%). The tunneler SYSTEMBC (7%) was the sixth most commonly observed malware in ransomware-related intrusions, though it was the fourth most commonly observed family in all investigations. Several of these also appear in the overall most frequently seen malware families: BEACON, RANSOMHUB, REDBIKE, BASTA, LOCKBIT, and SYSTEMBC. The overlap of most frequently seen families for both overall and ransomware-related intrusions highlights how pervasive and prolific ransomware-related intrusions are.\n\nCompared to global metrics, ransomware- related intrusions saw more malware category variation; however, the ransomware-related malware dataset contains a much smaller proportion of backdoors. Within ransomware operations, this likely coincides with threat actors continuing to rely on remote control and administration tools. Credential stealers also make up double the percentage of the dataset in ransomware-related intrusions in 2024 compared to the overall dataset. Threat actors using ransomware are more likely to rely on publicly available and legitimate tools, such as credential extraction tools (credential stealers and remote administration tools), to accom- plish their objectives.\n\n### **Most Frequently Seen Malware, 2024**\n\nRansomware-Related\n\n% of Intrusions\n\n15\n\n10\n\n5\n\n0 BEACON\n\n**15** %\n\nRANSOMHUB\n\n**10** %\n\nREDBIKE (Akira)\n\n**10** %\n\nBASTA\n\n**9** %\n\nLOCKBIT\n\n**9** %\n\nSYSTEMBC\n\n**7** %\n\n**Malware Family**\n\n**Most Frequently Seen Malware, 2024 Ransomware-Related**\n\n### **Ransomware**\n\n# **34**\n\n%\n\n### **Other**\n\n# **16**\n\n# % **12**\n\n%\n\n**Credential Stealer**\n\n**Backdoor**\n\n# **10**\n\n%\n\n**Utility**\n\n# **10**\n\n%\n\n**Tunneler**\n\n# **7**\n\n%\n\n**Downloader**\n\n# **6**\n\n%\n\n**Dropper**\n\n# **5**\n\n%\n\n**Observed Malware Families by Category, 2024 Ransomware-Related**","publish_date":null,"title":"34","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/m-trends-2025-en_p29.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/m-trends-2025-en_p29.pdf"},"request":{"example_id":"chart/m-trends-2025-en_p29","product_type":"parse","source_file_path":"m-trends-2025-en_p29.pdf"},"started_at":"2026-11-04T19:44:59.704416"} +{"completed_at":"2026-11-04T19:45:08.552366","latency_in_ms":7341,"output":{"example_id":"chart/m-trends-2025-en_p32","grounded_pages":[],"layout_pages":[],"markdown":"**Mandiant M-Trends 2025 Report** **32**\n\n**Data leak sites (DLS)** are websites that publish stolen data of companies that refuse to pay a ransom. While this data is skewed toward targets who refused to pay attackers’ ransom demands, it is still useful for understanding broad trends in extortion operations.\n\n# **Ransomware Operations**\n\n## **RANSOMHUB**\n\nThe RANSOMHUB ransomware-as-a-service (RaaS) and associated DLS launched in early 2024. By the second half of 2024, RANSOMHUB RaaS became the most prolific DLS that Mandiant tracks, taking the top spot from LockBit after its activity declined following law enforcement action. RANSOMHUB was also tied for most frequently observed ransomware in Mandiant incident response investigations performed in 2024. Mandiant currently tracks multiple threat clusters that have used this ransomware brand, including UNC2165, UNC5227, and others.\n\nJan Feb March April May June July Aug Sept Oct Nov Dec\n\nCount of DLS Listings\n\n100\n\n0\n\n150\n\n50\n\n200\n\n0\n\nRANSOMHUB\n\nLockBit\n\n**LockBit vs. RANSOMHUB DLS Listings, 2024**\n\n**UNC2165** ¹⁸ is a financially motivated threat cluster that has been active since at least 2019 and has conducted ransomware and data theft extortion operations using HADES, LOCKBIT, CONTI, and RANSOMHUB ransomware. UNC2165 has primarily gained access to victim organizations from FAKEUPDATES infections, although, since late 2020, some intrusions appeared to leverage stolen credentials. UNC2165 has used various methods to escalate privileges conducting Mimikatz and Kerberoasting attacks, targeting authentication data stored in the Windows registry, and searching for documents or files associated with password managers or that may contain plaintext credentials. Historically, UNC2165 operations heavily relied on BEACON for lateral movement and to maintain access to the victim environment; however, since late 2023, UNC2165 has used the MYTHIC post-exploitation framework and VIPERTUNNEL tunneler in intrusions. In most cases, UNC2165 has also stolen data from victims using Rclone or MEGASync.\n\n**UNC5227** is a financially motivated threat cluster active since at least November 2023 that has monetized access via ransomware deployment and data theft extortion. In some cases, UNC5227 has gained access to victim networks via brute-force attacks or stolen VPN credentials obtained from a separate threat cluster. UNC5227 relies on open-source tools, including MIMIKATZ and OPENSSH, to compromise additional accounts and move laterally through the network. They have also used PORTLIGHT, a custom Windows PowerShell utility for port-forwarding access using SecureShell (SSH) to maintain persistence, which may be exclusive to UNC5227. UNC5227 also uses EXMATTER, a private file upload tool, on compromised devices for data staging and theft before deploying ransomware. UNC5227 has deployed LOCKBIT.BLACK, ALPHV, RHYSIDA, and RANSOMHUB ransomware, based on direct observations as well as overlaps observed in the wild with EXMATTER and reverse Secure Shell (SSH) infrastructure.","pages":[{"line_numbers":[],"markdown":"**Mandiant M-Trends 2025 Report** **32**\n\n**Data leak sites (DLS)** are websites that publish stolen data of companies that refuse to pay a ransom. While this data is skewed toward targets who refused to pay attackers’ ransom demands, it is still useful for understanding broad trends in extortion operations.\n\n# **Ransomware Operations**\n\n## **RANSOMHUB**\n\nThe RANSOMHUB ransomware-as-a-service (RaaS) and associated DLS launched in early 2024. By the second half of 2024, RANSOMHUB RaaS became the most prolific DLS that Mandiant tracks, taking the top spot from LockBit after its activity declined following law enforcement action. RANSOMHUB was also tied for most frequently observed ransomware in Mandiant incident response investigations performed in 2024. Mandiant currently tracks multiple threat clusters that have used this ransomware brand, including UNC2165, UNC5227, and others.\n\nJan Feb March April May June July Aug Sept Oct Nov Dec\n\nCount of DLS Listings\n\n100\n\n0\n\n150\n\n50\n\n200\n\n0\n\nRANSOMHUB\n\nLockBit\n\n**LockBit vs. RANSOMHUB DLS Listings, 2024**\n\n**UNC2165** ¹⁸ is a financially motivated threat cluster that has been active since at least 2019 and has conducted ransomware and data theft extortion operations using HADES, LOCKBIT, CONTI, and RANSOMHUB ransomware. UNC2165 has primarily gained access to victim organizations from FAKEUPDATES infections, although, since late 2020, some intrusions appeared to leverage stolen credentials. UNC2165 has used various methods to escalate privileges conducting Mimikatz and Kerberoasting attacks, targeting authentication data stored in the Windows registry, and searching for documents or files associated with password managers or that may contain plaintext credentials. Historically, UNC2165 operations heavily relied on BEACON for lateral movement and to maintain access to the victim environment; however, since late 2023, UNC2165 has used the MYTHIC post-exploitation framework and VIPERTUNNEL tunneler in intrusions. In most cases, UNC2165 has also stolen data from victims using Rclone or MEGASync.\n\n**UNC5227** is a financially motivated threat cluster active since at least November 2023 that has monetized access via ransomware deployment and data theft extortion. In some cases, UNC5227 has gained access to victim networks via brute-force attacks or stolen VPN credentials obtained from a separate threat cluster. UNC5227 relies on open-source tools, including MIMIKATZ and OPENSSH, to compromise additional accounts and move laterally through the network. They have also used PORTLIGHT, a custom Windows PowerShell utility for port-forwarding access using SecureShell (SSH) to maintain persistence, which may be exclusive to UNC5227. UNC5227 also uses EXMATTER, a private file upload tool, on compromised devices for data staging and theft before deploying ransomware. UNC5227 has deployed LOCKBIT.BLACK, ALPHV, RHYSIDA, and RANSOMHUB ransomware, based on direct observations as well as overlaps observed in the wild with EXMATTER and reverse Secure Shell (SSH) infrastructure.","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**Mandiant M-Trends 2025 Report** **32**\n\n**Data leak sites (DLS)** are websites that publish stolen data of companies that refuse to pay a ransom. While this data is skewed toward targets who refused to pay attackers’ ransom demands, it is still useful for understanding broad trends in extortion operations.\n\n# **Ransomware Operations**\n\n## **RANSOMHUB**\n\nThe RANSOMHUB ransomware-as-a-service (RaaS) and associated DLS launched in early 2024. By the second half of 2024, RANSOMHUB RaaS became the most prolific DLS that Mandiant tracks, taking the top spot from LockBit after its activity declined following law enforcement action. RANSOMHUB was also tied for most frequently observed ransomware in Mandiant incident response investigations performed in 2024. Mandiant currently tracks multiple threat clusters that have used this ransomware brand, including UNC2165, UNC5227, and others.\n\nJan Feb March April May June July Aug Sept Oct Nov Dec\n\nCount of DLS Listings\n\n100\n\n0\n\n150\n\n50\n\n200\n\n0\n\nRANSOMHUB\n\nLockBit\n\n**LockBit vs. RANSOMHUB DLS Listings, 2024**\n\n**UNC2165** ¹⁸ is a financially motivated threat cluster that has been active since at least 2019 and has conducted ransomware and data theft extortion operations using HADES, LOCKBIT, CONTI, and RANSOMHUB ransomware. UNC2165 has primarily gained access to victim organizations from FAKEUPDATES infections, although, since late 2020, some intrusions appeared to leverage stolen credentials. UNC2165 has used various methods to escalate privileges conducting Mimikatz and Kerberoasting attacks, targeting authentication data stored in the Windows registry, and searching for documents or files associated with password managers or that may contain plaintext credentials. Historically, UNC2165 operations heavily relied on BEACON for lateral movement and to maintain access to the victim environment; however, since late 2023, UNC2165 has used the MYTHIC post-exploitation framework and VIPERTUNNEL tunneler in intrusions. In most cases, UNC2165 has also stolen data from victims using Rclone or MEGASync.\n\n**UNC5227** is a financially motivated threat cluster active since at least November 2023 that has monetized access via ransomware deployment and data theft extortion. In some cases, UNC5227 has gained access to victim networks via brute-force attacks or stolen VPN credentials obtained from a separate threat cluster. UNC5227 relies on open-source tools, including MIMIKATZ and OPENSSH, to compromise additional accounts and move laterally through the network. They have also used PORTLIGHT, a custom Windows PowerShell utility for port-forwarding access using SecureShell (SSH) to maintain persistence, which may be exclusive to UNC5227. UNC5227 also uses EXMATTER, a private file upload tool, on compromised devices for data staging and theft before deploying ransomware. UNC5227 has deployed LOCKBIT.BLACK, ALPHV, RHYSIDA, and RANSOMHUB ransomware, based on direct observations as well as overlaps observed in the wild with EXMATTER and reverse Secure Shell (SSH) infrastructure.","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/m-trends-2025-en_p32.pdf"]},"response":{"errors":[],"extract_id":"extract_9b3cb77bb246a249ba2c65bd190081d4","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**Mandiant M-Trends 2025 Report** **32**\n\n**Data leak sites (DLS)** are websites that publish stolen data of companies that refuse to pay a ransom. While this data is skewed toward targets who refused to pay attackers’ ransom demands, it is still useful for understanding broad trends in extortion operations.\n\n# **Ransomware Operations**\n\n## **RANSOMHUB**\n\nThe RANSOMHUB ransomware-as-a-service (RaaS) and associated DLS launched in early 2024. By the second half of 2024, RANSOMHUB RaaS became the most prolific DLS that Mandiant tracks, taking the top spot from LockBit after its activity declined following law enforcement action. RANSOMHUB was also tied for most frequently observed ransomware in Mandiant incident response investigations performed in 2024. Mandiant currently tracks multiple threat clusters that have used this ransomware brand, including UNC2165, UNC5227, and others.\n\nJan Feb March April May June July Aug Sept Oct Nov Dec\n\nCount of DLS Listings\n\n100\n\n0\n\n150\n\n50\n\n200\n\n0\n\nRANSOMHUB\n\nLockBit\n\n**LockBit vs. RANSOMHUB DLS Listings, 2024**\n\n**UNC2165** ¹⁸ is a financially motivated threat cluster that has been active since at least 2019 and has conducted ransomware and data theft extortion operations using HADES, LOCKBIT, CONTI, and RANSOMHUB ransomware. UNC2165 has primarily gained access to victim organizations from FAKEUPDATES infections, although, since late 2020, some intrusions appeared to leverage stolen credentials. UNC2165 has used various methods to escalate privileges conducting Mimikatz and Kerberoasting attacks, targeting authentication data stored in the Windows registry, and searching for documents or files associated with password managers or that may contain plaintext credentials. Historically, UNC2165 operations heavily relied on BEACON for lateral movement and to maintain access to the victim environment; however, since late 2023, UNC2165 has used the MYTHIC post-exploitation framework and VIPERTUNNEL tunneler in intrusions. In most cases, UNC2165 has also stolen data from victims using Rclone or MEGASync.\n\n**UNC5227** is a financially motivated threat cluster active since at least November 2023 that has monetized access via ransomware deployment and data theft extortion. In some cases, UNC5227 has gained access to victim networks via brute-force attacks or stolen VPN credentials obtained from a separate threat cluster. UNC5227 relies on open-source tools, including MIMIKATZ and OPENSSH, to compromise additional accounts and move laterally through the network. They have also used PORTLIGHT, a custom Windows PowerShell utility for port-forwarding access using SecureShell (SSH) to maintain persistence, which may be exclusive to UNC5227. UNC5227 also uses EXMATTER, a private file upload tool, on compromised devices for data staging and theft before deploying ransomware. UNC5227 has deployed LOCKBIT.BLACK, ALPHV, RHYSIDA, and RANSOMHUB ransomware, based on direct observations as well as overlaps observed in the wild with EXMATTER and reverse Secure Shell (SSH) infrastructure."],"full_content":"**Mandiant M-Trends 2025 Report** **32**\n\n**Data leak sites (DLS)** are websites that publish stolen data of companies that refuse to pay a ransom. While this data is skewed toward targets who refused to pay attackers’ ransom demands, it is still useful for understanding broad trends in extortion operations.\n\n# **Ransomware Operations**\n\n## **RANSOMHUB**\n\nThe RANSOMHUB ransomware-as-a-service (RaaS) and associated DLS launched in early 2024. By the second half of 2024, RANSOMHUB RaaS became the most prolific DLS that Mandiant tracks, taking the top spot from LockBit after its activity declined following law enforcement action. RANSOMHUB was also tied for most frequently observed ransomware in Mandiant incident response investigations performed in 2024. Mandiant currently tracks multiple threat clusters that have used this ransomware brand, including UNC2165, UNC5227, and others.\n\nJan Feb March April May June July Aug Sept Oct Nov Dec\n\nCount of DLS Listings\n\n100\n\n0\n\n150\n\n50\n\n200\n\n0\n\nRANSOMHUB\n\nLockBit\n\n**LockBit vs. RANSOMHUB DLS Listings, 2024**\n\n**UNC2165** ¹⁸ is a financially motivated threat cluster that has been active since at least 2019 and has conducted ransomware and data theft extortion operations using HADES, LOCKBIT, CONTI, and RANSOMHUB ransomware. UNC2165 has primarily gained access to victim organizations from FAKEUPDATES infections, although, since late 2020, some intrusions appeared to leverage stolen credentials. UNC2165 has used various methods to escalate privileges conducting Mimikatz and Kerberoasting attacks, targeting authentication data stored in the Windows registry, and searching for documents or files associated with password managers or that may contain plaintext credentials. Historically, UNC2165 operations heavily relied on BEACON for lateral movement and to maintain access to the victim environment; however, since late 2023, UNC2165 has used the MYTHIC post-exploitation framework and VIPERTUNNEL tunneler in intrusions. In most cases, UNC2165 has also stolen data from victims using Rclone or MEGASync.\n\n**UNC5227** is a financially motivated threat cluster active since at least November 2023 that has monetized access via ransomware deployment and data theft extortion. In some cases, UNC5227 has gained access to victim networks via brute-force attacks or stolen VPN credentials obtained from a separate threat cluster. UNC5227 relies on open-source tools, including MIMIKATZ and OPENSSH, to compromise additional accounts and move laterally through the network. They have also used PORTLIGHT, a custom Windows PowerShell utility for port-forwarding access using SecureShell (SSH) to maintain persistence, which may be exclusive to UNC5227. UNC5227 also uses EXMATTER, a private file upload tool, on compromised devices for data staging and theft before deploying ransomware. UNC5227 has deployed LOCKBIT.BLACK, ALPHV, RHYSIDA, and RANSOMHUB ransomware, based on direct observations as well as overlaps observed in the wild with EXMATTER and reverse Secure Shell (SSH) infrastructure.","publish_date":null,"title":"Ransomware Operations","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/m-trends-2025-en_p32.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/m-trends-2025-en_p32.pdf"},"request":{"example_id":"chart/m-trends-2025-en_p32","product_type":"parse","source_file_path":"m-trends-2025-en_p32.pdf"},"started_at":"2026-11-04T19:45:01.210466"} +{"completed_at":"2026-11-04T19:45:04.419431","latency_in_ms":2016,"output":{"example_id":"chart/m-trends-2025-en_p41","grounded_pages":[],"layout_pages":[],"markdown":"**Mandiant M-Trends 2025 Report** **41**\n\n**tial Infection Vectors by Region-EMEA Map**\n\n# **Targeted Attacks**\n\n# **Initial Infection Vector**\n\nThe most frequently identified initial infection vectors in Mandiant incident response investi- gations in EMEA in 2024 were exploits (39%), followed by email phishing (15%) and brute-force attacks (10%). In EMEA, email phishing and brute-force attacks represented larger proportions of observed initial infection vectors than Mandiant encountered in global investigations.\n\n# **Detection by Source**\n\nThe intrusions that Mandiant investigated in EMEA in 2024 were first discovered internally 41% of the time, while in 59% of cases, an external organization first notified organizations of a compromise. These figures are similar to the global numbers (43% internal and 57% external).\n\nThe metrics reported in this section are based on Mandiant Consulting investigations affecting organizations in Europe, the Middle East, and Africa (EMEA).\n\n# **EMEA**\n\nInternal Detection (Percent)\n\nExternal Detection (Percent)\n\n**EMEA Detection by Source, 2017-2024**\n\n100\n\n80\n\n60\n\n0\n\n40\n\n20\n\n100\n\n80\n\n60\n\n0\n\n40\n\n20\n\n2017 2018 2019 2020 2021 2022 2023 2024\n\nInternal\n\nAverage Trend\n\n**EMEA Detection by Source, 2017-2024**\n\n**EMEA Exploit 39%**\n\n**Email Phishing 15%**\n\n**Brute Force 10%**","pages":[{"line_numbers":[],"markdown":"**Mandiant M-Trends 2025 Report** **41**\n\n**tial Infection Vectors by Region-EMEA Map**\n\n# **Targeted Attacks**\n\n# **Initial Infection Vector**\n\nThe most frequently identified initial infection vectors in Mandiant incident response investi- gations in EMEA in 2024 were exploits (39%), followed by email phishing (15%) and brute-force attacks (10%). In EMEA, email phishing and brute-force attacks represented larger proportions of observed initial infection vectors than Mandiant encountered in global investigations.\n\n# **Detection by Source**\n\nThe intrusions that Mandiant investigated in EMEA in 2024 were first discovered internally 41% of the time, while in 59% of cases, an external organization first notified organizations of a compromise. These figures are similar to the global numbers (43% internal and 57% external).\n\nThe metrics reported in this section are based on Mandiant Consulting investigations affecting organizations in Europe, the Middle East, and Africa (EMEA).\n\n# **EMEA**\n\nInternal Detection (Percent)\n\nExternal Detection (Percent)\n\n**EMEA Detection by Source, 2017-2024**\n\n100\n\n80\n\n60\n\n0\n\n40\n\n20\n\n100\n\n80\n\n60\n\n0\n\n40\n\n20\n\n2017 2018 2019 2020 2021 2022 2023 2024\n\nInternal\n\nAverage Trend\n\n**EMEA Detection by Source, 2017-2024**\n\n**EMEA Exploit 39%**\n\n**Email Phishing 15%**\n\n**Brute Force 10%**","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**Mandiant M-Trends 2025 Report** **41**\n\n**tial Infection Vectors by Region-EMEA Map**\n\n# **Targeted Attacks**\n\n# **Initial Infection Vector**\n\nThe most frequently identified initial infection vectors in Mandiant incident response investi- gations in EMEA in 2024 were exploits (39%), followed by email phishing (15%) and brute-force attacks (10%). In EMEA, email phishing and brute-force attacks represented larger proportions of observed initial infection vectors than Mandiant encountered in global investigations.\n\n# **Detection by Source**\n\nThe intrusions that Mandiant investigated in EMEA in 2024 were first discovered internally 41% of the time, while in 59% of cases, an external organization first notified organizations of a compromise. These figures are similar to the global numbers (43% internal and 57% external).\n\nThe metrics reported in this section are based on Mandiant Consulting investigations affecting organizations in Europe, the Middle East, and Africa (EMEA).\n\n# **EMEA**\n\nInternal Detection (Percent)\n\nExternal Detection (Percent)\n\n**EMEA Detection by Source, 2017-2024**\n\n100\n\n80\n\n60\n\n0\n\n40\n\n20\n\n100\n\n80\n\n60\n\n0\n\n40\n\n20\n\n2017 2018 2019 2020 2021 2022 2023 2024\n\nInternal\n\nAverage Trend\n\n**EMEA Detection by Source, 2017-2024**\n\n**EMEA Exploit 39%**\n\n**Email Phishing 15%**\n\n**Brute Force 10%**","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/m-trends-2025-en_p41.pdf"]},"response":{"errors":[],"extract_id":"extract_dfcb0e51902b4d67725ba3f6084ad022","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**Mandiant M-Trends 2025 Report** **41**\n\n**tial Infection Vectors by Region-EMEA Map**\n\n# **Targeted Attacks**\n\n# **Initial Infection Vector**\n\nThe most frequently identified initial infection vectors in Mandiant incident response investi- gations in EMEA in 2024 were exploits (39%), followed by email phishing (15%) and brute-force attacks (10%). In EMEA, email phishing and brute-force attacks represented larger proportions of observed initial infection vectors than Mandiant encountered in global investigations.\n\n# **Detection by Source**\n\nThe intrusions that Mandiant investigated in EMEA in 2024 were first discovered internally 41% of the time, while in 59% of cases, an external organization first notified organizations of a compromise. These figures are similar to the global numbers (43% internal and 57% external).\n\nThe metrics reported in this section are based on Mandiant Consulting investigations affecting organizations in Europe, the Middle East, and Africa (EMEA).\n\n# **EMEA**\n\nInternal Detection (Percent)\n\nExternal Detection (Percent)\n\n**EMEA Detection by Source, 2017-2024**\n\n100\n\n80\n\n60\n\n0\n\n40\n\n20\n\n100\n\n80\n\n60\n\n0\n\n40\n\n20\n\n2017 2018 2019 2020 2021 2022 2023 2024\n\nInternal\n\nAverage Trend\n\n**EMEA Detection by Source, 2017-2024**\n\n**EMEA Exploit 39%**\n\n**Email Phishing 15%**\n\n**Brute Force 10%**"],"full_content":"**Mandiant M-Trends 2025 Report** **41**\n\n**tial Infection Vectors by Region-EMEA Map**\n\n# **Targeted Attacks**\n\n# **Initial Infection Vector**\n\nThe most frequently identified initial infection vectors in Mandiant incident response investi- gations in EMEA in 2024 were exploits (39%), followed by email phishing (15%) and brute-force attacks (10%). In EMEA, email phishing and brute-force attacks represented larger proportions of observed initial infection vectors than Mandiant encountered in global investigations.\n\n# **Detection by Source**\n\nThe intrusions that Mandiant investigated in EMEA in 2024 were first discovered internally 41% of the time, while in 59% of cases, an external organization first notified organizations of a compromise. These figures are similar to the global numbers (43% internal and 57% external).\n\nThe metrics reported in this section are based on Mandiant Consulting investigations affecting organizations in Europe, the Middle East, and Africa (EMEA).\n\n# **EMEA**\n\nInternal Detection (Percent)\n\nExternal Detection (Percent)\n\n**EMEA Detection by Source, 2017-2024**\n\n100\n\n80\n\n60\n\n0\n\n40\n\n20\n\n100\n\n80\n\n60\n\n0\n\n40\n\n20\n\n2017 2018 2019 2020 2021 2022 2023 2024\n\nInternal\n\nAverage Trend\n\n**EMEA Detection by Source, 2017-2024**\n\n**EMEA Exploit 39%**\n\n**Email Phishing 15%**\n\n**Brute Force 10%**","publish_date":null,"title":"EMEA","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/m-trends-2025-en_p41.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/m-trends-2025-en_p41.pdf"},"request":{"example_id":"chart/m-trends-2025-en_p41","product_type":"parse","source_file_path":"m-trends-2025-en_p41.pdf"},"started_at":"2026-11-04T19:45:02.402658"} +{"completed_at":"2026-11-04T19:31:08.441992","latency_in_ms":1311,"output":{"example_id":"chart/modeling-insured-catastrophe-loss-a-global-perspective-for-2025_p18","grounded_pages":[],"layout_pages":[],"markdown":"It’s not yet clear whether regional environmental trends have specific causes. The amplitude and pattern of some regional trends align with expected end-of-the- century impacts from global warming. However, this amplitude is larger than expected by the emerging understanding of how greenhouse gases would eventually affect severe thunderstorms in the long term. Some of these trends may reflect inherent variability in the ocean and atmosphere that spans decades. For instance, regional trends in severe thunderstorm environments show no global consistency that would point to a common cause.\n\nThe share of risk from frequency perils is increasing primarily because recent activity offers new data and clarity about elevated risk in certain geographic corridors.\n\nVerisk peril models reflect the state of the science from two perspectives: leveraging a variety of limited data sources and incorporating a process-level understanding of peril activity. This informs our view of near-present climate around the globe—that surprisingly persistent regional environmental factors drive elevated losses from frequency perils, thus complicating the discussion of how climate change affects trends in modeled catastrophe hazards.\n\n# 12%\n\n### increase in the contributions of frequency perils to the total modeled risk from 2025 over 2024\n\n_Note: 2024 USD adjusted for inflation and exposure growth. Some severe thunderstorm_ _and winter storm events may also include losses from inland flood._\n\nFigure 4. Comparison of insured losses exceeding $1 billion by peril between 2015–2019 and 2020–2024 (Source: Verisk, Aon, Swiss Re)\n\nTropical Cyclone\n\nEarthquake\n\nSevere Thunderstorm\n\nWinter Storm\n\nWildfire\n\nInland Flood","pages":[{"line_numbers":[],"markdown":"It’s not yet clear whether regional environmental trends have specific causes. The amplitude and pattern of some regional trends align with expected end-of-the- century impacts from global warming. However, this amplitude is larger than expected by the emerging understanding of how greenhouse gases would eventually affect severe thunderstorms in the long term. Some of these trends may reflect inherent variability in the ocean and atmosphere that spans decades. For instance, regional trends in severe thunderstorm environments show no global consistency that would point to a common cause.\n\nThe share of risk from frequency perils is increasing primarily because recent activity offers new data and clarity about elevated risk in certain geographic corridors.\n\nVerisk peril models reflect the state of the science from two perspectives: leveraging a variety of limited data sources and incorporating a process-level understanding of peril activity. This informs our view of near-present climate around the globe—that surprisingly persistent regional environmental factors drive elevated losses from frequency perils, thus complicating the discussion of how climate change affects trends in modeled catastrophe hazards.\n\n# 12%\n\n### increase in the contributions of frequency perils to the total modeled risk from 2025 over 2024\n\n_Note: 2024 USD adjusted for inflation and exposure growth. Some severe thunderstorm_ _and winter storm events may also include losses from inland flood._\n\nFigure 4. Comparison of insured losses exceeding $1 billion by peril between 2015–2019 and 2020–2024 (Source: Verisk, Aon, Swiss Re)\n\nTropical Cyclone\n\nEarthquake\n\nSevere Thunderstorm\n\nWinter Storm\n\nWildfire\n\nInland Flood","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"It’s not yet clear whether regional environmental trends have specific causes. The amplitude and pattern of some regional trends align with expected end-of-the- century impacts from global warming. However, this amplitude is larger than expected by the emerging understanding of how greenhouse gases would eventually affect severe thunderstorms in the long term. Some of these trends may reflect inherent variability in the ocean and atmosphere that spans decades. For instance, regional trends in severe thunderstorm environments show no global consistency that would point to a common cause.\n\nThe share of risk from frequency perils is increasing primarily because recent activity offers new data and clarity about elevated risk in certain geographic corridors.\n\nVerisk peril models reflect the state of the science from two perspectives: leveraging a variety of limited data sources and incorporating a process-level understanding of peril activity. This informs our view of near-present climate around the globe—that surprisingly persistent regional environmental factors drive elevated losses from frequency perils, thus complicating the discussion of how climate change affects trends in modeled catastrophe hazards.\n\n# 12%\n\n### increase in the contributions of frequency perils to the total modeled risk from 2025 over 2024\n\n_Note: 2024 USD adjusted for inflation and exposure growth. Some severe thunderstorm_ _and winter storm events may also include losses from inland flood._\n\nFigure 4. Comparison of insured losses exceeding $1 billion by peril between 2015–2019 and 2020–2024 (Source: Verisk, Aon, Swiss Re)\n\nTropical Cyclone\n\nEarthquake\n\nSevere Thunderstorm\n\nWinter Storm\n\nWildfire\n\nInland Flood","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/modeling-insured-catastrophe-loss-a-global-perspective-for-2025_p18.pdf"]},"response":{"errors":[],"extract_id":"extract_fef3697304c31ead61cf234e5499edf0","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["It’s not yet clear whether regional environmental trends have specific causes. The amplitude and pattern of some regional trends align with expected end-of-the- century impacts from global warming. However, this amplitude is larger than expected by the emerging understanding of how greenhouse gases would eventually affect severe thunderstorms in the long term. Some of these trends may reflect inherent variability in the ocean and atmosphere that spans decades. For instance, regional trends in severe thunderstorm environments show no global consistency that would point to a common cause.\n\nThe share of risk from frequency perils is increasing primarily because recent activity offers new data and clarity about elevated risk in certain geographic corridors.\n\nVerisk peril models reflect the state of the science from two perspectives: leveraging a variety of limited data sources and incorporating a process-level understanding of peril activity. This informs our view of near-present climate around the globe—that surprisingly persistent regional environmental factors drive elevated losses from frequency perils, thus complicating the discussion of how climate change affects trends in modeled catastrophe hazards.\n\n# 12%\n\n### increase in the contributions of frequency perils to the total modeled risk from 2025 over 2024\n\n_Note: 2024 USD adjusted for inflation and exposure growth. Some severe thunderstorm_ _and winter storm events may also include losses from inland flood._\n\nFigure 4. Comparison of insured losses exceeding $1 billion by peril between 2015–2019 and 2020–2024 (Source: Verisk, Aon, Swiss Re)\n\nTropical Cyclone\n\nEarthquake\n\nSevere Thunderstorm\n\nWinter Storm\n\nWildfire\n\nInland Flood"],"full_content":"It’s not yet clear whether regional environmental trends have specific causes. The amplitude and pattern of some regional trends align with expected end-of-the- century impacts from global warming. However, this amplitude is larger than expected by the emerging understanding of how greenhouse gases would eventually affect severe thunderstorms in the long term. Some of these trends may reflect inherent variability in the ocean and atmosphere that spans decades. For instance, regional trends in severe thunderstorm environments show no global consistency that would point to a common cause.\n\nThe share of risk from frequency perils is increasing primarily because recent activity offers new data and clarity about elevated risk in certain geographic corridors.\n\nVerisk peril models reflect the state of the science from two perspectives: leveraging a variety of limited data sources and incorporating a process-level understanding of peril activity. This informs our view of near-present climate around the globe—that surprisingly persistent regional environmental factors drive elevated losses from frequency perils, thus complicating the discussion of how climate change affects trends in modeled catastrophe hazards.\n\n# 12%\n\n### increase in the contributions of frequency perils to the total modeled risk from 2025 over 2024\n\n_Note: 2024 USD adjusted for inflation and exposure growth. Some severe thunderstorm_ _and winter storm events may also include losses from inland flood._\n\nFigure 4. Comparison of insured losses exceeding $1 billion by peril between 2015–2019 and 2020–2024 (Source: Verisk, Aon, Swiss Re)\n\nTropical Cyclone\n\nEarthquake\n\nSevere Thunderstorm\n\nWinter Storm\n\nWildfire\n\nInland Flood","publish_date":null,"title":"12","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/modeling-insured-catastrophe-loss-a-global-perspective-for-2025_p18.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/modeling-insured-catastrophe-loss-a-global-perspective-for-2025_p18.pdf"},"request":{"example_id":"chart/modeling-insured-catastrophe-loss-a-global-perspective-for-2025_p18","product_type":"parse","source_file_path":"modeling-insured-catastrophe-loss-a-global-perspective-for-2025_p18.pdf"},"started_at":"2026-11-04T19:31:07.130589"} +{"completed_at":"2026-11-04T19:45:08.477596","latency_in_ms":3444,"output":{"example_id":"chart/mts0625_p6","grounded_pages":[],"layout_pages":[],"markdown":"6 _Source_ _d_ _ata: Table 1_\n\nReceipts Outlays Deficit(-)/Surplus\n\n**2024 2025**\n\n**Oct**\n\n**Nov**\n\n**Dec**\n\n**Jan**\n\n**Feb**\n\n**Mar**\n\n**Apr**\n\n**May**\n\n**Jun**\n\n**Jul**\n\n**Aug**\n\n**Sep**\n\n**Oct**\n\n**Nov**\n\n**Dec**\n\n**Jan**\n\n**Feb**\n\n**Mar**\n\n**Apr**\n\n**May**\n\n**Jun**\n\n($1,000B)\n\n($600B)\n\n($200B)\n\n$200B\n\n$600B\n\n$1,000B\n\n($600B)\n\n($400B)\n\n($200B)\n\n$0B\n\n$200B\n\n**Figure 3. Monthly Receipts, Outlays, and Budget Deficit/Surplus of the U.S. Government, Fiscal Years 2024 and 2025**\n\n**2024 2025**\n\n**Oct**\n\n**Nov**\n\n**Dec**\n\n**Jan**\n\n**Feb**\n\n**Mar**\n\n**Apr**\n\n**May**\n\n**Jun**\n\n**Jul**\n\n**Aug**\n\n**Sep**\n\n**Oct**\n\n**Nov**\n\n**Dec**\n\n**Jan**\n\n**Feb**\n\n**Mar**\n\n**Apr**\n\n**May**\n\n**Jun**\n\n$1,000B $2,000B $3,000B $4,000B $5,000B\n\n($7,000B) ($5,000B)\n\n($3,000B)\n\n($1,000B)\n\n($1,000B)\n\n$0B\n\n**Figure 4. Monthly Receipts, Outlays, and Budget Deficit/Surplus of the U.S. Government, Cumulative, Fiscal Years 2024 and 2025**\n\nReceipts Outlays Deficit(-)/Surplus\n\n$0B","pages":[{"line_numbers":[],"markdown":"6 _Source_ _d_ _ata: Table 1_\n\nReceipts Outlays Deficit(-)/Surplus\n\n**2024 2025**\n\n**Oct**\n\n**Nov**\n\n**Dec**\n\n**Jan**\n\n**Feb**\n\n**Mar**\n\n**Apr**\n\n**May**\n\n**Jun**\n\n**Jul**\n\n**Aug**\n\n**Sep**\n\n**Oct**\n\n**Nov**\n\n**Dec**\n\n**Jan**\n\n**Feb**\n\n**Mar**\n\n**Apr**\n\n**May**\n\n**Jun**\n\n($1,000B)\n\n($600B)\n\n($200B)\n\n$200B\n\n$600B\n\n$1,000B\n\n($600B)\n\n($400B)\n\n($200B)\n\n$0B\n\n$200B\n\n**Figure 3. Monthly Receipts, Outlays, and Budget Deficit/Surplus of the U.S. Government, Fiscal Years 2024 and 2025**\n\n**2024 2025**\n\n**Oct**\n\n**Nov**\n\n**Dec**\n\n**Jan**\n\n**Feb**\n\n**Mar**\n\n**Apr**\n\n**May**\n\n**Jun**\n\n**Jul**\n\n**Aug**\n\n**Sep**\n\n**Oct**\n\n**Nov**\n\n**Dec**\n\n**Jan**\n\n**Feb**\n\n**Mar**\n\n**Apr**\n\n**May**\n\n**Jun**\n\n$1,000B $2,000B $3,000B $4,000B $5,000B\n\n($7,000B) ($5,000B)\n\n($3,000B)\n\n($1,000B)\n\n($1,000B)\n\n$0B\n\n**Figure 4. Monthly Receipts, Outlays, and Budget Deficit/Surplus of the U.S. Government, Cumulative, Fiscal Years 2024 and 2025**\n\nReceipts Outlays Deficit(-)/Surplus\n\n$0B","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"6 _Source_ _d_ _ata: Table 1_\n\nReceipts Outlays Deficit(-)/Surplus\n\n**2024 2025**\n\n**Oct**\n\n**Nov**\n\n**Dec**\n\n**Jan**\n\n**Feb**\n\n**Mar**\n\n**Apr**\n\n**May**\n\n**Jun**\n\n**Jul**\n\n**Aug**\n\n**Sep**\n\n**Oct**\n\n**Nov**\n\n**Dec**\n\n**Jan**\n\n**Feb**\n\n**Mar**\n\n**Apr**\n\n**May**\n\n**Jun**\n\n($1,000B)\n\n($600B)\n\n($200B)\n\n$200B\n\n$600B\n\n$1,000B\n\n($600B)\n\n($400B)\n\n($200B)\n\n$0B\n\n$200B\n\n**Figure 3. Monthly Receipts, Outlays, and Budget Deficit/Surplus of the U.S. Government, Fiscal Years 2024 and 2025**\n\n**2024 2025**\n\n**Oct**\n\n**Nov**\n\n**Dec**\n\n**Jan**\n\n**Feb**\n\n**Mar**\n\n**Apr**\n\n**May**\n\n**Jun**\n\n**Jul**\n\n**Aug**\n\n**Sep**\n\n**Oct**\n\n**Nov**\n\n**Dec**\n\n**Jan**\n\n**Feb**\n\n**Mar**\n\n**Apr**\n\n**May**\n\n**Jun**\n\n$1,000B $2,000B $3,000B $4,000B $5,000B\n\n($7,000B) ($5,000B)\n\n($3,000B)\n\n($1,000B)\n\n($1,000B)\n\n$0B\n\n**Figure 4. Monthly Receipts, Outlays, and Budget Deficit/Surplus of the U.S. Government, Cumulative, Fiscal Years 2024 and 2025**\n\nReceipts Outlays Deficit(-)/Surplus\n\n$0B","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/mts0625_p6.pdf"]},"response":{"errors":[],"extract_id":"extract_c2307762ec0f3e74bab22d4177ee40a8","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["6 _Source_ _d_ _ata: Table 1_\n\nReceipts Outlays Deficit(-)/Surplus\n\n**2024 2025**\n\n**Oct**\n\n**Nov**\n\n**Dec**\n\n**Jan**\n\n**Feb**\n\n**Mar**\n\n**Apr**\n\n**May**\n\n**Jun**\n\n**Jul**\n\n**Aug**\n\n**Sep**\n\n**Oct**\n\n**Nov**\n\n**Dec**\n\n**Jan**\n\n**Feb**\n\n**Mar**\n\n**Apr**\n\n**May**\n\n**Jun**\n\n($1,000B)\n\n($600B)\n\n($200B)\n\n$200B\n\n$600B\n\n$1,000B\n\n($600B)\n\n($400B)\n\n($200B)\n\n$0B\n\n$200B\n\n**Figure 3. Monthly Receipts, Outlays, and Budget Deficit/Surplus of the U.S. Government, Fiscal Years 2024 and 2025**\n\n**2024 2025**\n\n**Oct**\n\n**Nov**\n\n**Dec**\n\n**Jan**\n\n**Feb**\n\n**Mar**\n\n**Apr**\n\n**May**\n\n**Jun**\n\n**Jul**\n\n**Aug**\n\n**Sep**\n\n**Oct**\n\n**Nov**\n\n**Dec**\n\n**Jan**\n\n**Feb**\n\n**Mar**\n\n**Apr**\n\n**May**\n\n**Jun**\n\n$1,000B $2,000B $3,000B $4,000B $5,000B\n\n($7,000B) ($5,000B)\n\n($3,000B)\n\n($1,000B)\n\n($1,000B)\n\n$0B\n\n**Figure 4. Monthly Receipts, Outlays, and Budget Deficit/Surplus of the U.S. Government, Cumulative, Fiscal Years 2024 and 2025**\n\nReceipts Outlays Deficit(-)/Surplus\n\n$0B"],"full_content":"6 _Source_ _d_ _ata: Table 1_\n\nReceipts Outlays Deficit(-)/Surplus\n\n**2024 2025**\n\n**Oct**\n\n**Nov**\n\n**Dec**\n\n**Jan**\n\n**Feb**\n\n**Mar**\n\n**Apr**\n\n**May**\n\n**Jun**\n\n**Jul**\n\n**Aug**\n\n**Sep**\n\n**Oct**\n\n**Nov**\n\n**Dec**\n\n**Jan**\n\n**Feb**\n\n**Mar**\n\n**Apr**\n\n**May**\n\n**Jun**\n\n($1,000B)\n\n($600B)\n\n($200B)\n\n$200B\n\n$600B\n\n$1,000B\n\n($600B)\n\n($400B)\n\n($200B)\n\n$0B\n\n$200B\n\n**Figure 3. Monthly Receipts, Outlays, and Budget Deficit/Surplus of the U.S. Government, Fiscal Years 2024 and 2025**\n\n**2024 2025**\n\n**Oct**\n\n**Nov**\n\n**Dec**\n\n**Jan**\n\n**Feb**\n\n**Mar**\n\n**Apr**\n\n**May**\n\n**Jun**\n\n**Jul**\n\n**Aug**\n\n**Sep**\n\n**Oct**\n\n**Nov**\n\n**Dec**\n\n**Jan**\n\n**Feb**\n\n**Mar**\n\n**Apr**\n\n**May**\n\n**Jun**\n\n$1,000B $2,000B $3,000B $4,000B $5,000B\n\n($7,000B) ($5,000B)\n\n($3,000B)\n\n($1,000B)\n\n($1,000B)\n\n$0B\n\n**Figure 4. Monthly Receipts, Outlays, and Budget Deficit/Surplus of the U.S. Government, Cumulative, Fiscal Years 2024 and 2025**\n\nReceipts Outlays Deficit(-)/Surplus\n\n$0B","publish_date":null,"title":"Figure 3. Monthly Receipts, Outlays, and Budget Deficit/Surplus of the U.S. Government, Fiscal Years 2024 and 2025 2024 2025 $1,000B","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/mts0625_p6.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/mts0625_p6.pdf"},"request":{"example_id":"chart/mts0625_p6","product_type":"parse","source_file_path":"mts0625_p6.pdf"},"started_at":"2026-11-04T19:45:05.033035"} +{"completed_at":"2026-11-04T19:45:11.422054","latency_in_ms":4734,"output":{"example_id":"chart/natural-catastrophe-and-climate-report-2023_p16","grounded_pages":[],"layout_pages":[],"markdown":"Global Natural Catastrophe Overview\n\n16\n\n2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023\n\n25\n\n50\n\n75\n\n100\n\n125\n\n150\n\n175\n\n200\n\nUSD billion\n\nWeather / Climate Other Natural Perils\n\nBy Peril % Covered by Private Insurers Billion-Dollar Events\n\nSCS\n\nFL\n\nTC\n\nDR\n\nEQ\n\nWF\n\nEU WS\n\nWW\n\nOther\n\n$13B\n\n$71B\n\n$0B\n\n$11B\n\n$4B\n\n$9B\n\n$5B\n\n$3B\n\n$7B\n\nSCS\n\nFL\n\nEQ\n\nTC\n\nDR\n\nWF\n\nEU WS\n\nWW\n\nOther\n\n58%\n\n11%\n\n0%\n\n4%\n\n6%\n\n9%\n\n3%\n\n3%\n\n7%\n\nPrivate Public\n\n## 90%\n\n2023\n\n## 90%\n\n2000-2023\n\nUnited States 24\n\nWeather / Climate Other Natural Perils\n\nEurope\n\nOceania\n\nAsia\n\nLatin America\n\nAfrica\n\nMiddle East\n\nRest of N. America\n\n6\n\n2\n\n1\n\n1\n\n0\n\n0\n\n0\n\n34 All Natural Perils 33 Weather / Climate\n\nA 20-year view of global insured losses resulting from natural catastrophes is seen below in Figure 7. There has been an increasingly obvious upward\n\ntrend in insured losses during this time (+4.3% annualized rate of growth). It is unsurprising that the annual rate of insured losses has grown at a faster\n\npace than the overall economic total. As insurance penetration continues to expand into parts of the world with traditionally limited policy take-up, in\n\naddition to the growth of public-private partnerships to introduce parametric or cat bond solutions to guarantee financial recovery in the aftermath of\n\nan event, it is expected that more damage costs will be covered. We also see a greater portion of losses covered by government-sponsored insurance\n\nentities. Examples include the National Flood Insurance Program, or the USDA’s Risk Management Agency’s crop insurance program in the US, or\n\nthe Turkish Catastrophe Insurance Pool (TCIP). While the progress of insurance protection is promising, there remain considerable gaps in coverage\n\nthat leave many parts of the developing world highly vulnerable to catastrophe risk. The need for more guaranteed climate or natural catastrophe\n\nfinancing to mitigate or adapt to a more complex world of natural hazards becomes more critical by the day.\n\nFigure 7: Last 20 years of annual global insured losses; historical losses adjusted using US CPI and a construction + cost of labor factor | Data and Graphic: Gallagher Re\n\nFigure 8: 2023 global insured loss statistics | Data and Graphic: Gallagher Re\n\nEQ: Earthquake\n\nDR: Drought\n\nSCS: Severe Convective Storm\n\nEU WS: European Windstorm\n\nTC: Tropical Cyclone\n\nWF: Wildfire\n\nWW: Winter Weather\n\nFL: Flooding","pages":[{"line_numbers":[],"markdown":"Global Natural Catastrophe Overview\n\n16\n\n2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023\n\n25\n\n50\n\n75\n\n100\n\n125\n\n150\n\n175\n\n200\n\nUSD billion\n\nWeather / Climate Other Natural Perils\n\nBy Peril % Covered by Private Insurers Billion-Dollar Events\n\nSCS\n\nFL\n\nTC\n\nDR\n\nEQ\n\nWF\n\nEU WS\n\nWW\n\nOther\n\n$13B\n\n$71B\n\n$0B\n\n$11B\n\n$4B\n\n$9B\n\n$5B\n\n$3B\n\n$7B\n\nSCS\n\nFL\n\nEQ\n\nTC\n\nDR\n\nWF\n\nEU WS\n\nWW\n\nOther\n\n58%\n\n11%\n\n0%\n\n4%\n\n6%\n\n9%\n\n3%\n\n3%\n\n7%\n\nPrivate Public\n\n## 90%\n\n2023\n\n## 90%\n\n2000-2023\n\nUnited States 24\n\nWeather / Climate Other Natural Perils\n\nEurope\n\nOceania\n\nAsia\n\nLatin America\n\nAfrica\n\nMiddle East\n\nRest of N. America\n\n6\n\n2\n\n1\n\n1\n\n0\n\n0\n\n0\n\n34 All Natural Perils 33 Weather / Climate\n\nA 20-year view of global insured losses resulting from natural catastrophes is seen below in Figure 7. There has been an increasingly obvious upward\n\ntrend in insured losses during this time (+4.3% annualized rate of growth). It is unsurprising that the annual rate of insured losses has grown at a faster\n\npace than the overall economic total. As insurance penetration continues to expand into parts of the world with traditionally limited policy take-up, in\n\naddition to the growth of public-private partnerships to introduce parametric or cat bond solutions to guarantee financial recovery in the aftermath of\n\nan event, it is expected that more damage costs will be covered. We also see a greater portion of losses covered by government-sponsored insurance\n\nentities. Examples include the National Flood Insurance Program, or the USDA’s Risk Management Agency’s crop insurance program in the US, or\n\nthe Turkish Catastrophe Insurance Pool (TCIP). While the progress of insurance protection is promising, there remain considerable gaps in coverage\n\nthat leave many parts of the developing world highly vulnerable to catastrophe risk. The need for more guaranteed climate or natural catastrophe\n\nfinancing to mitigate or adapt to a more complex world of natural hazards becomes more critical by the day.\n\nFigure 7: Last 20 years of annual global insured losses; historical losses adjusted using US CPI and a construction + cost of labor factor | Data and Graphic: Gallagher Re\n\nFigure 8: 2023 global insured loss statistics | Data and Graphic: Gallagher Re\n\nEQ: Earthquake\n\nDR: Drought\n\nSCS: Severe Convective Storm\n\nEU WS: European Windstorm\n\nTC: Tropical Cyclone\n\nWF: Wildfire\n\nWW: Winter Weather\n\nFL: Flooding","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Global Natural Catastrophe Overview\n\n16\n\n2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023\n\n25\n\n50\n\n75\n\n100\n\n125\n\n150\n\n175\n\n200\n\nUSD billion\n\nWeather / Climate Other Natural Perils\n\nBy Peril % Covered by Private Insurers Billion-Dollar Events\n\nSCS\n\nFL\n\nTC\n\nDR\n\nEQ\n\nWF\n\nEU WS\n\nWW\n\nOther\n\n$13B\n\n$71B\n\n$0B\n\n$11B\n\n$4B\n\n$9B\n\n$5B\n\n$3B\n\n$7B\n\nSCS\n\nFL\n\nEQ\n\nTC\n\nDR\n\nWF\n\nEU WS\n\nWW\n\nOther\n\n58%\n\n11%\n\n0%\n\n4%\n\n6%\n\n9%\n\n3%\n\n3%\n\n7%\n\nPrivate Public\n\n## 90%\n\n2023\n\n## 90%\n\n2000-2023\n\nUnited States 24\n\nWeather / Climate Other Natural Perils\n\nEurope\n\nOceania\n\nAsia\n\nLatin America\n\nAfrica\n\nMiddle East\n\nRest of N. America\n\n6\n\n2\n\n1\n\n1\n\n0\n\n0\n\n0\n\n34 All Natural Perils 33 Weather / Climate\n\nA 20-year view of global insured losses resulting from natural catastrophes is seen below in Figure 7. There has been an increasingly obvious upward\n\ntrend in insured losses during this time (+4.3% annualized rate of growth). It is unsurprising that the annual rate of insured losses has grown at a faster\n\npace than the overall economic total. As insurance penetration continues to expand into parts of the world with traditionally limited policy take-up, in\n\naddition to the growth of public-private partnerships to introduce parametric or cat bond solutions to guarantee financial recovery in the aftermath of\n\nan event, it is expected that more damage costs will be covered. We also see a greater portion of losses covered by government-sponsored insurance\n\nentities. Examples include the National Flood Insurance Program, or the USDA’s Risk Management Agency’s crop insurance program in the US, or\n\nthe Turkish Catastrophe Insurance Pool (TCIP). While the progress of insurance protection is promising, there remain considerable gaps in coverage\n\nthat leave many parts of the developing world highly vulnerable to catastrophe risk. The need for more guaranteed climate or natural catastrophe\n\nfinancing to mitigate or adapt to a more complex world of natural hazards becomes more critical by the day.\n\nFigure 7: Last 20 years of annual global insured losses; historical losses adjusted using US CPI and a construction + cost of labor factor | Data and Graphic: Gallagher Re\n\nFigure 8: 2023 global insured loss statistics | Data and Graphic: Gallagher Re\n\nEQ: Earthquake\n\nDR: Drought\n\nSCS: Severe Convective Storm\n\nEU WS: European Windstorm\n\nTC: Tropical Cyclone\n\nWF: Wildfire\n\nWW: Winter Weather\n\nFL: Flooding","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/natural-catastrophe-and-climate-report-2023_p16.pdf"]},"response":{"errors":[],"extract_id":"extract_eacfb6f6ae660271fb422964b8d41cc8","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Global Natural Catastrophe Overview\n\n16\n\n2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023\n\n25\n\n50\n\n75\n\n100\n\n125\n\n150\n\n175\n\n200\n\nUSD billion\n\nWeather / Climate Other Natural Perils\n\nBy Peril % Covered by Private Insurers Billion-Dollar Events\n\nSCS\n\nFL\n\nTC\n\nDR\n\nEQ\n\nWF\n\nEU WS\n\nWW\n\nOther\n\n$13B\n\n$71B\n\n$0B\n\n$11B\n\n$4B\n\n$9B\n\n$5B\n\n$3B\n\n$7B\n\nSCS\n\nFL\n\nEQ\n\nTC\n\nDR\n\nWF\n\nEU WS\n\nWW\n\nOther\n\n58%\n\n11%\n\n0%\n\n4%\n\n6%\n\n9%\n\n3%\n\n3%\n\n7%\n\nPrivate Public\n\n## 90%\n\n2023\n\n## 90%\n\n2000-2023\n\nUnited States 24\n\nWeather / Climate Other Natural Perils\n\nEurope\n\nOceania\n\nAsia\n\nLatin America\n\nAfrica\n\nMiddle East\n\nRest of N. America\n\n6\n\n2\n\n1\n\n1\n\n0\n\n0\n\n0\n\n34 All Natural Perils 33 Weather / Climate\n\nA 20-year view of global insured losses resulting from natural catastrophes is seen below in Figure 7. There has been an increasingly obvious upward\n\ntrend in insured losses during this time (+4.3% annualized rate of growth). It is unsurprising that the annual rate of insured losses has grown at a faster\n\npace than the overall economic total. As insurance penetration continues to expand into parts of the world with traditionally limited policy take-up, in\n\naddition to the growth of public-private partnerships to introduce parametric or cat bond solutions to guarantee financial recovery in the aftermath of\n\nan event, it is expected that more damage costs will be covered. We also see a greater portion of losses covered by government-sponsored insurance\n\nentities. Examples include the National Flood Insurance Program, or the USDA’s Risk Management Agency’s crop insurance program in the US, or\n\nthe Turkish Catastrophe Insurance Pool (TCIP). While the progress of insurance protection is promising, there remain considerable gaps in coverage\n\nthat leave many parts of the developing world highly vulnerable to catastrophe risk. The need for more guaranteed climate or natural catastrophe\n\nfinancing to mitigate or adapt to a more complex world of natural hazards becomes more critical by the day.\n\nFigure 7: Last 20 years of annual global insured losses; historical losses adjusted using US CPI and a construction + cost of labor factor | Data and Graphic: Gallagher Re\n\nFigure 8: 2023 global insured loss statistics | Data and Graphic: Gallagher Re\n\nEQ: Earthquake\n\nDR: Drought\n\nSCS: Severe Convective Storm\n\nEU WS: European Windstorm\n\nTC: Tropical Cyclone\n\nWF: Wildfire\n\nWW: Winter Weather\n\nFL: Flooding"],"full_content":"Global Natural Catastrophe Overview\n\n16\n\n2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023\n\n25\n\n50\n\n75\n\n100\n\n125\n\n150\n\n175\n\n200\n\nUSD billion\n\nWeather / Climate Other Natural Perils\n\nBy Peril % Covered by Private Insurers Billion-Dollar Events\n\nSCS\n\nFL\n\nTC\n\nDR\n\nEQ\n\nWF\n\nEU WS\n\nWW\n\nOther\n\n$13B\n\n$71B\n\n$0B\n\n$11B\n\n$4B\n\n$9B\n\n$5B\n\n$3B\n\n$7B\n\nSCS\n\nFL\n\nEQ\n\nTC\n\nDR\n\nWF\n\nEU WS\n\nWW\n\nOther\n\n58%\n\n11%\n\n0%\n\n4%\n\n6%\n\n9%\n\n3%\n\n3%\n\n7%\n\nPrivate Public\n\n## 90%\n\n2023\n\n## 90%\n\n2000-2023\n\nUnited States 24\n\nWeather / Climate Other Natural Perils\n\nEurope\n\nOceania\n\nAsia\n\nLatin America\n\nAfrica\n\nMiddle East\n\nRest of N. America\n\n6\n\n2\n\n1\n\n1\n\n0\n\n0\n\n0\n\n34 All Natural Perils 33 Weather / Climate\n\nA 20-year view of global insured losses resulting from natural catastrophes is seen below in Figure 7. There has been an increasingly obvious upward\n\ntrend in insured losses during this time (+4.3% annualized rate of growth). It is unsurprising that the annual rate of insured losses has grown at a faster\n\npace than the overall economic total. As insurance penetration continues to expand into parts of the world with traditionally limited policy take-up, in\n\naddition to the growth of public-private partnerships to introduce parametric or cat bond solutions to guarantee financial recovery in the aftermath of\n\nan event, it is expected that more damage costs will be covered. We also see a greater portion of losses covered by government-sponsored insurance\n\nentities. Examples include the National Flood Insurance Program, or the USDA’s Risk Management Agency’s crop insurance program in the US, or\n\nthe Turkish Catastrophe Insurance Pool (TCIP). While the progress of insurance protection is promising, there remain considerable gaps in coverage\n\nthat leave many parts of the developing world highly vulnerable to catastrophe risk. The need for more guaranteed climate or natural catastrophe\n\nfinancing to mitigate or adapt to a more complex world of natural hazards becomes more critical by the day.\n\nFigure 7: Last 20 years of annual global insured losses; historical losses adjusted using US CPI and a construction + cost of labor factor | Data and Graphic: Gallagher Re\n\nFigure 8: 2023 global insured loss statistics | Data and Graphic: Gallagher Re\n\nEQ: Earthquake\n\nDR: Drought\n\nSCS: Severe Convective Storm\n\nEU WS: European Windstorm\n\nTC: Tropical Cyclone\n\nWF: Wildfire\n\nWW: Winter Weather\n\nFL: Flooding","publish_date":null,"title":"90% 90%","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/natural-catastrophe-and-climate-report-2023_p16.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/natural-catastrophe-and-climate-report-2023_p16.pdf"},"request":{"example_id":"chart/natural-catastrophe-and-climate-report-2023_p16","product_type":"parse","source_file_path":"natural-catastrophe-and-climate-report-2023_p16.pdf"},"started_at":"2026-11-04T19:45:06.687463"} +{"completed_at":"2026-11-04T19:45:09.299596","latency_in_ms":1215,"output":{"example_id":"chart/natural-catastrophe-and-climate-report-2023_p20","grounded_pages":[],"layout_pages":[],"markdown":"In Focus\n\n20\n\nThe result of this destabilization of the US insurance market has led\n\nstate regulators and governance agencies to request more data and\n\ntransparency from the insurance market on how they are accounting\n\nfor climate risk and where their portfolios are invested. This has come\n\nin the form of more disclosure requests from places such as the\n\nUS Federal Insurance Office, the National Association of Insurance\n\nCommissioners (NAIC), and individual state insurance departments.\n\nIn November 2023, the US Senate Budget Committee sent letters\n\nto 41 insurance companies requesting information on their plans to\n\naddress increased underwriting losses from climate disasters and on\n\ninvestment decisions involving fossil fuel projects. The US Securities\n\nand Exchange Commission (SEC) is also expected to release a long-\n\ndelayed series of climate-related risk disclosure rules in 2024, which\n\nwill include an accounting of companies’ greenhouse gas emissions\n\n(Scope 1, 2, and 3).\n\nThe US has trailed other parts of the world, notably Europe, in\n\nrequiring more climate disclosures. Nevertheless, as insurability\n\nbecomes a more complex issue for the private market, state, and\n\nfederal officials are trying to understand the underlying issues —\n\nleading to more questions. Most focus tends to be on the insurance\n\nmarkets in Florida, California, and Louisiana, but the reality is that\n\nthe insurability and affordability issue is a 50-state problem. Case in\n\npoint — A 2023 study noted that Oklahoma had the highest average\n\nhomeowner insurance premium at USD5,317 per year — or nearly\n\ndouble the national average of USD2,777. This was based on a policy\n\nwith USD300,000 of dwelling coverage, a USD300,000 liability limit,\n\nand a USD1,000 deductible.\n\nAs homeowners seek the most affordable insurance rates, there are\n\nsome strategies to consider. There are “risk-based insurance pricing”\n\noptions, which can be used as a form of incentivization. Insurers may\n\noffer lower premiums to customers if they improve the structural\n\nintegrity of their homes through retrofitting, for example, or use\n\nmodern construction practices that meet or exceed state or local\n\nbuilding code requirements. There are known technological and\n\nengineering improvements to limit risk to a home via higher-quality\n\nsiding or roofing. The Insurance Institute for Home and Business\n\nSafety (IBHS), has available impact-resistant shingle performance\n\nratings in the fight against hail damage. While not an explicit insurance\n\nproduct, there are incentivization programs now being offered by\n\nthe US federal government that promote clean energy transition and\n\nimproved construction practices. As an example, the 2022 Inflation\n\nReduction Act extended the clean energy Investment Tax Credit that\n\nprovides up to a 30% credit for various wind, solar, energy storage,\n\nand other renewable energy projects.\n\nFigure 11: Total global reinsurance dedicated capital | Data and Graphic: Gallagher Re\n\n2015 FY 2016 FY 2017 FY 2018 FY 2019 FY 2020 FY 2021 FY 2022 FY 2023 HY\n\n200\n\n400\n\n600\n\n800\n\nUSD billion\n\n449\n\n709\n\n486\n\n629\n\n526\n\n427\n\n672\n\n725\n\n615\n\nAlternative Capital (Excludes Life, Accident, and Health) Major Regional & Local Reinsurers + Pro-Rated Portion of Capital w/in Major Groups Index","pages":[{"line_numbers":[],"markdown":"In Focus\n\n20\n\nThe result of this destabilization of the US insurance market has led\n\nstate regulators and governance agencies to request more data and\n\ntransparency from the insurance market on how they are accounting\n\nfor climate risk and where their portfolios are invested. This has come\n\nin the form of more disclosure requests from places such as the\n\nUS Federal Insurance Office, the National Association of Insurance\n\nCommissioners (NAIC), and individual state insurance departments.\n\nIn November 2023, the US Senate Budget Committee sent letters\n\nto 41 insurance companies requesting information on their plans to\n\naddress increased underwriting losses from climate disasters and on\n\ninvestment decisions involving fossil fuel projects. The US Securities\n\nand Exchange Commission (SEC) is also expected to release a long-\n\ndelayed series of climate-related risk disclosure rules in 2024, which\n\nwill include an accounting of companies’ greenhouse gas emissions\n\n(Scope 1, 2, and 3).\n\nThe US has trailed other parts of the world, notably Europe, in\n\nrequiring more climate disclosures. Nevertheless, as insurability\n\nbecomes a more complex issue for the private market, state, and\n\nfederal officials are trying to understand the underlying issues —\n\nleading to more questions. Most focus tends to be on the insurance\n\nmarkets in Florida, California, and Louisiana, but the reality is that\n\nthe insurability and affordability issue is a 50-state problem. Case in\n\npoint — A 2023 study noted that Oklahoma had the highest average\n\nhomeowner insurance premium at USD5,317 per year — or nearly\n\ndouble the national average of USD2,777. This was based on a policy\n\nwith USD300,000 of dwelling coverage, a USD300,000 liability limit,\n\nand a USD1,000 deductible.\n\nAs homeowners seek the most affordable insurance rates, there are\n\nsome strategies to consider. There are “risk-based insurance pricing”\n\noptions, which can be used as a form of incentivization. Insurers may\n\noffer lower premiums to customers if they improve the structural\n\nintegrity of their homes through retrofitting, for example, or use\n\nmodern construction practices that meet or exceed state or local\n\nbuilding code requirements. There are known technological and\n\nengineering improvements to limit risk to a home via higher-quality\n\nsiding or roofing. The Insurance Institute for Home and Business\n\nSafety (IBHS), has available impact-resistant shingle performance\n\nratings in the fight against hail damage. While not an explicit insurance\n\nproduct, there are incentivization programs now being offered by\n\nthe US federal government that promote clean energy transition and\n\nimproved construction practices. As an example, the 2022 Inflation\n\nReduction Act extended the clean energy Investment Tax Credit that\n\nprovides up to a 30% credit for various wind, solar, energy storage,\n\nand other renewable energy projects.\n\nFigure 11: Total global reinsurance dedicated capital | Data and Graphic: Gallagher Re\n\n2015 FY 2016 FY 2017 FY 2018 FY 2019 FY 2020 FY 2021 FY 2022 FY 2023 HY\n\n200\n\n400\n\n600\n\n800\n\nUSD billion\n\n449\n\n709\n\n486\n\n629\n\n526\n\n427\n\n672\n\n725\n\n615\n\nAlternative Capital (Excludes Life, Accident, and Health) Major Regional & Local Reinsurers + Pro-Rated Portion of Capital w/in Major Groups Index","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"In Focus\n\n20\n\nThe result of this destabilization of the US insurance market has led\n\nstate regulators and governance agencies to request more data and\n\ntransparency from the insurance market on how they are accounting\n\nfor climate risk and where their portfolios are invested. This has come\n\nin the form of more disclosure requests from places such as the\n\nUS Federal Insurance Office, the National Association of Insurance\n\nCommissioners (NAIC), and individual state insurance departments.\n\nIn November 2023, the US Senate Budget Committee sent letters\n\nto 41 insurance companies requesting information on their plans to\n\naddress increased underwriting losses from climate disasters and on\n\ninvestment decisions involving fossil fuel projects. The US Securities\n\nand Exchange Commission (SEC) is also expected to release a long-\n\ndelayed series of climate-related risk disclosure rules in 2024, which\n\nwill include an accounting of companies’ greenhouse gas emissions\n\n(Scope 1, 2, and 3).\n\nThe US has trailed other parts of the world, notably Europe, in\n\nrequiring more climate disclosures. Nevertheless, as insurability\n\nbecomes a more complex issue for the private market, state, and\n\nfederal officials are trying to understand the underlying issues —\n\nleading to more questions. Most focus tends to be on the insurance\n\nmarkets in Florida, California, and Louisiana, but the reality is that\n\nthe insurability and affordability issue is a 50-state problem. Case in\n\npoint — A 2023 study noted that Oklahoma had the highest average\n\nhomeowner insurance premium at USD5,317 per year — or nearly\n\ndouble the national average of USD2,777. This was based on a policy\n\nwith USD300,000 of dwelling coverage, a USD300,000 liability limit,\n\nand a USD1,000 deductible.\n\nAs homeowners seek the most affordable insurance rates, there are\n\nsome strategies to consider. There are “risk-based insurance pricing”\n\noptions, which can be used as a form of incentivization. Insurers may\n\noffer lower premiums to customers if they improve the structural\n\nintegrity of their homes through retrofitting, for example, or use\n\nmodern construction practices that meet or exceed state or local\n\nbuilding code requirements. There are known technological and\n\nengineering improvements to limit risk to a home via higher-quality\n\nsiding or roofing. The Insurance Institute for Home and Business\n\nSafety (IBHS), has available impact-resistant shingle performance\n\nratings in the fight against hail damage. While not an explicit insurance\n\nproduct, there are incentivization programs now being offered by\n\nthe US federal government that promote clean energy transition and\n\nimproved construction practices. As an example, the 2022 Inflation\n\nReduction Act extended the clean energy Investment Tax Credit that\n\nprovides up to a 30% credit for various wind, solar, energy storage,\n\nand other renewable energy projects.\n\nFigure 11: Total global reinsurance dedicated capital | Data and Graphic: Gallagher Re\n\n2015 FY 2016 FY 2017 FY 2018 FY 2019 FY 2020 FY 2021 FY 2022 FY 2023 HY\n\n200\n\n400\n\n600\n\n800\n\nUSD billion\n\n449\n\n709\n\n486\n\n629\n\n526\n\n427\n\n672\n\n725\n\n615\n\nAlternative Capital (Excludes Life, Accident, and Health) Major Regional & Local Reinsurers + Pro-Rated Portion of Capital w/in Major Groups Index","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/natural-catastrophe-and-climate-report-2023_p20.pdf"]},"response":{"errors":[],"extract_id":"extract_bbca2e60db13e118ed9f12babdc69cc3","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["In Focus\n\n20\n\nThe result of this destabilization of the US insurance market has led\n\nstate regulators and governance agencies to request more data and\n\ntransparency from the insurance market on how they are accounting\n\nfor climate risk and where their portfolios are invested. This has come\n\nin the form of more disclosure requests from places such as the\n\nUS Federal Insurance Office, the National Association of Insurance\n\nCommissioners (NAIC), and individual state insurance departments.\n\nIn November 2023, the US Senate Budget Committee sent letters\n\nto 41 insurance companies requesting information on their plans to\n\naddress increased underwriting losses from climate disasters and on\n\ninvestment decisions involving fossil fuel projects. The US Securities\n\nand Exchange Commission (SEC) is also expected to release a long-\n\ndelayed series of climate-related risk disclosure rules in 2024, which\n\nwill include an accounting of companies’ greenhouse gas emissions\n\n(Scope 1, 2, and 3).\n\nThe US has trailed other parts of the world, notably Europe, in\n\nrequiring more climate disclosures. Nevertheless, as insurability\n\nbecomes a more complex issue for the private market, state, and\n\nfederal officials are trying to understand the underlying issues —\n\nleading to more questions. Most focus tends to be on the insurance\n\nmarkets in Florida, California, and Louisiana, but the reality is that\n\nthe insurability and affordability issue is a 50-state problem. Case in\n\npoint — A 2023 study noted that Oklahoma had the highest average\n\nhomeowner insurance premium at USD5,317 per year — or nearly\n\ndouble the national average of USD2,777. This was based on a policy\n\nwith USD300,000 of dwelling coverage, a USD300,000 liability limit,\n\nand a USD1,000 deductible.\n\nAs homeowners seek the most affordable insurance rates, there are\n\nsome strategies to consider. There are “risk-based insurance pricing”\n\noptions, which can be used as a form of incentivization. Insurers may\n\noffer lower premiums to customers if they improve the structural\n\nintegrity of their homes through retrofitting, for example, or use\n\nmodern construction practices that meet or exceed state or local\n\nbuilding code requirements. There are known technological and\n\nengineering improvements to limit risk to a home via higher-quality\n\nsiding or roofing. The Insurance Institute for Home and Business\n\nSafety (IBHS), has available impact-resistant shingle performance\n\nratings in the fight against hail damage. While not an explicit insurance\n\nproduct, there are incentivization programs now being offered by\n\nthe US federal government that promote clean energy transition and\n\nimproved construction practices. As an example, the 2022 Inflation\n\nReduction Act extended the clean energy Investment Tax Credit that\n\nprovides up to a 30% credit for various wind, solar, energy storage,\n\nand other renewable energy projects.\n\nFigure 11: Total global reinsurance dedicated capital | Data and Graphic: Gallagher Re\n\n2015 FY 2016 FY 2017 FY 2018 FY 2019 FY 2020 FY 2021 FY 2022 FY 2023 HY\n\n200\n\n400\n\n600\n\n800\n\nUSD billion\n\n449\n\n709\n\n486\n\n629\n\n526\n\n427\n\n672\n\n725\n\n615\n\nAlternative Capital (Excludes Life, Accident, and Health) Major Regional & Local Reinsurers + Pro-Rated Portion of Capital w/in Major Groups Index"],"full_content":"In Focus\n\n20\n\nThe result of this destabilization of the US insurance market has led\n\nstate regulators and governance agencies to request more data and\n\ntransparency from the insurance market on how they are accounting\n\nfor climate risk and where their portfolios are invested. This has come\n\nin the form of more disclosure requests from places such as the\n\nUS Federal Insurance Office, the National Association of Insurance\n\nCommissioners (NAIC), and individual state insurance departments.\n\nIn November 2023, the US Senate Budget Committee sent letters\n\nto 41 insurance companies requesting information on their plans to\n\naddress increased underwriting losses from climate disasters and on\n\ninvestment decisions involving fossil fuel projects. The US Securities\n\nand Exchange Commission (SEC) is also expected to release a long-\n\ndelayed series of climate-related risk disclosure rules in 2024, which\n\nwill include an accounting of companies’ greenhouse gas emissions\n\n(Scope 1, 2, and 3).\n\nThe US has trailed other parts of the world, notably Europe, in\n\nrequiring more climate disclosures. Nevertheless, as insurability\n\nbecomes a more complex issue for the private market, state, and\n\nfederal officials are trying to understand the underlying issues —\n\nleading to more questions. Most focus tends to be on the insurance\n\nmarkets in Florida, California, and Louisiana, but the reality is that\n\nthe insurability and affordability issue is a 50-state problem. Case in\n\npoint — A 2023 study noted that Oklahoma had the highest average\n\nhomeowner insurance premium at USD5,317 per year — or nearly\n\ndouble the national average of USD2,777. This was based on a policy\n\nwith USD300,000 of dwelling coverage, a USD300,000 liability limit,\n\nand a USD1,000 deductible.\n\nAs homeowners seek the most affordable insurance rates, there are\n\nsome strategies to consider. There are “risk-based insurance pricing”\n\noptions, which can be used as a form of incentivization. Insurers may\n\noffer lower premiums to customers if they improve the structural\n\nintegrity of their homes through retrofitting, for example, or use\n\nmodern construction practices that meet or exceed state or local\n\nbuilding code requirements. There are known technological and\n\nengineering improvements to limit risk to a home via higher-quality\n\nsiding or roofing. The Insurance Institute for Home and Business\n\nSafety (IBHS), has available impact-resistant shingle performance\n\nratings in the fight against hail damage. While not an explicit insurance\n\nproduct, there are incentivization programs now being offered by\n\nthe US federal government that promote clean energy transition and\n\nimproved construction practices. As an example, the 2022 Inflation\n\nReduction Act extended the clean energy Investment Tax Credit that\n\nprovides up to a 30% credit for various wind, solar, energy storage,\n\nand other renewable energy projects.\n\nFigure 11: Total global reinsurance dedicated capital | Data and Graphic: Gallagher Re\n\n2015 FY 2016 FY 2017 FY 2018 FY 2019 FY 2020 FY 2021 FY 2022 FY 2023 HY\n\n200\n\n400\n\n600\n\n800\n\nUSD billion\n\n449\n\n709\n\n486\n\n629\n\n526\n\n427\n\n672\n\n725\n\n615\n\nAlternative Capital (Excludes Life, Accident, and Health) Major Regional & Local Reinsurers + Pro-Rated Portion of Capital w/in Major Groups Index","publish_date":null,"title":"The result of this destabilization of the US insurance market has led As homeowners seek the most affordable insurance rates,","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/natural-catastrophe-and-climate-report-2023_p20.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/natural-catastrophe-and-climate-report-2023_p20.pdf"},"request":{"example_id":"chart/natural-catastrophe-and-climate-report-2023_p20","product_type":"parse","source_file_path":"natural-catastrophe-and-climate-report-2023_p20.pdf"},"started_at":"2026-11-04T19:45:08.083853"} +{"completed_at":"2026-11-04T19:45:12.055055","latency_in_ms":2752,"output":{"example_id":"chart/natural-catastrophe-and-climate-report-2023_p41","grounded_pages":[],"layout_pages":[],"markdown":"Regional Recaps\n\n41\n\nIn the US, elevated industry losses were driven by high frequency\n\nand record severe convective storm (SCS) activity throughout the\n\nyear. While large hail continued to produce the biggest share of SCS\n\nloss, notable tornadic activity, severe wind gusts, and heavy rainfall\n\naided in pushing totals higher. SCS insured losses in 2023 reached\n\nUSD60 billion. 2023 became only the third year on record to reach or\n\nexceed the USD40 billion threshold, alongside 2020 (USD44 billion)\n\nand 2011 (USD40 billion). In addition to SCS, notable events included\n\nthe Lahaina Wildfire in Hawaii, billion-dollar-plus economic flood\n\nlosses following record-setting 24-hour rainfall in Fort Lauderdale and\n\nNew York City, Hurricane Idalia’s Category 3 landfall in Florida, and\n\nexpansive drought conditions which predominantly impacted the\n\nSouthwest, Southern Plains, and Southeast.\n\n• 2023 became the costliest insured year on record for US SCS with losses reaching USD60 billion\n\n• 31 events with more than USD1 billion in economic losses, and 24 events with more than USD1 billion of insured losses\n\n• Hawaii’s Lahaina wildfire on Maui became the state’s second costliest event on record, only behind Hurricane Iniki (1992)\n\n• Exceptional rainfall in highly urbanized regions of Florida and New York/New Jersey led to billion-dollar economic losses\n\n• Hurricane Idalia was the only hurricane to make landfall in the US, generating an industry loss of at least USD1.25 billion\n\nUS SCS events dominated global insured loss totals\n\nHigh frequency and impactful SCS events in the US dominated global\n\ninsured losses in 2023, led by the hail sub-peril. Hail typically accounts\n\nfor 50% to 80% of SCS claims in any given year. While there has been\n\nno discernible long-term trend in the frequency of SCS outbreaks,\n\nelevated losses in recent years have been aided by expanding urban\n\nfootprints in hail prone regions and increased costs due to aging\n\nhousing stocks.\n\nLikewise, the average annual number of tornadoes has remained\n\ncomparatively constant, however, the number of tornadoes on a\n\nper event basis has shown an increasing trend alongside a shift in\n\ngeographical distribution. This has led to elevated tornado counts\n\nacross highly vulnerable regions of the mid-South and Southeast,\n\nwhere a disproportionate number of manufactured homes are\n\noccupied. Concurrently, there has been an upward trend in the\n\nintensity of extreme precipitation associated with severe storm\n\ncomplexes, and the times of the year in which favorable atmospheric\n\nconditions for SCS outbreaks are present have extended amid a\n\nwarming climate.\n\nFigure 23: Top 10 costliest years for insured losses on the US mainland (top) and\n\naggregated losses since 1990: severe convective storms and tropical cyclone\n\nData and Graphic: Gallagher Re\n\n1 2023\n\n2 2020\n\n3 2011\n\n4 2022\n\n5 2021\n\n6 2019\n\n7 2017\n\n8 2012\n\n9 2016\n\n10 2014\n\n$59.7B\n\n$44.0B\n\n$40.1B\n\n$31.5B\n\n$29.1B\n\n$28.2B\n\n$26.6B\n\n$22.4B\n\n$22.0B\n\n$19.4B\n\n1 2005\n\n2 2017\n\n3 2022\n\n4 2021\n\n5 2012\n\n6 2004\n\n7 1992\n\n8 2008\n\n9 2020\n\n10 2018\n\n$134.3B\n\n$68.3B\n\n$59.5B\n\n$42.8B\n\n$42.2B\n\n$39.6B\n\n$38.4B\n\n$31.6B\n\n$29.3B\n\n$23.1B\n\nUS Mainland\n\nTop 10 Costliest Years: Insured Loss\n\nSevere Convective Storm Tropical Cyclone\n\n1990 1995 2000 2005 2010 2015 2020 2025\n\n$100B\n\n$200B\n\n$300B\n\n$400B\n\n$500B\n\n$600B\n\nUSD billion\n\n566 583\n\n1 2023\n\n2 2020\n\n3 2011\n\n4 2022\n\n5 2021\n\n6 2019\n\n7 2017\n\n8 2012\n\n9 2016\n\n10 2014\n\n$59.7B\n\n$44.0B\n\n$40.1B\n\n$31.5B\n\n$29.1B\n\n$28.2B\n\n$26.6B\n\n$22.4B\n\n$22.0B\n\n$19.4B\n\n1 2005\n\n2 2017\n\n3 2022\n\n4 2021\n\n5 2012\n\n6 2004\n\n7 1992\n\n8 2008\n\n9 2020\n\n10 2018\n\n$134.3B\n\n$68.3B\n\n$59.5B\n\n$42.8B\n\n$42.2B\n\n$39.6B\n\n$38.4B\n\n$31.6B\n\n$29.3B\n\n$23.1B\n\nUS Mainland Top 10 Costliest Years: Insured Loss\n\nSevere Convective Storm Tropical Cyclone\n\nSCS TC\n\nCumulative US Insured Loss: SCS vs TC","pages":[{"line_numbers":[],"markdown":"Regional Recaps\n\n41\n\nIn the US, elevated industry losses were driven by high frequency\n\nand record severe convective storm (SCS) activity throughout the\n\nyear. While large hail continued to produce the biggest share of SCS\n\nloss, notable tornadic activity, severe wind gusts, and heavy rainfall\n\naided in pushing totals higher. SCS insured losses in 2023 reached\n\nUSD60 billion. 2023 became only the third year on record to reach or\n\nexceed the USD40 billion threshold, alongside 2020 (USD44 billion)\n\nand 2011 (USD40 billion). In addition to SCS, notable events included\n\nthe Lahaina Wildfire in Hawaii, billion-dollar-plus economic flood\n\nlosses following record-setting 24-hour rainfall in Fort Lauderdale and\n\nNew York City, Hurricane Idalia’s Category 3 landfall in Florida, and\n\nexpansive drought conditions which predominantly impacted the\n\nSouthwest, Southern Plains, and Southeast.\n\n• 2023 became the costliest insured year on record for US SCS with losses reaching USD60 billion\n\n• 31 events with more than USD1 billion in economic losses, and 24 events with more than USD1 billion of insured losses\n\n• Hawaii’s Lahaina wildfire on Maui became the state’s second costliest event on record, only behind Hurricane Iniki (1992)\n\n• Exceptional rainfall in highly urbanized regions of Florida and New York/New Jersey led to billion-dollar economic losses\n\n• Hurricane Idalia was the only hurricane to make landfall in the US, generating an industry loss of at least USD1.25 billion\n\nUS SCS events dominated global insured loss totals\n\nHigh frequency and impactful SCS events in the US dominated global\n\ninsured losses in 2023, led by the hail sub-peril. Hail typically accounts\n\nfor 50% to 80% of SCS claims in any given year. While there has been\n\nno discernible long-term trend in the frequency of SCS outbreaks,\n\nelevated losses in recent years have been aided by expanding urban\n\nfootprints in hail prone regions and increased costs due to aging\n\nhousing stocks.\n\nLikewise, the average annual number of tornadoes has remained\n\ncomparatively constant, however, the number of tornadoes on a\n\nper event basis has shown an increasing trend alongside a shift in\n\ngeographical distribution. This has led to elevated tornado counts\n\nacross highly vulnerable regions of the mid-South and Southeast,\n\nwhere a disproportionate number of manufactured homes are\n\noccupied. Concurrently, there has been an upward trend in the\n\nintensity of extreme precipitation associated with severe storm\n\ncomplexes, and the times of the year in which favorable atmospheric\n\nconditions for SCS outbreaks are present have extended amid a\n\nwarming climate.\n\nFigure 23: Top 10 costliest years for insured losses on the US mainland (top) and\n\naggregated losses since 1990: severe convective storms and tropical cyclone\n\nData and Graphic: Gallagher Re\n\n1 2023\n\n2 2020\n\n3 2011\n\n4 2022\n\n5 2021\n\n6 2019\n\n7 2017\n\n8 2012\n\n9 2016\n\n10 2014\n\n$59.7B\n\n$44.0B\n\n$40.1B\n\n$31.5B\n\n$29.1B\n\n$28.2B\n\n$26.6B\n\n$22.4B\n\n$22.0B\n\n$19.4B\n\n1 2005\n\n2 2017\n\n3 2022\n\n4 2021\n\n5 2012\n\n6 2004\n\n7 1992\n\n8 2008\n\n9 2020\n\n10 2018\n\n$134.3B\n\n$68.3B\n\n$59.5B\n\n$42.8B\n\n$42.2B\n\n$39.6B\n\n$38.4B\n\n$31.6B\n\n$29.3B\n\n$23.1B\n\nUS Mainland\n\nTop 10 Costliest Years: Insured Loss\n\nSevere Convective Storm Tropical Cyclone\n\n1990 1995 2000 2005 2010 2015 2020 2025\n\n$100B\n\n$200B\n\n$300B\n\n$400B\n\n$500B\n\n$600B\n\nUSD billion\n\n566 583\n\n1 2023\n\n2 2020\n\n3 2011\n\n4 2022\n\n5 2021\n\n6 2019\n\n7 2017\n\n8 2012\n\n9 2016\n\n10 2014\n\n$59.7B\n\n$44.0B\n\n$40.1B\n\n$31.5B\n\n$29.1B\n\n$28.2B\n\n$26.6B\n\n$22.4B\n\n$22.0B\n\n$19.4B\n\n1 2005\n\n2 2017\n\n3 2022\n\n4 2021\n\n5 2012\n\n6 2004\n\n7 1992\n\n8 2008\n\n9 2020\n\n10 2018\n\n$134.3B\n\n$68.3B\n\n$59.5B\n\n$42.8B\n\n$42.2B\n\n$39.6B\n\n$38.4B\n\n$31.6B\n\n$29.3B\n\n$23.1B\n\nUS Mainland Top 10 Costliest Years: Insured Loss\n\nSevere Convective Storm Tropical Cyclone\n\nSCS TC\n\nCumulative US Insured Loss: SCS vs TC","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Regional Recaps\n\n41\n\nIn the US, elevated industry losses were driven by high frequency\n\nand record severe convective storm (SCS) activity throughout the\n\nyear. While large hail continued to produce the biggest share of SCS\n\nloss, notable tornadic activity, severe wind gusts, and heavy rainfall\n\naided in pushing totals higher. SCS insured losses in 2023 reached\n\nUSD60 billion. 2023 became only the third year on record to reach or\n\nexceed the USD40 billion threshold, alongside 2020 (USD44 billion)\n\nand 2011 (USD40 billion). In addition to SCS, notable events included\n\nthe Lahaina Wildfire in Hawaii, billion-dollar-plus economic flood\n\nlosses following record-setting 24-hour rainfall in Fort Lauderdale and\n\nNew York City, Hurricane Idalia’s Category 3 landfall in Florida, and\n\nexpansive drought conditions which predominantly impacted the\n\nSouthwest, Southern Plains, and Southeast.\n\n• 2023 became the costliest insured year on record for US SCS with losses reaching USD60 billion\n\n• 31 events with more than USD1 billion in economic losses, and 24 events with more than USD1 billion of insured losses\n\n• Hawaii’s Lahaina wildfire on Maui became the state’s second costliest event on record, only behind Hurricane Iniki (1992)\n\n• Exceptional rainfall in highly urbanized regions of Florida and New York/New Jersey led to billion-dollar economic losses\n\n• Hurricane Idalia was the only hurricane to make landfall in the US, generating an industry loss of at least USD1.25 billion\n\nUS SCS events dominated global insured loss totals\n\nHigh frequency and impactful SCS events in the US dominated global\n\ninsured losses in 2023, led by the hail sub-peril. Hail typically accounts\n\nfor 50% to 80% of SCS claims in any given year. While there has been\n\nno discernible long-term trend in the frequency of SCS outbreaks,\n\nelevated losses in recent years have been aided by expanding urban\n\nfootprints in hail prone regions and increased costs due to aging\n\nhousing stocks.\n\nLikewise, the average annual number of tornadoes has remained\n\ncomparatively constant, however, the number of tornadoes on a\n\nper event basis has shown an increasing trend alongside a shift in\n\ngeographical distribution. This has led to elevated tornado counts\n\nacross highly vulnerable regions of the mid-South and Southeast,\n\nwhere a disproportionate number of manufactured homes are\n\noccupied. Concurrently, there has been an upward trend in the\n\nintensity of extreme precipitation associated with severe storm\n\ncomplexes, and the times of the year in which favorable atmospheric\n\nconditions for SCS outbreaks are present have extended amid a\n\nwarming climate.\n\nFigure 23: Top 10 costliest years for insured losses on the US mainland (top) and\n\naggregated losses since 1990: severe convective storms and tropical cyclone\n\nData and Graphic: Gallagher Re\n\n1 2023\n\n2 2020\n\n3 2011\n\n4 2022\n\n5 2021\n\n6 2019\n\n7 2017\n\n8 2012\n\n9 2016\n\n10 2014\n\n$59.7B\n\n$44.0B\n\n$40.1B\n\n$31.5B\n\n$29.1B\n\n$28.2B\n\n$26.6B\n\n$22.4B\n\n$22.0B\n\n$19.4B\n\n1 2005\n\n2 2017\n\n3 2022\n\n4 2021\n\n5 2012\n\n6 2004\n\n7 1992\n\n8 2008\n\n9 2020\n\n10 2018\n\n$134.3B\n\n$68.3B\n\n$59.5B\n\n$42.8B\n\n$42.2B\n\n$39.6B\n\n$38.4B\n\n$31.6B\n\n$29.3B\n\n$23.1B\n\nUS Mainland\n\nTop 10 Costliest Years: Insured Loss\n\nSevere Convective Storm Tropical Cyclone\n\n1990 1995 2000 2005 2010 2015 2020 2025\n\n$100B\n\n$200B\n\n$300B\n\n$400B\n\n$500B\n\n$600B\n\nUSD billion\n\n566 583\n\n1 2023\n\n2 2020\n\n3 2011\n\n4 2022\n\n5 2021\n\n6 2019\n\n7 2017\n\n8 2012\n\n9 2016\n\n10 2014\n\n$59.7B\n\n$44.0B\n\n$40.1B\n\n$31.5B\n\n$29.1B\n\n$28.2B\n\n$26.6B\n\n$22.4B\n\n$22.0B\n\n$19.4B\n\n1 2005\n\n2 2017\n\n3 2022\n\n4 2021\n\n5 2012\n\n6 2004\n\n7 1992\n\n8 2008\n\n9 2020\n\n10 2018\n\n$134.3B\n\n$68.3B\n\n$59.5B\n\n$42.8B\n\n$42.2B\n\n$39.6B\n\n$38.4B\n\n$31.6B\n\n$29.3B\n\n$23.1B\n\nUS Mainland Top 10 Costliest Years: Insured Loss\n\nSevere Convective Storm Tropical Cyclone\n\nSCS TC\n\nCumulative US Insured Loss: SCS vs TC","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/natural-catastrophe-and-climate-report-2023_p41.pdf"]},"response":{"errors":[],"extract_id":"extract_582aca753b6560a3401b907580477167","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Regional Recaps\n\n41\n\nIn the US, elevated industry losses were driven by high frequency\n\nand record severe convective storm (SCS) activity throughout the\n\nyear. While large hail continued to produce the biggest share of SCS\n\nloss, notable tornadic activity, severe wind gusts, and heavy rainfall\n\naided in pushing totals higher. SCS insured losses in 2023 reached\n\nUSD60 billion. 2023 became only the third year on record to reach or\n\nexceed the USD40 billion threshold, alongside 2020 (USD44 billion)\n\nand 2011 (USD40 billion). In addition to SCS, notable events included\n\nthe Lahaina Wildfire in Hawaii, billion-dollar-plus economic flood\n\nlosses following record-setting 24-hour rainfall in Fort Lauderdale and\n\nNew York City, Hurricane Idalia’s Category 3 landfall in Florida, and\n\nexpansive drought conditions which predominantly impacted the\n\nSouthwest, Southern Plains, and Southeast.\n\n• 2023 became the costliest insured year on record for US SCS with losses reaching USD60 billion\n\n• 31 events with more than USD1 billion in economic losses, and 24 events with more than USD1 billion of insured losses\n\n• Hawaii’s Lahaina wildfire on Maui became the state’s second costliest event on record, only behind Hurricane Iniki (1992)\n\n• Exceptional rainfall in highly urbanized regions of Florida and New York/New Jersey led to billion-dollar economic losses\n\n• Hurricane Idalia was the only hurricane to make landfall in the US, generating an industry loss of at least USD1.25 billion\n\nUS SCS events dominated global insured loss totals\n\nHigh frequency and impactful SCS events in the US dominated global\n\ninsured losses in 2023, led by the hail sub-peril. Hail typically accounts\n\nfor 50% to 80% of SCS claims in any given year. While there has been\n\nno discernible long-term trend in the frequency of SCS outbreaks,\n\nelevated losses in recent years have been aided by expanding urban\n\nfootprints in hail prone regions and increased costs due to aging\n\nhousing stocks.\n\nLikewise, the average annual number of tornadoes has remained\n\ncomparatively constant, however, the number of tornadoes on a\n\nper event basis has shown an increasing trend alongside a shift in\n\ngeographical distribution. This has led to elevated tornado counts\n\nacross highly vulnerable regions of the mid-South and Southeast,\n\nwhere a disproportionate number of manufactured homes are\n\noccupied. Concurrently, there has been an upward trend in the\n\nintensity of extreme precipitation associated with severe storm\n\ncomplexes, and the times of the year in which favorable atmospheric\n\nconditions for SCS outbreaks are present have extended amid a\n\nwarming climate.\n\nFigure 23: Top 10 costliest years for insured losses on the US mainland (top) and\n\naggregated losses since 1990: severe convective storms and tropical cyclone\n\nData and Graphic: Gallagher Re\n\n1 2023\n\n2 2020\n\n3 2011\n\n4 2022\n\n5 2021\n\n6 2019\n\n7 2017\n\n8 2012\n\n9 2016\n\n10 2014\n\n$59.7B\n\n$44.0B\n\n$40.1B\n\n$31.5B\n\n$29.1B\n\n$28.2B\n\n$26.6B\n\n$22.4B\n\n$22.0B\n\n$19.4B\n\n1 2005\n\n2 2017\n\n3 2022\n\n4 2021\n\n5 2012\n\n6 2004\n\n7 1992\n\n8 2008\n\n9 2020\n\n10 2018\n\n$134.3B\n\n$68.3B\n\n$59.5B\n\n$42.8B\n\n$42.2B\n\n$39.6B\n\n$38.4B\n\n$31.6B\n\n$29.3B\n\n$23.1B\n\nUS Mainland\n\nTop 10 Costliest Years: Insured Loss\n\nSevere Convective Storm Tropical Cyclone\n\n1990 1995 2000 2005 2010 2015 2020 2025\n\n$100B\n\n$200B\n\n$300B\n\n$400B\n\n$500B\n\n$600B\n\nUSD billion\n\n566 583\n\n1 2023\n\n2 2020\n\n3 2011\n\n4 2022\n\n5 2021\n\n6 2019\n\n7 2017\n\n8 2012\n\n9 2016\n\n10 2014\n\n$59.7B\n\n$44.0B\n\n$40.1B\n\n$31.5B\n\n$29.1B\n\n$28.2B\n\n$26.6B\n\n$22.4B\n\n$22.0B\n\n$19.4B\n\n1 2005\n\n2 2017\n\n3 2022\n\n4 2021\n\n5 2012\n\n6 2004\n\n7 1992\n\n8 2008\n\n9 2020\n\n10 2018\n\n$134.3B\n\n$68.3B\n\n$59.5B\n\n$42.8B\n\n$42.2B\n\n$39.6B\n\n$38.4B\n\n$31.6B\n\n$29.3B\n\n$23.1B\n\nUS Mainland Top 10 Costliest Years: Insured Loss\n\nSevere Convective Storm Tropical Cyclone\n\nSCS TC\n\nCumulative US Insured Loss: SCS vs TC"],"full_content":"Regional Recaps\n\n41\n\nIn the US, elevated industry losses were driven by high frequency\n\nand record severe convective storm (SCS) activity throughout the\n\nyear. While large hail continued to produce the biggest share of SCS\n\nloss, notable tornadic activity, severe wind gusts, and heavy rainfall\n\naided in pushing totals higher. SCS insured losses in 2023 reached\n\nUSD60 billion. 2023 became only the third year on record to reach or\n\nexceed the USD40 billion threshold, alongside 2020 (USD44 billion)\n\nand 2011 (USD40 billion). In addition to SCS, notable events included\n\nthe Lahaina Wildfire in Hawaii, billion-dollar-plus economic flood\n\nlosses following record-setting 24-hour rainfall in Fort Lauderdale and\n\nNew York City, Hurricane Idalia’s Category 3 landfall in Florida, and\n\nexpansive drought conditions which predominantly impacted the\n\nSouthwest, Southern Plains, and Southeast.\n\n• 2023 became the costliest insured year on record for US SCS with losses reaching USD60 billion\n\n• 31 events with more than USD1 billion in economic losses, and 24 events with more than USD1 billion of insured losses\n\n• Hawaii’s Lahaina wildfire on Maui became the state’s second costliest event on record, only behind Hurricane Iniki (1992)\n\n• Exceptional rainfall in highly urbanized regions of Florida and New York/New Jersey led to billion-dollar economic losses\n\n• Hurricane Idalia was the only hurricane to make landfall in the US, generating an industry loss of at least USD1.25 billion\n\nUS SCS events dominated global insured loss totals\n\nHigh frequency and impactful SCS events in the US dominated global\n\ninsured losses in 2023, led by the hail sub-peril. Hail typically accounts\n\nfor 50% to 80% of SCS claims in any given year. While there has been\n\nno discernible long-term trend in the frequency of SCS outbreaks,\n\nelevated losses in recent years have been aided by expanding urban\n\nfootprints in hail prone regions and increased costs due to aging\n\nhousing stocks.\n\nLikewise, the average annual number of tornadoes has remained\n\ncomparatively constant, however, the number of tornadoes on a\n\nper event basis has shown an increasing trend alongside a shift in\n\ngeographical distribution. This has led to elevated tornado counts\n\nacross highly vulnerable regions of the mid-South and Southeast,\n\nwhere a disproportionate number of manufactured homes are\n\noccupied. Concurrently, there has been an upward trend in the\n\nintensity of extreme precipitation associated with severe storm\n\ncomplexes, and the times of the year in which favorable atmospheric\n\nconditions for SCS outbreaks are present have extended amid a\n\nwarming climate.\n\nFigure 23: Top 10 costliest years for insured losses on the US mainland (top) and\n\naggregated losses since 1990: severe convective storms and tropical cyclone\n\nData and Graphic: Gallagher Re\n\n1 2023\n\n2 2020\n\n3 2011\n\n4 2022\n\n5 2021\n\n6 2019\n\n7 2017\n\n8 2012\n\n9 2016\n\n10 2014\n\n$59.7B\n\n$44.0B\n\n$40.1B\n\n$31.5B\n\n$29.1B\n\n$28.2B\n\n$26.6B\n\n$22.4B\n\n$22.0B\n\n$19.4B\n\n1 2005\n\n2 2017\n\n3 2022\n\n4 2021\n\n5 2012\n\n6 2004\n\n7 1992\n\n8 2008\n\n9 2020\n\n10 2018\n\n$134.3B\n\n$68.3B\n\n$59.5B\n\n$42.8B\n\n$42.2B\n\n$39.6B\n\n$38.4B\n\n$31.6B\n\n$29.3B\n\n$23.1B\n\nUS Mainland\n\nTop 10 Costliest Years: Insured Loss\n\nSevere Convective Storm Tropical Cyclone\n\n1990 1995 2000 2005 2010 2015 2020 2025\n\n$100B\n\n$200B\n\n$300B\n\n$400B\n\n$500B\n\n$600B\n\nUSD billion\n\n566 583\n\n1 2023\n\n2 2020\n\n3 2011\n\n4 2022\n\n5 2021\n\n6 2019\n\n7 2017\n\n8 2012\n\n9 2016\n\n10 2014\n\n$59.7B\n\n$44.0B\n\n$40.1B\n\n$31.5B\n\n$29.1B\n\n$28.2B\n\n$26.6B\n\n$22.4B\n\n$22.0B\n\n$19.4B\n\n1 2005\n\n2 2017\n\n3 2022\n\n4 2021\n\n5 2012\n\n6 2004\n\n7 1992\n\n8 2008\n\n9 2020\n\n10 2018\n\n$134.3B\n\n$68.3B\n\n$59.5B\n\n$42.8B\n\n$42.2B\n\n$39.6B\n\n$38.4B\n\n$31.6B\n\n$29.3B\n\n$23.1B\n\nUS Mainland Top 10 Costliest Years: Insured Loss\n\nSevere Convective Storm Tropical Cyclone\n\nSCS TC\n\nCumulative US Insured Loss: SCS vs TC","publish_date":null,"title":"US Mainland Cumulative US Insured Loss: SCS vs TC Top 10 Costliest Years: Insured Loss 1 1 2 2 3","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/natural-catastrophe-and-climate-report-2023_p41.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/natural-catastrophe-and-climate-report-2023_p41.pdf"},"request":{"example_id":"chart/natural-catastrophe-and-climate-report-2023_p41","product_type":"parse","source_file_path":"natural-catastrophe-and-climate-report-2023_p41.pdf"},"started_at":"2026-11-04T19:45:09.302226"} +{"completed_at":"2026-11-04T19:45:13.735843","latency_in_ms":2637,"output":{"example_id":"chart/natural-catastrophe-and-climate-report-2023_p54","grounded_pages":[],"layout_pages":[],"markdown":"Regional Recaps\n\n54\n\n## APAC\n\nFigure 36: Map of notable Asia-Pacific (APAC) events in 2023 | Data and Graphic: Gallagher Re\n\nFigure 37: Asia-Pacific (APAC) natural catastrophe statistics | Data and Graphic: Gallagher Re\n\nSingapore\n\nKuala Lumpur\n\nHong Kong\n\nNew Delhi\n\nBangkok Chennai\n\nMumbai\n\nJakarta\n\nBeijing\n\nManila\n\nTokyo\n\nAfghanistan Earthquakes\n\nOctober sequence in Herat Province; Four tremors measured at M6.3; Total Fatalities: 1,489\n\nTyphoon Doksuri\n\nRecord rainfall from storm's remnants led to catastrophic inland flooding; Eco: USD18.5bn / Ins: 1.7bn\n\nCyclone Michaung\n\nTorrential rain & flooding in Chennai Eco: USD1.9bn / Ins: USD240mn Cyclone Mocha\n\nCategory 4 landfall in Myanmar Economic Loss; USD2.2bn\n\nJapan: Seasonal Rains Torrential rains in July led to extensive flood damage Insured Loss: USD350mn\n\nChristchurch\n\nMelbourne Auckland\n\nBrisbane\n\nSydney\n\nPerth\n\nNorth Island (NZ) Floods\n\nEconomic Loss: USD3.1bn Insured Loss: USD1.9bn\n\nCyclone Gabrielle\n\nEconomic Loss; USD3.8bn Insured Loss: USD1.8bn\n\nAustralia Summer Heat\n\nSydney: 43.5°C / 110.3°F on Dec 9 City's hottest day since 1929\n\nCyclone Jasper Extensive flash flood damage in QLD; 2,252 mm / 88.66 in at Bairds; Wettest Australia TC on record\n\nAsia Oceania\n\nAnnual APAC Natural Peril Losses Aggregate APAC Weather / Climate Losses\n\n# 5\n\nNumber of Asian Countries w/ All-Time Heat Records Set in 2023 Countries: CN, TH, VN, LA, SG\n\n# 17\n\nNumber of Named Storms in the 2023 NW Pacific Typhoon Season; Third-Fewest Since 1980\n\nNotable Statistics in 2023\n\n# 744\\.8\n\nAmount of Rainfall (mm) in Beijing via Remnants of TY Doksuri; Heaviest Total Since 1883\n\n# 2\n\nWeeks Between New Zealand's Two Costliest Weather Disasters On Record\n\nEconomic Loss Insured Loss\n\n2000 2005 2010 2015 2020\n\n$100B\n\n$200B\n\n$300B\n\n$400B\n\n$500B\n\n$600B\n\nUSD billion\n\n2000 2005 2010 2015 2020\n\n$500B\n\n$1,000B\n\n$1,500B\n\n$2,000B\n\n$2,500B\n\nUSD billion\n\n$2,102B\n\n$243B","pages":[{"line_numbers":[],"markdown":"Regional Recaps\n\n54\n\n## APAC\n\nFigure 36: Map of notable Asia-Pacific (APAC) events in 2023 | Data and Graphic: Gallagher Re\n\nFigure 37: Asia-Pacific (APAC) natural catastrophe statistics | Data and Graphic: Gallagher Re\n\nSingapore\n\nKuala Lumpur\n\nHong Kong\n\nNew Delhi\n\nBangkok Chennai\n\nMumbai\n\nJakarta\n\nBeijing\n\nManila\n\nTokyo\n\nAfghanistan Earthquakes\n\nOctober sequence in Herat Province; Four tremors measured at M6.3; Total Fatalities: 1,489\n\nTyphoon Doksuri\n\nRecord rainfall from storm's remnants led to catastrophic inland flooding; Eco: USD18.5bn / Ins: 1.7bn\n\nCyclone Michaung\n\nTorrential rain & flooding in Chennai Eco: USD1.9bn / Ins: USD240mn Cyclone Mocha\n\nCategory 4 landfall in Myanmar Economic Loss; USD2.2bn\n\nJapan: Seasonal Rains Torrential rains in July led to extensive flood damage Insured Loss: USD350mn\n\nChristchurch\n\nMelbourne Auckland\n\nBrisbane\n\nSydney\n\nPerth\n\nNorth Island (NZ) Floods\n\nEconomic Loss: USD3.1bn Insured Loss: USD1.9bn\n\nCyclone Gabrielle\n\nEconomic Loss; USD3.8bn Insured Loss: USD1.8bn\n\nAustralia Summer Heat\n\nSydney: 43.5°C / 110.3°F on Dec 9 City's hottest day since 1929\n\nCyclone Jasper Extensive flash flood damage in QLD; 2,252 mm / 88.66 in at Bairds; Wettest Australia TC on record\n\nAsia Oceania\n\nAnnual APAC Natural Peril Losses Aggregate APAC Weather / Climate Losses\n\n# 5\n\nNumber of Asian Countries w/ All-Time Heat Records Set in 2023 Countries: CN, TH, VN, LA, SG\n\n# 17\n\nNumber of Named Storms in the 2023 NW Pacific Typhoon Season; Third-Fewest Since 1980\n\nNotable Statistics in 2023\n\n# 744\\.8\n\nAmount of Rainfall (mm) in Beijing via Remnants of TY Doksuri; Heaviest Total Since 1883\n\n# 2\n\nWeeks Between New Zealand's Two Costliest Weather Disasters On Record\n\nEconomic Loss Insured Loss\n\n2000 2005 2010 2015 2020\n\n$100B\n\n$200B\n\n$300B\n\n$400B\n\n$500B\n\n$600B\n\nUSD billion\n\n2000 2005 2010 2015 2020\n\n$500B\n\n$1,000B\n\n$1,500B\n\n$2,000B\n\n$2,500B\n\nUSD billion\n\n$2,102B\n\n$243B","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Regional Recaps\n\n54\n\n## APAC\n\nFigure 36: Map of notable Asia-Pacific (APAC) events in 2023 | Data and Graphic: Gallagher Re\n\nFigure 37: Asia-Pacific (APAC) natural catastrophe statistics | Data and Graphic: Gallagher Re\n\nSingapore\n\nKuala Lumpur\n\nHong Kong\n\nNew Delhi\n\nBangkok Chennai\n\nMumbai\n\nJakarta\n\nBeijing\n\nManila\n\nTokyo\n\nAfghanistan Earthquakes\n\nOctober sequence in Herat Province; Four tremors measured at M6.3; Total Fatalities: 1,489\n\nTyphoon Doksuri\n\nRecord rainfall from storm's remnants led to catastrophic inland flooding; Eco: USD18.5bn / Ins: 1.7bn\n\nCyclone Michaung\n\nTorrential rain & flooding in Chennai Eco: USD1.9bn / Ins: USD240mn Cyclone Mocha\n\nCategory 4 landfall in Myanmar Economic Loss; USD2.2bn\n\nJapan: Seasonal Rains Torrential rains in July led to extensive flood damage Insured Loss: USD350mn\n\nChristchurch\n\nMelbourne Auckland\n\nBrisbane\n\nSydney\n\nPerth\n\nNorth Island (NZ) Floods\n\nEconomic Loss: USD3.1bn Insured Loss: USD1.9bn\n\nCyclone Gabrielle\n\nEconomic Loss; USD3.8bn Insured Loss: USD1.8bn\n\nAustralia Summer Heat\n\nSydney: 43.5°C / 110.3°F on Dec 9 City's hottest day since 1929\n\nCyclone Jasper Extensive flash flood damage in QLD; 2,252 mm / 88.66 in at Bairds; Wettest Australia TC on record\n\nAsia Oceania\n\nAnnual APAC Natural Peril Losses Aggregate APAC Weather / Climate Losses\n\n# 5\n\nNumber of Asian Countries w/ All-Time Heat Records Set in 2023 Countries: CN, TH, VN, LA, SG\n\n# 17\n\nNumber of Named Storms in the 2023 NW Pacific Typhoon Season; Third-Fewest Since 1980\n\nNotable Statistics in 2023\n\n# 744\\.8\n\nAmount of Rainfall (mm) in Beijing via Remnants of TY Doksuri; Heaviest Total Since 1883\n\n# 2\n\nWeeks Between New Zealand's Two Costliest Weather Disasters On Record\n\nEconomic Loss Insured Loss\n\n2000 2005 2010 2015 2020\n\n$100B\n\n$200B\n\n$300B\n\n$400B\n\n$500B\n\n$600B\n\nUSD billion\n\n2000 2005 2010 2015 2020\n\n$500B\n\n$1,000B\n\n$1,500B\n\n$2,000B\n\n$2,500B\n\nUSD billion\n\n$2,102B\n\n$243B","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/natural-catastrophe-and-climate-report-2023_p54.pdf"]},"response":{"errors":[],"extract_id":"extract_4e2dc2f21dbd238557c6a18db3e33e5d","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Regional Recaps\n\n54\n\n## APAC\n\nFigure 36: Map of notable Asia-Pacific (APAC) events in 2023 | Data and Graphic: Gallagher Re\n\nFigure 37: Asia-Pacific (APAC) natural catastrophe statistics | Data and Graphic: Gallagher Re\n\nSingapore\n\nKuala Lumpur\n\nHong Kong\n\nNew Delhi\n\nBangkok Chennai\n\nMumbai\n\nJakarta\n\nBeijing\n\nManila\n\nTokyo\n\nAfghanistan Earthquakes\n\nOctober sequence in Herat Province; Four tremors measured at M6.3; Total Fatalities: 1,489\n\nTyphoon Doksuri\n\nRecord rainfall from storm's remnants led to catastrophic inland flooding; Eco: USD18.5bn / Ins: 1.7bn\n\nCyclone Michaung\n\nTorrential rain & flooding in Chennai Eco: USD1.9bn / Ins: USD240mn Cyclone Mocha\n\nCategory 4 landfall in Myanmar Economic Loss; USD2.2bn\n\nJapan: Seasonal Rains Torrential rains in July led to extensive flood damage Insured Loss: USD350mn\n\nChristchurch\n\nMelbourne Auckland\n\nBrisbane\n\nSydney\n\nPerth\n\nNorth Island (NZ) Floods\n\nEconomic Loss: USD3.1bn Insured Loss: USD1.9bn\n\nCyclone Gabrielle\n\nEconomic Loss; USD3.8bn Insured Loss: USD1.8bn\n\nAustralia Summer Heat\n\nSydney: 43.5°C / 110.3°F on Dec 9 City's hottest day since 1929\n\nCyclone Jasper Extensive flash flood damage in QLD; 2,252 mm / 88.66 in at Bairds; Wettest Australia TC on record\n\nAsia Oceania\n\nAnnual APAC Natural Peril Losses Aggregate APAC Weather / Climate Losses\n\n# 5\n\nNumber of Asian Countries w/ All-Time Heat Records Set in 2023 Countries: CN, TH, VN, LA, SG\n\n# 17\n\nNumber of Named Storms in the 2023 NW Pacific Typhoon Season; Third-Fewest Since 1980\n\nNotable Statistics in 2023\n\n# 744\\.8\n\nAmount of Rainfall (mm) in Beijing via Remnants of TY Doksuri; Heaviest Total Since 1883\n\n# 2\n\nWeeks Between New Zealand's Two Costliest Weather Disasters On Record\n\nEconomic Loss Insured Loss\n\n2000 2005 2010 2015 2020\n\n$100B\n\n$200B\n\n$300B\n\n$400B\n\n$500B\n\n$600B\n\nUSD billion\n\n2000 2005 2010 2015 2020\n\n$500B\n\n$1,000B\n\n$1,500B\n\n$2,000B\n\n$2,500B\n\nUSD billion\n\n$2,102B\n\n$243B"],"full_content":"Regional Recaps\n\n54\n\n## APAC\n\nFigure 36: Map of notable Asia-Pacific (APAC) events in 2023 | Data and Graphic: Gallagher Re\n\nFigure 37: Asia-Pacific (APAC) natural catastrophe statistics | Data and Graphic: Gallagher Re\n\nSingapore\n\nKuala Lumpur\n\nHong Kong\n\nNew Delhi\n\nBangkok Chennai\n\nMumbai\n\nJakarta\n\nBeijing\n\nManila\n\nTokyo\n\nAfghanistan Earthquakes\n\nOctober sequence in Herat Province; Four tremors measured at M6.3; Total Fatalities: 1,489\n\nTyphoon Doksuri\n\nRecord rainfall from storm's remnants led to catastrophic inland flooding; Eco: USD18.5bn / Ins: 1.7bn\n\nCyclone Michaung\n\nTorrential rain & flooding in Chennai Eco: USD1.9bn / Ins: USD240mn Cyclone Mocha\n\nCategory 4 landfall in Myanmar Economic Loss; USD2.2bn\n\nJapan: Seasonal Rains Torrential rains in July led to extensive flood damage Insured Loss: USD350mn\n\nChristchurch\n\nMelbourne Auckland\n\nBrisbane\n\nSydney\n\nPerth\n\nNorth Island (NZ) Floods\n\nEconomic Loss: USD3.1bn Insured Loss: USD1.9bn\n\nCyclone Gabrielle\n\nEconomic Loss; USD3.8bn Insured Loss: USD1.8bn\n\nAustralia Summer Heat\n\nSydney: 43.5°C / 110.3°F on Dec 9 City's hottest day since 1929\n\nCyclone Jasper Extensive flash flood damage in QLD; 2,252 mm / 88.66 in at Bairds; Wettest Australia TC on record\n\nAsia Oceania\n\nAnnual APAC Natural Peril Losses Aggregate APAC Weather / Climate Losses\n\n# 5\n\nNumber of Asian Countries w/ All-Time Heat Records Set in 2023 Countries: CN, TH, VN, LA, SG\n\n# 17\n\nNumber of Named Storms in the 2023 NW Pacific Typhoon Season; Third-Fewest Since 1980\n\nNotable Statistics in 2023\n\n# 744\\.8\n\nAmount of Rainfall (mm) in Beijing via Remnants of TY Doksuri; Heaviest Total Since 1883\n\n# 2\n\nWeeks Between New Zealand's Two Costliest Weather Disasters On Record\n\nEconomic Loss Insured Loss\n\n2000 2005 2010 2015 2020\n\n$100B\n\n$200B\n\n$300B\n\n$400B\n\n$500B\n\n$600B\n\nUSD billion\n\n2000 2005 2010 2015 2020\n\n$500B\n\n$1,000B\n\n$1,500B\n\n$2,000B\n\n$2,500B\n\nUSD billion\n\n$2,102B\n\n$243B","publish_date":null,"title":"5 17 744.8 2","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/natural-catastrophe-and-climate-report-2023_p54.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/natural-catastrophe-and-climate-report-2023_p54.pdf"},"request":{"example_id":"chart/natural-catastrophe-and-climate-report-2023_p54","product_type":"parse","source_file_path":"natural-catastrophe-and-climate-report-2023_p54.pdf"},"started_at":"2026-11-04T19:45:11.098216"} +{"completed_at":"2026-11-04T19:45:14.357018","latency_in_ms":1430,"output":{"example_id":"chart/natural-catastrophe-and-climate-report-2023_p72","grounded_pages":[],"layout_pages":[],"markdown":"Appendix\n\n72\n\nFigure 42: Total number of billion-dollar economic loss natural catastrophes; losses adjusted to 2023 USD | Data and Graphic: Gallagher Re\n\nFigure 43: Total number of billion-dollar insured loss natural catastrophes; losses adjusted to 2023 USD | Data and Graphic: Gallagher Re\n\n## Total Number of Billion-Dollar Events: Economic and Insured (1990–2023)\n\nWeather / Climate Other Natural Perils\n\n1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014 2016 2018 2020 2022\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\n70\n\n1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014 2016 2018 2020 2022\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n35\n\nWeather / Climate Other Natural Perils","pages":[{"line_numbers":[],"markdown":"Appendix\n\n72\n\nFigure 42: Total number of billion-dollar economic loss natural catastrophes; losses adjusted to 2023 USD | Data and Graphic: Gallagher Re\n\nFigure 43: Total number of billion-dollar insured loss natural catastrophes; losses adjusted to 2023 USD | Data and Graphic: Gallagher Re\n\n## Total Number of Billion-Dollar Events: Economic and Insured (1990–2023)\n\nWeather / Climate Other Natural Perils\n\n1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014 2016 2018 2020 2022\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\n70\n\n1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014 2016 2018 2020 2022\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n35\n\nWeather / Climate Other Natural Perils","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Appendix\n\n72\n\nFigure 42: Total number of billion-dollar economic loss natural catastrophes; losses adjusted to 2023 USD | Data and Graphic: Gallagher Re\n\nFigure 43: Total number of billion-dollar insured loss natural catastrophes; losses adjusted to 2023 USD | Data and Graphic: Gallagher Re\n\n## Total Number of Billion-Dollar Events: Economic and Insured (1990–2023)\n\nWeather / Climate Other Natural Perils\n\n1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014 2016 2018 2020 2022\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\n70\n\n1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014 2016 2018 2020 2022\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n35\n\nWeather / Climate Other Natural Perils","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/natural-catastrophe-and-climate-report-2023_p72.pdf"]},"response":{"errors":[],"extract_id":"extract_679231d849720c1a1fcdf15be747889f","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Appendix\n\n72\n\nFigure 42: Total number of billion-dollar economic loss natural catastrophes; losses adjusted to 2023 USD | Data and Graphic: Gallagher Re\n\nFigure 43: Total number of billion-dollar insured loss natural catastrophes; losses adjusted to 2023 USD | Data and Graphic: Gallagher Re\n\n## Total Number of Billion-Dollar Events: Economic and Insured (1990–2023)\n\nWeather / Climate Other Natural Perils\n\n1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014 2016 2018 2020 2022\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\n70\n\n1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014 2016 2018 2020 2022\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n35\n\nWeather / Climate Other Natural Perils"],"full_content":"Appendix\n\n72\n\nFigure 42: Total number of billion-dollar economic loss natural catastrophes; losses adjusted to 2023 USD | Data and Graphic: Gallagher Re\n\nFigure 43: Total number of billion-dollar insured loss natural catastrophes; losses adjusted to 2023 USD | Data and Graphic: Gallagher Re\n\n## Total Number of Billion-Dollar Events: Economic and Insured (1990–2023)\n\nWeather / Climate Other Natural Perils\n\n1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014 2016 2018 2020 2022\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\n70\n\n1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014 2016 2018 2020 2022\n\n5\n\n10\n\n15\n\n20\n\n25\n\n30\n\n35\n\nWeather / Climate Other Natural Perils","publish_date":null,"title":"Total Number of Billion-Dollar Events: Economic and Insured (1990–2023)","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/natural-catastrophe-and-climate-report-2023_p72.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/natural-catastrophe-and-climate-report-2023_p72.pdf"},"request":{"example_id":"chart/natural-catastrophe-and-climate-report-2023_p72","product_type":"parse","source_file_path":"natural-catastrophe-and-climate-report-2023_p72.pdf"},"started_at":"2026-11-04T19:45:12.926801"} +{"completed_at":"2026-11-04T19:45:16.539697","latency_in_ms":2405,"output":{"example_id":"chart/natural-catastrophe-and-climate-report-q3-2025_p23","grounded_pages":[],"layout_pages":[],"markdown":"A persistent characteristic of 2025 SCS outbreaks has been\n\ninstances of torrential rainfall spawning flash flooding. These SCS\n\ndriven flash-flooding events were aided by multiple factors, which\n\nincluded cut-off (stuck) low pressure systems in tandem with\n\npersistent flow from the Gulf and / or training (repeating) storms\n\nthat formed near urban regions or recent burn scars. Notable\n\nevents included flash flooding in the Texas Hill Country, Ruidoso\n\n(New Mexico), New York City (New York), Chicago (Illinois), and\n\nMilwaukee (Wisconsin) among others. In many instances, low\n\nNational Flood Insurance Program (NFIP) take up rates meant\n\nmost of the flood-related damage went uninsured.\n\nUnited States\n\nSCS losses in the US continued to drive a substantial portion\n\nof global insured SCS losses during the first nine months of\n\n2025\\. Most of the losses were incurred following a highly active\n\nperiod for the peril during the peak months between March and\n\nJune. US SCS accounted for 44% of global insured losses alone\n\nthrough Q3. To date, the US has experienced 13 billion-dollar\n\ninsured SCS loss events, of which five saw multi-billion-dollar\n\nindustry loss totals (USD2+ billion).\n\n2025 marked the third consecutive year in which US SCS insured\n\nlosses through September exceeded USD40 billion. When\n\nextending to the full calendar year, 2025 already ranks as the\n\nfourth-costliest year for insurers on record – only behind 2023\n\n(USD62 billion), 2024 (USD57 billion), and 2020 (USD47 billion).\n\nAs some of the larger 2025 outbreaks continue to be assessed\n\nand claims are processed, further loss progression is expected in\n\nthe months ahead.\n\n## US & Europe: Severe Convective Storm\n\nUS SCS YTD: Economic losses reached USD61 billion as the portion covered by insurance is now up to USD46 billion\n\nUS: Insured SCS losses already amount to the fourth-costliest year on record; only behind 2023, 2024, and 2020\n\nUS: At least 17 individual billion-dollar economic loss outbreaks; 13+ also resulted in a billion-dollar (or more) insured loss\n\nEurope: Generally benign SCS season compared to recent years; costliest stretch occurred in France during May and June\n\n# Major Event Reviews\n\n2000 2002 2004 2006 2008 2010 2012 2014 2016 2018 2020 2022 2024\n\n100\n\n200\n\n300\n\n400\n\n500\n\n600\n\n700\n\nInsured Loss (USD bn; Today's Dollars)\n\n648\n\n253\n\nUnited States Rest of World Previous year values in 2025 USD using US CPI and a construction / material / cost of labor factor adjustment. Data is thru September 2025.\n\nFigure 14: SCS insured losses from 2000 to 2025 YTD; US compared to the rest of the world Loss Data & Graphic: Gallagher Re\n\n23","pages":[{"line_numbers":[],"markdown":"A persistent characteristic of 2025 SCS outbreaks has been\n\ninstances of torrential rainfall spawning flash flooding. These SCS\n\ndriven flash-flooding events were aided by multiple factors, which\n\nincluded cut-off (stuck) low pressure systems in tandem with\n\npersistent flow from the Gulf and / or training (repeating) storms\n\nthat formed near urban regions or recent burn scars. Notable\n\nevents included flash flooding in the Texas Hill Country, Ruidoso\n\n(New Mexico), New York City (New York), Chicago (Illinois), and\n\nMilwaukee (Wisconsin) among others. In many instances, low\n\nNational Flood Insurance Program (NFIP) take up rates meant\n\nmost of the flood-related damage went uninsured.\n\nUnited States\n\nSCS losses in the US continued to drive a substantial portion\n\nof global insured SCS losses during the first nine months of\n\n2025\\. Most of the losses were incurred following a highly active\n\nperiod for the peril during the peak months between March and\n\nJune. US SCS accounted for 44% of global insured losses alone\n\nthrough Q3. To date, the US has experienced 13 billion-dollar\n\ninsured SCS loss events, of which five saw multi-billion-dollar\n\nindustry loss totals (USD2+ billion).\n\n2025 marked the third consecutive year in which US SCS insured\n\nlosses through September exceeded USD40 billion. When\n\nextending to the full calendar year, 2025 already ranks as the\n\nfourth-costliest year for insurers on record – only behind 2023\n\n(USD62 billion), 2024 (USD57 billion), and 2020 (USD47 billion).\n\nAs some of the larger 2025 outbreaks continue to be assessed\n\nand claims are processed, further loss progression is expected in\n\nthe months ahead.\n\n## US & Europe: Severe Convective Storm\n\nUS SCS YTD: Economic losses reached USD61 billion as the portion covered by insurance is now up to USD46 billion\n\nUS: Insured SCS losses already amount to the fourth-costliest year on record; only behind 2023, 2024, and 2020\n\nUS: At least 17 individual billion-dollar economic loss outbreaks; 13+ also resulted in a billion-dollar (or more) insured loss\n\nEurope: Generally benign SCS season compared to recent years; costliest stretch occurred in France during May and June\n\n# Major Event Reviews\n\n2000 2002 2004 2006 2008 2010 2012 2014 2016 2018 2020 2022 2024\n\n100\n\n200\n\n300\n\n400\n\n500\n\n600\n\n700\n\nInsured Loss (USD bn; Today's Dollars)\n\n648\n\n253\n\nUnited States Rest of World Previous year values in 2025 USD using US CPI and a construction / material / cost of labor factor adjustment. Data is thru September 2025.\n\nFigure 14: SCS insured losses from 2000 to 2025 YTD; US compared to the rest of the world Loss Data & Graphic: Gallagher Re\n\n23","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"A persistent characteristic of 2025 SCS outbreaks has been\n\ninstances of torrential rainfall spawning flash flooding. These SCS\n\ndriven flash-flooding events were aided by multiple factors, which\n\nincluded cut-off (stuck) low pressure systems in tandem with\n\npersistent flow from the Gulf and / or training (repeating) storms\n\nthat formed near urban regions or recent burn scars. Notable\n\nevents included flash flooding in the Texas Hill Country, Ruidoso\n\n(New Mexico), New York City (New York), Chicago (Illinois), and\n\nMilwaukee (Wisconsin) among others. In many instances, low\n\nNational Flood Insurance Program (NFIP) take up rates meant\n\nmost of the flood-related damage went uninsured.\n\nUnited States\n\nSCS losses in the US continued to drive a substantial portion\n\nof global insured SCS losses during the first nine months of\n\n2025\\. Most of the losses were incurred following a highly active\n\nperiod for the peril during the peak months between March and\n\nJune. US SCS accounted for 44% of global insured losses alone\n\nthrough Q3. To date, the US has experienced 13 billion-dollar\n\ninsured SCS loss events, of which five saw multi-billion-dollar\n\nindustry loss totals (USD2+ billion).\n\n2025 marked the third consecutive year in which US SCS insured\n\nlosses through September exceeded USD40 billion. When\n\nextending to the full calendar year, 2025 already ranks as the\n\nfourth-costliest year for insurers on record – only behind 2023\n\n(USD62 billion), 2024 (USD57 billion), and 2020 (USD47 billion).\n\nAs some of the larger 2025 outbreaks continue to be assessed\n\nand claims are processed, further loss progression is expected in\n\nthe months ahead.\n\n## US & Europe: Severe Convective Storm\n\nUS SCS YTD: Economic losses reached USD61 billion as the portion covered by insurance is now up to USD46 billion\n\nUS: Insured SCS losses already amount to the fourth-costliest year on record; only behind 2023, 2024, and 2020\n\nUS: At least 17 individual billion-dollar economic loss outbreaks; 13+ also resulted in a billion-dollar (or more) insured loss\n\nEurope: Generally benign SCS season compared to recent years; costliest stretch occurred in France during May and June\n\n# Major Event Reviews\n\n2000 2002 2004 2006 2008 2010 2012 2014 2016 2018 2020 2022 2024\n\n100\n\n200\n\n300\n\n400\n\n500\n\n600\n\n700\n\nInsured Loss (USD bn; Today's Dollars)\n\n648\n\n253\n\nUnited States Rest of World Previous year values in 2025 USD using US CPI and a construction / material / cost of labor factor adjustment. Data is thru September 2025.\n\nFigure 14: SCS insured losses from 2000 to 2025 YTD; US compared to the rest of the world Loss Data & Graphic: Gallagher Re\n\n23","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/natural-catastrophe-and-climate-report-q3-2025_p23.pdf"]},"response":{"errors":[],"extract_id":"extract_5117e74aa2036d7b5fe2e41e901b62a9","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["A persistent characteristic of 2025 SCS outbreaks has been\n\ninstances of torrential rainfall spawning flash flooding. These SCS\n\ndriven flash-flooding events were aided by multiple factors, which\n\nincluded cut-off (stuck) low pressure systems in tandem with\n\npersistent flow from the Gulf and / or training (repeating) storms\n\nthat formed near urban regions or recent burn scars. Notable\n\nevents included flash flooding in the Texas Hill Country, Ruidoso\n\n(New Mexico), New York City (New York), Chicago (Illinois), and\n\nMilwaukee (Wisconsin) among others. In many instances, low\n\nNational Flood Insurance Program (NFIP) take up rates meant\n\nmost of the flood-related damage went uninsured.\n\nUnited States\n\nSCS losses in the US continued to drive a substantial portion\n\nof global insured SCS losses during the first nine months of\n\n2025\\. Most of the losses were incurred following a highly active\n\nperiod for the peril during the peak months between March and\n\nJune. US SCS accounted for 44% of global insured losses alone\n\nthrough Q3. To date, the US has experienced 13 billion-dollar\n\ninsured SCS loss events, of which five saw multi-billion-dollar\n\nindustry loss totals (USD2+ billion).\n\n2025 marked the third consecutive year in which US SCS insured\n\nlosses through September exceeded USD40 billion. When\n\nextending to the full calendar year, 2025 already ranks as the\n\nfourth-costliest year for insurers on record – only behind 2023\n\n(USD62 billion), 2024 (USD57 billion), and 2020 (USD47 billion).\n\nAs some of the larger 2025 outbreaks continue to be assessed\n\nand claims are processed, further loss progression is expected in\n\nthe months ahead.\n\n## US & Europe: Severe Convective Storm\n\nUS SCS YTD: Economic losses reached USD61 billion as the portion covered by insurance is now up to USD46 billion\n\nUS: Insured SCS losses already amount to the fourth-costliest year on record; only behind 2023, 2024, and 2020\n\nUS: At least 17 individual billion-dollar economic loss outbreaks; 13+ also resulted in a billion-dollar (or more) insured loss\n\nEurope: Generally benign SCS season compared to recent years; costliest stretch occurred in France during May and June\n\n# Major Event Reviews\n\n2000 2002 2004 2006 2008 2010 2012 2014 2016 2018 2020 2022 2024\n\n100\n\n200\n\n300\n\n400\n\n500\n\n600\n\n700\n\nInsured Loss (USD bn; Today's Dollars)\n\n648\n\n253\n\nUnited States Rest of World Previous year values in 2025 USD using US CPI and a construction / material / cost of labor factor adjustment. Data is thru September 2025.\n\nFigure 14: SCS insured losses from 2000 to 2025 YTD; US compared to the rest of the world Loss Data & Graphic: Gallagher Re\n\n23"],"full_content":"A persistent characteristic of 2025 SCS outbreaks has been\n\ninstances of torrential rainfall spawning flash flooding. These SCS\n\ndriven flash-flooding events were aided by multiple factors, which\n\nincluded cut-off (stuck) low pressure systems in tandem with\n\npersistent flow from the Gulf and / or training (repeating) storms\n\nthat formed near urban regions or recent burn scars. Notable\n\nevents included flash flooding in the Texas Hill Country, Ruidoso\n\n(New Mexico), New York City (New York), Chicago (Illinois), and\n\nMilwaukee (Wisconsin) among others. In many instances, low\n\nNational Flood Insurance Program (NFIP) take up rates meant\n\nmost of the flood-related damage went uninsured.\n\nUnited States\n\nSCS losses in the US continued to drive a substantial portion\n\nof global insured SCS losses during the first nine months of\n\n2025\\. Most of the losses were incurred following a highly active\n\nperiod for the peril during the peak months between March and\n\nJune. US SCS accounted for 44% of global insured losses alone\n\nthrough Q3. To date, the US has experienced 13 billion-dollar\n\ninsured SCS loss events, of which five saw multi-billion-dollar\n\nindustry loss totals (USD2+ billion).\n\n2025 marked the third consecutive year in which US SCS insured\n\nlosses through September exceeded USD40 billion. When\n\nextending to the full calendar year, 2025 already ranks as the\n\nfourth-costliest year for insurers on record – only behind 2023\n\n(USD62 billion), 2024 (USD57 billion), and 2020 (USD47 billion).\n\nAs some of the larger 2025 outbreaks continue to be assessed\n\nand claims are processed, further loss progression is expected in\n\nthe months ahead.\n\n## US & Europe: Severe Convective Storm\n\nUS SCS YTD: Economic losses reached USD61 billion as the portion covered by insurance is now up to USD46 billion\n\nUS: Insured SCS losses already amount to the fourth-costliest year on record; only behind 2023, 2024, and 2020\n\nUS: At least 17 individual billion-dollar economic loss outbreaks; 13+ also resulted in a billion-dollar (or more) insured loss\n\nEurope: Generally benign SCS season compared to recent years; costliest stretch occurred in France during May and June\n\n# Major Event Reviews\n\n2000 2002 2004 2006 2008 2010 2012 2014 2016 2018 2020 2022 2024\n\n100\n\n200\n\n300\n\n400\n\n500\n\n600\n\n700\n\nInsured Loss (USD bn; Today's Dollars)\n\n648\n\n253\n\nUnited States Rest of World Previous year values in 2025 USD using US CPI and a construction / material / cost of labor factor adjustment. Data is thru September 2025.\n\nFigure 14: SCS insured losses from 2000 to 2025 YTD; US compared to the rest of the world Loss Data & Graphic: Gallagher Re\n\n23","publish_date":null,"title":"Major Event Reviews","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/natural-catastrophe-and-climate-report-q3-2025_p23.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/natural-catastrophe-and-climate-report-q3-2025_p23.pdf"},"request":{"example_id":"chart/natural-catastrophe-and-climate-report-q3-2025_p23","product_type":"parse","source_file_path":"natural-catastrophe-and-climate-report-q3-2025_p23.pdf"},"started_at":"2026-11-04T19:45:14.134532"} +{"completed_at":"2026-11-04T19:45:21.977377","latency_in_ms":6602,"output":{"example_id":"chart/natural_catastrophe_and_climate_report_2024_h1_p23","grounded_pages":[],"layout_pages":[],"markdown":"Weather / Climate Review\n\n## El Niño Officially Ends; NOAA Declares a “La Niña Watch” as Expected Transition Looms\n\nNOAA officially declared the end of El Niño in June as ocean waters\n\nacross the central and eastern Pacific Ocean had sufficiently cooled\n\nto mark the arrival of ENSO-neutral conditions. This signaled the\n\nend of the strongest El Niño phase since 2015/16. The agency\n\nadditionally noted the issuance of a “La Niña Watch” as the quick\n\ncooling of surface waters in the Pacific’s “Niño–3.4 Region” signaled\n\nthe pendulum swing back towards La Niña. NOAA noted that there\n\nwas a 70% likelihood that La Niña conditions would arrive by the\n\nAugust–October timeframe. This would directly coincide with the\n\npeak of the Atlantic hurricane season. The agency cited that La Niña\n\nwould be anticipated to persist into the Northern Hemisphere winter\n\nmonths in 2024/25.\n\nIt should be noted that a quick transition to La Niña is common in\n\nthe immediate aftermath of a strong phase of El Niño. This quick\n\nshift also occurred following historical strong El Niño events in\n\n1982/83 and 1997/98. The arrival of La Niña should help moderate\n\nthe accelerated rate of record land and ocean warmth that was\n\nobserved during the most recent El Niño.\n\nFor background context, the El Niño-Southern Oscillation (ENSO)\n\nis a pivotal seasonal climate phenomenon that affects large-scale\n\natmospheric and ocean circulations. This in turn greatly influences\n\nglobal temperature and precipitation patterns. ENSO is a recurring\n\noscillation involving changes in the temperature of waters in the\n\ncentral and eastern equatorial Pacific Ocean and their subsequent\n\nimpacts on atmospheric conditions. Warming periods are noted as\n\nEl Niño cycles and cooling periods are known as La Niña cycles. The\n\nENSO footprint historically has the greatest influence during the\n\nNorthern Hemisphere winter months (December to March).\n\nIt is important to note that to be considered an ENSO phase, NOAA\n\nrequires five consecutive three-month running mean averages of sea\n\nsurface temperature (SST) anomalies in the Niño–3.4 region to be\n\n\\+0.5°C (El Niño) or -0.5°C (La Niña).\n\nLarge-scale atmospheric circulations associated with La Niña\n\ninclude rising air in the western Pacific, which aids in increased\n\nstorminess and rainfall. The phenomenon is likewise linked to wetter\n\nthan normal episodes in parts of northern South America and\n\nsouthern Africa. La Niña tends to enhance the odds of drier and\n\nmilder conditions across the US Southwest which raises concerns\n\nof developing drought, decreased snowpack, and strained water\n\nresources. This often has links to drier conditions in California that\n\nmay exacerbate wildfire risk. Further, the impacts of La Niña usually\n\ncontinue into the warm months for Australia. The east coast can\n\nexperience severe flooding during La Niña summers. According to\n\nthe Australia Bureau of Meteorology (BoM), 12 out of 18 La Niña\n\nevents since 1900 have resulted in floods across the country.\n\nWhile NOAA is generally considered the most globally definitive\n\nauthority on tracking ENSO, there are other regional weather and\n\nclimate centers worldwide that follow the phenomenon. These\n\nagencies often adopt different thresholds and metrics in tracking\n\nENSO. Currently, all the international agencies agree on the\n\nlikelihood of a La Niña to develop in the second half of 2024.\n\nFor example, the BoM selects its own climate models. Three of\n\nits seven surveyed models suggest the possibility of sea surface\n\ntemperatures at La Niña levels beginning in September. The\n\nPhilippine Atmospheric, Geophysical and Astronomical Services\n\nAdministration (PAGASA) said that La Niña would become most\n\nlikely in the October to December timeframe.\n\nFigure 12: Monthly view of the Oceanic Niño Index (ONI) sea surface temperature anomalies in the Niño 3.4 region | Data: NOAA | Graphic: Gallagher Re\n\nValue (ONI (1950..\n\nEl Niño Neutral La Niña\n\n1950 1955 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 2020 2025\n\n\\-3\n\n\\-2\n\n\\-1\n\n0\n\n1\n\n2\n\n3\n\n1950 1955 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 2020 2025\n\n\\-3\n\n\\-2\n\n\\-1\n\n0\n\n1\n\n2\n\n3\n\nNiño3.4 Region\n\n1950 1955 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 2020 2025\n\n\\-3.0\n\n\\-2.0\n\n\\-1.0\n\n0\\.0\n\n1\\.0\n\n2\\.0\n\n3\\.0\n\nEl Niño Neutral La Niña\n\n23","pages":[{"line_numbers":[],"markdown":"Weather / Climate Review\n\n## El Niño Officially Ends; NOAA Declares a “La Niña Watch” as Expected Transition Looms\n\nNOAA officially declared the end of El Niño in June as ocean waters\n\nacross the central and eastern Pacific Ocean had sufficiently cooled\n\nto mark the arrival of ENSO-neutral conditions. This signaled the\n\nend of the strongest El Niño phase since 2015/16. The agency\n\nadditionally noted the issuance of a “La Niña Watch” as the quick\n\ncooling of surface waters in the Pacific’s “Niño–3.4 Region” signaled\n\nthe pendulum swing back towards La Niña. NOAA noted that there\n\nwas a 70% likelihood that La Niña conditions would arrive by the\n\nAugust–October timeframe. This would directly coincide with the\n\npeak of the Atlantic hurricane season. The agency cited that La Niña\n\nwould be anticipated to persist into the Northern Hemisphere winter\n\nmonths in 2024/25.\n\nIt should be noted that a quick transition to La Niña is common in\n\nthe immediate aftermath of a strong phase of El Niño. This quick\n\nshift also occurred following historical strong El Niño events in\n\n1982/83 and 1997/98. The arrival of La Niña should help moderate\n\nthe accelerated rate of record land and ocean warmth that was\n\nobserved during the most recent El Niño.\n\nFor background context, the El Niño-Southern Oscillation (ENSO)\n\nis a pivotal seasonal climate phenomenon that affects large-scale\n\natmospheric and ocean circulations. This in turn greatly influences\n\nglobal temperature and precipitation patterns. ENSO is a recurring\n\noscillation involving changes in the temperature of waters in the\n\ncentral and eastern equatorial Pacific Ocean and their subsequent\n\nimpacts on atmospheric conditions. Warming periods are noted as\n\nEl Niño cycles and cooling periods are known as La Niña cycles. The\n\nENSO footprint historically has the greatest influence during the\n\nNorthern Hemisphere winter months (December to March).\n\nIt is important to note that to be considered an ENSO phase, NOAA\n\nrequires five consecutive three-month running mean averages of sea\n\nsurface temperature (SST) anomalies in the Niño–3.4 region to be\n\n\\+0.5°C (El Niño) or -0.5°C (La Niña).\n\nLarge-scale atmospheric circulations associated with La Niña\n\ninclude rising air in the western Pacific, which aids in increased\n\nstorminess and rainfall. The phenomenon is likewise linked to wetter\n\nthan normal episodes in parts of northern South America and\n\nsouthern Africa. La Niña tends to enhance the odds of drier and\n\nmilder conditions across the US Southwest which raises concerns\n\nof developing drought, decreased snowpack, and strained water\n\nresources. This often has links to drier conditions in California that\n\nmay exacerbate wildfire risk. Further, the impacts of La Niña usually\n\ncontinue into the warm months for Australia. The east coast can\n\nexperience severe flooding during La Niña summers. According to\n\nthe Australia Bureau of Meteorology (BoM), 12 out of 18 La Niña\n\nevents since 1900 have resulted in floods across the country.\n\nWhile NOAA is generally considered the most globally definitive\n\nauthority on tracking ENSO, there are other regional weather and\n\nclimate centers worldwide that follow the phenomenon. These\n\nagencies often adopt different thresholds and metrics in tracking\n\nENSO. Currently, all the international agencies agree on the\n\nlikelihood of a La Niña to develop in the second half of 2024.\n\nFor example, the BoM selects its own climate models. Three of\n\nits seven surveyed models suggest the possibility of sea surface\n\ntemperatures at La Niña levels beginning in September. The\n\nPhilippine Atmospheric, Geophysical and Astronomical Services\n\nAdministration (PAGASA) said that La Niña would become most\n\nlikely in the October to December timeframe.\n\nFigure 12: Monthly view of the Oceanic Niño Index (ONI) sea surface temperature anomalies in the Niño 3.4 region | Data: NOAA | Graphic: Gallagher Re\n\nValue (ONI (1950..\n\nEl Niño Neutral La Niña\n\n1950 1955 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 2020 2025\n\n\\-3\n\n\\-2\n\n\\-1\n\n0\n\n1\n\n2\n\n3\n\n1950 1955 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 2020 2025\n\n\\-3\n\n\\-2\n\n\\-1\n\n0\n\n1\n\n2\n\n3\n\nNiño3.4 Region\n\n1950 1955 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 2020 2025\n\n\\-3.0\n\n\\-2.0\n\n\\-1.0\n\n0\\.0\n\n1\\.0\n\n2\\.0\n\n3\\.0\n\nEl Niño Neutral La Niña\n\n23","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Weather / Climate Review\n\n## El Niño Officially Ends; NOAA Declares a “La Niña Watch” as Expected Transition Looms\n\nNOAA officially declared the end of El Niño in June as ocean waters\n\nacross the central and eastern Pacific Ocean had sufficiently cooled\n\nto mark the arrival of ENSO-neutral conditions. This signaled the\n\nend of the strongest El Niño phase since 2015/16. The agency\n\nadditionally noted the issuance of a “La Niña Watch” as the quick\n\ncooling of surface waters in the Pacific’s “Niño–3.4 Region” signaled\n\nthe pendulum swing back towards La Niña. NOAA noted that there\n\nwas a 70% likelihood that La Niña conditions would arrive by the\n\nAugust–October timeframe. This would directly coincide with the\n\npeak of the Atlantic hurricane season. The agency cited that La Niña\n\nwould be anticipated to persist into the Northern Hemisphere winter\n\nmonths in 2024/25.\n\nIt should be noted that a quick transition to La Niña is common in\n\nthe immediate aftermath of a strong phase of El Niño. This quick\n\nshift also occurred following historical strong El Niño events in\n\n1982/83 and 1997/98. The arrival of La Niña should help moderate\n\nthe accelerated rate of record land and ocean warmth that was\n\nobserved during the most recent El Niño.\n\nFor background context, the El Niño-Southern Oscillation (ENSO)\n\nis a pivotal seasonal climate phenomenon that affects large-scale\n\natmospheric and ocean circulations. This in turn greatly influences\n\nglobal temperature and precipitation patterns. ENSO is a recurring\n\noscillation involving changes in the temperature of waters in the\n\ncentral and eastern equatorial Pacific Ocean and their subsequent\n\nimpacts on atmospheric conditions. Warming periods are noted as\n\nEl Niño cycles and cooling periods are known as La Niña cycles. The\n\nENSO footprint historically has the greatest influence during the\n\nNorthern Hemisphere winter months (December to March).\n\nIt is important to note that to be considered an ENSO phase, NOAA\n\nrequires five consecutive three-month running mean averages of sea\n\nsurface temperature (SST) anomalies in the Niño–3.4 region to be\n\n\\+0.5°C (El Niño) or -0.5°C (La Niña).\n\nLarge-scale atmospheric circulations associated with La Niña\n\ninclude rising air in the western Pacific, which aids in increased\n\nstorminess and rainfall. The phenomenon is likewise linked to wetter\n\nthan normal episodes in parts of northern South America and\n\nsouthern Africa. La Niña tends to enhance the odds of drier and\n\nmilder conditions across the US Southwest which raises concerns\n\nof developing drought, decreased snowpack, and strained water\n\nresources. This often has links to drier conditions in California that\n\nmay exacerbate wildfire risk. Further, the impacts of La Niña usually\n\ncontinue into the warm months for Australia. The east coast can\n\nexperience severe flooding during La Niña summers. According to\n\nthe Australia Bureau of Meteorology (BoM), 12 out of 18 La Niña\n\nevents since 1900 have resulted in floods across the country.\n\nWhile NOAA is generally considered the most globally definitive\n\nauthority on tracking ENSO, there are other regional weather and\n\nclimate centers worldwide that follow the phenomenon. These\n\nagencies often adopt different thresholds and metrics in tracking\n\nENSO. Currently, all the international agencies agree on the\n\nlikelihood of a La Niña to develop in the second half of 2024.\n\nFor example, the BoM selects its own climate models. Three of\n\nits seven surveyed models suggest the possibility of sea surface\n\ntemperatures at La Niña levels beginning in September. The\n\nPhilippine Atmospheric, Geophysical and Astronomical Services\n\nAdministration (PAGASA) said that La Niña would become most\n\nlikely in the October to December timeframe.\n\nFigure 12: Monthly view of the Oceanic Niño Index (ONI) sea surface temperature anomalies in the Niño 3.4 region | Data: NOAA | Graphic: Gallagher Re\n\nValue (ONI (1950..\n\nEl Niño Neutral La Niña\n\n1950 1955 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 2020 2025\n\n\\-3\n\n\\-2\n\n\\-1\n\n0\n\n1\n\n2\n\n3\n\n1950 1955 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 2020 2025\n\n\\-3\n\n\\-2\n\n\\-1\n\n0\n\n1\n\n2\n\n3\n\nNiño3.4 Region\n\n1950 1955 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 2020 2025\n\n\\-3.0\n\n\\-2.0\n\n\\-1.0\n\n0\\.0\n\n1\\.0\n\n2\\.0\n\n3\\.0\n\nEl Niño Neutral La Niña\n\n23","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/natural_catastrophe_and_climate_report_2024_h1_p23.pdf"]},"response":{"errors":[],"extract_id":"extract_e7506cf1d5b060237d50fb1ee70e10a2","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Weather / Climate Review\n\n## El Niño Officially Ends; NOAA Declares a “La Niña Watch” as Expected Transition Looms\n\nNOAA officially declared the end of El Niño in June as ocean waters\n\nacross the central and eastern Pacific Ocean had sufficiently cooled\n\nto mark the arrival of ENSO-neutral conditions. This signaled the\n\nend of the strongest El Niño phase since 2015/16. The agency\n\nadditionally noted the issuance of a “La Niña Watch” as the quick\n\ncooling of surface waters in the Pacific’s “Niño–3.4 Region” signaled\n\nthe pendulum swing back towards La Niña. NOAA noted that there\n\nwas a 70% likelihood that La Niña conditions would arrive by the\n\nAugust–October timeframe. This would directly coincide with the\n\npeak of the Atlantic hurricane season. The agency cited that La Niña\n\nwould be anticipated to persist into the Northern Hemisphere winter\n\nmonths in 2024/25.\n\nIt should be noted that a quick transition to La Niña is common in\n\nthe immediate aftermath of a strong phase of El Niño. This quick\n\nshift also occurred following historical strong El Niño events in\n\n1982/83 and 1997/98. The arrival of La Niña should help moderate\n\nthe accelerated rate of record land and ocean warmth that was\n\nobserved during the most recent El Niño.\n\nFor background context, the El Niño-Southern Oscillation (ENSO)\n\nis a pivotal seasonal climate phenomenon that affects large-scale\n\natmospheric and ocean circulations. This in turn greatly influences\n\nglobal temperature and precipitation patterns. ENSO is a recurring\n\noscillation involving changes in the temperature of waters in the\n\ncentral and eastern equatorial Pacific Ocean and their subsequent\n\nimpacts on atmospheric conditions. Warming periods are noted as\n\nEl Niño cycles and cooling periods are known as La Niña cycles. The\n\nENSO footprint historically has the greatest influence during the\n\nNorthern Hemisphere winter months (December to March).\n\nIt is important to note that to be considered an ENSO phase, NOAA\n\nrequires five consecutive three-month running mean averages of sea\n\nsurface temperature (SST) anomalies in the Niño–3.4 region to be\n\n\\+0.5°C (El Niño) or -0.5°C (La Niña).\n\nLarge-scale atmospheric circulations associated with La Niña\n\ninclude rising air in the western Pacific, which aids in increased\n\nstorminess and rainfall. The phenomenon is likewise linked to wetter\n\nthan normal episodes in parts of northern South America and\n\nsouthern Africa. La Niña tends to enhance the odds of drier and\n\nmilder conditions across the US Southwest which raises concerns\n\nof developing drought, decreased snowpack, and strained water\n\nresources. This often has links to drier conditions in California that\n\nmay exacerbate wildfire risk. Further, the impacts of La Niña usually\n\ncontinue into the warm months for Australia. The east coast can\n\nexperience severe flooding during La Niña summers. According to\n\nthe Australia Bureau of Meteorology (BoM), 12 out of 18 La Niña\n\nevents since 1900 have resulted in floods across the country.\n\nWhile NOAA is generally considered the most globally definitive\n\nauthority on tracking ENSO, there are other regional weather and\n\nclimate centers worldwide that follow the phenomenon. These\n\nagencies often adopt different thresholds and metrics in tracking\n\nENSO. Currently, all the international agencies agree on the\n\nlikelihood of a La Niña to develop in the second half of 2024.\n\nFor example, the BoM selects its own climate models. Three of\n\nits seven surveyed models suggest the possibility of sea surface\n\ntemperatures at La Niña levels beginning in September. The\n\nPhilippine Atmospheric, Geophysical and Astronomical Services\n\nAdministration (PAGASA) said that La Niña would become most\n\nlikely in the October to December timeframe.\n\nFigure 12: Monthly view of the Oceanic Niño Index (ONI) sea surface temperature anomalies in the Niño 3.4 region | Data: NOAA | Graphic: Gallagher Re\n\nValue (ONI (1950..\n\nEl Niño Neutral La Niña\n\n1950 1955 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 2020 2025\n\n\\-3\n\n\\-2\n\n\\-1\n\n0\n\n1\n\n2\n\n3\n\n1950 1955 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 2020 2025\n\n\\-3\n\n\\-2\n\n\\-1\n\n0\n\n1\n\n2\n\n3\n\nNiño3.4 Region\n\n1950 1955 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 2020 2025\n\n\\-3.0\n\n\\-2.0\n\n\\-1.0\n\n0\\.0\n\n1\\.0\n\n2\\.0\n\n3\\.0\n\nEl Niño Neutral La Niña\n\n23"],"full_content":"Weather / Climate Review\n\n## El Niño Officially Ends; NOAA Declares a “La Niña Watch” as Expected Transition Looms\n\nNOAA officially declared the end of El Niño in June as ocean waters\n\nacross the central and eastern Pacific Ocean had sufficiently cooled\n\nto mark the arrival of ENSO-neutral conditions. This signaled the\n\nend of the strongest El Niño phase since 2015/16. The agency\n\nadditionally noted the issuance of a “La Niña Watch” as the quick\n\ncooling of surface waters in the Pacific’s “Niño–3.4 Region” signaled\n\nthe pendulum swing back towards La Niña. NOAA noted that there\n\nwas a 70% likelihood that La Niña conditions would arrive by the\n\nAugust–October timeframe. This would directly coincide with the\n\npeak of the Atlantic hurricane season. The agency cited that La Niña\n\nwould be anticipated to persist into the Northern Hemisphere winter\n\nmonths in 2024/25.\n\nIt should be noted that a quick transition to La Niña is common in\n\nthe immediate aftermath of a strong phase of El Niño. This quick\n\nshift also occurred following historical strong El Niño events in\n\n1982/83 and 1997/98. The arrival of La Niña should help moderate\n\nthe accelerated rate of record land and ocean warmth that was\n\nobserved during the most recent El Niño.\n\nFor background context, the El Niño-Southern Oscillation (ENSO)\n\nis a pivotal seasonal climate phenomenon that affects large-scale\n\natmospheric and ocean circulations. This in turn greatly influences\n\nglobal temperature and precipitation patterns. ENSO is a recurring\n\noscillation involving changes in the temperature of waters in the\n\ncentral and eastern equatorial Pacific Ocean and their subsequent\n\nimpacts on atmospheric conditions. Warming periods are noted as\n\nEl Niño cycles and cooling periods are known as La Niña cycles. The\n\nENSO footprint historically has the greatest influence during the\n\nNorthern Hemisphere winter months (December to March).\n\nIt is important to note that to be considered an ENSO phase, NOAA\n\nrequires five consecutive three-month running mean averages of sea\n\nsurface temperature (SST) anomalies in the Niño–3.4 region to be\n\n\\+0.5°C (El Niño) or -0.5°C (La Niña).\n\nLarge-scale atmospheric circulations associated with La Niña\n\ninclude rising air in the western Pacific, which aids in increased\n\nstorminess and rainfall. The phenomenon is likewise linked to wetter\n\nthan normal episodes in parts of northern South America and\n\nsouthern Africa. La Niña tends to enhance the odds of drier and\n\nmilder conditions across the US Southwest which raises concerns\n\nof developing drought, decreased snowpack, and strained water\n\nresources. This often has links to drier conditions in California that\n\nmay exacerbate wildfire risk. Further, the impacts of La Niña usually\n\ncontinue into the warm months for Australia. The east coast can\n\nexperience severe flooding during La Niña summers. According to\n\nthe Australia Bureau of Meteorology (BoM), 12 out of 18 La Niña\n\nevents since 1900 have resulted in floods across the country.\n\nWhile NOAA is generally considered the most globally definitive\n\nauthority on tracking ENSO, there are other regional weather and\n\nclimate centers worldwide that follow the phenomenon. These\n\nagencies often adopt different thresholds and metrics in tracking\n\nENSO. Currently, all the international agencies agree on the\n\nlikelihood of a La Niña to develop in the second half of 2024.\n\nFor example, the BoM selects its own climate models. Three of\n\nits seven surveyed models suggest the possibility of sea surface\n\ntemperatures at La Niña levels beginning in September. The\n\nPhilippine Atmospheric, Geophysical and Astronomical Services\n\nAdministration (PAGASA) said that La Niña would become most\n\nlikely in the October to December timeframe.\n\nFigure 12: Monthly view of the Oceanic Niño Index (ONI) sea surface temperature anomalies in the Niño 3.4 region | Data: NOAA | Graphic: Gallagher Re\n\nValue (ONI (1950..\n\nEl Niño Neutral La Niña\n\n1950 1955 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 2020 2025\n\n\\-3\n\n\\-2\n\n\\-1\n\n0\n\n1\n\n2\n\n3\n\n1950 1955 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 2020 2025\n\n\\-3\n\n\\-2\n\n\\-1\n\n0\n\n1\n\n2\n\n3\n\nNiño3.4 Region\n\n1950 1955 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 2020 2025\n\n\\-3.0\n\n\\-2.0\n\n\\-1.0\n\n0\\.0\n\n1\\.0\n\n2\\.0\n\n3\\.0\n\nEl Niño Neutral La Niña\n\n23","publish_date":null,"title":"El Niño Officially Ends; NOAA Declares a “La Niña Watch” as Expected Transition Looms","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/natural_catastrophe_and_climate_report_2024_h1_p23.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/natural_catastrophe_and_climate_report_2024_h1_p23.pdf"},"request":{"example_id":"chart/natural_catastrophe_and_climate_report_2024_h1_p23","product_type":"parse","source_file_path":"natural_catastrophe_and_climate_report_2024_h1_p23.pdf"},"started_at":"2026-11-04T19:45:15.375214"} +{"completed_at":"2026-11-04T19:45:18.847577","latency_in_ms":2406,"output":{"example_id":"chart/pdf_61c93ea643bc_p10","grounded_pages":[],"layout_pages":[],"markdown":"## U.S. Liability Management Transactions: Quarterly Update Through Q4 2025\n\nCreditSights.com 9\n\n_Figure 2_\n\n### **III. Litigation Updates**\n\nA summary of the most notable ongoing court cases relating to liability management through Q4 2025 (and into Q1 2026) follows: _STG Logistics (NY Sup. Ct, County of New York; Bankr. D.N.J.)_ : STG Logistics’ fall 2024 LMT ran the gamut of LMT structures, and included drop-down, tiered non-pro rata exchanges, and double dip components. Despite closing with roughly 93% participation from existing creditors, two holdouts, Audax and Siemens Financial, sued the company, participating lenders, and Antares (as agent) alleging a variety of breach of contract claims, as well as claiming a violation of New York’s fraudulent transfer law (the Uniform Voidable Transactions Act), and a breach of the implied covenant of good faith and fair dealing. Defendants filed motions to dismiss in spring 2025 (including on the basis that the plaintiff lack standing due to the credit agreement’s no action clause) and oral arguments were held before Judge Anar Patel of the Supreme Court of New York in August 2025. In a victory for plaintiffs, Judge Patel ruled on January 5, 2026, that most of the plaintiffs’ claims would survive the motion to dismiss (with the notable exception of a fraudulent transfer claim). As noted above, the company filed for bankruptcy protection on January 12, 2025 in the District of New Jersey with a $150 mn DIP in place and a prearranged plan to shed 91% of its funded debt load. See also Five Notable Arguments Worth Paying Attention to in STG Logistics’ LMT Litigation and STG Logistics – Plaintiffs Survive the Motion to Dismiss Largely Unscathed.\n\n2\\.50\n\n2\\.75\n\n3\\.00\n\n3\\.25\n\n3\\.50\n\n3\\.75\n\n4\\.00\n\n4\\.25\n\n4\\.50\n\n1Q18 2Q18 3Q18 4Q18 1Q19 2Q19 3Q19 4Q19 1Q20 2Q20 3Q20 4Q20 1Q21 2Q21 3Q21 4Q21 1Q22 2Q22 3Q22 4Q22 1Q23 2Q23 3Q23 4Q23 1Q24 2Q24 3Q24 4Q24 1Q25 2Q25 3Q25 4Q25\n\n**PE-Backed** **│** **All Loans**\n\nSource: Covenant Review, a Fitch Solutions Company\n\nMore Protective\n\n**Average Covenant Review Documentation Score of Institutional New-Issue Loans by Quarter**","pages":[{"line_numbers":[],"markdown":"## U.S. Liability Management Transactions: Quarterly Update Through Q4 2025\n\nCreditSights.com 9\n\n_Figure 2_\n\n### **III. Litigation Updates**\n\nA summary of the most notable ongoing court cases relating to liability management through Q4 2025 (and into Q1 2026) follows: _STG Logistics (NY Sup. Ct, County of New York; Bankr. D.N.J.)_ : STG Logistics’ fall 2024 LMT ran the gamut of LMT structures, and included drop-down, tiered non-pro rata exchanges, and double dip components. Despite closing with roughly 93% participation from existing creditors, two holdouts, Audax and Siemens Financial, sued the company, participating lenders, and Antares (as agent) alleging a variety of breach of contract claims, as well as claiming a violation of New York’s fraudulent transfer law (the Uniform Voidable Transactions Act), and a breach of the implied covenant of good faith and fair dealing. Defendants filed motions to dismiss in spring 2025 (including on the basis that the plaintiff lack standing due to the credit agreement’s no action clause) and oral arguments were held before Judge Anar Patel of the Supreme Court of New York in August 2025. In a victory for plaintiffs, Judge Patel ruled on January 5, 2026, that most of the plaintiffs’ claims would survive the motion to dismiss (with the notable exception of a fraudulent transfer claim). As noted above, the company filed for bankruptcy protection on January 12, 2025 in the District of New Jersey with a $150 mn DIP in place and a prearranged plan to shed 91% of its funded debt load. See also Five Notable Arguments Worth Paying Attention to in STG Logistics’ LMT Litigation and STG Logistics – Plaintiffs Survive the Motion to Dismiss Largely Unscathed.\n\n2\\.50\n\n2\\.75\n\n3\\.00\n\n3\\.25\n\n3\\.50\n\n3\\.75\n\n4\\.00\n\n4\\.25\n\n4\\.50\n\n1Q18 2Q18 3Q18 4Q18 1Q19 2Q19 3Q19 4Q19 1Q20 2Q20 3Q20 4Q20 1Q21 2Q21 3Q21 4Q21 1Q22 2Q22 3Q22 4Q22 1Q23 2Q23 3Q23 4Q23 1Q24 2Q24 3Q24 4Q24 1Q25 2Q25 3Q25 4Q25\n\n**PE-Backed** **│** **All Loans**\n\nSource: Covenant Review, a Fitch Solutions Company\n\nMore Protective\n\n**Average Covenant Review Documentation Score of Institutional New-Issue Loans by Quarter**","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"## U.S. Liability Management Transactions: Quarterly Update Through Q4 2025\n\nCreditSights.com 9\n\n_Figure 2_\n\n### **III. Litigation Updates**\n\nA summary of the most notable ongoing court cases relating to liability management through Q4 2025 (and into Q1 2026) follows: _STG Logistics (NY Sup. Ct, County of New York; Bankr. D.N.J.)_ : STG Logistics’ fall 2024 LMT ran the gamut of LMT structures, and included drop-down, tiered non-pro rata exchanges, and double dip components. Despite closing with roughly 93% participation from existing creditors, two holdouts, Audax and Siemens Financial, sued the company, participating lenders, and Antares (as agent) alleging a variety of breach of contract claims, as well as claiming a violation of New York’s fraudulent transfer law (the Uniform Voidable Transactions Act), and a breach of the implied covenant of good faith and fair dealing. Defendants filed motions to dismiss in spring 2025 (including on the basis that the plaintiff lack standing due to the credit agreement’s no action clause) and oral arguments were held before Judge Anar Patel of the Supreme Court of New York in August 2025. In a victory for plaintiffs, Judge Patel ruled on January 5, 2026, that most of the plaintiffs’ claims would survive the motion to dismiss (with the notable exception of a fraudulent transfer claim). As noted above, the company filed for bankruptcy protection on January 12, 2025 in the District of New Jersey with a $150 mn DIP in place and a prearranged plan to shed 91% of its funded debt load. See also Five Notable Arguments Worth Paying Attention to in STG Logistics’ LMT Litigation and STG Logistics – Plaintiffs Survive the Motion to Dismiss Largely Unscathed.\n\n2\\.50\n\n2\\.75\n\n3\\.00\n\n3\\.25\n\n3\\.50\n\n3\\.75\n\n4\\.00\n\n4\\.25\n\n4\\.50\n\n1Q18 2Q18 3Q18 4Q18 1Q19 2Q19 3Q19 4Q19 1Q20 2Q20 3Q20 4Q20 1Q21 2Q21 3Q21 4Q21 1Q22 2Q22 3Q22 4Q22 1Q23 2Q23 3Q23 4Q23 1Q24 2Q24 3Q24 4Q24 1Q25 2Q25 3Q25 4Q25\n\n**PE-Backed** **│** **All Loans**\n\nSource: Covenant Review, a Fitch Solutions Company\n\nMore Protective\n\n**Average Covenant Review Documentation Score of Institutional New-Issue Loans by Quarter**","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/pdf_61c93ea643bc_p10.pdf"]},"response":{"errors":[],"extract_id":"extract_d04228c13fc1a8368ef732e9bd1c17b3","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["## U.S. Liability Management Transactions: Quarterly Update Through Q4 2025\n\nCreditSights.com 9\n\n_Figure 2_\n\n### **III. Litigation Updates**\n\nA summary of the most notable ongoing court cases relating to liability management through Q4 2025 (and into Q1 2026) follows: _STG Logistics (NY Sup. Ct, County of New York; Bankr. D.N.J.)_ : STG Logistics’ fall 2024 LMT ran the gamut of LMT structures, and included drop-down, tiered non-pro rata exchanges, and double dip components. Despite closing with roughly 93% participation from existing creditors, two holdouts, Audax and Siemens Financial, sued the company, participating lenders, and Antares (as agent) alleging a variety of breach of contract claims, as well as claiming a violation of New York’s fraudulent transfer law (the Uniform Voidable Transactions Act), and a breach of the implied covenant of good faith and fair dealing. Defendants filed motions to dismiss in spring 2025 (including on the basis that the plaintiff lack standing due to the credit agreement’s no action clause) and oral arguments were held before Judge Anar Patel of the Supreme Court of New York in August 2025. In a victory for plaintiffs, Judge Patel ruled on January 5, 2026, that most of the plaintiffs’ claims would survive the motion to dismiss (with the notable exception of a fraudulent transfer claim). As noted above, the company filed for bankruptcy protection on January 12, 2025 in the District of New Jersey with a $150 mn DIP in place and a prearranged plan to shed 91% of its funded debt load. See also Five Notable Arguments Worth Paying Attention to in STG Logistics’ LMT Litigation and STG Logistics – Plaintiffs Survive the Motion to Dismiss Largely Unscathed.\n\n2\\.50\n\n2\\.75\n\n3\\.00\n\n3\\.25\n\n3\\.50\n\n3\\.75\n\n4\\.00\n\n4\\.25\n\n4\\.50\n\n1Q18 2Q18 3Q18 4Q18 1Q19 2Q19 3Q19 4Q19 1Q20 2Q20 3Q20 4Q20 1Q21 2Q21 3Q21 4Q21 1Q22 2Q22 3Q22 4Q22 1Q23 2Q23 3Q23 4Q23 1Q24 2Q24 3Q24 4Q24 1Q25 2Q25 3Q25 4Q25\n\n**PE-Backed** **│** **All Loans**\n\nSource: Covenant Review, a Fitch Solutions Company\n\nMore Protective\n\n**Average Covenant Review Documentation Score of Institutional New-Issue Loans by Quarter**"],"full_content":"## U.S. Liability Management Transactions: Quarterly Update Through Q4 2025\n\nCreditSights.com 9\n\n_Figure 2_\n\n### **III. Litigation Updates**\n\nA summary of the most notable ongoing court cases relating to liability management through Q4 2025 (and into Q1 2026) follows: _STG Logistics (NY Sup. Ct, County of New York; Bankr. D.N.J.)_ : STG Logistics’ fall 2024 LMT ran the gamut of LMT structures, and included drop-down, tiered non-pro rata exchanges, and double dip components. Despite closing with roughly 93% participation from existing creditors, two holdouts, Audax and Siemens Financial, sued the company, participating lenders, and Antares (as agent) alleging a variety of breach of contract claims, as well as claiming a violation of New York’s fraudulent transfer law (the Uniform Voidable Transactions Act), and a breach of the implied covenant of good faith and fair dealing. Defendants filed motions to dismiss in spring 2025 (including on the basis that the plaintiff lack standing due to the credit agreement’s no action clause) and oral arguments were held before Judge Anar Patel of the Supreme Court of New York in August 2025. In a victory for plaintiffs, Judge Patel ruled on January 5, 2026, that most of the plaintiffs’ claims would survive the motion to dismiss (with the notable exception of a fraudulent transfer claim). As noted above, the company filed for bankruptcy protection on January 12, 2025 in the District of New Jersey with a $150 mn DIP in place and a prearranged plan to shed 91% of its funded debt load. See also Five Notable Arguments Worth Paying Attention to in STG Logistics’ LMT Litigation and STG Logistics – Plaintiffs Survive the Motion to Dismiss Largely Unscathed.\n\n2\\.50\n\n2\\.75\n\n3\\.00\n\n3\\.25\n\n3\\.50\n\n3\\.75\n\n4\\.00\n\n4\\.25\n\n4\\.50\n\n1Q18 2Q18 3Q18 4Q18 1Q19 2Q19 3Q19 4Q19 1Q20 2Q20 3Q20 4Q20 1Q21 2Q21 3Q21 4Q21 1Q22 2Q22 3Q22 4Q22 1Q23 2Q23 3Q23 4Q23 1Q24 2Q24 3Q24 4Q24 1Q25 2Q25 3Q25 4Q25\n\n**PE-Backed** **│** **All Loans**\n\nSource: Covenant Review, a Fitch Solutions Company\n\nMore Protective\n\n**Average Covenant Review Documentation Score of Institutional New-Issue Loans by Quarter**","publish_date":null,"title":"U.S. Liability Management Transactions: Quarterly Update Through Q4 2025","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/pdf_61c93ea643bc_p10.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/pdf_61c93ea643bc_p10.pdf"},"request":{"example_id":"chart/pdf_61c93ea643bc_p10","product_type":"parse","source_file_path":"pdf_61c93ea643bc_p10.pdf"},"started_at":"2026-11-04T19:45:16.440752"} +{"completed_at":"2026-11-04T19:45:19.844398","latency_in_ms":1535,"output":{"example_id":"chart/pdf_bae2c1684672_p26","grounded_pages":[],"layout_pages":[],"markdown":"B10 | Monday, February 2, 2026 **THE WALL STREET JOURNAL.**\n\n# **Luxury Comeback in China Isn’t Likely**\n\n### Even with a stock market outperforming the S&P 500, Chinese shoppers aren’t flocking to chic brands as they once did\n\n# HEARD\n\n## ON\n\n# THESTREET\n\nFINANCIAL ANALYSIS & COMMENTARY\n\nThe Louis Vuitton store in Shanghai is shaped like a cruise ship.\n\nRAUL ARIANO/BLOOMBERG NEWS\n\nChinese shoppers are returning to luxury stores, but with less ap- petite to spend and in greatly di- minished numbers. That is disap- pointing for high-end brands desperate for fresh growth. After five years of weak sales, some luxury brands said their China business recently turned a corner. **Richemont** , which owns Cartier and Van Cleef & Arpels jewelry, said sales started to grow again in China in the third quarter of 2025. British luxury trench coat maker **Burberry** also noticed an improvement last fall, and said de- mand accelerated further in the fi- nal three months of the year. The recent results of the world’s biggest luxury goods com- pany, **LVMH** , gave a more mixed picture. Sales to Chinese consum- ers are stabilizing for LVMH, but aren’t back to growth yet. The company is tweaking its footprint in the country. LVMH recently agreed to sell its duty-free business in China, but is investing, too. The company’s flag- ship brand, Louis Vuitton, in June opened a Shanghai boutique, which is drawing thousands of vis- itors a day. If the global luxury industry is to get back to the 6% annual growth it averaged over the past two decades, it needs China to re- cover. Luxury stocks used to be a favorite way for investors to get exposure to the country’s then- booming middle class. The Chinese went from barely being on luxury brands’ radar in the early 2000s to generating 35% of the industry’s global sales by 2019, Bernstein estimates. The ap- petite for luxury goods was so strong that Chinese tourists flew to European capitals for the expe- rience of buying—for example—a Chanel or Hermès purse at a flag- ship in Paris. The pandemic and China’s im- ploding housing market ended the good times. The Chinese share of\n\nglobal luxury spending dropped to 23% last year. They have been overtaken by Americans, who are now the industry’s most important customers. Last year, U.S. shop- pers generated 31% of global lux- ury sales, up from 22% before the pandemic. This leaves luxury brands heav- ily reliant on a clutch of wealthy Americans. The S&P 500 has gained 17% over the past 12 months, generating an additional $8.4 trillion of stock-market wealth. Most of the benefits have gone to the richest households, who have higher-than-average expo- sure to stocks. This is boosting the very top of the luxury market, for goods such as fine jewelry, or seats in first-class air cabins. Brands such as Gucci or Saint Lau- rent that are more dependent on aspirational shoppers in the U.S., defined as customers who spend up to $2,500 a year on luxury goods, are suffering. A recovery in China would pro- vide a welcome hedge. The Shang- hai Stock Exchange Composite In- dex has outperformed the S&P 500 since the start of 2025. But China’s equity markets are volatile and a less important driver of consumer spending than in the U.S. Stocks make up just a 10th of China’s household wealth according to Goldman Sachs. This compares to 35% in the U.S., data from the Fed- eral Reserve shows. Home values, which make up 60% of household wealth in China, are much more important for con- sumer spending. A property crash has wiped two-fifths off the value of Chinese real estate since 2021 in some parts of the country. Prices for existing homes fell 6% in 2025, and don’t appear to have bottomed out yet. However, prices are stabilizing in Tier 1 cities in China such as Shanghai and Shenzhen, where wealthy residents are also more\n\nlikely to own stocks. This could create pockets of demand for lux- ury goods this year. But a widespread recovery is a long shot. When middle-income Chinese consumers were splurging on designer goods several years ago, their homes were appreciat- ing in value. In the decade leading up to the pandemic, home prices in China’s top 70 cities rose more than 50%, data from the National Bureau of Statistics of China shows. Incomes were rising 10% a year on average over the period, more than double today’s rate. Chinese households are still prodigious savers. The average household is socking away one- third of its income, compared with less than 5% in the U.S. This will limit how much middle-class con- sumers are willing to spend on luxury goods. That said, brands perceived to offer value for the money are growing. **Ralph Lauren** said its sales in China rose more than 30% in its latest quarter. Coach’s Chi- nese business was up 21%. Bur- berry is having success with Gen Z consumers, thanks in part to a re- alistic pricing strategy. Local luxury players such as Songmont handbags are also win- ning over price-sensitive Chinese shoppers. Their goods sell for four to five times their manufacturing cost, compared with a markup of 10 times or higher for Western luxury brands, Bernstein notes. Richemont Chief Executive Ni- colas Bos said on the company’s latest investor call that the bar for selling luxury goods in China is high today. Chinese customers are “becoming much more demanding, discerning and differentiating when it comes to their choice of brands.” The days when shoppers queued around the block for a Chanel or Louis Vuitton handbag aren’t coming back soon. _—Carol Ryan_\n\n\\*Total exceeds 100 because of rounding.\n\nFeb. 2025 Jan. ’26 –20\n\n–10\n\n0\n\n10\n\n20\n\n30%\n\nShanghai Composite\n\nS&P 500\n\nIndex performance during the past year\n\nNew homes\n\nSecond-hand homes\n\nChina’s house prices, change from a year earlier\n\n**2025**\n\n2019\n\n0% 25 50 75 100\n\n**Chinese American** European\n\nJapanese\n\nOther Asian\n\nOther Global luxury spending by nationality\\*\n\nSources: Bernstein (spending); FactSet (stock market); National Bureau of Statistics of China, Bernstein analysis (house prices)\n\nJan. 2025 Dec. –10\n\n–8\n\n–6\n\n–4\n\n–2\n\n0%\n\n# Cruise Investors Got a Cabin Upgrade\n\nCruise fans love wave season, but this year’s prime booking win- dow was making investors queasy. They were concerned that de- mand would soften amid higher prices and increased capacity. In the six months ended last Wednes- day, shares of all three major oceangoing cruise lines were down, trailing the S&P 500 by 19% on av- erage. Weakest of all over that span was **Royal Caribbean** . Then it reported full-year results on Thursday. The stock surged 17% after management said Royal Ca- ribbean had just experienced the seven highest weeks for bookings in the company’s history. Five of those weeks fell during wave season, which continues through March. “What they heard from us today\n\nis that our guests continue to pri- oritize travel with us and that they’re willing to pay more,” said Royal Caribbean CEO Jason Liberty in an interview. Shares of competitors **Carnival**\n\nand **Norwegian Cruise Line** also jumped. The best-performing cruise stock recently had been **Viking** , which specializes in more intimate and sedate river cruises. That wasn’t so much because of where its vessels sail as who you’re likely to find on board. Michael Er- stad, senior consumer analyst at M Science, points out that their older, wealthier clients were seen as more financially secure. That cohort has kept spending on other products and services even as younger peo- ple with jobs and children have tightened their belts. But cruise lines keep confound- ing expectations. Indeed, after the industry’s near-death experience during the pandemic, they’ve be- come more popular than ever. Ship- yards that specialize in the vessels\n\nare booked up for years. Between the three big oceango- ing companies, combined passenger ticket sales are seen being almost 50% higher this fiscal year than in 2019. Earnings per share are still much lower for Carnival and Nor- wegian, though, because of finan- cial ballast they picked up to sur- vive the industry’s shutdown—debt and stock issued to just stay in business. Royal Caribbean navigated the crisis more skillfully. Now it is re- turning billions to shareholders with cash left over to order new ships. That includes expanding in an interesting direction: river cruis- ing. On Thursday, the company an- nounced the addition of 10 addi- tional vessels to its Celebrity River Cruises subsidiary.\n\nLiberty doesn’t view that as an economic hedge. He said that based on the copious data the company has on its many return customers, river cruises interest many of them. Ocean vacations are already often multigenerational trips, he says. Why has cruise demand been so buoyant, though? One explanation might be price. Even though they’ve become more expensive, land-based vacations have seen even steeper inflation. So how do they do it? This isn’t the Love Boat. Ships are massive entertainment complexes and in- creasingly sail to private beaches or islands owned by the cruise lines themselves. That all creates greater economies of scale. Cruisers seem to like it, too. _—Spencer Jakab_\n\nforced crypto investors to re- think their riskier holdings. Some traders are hoping for relief ahead. Kevin Warsh, President Trump’s Federal Reserve chair nominee, has made supportive comments about bitcoin in the past, but some investors fear he could support tighter monetary pol- icy and hurt riskier assets. Wall Street’s stomach for those riskier bets could be\n\ntested this week, when Strat- egy reports its earnings. Big tech names like **Palantir** , **Am- azon.com** and chip maker **Ad- vanced Micro Devices** are also slated to report their fi- nancial results, and a jobs re- port from the Labor Depart- ment is due this coming Friday. Investors have yanked a net $227 million out of spot bitcoin exchange-traded\n\n## **MARKETS**\n\nunderwater. “It’s a mess out there,” said Gennaro Salemme, a 32- year-old car salesman in Chi- cago. “The vibe right now is ‘stay alive.’” Salemme still keeps roughly 40% of his $200,000 portfolio in crypto invest- ments including ether, XRP and bitcoin ETFs. He first bought bitcoin nine years ago when the price was $3,000, and said he is “100%” still bullish on it for the long term. But holding on can be excruciating when everyone else is abandoning ship. “Everyone wants to leave when things go south,” Sa- lemme said. “A lot of this is emotionally driven.” After rallying in early 2025, digital assets surren- dered much of the spotlight to the artificial-intelligence race, which sucked up inves- tors’ capital and powered a third straight year of double- digit gains in stocks. A weak- ening job market and escalat- ing geopolitical tensions also\n\nfunds through Jan. 28, ac- cording to Morningstar Direct data. The bleak mood in crypto- land is in contrast to the start of last year, when investors cheered what many believed would be a new golden era for digital assets. Major Wall Street firms had already launched spot bitcoin ETFs, pushing digital tokens into the mainstream. And Donald Trump was back in the White House, promising to turn America into a “bitcoin su- perpower.” Then last summer, Trump signed into law a measure to regulate stablecoins, fueling expectations that clearer rules could help boost the crypto industry. “It was almost like it was 2021 all over again,” said Lewis Carr, a 19-year-old col- lege student in Texas who sold gaming consoles on Facebook Marketplace to fund his crypto investments. “Like you couldn’t lose money.” But the celebratory mood\n\neventually gave way to a hangover. Carr keeps about one-quar- ter of his roughly $6,000 portfolio in crypto and still puts about $50 a week into bitcoin, ether and XRP. But the slide has tempered his optimism. “Crypto right now is a to- tal gamble,” he said. “I think everybody realizes that.” Will Warden had similarly high expectations not so long ago. The commercial real-es- tate broker, 44, had watched crypto from the sidelines for years, then jumped in after Trump took office last Janu- ary. “My expectations were that by late spring, summer of 2025, I would have had a huge return,” he said. “But that didn’t happen.” Warden has decided to ride out the downturn, but his fa- ther and brother have cashed out of their crypto holdings. “A lot of people are ex- hausted, or irritated, or an- gry,” he said.\n\nThe party in the crypto market is starting to feel like a distant memory. After a disappointing end to 2025, major cryptocurren- cies have languished to start the new year. Bitcoin shed roughly one-third of its value since hitting a record in Octo- ber through Friday, and sank further on Saturday. The slump has continued even though stocks are near records, with the gains broad- ening throughout the market and investors betting the economy will keep growing. A weakened dollar pushed in- vestors toward alternative as- sets for safety in the so-called debasement trade. Yet prices for cryptocur- rencies have slid. On Saturday, bitcoin briefly dipped below the average price Michael Saylor’s **Strat- egy** has paid for the token, putting the crypto-treasury company’s bitcoin holdings\n\nBY DALVIN BROWN AND HANNAH ERIN LANG\n\n# The Vibe in the Crypto Market Right Now: ‘Stay Alive’\n\nS&P 500\n\nBitcoin\n\nAug. 2025 Sept. Oct. Nov. Dec. Jan. ’26\n\nCryptocurrency and index performance, past six months\n\nSources: FactSet (index); Kraken (ether); CoinDesk (bitcoin)\n\n–30\n\n–20\n\n–10\n\n0\n\n10\n\n20\n\n30%\n\nEther\n\nViking 8.8%\n\n–13.5\n\n–21.4\n\n–22.4\n\nSix-month total return through Wednesday (prior to Royal Caribbean results)\n\nSource: FactSet\n\nRoyal Caribbean\n\nNorwegian\n\nCarnival\n\nP2JW033000-0-B01000-1--------XA","pages":[{"line_numbers":[],"markdown":"B10 | Monday, February 2, 2026 **THE WALL STREET JOURNAL.**\n\n# **Luxury Comeback in China Isn’t Likely**\n\n### Even with a stock market outperforming the S&P 500, Chinese shoppers aren’t flocking to chic brands as they once did\n\n# HEARD\n\n## ON\n\n# THESTREET\n\nFINANCIAL ANALYSIS & COMMENTARY\n\nThe Louis Vuitton store in Shanghai is shaped like a cruise ship.\n\nRAUL ARIANO/BLOOMBERG NEWS\n\nChinese shoppers are returning to luxury stores, but with less ap- petite to spend and in greatly di- minished numbers. That is disap- pointing for high-end brands desperate for fresh growth. After five years of weak sales, some luxury brands said their China business recently turned a corner. **Richemont** , which owns Cartier and Van Cleef & Arpels jewelry, said sales started to grow again in China in the third quarter of 2025. British luxury trench coat maker **Burberry** also noticed an improvement last fall, and said de- mand accelerated further in the fi- nal three months of the year. The recent results of the world’s biggest luxury goods com- pany, **LVMH** , gave a more mixed picture. Sales to Chinese consum- ers are stabilizing for LVMH, but aren’t back to growth yet. The company is tweaking its footprint in the country. LVMH recently agreed to sell its duty-free business in China, but is investing, too. The company’s flag- ship brand, Louis Vuitton, in June opened a Shanghai boutique, which is drawing thousands of vis- itors a day. If the global luxury industry is to get back to the 6% annual growth it averaged over the past two decades, it needs China to re- cover. Luxury stocks used to be a favorite way for investors to get exposure to the country’s then- booming middle class. The Chinese went from barely being on luxury brands’ radar in the early 2000s to generating 35% of the industry’s global sales by 2019, Bernstein estimates. The ap- petite for luxury goods was so strong that Chinese tourists flew to European capitals for the expe- rience of buying—for example—a Chanel or Hermès purse at a flag- ship in Paris. The pandemic and China’s im- ploding housing market ended the good times. The Chinese share of\n\nglobal luxury spending dropped to 23% last year. They have been overtaken by Americans, who are now the industry’s most important customers. Last year, U.S. shop- pers generated 31% of global lux- ury sales, up from 22% before the pandemic. This leaves luxury brands heav- ily reliant on a clutch of wealthy Americans. The S&P 500 has gained 17% over the past 12 months, generating an additional $8.4 trillion of stock-market wealth. Most of the benefits have gone to the richest households, who have higher-than-average expo- sure to stocks. This is boosting the very top of the luxury market, for goods such as fine jewelry, or seats in first-class air cabins. Brands such as Gucci or Saint Lau- rent that are more dependent on aspirational shoppers in the U.S., defined as customers who spend up to $2,500 a year on luxury goods, are suffering. A recovery in China would pro- vide a welcome hedge. The Shang- hai Stock Exchange Composite In- dex has outperformed the S&P 500 since the start of 2025. But China’s equity markets are volatile and a less important driver of consumer spending than in the U.S. Stocks make up just a 10th of China’s household wealth according to Goldman Sachs. This compares to 35% in the U.S., data from the Fed- eral Reserve shows. Home values, which make up 60% of household wealth in China, are much more important for con- sumer spending. A property crash has wiped two-fifths off the value of Chinese real estate since 2021 in some parts of the country. Prices for existing homes fell 6% in 2025, and don’t appear to have bottomed out yet. However, prices are stabilizing in Tier 1 cities in China such as Shanghai and Shenzhen, where wealthy residents are also more\n\nlikely to own stocks. This could create pockets of demand for lux- ury goods this year. But a widespread recovery is a long shot. When middle-income Chinese consumers were splurging on designer goods several years ago, their homes were appreciat- ing in value. In the decade leading up to the pandemic, home prices in China’s top 70 cities rose more than 50%, data from the National Bureau of Statistics of China shows. Incomes were rising 10% a year on average over the period, more than double today’s rate. Chinese households are still prodigious savers. The average household is socking away one- third of its income, compared with less than 5% in the U.S. This will limit how much middle-class con- sumers are willing to spend on luxury goods. That said, brands perceived to offer value for the money are growing. **Ralph Lauren** said its sales in China rose more than 30% in its latest quarter. Coach’s Chi- nese business was up 21%. Bur- berry is having success with Gen Z consumers, thanks in part to a re- alistic pricing strategy. Local luxury players such as Songmont handbags are also win- ning over price-sensitive Chinese shoppers. Their goods sell for four to five times their manufacturing cost, compared with a markup of 10 times or higher for Western luxury brands, Bernstein notes. Richemont Chief Executive Ni- colas Bos said on the company’s latest investor call that the bar for selling luxury goods in China is high today. Chinese customers are “becoming much more demanding, discerning and differentiating when it comes to their choice of brands.” The days when shoppers queued around the block for a Chanel or Louis Vuitton handbag aren’t coming back soon. _—Carol Ryan_\n\n\\*Total exceeds 100 because of rounding.\n\nFeb. 2025 Jan. ’26 –20\n\n–10\n\n0\n\n10\n\n20\n\n30%\n\nShanghai Composite\n\nS&P 500\n\nIndex performance during the past year\n\nNew homes\n\nSecond-hand homes\n\nChina’s house prices, change from a year earlier\n\n**2025**\n\n2019\n\n0% 25 50 75 100\n\n**Chinese American** European\n\nJapanese\n\nOther Asian\n\nOther Global luxury spending by nationality\\*\n\nSources: Bernstein (spending); FactSet (stock market); National Bureau of Statistics of China, Bernstein analysis (house prices)\n\nJan. 2025 Dec. –10\n\n–8\n\n–6\n\n–4\n\n–2\n\n0%\n\n# Cruise Investors Got a Cabin Upgrade\n\nCruise fans love wave season, but this year’s prime booking win- dow was making investors queasy. They were concerned that de- mand would soften amid higher prices and increased capacity. In the six months ended last Wednes- day, shares of all three major oceangoing cruise lines were down, trailing the S&P 500 by 19% on av- erage. Weakest of all over that span was **Royal Caribbean** . Then it reported full-year results on Thursday. The stock surged 17% after management said Royal Ca- ribbean had just experienced the seven highest weeks for bookings in the company’s history. Five of those weeks fell during wave season, which continues through March. “What they heard from us today\n\nis that our guests continue to pri- oritize travel with us and that they’re willing to pay more,” said Royal Caribbean CEO Jason Liberty in an interview. Shares of competitors **Carnival**\n\nand **Norwegian Cruise Line** also jumped. The best-performing cruise stock recently had been **Viking** , which specializes in more intimate and sedate river cruises. That wasn’t so much because of where its vessels sail as who you’re likely to find on board. Michael Er- stad, senior consumer analyst at M Science, points out that their older, wealthier clients were seen as more financially secure. That cohort has kept spending on other products and services even as younger peo- ple with jobs and children have tightened their belts. But cruise lines keep confound- ing expectations. Indeed, after the industry’s near-death experience during the pandemic, they’ve be- come more popular than ever. Ship- yards that specialize in the vessels\n\nare booked up for years. Between the three big oceango- ing companies, combined passenger ticket sales are seen being almost 50% higher this fiscal year than in 2019. Earnings per share are still much lower for Carnival and Nor- wegian, though, because of finan- cial ballast they picked up to sur- vive the industry’s shutdown—debt and stock issued to just stay in business. Royal Caribbean navigated the crisis more skillfully. Now it is re- turning billions to shareholders with cash left over to order new ships. That includes expanding in an interesting direction: river cruis- ing. On Thursday, the company an- nounced the addition of 10 addi- tional vessels to its Celebrity River Cruises subsidiary.\n\nLiberty doesn’t view that as an economic hedge. He said that based on the copious data the company has on its many return customers, river cruises interest many of them. Ocean vacations are already often multigenerational trips, he says. Why has cruise demand been so buoyant, though? One explanation might be price. Even though they’ve become more expensive, land-based vacations have seen even steeper inflation. So how do they do it? This isn’t the Love Boat. Ships are massive entertainment complexes and in- creasingly sail to private beaches or islands owned by the cruise lines themselves. That all creates greater economies of scale. Cruisers seem to like it, too. _—Spencer Jakab_\n\nforced crypto investors to re- think their riskier holdings. Some traders are hoping for relief ahead. Kevin Warsh, President Trump’s Federal Reserve chair nominee, has made supportive comments about bitcoin in the past, but some investors fear he could support tighter monetary pol- icy and hurt riskier assets. Wall Street’s stomach for those riskier bets could be\n\ntested this week, when Strat- egy reports its earnings. Big tech names like **Palantir** , **Am- azon.com** and chip maker **Ad- vanced Micro Devices** are also slated to report their fi- nancial results, and a jobs re- port from the Labor Depart- ment is due this coming Friday. Investors have yanked a net $227 million out of spot bitcoin exchange-traded\n\n## **MARKETS**\n\nunderwater. “It’s a mess out there,” said Gennaro Salemme, a 32- year-old car salesman in Chi- cago. “The vibe right now is ‘stay alive.’” Salemme still keeps roughly 40% of his $200,000 portfolio in crypto invest- ments including ether, XRP and bitcoin ETFs. He first bought bitcoin nine years ago when the price was $3,000, and said he is “100%” still bullish on it for the long term. But holding on can be excruciating when everyone else is abandoning ship. “Everyone wants to leave when things go south,” Sa- lemme said. “A lot of this is emotionally driven.” After rallying in early 2025, digital assets surren- dered much of the spotlight to the artificial-intelligence race, which sucked up inves- tors’ capital and powered a third straight year of double- digit gains in stocks. A weak- ening job market and escalat- ing geopolitical tensions also\n\nfunds through Jan. 28, ac- cording to Morningstar Direct data. The bleak mood in crypto- land is in contrast to the start of last year, when investors cheered what many believed would be a new golden era for digital assets. Major Wall Street firms had already launched spot bitcoin ETFs, pushing digital tokens into the mainstream. And Donald Trump was back in the White House, promising to turn America into a “bitcoin su- perpower.” Then last summer, Trump signed into law a measure to regulate stablecoins, fueling expectations that clearer rules could help boost the crypto industry. “It was almost like it was 2021 all over again,” said Lewis Carr, a 19-year-old col- lege student in Texas who sold gaming consoles on Facebook Marketplace to fund his crypto investments. “Like you couldn’t lose money.” But the celebratory mood\n\neventually gave way to a hangover. Carr keeps about one-quar- ter of his roughly $6,000 portfolio in crypto and still puts about $50 a week into bitcoin, ether and XRP. But the slide has tempered his optimism. “Crypto right now is a to- tal gamble,” he said. “I think everybody realizes that.” Will Warden had similarly high expectations not so long ago. The commercial real-es- tate broker, 44, had watched crypto from the sidelines for years, then jumped in after Trump took office last Janu- ary. “My expectations were that by late spring, summer of 2025, I would have had a huge return,” he said. “But that didn’t happen.” Warden has decided to ride out the downturn, but his fa- ther and brother have cashed out of their crypto holdings. “A lot of people are ex- hausted, or irritated, or an- gry,” he said.\n\nThe party in the crypto market is starting to feel like a distant memory. After a disappointing end to 2025, major cryptocurren- cies have languished to start the new year. Bitcoin shed roughly one-third of its value since hitting a record in Octo- ber through Friday, and sank further on Saturday. The slump has continued even though stocks are near records, with the gains broad- ening throughout the market and investors betting the economy will keep growing. A weakened dollar pushed in- vestors toward alternative as- sets for safety in the so-called debasement trade. Yet prices for cryptocur- rencies have slid. On Saturday, bitcoin briefly dipped below the average price Michael Saylor’s **Strat- egy** has paid for the token, putting the crypto-treasury company’s bitcoin holdings\n\nBY DALVIN BROWN AND HANNAH ERIN LANG\n\n# The Vibe in the Crypto Market Right Now: ‘Stay Alive’\n\nS&P 500\n\nBitcoin\n\nAug. 2025 Sept. Oct. Nov. Dec. Jan. ’26\n\nCryptocurrency and index performance, past six months\n\nSources: FactSet (index); Kraken (ether); CoinDesk (bitcoin)\n\n–30\n\n–20\n\n–10\n\n0\n\n10\n\n20\n\n30%\n\nEther\n\nViking 8.8%\n\n–13.5\n\n–21.4\n\n–22.4\n\nSix-month total return through Wednesday (prior to Royal Caribbean results)\n\nSource: FactSet\n\nRoyal Caribbean\n\nNorwegian\n\nCarnival\n\nP2JW033000-0-B01000-1--------XA","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"B10 | Monday, February 2, 2026 **THE WALL STREET JOURNAL.**\n\n# **Luxury Comeback in China Isn’t Likely**\n\n### Even with a stock market outperforming the S&P 500, Chinese shoppers aren’t flocking to chic brands as they once did\n\n# HEARD\n\n## ON\n\n# THESTREET\n\nFINANCIAL ANALYSIS & COMMENTARY\n\nThe Louis Vuitton store in Shanghai is shaped like a cruise ship.\n\nRAUL ARIANO/BLOOMBERG NEWS\n\nChinese shoppers are returning to luxury stores, but with less ap- petite to spend and in greatly di- minished numbers. That is disap- pointing for high-end brands desperate for fresh growth. After five years of weak sales, some luxury brands said their China business recently turned a corner. **Richemont** , which owns Cartier and Van Cleef & Arpels jewelry, said sales started to grow again in China in the third quarter of 2025. British luxury trench coat maker **Burberry** also noticed an improvement last fall, and said de- mand accelerated further in the fi- nal three months of the year. The recent results of the world’s biggest luxury goods com- pany, **LVMH** , gave a more mixed picture. Sales to Chinese consum- ers are stabilizing for LVMH, but aren’t back to growth yet. The company is tweaking its footprint in the country. LVMH recently agreed to sell its duty-free business in China, but is investing, too. The company’s flag- ship brand, Louis Vuitton, in June opened a Shanghai boutique, which is drawing thousands of vis- itors a day. If the global luxury industry is to get back to the 6% annual growth it averaged over the past two decades, it needs China to re- cover. Luxury stocks used to be a favorite way for investors to get exposure to the country’s then- booming middle class. The Chinese went from barely being on luxury brands’ radar in the early 2000s to generating 35% of the industry’s global sales by 2019, Bernstein estimates. The ap- petite for luxury goods was so strong that Chinese tourists flew to European capitals for the expe- rience of buying—for example—a Chanel or Hermès purse at a flag- ship in Paris. The pandemic and China’s im- ploding housing market ended the good times. The Chinese share of\n\nglobal luxury spending dropped to 23% last year. They have been overtaken by Americans, who are now the industry’s most important customers. Last year, U.S. shop- pers generated 31% of global lux- ury sales, up from 22% before the pandemic. This leaves luxury brands heav- ily reliant on a clutch of wealthy Americans. The S&P 500 has gained 17% over the past 12 months, generating an additional $8.4 trillion of stock-market wealth. Most of the benefits have gone to the richest households, who have higher-than-average expo- sure to stocks. This is boosting the very top of the luxury market, for goods such as fine jewelry, or seats in first-class air cabins. Brands such as Gucci or Saint Lau- rent that are more dependent on aspirational shoppers in the U.S., defined as customers who spend up to $2,500 a year on luxury goods, are suffering. A recovery in China would pro- vide a welcome hedge. The Shang- hai Stock Exchange Composite In- dex has outperformed the S&P 500 since the start of 2025. But China’s equity markets are volatile and a less important driver of consumer spending than in the U.S. Stocks make up just a 10th of China’s household wealth according to Goldman Sachs. This compares to 35% in the U.S., data from the Fed- eral Reserve shows. Home values, which make up 60% of household wealth in China, are much more important for con- sumer spending. A property crash has wiped two-fifths off the value of Chinese real estate since 2021 in some parts of the country. Prices for existing homes fell 6% in 2025, and don’t appear to have bottomed out yet. However, prices are stabilizing in Tier 1 cities in China such as Shanghai and Shenzhen, where wealthy residents are also more\n\nlikely to own stocks. This could create pockets of demand for lux- ury goods this year. But a widespread recovery is a long shot. When middle-income Chinese consumers were splurging on designer goods several years ago, their homes were appreciat- ing in value. In the decade leading up to the pandemic, home prices in China’s top 70 cities rose more than 50%, data from the National Bureau of Statistics of China shows. Incomes were rising 10% a year on average over the period, more than double today’s rate. Chinese households are still prodigious savers. The average household is socking away one- third of its income, compared with less than 5% in the U.S. This will limit how much middle-class con- sumers are willing to spend on luxury goods. That said, brands perceived to offer value for the money are growing. **Ralph Lauren** said its sales in China rose more than 30% in its latest quarter. Coach’s Chi- nese business was up 21%. Bur- berry is having success with Gen Z consumers, thanks in part to a re- alistic pricing strategy. Local luxury players such as Songmont handbags are also win- ning over price-sensitive Chinese shoppers. Their goods sell for four to five times their manufacturing cost, compared with a markup of 10 times or higher for Western luxury brands, Bernstein notes. Richemont Chief Executive Ni- colas Bos said on the company’s latest investor call that the bar for selling luxury goods in China is high today. Chinese customers are “becoming much more demanding, discerning and differentiating when it comes to their choice of brands.” The days when shoppers queued around the block for a Chanel or Louis Vuitton handbag aren’t coming back soon. _—Carol Ryan_\n\n\\*Total exceeds 100 because of rounding.\n\nFeb. 2025 Jan. ’26 –20\n\n–10\n\n0\n\n10\n\n20\n\n30%\n\nShanghai Composite\n\nS&P 500\n\nIndex performance during the past year\n\nNew homes\n\nSecond-hand homes\n\nChina’s house prices, change from a year earlier\n\n**2025**\n\n2019\n\n0% 25 50 75 100\n\n**Chinese American** European\n\nJapanese\n\nOther Asian\n\nOther Global luxury spending by nationality\\*\n\nSources: Bernstein (spending); FactSet (stock market); National Bureau of Statistics of China, Bernstein analysis (house prices)\n\nJan. 2025 Dec. –10\n\n–8\n\n–6\n\n–4\n\n–2\n\n0%\n\n# Cruise Investors Got a Cabin Upgrade\n\nCruise fans love wave season, but this year’s prime booking win- dow was making investors queasy. They were concerned that de- mand would soften amid higher prices and increased capacity. In the six months ended last Wednes- day, shares of all three major oceangoing cruise lines were down, trailing the S&P 500 by 19% on av- erage. Weakest of all over that span was **Royal Caribbean** . Then it reported full-year results on Thursday. The stock surged 17% after management said Royal Ca- ribbean had just experienced the seven highest weeks for bookings in the company’s history. Five of those weeks fell during wave season, which continues through March. “What they heard from us today\n\nis that our guests continue to pri- oritize travel with us and that they’re willing to pay more,” said Royal Caribbean CEO Jason Liberty in an interview. Shares of competitors **Carnival**\n\nand **Norwegian Cruise Line** also jumped. The best-performing cruise stock recently had been **Viking** , which specializes in more intimate and sedate river cruises. That wasn’t so much because of where its vessels sail as who you’re likely to find on board. Michael Er- stad, senior consumer analyst at M Science, points out that their older, wealthier clients were seen as more financially secure. That cohort has kept spending on other products and services even as younger peo- ple with jobs and children have tightened their belts. But cruise lines keep confound- ing expectations. Indeed, after the industry’s near-death experience during the pandemic, they’ve be- come more popular than ever. Ship- yards that specialize in the vessels\n\nare booked up for years. Between the three big oceango- ing companies, combined passenger ticket sales are seen being almost 50% higher this fiscal year than in 2019. Earnings per share are still much lower for Carnival and Nor- wegian, though, because of finan- cial ballast they picked up to sur- vive the industry’s shutdown—debt and stock issued to just stay in business. Royal Caribbean navigated the crisis more skillfully. Now it is re- turning billions to shareholders with cash left over to order new ships. That includes expanding in an interesting direction: river cruis- ing. On Thursday, the company an- nounced the addition of 10 addi- tional vessels to its Celebrity River Cruises subsidiary.\n\nLiberty doesn’t view that as an economic hedge. He said that based on the copious data the company has on its many return customers, river cruises interest many of them. Ocean vacations are already often multigenerational trips, he says. Why has cruise demand been so buoyant, though? One explanation might be price. Even though they’ve become more expensive, land-based vacations have seen even steeper inflation. So how do they do it? This isn’t the Love Boat. Ships are massive entertainment complexes and in- creasingly sail to private beaches or islands owned by the cruise lines themselves. That all creates greater economies of scale. Cruisers seem to like it, too. _—Spencer Jakab_\n\nforced crypto investors to re- think their riskier holdings. Some traders are hoping for relief ahead. Kevin Warsh, President Trump’s Federal Reserve chair nominee, has made supportive comments about bitcoin in the past, but some investors fear he could support tighter monetary pol- icy and hurt riskier assets. Wall Street’s stomach for those riskier bets could be\n\ntested this week, when Strat- egy reports its earnings. Big tech names like **Palantir** , **Am- azon.com** and chip maker **Ad- vanced Micro Devices** are also slated to report their fi- nancial results, and a jobs re- port from the Labor Depart- ment is due this coming Friday. Investors have yanked a net $227 million out of spot bitcoin exchange-traded\n\n## **MARKETS**\n\nunderwater. “It’s a mess out there,” said Gennaro Salemme, a 32- year-old car salesman in Chi- cago. “The vibe right now is ‘stay alive.’” Salemme still keeps roughly 40% of his $200,000 portfolio in crypto invest- ments including ether, XRP and bitcoin ETFs. He first bought bitcoin nine years ago when the price was $3,000, and said he is “100%” still bullish on it for the long term. But holding on can be excruciating when everyone else is abandoning ship. “Everyone wants to leave when things go south,” Sa- lemme said. “A lot of this is emotionally driven.” After rallying in early 2025, digital assets surren- dered much of the spotlight to the artificial-intelligence race, which sucked up inves- tors’ capital and powered a third straight year of double- digit gains in stocks. A weak- ening job market and escalat- ing geopolitical tensions also\n\nfunds through Jan. 28, ac- cording to Morningstar Direct data. The bleak mood in crypto- land is in contrast to the start of last year, when investors cheered what many believed would be a new golden era for digital assets. Major Wall Street firms had already launched spot bitcoin ETFs, pushing digital tokens into the mainstream. And Donald Trump was back in the White House, promising to turn America into a “bitcoin su- perpower.” Then last summer, Trump signed into law a measure to regulate stablecoins, fueling expectations that clearer rules could help boost the crypto industry. “It was almost like it was 2021 all over again,” said Lewis Carr, a 19-year-old col- lege student in Texas who sold gaming consoles on Facebook Marketplace to fund his crypto investments. “Like you couldn’t lose money.” But the celebratory mood\n\neventually gave way to a hangover. Carr keeps about one-quar- ter of his roughly $6,000 portfolio in crypto and still puts about $50 a week into bitcoin, ether and XRP. But the slide has tempered his optimism. “Crypto right now is a to- tal gamble,” he said. “I think everybody realizes that.” Will Warden had similarly high expectations not so long ago. The commercial real-es- tate broker, 44, had watched crypto from the sidelines for years, then jumped in after Trump took office last Janu- ary. “My expectations were that by late spring, summer of 2025, I would have had a huge return,” he said. “But that didn’t happen.” Warden has decided to ride out the downturn, but his fa- ther and brother have cashed out of their crypto holdings. “A lot of people are ex- hausted, or irritated, or an- gry,” he said.\n\nThe party in the crypto market is starting to feel like a distant memory. After a disappointing end to 2025, major cryptocurren- cies have languished to start the new year. Bitcoin shed roughly one-third of its value since hitting a record in Octo- ber through Friday, and sank further on Saturday. The slump has continued even though stocks are near records, with the gains broad- ening throughout the market and investors betting the economy will keep growing. A weakened dollar pushed in- vestors toward alternative as- sets for safety in the so-called debasement trade. Yet prices for cryptocur- rencies have slid. On Saturday, bitcoin briefly dipped below the average price Michael Saylor’s **Strat- egy** has paid for the token, putting the crypto-treasury company’s bitcoin holdings\n\nBY DALVIN BROWN AND HANNAH ERIN LANG\n\n# The Vibe in the Crypto Market Right Now: ‘Stay Alive’\n\nS&P 500\n\nBitcoin\n\nAug. 2025 Sept. Oct. Nov. Dec. Jan. ’26\n\nCryptocurrency and index performance, past six months\n\nSources: FactSet (index); Kraken (ether); CoinDesk (bitcoin)\n\n–30\n\n–20\n\n–10\n\n0\n\n10\n\n20\n\n30%\n\nEther\n\nViking 8.8%\n\n–13.5\n\n–21.4\n\n–22.4\n\nSix-month total return through Wednesday (prior to Royal Caribbean results)\n\nSource: FactSet\n\nRoyal Caribbean\n\nNorwegian\n\nCarnival\n\nP2JW033000-0-B01000-1--------XA","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/pdf_bae2c1684672_p26.pdf"]},"response":{"errors":[],"extract_id":"extract_168e61948404a1d09877d43b31aa8be3","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["B10 | Monday, February 2, 2026 **THE WALL STREET JOURNAL.**\n\n# **Luxury Comeback in China Isn’t Likely**\n\n### Even with a stock market outperforming the S&P 500, Chinese shoppers aren’t flocking to chic brands as they once did\n\n# HEARD\n\n## ON\n\n# THESTREET\n\nFINANCIAL ANALYSIS & COMMENTARY\n\nThe Louis Vuitton store in Shanghai is shaped like a cruise ship.\n\nRAUL ARIANO/BLOOMBERG NEWS\n\nChinese shoppers are returning to luxury stores, but with less ap- petite to spend and in greatly di- minished numbers. That is disap- pointing for high-end brands desperate for fresh growth. After five years of weak sales, some luxury brands said their China business recently turned a corner. **Richemont** , which owns Cartier and Van Cleef & Arpels jewelry, said sales started to grow again in China in the third quarter of 2025. British luxury trench coat maker **Burberry** also noticed an improvement last fall, and said de- mand accelerated further in the fi- nal three months of the year. The recent results of the world’s biggest luxury goods com- pany, **LVMH** , gave a more mixed picture. Sales to Chinese consum- ers are stabilizing for LVMH, but aren’t back to growth yet. The company is tweaking its footprint in the country. LVMH recently agreed to sell its duty-free business in China, but is investing, too. The company’s flag- ship brand, Louis Vuitton, in June opened a Shanghai boutique, which is drawing thousands of vis- itors a day. If the global luxury industry is to get back to the 6% annual growth it averaged over the past two decades, it needs China to re- cover. Luxury stocks used to be a favorite way for investors to get exposure to the country’s then- booming middle class. The Chinese went from barely being on luxury brands’ radar in the early 2000s to generating 35% of the industry’s global sales by 2019, Bernstein estimates. The ap- petite for luxury goods was so strong that Chinese tourists flew to European capitals for the expe- rience of buying—for example—a Chanel or Hermès purse at a flag- ship in Paris. The pandemic and China’s im- ploding housing market ended the good times. The Chinese share of\n\nglobal luxury spending dropped to 23% last year. They have been overtaken by Americans, who are now the industry’s most important customers. Last year, U.S. shop- pers generated 31% of global lux- ury sales, up from 22% before the pandemic. This leaves luxury brands heav- ily reliant on a clutch of wealthy Americans. The S&P 500 has gained 17% over the past 12 months, generating an additional $8.4 trillion of stock-market wealth. Most of the benefits have gone to the richest households, who have higher-than-average expo- sure to stocks. This is boosting the very top of the luxury market, for goods such as fine jewelry, or seats in first-class air cabins. Brands such as Gucci or Saint Lau- rent that are more dependent on aspirational shoppers in the U.S., defined as customers who spend up to $2,500 a year on luxury goods, are suffering. A recovery in China would pro- vide a welcome hedge. The Shang- hai Stock Exchange Composite In- dex has outperformed the S&P 500 since the start of 2025. But China’s equity markets are volatile and a less important driver of consumer spending than in the U.S. Stocks make up just a 10th of China’s household wealth according to Goldman Sachs. This compares to 35% in the U.S., data from the Fed- eral Reserve shows. Home values, which make up 60% of household wealth in China, are much more important for con- sumer spending. A property crash has wiped two-fifths off the value of Chinese real estate since 2021 in some parts of the country. Prices for existing homes fell 6% in 2025, and don’t appear to have bottomed out yet. However, prices are stabilizing in Tier 1 cities in China such as Shanghai and Shenzhen, where wealthy residents are also more\n\nlikely to own stocks. This could create pockets of demand for lux- ury goods this year. But a widespread recovery is a long shot. When middle-income Chinese consumers were splurging on designer goods several years ago, their homes were appreciat- ing in value. In the decade leading up to the pandemic, home prices in China’s top 70 cities rose more than 50%, data from the National Bureau of Statistics of China shows. Incomes were rising 10% a year on average over the period, more than double today’s rate. Chinese households are still prodigious savers. The average household is socking away one- third of its income, compared with less than 5% in the U.S. This will limit how much middle-class con- sumers are willing to spend on luxury goods. That said, brands perceived to offer value for the money are growing. **Ralph Lauren** said its sales in China rose more than 30% in its latest quarter. Coach’s Chi- nese business was up 21%. Bur- berry is having success with Gen Z consumers, thanks in part to a re- alistic pricing strategy. Local luxury players such as Songmont handbags are also win- ning over price-sensitive Chinese shoppers. Their goods sell for four to five times their manufacturing cost, compared with a markup of 10 times or higher for Western luxury brands, Bernstein notes. Richemont Chief Executive Ni- colas Bos said on the company’s latest investor call that the bar for selling luxury goods in China is high today. Chinese customers are “becoming much more demanding, discerning and differentiating when it comes to their choice of brands.” The days when shoppers queued around the block for a Chanel or Louis Vuitton handbag aren’t coming back soon. _—Carol Ryan_\n\n\\*Total exceeds 100 because of rounding.\n\nFeb. 2025 Jan. ’26 –20\n\n–10\n\n0\n\n10\n\n20\n\n30%\n\nShanghai Composite\n\nS&P 500\n\nIndex performance during the past year\n\nNew homes\n\nSecond-hand homes\n\nChina’s house prices, change from a year earlier\n\n**2025**\n\n2019\n\n0% 25 50 75 100\n\n**Chinese American** European\n\nJapanese\n\nOther Asian\n\nOther Global luxury spending by nationality\\*\n\nSources: Bernstein (spending); FactSet (stock market); National Bureau of Statistics of China, Bernstein analysis (house prices)\n\nJan. 2025 Dec. –10\n\n–8\n\n–6\n\n–4\n\n–2\n\n0%\n\n# Cruise Investors Got a Cabin Upgrade\n\nCruise fans love wave season, but this year’s prime booking win- dow was making investors queasy. They were concerned that de- mand would soften amid higher prices and increased capacity. In the six months ended last Wednes- day, shares of all three major oceangoing cruise lines were down, trailing the S&P 500 by 19% on av- erage. Weakest of all over that span was **Royal Caribbean** . Then it reported full-year results on Thursday. The stock surged 17% after management said Royal Ca- ribbean had just experienced the seven highest weeks for bookings in the company’s history. Five of those weeks fell during wave season, which continues through March. “What they heard from us today\n\nis that our guests continue to pri- oritize travel with us and that they’re willing to pay more,” said Royal Caribbean CEO Jason Liberty in an interview. Shares of competitors **Carnival**\n\nand **Norwegian Cruise Line** also jumped. The best-performing cruise stock recently had been **Viking** , which specializes in more intimate and sedate river cruises. That wasn’t so much because of where its vessels sail as who you’re likely to find on board. Michael Er- stad, senior consumer analyst at M Science, points out that their older, wealthier clients were seen as more financially secure. That cohort has kept spending on other products and services even as younger peo- ple with jobs and children have tightened their belts. But cruise lines keep confound- ing expectations. Indeed, after the industry’s near-death experience during the pandemic, they’ve be- come more popular than ever. Ship- yards that specialize in the vessels\n\nare booked up for years. Between the three big oceango- ing companies, combined passenger ticket sales are seen being almost 50% higher this fiscal year than in 2019. Earnings per share are still much lower for Carnival and Nor- wegian, though, because of finan- cial ballast they picked up to sur- vive the industry’s shutdown—debt and stock issued to just stay in business. Royal Caribbean navigated the crisis more skillfully. Now it is re- turning billions to shareholders with cash left over to order new ships. That includes expanding in an interesting direction: river cruis- ing. On Thursday, the company an- nounced the addition of 10 addi- tional vessels to its Celebrity River Cruises subsidiary.\n\nLiberty doesn’t view that as an economic hedge. He said that based on the copious data the company has on its many return customers, river cruises interest many of them. Ocean vacations are already often multigenerational trips, he says. Why has cruise demand been so buoyant, though? One explanation might be price. Even though they’ve become more expensive, land-based vacations have seen even steeper inflation. So how do they do it? This isn’t the Love Boat. Ships are massive entertainment complexes and in- creasingly sail to private beaches or islands owned by the cruise lines themselves. That all creates greater economies of scale. Cruisers seem to like it, too. _—Spencer Jakab_\n\nforced crypto investors to re- think their riskier holdings. Some traders are hoping for relief ahead. Kevin Warsh, President Trump’s Federal Reserve chair nominee, has made supportive comments about bitcoin in the past, but some investors fear he could support tighter monetary pol- icy and hurt riskier assets. Wall Street’s stomach for those riskier bets could be\n\ntested this week, when Strat- egy reports its earnings. Big tech names like **Palantir** , **Am- azon.com** and chip maker **Ad- vanced Micro Devices** are also slated to report their fi- nancial results, and a jobs re- port from the Labor Depart- ment is due this coming Friday. Investors have yanked a net $227 million out of spot bitcoin exchange-traded\n\n## **MARKETS**\n\nunderwater. “It’s a mess out there,” said Gennaro Salemme, a 32- year-old car salesman in Chi- cago. “The vibe right now is ‘stay alive.’” Salemme still keeps roughly 40% of his $200,000 portfolio in crypto invest- ments including ether, XRP and bitcoin ETFs. He first bought bitcoin nine years ago when the price was $3,000, and said he is “100%” still bullish on it for the long term. But holding on can be excruciating when everyone else is abandoning ship. “Everyone wants to leave when things go south,” Sa- lemme said. “A lot of this is emotionally driven.” After rallying in early 2025, digital assets surren- dered much of the spotlight to the artificial-intelligence race, which sucked up inves- tors’ capital and powered a third straight year of double- digit gains in stocks. A weak- ening job market and escalat- ing geopolitical tensions also\n\nfunds through Jan. 28, ac- cording to Morningstar Direct data. The bleak mood in crypto- land is in contrast to the start of last year, when investors cheered what many believed would be a new golden era for digital assets. Major Wall Street firms had already launched spot bitcoin ETFs, pushing digital tokens into the mainstream. And Donald Trump was back in the White House, promising to turn America into a “bitcoin su- perpower.” Then last summer, Trump signed into law a measure to regulate stablecoins, fueling expectations that clearer rules could help boost the crypto industry. “It was almost like it was 2021 all over again,” said Lewis Carr, a 19-year-old col- lege student in Texas who sold gaming consoles on Facebook Marketplace to fund his crypto investments. “Like you couldn’t lose money.” But the celebratory mood\n\neventually gave way to a hangover. Carr keeps about one-quar- ter of his roughly $6,000 portfolio in crypto and still puts about $50 a week into bitcoin, ether and XRP. But the slide has tempered his optimism. “Crypto right now is a to- tal gamble,” he said. “I think everybody realizes that.” Will Warden had similarly high expectations not so long ago. The commercial real-es- tate broker, 44, had watched crypto from the sidelines for years, then jumped in after Trump took office last Janu- ary. “My expectations were that by late spring, summer of 2025, I would have had a huge return,” he said. “But that didn’t happen.” Warden has decided to ride out the downturn, but his fa- ther and brother have cashed out of their crypto holdings. “A lot of people are ex- hausted, or irritated, or an- gry,” he said.\n\nThe party in the crypto market is starting to feel like a distant memory. After a disappointing end to 2025, major cryptocurren- cies have languished to start the new year. Bitcoin shed roughly one-third of its value since hitting a record in Octo- ber through Friday, and sank further on Saturday. The slump has continued even though stocks are near records, with the gains broad- ening throughout the market and investors betting the economy will keep growing. A weakened dollar pushed in- vestors toward alternative as- sets for safety in the so-called debasement trade. Yet prices for cryptocur- rencies have slid. On Saturday, bitcoin briefly dipped below the average price Michael Saylor’s **Strat- egy** has paid for the token, putting the crypto-treasury company’s bitcoin holdings\n\nBY DALVIN BROWN AND HANNAH ERIN LANG\n\n# The Vibe in the Crypto Market Right Now: ‘Stay Alive’\n\nS&P 500\n\nBitcoin\n\nAug. 2025 Sept. Oct. Nov. Dec. Jan. ’26\n\nCryptocurrency and index performance, past six months\n\nSources: FactSet (index); Kraken (ether); CoinDesk (bitcoin)\n\n–30\n\n–20\n\n–10\n\n0\n\n10\n\n20\n\n30%\n\nEther\n\nViking 8.8%\n\n–13.5\n\n–21.4\n\n–22.4\n\nSix-month total return through Wednesday (prior to Royal Caribbean results)\n\nSource: FactSet\n\nRoyal Caribbean\n\nNorwegian\n\nCarnival\n\nP2JW033000-0-B01000-1--------XA"],"full_content":"B10 | Monday, February 2, 2026 **THE WALL STREET JOURNAL.**\n\n# **Luxury Comeback in China Isn’t Likely**\n\n### Even with a stock market outperforming the S&P 500, Chinese shoppers aren’t flocking to chic brands as they once did\n\n# HEARD\n\n## ON\n\n# THESTREET\n\nFINANCIAL ANALYSIS & COMMENTARY\n\nThe Louis Vuitton store in Shanghai is shaped like a cruise ship.\n\nRAUL ARIANO/BLOOMBERG NEWS\n\nChinese shoppers are returning to luxury stores, but with less ap- petite to spend and in greatly di- minished numbers. That is disap- pointing for high-end brands desperate for fresh growth. After five years of weak sales, some luxury brands said their China business recently turned a corner. **Richemont** , which owns Cartier and Van Cleef & Arpels jewelry, said sales started to grow again in China in the third quarter of 2025. British luxury trench coat maker **Burberry** also noticed an improvement last fall, and said de- mand accelerated further in the fi- nal three months of the year. The recent results of the world’s biggest luxury goods com- pany, **LVMH** , gave a more mixed picture. Sales to Chinese consum- ers are stabilizing for LVMH, but aren’t back to growth yet. The company is tweaking its footprint in the country. LVMH recently agreed to sell its duty-free business in China, but is investing, too. The company’s flag- ship brand, Louis Vuitton, in June opened a Shanghai boutique, which is drawing thousands of vis- itors a day. If the global luxury industry is to get back to the 6% annual growth it averaged over the past two decades, it needs China to re- cover. Luxury stocks used to be a favorite way for investors to get exposure to the country’s then- booming middle class. The Chinese went from barely being on luxury brands’ radar in the early 2000s to generating 35% of the industry’s global sales by 2019, Bernstein estimates. The ap- petite for luxury goods was so strong that Chinese tourists flew to European capitals for the expe- rience of buying—for example—a Chanel or Hermès purse at a flag- ship in Paris. The pandemic and China’s im- ploding housing market ended the good times. The Chinese share of\n\nglobal luxury spending dropped to 23% last year. They have been overtaken by Americans, who are now the industry’s most important customers. Last year, U.S. shop- pers generated 31% of global lux- ury sales, up from 22% before the pandemic. This leaves luxury brands heav- ily reliant on a clutch of wealthy Americans. The S&P 500 has gained 17% over the past 12 months, generating an additional $8.4 trillion of stock-market wealth. Most of the benefits have gone to the richest households, who have higher-than-average expo- sure to stocks. This is boosting the very top of the luxury market, for goods such as fine jewelry, or seats in first-class air cabins. Brands such as Gucci or Saint Lau- rent that are more dependent on aspirational shoppers in the U.S., defined as customers who spend up to $2,500 a year on luxury goods, are suffering. A recovery in China would pro- vide a welcome hedge. The Shang- hai Stock Exchange Composite In- dex has outperformed the S&P 500 since the start of 2025. But China’s equity markets are volatile and a less important driver of consumer spending than in the U.S. Stocks make up just a 10th of China’s household wealth according to Goldman Sachs. This compares to 35% in the U.S., data from the Fed- eral Reserve shows. Home values, which make up 60% of household wealth in China, are much more important for con- sumer spending. A property crash has wiped two-fifths off the value of Chinese real estate since 2021 in some parts of the country. Prices for existing homes fell 6% in 2025, and don’t appear to have bottomed out yet. However, prices are stabilizing in Tier 1 cities in China such as Shanghai and Shenzhen, where wealthy residents are also more\n\nlikely to own stocks. This could create pockets of demand for lux- ury goods this year. But a widespread recovery is a long shot. When middle-income Chinese consumers were splurging on designer goods several years ago, their homes were appreciat- ing in value. In the decade leading up to the pandemic, home prices in China’s top 70 cities rose more than 50%, data from the National Bureau of Statistics of China shows. Incomes were rising 10% a year on average over the period, more than double today’s rate. Chinese households are still prodigious savers. The average household is socking away one- third of its income, compared with less than 5% in the U.S. This will limit how much middle-class con- sumers are willing to spend on luxury goods. That said, brands perceived to offer value for the money are growing. **Ralph Lauren** said its sales in China rose more than 30% in its latest quarter. Coach’s Chi- nese business was up 21%. Bur- berry is having success with Gen Z consumers, thanks in part to a re- alistic pricing strategy. Local luxury players such as Songmont handbags are also win- ning over price-sensitive Chinese shoppers. Their goods sell for four to five times their manufacturing cost, compared with a markup of 10 times or higher for Western luxury brands, Bernstein notes. Richemont Chief Executive Ni- colas Bos said on the company’s latest investor call that the bar for selling luxury goods in China is high today. Chinese customers are “becoming much more demanding, discerning and differentiating when it comes to their choice of brands.” The days when shoppers queued around the block for a Chanel or Louis Vuitton handbag aren’t coming back soon. _—Carol Ryan_\n\n\\*Total exceeds 100 because of rounding.\n\nFeb. 2025 Jan. ’26 –20\n\n–10\n\n0\n\n10\n\n20\n\n30%\n\nShanghai Composite\n\nS&P 500\n\nIndex performance during the past year\n\nNew homes\n\nSecond-hand homes\n\nChina’s house prices, change from a year earlier\n\n**2025**\n\n2019\n\n0% 25 50 75 100\n\n**Chinese American** European\n\nJapanese\n\nOther Asian\n\nOther Global luxury spending by nationality\\*\n\nSources: Bernstein (spending); FactSet (stock market); National Bureau of Statistics of China, Bernstein analysis (house prices)\n\nJan. 2025 Dec. –10\n\n–8\n\n–6\n\n–4\n\n–2\n\n0%\n\n# Cruise Investors Got a Cabin Upgrade\n\nCruise fans love wave season, but this year’s prime booking win- dow was making investors queasy. They were concerned that de- mand would soften amid higher prices and increased capacity. In the six months ended last Wednes- day, shares of all three major oceangoing cruise lines were down, trailing the S&P 500 by 19% on av- erage. Weakest of all over that span was **Royal Caribbean** . Then it reported full-year results on Thursday. The stock surged 17% after management said Royal Ca- ribbean had just experienced the seven highest weeks for bookings in the company’s history. Five of those weeks fell during wave season, which continues through March. “What they heard from us today\n\nis that our guests continue to pri- oritize travel with us and that they’re willing to pay more,” said Royal Caribbean CEO Jason Liberty in an interview. Shares of competitors **Carnival**\n\nand **Norwegian Cruise Line** also jumped. The best-performing cruise stock recently had been **Viking** , which specializes in more intimate and sedate river cruises. That wasn’t so much because of where its vessels sail as who you’re likely to find on board. Michael Er- stad, senior consumer analyst at M Science, points out that their older, wealthier clients were seen as more financially secure. That cohort has kept spending on other products and services even as younger peo- ple with jobs and children have tightened their belts. But cruise lines keep confound- ing expectations. Indeed, after the industry’s near-death experience during the pandemic, they’ve be- come more popular than ever. Ship- yards that specialize in the vessels\n\nare booked up for years. Between the three big oceango- ing companies, combined passenger ticket sales are seen being almost 50% higher this fiscal year than in 2019. Earnings per share are still much lower for Carnival and Nor- wegian, though, because of finan- cial ballast they picked up to sur- vive the industry’s shutdown—debt and stock issued to just stay in business. Royal Caribbean navigated the crisis more skillfully. Now it is re- turning billions to shareholders with cash left over to order new ships. That includes expanding in an interesting direction: river cruis- ing. On Thursday, the company an- nounced the addition of 10 addi- tional vessels to its Celebrity River Cruises subsidiary.\n\nLiberty doesn’t view that as an economic hedge. He said that based on the copious data the company has on its many return customers, river cruises interest many of them. Ocean vacations are already often multigenerational trips, he says. Why has cruise demand been so buoyant, though? One explanation might be price. Even though they’ve become more expensive, land-based vacations have seen even steeper inflation. So how do they do it? This isn’t the Love Boat. Ships are massive entertainment complexes and in- creasingly sail to private beaches or islands owned by the cruise lines themselves. That all creates greater economies of scale. Cruisers seem to like it, too. _—Spencer Jakab_\n\nforced crypto investors to re- think their riskier holdings. Some traders are hoping for relief ahead. Kevin Warsh, President Trump’s Federal Reserve chair nominee, has made supportive comments about bitcoin in the past, but some investors fear he could support tighter monetary pol- icy and hurt riskier assets. Wall Street’s stomach for those riskier bets could be\n\ntested this week, when Strat- egy reports its earnings. Big tech names like **Palantir** , **Am- azon.com** and chip maker **Ad- vanced Micro Devices** are also slated to report their fi- nancial results, and a jobs re- port from the Labor Depart- ment is due this coming Friday. Investors have yanked a net $227 million out of spot bitcoin exchange-traded\n\n## **MARKETS**\n\nunderwater. “It’s a mess out there,” said Gennaro Salemme, a 32- year-old car salesman in Chi- cago. “The vibe right now is ‘stay alive.’” Salemme still keeps roughly 40% of his $200,000 portfolio in crypto invest- ments including ether, XRP and bitcoin ETFs. He first bought bitcoin nine years ago when the price was $3,000, and said he is “100%” still bullish on it for the long term. But holding on can be excruciating when everyone else is abandoning ship. “Everyone wants to leave when things go south,” Sa- lemme said. “A lot of this is emotionally driven.” After rallying in early 2025, digital assets surren- dered much of the spotlight to the artificial-intelligence race, which sucked up inves- tors’ capital and powered a third straight year of double- digit gains in stocks. A weak- ening job market and escalat- ing geopolitical tensions also\n\nfunds through Jan. 28, ac- cording to Morningstar Direct data. The bleak mood in crypto- land is in contrast to the start of last year, when investors cheered what many believed would be a new golden era for digital assets. Major Wall Street firms had already launched spot bitcoin ETFs, pushing digital tokens into the mainstream. And Donald Trump was back in the White House, promising to turn America into a “bitcoin su- perpower.” Then last summer, Trump signed into law a measure to regulate stablecoins, fueling expectations that clearer rules could help boost the crypto industry. “It was almost like it was 2021 all over again,” said Lewis Carr, a 19-year-old col- lege student in Texas who sold gaming consoles on Facebook Marketplace to fund his crypto investments. “Like you couldn’t lose money.” But the celebratory mood\n\neventually gave way to a hangover. Carr keeps about one-quar- ter of his roughly $6,000 portfolio in crypto and still puts about $50 a week into bitcoin, ether and XRP. But the slide has tempered his optimism. “Crypto right now is a to- tal gamble,” he said. “I think everybody realizes that.” Will Warden had similarly high expectations not so long ago. The commercial real-es- tate broker, 44, had watched crypto from the sidelines for years, then jumped in after Trump took office last Janu- ary. “My expectations were that by late spring, summer of 2025, I would have had a huge return,” he said. “But that didn’t happen.” Warden has decided to ride out the downturn, but his fa- ther and brother have cashed out of their crypto holdings. “A lot of people are ex- hausted, or irritated, or an- gry,” he said.\n\nThe party in the crypto market is starting to feel like a distant memory. After a disappointing end to 2025, major cryptocurren- cies have languished to start the new year. Bitcoin shed roughly one-third of its value since hitting a record in Octo- ber through Friday, and sank further on Saturday. The slump has continued even though stocks are near records, with the gains broad- ening throughout the market and investors betting the economy will keep growing. A weakened dollar pushed in- vestors toward alternative as- sets for safety in the so-called debasement trade. Yet prices for cryptocur- rencies have slid. On Saturday, bitcoin briefly dipped below the average price Michael Saylor’s **Strat- egy** has paid for the token, putting the crypto-treasury company’s bitcoin holdings\n\nBY DALVIN BROWN AND HANNAH ERIN LANG\n\n# The Vibe in the Crypto Market Right Now: ‘Stay Alive’\n\nS&P 500\n\nBitcoin\n\nAug. 2025 Sept. Oct. Nov. Dec. Jan. ’26\n\nCryptocurrency and index performance, past six months\n\nSources: FactSet (index); Kraken (ether); CoinDesk (bitcoin)\n\n–30\n\n–20\n\n–10\n\n0\n\n10\n\n20\n\n30%\n\nEther\n\nViking 8.8%\n\n–13.5\n\n–21.4\n\n–22.4\n\nSix-month total return through Wednesday (prior to Royal Caribbean results)\n\nSource: FactSet\n\nRoyal Caribbean\n\nNorwegian\n\nCarnival\n\nP2JW033000-0-B01000-1--------XA","publish_date":null,"title":"HEARD STREET","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/pdf_bae2c1684672_p26.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/pdf_bae2c1684672_p26.pdf"},"request":{"example_id":"chart/pdf_bae2c1684672_p26","product_type":"parse","source_file_path":"pdf_bae2c1684672_p26.pdf"},"started_at":"2026-11-04T19:45:18.309013"} +{"completed_at":"2026-11-04T19:45:20.011907","latency_in_ms":1052,"output":{"example_id":"chart/pdf_d2e0103f64cd_p29","grounded_pages":[],"layout_pages":[],"markdown":"The following graph illustrates the average occupancy rates in the Singapore CBD for the period from 2010 to the third quarter of 2013.\n\n**Average Occupancy Rates in the Singapore CBD (%)**\n\n65%\n\n70%\n\n75%\n\n80%\n\n85%\n\n90%\n\n95%\n\n100%\n\nShenton Way/Robinson Rd/Cecil Street Raffles Place Marina Bay\n\nQ3 2013 Q2 2013 Q1 2013 2012 2011 2010\n\n96\\.8% 94.4%\n\n90\\.4%\n\n70\\.3%\n\n83\\.6%\n\n90\\.9%\n\n84\\.1% (1)\n\n95\\.3%\n\n97\\.7%\n\n94\\.8% 94.3%\n\n95\\.2%\n\n91\\.6%\n\n93\\.3% 93.6%\n\n88\\.5%\n\n91\\.2%\n\n93\\.7%\n\n**Note:**\n\n(1) Due to the completion of Asia Square Tower 2 in the third quarter of 2013.\n\n_Source: Independent Market Research Report._\n\nAverage monthly passing rent for office space in the Singapore CBD had eased in 2011 due to uncertain global economic conditions, but began to recover in the third quarter of 2013, with rates at Marina Bay and Raffles Place growing 4.8% and 1.3% quarter-on-quarter to S$11.0 per sq ft per month and S$9.4 per sq ft per month, respectively.\n\nThe following graph illustrates the average rental rates in the Singapore CBD for the period from 2010 to the third quarter of 2013.\n\n**Average Rental Rates in the Singapore CBD (S$ per sq ft per month)**\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\nShenton Way/ Robinson Rd/ Cecil Street Raffles Place Marina Bay\n\nQ3 2013 Q2 2013 Q1 2013 2012 2011 2010\n\n12\\.0\n\n10\\.5 10.5 10.5 11.0\n\n9\\.4\n\n7\\.5\n\n9\\.3\n\n7\\.3 7.3\n\n9\\.3 9.3\n\n7\\.3\n\n9\\.8\n\n7\\.8 9.0\n\n6\\.5\n\nS$\n\n_Source: Independent Market Research Report._\n\n13","pages":[{"line_numbers":[],"markdown":"The following graph illustrates the average occupancy rates in the Singapore CBD for the period from 2010 to the third quarter of 2013.\n\n**Average Occupancy Rates in the Singapore CBD (%)**\n\n65%\n\n70%\n\n75%\n\n80%\n\n85%\n\n90%\n\n95%\n\n100%\n\nShenton Way/Robinson Rd/Cecil Street Raffles Place Marina Bay\n\nQ3 2013 Q2 2013 Q1 2013 2012 2011 2010\n\n96\\.8% 94.4%\n\n90\\.4%\n\n70\\.3%\n\n83\\.6%\n\n90\\.9%\n\n84\\.1% (1)\n\n95\\.3%\n\n97\\.7%\n\n94\\.8% 94.3%\n\n95\\.2%\n\n91\\.6%\n\n93\\.3% 93.6%\n\n88\\.5%\n\n91\\.2%\n\n93\\.7%\n\n**Note:**\n\n(1) Due to the completion of Asia Square Tower 2 in the third quarter of 2013.\n\n_Source: Independent Market Research Report._\n\nAverage monthly passing rent for office space in the Singapore CBD had eased in 2011 due to uncertain global economic conditions, but began to recover in the third quarter of 2013, with rates at Marina Bay and Raffles Place growing 4.8% and 1.3% quarter-on-quarter to S$11.0 per sq ft per month and S$9.4 per sq ft per month, respectively.\n\nThe following graph illustrates the average rental rates in the Singapore CBD for the period from 2010 to the third quarter of 2013.\n\n**Average Rental Rates in the Singapore CBD (S$ per sq ft per month)**\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\nShenton Way/ Robinson Rd/ Cecil Street Raffles Place Marina Bay\n\nQ3 2013 Q2 2013 Q1 2013 2012 2011 2010\n\n12\\.0\n\n10\\.5 10.5 10.5 11.0\n\n9\\.4\n\n7\\.5\n\n9\\.3\n\n7\\.3 7.3\n\n9\\.3 9.3\n\n7\\.3\n\n9\\.8\n\n7\\.8 9.0\n\n6\\.5\n\nS$\n\n_Source: Independent Market Research Report._\n\n13","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"The following graph illustrates the average occupancy rates in the Singapore CBD for the period from 2010 to the third quarter of 2013.\n\n**Average Occupancy Rates in the Singapore CBD (%)**\n\n65%\n\n70%\n\n75%\n\n80%\n\n85%\n\n90%\n\n95%\n\n100%\n\nShenton Way/Robinson Rd/Cecil Street Raffles Place Marina Bay\n\nQ3 2013 Q2 2013 Q1 2013 2012 2011 2010\n\n96\\.8% 94.4%\n\n90\\.4%\n\n70\\.3%\n\n83\\.6%\n\n90\\.9%\n\n84\\.1% (1)\n\n95\\.3%\n\n97\\.7%\n\n94\\.8% 94.3%\n\n95\\.2%\n\n91\\.6%\n\n93\\.3% 93.6%\n\n88\\.5%\n\n91\\.2%\n\n93\\.7%\n\n**Note:**\n\n(1) Due to the completion of Asia Square Tower 2 in the third quarter of 2013.\n\n_Source: Independent Market Research Report._\n\nAverage monthly passing rent for office space in the Singapore CBD had eased in 2011 due to uncertain global economic conditions, but began to recover in the third quarter of 2013, with rates at Marina Bay and Raffles Place growing 4.8% and 1.3% quarter-on-quarter to S$11.0 per sq ft per month and S$9.4 per sq ft per month, respectively.\n\nThe following graph illustrates the average rental rates in the Singapore CBD for the period from 2010 to the third quarter of 2013.\n\n**Average Rental Rates in the Singapore CBD (S$ per sq ft per month)**\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\nShenton Way/ Robinson Rd/ Cecil Street Raffles Place Marina Bay\n\nQ3 2013 Q2 2013 Q1 2013 2012 2011 2010\n\n12\\.0\n\n10\\.5 10.5 10.5 11.0\n\n9\\.4\n\n7\\.5\n\n9\\.3\n\n7\\.3 7.3\n\n9\\.3 9.3\n\n7\\.3\n\n9\\.8\n\n7\\.8 9.0\n\n6\\.5\n\nS$\n\n_Source: Independent Market Research Report._\n\n13","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/pdf_d2e0103f64cd_p29.pdf"]},"response":{"errors":[],"extract_id":"extract_e988607485d7460242c7095d564d3662","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["The following graph illustrates the average occupancy rates in the Singapore CBD for the period from 2010 to the third quarter of 2013.\n\n**Average Occupancy Rates in the Singapore CBD (%)**\n\n65%\n\n70%\n\n75%\n\n80%\n\n85%\n\n90%\n\n95%\n\n100%\n\nShenton Way/Robinson Rd/Cecil Street Raffles Place Marina Bay\n\nQ3 2013 Q2 2013 Q1 2013 2012 2011 2010\n\n96\\.8% 94.4%\n\n90\\.4%\n\n70\\.3%\n\n83\\.6%\n\n90\\.9%\n\n84\\.1% (1)\n\n95\\.3%\n\n97\\.7%\n\n94\\.8% 94.3%\n\n95\\.2%\n\n91\\.6%\n\n93\\.3% 93.6%\n\n88\\.5%\n\n91\\.2%\n\n93\\.7%\n\n**Note:**\n\n(1) Due to the completion of Asia Square Tower 2 in the third quarter of 2013.\n\n_Source: Independent Market Research Report._\n\nAverage monthly passing rent for office space in the Singapore CBD had eased in 2011 due to uncertain global economic conditions, but began to recover in the third quarter of 2013, with rates at Marina Bay and Raffles Place growing 4.8% and 1.3% quarter-on-quarter to S$11.0 per sq ft per month and S$9.4 per sq ft per month, respectively.\n\nThe following graph illustrates the average rental rates in the Singapore CBD for the period from 2010 to the third quarter of 2013.\n\n**Average Rental Rates in the Singapore CBD (S$ per sq ft per month)**\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\nShenton Way/ Robinson Rd/ Cecil Street Raffles Place Marina Bay\n\nQ3 2013 Q2 2013 Q1 2013 2012 2011 2010\n\n12\\.0\n\n10\\.5 10.5 10.5 11.0\n\n9\\.4\n\n7\\.5\n\n9\\.3\n\n7\\.3 7.3\n\n9\\.3 9.3\n\n7\\.3\n\n9\\.8\n\n7\\.8 9.0\n\n6\\.5\n\nS$\n\n_Source: Independent Market Research Report._\n\n13"],"full_content":"The following graph illustrates the average occupancy rates in the Singapore CBD for the period from 2010 to the third quarter of 2013.\n\n**Average Occupancy Rates in the Singapore CBD (%)**\n\n65%\n\n70%\n\n75%\n\n80%\n\n85%\n\n90%\n\n95%\n\n100%\n\nShenton Way/Robinson Rd/Cecil Street Raffles Place Marina Bay\n\nQ3 2013 Q2 2013 Q1 2013 2012 2011 2010\n\n96\\.8% 94.4%\n\n90\\.4%\n\n70\\.3%\n\n83\\.6%\n\n90\\.9%\n\n84\\.1% (1)\n\n95\\.3%\n\n97\\.7%\n\n94\\.8% 94.3%\n\n95\\.2%\n\n91\\.6%\n\n93\\.3% 93.6%\n\n88\\.5%\n\n91\\.2%\n\n93\\.7%\n\n**Note:**\n\n(1) Due to the completion of Asia Square Tower 2 in the third quarter of 2013.\n\n_Source: Independent Market Research Report._\n\nAverage monthly passing rent for office space in the Singapore CBD had eased in 2011 due to uncertain global economic conditions, but began to recover in the third quarter of 2013, with rates at Marina Bay and Raffles Place growing 4.8% and 1.3% quarter-on-quarter to S$11.0 per sq ft per month and S$9.4 per sq ft per month, respectively.\n\nThe following graph illustrates the average rental rates in the Singapore CBD for the period from 2010 to the third quarter of 2013.\n\n**Average Rental Rates in the Singapore CBD (S$ per sq ft per month)**\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n14\n\nShenton Way/ Robinson Rd/ Cecil Street Raffles Place Marina Bay\n\nQ3 2013 Q2 2013 Q1 2013 2012 2011 2010\n\n12\\.0\n\n10\\.5 10.5 10.5 11.0\n\n9\\.4\n\n7\\.5\n\n9\\.3\n\n7\\.3 7.3\n\n9\\.3 9.3\n\n7\\.3\n\n9\\.8\n\n7\\.8 9.0\n\n6\\.5\n\nS$\n\n_Source: Independent Market Research Report._\n\n13","publish_date":null,"title":"The following graph illustrates the average occupancy rates in the Singapore CBD for the period from 2010 to the third","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/pdf_d2e0103f64cd_p29.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/pdf_d2e0103f64cd_p29.pdf"},"request":{"example_id":"chart/pdf_d2e0103f64cd_p29","product_type":"parse","source_file_path":"pdf_d2e0103f64cd_p29.pdf"},"started_at":"2026-11-04T19:45:18.959085"} +{"completed_at":"2026-11-04T19:45:26.113600","latency_in_ms":6021,"output":{"example_id":"chart/pdf_f270146f8ca7_p17","grounded_pages":[],"layout_pages":[],"markdown":"BY REBECCA FENG AND JOE WALLACE\n\nBY TIMOTHY W. MARTIN AND JIYOUNG SOHN\n\n# **Hyundai’s CEO Leads U.S. Push, Sets Sights On Robots**\n\nBY JON EMONT\n\nLastWeek: **S&P** 6939\\.03 À 0.34% **S&P FIN** À 0.69% **S&P IT** g 0.36% **DJ TRANS** À 0.55% **WSJ $ IDX** g 0.35% **2–YR. TREAS.** yield 3.527% **NIKKEI** 53322\\.85 g 0.97% **Seemore atWSJ.com/Markets**\n\n# BUSINESS&FINANCE\n\n**BUSINESS NEWS** Labor shortage hits broadband industry as demand for fiber-optic cable surges. **B9**\n\n**HEARD ON THE STREET** Luxury business is unlikely to rebound in China. **B10**\n\nTwo CEOs waited by the el- evators this summer in an of- fice building in California farm country where their compa- nies made their headquarters, on different floors. One ran a multinational salad company, the other a tiny bank with four branches. Their usual 8 a.m. small talk took a turn when Taylor Fresh Foods Chief Executive Bruce Taylor offered **Pacific Valley Bancorp** CEO Anker Fa- noe an investment of between $15 million and $20 million. “Would that help you guys?” Taylor asked. Fanoe thought it had to be a joke and moved on with his day. But truth be told, Fanoe\n\nthought his bank could use it. Pacific Valley had recently be- come the last locally owned and headquartered bank in Monterey County after com- petitors were eaten up by ac- quisitions, he said. Activist investors scouring the country for new targets in the industry had also rattled his bank. Pacific Valley’s profit had dropped double-digits af- ter costs jumped and a busi- ness loan went bad. Taylor Fresh Foods, which sells prepackaged salad prod- ucts and rakes in billions of dollars in revenue each year, announced a roughly 16% stake in Pacific Valley several months later. That made it the largest investor in the bank, with a seat on its board. Tay- lor was already one of its larg-\n\nest depositors and the owner of the five-story Taylor Build- ing where the bank leases its headquarters. The unusual investment in Pacific Valley by a salad com- pany underscores the mount- ing challenges that thousands of community banks around the country face. Pressures in- clude the emergence of com- petitors such as stablecoin and fintech firms, in addition to a steady rise in technology costs. “It’s hard to be a bank CEO and not be worried,” Fanoe said. Based in Salinas, Calif., a blue-collar city outside of Monterey, the floors of Pacific Valley are frequently covered in dirt that farmers track in, dressed in Wrangler jeans and\n\nwork boots. Its branches are staffed with bilingual speak- ers, with Salinas majority His- panic. When the rodeo comes to town, the bank puts out bales\n\nof hay and a saddle to cele- brate. Community banks like Pa- cific Valley often serve as stewards of local economies that rely on them for plain-va-\n\nnilla loans. These banks tend to win deeply loyal customers who keep deposits with them. At Pacific Valley, local knowledge and flexibility al- _PleaseturntopageB2_\n\nBY GINA HEEB\n\nBorder police stopped a car last week crossing from Hong Kong into the mainland Chi- nese city of Shenzhen. The sus- pected crime: silver smuggling. Two men in their 40s hid nearly 500 pounds of silver bars in 10 cookie tins, six milk-powder tins and nine crispy egg-roll boxes. The esti- mated market value was roughly $782,000, according to Hong Kong customs author- ities, who arrested the men and seized the bars.\n\nThe duo were likely headed for one of the largest bullion markets in China, the Shen- zhen Shuibei Gold and Jewelry Market, hoping to cash in on the silver boom. The market, where prices trade at a pre- mium compared with other countries, has become an epi- center of a metals mania that has captivated traders. Silver prices shot up more than 250% in the past year to as high as $114 a troy ounce in New York trading. At their peak, they were up 63% in January. In China, where the metal is needed for solar- panel production and coveted by speculators, silver futures rose to even higher levels, above $140 on Thursday. Silver’s surge crumbled on Friday, when prices plunged _PleaseturntopageB9_\n\n# **Silver Market’s Swings Rattle Companies**\n\n### Metal’s boom, slowed by a plunge, caused frenzy and pain for refiners and users\n\n2016 ’17 ’18 ’19 ’20 ’21 ’22 ’23 ’24 ’25\n\n0\n\n50\n\n100\n\n150\n\n200\n\n250million ounces Photovoltaics Jewelry\n\nSilver demand, by source\n\nJan. 2025 Jan. ’26\n\n3\\.0\n\n4\\.0\n\n5\\.0\n\n3\\.5\n\n4\\.5\n\n5\\.5 cents per watt Solar cells price\\*\n\n\\*Prices for TOPCon M10 cells in China\n\nFeb. 2024 ’25\n\nFront-month contract\n\n’26\n\n0\n\n20\n\n40\n\n60\n\n80\n\n100\n\n$120 a troy ounce\n\nSilver futures price\n\nSources: The Silver Institute (demand); OPIS (solar cells); Dow Jones Market Data (futures)\n\nFriday\n\n### $78.29\n\n▼31%\n\n# Generals Are Hot Commodity in Mining **INSIDE**\n\nGOYANG, South Korea—If the job of a car executive is to sell cars, it is hard to argue with the results of **Hyundai Motor** ’s José Muñoz. Last year, Hyundai sold about a million vehicles in the U.S.—39% more than when he joined the company in 2019— and more than four million worldwide. Yet Muñoz, the first non- Korean chief executive of the South Korean carmaker, seems to be worried that Hyundai is moving too slowly in an era of tariffs and technological change. He wants to shake up its hierarchical culture, expand into robots and flying cars, and build factories in the U.S. faster. “We’re focusing on acceler- ating,” he said of the com- pany’s roughly $26 billion plan to invest in the U.S., in an in- terview. “The sooner the bet- ter, so we can enjoy our in- vestments.” So far the stock market likes his gambles on high-tech moonshots. When Hyundai showed off its Atlas humanoid robot in January, the stock price skyrocketed. It is up 80% in just the past month. On Thursday, Hyundai re- ported a 22% drop in net in- come for 2025, stung by Presi- dent Trump’s tariffs on imported cars such as Korean- made Hyundais, although global revenue hit a record. Trump stoked uncertainty again last Monday by threat- ening to restore 25% U.S. tar- iffs on South Korea just three months after the two coun- tries had agreed to lower the rate to 15%. Muñoz, who has met Trump several times, said he believed the president understood Hyundai’s commitment to the U.S. In the unpredictable Trump age, the CEO wants his people to be ready for anything—and he reflects that in his manage- ment style. When he addressed roughly 1,000 local sales employees early this year in this Seoul suburb, aides had Korean translations of his English-lan- guage speech ready to display on a large screen. But Muñoz scrapped the canned speech. He grabbed a microphone and invited an interpreter to join him on stage, where he told the troops that 2026 couldn’t be just another year of busi- ness as usual. Muñoz, a 60-year-old Span- ish-U.S. dual national, didn’t start out as an auto-industry man. He earned a Ph.D. in nu- clear engineering in his native Spain during the 1980s and worked in the country’s nu- clear-energy sector. Back then, he didn’t even own a car. _PleaseturntopageB2_\n\nCompanies that mine criti- cal minerals are looking more like defense contractors these days. One sign: Many are hir- ing retired generals. Western miners have added at least a dozen former Ameri- can generals, colonels and ad- mirals to boards or to senior leadership roles over the past year, hoping to tap the billions of dollars the Defense Depart- ment is shelling out to rebuild domestic supply chains after China cut the U.S. off for much of last year. The retired generals guide the geologists and finance ex- ecutives who run mining com- panies through the less-famil- iar terrain of Washington. They make introductions to defense companies, scrutinize business deals with foreign partners to make sure they won’t raise\n\nflags with the Pentagon and give advice about whom to meet on Capitol Hill. **Almonty Industries** , a North American company that mines tungsten in Europe and South Korea, picked up two re-\n\ntired American generals last year. Retired four-star Army Gen. Gustave Perna, a logistics expert, joined the company’s board in March and retired Army Brig. Gen. Steven Allen enlisted as chief operations of-\n\nficer in December. Board positions can earn former generals tens of thou- sands to hundreds of thou- sands of dollars a year. Retired Gen. Richard B. Myers, a board member at rare-earth miner **MP Materials** , brought in $240,000 in 2024, based on fees and stock awards. **United States Antimony** , whose namesake metal is used to harden bullets, brought a re- tired four-star Army general, Jack Keane, onto its board in August, after a search that stretched for longer than a year to find the right former military official. **Critical Met- als** , a British Virgin Islands- based company developing a rare-earth mine in Greenland, last year appointed a retired four-star general and a retired rear admiral to an advisory board. _PleaseturntopageB2_\n\n# A Global Salad Chain Throws A Lifeline to Tiny Local Bank\n\nSTREETWISE | By James Mackintosh\n\n# **Parsing Boom and Bust in Gold, Silver** **W**\n\nhat if—and bear with me here—the wild gyrations in gold and silver prices are actually telling us some- thing? There are three plausible messages from the rise in gold prices. I’m not con- vinced that any of them provide a full explanation for what’s going on. But all of them have a grain of truth, and tell us something about how inves- tors in stocks, bonds and currencies are navigating a very uncertain time in the world\n\n**1) Investors want gold as a dollar alternative**\n\nCountries that fear falling afoul of Western sanctions have been buying gold rather than dollars for their foreign-exchange reserves for several years. But last year, central banks reduced their buying as prices jumped, according to the World Gold Council. Instead, buying lately has been dominated by private investors, especially via ex- change-traded funds. They might simply be scooping up gold expecting worried central banks will intensify\n\ntheir buying at a still-higher price, but at best that’s a hope. If they’re right, gold should be ris- ing as the dol- lar falls against other currencies. In fact, gold moved indepen- dently of the dollar day to day in the past year, al- though overall the dollar is down a lot while gold went wild. Treasury yields should also rise relative to other\n\ncountries as money leaves America. But U.S. 10- year Treasury yields have fallen a little since the start of last year, while those of Japan, France, Germany and the U.K. are all up, in some cases by a lot.\n\n**2) The ‘debasement trade’** An idea that appeals to a lot of those scarred by the last bout of inflation is that another is on the way,\n\ncaused by hefty government stimulus and a deliberate policy of weakening the dol- lar. A weak dollar is mechani- cally good for gold, since it is priced in dollars, while gold is often touted as a safe asset to sit out infla- tion. The Swiss franc, the clos- est thing to a haven cur- rency, also has performed well in the past couple of weeks, along with the low- debt oil-exposed Norwegian krone. Friday’s plunge in gold and full-blown crash in sil- _PleaseturntopageB9_\n\n### Friday’s plunge indicates investors were worried about debasement.\n\nRetired four-star Army Gen. Gustave Perna, left, and retired Army Brig. Gen. Steven Allen play roles at Almonty Industries.\n\nYONHAP NEWS/NEWSCOM/ZUMA PRESS\n\nCEO Bruce Taylor of Taylor Fresh Foods, which sells prepackaged salad products.\n\nCOLBY TARSITANO FOR WSJ\n\nPATRICK SEMANSKY/AP\n\nP2JW033000-0-B00100-1--------XA","pages":[{"line_numbers":[],"markdown":"BY REBECCA FENG AND JOE WALLACE\n\nBY TIMOTHY W. MARTIN AND JIYOUNG SOHN\n\n# **Hyundai’s CEO Leads U.S. Push, Sets Sights On Robots**\n\nBY JON EMONT\n\nLastWeek: **S&P** 6939\\.03 À 0.34% **S&P FIN** À 0.69% **S&P IT** g 0.36% **DJ TRANS** À 0.55% **WSJ $ IDX** g 0.35% **2–YR. TREAS.** yield 3.527% **NIKKEI** 53322\\.85 g 0.97% **Seemore atWSJ.com/Markets**\n\n# BUSINESS&FINANCE\n\n**BUSINESS NEWS** Labor shortage hits broadband industry as demand for fiber-optic cable surges. **B9**\n\n**HEARD ON THE STREET** Luxury business is unlikely to rebound in China. **B10**\n\nTwo CEOs waited by the el- evators this summer in an of- fice building in California farm country where their compa- nies made their headquarters, on different floors. One ran a multinational salad company, the other a tiny bank with four branches. Their usual 8 a.m. small talk took a turn when Taylor Fresh Foods Chief Executive Bruce Taylor offered **Pacific Valley Bancorp** CEO Anker Fa- noe an investment of between $15 million and $20 million. “Would that help you guys?” Taylor asked. Fanoe thought it had to be a joke and moved on with his day. But truth be told, Fanoe\n\nthought his bank could use it. Pacific Valley had recently be- come the last locally owned and headquartered bank in Monterey County after com- petitors were eaten up by ac- quisitions, he said. Activist investors scouring the country for new targets in the industry had also rattled his bank. Pacific Valley’s profit had dropped double-digits af- ter costs jumped and a busi- ness loan went bad. Taylor Fresh Foods, which sells prepackaged salad prod- ucts and rakes in billions of dollars in revenue each year, announced a roughly 16% stake in Pacific Valley several months later. That made it the largest investor in the bank, with a seat on its board. Tay- lor was already one of its larg-\n\nest depositors and the owner of the five-story Taylor Build- ing where the bank leases its headquarters. The unusual investment in Pacific Valley by a salad com- pany underscores the mount- ing challenges that thousands of community banks around the country face. Pressures in- clude the emergence of com- petitors such as stablecoin and fintech firms, in addition to a steady rise in technology costs. “It’s hard to be a bank CEO and not be worried,” Fanoe said. Based in Salinas, Calif., a blue-collar city outside of Monterey, the floors of Pacific Valley are frequently covered in dirt that farmers track in, dressed in Wrangler jeans and\n\nwork boots. Its branches are staffed with bilingual speak- ers, with Salinas majority His- panic. When the rodeo comes to town, the bank puts out bales\n\nof hay and a saddle to cele- brate. Community banks like Pa- cific Valley often serve as stewards of local economies that rely on them for plain-va-\n\nnilla loans. These banks tend to win deeply loyal customers who keep deposits with them. At Pacific Valley, local knowledge and flexibility al- _PleaseturntopageB2_\n\nBY GINA HEEB\n\nBorder police stopped a car last week crossing from Hong Kong into the mainland Chi- nese city of Shenzhen. The sus- pected crime: silver smuggling. Two men in their 40s hid nearly 500 pounds of silver bars in 10 cookie tins, six milk-powder tins and nine crispy egg-roll boxes. The esti- mated market value was roughly $782,000, according to Hong Kong customs author- ities, who arrested the men and seized the bars.\n\nThe duo were likely headed for one of the largest bullion markets in China, the Shen- zhen Shuibei Gold and Jewelry Market, hoping to cash in on the silver boom. The market, where prices trade at a pre- mium compared with other countries, has become an epi- center of a metals mania that has captivated traders. Silver prices shot up more than 250% in the past year to as high as $114 a troy ounce in New York trading. At their peak, they were up 63% in January. In China, where the metal is needed for solar- panel production and coveted by speculators, silver futures rose to even higher levels, above $140 on Thursday. Silver’s surge crumbled on Friday, when prices plunged _PleaseturntopageB9_\n\n# **Silver Market’s Swings Rattle Companies**\n\n### Metal’s boom, slowed by a plunge, caused frenzy and pain for refiners and users\n\n2016 ’17 ’18 ’19 ’20 ’21 ’22 ’23 ’24 ’25\n\n0\n\n50\n\n100\n\n150\n\n200\n\n250million ounces Photovoltaics Jewelry\n\nSilver demand, by source\n\nJan. 2025 Jan. ’26\n\n3\\.0\n\n4\\.0\n\n5\\.0\n\n3\\.5\n\n4\\.5\n\n5\\.5 cents per watt Solar cells price\\*\n\n\\*Prices for TOPCon M10 cells in China\n\nFeb. 2024 ’25\n\nFront-month contract\n\n��26\n\n0\n\n20\n\n40\n\n60\n\n80\n\n100\n\n$120 a troy ounce\n\nSilver futures price\n\nSources: The Silver Institute (demand); OPIS (solar cells); Dow Jones Market Data (futures)\n\nFriday\n\n### $78.29\n\n▼31%\n\n# Generals Are Hot Commodity in Mining **INSIDE**\n\nGOYANG, South Korea—If the job of a car executive is to sell cars, it is hard to argue with the results of **Hyundai Motor** ’s José Muñoz. Last year, Hyundai sold about a million vehicles in the U.S.—39% more than when he joined the company in 2019— and more than four million worldwide. Yet Muñoz, the first non- Korean chief executive of the South Korean carmaker, seems to be worried that Hyundai is moving too slowly in an era of tariffs and technological change. He wants to shake up its hierarchical culture, expand into robots and flying cars, and build factories in the U.S. faster. “We’re focusing on acceler- ating,” he said of the com- pany’s roughly $26 billion plan to invest in the U.S., in an in- terview. “The sooner the bet- ter, so we can enjoy our in- vestments.” So far the stock market likes his gambles on high-tech moonshots. When Hyundai showed off its Atlas humanoid robot in January, the stock price skyrocketed. It is up 80% in just the past month. On Thursday, Hyundai re- ported a 22% drop in net in- come for 2025, stung by Presi- dent Trump’s tariffs on imported cars such as Korean- made Hyundais, although global revenue hit a record. Trump stoked uncertainty again last Monday by threat- ening to restore 25% U.S. tar- iffs on South Korea just three months after the two coun- tries had agreed to lower the rate to 15%. Muñoz, who has met Trump several times, said he believed the president understood Hyundai’s commitment to the U.S. In the unpredictable Trump age, the CEO wants his people to be ready for anything—and he reflects that in his manage- ment style. When he addressed roughly 1,000 local sales employees early this year in this Seoul suburb, aides had Korean translations of his English-lan- guage speech ready to display on a large screen. But Muñoz scrapped the canned speech. He grabbed a microphone and invited an interpreter to join him on stage, where he told the troops that 2026 couldn’t be just another year of busi- ness as usual. Muñoz, a 60-year-old Span- ish-U.S. dual national, didn’t start out as an auto-industry man. He earned a Ph.D. in nu- clear engineering in his native Spain during the 1980s and worked in the country’s nu- clear-energy sector. Back then, he didn’t even own a car. _PleaseturntopageB2_\n\nCompanies that mine criti- cal minerals are looking more like defense contractors these days. One sign: Many are hir- ing retired generals. Western miners have added at least a dozen former Ameri- can generals, colonels and ad- mirals to boards or to senior leadership roles over the past year, hoping to tap the billions of dollars the Defense Depart- ment is shelling out to rebuild domestic supply chains after China cut the U.S. off for much of last year. The retired generals guide the geologists and finance ex- ecutives who run mining com- panies through the less-famil- iar terrain of Washington. They make introductions to defense companies, scrutinize business deals with foreign partners to make sure they won’t raise\n\nflags with the Pentagon and give advice about whom to meet on Capitol Hill. **Almonty Industries** , a North American company that mines tungsten in Europe and South Korea, picked up two re-\n\ntired American generals last year. Retired four-star Army Gen. Gustave Perna, a logistics expert, joined the company’s board in March and retired Army Brig. Gen. Steven Allen enlisted as chief operations of-\n\nficer in December. Board positions can earn former generals tens of thou- sands to hundreds of thou- sands of dollars a year. Retired Gen. Richard B. Myers, a board member at rare-earth miner **MP Materials** , brought in $240,000 in 2024, based on fees and stock awards. **United States Antimony** , whose namesake metal is used to harden bullets, brought a re- tired four-star Army general, Jack Keane, onto its board in August, after a search that stretched for longer than a year to find the right former military official. **Critical Met- als** , a British Virgin Islands- based company developing a rare-earth mine in Greenland, last year appointed a retired four-star general and a retired rear admiral to an advisory board. _PleaseturntopageB2_\n\n# A Global Salad Chain Throws A Lifeline to Tiny Local Bank\n\nSTREETWISE | By James Mackintosh\n\n# **Parsing Boom and Bust in Gold, Silver** **W**\n\nhat if—and bear with me here—the wild gyrations in gold and silver prices are actually telling us some- thing? There are three plausible messages from the rise in gold prices. I’m not con- vinced that any of them provide a full explanation for what’s going on. But all of them have a grain of truth, and tell us something about how inves- tors in stocks, bonds and currencies are navigating a very uncertain time in the world\n\n**1) Investors want gold as a dollar alternative**\n\nCountries that fear falling afoul of Western sanctions have been buying gold rather than dollars for their foreign-exchange reserves for several years. But last year, central banks reduced their buying as prices jumped, according to the World Gold Council. Instead, buying lately has been dominated by private investors, especially via ex- change-traded funds. They might simply be scooping up gold expecting worried central banks will intensify\n\ntheir buying at a still-higher price, but at best that’s a hope. If they’re right, gold should be ris- ing as the dol- lar falls against other currencies. In fact, gold moved indepen- dently of the dollar day to day in the past year, al- though overall the dollar is down a lot while gold went wild. Treasury yields should also rise relative to other\n\ncountries as money leaves America. But U.S. 10- year Treasury yields have fallen a little since the start of last year, while those of Japan, France, Germany and the U.K. are all up, in some cases by a lot.\n\n**2) The ‘debasement trade’** An idea that appeals to a lot of those scarred by the last bout of inflation is that another is on the way,\n\ncaused by hefty government stimulus and a deliberate policy of weakening the dol- lar. A weak dollar is mechani- cally good for gold, since it is priced in dollars, while gold is often touted as a safe asset to sit out infla- tion. The Swiss franc, the clos- est thing to a haven cur- rency, also has performed well in the past couple of weeks, along with the low- debt oil-exposed Norwegian krone. Friday’s plunge in gold and full-blown crash in sil- _PleaseturntopageB9_\n\n### Friday’s plunge indicates investors were worried about debasement.\n\nRetired four-star Army Gen. Gustave Perna, left, and retired Army Brig. Gen. Steven Allen play roles at Almonty Industries.\n\nYONHAP NEWS/NEWSCOM/ZUMA PRESS\n\nCEO Bruce Taylor of Taylor Fresh Foods, which sells prepackaged salad products.\n\nCOLBY TARSITANO FOR WSJ\n\nPATRICK SEMANSKY/AP\n\nP2JW033000-0-B00100-1--------XA","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"BY REBECCA FENG AND JOE WALLACE\n\nBY TIMOTHY W. MARTIN AND JIYOUNG SOHN\n\n# **Hyundai’s CEO Leads U.S. Push, Sets Sights On Robots**\n\nBY JON EMONT\n\nLastWeek: **S&P** 6939\\.03 À 0.34% **S&P FIN** À 0.69% **S&P IT** g 0.36% **DJ TRANS** À 0.55% **WSJ $ IDX** g 0.35% **2–YR. TREAS.** yield 3.527% **NIKKEI** 53322\\.85 g 0.97% **Seemore atWSJ.com/Markets**\n\n# BUSINESS&FINANCE\n\n**BUSINESS NEWS** Labor shortage hits broadband industry as demand for fiber-optic cable surges. **B9**\n\n**HEARD ON THE STREET** Luxury business is unlikely to rebound in China. **B10**\n\nTwo CEOs waited by the el- evators this summer in an of- fice building in California farm country where their compa- nies made their headquarters, on different floors. One ran a multinational salad company, the other a tiny bank with four branches. Their usual 8 a.m. small talk took a turn when Taylor Fresh Foods Chief Executive Bruce Taylor offered **Pacific Valley Bancorp** CEO Anker Fa- noe an investment of between $15 million and $20 million. “Would that help you guys?” Taylor asked. Fanoe thought it had to be a joke and moved on with his day. But truth be told, Fanoe\n\nthought his bank could use it. Pacific Valley had recently be- come the last locally owned and headquartered bank in Monterey County after com- petitors were eaten up by ac- quisitions, he said. Activist investors scouring the country for new targets in the industry had also rattled his bank. Pacific Valley’s profit had dropped double-digits af- ter costs jumped and a busi- ness loan went bad. Taylor Fresh Foods, which sells prepackaged salad prod- ucts and rakes in billions of dollars in revenue each year, announced a roughly 16% stake in Pacific Valley several months later. That made it the largest investor in the bank, with a seat on its board. Tay- lor was already one of its larg-\n\nest depositors and the owner of the five-story Taylor Build- ing where the bank leases its headquarters. The unusual investment in Pacific Valley by a salad com- pany underscores the mount- ing challenges that thousands of community banks around the country face. Pressures in- clude the emergence of com- petitors such as stablecoin and fintech firms, in addition to a steady rise in technology costs. “It’s hard to be a bank CEO and not be worried,” Fanoe said. Based in Salinas, Calif., a blue-collar city outside of Monterey, the floors of Pacific Valley are frequently covered in dirt that farmers track in, dressed in Wrangler jeans and\n\nwork boots. Its branches are staffed with bilingual speak- ers, with Salinas majority His- panic. When the rodeo comes to town, the bank puts out bales\n\nof hay and a saddle to cele- brate. Community banks like Pa- cific Valley often serve as stewards of local economies that rely on them for plain-va-\n\nnilla loans. These banks tend to win deeply loyal customers who keep deposits with them. At Pacific Valley, local knowledge and flexibility al- _PleaseturntopageB2_\n\nBY GINA HEEB\n\nBorder police stopped a car last week crossing from Hong Kong into the mainland Chi- nese city of Shenzhen. The sus- pected crime: silver smuggling. Two men in their 40s hid nearly 500 pounds of silver bars in 10 cookie tins, six milk-powder tins and nine crispy egg-roll boxes. The esti- mated market value was roughly $782,000, according to Hong Kong customs author- ities, who arrested the men and seized the bars.\n\nThe duo were likely headed for one of the largest bullion markets in China, the Shen- zhen Shuibei Gold and Jewelry Market, hoping to cash in on the silver boom. The market, where prices trade at a pre- mium compared with other countries, has become an epi- center of a metals mania that has captivated traders. Silver prices shot up more than 250% in the past year to as high as $114 a troy ounce in New York trading. At their peak, they were up 63% in January. In China, where the metal is needed for solar- panel production and coveted by speculators, silver futures rose to even higher levels, above $140 on Thursday. Silver’s surge crumbled on Friday, when prices plunged _PleaseturntopageB9_\n\n# **Silver Market’s Swings Rattle Companies**\n\n### Metal’s boom, slowed by a plunge, caused frenzy and pain for refiners and users\n\n2016 ’17 ’18 ’19 ’20 ’21 ’22 ’23 ’24 ’25\n\n0\n\n50\n\n100\n\n150\n\n200\n\n250million ounces Photovoltaics Jewelry\n\nSilver demand, by source\n\nJan. 2025 Jan. ’26\n\n3\\.0\n\n4\\.0\n\n5\\.0\n\n3\\.5\n\n4\\.5\n\n5\\.5 cents per watt Solar cells price\\*\n\n\\*Prices for TOPCon M10 cells in China\n\nFeb. 2024 ’25\n\nFront-month contract\n\n’26\n\n0\n\n20\n\n40\n\n60\n\n80\n\n100\n\n$120 a troy ounce\n\nSilver futures price\n\nSources: The Silver Institute (demand); OPIS (solar cells); Dow Jones Market Data (futures)\n\nFriday\n\n### $78.29\n\n▼31%\n\n# Generals Are Hot Commodity in Mining **INSIDE**\n\nGOYANG, South Korea—If the job of a car executive is to sell cars, it is hard to argue with the results of **Hyundai Motor** ’s José Muñoz. Last year, Hyundai sold about a million vehicles in the U.S.—39% more than when he joined the company in 2019— and more than four million worldwide. Yet Muñoz, the first non- Korean chief executive of the South Korean carmaker, seems to be worried that Hyundai is moving too slowly in an era of tariffs and technological change. He wants to shake up its hierarchical culture, expand into robots and flying cars, and build factories in the U.S. faster. “We’re focusing on acceler- ating,” he said of the com- pany’s roughly $26 billion plan to invest in the U.S., in an in- terview. “The sooner the bet- ter, so we can enjoy our in- vestments.” So far the stock market likes his gambles on high-tech moonshots. When Hyundai showed off its Atlas humanoid robot in January, the stock price skyrocketed. It is up 80% in just the past month. On Thursday, Hyundai re- ported a 22% drop in net in- come for 2025, stung by Presi- dent Trump’s tariffs on imported cars such as Korean- made Hyundais, although global revenue hit a record. Trump stoked uncertainty again last Monday by threat- ening to restore 25% U.S. tar- iffs on South Korea just three months after the two coun- tries had agreed to lower the rate to 15%. Muñoz, who has met Trump several times, said he believed the president understood Hyundai’s commitment to the U.S. In the unpredictable Trump age, the CEO wants his people to be ready for anything—and he reflects that in his manage- ment style. When he addressed roughly 1,000 local sales employees early this year in this Seoul suburb, aides had Korean translations of his English-lan- guage speech ready to display on a large screen. But Muñoz scrapped the canned speech. He grabbed a microphone and invited an interpreter to join him on stage, where he told the troops that 2026 couldn’t be just another year of busi- ness as usual. Muñoz, a 60-year-old Span- ish-U.S. dual national, didn’t start out as an auto-industry man. He earned a Ph.D. in nu- clear engineering in his native Spain during the 1980s and worked in the country’s nu- clear-energy sector. Back then, he didn’t even own a car. _PleaseturntopageB2_\n\nCompanies that mine criti- cal minerals are looking more like defense contractors these days. One sign: Many are hir- ing retired generals. Western miners have added at least a dozen former Ameri- can generals, colonels and ad- mirals to boards or to senior leadership roles over the past year, hoping to tap the billions of dollars the Defense Depart- ment is shelling out to rebuild domestic supply chains after China cut the U.S. off for much of last year. The retired generals guide the geologists and finance ex- ecutives who run mining com- panies through the less-famil- iar terrain of Washington. They make introductions to defense companies, scrutinize business deals with foreign partners to make sure they won’t raise\n\nflags with the Pentagon and give advice about whom to meet on Capitol Hill. **Almonty Industries** , a North American company that mines tungsten in Europe and South Korea, picked up two re-\n\ntired American generals last year. Retired four-star Army Gen. Gustave Perna, a logistics expert, joined the company’s board in March and retired Army Brig. Gen. Steven Allen enlisted as chief operations of-\n\nficer in December. Board positions can earn former generals tens of thou- sands to hundreds of thou- sands of dollars a year. Retired Gen. Richard B. Myers, a board member at rare-earth miner **MP Materials** , brought in $240,000 in 2024, based on fees and stock awards. **United States Antimony** , whose namesake metal is used to harden bullets, brought a re- tired four-star Army general, Jack Keane, onto its board in August, after a search that stretched for longer than a year to find the right former military official. **Critical Met- als** , a British Virgin Islands- based company developing a rare-earth mine in Greenland, last year appointed a retired four-star general and a retired rear admiral to an advisory board. _PleaseturntopageB2_\n\n# A Global Salad Chain Throws A Lifeline to Tiny Local Bank\n\nSTREETWISE | By James Mackintosh\n\n# **Parsing Boom and Bust in Gold, Silver** **W**\n\nhat if—and bear with me here—the wild gyrations in gold and silver prices are actually telling us some- thing? There are three plausible messages from the rise in gold prices. I’m not con- vinced that any of them provide a full explanation for what’s going on. But all of them have a grain of truth, and tell us something about how inves- tors in stocks, bonds and currencies are navigating a very uncertain time in the world\n\n**1) Investors want gold as a dollar alternative**\n\nCountries that fear falling afoul of Western sanctions have been buying gold rather than dollars for their foreign-exchange reserves for several years. But last year, central banks reduced their buying as prices jumped, according to the World Gold Council. Instead, buying lately has been dominated by private investors, especially via ex- change-traded funds. They might simply be scooping up gold expecting worried central banks will intensify\n\ntheir buying at a still-higher price, but at best that’s a hope. If they’re right, gold should be ris- ing as the dol- lar falls against other currencies. In fact, gold moved indepen- dently of the dollar day to day in the past year, al- though overall the dollar is down a lot while gold went wild. Treasury yields should also rise relative to other\n\ncountries as money leaves America. But U.S. 10- year Treasury yields have fallen a little since the start of last year, while those of Japan, France, Germany and the U.K. are all up, in some cases by a lot.\n\n**2) The ‘debasement trade’** An idea that appeals to a lot of those scarred by the last bout of inflation is that another is on the way,\n\ncaused by hefty government stimulus and a deliberate policy of weakening the dol- lar. A weak dollar is mechani- cally good for gold, since it is priced in dollars, while gold is often touted as a safe asset to sit out infla- tion. The Swiss franc, the clos- est thing to a haven cur- rency, also has performed well in the past couple of weeks, along with the low- debt oil-exposed Norwegian krone. Friday’s plunge in gold and full-blown crash in sil- _PleaseturntopageB9_\n\n### Friday’s plunge indicates investors were worried about debasement.\n\nRetired four-star Army Gen. Gustave Perna, left, and retired Army Brig. Gen. Steven Allen play roles at Almonty Industries.\n\nYONHAP NEWS/NEWSCOM/ZUMA PRESS\n\nCEO Bruce Taylor of Taylor Fresh Foods, which sells prepackaged salad products.\n\nCOLBY TARSITANO FOR WSJ\n\nPATRICK SEMANSKY/AP\n\nP2JW033000-0-B00100-1--------XA","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/pdf_f270146f8ca7_p17.pdf"]},"response":{"errors":[],"extract_id":"extract_7e01d7d7c8d5c7a80aed44cd6eedc66d","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["BY REBECCA FENG AND JOE WALLACE\n\nBY TIMOTHY W. MARTIN AND JIYOUNG SOHN\n\n# **Hyundai’s CEO Leads U.S. Push, Sets Sights On Robots**\n\nBY JON EMONT\n\nLastWeek: **S&P** 6939\\.03 À 0.34% **S&P FIN** À 0.69% **S&P IT** g 0.36% **DJ TRANS** À 0.55% **WSJ $ IDX** g 0.35% **2–YR. TREAS.** yield 3.527% **NIKKEI** 53322\\.85 g 0.97% **Seemore atWSJ.com/Markets**\n\n# BUSINESS&FINANCE\n\n**BUSINESS NEWS** Labor shortage hits broadband industry as demand for fiber-optic cable surges. **B9**\n\n**HEARD ON THE STREET** Luxury business is unlikely to rebound in China. **B10**\n\nTwo CEOs waited by the el- evators this summer in an of- fice building in California farm country where their compa- nies made their headquarters, on different floors. One ran a multinational salad company, the other a tiny bank with four branches. Their usual 8 a.m. small talk took a turn when Taylor Fresh Foods Chief Executive Bruce Taylor offered **Pacific Valley Bancorp** CEO Anker Fa- noe an investment of between $15 million and $20 million. “Would that help you guys?” Taylor asked. Fanoe thought it had to be a joke and moved on with his day. But truth be told, Fanoe\n\nthought his bank could use it. Pacific Valley had recently be- come the last locally owned and headquartered bank in Monterey County after com- petitors were eaten up by ac- quisitions, he said. Activist investors scouring the country for new targets in the industry had also rattled his bank. Pacific Valley’s profit had dropped double-digits af- ter costs jumped and a busi- ness loan went bad. Taylor Fresh Foods, which sells prepackaged salad prod- ucts and rakes in billions of dollars in revenue each year, announced a roughly 16% stake in Pacific Valley several months later. That made it the largest investor in the bank, with a seat on its board. Tay- lor was already one of its larg-\n\nest depositors and the owner of the five-story Taylor Build- ing where the bank leases its headquarters. The unusual investment in Pacific Valley by a salad com- pany underscores the mount- ing challenges that thousands of community banks around the country face. Pressures in- clude the emergence of com- petitors such as stablecoin and fintech firms, in addition to a steady rise in technology costs. “It’s hard to be a bank CEO and not be worried,” Fanoe said. Based in Salinas, Calif., a blue-collar city outside of Monterey, the floors of Pacific Valley are frequently covered in dirt that farmers track in, dressed in Wrangler jeans and\n\nwork boots. Its branches are staffed with bilingual speak- ers, with Salinas majority His- panic. When the rodeo comes to town, the bank puts out bales\n\nof hay and a saddle to cele- brate. Community banks like Pa- cific Valley often serve as stewards of local economies that rely on them for plain-va-\n\nnilla loans. These banks tend to win deeply loyal customers who keep deposits with them. At Pacific Valley, local knowledge and flexibility al- _PleaseturntopageB2_\n\nBY GINA HEEB\n\nBorder police stopped a car last week crossing from Hong Kong into the mainland Chi- nese city of Shenzhen. The sus- pected crime: silver smuggling. Two men in their 40s hid nearly 500 pounds of silver bars in 10 cookie tins, six milk-powder tins and nine crispy egg-roll boxes. The esti- mated market value was roughly $782,000, according to Hong Kong customs author- ities, who arrested the men and seized the bars.\n\nThe duo were likely headed for one of the largest bullion markets in China, the Shen- zhen Shuibei Gold and Jewelry Market, hoping to cash in on the silver boom. The market, where prices trade at a pre- mium compared with other countries, has become an epi- center of a metals mania that has captivated traders. Silver prices shot up more than 250% in the past year to as high as $114 a troy ounce in New York trading. At their peak, they were up 63% in January. In China, where the metal is needed for solar- panel production and coveted by speculators, silver futures rose to even higher levels, above $140 on Thursday. Silver’s surge crumbled on Friday, when prices plunged _PleaseturntopageB9_\n\n# **Silver Market’s Swings Rattle Companies**\n\n### Metal’s boom, slowed by a plunge, caused frenzy and pain for refiners and users\n\n2016 ’17 ’18 ’19 ’20 ’21 ’22 ’23 ’24 ’25\n\n0\n\n50\n\n100\n\n150\n\n200\n\n250million ounces Photovoltaics Jewelry\n\nSilver demand, by source\n\nJan. 2025 Jan. ’26\n\n3\\.0\n\n4\\.0\n\n5\\.0\n\n3\\.5\n\n4\\.5\n\n5\\.5 cents per watt Solar cells price\\*\n\n\\*Prices for TOPCon M10 cells in China\n\nFeb. 2024 ’25\n\nFront-month contract\n\n’26\n\n0\n\n20\n\n40\n\n60\n\n80\n\n100\n\n$120 a troy ounce\n\nSilver futures price\n\nSources: The Silver Institute (demand); OPIS (solar cells); Dow Jones Market Data (futures)\n\nFriday\n\n### $78.29\n\n▼31%\n\n# Generals Are Hot Commodity in Mining **INSIDE**\n\nGOYANG, South Korea—If the job of a car executive is to sell cars, it is hard to argue with the results of **Hyundai Motor** ’s José Muñoz. Last year, Hyundai sold about a million vehicles in the U.S.—39% more than when he joined the company in 2019— and more than four million worldwide. Yet Muñoz, the first non- Korean chief executive of the South Korean carmaker, seems to be worried that Hyundai is moving too slowly in an era of tariffs and technological change. He wants to shake up its hierarchical culture, expand into robots and flying cars, and build factories in the U.S. faster. “We’re focusing on acceler- ating,” he said of the com- pany’s roughly $26 billion plan to invest in the U.S., in an in- terview. “The sooner the bet- ter, so we can enjoy our in- vestments.” So far the stock market likes his gambles on high-tech moonshots. When Hyundai showed off its Atlas humanoid robot in January, the stock price skyrocketed. It is up 80% in just the past month. On Thursday, Hyundai re- ported a 22% drop in net in- come for 2025, stung by Presi- dent Trump’s tariffs on imported cars such as Korean- made Hyundais, although global revenue hit a record. Trump stoked uncertainty again last Monday by threat- ening to restore 25% U.S. tar- iffs on South Korea just three months after the two coun- tries had agreed to lower the rate to 15%. Muñoz, who has met Trump several times, said he believed the president understood Hyundai’s commitment to the U.S. In the unpredictable Trump age, the CEO wants his people to be ready for anything—and he reflects that in his manage- ment style. When he addressed roughly 1,000 local sales employees early this year in this Seoul suburb, aides had Korean translations of his English-lan- guage speech ready to display on a large screen. But Muñoz scrapped the canned speech. He grabbed a microphone and invited an interpreter to join him on stage, where he told the troops that 2026 couldn’t be just another year of busi- ness as usual. Muñoz, a 60-year-old Span- ish-U.S. dual national, didn’t start out as an auto-industry man. He earned a Ph.D. in nu- clear engineering in his native Spain during the 1980s and worked in the country’s nu- clear-energy sector. Back then, he didn’t even own a car. _PleaseturntopageB2_\n\nCompanies that mine criti- cal minerals are looking more like defense contractors these days. One sign: Many are hir- ing retired generals. Western miners have added at least a dozen former Ameri- can generals, colonels and ad- mirals to boards or to senior leadership roles over the past year, hoping to tap the billions of dollars the Defense Depart- ment is shelling out to rebuild domestic supply chains after China cut the U.S. off for much of last year. The retired generals guide the geologists and finance ex- ecutives who run mining com- panies through the less-famil- iar terrain of Washington. They make introductions to defense companies, scrutinize business deals with foreign partners to make sure they won’t raise\n\nflags with the Pentagon and give advice about whom to meet on Capitol Hill. **Almonty Industries** , a North American company that mines tungsten in Europe and South Korea, picked up two re-\n\ntired American generals last year. Retired four-star Army Gen. Gustave Perna, a logistics expert, joined the company’s board in March and retired Army Brig. Gen. Steven Allen enlisted as chief operations of-\n\nficer in December. Board positions can earn former generals tens of thou- sands to hundreds of thou- sands of dollars a year. Retired Gen. Richard B. Myers, a board member at rare-earth miner **MP Materials** , brought in $240,000 in 2024, based on fees and stock awards. **United States Antimony** , whose namesake metal is used to harden bullets, brought a re- tired four-star Army general, Jack Keane, onto its board in August, after a search that stretched for longer than a year to find the right former military official. **Critical Met- als** , a British Virgin Islands- based company developing a rare-earth mine in Greenland, last year appointed a retired four-star general and a retired rear admiral to an advisory board. _PleaseturntopageB2_\n\n# A Global Salad Chain Throws A Lifeline to Tiny Local Bank\n\nSTREETWISE | By James Mackintosh\n\n# **Parsing Boom and Bust in Gold, Silver** **W**\n\nhat if—and bear with me here—the wild gyrations in gold and silver prices are actually telling us some- thing? There are three plausible messages from the rise in gold prices. I’m not con- vinced that any of them provide a full explanation for what’s going on. But all of them have a grain of truth, and tell us something about how inves- tors in stocks, bonds and currencies are navigating a very uncertain time in the world\n\n**1) Investors want gold as a dollar alternative**\n\nCountries that fear falling afoul of Western sanctions have been buying gold rather than dollars for their foreign-exchange reserves for several years. But last year, central banks reduced their buying as prices jumped, according to the World Gold Council. Instead, buying lately has been dominated by private investors, especially via ex- change-traded funds. They might simply be scooping up gold expecting worried central banks will intensify\n\ntheir buying at a still-higher price, but at best that’s a hope. If they’re right, gold should be ris- ing as the dol- lar falls against other currencies. In fact, gold moved indepen- dently of the dollar day to day in the past year, al- though overall the dollar is down a lot while gold went wild. Treasury yields should also rise relative to other\n\ncountries as money leaves America. But U.S. 10- year Treasury yields have fallen a little since the start of last year, while those of Japan, France, Germany and the U.K. are all up, in some cases by a lot.\n\n**2) The ‘debasement trade’** An idea that appeals to a lot of those scarred by the last bout of inflation is that another is on the way,\n\ncaused by hefty government stimulus and a deliberate policy of weakening the dol- lar. A weak dollar is mechani- cally good for gold, since it is priced in dollars, while gold is often touted as a safe asset to sit out infla- tion. The Swiss franc, the clos- est thing to a haven cur- rency, also has performed well in the past couple of weeks, along with the low- debt oil-exposed Norwegian krone. Friday’s plunge in gold and full-blown crash in sil- _PleaseturntopageB9_\n\n### Friday’s plunge indicates investors were worried about debasement.\n\nRetired four-star Army Gen. Gustave Perna, left, and retired Army Brig. Gen. Steven Allen play roles at Almonty Industries.\n\nYONHAP NEWS/NEWSCOM/ZUMA PRESS\n\nCEO Bruce Taylor of Taylor Fresh Foods, which sells prepackaged salad products.\n\nCOLBY TARSITANO FOR WSJ\n\nPATRICK SEMANSKY/AP\n\nP2JW033000-0-B00100-1--------XA"],"full_content":"BY REBECCA FENG AND JOE WALLACE\n\nBY TIMOTHY W. MARTIN AND JIYOUNG SOHN\n\n# **Hyundai’s CEO Leads U.S. Push, Sets Sights On Robots**\n\nBY JON EMONT\n\nLastWeek: **S&P** 6939\\.03 À 0.34% **S&P FIN** À 0.69% **S&P IT** g 0.36% **DJ TRANS** À 0.55% **WSJ $ IDX** g 0.35% **2–YR. TREAS.** yield 3.527% **NIKKEI** 53322\\.85 g 0.97% **Seemore atWSJ.com/Markets**\n\n# BUSINESS&FINANCE\n\n**BUSINESS NEWS** Labor shortage hits broadband industry as demand for fiber-optic cable surges. **B9**\n\n**HEARD ON THE STREET** Luxury business is unlikely to rebound in China. **B10**\n\nTwo CEOs waited by the el- evators this summer in an of- fice building in California farm country where their compa- nies made their headquarters, on different floors. One ran a multinational salad company, the other a tiny bank with four branches. Their usual 8 a.m. small talk took a turn when Taylor Fresh Foods Chief Executive Bruce Taylor offered **Pacific Valley Bancorp** CEO Anker Fa- noe an investment of between $15 million and $20 million. “Would that help you guys?” Taylor asked. Fanoe thought it had to be a joke and moved on with his day. But truth be told, Fanoe\n\nthought his bank could use it. Pacific Valley had recently be- come the last locally owned and headquartered bank in Monterey County after com- petitors were eaten up by ac- quisitions, he said. Activist investors scouring the country for new targets in the industry had also rattled his bank. Pacific Valley’s profit had dropped double-digits af- ter costs jumped and a busi- ness loan went bad. Taylor Fresh Foods, which sells prepackaged salad prod- ucts and rakes in billions of dollars in revenue each year, announced a roughly 16% stake in Pacific Valley several months later. That made it the largest investor in the bank, with a seat on its board. Tay- lor was already one of its larg-\n\nest depositors and the owner of the five-story Taylor Build- ing where the bank leases its headquarters. The unusual investment in Pacific Valley by a salad com- pany underscores the mount- ing challenges that thousands of community banks around the country face. Pressures in- clude the emergence of com- petitors such as stablecoin and fintech firms, in addition to a steady rise in technology costs. “It’s hard to be a bank CEO and not be worried,” Fanoe said. Based in Salinas, Calif., a blue-collar city outside of Monterey, the floors of Pacific Valley are frequently covered in dirt that farmers track in, dressed in Wrangler jeans and\n\nwork boots. Its branches are staffed with bilingual speak- ers, with Salinas majority His- panic. When the rodeo comes to town, the bank puts out bales\n\nof hay and a saddle to cele- brate. Community banks like Pa- cific Valley often serve as stewards of local economies that rely on them for plain-va-\n\nnilla loans. These banks tend to win deeply loyal customers who keep deposits with them. At Pacific Valley, local knowledge and flexibility al- _PleaseturntopageB2_\n\nBY GINA HEEB\n\nBorder police stopped a car last week crossing from Hong Kong into the mainland Chi- nese city of Shenzhen. The sus- pected crime: silver smuggling. Two men in their 40s hid nearly 500 pounds of silver bars in 10 cookie tins, six milk-powder tins and nine crispy egg-roll boxes. The esti- mated market value was roughly $782,000, according to Hong Kong customs author- ities, who arrested the men and seized the bars.\n\nThe duo were likely headed for one of the largest bullion markets in China, the Shen- zhen Shuibei Gold and Jewelry Market, hoping to cash in on the silver boom. The market, where prices trade at a pre- mium compared with other countries, has become an epi- center of a metals mania that has captivated traders. Silver prices shot up more than 250% in the past year to as high as $114 a troy ounce in New York trading. At their peak, they were up 63% in January. In China, where the metal is needed for solar- panel production and coveted by speculators, silver futures rose to even higher levels, above $140 on Thursday. Silver’s surge crumbled on Friday, when prices plunged _PleaseturntopageB9_\n\n# **Silver Market’s Swings Rattle Companies**\n\n### Metal’s boom, slowed by a plunge, caused frenzy and pain for refiners and users\n\n2016 ’17 ’18 ’19 ’20 ’21 ’22 ’23 ’24 ’25\n\n0\n\n50\n\n100\n\n150\n\n200\n\n250million ounces Photovoltaics Jewelry\n\nSilver demand, by source\n\nJan. 2025 Jan. ’26\n\n3\\.0\n\n4\\.0\n\n5\\.0\n\n3\\.5\n\n4\\.5\n\n5\\.5 cents per watt Solar cells price\\*\n\n\\*Prices for TOPCon M10 cells in China\n\nFeb. 2024 ’25\n\nFront-month contract\n\n’26\n\n0\n\n20\n\n40\n\n60\n\n80\n\n100\n\n$120 a troy ounce\n\nSilver futures price\n\nSources: The Silver Institute (demand); OPIS (solar cells); Dow Jones Market Data (futures)\n\nFriday\n\n### $78.29\n\n▼31%\n\n# Generals Are Hot Commodity in Mining **INSIDE**\n\nGOYANG, South Korea—If the job of a car executive is to sell cars, it is hard to argue with the results of **Hyundai Motor** ’s José Muñoz. Last year, Hyundai sold about a million vehicles in the U.S.—39% more than when he joined the company in 2019— and more than four million worldwide. Yet Muñoz, the first non- Korean chief executive of the South Korean carmaker, seems to be worried that Hyundai is moving too slowly in an era of tariffs and technological change. He wants to shake up its hierarchical culture, expand into robots and flying cars, and build factories in the U.S. faster. “We’re focusing on acceler- ating,” he said of the com- pany’s roughly $26 billion plan to invest in the U.S., in an in- terview. “The sooner the bet- ter, so we can enjoy our in- vestments.” So far the stock market likes his gambles on high-tech moonshots. When Hyundai showed off its Atlas humanoid robot in January, the stock price skyrocketed. It is up 80% in just the past month. On Thursday, Hyundai re- ported a 22% drop in net in- come for 2025, stung by Presi- dent Trump’s tariffs on imported cars such as Korean- made Hyundais, although global revenue hit a record. Trump stoked uncertainty again last Monday by threat- ening to restore 25% U.S. tar- iffs on South Korea just three months after the two coun- tries had agreed to lower the rate to 15%. Muñoz, who has met Trump several times, said he believed the president understood Hyundai’s commitment to the U.S. In the unpredictable Trump age, the CEO wants his people to be ready for anything—and he reflects that in his manage- ment style. When he addressed roughly 1,000 local sales employees early this year in this Seoul suburb, aides had Korean translations of his English-lan- guage speech ready to display on a large screen. But Muñoz scrapped the canned speech. He grabbed a microphone and invited an interpreter to join him on stage, where he told the troops that 2026 couldn’t be just another year of busi- ness as usual. Muñoz, a 60-year-old Span- ish-U.S. dual national, didn’t start out as an auto-industry man. He earned a Ph.D. in nu- clear engineering in his native Spain during the 1980s and worked in the country’s nu- clear-energy sector. Back then, he didn’t even own a car. _PleaseturntopageB2_\n\nCompanies that mine criti- cal minerals are looking more like defense contractors these days. One sign: Many are hir- ing retired generals. Western miners have added at least a dozen former Ameri- can generals, colonels and ad- mirals to boards or to senior leadership roles over the past year, hoping to tap the billions of dollars the Defense Depart- ment is shelling out to rebuild domestic supply chains after China cut the U.S. off for much of last year. The retired generals guide the geologists and finance ex- ecutives who run mining com- panies through the less-famil- iar terrain of Washington. They make introductions to defense companies, scrutinize business deals with foreign partners to make sure they won’t raise\n\nflags with the Pentagon and give advice about whom to meet on Capitol Hill. **Almonty Industries** , a North American company that mines tungsten in Europe and South Korea, picked up two re-\n\ntired American generals last year. Retired four-star Army Gen. Gustave Perna, a logistics expert, joined the company’s board in March and retired Army Brig. Gen. Steven Allen enlisted as chief operations of-\n\nficer in December. Board positions can earn former generals tens of thou- sands to hundreds of thou- sands of dollars a year. Retired Gen. Richard B. Myers, a board member at rare-earth miner **MP Materials** , brought in $240,000 in 2024, based on fees and stock awards. **United States Antimony** , whose namesake metal is used to harden bullets, brought a re- tired four-star Army general, Jack Keane, onto its board in August, after a search that stretched for longer than a year to find the right former military official. **Critical Met- als** , a British Virgin Islands- based company developing a rare-earth mine in Greenland, last year appointed a retired four-star general and a retired rear admiral to an advisory board. _PleaseturntopageB2_\n\n# A Global Salad Chain Throws A Lifeline to Tiny Local Bank\n\nSTREETWISE | By James Mackintosh\n\n# **Parsing Boom and Bust in Gold, Silver** **W**\n\nhat if—and bear with me here—the wild gyrations in gold and silver prices are actually telling us some- thing? There are three plausible messages from the rise in gold prices. I’m not con- vinced that any of them provide a full explanation for what’s going on. But all of them have a grain of truth, and tell us something about how inves- tors in stocks, bonds and currencies are navigating a very uncertain time in the world\n\n**1) Investors want gold as a dollar alternative**\n\nCountries that fear falling afoul of Western sanctions have been buying gold rather than dollars for their foreign-exchange reserves for several years. But last year, central banks reduced their buying as prices jumped, according to the World Gold Council. Instead, buying lately has been dominated by private investors, especially via ex- change-traded funds. They might simply be scooping up gold expecting worried central banks will intensify\n\ntheir buying at a still-higher price, but at best that’s a hope. If they’re right, gold should be ris- ing as the dol- lar falls against other currencies. In fact, gold moved indepen- dently of the dollar day to day in the past year, al- though overall the dollar is down a lot while gold went wild. Treasury yields should also rise relative to other\n\ncountries as money leaves America. But U.S. 10- year Treasury yields have fallen a little since the start of last year, while those of Japan, France, Germany and the U.K. are all up, in some cases by a lot.\n\n**2) The ‘debasement trade’** An idea that appeals to a lot of those scarred by the last bout of inflation is that another is on the way,\n\ncaused by hefty government stimulus and a deliberate policy of weakening the dol- lar. A weak dollar is mechani- cally good for gold, since it is priced in dollars, while gold is often touted as a safe asset to sit out infla- tion. The Swiss franc, the clos- est thing to a haven cur- rency, also has performed well in the past couple of weeks, along with the low- debt oil-exposed Norwegian krone. Friday’s plunge in gold and full-blown crash in sil- _PleaseturntopageB9_\n\n### Friday’s plunge indicates investors were worried about debasement.\n\nRetired four-star Army Gen. Gustave Perna, left, and retired Army Brig. Gen. Steven Allen play roles at Almonty Industries.\n\nYONHAP NEWS/NEWSCOM/ZUMA PRESS\n\nCEO Bruce Taylor of Taylor Fresh Foods, which sells prepackaged salad products.\n\nCOLBY TARSITANO FOR WSJ\n\nPATRICK SEMANSKY/AP\n\nP2JW033000-0-B00100-1--------XA","publish_date":null,"title":"BUSINESS & FINANCE","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/pdf_f270146f8ca7_p17.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/pdf_f270146f8ca7_p17.pdf"},"request":{"example_id":"chart/pdf_f270146f8ca7_p17","product_type":"parse","source_file_path":"pdf_f270146f8ca7_p17.pdf"},"started_at":"2026-11-04T19:45:20.092580"} +{"completed_at":"2026-11-04T19:45:24.742813","latency_in_ms":3488,"output":{"example_id":"chart/pdf_f270146f8ca7_p25","grounded_pages":[],"layout_pages":[],"markdown":"**THE WALL STREET JOURNAL.** Monday, February 2, 2026 | B9\n\nBY PATIENCE HAGGIN\n\nA JKL Associates work site where fiber-optic cabling is being fed through a conduit.\n\nSilver is now the largest cost component of solar cells.\n\nCHINATOPIX/AP\n\n### **Monday**\n\nISM mfg. index Dec., previous 51.9 Jan., expected n.a.\n\nEarningsexpected Estimate/Year Ago\n\nIDEXX Laboratories 2.93/2.62 NXPSemiconductors 3.31/3.18 PalantirTechnologies 0.23/0.14 Simon Property Group 1.90/2.04 Teradyne 1.36/0.95 WaltDisney 1.57/1.76\n\n### **Tuesday**\n\nEarnings expected Estimate/Year Ago\n\nAdvanced Micro Devices 1.32/1.09 Amgen 4.73/5.31 Eaton 3.31/2.83 Merck 2.01/1.72 PepsiCo 2.24/1.96 Pfizer 0.57/0.63\n\n### **Wednesday**\n\nEIA status report Previous change in stocks in millions ofbarrels\n\nCrude-oilstocks down 2.3 Gasoline stocks up 0.2 Distillatesup 0.3\n\nMort.bankersindexes Purch.,previous down0.4% Refinan.,prev. down16.0%\n\nEarningsexpected Estimate/YearAgo\n\nAbbVie 2.65/2.16 Alphabet 2.64/2.15 BostonScientific 0.78/0.70 Eli Lilly 6.95/5.32 Qualcomm 3.39/3.41 Uber Technologies 0.78/3.21\n\n### **Thursday**\n\nEIA report:natural-gas Previous changein stocks in billionsof cubic feet down242\n\nInitial joblessclaims Previous 209,000 Expected 212,000\n\nFreddieMac mortgage survey Previousweeklyaverages\n\n30-yearfixed 6.10% 15-yearfixed 5.49%\n\nEarningsexpected Estimate/YearAgo\n\nAmazon.com 1.96/1.86 Bristol-MyersSquibb 1.23/1.67\n\nConocoPhillips 1.07/1.98 Intercontinental Exchange 1.68/1.52 KKR 1.16/1.32 Linde 4.18/3.97\n\n### **Friday**\n\nNonfarm payrolls Dec.,previous 50,000 Jan., expected 55,000\n\nUnemploymentrate Dec.,previous 4.4% Jan., expected 4.4%\n\nU.Mich. consumer index Jan., final 56.4 Feb.,prelim 54\n\nConsumerCredit Nov.,previous up$4.2bil. Dec.,expected up$8.0 bil.\n\nEarnings expected Estimate/Year Ago\n\nBiogen 1.63/3.44 Carlyle Group 1.00/0.92 CboeGlobal Markets 2.94/2.10 Centene (1.22)/0.80 PhilipMorris International 1.70/1.55 RoivantSciences (0.30)/0.23\n\n\\* FactSet Estimates earnings-per-share estimates don’t include extraordinary items (Losses in paren- theses) u Adjusted for stock split Note: Forecasts are from Dow Jones weekly survey of economists\n\nEAGAR, Ariz.—Sammy Col- lins twists and pulls on a direc- tional drill’s controls. He uses the 15,000-pound vehicle to bore a path through the soil for future fiber-optic lines. About 40 other members of his crew across this rural corner of Apache County dig ditches and hang fiber cables on poles. Collins, who previously worked at an addiction treat- ment facility, joined this crew six months ago, notching two promotions and boosting his pay by more than 22%. Experi- enced drillers can make any- where from the high $20s to mid-$40s an hour. Fiber cables were laid at a record number of homes last year, fueled by billions in fed- eral grants for broadband in- ternet. Technology companies’ appetite for data centers is boosting demand further. Yet the rush to build is hit- ting a human bottleneck, par- ticularly among workers who can piece together the fiber networks at the heart of the technology. Crews can’t hire fast enough and workers often are poached by competitors.\n\nand expects pay to rise even faster, said Chase Lapcinski, Push president. Entry-level fiber workers at Jonesburg, Mo.-based Sellen- riek Construction made about $60,000 including overtime last year. That can jump by 25% to 30% in the first three years, said Chief Executive Steve Sellenriek, and site su- perintendents can make six figures. Sellenriek said he is seeing an unprecedented rush of clients trying to lock in con- tracts early, before wages rise even further. Many workers join crews\n\nwithout prior fiber experience, the way Collins did. Others take training courses before getting hired. Learning Alli- ance in Tampa, Fla., offers a six-week to eight-week boot camp, funded by military bene- fits and corporate donations, that trains students to install fiber. The program’s 2025 graduates made about $26 an hour in starting pay, plus over- time, said Cesar Ruiz, Learning Alliance’s president and CEO. He has seen that rate increase by about 13% in two years. Fiber crews often work more than 10 hours a day out-\n\ndoors in nearly all weather. They may relocate to a project area for months at a time. And winter can bring less work in parts of the country, because of challenges drilling through snow and frozen ground. Mitch Grant, a fiber splicer with K&S Communications, has been spending time lately in Springerville, Ariz. Most locals there don’t have high-speed fi- ber internet, and he says they are eager to get connected—so eager, they stop him at work and ask for updates. (The ser- vice, provided by Commnet’s Ethos Broadband, is expected\n\n# **BUSINESS & FINANCE**\n\n“We just don’t have the peo- ple to do the work,” said Kyle Braude, who manages fiber build projects for JKL Associ- ates, a subsidiary of utility- construction contractor Push. The biggest holdup in the fiber boom, he said, is human capital for roles including drillers, foremen, splicers and aerial linemen. The demand is expected to create 58,000 new jobs be- tween 2025 and 2032, accord- ing to a 2024 report by the Fi- ber Broadband Association and the Power & Communication Contractors Association. With 120,000 workers expected to leave the field in the same pe- riod—mainly through retire- ment—the report forecasts a combined shortage of 178,000 workers. As a result, contrac- tors and analysts predict our AI future might take longer to achieve and a goal of nation- wide high-speed internet could remain out of reach. For those with fiber-laying skills, the work can pay good money, no college needed. Telecommunications line in- stallers and repairers made an- nual median wages of $70,500 for the year ended May 2024, the latest figure available, ac- cording to the Bureau of Labor Statistics. That figure doesn’t include overtime pay. Push raised hourly wages for fiber crews by 5% to 8% in each of the past several years,\n\nto be in place within two to four months.) Grant’s work is delicate: He cleaves each strand of a fiber- optic cable for a clean edge, then uses a machine to fuse the hair-thin filaments, which carry digital data at the speed of light. A slight misalignment, or even a speck of oil from the technician’s hands, and the strands won’t join. “When you first start doing it, it’s like, ‘How can you possi- bly do it?’ But then you get muscle memory,” Grant said. Four years ago, he left cell tower work to manage fiber crews, then saw how much this particular specialty makes— sometimes upward of $30 per splice for projects like this— and moved into that role. Skilled trades across the board are suffering a work- force pinch, and fiber doesn’t have the training pipeline that trades like auto repair and electrical power work enjoy. Lapcinski of Push is working to change that. He hosts hands-on skill days for high- school students in his home state of Wisconsin, so teens can get exposure. He hired farmers, former inmates and white-collar workers tired of office life. “AI isn’t going to replace this,” Lapcinski said. “It will accelerate and support some of it, but it’s not going to go out and dig the hole.”\n\n# **Labor Shortage Hits Broadband Boom**\n\n### Record demand for fiber-optic cables means workers can pull in soaring pay\n\nPAUL RATJE FOR WSJ\n\nowns Apmex, an online pre- cious-metals market. He sold millions of dollars of silver to dealers from Hong Kong, Singa- pore and Japan in recent days. The company’s mints, which source silver from local refineries in North America, are backlogged with orders. “Our company’s revenue before January was 70% gold, 30% silver. And January is probably going to be 70% sil- ver, 30% gold,” he said. On the trading floor of pre- cious-metals group Heraeus in Hanau, Germany, Dominik Sper- zel is fielding calls from inves- tors hunting for silver bars. Heraeus’ refineries, like those elsewhere in Europe, can’t keep up with demand and people cashing in the family silver. “The run on this, it is amaz- ing,” said Sperzel, head of trad- ing for Germany, adding that the buying spree is a surprise because silver, unlike gold, in- curs a sales tax in Germany. Manufacturers that rely on silver are stung by the metal’s explosive rally. Silver is a critical component in solar cells, of which 80% globally are made in China. The solar sector is estimated to have contributed about 17% of total silver demand in 2025, more than double its share from a decade ago and on par with what is used to make jewelry, according to the Silver Insti- tute, an industry group. As silver prices surged, the world’s largest solar-panel makers saw profitability plunge. Some are raising prices and speeding up plans to sub-\n\nstitute silver with cheaper ma- terials such as copper. An average 400-watt resi- dential solar panel contains about 4 to 6 grams of silver, according to Sagar Chopra, a senior research analyst at en- ergy consulting firm Wood Mackenzie. The precious metal is applied as silver paste to form busbars—solid metal strips that conduct electricity. For years, silver accounted for 5% to 7% of the cost of manufacturing a solar panel, Chopra said. Now it has be- come the largest cost compo- nent of solar-cell production, accounting for 15% to 20%. Some of the world’s largest solar-panel companies have is- sued silver-related profit warnings. Tongwei Solar, which boasts 14% of the global solar-cell market, said it would report a loss of up to $1.4 bil- lion in its 2025 results because\n\nof surging silver prices. Since January, prices for a typical solar cell in China have risen 10% from 4.96 cents per watt to 5.47 cents per watt, according to OPIS, an energy- data and analytics provider owned by the same parent as The Wall Street Journal. Eventually, refiners might process enough scrap to bring silver prices down. Industrial users are looking to minimize their silver consumption. Esti- mated industrial demand for sil- ver fell 2% last year thanks to a sharp drop in the amount of sil- ver used in each module, ac- cording to the Silver Institute. The constraints fueling wild moves in silver prices are financial as well as physical. The cost to borrow silver—es- pecially in London—soared last year after tons of silver moved across the Atlantic on fears that President Trump was going to impose tariffs on the metal. He didn’t impose the tariffs, but the borrowing costs remain high. The high borrowing costs, combined with the soaring price of the metal, has stretched the balance sheets of the banks, refiners and re- cycling firms that grease pre- cious-metal markets. That has curtailed the ability of finan- cial institutions to step in and temper big price moves. Changes in the direction of silver prices, as evidenced by the price plunge on Friday, “could be very sharp indeed,” said Matthew Piggott, director of gold and silver at consulting firm Metals Focus in London.\n\nmore than 30% off the highs, below $80 a troy ounce. The metal’s rise, analysts say, had been supported less by indus- trial demand or lack of supply, and more by speculators bet- ting on a shifting global order and distrust of the U.S. dollar. Some market participants say silver is behaving like “gold squared” and “gold on steroids.” Others compared it to meme-stock mania, or called silver the “devil’s metal,” arguing that a future price drop will surely send many believers to hell. The episode is reminiscent of the 1980 attempt by the Hunt brothers to corner the market. The price then peaked near $50, plunged and failed to attain those levels for four decades until last October. At times in recent months, regional gaps in prices got so wide that it has been profit- able for traders to move silver by air. It normally travels by ship because it is heavy and worth less than gold. “You know where your lug- gage is sitting underneath? You got gold and silver sitting un- derneath there, too,” said Ken Lewis, chief executive of Bullion International Group, which\n\n_ContinuedfrompageB1_\n\n# _Silver Rattles Companies_\n\n### **THE TICKER** | MARKET EVENTS COMING THIS WEEK\n\nver prices support the idea that investors were worried about debasement. The drops came as President Trump chose Kevin Warsh to chair the Federal Re- serve. He is seen as less likely to slash interest rates than the other Kevin who is in the running, Kevin Has- sett. Prices moved on the day mostly as you should expect from the selection of a hard-money hawk, or the avoidance of a soft-money dove: Stocks, gold and silver declined; the dollar and long-term Treasury yields rose. The exception came from the nuance that Warsh wants to sell the central bank’s stockpile of Trea- surys, but he has turned dovish on interest rates, so 2-year Treasury yields fell a little. Stimulus from President Trump’s tax cuts should be in the tens of billions of dol- lars this year, something likely to boost consumer\n\n_ContinuedfrompageB1_\n\nspending, shore up the economy and make inflation a little higher than it other- wise would be. The trouble is that infla- tion fears should have shown up in bond markets, and they just haven’t. Long-term inflation ex- pectations, measured by the break-even for the five years starting in five years’ time, have actually dropped this year and are lower than at the start of last year. Why would debasement fears only show up in pre- cious metals? Indeed, the Swiss franc moved against the dollar ex- actly in line with the euro until Japan’s snap election two weeks ago, suggesting no obvious investor flight to currencies perceived as de- basement-proof before that. Fear of a far weaker dol- lar in the future ought to help U.S. stocks, too. While hardly a safe haven from inflation, stocks ought to have some protection both because of foreign earnings and because sales should rise with consumer prices. Instead, U.S. stocks lagged far behind overseas stocks both this year and last year.\n\n**3) A global boom ignites inflation** The growing confidence in global growth has led to\n\nmarkets that act a lot like the years before the 2008-09 global financial crisis. From 2001 to 2007, inves- tors chose foreign over U.S. stocks, smaller companies over larger, and preferred cheap value stocks to growth stocks. Strong economies boosted copper demand, and the price soared. Gold thrived, too, rising from $273 an ounce at the start of 2001 to $634 by the start of 2007, before the cri- sis came into view. The same happened for the past cou- ple of months as investors fell out of love with big tech- nology compa- nies’ spending on artificial intelligence. Then this year small stocks smashed the biggest, silver surfed a wave of pri- vate buying and gold jumped 21.8% in the first 21 trading days of the year, the most over such a period since late 1999. Copper was already on a roll thanks to data-center construction, but since No- vember soared another 20%, before having its worst day on Friday since the Trump\n\ntariffs last April. Stronger global growth is a reasonable thing to ar- gue for, given Japan’s promised tax cuts, huge military spending in Ger- many, hopes for at least some China stimulus and the possibility of peace in Ukraine. But it doesn’t really ex- plain why the dollar fell this year exactly as much against the yen, where new stimulus is the driving force, as against the euro, where stimu- lus was baked into the price last year, or sterling, where nothing changed. It certainly doesn’t justify the tripling in the price of silver in 12 months. Gold has a role in all these stories, but the scale of the price moves, com- bined with the soaring and crashing of silver, shows a lot of froth. Like the grit in the oys- ter, there’s a solid core of truth inside the vanity of wild price swings. But Friday’s plummeting precious metals prices show the danger of being late to a rally that has gone far be- yond that truth.\n\n# _Parsing Gold, Silver Boom, Bust_\n\n2010 ’15 ’20 ’25\n\n–100\n\n0\n\n100\n\n200\n\n300\n\n400\n\n500\n\n600%\n\nSilver\n\nGold\n\nFutures price performance since the end of 2009\n\nSources: FactSet (futures); LSEG (relative performance)\n\nJan. 2026 9 16 23 30\n\n0\n\n20\n\n40\n\n60%\n\nFutures price performance this year\n\nSilver\n\nGold\n\nTop 50 vs. Russell 2000\n\nRussell 1000 Growth vs. Value\n\nJan. 2026\n\n–12\n\n–10\n\n–8\n\n–6\n\n–4\n\n–2\n\n0%\n\nRelative performance this year\n\n9 16 23 30\n\nFY2016 ’20 ’25\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12 million\n\nNew homes reached by fiber cables\n\nSource: Fiber Broadband Association Note: Fiscal year ends Sept. 30\n\n### Like the grit in oysters, there’s a core of truth in the vanity of price swings.\n\nP2JW033000-0-B00900-1--------XA","pages":[{"line_numbers":[],"markdown":"**THE WALL STREET JOURNAL.** Monday, February 2, 2026 | B9\n\nBY PATIENCE HAGGIN\n\nA JKL Associates work site where fiber-optic cabling is being fed through a conduit.\n\nSilver is now the largest cost component of solar cells.\n\nCHINATOPIX/AP\n\n### **Monday**\n\nISM mfg. index Dec., previous 51.9 Jan., expected n.a.\n\nEarningsexpected Estimate/Year Ago\n\nIDEXX Laboratories 2.93/2.62 NXPSemiconductors 3.31/3.18 PalantirTechnologies 0.23/0.14 Simon Property Group 1.90/2.04 Teradyne 1.36/0.95 WaltDisney 1.57/1.76\n\n### **Tuesday**\n\nEarnings expected Estimate/Year Ago\n\nAdvanced Micro Devices 1.32/1.09 Amgen 4.73/5.31 Eaton 3.31/2.83 Merck 2.01/1.72 PepsiCo 2.24/1.96 Pfizer 0.57/0.63\n\n### **Wednesday**\n\nEIA status report Previous change in stocks in millions ofbarrels\n\nCrude-oilstocks down 2.3 Gasoline stocks up 0.2 Distillatesup 0.3\n\nMort.bankersindexes Purch.,previous down0.4% Refinan.,prev. down16.0%\n\nEarningsexpected Estimate/YearAgo\n\nAbbVie 2.65/2.16 Alphabet 2.64/2.15 BostonScientific 0.78/0.70 Eli Lilly 6.95/5.32 Qualcomm 3.39/3.41 Uber Technologies 0.78/3.21\n\n### **Thursday**\n\nEIA report:natural-gas Previous changein stocks in billionsof cubic feet down242\n\nInitial joblessclaims Previous 209,000 Expected 212,000\n\nFreddieMac mortgage survey Previousweeklyaverages\n\n30-yearfixed 6.10% 15-yearfixed 5.49%\n\nEarningsexpected Estimate/YearAgo\n\nAmazon.com 1.96/1.86 Bristol-MyersSquibb 1.23/1.67\n\nConocoPhillips 1.07/1.98 Intercontinental Exchange 1.68/1.52 KKR 1.16/1.32 Linde 4.18/3.97\n\n### **Friday**\n\nNonfarm payrolls Dec.,previous 50,000 Jan., expected 55,000\n\nUnemploymentrate Dec.,previous 4.4% Jan., expected 4.4%\n\nU.Mich. consumer index Jan., final 56.4 Feb.,prelim 54\n\nConsumerCredit Nov.,previous up$4.2bil. Dec.,expected up$8.0 bil.\n\nEarnings expected Estimate/Year Ago\n\nBiogen 1.63/3.44 Carlyle Group 1.00/0.92 CboeGlobal Markets 2.94/2.10 Centene (1.22)/0.80 PhilipMorris International 1.70/1.55 RoivantSciences (0.30)/0.23\n\n\\* FactSet Estimates earnings-per-share estimates don’t include extraordinary items (Losses in paren- theses) u Adjusted for stock split Note: Forecasts are from Dow Jones weekly survey of economists\n\nEAGAR, Ariz.—Sammy Col- lins twists and pulls on a direc- tional drill’s controls. He uses the 15,000-pound vehicle to bore a path through the soil for future fiber-optic lines. About 40 other members of his crew across this rural corner of Apache County dig ditches and hang fiber cables on poles. Collins, who previously worked at an addiction treat- ment facility, joined this crew six months ago, notching two promotions and boosting his pay by more than 22%. Experi- enced drillers can make any- where from the high $20s to mid-$40s an hour. Fiber cables were laid at a record number of homes last year, fueled by billions in fed- eral grants for broadband in- ternet. Technology companies’ appetite for data centers is boosting demand further. Yet the rush to build is hit- ting a human bottleneck, par- ticularly among workers who can piece together the fiber networks at the heart of the technology. Crews can’t hire fast enough and workers often are poached by competitors.\n\nand expects pay to rise even faster, said Chase Lapcinski, Push president. Entry-level fiber workers at Jonesburg, Mo.-based Sellen- riek Construction made about $60,000 including overtime last year. That can jump by 25% to 30% in the first three years, said Chief Executive Steve Sellenriek, and site su- perintendents can make six figures. Sellenriek said he is seeing an unprecedented rush of clients trying to lock in con- tracts early, before wages rise even further. Many workers join crews\n\nwithout prior fiber experience, the way Collins did. Others take training courses before getting hired. Learning Alli- ance in Tampa, Fla., offers a six-week to eight-week boot camp, funded by military bene- fits and corporate donations, that trains students to install fiber. The program’s 2025 graduates made about $26 an hour in starting pay, plus over- time, said Cesar Ruiz, Learning Alliance’s president and CEO. He has seen that rate increase by about 13% in two years. Fiber crews often work more than 10 hours a day out-\n\ndoors in nearly all weather. They may relocate to a project area for months at a time. And winter can bring less work in parts of the country, because of challenges drilling through snow and frozen ground. Mitch Grant, a fiber splicer with K&S Communications, has been spending time lately in Springerville, Ariz. Most locals there don’t have high-speed fi- ber internet, and he says they are eager to get connected—so eager, they stop him at work and ask for updates. (The ser- vice, provided by Commnet’s Ethos Broadband, is expected\n\n# **BUSINESS & FINANCE**\n\n“We just don’t have the peo- ple to do the work,” said Kyle Braude, who manages fiber build projects for JKL Associ- ates, a subsidiary of utility- construction contractor Push. The biggest holdup in the fiber boom, he said, is human capital for roles including drillers, foremen, splicers and aerial linemen. The demand is expected to create 58,000 new jobs be- tween 2025 and 2032, accord- ing to a 2024 report by the Fi- ber Broadband Association and the Power & Communication Contractors Association. With 120,000 workers expected to leave the field in the same pe- riod—mainly through retire- ment—the report forecasts a combined shortage of 178,000 workers. As a result, contrac- tors and analysts predict our AI future might take longer to achieve and a goal of nation- wide high-speed internet could remain out of reach. For those with fiber-laying skills, the work can pay good money, no college needed. Telecommunications line in- stallers and repairers made an- nual median wages of $70,500 for the year ended May 2024, the latest figure available, ac- cording to the Bureau of Labor Statistics. That figure doesn’t include overtime pay. Push raised hourly wages for fiber crews by 5% to 8% in each of the past several years,\n\nto be in place within two to four months.) Grant’s work is delicate: He cleaves each strand of a fiber- optic cable for a clean edge, then uses a machine to fuse the hair-thin filaments, which carry digital data at the speed of light. A slight misalignment, or even a speck of oil from the technician’s hands, and the strands won’t join. “When you first start doing it, it’s like, ‘How can you possi- bly do it?’ But then you get muscle memory,” Grant said. Four years ago, he left cell tower work to manage fiber crews, then saw how much this particular specialty makes— sometimes upward of $30 per splice for projects like this— and moved into that role. Skilled trades across the board are suffering a work- force pinch, and fiber doesn’t have the training pipeline that trades like auto repair and electrical power work enjoy. Lapcinski of Push is working to change that. He hosts hands-on skill days for high- school students in his home state of Wisconsin, so teens can get exposure. He hired farmers, former inmates and white-collar workers tired of office life. “AI isn’t going to replace this,” Lapcinski said. “It will accelerate and support some of it, but it’s not going to go out and dig the hole.”\n\n# **Labor Shortage Hits Broadband Boom**\n\n### Record demand for fiber-optic cables means workers can pull in soaring pay\n\nPAUL RATJE FOR WSJ\n\nowns Apmex, an online pre- cious-metals market. He sold millions of dollars of silver to dealers from Hong Kong, Singa- pore and Japan in recent days. The company’s mints, which source silver from local refineries in North America, are backlogged with orders. “Our company’s revenue before January was 70% gold, 30% silver. And January is probably going to be 70% sil- ver, 30% gold,” he said. On the trading floor of pre- cious-metals group Heraeus in Hanau, Germany, Dominik Sper- zel is fielding calls from inves- tors hunting for silver bars. Heraeus’ refineries, like those elsewhere in Europe, can’t keep up with demand and people cashing in the family silver. “The run on this, it is amaz- ing,” said Sperzel, head of trad- ing for Germany, adding that the buying spree is a surprise because silver, unlike gold, in- curs a sales tax in Germany. Manufacturers that rely on silver are stung by the metal’s explosive rally. Silver is a critical component in solar cells, of which 80% globally are made in China. The solar sector is estimated to have contributed about 17% of total silver demand in 2025, more than double its share from a decade ago and on par with what is used to make jewelry, according to the Silver Insti- tute, an industry group. As silver prices surged, the world’s largest solar-panel makers saw profitability plunge. Some are raising prices and speeding up plans to sub-\n\nstitute silver with cheaper ma- terials such as copper. An average 400-watt resi- dential solar panel contains about 4 to 6 grams of silver, according to Sagar Chopra, a senior research analyst at en- ergy consulting firm Wood Mackenzie. The precious metal is applied as silver paste to form busbars—solid metal strips that conduct electricity. For years, silver accounted for 5% to 7% of the cost of manufacturing a solar panel, Chopra said. Now it has be- come the largest cost compo- nent of solar-cell production, accounting for 15% to 20%. Some of the world’s largest solar-panel companies have is- sued silver-related profit warnings. Tongwei Solar, which boasts 14% of the global solar-cell market, said it would report a loss of up to $1.4 bil- lion in its 2025 results because\n\nof surging silver prices. Since January, prices for a typical solar cell in China have risen 10% from 4.96 cents per watt to 5.47 cents per watt, according to OPIS, an energy- data and analytics provider owned by the same parent as The Wall Street Journal. Eventually, refiners might process enough scrap to bring silver prices down. Industrial users are looking to minimize their silver consumption. Esti- mated industrial demand for sil- ver fell 2% last year thanks to a sharp drop in the amount of sil- ver used in each module, ac- cording to the Silver Institute. The constraints fueling wild moves in silver prices are financial as well as physical. The cost to borrow silver—es- pecially in London—soared last year after tons of silver moved across the Atlantic on fears that President Trump was going to impose tariffs on the metal. He didn’t impose the tariffs, but the borrowing costs remain high. The high borrowing costs, combined with the soaring price of the metal, has stretched the balance sheets of the banks, refiners and re- cycling firms that grease pre- cious-metal markets. That has curtailed the ability of finan- cial institutions to step in and temper big price moves. Changes in the direction of silver prices, as evidenced by the price plunge on Friday, “could be very sharp indeed,” said Matthew Piggott, director of gold and silver at consulting firm Metals Focus in London.\n\nmore than 30% off the highs, below $80 a troy ounce. The metal’s rise, analysts say, had been supported less by indus- trial demand or lack of supply, and more by speculators bet- ting on a shifting global order and distrust of the U.S. dollar. Some market participants say silver is behaving like “gold squared” and “gold on steroids.” Others compared it to meme-stock mania, or called silver the “devil’s metal,” arguing that a future price drop will surely send many believers to hell. The episode is reminiscent of the 1980 attempt by the Hunt brothers to corner the market. The price then peaked near $50, plunged and failed to attain those levels for four decades until last October. At times in recent months, regional gaps in prices got so wide that it has been profit- able for traders to move silver by air. It normally travels by ship because it is heavy and worth less than gold. “You know where your lug- gage is sitting underneath? You got gold and silver sitting un- derneath there, too,” said Ken Lewis, chief executive of Bullion International Group, which\n\n_ContinuedfrompageB1_\n\n# _Silver Rattles Companies_\n\n### **THE TICKER** | MARKET EVENTS COMING THIS WEEK\n\nver prices support the idea that investors were worried about debasement. The drops came as President Trump chose Kevin Warsh to chair the Federal Re- serve. He is seen as less likely to slash interest rates than the other Kevin who is in the running, Kevin Has- sett. Prices moved on the day mostly as you should expect from the selection of a hard-money hawk, or the avoidance of a soft-money dove: Stocks, gold and silver declined; the dollar and long-term Treasury yields rose. The exception came from the nuance that Warsh wants to sell the central bank’s stockpile of Trea- surys, but he has turned dovish on interest rates, so 2-year Treasury yields fell a little. Stimulus from President Trump’s tax cuts should be in the tens of billions of dol- lars this year, something likely to boost consumer\n\n_ContinuedfrompageB1_\n\nspending, shore up the economy and make inflation a little higher than it other- wise would be. The trouble is that infla- tion fears should have shown up in bond markets, and they just haven’t. Long-term inflation ex- pectations, measured by the break-even for the five years starting in five years’ time, have actually dropped this year and are lower than at the start of last year. Why would debasement fears only show up in pre- cious metals? Indeed, the Swiss franc moved against the dollar ex- actly in line with the euro until Japan’s snap election two weeks ago, suggesting no obvious investor flight to currencies perceived as de- basement-proof before that. Fear of a far weaker dol- lar in the future ought to help U.S. stocks, too. While hardly a safe haven from inflation, stocks ought to have some protection both because of foreign earnings and because sales should rise with consumer prices. Instead, U.S. stocks lagged far behind overseas stocks both this year and last year.\n\n**3) A global boom ignites inflation** The growing confidence in global growth has led to\n\nmarkets that act a lot like the years before the 2008-09 global financial crisis. From 2001 to 2007, inves- tors chose foreign over U.S. stocks, smaller companies over larger, and preferred cheap value stocks to growth stocks. Strong economies boosted copper demand, and the price soared. Gold thrived, too, rising from $273 an ounce at the start of 2001 to $634 by the start of 2007, before the cri- sis came into view. The same happened for the past cou- ple of months as investors fell out of love with big tech- nology compa- nies’ spending on artificial intelligence. Then this year small stocks smashed the biggest, silver surfed a wave of pri- vate buying and gold jumped 21.8% in the first 21 trading days of the year, the most over such a period since late 1999. Copper was already on a roll thanks to data-center construction, but since No- vember soared another 20%, before having its worst day on Friday since the Trump\n\ntariffs last April. Stronger global growth is a reasonable thing to ar- gue for, given Japan’s promised tax cuts, huge military spending in Ger- many, hopes for at least some China stimulus and the possibility of peace in Ukraine. But it doesn’t really ex- plain why the dollar fell this year exactly as much against the yen, where new stimulus is the driving force, as against the euro, where stimu- lus was baked into the price last year, or sterling, where nothing changed. It certainly doesn’t justify the tripling in the price of silver in 12 months. Gold has a role in all these stories, but the scale of the price moves, com- bined with the soaring and crashing of silver, shows a lot of froth. Like the grit in the oys- ter, there’s a solid core of truth inside the vanity of wild price swings. But Friday’s plummeting precious metals prices show the danger of being late to a rally that has gone far be- yond that truth.\n\n# _Parsing Gold, Silver Boom, Bust_\n\n2010 ’15 ’20 ’25\n\n–100\n\n0\n\n100\n\n200\n\n300\n\n400\n\n500\n\n600%\n\nSilver\n\nGold\n\nFutures price performance since the end of 2009\n\nSources: FactSet (futures); LSEG (relative performance)\n\nJan. 2026 9 16 23 30\n\n0\n\n20\n\n40\n\n60%\n\nFutures price performance this year\n\nSilver\n\nGold\n\nTop 50 vs. Russell 2000\n\nRussell 1000 Growth vs. Value\n\nJan. 2026\n\n–12\n\n–10\n\n–8\n\n–6\n\n–4\n\n–2\n\n0%\n\nRelative performance this year\n\n9 16 23 30\n\nFY2016 ’20 ’25\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12 million\n\nNew homes reached by fiber cables\n\nSource: Fiber Broadband Association Note: Fiscal year ends Sept. 30\n\n### Like the grit in oysters, there’s a core of truth in the vanity of price swings.\n\nP2JW033000-0-B00900-1--------XA","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**THE WALL STREET JOURNAL.** Monday, February 2, 2026 | B9\n\nBY PATIENCE HAGGIN\n\nA JKL Associates work site where fiber-optic cabling is being fed through a conduit.\n\nSilver is now the largest cost component of solar cells.\n\nCHINATOPIX/AP\n\n### **Monday**\n\nISM mfg. index Dec., previous 51.9 Jan., expected n.a.\n\nEarningsexpected Estimate/Year Ago\n\nIDEXX Laboratories 2.93/2.62 NXPSemiconductors 3.31/3.18 PalantirTechnologies 0.23/0.14 Simon Property Group 1.90/2.04 Teradyne 1.36/0.95 WaltDisney 1.57/1.76\n\n### **Tuesday**\n\nEarnings expected Estimate/Year Ago\n\nAdvanced Micro Devices 1.32/1.09 Amgen 4.73/5.31 Eaton 3.31/2.83 Merck 2.01/1.72 PepsiCo 2.24/1.96 Pfizer 0.57/0.63\n\n### **Wednesday**\n\nEIA status report Previous change in stocks in millions ofbarrels\n\nCrude-oilstocks down 2.3 Gasoline stocks up 0.2 Distillatesup 0.3\n\nMort.bankersindexes Purch.,previous down0.4% Refinan.,prev. down16.0%\n\nEarningsexpected Estimate/YearAgo\n\nAbbVie 2.65/2.16 Alphabet 2.64/2.15 BostonScientific 0.78/0.70 Eli Lilly 6.95/5.32 Qualcomm 3.39/3.41 Uber Technologies 0.78/3.21\n\n### **Thursday**\n\nEIA report:natural-gas Previous changein stocks in billionsof cubic feet down242\n\nInitial joblessclaims Previous 209,000 Expected 212,000\n\nFreddieMac mortgage survey Previousweeklyaverages\n\n30-yearfixed 6.10% 15-yearfixed 5.49%\n\nEarningsexpected Estimate/YearAgo\n\nAmazon.com 1.96/1.86 Bristol-MyersSquibb 1.23/1.67\n\nConocoPhillips 1.07/1.98 Intercontinental Exchange 1.68/1.52 KKR 1.16/1.32 Linde 4.18/3.97\n\n### **Friday**\n\nNonfarm payrolls Dec.,previous 50,000 Jan., expected 55,000\n\nUnemploymentrate Dec.,previous 4.4% Jan., expected 4.4%\n\nU.Mich. consumer index Jan., final 56.4 Feb.,prelim 54\n\nConsumerCredit Nov.,previous up$4.2bil. Dec.,expected up$8.0 bil.\n\nEarnings expected Estimate/Year Ago\n\nBiogen 1.63/3.44 Carlyle Group 1.00/0.92 CboeGlobal Markets 2.94/2.10 Centene (1.22)/0.80 PhilipMorris International 1.70/1.55 RoivantSciences (0.30)/0.23\n\n\\* FactSet Estimates earnings-per-share estimates don’t include extraordinary items (Losses in paren- theses) u Adjusted for stock split Note: Forecasts are from Dow Jones weekly survey of economists\n\nEAGAR, Ariz.—Sammy Col- lins twists and pulls on a direc- tional drill’s controls. He uses the 15,000-pound vehicle to bore a path through the soil for future fiber-optic lines. About 40 other members of his crew across this rural corner of Apache County dig ditches and hang fiber cables on poles. Collins, who previously worked at an addiction treat- ment facility, joined this crew six months ago, notching two promotions and boosting his pay by more than 22%. Experi- enced drillers can make any- where from the high $20s to mid-$40s an hour. Fiber cables were laid at a record number of homes last year, fueled by billions in fed- eral grants for broadband in- ternet. Technology companies’ appetite for data centers is boosting demand further. Yet the rush to build is hit- ting a human bottleneck, par- ticularly among workers who can piece together the fiber networks at the heart of the technology. Crews can’t hire fast enough and workers often are poached by competitors.\n\nand expects pay to rise even faster, said Chase Lapcinski, Push president. Entry-level fiber workers at Jonesburg, Mo.-based Sellen- riek Construction made about $60,000 including overtime last year. That can jump by 25% to 30% in the first three years, said Chief Executive Steve Sellenriek, and site su- perintendents can make six figures. Sellenriek said he is seeing an unprecedented rush of clients trying to lock in con- tracts early, before wages rise even further. Many workers join crews\n\nwithout prior fiber experience, the way Collins did. Others take training courses before getting hired. Learning Alli- ance in Tampa, Fla., offers a six-week to eight-week boot camp, funded by military bene- fits and corporate donations, that trains students to install fiber. The program’s 2025 graduates made about $26 an hour in starting pay, plus over- time, said Cesar Ruiz, Learning Alliance’s president and CEO. He has seen that rate increase by about 13% in two years. Fiber crews often work more than 10 hours a day out-\n\ndoors in nearly all weather. They may relocate to a project area for months at a time. And winter can bring less work in parts of the country, because of challenges drilling through snow and frozen ground. Mitch Grant, a fiber splicer with K&S Communications, has been spending time lately in Springerville, Ariz. Most locals there don’t have high-speed fi- ber internet, and he says they are eager to get connected—so eager, they stop him at work and ask for updates. (The ser- vice, provided by Commnet’s Ethos Broadband, is expected\n\n# **BUSINESS & FINANCE**\n\n“We just don’t have the peo- ple to do the work,” said Kyle Braude, who manages fiber build projects for JKL Associ- ates, a subsidiary of utility- construction contractor Push. The biggest holdup in the fiber boom, he said, is human capital for roles including drillers, foremen, splicers and aerial linemen. The demand is expected to create 58,000 new jobs be- tween 2025 and 2032, accord- ing to a 2024 report by the Fi- ber Broadband Association and the Power & Communication Contractors Association. With 120,000 workers expected to leave the field in the same pe- riod—mainly through retire- ment—the report forecasts a combined shortage of 178,000 workers. As a result, contrac- tors and analysts predict our AI future might take longer to achieve and a goal of nation- wide high-speed internet could remain out of reach. For those with fiber-laying skills, the work can pay good money, no college needed. Telecommunications line in- stallers and repairers made an- nual median wages of $70,500 for the year ended May 2024, the latest figure available, ac- cording to the Bureau of Labor Statistics. That figure doesn’t include overtime pay. Push raised hourly wages for fiber crews by 5% to 8% in each of the past several years,\n\nto be in place within two to four months.) Grant’s work is delicate: He cleaves each strand of a fiber- optic cable for a clean edge, then uses a machine to fuse the hair-thin filaments, which carry digital data at the speed of light. A slight misalignment, or even a speck of oil from the technician’s hands, and the strands won’t join. “When you first start doing it, it’s like, ‘How can you possi- bly do it?’ But then you get muscle memory,” Grant said. Four years ago, he left cell tower work to manage fiber crews, then saw how much this particular specialty makes— sometimes upward of $30 per splice for projects like this— and moved into that role. Skilled trades across the board are suffering a work- force pinch, and fiber doesn’t have the training pipeline that trades like auto repair and electrical power work enjoy. Lapcinski of Push is working to change that. He hosts hands-on skill days for high- school students in his home state of Wisconsin, so teens can get exposure. He hired farmers, former inmates and white-collar workers tired of office life. “AI isn’t going to replace this,” Lapcinski said. “It will accelerate and support some of it, but it’s not going to go out and dig the hole.”\n\n# **Labor Shortage Hits Broadband Boom**\n\n### Record demand for fiber-optic cables means workers can pull in soaring pay\n\nPAUL RATJE FOR WSJ\n\nowns Apmex, an online pre- cious-metals market. He sold millions of dollars of silver to dealers from Hong Kong, Singa- pore and Japan in recent days. The company’s mints, which source silver from local refineries in North America, are backlogged with orders. “Our company’s revenue before January was 70% gold, 30% silver. And January is probably going to be 70% sil- ver, 30% gold,” he said. On the trading floor of pre- cious-metals group Heraeus in Hanau, Germany, Dominik Sper- zel is fielding calls from inves- tors hunting for silver bars. Heraeus’ refineries, like those elsewhere in Europe, can’t keep up with demand and people cashing in the family silver. “The run on this, it is amaz- ing,” said Sperzel, head of trad- ing for Germany, adding that the buying spree is a surprise because silver, unlike gold, in- curs a sales tax in Germany. Manufacturers that rely on silver are stung by the metal’s explosive rally. Silver is a critical component in solar cells, of which 80% globally are made in China. The solar sector is estimated to have contributed about 17% of total silver demand in 2025, more than double its share from a decade ago and on par with what is used to make jewelry, according to the Silver Insti- tute, an industry group. As silver prices surged, the world’s largest solar-panel makers saw profitability plunge. Some are raising prices and speeding up plans to sub-\n\nstitute silver with cheaper ma- terials such as copper. An average 400-watt resi- dential solar panel contains about 4 to 6 grams of silver, according to Sagar Chopra, a senior research analyst at en- ergy consulting firm Wood Mackenzie. The precious metal is applied as silver paste to form busbars—solid metal strips that conduct electricity. For years, silver accounted for 5% to 7% of the cost of manufacturing a solar panel, Chopra said. Now it has be- come the largest cost compo- nent of solar-cell production, accounting for 15% to 20%. Some of the world’s largest solar-panel companies have is- sued silver-related profit warnings. Tongwei Solar, which boasts 14% of the global solar-cell market, said it would report a loss of up to $1.4 bil- lion in its 2025 results because\n\nof surging silver prices. Since January, prices for a typical solar cell in China have risen 10% from 4.96 cents per watt to 5.47 cents per watt, according to OPIS, an energy- data and analytics provider owned by the same parent as The Wall Street Journal. Eventually, refiners might process enough scrap to bring silver prices down. Industrial users are looking to minimize their silver consumption. Esti- mated industrial demand for sil- ver fell 2% last year thanks to a sharp drop in the amount of sil- ver used in each module, ac- cording to the Silver Institute. The constraints fueling wild moves in silver prices are financial as well as physical. The cost to borrow silver—es- pecially in London—soared last year after tons of silver moved across the Atlantic on fears that President Trump was going to impose tariffs on the metal. He didn’t impose the tariffs, but the borrowing costs remain high. The high borrowing costs, combined with the soaring price of the metal, has stretched the balance sheets of the banks, refiners and re- cycling firms that grease pre- cious-metal markets. That has curtailed the ability of finan- cial institutions to step in and temper big price moves. Changes in the direction of silver prices, as evidenced by the price plunge on Friday, “could be very sharp indeed,” said Matthew Piggott, director of gold and silver at consulting firm Metals Focus in London.\n\nmore than 30% off the highs, below $80 a troy ounce. The metal’s rise, analysts say, had been supported less by indus- trial demand or lack of supply, and more by speculators bet- ting on a shifting global order and distrust of the U.S. dollar. Some market participants say silver is behaving like “gold squared” and “gold on steroids.” Others compared it to meme-stock mania, or called silver the “devil’s metal,” arguing that a future price drop will surely send many believers to hell. The episode is reminiscent of the 1980 attempt by the Hunt brothers to corner the market. The price then peaked near $50, plunged and failed to attain those levels for four decades until last October. At times in recent months, regional gaps in prices got so wide that it has been profit- able for traders to move silver by air. It normally travels by ship because it is heavy and worth less than gold. “You know where your lug- gage is sitting underneath? You got gold and silver sitting un- derneath there, too,” said Ken Lewis, chief executive of Bullion International Group, which\n\n_ContinuedfrompageB1_\n\n# _Silver Rattles Companies_\n\n### **THE TICKER** | MARKET EVENTS COMING THIS WEEK\n\nver prices support the idea that investors were worried about debasement. The drops came as President Trump chose Kevin Warsh to chair the Federal Re- serve. He is seen as less likely to slash interest rates than the other Kevin who is in the running, Kevin Has- sett. Prices moved on the day mostly as you should expect from the selection of a hard-money hawk, or the avoidance of a soft-money dove: Stocks, gold and silver declined; the dollar and long-term Treasury yields rose. The exception came from the nuance that Warsh wants to sell the central bank’s stockpile of Trea- surys, but he has turned dovish on interest rates, so 2-year Treasury yields fell a little. Stimulus from President Trump’s tax cuts should be in the tens of billions of dol- lars this year, something likely to boost consumer\n\n_ContinuedfrompageB1_\n\nspending, shore up the economy and make inflation a little higher than it other- wise would be. The trouble is that infla- tion fears should have shown up in bond markets, and they just haven’t. Long-term inflation ex- pectations, measured by the break-even for the five years starting in five years’ time, have actually dropped this year and are lower than at the start of last year. Why would debasement fears only show up in pre- cious metals? Indeed, the Swiss franc moved against the dollar ex- actly in line with the euro until Japan’s snap election two weeks ago, suggesting no obvious investor flight to currencies perceived as de- basement-proof before that. Fear of a far weaker dol- lar in the future ought to help U.S. stocks, too. While hardly a safe haven from inflation, stocks ought to have some protection both because of foreign earnings and because sales should rise with consumer prices. Instead, U.S. stocks lagged far behind overseas stocks both this year and last year.\n\n**3) A global boom ignites inflation** The growing confidence in global growth has led to\n\nmarkets that act a lot like the years before the 2008-09 global financial crisis. From 2001 to 2007, inves- tors chose foreign over U.S. stocks, smaller companies over larger, and preferred cheap value stocks to growth stocks. Strong economies boosted copper demand, and the price soared. Gold thrived, too, rising from $273 an ounce at the start of 2001 to $634 by the start of 2007, before the cri- sis came into view. The same happened for the past cou- ple of months as investors fell out of love with big tech- nology compa- nies’ spending on artificial intelligence. Then this year small stocks smashed the biggest, silver surfed a wave of pri- vate buying and gold jumped 21.8% in the first 21 trading days of the year, the most over such a period since late 1999. Copper was already on a roll thanks to data-center construction, but since No- vember soared another 20%, before having its worst day on Friday since the Trump\n\ntariffs last April. Stronger global growth is a reasonable thing to ar- gue for, given Japan’s promised tax cuts, huge military spending in Ger- many, hopes for at least some China stimulus and the possibility of peace in Ukraine. But it doesn’t really ex- plain why the dollar fell this year exactly as much against the yen, where new stimulus is the driving force, as against the euro, where stimu- lus was baked into the price last year, or sterling, where nothing changed. It certainly doesn’t justify the tripling in the price of silver in 12 months. Gold has a role in all these stories, but the scale of the price moves, com- bined with the soaring and crashing of silver, shows a lot of froth. Like the grit in the oys- ter, there’s a solid core of truth inside the vanity of wild price swings. But Friday’s plummeting precious metals prices show the danger of being late to a rally that has gone far be- yond that truth.\n\n# _Parsing Gold, Silver Boom, Bust_\n\n2010 ’15 ’20 ’25\n\n–100\n\n0\n\n100\n\n200\n\n300\n\n400\n\n500\n\n600%\n\nSilver\n\nGold\n\nFutures price performance since the end of 2009\n\nSources: FactSet (futures); LSEG (relative performance)\n\nJan. 2026 9 16 23 30\n\n0\n\n20\n\n40\n\n60%\n\nFutures price performance this year\n\nSilver\n\nGold\n\nTop 50 vs. Russell 2000\n\nRussell 1000 Growth vs. Value\n\nJan. 2026\n\n–12\n\n–10\n\n–8\n\n–6\n\n–4\n\n–2\n\n0%\n\nRelative performance this year\n\n9 16 23 30\n\nFY2016 ’20 ’25\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12 million\n\nNew homes reached by fiber cables\n\nSource: Fiber Broadband Association Note: Fiscal year ends Sept. 30\n\n### Like the grit in oysters, there’s a core of truth in the vanity of price swings.\n\nP2JW033000-0-B00900-1--------XA","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/pdf_f270146f8ca7_p25.pdf"]},"response":{"errors":[],"extract_id":"extract_a6565b401079c721a2b936b7ec87d749","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**THE WALL STREET JOURNAL.** Monday, February 2, 2026 | B9\n\nBY PATIENCE HAGGIN\n\nA JKL Associates work site where fiber-optic cabling is being fed through a conduit.\n\nSilver is now the largest cost component of solar cells.\n\nCHINATOPIX/AP\n\n### **Monday**\n\nISM mfg. index Dec., previous 51.9 Jan., expected n.a.\n\nEarningsexpected Estimate/Year Ago\n\nIDEXX Laboratories 2.93/2.62 NXPSemiconductors 3.31/3.18 PalantirTechnologies 0.23/0.14 Simon Property Group 1.90/2.04 Teradyne 1.36/0.95 WaltDisney 1.57/1.76\n\n### **Tuesday**\n\nEarnings expected Estimate/Year Ago\n\nAdvanced Micro Devices 1.32/1.09 Amgen 4.73/5.31 Eaton 3.31/2.83 Merck 2.01/1.72 PepsiCo 2.24/1.96 Pfizer 0.57/0.63\n\n### **Wednesday**\n\nEIA status report Previous change in stocks in millions ofbarrels\n\nCrude-oilstocks down 2.3 Gasoline stocks up 0.2 Distillatesup 0.3\n\nMort.bankersindexes Purch.,previous down0.4% Refinan.,prev. down16.0%\n\nEarningsexpected Estimate/YearAgo\n\nAbbVie 2.65/2.16 Alphabet 2.64/2.15 BostonScientific 0.78/0.70 Eli Lilly 6.95/5.32 Qualcomm 3.39/3.41 Uber Technologies 0.78/3.21\n\n### **Thursday**\n\nEIA report:natural-gas Previous changein stocks in billionsof cubic feet down242\n\nInitial joblessclaims Previous 209,000 Expected 212,000\n\nFreddieMac mortgage survey Previousweeklyaverages\n\n30-yearfixed 6.10% 15-yearfixed 5.49%\n\nEarningsexpected Estimate/YearAgo\n\nAmazon.com 1.96/1.86 Bristol-MyersSquibb 1.23/1.67\n\nConocoPhillips 1.07/1.98 Intercontinental Exchange 1.68/1.52 KKR 1.16/1.32 Linde 4.18/3.97\n\n### **Friday**\n\nNonfarm payrolls Dec.,previous 50,000 Jan., expected 55,000\n\nUnemploymentrate Dec.,previous 4.4% Jan., expected 4.4%\n\nU.Mich. consumer index Jan., final 56.4 Feb.,prelim 54\n\nConsumerCredit Nov.,previous up$4.2bil. Dec.,expected up$8.0 bil.\n\nEarnings expected Estimate/Year Ago\n\nBiogen 1.63/3.44 Carlyle Group 1.00/0.92 CboeGlobal Markets 2.94/2.10 Centene (1.22)/0.80 PhilipMorris International 1.70/1.55 RoivantSciences (0.30)/0.23\n\n\\* FactSet Estimates earnings-per-share estimates don’t include extraordinary items (Losses in paren- theses) u Adjusted for stock split Note: Forecasts are from Dow Jones weekly survey of economists\n\nEAGAR, Ariz.—Sammy Col- lins twists and pulls on a direc- tional drill’s controls. He uses the 15,000-pound vehicle to bore a path through the soil for future fiber-optic lines. About 40 other members of his crew across this rural corner of Apache County dig ditches and hang fiber cables on poles. Collins, who previously worked at an addiction treat- ment facility, joined this crew six months ago, notching two promotions and boosting his pay by more than 22%. Experi- enced drillers can make any- where from the high $20s to mid-$40s an hour. Fiber cables were laid at a record number of homes last year, fueled by billions in fed- eral grants for broadband in- ternet. Technology companies’ appetite for data centers is boosting demand further. Yet the rush to build is hit- ting a human bottleneck, par- ticularly among workers who can piece together the fiber networks at the heart of the technology. Crews can’t hire fast enough and workers often are poached by competitors.\n\nand expects pay to rise even faster, said Chase Lapcinski, Push president. Entry-level fiber workers at Jonesburg, Mo.-based Sellen- riek Construction made about $60,000 including overtime last year. That can jump by 25% to 30% in the first three years, said Chief Executive Steve Sellenriek, and site su- perintendents can make six figures. Sellenriek said he is seeing an unprecedented rush of clients trying to lock in con- tracts early, before wages rise even further. Many workers join crews\n\nwithout prior fiber experience, the way Collins did. Others take training courses before getting hired. Learning Alli- ance in Tampa, Fla., offers a six-week to eight-week boot camp, funded by military bene- fits and corporate donations, that trains students to install fiber. The program’s 2025 graduates made about $26 an hour in starting pay, plus over- time, said Cesar Ruiz, Learning Alliance’s president and CEO. He has seen that rate increase by about 13% in two years. Fiber crews often work more than 10 hours a day out-\n\ndoors in nearly all weather. They may relocate to a project area for months at a time. And winter can bring less work in parts of the country, because of challenges drilling through snow and frozen ground. Mitch Grant, a fiber splicer with K&S Communications, has been spending time lately in Springerville, Ariz. Most locals there don’t have high-speed fi- ber internet, and he says they are eager to get connected—so eager, they stop him at work and ask for updates. (The ser- vice, provided by Commnet’s Ethos Broadband, is expected\n\n# **BUSINESS & FINANCE**\n\n“We just don’t have the peo- ple to do the work,” said Kyle Braude, who manages fiber build projects for JKL Associ- ates, a subsidiary of utility- construction contractor Push. The biggest holdup in the fiber boom, he said, is human capital for roles including drillers, foremen, splicers and aerial linemen. The demand is expected to create 58,000 new jobs be- tween 2025 and 2032, accord- ing to a 2024 report by the Fi- ber Broadband Association and the Power & Communication Contractors Association. With 120,000 workers expected to leave the field in the same pe- riod—mainly through retire- ment—the report forecasts a combined shortage of 178,000 workers. As a result, contrac- tors and analysts predict our AI future might take longer to achieve and a goal of nation- wide high-speed internet could remain out of reach. For those with fiber-laying skills, the work can pay good money, no college needed. Telecommunications line in- stallers and repairers made an- nual median wages of $70,500 for the year ended May 2024, the latest figure available, ac- cording to the Bureau of Labor Statistics. That figure doesn’t include overtime pay. Push raised hourly wages for fiber crews by 5% to 8% in each of the past several years,\n\nto be in place within two to four months.) Grant’s work is delicate: He cleaves each strand of a fiber- optic cable for a clean edge, then uses a machine to fuse the hair-thin filaments, which carry digital data at the speed of light. A slight misalignment, or even a speck of oil from the technician’s hands, and the strands won’t join. “When you first start doing it, it’s like, ‘How can you possi- bly do it?’ But then you get muscle memory,” Grant said. Four years ago, he left cell tower work to manage fiber crews, then saw how much this particular specialty makes— sometimes upward of $30 per splice for projects like this— and moved into that role. Skilled trades across the board are suffering a work- force pinch, and fiber doesn’t have the training pipeline that trades like auto repair and electrical power work enjoy. Lapcinski of Push is working to change that. He hosts hands-on skill days for high- school students in his home state of Wisconsin, so teens can get exposure. He hired farmers, former inmates and white-collar workers tired of office life. “AI isn’t going to replace this,” Lapcinski said. “It will accelerate and support some of it, but it’s not going to go out and dig the hole.”\n\n# **Labor Shortage Hits Broadband Boom**\n\n### Record demand for fiber-optic cables means workers can pull in soaring pay\n\nPAUL RATJE FOR WSJ\n\nowns Apmex, an online pre- cious-metals market. He sold millions of dollars of silver to dealers from Hong Kong, Singa- pore and Japan in recent days. The company’s mints, which source silver from local refineries in North America, are backlogged with orders. “Our company’s revenue before January was 70% gold, 30% silver. And January is probably going to be 70% sil- ver, 30% gold,” he said. On the trading floor of pre- cious-metals group Heraeus in Hanau, Germany, Dominik Sper- zel is fielding calls from inves- tors hunting for silver bars. Heraeus’ refineries, like those elsewhere in Europe, can’t keep up with demand and people cashing in the family silver. “The run on this, it is amaz- ing,” said Sperzel, head of trad- ing for Germany, adding that the buying spree is a surprise because silver, unlike gold, in- curs a sales tax in Germany. Manufacturers that rely on silver are stung by the metal’s explosive rally. Silver is a critical component in solar cells, of which 80% globally are made in China. The solar sector is estimated to have contributed about 17% of total silver demand in 2025, more than double its share from a decade ago and on par with what is used to make jewelry, according to the Silver Insti- tute, an industry group. As silver prices surged, the world’s largest solar-panel makers saw profitability plunge. Some are raising prices and speeding up plans to sub-\n\nstitute silver with cheaper ma- terials such as copper. An average 400-watt resi- dential solar panel contains about 4 to 6 grams of silver, according to Sagar Chopra, a senior research analyst at en- ergy consulting firm Wood Mackenzie. The precious metal is applied as silver paste to form busbars—solid metal strips that conduct electricity. For years, silver accounted for 5% to 7% of the cost of manufacturing a solar panel, Chopra said. Now it has be- come the largest cost compo- nent of solar-cell production, accounting for 15% to 20%. Some of the world’s largest solar-panel companies have is- sued silver-related profit warnings. Tongwei Solar, which boasts 14% of the global solar-cell market, said it would report a loss of up to $1.4 bil- lion in its 2025 results because\n\nof surging silver prices. Since January, prices for a typical solar cell in China have risen 10% from 4.96 cents per watt to 5.47 cents per watt, according to OPIS, an energy- data and analytics provider owned by the same parent as The Wall Street Journal. Eventually, refiners might process enough scrap to bring silver prices down. Industrial users are looking to minimize their silver consumption. Esti- mated industrial demand for sil- ver fell 2% last year thanks to a sharp drop in the amount of sil- ver used in each module, ac- cording to the Silver Institute. The constraints fueling wild moves in silver prices are financial as well as physical. The cost to borrow silver—es- pecially in London—soared last year after tons of silver moved across the Atlantic on fears that President Trump was going to impose tariffs on the metal. He didn’t impose the tariffs, but the borrowing costs remain high. The high borrowing costs, combined with the soaring price of the metal, has stretched the balance sheets of the banks, refiners and re- cycling firms that grease pre- cious-metal markets. That has curtailed the ability of finan- cial institutions to step in and temper big price moves. Changes in the direction of silver prices, as evidenced by the price plunge on Friday, “could be very sharp indeed,” said Matthew Piggott, director of gold and silver at consulting firm Metals Focus in London.\n\nmore than 30% off the highs, below $80 a troy ounce. The metal’s rise, analysts say, had been supported less by indus- trial demand or lack of supply, and more by speculators bet- ting on a shifting global order and distrust of the U.S. dollar. Some market participants say silver is behaving like “gold squared” and “gold on steroids.” Others compared it to meme-stock mania, or called silver the “devil’s metal,” arguing that a future price drop will surely send many believers to hell. The episode is reminiscent of the 1980 attempt by the Hunt brothers to corner the market. The price then peaked near $50, plunged and failed to attain those levels for four decades until last October. At times in recent months, regional gaps in prices got so wide that it has been profit- able for traders to move silver by air. It normally travels by ship because it is heavy and worth less than gold. “You know where your lug- gage is sitting underneath? You got gold and silver sitting un- derneath there, too,” said Ken Lewis, chief executive of Bullion International Group, which\n\n_ContinuedfrompageB1_\n\n# _Silver Rattles Companies_\n\n### **THE TICKER** | MARKET EVENTS COMING THIS WEEK\n\nver prices support the idea that investors were worried about debasement. The drops came as President Trump chose Kevin Warsh to chair the Federal Re- serve. He is seen as less likely to slash interest rates than the other Kevin who is in the running, Kevin Has- sett. Prices moved on the day mostly as you should expect from the selection of a hard-money hawk, or the avoidance of a soft-money dove: Stocks, gold and silver declined; the dollar and long-term Treasury yields rose. The exception came from the nuance that Warsh wants to sell the central bank’s stockpile of Trea- surys, but he has turned dovish on interest rates, so 2-year Treasury yields fell a little. Stimulus from President Trump’s tax cuts should be in the tens of billions of dol- lars this year, something likely to boost consumer\n\n_ContinuedfrompageB1_\n\nspending, shore up the economy and make inflation a little higher than it other- wise would be. The trouble is that infla- tion fears should have shown up in bond markets, and they just haven’t. Long-term inflation ex- pectations, measured by the break-even for the five years starting in five years’ time, have actually dropped this year and are lower than at the start of last year. Why would debasement fears only show up in pre- cious metals? Indeed, the Swiss franc moved against the dollar ex- actly in line with the euro until Japan’s snap election two weeks ago, suggesting no obvious investor flight to currencies perceived as de- basement-proof before that. Fear of a far weaker dol- lar in the future ought to help U.S. stocks, too. While hardly a safe haven from inflation, stocks ought to have some protection both because of foreign earnings and because sales should rise with consumer prices. Instead, U.S. stocks lagged far behind overseas stocks both this year and last year.\n\n**3) A global boom ignites inflation** The growing confidence in global growth has led to\n\nmarkets that act a lot like the years before the 2008-09 global financial crisis. From 2001 to 2007, inves- tors chose foreign over U.S. stocks, smaller companies over larger, and preferred cheap value stocks to growth stocks. Strong economies boosted copper demand, and the price soared. Gold thrived, too, rising from $273 an ounce at the start of 2001 to $634 by the start of 2007, before the cri- sis came into view. The same happened for the past cou- ple of months as investors fell out of love with big tech- nology compa- nies’ spending on artificial intelligence. Then this year small stocks smashed the biggest, silver surfed a wave of pri- vate buying and gold jumped 21.8% in the first 21 trading days of the year, the most over such a period since late 1999. Copper was already on a roll thanks to data-center construction, but since No- vember soared another 20%, before having its worst day on Friday since the Trump\n\ntariffs last April. Stronger global growth is a reasonable thing to ar- gue for, given Japan’s promised tax cuts, huge military spending in Ger- many, hopes for at least some China stimulus and the possibility of peace in Ukraine. But it doesn’t really ex- plain why the dollar fell this year exactly as much against the yen, where new stimulus is the driving force, as against the euro, where stimu- lus was baked into the price last year, or sterling, where nothing changed. It certainly doesn’t justify the tripling in the price of silver in 12 months. Gold has a role in all these stories, but the scale of the price moves, com- bined with the soaring and crashing of silver, shows a lot of froth. Like the grit in the oys- ter, there’s a solid core of truth inside the vanity of wild price swings. But Friday’s plummeting precious metals prices show the danger of being late to a rally that has gone far be- yond that truth.\n\n# _Parsing Gold, Silver Boom, Bust_\n\n2010 ’15 ’20 ’25\n\n–100\n\n0\n\n100\n\n200\n\n300\n\n400\n\n500\n\n600%\n\nSilver\n\nGold\n\nFutures price performance since the end of 2009\n\nSources: FactSet (futures); LSEG (relative performance)\n\nJan. 2026 9 16 23 30\n\n0\n\n20\n\n40\n\n60%\n\nFutures price performance this year\n\nSilver\n\nGold\n\nTop 50 vs. Russell 2000\n\nRussell 1000 Growth vs. Value\n\nJan. 2026\n\n–12\n\n–10\n\n–8\n\n–6\n\n–4\n\n–2\n\n0%\n\nRelative performance this year\n\n9 16 23 30\n\nFY2016 ’20 ’25\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12 million\n\nNew homes reached by fiber cables\n\nSource: Fiber Broadband Association Note: Fiscal year ends Sept. 30\n\n### Like the grit in oysters, there’s a core of truth in the vanity of price swings.\n\nP2JW033000-0-B00900-1--------XA"],"full_content":"**THE WALL STREET JOURNAL.** Monday, February 2, 2026 | B9\n\nBY PATIENCE HAGGIN\n\nA JKL Associates work site where fiber-optic cabling is being fed through a conduit.\n\nSilver is now the largest cost component of solar cells.\n\nCHINATOPIX/AP\n\n### **Monday**\n\nISM mfg. index Dec., previous 51.9 Jan., expected n.a.\n\nEarningsexpected Estimate/Year Ago\n\nIDEXX Laboratories 2.93/2.62 NXPSemiconductors 3.31/3.18 PalantirTechnologies 0.23/0.14 Simon Property Group 1.90/2.04 Teradyne 1.36/0.95 WaltDisney 1.57/1.76\n\n### **Tuesday**\n\nEarnings expected Estimate/Year Ago\n\nAdvanced Micro Devices 1.32/1.09 Amgen 4.73/5.31 Eaton 3.31/2.83 Merck 2.01/1.72 PepsiCo 2.24/1.96 Pfizer 0.57/0.63\n\n### **Wednesday**\n\nEIA status report Previous change in stocks in millions ofbarrels\n\nCrude-oilstocks down 2.3 Gasoline stocks up 0.2 Distillatesup 0.3\n\nMort.bankersindexes Purch.,previous down0.4% Refinan.,prev. down16.0%\n\nEarningsexpected Estimate/YearAgo\n\nAbbVie 2.65/2.16 Alphabet 2.64/2.15 BostonScientific 0.78/0.70 Eli Lilly 6.95/5.32 Qualcomm 3.39/3.41 Uber Technologies 0.78/3.21\n\n### **Thursday**\n\nEIA report:natural-gas Previous changein stocks in billionsof cubic feet down242\n\nInitial joblessclaims Previous 209,000 Expected 212,000\n\nFreddieMac mortgage survey Previousweeklyaverages\n\n30-yearfixed 6.10% 15-yearfixed 5.49%\n\nEarningsexpected Estimate/YearAgo\n\nAmazon.com 1.96/1.86 Bristol-MyersSquibb 1.23/1.67\n\nConocoPhillips 1.07/1.98 Intercontinental Exchange 1.68/1.52 KKR 1.16/1.32 Linde 4.18/3.97\n\n### **Friday**\n\nNonfarm payrolls Dec.,previous 50,000 Jan., expected 55,000\n\nUnemploymentrate Dec.,previous 4.4% Jan., expected 4.4%\n\nU.Mich. consumer index Jan., final 56.4 Feb.,prelim 54\n\nConsumerCredit Nov.,previous up$4.2bil. Dec.,expected up$8.0 bil.\n\nEarnings expected Estimate/Year Ago\n\nBiogen 1.63/3.44 Carlyle Group 1.00/0.92 CboeGlobal Markets 2.94/2.10 Centene (1.22)/0.80 PhilipMorris International 1.70/1.55 RoivantSciences (0.30)/0.23\n\n\\* FactSet Estimates earnings-per-share estimates don’t include extraordinary items (Losses in paren- theses) u Adjusted for stock split Note: Forecasts are from Dow Jones weekly survey of economists\n\nEAGAR, Ariz.—Sammy Col- lins twists and pulls on a direc- tional drill’s controls. He uses the 15,000-pound vehicle to bore a path through the soil for future fiber-optic lines. About 40 other members of his crew across this rural corner of Apache County dig ditches and hang fiber cables on poles. Collins, who previously worked at an addiction treat- ment facility, joined this crew six months ago, notching two promotions and boosting his pay by more than 22%. Experi- enced drillers can make any- where from the high $20s to mid-$40s an hour. Fiber cables were laid at a record number of homes last year, fueled by billions in fed- eral grants for broadband in- ternet. Technology companies’ appetite for data centers is boosting demand further. Yet the rush to build is hit- ting a human bottleneck, par- ticularly among workers who can piece together the fiber networks at the heart of the technology. Crews can’t hire fast enough and workers often are poached by competitors.\n\nand expects pay to rise even faster, said Chase Lapcinski, Push president. Entry-level fiber workers at Jonesburg, Mo.-based Sellen- riek Construction made about $60,000 including overtime last year. That can jump by 25% to 30% in the first three years, said Chief Executive Steve Sellenriek, and site su- perintendents can make six figures. Sellenriek said he is seeing an unprecedented rush of clients trying to lock in con- tracts early, before wages rise even further. Many workers join crews\n\nwithout prior fiber experience, the way Collins did. Others take training courses before getting hired. Learning Alli- ance in Tampa, Fla., offers a six-week to eight-week boot camp, funded by military bene- fits and corporate donations, that trains students to install fiber. The program’s 2025 graduates made about $26 an hour in starting pay, plus over- time, said Cesar Ruiz, Learning Alliance’s president and CEO. He has seen that rate increase by about 13% in two years. Fiber crews often work more than 10 hours a day out-\n\ndoors in nearly all weather. They may relocate to a project area for months at a time. And winter can bring less work in parts of the country, because of challenges drilling through snow and frozen ground. Mitch Grant, a fiber splicer with K&S Communications, has been spending time lately in Springerville, Ariz. Most locals there don’t have high-speed fi- ber internet, and he says they are eager to get connected—so eager, they stop him at work and ask for updates. (The ser- vice, provided by Commnet’s Ethos Broadband, is expected\n\n# **BUSINESS & FINANCE**\n\n“We just don’t have the peo- ple to do the work,” said Kyle Braude, who manages fiber build projects for JKL Associ- ates, a subsidiary of utility- construction contractor Push. The biggest holdup in the fiber boom, he said, is human capital for roles including drillers, foremen, splicers and aerial linemen. The demand is expected to create 58,000 new jobs be- tween 2025 and 2032, accord- ing to a 2024 report by the Fi- ber Broadband Association and the Power & Communication Contractors Association. With 120,000 workers expected to leave the field in the same pe- riod—mainly through retire- ment—the report forecasts a combined shortage of 178,000 workers. As a result, contrac- tors and analysts predict our AI future might take longer to achieve and a goal of nation- wide high-speed internet could remain out of reach. For those with fiber-laying skills, the work can pay good money, no college needed. Telecommunications line in- stallers and repairers made an- nual median wages of $70,500 for the year ended May 2024, the latest figure available, ac- cording to the Bureau of Labor Statistics. That figure doesn’t include overtime pay. Push raised hourly wages for fiber crews by 5% to 8% in each of the past several years,\n\nto be in place within two to four months.) Grant’s work is delicate: He cleaves each strand of a fiber- optic cable for a clean edge, then uses a machine to fuse the hair-thin filaments, which carry digital data at the speed of light. A slight misalignment, or even a speck of oil from the technician’s hands, and the strands won’t join. “When you first start doing it, it’s like, ‘How can you possi- bly do it?’ But then you get muscle memory,” Grant said. Four years ago, he left cell tower work to manage fiber crews, then saw how much this particular specialty makes— sometimes upward of $30 per splice for projects like this— and moved into that role. Skilled trades across the board are suffering a work- force pinch, and fiber doesn’t have the training pipeline that trades like auto repair and electrical power work enjoy. Lapcinski of Push is working to change that. He hosts hands-on skill days for high- school students in his home state of Wisconsin, so teens can get exposure. He hired farmers, former inmates and white-collar workers tired of office life. “AI isn’t going to replace this,” Lapcinski said. “It will accelerate and support some of it, but it’s not going to go out and dig the hole.”\n\n# **Labor Shortage Hits Broadband Boom**\n\n### Record demand for fiber-optic cables means workers can pull in soaring pay\n\nPAUL RATJE FOR WSJ\n\nowns Apmex, an online pre- cious-metals market. He sold millions of dollars of silver to dealers from Hong Kong, Singa- pore and Japan in recent days. The company’s mints, which source silver from local refineries in North America, are backlogged with orders. “Our company’s revenue before January was 70% gold, 30% silver. And January is probably going to be 70% sil- ver, 30% gold,” he said. On the trading floor of pre- cious-metals group Heraeus in Hanau, Germany, Dominik Sper- zel is fielding calls from inves- tors hunting for silver bars. Heraeus’ refineries, like those elsewhere in Europe, can’t keep up with demand and people cashing in the family silver. “The run on this, it is amaz- ing,” said Sperzel, head of trad- ing for Germany, adding that the buying spree is a surprise because silver, unlike gold, in- curs a sales tax in Germany. Manufacturers that rely on silver are stung by the metal’s explosive rally. Silver is a critical component in solar cells, of which 80% globally are made in China. The solar sector is estimated to have contributed about 17% of total silver demand in 2025, more than double its share from a decade ago and on par with what is used to make jewelry, according to the Silver Insti- tute, an industry group. As silver prices surged, the world’s largest solar-panel makers saw profitability plunge. Some are raising prices and speeding up plans to sub-\n\nstitute silver with cheaper ma- terials such as copper. An average 400-watt resi- dential solar panel contains about 4 to 6 grams of silver, according to Sagar Chopra, a senior research analyst at en- ergy consulting firm Wood Mackenzie. The precious metal is applied as silver paste to form busbars—solid metal strips that conduct electricity. For years, silver accounted for 5% to 7% of the cost of manufacturing a solar panel, Chopra said. Now it has be- come the largest cost compo- nent of solar-cell production, accounting for 15% to 20%. Some of the world’s largest solar-panel companies have is- sued silver-related profit warnings. Tongwei Solar, which boasts 14% of the global solar-cell market, said it would report a loss of up to $1.4 bil- lion in its 2025 results because\n\nof surging silver prices. Since January, prices for a typical solar cell in China have risen 10% from 4.96 cents per watt to 5.47 cents per watt, according to OPIS, an energy- data and analytics provider owned by the same parent as The Wall Street Journal. Eventually, refiners might process enough scrap to bring silver prices down. Industrial users are looking to minimize their silver consumption. Esti- mated industrial demand for sil- ver fell 2% last year thanks to a sharp drop in the amount of sil- ver used in each module, ac- cording to the Silver Institute. The constraints fueling wild moves in silver prices are financial as well as physical. The cost to borrow silver—es- pecially in London—soared last year after tons of silver moved across the Atlantic on fears that President Trump was going to impose tariffs on the metal. He didn’t impose the tariffs, but the borrowing costs remain high. The high borrowing costs, combined with the soaring price of the metal, has stretched the balance sheets of the banks, refiners and re- cycling firms that grease pre- cious-metal markets. That has curtailed the ability of finan- cial institutions to step in and temper big price moves. Changes in the direction of silver prices, as evidenced by the price plunge on Friday, “could be very sharp indeed,” said Matthew Piggott, director of gold and silver at consulting firm Metals Focus in London.\n\nmore than 30% off the highs, below $80 a troy ounce. The metal’s rise, analysts say, had been supported less by indus- trial demand or lack of supply, and more by speculators bet- ting on a shifting global order and distrust of the U.S. dollar. Some market participants say silver is behaving like “gold squared” and “gold on steroids.” Others compared it to meme-stock mania, or called silver the “devil’s metal,” arguing that a future price drop will surely send many believers to hell. The episode is reminiscent of the 1980 attempt by the Hunt brothers to corner the market. The price then peaked near $50, plunged and failed to attain those levels for four decades until last October. At times in recent months, regional gaps in prices got so wide that it has been profit- able for traders to move silver by air. It normally travels by ship because it is heavy and worth less than gold. “You know where your lug- gage is sitting underneath? You got gold and silver sitting un- derneath there, too,” said Ken Lewis, chief executive of Bullion International Group, which\n\n_ContinuedfrompageB1_\n\n# _Silver Rattles Companies_\n\n### **THE TICKER** | MARKET EVENTS COMING THIS WEEK\n\nver prices support the idea that investors were worried about debasement. The drops came as President Trump chose Kevin Warsh to chair the Federal Re- serve. He is seen as less likely to slash interest rates than the other Kevin who is in the running, Kevin Has- sett. Prices moved on the day mostly as you should expect from the selection of a hard-money hawk, or the avoidance of a soft-money dove: Stocks, gold and silver declined; the dollar and long-term Treasury yields rose. The exception came from the nuance that Warsh wants to sell the central bank’s stockpile of Trea- surys, but he has turned dovish on interest rates, so 2-year Treasury yields fell a little. Stimulus from President Trump’s tax cuts should be in the tens of billions of dol- lars this year, something likely to boost consumer\n\n_ContinuedfrompageB1_\n\nspending, shore up the economy and make inflation a little higher than it other- wise would be. The trouble is that infla- tion fears should have shown up in bond markets, and they just haven’t. Long-term inflation ex- pectations, measured by the break-even for the five years starting in five years’ time, have actually dropped this year and are lower than at the start of last year. Why would debasement fears only show up in pre- cious metals? Indeed, the Swiss franc moved against the dollar ex- actly in line with the euro until Japan’s snap election two weeks ago, suggesting no obvious investor flight to currencies perceived as de- basement-proof before that. Fear of a far weaker dol- lar in the future ought to help U.S. stocks, too. While hardly a safe haven from inflation, stocks ought to have some protection both because of foreign earnings and because sales should rise with consumer prices. Instead, U.S. stocks lagged far behind overseas stocks both this year and last year.\n\n**3) A global boom ignites inflation** The growing confidence in global growth has led to\n\nmarkets that act a lot like the years before the 2008-09 global financial crisis. From 2001 to 2007, inves- tors chose foreign over U.S. stocks, smaller companies over larger, and preferred cheap value stocks to growth stocks. Strong economies boosted copper demand, and the price soared. Gold thrived, too, rising from $273 an ounce at the start of 2001 to $634 by the start of 2007, before the cri- sis came into view. The same happened for the past cou- ple of months as investors fell out of love with big tech- nology compa- nies’ spending on artificial intelligence. Then this year small stocks smashed the biggest, silver surfed a wave of pri- vate buying and gold jumped 21.8% in the first 21 trading days of the year, the most over such a period since late 1999. Copper was already on a roll thanks to data-center construction, but since No- vember soared another 20%, before having its worst day on Friday since the Trump\n\ntariffs last April. Stronger global growth is a reasonable thing to ar- gue for, given Japan’s promised tax cuts, huge military spending in Ger- many, hopes for at least some China stimulus and the possibility of peace in Ukraine. But it doesn’t really ex- plain why the dollar fell this year exactly as much against the yen, where new stimulus is the driving force, as against the euro, where stimu- lus was baked into the price last year, or sterling, where nothing changed. It certainly doesn’t justify the tripling in the price of silver in 12 months. Gold has a role in all these stories, but the scale of the price moves, com- bined with the soaring and crashing of silver, shows a lot of froth. Like the grit in the oys- ter, there’s a solid core of truth inside the vanity of wild price swings. But Friday’s plummeting precious metals prices show the danger of being late to a rally that has gone far be- yond that truth.\n\n# _Parsing Gold, Silver Boom, Bust_\n\n2010 ’15 ’20 ’25\n\n–100\n\n0\n\n100\n\n200\n\n300\n\n400\n\n500\n\n600%\n\nSilver\n\nGold\n\nFutures price performance since the end of 2009\n\nSources: FactSet (futures); LSEG (relative performance)\n\nJan. 2026 9 16 23 30\n\n0\n\n20\n\n40\n\n60%\n\nFutures price performance this year\n\nSilver\n\nGold\n\nTop 50 vs. Russell 2000\n\nRussell 1000 Growth vs. Value\n\nJan. 2026\n\n–12\n\n–10\n\n–8\n\n–6\n\n–4\n\n–2\n\n0%\n\nRelative performance this year\n\n9 16 23 30\n\nFY2016 ’20 ’25\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12 million\n\nNew homes reached by fiber cables\n\nSource: Fiber Broadband Association Note: Fiscal year ends Sept. 30\n\n### Like the grit in oysters, there’s a core of truth in the vanity of price swings.\n\nP2JW033000-0-B00900-1--------XA","publish_date":null,"title":"Labor Shortage Hits Broadband Boom","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/pdf_f270146f8ca7_p25.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/pdf_f270146f8ca7_p25.pdf"},"request":{"example_id":"chart/pdf_f270146f8ca7_p25","product_type":"parse","source_file_path":"pdf_f270146f8ca7_p25.pdf"},"started_at":"2026-11-04T19:45:21.253952"} +{"completed_at":"2026-11-04T19:45:26.300146","latency_in_ms":3575,"output":{"example_id":"chart/pdf_f270146f8ca7_p4","grounded_pages":[],"layout_pages":[],"markdown":"A4 | Monday, February 2, 2026 **P W L C 10 11 12 H T G K R F A M 1 2 3 4 5 6 7 8 9 O I X X** \\* \\* \\* \\* \\* **THE WALL STREET JOURNAL.**\n\nWASHINGTON—President Trump’s lawsuit against the government that he runs pres- ents a mind-bending minefield of conflicts that could end with the president’s appoin- tees approving a federal pay- out to him.\n\nTrump is simultaneously supervising the private attor- neys demanding at least $10 billion in damages over the past disclosure of his tax records, while also overseeing the senior Justice Department attorneys with authority to settle the case and authorize a payment. “This creates the risk of the most collusive lawsuit of all time, because it is ulti- mately the president suing a defendant whom he says has to do whatever he directs,” said Peter Keisler, who served as acting attorney general in the George W. Bush adminis- tration. Trump, two of his sons and his businesses sued the gov- ernment in federal court in Miami last week, arguing that they were harmed when for- mer IRS contractor Charles\n\n_By_ _Richard Rubin_ _,_ _C. Ryan Barber_ _and_ _Annie Linskey_\n\n# **U.S. NEWS**\n\nLittlejohn illegally disclosed their tax information to the New York Times and ProPub- lica in 2019 and 2020. “Defendants have caused Plaintiffs reputational and fi- nancial harm, public embar- rassment, unfairly tarnished their business reputations, portrayed them in a false light, and negatively affected President Trump, and the other Plaintiffs’ public stand- ing,” the lawsuit said. Littlejohn, who is serving a five-year sentence in federal prison, pleaded guilty in 2023 and said he intentionally leaked Trump’s information before the 2020 election. The disclosures showed Trump regularly reported losses and paid just $750 in federal in- come taxes in 2016 and in 2017. Littlejohn also disclosed tax data of thousands of other wealthy Americans. Trump said on Saturday that he recognized the “strange position” created by the lawsuit he filed and said he would likely donate pro- ceeds from a settlement to charities such as the American Cancer Society. “If I pay myself, that some- how will never look good,” he said. “Nobody cares how much if it goes to good charities. So you settle by giving charities a lot of money. And I think we’re going to do something like that.” The Justice Department hasn’t yet responded to Trump’s lawsuit, which was\n\nassigned to Judge Kathleen Williams, an appointee of for- mer President Barack Obama. The Treasury Department and Internal Revenue Service didn’t comment. On one crucial question, there’s already a conflict be- tween the president’s stance and the government’s previ- ously stated position. If Little- john was an IRS employee, the government can be sued. If he was a contractor, Trump can’t sue the government. Trump’s lawsuit notes Lit- tlejohn’s vast access to tax- payer information and day-to- day supervision by IRS employees as evidence that he was effectively a government employee. In a separate case brought by billionaire Ken Griffin in 2022, a federal judge let that argument proceed through early stages of litiga- tion. That case was settled with an apology from the gov- ernment to Griffin, before the judge reached a decision. The DOJ argued then that Littlejohn wasn’t an IRS em- ployee. It reiterated that posi- tion as recently as two months ago—during the Trump administration. “At all times, Mr. Littlejohn was a contractor who wrong- fully disclosed taxpayer return information,” the government wrote in a November 2025 court filing in Maryland. “The United States cannot be held liable for his illegal conduct.” Moreover, Littlejohn’s ille- gal conduct happened during\n\nTrump’s first term. That is, the plaintiff’s complaints about IRS failures and negli- gence pin the blame on offi- cials who were supervised by that very plaintiff. Normally, Justice Depart- ment tax attorneys who focus on this area of the law would handle the case. They would likely raise several challenges to Trump’s claims beyond ar- guing that Littlejohn wasn’t an IRS employee. The government could also argue that Trump filed the case too late, said Francesca Ugolini, former chief of the appellate section in DOJ’s tax division. The statute says suits can be filed within two years of the “date of discovery” of the dis- closure. The first New York Times story stemming from\n\nLittlejohn’s disclosure was published in September 2020, though the documents’ prove- nance wasn’t clear then. Little- john was charged in September 2023. Trump, who broke a tra- dition of voluntary-tax-return disclosure by presidents and candidates, filed the lawsuit exactly two years after receiv- ing a formal IRS notice about the unauthorized release. Another point of contention would be the damages calcula- tion. The $10 billion figure isn’t punitive damages. In- stead, it’s a $1,000-per-disclo- sure level specified in the law, multiplied by an estimate of the number of times that Trump’s tax information was published or broadcast. However, the law specifies $1,000 for each unauthorized disclosure. The government could argue that the only un- authorized disclosures were the ones that Littlejohn made. Ugolini said the DOJ’s model should be its handling of Hunter Biden’s tax-disclo- sure suit against the govern- ment. While his father was president, the son sought damages for comments by IRS agents who were raising con- cerns to Congress about the handling of the government’s criminal investigation into him. “The career attorneys at DOJ were allowed to handle the case in the normal way, raising defenses for the United States,” said Ugolini, now at SouthBank Legal.\n\nHunter Biden pleaded guilty to tax crimes in 2024. He dropped the lawsuit against the government last April, just months after his fa- ther pardoned him and Trump returned to the White House. Another potential route is a settlement. Trump has filed other law- suits seeking billions of dol- lars in damages, including one against the publisher of The Wall Street Journal. His gov- ernmental power can create incentives for defendants to settle. Since his 2024 election victory, Disney’s ABC News, Meta Platforms, Alphabet’s YouTube and Paramount Global—the owner of CBS— have agreed to multimillion- dollar payments to settle liti- gation initiated by Trump. The tax case is distinct be- cause Trump doesn’t just have leverage over the defendants. He has control. Trump has separately filed administrative claims demanding millions of dollars from the Justice De- partment over past criminal investigations into him. Asked in October about re- porting that he was seeking more than $200 million, Trump appeared to acknowl- edge the awkwardness of pur- suing payment from the gov- ernment he leads. Trump said then that he was “damaged very greatly” and would give any money to charity. He said on Saturday that he still needs to work out a settlement in that matter.\n\n# **Trump Straddles Both Sides of IRS Suit**\n\n### President is plaintiff, but also oversees people who could decide on a payout\n\non Democrats in the House, a person familiar with the mat- ter said. That meant the cham- ber couldn’t quickly pass the package on Monday through a fast-track provision on a two- thirds vote, and would instead have to go through the “rule” process that would push a vote until Tuesday at the earliest. Jeffries said the Senate vote was a step in the right direc- tion but argued that House Democrats weren’t a part of any Senate deal. He ticked off a series of changes demanded by Democrats, such as a higher threshold for searches and ar- rests, mandated use of body cameras and a ban on masks.\n\nJeffries echoed calls from many Democrats who say even two weeks of funding for DHS is too much. Some Democrats have called for Immigration and Customs Enforcement to be eliminated altogether, after a month in which immigration agents surged into Minneapo- lis, fatally shooting two U.S. citizens to death and prompt- ing protests. “The administration can’t just talk the talk. They need to walk the walk. That should be- gin today—not in two weeks, today,” Jeffries said on Sunday on ABC. Johnson, in an interview on Fox News, dismissed the calls\n\nfor no masks and a require- ment for judicial warrants for apprehensions and searches, but he did speak in favor of body cameras on ICE officers. In a separate appearance on NBC, Johnson said some of the Democrats’ conditions “are ob- viously reasonable and should happen. But others are going to require a lot more negotiation.” The House Democrats’ deci- sion to play hardball has thrown new uncertainty into resolving the partial shutdown. Just last month, Johnson celebrated passing the final six of the 12 annual individual spending bills and sent them to the Senate, a rare occurrence in recent de-\n\ncades when terms like omnibus became commonplace. But an- ger over the death of Pretti made the path to passage more difficult, prompting the Senate to change the House package and send it back this week. Johnson’s tiny vote margin gives any one or two GOP members leverage if they want changes or concessions. John- son as soon as Monday will swear in Christian Menefee, a Democrat who won a special election to represent Texas’ 18th Congressional District af- ter Rep. Sylvester Turner’s death last year. That will bring the majority to 218-214. And the expected vote comes as some conservatives aren’t happy with the two- week extension of the DHS bill, instead wanting a four- to six- week extension. The conserva- tive House Freedom Caucus in a letter to Trump on Tuesday backed the efforts of ICE and urged the administration to “use all tools necessary.” Other GOP lawmakers have said they would demand that the VOTE Act—a measure that imposes stricter requirements for vot- ing, among other changes—be included in any package. Activists and rank-and-file Democrats have rallied around denying ICE any funding. “I’m not a no. I’m a firm no,” said Ro Khanna (D., Calif.) on NBC. “I just in good con- science cannot vote to give more money to ICE agents as they’re violating our constitu- tional rights,” he said. Trump last week moved to\n\npull back on his administra- tion’s immigration-enforce- ment campaign, including re- placing the leader of the Minneapolis operation. But he hasn’t made clear what kind of changes he would accept on immigration-enforcement prac- tices. The White House, like many congressional Republi- cans, had initially pushed for administration changes or ex- ecutive actions, rather than new law passed by Congress. On Saturday night, Trump backed ICE and gave a warning to protesters. “If they do any- thing bad to our people then they will have to suffer,” he said. Asked about body cam- eras for immigration agents, he said Sunday: “We’ll be talking about that in the near future.” Some Democrats said they became further dug in on op- posing any ICE funding after Trump returned to his more forceful rhetoric against pro- testers, and continued to back Homeland Security Secretary Kristi Noem, who many Demo- crats and some Republicans have said should resign. The partial shutdown that started Saturday morning af- fects the agencies attached to the six bills that haven’t yet passed the House, including DHS and the departments of Defense and Transportation. In departments where funding lapses, federal employees are set to be furloughed or work without pay until funding is restored. Federal employees typically get back pay when a shutdown ends.\n\nthem that we need to imple- ment the president’s play call,” giving the White House time to discuss changes to DHS with Democrats. House Republicans will have to swallow frustration over what they see as being painted into a corner by Democrats, and some have floated their own demands. Asked if he ex- pects Trump to help twist arms to get the necessary sup- port, the GOP leader replied: “I’m sure he will get on the phone with them.” The Senate on Friday passed the bipartisan plan struck between Senate Demo- crats and the White House to fund the government, just days after the killing of Alex Pretti by immigration agents in Min- neapolis the previous weekend. But with the House out and not returning until Monday, a funding lapse was triggered Saturday morning. While the Senate vote drew the support of Minority Leader Chuck Schumer (D., N.Y.) and about half of the Senate Demo- cratic caucus, House Minority Leader Hakeem Jeffries (D., N.Y.) told Johnson in a phone call on Saturday not to count\n\n_ContinuedfromPageOne_\n\n# _Funding Runs Into Pushback_\n\nvideos, called NAND, as well as the memory used to run apps quickly, called DRAM. By the end of this year, the price of DRAM will quadruple from 2023 levels, and NAND will more than triple, estimates TechInsights. Howard estimates that Ap- ple could pay $57 more for the two types of memory that go into the base-model iPhone 18 due this fall compared with the base model iPhone 17 cur- rently on sale. For a device that retails for $799, that would be a big hit to profit margins. Apple’s purchasing power and expertise in designing ad- vanced electronics long made it an unrivaled Goliath among the Asian companies that make most of the iPhone’s parts and assemble the device. Apple spends billions of dollars a year on NAND, for in- stance, people familiar with the figures said, likely making it the single biggest buyer globally. Suppliers flocked to win Apple’s business, hoping to leverage its know-how and prestige to attract other cus- tomers. These days, however, “the companies now pushing the boundaries of human‑scale en- gineering are the ones like Nvidia,” said Ming-chi Kuo, an analyst with TF International Securities.\n\nagainst the iPhone maker. Apple signs long-term con- tracts for memory, but it has used its heft to squeeze suppli- ers. Its contracts have empow- ered it to negotiate prices as often as weekly, and to even refuse to buy any memory from a supplier if Apple didn’t view the price as favorable, people familiar with its mem- ory purchases said. To boost leverage with suppliers, Apple even began stocking more in- ventory of memory. That was atypical for Cook, who nor- mally cuts inventory to the bone to maximize Apple’s cash flow. Apple is fighting not only for current deliveries but also for the attention of engineers at suppliers. Glass scientists who worked on developing the smoothest and lightest smart- phone displays are now also spending time on specialized glass for packaging advanced AI processing chips, industry executives said. Makers of sensors and other gizmos inside the iPhone are winning new business from AI companies such as OpenAI that are developing their own hardware. Still, suppliers said they were far from giving up on business with Apple. Working with Apple is a form of educa- tion, they said, because it re- mains one of the most de-\n\nmanding and disciplined customers in the industry. TSMC, the Taiwanese chip manufacturer, has built suc- cessive generations of its most advanced chips with Apple as its lead customer, relying on the big predictable demand for iPhones. Now that TSMC is do- ing more business with Nvidia and other AI companies, peo- ple with knowledge of the chip supply chain said Apple was exploring whether some lower- end processors could be made by someone other than TSMC. One of Apple’s biggest profit spinners is selling extra memory for far more than the memory chips cost the com- pany. Last fall, Apple discon- tinued the iPhone Pro model with 128 gigabytes of storage. Customers who want that model must now start at 256 gigabytes and pay $100 more—the type of move that could be repeated this year to help Apple offset higher costs, wrote Craig Moffett, an ana- lyst at Moffett Nathanson, in an investor note. However, Apple isn’t ex- pected to raise the price of its next iPhone models over simi- larly equipped iPhone 17s, said Kuo, the analyst. News Corp, owner of The Wall Street Journal, has a commercial agreement to sup- ply news through Apple ser- vices.\n\ntor Manufacturing, Nvidia Chief Executive Jensen Huang said on a podcast. Apple had been TSMC’s biggest customer by a wide margin for years. TSMC is the world’s leading manufacturer of advanced chips for AI servers, smart- phones and other computing devices. Spokesmen for Apple and TSMC declined to comment. The big computers that handle AI tasks don’t look like the smartphones consumers own, but many companies sup- ply components for both. In particular, memory chips are in short supply as companies such as OpenAI, Alphabet’s Google, Meta, Microsoft and others collectively spend hun- dreds of billions of dollars to build AI computing capacity. “The rate of increase in the price of memory is unprece- dented,” said Mike Howard, an analyst for research firm TechInsights. That applies both to the flash memory chips that store photos and\n\n_ContinuedfromPageOne_\n\n# _Apple’s Margins Threatened_\n\nDemand for AI hardware is poised to keep growing rap- idly. Apple’s spending growth is modest in comparison with what is being spent to fill up AI data centers, even though it is breaking records with huge sales of the iPhone 17. Samsung Electronics and\n\nSK Hynix are raising the price of a type of DRAM chip for Apple, people familiar with Apple’s supply chain said. Big AI companies pay generously and are willing to lock in sup- ply and make upfront pay- ments, giving the South Ko- rean chip makers leverage\n\nFY2023 ’24 ’25 ’26\n\n0\n\n2\n\n4\n\n6\n\n$8 Price per gigabyte for DRAM, the main system memory in AI servers and smartphones, quarterly\n\n2020 ’21 ’22 ’23 ’24 ’25\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60% High Performance Computing **\\*** Smartphone\n\nProportion of revenue by market segment, quarterly\n\nSources: TechInsights (price per gigabyte for DRAM); TSMC (revenue) \\*Includes AI-related chips\n\nProjections\n\nPresident Trump speaking at Mar-a-Lago on Sunday.\n\nAL DRAGO/GETTY IMAGES\n\nHouse Minority Leader Hakeem Jeffries (D., N.Y.) is seeking to rein in immigration agents’ tactics.\n\nGRAEME SLOAN/BLOOMBERG NEWS\n\nP2JW033000-5-A00400-16DDF61078D","pages":[{"line_numbers":[],"markdown":"A4 | Monday, February 2, 2026 **P W L C 10 11 12 H T G K R F A M 1 2 3 4 5 6 7 8 9 O I X X** \\* \\* \\* \\* \\* **THE WALL STREET JOURNAL.**\n\nWASHINGTON—President Trump’s lawsuit against the government that he runs pres- ents a mind-bending minefield of conflicts that could end with the president’s appoin- tees approving a federal pay- out to him.\n\nTrump is simultaneously supervising the private attor- neys demanding at least $10 billion in damages over the past disclosure of his tax records, while also overseeing the senior Justice Department attorneys with authority to settle the case and authorize a payment. “This creates the risk of the most collusive lawsuit of all time, because it is ulti- mately the president suing a defendant whom he says has to do whatever he directs,” said Peter Keisler, who served as acting attorney general in the George W. Bush adminis- tration. Trump, two of his sons and his businesses sued the gov- ernment in federal court in Miami last week, arguing that they were harmed when for- mer IRS contractor Charles\n\n_By_ _Richard Rubin_ _,_ _C. Ryan Barber_ _and_ _Annie Linskey_\n\n# **U.S. NEWS**\n\nLittlejohn illegally disclosed their tax information to the New York Times and ProPub- lica in 2019 and 2020. “Defendants have caused Plaintiffs reputational and fi- nancial harm, public embar- rassment, unfairly tarnished their business reputations, portrayed them in a false light, and negatively affected President Trump, and the other Plaintiffs’ public stand- ing,” the lawsuit said. Littlejohn, who is serving a five-year sentence in federal prison, pleaded guilty in 2023 and said he intentionally leaked Trump’s information before the 2020 election. The disclosures showed Trump regularly reported losses and paid just $750 in federal in- come taxes in 2016 and in 2017. Littlejohn also disclosed tax data of thousands of other wealthy Americans. Trump said on Saturday that he recognized the “strange position” created by the lawsuit he filed and said he would likely donate pro- ceeds from a settlement to charities such as the American Cancer Society. “If I pay myself, that some- how will never look good,” he said. “Nobody cares how much if it goes to good charities. So you settle by giving charities a lot of money. And I think we’re going to do something like that.” The Justice Department hasn’t yet responded to Trump’s lawsuit, which was\n\nassigned to Judge Kathleen Williams, an appointee of for- mer President Barack Obama. The Treasury Department and Internal Revenue Service didn’t comment. On one crucial question, there’s already a conflict be- tween the president’s stance and the government’s previ- ously stated position. If Little- john was an IRS employee, the government can be sued. If he was a contractor, Trump can’t sue the government. Trump’s lawsuit notes Lit- tlejohn’s vast access to tax- payer information and day-to- day supervision by IRS employees as evidence that he was effectively a government employee. In a separate case brought by billionaire Ken Griffin in 2022, a federal judge let that argument proceed through early stages of litiga- tion. That case was settled with an apology from the gov- ernment to Griffin, before the judge reached a decision. The DOJ argued then that Littlejohn wasn’t an IRS em- ployee. It reiterated that posi- tion as recently as two months ago—during the Trump administration. “At all times, Mr. Littlejohn was a contractor who wrong- fully disclosed taxpayer return information,” the government wrote in a November 2025 court filing in Maryland. “The United States cannot be held liable for his illegal conduct.” Moreover, Littlejohn’s ille- gal conduct happened during\n\nTrump’s first term. That is, the plaintiff’s complaints about IRS failures and negli- gence pin the blame on offi- cials who were supervised by that very plaintiff. Normally, Justice Depart- ment tax attorneys who focus on this area of the law would handle the case. They would likely raise several challenges to Trump’s claims beyond ar- guing that Littlejohn wasn’t an IRS employee. The government could also argue that Trump filed the case too late, said Francesca Ugolini, former chief of the appellate section in DOJ’s tax division. The statute says suits can be filed within two years of the “date of discovery” of the dis- closure. The first New York Times story stemming from\n\nLittlejohn’s disclosure was published in September 2020, though the documents’ prove- nance wasn’t clear then. Little- john was charged in September 2023. Trump, who broke a tra- dition of voluntary-tax-return disclosure by presidents and candidates, filed the lawsuit exactly two years after receiv- ing a formal IRS notice about the unauthorized release. Another point of contention would be the damages calcula- tion. The $10 billion figure isn’t punitive damages. In- stead, it’s a $1,000-per-disclo- sure level specified in the law, multiplied by an estimate of the number of times that Trump’s tax information was published or broadcast. However, the law specifies $1,000 for each unauthorized disclosure. The government could argue that the only un- authorized disclosures were the ones that Littlejohn made. Ugolini said the DOJ’s model should be its handling of Hunter Biden’s tax-disclo- sure suit against the govern- ment. While his father was president, the son sought damages for comments by IRS agents who were raising con- cerns to Congress about the handling of the government’s criminal investigation into him. “The career attorneys at DOJ were allowed to handle the case in the normal way, raising defenses for the United States,” said Ugolini, now at SouthBank Legal.\n\nHunter Biden pleaded guilty to tax crimes in 2024. He dropped the lawsuit against the government last April, just months after his fa- ther pardoned him and Trump returned to the White House. Another potential route is a settlement. Trump has filed other law- suits seeking billions of dol- lars in damages, including one against the publisher of The Wall Street Journal. His gov- ernmental power can create incentives for defendants to settle. Since his 2024 election victory, Disney’s ABC News, Meta Platforms, Alphabet’s YouTube and Paramount Global—the owner of CBS— have agreed to multimillion- dollar payments to settle liti- gation initiated by Trump. The tax case is distinct be- cause Trump doesn’t just have leverage over the defendants. He has control. Trump has separately filed administrative claims demanding millions of dollars from the Justice De- partment over past criminal investigations into him. Asked in October about re- porting that he was seeking more than $200 million, Trump appeared to acknowl- edge the awkwardness of pur- suing payment from the gov- ernment he leads. Trump said then that he was “damaged very greatly” and would give any money to charity. He said on Saturday that he still needs to work out a settlement in that matter.\n\n# **Trump Straddles Both Sides of IRS Suit**\n\n### President is plaintiff, but also oversees people who could decide on a payout\n\non Democrats in the House, a person familiar with the mat- ter said. That meant the cham- ber couldn’t quickly pass the package on Monday through a fast-track provision on a two- thirds vote, and would instead have to go through the “rule” process that would push a vote until Tuesday at the earliest. Jeffries said the Senate vote was a step in the right direc- tion but argued that House Democrats weren’t a part of any Senate deal. He ticked off a series of changes demanded by Democrats, such as a higher threshold for searches and ar- rests, mandated use of body cameras and a ban on masks.\n\nJeffries echoed calls from many Democrats who say even two weeks of funding for DHS is too much. Some Democrats have called for Immigration and Customs Enforcement to be eliminated altogether, after a month in which immigration agents surged into Minneapo- lis, fatally shooting two U.S. citizens to death and prompt- ing protests. “The administration can’t just talk the talk. They need to walk the walk. That should be- gin today—not in two weeks, today,” Jeffries said on Sunday on ABC. Johnson, in an interview on Fox News, dismissed the calls\n\nfor no masks and a require- ment for judicial warrants for apprehensions and searches, but he did speak in favor of body cameras on ICE officers. In a separate appearance on NBC, Johnson said some of the Democrats’ conditions “are ob- viously reasonable and should happen. But others are going to require a lot more negotiation.” The House Democrats’ deci- sion to play hardball has thrown new uncertainty into resolving the partial shutdown. Just last month, Johnson celebrated passing the final six of the 12 annual individual spending bills and sent them to the Senate, a rare occurrence in recent de-\n\ncades when terms like omnibus became commonplace. But an- ger over the death of Pretti made the path to passage more difficult, prompting the Senate to change the House package and send it back this week. Johnson’s tiny vote margin gives any one or two GOP members leverage if they want changes or concessions. John- son as soon as Monday will swear in Christian Menefee, a Democrat who won a special election to represent Texas’ 18th Congressional District af- ter Rep. Sylvester Turner’s death last year. That will bring the majority to 218-214. And the expected vote comes as some conservatives aren’t happy with the two- week extension of the DHS bill, instead wanting a four- to six- week extension. The conserva- tive House Freedom Caucus in a letter to Trump on Tuesday backed the efforts of ICE and urged the administration to “use all tools necessary.” Other GOP lawmakers have said they would demand that the VOTE Act—a measure that imposes stricter requirements for vot- ing, among other changes—be included in any package. Activists and rank-and-file Democrats have rallied around denying ICE any funding. “I’m not a no. I’m a firm no,” said Ro Khanna (D., Calif.) on NBC. “I just in good con- science cannot vote to give more money to ICE agents as they’re violating our constitu- tional rights,” he said. Trump last week moved to\n\npull back on his administra- tion’s immigration-enforce- ment campaign, including re- placing the leader of the Minneapolis operation. But he hasn’t made clear what kind of changes he would accept on immigration-enforcement prac- tices. The White House, like many congressional Republi- cans, had initially pushed for administration changes or ex- ecutive actions, rather than new law passed by Congress. On Saturday night, Trump backed ICE and gave a warning to protesters. “If they do any- thing bad to our people then they will have to suffer,” he said. Asked about body cam- eras for immigration agents, he said Sunday: “We’ll be talking about that in the near future.” Some Democrats said they became further dug in on op- posing any ICE funding after Trump returned to his more forceful rhetoric against pro- testers, and continued to back Homeland Security Secretary Kristi Noem, who many Demo- crats and some Republicans have said should resign. The partial shutdown that started Saturday morning af- fects the agencies attached to the six bills that haven’t yet passed the House, including DHS and the departments of Defense and Transportation. In departments where funding lapses, federal employees are set to be furloughed or work without pay until funding is restored. Federal employees typically get back pay when a shutdown ends.\n\nthem that we need to imple- ment the president’s play call,” giving the White House time to discuss changes to DHS with Democrats. House Republicans will have to swallow frustration over what they see as being painted into a corner by Democrats, and some have floated their own demands. Asked if he ex- pects Trump to help twist arms to get the necessary sup- port, the GOP leader replied: “I’m sure he will get on the phone with them.” The Senate on Friday passed the bipartisan plan struck between Senate Demo- crats and the White House to fund the government, just days after the killing of Alex Pretti by immigration agents in Min- neapolis the previous weekend. But with the House out and not returning until Monday, a funding lapse was triggered Saturday morning. While the Senate vote drew the support of Minority Leader Chuck Schumer (D., N.Y.) and about half of the Senate Demo- cratic caucus, House Minority Leader Hakeem Jeffries (D., N.Y.) told Johnson in a phone call on Saturday not to count\n\n_ContinuedfromPageOne_\n\n# _Funding Runs Into Pushback_\n\nvideos, called NAND, as well as the memory used to run apps quickly, called DRAM. By the end of this year, the price of DRAM will quadruple from 2023 levels, and NAND will more than triple, estimates TechInsights. Howard estimates that Ap- ple could pay $57 more for the two types of memory that go into the base-model iPhone 18 due this fall compared with the base model iPhone 17 cur- rently on sale. For a device that retails for $799, that would be a big hit to profit margins. Apple’s purchasing power and expertise in designing ad- vanced electronics long made it an unrivaled Goliath among the Asian companies that make most of the iPhone’s parts and assemble the device. Apple spends billions of dollars a year on NAND, for in- stance, people familiar with the figures said, likely making it the single biggest buyer globally. Suppliers flocked to win Apple’s business, hoping to leverage its know-how and prestige to attract other cus- tomers. These days, however, “the companies now pushing the boundaries of human‑scale en- gineering are the ones like Nvidia,” said Ming-chi Kuo, an analyst with TF International Securities.\n\nagainst the iPhone maker. Apple signs long-term con- tracts for memory, but it has used its heft to squeeze suppli- ers. Its contracts have empow- ered it to negotiate prices as often as weekly, and to even refuse to buy any memory from a supplier if Apple didn’t view the price as favorable, people familiar with its mem- ory purchases said. To boost leverage with suppliers, Apple even began stocking more in- ventory of memory. That was atypical for Cook, who nor- mally cuts inventory to the bone to maximize Apple’s cash flow. Apple is fighting not only for current deliveries but also for the attention of engineers at suppliers. Glass scientists who worked on developing the smoothest and lightest smart- phone displays are now also spending time on specialized glass for packaging advanced AI processing chips, industry executives said. Makers of sensors and other gizmos inside the iPhone are winning new business from AI companies such as OpenAI that are developing their own hardware. Still, suppliers said they were far from giving up on business with Apple. Working with Apple is a form of educa- tion, they said, because it re- mains one of the most de-\n\nmanding and disciplined customers in the industry. TSMC, the Taiwanese chip manufacturer, has built suc- cessive generations of its most advanced chips with Apple as its lead customer, relying on the big predictable demand for iPhones. Now that TSMC is do- ing more business with Nvidia and other AI companies, peo- ple with knowledge of the chip supply chain said Apple was exploring whether some lower- end processors could be made by someone other than TSMC. One of Apple’s biggest profit spinners is selling extra memory for far more than the memory chips cost the com- pany. Last fall, Apple discon- tinued the iPhone Pro model with 128 gigabytes of storage. Customers who want that model must now start at 256 gigabytes and pay $100 more—the type of move that could be repeated this year to help Apple offset higher costs, wrote Craig Moffett, an ana- lyst at Moffett Nathanson, in an investor note. However, Apple isn’t ex- pected to raise the price of its next iPhone models over simi- larly equipped iPhone 17s, said Kuo, the analyst. News Corp, owner of The Wall Street Journal, has a commercial agreement to sup- ply news through Apple ser- vices.\n\ntor Manufacturing, Nvidia Chief Executive Jensen Huang said on a podcast. Apple had been TSMC’s biggest customer by a wide margin for years. TSMC is the world’s leading manufacturer of advanced chips for AI servers, smart- phones and other computing devices. Spokesmen for Apple and TSMC declined to comment. The big computers that handle AI tasks don’t look like the smartphones consumers own, but many companies sup- ply components for both. In particular, memory chips are in short supply as companies such as OpenAI, Alphabet’s Google, Meta, Microsoft and others collectively spend hun- dreds of billions of dollars to build AI computing capacity. “The rate of increase in the price of memory is unprece- dented,” said Mike Howard, an analyst for research firm TechInsights. That applies both to the flash memory chips that store photos and\n\n_ContinuedfromPageOne_\n\n# _Apple’s Margins Threatened_\n\nDemand for AI hardware is poised to keep growing rap- idly. Apple’s spending growth is modest in comparison with what is being spent to fill up AI data centers, even though it is breaking records with huge sales of the iPhone 17. Samsung Electronics and\n\nSK Hynix are raising the price of a type of DRAM chip for Apple, people familiar with Apple’s supply chain said. Big AI companies pay generously and are willing to lock in sup- ply and make upfront pay- ments, giving the South Ko- rean chip makers leverage\n\nFY2023 ’24 ’25 ’26\n\n0\n\n2\n\n4\n\n6\n\n$8 Price per gigabyte for DRAM, the main system memory in AI servers and smartphones, quarterly\n\n2020 ’21 ’22 ’23 ’24 ’25\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60% High Performance Computing **\\*** Smartphone\n\nProportion of revenue by market segment, quarterly\n\nSources: TechInsights (price per gigabyte for DRAM); TSMC (revenue) \\*Includes AI-related chips\n\nProjections\n\nPresident Trump speaking at Mar-a-Lago on Sunday.\n\nAL DRAGO/GETTY IMAGES\n\nHouse Minority Leader Hakeem Jeffries (D., N.Y.) is seeking to rein in immigration agents’ tactics.\n\nGRAEME SLOAN/BLOOMBERG NEWS\n\nP2JW033000-5-A00400-16DDF61078D","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"A4 | Monday, February 2, 2026 **P W L C 10 11 12 H T G K R F A M 1 2 3 4 5 6 7 8 9 O I X X** \\* \\* \\* \\* \\* **THE WALL STREET JOURNAL.**\n\nWASHINGTON—President Trump’s lawsuit against the government that he runs pres- ents a mind-bending minefield of conflicts that could end with the president’s appoin- tees approving a federal pay- out to him.\n\nTrump is simultaneously supervising the private attor- neys demanding at least $10 billion in damages over the past disclosure of his tax records, while also overseeing the senior Justice Department attorneys with authority to settle the case and authorize a payment. “This creates the risk of the most collusive lawsuit of all time, because it is ulti- mately the president suing a defendant whom he says has to do whatever he directs,” said Peter Keisler, who served as acting attorney general in the George W. Bush adminis- tration. Trump, two of his sons and his businesses sued the gov- ernment in federal court in Miami last week, arguing that they were harmed when for- mer IRS contractor Charles\n\n_By_ _Richard Rubin_ _,_ _C. Ryan Barber_ _and_ _Annie Linskey_\n\n# **U.S. NEWS**\n\nLittlejohn illegally disclosed their tax information to the New York Times and ProPub- lica in 2019 and 2020. “Defendants have caused Plaintiffs reputational and fi- nancial harm, public embar- rassment, unfairly tarnished their business reputations, portrayed them in a false light, and negatively affected President Trump, and the other Plaintiffs’ public stand- ing,” the lawsuit said. Littlejohn, who is serving a five-year sentence in federal prison, pleaded guilty in 2023 and said he intentionally leaked Trump’s information before the 2020 election. The disclosures showed Trump regularly reported losses and paid just $750 in federal in- come taxes in 2016 and in 2017. Littlejohn also disclosed tax data of thousands of other wealthy Americans. Trump said on Saturday that he recognized the “strange position” created by the lawsuit he filed and said he would likely donate pro- ceeds from a settlement to charities such as the American Cancer Society. “If I pay myself, that some- how will never look good,” he said. “Nobody cares how much if it goes to good charities. So you settle by giving charities a lot of money. And I think we’re going to do something like that.” The Justice Department hasn’t yet responded to Trump’s lawsuit, which was\n\nassigned to Judge Kathleen Williams, an appointee of for- mer President Barack Obama. The Treasury Department and Internal Revenue Service didn’t comment. On one crucial question, there’s already a conflict be- tween the president’s stance and the government’s previ- ously stated position. If Little- john was an IRS employee, the government can be sued. If he was a contractor, Trump can’t sue the government. Trump’s lawsuit notes Lit- tlejohn’s vast access to tax- payer information and day-to- day supervision by IRS employees as evidence that he was effectively a government employee. In a separate case brought by billionaire Ken Griffin in 2022, a federal judge let that argument proceed through early stages of litiga- tion. That case was settled with an apology from the gov- ernment to Griffin, before the judge reached a decision. The DOJ argued then that Littlejohn wasn’t an IRS em- ployee. It reiterated that posi- tion as recently as two months ago—during the Trump administration. “At all times, Mr. Littlejohn was a contractor who wrong- fully disclosed taxpayer return information,” the government wrote in a November 2025 court filing in Maryland. “The United States cannot be held liable for his illegal conduct.” Moreover, Littlejohn’s ille- gal conduct happened during\n\nTrump’s first term. That is, the plaintiff’s complaints about IRS failures and negli- gence pin the blame on offi- cials who were supervised by that very plaintiff. Normally, Justice Depart- ment tax attorneys who focus on this area of the law would handle the case. They would likely raise several challenges to Trump’s claims beyond ar- guing that Littlejohn wasn’t an IRS employee. The government could also argue that Trump filed the case too late, said Francesca Ugolini, former chief of the appellate section in DOJ’s tax division. The statute says suits can be filed within two years of the “date of discovery” of the dis- closure. The first New York Times story stemming from\n\nLittlejohn’s disclosure was published in September 2020, though the documents’ prove- nance wasn’t clear then. Little- john was charged in September 2023. Trump, who broke a tra- dition of voluntary-tax-return disclosure by presidents and candidates, filed the lawsuit exactly two years after receiv- ing a formal IRS notice about the unauthorized release. Another point of contention would be the damages calcula- tion. The $10 billion figure isn’t punitive damages. In- stead, it’s a $1,000-per-disclo- sure level specified in the law, multiplied by an estimate of the number of times that Trump’s tax information was published or broadcast. However, the law specifies $1,000 for each unauthorized disclosure. The government could argue that the only un- authorized disclosures were the ones that Littlejohn made. Ugolini said the DOJ’s model should be its handling of Hunter Biden’s tax-disclo- sure suit against the govern- ment. While his father was president, the son sought damages for comments by IRS agents who were raising con- cerns to Congress about the handling of the government’s criminal investigation into him. “The career attorneys at DOJ were allowed to handle the case in the normal way, raising defenses for the United States,” said Ugolini, now at SouthBank Legal.\n\nHunter Biden pleaded guilty to tax crimes in 2024. He dropped the lawsuit against the government last April, just months after his fa- ther pardoned him and Trump returned to the White House. Another potential route is a settlement. Trump has filed other law- suits seeking billions of dol- lars in damages, including one against the publisher of The Wall Street Journal. His gov- ernmental power can create incentives for defendants to settle. Since his 2024 election victory, Disney’s ABC News, Meta Platforms, Alphabet’s YouTube and Paramount Global—the owner of CBS— have agreed to multimillion- dollar payments to settle liti- gation initiated by Trump. The tax case is distinct be- cause Trump doesn’t just have leverage over the defendants. He has control. Trump has separately filed administrative claims demanding millions of dollars from the Justice De- partment over past criminal investigations into him. Asked in October about re- porting that he was seeking more than $200 million, Trump appeared to acknowl- edge the awkwardness of pur- suing payment from the gov- ernment he leads. Trump said then that he was “damaged very greatly” and would give any money to charity. He said on Saturday that he still needs to work out a settlement in that matter.\n\n# **Trump Straddles Both Sides of IRS Suit**\n\n### President is plaintiff, but also oversees people who could decide on a payout\n\non Democrats in the House, a person familiar with the mat- ter said. That meant the cham- ber couldn’t quickly pass the package on Monday through a fast-track provision on a two- thirds vote, and would instead have to go through the “rule” process that would push a vote until Tuesday at the earliest. Jeffries said the Senate vote was a step in the right direc- tion but argued that House Democrats weren’t a part of any Senate deal. He ticked off a series of changes demanded by Democrats, such as a higher threshold for searches and ar- rests, mandated use of body cameras and a ban on masks.\n\nJeffries echoed calls from many Democrats who say even two weeks of funding for DHS is too much. Some Democrats have called for Immigration and Customs Enforcement to be eliminated altogether, after a month in which immigration agents surged into Minneapo- lis, fatally shooting two U.S. citizens to death and prompt- ing protests. “The administration can’t just talk the talk. They need to walk the walk. That should be- gin today—not in two weeks, today,” Jeffries said on Sunday on ABC. Johnson, in an interview on Fox News, dismissed the calls\n\nfor no masks and a require- ment for judicial warrants for apprehensions and searches, but he did speak in favor of body cameras on ICE officers. In a separate appearance on NBC, Johnson said some of the Democrats’ conditions “are ob- viously reasonable and should happen. But others are going to require a lot more negotiation.” The House Democrats’ deci- sion to play hardball has thrown new uncertainty into resolving the partial shutdown. Just last month, Johnson celebrated passing the final six of the 12 annual individual spending bills and sent them to the Senate, a rare occurrence in recent de-\n\ncades when terms like omnibus became commonplace. But an- ger over the death of Pretti made the path to passage more difficult, prompting the Senate to change the House package and send it back this week. Johnson’s tiny vote margin gives any one or two GOP members leverage if they want changes or concessions. John- son as soon as Monday will swear in Christian Menefee, a Democrat who won a special election to represent Texas’ 18th Congressional District af- ter Rep. Sylvester Turner’s death last year. That will bring the majority to 218-214. And the expected vote comes as some conservatives aren’t happy with the two- week extension of the DHS bill, instead wanting a four- to six- week extension. The conserva- tive House Freedom Caucus in a letter to Trump on Tuesday backed the efforts of ICE and urged the administration to “use all tools necessary.” Other GOP lawmakers have said they would demand that the VOTE Act—a measure that imposes stricter requirements for vot- ing, among other changes—be included in any package. Activists and rank-and-file Democrats have rallied around denying ICE any funding. “I’m not a no. I’m a firm no,” said Ro Khanna (D., Calif.) on NBC. “I just in good con- science cannot vote to give more money to ICE agents as they’re violating our constitu- tional rights,” he said. Trump last week moved to\n\npull back on his administra- tion’s immigration-enforce- ment campaign, including re- placing the leader of the Minneapolis operation. But he hasn’t made clear what kind of changes he would accept on immigration-enforcement prac- tices. The White House, like many congressional Republi- cans, had initially pushed for administration changes or ex- ecutive actions, rather than new law passed by Congress. On Saturday night, Trump backed ICE and gave a warning to protesters. “If they do any- thing bad to our people then they will have to suffer,” he said. Asked about body cam- eras for immigration agents, he said Sunday: “We’ll be talking about that in the near future.” Some Democrats said they became further dug in on op- posing any ICE funding after Trump returned to his more forceful rhetoric against pro- testers, and continued to back Homeland Security Secretary Kristi Noem, who many Demo- crats and some Republicans have said should resign. The partial shutdown that started Saturday morning af- fects the agencies attached to the six bills that haven’t yet passed the House, including DHS and the departments of Defense and Transportation. In departments where funding lapses, federal employees are set to be furloughed or work without pay until funding is restored. Federal employees typically get back pay when a shutdown ends.\n\nthem that we need to imple- ment the president’s play call,” giving the White House time to discuss changes to DHS with Democrats. House Republicans will have to swallow frustration over what they see as being painted into a corner by Democrats, and some have floated their own demands. Asked if he ex- pects Trump to help twist arms to get the necessary sup- port, the GOP leader replied: “I’m sure he will get on the phone with them.” The Senate on Friday passed the bipartisan plan struck between Senate Demo- crats and the White House to fund the government, just days after the killing of Alex Pretti by immigration agents in Min- neapolis the previous weekend. But with the House out and not returning until Monday, a funding lapse was triggered Saturday morning. While the Senate vote drew the support of Minority Leader Chuck Schumer (D., N.Y.) and about half of the Senate Demo- cratic caucus, House Minority Leader Hakeem Jeffries (D., N.Y.) told Johnson in a phone call on Saturday not to count\n\n_ContinuedfromPageOne_\n\n# _Funding Runs Into Pushback_\n\nvideos, called NAND, as well as the memory used to run apps quickly, called DRAM. By the end of this year, the price of DRAM will quadruple from 2023 levels, and NAND will more than triple, estimates TechInsights. Howard estimates that Ap- ple could pay $57 more for the two types of memory that go into the base-model iPhone 18 due this fall compared with the base model iPhone 17 cur- rently on sale. For a device that retails for $799, that would be a big hit to profit margins. Apple’s purchasing power and expertise in designing ad- vanced electronics long made it an unrivaled Goliath among the Asian companies that make most of the iPhone’s parts and assemble the device. Apple spends billions of dollars a year on NAND, for in- stance, people familiar with the figures said, likely making it the single biggest buyer globally. Suppliers flocked to win Apple’s business, hoping to leverage its know-how and prestige to attract other cus- tomers. These days, however, “the companies now pushing the boundaries of human‑scale en- gineering are the ones like Nvidia,” said Ming-chi Kuo, an analyst with TF International Securities.\n\nagainst the iPhone maker. Apple signs long-term con- tracts for memory, but it has used its heft to squeeze suppli- ers. Its contracts have empow- ered it to negotiate prices as often as weekly, and to even refuse to buy any memory from a supplier if Apple didn’t view the price as favorable, people familiar with its mem- ory purchases said. To boost leverage with suppliers, Apple even began stocking more in- ventory of memory. That was atypical for Cook, who nor- mally cuts inventory to the bone to maximize Apple’s cash flow. Apple is fighting not only for current deliveries but also for the attention of engineers at suppliers. Glass scientists who worked on developing the smoothest and lightest smart- phone displays are now also spending time on specialized glass for packaging advanced AI processing chips, industry executives said. Makers of sensors and other gizmos inside the iPhone are winning new business from AI companies such as OpenAI that are developing their own hardware. Still, suppliers said they were far from giving up on business with Apple. Working with Apple is a form of educa- tion, they said, because it re- mains one of the most de-\n\nmanding and disciplined customers in the industry. TSMC, the Taiwanese chip manufacturer, has built suc- cessive generations of its most advanced chips with Apple as its lead customer, relying on the big predictable demand for iPhones. Now that TSMC is do- ing more business with Nvidia and other AI companies, peo- ple with knowledge of the chip supply chain said Apple was exploring whether some lower- end processors could be made by someone other than TSMC. One of Apple’s biggest profit spinners is selling extra memory for far more than the memory chips cost the com- pany. Last fall, Apple discon- tinued the iPhone Pro model with 128 gigabytes of storage. Customers who want that model must now start at 256 gigabytes and pay $100 more—the type of move that could be repeated this year to help Apple offset higher costs, wrote Craig Moffett, an ana- lyst at Moffett Nathanson, in an investor note. However, Apple isn’t ex- pected to raise the price of its next iPhone models over simi- larly equipped iPhone 17s, said Kuo, the analyst. News Corp, owner of The Wall Street Journal, has a commercial agreement to sup- ply news through Apple ser- vices.\n\ntor Manufacturing, Nvidia Chief Executive Jensen Huang said on a podcast. Apple had been TSMC’s biggest customer by a wide margin for years. TSMC is the world’s leading manufacturer of advanced chips for AI servers, smart- phones and other computing devices. Spokesmen for Apple and TSMC declined to comment. The big computers that handle AI tasks don’t look like the smartphones consumers own, but many companies sup- ply components for both. In particular, memory chips are in short supply as companies such as OpenAI, Alphabet’s Google, Meta, Microsoft and others collectively spend hun- dreds of billions of dollars to build AI computing capacity. “The rate of increase in the price of memory is unprece- dented,” said Mike Howard, an analyst for research firm TechInsights. That applies both to the flash memory chips that store photos and\n\n_ContinuedfromPageOne_\n\n# _Apple’s Margins Threatened_\n\nDemand for AI hardware is poised to keep growing rap- idly. Apple’s spending growth is modest in comparison with what is being spent to fill up AI data centers, even though it is breaking records with huge sales of the iPhone 17. Samsung Electronics and\n\nSK Hynix are raising the price of a type of DRAM chip for Apple, people familiar with Apple’s supply chain said. Big AI companies pay generously and are willing to lock in sup- ply and make upfront pay- ments, giving the South Ko- rean chip makers leverage\n\nFY2023 ’24 ’25 ’26\n\n0\n\n2\n\n4\n\n6\n\n$8 Price per gigabyte for DRAM, the main system memory in AI servers and smartphones, quarterly\n\n2020 ’21 ’22 ’23 ’24 ’25\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60% High Performance Computing **\\*** Smartphone\n\nProportion of revenue by market segment, quarterly\n\nSources: TechInsights (price per gigabyte for DRAM); TSMC (revenue) \\*Includes AI-related chips\n\nProjections\n\nPresident Trump speaking at Mar-a-Lago on Sunday.\n\nAL DRAGO/GETTY IMAGES\n\nHouse Minority Leader Hakeem Jeffries (D., N.Y.) is seeking to rein in immigration agents’ tactics.\n\nGRAEME SLOAN/BLOOMBERG NEWS\n\nP2JW033000-5-A00400-16DDF61078D","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/pdf_f270146f8ca7_p4.pdf"]},"response":{"errors":[],"extract_id":"extract_d5dd21ff9234428f00b022dd94c02956","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["A4 | Monday, February 2, 2026 **P W L C 10 11 12 H T G K R F A M 1 2 3 4 5 6 7 8 9 O I X X** \\* \\* \\* \\* \\* **THE WALL STREET JOURNAL.**\n\nWASHINGTON—President Trump’s lawsuit against the government that he runs pres- ents a mind-bending minefield of conflicts that could end with the president’s appoin- tees approving a federal pay- out to him.\n\nTrump is simultaneously supervising the private attor- neys demanding at least $10 billion in damages over the past disclosure of his tax records, while also overseeing the senior Justice Department attorneys with authority to settle the case and authorize a payment. “This creates the risk of the most collusive lawsuit of all time, because it is ulti- mately the president suing a defendant whom he says has to do whatever he directs,” said Peter Keisler, who served as acting attorney general in the George W. Bush adminis- tration. Trump, two of his sons and his businesses sued the gov- ernment in federal court in Miami last week, arguing that they were harmed when for- mer IRS contractor Charles\n\n_By_ _Richard Rubin_ _,_ _C. Ryan Barber_ _and_ _Annie Linskey_\n\n# **U.S. NEWS**\n\nLittlejohn illegally disclosed their tax information to the New York Times and ProPub- lica in 2019 and 2020. “Defendants have caused Plaintiffs reputational and fi- nancial harm, public embar- rassment, unfairly tarnished their business reputations, portrayed them in a false light, and negatively affected President Trump, and the other Plaintiffs’ public stand- ing,” the lawsuit said. Littlejohn, who is serving a five-year sentence in federal prison, pleaded guilty in 2023 and said he intentionally leaked Trump’s information before the 2020 election. The disclosures showed Trump regularly reported losses and paid just $750 in federal in- come taxes in 2016 and in 2017. Littlejohn also disclosed tax data of thousands of other wealthy Americans. Trump said on Saturday that he recognized the “strange position” created by the lawsuit he filed and said he would likely donate pro- ceeds from a settlement to charities such as the American Cancer Society. “If I pay myself, that some- how will never look good,” he said. “Nobody cares how much if it goes to good charities. So you settle by giving charities a lot of money. And I think we’re going to do something like that.” The Justice Department hasn’t yet responded to Trump’s lawsuit, which was\n\nassigned to Judge Kathleen Williams, an appointee of for- mer President Barack Obama. The Treasury Department and Internal Revenue Service didn’t comment. On one crucial question, there’s already a conflict be- tween the president’s stance and the government’s previ- ously stated position. If Little- john was an IRS employee, the government can be sued. If he was a contractor, Trump can’t sue the government. Trump’s lawsuit notes Lit- tlejohn’s vast access to tax- payer information and day-to- day supervision by IRS employees as evidence that he was effectively a government employee. In a separate case brought by billionaire Ken Griffin in 2022, a federal judge let that argument proceed through early stages of litiga- tion. That case was settled with an apology from the gov- ernment to Griffin, before the judge reached a decision. The DOJ argued then that Littlejohn wasn’t an IRS em- ployee. It reiterated that posi- tion as recently as two months ago—during the Trump administration. “At all times, Mr. Littlejohn was a contractor who wrong- fully disclosed taxpayer return information,” the government wrote in a November 2025 court filing in Maryland. “The United States cannot be held liable for his illegal conduct.” Moreover, Littlejohn’s ille- gal conduct happened during\n\nTrump’s first term. That is, the plaintiff’s complaints about IRS failures and negli- gence pin the blame on offi- cials who were supervised by that very plaintiff. Normally, Justice Depart- ment tax attorneys who focus on this area of the law would handle the case. They would likely raise several challenges to Trump’s claims beyond ar- guing that Littlejohn wasn’t an IRS employee. The government could also argue that Trump filed the case too late, said Francesca Ugolini, former chief of the appellate section in DOJ’s tax division. The statute says suits can be filed within two years of the “date of discovery” of the dis- closure. The first New York Times story stemming from\n\nLittlejohn’s disclosure was published in September 2020, though the documents’ prove- nance wasn’t clear then. Little- john was charged in September 2023. Trump, who broke a tra- dition of voluntary-tax-return disclosure by presidents and candidates, filed the lawsuit exactly two years after receiv- ing a formal IRS notice about the unauthorized release. Another point of contention would be the damages calcula- tion. The $10 billion figure isn’t punitive damages. In- stead, it’s a $1,000-per-disclo- sure level specified in the law, multiplied by an estimate of the number of times that Trump’s tax information was published or broadcast. However, the law specifies $1,000 for each unauthorized disclosure. The government could argue that the only un- authorized disclosures were the ones that Littlejohn made. Ugolini said the DOJ’s model should be its handling of Hunter Biden’s tax-disclo- sure suit against the govern- ment. While his father was president, the son sought damages for comments by IRS agents who were raising con- cerns to Congress about the handling of the government’s criminal investigation into him. “The career attorneys at DOJ were allowed to handle the case in the normal way, raising defenses for the United States,” said Ugolini, now at SouthBank Legal.\n\nHunter Biden pleaded guilty to tax crimes in 2024. He dropped the lawsuit against the government last April, just months after his fa- ther pardoned him and Trump returned to the White House. Another potential route is a settlement. Trump has filed other law- suits seeking billions of dol- lars in damages, including one against the publisher of The Wall Street Journal. His gov- ernmental power can create incentives for defendants to settle. Since his 2024 election victory, Disney’s ABC News, Meta Platforms, Alphabet’s YouTube and Paramount Global—the owner of CBS— have agreed to multimillion- dollar payments to settle liti- gation initiated by Trump. The tax case is distinct be- cause Trump doesn’t just have leverage over the defendants. He has control. Trump has separately filed administrative claims demanding millions of dollars from the Justice De- partment over past criminal investigations into him. Asked in October about re- porting that he was seeking more than $200 million, Trump appeared to acknowl- edge the awkwardness of pur- suing payment from the gov- ernment he leads. Trump said then that he was “damaged very greatly” and would give any money to charity. He said on Saturday that he still needs to work out a settlement in that matter.\n\n# **Trump Straddles Both Sides of IRS Suit**\n\n### President is plaintiff, but also oversees people who could decide on a payout\n\non Democrats in the House, a person familiar with the mat- ter said. That meant the cham- ber couldn’t quickly pass the package on Monday through a fast-track provision on a two- thirds vote, and would instead have to go through the “rule” process that would push a vote until Tuesday at the earliest. Jeffries said the Senate vote was a step in the right direc- tion but argued that House Democrats weren’t a part of any Senate deal. He ticked off a series of changes demanded by Democrats, such as a higher threshold for searches and ar- rests, mandated use of body cameras and a ban on masks.\n\nJeffries echoed calls from many Democrats who say even two weeks of funding for DHS is too much. Some Democrats have called for Immigration and Customs Enforcement to be eliminated altogether, after a month in which immigration agents surged into Minneapo- lis, fatally shooting two U.S. citizens to death and prompt- ing protests. “The administration can’t just talk the talk. They need to walk the walk. That should be- gin today—not in two weeks, today,” Jeffries said on Sunday on ABC. Johnson, in an interview on Fox News, dismissed the calls\n\nfor no masks and a require- ment for judicial warrants for apprehensions and searches, but he did speak in favor of body cameras on ICE officers. In a separate appearance on NBC, Johnson said some of the Democrats’ conditions “are ob- viously reasonable and should happen. But others are going to require a lot more negotiation.” The House Democrats’ deci- sion to play hardball has thrown new uncertainty into resolving the partial shutdown. Just last month, Johnson celebrated passing the final six of the 12 annual individual spending bills and sent them to the Senate, a rare occurrence in recent de-\n\ncades when terms like omnibus became commonplace. But an- ger over the death of Pretti made the path to passage more difficult, prompting the Senate to change the House package and send it back this week. Johnson’s tiny vote margin gives any one or two GOP members leverage if they want changes or concessions. John- son as soon as Monday will swear in Christian Menefee, a Democrat who won a special election to represent Texas’ 18th Congressional District af- ter Rep. Sylvester Turner’s death last year. That will bring the majority to 218-214. And the expected vote comes as some conservatives aren’t happy with the two- week extension of the DHS bill, instead wanting a four- to six- week extension. The conserva- tive House Freedom Caucus in a letter to Trump on Tuesday backed the efforts of ICE and urged the administration to “use all tools necessary.” Other GOP lawmakers have said they would demand that the VOTE Act—a measure that imposes stricter requirements for vot- ing, among other changes—be included in any package. Activists and rank-and-file Democrats have rallied around denying ICE any funding. “I’m not a no. I’m a firm no,” said Ro Khanna (D., Calif.) on NBC. “I just in good con- science cannot vote to give more money to ICE agents as they’re violating our constitu- tional rights,” he said. Trump last week moved to\n\npull back on his administra- tion’s immigration-enforce- ment campaign, including re- placing the leader of the Minneapolis operation. But he hasn’t made clear what kind of changes he would accept on immigration-enforcement prac- tices. The White House, like many congressional Republi- cans, had initially pushed for administration changes or ex- ecutive actions, rather than new law passed by Congress. On Saturday night, Trump backed ICE and gave a warning to protesters. “If they do any- thing bad to our people then they will have to suffer,” he said. Asked about body cam- eras for immigration agents, he said Sunday: “We’ll be talking about that in the near future.” Some Democrats said they became further dug in on op- posing any ICE funding after Trump returned to his more forceful rhetoric against pro- testers, and continued to back Homeland Security Secretary Kristi Noem, who many Demo- crats and some Republicans have said should resign. The partial shutdown that started Saturday morning af- fects the agencies attached to the six bills that haven’t yet passed the House, including DHS and the departments of Defense and Transportation. In departments where funding lapses, federal employees are set to be furloughed or work without pay until funding is restored. Federal employees typically get back pay when a shutdown ends.\n\nthem that we need to imple- ment the president’s play call,” giving the White House time to discuss changes to DHS with Democrats. House Republicans will have to swallow frustration over what they see as being painted into a corner by Democrats, and some have floated their own demands. Asked if he ex- pects Trump to help twist arms to get the necessary sup- port, the GOP leader replied: “I’m sure he will get on the phone with them.” The Senate on Friday passed the bipartisan plan struck between Senate Demo- crats and the White House to fund the government, just days after the killing of Alex Pretti by immigration agents in Min- neapolis the previous weekend. But with the House out and not returning until Monday, a funding lapse was triggered Saturday morning. While the Senate vote drew the support of Minority Leader Chuck Schumer (D., N.Y.) and about half of the Senate Demo- cratic caucus, House Minority Leader Hakeem Jeffries (D., N.Y.) told Johnson in a phone call on Saturday not to count\n\n_ContinuedfromPageOne_\n\n# _Funding Runs Into Pushback_\n\nvideos, called NAND, as well as the memory used to run apps quickly, called DRAM. By the end of this year, the price of DRAM will quadruple from 2023 levels, and NAND will more than triple, estimates TechInsights. Howard estimates that Ap- ple could pay $57 more for the two types of memory that go into the base-model iPhone 18 due this fall compared with the base model iPhone 17 cur- rently on sale. For a device that retails for $799, that would be a big hit to profit margins. Apple’s purchasing power and expertise in designing ad- vanced electronics long made it an unrivaled Goliath among the Asian companies that make most of the iPhone’s parts and assemble the device. Apple spends billions of dollars a year on NAND, for in- stance, people familiar with the figures said, likely making it the single biggest buyer globally. Suppliers flocked to win Apple’s business, hoping to leverage its know-how and prestige to attract other cus- tomers. These days, however, “the companies now pushing the boundaries of human‑scale en- gineering are the ones like Nvidia,” said Ming-chi Kuo, an analyst with TF International Securities.\n\nagainst the iPhone maker. Apple signs long-term con- tracts for memory, but it has used its heft to squeeze suppli- ers. Its contracts have empow- ered it to negotiate prices as often as weekly, and to even refuse to buy any memory from a supplier if Apple didn’t view the price as favorable, people familiar with its mem- ory purchases said. To boost leverage with suppliers, Apple even began stocking more in- ventory of memory. That was atypical for Cook, who nor- mally cuts inventory to the bone to maximize Apple’s cash flow. Apple is fighting not only for current deliveries but also for the attention of engineers at suppliers. Glass scientists who worked on developing the smoothest and lightest smart- phone displays are now also spending time on specialized glass for packaging advanced AI processing chips, industry executives said. Makers of sensors and other gizmos inside the iPhone are winning new business from AI companies such as OpenAI that are developing their own hardware. Still, suppliers said they were far from giving up on business with Apple. Working with Apple is a form of educa- tion, they said, because it re- mains one of the most de-\n\nmanding and disciplined customers in the industry. TSMC, the Taiwanese chip manufacturer, has built suc- cessive generations of its most advanced chips with Apple as its lead customer, relying on the big predictable demand for iPhones. Now that TSMC is do- ing more business with Nvidia and other AI companies, peo- ple with knowledge of the chip supply chain said Apple was exploring whether some lower- end processors could be made by someone other than TSMC. One of Apple’s biggest profit spinners is selling extra memory for far more than the memory chips cost the com- pany. Last fall, Apple discon- tinued the iPhone Pro model with 128 gigabytes of storage. Customers who want that model must now start at 256 gigabytes and pay $100 more—the type of move that could be repeated this year to help Apple offset higher costs, wrote Craig Moffett, an ana- lyst at Moffett Nathanson, in an investor note. However, Apple isn’t ex- pected to raise the price of its next iPhone models over simi- larly equipped iPhone 17s, said Kuo, the analyst. News Corp, owner of The Wall Street Journal, has a commercial agreement to sup- ply news through Apple ser- vices.\n\ntor Manufacturing, Nvidia Chief Executive Jensen Huang said on a podcast. Apple had been TSMC’s biggest customer by a wide margin for years. TSMC is the world’s leading manufacturer of advanced chips for AI servers, smart- phones and other computing devices. Spokesmen for Apple and TSMC declined to comment. The big computers that handle AI tasks don’t look like the smartphones consumers own, but many companies sup- ply components for both. In particular, memory chips are in short supply as companies such as OpenAI, Alphabet’s Google, Meta, Microsoft and others collectively spend hun- dreds of billions of dollars to build AI computing capacity. “The rate of increase in the price of memory is unprece- dented,” said Mike Howard, an analyst for research firm TechInsights. That applies both to the flash memory chips that store photos and\n\n_ContinuedfromPageOne_\n\n# _Apple’s Margins Threatened_\n\nDemand for AI hardware is poised to keep growing rap- idly. Apple’s spending growth is modest in comparison with what is being spent to fill up AI data centers, even though it is breaking records with huge sales of the iPhone 17. Samsung Electronics and\n\nSK Hynix are raising the price of a type of DRAM chip for Apple, people familiar with Apple’s supply chain said. Big AI companies pay generously and are willing to lock in sup- ply and make upfront pay- ments, giving the South Ko- rean chip makers leverage\n\nFY2023 ’24 ’25 ’26\n\n0\n\n2\n\n4\n\n6\n\n$8 Price per gigabyte for DRAM, the main system memory in AI servers and smartphones, quarterly\n\n2020 ’21 ’22 ’23 ’24 ’25\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60% High Performance Computing **\\*** Smartphone\n\nProportion of revenue by market segment, quarterly\n\nSources: TechInsights (price per gigabyte for DRAM); TSMC (revenue) \\*Includes AI-related chips\n\nProjections\n\nPresident Trump speaking at Mar-a-Lago on Sunday.\n\nAL DRAGO/GETTY IMAGES\n\nHouse Minority Leader Hakeem Jeffries (D., N.Y.) is seeking to rein in immigration agents’ tactics.\n\nGRAEME SLOAN/BLOOMBERG NEWS\n\nP2JW033000-5-A00400-16DDF61078D"],"full_content":"A4 | Monday, February 2, 2026 **P W L C 10 11 12 H T G K R F A M 1 2 3 4 5 6 7 8 9 O I X X** \\* \\* \\* \\* \\* **THE WALL STREET JOURNAL.**\n\nWASHINGTON—President Trump’s lawsuit against the government that he runs pres- ents a mind-bending minefield of conflicts that could end with the president’s appoin- tees approving a federal pay- out to him.\n\nTrump is simultaneously supervising the private attor- neys demanding at least $10 billion in damages over the past disclosure of his tax records, while also overseeing the senior Justice Department attorneys with authority to settle the case and authorize a payment. “This creates the risk of the most collusive lawsuit of all time, because it is ulti- mately the president suing a defendant whom he says has to do whatever he directs,” said Peter Keisler, who served as acting attorney general in the George W. Bush adminis- tration. Trump, two of his sons and his businesses sued the gov- ernment in federal court in Miami last week, arguing that they were harmed when for- mer IRS contractor Charles\n\n_By_ _Richard Rubin_ _,_ _C. Ryan Barber_ _and_ _Annie Linskey_\n\n# **U.S. NEWS**\n\nLittlejohn illegally disclosed their tax information to the New York Times and ProPub- lica in 2019 and 2020. “Defendants have caused Plaintiffs reputational and fi- nancial harm, public embar- rassment, unfairly tarnished their business reputations, portrayed them in a false light, and negatively affected President Trump, and the other Plaintiffs’ public stand- ing,” the lawsuit said. Littlejohn, who is serving a five-year sentence in federal prison, pleaded guilty in 2023 and said he intentionally leaked Trump’s information before the 2020 election. The disclosures showed Trump regularly reported losses and paid just $750 in federal in- come taxes in 2016 and in 2017. Littlejohn also disclosed tax data of thousands of other wealthy Americans. Trump said on Saturday that he recognized the “strange position” created by the lawsuit he filed and said he would likely donate pro- ceeds from a settlement to charities such as the American Cancer Society. “If I pay myself, that some- how will never look good,” he said. “Nobody cares how much if it goes to good charities. So you settle by giving charities a lot of money. And I think we’re going to do something like that.” The Justice Department hasn’t yet responded to Trump’s lawsuit, which was\n\nassigned to Judge Kathleen Williams, an appointee of for- mer President Barack Obama. The Treasury Department and Internal Revenue Service didn’t comment. On one crucial question, there’s already a conflict be- tween the president’s stance and the government’s previ- ously stated position. If Little- john was an IRS employee, the government can be sued. If he was a contractor, Trump can’t sue the government. Trump’s lawsuit notes Lit- tlejohn’s vast access to tax- payer information and day-to- day supervision by IRS employees as evidence that he was effectively a government employee. In a separate case brought by billionaire Ken Griffin in 2022, a federal judge let that argument proceed through early stages of litiga- tion. That case was settled with an apology from the gov- ernment to Griffin, before the judge reached a decision. The DOJ argued then that Littlejohn wasn’t an IRS em- ployee. It reiterated that posi- tion as recently as two months ago—during the Trump administration. “At all times, Mr. Littlejohn was a contractor who wrong- fully disclosed taxpayer return information,” the government wrote in a November 2025 court filing in Maryland. “The United States cannot be held liable for his illegal conduct.” Moreover, Littlejohn’s ille- gal conduct happened during\n\nTrump’s first term. That is, the plaintiff’s complaints about IRS failures and negli- gence pin the blame on offi- cials who were supervised by that very plaintiff. Normally, Justice Depart- ment tax attorneys who focus on this area of the law would handle the case. They would likely raise several challenges to Trump’s claims beyond ar- guing that Littlejohn wasn’t an IRS employee. The government could also argue that Trump filed the case too late, said Francesca Ugolini, former chief of the appellate section in DOJ’s tax division. The statute says suits can be filed within two years of the “date of discovery” of the dis- closure. The first New York Times story stemming from\n\nLittlejohn’s disclosure was published in September 2020, though the documents’ prove- nance wasn’t clear then. Little- john was charged in September 2023. Trump, who broke a tra- dition of voluntary-tax-return disclosure by presidents and candidates, filed the lawsuit exactly two years after receiv- ing a formal IRS notice about the unauthorized release. Another point of contention would be the damages calcula- tion. The $10 billion figure isn’t punitive damages. In- stead, it’s a $1,000-per-disclo- sure level specified in the law, multiplied by an estimate of the number of times that Trump’s tax information was published or broadcast. However, the law specifies $1,000 for each unauthorized disclosure. The government could argue that the only un- authorized disclosures were the ones that Littlejohn made. Ugolini said the DOJ’s model should be its handling of Hunter Biden’s tax-disclo- sure suit against the govern- ment. While his father was president, the son sought damages for comments by IRS agents who were raising con- cerns to Congress about the handling of the government’s criminal investigation into him. “The career attorneys at DOJ were allowed to handle the case in the normal way, raising defenses for the United States,” said Ugolini, now at SouthBank Legal.\n\nHunter Biden pleaded guilty to tax crimes in 2024. He dropped the lawsuit against the government last April, just months after his fa- ther pardoned him and Trump returned to the White House. Another potential route is a settlement. Trump has filed other law- suits seeking billions of dol- lars in damages, including one against the publisher of The Wall Street Journal. His gov- ernmental power can create incentives for defendants to settle. Since his 2024 election victory, Disney’s ABC News, Meta Platforms, Alphabet’s YouTube and Paramount Global—the owner of CBS— have agreed to multimillion- dollar payments to settle liti- gation initiated by Trump. The tax case is distinct be- cause Trump doesn’t just have leverage over the defendants. He has control. Trump has separately filed administrative claims demanding millions of dollars from the Justice De- partment over past criminal investigations into him. Asked in October about re- porting that he was seeking more than $200 million, Trump appeared to acknowl- edge the awkwardness of pur- suing payment from the gov- ernment he leads. Trump said then that he was “damaged very greatly” and would give any money to charity. He said on Saturday that he still needs to work out a settlement in that matter.\n\n# **Trump Straddles Both Sides of IRS Suit**\n\n### President is plaintiff, but also oversees people who could decide on a payout\n\non Democrats in the House, a person familiar with the mat- ter said. That meant the cham- ber couldn’t quickly pass the package on Monday through a fast-track provision on a two- thirds vote, and would instead have to go through the “rule” process that would push a vote until Tuesday at the earliest. Jeffries said the Senate vote was a step in the right direc- tion but argued that House Democrats weren’t a part of any Senate deal. He ticked off a series of changes demanded by Democrats, such as a higher threshold for searches and ar- rests, mandated use of body cameras and a ban on masks.\n\nJeffries echoed calls from many Democrats who say even two weeks of funding for DHS is too much. Some Democrats have called for Immigration and Customs Enforcement to be eliminated altogether, after a month in which immigration agents surged into Minneapo- lis, fatally shooting two U.S. citizens to death and prompt- ing protests. “The administration can’t just talk the talk. They need to walk the walk. That should be- gin today—not in two weeks, today,” Jeffries said on Sunday on ABC. Johnson, in an interview on Fox News, dismissed the calls\n\nfor no masks and a require- ment for judicial warrants for apprehensions and searches, but he did speak in favor of body cameras on ICE officers. In a separate appearance on NBC, Johnson said some of the Democrats’ conditions “are ob- viously reasonable and should happen. But others are going to require a lot more negotiation.” The House Democrats’ deci- sion to play hardball has thrown new uncertainty into resolving the partial shutdown. Just last month, Johnson celebrated passing the final six of the 12 annual individual spending bills and sent them to the Senate, a rare occurrence in recent de-\n\ncades when terms like omnibus became commonplace. But an- ger over the death of Pretti made the path to passage more difficult, prompting the Senate to change the House package and send it back this week. Johnson’s tiny vote margin gives any one or two GOP members leverage if they want changes or concessions. John- son as soon as Monday will swear in Christian Menefee, a Democrat who won a special election to represent Texas’ 18th Congressional District af- ter Rep. Sylvester Turner’s death last year. That will bring the majority to 218-214. And the expected vote comes as some conservatives aren’t happy with the two- week extension of the DHS bill, instead wanting a four- to six- week extension. The conserva- tive House Freedom Caucus in a letter to Trump on Tuesday backed the efforts of ICE and urged the administration to “use all tools necessary.” Other GOP lawmakers have said they would demand that the VOTE Act—a measure that imposes stricter requirements for vot- ing, among other changes—be included in any package. Activists and rank-and-file Democrats have rallied around denying ICE any funding. “I’m not a no. I’m a firm no,” said Ro Khanna (D., Calif.) on NBC. “I just in good con- science cannot vote to give more money to ICE agents as they’re violating our constitu- tional rights,” he said. Trump last week moved to\n\npull back on his administra- tion’s immigration-enforce- ment campaign, including re- placing the leader of the Minneapolis operation. But he hasn’t made clear what kind of changes he would accept on immigration-enforcement prac- tices. The White House, like many congressional Republi- cans, had initially pushed for administration changes or ex- ecutive actions, rather than new law passed by Congress. On Saturday night, Trump backed ICE and gave a warning to protesters. “If they do any- thing bad to our people then they will have to suffer,” he said. Asked about body cam- eras for immigration agents, he said Sunday: “We’ll be talking about that in the near future.” Some Democrats said they became further dug in on op- posing any ICE funding after Trump returned to his more forceful rhetoric against pro- testers, and continued to back Homeland Security Secretary Kristi Noem, who many Demo- crats and some Republicans have said should resign. The partial shutdown that started Saturday morning af- fects the agencies attached to the six bills that haven’t yet passed the House, including DHS and the departments of Defense and Transportation. In departments where funding lapses, federal employees are set to be furloughed or work without pay until funding is restored. Federal employees typically get back pay when a shutdown ends.\n\nthem that we need to imple- ment the president’s play call,” giving the White House time to discuss changes to DHS with Democrats. House Republicans will have to swallow frustration over what they see as being painted into a corner by Democrats, and some have floated their own demands. Asked if he ex- pects Trump to help twist arms to get the necessary sup- port, the GOP leader replied: “I’m sure he will get on the phone with them.” The Senate on Friday passed the bipartisan plan struck between Senate Demo- crats and the White House to fund the government, just days after the killing of Alex Pretti by immigration agents in Min- neapolis the previous weekend. But with the House out and not returning until Monday, a funding lapse was triggered Saturday morning. While the Senate vote drew the support of Minority Leader Chuck Schumer (D., N.Y.) and about half of the Senate Demo- cratic caucus, House Minority Leader Hakeem Jeffries (D., N.Y.) told Johnson in a phone call on Saturday not to count\n\n_ContinuedfromPageOne_\n\n# _Funding Runs Into Pushback_\n\nvideos, called NAND, as well as the memory used to run apps quickly, called DRAM. By the end of this year, the price of DRAM will quadruple from 2023 levels, and NAND will more than triple, estimates TechInsights. Howard estimates that Ap- ple could pay $57 more for the two types of memory that go into the base-model iPhone 18 due this fall compared with the base model iPhone 17 cur- rently on sale. For a device that retails for $799, that would be a big hit to profit margins. Apple’s purchasing power and expertise in designing ad- vanced electronics long made it an unrivaled Goliath among the Asian companies that make most of the iPhone’s parts and assemble the device. Apple spends billions of dollars a year on NAND, for in- stance, people familiar with the figures said, likely making it the single biggest buyer globally. Suppliers flocked to win Apple’s business, hoping to leverage its know-how and prestige to attract other cus- tomers. These days, however, “the companies now pushing the boundaries of human‑scale en- gineering are the ones like Nvidia,” said Ming-chi Kuo, an analyst with TF International Securities.\n\nagainst the iPhone maker. Apple signs long-term con- tracts for memory, but it has used its heft to squeeze suppli- ers. Its contracts have empow- ered it to negotiate prices as often as weekly, and to even refuse to buy any memory from a supplier if Apple didn’t view the price as favorable, people familiar with its mem- ory purchases said. To boost leverage with suppliers, Apple even began stocking more in- ventory of memory. That was atypical for Cook, who nor- mally cuts inventory to the bone to maximize Apple’s cash flow. Apple is fighting not only for current deliveries but also for the attention of engineers at suppliers. Glass scientists who worked on developing the smoothest and lightest smart- phone displays are now also spending time on specialized glass for packaging advanced AI processing chips, industry executives said. Makers of sensors and other gizmos inside the iPhone are winning new business from AI companies such as OpenAI that are developing their own hardware. Still, suppliers said they were far from giving up on business with Apple. Working with Apple is a form of educa- tion, they said, because it re- mains one of the most de-\n\nmanding and disciplined customers in the industry. TSMC, the Taiwanese chip manufacturer, has built suc- cessive generations of its most advanced chips with Apple as its lead customer, relying on the big predictable demand for iPhones. Now that TSMC is do- ing more business with Nvidia and other AI companies, peo- ple with knowledge of the chip supply chain said Apple was exploring whether some lower- end processors could be made by someone other than TSMC. One of Apple’s biggest profit spinners is selling extra memory for far more than the memory chips cost the com- pany. Last fall, Apple discon- tinued the iPhone Pro model with 128 gigabytes of storage. Customers who want that model must now start at 256 gigabytes and pay $100 more—the type of move that could be repeated this year to help Apple offset higher costs, wrote Craig Moffett, an ana- lyst at Moffett Nathanson, in an investor note. However, Apple isn’t ex- pected to raise the price of its next iPhone models over simi- larly equipped iPhone 17s, said Kuo, the analyst. News Corp, owner of The Wall Street Journal, has a commercial agreement to sup- ply news through Apple ser- vices.\n\ntor Manufacturing, Nvidia Chief Executive Jensen Huang said on a podcast. Apple had been TSMC’s biggest customer by a wide margin for years. TSMC is the world’s leading manufacturer of advanced chips for AI servers, smart- phones and other computing devices. Spokesmen for Apple and TSMC declined to comment. The big computers that handle AI tasks don’t look like the smartphones consumers own, but many companies sup- ply components for both. In particular, memory chips are in short supply as companies such as OpenAI, Alphabet’s Google, Meta, Microsoft and others collectively spend hun- dreds of billions of dollars to build AI computing capacity. “The rate of increase in the price of memory is unprece- dented,” said Mike Howard, an analyst for research firm TechInsights. That applies both to the flash memory chips that store photos and\n\n_ContinuedfromPageOne_\n\n# _Apple’s Margins Threatened_\n\nDemand for AI hardware is poised to keep growing rap- idly. Apple’s spending growth is modest in comparison with what is being spent to fill up AI data centers, even though it is breaking records with huge sales of the iPhone 17. Samsung Electronics and\n\nSK Hynix are raising the price of a type of DRAM chip for Apple, people familiar with Apple’s supply chain said. Big AI companies pay generously and are willing to lock in sup- ply and make upfront pay- ments, giving the South Ko- rean chip makers leverage\n\nFY2023 ’24 ’25 ’26\n\n0\n\n2\n\n4\n\n6\n\n$8 Price per gigabyte for DRAM, the main system memory in AI servers and smartphones, quarterly\n\n2020 ’21 ’22 ’23 ’24 ’25\n\n0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60% High Performance Computing **\\*** Smartphone\n\nProportion of revenue by market segment, quarterly\n\nSources: TechInsights (price per gigabyte for DRAM); TSMC (revenue) \\*Includes AI-related chips\n\nProjections\n\nPresident Trump speaking at Mar-a-Lago on Sunday.\n\nAL DRAGO/GETTY IMAGES\n\nHouse Minority Leader Hakeem Jeffries (D., N.Y.) is seeking to rein in immigration agents’ tactics.\n\nGRAEME SLOAN/BLOOMBERG NEWS\n\nP2JW033000-5-A00400-16DDF61078D","publish_date":null,"title":"Trump Straddles Both Sides of IRS Suit","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/pdf_f270146f8ca7_p4.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/pdf_f270146f8ca7_p4.pdf"},"request":{"example_id":"chart/pdf_f270146f8ca7_p4","product_type":"parse","source_file_path":"pdf_f270146f8ca7_p4.pdf"},"started_at":"2026-11-04T19:45:22.724920"} +{"completed_at":"2026-11-04T19:45:25.582760","latency_in_ms":1267,"output":{"example_id":"chart/professional-services-industry-update_p5","grounded_pages":[],"layout_pages":[],"markdown":"4\n\n**806**\n\n**1,090**\n\n**1,398** **1,528**\n\n**1,352** **1,437 1,466 1,527**\n\n**1,329 1,342 1,286 1,178 1,151**\n\n**999 940**\n\n**1,096** **1,153**\n\n**Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2**\n\n**2020 2021 2022 2023 2024**\n\n# **Professional Services M&A Market Commentary**\n\n• Announced year-over-year Q2 2024 global M&A transaction value increased to $56.8 bn from $31.5 bn in Q2 2023. Over the same period, announced transaction volume experienced a slight increase of 0.2%\n\n• Announced Quarter-over-quarter, the global Q2 2024 transaction volume increased 5.2% compared to Q1 2024. Over the same period, announced transaction value increased from $28.1 bn in Q1 2024 to $56.8 bn in Q2 2024\n\n• The spike in transaction value for Q2 2024 was driven the announcement of several large acquisitions, including International Business Machines Corporation’s $7.7bn acquisition of HashiCorp. and Permira Advisers $7.3bn acquisition of Squarespace.\n\n**RecentNotableU.S.Professional ServicesDeals**\n\n**Professional ServicesAnnouncedDealVolumeandValue** **(1)**\n\nDeal Volume (# of transactions); Deal Value ($ in bn)\n\nJune 2024\n\n**iQor / Mill Point Capital** Mill Point Capital, a private equity firm mainly targeting control investments in lower-middle market companies, has acquired iQor, a provider of business process outsourcing solutions. Transaction terms were undisclosed.\n\nJune 2024\n\n**Motion Recruitment Partners / Kelly Services (NasdaqGS:KELY.A)** Kelly Services, a staffing and workforce solutions firm, has acquired Motion Recruitment Partners for $485 mn. Motion Recruitment Partners operates as a recruitment and talent solutions firm focused on serving IT and technology companies.\n\nJuly 2024\n\n**Amplify Consulting Partners / Sparq Holdings (Harvest Partners and Bain Capital)** Sparq, backed by Harvest Partners and Bain Capital, operates as an outsourced digital engineering services provider. Sparq has acquired Amplify Consulting, a consulting firm offering data engineering and visual analytics, for an undisclosed sum.\n\nJune 2024\n\n**PrincetonOne / Hueman People Solutions (Shore Capital Partners)** Hueman People Solutions a recruitment process outsourcing (“RPO”) and talent acquisition firm, has acquired PrincetonOne for an undisclosed value. PrincetonOne operates as a recruitment process outsourcing services provider. Hueman People is backed by Shore Capital Partners.\n\nJuly 2024\n\n**Tandem HR / Vensure Employer Services (Stone Point Capital)** Vensure Employer Services (Vensure HR) has acquired Tandem HR. Tandem offers outsourced human resource and professional employer organization (“PEO”) solutions. Vensure HR is backed by Stone Point Capital and the acquisition of Tandem HR marks the company’s fourth acquisition over the past two months.\n\n(1) Market statistics sourced from Capital IQ, Merger Market, Pitchbook, Wall Street research, press releases. Notes: M&A analysis based on deals announced during respective periods. Deal value represents aggregate deal value for disclosed transactions for time periods noted above\n\n**_$65.0 $66.9 $69.6 $61.4 $70.0 $32.3 $81.3 $141.3 $66.1 $80.3 $24.9 $27.3 $31.5 $51.2 $49.3 $28.1_**\n\nJuly 2024\n\n**Logic Information Systems / Accenture (NYSE:ACN)** Accenture, a professional services company that offers strategy and consulting, interactive, and technology and operations services, has announced the acquisition of Logic Information Systems, a strategy and consulting firm serving the retail industry. Transaction terms were undisclosed.\n\nJuly 2024\n\n**Jamesbeck Global Partners / ZRG Partners (RFE Investment Partners)** ZRG Partners, backed by RFE Investment Partners, is a talent advisory firm providing executive search, interim solutions, and RPO services. ZRG has acquired Jamesbeck Global Partners for an undisclosed sum. Jamesbeck is an executive recruitment firm.\n\nMay 2024\n\n**AuditBoard / HgCapital** HgCapital, an private equity firm focused on software and business services sectors, has announced the acquisition of AuditBoard for $3.0 bn and 15.00x LTM Revenue. AuditBoard provides a cloud- based audit, risk, and compliance management platform.\n\n**_$56.8_**","pages":[{"line_numbers":[],"markdown":"4\n\n**806**\n\n**1,090**\n\n**1,398** **1,528**\n\n**1,352** **1,437 1,466 1,527**\n\n**1,329 1,342 1,286 1,178 1,151**\n\n**999 940**\n\n**1,096** **1,153**\n\n**Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2**\n\n**2020 2021 2022 2023 2024**\n\n# **Professional Services M&A Market Commentary**\n\n• Announced year-over-year Q2 2024 global M&A transaction value increased to $56.8 bn from $31.5 bn in Q2 2023. Over the same period, announced transaction volume experienced a slight increase of 0.2%\n\n• Announced Quarter-over-quarter, the global Q2 2024 transaction volume increased 5.2% compared to Q1 2024. Over the same period, announced transaction value increased from $28.1 bn in Q1 2024 to $56.8 bn in Q2 2024\n\n• The spike in transaction value for Q2 2024 was driven the announcement of several large acquisitions, including International Business Machines Corporation’s $7.7bn acquisition of HashiCorp. and Permira Advisers $7.3bn acquisition of Squarespace.\n\n**RecentNotableU.S.Professional ServicesDeals**\n\n**Professional ServicesAnnouncedDealVolumeandValue** **(1)**\n\nDeal Volume (# of transactions); Deal Value ($ in bn)\n\nJune 2024\n\n**iQor / Mill Point Capital** Mill Point Capital, a private equity firm mainly targeting control investments in lower-middle market companies, has acquired iQor, a provider of business process outsourcing solutions. Transaction terms were undisclosed.\n\nJune 2024\n\n**Motion Recruitment Partners / Kelly Services (NasdaqGS:KELY.A)** Kelly Services, a staffing and workforce solutions firm, has acquired Motion Recruitment Partners for $485 mn. Motion Recruitment Partners operates as a recruitment and talent solutions firm focused on serving IT and technology companies.\n\nJuly 2024\n\n**Amplify Consulting Partners / Sparq Holdings (Harvest Partners and Bain Capital)** Sparq, backed by Harvest Partners and Bain Capital, operates as an outsourced digital engineering services provider. Sparq has acquired Amplify Consulting, a consulting firm offering data engineering and visual analytics, for an undisclosed sum.\n\nJune 2024\n\n**PrincetonOne / Hueman People Solutions (Shore Capital Partners)** Hueman People Solutions a recruitment process outsourcing (“RPO”) and talent acquisition firm, has acquired PrincetonOne for an undisclosed value. PrincetonOne operates as a recruitment process outsourcing services provider. Hueman People is backed by Shore Capital Partners.\n\nJuly 2024\n\n**Tandem HR / Vensure Employer Services (Stone Point Capital)** Vensure Employer Services (Vensure HR) has acquired Tandem HR. Tandem offers outsourced human resource and professional employer organization (“PEO”) solutions. Vensure HR is backed by Stone Point Capital and the acquisition of Tandem HR marks the company’s fourth acquisition over the past two months.\n\n(1) Market statistics sourced from Capital IQ, Merger Market, Pitchbook, Wall Street research, press releases. Notes: M&A analysis based on deals announced during respective periods. Deal value represents aggregate deal value for disclosed transactions for time periods noted above\n\n**_$65.0 $66.9 $69.6 $61.4 $70.0 $32.3 $81.3 $141.3 $66.1 $80.3 $24.9 $27.3 $31.5 $51.2 $49.3 $28.1_**\n\nJuly 2024\n\n**Logic Information Systems / Accenture (NYSE:ACN)** Accenture, a professional services company that offers strategy and consulting, interactive, and technology and operations services, has announced the acquisition of Logic Information Systems, a strategy and consulting firm serving the retail industry. Transaction terms were undisclosed.\n\nJuly 2024\n\n**Jamesbeck Global Partners / ZRG Partners (RFE Investment Partners)** ZRG Partners, backed by RFE Investment Partners, is a talent advisory firm providing executive search, interim solutions, and RPO services. ZRG has acquired Jamesbeck Global Partners for an undisclosed sum. Jamesbeck is an executive recruitment firm.\n\nMay 2024\n\n**AuditBoard / HgCapital** HgCapital, an private equity firm focused on software and business services sectors, has announced the acquisition of AuditBoard for $3.0 bn and 15.00x LTM Revenue. AuditBoard provides a cloud- based audit, risk, and compliance management platform.\n\n**_$56.8_**","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"4\n\n**806**\n\n**1,090**\n\n**1,398** **1,528**\n\n**1,352** **1,437 1,466 1,527**\n\n**1,329 1,342 1,286 1,178 1,151**\n\n**999 940**\n\n**1,096** **1,153**\n\n**Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2**\n\n**2020 2021 2022 2023 2024**\n\n# **Professional Services M&A Market Commentary**\n\n• Announced year-over-year Q2 2024 global M&A transaction value increased to $56.8 bn from $31.5 bn in Q2 2023. Over the same period, announced transaction volume experienced a slight increase of 0.2%\n\n• Announced Quarter-over-quarter, the global Q2 2024 transaction volume increased 5.2% compared to Q1 2024. Over the same period, announced transaction value increased from $28.1 bn in Q1 2024 to $56.8 bn in Q2 2024\n\n• The spike in transaction value for Q2 2024 was driven the announcement of several large acquisitions, including International Business Machines Corporation’s $7.7bn acquisition of HashiCorp. and Permira Advisers $7.3bn acquisition of Squarespace.\n\n**RecentNotableU.S.Professional ServicesDeals**\n\n**Professional ServicesAnnouncedDealVolumeandValue** **(1)**\n\nDeal Volume (# of transactions); Deal Value ($ in bn)\n\nJune 2024\n\n**iQor / Mill Point Capital** Mill Point Capital, a private equity firm mainly targeting control investments in lower-middle market companies, has acquired iQor, a provider of business process outsourcing solutions. Transaction terms were undisclosed.\n\nJune 2024\n\n**Motion Recruitment Partners / Kelly Services (NasdaqGS:KELY.A)** Kelly Services, a staffing and workforce solutions firm, has acquired Motion Recruitment Partners for $485 mn. Motion Recruitment Partners operates as a recruitment and talent solutions firm focused on serving IT and technology companies.\n\nJuly 2024\n\n**Amplify Consulting Partners / Sparq Holdings (Harvest Partners and Bain Capital)** Sparq, backed by Harvest Partners and Bain Capital, operates as an outsourced digital engineering services provider. Sparq has acquired Amplify Consulting, a consulting firm offering data engineering and visual analytics, for an undisclosed sum.\n\nJune 2024\n\n**PrincetonOne / Hueman People Solutions (Shore Capital Partners)** Hueman People Solutions a recruitment process outsourcing (“RPO”) and talent acquisition firm, has acquired PrincetonOne for an undisclosed value. PrincetonOne operates as a recruitment process outsourcing services provider. Hueman People is backed by Shore Capital Partners.\n\nJuly 2024\n\n**Tandem HR / Vensure Employer Services (Stone Point Capital)** Vensure Employer Services (Vensure HR) has acquired Tandem HR. Tandem offers outsourced human resource and professional employer organization (“PEO”) solutions. Vensure HR is backed by Stone Point Capital and the acquisition of Tandem HR marks the company’s fourth acquisition over the past two months.\n\n(1) Market statistics sourced from Capital IQ, Merger Market, Pitchbook, Wall Street research, press releases. Notes: M&A analysis based on deals announced during respective periods. Deal value represents aggregate deal value for disclosed transactions for time periods noted above\n\n**_$65.0 $66.9 $69.6 $61.4 $70.0 $32.3 $81.3 $141.3 $66.1 $80.3 $24.9 $27.3 $31.5 $51.2 $49.3 $28.1_**\n\nJuly 2024\n\n**Logic Information Systems / Accenture (NYSE:ACN)** Accenture, a professional services company that offers strategy and consulting, interactive, and technology and operations services, has announced the acquisition of Logic Information Systems, a strategy and consulting firm serving the retail industry. Transaction terms were undisclosed.\n\nJuly 2024\n\n**Jamesbeck Global Partners / ZRG Partners (RFE Investment Partners)** ZRG Partners, backed by RFE Investment Partners, is a talent advisory firm providing executive search, interim solutions, and RPO services. ZRG has acquired Jamesbeck Global Partners for an undisclosed sum. Jamesbeck is an executive recruitment firm.\n\nMay 2024\n\n**AuditBoard / HgCapital** HgCapital, an private equity firm focused on software and business services sectors, has announced the acquisition of AuditBoard for $3.0 bn and 15.00x LTM Revenue. AuditBoard provides a cloud- based audit, risk, and compliance management platform.\n\n**_$56.8_**","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/professional-services-industry-update_p5.pdf"]},"response":{"errors":[],"extract_id":"extract_1fadae0cf8fcf98ceaf1db1575574df7","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["4\n\n**806**\n\n**1,090**\n\n**1,398** **1,528**\n\n**1,352** **1,437 1,466 1,527**\n\n**1,329 1,342 1,286 1,178 1,151**\n\n**999 940**\n\n**1,096** **1,153**\n\n**Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2**\n\n**2020 2021 2022 2023 2024**\n\n# **Professional Services M&A Market Commentary**\n\n• Announced year-over-year Q2 2024 global M&A transaction value increased to $56.8 bn from $31.5 bn in Q2 2023. Over the same period, announced transaction volume experienced a slight increase of 0.2%\n\n• Announced Quarter-over-quarter, the global Q2 2024 transaction volume increased 5.2% compared to Q1 2024. Over the same period, announced transaction value increased from $28.1 bn in Q1 2024 to $56.8 bn in Q2 2024\n\n• The spike in transaction value for Q2 2024 was driven the announcement of several large acquisitions, including International Business Machines Corporation’s $7.7bn acquisition of HashiCorp. and Permira Advisers $7.3bn acquisition of Squarespace.\n\n**RecentNotableU.S.Professional ServicesDeals**\n\n**Professional ServicesAnnouncedDealVolumeandValue** **(1)**\n\nDeal Volume (# of transactions); Deal Value ($ in bn)\n\nJune 2024\n\n**iQor / Mill Point Capital** Mill Point Capital, a private equity firm mainly targeting control investments in lower-middle market companies, has acquired iQor, a provider of business process outsourcing solutions. Transaction terms were undisclosed.\n\nJune 2024\n\n**Motion Recruitment Partners / Kelly Services (NasdaqGS:KELY.A)** Kelly Services, a staffing and workforce solutions firm, has acquired Motion Recruitment Partners for $485 mn. Motion Recruitment Partners operates as a recruitment and talent solutions firm focused on serving IT and technology companies.\n\nJuly 2024\n\n**Amplify Consulting Partners / Sparq Holdings (Harvest Partners and Bain Capital)** Sparq, backed by Harvest Partners and Bain Capital, operates as an outsourced digital engineering services provider. Sparq has acquired Amplify Consulting, a consulting firm offering data engineering and visual analytics, for an undisclosed sum.\n\nJune 2024\n\n**PrincetonOne / Hueman People Solutions (Shore Capital Partners)** Hueman People Solutions a recruitment process outsourcing (“RPO”) and talent acquisition firm, has acquired PrincetonOne for an undisclosed value. PrincetonOne operates as a recruitment process outsourcing services provider. Hueman People is backed by Shore Capital Partners.\n\nJuly 2024\n\n**Tandem HR / Vensure Employer Services (Stone Point Capital)** Vensure Employer Services (Vensure HR) has acquired Tandem HR. Tandem offers outsourced human resource and professional employer organization (“PEO”) solutions. Vensure HR is backed by Stone Point Capital and the acquisition of Tandem HR marks the company’s fourth acquisition over the past two months.\n\n(1) Market statistics sourced from Capital IQ, Merger Market, Pitchbook, Wall Street research, press releases. Notes: M&A analysis based on deals announced during respective periods. Deal value represents aggregate deal value for disclosed transactions for time periods noted above\n\n**_$65.0 $66.9 $69.6 $61.4 $70.0 $32.3 $81.3 $141.3 $66.1 $80.3 $24.9 $27.3 $31.5 $51.2 $49.3 $28.1_**\n\nJuly 2024\n\n**Logic Information Systems / Accenture (NYSE:ACN)** Accenture, a professional services company that offers strategy and consulting, interactive, and technology and operations services, has announced the acquisition of Logic Information Systems, a strategy and consulting firm serving the retail industry. Transaction terms were undisclosed.\n\nJuly 2024\n\n**Jamesbeck Global Partners / ZRG Partners (RFE Investment Partners)** ZRG Partners, backed by RFE Investment Partners, is a talent advisory firm providing executive search, interim solutions, and RPO services. ZRG has acquired Jamesbeck Global Partners for an undisclosed sum. Jamesbeck is an executive recruitment firm.\n\nMay 2024\n\n**AuditBoard / HgCapital** HgCapital, an private equity firm focused on software and business services sectors, has announced the acquisition of AuditBoard for $3.0 bn and 15.00x LTM Revenue. AuditBoard provides a cloud- based audit, risk, and compliance management platform.\n\n**_$56.8_**"],"full_content":"4\n\n**806**\n\n**1,090**\n\n**1,398** **1,528**\n\n**1,352** **1,437 1,466 1,527**\n\n**1,329 1,342 1,286 1,178 1,151**\n\n**999 940**\n\n**1,096** **1,153**\n\n**Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2**\n\n**2020 2021 2022 2023 2024**\n\n# **Professional Services M&A Market Commentary**\n\n• Announced year-over-year Q2 2024 global M&A transaction value increased to $56.8 bn from $31.5 bn in Q2 2023. Over the same period, announced transaction volume experienced a slight increase of 0.2%\n\n• Announced Quarter-over-quarter, the global Q2 2024 transaction volume increased 5.2% compared to Q1 2024. Over the same period, announced transaction value increased from $28.1 bn in Q1 2024 to $56.8 bn in Q2 2024\n\n• The spike in transaction value for Q2 2024 was driven the announcement of several large acquisitions, including International Business Machines Corporation’s $7.7bn acquisition of HashiCorp. and Permira Advisers $7.3bn acquisition of Squarespace.\n\n**RecentNotableU.S.Professional ServicesDeals**\n\n**Professional ServicesAnnouncedDealVolumeandValue** **(1)**\n\nDeal Volume (# of transactions); Deal Value ($ in bn)\n\nJune 2024\n\n**iQor / Mill Point Capital** Mill Point Capital, a private equity firm mainly targeting control investments in lower-middle market companies, has acquired iQor, a provider of business process outsourcing solutions. Transaction terms were undisclosed.\n\nJune 2024\n\n**Motion Recruitment Partners / Kelly Services (NasdaqGS:KELY.A)** Kelly Services, a staffing and workforce solutions firm, has acquired Motion Recruitment Partners for $485 mn. Motion Recruitment Partners operates as a recruitment and talent solutions firm focused on serving IT and technology companies.\n\nJuly 2024\n\n**Amplify Consulting Partners / Sparq Holdings (Harvest Partners and Bain Capital)** Sparq, backed by Harvest Partners and Bain Capital, operates as an outsourced digital engineering services provider. Sparq has acquired Amplify Consulting, a consulting firm offering data engineering and visual analytics, for an undisclosed sum.\n\nJune 2024\n\n**PrincetonOne / Hueman People Solutions (Shore Capital Partners)** Hueman People Solutions a recruitment process outsourcing (“RPO”) and talent acquisition firm, has acquired PrincetonOne for an undisclosed value. PrincetonOne operates as a recruitment process outsourcing services provider. Hueman People is backed by Shore Capital Partners.\n\nJuly 2024\n\n**Tandem HR / Vensure Employer Services (Stone Point Capital)** Vensure Employer Services (Vensure HR) has acquired Tandem HR. Tandem offers outsourced human resource and professional employer organization (“PEO”) solutions. Vensure HR is backed by Stone Point Capital and the acquisition of Tandem HR marks the company’s fourth acquisition over the past two months.\n\n(1) Market statistics sourced from Capital IQ, Merger Market, Pitchbook, Wall Street research, press releases. Notes: M&A analysis based on deals announced during respective periods. Deal value represents aggregate deal value for disclosed transactions for time periods noted above\n\n**_$65.0 $66.9 $69.6 $61.4 $70.0 $32.3 $81.3 $141.3 $66.1 $80.3 $24.9 $27.3 $31.5 $51.2 $49.3 $28.1_**\n\nJuly 2024\n\n**Logic Information Systems / Accenture (NYSE:ACN)** Accenture, a professional services company that offers strategy and consulting, interactive, and technology and operations services, has announced the acquisition of Logic Information Systems, a strategy and consulting firm serving the retail industry. Transaction terms were undisclosed.\n\nJuly 2024\n\n**Jamesbeck Global Partners / ZRG Partners (RFE Investment Partners)** ZRG Partners, backed by RFE Investment Partners, is a talent advisory firm providing executive search, interim solutions, and RPO services. ZRG has acquired Jamesbeck Global Partners for an undisclosed sum. Jamesbeck is an executive recruitment firm.\n\nMay 2024\n\n**AuditBoard / HgCapital** HgCapital, an private equity firm focused on software and business services sectors, has announced the acquisition of AuditBoard for $3.0 bn and 15.00x LTM Revenue. AuditBoard provides a cloud- based audit, risk, and compliance management platform.\n\n**_$56.8_**","publish_date":null,"title":"Professional Services M&A Market Commentary","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/professional-services-industry-update_p5.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/professional-services-industry-update_p5.pdf"},"request":{"example_id":"chart/professional-services-industry-update_p5","product_type":"parse","source_file_path":"professional-services-industry-update_p5.pdf"},"started_at":"2026-11-04T19:45:24.315709"} +{"completed_at":"2026-11-04T19:45:26.760842","latency_in_ms":1241,"output":{"example_id":"chart/pub-derivatives-quarterly-qtr1-2018_p5","grounded_pages":[],"layout_pages":[],"markdown":"**Quarterly Report on Bank Trading and Derivatives Activities, First Quarter 2018**\n\n\\- 5 -\n\n**Figure 1. Bank Trading Revenue as a Percentage of Consolidated Holding Company Trading Revenue**\n\nSource: Consolidated Financial Statements for Holding Companies—FR Y-9C (Schedule HI) and call report (Schedule RI)","pages":[{"line_numbers":[],"markdown":"**Quarterly Report on Bank Trading and Derivatives Activities, First Quarter 2018**\n\n\\- 5 -\n\n**Figure 1. Bank Trading Revenue as a Percentage of Consolidated Holding Company Trading Revenue**\n\nSource: Consolidated Financial Statements for Holding Companies—FR Y-9C (Schedule HI) and call report (Schedule RI)","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**Quarterly Report on Bank Trading and Derivatives Activities, First Quarter 2018**\n\n\\- 5 -\n\n**Figure 1. Bank Trading Revenue as a Percentage of Consolidated Holding Company Trading Revenue**\n\nSource: Consolidated Financial Statements for Holding Companies—FR Y-9C (Schedule HI) and call report (Schedule RI)","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/pub-derivatives-quarterly-qtr1-2018_p5.pdf"]},"response":{"errors":[],"extract_id":"extract_642c7e64fa94d3cdb61c3c1455f552bc","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**Quarterly Report on Bank Trading and Derivatives Activities, First Quarter 2018**\n\n\\- 5 -\n\n**Figure 1. Bank Trading Revenue as a Percentage of Consolidated Holding Company Trading Revenue**\n\nSource: Consolidated Financial Statements for Holding Companies—FR Y-9C (Schedule HI) and call report (Schedule RI)"],"full_content":"**Quarterly Report on Bank Trading and Derivatives Activities, First Quarter 2018**\n\n\\- 5 -\n\n**Figure 1. Bank Trading Revenue as a Percentage of Consolidated Holding Company Trading Revenue**\n\nSource: Consolidated Financial Statements for Holding Companies—FR Y-9C (Schedule HI) and call report (Schedule RI)","publish_date":null,"title":"Quarterly Report on Bank Trading and Derivatives Activities, First Quarter 2018 Figure 1. Bank Trading Revenue as a Percentage of","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/pub-derivatives-quarterly-qtr1-2018_p5.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/pub-derivatives-quarterly-qtr1-2018_p5.pdf"},"request":{"example_id":"chart/pub-derivatives-quarterly-qtr1-2018_p5","product_type":"parse","source_file_path":"pub-derivatives-quarterly-qtr1-2018_p5.pdf"},"started_at":"2026-11-04T19:45:25.519327"} +{"completed_at":"2026-11-04T19:45:28.248593","latency_in_ms":1604,"output":{"example_id":"chart/pwc-semiconductor-and-beyond-2026-full-report_p17","grounded_pages":[],"layout_pages":[],"markdown":"**PwC** 17 Semiconductor and beyond 2026\n\n# Faster, bigger, smarter data centers\n\n### It is cliché, but true, that we are living in the world of data and connectivity now. More and more devices are interconnected than ever before, including cars, home appliances, smartphones, and PCs. In addition to the growth of number of devices connected, consumers are demanding higher-quality entertainment such as AR/VR/XR gaming and seamless video streaming. Moreover, the introduction of ‘ChatGPT’ in November 2022 became a motivator for both companies and individuals to actively utilize AI services in various applications they can possibly imagine.\n\n### These applications generate and require astronomical amount of data, and we’ve witnessed just the start of it. With high demand for gaming and video streaming, and most importantly, with the rising demand of AI, global data centers are expected to more than double their power consumption by 2030.\n\n### Data centers are a crucial resource for storing, processing, and managing data. They used to focus on providing services for enterprises, but with rising demand, they have achieved hyperscale status, providing internet as a service. Now with the demand of AI- specific applications, Data centers are once again evolving into AI- data centers, enhancing managers’ ability to provide lossless services to data center users.\n\n17\n\n### **Global data center power consumption** (Unit: GW)\n\n’19 ’20 ’21 ’22 ’23 ’24 ’25F ’26F ’27F ’28F ’29F ’30F\n\nAI-driven demand\n\nConventional demand\n\n0\n\n20\n\n40\n\n60\n\n80\n\n100\n\n120\n\n140\n\n160\n\n180\n\n### **Rising AI data center power consumption signals growing AI compute demand and chipset needs**\n\nSource: IEA, PwC analysis","pages":[{"line_numbers":[],"markdown":"**PwC** 17 Semiconductor and beyond 2026\n\n# Faster, bigger, smarter data centers\n\n### It is cliché, but true, that we are living in the world of data and connectivity now. More and more devices are interconnected than ever before, including cars, home appliances, smartphones, and PCs. In addition to the growth of number of devices connected, consumers are demanding higher-quality entertainment such as AR/VR/XR gaming and seamless video streaming. Moreover, the introduction of ‘ChatGPT’ in November 2022 became a motivator for both companies and individuals to actively utilize AI services in various applications they can possibly imagine.\n\n### These applications generate and require astronomical amount of data, and we’ve witnessed just the start of it. With high demand for gaming and video streaming, and most importantly, with the rising demand of AI, global data centers are expected to more than double their power consumption by 2030.\n\n### Data centers are a crucial resource for storing, processing, and managing data. They used to focus on providing services for enterprises, but with rising demand, they have achieved hyperscale status, providing internet as a service. Now with the demand of AI- specific applications, Data centers are once again evolving into AI- data centers, enhancing managers’ ability to provide lossless services to data center users.\n\n17\n\n### **Global data center power consumption** (Unit: GW)\n\n’19 ’20 ’21 ’22 ’23 ’24 ’25F ’26F ’27F ’28F ’29F ’30F\n\nAI-driven demand\n\nConventional demand\n\n0\n\n20\n\n40\n\n60\n\n80\n\n100\n\n120\n\n140\n\n160\n\n180\n\n### **Rising AI data center power consumption signals growing AI compute demand and chipset needs**\n\nSource: IEA, PwC analysis","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**PwC** 17 Semiconductor and beyond 2026\n\n# Faster, bigger, smarter data centers\n\n### It is cliché, but true, that we are living in the world of data and connectivity now. More and more devices are interconnected than ever before, including cars, home appliances, smartphones, and PCs. In addition to the growth of number of devices connected, consumers are demanding higher-quality entertainment such as AR/VR/XR gaming and seamless video streaming. Moreover, the introduction of ‘ChatGPT’ in November 2022 became a motivator for both companies and individuals to actively utilize AI services in various applications they can possibly imagine.\n\n### These applications generate and require astronomical amount of data, and we’ve witnessed just the start of it. With high demand for gaming and video streaming, and most importantly, with the rising demand of AI, global data centers are expected to more than double their power consumption by 2030.\n\n### Data centers are a crucial resource for storing, processing, and managing data. They used to focus on providing services for enterprises, but with rising demand, they have achieved hyperscale status, providing internet as a service. Now with the demand of AI- specific applications, Data centers are once again evolving into AI- data centers, enhancing managers’ ability to provide lossless services to data center users.\n\n17\n\n### **Global data center power consumption** (Unit: GW)\n\n’19 ’20 ’21 ’22 ’23 ’24 ’25F ’26F ’27F ’28F ’29F ’30F\n\nAI-driven demand\n\nConventional demand\n\n0\n\n20\n\n40\n\n60\n\n80\n\n100\n\n120\n\n140\n\n160\n\n180\n\n### **Rising AI data center power consumption signals growing AI compute demand and chipset needs**\n\nSource: IEA, PwC analysis","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/pwc-semiconductor-and-beyond-2026-full-report_p17.pdf"]},"response":{"errors":[],"extract_id":"extract_0c9b6c6d845568c74cf47c50c8f95777","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**PwC** 17 Semiconductor and beyond 2026\n\n# Faster, bigger, smarter data centers\n\n### It is cliché, but true, that we are living in the world of data and connectivity now. More and more devices are interconnected than ever before, including cars, home appliances, smartphones, and PCs. In addition to the growth of number of devices connected, consumers are demanding higher-quality entertainment such as AR/VR/XR gaming and seamless video streaming. Moreover, the introduction of ‘ChatGPT’ in November 2022 became a motivator for both companies and individuals to actively utilize AI services in various applications they can possibly imagine.\n\n### These applications generate and require astronomical amount of data, and we’ve witnessed just the start of it. With high demand for gaming and video streaming, and most importantly, with the rising demand of AI, global data centers are expected to more than double their power consumption by 2030.\n\n### Data centers are a crucial resource for storing, processing, and managing data. They used to focus on providing services for enterprises, but with rising demand, they have achieved hyperscale status, providing internet as a service. Now with the demand of AI- specific applications, Data centers are once again evolving into AI- data centers, enhancing managers’ ability to provide lossless services to data center users.\n\n17\n\n### **Global data center power consumption** (Unit: GW)\n\n’19 ’20 ’21 ’22 ’23 ’24 ’25F ’26F ’27F ’28F ’29F ’30F\n\nAI-driven demand\n\nConventional demand\n\n0\n\n20\n\n40\n\n60\n\n80\n\n100\n\n120\n\n140\n\n160\n\n180\n\n### **Rising AI data center power consumption signals growing AI compute demand and chipset needs**\n\nSource: IEA, PwC analysis"],"full_content":"**PwC** 17 Semiconductor and beyond 2026\n\n# Faster, bigger, smarter data centers\n\n### It is cliché, but true, that we are living in the world of data and connectivity now. More and more devices are interconnected than ever before, including cars, home appliances, smartphones, and PCs. In addition to the growth of number of devices connected, consumers are demanding higher-quality entertainment such as AR/VR/XR gaming and seamless video streaming. Moreover, the introduction of ‘ChatGPT’ in November 2022 became a motivator for both companies and individuals to actively utilize AI services in various applications they can possibly imagine.\n\n### These applications generate and require astronomical amount of data, and we’ve witnessed just the start of it. With high demand for gaming and video streaming, and most importantly, with the rising demand of AI, global data centers are expected to more than double their power consumption by 2030.\n\n### Data centers are a crucial resource for storing, processing, and managing data. They used to focus on providing services for enterprises, but with rising demand, they have achieved hyperscale status, providing internet as a service. Now with the demand of AI- specific applications, Data centers are once again evolving into AI- data centers, enhancing managers’ ability to provide lossless services to data center users.\n\n17\n\n### **Global data center power consumption** (Unit: GW)\n\n’19 ’20 ’21 ’22 ’23 ’24 ’25F ’26F ’27F ’28F ’29F ’30F\n\nAI-driven demand\n\nConventional demand\n\n0\n\n20\n\n40\n\n60\n\n80\n\n100\n\n120\n\n140\n\n160\n\n180\n\n### **Rising AI data center power consumption signals growing AI compute demand and chipset needs**\n\nSource: IEA, PwC analysis","publish_date":null,"title":"Faster, bigger, smarter data centers","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/pwc-semiconductor-and-beyond-2026-full-report_p17.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/pwc-semiconductor-and-beyond-2026-full-report_p17.pdf"},"request":{"example_id":"chart/pwc-semiconductor-and-beyond-2026-full-report_p17","product_type":"parse","source_file_path":"pwc-semiconductor-and-beyond-2026-full-report_p17.pdf"},"started_at":"2026-11-04T19:45:26.644061"} +{"completed_at":"2026-11-04T19:45:31.168541","latency_in_ms":3472,"output":{"example_id":"chart/pwc-semiconductor-and-beyond-2026-full-report_p18","grounded_pages":[],"layout_pages":[],"markdown":"**PwC** 18 Semiconductor and beyond 2026\n\n# The future of intelligent infrastructure\n\n### As AI applications push the required data volume for processing, and as the scale of data centers grows in response, operational expenses for cooling and electricity have reached astronomical levels. Companies are now seeking ways to operate more cost-effectively.\n\n### The first way to do so while boosting AI performance is to utilize data-center-specific chips. These chips are crucial for achieving high performance because they are built to handle the intensive computational demands of data processing more efficiently than general- purpose processors. To get the desired level of performance, companies are turning to these specialized chips to meet their needs.\n\n### However, even if designed specifically for data centers, standard chips are designed for multiple customers and thus contain functions that a given customer might not use. Therefore, big techs like cloud service providers are developing their own AI accelerators specifically designed for their own data center applications.\n\n### By developing AI chips tailored to specific workload, companies can reduce cost and power while achieving higher performance. The demand for AI accelerators is expected to rise as the need for cost reduction grows with increased data-processing demand.\n\n### Therefore, the revenue share of AI accelerators among chips used in data centers can grow at a very fast pace, reaching around 50% of total data center chips.\n\n### In addition to AI accelerators, other data- center-specific chips such as data processing units (DPUs) and advanced memory chips such as HBM can also rise. HBM reduces a data processing bottleneck, supporting high- performance GPUs, and data processing units (DPUs) offload CPU networking workloads by handling data transmission. Sales of these data- center-specific chips may continue to grow as they become essential.\n\n18\n\n### **AI accelerators in the data center** (Unit: $ Billion)\n\n50\n\n100\n\n150\n\n200\n\n250\n\n300\n\n40\n\n0\n\n10\n\n20\n\n30\n\n0\n\n50\n\n60\n\n70\n\n80\n\n90\n\n100\n\n35%\n\n’24 ’25F ’26F ’27F ’28F ’29F\n\n52%\n\n’30F\n\n(%)\n\nData center semiconductors Portion of AI accelerator\n\nSource: PwC analysis","pages":[{"line_numbers":[],"markdown":"**PwC** 18 Semiconductor and beyond 2026\n\n# The future of intelligent infrastructure\n\n### As AI applications push the required data volume for processing, and as the scale of data centers grows in response, operational expenses for cooling and electricity have reached astronomical levels. Companies are now seeking ways to operate more cost-effectively.\n\n### The first way to do so while boosting AI performance is to utilize data-center-specific chips. These chips are crucial for achieving high performance because they are built to handle the intensive computational demands of data processing more efficiently than general- purpose processors. To get the desired level of performance, companies are turning to these specialized chips to meet their needs.\n\n### However, even if designed specifically for data centers, standard chips are designed for multiple customers and thus contain functions that a given customer might not use. Therefore, big techs like cloud service providers are developing their own AI accelerators specifically designed for their own data center applications.\n\n### By developing AI chips tailored to specific workload, companies can reduce cost and power while achieving higher performance. The demand for AI accelerators is expected to rise as the need for cost reduction grows with increased data-processing demand.\n\n### Therefore, the revenue share of AI accelerators among chips used in data centers can grow at a very fast pace, reaching around 50% of total data center chips.\n\n### In addition to AI accelerators, other data- center-specific chips such as data processing units (DPUs) and advanced memory chips such as HBM can also rise. HBM reduces a data processing bottleneck, supporting high- performance GPUs, and data processing units (DPUs) offload CPU networking workloads by handling data transmission. Sales of these data- center-specific chips may continue to grow as they become essential.\n\n18\n\n### **AI accelerators in the data center** (Unit: $ Billion)\n\n50\n\n100\n\n150\n\n200\n\n250\n\n300\n\n40\n\n0\n\n10\n\n20\n\n30\n\n0\n\n50\n\n60\n\n70\n\n80\n\n90\n\n100\n\n35%\n\n’24 ’25F ’26F ’27F ’28F ’29F\n\n52%\n\n’30F\n\n(%)\n\nData center semiconductors Portion of AI accelerator\n\nSource: PwC analysis","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**PwC** 18 Semiconductor and beyond 2026\n\n# The future of intelligent infrastructure\n\n### As AI applications push the required data volume for processing, and as the scale of data centers grows in response, operational expenses for cooling and electricity have reached astronomical levels. Companies are now seeking ways to operate more cost-effectively.\n\n### The first way to do so while boosting AI performance is to utilize data-center-specific chips. These chips are crucial for achieving high performance because they are built to handle the intensive computational demands of data processing more efficiently than general- purpose processors. To get the desired level of performance, companies are turning to these specialized chips to meet their needs.\n\n### However, even if designed specifically for data centers, standard chips are designed for multiple customers and thus contain functions that a given customer might not use. Therefore, big techs like cloud service providers are developing their own AI accelerators specifically designed for their own data center applications.\n\n### By developing AI chips tailored to specific workload, companies can reduce cost and power while achieving higher performance. The demand for AI accelerators is expected to rise as the need for cost reduction grows with increased data-processing demand.\n\n### Therefore, the revenue share of AI accelerators among chips used in data centers can grow at a very fast pace, reaching around 50% of total data center chips.\n\n### In addition to AI accelerators, other data- center-specific chips such as data processing units (DPUs) and advanced memory chips such as HBM can also rise. HBM reduces a data processing bottleneck, supporting high- performance GPUs, and data processing units (DPUs) offload CPU networking workloads by handling data transmission. Sales of these data- center-specific chips may continue to grow as they become essential.\n\n18\n\n### **AI accelerators in the data center** (Unit: $ Billion)\n\n50\n\n100\n\n150\n\n200\n\n250\n\n300\n\n40\n\n0\n\n10\n\n20\n\n30\n\n0\n\n50\n\n60\n\n70\n\n80\n\n90\n\n100\n\n35%\n\n’24 ’25F ’26F ’27F ’28F ’29F\n\n52%\n\n’30F\n\n(%)\n\nData center semiconductors Portion of AI accelerator\n\nSource: PwC analysis","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/pwc-semiconductor-and-beyond-2026-full-report_p18.pdf"]},"response":{"errors":[],"extract_id":"extract_99c90c82e7f8f3c62afdaf94a7c14d47","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**PwC** 18 Semiconductor and beyond 2026\n\n# The future of intelligent infrastructure\n\n### As AI applications push the required data volume for processing, and as the scale of data centers grows in response, operational expenses for cooling and electricity have reached astronomical levels. Companies are now seeking ways to operate more cost-effectively.\n\n### The first way to do so while boosting AI performance is to utilize data-center-specific chips. These chips are crucial for achieving high performance because they are built to handle the intensive computational demands of data processing more efficiently than general- purpose processors. To get the desired level of performance, companies are turning to these specialized chips to meet their needs.\n\n### However, even if designed specifically for data centers, standard chips are designed for multiple customers and thus contain functions that a given customer might not use. Therefore, big techs like cloud service providers are developing their own AI accelerators specifically designed for their own data center applications.\n\n### By developing AI chips tailored to specific workload, companies can reduce cost and power while achieving higher performance. The demand for AI accelerators is expected to rise as the need for cost reduction grows with increased data-processing demand.\n\n### Therefore, the revenue share of AI accelerators among chips used in data centers can grow at a very fast pace, reaching around 50% of total data center chips.\n\n### In addition to AI accelerators, other data- center-specific chips such as data processing units (DPUs) and advanced memory chips such as HBM can also rise. HBM reduces a data processing bottleneck, supporting high- performance GPUs, and data processing units (DPUs) offload CPU networking workloads by handling data transmission. Sales of these data- center-specific chips may continue to grow as they become essential.\n\n18\n\n### **AI accelerators in the data center** (Unit: $ Billion)\n\n50\n\n100\n\n150\n\n200\n\n250\n\n300\n\n40\n\n0\n\n10\n\n20\n\n30\n\n0\n\n50\n\n60\n\n70\n\n80\n\n90\n\n100\n\n35%\n\n’24 ’25F ’26F ’27F ’28F ’29F\n\n52%\n\n’30F\n\n(%)\n\nData center semiconductors Portion of AI accelerator\n\nSource: PwC analysis"],"full_content":"**PwC** 18 Semiconductor and beyond 2026\n\n# The future of intelligent infrastructure\n\n### As AI applications push the required data volume for processing, and as the scale of data centers grows in response, operational expenses for cooling and electricity have reached astronomical levels. Companies are now seeking ways to operate more cost-effectively.\n\n### The first way to do so while boosting AI performance is to utilize data-center-specific chips. These chips are crucial for achieving high performance because they are built to handle the intensive computational demands of data processing more efficiently than general- purpose processors. To get the desired level of performance, companies are turning to these specialized chips to meet their needs.\n\n### However, even if designed specifically for data centers, standard chips are designed for multiple customers and thus contain functions that a given customer might not use. Therefore, big techs like cloud service providers are developing their own AI accelerators specifically designed for their own data center applications.\n\n### By developing AI chips tailored to specific workload, companies can reduce cost and power while achieving higher performance. The demand for AI accelerators is expected to rise as the need for cost reduction grows with increased data-processing demand.\n\n### Therefore, the revenue share of AI accelerators among chips used in data centers can grow at a very fast pace, reaching around 50% of total data center chips.\n\n### In addition to AI accelerators, other data- center-specific chips such as data processing units (DPUs) and advanced memory chips such as HBM can also rise. HBM reduces a data processing bottleneck, supporting high- performance GPUs, and data processing units (DPUs) offload CPU networking workloads by handling data transmission. Sales of these data- center-specific chips may continue to grow as they become essential.\n\n18\n\n### **AI accelerators in the data center** (Unit: $ Billion)\n\n50\n\n100\n\n150\n\n200\n\n250\n\n300\n\n40\n\n0\n\n10\n\n20\n\n30\n\n0\n\n50\n\n60\n\n70\n\n80\n\n90\n\n100\n\n35%\n\n’24 ’25F ’26F ’27F ’28F ’29F\n\n52%\n\n’30F\n\n(%)\n\nData center semiconductors Portion of AI accelerator\n\nSource: PwC analysis","publish_date":null,"title":"The future of intelligent infrastructure","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/pwc-semiconductor-and-beyond-2026-full-report_p18.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/pwc-semiconductor-and-beyond-2026-full-report_p18.pdf"},"request":{"example_id":"chart/pwc-semiconductor-and-beyond-2026-full-report_p18","product_type":"parse","source_file_path":"pwc-semiconductor-and-beyond-2026-full-report_p18.pdf"},"started_at":"2026-11-04T19:45:27.696137"} +{"completed_at":"2026-11-04T19:45:32.016475","latency_in_ms":2740,"output":{"example_id":"chart/r_qt1212e_p11","grounded_pages":[],"layout_pages":[],"markdown":"BIS Quarterly Review, December 2012 33\n\nto other bonds, the possibility of total loss is part of the arrangement from inception, and is compensated ex ante by a higher coupon.\n\nDespite experts’ high expectations, the catastrophe bond market has remained relatively small. Bond issuance has never exceeded $7 billion per year, limiting the outstanding capital at risk to $14 billion (Graph 8). Very few catastrophe bonds have been triggered to date. The 2005 Gulf Coast hurricanes activated payouts from only one of nine catastrophe bonds outstanding at the time (IAIS (2009)). Likewise, the 2011 Japan earthquake and tsunami triggered one known catastrophe bond, resulting in a payout of less than $300 million. Payouts to reinsurers from these bonds are small when compared to the sum of insured losses ($116 billion in 2005 and $110 billion in 2011).\n\nThe global financial crisis has also dealt a blow to this market. The year 2008 saw a rapid decline in catastrophe bond issuance, reflecting generalised funding pressure and investor concern over the vulnerability of insurance entities. The crisis also demonstrated that securitisation structures introduce additional risk through linkages between financial entities. A case in point was the Lehman Brothers bankruptcy in September 2008. Four catastrophe bonds were impaired – not due to natural catastrophes, but because they included a total return swap with Lehman Brothers acting as a counterparty. Following Lehman’s failure, these securitisation arrangements were no longer fully funded, and their market value plunged. Investors thus learned that catastrophe bonds are not immune to “unnatural” disasters such as major institutional failures. ¹³\n\nA further set of financial linkages arises with other financial institutions through cross-holdings of debt and equity. Insurance companies hold large positions in fixed income instruments, including bank bonds. At the same time, other financial entities own bonds and stocks in insurance companies. For instance, the two largest reinsurance companies stated in their latest (2011) annual reports that Warren\n\n13 Following this episode, sponsors of catastrophe bonds employed other types of collateral arrangements in lieu of total return swaps. There has recently been a shift towards the use of government bonds as collateral.\n\nCatastrophe bond issuance by type of risk ¹ Graph 8\n\nUSD bn\n\n1 Data before 2003 are not broken down by type. 2 Includes mortality, peril, life and worldwide risks. 3 Values are year-to-date, thus the value for 2012 is not final.\n\nSources: Artemis; Guy Carpenter.\n\n0\n\n3\n\n6\n\n9\n\n12\n\n96 97 98 99 00 01 02 03 04 05 06 07 08 09 10 11 12\n\nEarthquake Weather\n\nWeather/earthquake Weather/mortality\n\nLife and other ² Risk capital outstanding at year-end ³","pages":[{"line_numbers":[],"markdown":"BIS Quarterly Review, December 2012 33\n\nto other bonds, the possibility of total loss is part of the arrangement from inception, and is compensated ex ante by a higher coupon.\n\nDespite experts’ high expectations, the catastrophe bond market has remained relatively small. Bond issuance has never exceeded $7 billion per year, limiting the outstanding capital at risk to $14 billion (Graph 8). Very few catastrophe bonds have been triggered to date. The 2005 Gulf Coast hurricanes activated payouts from only one of nine catastrophe bonds outstanding at the time (IAIS (2009)). Likewise, the 2011 Japan earthquake and tsunami triggered one known catastrophe bond, resulting in a payout of less than $300 million. Payouts to reinsurers from these bonds are small when compared to the sum of insured losses ($116 billion in 2005 and $110 billion in 2011).\n\nThe global financial crisis has also dealt a blow to this market. The year 2008 saw a rapid decline in catastrophe bond issuance, reflecting generalised funding pressure and investor concern over the vulnerability of insurance entities. The crisis also demonstrated that securitisation structures introduce additional risk through linkages between financial entities. A case in point was the Lehman Brothers bankruptcy in September 2008. Four catastrophe bonds were impaired – not due to natural catastrophes, but because they included a total return swap with Lehman Brothers acting as a counterparty. Following Lehman’s failure, these securitisation arrangements were no longer fully funded, and their market value plunged. Investors thus learned that catastrophe bonds are not immune to “unnatural” disasters such as major institutional failures. ¹³\n\nA further set of financial linkages arises with other financial institutions through cross-holdings of debt and equity. Insurance companies hold large positions in fixed income instruments, including bank bonds. At the same time, other financial entities own bonds and stocks in insurance companies. For instance, the two largest reinsurance companies stated in their latest (2011) annual reports that Warren\n\n13 Following this episode, sponsors of catastrophe bonds employed other types of collateral arrangements in lieu of total return swaps. There has recently been a shift towards the use of government bonds as collateral.\n\nCatastrophe bond issuance by type of risk ¹ Graph 8\n\nUSD bn\n\n1 Data before 2003 are not broken down by type. 2 Includes mortality, peril, life and worldwide risks. 3 Values are year-to-date, thus the value for 2012 is not final.\n\nSources: Artemis; Guy Carpenter.\n\n0\n\n3\n\n6\n\n9\n\n12\n\n96 97 98 99 00 01 02 03 04 05 06 07 08 09 10 11 12\n\nEarthquake Weather\n\nWeather/earthquake Weather/mortality\n\nLife and other ² Risk capital outstanding at year-end ³","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"BIS Quarterly Review, December 2012 33\n\nto other bonds, the possibility of total loss is part of the arrangement from inception, and is compensated ex ante by a higher coupon.\n\nDespite experts’ high expectations, the catastrophe bond market has remained relatively small. Bond issuance has never exceeded $7 billion per year, limiting the outstanding capital at risk to $14 billion (Graph 8). Very few catastrophe bonds have been triggered to date. The 2005 Gulf Coast hurricanes activated payouts from only one of nine catastrophe bonds outstanding at the time (IAIS (2009)). Likewise, the 2011 Japan earthquake and tsunami triggered one known catastrophe bond, resulting in a payout of less than $300 million. Payouts to reinsurers from these bonds are small when compared to the sum of insured losses ($116 billion in 2005 and $110 billion in 2011).\n\nThe global financial crisis has also dealt a blow to this market. The year 2008 saw a rapid decline in catastrophe bond issuance, reflecting generalised funding pressure and investor concern over the vulnerability of insurance entities. The crisis also demonstrated that securitisation structures introduce additional risk through linkages between financial entities. A case in point was the Lehman Brothers bankruptcy in September 2008. Four catastrophe bonds were impaired – not due to natural catastrophes, but because they included a total return swap with Lehman Brothers acting as a counterparty. Following Lehman’s failure, these securitisation arrangements were no longer fully funded, and their market value plunged. Investors thus learned that catastrophe bonds are not immune to “unnatural” disasters such as major institutional failures. ¹³\n\nA further set of financial linkages arises with other financial institutions through cross-holdings of debt and equity. Insurance companies hold large positions in fixed income instruments, including bank bonds. At the same time, other financial entities own bonds and stocks in insurance companies. For instance, the two largest reinsurance companies stated in their latest (2011) annual reports that Warren\n\n13 Following this episode, sponsors of catastrophe bonds employed other types of collateral arrangements in lieu of total return swaps. There has recently been a shift towards the use of government bonds as collateral.\n\nCatastrophe bond issuance by type of risk ¹ Graph 8\n\nUSD bn\n\n1 Data before 2003 are not broken down by type. 2 Includes mortality, peril, life and worldwide risks. 3 Values are year-to-date, thus the value for 2012 is not final.\n\nSources: Artemis; Guy Carpenter.\n\n0\n\n3\n\n6\n\n9\n\n12\n\n96 97 98 99 00 01 02 03 04 05 06 07 08 09 10 11 12\n\nEarthquake Weather\n\nWeather/earthquake Weather/mortality\n\nLife and other ² Risk capital outstanding at year-end ³","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/r_qt1212e_p11.pdf"]},"response":{"errors":[],"extract_id":"extract_19e4fad34c9e976e0458697e342e576d","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["BIS Quarterly Review, December 2012 33\n\nto other bonds, the possibility of total loss is part of the arrangement from inception, and is compensated ex ante by a higher coupon.\n\nDespite experts’ high expectations, the catastrophe bond market has remained relatively small. Bond issuance has never exceeded $7 billion per year, limiting the outstanding capital at risk to $14 billion (Graph 8). Very few catastrophe bonds have been triggered to date. The 2005 Gulf Coast hurricanes activated payouts from only one of nine catastrophe bonds outstanding at the time (IAIS (2009)). Likewise, the 2011 Japan earthquake and tsunami triggered one known catastrophe bond, resulting in a payout of less than $300 million. Payouts to reinsurers from these bonds are small when compared to the sum of insured losses ($116 billion in 2005 and $110 billion in 2011).\n\nThe global financial crisis has also dealt a blow to this market. The year 2008 saw a rapid decline in catastrophe bond issuance, reflecting generalised funding pressure and investor concern over the vulnerability of insurance entities. The crisis also demonstrated that securitisation structures introduce additional risk through linkages between financial entities. A case in point was the Lehman Brothers bankruptcy in September 2008. Four catastrophe bonds were impaired – not due to natural catastrophes, but because they included a total return swap with Lehman Brothers acting as a counterparty. Following Lehman’s failure, these securitisation arrangements were no longer fully funded, and their market value plunged. Investors thus learned that catastrophe bonds are not immune to “unnatural” disasters such as major institutional failures. ¹³\n\nA further set of financial linkages arises with other financial institutions through cross-holdings of debt and equity. Insurance companies hold large positions in fixed income instruments, including bank bonds. At the same time, other financial entities own bonds and stocks in insurance companies. For instance, the two largest reinsurance companies stated in their latest (2011) annual reports that Warren\n\n13 Following this episode, sponsors of catastrophe bonds employed other types of collateral arrangements in lieu of total return swaps. There has recently been a shift towards the use of government bonds as collateral.\n\nCatastrophe bond issuance by type of risk ¹ Graph 8\n\nUSD bn\n\n1 Data before 2003 are not broken down by type. 2 Includes mortality, peril, life and worldwide risks. 3 Values are year-to-date, thus the value for 2012 is not final.\n\nSources: Artemis; Guy Carpenter.\n\n0\n\n3\n\n6\n\n9\n\n12\n\n96 97 98 99 00 01 02 03 04 05 06 07 08 09 10 11 12\n\nEarthquake Weather\n\nWeather/earthquake Weather/mortality\n\nLife and other ² Risk capital outstanding at year-end ³"],"full_content":"BIS Quarterly Review, December 2012 33\n\nto other bonds, the possibility of total loss is part of the arrangement from inception, and is compensated ex ante by a higher coupon.\n\nDespite experts’ high expectations, the catastrophe bond market has remained relatively small. Bond issuance has never exceeded $7 billion per year, limiting the outstanding capital at risk to $14 billion (Graph 8). Very few catastrophe bonds have been triggered to date. The 2005 Gulf Coast hurricanes activated payouts from only one of nine catastrophe bonds outstanding at the time (IAIS (2009)). Likewise, the 2011 Japan earthquake and tsunami triggered one known catastrophe bond, resulting in a payout of less than $300 million. Payouts to reinsurers from these bonds are small when compared to the sum of insured losses ($116 billion in 2005 and $110 billion in 2011).\n\nThe global financial crisis has also dealt a blow to this market. The year 2008 saw a rapid decline in catastrophe bond issuance, reflecting generalised funding pressure and investor concern over the vulnerability of insurance entities. The crisis also demonstrated that securitisation structures introduce additional risk through linkages between financial entities. A case in point was the Lehman Brothers bankruptcy in September 2008. Four catastrophe bonds were impaired – not due to natural catastrophes, but because they included a total return swap with Lehman Brothers acting as a counterparty. Following Lehman’s failure, these securitisation arrangements were no longer fully funded, and their market value plunged. Investors thus learned that catastrophe bonds are not immune to “unnatural” disasters such as major institutional failures. ¹³\n\nA further set of financial linkages arises with other financial institutions through cross-holdings of debt and equity. Insurance companies hold large positions in fixed income instruments, including bank bonds. At the same time, other financial entities own bonds and stocks in insurance companies. For instance, the two largest reinsurance companies stated in their latest (2011) annual reports that Warren\n\n13 Following this episode, sponsors of catastrophe bonds employed other types of collateral arrangements in lieu of total return swaps. There has recently been a shift towards the use of government bonds as collateral.\n\nCatastrophe bond issuance by type of risk ¹ Graph 8\n\nUSD bn\n\n1 Data before 2003 are not broken down by type. 2 Includes mortality, peril, life and worldwide risks. 3 Values are year-to-date, thus the value for 2012 is not final.\n\nSources: Artemis; Guy Carpenter.\n\n0\n\n3\n\n6\n\n9\n\n12\n\n96 97 98 99 00 01 02 03 04 05 06 07 08 09 10 11 12\n\nEarthquake Weather\n\nWeather/earthquake Weather/mortality\n\nLife and other ² Risk capital outstanding at year-end ³","publish_date":null,"title":"Catastrophe bond issuance by type of risk to other bonds, the possibility of total loss is part of the arrangement","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/r_qt1212e_p11.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/r_qt1212e_p11.pdf"},"request":{"example_id":"chart/r_qt1212e_p11","product_type":"parse","source_file_path":"r_qt1212e_p11.pdf"},"started_at":"2026-11-04T19:45:29.275861"} +{"completed_at":"2026-11-04T19:45:31.387060","latency_in_ms":1320,"output":{"example_id":"chart/r_qt1212e_p9","grounded_pages":[],"layout_pages":[],"markdown":"BIS Quarterly Review, December 2012 31\n\nIn spite of the reinsurance market’s size and concentration, failures of reinsurance companies have remained limited in scope. The largest failures to date, comprising two bankruptcies in 2003, led to an essentially inconsequential reduction in available reinsurance capacity of 0.4% (Graph 6, left-hand panel). That said, any failure of a reinsurer leads to a loss of reinsurance recoverables by primary insurers, and could cause broader market tensions in the event of a disorderly liquidation of large portfolios.\n\nIn this respect, the degree of connectedness within the global insurance market plays an important role. Based on their business model, reinsurers enter into contracts with a large number of primary insurance companies, giving rise to numerous vertical links (Graph 3). In addition, risk transfer between reinsurers leads to horizontal linkages. ¹⁰ We estimate that 12% of natural catastrophe-related risk accepted by reinsurers is transferred within the reinsurance industry, which implies that the industry as a whole retains most of the risks it contracts. In 2011, reinsurers paid 3% of earned premiums to cede catastrophe risk to entities _outside_ the insurance sector. Judging by premium volume, the global insurance market transfers a similarly small share of accepted risk to other financial institutions and the wider financial markets.\n\n### Linkages with financial markets\n\nArrangements designed to transfer risk out of the insurance sector create linkages with other financial market participants. Retrocession to other financial institutions uses contractual arrangements similar to those between reinsurers, and commits\n\n10 For example, a reinsurer might exchange some of its exposure to earthquake risk in Japan for US flood risk with another reinsurer.\n\nReinsurance market developments Graph 6\n\nReinsurance market concentration\n\nPer cent\n\nFailures of reinsurance companies\n\nPer cent of premiums Number of companies\n\n1 Market share of the 10 largest reinsurance companies, measured as a share of gross premiums written by reinsurance companies worldwide in the non-life (property and casualty) business. ² In relation to total market size as measured by gross premiums written (premiums ceded by insurers to reinsurance companies). ³ Number of failures of reinsurance companies worldwide, per calendar year.\n\nSources: IAIS, based on industry data; authors’ calculations.\n\n30\n\n40\n\n50\n\n60\n\n00 01 02 03 04 05 06 07 08 09 10 11\n\nTop 10 market share ¹\n\n0\\.0\n\n0\\.1\n\n0\\.2\n\n0\\.3\n\n0\\.4\n\n0\n\n1\n\n2\n\n3\n\n4\n\n80 83 86 89 92 95 98 01 04 07 10\n\nMarket share of failed companies ² Lhs:\n\nNumber of failures ³ Rhs:","pages":[{"line_numbers":[],"markdown":"BIS Quarterly Review, December 2012 31\n\nIn spite of the reinsurance market’s size and concentration, failures of reinsurance companies have remained limited in scope. The largest failures to date, comprising two bankruptcies in 2003, led to an essentially inconsequential reduction in available reinsurance capacity of 0.4% (Graph 6, left-hand panel). That said, any failure of a reinsurer leads to a loss of reinsurance recoverables by primary insurers, and could cause broader market tensions in the event of a disorderly liquidation of large portfolios.\n\nIn this respect, the degree of connectedness within the global insurance market plays an important role. Based on their business model, reinsurers enter into contracts with a large number of primary insurance companies, giving rise to numerous vertical links (Graph 3). In addition, risk transfer between reinsurers leads to horizontal linkages. ¹⁰ We estimate that 12% of natural catastrophe-related risk accepted by reinsurers is transferred within the reinsurance industry, which implies that the industry as a whole retains most of the risks it contracts. In 2011, reinsurers paid 3% of earned premiums to cede catastrophe risk to entities _outside_ the insurance sector. Judging by premium volume, the global insurance market transfers a similarly small share of accepted risk to other financial institutions and the wider financial markets.\n\n### Linkages with financial markets\n\nArrangements designed to transfer risk out of the insurance sector create linkages with other financial market participants. Retrocession to other financial institutions uses contractual arrangements similar to those between reinsurers, and commits\n\n10 For example, a reinsurer might exchange some of its exposure to earthquake risk in Japan for US flood risk with another reinsurer.\n\nReinsurance market developments Graph 6\n\nReinsurance market concentration\n\nPer cent\n\nFailures of reinsurance companies\n\nPer cent of premiums Number of companies\n\n1 Market share of the 10 largest reinsurance companies, measured as a share of gross premiums written by reinsurance companies worldwide in the non-life (property and casualty) business. ² In relation to total market size as measured by gross premiums written (premiums ceded by insurers to reinsurance companies). ³ Number of failures of reinsurance companies worldwide, per calendar year.\n\nSources: IAIS, based on industry data; authors’ calculations.\n\n30\n\n40\n\n50\n\n60\n\n00 01 02 03 04 05 06 07 08 09 10 11\n\nTop 10 market share ¹\n\n0\\.0\n\n0\\.1\n\n0\\.2\n\n0\\.3\n\n0\\.4\n\n0\n\n1\n\n2\n\n3\n\n4\n\n80 83 86 89 92 95 98 01 04 07 10\n\nMarket share of failed companies ² Lhs:\n\nNumber of failures ³ Rhs:","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"BIS Quarterly Review, December 2012 31\n\nIn spite of the reinsurance market’s size and concentration, failures of reinsurance companies have remained limited in scope. The largest failures to date, comprising two bankruptcies in 2003, led to an essentially inconsequential reduction in available reinsurance capacity of 0.4% (Graph 6, left-hand panel). That said, any failure of a reinsurer leads to a loss of reinsurance recoverables by primary insurers, and could cause broader market tensions in the event of a disorderly liquidation of large portfolios.\n\nIn this respect, the degree of connectedness within the global insurance market plays an important role. Based on their business model, reinsurers enter into contracts with a large number of primary insurance companies, giving rise to numerous vertical links (Graph 3). In addition, risk transfer between reinsurers leads to horizontal linkages. ¹⁰ We estimate that 12% of natural catastrophe-related risk accepted by reinsurers is transferred within the reinsurance industry, which implies that the industry as a whole retains most of the risks it contracts. In 2011, reinsurers paid 3% of earned premiums to cede catastrophe risk to entities _outside_ the insurance sector. Judging by premium volume, the global insurance market transfers a similarly small share of accepted risk to other financial institutions and the wider financial markets.\n\n### Linkages with financial markets\n\nArrangements designed to transfer risk out of the insurance sector create linkages with other financial market participants. Retrocession to other financial institutions uses contractual arrangements similar to those between reinsurers, and commits\n\n10 For example, a reinsurer might exchange some of its exposure to earthquake risk in Japan for US flood risk with another reinsurer.\n\nReinsurance market developments Graph 6\n\nReinsurance market concentration\n\nPer cent\n\nFailures of reinsurance companies\n\nPer cent of premiums Number of companies\n\n1 Market share of the 10 largest reinsurance companies, measured as a share of gross premiums written by reinsurance companies worldwide in the non-life (property and casualty) business. ² In relation to total market size as measured by gross premiums written (premiums ceded by insurers to reinsurance companies). ³ Number of failures of reinsurance companies worldwide, per calendar year.\n\nSources: IAIS, based on industry data; authors’ calculations.\n\n30\n\n40\n\n50\n\n60\n\n00 01 02 03 04 05 06 07 08 09 10 11\n\nTop 10 market share ¹\n\n0\\.0\n\n0\\.1\n\n0\\.2\n\n0\\.3\n\n0\\.4\n\n0\n\n1\n\n2\n\n3\n\n4\n\n80 83 86 89 92 95 98 01 04 07 10\n\nMarket share of failed companies ² Lhs:\n\nNumber of failures ³ Rhs:","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/r_qt1212e_p9.pdf"]},"response":{"errors":[],"extract_id":"extract_0dfd1e56bb9e01bfc6318b88f6fb24b9","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["BIS Quarterly Review, December 2012 31\n\nIn spite of the reinsurance market’s size and concentration, failures of reinsurance companies have remained limited in scope. The largest failures to date, comprising two bankruptcies in 2003, led to an essentially inconsequential reduction in available reinsurance capacity of 0.4% (Graph 6, left-hand panel). That said, any failure of a reinsurer leads to a loss of reinsurance recoverables by primary insurers, and could cause broader market tensions in the event of a disorderly liquidation of large portfolios.\n\nIn this respect, the degree of connectedness within the global insurance market plays an important role. Based on their business model, reinsurers enter into contracts with a large number of primary insurance companies, giving rise to numerous vertical links (Graph 3). In addition, risk transfer between reinsurers leads to horizontal linkages. ¹⁰ We estimate that 12% of natural catastrophe-related risk accepted by reinsurers is transferred within the reinsurance industry, which implies that the industry as a whole retains most of the risks it contracts. In 2011, reinsurers paid 3% of earned premiums to cede catastrophe risk to entities _outside_ the insurance sector. Judging by premium volume, the global insurance market transfers a similarly small share of accepted risk to other financial institutions and the wider financial markets.\n\n### Linkages with financial markets\n\nArrangements designed to transfer risk out of the insurance sector create linkages with other financial market participants. Retrocession to other financial institutions uses contractual arrangements similar to those between reinsurers, and commits\n\n10 For example, a reinsurer might exchange some of its exposure to earthquake risk in Japan for US flood risk with another reinsurer.\n\nReinsurance market developments Graph 6\n\nReinsurance market concentration\n\nPer cent\n\nFailures of reinsurance companies\n\nPer cent of premiums Number of companies\n\n1 Market share of the 10 largest reinsurance companies, measured as a share of gross premiums written by reinsurance companies worldwide in the non-life (property and casualty) business. ² In relation to total market size as measured by gross premiums written (premiums ceded by insurers to reinsurance companies). ³ Number of failures of reinsurance companies worldwide, per calendar year.\n\nSources: IAIS, based on industry data; authors’ calculations.\n\n30\n\n40\n\n50\n\n60\n\n00 01 02 03 04 05 06 07 08 09 10 11\n\nTop 10 market share ¹\n\n0\\.0\n\n0\\.1\n\n0\\.2\n\n0\\.3\n\n0\\.4\n\n0\n\n1\n\n2\n\n3\n\n4\n\n80 83 86 89 92 95 98 01 04 07 10\n\nMarket share of failed companies ² Lhs:\n\nNumber of failures ³ Rhs:"],"full_content":"BIS Quarterly Review, December 2012 31\n\nIn spite of the reinsurance market’s size and concentration, failures of reinsurance companies have remained limited in scope. The largest failures to date, comprising two bankruptcies in 2003, led to an essentially inconsequential reduction in available reinsurance capacity of 0.4% (Graph 6, left-hand panel). That said, any failure of a reinsurer leads to a loss of reinsurance recoverables by primary insurers, and could cause broader market tensions in the event of a disorderly liquidation of large portfolios.\n\nIn this respect, the degree of connectedness within the global insurance market plays an important role. Based on their business model, reinsurers enter into contracts with a large number of primary insurance companies, giving rise to numerous vertical links (Graph 3). In addition, risk transfer between reinsurers leads to horizontal linkages. ¹⁰ We estimate that 12% of natural catastrophe-related risk accepted by reinsurers is transferred within the reinsurance industry, which implies that the industry as a whole retains most of the risks it contracts. In 2011, reinsurers paid 3% of earned premiums to cede catastrophe risk to entities _outside_ the insurance sector. Judging by premium volume, the global insurance market transfers a similarly small share of accepted risk to other financial institutions and the wider financial markets.\n\n### Linkages with financial markets\n\nArrangements designed to transfer risk out of the insurance sector create linkages with other financial market participants. Retrocession to other financial institutions uses contractual arrangements similar to those between reinsurers, and commits\n\n10 For example, a reinsurer might exchange some of its exposure to earthquake risk in Japan for US flood risk with another reinsurer.\n\nReinsurance market developments Graph 6\n\nReinsurance market concentration\n\nPer cent\n\nFailures of reinsurance companies\n\nPer cent of premiums Number of companies\n\n1 Market share of the 10 largest reinsurance companies, measured as a share of gross premiums written by reinsurance companies worldwide in the non-life (property and casualty) business. ² In relation to total market size as measured by gross premiums written (premiums ceded by insurers to reinsurance companies). ³ Number of failures of reinsurance companies worldwide, per calendar year.\n\nSources: IAIS, based on industry data; authors’ calculations.\n\n30\n\n40\n\n50\n\n60\n\n00 01 02 03 04 05 06 07 08 09 10 11\n\nTop 10 market share ¹\n\n0\\.0\n\n0\\.1\n\n0\\.2\n\n0\\.3\n\n0\\.4\n\n0\n\n1\n\n2\n\n3\n\n4\n\n80 83 86 89 92 95 98 01 04 07 10\n\nMarket share of failed companies ² Lhs:\n\nNumber of failures ³ Rhs:","publish_date":null,"title":"Linkages with financial markets","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/r_qt1212e_p9.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/r_qt1212e_p9.pdf"},"request":{"example_id":"chart/r_qt1212e_p9","product_type":"parse","source_file_path":"r_qt1212e_p9.pdf"},"started_at":"2026-11-04T19:45:30.066441"} +{"completed_at":"2026-11-04T19:45:32.996934","latency_in_ms":2122,"output":{"example_id":"chart/salesforce-state-of-data-and-analytics-2nd-edition_p21","grounded_pages":[],"layout_pages":[],"markdown":"**46%**\n\n**43%**\n\n**42%**\n\n**44%**\n\n**47%**\n\n**43%**\n\n**46%**\n\n**47%**\n\n**46%**\n\n**46%**\n\n**SALESFORCE** **21** **STATE OF DATA & ANALY TICS, 2** **ND** **EDITION**\n\n# **Bad Data: The Hidden Reason AI Projects Fail**\n\n### A whopping majority of data and analytics leaders believe their data strategies — how they collect, manage, store, and make available their organizations' data — aren’t up to the task of bridging their AI initiatives with business needs.\n\n## 84% of data and analytics leaders believe their data strategies need overhauls for successful AI.\n\n### The general dissatisfaction with AI outputs can be traced in large part to shortcomings in the data that underpin them. Unfortunately, most respondents continue to lack full confidence in various aspects of their data — ranging from integration and harmonization to governance and management practices and overall quality — and progress has been incremental at best since our 2023 survey.\n\n**2023**\n\n**2025**\n\n# 03\n\n## **Data Foundations Are Barely Improving in the Agentic Era**\n\nQuality\n\nIntegration\n\nHarmonization\n\nInfrastructure\n\nManagement\n\nSecurity/compliance\n\nGovernance\n\nTimeliness\\*\n\n### Bridge data strategy, culture, and execution to build an agentic enterprise. Get the guide\n\n## **Salesforce Perspective**\n\n### **Data and Analytics Leaders Who Rate the Following Data Aspects as Fully Prepared for AI**\n\n**45%**\n\n**44%**\n\n**47%**\n\n**46%**\n\n**49%**\n\n\\* Not included as a variable in 2023.","pages":[{"line_numbers":[],"markdown":"**46%**\n\n**43%**\n\n**42%**\n\n**44%**\n\n**47%**\n\n**43%**\n\n**46%**\n\n**47%**\n\n**46%**\n\n**46%**\n\n**SALESFORCE** **21** **STATE OF DATA & ANALY TICS, 2** **ND** **EDITION**\n\n# **Bad Data: The Hidden Reason AI Projects Fail**\n\n### A whopping majority of data and analytics leaders believe their data strategies — how they collect, manage, store, and make available their organizations' data — aren’t up to the task of bridging their AI initiatives with business needs.\n\n## 84% of data and analytics leaders believe their data strategies need overhauls for successful AI.\n\n### The general dissatisfaction with AI outputs can be traced in large part to shortcomings in the data that underpin them. Unfortunately, most respondents continue to lack full confidence in various aspects of their data — ranging from integration and harmonization to governance and management practices and overall quality — and progress has been incremental at best since our 2023 survey.\n\n**2023**\n\n**2025**\n\n# 03\n\n## **Data Foundations Are Barely Improving in the Agentic Era**\n\nQuality\n\nIntegration\n\nHarmonization\n\nInfrastructure\n\nManagement\n\nSecurity/compliance\n\nGovernance\n\nTimeliness\\*\n\n### Bridge data strategy, culture, and execution to build an agentic enterprise. Get the guide\n\n## **Salesforce Perspective**\n\n### **Data and Analytics Leaders Who Rate the Following Data Aspects as Fully Prepared for AI**\n\n**45%**\n\n**44%**\n\n**47%**\n\n**46%**\n\n**49%**\n\n\\* Not included as a variable in 2023.","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**46%**\n\n**43%**\n\n**42%**\n\n**44%**\n\n**47%**\n\n**43%**\n\n**46%**\n\n**47%**\n\n**46%**\n\n**46%**\n\n**SALESFORCE** **21** **STATE OF DATA & ANALY TICS, 2** **ND** **EDITION**\n\n# **Bad Data: The Hidden Reason AI Projects Fail**\n\n### A whopping majority of data and analytics leaders believe their data strategies — how they collect, manage, store, and make available their organizations' data — aren’t up to the task of bridging their AI initiatives with business needs.\n\n## 84% of data and analytics leaders believe their data strategies need overhauls for successful AI.\n\n### The general dissatisfaction with AI outputs can be traced in large part to shortcomings in the data that underpin them. Unfortunately, most respondents continue to lack full confidence in various aspects of their data — ranging from integration and harmonization to governance and management practices and overall quality — and progress has been incremental at best since our 2023 survey.\n\n**2023**\n\n**2025**\n\n# 03\n\n## **Data Foundations Are Barely Improving in the Agentic Era**\n\nQuality\n\nIntegration\n\nHarmonization\n\nInfrastructure\n\nManagement\n\nSecurity/compliance\n\nGovernance\n\nTimeliness\\*\n\n### Bridge data strategy, culture, and execution to build an agentic enterprise. Get the guide\n\n## **Salesforce Perspective**\n\n### **Data and Analytics Leaders Who Rate the Following Data Aspects as Fully Prepared for AI**\n\n**45%**\n\n**44%**\n\n**47%**\n\n**46%**\n\n**49%**\n\n\\* Not included as a variable in 2023.","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/salesforce-state-of-data-and-analytics-2nd-edition_p21.pdf"]},"response":{"errors":[],"extract_id":"extract_5e59cc51444a0efb42437adc466d2322","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**46%**\n\n**43%**\n\n**42%**\n\n**44%**\n\n**47%**\n\n**43%**\n\n**46%**\n\n**47%**\n\n**46%**\n\n**46%**\n\n**SALESFORCE** **21** **STATE OF DATA & ANALY TICS, 2** **ND** **EDITION**\n\n# **Bad Data: The Hidden Reason AI Projects Fail**\n\n### A whopping majority of data and analytics leaders believe their data strategies — how they collect, manage, store, and make available their organizations' data — aren’t up to the task of bridging their AI initiatives with business needs.\n\n## 84% of data and analytics leaders believe their data strategies need overhauls for successful AI.\n\n### The general dissatisfaction with AI outputs can be traced in large part to shortcomings in the data that underpin them. Unfortunately, most respondents continue to lack full confidence in various aspects of their data — ranging from integration and harmonization to governance and management practices and overall quality — and progress has been incremental at best since our 2023 survey.\n\n**2023**\n\n**2025**\n\n# 03\n\n## **Data Foundations Are Barely Improving in the Agentic Era**\n\nQuality\n\nIntegration\n\nHarmonization\n\nInfrastructure\n\nManagement\n\nSecurity/compliance\n\nGovernance\n\nTimeliness\\*\n\n### Bridge data strategy, culture, and execution to build an agentic enterprise. Get the guide\n\n## **Salesforce Perspective**\n\n### **Data and Analytics Leaders Who Rate the Following Data Aspects as Fully Prepared for AI**\n\n**45%**\n\n**44%**\n\n**47%**\n\n**46%**\n\n**49%**\n\n\\* Not included as a variable in 2023."],"full_content":"**46%**\n\n**43%**\n\n**42%**\n\n**44%**\n\n**47%**\n\n**43%**\n\n**46%**\n\n**47%**\n\n**46%**\n\n**46%**\n\n**SALESFORCE** **21** **STATE OF DATA & ANALY TICS, 2** **ND** **EDITION**\n\n# **Bad Data: The Hidden Reason AI Projects Fail**\n\n### A whopping majority of data and analytics leaders believe their data strategies — how they collect, manage, store, and make available their organizations' data — aren’t up to the task of bridging their AI initiatives with business needs.\n\n## 84% of data and analytics leaders believe their data strategies need overhauls for successful AI.\n\n### The general dissatisfaction with AI outputs can be traced in large part to shortcomings in the data that underpin them. Unfortunately, most respondents continue to lack full confidence in various aspects of their data — ranging from integration and harmonization to governance and management practices and overall quality — and progress has been incremental at best since our 2023 survey.\n\n**2023**\n\n**2025**\n\n# 03\n\n## **Data Foundations Are Barely Improving in the Agentic Era**\n\nQuality\n\nIntegration\n\nHarmonization\n\nInfrastructure\n\nManagement\n\nSecurity/compliance\n\nGovernance\n\nTimeliness\\*\n\n### Bridge data strategy, culture, and execution to build an agentic enterprise. Get the guide\n\n## **Salesforce Perspective**\n\n### **Data and Analytics Leaders Who Rate the Following Data Aspects as Fully Prepared for AI**\n\n**45%**\n\n**44%**\n\n**47%**\n\n**46%**\n\n**49%**\n\n\\* Not included as a variable in 2023.","publish_date":null,"title":"Bad Data: The Hidden Reason AI Projects Fail","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/salesforce-state-of-data-and-analytics-2nd-edition_p21.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/salesforce-state-of-data-and-analytics-2nd-edition_p21.pdf"},"request":{"example_id":"chart/salesforce-state-of-data-and-analytics-2nd-edition_p21","product_type":"parse","source_file_path":"salesforce-state-of-data-and-analytics-2nd-edition_p21.pdf"},"started_at":"2026-11-04T19:45:30.874678"} +{"completed_at":"2026-11-04T19:45:35.257891","latency_in_ms":3240,"output":{"example_id":"chart/semiconductor_sector_study_p15","grounded_pages":[],"layout_pages":[],"markdown":"PERSPECTIVE ECONOMICS\n\nFigure 2.5 – Dedicated Companies by Sector (UK HQ | Intl. HQ)\n\nSource: Glass.ai _(note: totals may not sum due to rounding)_\n\n2\\.2.5. Regional Semiconductor Supply Chain Activity\n\nApproximately half of all dedicated semiconductor companies (including both internationally and UK headquartered companies) are located outside of London, the South East and the East of England.\n\nArm accounts for a significant proportion of both total revenue, and total employment among dedicated, UK headquartered semiconductor companies (54% and 32% of total revenue and employment respectively). With its headquarters in Cambridge, Arm makes the East of England a focal point of semiconductor sector activity in the UK.\n\nTo better understand the geographic composition of semiconductor activity among other companies, the study looked specifically at a subset of dedicated UK headquartered companies excluding Arm vi . This subsequent analysis emphasises the significance of regions outside London and the South East to the UK’s semiconductor sector (Table 2.2). For example, within this subset, Wales is home to less than 10% of dedicated, UK headquartered companies, but accounts for 14% of employment and 30% of revenues.","pages":[{"line_numbers":[],"markdown":"PERSPECTIVE ECONOMICS\n\nFigure 2.5 – Dedicated Companies by Sector (UK HQ | Intl. HQ)\n\nSource: Glass.ai _(note: totals may not sum due to rounding)_\n\n2\\.2.5. Regional Semiconductor Supply Chain Activity\n\nApproximately half of all dedicated semiconductor companies (including both internationally and UK headquartered companies) are located outside of London, the South East and the East of England.\n\nArm accounts for a significant proportion of both total revenue, and total employment among dedicated, UK headquartered semiconductor companies (54% and 32% of total revenue and employment respectively). With its headquarters in Cambridge, Arm makes the East of England a focal point of semiconductor sector activity in the UK.\n\nTo better understand the geographic composition of semiconductor activity among other companies, the study looked specifically at a subset of dedicated UK headquartered companies excluding Arm vi . This subsequent analysis emphasises the significance of regions outside London and the South East to the UK’s semiconductor sector (Table 2.2). For example, within this subset, Wales is home to less than 10% of dedicated, UK headquartered companies, but accounts for 14% of employment and 30% of revenues.","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"PERSPECTIVE ECONOMICS\n\nFigure 2.5 – Dedicated Companies by Sector (UK HQ | Intl. HQ)\n\nSource: Glass.ai _(note: totals may not sum due to rounding)_\n\n2\\.2.5. Regional Semiconductor Supply Chain Activity\n\nApproximately half of all dedicated semiconductor companies (including both internationally and UK headquartered companies) are located outside of London, the South East and the East of England.\n\nArm accounts for a significant proportion of both total revenue, and total employment among dedicated, UK headquartered semiconductor companies (54% and 32% of total revenue and employment respectively). With its headquarters in Cambridge, Arm makes the East of England a focal point of semiconductor sector activity in the UK.\n\nTo better understand the geographic composition of semiconductor activity among other companies, the study looked specifically at a subset of dedicated UK headquartered companies excluding Arm vi . This subsequent analysis emphasises the significance of regions outside London and the South East to the UK’s semiconductor sector (Table 2.2). For example, within this subset, Wales is home to less than 10% of dedicated, UK headquartered companies, but accounts for 14% of employment and 30% of revenues.","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/semiconductor_sector_study_p15.pdf"]},"response":{"errors":[],"extract_id":"extract_0baf52e49a665634aec7a26ed33bc554","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["PERSPECTIVE ECONOMICS\n\nFigure 2.5 – Dedicated Companies by Sector (UK HQ | Intl. HQ)\n\nSource: Glass.ai _(note: totals may not sum due to rounding)_\n\n2\\.2.5. Regional Semiconductor Supply Chain Activity\n\nApproximately half of all dedicated semiconductor companies (including both internationally and UK headquartered companies) are located outside of London, the South East and the East of England.\n\nArm accounts for a significant proportion of both total revenue, and total employment among dedicated, UK headquartered semiconductor companies (54% and 32% of total revenue and employment respectively). With its headquarters in Cambridge, Arm makes the East of England a focal point of semiconductor sector activity in the UK.\n\nTo better understand the geographic composition of semiconductor activity among other companies, the study looked specifically at a subset of dedicated UK headquartered companies excluding Arm vi . This subsequent analysis emphasises the significance of regions outside London and the South East to the UK’s semiconductor sector (Table 2.2). For example, within this subset, Wales is home to less than 10% of dedicated, UK headquartered companies, but accounts for 14% of employment and 30% of revenues."],"full_content":"PERSPECTIVE ECONOMICS\n\nFigure 2.5 – Dedicated Companies by Sector (UK HQ | Intl. HQ)\n\nSource: Glass.ai _(note: totals may not sum due to rounding)_\n\n2\\.2.5. Regional Semiconductor Supply Chain Activity\n\nApproximately half of all dedicated semiconductor companies (including both internationally and UK headquartered companies) are located outside of London, the South East and the East of England.\n\nArm accounts for a significant proportion of both total revenue, and total employment among dedicated, UK headquartered semiconductor companies (54% and 32% of total revenue and employment respectively). With its headquarters in Cambridge, Arm makes the East of England a focal point of semiconductor sector activity in the UK.\n\nTo better understand the geographic composition of semiconductor activity among other companies, the study looked specifically at a subset of dedicated UK headquartered companies excluding Arm vi . This subsequent analysis emphasises the significance of regions outside London and the South East to the UK’s semiconductor sector (Table 2.2). For example, within this subset, Wales is home to less than 10% of dedicated, UK headquartered companies, but accounts for 14% of employment and 30% of revenues.","publish_date":null,"title":"Figure 2.5 – Dedicated Companies by Sector (UK HQ | Intl. HQ) (note: totals may not sum due to rounding)","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/semiconductor_sector_study_p15.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/semiconductor_sector_study_p15.pdf"},"request":{"example_id":"chart/semiconductor_sector_study_p15","product_type":"parse","source_file_path":"semiconductor_sector_study_p15.pdf"},"started_at":"2026-11-04T19:45:32.017648"} +{"completed_at":"2026-11-04T19:45:34.695001","latency_in_ms":2071,"output":{"example_id":"chart/semiconductor_sector_study_p22","grounded_pages":[],"layout_pages":[],"markdown":"PERSPECTIVE ECONOMICS\n\n# 4\\. Investment\n\n### 4\\.1. Investment to Date\n\nDedicated UK semiconductor companies have secured a total of £1.7bn in grants and fundraisings xii . This includes £291m in grants and £1.4bn in fundraisings. Forty-four companies have secured fundraisings, with approximately 70% of total fundraisings having been secured by five companies.\n\nFigure 4.1 shows the value of fundraisings secured by dedicated semiconductor companies within their latest fundraising rounds. N.B. previous fundraising values are not shown and totals therefore do not sum to the total £1.4bn.\n\nFigure 4.1 – Latest Fundraisings\n\nSource: Beauhurst (all latest fundraisings, 2007 - 2023)\n\n4\\.1.1. Investment by Stage of Evolution\n\nJust under three quarters of total _fundraisings_ have been secured by growth-stage semiconductor companies. The value of _grants_ secured by dedicated companies is more evenly spread across stages of evolution, including notably higher proportions secured by","pages":[{"line_numbers":[],"markdown":"PERSPECTIVE ECONOMICS\n\n# 4\\. Investment\n\n### 4\\.1. Investment to Date\n\nDedicated UK semiconductor companies have secured a total of £1.7bn in grants and fundraisings xii . This includes £291m in grants and £1.4bn in fundraisings. Forty-four companies have secured fundraisings, with approximately 70% of total fundraisings having been secured by five companies.\n\nFigure 4.1 shows the value of fundraisings secured by dedicated semiconductor companies within their latest fundraising rounds. N.B. previous fundraising values are not shown and totals therefore do not sum to the total £1.4bn.\n\nFigure 4.1 – Latest Fundraisings\n\nSource: Beauhurst (all latest fundraisings, 2007 - 2023)\n\n4\\.1.1. Investment by Stage of Evolution\n\nJust under three quarters of total _fundraisings_ have been secured by growth-stage semiconductor companies. The value of _grants_ secured by dedicated companies is more evenly spread across stages of evolution, including notably higher proportions secured by","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"PERSPECTIVE ECONOMICS\n\n# 4\\. Investment\n\n### 4\\.1. Investment to Date\n\nDedicated UK semiconductor companies have secured a total of £1.7bn in grants and fundraisings xii . This includes £291m in grants and £1.4bn in fundraisings. Forty-four companies have secured fundraisings, with approximately 70% of total fundraisings having been secured by five companies.\n\nFigure 4.1 shows the value of fundraisings secured by dedicated semiconductor companies within their latest fundraising rounds. N.B. previous fundraising values are not shown and totals therefore do not sum to the total £1.4bn.\n\nFigure 4.1 – Latest Fundraisings\n\nSource: Beauhurst (all latest fundraisings, 2007 - 2023)\n\n4\\.1.1. Investment by Stage of Evolution\n\nJust under three quarters of total _fundraisings_ have been secured by growth-stage semiconductor companies. The value of _grants_ secured by dedicated companies is more evenly spread across stages of evolution, including notably higher proportions secured by","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/semiconductor_sector_study_p22.pdf"]},"response":{"errors":[],"extract_id":"extract_0cd74e3ed44a65fb243657510056b079","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["PERSPECTIVE ECONOMICS\n\n# 4\\. Investment\n\n### 4\\.1. Investment to Date\n\nDedicated UK semiconductor companies have secured a total of £1.7bn in grants and fundraisings xii . This includes £291m in grants and £1.4bn in fundraisings. Forty-four companies have secured fundraisings, with approximately 70% of total fundraisings having been secured by five companies.\n\nFigure 4.1 shows the value of fundraisings secured by dedicated semiconductor companies within their latest fundraising rounds. N.B. previous fundraising values are not shown and totals therefore do not sum to the total £1.4bn.\n\nFigure 4.1 – Latest Fundraisings\n\nSource: Beauhurst (all latest fundraisings, 2007 - 2023)\n\n4\\.1.1. Investment by Stage of Evolution\n\nJust under three quarters of total _fundraisings_ have been secured by growth-stage semiconductor companies. The value of _grants_ secured by dedicated companies is more evenly spread across stages of evolution, including notably higher proportions secured by"],"full_content":"PERSPECTIVE ECONOMICS\n\n# 4\\. Investment\n\n### 4\\.1. Investment to Date\n\nDedicated UK semiconductor companies have secured a total of £1.7bn in grants and fundraisings xii . This includes £291m in grants and £1.4bn in fundraisings. Forty-four companies have secured fundraisings, with approximately 70% of total fundraisings having been secured by five companies.\n\nFigure 4.1 shows the value of fundraisings secured by dedicated semiconductor companies within their latest fundraising rounds. N.B. previous fundraising values are not shown and totals therefore do not sum to the total £1.4bn.\n\nFigure 4.1 – Latest Fundraisings\n\nSource: Beauhurst (all latest fundraisings, 2007 - 2023)\n\n4\\.1.1. Investment by Stage of Evolution\n\nJust under three quarters of total _fundraisings_ have been secured by growth-stage semiconductor companies. The value of _grants_ secured by dedicated companies is more evenly spread across stages of evolution, including notably higher proportions secured by","publish_date":null,"title":"4. Investment","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/semiconductor_sector_study_p22.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/semiconductor_sector_study_p22.pdf"},"request":{"example_id":"chart/semiconductor_sector_study_p22","product_type":"parse","source_file_path":"semiconductor_sector_study_p22.pdf"},"started_at":"2026-11-04T19:45:32.623728"} +{"completed_at":"2026-11-04T19:45:36.121694","latency_in_ms":2172,"output":{"example_id":"chart/sigma-1-2021-en_p12","grounded_pages":[],"layout_pages":[],"markdown":"10 **Swiss Re** _sigma_ No 1/2021\n\n**Insuring subsidence: the French example** Subsidence risk has long been on the policy agenda in France, more so than in other markets. A national natural catastrophe insurance programme – CatNat – has covered property losses caused by subsidence since 1989. From 2016, inflation- corrected insured losses from subsidence have exceeded EUR 550 million each year (see Figure 6), with the average annual loss standing at EUR 810 million. ¹⁴ That corresponds to about 45% of the annual CatNat premiums collected, making subsidence the costliest natural hazard in France. ¹⁵ Compared with an average annual loss of EUR 310 million in 2000–2015, the sudden and persistent yearly explosion in claims post-2016 is puzzling.\n\nA Swiss Re study in 2011 suggested that climate change would lead to more droughts and increasing subsidence losses in France. Considering the dry years of 2017–2019, it seems this prediction has been confirmed. ¹⁶ That said, we would expect climate change influence to appear in a gradual upward trend of annual losses rather than a sudden, step-like increase. The loss pattern post-2016 suggests additional factors at play.\n\nOne of the drivers may be the French insurance scheme itself, a key feature of which is that claimants are compensated for subsidence losses only if their municipality has been awarded a so-called CatNat declaration. ¹⁷ The mayor of each municipality must first request such a declaration, the grant of which is dependent on whether the region has experienced what is determined to be a drought, and if specific geological prerequisites are met. In recent years, there has been an increase in the number of municipalities requesting the declaration. In 2018 and 2019, of all requests, 23% came from first-time applicants (see Figure 7). This cannot be explained by an increase in drought conditions alone. ¹⁸ It may also be due to increasing awareness of the possibility to receive compensation for subsidence losses.\n\nWe believe awareness of the opportunity to claim for subsidence losses will likely continue to rise. Of all municipalities, 62% have yet to request a CatNat declaration\n\n14 _Les catastrophes naturelles en France_ , Bilan 1982–2019, 2020, Caisse Centrale de Réassurance 15 In 2018, EUR 1 649 million of CatNat premiums were collected in France, most as cover for flood, subsidence, and earthquake risk. _L’assurance des catastrophes naturelles en 2018_ , FFA, 2018. 16 _The hidden risks of climate change: An increase in property damage from soil subsidence in Europe_ , Swiss Re, 2011. 17 Based on pre-defined criteria, the French administration provides CatNat declarations to municipalities that have been affected by a natural disaster. 18 “arrêtés de catastrophe naturelle”, _CCR_ , 2020.\n\nFrance has had a nationwide insurance scheme for subsidence since 1989.\n\n**Figure 6** Annual insured losses from subsidence in France, EUR million\n\nSource: Les catastrophes naturelles en France, Bilan 1982–2019, Caisse Centrale de Réassurance, 2020\n\n0\n\n500\n\n1 000\n\n1 500\n\n2 000\n\nErr\n\nCC\n\nme\n\nme\n\n2019 2017 2015 2013 2011 2009 2007 2005 2003 2001\n\nCCR 2000–2015 2016–2019 Mean loss: Error range\n\nClimate change will likely exacerbate soil subsidence…\n\n…but recent spikes in insured losses are likely due to growing awareness of the nationwide scheme.\n\nIf climate conditions remain dry, claims will likely continue to rise.\n\n### **Exploring the secondary perils universe**","pages":[{"line_numbers":[],"markdown":"10 **Swiss Re** _sigma_ No 1/2021\n\n**Insuring subsidence: the French example** Subsidence risk has long been on the policy agenda in France, more so than in other markets. A national natural catastrophe insurance programme – CatNat – has covered property losses caused by subsidence since 1989. From 2016, inflation- corrected insured losses from subsidence have exceeded EUR 550 million each year (see Figure 6), with the average annual loss standing at EUR 810 million. ¹⁴ That corresponds to about 45% of the annual CatNat premiums collected, making subsidence the costliest natural hazard in France. ¹⁵ Compared with an average annual loss of EUR 310 million in 2000–2015, the sudden and persistent yearly explosion in claims post-2016 is puzzling.\n\nA Swiss Re study in 2011 suggested that climate change would lead to more droughts and increasing subsidence losses in France. Considering the dry years of 2017–2019, it seems this prediction has been confirmed. ¹⁶ That said, we would expect climate change influence to appear in a gradual upward trend of annual losses rather than a sudden, step-like increase. The loss pattern post-2016 suggests additional factors at play.\n\nOne of the drivers may be the French insurance scheme itself, a key feature of which is that claimants are compensated for subsidence losses only if their municipality has been awarded a so-called CatNat declaration. ¹⁷ The mayor of each municipality must first request such a declaration, the grant of which is dependent on whether the region has experienced what is determined to be a drought, and if specific geological prerequisites are met. In recent years, there has been an increase in the number of municipalities requesting the declaration. In 2018 and 2019, of all requests, 23% came from first-time applicants (see Figure 7). This cannot be explained by an increase in drought conditions alone. ¹⁸ It may also be due to increasing awareness of the possibility to receive compensation for subsidence losses.\n\nWe believe awareness of the opportunity to claim for subsidence losses will likely continue to rise. Of all municipalities, 62% have yet to request a CatNat declaration\n\n14 _Les catastrophes naturelles en France_ , Bilan 1982–2019, 2020, Caisse Centrale de Réassurance 15 In 2018, EUR 1 649 million of CatNat premiums were collected in France, most as cover for flood, subsidence, and earthquake risk. _L’assurance des catastrophes naturelles en 2018_ , FFA, 2018. 16 _The hidden risks of climate change: An increase in property damage from soil subsidence in Europe_ , Swiss Re, 2011. 17 Based on pre-defined criteria, the French administration provides CatNat declarations to municipalities that have been affected by a natural disaster. 18 “arrêtés de catastrophe naturelle”, _CCR_ , 2020.\n\nFrance has had a nationwide insurance scheme for subsidence since 1989.\n\n**Figure 6** Annual insured losses from subsidence in France, EUR million\n\nSource: Les catastrophes naturelles en France, Bilan 1982–2019, Caisse Centrale de Réassurance, 2020\n\n0\n\n500\n\n1 000\n\n1 500\n\n2 000\n\nErr\n\nCC\n\nme\n\nme\n\n2019 2017 2015 2013 2011 2009 2007 2005 2003 2001\n\nCCR 2000–2015 2016–2019 Mean loss: Error range\n\nClimate change will likely exacerbate soil subsidence…\n\n…but recent spikes in insured losses are likely due to growing awareness of the nationwide scheme.\n\nIf climate conditions remain dry, claims will likely continue to rise.\n\n### **Exploring the secondary perils universe**","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"10 **Swiss Re** _sigma_ No 1/2021\n\n**Insuring subsidence: the French example** Subsidence risk has long been on the policy agenda in France, more so than in other markets. A national natural catastrophe insurance programme – CatNat – has covered property losses caused by subsidence since 1989. From 2016, inflation- corrected insured losses from subsidence have exceeded EUR 550 million each year (see Figure 6), with the average annual loss standing at EUR 810 million. ¹⁴ That corresponds to about 45% of the annual CatNat premiums collected, making subsidence the costliest natural hazard in France. ¹⁵ Compared with an average annual loss of EUR 310 million in 2000–2015, the sudden and persistent yearly explosion in claims post-2016 is puzzling.\n\nA Swiss Re study in 2011 suggested that climate change would lead to more droughts and increasing subsidence losses in France. Considering the dry years of 2017–2019, it seems this prediction has been confirmed. ¹⁶ That said, we would expect climate change influence to appear in a gradual upward trend of annual losses rather than a sudden, step-like increase. The loss pattern post-2016 suggests additional factors at play.\n\nOne of the drivers may be the French insurance scheme itself, a key feature of which is that claimants are compensated for subsidence losses only if their municipality has been awarded a so-called CatNat declaration. ¹⁷ The mayor of each municipality must first request such a declaration, the grant of which is dependent on whether the region has experienced what is determined to be a drought, and if specific geological prerequisites are met. In recent years, there has been an increase in the number of municipalities requesting the declaration. In 2018 and 2019, of all requests, 23% came from first-time applicants (see Figure 7). This cannot be explained by an increase in drought conditions alone. ¹⁸ It may also be due to increasing awareness of the possibility to receive compensation for subsidence losses.\n\nWe believe awareness of the opportunity to claim for subsidence losses will likely continue to rise. Of all municipalities, 62% have yet to request a CatNat declaration\n\n14 _Les catastrophes naturelles en France_ , Bilan 1982–2019, 2020, Caisse Centrale de Réassurance 15 In 2018, EUR 1 649 million of CatNat premiums were collected in France, most as cover for flood, subsidence, and earthquake risk. _L’assurance des catastrophes naturelles en 2018_ , FFA, 2018. 16 _The hidden risks of climate change: An increase in property damage from soil subsidence in Europe_ , Swiss Re, 2011. 17 Based on pre-defined criteria, the French administration provides CatNat declarations to municipalities that have been affected by a natural disaster. 18 “arrêtés de catastrophe naturelle”, _CCR_ , 2020.\n\nFrance has had a nationwide insurance scheme for subsidence since 1989.\n\n**Figure 6** Annual insured losses from subsidence in France, EUR million\n\nSource: Les catastrophes naturelles en France, Bilan 1982–2019, Caisse Centrale de Réassurance, 2020\n\n0\n\n500\n\n1 000\n\n1 500\n\n2 000\n\nErr\n\nCC\n\nme\n\nme\n\n2019 2017 2015 2013 2011 2009 2007 2005 2003 2001\n\nCCR 2000–2015 2016–2019 Mean loss: Error range\n\nClimate change will likely exacerbate soil subsidence…\n\n…but recent spikes in insured losses are likely due to growing awareness of the nationwide scheme.\n\nIf climate conditions remain dry, claims will likely continue to rise.\n\n### **Exploring the secondary perils universe**","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/sigma-1-2021-en_p12.pdf"]},"response":{"errors":[],"extract_id":"extract_a62fdcb73d63b0951a1ce29d462941f9","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["10 **Swiss Re** _sigma_ No 1/2021\n\n**Insuring subsidence: the French example** Subsidence risk has long been on the policy agenda in France, more so than in other markets. A national natural catastrophe insurance programme – CatNat – has covered property losses caused by subsidence since 1989. From 2016, inflation- corrected insured losses from subsidence have exceeded EUR 550 million each year (see Figure 6), with the average annual loss standing at EUR 810 million. ¹⁴ That corresponds to about 45% of the annual CatNat premiums collected, making subsidence the costliest natural hazard in France. ¹⁵ Compared with an average annual loss of EUR 310 million in 2000–2015, the sudden and persistent yearly explosion in claims post-2016 is puzzling.\n\nA Swiss Re study in 2011 suggested that climate change would lead to more droughts and increasing subsidence losses in France. Considering the dry years of 2017–2019, it seems this prediction has been confirmed. ¹⁶ That said, we would expect climate change influence to appear in a gradual upward trend of annual losses rather than a sudden, step-like increase. The loss pattern post-2016 suggests additional factors at play.\n\nOne of the drivers may be the French insurance scheme itself, a key feature of which is that claimants are compensated for subsidence losses only if their municipality has been awarded a so-called CatNat declaration. ¹⁷ The mayor of each municipality must first request such a declaration, the grant of which is dependent on whether the region has experienced what is determined to be a drought, and if specific geological prerequisites are met. In recent years, there has been an increase in the number of municipalities requesting the declaration. In 2018 and 2019, of all requests, 23% came from first-time applicants (see Figure 7). This cannot be explained by an increase in drought conditions alone. ¹⁸ It may also be due to increasing awareness of the possibility to receive compensation for subsidence losses.\n\nWe believe awareness of the opportunity to claim for subsidence losses will likely continue to rise. Of all municipalities, 62% have yet to request a CatNat declaration\n\n14 _Les catastrophes naturelles en France_ , Bilan 1982–2019, 2020, Caisse Centrale de Réassurance 15 In 2018, EUR 1 649 million of CatNat premiums were collected in France, most as cover for flood, subsidence, and earthquake risk. _L’assurance des catastrophes naturelles en 2018_ , FFA, 2018. 16 _The hidden risks of climate change: An increase in property damage from soil subsidence in Europe_ , Swiss Re, 2011. 17 Based on pre-defined criteria, the French administration provides CatNat declarations to municipalities that have been affected by a natural disaster. 18 “arrêtés de catastrophe naturelle”, _CCR_ , 2020.\n\nFrance has had a nationwide insurance scheme for subsidence since 1989.\n\n**Figure 6** Annual insured losses from subsidence in France, EUR million\n\nSource: Les catastrophes naturelles en France, Bilan 1982–2019, Caisse Centrale de Réassurance, 2020\n\n0\n\n500\n\n1 000\n\n1 500\n\n2 000\n\nErr\n\nCC\n\nme\n\nme\n\n2019 2017 2015 2013 2011 2009 2007 2005 2003 2001\n\nCCR 2000–2015 2016–2019 Mean loss: Error range\n\nClimate change will likely exacerbate soil subsidence…\n\n…but recent spikes in insured losses are likely due to growing awareness of the nationwide scheme.\n\nIf climate conditions remain dry, claims will likely continue to rise.\n\n### **Exploring the secondary perils universe**"],"full_content":"10 **Swiss Re** _sigma_ No 1/2021\n\n**Insuring subsidence: the French example** Subsidence risk has long been on the policy agenda in France, more so than in other markets. A national natural catastrophe insurance programme – CatNat – has covered property losses caused by subsidence since 1989. From 2016, inflation- corrected insured losses from subsidence have exceeded EUR 550 million each year (see Figure 6), with the average annual loss standing at EUR 810 million. ¹⁴ That corresponds to about 45% of the annual CatNat premiums collected, making subsidence the costliest natural hazard in France. ¹⁵ Compared with an average annual loss of EUR 310 million in 2000–2015, the sudden and persistent yearly explosion in claims post-2016 is puzzling.\n\nA Swiss Re study in 2011 suggested that climate change would lead to more droughts and increasing subsidence losses in France. Considering the dry years of 2017–2019, it seems this prediction has been confirmed. ¹⁶ That said, we would expect climate change influence to appear in a gradual upward trend of annual losses rather than a sudden, step-like increase. The loss pattern post-2016 suggests additional factors at play.\n\nOne of the drivers may be the French insurance scheme itself, a key feature of which is that claimants are compensated for subsidence losses only if their municipality has been awarded a so-called CatNat declaration. ¹⁷ The mayor of each municipality must first request such a declaration, the grant of which is dependent on whether the region has experienced what is determined to be a drought, and if specific geological prerequisites are met. In recent years, there has been an increase in the number of municipalities requesting the declaration. In 2018 and 2019, of all requests, 23% came from first-time applicants (see Figure 7). This cannot be explained by an increase in drought conditions alone. ¹⁸ It may also be due to increasing awareness of the possibility to receive compensation for subsidence losses.\n\nWe believe awareness of the opportunity to claim for subsidence losses will likely continue to rise. Of all municipalities, 62% have yet to request a CatNat declaration\n\n14 _Les catastrophes naturelles en France_ , Bilan 1982–2019, 2020, Caisse Centrale de Réassurance 15 In 2018, EUR 1 649 million of CatNat premiums were collected in France, most as cover for flood, subsidence, and earthquake risk. _L’assurance des catastrophes naturelles en 2018_ , FFA, 2018. 16 _The hidden risks of climate change: An increase in property damage from soil subsidence in Europe_ , Swiss Re, 2011. 17 Based on pre-defined criteria, the French administration provides CatNat declarations to municipalities that have been affected by a natural disaster. 18 “arrêtés de catastrophe naturelle”, _CCR_ , 2020.\n\nFrance has had a nationwide insurance scheme for subsidence since 1989.\n\n**Figure 6** Annual insured losses from subsidence in France, EUR million\n\nSource: Les catastrophes naturelles en France, Bilan 1982–2019, Caisse Centrale de Réassurance, 2020\n\n0\n\n500\n\n1 000\n\n1 500\n\n2 000\n\nErr\n\nCC\n\nme\n\nme\n\n2019 2017 2015 2013 2011 2009 2007 2005 2003 2001\n\nCCR 2000–2015 2016–2019 Mean loss: Error range\n\nClimate change will likely exacerbate soil subsidence…\n\n…but recent spikes in insured losses are likely due to growing awareness of the nationwide scheme.\n\nIf climate conditions remain dry, claims will likely continue to rise.\n\n### **Exploring the secondary perils universe**","publish_date":null,"title":"Exploring the secondary perils universe","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/sigma-1-2021-en_p12.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/sigma-1-2021-en_p12.pdf"},"request":{"example_id":"chart/sigma-1-2021-en_p12","product_type":"parse","source_file_path":"sigma-1-2021-en_p12.pdf"},"started_at":"2026-11-04T19:45:33.948721"} +{"completed_at":"2026-11-04T19:45:36.172783","latency_in_ms":1103,"output":{"example_id":"chart/sigma-1-2021-en_p20","grounded_pages":[],"layout_pages":[],"markdown":"18 **Swiss Re** _sigma_ No 1/2021\n\n# **Primary and secondary perils, and the way forward**\n\n### Don’t forget about primary-peril risks\n\nAccording to _sigma_ records, from 1970 total (primary + secondary peril) annual insured losses exceeded USD 80 billion in only six years, all since 2005. Losses were highest when large primary perils events occurred, the peak-loss years being 2017 (Hurricanes Harvey, Irma, Maria), 2011 (earthquakes in Japan and New Zealand, flood in Thailand), 2005 (Hurricanes Katrina, Wilma, Rita) and 2012 (Hurricane Sandy). There has been a growing contribution of secondary perils to total losses in off-peak years, notably in 2018 and 2020. More than 50% of all catastrophe-related insured losses in 2016–2020 were from secondary perils. In 2020 alone, the share of secondary-related was more than 70% of total losses.\n\nHowever, this does not mean secondary perils are becoming primary loss drivers. Over the past decades, the contribution of secondary perils to total annual insurance losses has fluctuated between 20–90%, without a clear trend in either direction. In years like 2020, secondary perils have been in the spotlight, but this is no reason to believe that peak primary-peril loss events will not occur again. Since 1970, losses from both peril groups have been accumulating more or less in sync (see Figure 11). And while secondary peril losses have accumulated faster in specific years, history shows that it can take just a few peak-event strikes to demonstrate the very large magnitude of losses that primary peril risk poses, as in the case of Hurricanes Katrina, Wilma and Rita in 2005.\n\n**Insured losses from primary and secondary perils have been on an upward trajectory since 1970. In the last decade, in years absent major losses from primary perils, accumulated losses from secondary perils alone have been high, 2020 being a case in point. The risk trends impacting both peril types remain the same, and rising losses from secondary perils are a warning that future peak-loss scenarios could also grow significantly. However, insurance industry risk knowledge of secondary perils lags that of primary perils. Better understanding of the complex interplay of factors influencing secondary peril event losses – natural and socio-economic – is essential to improve modelling capabilities and better inform forward-looking risk mitigation decisions.**\n\nLosses from secondary peril have been rising….\n\n...but the huge loss potential from primary peril strikes remains an ever-present danger.\n\n**Figure 11** Cumulative insured losses from primary and secondary perils since 1970, in USD billion at 2020 prices\n\nSource: Swiss Re Institute\n\n0\n\n200\n\n400\n\n600\n\n800\n\n1 000\n\n2020 2015 2010 2005 2000 1995 1990 1985 1980 1975 1970\n\nPrimary perils Secondary perils","pages":[{"line_numbers":[],"markdown":"18 **Swiss Re** _sigma_ No 1/2021\n\n# **Primary and secondary perils, and the way forward**\n\n### Don’t forget about primary-peril risks\n\nAccording to _sigma_ records, from 1970 total (primary + secondary peril) annual insured losses exceeded USD 80 billion in only six years, all since 2005. Losses were highest when large primary perils events occurred, the peak-loss years being 2017 (Hurricanes Harvey, Irma, Maria), 2011 (earthquakes in Japan and New Zealand, flood in Thailand), 2005 (Hurricanes Katrina, Wilma, Rita) and 2012 (Hurricane Sandy). There has been a growing contribution of secondary perils to total losses in off-peak years, notably in 2018 and 2020. More than 50% of all catastrophe-related insured losses in 2016–2020 were from secondary perils. In 2020 alone, the share of secondary-related was more than 70% of total losses.\n\nHowever, this does not mean secondary perils are becoming primary loss drivers. Over the past decades, the contribution of secondary perils to total annual insurance losses has fluctuated between 20–90%, without a clear trend in either direction. In years like 2020, secondary perils have been in the spotlight, but this is no reason to believe that peak primary-peril loss events will not occur again. Since 1970, losses from both peril groups have been accumulating more or less in sync (see Figure 11). And while secondary peril losses have accumulated faster in specific years, history shows that it can take just a few peak-event strikes to demonstrate the very large magnitude of losses that primary peril risk poses, as in the case of Hurricanes Katrina, Wilma and Rita in 2005.\n\n**Insured losses from primary and secondary perils have been on an upward trajectory since 1970. In the last decade, in years absent major losses from primary perils, accumulated losses from secondary perils alone have been high, 2020 being a case in point. The risk trends impacting both peril types remain the same, and rising losses from secondary perils are a warning that future peak-loss scenarios could also grow significantly. However, insurance industry risk knowledge of secondary perils lags that of primary perils. Better understanding of the complex interplay of factors influencing secondary peril event losses – natural and socio-economic – is essential to improve modelling capabilities and better inform forward-looking risk mitigation decisions.**\n\nLosses from secondary peril have been rising….\n\n...but the huge loss potential from primary peril strikes remains an ever-present danger.\n\n**Figure 11** Cumulative insured losses from primary and secondary perils since 1970, in USD billion at 2020 prices\n\nSource: Swiss Re Institute\n\n0\n\n200\n\n400\n\n600\n\n800\n\n1 000\n\n2020 2015 2010 2005 2000 1995 1990 1985 1980 1975 1970\n\nPrimary perils Secondary perils","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"18 **Swiss Re** _sigma_ No 1/2021\n\n# **Primary and secondary perils, and the way forward**\n\n### Don’t forget about primary-peril risks\n\nAccording to _sigma_ records, from 1970 total (primary + secondary peril) annual insured losses exceeded USD 80 billion in only six years, all since 2005. Losses were highest when large primary perils events occurred, the peak-loss years being 2017 (Hurricanes Harvey, Irma, Maria), 2011 (earthquakes in Japan and New Zealand, flood in Thailand), 2005 (Hurricanes Katrina, Wilma, Rita) and 2012 (Hurricane Sandy). There has been a growing contribution of secondary perils to total losses in off-peak years, notably in 2018 and 2020. More than 50% of all catastrophe-related insured losses in 2016–2020 were from secondary perils. In 2020 alone, the share of secondary-related was more than 70% of total losses.\n\nHowever, this does not mean secondary perils are becoming primary loss drivers. Over the past decades, the contribution of secondary perils to total annual insurance losses has fluctuated between 20–90%, without a clear trend in either direction. In years like 2020, secondary perils have been in the spotlight, but this is no reason to believe that peak primary-peril loss events will not occur again. Since 1970, losses from both peril groups have been accumulating more or less in sync (see Figure 11). And while secondary peril losses have accumulated faster in specific years, history shows that it can take just a few peak-event strikes to demonstrate the very large magnitude of losses that primary peril risk poses, as in the case of Hurricanes Katrina, Wilma and Rita in 2005.\n\n**Insured losses from primary and secondary perils have been on an upward trajectory since 1970. In the last decade, in years absent major losses from primary perils, accumulated losses from secondary perils alone have been high, 2020 being a case in point. The risk trends impacting both peril types remain the same, and rising losses from secondary perils are a warning that future peak-loss scenarios could also grow significantly. However, insurance industry risk knowledge of secondary perils lags that of primary perils. Better understanding of the complex interplay of factors influencing secondary peril event losses – natural and socio-economic – is essential to improve modelling capabilities and better inform forward-looking risk mitigation decisions.**\n\nLosses from secondary peril have been rising….\n\n...but the huge loss potential from primary peril strikes remains an ever-present danger.\n\n**Figure 11** Cumulative insured losses from primary and secondary perils since 1970, in USD billion at 2020 prices\n\nSource: Swiss Re Institute\n\n0\n\n200\n\n400\n\n600\n\n800\n\n1 000\n\n2020 2015 2010 2005 2000 1995 1990 1985 1980 1975 1970\n\nPrimary perils Secondary perils","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/sigma-1-2021-en_p20.pdf"]},"response":{"errors":[],"extract_id":"extract_9974e610b419ff733c1736b7c1f0bf3f","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["18 **Swiss Re** _sigma_ No 1/2021\n\n# **Primary and secondary perils, and the way forward**\n\n### Don’t forget about primary-peril risks\n\nAccording to _sigma_ records, from 1970 total (primary + secondary peril) annual insured losses exceeded USD 80 billion in only six years, all since 2005. Losses were highest when large primary perils events occurred, the peak-loss years being 2017 (Hurricanes Harvey, Irma, Maria), 2011 (earthquakes in Japan and New Zealand, flood in Thailand), 2005 (Hurricanes Katrina, Wilma, Rita) and 2012 (Hurricane Sandy). There has been a growing contribution of secondary perils to total losses in off-peak years, notably in 2018 and 2020. More than 50% of all catastrophe-related insured losses in 2016–2020 were from secondary perils. In 2020 alone, the share of secondary-related was more than 70% of total losses.\n\nHowever, this does not mean secondary perils are becoming primary loss drivers. Over the past decades, the contribution of secondary perils to total annual insurance losses has fluctuated between 20–90%, without a clear trend in either direction. In years like 2020, secondary perils have been in the spotlight, but this is no reason to believe that peak primary-peril loss events will not occur again. Since 1970, losses from both peril groups have been accumulating more or less in sync (see Figure 11). And while secondary peril losses have accumulated faster in specific years, history shows that it can take just a few peak-event strikes to demonstrate the very large magnitude of losses that primary peril risk poses, as in the case of Hurricanes Katrina, Wilma and Rita in 2005.\n\n**Insured losses from primary and secondary perils have been on an upward trajectory since 1970. In the last decade, in years absent major losses from primary perils, accumulated losses from secondary perils alone have been high, 2020 being a case in point. The risk trends impacting both peril types remain the same, and rising losses from secondary perils are a warning that future peak-loss scenarios could also grow significantly. However, insurance industry risk knowledge of secondary perils lags that of primary perils. Better understanding of the complex interplay of factors influencing secondary peril event losses – natural and socio-economic – is essential to improve modelling capabilities and better inform forward-looking risk mitigation decisions.**\n\nLosses from secondary peril have been rising….\n\n...but the huge loss potential from primary peril strikes remains an ever-present danger.\n\n**Figure 11** Cumulative insured losses from primary and secondary perils since 1970, in USD billion at 2020 prices\n\nSource: Swiss Re Institute\n\n0\n\n200\n\n400\n\n600\n\n800\n\n1 000\n\n2020 2015 2010 2005 2000 1995 1990 1985 1980 1975 1970\n\nPrimary perils Secondary perils"],"full_content":"18 **Swiss Re** _sigma_ No 1/2021\n\n# **Primary and secondary perils, and the way forward**\n\n### Don’t forget about primary-peril risks\n\nAccording to _sigma_ records, from 1970 total (primary + secondary peril) annual insured losses exceeded USD 80 billion in only six years, all since 2005. Losses were highest when large primary perils events occurred, the peak-loss years being 2017 (Hurricanes Harvey, Irma, Maria), 2011 (earthquakes in Japan and New Zealand, flood in Thailand), 2005 (Hurricanes Katrina, Wilma, Rita) and 2012 (Hurricane Sandy). There has been a growing contribution of secondary perils to total losses in off-peak years, notably in 2018 and 2020. More than 50% of all catastrophe-related insured losses in 2016–2020 were from secondary perils. In 2020 alone, the share of secondary-related was more than 70% of total losses.\n\nHowever, this does not mean secondary perils are becoming primary loss drivers. Over the past decades, the contribution of secondary perils to total annual insurance losses has fluctuated between 20–90%, without a clear trend in either direction. In years like 2020, secondary perils have been in the spotlight, but this is no reason to believe that peak primary-peril loss events will not occur again. Since 1970, losses from both peril groups have been accumulating more or less in sync (see Figure 11). And while secondary peril losses have accumulated faster in specific years, history shows that it can take just a few peak-event strikes to demonstrate the very large magnitude of losses that primary peril risk poses, as in the case of Hurricanes Katrina, Wilma and Rita in 2005.\n\n**Insured losses from primary and secondary perils have been on an upward trajectory since 1970. In the last decade, in years absent major losses from primary perils, accumulated losses from secondary perils alone have been high, 2020 being a case in point. The risk trends impacting both peril types remain the same, and rising losses from secondary perils are a warning that future peak-loss scenarios could also grow significantly. However, insurance industry risk knowledge of secondary perils lags that of primary perils. Better understanding of the complex interplay of factors influencing secondary peril event losses – natural and socio-economic – is essential to improve modelling capabilities and better inform forward-looking risk mitigation decisions.**\n\nLosses from secondary peril have been rising….\n\n...but the huge loss potential from primary peril strikes remains an ever-present danger.\n\n**Figure 11** Cumulative insured losses from primary and secondary perils since 1970, in USD billion at 2020 prices\n\nSource: Swiss Re Institute\n\n0\n\n200\n\n400\n\n600\n\n800\n\n1 000\n\n2020 2015 2010 2005 2000 1995 1990 1985 1980 1975 1970\n\nPrimary perils Secondary perils","publish_date":null,"title":"Primary and secondary perils, and the way forward","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/sigma-1-2021-en_p20.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/sigma-1-2021-en_p20.pdf"},"request":{"example_id":"chart/sigma-1-2021-en_p20","product_type":"parse","source_file_path":"sigma-1-2021-en_p20.pdf"},"started_at":"2026-11-04T19:45:35.069774"} +{"completed_at":"2026-11-04T19:45:37.995447","latency_in_ms":1712,"output":{"example_id":"chart/sigma-1-2021-en_p4","grounded_pages":[],"layout_pages":[],"markdown":"2 **Swiss Re** _sigma_ No 1/2021\n\n# **Key takeaways**\n\n### 2020 in a nutshell\n\n### Uninflated, inflated (2020 prices) and normalised economic losses from natural catastrophes, USD billion\n\nThe annual growth rate of normalised losses from natural catastrophes on a 10-year moving average basis between 1970 and 2020 was 1.3%. Normalisation adjusts to show that an event in the past, if it were to occur in today's world of higher levels of asset values, would cause more damage. This is due to the accumulation of human and economic value (physical assets) in the intervening years.\n\n0\n\n100\n\n200\n\n300\n\n400\n\n500\n\n600\n\n2020 2015 2010 2005 2000 1995 1990 1985 1980\n\nUninflated Inflated Normalised Economic losses:\n\nNote: normalised by GDP (country real GDP + US inflation); loss-data quality prior to 1990 poor. Source: Swiss Re Institute\n\n**Economic losses**\n\nTotal\n\nNatural catastophes **USD 190 billion**\n\n10-year average: 0.26%\n\n2019: USD 150 bn 10-year average: USD 222 bn\n\nMan-made **USD 12 billion**\n\n## **USD 202 billion**\n\n**0\\.24%** of global GDP\n\n2019\n\n**USD 87 bn USD 113 bn**\n\n**USD 143 bn**\n\n2020 10-year average\n\n**Insured losses**\n\nTotal\n\nNatural catastophes **USD 81 billion**\n\n10-year average: 4.7%\n\n2019: USD 22.5 bn\n\n2019: USD 31.9 bn\n\n2019: USD 63 bn 10-year average: USD 79 bn\n\nMan-made **USD 8 billion**\n\n## **USD 89 billion**\n\n**4\\.9%** of global property direct premiums written\n\n**Victims**\n\n## **7 993**\n\n**Catastrophe events**\n\n## **274**\n\n**Natural catastrophe insured losses**\n\nTotal\n\nSecondary peril **USD 57.4 billion** (71%)\n\n2019: USD 54 bn 10-year average: USD 74 bn\n\nPrimary peril **USD 23.2 billion** (29%)\n\n## **USD 81 billion**\n\n**Global protection gap**\n\nSource: Swiss Re Institute","pages":[{"line_numbers":[],"markdown":"2 **Swiss Re** _sigma_ No 1/2021\n\n# **Key takeaways**\n\n### 2020 in a nutshell\n\n### Uninflated, inflated (2020 prices) and normalised economic losses from natural catastrophes, USD billion\n\nThe annual growth rate of normalised losses from natural catastrophes on a 10-year moving average basis between 1970 and 2020 was 1.3%. Normalisation adjusts to show that an event in the past, if it were to occur in today's world of higher levels of asset values, would cause more damage. This is due to the accumulation of human and economic value (physical assets) in the intervening years.\n\n0\n\n100\n\n200\n\n300\n\n400\n\n500\n\n600\n\n2020 2015 2010 2005 2000 1995 1990 1985 1980\n\nUninflated Inflated Normalised Economic losses:\n\nNote: normalised by GDP (country real GDP + US inflation); loss-data quality prior to 1990 poor. Source: Swiss Re Institute\n\n**Economic losses**\n\nTotal\n\nNatural catastophes **USD 190 billion**\n\n10-year average: 0.26%\n\n2019: USD 150 bn 10-year average: USD 222 bn\n\nMan-made **USD 12 billion**\n\n## **USD 202 billion**\n\n**0\\.24%** of global GDP\n\n2019\n\n**USD 87 bn USD 113 bn**\n\n**USD 143 bn**\n\n2020 10-year average\n\n**Insured losses**\n\nTotal\n\nNatural catastophes **USD 81 billion**\n\n10-year average: 4.7%\n\n2019: USD 22.5 bn\n\n2019: USD 31.9 bn\n\n2019: USD 63 bn 10-year average: USD 79 bn\n\nMan-made **USD 8 billion**\n\n## **USD 89 billion**\n\n**4\\.9%** of global property direct premiums written\n\n**Victims**\n\n## **7 993**\n\n**Catastrophe events**\n\n## **274**\n\n**Natural catastrophe insured losses**\n\nTotal\n\nSecondary peril **USD 57.4 billion** (71%)\n\n2019: USD 54 bn 10-year average: USD 74 bn\n\nPrimary peril **USD 23.2 billion** (29%)\n\n## **USD 81 billion**\n\n**Global protection gap**\n\nSource: Swiss Re Institute","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"2 **Swiss Re** _sigma_ No 1/2021\n\n# **Key takeaways**\n\n### 2020 in a nutshell\n\n### Uninflated, inflated (2020 prices) and normalised economic losses from natural catastrophes, USD billion\n\nThe annual growth rate of normalised losses from natural catastrophes on a 10-year moving average basis between 1970 and 2020 was 1.3%. Normalisation adjusts to show that an event in the past, if it were to occur in today's world of higher levels of asset values, would cause more damage. This is due to the accumulation of human and economic value (physical assets) in the intervening years.\n\n0\n\n100\n\n200\n\n300\n\n400\n\n500\n\n600\n\n2020 2015 2010 2005 2000 1995 1990 1985 1980\n\nUninflated Inflated Normalised Economic losses:\n\nNote: normalised by GDP (country real GDP + US inflation); loss-data quality prior to 1990 poor. Source: Swiss Re Institute\n\n**Economic losses**\n\nTotal\n\nNatural catastophes **USD 190 billion**\n\n10-year average: 0.26%\n\n2019: USD 150 bn 10-year average: USD 222 bn\n\nMan-made **USD 12 billion**\n\n## **USD 202 billion**\n\n**0\\.24%** of global GDP\n\n2019\n\n**USD 87 bn USD 113 bn**\n\n**USD 143 bn**\n\n2020 10-year average\n\n**Insured losses**\n\nTotal\n\nNatural catastophes **USD 81 billion**\n\n10-year average: 4.7%\n\n2019: USD 22.5 bn\n\n2019: USD 31.9 bn\n\n2019: USD 63 bn 10-year average: USD 79 bn\n\nMan-made **USD 8 billion**\n\n## **USD 89 billion**\n\n**4\\.9%** of global property direct premiums written\n\n**Victims**\n\n## **7 993**\n\n**Catastrophe events**\n\n## **274**\n\n**Natural catastrophe insured losses**\n\nTotal\n\nSecondary peril **USD 57.4 billion** (71%)\n\n2019: USD 54 bn 10-year average: USD 74 bn\n\nPrimary peril **USD 23.2 billion** (29%)\n\n## **USD 81 billion**\n\n**Global protection gap**\n\nSource: Swiss Re Institute","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/sigma-1-2021-en_p4.pdf"]},"response":{"errors":[],"extract_id":"extract_3f5ea124281d6867de166e784ca196c2","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["2 **Swiss Re** _sigma_ No 1/2021\n\n# **Key takeaways**\n\n### 2020 in a nutshell\n\n### Uninflated, inflated (2020 prices) and normalised economic losses from natural catastrophes, USD billion\n\nThe annual growth rate of normalised losses from natural catastrophes on a 10-year moving average basis between 1970 and 2020 was 1.3%. Normalisation adjusts to show that an event in the past, if it were to occur in today's world of higher levels of asset values, would cause more damage. This is due to the accumulation of human and economic value (physical assets) in the intervening years.\n\n0\n\n100\n\n200\n\n300\n\n400\n\n500\n\n600\n\n2020 2015 2010 2005 2000 1995 1990 1985 1980\n\nUninflated Inflated Normalised Economic losses:\n\nNote: normalised by GDP (country real GDP + US inflation); loss-data quality prior to 1990 poor. Source: Swiss Re Institute\n\n**Economic losses**\n\nTotal\n\nNatural catastophes **USD 190 billion**\n\n10-year average: 0.26%\n\n2019: USD 150 bn 10-year average: USD 222 bn\n\nMan-made **USD 12 billion**\n\n## **USD 202 billion**\n\n**0\\.24%** of global GDP\n\n2019\n\n**USD 87 bn USD 113 bn**\n\n**USD 143 bn**\n\n2020 10-year average\n\n**Insured losses**\n\nTotal\n\nNatural catastophes **USD 81 billion**\n\n10-year average: 4.7%\n\n2019: USD 22.5 bn\n\n2019: USD 31.9 bn\n\n2019: USD 63 bn 10-year average: USD 79 bn\n\nMan-made **USD 8 billion**\n\n## **USD 89 billion**\n\n**4\\.9%** of global property direct premiums written\n\n**Victims**\n\n## **7 993**\n\n**Catastrophe events**\n\n## **274**\n\n**Natural catastrophe insured losses**\n\nTotal\n\nSecondary peril **USD 57.4 billion** (71%)\n\n2019: USD 54 bn 10-year average: USD 74 bn\n\nPrimary peril **USD 23.2 billion** (29%)\n\n## **USD 81 billion**\n\n**Global protection gap**\n\nSource: Swiss Re Institute"],"full_content":"2 **Swiss Re** _sigma_ No 1/2021\n\n# **Key takeaways**\n\n### 2020 in a nutshell\n\n### Uninflated, inflated (2020 prices) and normalised economic losses from natural catastrophes, USD billion\n\nThe annual growth rate of normalised losses from natural catastrophes on a 10-year moving average basis between 1970 and 2020 was 1.3%. Normalisation adjusts to show that an event in the past, if it were to occur in today's world of higher levels of asset values, would cause more damage. This is due to the accumulation of human and economic value (physical assets) in the intervening years.\n\n0\n\n100\n\n200\n\n300\n\n400\n\n500\n\n600\n\n2020 2015 2010 2005 2000 1995 1990 1985 1980\n\nUninflated Inflated Normalised Economic losses:\n\nNote: normalised by GDP (country real GDP + US inflation); loss-data quality prior to 1990 poor. Source: Swiss Re Institute\n\n**Economic losses**\n\nTotal\n\nNatural catastophes **USD 190 billion**\n\n10-year average: 0.26%\n\n2019: USD 150 bn 10-year average: USD 222 bn\n\nMan-made **USD 12 billion**\n\n## **USD 202 billion**\n\n**0\\.24%** of global GDP\n\n2019\n\n**USD 87 bn USD 113 bn**\n\n**USD 143 bn**\n\n2020 10-year average\n\n**Insured losses**\n\nTotal\n\nNatural catastophes **USD 81 billion**\n\n10-year average: 4.7%\n\n2019: USD 22.5 bn\n\n2019: USD 31.9 bn\n\n2019: USD 63 bn 10-year average: USD 79 bn\n\nMan-made **USD 8 billion**\n\n## **USD 89 billion**\n\n**4\\.9%** of global property direct premiums written\n\n**Victims**\n\n## **7 993**\n\n**Catastrophe events**\n\n## **274**\n\n**Natural catastrophe insured losses**\n\nTotal\n\nSecondary peril **USD 57.4 billion** (71%)\n\n2019: USD 54 bn 10-year average: USD 74 bn\n\nPrimary peril **USD 23.2 billion** (29%)\n\n## **USD 81 billion**\n\n**Global protection gap**\n\nSource: Swiss Re Institute","publish_date":null,"title":"Key takeaways","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/sigma-1-2021-en_p4.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/sigma-1-2021-en_p4.pdf"},"request":{"example_id":"chart/sigma-1-2021-en_p4","product_type":"parse","source_file_path":"sigma-1-2021-en_p4.pdf"},"started_at":"2026-11-04T19:45:36.282682"} +{"completed_at":"2026-11-04T19:45:38.913864","latency_in_ms":1201,"output":{"example_id":"chart/sigma-1-2021-en_p6","grounded_pages":[],"layout_pages":[],"markdown":"4 **Swiss Re** _sigma_ No 1/2021\n\n**Last year’s loss experience reaffirmed the significant threat presented by secondary perils. Secondary perils caused more than 70% of insured losses from all natural catastrophes (USD 81 billion) in 2020.** **1** **Accumulated losses from many small- and mid-size events made it the fifth-costliest year on record in insured loss terms. The main drivers were SCS and wildfires in the US and Australia. In all regions where extreme weather events occurred, the experience was shaped by the complex interplay of natural-world phenomena, including climate change, and socio-economic trends.**\n\n### Secondary and primary peril loss developments\n\nThe term _primary perils_ references catastrophes of larger scale, notably tropical cyclones, earthquakes and European winter storms. Events tend to be less frequent, but resulting losses can be extreme. We use _secondary perils_ as an umbrella term for natural catastrophes that typically generate losses of low to medium magnitude, but that can happen relatively frequently. These include “independent secondary perils” such as SCS (including thunderstorms, hail and tornadoes), drought, wildfire, snow, flash floods and landslides. Primary perils can also trigger causally-linked consequential losses, for instance from heavy rains in the wake of a tropical cyclone, a storm surge induced by a winter storm, or fire outbreak after an earthquake. We consider these to be “secondary effects of primary perils” that also fall into the secondary perils bucket. Losses from primary and secondary perils have both risen continuously since 1970 (see Figure 1).\n\nThe largest annual insured loss total from independent secondary peril events – USD 62 billion – occurred in 2011, the year of devastating floods in Thailand and high-loss tornado outbreaks in the US. More recently, three of the four costliest secondary perils years occurred during 2017–2020. Primary perils are well- monitored risks and the re/insurance industry’s modelling capabilities of these exposures are strong. Their secondary effects, however, are not always fully captured. Likewise, independent secondary events are often also not fully monitored nor modelled. Given the rise of their associated losses in absolute terms, secondary perils need to be better understood for the purpose of more complete risk assessment and sustainable underwriting of natural catastrophe risks overall.\n\n1 Unless otherwise stated in the text, all loss numbers in the report as inflation-adjusted.\n\nSecondary perils are high frequency, low- to medium-severity events.\n\n**Figure 1** Annual primary and secondary peril insured losses since 1970 (USD billion) at 2020 prices, and share of secondary peril in total insured losses (five-years accumulated)\n\nSource: Swiss Re Institute\n\n0\n\n25\n\n50\n\n75\n\n100\n\n125\n\n150\n\n175\n\n2020 2015 2010 2005 2000 1995 1990 1985 1980 1975 1970\n\n0%\n\n20%\n\n40%\n\n60%\n\n80%\n\n100%\n\nInsured losses\n\nSecondary perils share\n\nIndependent secondary perils 5-years accumulated Primary perils Secondary perils share: Secondary effects of primary perils Insured losses:\n\nPrimary perils are better understood than secondary peril risks.\n\n# **Exploring the secondary perils universe**","pages":[{"line_numbers":[],"markdown":"4 **Swiss Re** _sigma_ No 1/2021\n\n**Last year’s loss experience reaffirmed the significant threat presented by secondary perils. Secondary perils caused more than 70% of insured losses from all natural catastrophes (USD 81 billion) in 2020.** **1** **Accumulated losses from many small- and mid-size events made it the fifth-costliest year on record in insured loss terms. The main drivers were SCS and wildfires in the US and Australia. In all regions where extreme weather events occurred, the experience was shaped by the complex interplay of natural-world phenomena, including climate change, and socio-economic trends.**\n\n### Secondary and primary peril loss developments\n\nThe term _primary perils_ references catastrophes of larger scale, notably tropical cyclones, earthquakes and European winter storms. Events tend to be less frequent, but resulting losses can be extreme. We use _secondary perils_ as an umbrella term for natural catastrophes that typically generate losses of low to medium magnitude, but that can happen relatively frequently. These include “independent secondary perils” such as SCS (including thunderstorms, hail and tornadoes), drought, wildfire, snow, flash floods and landslides. Primary perils can also trigger causally-linked consequential losses, for instance from heavy rains in the wake of a tropical cyclone, a storm surge induced by a winter storm, or fire outbreak after an earthquake. We consider these to be “secondary effects of primary perils” that also fall into the secondary perils bucket. Losses from primary and secondary perils have both risen continuously since 1970 (see Figure 1).\n\nThe largest annual insured loss total from independent secondary peril events – USD 62 billion – occurred in 2011, the year of devastating floods in Thailand and high-loss tornado outbreaks in the US. More recently, three of the four costliest secondary perils years occurred during 2017–2020. Primary perils are well- monitored risks and the re/insurance industry’s modelling capabilities of these exposures are strong. Their secondary effects, however, are not always fully captured. Likewise, independent secondary events are often also not fully monitored nor modelled. Given the rise of their associated losses in absolute terms, secondary perils need to be better understood for the purpose of more complete risk assessment and sustainable underwriting of natural catastrophe risks overall.\n\n1 Unless otherwise stated in the text, all loss numbers in the report as inflation-adjusted.\n\nSecondary perils are high frequency, low- to medium-severity events.\n\n**Figure 1** Annual primary and secondary peril insured losses since 1970 (USD billion) at 2020 prices, and share of secondary peril in total insured losses (five-years accumulated)\n\nSource: Swiss Re Institute\n\n0\n\n25\n\n50\n\n75\n\n100\n\n125\n\n150\n\n175\n\n2020 2015 2010 2005 2000 1995 1990 1985 1980 1975 1970\n\n0%\n\n20%\n\n40%\n\n60%\n\n80%\n\n100%\n\nInsured losses\n\nSecondary perils share\n\nIndependent secondary perils 5-years accumulated Primary perils Secondary perils share: Secondary effects of primary perils Insured losses:\n\nPrimary perils are better understood than secondary peril risks.\n\n# **Exploring the secondary perils universe**","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"4 **Swiss Re** _sigma_ No 1/2021\n\n**Last year’s loss experience reaffirmed the significant threat presented by secondary perils. Secondary perils caused more than 70% of insured losses from all natural catastrophes (USD 81 billion) in 2020.** **1** **Accumulated losses from many small- and mid-size events made it the fifth-costliest year on record in insured loss terms. The main drivers were SCS and wildfires in the US and Australia. In all regions where extreme weather events occurred, the experience was shaped by the complex interplay of natural-world phenomena, including climate change, and socio-economic trends.**\n\n### Secondary and primary peril loss developments\n\nThe term _primary perils_ references catastrophes of larger scale, notably tropical cyclones, earthquakes and European winter storms. Events tend to be less frequent, but resulting losses can be extreme. We use _secondary perils_ as an umbrella term for natural catastrophes that typically generate losses of low to medium magnitude, but that can happen relatively frequently. These include “independent secondary perils” such as SCS (including thunderstorms, hail and tornadoes), drought, wildfire, snow, flash floods and landslides. Primary perils can also trigger causally-linked consequential losses, for instance from heavy rains in the wake of a tropical cyclone, a storm surge induced by a winter storm, or fire outbreak after an earthquake. We consider these to be “secondary effects of primary perils” that also fall into the secondary perils bucket. Losses from primary and secondary perils have both risen continuously since 1970 (see Figure 1).\n\nThe largest annual insured loss total from independent secondary peril events – USD 62 billion – occurred in 2011, the year of devastating floods in Thailand and high-loss tornado outbreaks in the US. More recently, three of the four costliest secondary perils years occurred during 2017–2020. Primary perils are well- monitored risks and the re/insurance industry’s modelling capabilities of these exposures are strong. Their secondary effects, however, are not always fully captured. Likewise, independent secondary events are often also not fully monitored nor modelled. Given the rise of their associated losses in absolute terms, secondary perils need to be better understood for the purpose of more complete risk assessment and sustainable underwriting of natural catastrophe risks overall.\n\n1 Unless otherwise stated in the text, all loss numbers in the report as inflation-adjusted.\n\nSecondary perils are high frequency, low- to medium-severity events.\n\n**Figure 1** Annual primary and secondary peril insured losses since 1970 (USD billion) at 2020 prices, and share of secondary peril in total insured losses (five-years accumulated)\n\nSource: Swiss Re Institute\n\n0\n\n25\n\n50\n\n75\n\n100\n\n125\n\n150\n\n175\n\n2020 2015 2010 2005 2000 1995 1990 1985 1980 1975 1970\n\n0%\n\n20%\n\n40%\n\n60%\n\n80%\n\n100%\n\nInsured losses\n\nSecondary perils share\n\nIndependent secondary perils 5-years accumulated Primary perils Secondary perils share: Secondary effects of primary perils Insured losses:\n\nPrimary perils are better understood than secondary peril risks.\n\n# **Exploring the secondary perils universe**","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/sigma-1-2021-en_p6.pdf"]},"response":{"errors":[],"extract_id":"extract_302b86030db00042a37a5afbd1384ec5","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["4 **Swiss Re** _sigma_ No 1/2021\n\n**Last year’s loss experience reaffirmed the significant threat presented by secondary perils. Secondary perils caused more than 70% of insured losses from all natural catastrophes (USD 81 billion) in 2020.** **1** **Accumulated losses from many small- and mid-size events made it the fifth-costliest year on record in insured loss terms. The main drivers were SCS and wildfires in the US and Australia. In all regions where extreme weather events occurred, the experience was shaped by the complex interplay of natural-world phenomena, including climate change, and socio-economic trends.**\n\n### Secondary and primary peril loss developments\n\nThe term _primary perils_ references catastrophes of larger scale, notably tropical cyclones, earthquakes and European winter storms. Events tend to be less frequent, but resulting losses can be extreme. We use _secondary perils_ as an umbrella term for natural catastrophes that typically generate losses of low to medium magnitude, but that can happen relatively frequently. These include “independent secondary perils” such as SCS (including thunderstorms, hail and tornadoes), drought, wildfire, snow, flash floods and landslides. Primary perils can also trigger causally-linked consequential losses, for instance from heavy rains in the wake of a tropical cyclone, a storm surge induced by a winter storm, or fire outbreak after an earthquake. We consider these to be “secondary effects of primary perils” that also fall into the secondary perils bucket. Losses from primary and secondary perils have both risen continuously since 1970 (see Figure 1).\n\nThe largest annual insured loss total from independent secondary peril events – USD 62 billion – occurred in 2011, the year of devastating floods in Thailand and high-loss tornado outbreaks in the US. More recently, three of the four costliest secondary perils years occurred during 2017–2020. Primary perils are well- monitored risks and the re/insurance industry’s modelling capabilities of these exposures are strong. Their secondary effects, however, are not always fully captured. Likewise, independent secondary events are often also not fully monitored nor modelled. Given the rise of their associated losses in absolute terms, secondary perils need to be better understood for the purpose of more complete risk assessment and sustainable underwriting of natural catastrophe risks overall.\n\n1 Unless otherwise stated in the text, all loss numbers in the report as inflation-adjusted.\n\nSecondary perils are high frequency, low- to medium-severity events.\n\n**Figure 1** Annual primary and secondary peril insured losses since 1970 (USD billion) at 2020 prices, and share of secondary peril in total insured losses (five-years accumulated)\n\nSource: Swiss Re Institute\n\n0\n\n25\n\n50\n\n75\n\n100\n\n125\n\n150\n\n175\n\n2020 2015 2010 2005 2000 1995 1990 1985 1980 1975 1970\n\n0%\n\n20%\n\n40%\n\n60%\n\n80%\n\n100%\n\nInsured losses\n\nSecondary perils share\n\nIndependent secondary perils 5-years accumulated Primary perils Secondary perils share: Secondary effects of primary perils Insured losses:\n\nPrimary perils are better understood than secondary peril risks.\n\n# **Exploring the secondary perils universe**"],"full_content":"4 **Swiss Re** _sigma_ No 1/2021\n\n**Last year’s loss experience reaffirmed the significant threat presented by secondary perils. Secondary perils caused more than 70% of insured losses from all natural catastrophes (USD 81 billion) in 2020.** **1** **Accumulated losses from many small- and mid-size events made it the fifth-costliest year on record in insured loss terms. The main drivers were SCS and wildfires in the US and Australia. In all regions where extreme weather events occurred, the experience was shaped by the complex interplay of natural-world phenomena, including climate change, and socio-economic trends.**\n\n### Secondary and primary peril loss developments\n\nThe term _primary perils_ references catastrophes of larger scale, notably tropical cyclones, earthquakes and European winter storms. Events tend to be less frequent, but resulting losses can be extreme. We use _secondary perils_ as an umbrella term for natural catastrophes that typically generate losses of low to medium magnitude, but that can happen relatively frequently. These include “independent secondary perils” such as SCS (including thunderstorms, hail and tornadoes), drought, wildfire, snow, flash floods and landslides. Primary perils can also trigger causally-linked consequential losses, for instance from heavy rains in the wake of a tropical cyclone, a storm surge induced by a winter storm, or fire outbreak after an earthquake. We consider these to be “secondary effects of primary perils” that also fall into the secondary perils bucket. Losses from primary and secondary perils have both risen continuously since 1970 (see Figure 1).\n\nThe largest annual insured loss total from independent secondary peril events – USD 62 billion – occurred in 2011, the year of devastating floods in Thailand and high-loss tornado outbreaks in the US. More recently, three of the four costliest secondary perils years occurred during 2017–2020. Primary perils are well- monitored risks and the re/insurance industry’s modelling capabilities of these exposures are strong. Their secondary effects, however, are not always fully captured. Likewise, independent secondary events are often also not fully monitored nor modelled. Given the rise of their associated losses in absolute terms, secondary perils need to be better understood for the purpose of more complete risk assessment and sustainable underwriting of natural catastrophe risks overall.\n\n1 Unless otherwise stated in the text, all loss numbers in the report as inflation-adjusted.\n\nSecondary perils are high frequency, low- to medium-severity events.\n\n**Figure 1** Annual primary and secondary peril insured losses since 1970 (USD billion) at 2020 prices, and share of secondary peril in total insured losses (five-years accumulated)\n\nSource: Swiss Re Institute\n\n0\n\n25\n\n50\n\n75\n\n100\n\n125\n\n150\n\n175\n\n2020 2015 2010 2005 2000 1995 1990 1985 1980 1975 1970\n\n0%\n\n20%\n\n40%\n\n60%\n\n80%\n\n100%\n\nInsured losses\n\nSecondary perils share\n\nIndependent secondary perils 5-years accumulated Primary perils Secondary perils share: Secondary effects of primary perils Insured losses:\n\nPrimary perils are better understood than secondary peril risks.\n\n# **Exploring the secondary perils universe**","publish_date":null,"title":"Exploring the secondary perils universe","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/sigma-1-2021-en_p6.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/sigma-1-2021-en_p6.pdf"},"request":{"example_id":"chart/sigma-1-2021-en_p6","product_type":"parse","source_file_path":"sigma-1-2021-en_p6.pdf"},"started_at":"2026-11-04T19:45:37.712774"} +{"completed_at":"2026-11-04T19:44:48.606104","latency_in_ms":2457,"output":{"example_id":"chart/sigma-1-2021-en_p9","grounded_pages":[],"layout_pages":[],"markdown":"**Swiss Re** _sigma_ No 1/2021 7\n\n_Implications for insurance_ Climate conditions help explain the physics leading to the weather event sequence. From the insurance perspective, important questions are: was the 2019/20 weather story in Australia a consequence of climate change, and was it an exceptional year? These questions cannot be answered with a definitive yes or no, but do raise important considerations for risk assessment and the pricing assumptions for standard natural catastrophe protection insurance covers.\n\nThere have been previous occurrences in Australia’s climate history of multiple extreme events coming together in clusters in highly variable and strong conditions (eg, in 2011). Severe weather events in Australia remain predominantly influenced by climate drivers like IOD, SAM and ENSO in the short term. As such, it is not easy to attribute a specific event or even an intensive natural catastrophe season to climate change. That said, long-term trends in risk factors like fire-inducing dry weather or extreme rainfall are very likely influenced by the rise in global temperatures. ⁶ The spring-summer of 2019/20 was an extreme season with several climate signals feeding off each other. And, with climate change set to continue if no mitigating actions are taken, such seasons are less likely to be exceptional.\n\nSecondary perils account for roughly 75% of insured losses in Australia. After a benign 2000–09, the last decade has seen an increased number of severe insurance losses, at AUD 1 billion or higher, many times in a single year across all secondary perils. ⁷ Short-term climate signals and the effects of climate change alone do not explain this increase. Climate change loss-impacts are being compounded by increasing exposure and expanding cities in coastal areas and in the wilderness. What may previously been a small or even unnoticed natural disturbance can today lead to significant insured losses.\n\nThe socioeconomic factors, and short- and long-term climate drivers are here to stay, warranting careful reconsideration of commercially sustainable insurance offerings. Risk assessment needs to account for frequency and severity of individual perils as in a traditional natural catastrophe modelling. However, it also needs to explicitly consider the compounding effects of climate variability. These can trigger multiple and diverse secondary perils in the same season, adding significant annual variability to aggregate losses in underwriting. In view of dynamic nature of risk, taking a forward-looking stance versus averaging a multi-decadal past, is paramount.\n\n6 _Severe weather in a changing climate – 2_ _nd_ _ed_ , IAG, National Center for Atmospheric Research, 2020. 7 Insurance Council of Australia.\n\n**Figure 3** Timeline of events against trends of various climate drivers\n\nSources: Bureau of Meteorology (for IOD and ENSO data), British Antarctic Survey (for SAM data), Insurance Council of Australia (for loss numbers), Swiss Re Institute\n\n–5 –4 –3 –2 –1 0 1 2 3 4 5 6 7\n\nMay 2020 Jan 2020 Sep 2019 May 2019 Jan 2019\n\n–1.5\n\n–1.0\n\n–0.5\n\n0\\.0\n\n0\\.5\n\n1\\.0\n\n1\\.5\n\n2\\.0\n\n2\\.5\n\nSAM\n\nENSO/IOD\n\nSAM ENSO IOD Climate index: Jun 2019–Mar 2020\n\nBushfires: AUD 2.3 bn\n\nJan 20, 2020\n\nACT Hail: AUD 1.66 bn\n\nFeb 7–10, 2020\n\nEast Coast Flood: AUD 0.97 bn\n\nEvents:\n\nThe 2019/20 spring-summer season in Australia brought many severe weather events.\n\nClimate change cannot be cited as the driving cause.\n\nSocio-economic factors also play a significant role.\n\nRisk assessment should take a forward-looking stance of the multiple factor influencing losses.","pages":[{"line_numbers":[],"markdown":"**Swiss Re** _sigma_ No 1/2021 7\n\n_Implications for insurance_ Climate conditions help explain the physics leading to the weather event sequence. From the insurance perspective, important questions are: was the 2019/20 weather story in Australia a consequence of climate change, and was it an exceptional year? These questions cannot be answered with a definitive yes or no, but do raise important considerations for risk assessment and the pricing assumptions for standard natural catastrophe protection insurance covers.\n\nThere have been previous occurrences in Australia’s climate history of multiple extreme events coming together in clusters in highly variable and strong conditions (eg, in 2011). Severe weather events in Australia remain predominantly influenced by climate drivers like IOD, SAM and ENSO in the short term. As such, it is not easy to attribute a specific event or even an intensive natural catastrophe season to climate change. That said, long-term trends in risk factors like fire-inducing dry weather or extreme rainfall are very likely influenced by the rise in global temperatures. ⁶ The spring-summer of 2019/20 was an extreme season with several climate signals feeding off each other. And, with climate change set to continue if no mitigating actions are taken, such seasons are less likely to be exceptional.\n\nSecondary perils account for roughly 75% of insured losses in Australia. After a benign 2000–09, the last decade has seen an increased number of severe insurance losses, at AUD 1 billion or higher, many times in a single year across all secondary perils. ⁷ Short-term climate signals and the effects of climate change alone do not explain this increase. Climate change loss-impacts are being compounded by increasing exposure and expanding cities in coastal areas and in the wilderness. What may previously been a small or even unnoticed natural disturbance can today lead to significant insured losses.\n\nThe socioeconomic factors, and short- and long-term climate drivers are here to stay, warranting careful reconsideration of commercially sustainable insurance offerings. Risk assessment needs to account for frequency and severity of individual perils as in a traditional natural catastrophe modelling. However, it also needs to explicitly consider the compounding effects of climate variability. These can trigger multiple and diverse secondary perils in the same season, adding significant annual variability to aggregate losses in underwriting. In view of dynamic nature of risk, taking a forward-looking stance versus averaging a multi-decadal past, is paramount.\n\n6 _Severe weather in a changing climate – 2_ _nd_ _ed_ , IAG, National Center for Atmospheric Research, 2020. 7 Insurance Council of Australia.\n\n**Figure 3** Timeline of events against trends of various climate drivers\n\nSources: Bureau of Meteorology (for IOD and ENSO data), British Antarctic Survey (for SAM data), Insurance Council of Australia (for loss numbers), Swiss Re Institute\n\n–5 –4 –3 –2 –1 0 1 2 3 4 5 6 7\n\nMay 2020 Jan 2020 Sep 2019 May 2019 Jan 2019\n\n–1.5\n\n–1.0\n\n–0.5\n\n0\\.0\n\n0\\.5\n\n1\\.0\n\n1\\.5\n\n2\\.0\n\n2\\.5\n\nSAM\n\nENSO/IOD\n\nSAM ENSO IOD Climate index: Jun 2019–Mar 2020\n\nBushfires: AUD 2.3 bn\n\nJan 20, 2020\n\nACT Hail: AUD 1.66 bn\n\nFeb 7–10, 2020\n\nEast Coast Flood: AUD 0.97 bn\n\nEvents:\n\nThe 2019/20 spring-summer season in Australia brought many severe weather events.\n\nClimate change cannot be cited as the driving cause.\n\nSocio-economic factors also play a significant role.\n\nRisk assessment should take a forward-looking stance of the multiple factor influencing losses.","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**Swiss Re** _sigma_ No 1/2021 7\n\n_Implications for insurance_ Climate conditions help explain the physics leading to the weather event sequence. From the insurance perspective, important questions are: was the 2019/20 weather story in Australia a consequence of climate change, and was it an exceptional year? These questions cannot be answered with a definitive yes or no, but do raise important considerations for risk assessment and the pricing assumptions for standard natural catastrophe protection insurance covers.\n\nThere have been previous occurrences in Australia’s climate history of multiple extreme events coming together in clusters in highly variable and strong conditions (eg, in 2011). Severe weather events in Australia remain predominantly influenced by climate drivers like IOD, SAM and ENSO in the short term. As such, it is not easy to attribute a specific event or even an intensive natural catastrophe season to climate change. That said, long-term trends in risk factors like fire-inducing dry weather or extreme rainfall are very likely influenced by the rise in global temperatures. ⁶ The spring-summer of 2019/20 was an extreme season with several climate signals feeding off each other. And, with climate change set to continue if no mitigating actions are taken, such seasons are less likely to be exceptional.\n\nSecondary perils account for roughly 75% of insured losses in Australia. After a benign 2000–09, the last decade has seen an increased number of severe insurance losses, at AUD 1 billion or higher, many times in a single year across all secondary perils. ⁷ Short-term climate signals and the effects of climate change alone do not explain this increase. Climate change loss-impacts are being compounded by increasing exposure and expanding cities in coastal areas and in the wilderness. What may previously been a small or even unnoticed natural disturbance can today lead to significant insured losses.\n\nThe socioeconomic factors, and short- and long-term climate drivers are here to stay, warranting careful reconsideration of commercially sustainable insurance offerings. Risk assessment needs to account for frequency and severity of individual perils as in a traditional natural catastrophe modelling. However, it also needs to explicitly consider the compounding effects of climate variability. These can trigger multiple and diverse secondary perils in the same season, adding significant annual variability to aggregate losses in underwriting. In view of dynamic nature of risk, taking a forward-looking stance versus averaging a multi-decadal past, is paramount.\n\n6 _Severe weather in a changing climate – 2_ _nd_ _ed_ , IAG, National Center for Atmospheric Research, 2020. 7 Insurance Council of Australia.\n\n**Figure 3** Timeline of events against trends of various climate drivers\n\nSources: Bureau of Meteorology (for IOD and ENSO data), British Antarctic Survey (for SAM data), Insurance Council of Australia (for loss numbers), Swiss Re Institute\n\n–5 –4 –3 –2 –1 0 1 2 3 4 5 6 7\n\nMay 2020 Jan 2020 Sep 2019 May 2019 Jan 2019\n\n–1.5\n\n–1.0\n\n–0.5\n\n0\\.0\n\n0\\.5\n\n1\\.0\n\n1\\.5\n\n2\\.0\n\n2\\.5\n\nSAM\n\nENSO/IOD\n\nSAM ENSO IOD Climate index: Jun 2019–Mar 2020\n\nBushfires: AUD 2.3 bn\n\nJan 20, 2020\n\nACT Hail: AUD 1.66 bn\n\nFeb 7–10, 2020\n\nEast Coast Flood: AUD 0.97 bn\n\nEvents:\n\nThe 2019/20 spring-summer season in Australia brought many severe weather events.\n\nClimate change cannot be cited as the driving cause.\n\nSocio-economic factors also play a significant role.\n\nRisk assessment should take a forward-looking stance of the multiple factor influencing losses.","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/sigma-1-2021-en_p9.pdf"]},"response":{"errors":[],"extract_id":"extract_3516ffe134b0557213ebccb4a73ff584","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**Swiss Re** _sigma_ No 1/2021 7\n\n_Implications for insurance_ Climate conditions help explain the physics leading to the weather event sequence. From the insurance perspective, important questions are: was the 2019/20 weather story in Australia a consequence of climate change, and was it an exceptional year? These questions cannot be answered with a definitive yes or no, but do raise important considerations for risk assessment and the pricing assumptions for standard natural catastrophe protection insurance covers.\n\nThere have been previous occurrences in Australia’s climate history of multiple extreme events coming together in clusters in highly variable and strong conditions (eg, in 2011). Severe weather events in Australia remain predominantly influenced by climate drivers like IOD, SAM and ENSO in the short term. As such, it is not easy to attribute a specific event or even an intensive natural catastrophe season to climate change. That said, long-term trends in risk factors like fire-inducing dry weather or extreme rainfall are very likely influenced by the rise in global temperatures. ⁶ The spring-summer of 2019/20 was an extreme season with several climate signals feeding off each other. And, with climate change set to continue if no mitigating actions are taken, such seasons are less likely to be exceptional.\n\nSecondary perils account for roughly 75% of insured losses in Australia. After a benign 2000–09, the last decade has seen an increased number of severe insurance losses, at AUD 1 billion or higher, many times in a single year across all secondary perils. ⁷ Short-term climate signals and the effects of climate change alone do not explain this increase. Climate change loss-impacts are being compounded by increasing exposure and expanding cities in coastal areas and in the wilderness. What may previously been a small or even unnoticed natural disturbance can today lead to significant insured losses.\n\nThe socioeconomic factors, and short- and long-term climate drivers are here to stay, warranting careful reconsideration of commercially sustainable insurance offerings. Risk assessment needs to account for frequency and severity of individual perils as in a traditional natural catastrophe modelling. However, it also needs to explicitly consider the compounding effects of climate variability. These can trigger multiple and diverse secondary perils in the same season, adding significant annual variability to aggregate losses in underwriting. In view of dynamic nature of risk, taking a forward-looking stance versus averaging a multi-decadal past, is paramount.\n\n6 _Severe weather in a changing climate – 2_ _nd_ _ed_ , IAG, National Center for Atmospheric Research, 2020. 7 Insurance Council of Australia.\n\n**Figure 3** Timeline of events against trends of various climate drivers\n\nSources: Bureau of Meteorology (for IOD and ENSO data), British Antarctic Survey (for SAM data), Insurance Council of Australia (for loss numbers), Swiss Re Institute\n\n–5 –4 –3 –2 –1 0 1 2 3 4 5 6 7\n\nMay 2020 Jan 2020 Sep 2019 May 2019 Jan 2019\n\n–1.5\n\n–1.0\n\n–0.5\n\n0\\.0\n\n0\\.5\n\n1\\.0\n\n1\\.5\n\n2\\.0\n\n2\\.5\n\nSAM\n\nENSO/IOD\n\nSAM ENSO IOD Climate index: Jun 2019–Mar 2020\n\nBushfires: AUD 2.3 bn\n\nJan 20, 2020\n\nACT Hail: AUD 1.66 bn\n\nFeb 7–10, 2020\n\nEast Coast Flood: AUD 0.97 bn\n\nEvents:\n\nThe 2019/20 spring-summer season in Australia brought many severe weather events.\n\nClimate change cannot be cited as the driving cause.\n\nSocio-economic factors also play a significant role.\n\nRisk assessment should take a forward-looking stance of the multiple factor influencing losses."],"full_content":"**Swiss Re** _sigma_ No 1/2021 7\n\n_Implications for insurance_ Climate conditions help explain the physics leading to the weather event sequence. From the insurance perspective, important questions are: was the 2019/20 weather story in Australia a consequence of climate change, and was it an exceptional year? These questions cannot be answered with a definitive yes or no, but do raise important considerations for risk assessment and the pricing assumptions for standard natural catastrophe protection insurance covers.\n\nThere have been previous occurrences in Australia’s climate history of multiple extreme events coming together in clusters in highly variable and strong conditions (eg, in 2011). Severe weather events in Australia remain predominantly influenced by climate drivers like IOD, SAM and ENSO in the short term. As such, it is not easy to attribute a specific event or even an intensive natural catastrophe season to climate change. That said, long-term trends in risk factors like fire-inducing dry weather or extreme rainfall are very likely influenced by the rise in global temperatures. ⁶ The spring-summer of 2019/20 was an extreme season with several climate signals feeding off each other. And, with climate change set to continue if no mitigating actions are taken, such seasons are less likely to be exceptional.\n\nSecondary perils account for roughly 75% of insured losses in Australia. After a benign 2000–09, the last decade has seen an increased number of severe insurance losses, at AUD 1 billion or higher, many times in a single year across all secondary perils. ⁷ Short-term climate signals and the effects of climate change alone do not explain this increase. Climate change loss-impacts are being compounded by increasing exposure and expanding cities in coastal areas and in the wilderness. What may previously been a small or even unnoticed natural disturbance can today lead to significant insured losses.\n\nThe socioeconomic factors, and short- and long-term climate drivers are here to stay, warranting careful reconsideration of commercially sustainable insurance offerings. Risk assessment needs to account for frequency and severity of individual perils as in a traditional natural catastrophe modelling. However, it also needs to explicitly consider the compounding effects of climate variability. These can trigger multiple and diverse secondary perils in the same season, adding significant annual variability to aggregate losses in underwriting. In view of dynamic nature of risk, taking a forward-looking stance versus averaging a multi-decadal past, is paramount.\n\n6 _Severe weather in a changing climate – 2_ _nd_ _ed_ , IAG, National Center for Atmospheric Research, 2020. 7 Insurance Council of Australia.\n\n**Figure 3** Timeline of events against trends of various climate drivers\n\nSources: Bureau of Meteorology (for IOD and ENSO data), British Antarctic Survey (for SAM data), Insurance Council of Australia (for loss numbers), Swiss Re Institute\n\n–5 –4 –3 –2 –1 0 1 2 3 4 5 6 7\n\nMay 2020 Jan 2020 Sep 2019 May 2019 Jan 2019\n\n–1.5\n\n–1.0\n\n–0.5\n\n0\\.0\n\n0\\.5\n\n1\\.0\n\n1\\.5\n\n2\\.0\n\n2\\.5\n\nSAM\n\nENSO/IOD\n\nSAM ENSO IOD Climate index: Jun 2019–Mar 2020\n\nBushfires: AUD 2.3 bn\n\nJan 20, 2020\n\nACT Hail: AUD 1.66 bn\n\nFeb 7–10, 2020\n\nEast Coast Flood: AUD 0.97 bn\n\nEvents:\n\nThe 2019/20 spring-summer season in Australia brought many severe weather events.\n\nClimate change cannot be cited as the driving cause.\n\nSocio-economic factors also play a significant role.\n\nRisk assessment should take a forward-looking stance of the multiple factor influencing losses.","publish_date":null,"title":"Implications for insurance","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/sigma-1-2021-en_p9.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/sigma-1-2021-en_p9.pdf"},"request":{"example_id":"chart/sigma-1-2021-en_p9","product_type":"parse","source_file_path":"sigma-1-2021-en_p9.pdf"},"started_at":"2026-11-04T19:44:46.148602"} +{"completed_at":"2026-11-04T19:44:49.107286","latency_in_ms":993,"output":{"example_id":"chart/sri-sigma-natural-catastrophes-1-2025_p10","grounded_pages":[],"layout_pages":[],"markdown":"10 **Swiss Re Institute** _sigma_ No 1/2025 Insured losses on trend\n\nThe absence of peak-loss years since 2017 should not be read as a signal that primary peril loss events, especially large “tail” events, are becoming less likely. History shows that annual losses from primary perils vary considerably, punctuated by peak loss years. The experience of Hurricanes Katrina, Wilma and Rita in 2005, or Harvey, Irma and Maria in 2017 illustrates the scale of huge losses that can result from primary perils. And also that in years when low-frequency high-severity events do strike, particularly in regions of high economic exposure, insured losses from all natural catastrophes can reach well above trend levels, sometimes 100% or more above trend. In the next chapter, we estimate and demonstrate the potential magnitude of insured losses in a next peak- loss year and highlight scenarios that can cause such losses.\n\n**Figure 6** Distribution of annual losses from primary and secondary perils in % deviation from trend (1995 ‒ 2024)\n\nSource: Swiss Re Institute\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n700% 675% 650% 625% 600% 575% 550% 525% 500% 475% 450% 425% 400% 375% 350% 325% 300% 275% 250% 225% 200% 175% 150% 125% 100%\n\n75% 50% 25% 0% –25% –50% –75% –100% –125% –150%\n\nPrimary perils Secondary perils\n\nDeviations from tend\n\n2011 1999 2004 2005\n\n2011, 2017\n\nNumber of years\n\n…and it is a matter of _when_ , not if, the next peak loss year will occur.","pages":[{"line_numbers":[],"markdown":"10 **Swiss Re Institute** _sigma_ No 1/2025 Insured losses on trend\n\nThe absence of peak-loss years since 2017 should not be read as a signal that primary peril loss events, especially large “tail” events, are becoming less likely. History shows that annual losses from primary perils vary considerably, punctuated by peak loss years. The experience of Hurricanes Katrina, Wilma and Rita in 2005, or Harvey, Irma and Maria in 2017 illustrates the scale of huge losses that can result from primary perils. And also that in years when low-frequency high-severity events do strike, particularly in regions of high economic exposure, insured losses from all natural catastrophes can reach well above trend levels, sometimes 100% or more above trend. In the next chapter, we estimate and demonstrate the potential magnitude of insured losses in a next peak- loss year and highlight scenarios that can cause such losses.\n\n**Figure 6** Distribution of annual losses from primary and secondary perils in % deviation from trend (1995 ‒ 2024)\n\nSource: Swiss Re Institute\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n700% 675% 650% 625% 600% 575% 550% 525% 500% 475% 450% 425% 400% 375% 350% 325% 300% 275% 250% 225% 200% 175% 150% 125% 100%\n\n75% 50% 25% 0% –25% –50% –75% –100% –125% –150%\n\nPrimary perils Secondary perils\n\nDeviations from tend\n\n2011 1999 2004 2005\n\n2011, 2017\n\nNumber of years\n\n…and it is a matter of _when_ , not if, the next peak loss year will occur.","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"10 **Swiss Re Institute** _sigma_ No 1/2025 Insured losses on trend\n\nThe absence of peak-loss years since 2017 should not be read as a signal that primary peril loss events, especially large “tail” events, are becoming less likely. History shows that annual losses from primary perils vary considerably, punctuated by peak loss years. The experience of Hurricanes Katrina, Wilma and Rita in 2005, or Harvey, Irma and Maria in 2017 illustrates the scale of huge losses that can result from primary perils. And also that in years when low-frequency high-severity events do strike, particularly in regions of high economic exposure, insured losses from all natural catastrophes can reach well above trend levels, sometimes 100% or more above trend. In the next chapter, we estimate and demonstrate the potential magnitude of insured losses in a next peak- loss year and highlight scenarios that can cause such losses.\n\n**Figure 6** Distribution of annual losses from primary and secondary perils in % deviation from trend (1995 ‒ 2024)\n\nSource: Swiss Re Institute\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n700% 675% 650% 625% 600% 575% 550% 525% 500% 475% 450% 425% 400% 375% 350% 325% 300% 275% 250% 225% 200% 175% 150% 125% 100%\n\n75% 50% 25% 0% –25% –50% –75% –100% –125% –150%\n\nPrimary perils Secondary perils\n\nDeviations from tend\n\n2011 1999 2004 2005\n\n2011, 2017\n\nNumber of years\n\n…and it is a matter of _when_ , not if, the next peak loss year will occur.","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/sri-sigma-natural-catastrophes-1-2025_p10.pdf"]},"response":{"errors":[],"extract_id":"extract_b773384ac203c7d3226a23f41e60c36f","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["10 **Swiss Re Institute** _sigma_ No 1/2025 Insured losses on trend\n\nThe absence of peak-loss years since 2017 should not be read as a signal that primary peril loss events, especially large “tail” events, are becoming less likely. History shows that annual losses from primary perils vary considerably, punctuated by peak loss years. The experience of Hurricanes Katrina, Wilma and Rita in 2005, or Harvey, Irma and Maria in 2017 illustrates the scale of huge losses that can result from primary perils. And also that in years when low-frequency high-severity events do strike, particularly in regions of high economic exposure, insured losses from all natural catastrophes can reach well above trend levels, sometimes 100% or more above trend. In the next chapter, we estimate and demonstrate the potential magnitude of insured losses in a next peak- loss year and highlight scenarios that can cause such losses.\n\n**Figure 6** Distribution of annual losses from primary and secondary perils in % deviation from trend (1995 ‒ 2024)\n\nSource: Swiss Re Institute\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n700% 675% 650% 625% 600% 575% 550% 525% 500% 475% 450% 425% 400% 375% 350% 325% 300% 275% 250% 225% 200% 175% 150% 125% 100%\n\n75% 50% 25% 0% –25% –50% –75% –100% –125% –150%\n\nPrimary perils Secondary perils\n\nDeviations from tend\n\n2011 1999 2004 2005\n\n2011, 2017\n\nNumber of years\n\n…and it is a matter of _when_ , not if, the next peak loss year will occur."],"full_content":"10 **Swiss Re Institute** _sigma_ No 1/2025 Insured losses on trend\n\nThe absence of peak-loss years since 2017 should not be read as a signal that primary peril loss events, especially large “tail” events, are becoming less likely. History shows that annual losses from primary perils vary considerably, punctuated by peak loss years. The experience of Hurricanes Katrina, Wilma and Rita in 2005, or Harvey, Irma and Maria in 2017 illustrates the scale of huge losses that can result from primary perils. And also that in years when low-frequency high-severity events do strike, particularly in regions of high economic exposure, insured losses from all natural catastrophes can reach well above trend levels, sometimes 100% or more above trend. In the next chapter, we estimate and demonstrate the potential magnitude of insured losses in a next peak- loss year and highlight scenarios that can cause such losses.\n\n**Figure 6** Distribution of annual losses from primary and secondary perils in % deviation from trend (1995 ‒ 2024)\n\nSource: Swiss Re Institute\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n700% 675% 650% 625% 600% 575% 550% 525% 500% 475% 450% 425% 400% 375% 350% 325% 300% 275% 250% 225% 200% 175% 150% 125% 100%\n\n75% 50% 25% 0% –25% –50% –75% –100% –125% –150%\n\nPrimary perils Secondary perils\n\nDeviations from tend\n\n2011 1999 2004 2005\n\n2011, 2017\n\nNumber of years\n\n…and it is a matter of _when_ , not if, the next peak loss year will occur.","publish_date":null,"title":"The absence of peak-loss years since 2017 should not be read as a signal that primary peril loss events, especially","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/sri-sigma-natural-catastrophes-1-2025_p10.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/sri-sigma-natural-catastrophes-1-2025_p10.pdf"},"request":{"example_id":"chart/sri-sigma-natural-catastrophes-1-2025_p10","product_type":"parse","source_file_path":"sri-sigma-natural-catastrophes-1-2025_p10.pdf"},"started_at":"2026-11-04T19:44:48.113686"} +{"completed_at":"2026-11-04T19:44:50.237139","latency_in_ms":1128,"output":{"example_id":"chart/sri-sigma-natural-catastrophes-1-2025_p19","grounded_pages":[],"layout_pages":[],"markdown":"Insurability and homeowner insurance in the US _sigma_ No 1/2025 **Swiss Re Institute** 19\n\nWritten premiums to the CFP grew by 199% between September 2021 and September 2024 (USD 1.3 billion), ³⁷ making the association the third-largest home insurer in California. Far from the intended role of insurer-of-last-resort, the CFP is rapidly becoming a leading insurance entity and often the only option available for homeowners. ³⁸ Shrinking availability in the voluntary market is particularly pronounced in high-risk counties, and this is where the CFP bears high exposure (see Figure 11). State-wide exposure of the CFP grew from USD 195 billion to USD 468 billion in the same three-year period as above. This includes significant exposure in the Palisades neighbourhood, to the tune of USD 5.9 billion and in Eaton of USD 4.7 billion. ³⁹ In the Palisades fires, around 22% of the structures destroyed were covered by the CFP. In the Eaton Fires the share was around 12%.\n\n### Homeowner insurance in the US\n\nClaims costs have risen in recent years on the back of inflationary pressures, particularly in high-risk areas. Net incurred losses of homeowners insurers’ nationwide rose at a compound annual growth rate (CAGR) of 8% between 2018 and 2024. ⁴⁰ The key economic driver was construction costs, up around 40% between 2020 and 2022. This was far in excess of the consumer price index, which pushed up claim costs for insurers (see Figure 12). Construction costs have largely stabilised since 2023 but remain much higher than pre-COVID-19 levels.\n\n37 _Key Statistics & Data,_ The California FAIR Plan.\n\n38 Alternatively, cover is also available from the less regulated Excess & Surplus market. However, this is not suited to average homeowners; it is the purview of owners of high net-worth real estate.\n\n39 The California FAIR Plan, op. cit.\n\n40 _Mortgage Monitor report_ , Intercontinental Exchange (ICE), October 2024.\n\nThe CFP has assumed many of the non- renewal policies from insurers, particularly in high-risk areas.\n\n**Figure 11** Number of California Fair Plan new and renewed policies as share of total homeowner policies\n\nSource: California Department of Insurance, Swiss Re Institute\n\n0%\n\n5%\n\n10%\n\n15%\n\n20%\n\n25%\n\n30%\n\nTop 10 high fire risk counties\n\nStatewide\n\n2023 2022 2021 2020 2019 2018\n\nStatewide Top 10 high fire risk counties\n\nThe post pandemic inflation shock has been a main force behind the increases in homeowner insurance premiums in the US in recent years.","pages":[{"line_numbers":[],"markdown":"Insurability and homeowner insurance in the US _sigma_ No 1/2025 **Swiss Re Institute** 19\n\nWritten premiums to the CFP grew by 199% between September 2021 and September 2024 (USD 1.3 billion), ³⁷ making the association the third-largest home insurer in California. Far from the intended role of insurer-of-last-resort, the CFP is rapidly becoming a leading insurance entity and often the only option available for homeowners. ³⁸ Shrinking availability in the voluntary market is particularly pronounced in high-risk counties, and this is where the CFP bears high exposure (see Figure 11). State-wide exposure of the CFP grew from USD 195 billion to USD 468 billion in the same three-year period as above. This includes significant exposure in the Palisades neighbourhood, to the tune of USD 5.9 billion and in Eaton of USD 4.7 billion. ³⁹ In the Palisades fires, around 22% of the structures destroyed were covered by the CFP. In the Eaton Fires the share was around 12%.\n\n### Homeowner insurance in the US\n\nClaims costs have risen in recent years on the back of inflationary pressures, particularly in high-risk areas. Net incurred losses of homeowners insurers’ nationwide rose at a compound annual growth rate (CAGR) of 8% between 2018 and 2024. ⁴⁰ The key economic driver was construction costs, up around 40% between 2020 and 2022. This was far in excess of the consumer price index, which pushed up claim costs for insurers (see Figure 12). Construction costs have largely stabilised since 2023 but remain much higher than pre-COVID-19 levels.\n\n37 _Key Statistics & Data,_ The California FAIR Plan.\n\n38 Alternatively, cover is also available from the less regulated Excess & Surplus market. However, this is not suited to average homeowners; it is the purview of owners of high net-worth real estate.\n\n39 The California FAIR Plan, op. cit.\n\n40 _Mortgage Monitor report_ , Intercontinental Exchange (ICE), October 2024.\n\nThe CFP has assumed many of the non- renewal policies from insurers, particularly in high-risk areas.\n\n**Figure 11** Number of California Fair Plan new and renewed policies as share of total homeowner policies\n\nSource: California Department of Insurance, Swiss Re Institute\n\n0%\n\n5%\n\n10%\n\n15%\n\n20%\n\n25%\n\n30%\n\nTop 10 high fire risk counties\n\nStatewide\n\n2023 2022 2021 2020 2019 2018\n\nStatewide Top 10 high fire risk counties\n\nThe post pandemic inflation shock has been a main force behind the increases in homeowner insurance premiums in the US in recent years.","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Insurability and homeowner insurance in the US _sigma_ No 1/2025 **Swiss Re Institute** 19\n\nWritten premiums to the CFP grew by 199% between September 2021 and September 2024 (USD 1.3 billion), ³⁷ making the association the third-largest home insurer in California. Far from the intended role of insurer-of-last-resort, the CFP is rapidly becoming a leading insurance entity and often the only option available for homeowners. ³⁸ Shrinking availability in the voluntary market is particularly pronounced in high-risk counties, and this is where the CFP bears high exposure (see Figure 11). State-wide exposure of the CFP grew from USD 195 billion to USD 468 billion in the same three-year period as above. This includes significant exposure in the Palisades neighbourhood, to the tune of USD 5.9 billion and in Eaton of USD 4.7 billion. ³⁹ In the Palisades fires, around 22% of the structures destroyed were covered by the CFP. In the Eaton Fires the share was around 12%.\n\n### Homeowner insurance in the US\n\nClaims costs have risen in recent years on the back of inflationary pressures, particularly in high-risk areas. Net incurred losses of homeowners insurers’ nationwide rose at a compound annual growth rate (CAGR) of 8% between 2018 and 2024. ⁴⁰ The key economic driver was construction costs, up around 40% between 2020 and 2022. This was far in excess of the consumer price index, which pushed up claim costs for insurers (see Figure 12). Construction costs have largely stabilised since 2023 but remain much higher than pre-COVID-19 levels.\n\n37 _Key Statistics & Data,_ The California FAIR Plan.\n\n38 Alternatively, cover is also available from the less regulated Excess & Surplus market. However, this is not suited to average homeowners; it is the purview of owners of high net-worth real estate.\n\n39 The California FAIR Plan, op. cit.\n\n40 _Mortgage Monitor report_ , Intercontinental Exchange (ICE), October 2024.\n\nThe CFP has assumed many of the non- renewal policies from insurers, particularly in high-risk areas.\n\n**Figure 11** Number of California Fair Plan new and renewed policies as share of total homeowner policies\n\nSource: California Department of Insurance, Swiss Re Institute\n\n0%\n\n5%\n\n10%\n\n15%\n\n20%\n\n25%\n\n30%\n\nTop 10 high fire risk counties\n\nStatewide\n\n2023 2022 2021 2020 2019 2018\n\nStatewide Top 10 high fire risk counties\n\nThe post pandemic inflation shock has been a main force behind the increases in homeowner insurance premiums in the US in recent years.","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/sri-sigma-natural-catastrophes-1-2025_p19.pdf"]},"response":{"errors":[],"extract_id":"extract_39e8a7e1e929811b0afca27a00cfc881","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Insurability and homeowner insurance in the US _sigma_ No 1/2025 **Swiss Re Institute** 19\n\nWritten premiums to the CFP grew by 199% between September 2021 and September 2024 (USD 1.3 billion), ³⁷ making the association the third-largest home insurer in California. Far from the intended role of insurer-of-last-resort, the CFP is rapidly becoming a leading insurance entity and often the only option available for homeowners. ³⁸ Shrinking availability in the voluntary market is particularly pronounced in high-risk counties, and this is where the CFP bears high exposure (see Figure 11). State-wide exposure of the CFP grew from USD 195 billion to USD 468 billion in the same three-year period as above. This includes significant exposure in the Palisades neighbourhood, to the tune of USD 5.9 billion and in Eaton of USD 4.7 billion. ³⁹ In the Palisades fires, around 22% of the structures destroyed were covered by the CFP. In the Eaton Fires the share was around 12%.\n\n### Homeowner insurance in the US\n\nClaims costs have risen in recent years on the back of inflationary pressures, particularly in high-risk areas. Net incurred losses of homeowners insurers’ nationwide rose at a compound annual growth rate (CAGR) of 8% between 2018 and 2024. ⁴⁰ The key economic driver was construction costs, up around 40% between 2020 and 2022. This was far in excess of the consumer price index, which pushed up claim costs for insurers (see Figure 12). Construction costs have largely stabilised since 2023 but remain much higher than pre-COVID-19 levels.\n\n37 _Key Statistics & Data,_ The California FAIR Plan.\n\n38 Alternatively, cover is also available from the less regulated Excess & Surplus market. However, this is not suited to average homeowners; it is the purview of owners of high net-worth real estate.\n\n39 The California FAIR Plan, op. cit.\n\n40 _Mortgage Monitor report_ , Intercontinental Exchange (ICE), October 2024.\n\nThe CFP has assumed many of the non- renewal policies from insurers, particularly in high-risk areas.\n\n**Figure 11** Number of California Fair Plan new and renewed policies as share of total homeowner policies\n\nSource: California Department of Insurance, Swiss Re Institute\n\n0%\n\n5%\n\n10%\n\n15%\n\n20%\n\n25%\n\n30%\n\nTop 10 high fire risk counties\n\nStatewide\n\n2023 2022 2021 2020 2019 2018\n\nStatewide Top 10 high fire risk counties\n\nThe post pandemic inflation shock has been a main force behind the increases in homeowner insurance premiums in the US in recent years."],"full_content":"Insurability and homeowner insurance in the US _sigma_ No 1/2025 **Swiss Re Institute** 19\n\nWritten premiums to the CFP grew by 199% between September 2021 and September 2024 (USD 1.3 billion), ³⁷ making the association the third-largest home insurer in California. Far from the intended role of insurer-of-last-resort, the CFP is rapidly becoming a leading insurance entity and often the only option available for homeowners. ³⁸ Shrinking availability in the voluntary market is particularly pronounced in high-risk counties, and this is where the CFP bears high exposure (see Figure 11). State-wide exposure of the CFP grew from USD 195 billion to USD 468 billion in the same three-year period as above. This includes significant exposure in the Palisades neighbourhood, to the tune of USD 5.9 billion and in Eaton of USD 4.7 billion. ³⁹ In the Palisades fires, around 22% of the structures destroyed were covered by the CFP. In the Eaton Fires the share was around 12%.\n\n### Homeowner insurance in the US\n\nClaims costs have risen in recent years on the back of inflationary pressures, particularly in high-risk areas. Net incurred losses of homeowners insurers’ nationwide rose at a compound annual growth rate (CAGR) of 8% between 2018 and 2024. ⁴⁰ The key economic driver was construction costs, up around 40% between 2020 and 2022. This was far in excess of the consumer price index, which pushed up claim costs for insurers (see Figure 12). Construction costs have largely stabilised since 2023 but remain much higher than pre-COVID-19 levels.\n\n37 _Key Statistics & Data,_ The California FAIR Plan.\n\n38 Alternatively, cover is also available from the less regulated Excess & Surplus market. However, this is not suited to average homeowners; it is the purview of owners of high net-worth real estate.\n\n39 The California FAIR Plan, op. cit.\n\n40 _Mortgage Monitor report_ , Intercontinental Exchange (ICE), October 2024.\n\nThe CFP has assumed many of the non- renewal policies from insurers, particularly in high-risk areas.\n\n**Figure 11** Number of California Fair Plan new and renewed policies as share of total homeowner policies\n\nSource: California Department of Insurance, Swiss Re Institute\n\n0%\n\n5%\n\n10%\n\n15%\n\n20%\n\n25%\n\n30%\n\nTop 10 high fire risk counties\n\nStatewide\n\n2023 2022 2021 2020 2019 2018\n\nStatewide Top 10 high fire risk counties\n\nThe post pandemic inflation shock has been a main force behind the increases in homeowner insurance premiums in the US in recent years.","publish_date":null,"title":"Homeowner insurance in the US","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/sri-sigma-natural-catastrophes-1-2025_p19.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/sri-sigma-natural-catastrophes-1-2025_p19.pdf"},"request":{"example_id":"chart/sri-sigma-natural-catastrophes-1-2025_p19","product_type":"parse","source_file_path":"sri-sigma-natural-catastrophes-1-2025_p19.pdf"},"started_at":"2026-11-04T19:44:49.108551"} +{"completed_at":"2026-11-04T19:44:52.461632","latency_in_ms":1420,"output":{"example_id":"chart/sri-sigma-natural-catastrophes-1-2025_p20","grounded_pages":[],"layout_pages":[],"markdown":"20 **Swiss Re Institute** _sigma_ No 1/2025 Insurability and homeowner insurance in the US\n\nThe surge in construction and rebuilding costs adversely impacted underwriting results in property insurance. From 2020 to 2024, the segment’s combined ratio was more than 100%, reaching a decade high of 110% in 2023. This indicates that insurers paid out more in claims and expenses than they collected in premiums (see Figure 13). ⁴¹ On a net basis, homeowner insurers have incurred underwriting losses of USD 35 billion over the last 10 years. They experienced underwriting profit in just three of those years.\n\nRecent increases in homeowner premiums represent a catch-up effect to inflation gains, although with a lag due to reporting and regulatory processes. After several years of underwriting losses, homeowners’ premiums only recently caught up with housing replacement costs. However, ongoing underwriting losses suggest that premiums are still not commensurate with the risk, and that further alignment is necessary to sustain insurance business.\n\n41 _12th national risk assessment: property prices in peril_ , First Street, February 2025.\n\n**Figure 12** US Construction cost indices\n\nSource: Bureau of Labor Statistics, Census Bureau, S&P Capital IQ, Swiss Re Institute\n\n100\n\n110\n\n120\n\n130\n\n140\n\n150\n\nConstruction Indicators, One-Family Houses, Construction Price, Fisher (Deflator), Houses u\n\nBLS, PPI, Commodity Data, Intermediate Demand by Commodity Type, 12, Materials & Com\n\nProducer Price Index, FD-ID, Final Demand, Construction, Index\n\nConsumer Price Index, All Urban Consumers, U.S. City Average, All Items, Index\n\n2025 2024 2023 2022 2021 2020\n\nCPI PPI, Materials & Components for Construction PPI, Final Demand, Construction\n\nOne-Family Houses under Construction, Census\n\nHomeowner insurers have experienced many years of underwriting losses due to increased claims costs.\n\n**Figure 13** US homeowner insurance underwriting result\n\nSource: AM Best, Swiss Re Institute\n\n–15%\n\n–10%\n\n–5%\n\n0%\n\n5%\n\n10%\n\n2024 2023 2022 2021 2020 2019 2018 2017 2016 2015 2014\n\nUnderwriting profit Underwriting loss\n\nOnly recently have premiums come back on par with housing replacement costs.","pages":[{"line_numbers":[],"markdown":"20 **Swiss Re Institute** _sigma_ No 1/2025 Insurability and homeowner insurance in the US\n\nThe surge in construction and rebuilding costs adversely impacted underwriting results in property insurance. From 2020 to 2024, the segment’s combined ratio was more than 100%, reaching a decade high of 110% in 2023. This indicates that insurers paid out more in claims and expenses than they collected in premiums (see Figure 13). ⁴¹ On a net basis, homeowner insurers have incurred underwriting losses of USD 35 billion over the last 10 years. They experienced underwriting profit in just three of those years.\n\nRecent increases in homeowner premiums represent a catch-up effect to inflation gains, although with a lag due to reporting and regulatory processes. After several years of underwriting losses, homeowners’ premiums only recently caught up with housing replacement costs. However, ongoing underwriting losses suggest that premiums are still not commensurate with the risk, and that further alignment is necessary to sustain insurance business.\n\n41 _12th national risk assessment: property prices in peril_ , First Street, February 2025.\n\n**Figure 12** US Construction cost indices\n\nSource: Bureau of Labor Statistics, Census Bureau, S&P Capital IQ, Swiss Re Institute\n\n100\n\n110\n\n120\n\n130\n\n140\n\n150\n\nConstruction Indicators, One-Family Houses, Construction Price, Fisher (Deflator), Houses u\n\nBLS, PPI, Commodity Data, Intermediate Demand by Commodity Type, 12, Materials & Com\n\nProducer Price Index, FD-ID, Final Demand, Construction, Index\n\nConsumer Price Index, All Urban Consumers, U.S. City Average, All Items, Index\n\n2025 2024 2023 2022 2021 2020\n\nCPI PPI, Materials & Components for Construction PPI, Final Demand, Construction\n\nOne-Family Houses under Construction, Census\n\nHomeowner insurers have experienced many years of underwriting losses due to increased claims costs.\n\n**Figure 13** US homeowner insurance underwriting result\n\nSource: AM Best, Swiss Re Institute\n\n–15%\n\n–10%\n\n–5%\n\n0%\n\n5%\n\n10%\n\n2024 2023 2022 2021 2020 2019 2018 2017 2016 2015 2014\n\nUnderwriting profit Underwriting loss\n\nOnly recently have premiums come back on par with housing replacement costs.","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"20 **Swiss Re Institute** _sigma_ No 1/2025 Insurability and homeowner insurance in the US\n\nThe surge in construction and rebuilding costs adversely impacted underwriting results in property insurance. From 2020 to 2024, the segment’s combined ratio was more than 100%, reaching a decade high of 110% in 2023. This indicates that insurers paid out more in claims and expenses than they collected in premiums (see Figure 13). ⁴¹ On a net basis, homeowner insurers have incurred underwriting losses of USD 35 billion over the last 10 years. They experienced underwriting profit in just three of those years.\n\nRecent increases in homeowner premiums represent a catch-up effect to inflation gains, although with a lag due to reporting and regulatory processes. After several years of underwriting losses, homeowners’ premiums only recently caught up with housing replacement costs. However, ongoing underwriting losses suggest that premiums are still not commensurate with the risk, and that further alignment is necessary to sustain insurance business.\n\n41 _12th national risk assessment: property prices in peril_ , First Street, February 2025.\n\n**Figure 12** US Construction cost indices\n\nSource: Bureau of Labor Statistics, Census Bureau, S&P Capital IQ, Swiss Re Institute\n\n100\n\n110\n\n120\n\n130\n\n140\n\n150\n\nConstruction Indicators, One-Family Houses, Construction Price, Fisher (Deflator), Houses u\n\nBLS, PPI, Commodity Data, Intermediate Demand by Commodity Type, 12, Materials & Com\n\nProducer Price Index, FD-ID, Final Demand, Construction, Index\n\nConsumer Price Index, All Urban Consumers, U.S. City Average, All Items, Index\n\n2025 2024 2023 2022 2021 2020\n\nCPI PPI, Materials & Components for Construction PPI, Final Demand, Construction\n\nOne-Family Houses under Construction, Census\n\nHomeowner insurers have experienced many years of underwriting losses due to increased claims costs.\n\n**Figure 13** US homeowner insurance underwriting result\n\nSource: AM Best, Swiss Re Institute\n\n–15%\n\n–10%\n\n–5%\n\n0%\n\n5%\n\n10%\n\n2024 2023 2022 2021 2020 2019 2018 2017 2016 2015 2014\n\nUnderwriting profit Underwriting loss\n\nOnly recently have premiums come back on par with housing replacement costs.","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/sri-sigma-natural-catastrophes-1-2025_p20.pdf"]},"response":{"errors":[],"extract_id":"extract_a3b6eb80c9c160e802d8c0873914560f","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["20 **Swiss Re Institute** _sigma_ No 1/2025 Insurability and homeowner insurance in the US\n\nThe surge in construction and rebuilding costs adversely impacted underwriting results in property insurance. From 2020 to 2024, the segment’s combined ratio was more than 100%, reaching a decade high of 110% in 2023. This indicates that insurers paid out more in claims and expenses than they collected in premiums (see Figure 13). ⁴¹ On a net basis, homeowner insurers have incurred underwriting losses of USD 35 billion over the last 10 years. They experienced underwriting profit in just three of those years.\n\nRecent increases in homeowner premiums represent a catch-up effect to inflation gains, although with a lag due to reporting and regulatory processes. After several years of underwriting losses, homeowners’ premiums only recently caught up with housing replacement costs. However, ongoing underwriting losses suggest that premiums are still not commensurate with the risk, and that further alignment is necessary to sustain insurance business.\n\n41 _12th national risk assessment: property prices in peril_ , First Street, February 2025.\n\n**Figure 12** US Construction cost indices\n\nSource: Bureau of Labor Statistics, Census Bureau, S&P Capital IQ, Swiss Re Institute\n\n100\n\n110\n\n120\n\n130\n\n140\n\n150\n\nConstruction Indicators, One-Family Houses, Construction Price, Fisher (Deflator), Houses u\n\nBLS, PPI, Commodity Data, Intermediate Demand by Commodity Type, 12, Materials & Com\n\nProducer Price Index, FD-ID, Final Demand, Construction, Index\n\nConsumer Price Index, All Urban Consumers, U.S. City Average, All Items, Index\n\n2025 2024 2023 2022 2021 2020\n\nCPI PPI, Materials & Components for Construction PPI, Final Demand, Construction\n\nOne-Family Houses under Construction, Census\n\nHomeowner insurers have experienced many years of underwriting losses due to increased claims costs.\n\n**Figure 13** US homeowner insurance underwriting result\n\nSource: AM Best, Swiss Re Institute\n\n–15%\n\n–10%\n\n–5%\n\n0%\n\n5%\n\n10%\n\n2024 2023 2022 2021 2020 2019 2018 2017 2016 2015 2014\n\nUnderwriting profit Underwriting loss\n\nOnly recently have premiums come back on par with housing replacement costs."],"full_content":"20 **Swiss Re Institute** _sigma_ No 1/2025 Insurability and homeowner insurance in the US\n\nThe surge in construction and rebuilding costs adversely impacted underwriting results in property insurance. From 2020 to 2024, the segment’s combined ratio was more than 100%, reaching a decade high of 110% in 2023. This indicates that insurers paid out more in claims and expenses than they collected in premiums (see Figure 13). ⁴¹ On a net basis, homeowner insurers have incurred underwriting losses of USD 35 billion over the last 10 years. They experienced underwriting profit in just three of those years.\n\nRecent increases in homeowner premiums represent a catch-up effect to inflation gains, although with a lag due to reporting and regulatory processes. After several years of underwriting losses, homeowners’ premiums only recently caught up with housing replacement costs. However, ongoing underwriting losses suggest that premiums are still not commensurate with the risk, and that further alignment is necessary to sustain insurance business.\n\n41 _12th national risk assessment: property prices in peril_ , First Street, February 2025.\n\n**Figure 12** US Construction cost indices\n\nSource: Bureau of Labor Statistics, Census Bureau, S&P Capital IQ, Swiss Re Institute\n\n100\n\n110\n\n120\n\n130\n\n140\n\n150\n\nConstruction Indicators, One-Family Houses, Construction Price, Fisher (Deflator), Houses u\n\nBLS, PPI, Commodity Data, Intermediate Demand by Commodity Type, 12, Materials & Com\n\nProducer Price Index, FD-ID, Final Demand, Construction, Index\n\nConsumer Price Index, All Urban Consumers, U.S. City Average, All Items, Index\n\n2025 2024 2023 2022 2021 2020\n\nCPI PPI, Materials & Components for Construction PPI, Final Demand, Construction\n\nOne-Family Houses under Construction, Census\n\nHomeowner insurers have experienced many years of underwriting losses due to increased claims costs.\n\n**Figure 13** US homeowner insurance underwriting result\n\nSource: AM Best, Swiss Re Institute\n\n–15%\n\n–10%\n\n–5%\n\n0%\n\n5%\n\n10%\n\n2024 2023 2022 2021 2020 2019 2018 2017 2016 2015 2014\n\nUnderwriting profit Underwriting loss\n\nOnly recently have premiums come back on par with housing replacement costs.","publish_date":null,"title":"The surge in construction and rebuilding costs adversely impacted underwriting results in property insurance. From 2020 to 2024, the segment’s","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/sri-sigma-natural-catastrophes-1-2025_p20.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/sri-sigma-natural-catastrophes-1-2025_p20.pdf"},"request":{"example_id":"chart/sri-sigma-natural-catastrophes-1-2025_p20","product_type":"parse","source_file_path":"sri-sigma-natural-catastrophes-1-2025_p20.pdf"},"started_at":"2026-11-04T19:44:51.040793"} +{"completed_at":"2026-11-04T19:44:54.320346","latency_in_ms":2461,"output":{"example_id":"chart/sri-sigma-natural-catastrophes-1-2025_p26","grounded_pages":[],"layout_pages":[],"markdown":"26 **Swiss Re Institute** _sigma_ No 1/2025\n\n### Number of catastrophic events: 336\n\nIn terms of _sigma_ criteria, there were 336 catastrophes worldwide in 2024, up from 334 in 2023. There were 228 natural catastrophes (up from 219 in 2023), and 108 man- made disasters (down from 115 in 2023).\n\n### Number of victims: 14 083\n\nWorldwide, 14 083 people are believed to have died or gone missing in disaster events in 2024. Natural catastrophes claimed 10 822 victims. Man-made disasters claimed a further 3 261 victims.\n\n# Appendix: facts and figures\n\n**Figure 16** Number of catastrophic events, 1970 ‒ 2024\n\nSources: Swiss Re Institute\n\n0\n\n50\n\n100\n\n150\n\n200\n\n250\n\n300\n\n2024 2021 2018 2015 2012 2009 2006 2003 2000 1997 1994 1991 1988 1985 1982 1979 1976 1973 1970\n\nNatural catastrophes Man-made disasters\n\n**Figure 17** Number of victims, 1970 - 2024\n\nNote: Scale is logarithmic: the number of victims increases tenfold per band. Source: Swiss Re Institute\n\n1 000\n\n10 000\n\n100 000\n\n1 000 000\n\n10 000 000\n\n2021 2024 2018 2015 2012 2009 2006 2003 2000 1997 1994 1991 1988 1985 1982 1979 1976 1973 1970\n\nMan-made disasters Natural catastrophes\n\n1970: Bangladesh storm, Peru earthquake\n\n1976: Tangshan Earthquake, China\n\n1991: Cyclone Gorky, Bangladesh\n\n2004: Indian Ocean Earthquake and tsunami\n\n2008: Cyclone Nargis, Myanmar\n\n2010: Haiti Earthquake 2023: Turkey & Syria earthquake","pages":[{"line_numbers":[],"markdown":"26 **Swiss Re Institute** _sigma_ No 1/2025\n\n### Number of catastrophic events: 336\n\nIn terms of _sigma_ criteria, there were 336 catastrophes worldwide in 2024, up from 334 in 2023. There were 228 natural catastrophes (up from 219 in 2023), and 108 man- made disasters (down from 115 in 2023).\n\n### Number of victims: 14 083\n\nWorldwide, 14 083 people are believed to have died or gone missing in disaster events in 2024. Natural catastrophes claimed 10 822 victims. Man-made disasters claimed a further 3 261 victims.\n\n# Appendix: facts and figures\n\n**Figure 16** Number of catastrophic events, 1970 ‒ 2024\n\nSources: Swiss Re Institute\n\n0\n\n50\n\n100\n\n150\n\n200\n\n250\n\n300\n\n2024 2021 2018 2015 2012 2009 2006 2003 2000 1997 1994 1991 1988 1985 1982 1979 1976 1973 1970\n\nNatural catastrophes Man-made disasters\n\n**Figure 17** Number of victims, 1970 - 2024\n\nNote: Scale is logarithmic: the number of victims increases tenfold per band. Source: Swiss Re Institute\n\n1 000\n\n10 000\n\n100 000\n\n1 000 000\n\n10 000 000\n\n2021 2024 2018 2015 2012 2009 2006 2003 2000 1997 1994 1991 1988 1985 1982 1979 1976 1973 1970\n\nMan-made disasters Natural catastrophes\n\n1970: Bangladesh storm, Peru earthquake\n\n1976: Tangshan Earthquake, China\n\n1991: Cyclone Gorky, Bangladesh\n\n2004: Indian Ocean Earthquake and tsunami\n\n2008: Cyclone Nargis, Myanmar\n\n2010: Haiti Earthquake 2023: Turkey & Syria earthquake","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"26 **Swiss Re Institute** _sigma_ No 1/2025\n\n### Number of catastrophic events: 336\n\nIn terms of _sigma_ criteria, there were 336 catastrophes worldwide in 2024, up from 334 in 2023. There were 228 natural catastrophes (up from 219 in 2023), and 108 man- made disasters (down from 115 in 2023).\n\n### Number of victims: 14 083\n\nWorldwide, 14 083 people are believed to have died or gone missing in disaster events in 2024. Natural catastrophes claimed 10 822 victims. Man-made disasters claimed a further 3 261 victims.\n\n# Appendix: facts and figures\n\n**Figure 16** Number of catastrophic events, 1970 ‒ 2024\n\nSources: Swiss Re Institute\n\n0\n\n50\n\n100\n\n150\n\n200\n\n250\n\n300\n\n2024 2021 2018 2015 2012 2009 2006 2003 2000 1997 1994 1991 1988 1985 1982 1979 1976 1973 1970\n\nNatural catastrophes Man-made disasters\n\n**Figure 17** Number of victims, 1970 - 2024\n\nNote: Scale is logarithmic: the number of victims increases tenfold per band. Source: Swiss Re Institute\n\n1 000\n\n10 000\n\n100 000\n\n1 000 000\n\n10 000 000\n\n2021 2024 2018 2015 2012 2009 2006 2003 2000 1997 1994 1991 1988 1985 1982 1979 1976 1973 1970\n\nMan-made disasters Natural catastrophes\n\n1970: Bangladesh storm, Peru earthquake\n\n1976: Tangshan Earthquake, China\n\n1991: Cyclone Gorky, Bangladesh\n\n2004: Indian Ocean Earthquake and tsunami\n\n2008: Cyclone Nargis, Myanmar\n\n2010: Haiti Earthquake 2023: Turkey & Syria earthquake","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/sri-sigma-natural-catastrophes-1-2025_p26.pdf"]},"response":{"errors":[],"extract_id":"extract_122959f9d0f293cfbd5a9dc31e2af636","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["26 **Swiss Re Institute** _sigma_ No 1/2025\n\n### Number of catastrophic events: 336\n\nIn terms of _sigma_ criteria, there were 336 catastrophes worldwide in 2024, up from 334 in 2023. There were 228 natural catastrophes (up from 219 in 2023), and 108 man- made disasters (down from 115 in 2023).\n\n### Number of victims: 14 083\n\nWorldwide, 14 083 people are believed to have died or gone missing in disaster events in 2024. Natural catastrophes claimed 10 822 victims. Man-made disasters claimed a further 3 261 victims.\n\n# Appendix: facts and figures\n\n**Figure 16** Number of catastrophic events, 1970 ‒ 2024\n\nSources: Swiss Re Institute\n\n0\n\n50\n\n100\n\n150\n\n200\n\n250\n\n300\n\n2024 2021 2018 2015 2012 2009 2006 2003 2000 1997 1994 1991 1988 1985 1982 1979 1976 1973 1970\n\nNatural catastrophes Man-made disasters\n\n**Figure 17** Number of victims, 1970 - 2024\n\nNote: Scale is logarithmic: the number of victims increases tenfold per band. Source: Swiss Re Institute\n\n1 000\n\n10 000\n\n100 000\n\n1 000 000\n\n10 000 000\n\n2021 2024 2018 2015 2012 2009 2006 2003 2000 1997 1994 1991 1988 1985 1982 1979 1976 1973 1970\n\nMan-made disasters Natural catastrophes\n\n1970: Bangladesh storm, Peru earthquake\n\n1976: Tangshan Earthquake, China\n\n1991: Cyclone Gorky, Bangladesh\n\n2004: Indian Ocean Earthquake and tsunami\n\n2008: Cyclone Nargis, Myanmar\n\n2010: Haiti Earthquake 2023: Turkey & Syria earthquake"],"full_content":"26 **Swiss Re Institute** _sigma_ No 1/2025\n\n### Number of catastrophic events: 336\n\nIn terms of _sigma_ criteria, there were 336 catastrophes worldwide in 2024, up from 334 in 2023. There were 228 natural catastrophes (up from 219 in 2023), and 108 man- made disasters (down from 115 in 2023).\n\n### Number of victims: 14 083\n\nWorldwide, 14 083 people are believed to have died or gone missing in disaster events in 2024. Natural catastrophes claimed 10 822 victims. Man-made disasters claimed a further 3 261 victims.\n\n# Appendix: facts and figures\n\n**Figure 16** Number of catastrophic events, 1970 ‒ 2024\n\nSources: Swiss Re Institute\n\n0\n\n50\n\n100\n\n150\n\n200\n\n250\n\n300\n\n2024 2021 2018 2015 2012 2009 2006 2003 2000 1997 1994 1991 1988 1985 1982 1979 1976 1973 1970\n\nNatural catastrophes Man-made disasters\n\n**Figure 17** Number of victims, 1970 - 2024\n\nNote: Scale is logarithmic: the number of victims increases tenfold per band. Source: Swiss Re Institute\n\n1 000\n\n10 000\n\n100 000\n\n1 000 000\n\n10 000 000\n\n2021 2024 2018 2015 2012 2009 2006 2003 2000 1997 1994 1991 1988 1985 1982 1979 1976 1973 1970\n\nMan-made disasters Natural catastrophes\n\n1970: Bangladesh storm, Peru earthquake\n\n1976: Tangshan Earthquake, China\n\n1991: Cyclone Gorky, Bangladesh\n\n2004: Indian Ocean Earthquake and tsunami\n\n2008: Cyclone Nargis, Myanmar\n\n2010: Haiti Earthquake 2023: Turkey & Syria earthquake","publish_date":null,"title":"Appendix: facts and figures","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/sri-sigma-natural-catastrophes-1-2025_p26.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/sri-sigma-natural-catastrophes-1-2025_p26.pdf"},"request":{"example_id":"chart/sri-sigma-natural-catastrophes-1-2025_p26","product_type":"parse","source_file_path":"sri-sigma-natural-catastrophes-1-2025_p26.pdf"},"started_at":"2026-11-04T19:44:51.859179"} +{"completed_at":"2026-11-04T19:44:54.861760","latency_in_ms":1980,"output":{"example_id":"chart/sri-sigma-natural-catastrophes-1-2025_p27","grounded_pages":[],"layout_pages":[],"markdown":"Appendix: facts and figures _sigma_ No 1/2025 **Swiss Re Institute** 27\n\n### Global insured losses\n\nHurricane Helene and Milton, the second highest severe convective storms (SCS) and large-scale urban floods were the main events driving global insured losses from natural disasters of USD 137 billion in 2024. Man-made disasters triggered an additional USD 9 billion losses.\n\n### Primary and secondary perils\n\nHurricane Helene and Milton brought the share of insured losses from so-called primary perils to roughly 41% in 2024, more than double the share in 2023 (19%).\n\n**Figure 18** Insured catastrophe losses, 1970 – 2024 (USD bn, 2024 prices)\n\nSource: Swiss Re Institute\n\n0\n\n50\n\n100\n\n150\n\n200\n\n250\n\n2024 2021 2018 2015 2012 2009 2006 2003 2000 1997 1994 1991 1988 1985 1982 1979 1976 1973 1970\n\nEarthquake/tsunami Weather-related catastrophes Man-made disasters\n\nHurricane Andrew\n\nWinter Storm Lothar WTC\n\nHurricanes Ike, Gustav\n\nHurricanes Ivan, Charley, Frances\n\nHurricanes Katrina, Rita, Wilma\n\nJapan, NZ earthquakes, Thailand flood\n\nHurricanes Harvey, Irma, Maria\n\nHurricane Ian\n\nHurricanes Helene, Milton\n\nHurricane Sandy\n\n**Figure 19** Global insured losses from primary and secondary perils, 1970 – 2024, in USD billion at 2024 prices\n\nSource: Swiss Re Institute\n\n0\n\n25\n\n50\n\n75\n\n100\n\n125\n\n150\n\n175\n\n200\n\n2024 2021 2018 2015 2012 2009 2006 2003 2000 1997 1994 1991 1988 1985 1982 1979 1976 1973 1970\n\nInsured losses (USD bn)\n\nExponential trend (total natcat losses) Primary perils (incl secondary effects) Secondary perils","pages":[{"line_numbers":[],"markdown":"Appendix: facts and figures _sigma_ No 1/2025 **Swiss Re Institute** 27\n\n### Global insured losses\n\nHurricane Helene and Milton, the second highest severe convective storms (SCS) and large-scale urban floods were the main events driving global insured losses from natural disasters of USD 137 billion in 2024. Man-made disasters triggered an additional USD 9 billion losses.\n\n### Primary and secondary perils\n\nHurricane Helene and Milton brought the share of insured losses from so-called primary perils to roughly 41% in 2024, more than double the share in 2023 (19%).\n\n**Figure 18** Insured catastrophe losses, 1970 – 2024 (USD bn, 2024 prices)\n\nSource: Swiss Re Institute\n\n0\n\n50\n\n100\n\n150\n\n200\n\n250\n\n2024 2021 2018 2015 2012 2009 2006 2003 2000 1997 1994 1991 1988 1985 1982 1979 1976 1973 1970\n\nEarthquake/tsunami Weather-related catastrophes Man-made disasters\n\nHurricane Andrew\n\nWinter Storm Lothar WTC\n\nHurricanes Ike, Gustav\n\nHurricanes Ivan, Charley, Frances\n\nHurricanes Katrina, Rita, Wilma\n\nJapan, NZ earthquakes, Thailand flood\n\nHurricanes Harvey, Irma, Maria\n\nHurricane Ian\n\nHurricanes Helene, Milton\n\nHurricane Sandy\n\n**Figure 19** Global insured losses from primary and secondary perils, 1970 – 2024, in USD billion at 2024 prices\n\nSource: Swiss Re Institute\n\n0\n\n25\n\n50\n\n75\n\n100\n\n125\n\n150\n\n175\n\n200\n\n2024 2021 2018 2015 2012 2009 2006 2003 2000 1997 1994 1991 1988 1985 1982 1979 1976 1973 1970\n\nInsured losses (USD bn)\n\nExponential trend (total natcat losses) Primary perils (incl secondary effects) Secondary perils","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Appendix: facts and figures _sigma_ No 1/2025 **Swiss Re Institute** 27\n\n### Global insured losses\n\nHurricane Helene and Milton, the second highest severe convective storms (SCS) and large-scale urban floods were the main events driving global insured losses from natural disasters of USD 137 billion in 2024. Man-made disasters triggered an additional USD 9 billion losses.\n\n### Primary and secondary perils\n\nHurricane Helene and Milton brought the share of insured losses from so-called primary perils to roughly 41% in 2024, more than double the share in 2023 (19%).\n\n**Figure 18** Insured catastrophe losses, 1970 – 2024 (USD bn, 2024 prices)\n\nSource: Swiss Re Institute\n\n0\n\n50\n\n100\n\n150\n\n200\n\n250\n\n2024 2021 2018 2015 2012 2009 2006 2003 2000 1997 1994 1991 1988 1985 1982 1979 1976 1973 1970\n\nEarthquake/tsunami Weather-related catastrophes Man-made disasters\n\nHurricane Andrew\n\nWinter Storm Lothar WTC\n\nHurricanes Ike, Gustav\n\nHurricanes Ivan, Charley, Frances\n\nHurricanes Katrina, Rita, Wilma\n\nJapan, NZ earthquakes, Thailand flood\n\nHurricanes Harvey, Irma, Maria\n\nHurricane Ian\n\nHurricanes Helene, Milton\n\nHurricane Sandy\n\n**Figure 19** Global insured losses from primary and secondary perils, 1970 – 2024, in USD billion at 2024 prices\n\nSource: Swiss Re Institute\n\n0\n\n25\n\n50\n\n75\n\n100\n\n125\n\n150\n\n175\n\n200\n\n2024 2021 2018 2015 2012 2009 2006 2003 2000 1997 1994 1991 1988 1985 1982 1979 1976 1973 1970\n\nInsured losses (USD bn)\n\nExponential trend (total natcat losses) Primary perils (incl secondary effects) Secondary perils","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/sri-sigma-natural-catastrophes-1-2025_p27.pdf"]},"response":{"errors":[],"extract_id":"extract_03cf00239a8a942244686f3c714697f4","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Appendix: facts and figures _sigma_ No 1/2025 **Swiss Re Institute** 27\n\n### Global insured losses\n\nHurricane Helene and Milton, the second highest severe convective storms (SCS) and large-scale urban floods were the main events driving global insured losses from natural disasters of USD 137 billion in 2024. Man-made disasters triggered an additional USD 9 billion losses.\n\n### Primary and secondary perils\n\nHurricane Helene and Milton brought the share of insured losses from so-called primary perils to roughly 41% in 2024, more than double the share in 2023 (19%).\n\n**Figure 18** Insured catastrophe losses, 1970 – 2024 (USD bn, 2024 prices)\n\nSource: Swiss Re Institute\n\n0\n\n50\n\n100\n\n150\n\n200\n\n250\n\n2024 2021 2018 2015 2012 2009 2006 2003 2000 1997 1994 1991 1988 1985 1982 1979 1976 1973 1970\n\nEarthquake/tsunami Weather-related catastrophes Man-made disasters\n\nHurricane Andrew\n\nWinter Storm Lothar WTC\n\nHurricanes Ike, Gustav\n\nHurricanes Ivan, Charley, Frances\n\nHurricanes Katrina, Rita, Wilma\n\nJapan, NZ earthquakes, Thailand flood\n\nHurricanes Harvey, Irma, Maria\n\nHurricane Ian\n\nHurricanes Helene, Milton\n\nHurricane Sandy\n\n**Figure 19** Global insured losses from primary and secondary perils, 1970 – 2024, in USD billion at 2024 prices\n\nSource: Swiss Re Institute\n\n0\n\n25\n\n50\n\n75\n\n100\n\n125\n\n150\n\n175\n\n200\n\n2024 2021 2018 2015 2012 2009 2006 2003 2000 1997 1994 1991 1988 1985 1982 1979 1976 1973 1970\n\nInsured losses (USD bn)\n\nExponential trend (total natcat losses) Primary perils (incl secondary effects) Secondary perils"],"full_content":"Appendix: facts and figures _sigma_ No 1/2025 **Swiss Re Institute** 27\n\n### Global insured losses\n\nHurricane Helene and Milton, the second highest severe convective storms (SCS) and large-scale urban floods were the main events driving global insured losses from natural disasters of USD 137 billion in 2024. Man-made disasters triggered an additional USD 9 billion losses.\n\n### Primary and secondary perils\n\nHurricane Helene and Milton brought the share of insured losses from so-called primary perils to roughly 41% in 2024, more than double the share in 2023 (19%).\n\n**Figure 18** Insured catastrophe losses, 1970 – 2024 (USD bn, 2024 prices)\n\nSource: Swiss Re Institute\n\n0\n\n50\n\n100\n\n150\n\n200\n\n250\n\n2024 2021 2018 2015 2012 2009 2006 2003 2000 1997 1994 1991 1988 1985 1982 1979 1976 1973 1970\n\nEarthquake/tsunami Weather-related catastrophes Man-made disasters\n\nHurricane Andrew\n\nWinter Storm Lothar WTC\n\nHurricanes Ike, Gustav\n\nHurricanes Ivan, Charley, Frances\n\nHurricanes Katrina, Rita, Wilma\n\nJapan, NZ earthquakes, Thailand flood\n\nHurricanes Harvey, Irma, Maria\n\nHurricane Ian\n\nHurricanes Helene, Milton\n\nHurricane Sandy\n\n**Figure 19** Global insured losses from primary and secondary perils, 1970 – 2024, in USD billion at 2024 prices\n\nSource: Swiss Re Institute\n\n0\n\n25\n\n50\n\n75\n\n100\n\n125\n\n150\n\n175\n\n200\n\n2024 2021 2018 2015 2012 2009 2006 2003 2000 1997 1994 1991 1988 1985 1982 1979 1976 1973 1970\n\nInsured losses (USD bn)\n\nExponential trend (total natcat losses) Primary perils (incl secondary effects) Secondary perils","publish_date":null,"title":"Global insured losses Primary and secondary perils","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/sri-sigma-natural-catastrophes-1-2025_p27.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/sri-sigma-natural-catastrophes-1-2025_p27.pdf"},"request":{"example_id":"chart/sri-sigma-natural-catastrophes-1-2025_p27","product_type":"parse","source_file_path":"sri-sigma-natural-catastrophes-1-2025_p27.pdf"},"started_at":"2026-11-04T19:44:52.881454"} +{"completed_at":"2026-11-04T19:44:55.916653","latency_in_ms":1923,"output":{"example_id":"chart/sri-sigma-natural-catastrophes-1-2025_p28","grounded_pages":[],"layout_pages":[],"markdown":"28 **Swiss Re Institute** _sigma_ No 1/2025 Appendix: facts and figures\n\n### Total economic losses: USD 328 billion\n\nEconomic losses were highest in North America, driven by Hurricane Helene (USD 79 billion), and severe convective storms.\n\n### Global natural catastrophe protection gap: USD 181 billion\n\nFigure 20 shows global economic and insured losses over time. This highlights the insurance protection gap, ie the financial loss generated by natural catastrophes not covered by insurance. In 2024, the global natural catastrophe protection gap was around USD 181 billion, up from 177 in 2023 and from the previous 10-year average of USD 143 billion.\n\n**Table 2** Economic losses, in USD billion and as a % of global GDP, 2024\n\nNote: some percentages may not add up to 100 due to rounding.\n\nSource: Swiss Re Institute\n\n**Regions in USD bn in % of GDP**\n\nNorth America 212 0.69%\n\nEurope 33 0.12%\n\nAsia 67 0.17%\n\nLatin America 12 0.17%\n\nAfrica 2 0.07%\n\nOceania/Australia 2 0.11%\n\nWorld total 328 0.30%\n\n**Figure 20** Insured vs uninsured losses, 1970 – 2024, in USD billion at 2024 prices\n\nEconomic losses = insured + uninsured losses. Source: Swiss Re Institute\n\n0\n\n100\n\n200\n\n300\n\n400\n\n500\n\n600\n\n2024 2021 2018 2015 2012 2009 2006 2003 2000 1997 1994 1991 1988 1985 1982 1979 1976 1973 1970\n\nUninsured losses Insured losses","pages":[{"line_numbers":[],"markdown":"28 **Swiss Re Institute** _sigma_ No 1/2025 Appendix: facts and figures\n\n### Total economic losses: USD 328 billion\n\nEconomic losses were highest in North America, driven by Hurricane Helene (USD 79 billion), and severe convective storms.\n\n### Global natural catastrophe protection gap: USD 181 billion\n\nFigure 20 shows global economic and insured losses over time. This highlights the insurance protection gap, ie the financial loss generated by natural catastrophes not covered by insurance. In 2024, the global natural catastrophe protection gap was around USD 181 billion, up from 177 in 2023 and from the previous 10-year average of USD 143 billion.\n\n**Table 2** Economic losses, in USD billion and as a % of global GDP, 2024\n\nNote: some percentages may not add up to 100 due to rounding.\n\nSource: Swiss Re Institute\n\n**Regions in USD bn in % of GDP**\n\nNorth America 212 0.69%\n\nEurope 33 0.12%\n\nAsia 67 0.17%\n\nLatin America 12 0.17%\n\nAfrica 2 0.07%\n\nOceania/Australia 2 0.11%\n\nWorld total 328 0.30%\n\n**Figure 20** Insured vs uninsured losses, 1970 – 2024, in USD billion at 2024 prices\n\nEconomic losses = insured + uninsured losses. Source: Swiss Re Institute\n\n0\n\n100\n\n200\n\n300\n\n400\n\n500\n\n600\n\n2024 2021 2018 2015 2012 2009 2006 2003 2000 1997 1994 1991 1988 1985 1982 1979 1976 1973 1970\n\nUninsured losses Insured losses","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"28 **Swiss Re Institute** _sigma_ No 1/2025 Appendix: facts and figures\n\n### Total economic losses: USD 328 billion\n\nEconomic losses were highest in North America, driven by Hurricane Helene (USD 79 billion), and severe convective storms.\n\n### Global natural catastrophe protection gap: USD 181 billion\n\nFigure 20 shows global economic and insured losses over time. This highlights the insurance protection gap, ie the financial loss generated by natural catastrophes not covered by insurance. In 2024, the global natural catastrophe protection gap was around USD 181 billion, up from 177 in 2023 and from the previous 10-year average of USD 143 billion.\n\n**Table 2** Economic losses, in USD billion and as a % of global GDP, 2024\n\nNote: some percentages may not add up to 100 due to rounding.\n\nSource: Swiss Re Institute\n\n**Regions in USD bn in % of GDP**\n\nNorth America 212 0.69%\n\nEurope 33 0.12%\n\nAsia 67 0.17%\n\nLatin America 12 0.17%\n\nAfrica 2 0.07%\n\nOceania/Australia 2 0.11%\n\nWorld total 328 0.30%\n\n**Figure 20** Insured vs uninsured losses, 1970 – 2024, in USD billion at 2024 prices\n\nEconomic losses = insured + uninsured losses. Source: Swiss Re Institute\n\n0\n\n100\n\n200\n\n300\n\n400\n\n500\n\n600\n\n2024 2021 2018 2015 2012 2009 2006 2003 2000 1997 1994 1991 1988 1985 1982 1979 1976 1973 1970\n\nUninsured losses Insured losses","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/sri-sigma-natural-catastrophes-1-2025_p28.pdf"]},"response":{"errors":[],"extract_id":"extract_1f0ded8b8791a7180d8e1db1843eb15a","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["28 **Swiss Re Institute** _sigma_ No 1/2025 Appendix: facts and figures\n\n### Total economic losses: USD 328 billion\n\nEconomic losses were highest in North America, driven by Hurricane Helene (USD 79 billion), and severe convective storms.\n\n### Global natural catastrophe protection gap: USD 181 billion\n\nFigure 20 shows global economic and insured losses over time. This highlights the insurance protection gap, ie the financial loss generated by natural catastrophes not covered by insurance. In 2024, the global natural catastrophe protection gap was around USD 181 billion, up from 177 in 2023 and from the previous 10-year average of USD 143 billion.\n\n**Table 2** Economic losses, in USD billion and as a % of global GDP, 2024\n\nNote: some percentages may not add up to 100 due to rounding.\n\nSource: Swiss Re Institute\n\n**Regions in USD bn in % of GDP**\n\nNorth America 212 0.69%\n\nEurope 33 0.12%\n\nAsia 67 0.17%\n\nLatin America 12 0.17%\n\nAfrica 2 0.07%\n\nOceania/Australia 2 0.11%\n\nWorld total 328 0.30%\n\n**Figure 20** Insured vs uninsured losses, 1970 – 2024, in USD billion at 2024 prices\n\nEconomic losses = insured + uninsured losses. Source: Swiss Re Institute\n\n0\n\n100\n\n200\n\n300\n\n400\n\n500\n\n600\n\n2024 2021 2018 2015 2012 2009 2006 2003 2000 1997 1994 1991 1988 1985 1982 1979 1976 1973 1970\n\nUninsured losses Insured losses"],"full_content":"28 **Swiss Re Institute** _sigma_ No 1/2025 Appendix: facts and figures\n\n### Total economic losses: USD 328 billion\n\nEconomic losses were highest in North America, driven by Hurricane Helene (USD 79 billion), and severe convective storms.\n\n### Global natural catastrophe protection gap: USD 181 billion\n\nFigure 20 shows global economic and insured losses over time. This highlights the insurance protection gap, ie the financial loss generated by natural catastrophes not covered by insurance. In 2024, the global natural catastrophe protection gap was around USD 181 billion, up from 177 in 2023 and from the previous 10-year average of USD 143 billion.\n\n**Table 2** Economic losses, in USD billion and as a % of global GDP, 2024\n\nNote: some percentages may not add up to 100 due to rounding.\n\nSource: Swiss Re Institute\n\n**Regions in USD bn in % of GDP**\n\nNorth America 212 0.69%\n\nEurope 33 0.12%\n\nAsia 67 0.17%\n\nLatin America 12 0.17%\n\nAfrica 2 0.07%\n\nOceania/Australia 2 0.11%\n\nWorld total 328 0.30%\n\n**Figure 20** Insured vs uninsured losses, 1970 – 2024, in USD billion at 2024 prices\n\nEconomic losses = insured + uninsured losses. Source: Swiss Re Institute\n\n0\n\n100\n\n200\n\n300\n\n400\n\n500\n\n600\n\n2024 2021 2018 2015 2012 2009 2006 2003 2000 1997 1994 1991 1988 1985 1982 1979 1976 1973 1970\n\nUninsured losses Insured losses","publish_date":null,"title":"Total economic losses: USD 328 billion Global natural catastrophe protection gap: USD 181 billion","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/sri-sigma-natural-catastrophes-1-2025_p28.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/sri-sigma-natural-catastrophes-1-2025_p28.pdf"},"request":{"example_id":"chart/sri-sigma-natural-catastrophes-1-2025_p28","product_type":"parse","source_file_path":"sri-sigma-natural-catastrophes-1-2025_p28.pdf"},"started_at":"2026-11-04T19:44:53.993090"} +{"completed_at":"2026-11-04T19:44:56.289072","latency_in_ms":983,"output":{"example_id":"chart/sri-sigma-natural-catastrophes-1-2025_p4","grounded_pages":[],"layout_pages":[],"markdown":"4 **Swiss Re Institute** _sigma_ No 1/2025 Key takeaways\n\n### Losses from primary perils remain an ever present risk and drive most deviations from trend\n\nWhile losses from both primary and secondary perils are on an upward trajectory, history shows that annual losses from primary perils vary considerably from year-to-year, punctuated by peak loss years. In the past 30 years, there have been five such years where losses from primary perils were more than double the trend (>100% on X-axis). These years also happen to be years in terms of global losses from all natural catastrophe events were highest, affirming the strong influence of primary-peril losses.\n\nDistribution of annual losses from primary and secondary perils in % deviation from trend (1995 ‒ 2024)\n\nSource: Swiss Re Institute\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n700% 675% 650% 625% 600% 575% 550% 525% 500% 475% 450% 425% 400% 375% 350% 325% 300% 275% 250% 225% 200% 175% 150% 125% 100%\n\n75% 50% 25% 0% –25% –50% –75% –100% –125% –150%\n\nPrimary perils Secondary perils\n\nDeviations from tend\n\n2011 1999 2004 2005\n\n2011, 2017\n\nNumber of years\n\n### Exposure to natural catastrophe risks a key factor in driving claims costs\n\nAnalysis of the link between catastrophe losses per policy and homeowner insurance prices at state level shows a strong correlation. States with highest exposures to natural catastrophes risks (according to insurance loss history), are also where, in general, homeowner premiums are highest. In states with higher average premiums, extremely high premium levels are localised. Like in Florida, for instance, where average premium per policy is amongst the highest of all states. In 11% of the Florida’s zip codes, premiums per policy were above USD 10 000. These are typically coastal zip codes with highest risk of storm surge and/or wind damage.\n\nDistribution of premium per policy in Florida at zip-code level, USD, 2022\n\nSource: FIO Data, Swiss Re Institute\n\n0\n\n20\n\n40\n\n60\n\n80\n\n100\n\n120\n\n140\n\n>15000 14500 13500 12500 11500 10500 9500 8500 7500 6500 5500 4500 3500 2500 1500\n\nNumber of zip codes in Florida\n\nPremium per policy in USD","pages":[{"line_numbers":[],"markdown":"4 **Swiss Re Institute** _sigma_ No 1/2025 Key takeaways\n\n### Losses from primary perils remain an ever present risk and drive most deviations from trend\n\nWhile losses from both primary and secondary perils are on an upward trajectory, history shows that annual losses from primary perils vary considerably from year-to-year, punctuated by peak loss years. In the past 30 years, there have been five such years where losses from primary perils were more than double the trend (>100% on X-axis). These years also happen to be years in terms of global losses from all natural catastrophe events were highest, affirming the strong influence of primary-peril losses.\n\nDistribution of annual losses from primary and secondary perils in % deviation from trend (1995 ‒ 2024)\n\nSource: Swiss Re Institute\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n700% 675% 650% 625% 600% 575% 550% 525% 500% 475% 450% 425% 400% 375% 350% 325% 300% 275% 250% 225% 200% 175% 150% 125% 100%\n\n75% 50% 25% 0% –25% –50% –75% –100% –125% –150%\n\nPrimary perils Secondary perils\n\nDeviations from tend\n\n2011 1999 2004 2005\n\n2011, 2017\n\nNumber of years\n\n### Exposure to natural catastrophe risks a key factor in driving claims costs\n\nAnalysis of the link between catastrophe losses per policy and homeowner insurance prices at state level shows a strong correlation. States with highest exposures to natural catastrophes risks (according to insurance loss history), are also where, in general, homeowner premiums are highest. In states with higher average premiums, extremely high premium levels are localised. Like in Florida, for instance, where average premium per policy is amongst the highest of all states. In 11% of the Florida’s zip codes, premiums per policy were above USD 10 000. These are typically coastal zip codes with highest risk of storm surge and/or wind damage.\n\nDistribution of premium per policy in Florida at zip-code level, USD, 2022\n\nSource: FIO Data, Swiss Re Institute\n\n0\n\n20\n\n40\n\n60\n\n80\n\n100\n\n120\n\n140\n\n>15000 14500 13500 12500 11500 10500 9500 8500 7500 6500 5500 4500 3500 2500 1500\n\nNumber of zip codes in Florida\n\nPremium per policy in USD","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"4 **Swiss Re Institute** _sigma_ No 1/2025 Key takeaways\n\n### Losses from primary perils remain an ever present risk and drive most deviations from trend\n\nWhile losses from both primary and secondary perils are on an upward trajectory, history shows that annual losses from primary perils vary considerably from year-to-year, punctuated by peak loss years. In the past 30 years, there have been five such years where losses from primary perils were more than double the trend (>100% on X-axis). These years also happen to be years in terms of global losses from all natural catastrophe events were highest, affirming the strong influence of primary-peril losses.\n\nDistribution of annual losses from primary and secondary perils in % deviation from trend (1995 ‒ 2024)\n\nSource: Swiss Re Institute\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n700% 675% 650% 625% 600% 575% 550% 525% 500% 475% 450% 425% 400% 375% 350% 325% 300% 275% 250% 225% 200% 175% 150% 125% 100%\n\n75% 50% 25% 0% –25% –50% –75% –100% –125% –150%\n\nPrimary perils Secondary perils\n\nDeviations from tend\n\n2011 1999 2004 2005\n\n2011, 2017\n\nNumber of years\n\n### Exposure to natural catastrophe risks a key factor in driving claims costs\n\nAnalysis of the link between catastrophe losses per policy and homeowner insurance prices at state level shows a strong correlation. States with highest exposures to natural catastrophes risks (according to insurance loss history), are also where, in general, homeowner premiums are highest. In states with higher average premiums, extremely high premium levels are localised. Like in Florida, for instance, where average premium per policy is amongst the highest of all states. In 11% of the Florida’s zip codes, premiums per policy were above USD 10 000. These are typically coastal zip codes with highest risk of storm surge and/or wind damage.\n\nDistribution of premium per policy in Florida at zip-code level, USD, 2022\n\nSource: FIO Data, Swiss Re Institute\n\n0\n\n20\n\n40\n\n60\n\n80\n\n100\n\n120\n\n140\n\n>15000 14500 13500 12500 11500 10500 9500 8500 7500 6500 5500 4500 3500 2500 1500\n\nNumber of zip codes in Florida\n\nPremium per policy in USD","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/sri-sigma-natural-catastrophes-1-2025_p4.pdf"]},"response":{"errors":[],"extract_id":"extract_bfd772446051c284a27badd639bace76","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["4 **Swiss Re Institute** _sigma_ No 1/2025 Key takeaways\n\n### Losses from primary perils remain an ever present risk and drive most deviations from trend\n\nWhile losses from both primary and secondary perils are on an upward trajectory, history shows that annual losses from primary perils vary considerably from year-to-year, punctuated by peak loss years. In the past 30 years, there have been five such years where losses from primary perils were more than double the trend (>100% on X-axis). These years also happen to be years in terms of global losses from all natural catastrophe events were highest, affirming the strong influence of primary-peril losses.\n\nDistribution of annual losses from primary and secondary perils in % deviation from trend (1995 ‒ 2024)\n\nSource: Swiss Re Institute\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n700% 675% 650% 625% 600% 575% 550% 525% 500% 475% 450% 425% 400% 375% 350% 325% 300% 275% 250% 225% 200% 175% 150% 125% 100%\n\n75% 50% 25% 0% –25% –50% –75% –100% –125% –150%\n\nPrimary perils Secondary perils\n\nDeviations from tend\n\n2011 1999 2004 2005\n\n2011, 2017\n\nNumber of years\n\n### Exposure to natural catastrophe risks a key factor in driving claims costs\n\nAnalysis of the link between catastrophe losses per policy and homeowner insurance prices at state level shows a strong correlation. States with highest exposures to natural catastrophes risks (according to insurance loss history), are also where, in general, homeowner premiums are highest. In states with higher average premiums, extremely high premium levels are localised. Like in Florida, for instance, where average premium per policy is amongst the highest of all states. In 11% of the Florida’s zip codes, premiums per policy were above USD 10 000. These are typically coastal zip codes with highest risk of storm surge and/or wind damage.\n\nDistribution of premium per policy in Florida at zip-code level, USD, 2022\n\nSource: FIO Data, Swiss Re Institute\n\n0\n\n20\n\n40\n\n60\n\n80\n\n100\n\n120\n\n140\n\n>15000 14500 13500 12500 11500 10500 9500 8500 7500 6500 5500 4500 3500 2500 1500\n\nNumber of zip codes in Florida\n\nPremium per policy in USD"],"full_content":"4 **Swiss Re Institute** _sigma_ No 1/2025 Key takeaways\n\n### Losses from primary perils remain an ever present risk and drive most deviations from trend\n\nWhile losses from both primary and secondary perils are on an upward trajectory, history shows that annual losses from primary perils vary considerably from year-to-year, punctuated by peak loss years. In the past 30 years, there have been five such years where losses from primary perils were more than double the trend (>100% on X-axis). These years also happen to be years in terms of global losses from all natural catastrophe events were highest, affirming the strong influence of primary-peril losses.\n\nDistribution of annual losses from primary and secondary perils in % deviation from trend (1995 ‒ 2024)\n\nSource: Swiss Re Institute\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n12\n\n700% 675% 650% 625% 600% 575% 550% 525% 500% 475% 450% 425% 400% 375% 350% 325% 300% 275% 250% 225% 200% 175% 150% 125% 100%\n\n75% 50% 25% 0% –25% –50% –75% –100% –125% –150%\n\nPrimary perils Secondary perils\n\nDeviations from tend\n\n2011 1999 2004 2005\n\n2011, 2017\n\nNumber of years\n\n### Exposure to natural catastrophe risks a key factor in driving claims costs\n\nAnalysis of the link between catastrophe losses per policy and homeowner insurance prices at state level shows a strong correlation. States with highest exposures to natural catastrophes risks (according to insurance loss history), are also where, in general, homeowner premiums are highest. In states with higher average premiums, extremely high premium levels are localised. Like in Florida, for instance, where average premium per policy is amongst the highest of all states. In 11% of the Florida’s zip codes, premiums per policy were above USD 10 000. These are typically coastal zip codes with highest risk of storm surge and/or wind damage.\n\nDistribution of premium per policy in Florida at zip-code level, USD, 2022\n\nSource: FIO Data, Swiss Re Institute\n\n0\n\n20\n\n40\n\n60\n\n80\n\n100\n\n120\n\n140\n\n>15000 14500 13500 12500 11500 10500 9500 8500 7500 6500 5500 4500 3500 2500 1500\n\nNumber of zip codes in Florida\n\nPremium per policy in USD","publish_date":null,"title":"Losses from primary perils remain an ever present risk and drive most deviations from trend Exposure to natural catastrophe risks","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/sri-sigma-natural-catastrophes-1-2025_p4.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/sri-sigma-natural-catastrophes-1-2025_p4.pdf"},"request":{"example_id":"chart/sri-sigma-natural-catastrophes-1-2025_p4","product_type":"parse","source_file_path":"sri-sigma-natural-catastrophes-1-2025_p4.pdf"},"started_at":"2026-11-04T19:44:55.305619"} +{"completed_at":"2026-11-04T19:44:57.700037","latency_in_ms":1396,"output":{"example_id":"chart/sri-sigma-natural-catastrophes-1-2025_p7","grounded_pages":[],"layout_pages":[],"markdown":"Insured losses on trend _sigma_ No 1/2025 **Swiss Re Institute** 7\n\nAdaptation measures like strengthening building codes or flood protection infrastructure help reduce exposures, but often such actions are insufficient to compensate for rising losses from value accumulation. A case in point is that even with major advancements in building codes, in 2022 Hurricane Ian still inflicted insured losses of USD 64 billion (in 2024 prices). ⁵ The main driver of these losses was the 620% growth in population in the landfall area since 1975. ⁶ Losses can be even more severe when the establishment of loss reduction measures lags behind economic growth, as seen for instance in the case of insufficient drainage infrastructure in the UAE and also in southern Spain during last year’s floodings there. New technologies can also introduce adverse developments to vulnerability. For example, the growing number of solar power installations on roofs, often the most exposed parts of a building, has been one of the driving forces behind the steep rise in SCS losses in recent years. ⁷\n\n### 2024 in more detail\n\nThe biggest single loss events of 2024 were Hurricanes Helene and Milton in the US. Coming in quick succession in September and October, they triggered combined insured losses of USD 44 billion. Helene unleashed huge flooding in mountainous areas across Appalachia, inflicting much damage in areas where flood insurance is not mandatory. Combined economic losses from the two hurricanes were USD 113 billion, indicating economic losses of more than double the insured losses. Less than 2% of households in Georgia are currently covered by the National Flood Insurance Program (NFIP), and less than 1% in the hardest-hit areas like Buncombe County, North Carolina, including Asheville, the biggest city in the region. Coverage for North Carolina overall is around 3%, and for South Carolina around 9%. ⁸ Hurricane Milton, though a relatively small tropical cyclone (TC) loss event, coming soon after Helene is a reminder of the potential for large losses if two hurricanes strike a city like Tampa or Miami in quick succession.\n\nOnce again SCS were main drivers of last year’s losses, with the overall cost to the global insurance industry mounting to USD 53 billion (see Figure 3). It was the second year running that SCS-related losses were above USD 50 billion. It was also the second costliest year for this peril after 2023, when insured losses were USD 70 billion (in 2024 prices). The US continues to see the largest SCS-related losses, but these are rising in other countries also. For instance, with insured losses of USD 2.3 billion, the hailstorm in Calgary last year was the costliest ever seen in Canada. SCS-triggered insured losses are\n\n5 _Ian revisited: Disentangling the drivers of US hurricane losses,_ Swiss Re, 25 September 2023 bid.\n\n6 _sigma_ 1/2023: Natural catastrophes and inflation; a perfect storm, Swiss Re Institute.\n\n7 _sigma_ 1/2024, Swiss Re Institute, op. cit.\n\n8 See _Flood Insurance Data and Analytics_ , FEMA.\n\nAdaptation measures can help temper losses.\n\nHurricane Helene led to large-scale inundation and losses in areas where flood insurance is not mandatory.\n\n**Figure 3** Global insured losses from natural catastrophes in 2024 and previous year averages (USD bn, 2024 prices)\n\nSource: Swiss Re Institute\n\n0 10 20 30 40 50 60 70 80 90 100 110 120 130 140\n\n2024\n\n5-year average (2019-2023)\n\n10-year average (2014-2023)\n\n15-year average (2009-2023)\n\n30-year average (1994-2023)\n\nEarthquakes European winter storms Tropical cyclones Wildfires Severe convective storms\n\nFloods Other secondary perils\n\nSCS continued their recent trend of causing elevated losses.","pages":[{"line_numbers":[],"markdown":"Insured losses on trend _sigma_ No 1/2025 **Swiss Re Institute** 7\n\nAdaptation measures like strengthening building codes or flood protection infrastructure help reduce exposures, but often such actions are insufficient to compensate for rising losses from value accumulation. A case in point is that even with major advancements in building codes, in 2022 Hurricane Ian still inflicted insured losses of USD 64 billion (in 2024 prices). �� The main driver of these losses was the 620% growth in population in the landfall area since 1975. ⁶ Losses can be even more severe when the establishment of loss reduction measures lags behind economic growth, as seen for instance in the case of insufficient drainage infrastructure in the UAE and also in southern Spain during last year’s floodings there. New technologies can also introduce adverse developments to vulnerability. For example, the growing number of solar power installations on roofs, often the most exposed parts of a building, has been one of the driving forces behind the steep rise in SCS losses in recent years. ⁷\n\n### 2024 in more detail\n\nThe biggest single loss events of 2024 were Hurricanes Helene and Milton in the US. Coming in quick succession in September and October, they triggered combined insured losses of USD 44 billion. Helene unleashed huge flooding in mountainous areas across Appalachia, inflicting much damage in areas where flood insurance is not mandatory. Combined economic losses from the two hurricanes were USD 113 billion, indicating economic losses of more than double the insured losses. Less than 2% of households in Georgia are currently covered by the National Flood Insurance Program (NFIP), and less than 1% in the hardest-hit areas like Buncombe County, North Carolina, including Asheville, the biggest city in the region. Coverage for North Carolina overall is around 3%, and for South Carolina around 9%. ⁸ Hurricane Milton, though a relatively small tropical cyclone (TC) loss event, coming soon after Helene is a reminder of the potential for large losses if two hurricanes strike a city like Tampa or Miami in quick succession.\n\nOnce again SCS were main drivers of last year’s losses, with the overall cost to the global insurance industry mounting to USD 53 billion (see Figure 3). It was the second year running that SCS-related losses were above USD 50 billion. It was also the second costliest year for this peril after 2023, when insured losses were USD 70 billion (in 2024 prices). The US continues to see the largest SCS-related losses, but these are rising in other countries also. For instance, with insured losses of USD 2.3 billion, the hailstorm in Calgary last year was the costliest ever seen in Canada. SCS-triggered insured losses are\n\n5 _Ian revisited: Disentangling the drivers of US hurricane losses,_ Swiss Re, 25 September 2023 bid.\n\n6 _sigma_ 1/2023: Natural catastrophes and inflation; a perfect storm, Swiss Re Institute.\n\n7 _sigma_ 1/2024, Swiss Re Institute, op. cit.\n\n8 See _Flood Insurance Data and Analytics_ , FEMA.\n\nAdaptation measures can help temper losses.\n\nHurricane Helene led to large-scale inundation and losses in areas where flood insurance is not mandatory.\n\n**Figure 3** Global insured losses from natural catastrophes in 2024 and previous year averages (USD bn, 2024 prices)\n\nSource: Swiss Re Institute\n\n0 10 20 30 40 50 60 70 80 90 100 110 120 130 140\n\n2024\n\n5-year average (2019-2023)\n\n10-year average (2014-2023)\n\n15-year average (2009-2023)\n\n30-year average (1994-2023)\n\nEarthquakes European winter storms Tropical cyclones Wildfires Severe convective storms\n\nFloods Other secondary perils\n\nSCS continued their recent trend of causing elevated losses.","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"Insured losses on trend _sigma_ No 1/2025 **Swiss Re Institute** 7\n\nAdaptation measures like strengthening building codes or flood protection infrastructure help reduce exposures, but often such actions are insufficient to compensate for rising losses from value accumulation. A case in point is that even with major advancements in building codes, in 2022 Hurricane Ian still inflicted insured losses of USD 64 billion (in 2024 prices). ⁵ The main driver of these losses was the 620% growth in population in the landfall area since 1975. ⁶ Losses can be even more severe when the establishment of loss reduction measures lags behind economic growth, as seen for instance in the case of insufficient drainage infrastructure in the UAE and also in southern Spain during last year’s floodings there. New technologies can also introduce adverse developments to vulnerability. For example, the growing number of solar power installations on roofs, often the most exposed parts of a building, has been one of the driving forces behind the steep rise in SCS losses in recent years. ⁷\n\n### 2024 in more detail\n\nThe biggest single loss events of 2024 were Hurricanes Helene and Milton in the US. Coming in quick succession in September and October, they triggered combined insured losses of USD 44 billion. Helene unleashed huge flooding in mountainous areas across Appalachia, inflicting much damage in areas where flood insurance is not mandatory. Combined economic losses from the two hurricanes were USD 113 billion, indicating economic losses of more than double the insured losses. Less than 2% of households in Georgia are currently covered by the National Flood Insurance Program (NFIP), and less than 1% in the hardest-hit areas like Buncombe County, North Carolina, including Asheville, the biggest city in the region. Coverage for North Carolina overall is around 3%, and for South Carolina around 9%. ⁸ Hurricane Milton, though a relatively small tropical cyclone (TC) loss event, coming soon after Helene is a reminder of the potential for large losses if two hurricanes strike a city like Tampa or Miami in quick succession.\n\nOnce again SCS were main drivers of last year’s losses, with the overall cost to the global insurance industry mounting to USD 53 billion (see Figure 3). It was the second year running that SCS-related losses were above USD 50 billion. It was also the second costliest year for this peril after 2023, when insured losses were USD 70 billion (in 2024 prices). The US continues to see the largest SCS-related losses, but these are rising in other countries also. For instance, with insured losses of USD 2.3 billion, the hailstorm in Calgary last year was the costliest ever seen in Canada. SCS-triggered insured losses are\n\n5 _Ian revisited: Disentangling the drivers of US hurricane losses,_ Swiss Re, 25 September 2023 bid.\n\n6 _sigma_ 1/2023: Natural catastrophes and inflation; a perfect storm, Swiss Re Institute.\n\n7 _sigma_ 1/2024, Swiss Re Institute, op. cit.\n\n8 See _Flood Insurance Data and Analytics_ , FEMA.\n\nAdaptation measures can help temper losses.\n\nHurricane Helene led to large-scale inundation and losses in areas where flood insurance is not mandatory.\n\n**Figure 3** Global insured losses from natural catastrophes in 2024 and previous year averages (USD bn, 2024 prices)\n\nSource: Swiss Re Institute\n\n0 10 20 30 40 50 60 70 80 90 100 110 120 130 140\n\n2024\n\n5-year average (2019-2023)\n\n10-year average (2014-2023)\n\n15-year average (2009-2023)\n\n30-year average (1994-2023)\n\nEarthquakes European winter storms Tropical cyclones Wildfires Severe convective storms\n\nFloods Other secondary perils\n\nSCS continued their recent trend of causing elevated losses.","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/sri-sigma-natural-catastrophes-1-2025_p7.pdf"]},"response":{"errors":[],"extract_id":"extract_f5e307a7b53f770173636d32bd16f065","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["Insured losses on trend _sigma_ No 1/2025 **Swiss Re Institute** 7\n\nAdaptation measures like strengthening building codes or flood protection infrastructure help reduce exposures, but often such actions are insufficient to compensate for rising losses from value accumulation. A case in point is that even with major advancements in building codes, in 2022 Hurricane Ian still inflicted insured losses of USD 64 billion (in 2024 prices). ⁵ The main driver of these losses was the 620% growth in population in the landfall area since 1975. ⁶ Losses can be even more severe when the establishment of loss reduction measures lags behind economic growth, as seen for instance in the case of insufficient drainage infrastructure in the UAE and also in southern Spain during last year’s floodings there. New technologies can also introduce adverse developments to vulnerability. For example, the growing number of solar power installations on roofs, often the most exposed parts of a building, has been one of the driving forces behind the steep rise in SCS losses in recent years. ⁷\n\n### 2024 in more detail\n\nThe biggest single loss events of 2024 were Hurricanes Helene and Milton in the US. Coming in quick succession in September and October, they triggered combined insured losses of USD 44 billion. Helene unleashed huge flooding in mountainous areas across Appalachia, inflicting much damage in areas where flood insurance is not mandatory. Combined economic losses from the two hurricanes were USD 113 billion, indicating economic losses of more than double the insured losses. Less than 2% of households in Georgia are currently covered by the National Flood Insurance Program (NFIP), and less than 1% in the hardest-hit areas like Buncombe County, North Carolina, including Asheville, the biggest city in the region. Coverage for North Carolina overall is around 3%, and for South Carolina around 9%. ⁸ Hurricane Milton, though a relatively small tropical cyclone (TC) loss event, coming soon after Helene is a reminder of the potential for large losses if two hurricanes strike a city like Tampa or Miami in quick succession.\n\nOnce again SCS were main drivers of last year’s losses, with the overall cost to the global insurance industry mounting to USD 53 billion (see Figure 3). It was the second year running that SCS-related losses were above USD 50 billion. It was also the second costliest year for this peril after 2023, when insured losses were USD 70 billion (in 2024 prices). The US continues to see the largest SCS-related losses, but these are rising in other countries also. For instance, with insured losses of USD 2.3 billion, the hailstorm in Calgary last year was the costliest ever seen in Canada. SCS-triggered insured losses are\n\n5 _Ian revisited: Disentangling the drivers of US hurricane losses,_ Swiss Re, 25 September 2023 bid.\n\n6 _sigma_ 1/2023: Natural catastrophes and inflation; a perfect storm, Swiss Re Institute.\n\n7 _sigma_ 1/2024, Swiss Re Institute, op. cit.\n\n8 See _Flood Insurance Data and Analytics_ , FEMA.\n\nAdaptation measures can help temper losses.\n\nHurricane Helene led to large-scale inundation and losses in areas where flood insurance is not mandatory.\n\n**Figure 3** Global insured losses from natural catastrophes in 2024 and previous year averages (USD bn, 2024 prices)\n\nSource: Swiss Re Institute\n\n0 10 20 30 40 50 60 70 80 90 100 110 120 130 140\n\n2024\n\n5-year average (2019-2023)\n\n10-year average (2014-2023)\n\n15-year average (2009-2023)\n\n30-year average (1994-2023)\n\nEarthquakes European winter storms Tropical cyclones Wildfires Severe convective storms\n\nFloods Other secondary perils\n\nSCS continued their recent trend of causing elevated losses."],"full_content":"Insured losses on trend _sigma_ No 1/2025 **Swiss Re Institute** 7\n\nAdaptation measures like strengthening building codes or flood protection infrastructure help reduce exposures, but often such actions are insufficient to compensate for rising losses from value accumulation. A case in point is that even with major advancements in building codes, in 2022 Hurricane Ian still inflicted insured losses of USD 64 billion (in 2024 prices). ⁵ The main driver of these losses was the 620% growth in population in the landfall area since 1975. ⁶ Losses can be even more severe when the establishment of loss reduction measures lags behind economic growth, as seen for instance in the case of insufficient drainage infrastructure in the UAE and also in southern Spain during last year’s floodings there. New technologies can also introduce adverse developments to vulnerability. For example, the growing number of solar power installations on roofs, often the most exposed parts of a building, has been one of the driving forces behind the steep rise in SCS losses in recent years. ⁷\n\n### 2024 in more detail\n\nThe biggest single loss events of 2024 were Hurricanes Helene and Milton in the US. Coming in quick succession in September and October, they triggered combined insured losses of USD 44 billion. Helene unleashed huge flooding in mountainous areas across Appalachia, inflicting much damage in areas where flood insurance is not mandatory. Combined economic losses from the two hurricanes were USD 113 billion, indicating economic losses of more than double the insured losses. Less than 2% of households in Georgia are currently covered by the National Flood Insurance Program (NFIP), and less than 1% in the hardest-hit areas like Buncombe County, North Carolina, including Asheville, the biggest city in the region. Coverage for North Carolina overall is around 3%, and for South Carolina around 9%. ⁸ Hurricane Milton, though a relatively small tropical cyclone (TC) loss event, coming soon after Helene is a reminder of the potential for large losses if two hurricanes strike a city like Tampa or Miami in quick succession.\n\nOnce again SCS were main drivers of last year’s losses, with the overall cost to the global insurance industry mounting to USD 53 billion (see Figure 3). It was the second year running that SCS-related losses were above USD 50 billion. It was also the second costliest year for this peril after 2023, when insured losses were USD 70 billion (in 2024 prices). The US continues to see the largest SCS-related losses, but these are rising in other countries also. For instance, with insured losses of USD 2.3 billion, the hailstorm in Calgary last year was the costliest ever seen in Canada. SCS-triggered insured losses are\n\n5 _Ian revisited: Disentangling the drivers of US hurricane losses,_ Swiss Re, 25 September 2023 bid.\n\n6 _sigma_ 1/2023: Natural catastrophes and inflation; a perfect storm, Swiss Re Institute.\n\n7 _sigma_ 1/2024, Swiss Re Institute, op. cit.\n\n8 See _Flood Insurance Data and Analytics_ , FEMA.\n\nAdaptation measures can help temper losses.\n\nHurricane Helene led to large-scale inundation and losses in areas where flood insurance is not mandatory.\n\n**Figure 3** Global insured losses from natural catastrophes in 2024 and previous year averages (USD bn, 2024 prices)\n\nSource: Swiss Re Institute\n\n0 10 20 30 40 50 60 70 80 90 100 110 120 130 140\n\n2024\n\n5-year average (2019-2023)\n\n10-year average (2014-2023)\n\n15-year average (2009-2023)\n\n30-year average (1994-2023)\n\nEarthquakes European winter storms Tropical cyclones Wildfires Severe convective storms\n\nFloods Other secondary perils\n\nSCS continued their recent trend of causing elevated losses.","publish_date":null,"title":"2024 in more detail","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/sri-sigma-natural-catastrophes-1-2025_p7.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/sri-sigma-natural-catastrophes-1-2025_p7.pdf"},"request":{"example_id":"chart/sri-sigma-natural-catastrophes-1-2025_p7","product_type":"parse","source_file_path":"sri-sigma-natural-catastrophes-1-2025_p7.pdf"},"started_at":"2026-11-04T19:44:56.303072"} +{"completed_at":"2026-11-04T19:44:58.924134","latency_in_ms":1502,"output":{"example_id":"chart/state-of-the-semiconductor-industry-report_p28","grounded_pages":[],"layout_pages":[],"markdown":"The race for scale in AI, which peaked between 2019 and 2021, has slowed, with Google’s 1.6 trillion parameter Switch-C setting the record back in 2021. ¹¹ Although GPT-4 and similar models have not surpassed this record, their typical model size is increasing. Since Meta’s LLaMa model leak in February 2023 ¹² , the proliferation of smaller AI models (five to 70 billion parameters) has surged, particularly in the open-source community. This “missing middle” category is now a hotbed for innovation, as these models are small enough for researchers and individuals to explore with high-end personal hardware.\n\n**Transforming data centers with AI accelerators** AI adoption, particularly the shift from predictive to generative AI, is leading to burgeoning demand for semiconductors, specifically AI accelerators and high-bandwidth memory (HBM). Traditional machine learning models and smaller neural networks in the sub-500M weight class, such as YOLO-v5, are being replaced by larger models in the 5B–70B weight class, such as Phi, Gemma, and Mistral-7B. As AI computing demand surges, NVIDIA’s data center business is expected to generate more than $80 billion this year, despite not all revenue being strictly related to compute or AI. This projection is particularly striking given that, prior to its standout 2Q23 results, the company’s total revenue was less than half of that figure.\n\nThe rapid growth in demand for AI silicon, especially in data centers, has shifted the traditional relationship between CPUs and co-processors such as GPUs or domain-specific ASICs. Increasingly, CPUs are becoming the business logic co-processors for GPUs, or are acting as orchestrators for a diverse range of specialized accelerators ( _see Exhibit 15_ ).\n\n**Exhibit 15 AI processors market for cloud and data center, 2021** **–** **2029 ($bn)**\n\n**50%**\n\n**74%**\n\n**80%**\n\n**116**\n\n**84%**\n\n**130**\n\n**84%**\n\n**141**\n\n**85%**\n\n**150**\n\n**81%**\n\n**152**\n\n**93**\n\n**35**\n\n**6**\n\n2021\n\n**10**\n\n2022 2023 2024F 2025F 2026F 2027F 2028F 2029F\n\nAI processor revenue\n\n% of data center TAM\n\nSource: Omdia Q3 2024\n\nPwC | State of the Semiconductor Industry 28","pages":[{"line_numbers":[],"markdown":"The race for scale in AI, which peaked between 2019 and 2021, has slowed, with Google’s 1.6 trillion parameter Switch-C setting the record back in 2021. ¹¹ Although GPT-4 and similar models have not surpassed this record, their typical model size is increasing. Since Meta’s LLaMa model leak in February 2023 ¹² , the proliferation of smaller AI models (five to 70 billion parameters) has surged, particularly in the open-source community. This “missing middle” category is now a hotbed for innovation, as these models are small enough for researchers and individuals to explore with high-end personal hardware.\n\n**Transforming data centers with AI accelerators** AI adoption, particularly the shift from predictive to generative AI, is leading to burgeoning demand for semiconductors, specifically AI accelerators and high-bandwidth memory (HBM). Traditional machine learning models and smaller neural networks in the sub-500M weight class, such as YOLO-v5, are being replaced by larger models in the 5B–70B weight class, such as Phi, Gemma, and Mistral-7B. As AI computing demand surges, NVIDIA’s data center business is expected to generate more than $80 billion this year, despite not all revenue being strictly related to compute or AI. This projection is particularly striking given that, prior to its standout 2Q23 results, the company’s total revenue was less than half of that figure.\n\nThe rapid growth in demand for AI silicon, especially in data centers, has shifted the traditional relationship between CPUs and co-processors such as GPUs or domain-specific ASICs. Increasingly, CPUs are becoming the business logic co-processors for GPUs, or are acting as orchestrators for a diverse range of specialized accelerators ( _see Exhibit 15_ ).\n\n**Exhibit 15 AI processors market for cloud and data center, 2021** **–** **2029 ($bn)**\n\n**50%**\n\n**74%**\n\n**80%**\n\n**116**\n\n**84%**\n\n**130**\n\n**84%**\n\n**141**\n\n**85%**\n\n**150**\n\n**81%**\n\n**152**\n\n**93**\n\n**35**\n\n**6**\n\n2021\n\n**10**\n\n2022 2023 2024F 2025F 2026F 2027F 2028F 2029F\n\nAI processor revenue\n\n% of data center TAM\n\nSource: Omdia Q3 2024\n\nPwC | State of the Semiconductor Industry 28","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"The race for scale in AI, which peaked between 2019 and 2021, has slowed, with Google’s 1.6 trillion parameter Switch-C setting the record back in 2021. ¹¹ Although GPT-4 and similar models have not surpassed this record, their typical model size is increasing. Since Meta’s LLaMa model leak in February 2023 ¹² , the proliferation of smaller AI models (five to 70 billion parameters) has surged, particularly in the open-source community. This “missing middle” category is now a hotbed for innovation, as these models are small enough for researchers and individuals to explore with high-end personal hardware.\n\n**Transforming data centers with AI accelerators** AI adoption, particularly the shift from predictive to generative AI, is leading to burgeoning demand for semiconductors, specifically AI accelerators and high-bandwidth memory (HBM). Traditional machine learning models and smaller neural networks in the sub-500M weight class, such as YOLO-v5, are being replaced by larger models in the 5B–70B weight class, such as Phi, Gemma, and Mistral-7B. As AI computing demand surges, NVIDIA’s data center business is expected to generate more than $80 billion this year, despite not all revenue being strictly related to compute or AI. This projection is particularly striking given that, prior to its standout 2Q23 results, the company’s total revenue was less than half of that figure.\n\nThe rapid growth in demand for AI silicon, especially in data centers, has shifted the traditional relationship between CPUs and co-processors such as GPUs or domain-specific ASICs. Increasingly, CPUs are becoming the business logic co-processors for GPUs, or are acting as orchestrators for a diverse range of specialized accelerators ( _see Exhibit 15_ ).\n\n**Exhibit 15 AI processors market for cloud and data center, 2021** **–** **2029 ($bn)**\n\n**50%**\n\n**74%**\n\n**80%**\n\n**116**\n\n**84%**\n\n**130**\n\n**84%**\n\n**141**\n\n**85%**\n\n**150**\n\n**81%**\n\n**152**\n\n**93**\n\n**35**\n\n**6**\n\n2021\n\n**10**\n\n2022 2023 2024F 2025F 2026F 2027F 2028F 2029F\n\nAI processor revenue\n\n% of data center TAM\n\nSource: Omdia Q3 2024\n\nPwC | State of the Semiconductor Industry 28","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/state-of-the-semiconductor-industry-report_p28.pdf"]},"response":{"errors":[],"extract_id":"extract_898d847ea849f48f9a4b04824e2fd1ff","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["The race for scale in AI, which peaked between 2019 and 2021, has slowed, with Google’s 1.6 trillion parameter Switch-C setting the record back in 2021. ¹¹ Although GPT-4 and similar models have not surpassed this record, their typical model size is increasing. Since Meta’s LLaMa model leak in February 2023 ¹² , the proliferation of smaller AI models (five to 70 billion parameters) has surged, particularly in the open-source community. This “missing middle” category is now a hotbed for innovation, as these models are small enough for researchers and individuals to explore with high-end personal hardware.\n\n**Transforming data centers with AI accelerators** AI adoption, particularly the shift from predictive to generative AI, is leading to burgeoning demand for semiconductors, specifically AI accelerators and high-bandwidth memory (HBM). Traditional machine learning models and smaller neural networks in the sub-500M weight class, such as YOLO-v5, are being replaced by larger models in the 5B–70B weight class, such as Phi, Gemma, and Mistral-7B. As AI computing demand surges, NVIDIA’s data center business is expected to generate more than $80 billion this year, despite not all revenue being strictly related to compute or AI. This projection is particularly striking given that, prior to its standout 2Q23 results, the company’s total revenue was less than half of that figure.\n\nThe rapid growth in demand for AI silicon, especially in data centers, has shifted the traditional relationship between CPUs and co-processors such as GPUs or domain-specific ASICs. Increasingly, CPUs are becoming the business logic co-processors for GPUs, or are acting as orchestrators for a diverse range of specialized accelerators ( _see Exhibit 15_ ).\n\n**Exhibit 15 AI processors market for cloud and data center, 2021** **–** **2029 ($bn)**\n\n**50%**\n\n**74%**\n\n**80%**\n\n**116**\n\n**84%**\n\n**130**\n\n**84%**\n\n**141**\n\n**85%**\n\n**150**\n\n**81%**\n\n**152**\n\n**93**\n\n**35**\n\n**6**\n\n2021\n\n**10**\n\n2022 2023 2024F 2025F 2026F 2027F 2028F 2029F\n\nAI processor revenue\n\n% of data center TAM\n\nSource: Omdia Q3 2024\n\nPwC | State of the Semiconductor Industry 28"],"full_content":"The race for scale in AI, which peaked between 2019 and 2021, has slowed, with Google’s 1.6 trillion parameter Switch-C setting the record back in 2021. ¹¹ Although GPT-4 and similar models have not surpassed this record, their typical model size is increasing. Since Meta’s LLaMa model leak in February 2023 ¹² , the proliferation of smaller AI models (five to 70 billion parameters) has surged, particularly in the open-source community. This “missing middle” category is now a hotbed for innovation, as these models are small enough for researchers and individuals to explore with high-end personal hardware.\n\n**Transforming data centers with AI accelerators** AI adoption, particularly the shift from predictive to generative AI, is leading to burgeoning demand for semiconductors, specifically AI accelerators and high-bandwidth memory (HBM). Traditional machine learning models and smaller neural networks in the sub-500M weight class, such as YOLO-v5, are being replaced by larger models in the 5B–70B weight class, such as Phi, Gemma, and Mistral-7B. As AI computing demand surges, NVIDIA’s data center business is expected to generate more than $80 billion this year, despite not all revenue being strictly related to compute or AI. This projection is particularly striking given that, prior to its standout 2Q23 results, the company’s total revenue was less than half of that figure.\n\nThe rapid growth in demand for AI silicon, especially in data centers, has shifted the traditional relationship between CPUs and co-processors such as GPUs or domain-specific ASICs. Increasingly, CPUs are becoming the business logic co-processors for GPUs, or are acting as orchestrators for a diverse range of specialized accelerators ( _see Exhibit 15_ ).\n\n**Exhibit 15 AI processors market for cloud and data center, 2021** **–** **2029 ($bn)**\n\n**50%**\n\n**74%**\n\n**80%**\n\n**116**\n\n**84%**\n\n**130**\n\n**84%**\n\n**141**\n\n**85%**\n\n**150**\n\n**81%**\n\n**152**\n\n**93**\n\n**35**\n\n**6**\n\n2021\n\n**10**\n\n2022 2023 2024F 2025F 2026F 2027F 2028F 2029F\n\nAI processor revenue\n\n% of data center TAM\n\nSource: Omdia Q3 2024\n\nPwC | State of the Semiconductor Industry 28","publish_date":null,"title":"The race for scale in AI, which peaked between 2019 and 2021, has slowed, with Google’s 1.6 trillion parameter Switch-C","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/state-of-the-semiconductor-industry-report_p28.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/state-of-the-semiconductor-industry-report_p28.pdf"},"request":{"example_id":"chart/state-of-the-semiconductor-industry-report_p28","product_type":"parse","source_file_path":"state-of-the-semiconductor-industry-report_p28.pdf"},"started_at":"2026-11-04T19:44:57.421954"} +{"completed_at":"2026-11-04T19:45:01.813662","latency_in_ms":3036,"output":{"example_id":"chart/tdr2023ch1_en_p19","grounded_pages":[],"layout_pages":[],"markdown":"19\n\nCurrent Trends and Challenges in the Global Economy\n\n# F. INFLATION AND DISTRIBUTION\n\nAfter 2020, inflation accelerated along similar trajectories in most countries, but in 2023, as inflation began to decrease, the paths diverged.\n\nSigns of an inflation pick-up began to appear in the United States in the early months of 2021. Various temporary and lasting triggers contributed to this. These include four main factors: (a) changes to global trade patterns that impacted import costs; (b) a surge in consumption expenditure by the wealthy, who benefited from the stock market gains fuelled by extra-loose monetary policy; (c) small increases in the real wages of the lowest paid occupations; and (d) the ability of producers and retailers to raise prices in order to recoup cost increases and increase profit margins (Bivens, 2022; Konczal and Lusiani, 2022; Hayes and Jung, 2022; Schnabel, 2022; Storm, 2022; Weber and Wasner, 2023).\n\nConcerns that inflation was extending beyond the expected transitory period that would normally accompany recovery from a deep economic shock only began towards the end of 2021, as an initial easing of rising prices was reversed. Failure to distribute effective vaccines worldwide prolonged the pandemic, causing temporary factors to linger on, eventually interacting with a largely anticipated initial increase in commodity prices. Then, as the war in Ukraine began, some commodity prices spiked, raising inflation rates further, especially in the European Union (see TDR, 2022 for a discussion). Despite the supply-side origins of this new round of inflationary pressure, leading central banks, beginning with the United States Federal Reserve, embarked on monetary tightening sooner than had been previously signalled.\n\n**Figure I.8 Inflation rates have remained in line with historical standards in most developing countries** Monthly consumer price index growth, selected developing countries (Year-on-year percentage change)\n\n2006 0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\n70\n\n80\n\n90\n\n\\-5\n\n0\n\n5\n\n10\n\n15\n\n20\n\n10 15 20\n\nChile Mexico Brazil China India Indonesia\n\nSouth Africa Türkiye (right axis)\n\nJul. 23\n\n_Source:_ OECD Statistics and national sources.\n\n## **I**","pages":[{"line_numbers":[],"markdown":"19\n\nCurrent Trends and Challenges in the Global Economy\n\n# F. INFLATION AND DISTRIBUTION\n\nAfter 2020, inflation accelerated along similar trajectories in most countries, but in 2023, as inflation began to decrease, the paths diverged.\n\nSigns of an inflation pick-up began to appear in the United States in the early months of 2021. Various temporary and lasting triggers contributed to this. These include four main factors: (a) changes to global trade patterns that impacted import costs; (b) a surge in consumption expenditure by the wealthy, who benefited from the stock market gains fuelled by extra-loose monetary policy; (c) small increases in the real wages of the lowest paid occupations; and (d) the ability of producers and retailers to raise prices in order to recoup cost increases and increase profit margins (Bivens, 2022; Konczal and Lusiani, 2022; Hayes and Jung, 2022; Schnabel, 2022; Storm, 2022; Weber and Wasner, 2023).\n\nConcerns that inflation was extending beyond the expected transitory period that would normally accompany recovery from a deep economic shock only began towards the end of 2021, as an initial easing of rising prices was reversed. Failure to distribute effective vaccines worldwide prolonged the pandemic, causing temporary factors to linger on, eventually interacting with a largely anticipated initial increase in commodity prices. Then, as the war in Ukraine began, some commodity prices spiked, raising inflation rates further, especially in the European Union (see TDR, 2022 for a discussion). Despite the supply-side origins of this new round of inflationary pressure, leading central banks, beginning with the United States Federal Reserve, embarked on monetary tightening sooner than had been previously signalled.\n\n**Figure I.8 Inflation rates have remained in line with historical standards in most developing countries** Monthly consumer price index growth, selected developing countries (Year-on-year percentage change)\n\n2006 0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\n70\n\n80\n\n90\n\n\\-5\n\n0\n\n5\n\n10\n\n15\n\n20\n\n10 15 20\n\nChile Mexico Brazil China India Indonesia\n\nSouth Africa Türkiye (right axis)\n\nJul. 23\n\n_Source:_ OECD Statistics and national sources.\n\n## **I**","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"19\n\nCurrent Trends and Challenges in the Global Economy\n\n# F. INFLATION AND DISTRIBUTION\n\nAfter 2020, inflation accelerated along similar trajectories in most countries, but in 2023, as inflation began to decrease, the paths diverged.\n\nSigns of an inflation pick-up began to appear in the United States in the early months of 2021. Various temporary and lasting triggers contributed to this. These include four main factors: (a) changes to global trade patterns that impacted import costs; (b) a surge in consumption expenditure by the wealthy, who benefited from the stock market gains fuelled by extra-loose monetary policy; (c) small increases in the real wages of the lowest paid occupations; and (d) the ability of producers and retailers to raise prices in order to recoup cost increases and increase profit margins (Bivens, 2022; Konczal and Lusiani, 2022; Hayes and Jung, 2022; Schnabel, 2022; Storm, 2022; Weber and Wasner, 2023).\n\nConcerns that inflation was extending beyond the expected transitory period that would normally accompany recovery from a deep economic shock only began towards the end of 2021, as an initial easing of rising prices was reversed. Failure to distribute effective vaccines worldwide prolonged the pandemic, causing temporary factors to linger on, eventually interacting with a largely anticipated initial increase in commodity prices. Then, as the war in Ukraine began, some commodity prices spiked, raising inflation rates further, especially in the European Union (see TDR, 2022 for a discussion). Despite the supply-side origins of this new round of inflationary pressure, leading central banks, beginning with the United States Federal Reserve, embarked on monetary tightening sooner than had been previously signalled.\n\n**Figure I.8 Inflation rates have remained in line with historical standards in most developing countries** Monthly consumer price index growth, selected developing countries (Year-on-year percentage change)\n\n2006 0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\n70\n\n80\n\n90\n\n\\-5\n\n0\n\n5\n\n10\n\n15\n\n20\n\n10 15 20\n\nChile Mexico Brazil China India Indonesia\n\nSouth Africa Türkiye (right axis)\n\nJul. 23\n\n_Source:_ OECD Statistics and national sources.\n\n## **I**","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/tdr2023ch1_en_p19.pdf"]},"response":{"errors":[],"extract_id":"extract_012fafd84cca8d99d66021a8cf66a6de","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["19\n\nCurrent Trends and Challenges in the Global Economy\n\n# F. INFLATION AND DISTRIBUTION\n\nAfter 2020, inflation accelerated along similar trajectories in most countries, but in 2023, as inflation began to decrease, the paths diverged.\n\nSigns of an inflation pick-up began to appear in the United States in the early months of 2021. Various temporary and lasting triggers contributed to this. These include four main factors: (a) changes to global trade patterns that impacted import costs; (b) a surge in consumption expenditure by the wealthy, who benefited from the stock market gains fuelled by extra-loose monetary policy; (c) small increases in the real wages of the lowest paid occupations; and (d) the ability of producers and retailers to raise prices in order to recoup cost increases and increase profit margins (Bivens, 2022; Konczal and Lusiani, 2022; Hayes and Jung, 2022; Schnabel, 2022; Storm, 2022; Weber and Wasner, 2023).\n\nConcerns that inflation was extending beyond the expected transitory period that would normally accompany recovery from a deep economic shock only began towards the end of 2021, as an initial easing of rising prices was reversed. Failure to distribute effective vaccines worldwide prolonged the pandemic, causing temporary factors to linger on, eventually interacting with a largely anticipated initial increase in commodity prices. Then, as the war in Ukraine began, some commodity prices spiked, raising inflation rates further, especially in the European Union (see TDR, 2022 for a discussion). Despite the supply-side origins of this new round of inflationary pressure, leading central banks, beginning with the United States Federal Reserve, embarked on monetary tightening sooner than had been previously signalled.\n\n**Figure I.8 Inflation rates have remained in line with historical standards in most developing countries** Monthly consumer price index growth, selected developing countries (Year-on-year percentage change)\n\n2006 0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\n70\n\n80\n\n90\n\n\\-5\n\n0\n\n5\n\n10\n\n15\n\n20\n\n10 15 20\n\nChile Mexico Brazil China India Indonesia\n\nSouth Africa Türkiye (right axis)\n\nJul. 23\n\n_Source:_ OECD Statistics and national sources.\n\n## **I**"],"full_content":"19\n\nCurrent Trends and Challenges in the Global Economy\n\n# F. INFLATION AND DISTRIBUTION\n\nAfter 2020, inflation accelerated along similar trajectories in most countries, but in 2023, as inflation began to decrease, the paths diverged.\n\nSigns of an inflation pick-up began to appear in the United States in the early months of 2021. Various temporary and lasting triggers contributed to this. These include four main factors: (a) changes to global trade patterns that impacted import costs; (b) a surge in consumption expenditure by the wealthy, who benefited from the stock market gains fuelled by extra-loose monetary policy; (c) small increases in the real wages of the lowest paid occupations; and (d) the ability of producers and retailers to raise prices in order to recoup cost increases and increase profit margins (Bivens, 2022; Konczal and Lusiani, 2022; Hayes and Jung, 2022; Schnabel, 2022; Storm, 2022; Weber and Wasner, 2023).\n\nConcerns that inflation was extending beyond the expected transitory period that would normally accompany recovery from a deep economic shock only began towards the end of 2021, as an initial easing of rising prices was reversed. Failure to distribute effective vaccines worldwide prolonged the pandemic, causing temporary factors to linger on, eventually interacting with a largely anticipated initial increase in commodity prices. Then, as the war in Ukraine began, some commodity prices spiked, raising inflation rates further, especially in the European Union (see TDR, 2022 for a discussion). Despite the supply-side origins of this new round of inflationary pressure, leading central banks, beginning with the United States Federal Reserve, embarked on monetary tightening sooner than had been previously signalled.\n\n**Figure I.8 Inflation rates have remained in line with historical standards in most developing countries** Monthly consumer price index growth, selected developing countries (Year-on-year percentage change)\n\n2006 0\n\n10\n\n20\n\n30\n\n40\n\n50\n\n60\n\n70\n\n80\n\n90\n\n\\-5\n\n0\n\n5\n\n10\n\n15\n\n20\n\n10 15 20\n\nChile Mexico Brazil China India Indonesia\n\nSouth Africa Türkiye (right axis)\n\nJul. 23\n\n_Source:_ OECD Statistics and national sources.\n\n## **I**","publish_date":null,"title":"F. INFLATION AND DISTRIBUTION","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/tdr2023ch1_en_p19.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/tdr2023ch1_en_p19.pdf"},"request":{"example_id":"chart/tdr2023ch1_en_p19","product_type":"parse","source_file_path":"tdr2023ch1_en_p19.pdf"},"started_at":"2026-11-04T19:44:58.776886"} +{"completed_at":"2026-11-04T19:45:02.400785","latency_in_ms":2494,"output":{"example_id":"chart/tdr2023ch1_en_p22","grounded_pages":[],"layout_pages":[],"markdown":"TRADE AND DEVELOPMENT REPORT 2023 Growth, Debt and Climate: Realigning the Global Financial Architecture\n\n22\n\nThree factors need to be considered to fully understand recent price dynamics.\n\n1\\. As the cost of key inputs accelerates, several circumstances allow firms to gain higher profits by setting their prices following the general increasing trend, even if the goods were produced when inputs were cheaper. Oligopolies and vertically integrated firms in particular were in a privileged position and have taken advantage of the general inflation to increase their profit margins. For them, the increased cost of credit has had a very limited impact, as the growth of their revenues has more than kept up with it (figure I.10).\n\n2\\. Second, falling energy and food prices in international markets may well reduce inflation, but this does not signal a decrease in the price of most retail goods and services: at best these will remain stable at their elevated levels. Furthermore, depending on market power and regulation, domestic prices of food and energy will keep increasing even if the international prices of the commodities they use as key inputs decrease. This means that wages will have to increase to regain the real purchasing power lost with inflation. Whether current policies are consistent with this scenario is doubtful (see section E above). Policymakers should consider how to tackle income inequality while also addressing the unchecked capacity of businesses in critical sectors of the economy to pass higher labour costs through to increased prices.\n\n3\\. Third, key factors of uncertainty and instability in international markets have not been addressed. The emergence of new players in commodities trade, such as the United States, now a net exporter of oil and gas, and of new restrictions to manufacturing trade, such as those relating to chips and semiconductors, evidently matter. But the structural problem relates to the organization of markets and trade, which are heavily exposed to asymmetric regulations and profiteering (chapters II and III, also TDR, 2022). Food prices, for instance, remain well above pre-pandemic averages: a largely unsustainable level for many households, especially in developing countries that are net importers of food.\n\n**Figure I.10 Profit shares have increased above their long-term rising trends** Income from profits and rents (Percentage of GDP)\n\n34 2003 08 13 18 23\n\n36\n\n38\n\n40\n\n42\n\n44\n\n46\n\nDeveloped countries Developing countries\n\nLinear (Developed countries) Linear (Developing countries)\n\n_Source:_ UNCTAD calculations based on the United Nations Global Policy Model and database.\n\n_Note:_ GDP at constant 2015 prices, PPP.\n\n_“Food prices remain well above pre- pandemic averages: an unsustainable level for many households, especially in developing countries that are net importers of food.”_\n\n_“Policymakers should consider how to tackle income inequality while addressing the unchecked capacity of businesses to translate higher labour costs into increased prices.”_","pages":[{"line_numbers":[],"markdown":"TRADE AND DEVELOPMENT REPORT 2023 Growth, Debt and Climate: Realigning the Global Financial Architecture\n\n22\n\nThree factors need to be considered to fully understand recent price dynamics.\n\n1\\. As the cost of key inputs accelerates, several circumstances allow firms to gain higher profits by setting their prices following the general increasing trend, even if the goods were produced when inputs were cheaper. Oligopolies and vertically integrated firms in particular were in a privileged position and have taken advantage of the general inflation to increase their profit margins. For them, the increased cost of credit has had a very limited impact, as the growth of their revenues has more than kept up with it (figure I.10).\n\n2\\. Second, falling energy and food prices in international markets may well reduce inflation, but this does not signal a decrease in the price of most retail goods and services: at best these will remain stable at their elevated levels. Furthermore, depending on market power and regulation, domestic prices of food and energy will keep increasing even if the international prices of the commodities they use as key inputs decrease. This means that wages will have to increase to regain the real purchasing power lost with inflation. Whether current policies are consistent with this scenario is doubtful (see section E above). Policymakers should consider how to tackle income inequality while also addressing the unchecked capacity of businesses in critical sectors of the economy to pass higher labour costs through to increased prices.\n\n3\\. Third, key factors of uncertainty and instability in international markets have not been addressed. The emergence of new players in commodities trade, such as the United States, now a net exporter of oil and gas, and of new restrictions to manufacturing trade, such as those relating to chips and semiconductors, evidently matter. But the structural problem relates to the organization of markets and trade, which are heavily exposed to asymmetric regulations and profiteering (chapters II and III, also TDR, 2022). Food prices, for instance, remain well above pre-pandemic averages: a largely unsustainable level for many households, especially in developing countries that are net importers of food.\n\n**Figure I.10 Profit shares have increased above their long-term rising trends** Income from profits and rents (Percentage of GDP)\n\n34 2003 08 13 18 23\n\n36\n\n38\n\n40\n\n42\n\n44\n\n46\n\nDeveloped countries Developing countries\n\nLinear (Developed countries) Linear (Developing countries)\n\n_Source:_ UNCTAD calculations based on the United Nations Global Policy Model and database.\n\n_Note:_ GDP at constant 2015 prices, PPP.\n\n_“Food prices remain well above pre- pandemic averages: an unsustainable level for many households, especially in developing countries that are net importers of food.”_\n\n_“Policymakers should consider how to tackle income inequality while addressing the unchecked capacity of businesses to translate higher labour costs into increased prices.”_","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"TRADE AND DEVELOPMENT REPORT 2023 Growth, Debt and Climate: Realigning the Global Financial Architecture\n\n22\n\nThree factors need to be considered to fully understand recent price dynamics.\n\n1\\. As the cost of key inputs accelerates, several circumstances allow firms to gain higher profits by setting their prices following the general increasing trend, even if the goods were produced when inputs were cheaper. Oligopolies and vertically integrated firms in particular were in a privileged position and have taken advantage of the general inflation to increase their profit margins. For them, the increased cost of credit has had a very limited impact, as the growth of their revenues has more than kept up with it (figure I.10).\n\n2\\. Second, falling energy and food prices in international markets may well reduce inflation, but this does not signal a decrease in the price of most retail goods and services: at best these will remain stable at their elevated levels. Furthermore, depending on market power and regulation, domestic prices of food and energy will keep increasing even if the international prices of the commodities they use as key inputs decrease. This means that wages will have to increase to regain the real purchasing power lost with inflation. Whether current policies are consistent with this scenario is doubtful (see section E above). Policymakers should consider how to tackle income inequality while also addressing the unchecked capacity of businesses in critical sectors of the economy to pass higher labour costs through to increased prices.\n\n3\\. Third, key factors of uncertainty and instability in international markets have not been addressed. The emergence of new players in commodities trade, such as the United States, now a net exporter of oil and gas, and of new restrictions to manufacturing trade, such as those relating to chips and semiconductors, evidently matter. But the structural problem relates to the organization of markets and trade, which are heavily exposed to asymmetric regulations and profiteering (chapters II and III, also TDR, 2022). Food prices, for instance, remain well above pre-pandemic averages: a largely unsustainable level for many households, especially in developing countries that are net importers of food.\n\n**Figure I.10 Profit shares have increased above their long-term rising trends** Income from profits and rents (Percentage of GDP)\n\n34 2003 08 13 18 23\n\n36\n\n38\n\n40\n\n42\n\n44\n\n46\n\nDeveloped countries Developing countries\n\nLinear (Developed countries) Linear (Developing countries)\n\n_Source:_ UNCTAD calculations based on the United Nations Global Policy Model and database.\n\n_Note:_ GDP at constant 2015 prices, PPP.\n\n_“Food prices remain well above pre- pandemic averages: an unsustainable level for many households, especially in developing countries that are net importers of food.”_\n\n_“Policymakers should consider how to tackle income inequality while addressing the unchecked capacity of businesses to translate higher labour costs into increased prices.”_","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/tdr2023ch1_en_p22.pdf"]},"response":{"errors":[],"extract_id":"extract_d8628822bb0063a0e9690e09148e2c50","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["TRADE AND DEVELOPMENT REPORT 2023 Growth, Debt and Climate: Realigning the Global Financial Architecture\n\n22\n\nThree factors need to be considered to fully understand recent price dynamics.\n\n1\\. As the cost of key inputs accelerates, several circumstances allow firms to gain higher profits by setting their prices following the general increasing trend, even if the goods were produced when inputs were cheaper. Oligopolies and vertically integrated firms in particular were in a privileged position and have taken advantage of the general inflation to increase their profit margins. For them, the increased cost of credit has had a very limited impact, as the growth of their revenues has more than kept up with it (figure I.10).\n\n2\\. Second, falling energy and food prices in international markets may well reduce inflation, but this does not signal a decrease in the price of most retail goods and services: at best these will remain stable at their elevated levels. Furthermore, depending on market power and regulation, domestic prices of food and energy will keep increasing even if the international prices of the commodities they use as key inputs decrease. This means that wages will have to increase to regain the real purchasing power lost with inflation. Whether current policies are consistent with this scenario is doubtful (see section E above). Policymakers should consider how to tackle income inequality while also addressing the unchecked capacity of businesses in critical sectors of the economy to pass higher labour costs through to increased prices.\n\n3\\. Third, key factors of uncertainty and instability in international markets have not been addressed. The emergence of new players in commodities trade, such as the United States, now a net exporter of oil and gas, and of new restrictions to manufacturing trade, such as those relating to chips and semiconductors, evidently matter. But the structural problem relates to the organization of markets and trade, which are heavily exposed to asymmetric regulations and profiteering (chapters II and III, also TDR, 2022). Food prices, for instance, remain well above pre-pandemic averages: a largely unsustainable level for many households, especially in developing countries that are net importers of food.\n\n**Figure I.10 Profit shares have increased above their long-term rising trends** Income from profits and rents (Percentage of GDP)\n\n34 2003 08 13 18 23\n\n36\n\n38\n\n40\n\n42\n\n44\n\n46\n\nDeveloped countries Developing countries\n\nLinear (Developed countries) Linear (Developing countries)\n\n_Source:_ UNCTAD calculations based on the United Nations Global Policy Model and database.\n\n_Note:_ GDP at constant 2015 prices, PPP.\n\n_“Food prices remain well above pre- pandemic averages: an unsustainable level for many households, especially in developing countries that are net importers of food.”_\n\n_“Policymakers should consider how to tackle income inequality while addressing the unchecked capacity of businesses to translate higher labour costs into increased prices.”_"],"full_content":"TRADE AND DEVELOPMENT REPORT 2023 Growth, Debt and Climate: Realigning the Global Financial Architecture\n\n22\n\nThree factors need to be considered to fully understand recent price dynamics.\n\n1\\. As the cost of key inputs accelerates, several circumstances allow firms to gain higher profits by setting their prices following the general increasing trend, even if the goods were produced when inputs were cheaper. Oligopolies and vertically integrated firms in particular were in a privileged position and have taken advantage of the general inflation to increase their profit margins. For them, the increased cost of credit has had a very limited impact, as the growth of their revenues has more than kept up with it (figure I.10).\n\n2\\. Second, falling energy and food prices in international markets may well reduce inflation, but this does not signal a decrease in the price of most retail goods and services: at best these will remain stable at their elevated levels. Furthermore, depending on market power and regulation, domestic prices of food and energy will keep increasing even if the international prices of the commodities they use as key inputs decrease. This means that wages will have to increase to regain the real purchasing power lost with inflation. Whether current policies are consistent with this scenario is doubtful (see section E above). Policymakers should consider how to tackle income inequality while also addressing the unchecked capacity of businesses in critical sectors of the economy to pass higher labour costs through to increased prices.\n\n3\\. Third, key factors of uncertainty and instability in international markets have not been addressed. The emergence of new players in commodities trade, such as the United States, now a net exporter of oil and gas, and of new restrictions to manufacturing trade, such as those relating to chips and semiconductors, evidently matter. But the structural problem relates to the organization of markets and trade, which are heavily exposed to asymmetric regulations and profiteering (chapters II and III, also TDR, 2022). Food prices, for instance, remain well above pre-pandemic averages: a largely unsustainable level for many households, especially in developing countries that are net importers of food.\n\n**Figure I.10 Profit shares have increased above their long-term rising trends** Income from profits and rents (Percentage of GDP)\n\n34 2003 08 13 18 23\n\n36\n\n38\n\n40\n\n42\n\n44\n\n46\n\nDeveloped countries Developing countries\n\nLinear (Developed countries) Linear (Developing countries)\n\n_Source:_ UNCTAD calculations based on the United Nations Global Policy Model and database.\n\n_Note:_ GDP at constant 2015 prices, PPP.\n\n_“Food prices remain well above pre- pandemic averages: an unsustainable level for many households, especially in developing countries that are net importers of food.”_\n\n_“Policymakers should consider how to tackle income inequality while addressing the unchecked capacity of businesses to translate higher labour costs into increased prices.”_","publish_date":null,"title":"“Food prices remain well above pre- pandemic averages: an unsustainable level for many households, especially in developing countries that are","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/tdr2023ch1_en_p22.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/tdr2023ch1_en_p22.pdf"},"request":{"example_id":"chart/tdr2023ch1_en_p22","product_type":"parse","source_file_path":"tdr2023ch1_en_p22.pdf"},"started_at":"2026-11-04T19:44:59.906249"} +{"completed_at":"2026-11-04T19:45:02.570266","latency_in_ms":1111,"output":{"example_id":"chart/tdr2023ch1_en_p5","grounded_pages":[],"layout_pages":[],"markdown":"5\n\nCurrent Trends and Challenges in the Global Economy\n\nThe chapter is structured as follows. Section B examines the emerging risks to post-COVID-19 growth trend at the global level. It finds that divergence within key regional blocks and between major economies clouds the fragile growth of 2023, with downside risks lingering into 2024. Section C analyses the sectoral contribution to global demand growth in G20 economies. Section D identifies some of the key dimensions of the asymmetry between growing corporate concentration on the one hand and thinning fiscal policy space on the other. Section E discusses credit, investment and the impact of monetary policy on income and wealth inequality. Section F explores inflation, distribution and the easing or persistence of inflationary trajectories. Section G looks at labour costs and inequality. Section H concludes.\n\n# B. GLOBAL GROWTH LANDSCAPE: DIVERGENCE\n\n# UNDER THE CLOUDS OF UNCERTAINTY\n\nThe growth of the world economic output is expected to decelerate to 2.4 per cent in 2023, before registering a small uptick to 2.5 per cent in 2024. (table I.1). These are among the lowest growth rates of the last four decades, outside of crisis years. Moreover, the figure for 2023 is below the conventional threshold of 2.5 per cent which marks recession in the global economy. These projections are subject to downside risks which have increased in recent months.\n\nAll regions, except for East and Central Asia, are expected to post slower growth this year than in 2022, with the largest drop (2.3 points) occurring in Europe. Likewise, among G20 countries, only Brazil, China, Japan, Mexico and the Russian Federation are expected to see a growth improvement, with considerable variation. Of particular concern, given the ambitious development and\n\nclimate targets set by the international community with a 2030 delivery date, growth in 2023 and 2024 is also set to fall below the average for the five- year period before the pandemic, in all regions. Latin America is the exception, where growth in the earlier period was particularly weak (ECLAC, 2023).\n\nIn 2023, global growth showed\n\nuneven deceleration. Larger emerging economies are unlikely to provide a robust offset to slower growth in advanced economies. With tighter monetary\n\npolicy, low investment (figure I.1), and limited government spending, the world economy is experiencing a lacklustre recovery reminiscent of the aftermath of the 2008–2009 financial crisis.\n\n**Figure I.1 Private investment is slowing down sharply again** Growth of private investment (Percentage change)\n\n2015 17 19 21 23\n\nDeveloped countries Developing countries\n\n\\-6\n\n\\-4\n\n\\-2\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n_Source:_ UNCTAD calculations based on the United Nations Global Policy Model\n\nand database.\n\n_Note:_ GDP at constant 2015 prices, PPP.\n\n## **I**","pages":[{"line_numbers":[],"markdown":"5\n\nCurrent Trends and Challenges in the Global Economy\n\nThe chapter is structured as follows. Section B examines the emerging risks to post-COVID-19 growth trend at the global level. It finds that divergence within key regional blocks and between major economies clouds the fragile growth of 2023, with downside risks lingering into 2024. Section C analyses the sectoral contribution to global demand growth in G20 economies. Section D identifies some of the key dimensions of the asymmetry between growing corporate concentration on the one hand and thinning fiscal policy space on the other. Section E discusses credit, investment and the impact of monetary policy on income and wealth inequality. Section F explores inflation, distribution and the easing or persistence of inflationary trajectories. Section G looks at labour costs and inequality. Section H concludes.\n\n# B. GLOBAL GROWTH LANDSCAPE: DIVERGENCE\n\n# UNDER THE CLOUDS OF UNCERTAINTY\n\nThe growth of the world economic output is expected to decelerate to 2.4 per cent in 2023, before registering a small uptick to 2.5 per cent in 2024. (table I.1). These are among the lowest growth rates of the last four decades, outside of crisis years. Moreover, the figure for 2023 is below the conventional threshold of 2.5 per cent which marks recession in the global economy. These projections are subject to downside risks which have increased in recent months.\n\nAll regions, except for East and Central Asia, are expected to post slower growth this year than in 2022, with the largest drop (2.3 points) occurring in Europe. Likewise, among G20 countries, only Brazil, China, Japan, Mexico and the Russian Federation are expected to see a growth improvement, with considerable variation. Of particular concern, given the ambitious development and\n\nclimate targets set by the international community with a 2030 delivery date, growth in 2023 and 2024 is also set to fall below the average for the five- year period before the pandemic, in all regions. Latin America is the exception, where growth in the earlier period was particularly weak (ECLAC, 2023).\n\nIn 2023, global growth showed\n\nuneven deceleration. Larger emerging economies are unlikely to provide a robust offset to slower growth in advanced economies. With tighter monetary\n\npolicy, low investment (figure I.1), and limited government spending, the world economy is experiencing a lacklustre recovery reminiscent of the aftermath of the 2008–2009 financial crisis.\n\n**Figure I.1 Private investment is slowing down sharply again** Growth of private investment (Percentage change)\n\n2015 17 19 21 23\n\nDeveloped countries Developing countries\n\n\\-6\n\n\\-4\n\n\\-2\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n_Source:_ UNCTAD calculations based on the United Nations Global Policy Model\n\nand database.\n\n_Note:_ GDP at constant 2015 prices, PPP.\n\n## **I**","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"5\n\nCurrent Trends and Challenges in the Global Economy\n\nThe chapter is structured as follows. Section B examines the emerging risks to post-COVID-19 growth trend at the global level. It finds that divergence within key regional blocks and between major economies clouds the fragile growth of 2023, with downside risks lingering into 2024. Section C analyses the sectoral contribution to global demand growth in G20 economies. Section D identifies some of the key dimensions of the asymmetry between growing corporate concentration on the one hand and thinning fiscal policy space on the other. Section E discusses credit, investment and the impact of monetary policy on income and wealth inequality. Section F explores inflation, distribution and the easing or persistence of inflationary trajectories. Section G looks at labour costs and inequality. Section H concludes.\n\n# B. GLOBAL GROWTH LANDSCAPE: DIVERGENCE\n\n# UNDER THE CLOUDS OF UNCERTAINTY\n\nThe growth of the world economic output is expected to decelerate to 2.4 per cent in 2023, before registering a small uptick to 2.5 per cent in 2024. (table I.1). These are among the lowest growth rates of the last four decades, outside of crisis years. Moreover, the figure for 2023 is below the conventional threshold of 2.5 per cent which marks recession in the global economy. These projections are subject to downside risks which have increased in recent months.\n\nAll regions, except for East and Central Asia, are expected to post slower growth this year than in 2022, with the largest drop (2.3 points) occurring in Europe. Likewise, among G20 countries, only Brazil, China, Japan, Mexico and the Russian Federation are expected to see a growth improvement, with considerable variation. Of particular concern, given the ambitious development and\n\nclimate targets set by the international community with a 2030 delivery date, growth in 2023 and 2024 is also set to fall below the average for the five- year period before the pandemic, in all regions. Latin America is the exception, where growth in the earlier period was particularly weak (ECLAC, 2023).\n\nIn 2023, global growth showed\n\nuneven deceleration. Larger emerging economies are unlikely to provide a robust offset to slower growth in advanced economies. With tighter monetary\n\npolicy, low investment (figure I.1), and limited government spending, the world economy is experiencing a lacklustre recovery reminiscent of the aftermath of the 2008–2009 financial crisis.\n\n**Figure I.1 Private investment is slowing down sharply again** Growth of private investment (Percentage change)\n\n2015 17 19 21 23\n\nDeveloped countries Developing countries\n\n\\-6\n\n\\-4\n\n\\-2\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n_Source:_ UNCTAD calculations based on the United Nations Global Policy Model\n\nand database.\n\n_Note:_ GDP at constant 2015 prices, PPP.\n\n## **I**","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/tdr2023ch1_en_p5.pdf"]},"response":{"errors":[],"extract_id":"extract_0d416f8f3bf017409044cc1d4d08d5c3","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["5\n\nCurrent Trends and Challenges in the Global Economy\n\nThe chapter is structured as follows. Section B examines the emerging risks to post-COVID-19 growth trend at the global level. It finds that divergence within key regional blocks and between major economies clouds the fragile growth of 2023, with downside risks lingering into 2024. Section C analyses the sectoral contribution to global demand growth in G20 economies. Section D identifies some of the key dimensions of the asymmetry between growing corporate concentration on the one hand and thinning fiscal policy space on the other. Section E discusses credit, investment and the impact of monetary policy on income and wealth inequality. Section F explores inflation, distribution and the easing or persistence of inflationary trajectories. Section G looks at labour costs and inequality. Section H concludes.\n\n# B. GLOBAL GROWTH LANDSCAPE: DIVERGENCE\n\n# UNDER THE CLOUDS OF UNCERTAINTY\n\nThe growth of the world economic output is expected to decelerate to 2.4 per cent in 2023, before registering a small uptick to 2.5 per cent in 2024. (table I.1). These are among the lowest growth rates of the last four decades, outside of crisis years. Moreover, the figure for 2023 is below the conventional threshold of 2.5 per cent which marks recession in the global economy. These projections are subject to downside risks which have increased in recent months.\n\nAll regions, except for East and Central Asia, are expected to post slower growth this year than in 2022, with the largest drop (2.3 points) occurring in Europe. Likewise, among G20 countries, only Brazil, China, Japan, Mexico and the Russian Federation are expected to see a growth improvement, with considerable variation. Of particular concern, given the ambitious development and\n\nclimate targets set by the international community with a 2030 delivery date, growth in 2023 and 2024 is also set to fall below the average for the five- year period before the pandemic, in all regions. Latin America is the exception, where growth in the earlier period was particularly weak (ECLAC, 2023).\n\nIn 2023, global growth showed\n\nuneven deceleration. Larger emerging economies are unlikely to provide a robust offset to slower growth in advanced economies. With tighter monetary\n\npolicy, low investment (figure I.1), and limited government spending, the world economy is experiencing a lacklustre recovery reminiscent of the aftermath of the 2008–2009 financial crisis.\n\n**Figure I.1 Private investment is slowing down sharply again** Growth of private investment (Percentage change)\n\n2015 17 19 21 23\n\nDeveloped countries Developing countries\n\n\\-6\n\n\\-4\n\n\\-2\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n_Source:_ UNCTAD calculations based on the United Nations Global Policy Model\n\nand database.\n\n_Note:_ GDP at constant 2015 prices, PPP.\n\n## **I**"],"full_content":"5\n\nCurrent Trends and Challenges in the Global Economy\n\nThe chapter is structured as follows. Section B examines the emerging risks to post-COVID-19 growth trend at the global level. It finds that divergence within key regional blocks and between major economies clouds the fragile growth of 2023, with downside risks lingering into 2024. Section C analyses the sectoral contribution to global demand growth in G20 economies. Section D identifies some of the key dimensions of the asymmetry between growing corporate concentration on the one hand and thinning fiscal policy space on the other. Section E discusses credit, investment and the impact of monetary policy on income and wealth inequality. Section F explores inflation, distribution and the easing or persistence of inflationary trajectories. Section G looks at labour costs and inequality. Section H concludes.\n\n# B. GLOBAL GROWTH LANDSCAPE: DIVERGENCE\n\n# UNDER THE CLOUDS OF UNCERTAINTY\n\nThe growth of the world economic output is expected to decelerate to 2.4 per cent in 2023, before registering a small uptick to 2.5 per cent in 2024. (table I.1). These are among the lowest growth rates of the last four decades, outside of crisis years. Moreover, the figure for 2023 is below the conventional threshold of 2.5 per cent which marks recession in the global economy. These projections are subject to downside risks which have increased in recent months.\n\nAll regions, except for East and Central Asia, are expected to post slower growth this year than in 2022, with the largest drop (2.3 points) occurring in Europe. Likewise, among G20 countries, only Brazil, China, Japan, Mexico and the Russian Federation are expected to see a growth improvement, with considerable variation. Of particular concern, given the ambitious development and\n\nclimate targets set by the international community with a 2030 delivery date, growth in 2023 and 2024 is also set to fall below the average for the five- year period before the pandemic, in all regions. Latin America is the exception, where growth in the earlier period was particularly weak (ECLAC, 2023).\n\nIn 2023, global growth showed\n\nuneven deceleration. Larger emerging economies are unlikely to provide a robust offset to slower growth in advanced economies. With tighter monetary\n\npolicy, low investment (figure I.1), and limited government spending, the world economy is experiencing a lacklustre recovery reminiscent of the aftermath of the 2008–2009 financial crisis.\n\n**Figure I.1 Private investment is slowing down sharply again** Growth of private investment (Percentage change)\n\n2015 17 19 21 23\n\nDeveloped countries Developing countries\n\n\\-6\n\n\\-4\n\n\\-2\n\n0\n\n2\n\n4\n\n6\n\n8\n\n10\n\n_Source:_ UNCTAD calculations based on the United Nations Global Policy Model\n\nand database.\n\n_Note:_ GDP at constant 2015 prices, PPP.\n\n## **I**","publish_date":null,"title":"B. GLOBAL GROWTH LANDSCAPE: DIVERGENCE UNDER THE CLOUDS OF UNCERTAINTY","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/tdr2023ch1_en_p5.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/tdr2023ch1_en_p5.pdf"},"request":{"example_id":"chart/tdr2023ch1_en_p5","product_type":"parse","source_file_path":"tdr2023ch1_en_p5.pdf"},"started_at":"2026-11-04T19:45:01.459156"} +{"completed_at":"2026-11-04T19:31:09.630709","latency_in_ms":1926,"output":{"example_id":"chart/timevalue_p9","grounded_pages":[],"layout_pages":[],"markdown":"## 9\n\n# Holding Period and Value\n\n0\n\n500\n\n1000\n\n1500\n\n2000\n\n2500\n\n3000\n\n3500\n\n4000\n\n4500\n\n1 5 10 20 30 40\n\n**Future value of $100 investment**\n\n**Holding period (in years)**\n\nStocks ($)\n\nTreasury Bonds ($)\n\nTreasury Bills ($)","pages":[{"line_numbers":[],"markdown":"## 9\n\n# Holding Period and Value\n\n0\n\n500\n\n1000\n\n1500\n\n2000\n\n2500\n\n3000\n\n3500\n\n4000\n\n4500\n\n1 5 10 20 30 40\n\n**Future value of $100 investment**\n\n**Holding period (in years)**\n\nStocks ($)\n\nTreasury Bonds ($)\n\nTreasury Bills ($)","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"## 9\n\n# Holding Period and Value\n\n0\n\n500\n\n1000\n\n1500\n\n2000\n\n2500\n\n3000\n\n3500\n\n4000\n\n4500\n\n1 5 10 20 30 40\n\n**Future value of $100 investment**\n\n**Holding period (in years)**\n\nStocks ($)\n\nTreasury Bonds ($)\n\nTreasury Bills ($)","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/timevalue_p9.pdf"]},"response":{"errors":[],"extract_id":"extract_0e95701bb7314a5e6391cd4050bd5f93","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["## 9\n\n# Holding Period and Value\n\n0\n\n500\n\n1000\n\n1500\n\n2000\n\n2500\n\n3000\n\n3500\n\n4000\n\n4500\n\n1 5 10 20 30 40\n\n**Future value of $100 investment**\n\n**Holding period (in years)**\n\nStocks ($)\n\nTreasury Bonds ($)\n\nTreasury Bills ($)"],"full_content":"## 9\n\n# Holding Period and Value\n\n0\n\n500\n\n1000\n\n1500\n\n2000\n\n2500\n\n3000\n\n3500\n\n4000\n\n4500\n\n1 5 10 20 30 40\n\n**Future value of $100 investment**\n\n**Holding period (in years)**\n\nStocks ($)\n\nTreasury Bonds ($)\n\nTreasury Bills ($)","publish_date":null,"title":"Holding Period and Value","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/timevalue_p9.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/timevalue_p9.pdf"},"request":{"example_id":"chart/timevalue_p9","product_type":"parse","source_file_path":"timevalue_p9.pdf"},"started_at":"2026-11-04T19:31:07.704661"} +{"completed_at":"2026-11-04T19:45:05.029300","latency_in_ms":1096,"output":{"example_id":"chart/tsla-20241231-gen_p34","grounded_pages":[],"layout_pages":[],"markdown":"**Unregistered Sales of Equity Securities and Use of Proceeds**\n\nIn connection with the offering of 2.00% Convertible Senior Notes due 2024 in May 2019, we sold warrants to each of Société Générale, Wells Fargo\n\nBank, National Association, Credit Suisse Capital LLC (later assigned to UBS AG, London Branch) and Goldman, Sachs & Co. LLC (together, the “2019 Warrantholders”). Between October 1, 2024 and October 15, 2024, we issued an aggregate of 2,894,424 shares of our common stock to the 2019 Warrantholders pursuant to their exercise of such warrants, which were net of the applicable exercise prices. Such shares were issued pursuant to an exemption from registration provided by Rule 3(a)(9) of the Securities Act of 1933.\n\n**Purchases of Equity Securities by the Issuer and Affiliated Purchasers**\n\nNone.\n\n**ITEM 6. [RESERVED]**\n\n31","pages":[{"line_numbers":[],"markdown":"**Unregistered Sales of Equity Securities and Use of Proceeds**\n\nIn connection with the offering of 2.00% Convertible Senior Notes due 2024 in May 2019, we sold warrants to each of Société Générale, Wells Fargo\n\nBank, National Association, Credit Suisse Capital LLC (later assigned to UBS AG, London Branch) and Goldman, Sachs & Co. LLC (together, the “2019 Warrantholders”). Between October 1, 2024 and October 15, 2024, we issued an aggregate of 2,894,424 shares of our common stock to the 2019 Warrantholders pursuant to their exercise of such warrants, which were net of the applicable exercise prices. Such shares were issued pursuant to an exemption from registration provided by Rule 3(a)(9) of the Securities Act of 1933.\n\n**Purchases of Equity Securities by the Issuer and Affiliated Purchasers**\n\nNone.\n\n**ITEM 6. [RESERVED]**\n\n31","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**Unregistered Sales of Equity Securities and Use of Proceeds**\n\nIn connection with the offering of 2.00% Convertible Senior Notes due 2024 in May 2019, we sold warrants to each of Société Générale, Wells Fargo\n\nBank, National Association, Credit Suisse Capital LLC (later assigned to UBS AG, London Branch) and Goldman, Sachs & Co. LLC (together, the “2019 Warrantholders”). Between October 1, 2024 and October 15, 2024, we issued an aggregate of 2,894,424 shares of our common stock to the 2019 Warrantholders pursuant to their exercise of such warrants, which were net of the applicable exercise prices. Such shares were issued pursuant to an exemption from registration provided by Rule 3(a)(9) of the Securities Act of 1933.\n\n**Purchases of Equity Securities by the Issuer and Affiliated Purchasers**\n\nNone.\n\n**ITEM 6. [RESERVED]**\n\n31","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/tsla-20241231-gen_p34.pdf"]},"response":{"errors":[],"extract_id":"extract_e348e933ea7de8f168f22ef8660e7635","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**Unregistered Sales of Equity Securities and Use of Proceeds**\n\nIn connection with the offering of 2.00% Convertible Senior Notes due 2024 in May 2019, we sold warrants to each of Société Générale, Wells Fargo\n\nBank, National Association, Credit Suisse Capital LLC (later assigned to UBS AG, London Branch) and Goldman, Sachs & Co. LLC (together, the “2019 Warrantholders”). Between October 1, 2024 and October 15, 2024, we issued an aggregate of 2,894,424 shares of our common stock to the 2019 Warrantholders pursuant to their exercise of such warrants, which were net of the applicable exercise prices. Such shares were issued pursuant to an exemption from registration provided by Rule 3(a)(9) of the Securities Act of 1933.\n\n**Purchases of Equity Securities by the Issuer and Affiliated Purchasers**\n\nNone.\n\n**ITEM 6. [RESERVED]**\n\n31"],"full_content":"**Unregistered Sales of Equity Securities and Use of Proceeds**\n\nIn connection with the offering of 2.00% Convertible Senior Notes due 2024 in May 2019, we sold warrants to each of Société Générale, Wells Fargo\n\nBank, National Association, Credit Suisse Capital LLC (later assigned to UBS AG, London Branch) and Goldman, Sachs & Co. LLC (together, the “2019 Warrantholders”). Between October 1, 2024 and October 15, 2024, we issued an aggregate of 2,894,424 shares of our common stock to the 2019 Warrantholders pursuant to their exercise of such warrants, which were net of the applicable exercise prices. Such shares were issued pursuant to an exemption from registration provided by Rule 3(a)(9) of the Securities Act of 1933.\n\n**Purchases of Equity Securities by the Issuer and Affiliated Purchasers**\n\nNone.\n\n**ITEM 6. [RESERVED]**\n\n31","publish_date":null,"title":"Unregistered Sales of Equity Securities and Use of Proceeds In connection with the offering of 2.00% Convertible Senior Notes due","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/tsla-20241231-gen_p34.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/tsla-20241231-gen_p34.pdf"},"request":{"example_id":"chart/tsla-20241231-gen_p34","product_type":"parse","source_file_path":"tsla-20241231-gen_p34.pdf"},"started_at":"2026-11-04T19:45:03.932789"} +{"completed_at":"2026-11-04T19:45:06.682598","latency_in_ms":1201,"output":{"example_id":"chart/us-global-outsourcing-survey-2024-report_p13","grounded_pages":[],"layout_pages":[],"markdown":"**Global Outsourcing Survey 2024 |** Multidimensional sourcing: Orchestrating the extended workforce ecosystem\n\n13\n\n**Focus on value from outsourcing front-office and core capabilities** Investment in third-party outsourcing continues to grow, despite increased adoption of insourcing and GICs.\n\n**% of organizations outsourcing by scope area**\n\n65% Helpdesk and End-User Computing\n\n40% Business / Operations\n\n49% Procurement / Sourcing\n\n52% Supply Chain / Manufacturing\n\n54% Finance\n\n57% Human Resources (HR)\n\n60% Tax\n\n68% Legal\n\n77% IT Infrastructure Services\n\n72% Data and Analytics\n\n72% Application / Software Development\n\n69% Application Support\n\n75% Innovative Technologies (GenAI, Blockchain, IoT, etc.)\n\n77% Cybersecurity\n\nNo change\n\nDecrease moderately\n\nIncrease significantly\n\nDecrease significantly\n\nIncrease moderately\n\nBack-office outsourcing remains a fundamental component for many organizations. These functions are being outsourced at high rates, allowing businesses to focus on their core competencies while benefiting from providers’ expertise to continue driving operational efficiency.\n\nOrganizations are also outsourcing front-office and core business capabilities such as sales, marketing, customer service, and order fulfillment for additional value: 56% of organizations now outsource front-office functions, and 46% outsource research and development (R&D) activities. These figures represent an approximate 10% increase relative to our 2022 findings. The outsourcing of front-office and core capabilities is common in organizations globally and across industries—with real estate, energy, and consumer products showing the highest level of interest. This points to a continued focus on leveraging outsourced services for value creation in core business capabilities.\n\n**17%**\n\n**3%**\n\n**39%**\n\n**30%**\n\n**11%**","pages":[{"line_numbers":[],"markdown":"**Global Outsourcing Survey 2024 |** Multidimensional sourcing: Orchestrating the extended workforce ecosystem\n\n13\n\n**Focus on value from outsourcing front-office and core capabilities** Investment in third-party outsourcing continues to grow, despite increased adoption of insourcing and GICs.\n\n**% of organizations outsourcing by scope area**\n\n65% Helpdesk and End-User Computing\n\n40% Business / Operations\n\n49% Procurement / Sourcing\n\n52% Supply Chain / Manufacturing\n\n54% Finance\n\n57% Human Resources (HR)\n\n60% Tax\n\n68% Legal\n\n77% IT Infrastructure Services\n\n72% Data and Analytics\n\n72% Application / Software Development\n\n69% Application Support\n\n75% Innovative Technologies (GenAI, Blockchain, IoT, etc.)\n\n77% Cybersecurity\n\nNo change\n\nDecrease moderately\n\nIncrease significantly\n\nDecrease significantly\n\nIncrease moderately\n\nBack-office outsourcing remains a fundamental component for many organizations. These functions are being outsourced at high rates, allowing businesses to focus on their core competencies while benefiting from providers’ expertise to continue driving operational efficiency.\n\nOrganizations are also outsourcing front-office and core business capabilities such as sales, marketing, customer service, and order fulfillment for additional value: 56% of organizations now outsource front-office functions, and 46% outsource research and development (R&D) activities. These figures represent an approximate 10% increase relative to our 2022 findings. The outsourcing of front-office and core capabilities is common in organizations globally and across industries—with real estate, energy, and consumer products showing the highest level of interest. This points to a continued focus on leveraging outsourced services for value creation in core business capabilities.\n\n**17%**\n\n**3%**\n\n**39%**\n\n**30%**\n\n**11%**","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**Global Outsourcing Survey 2024 |** Multidimensional sourcing: Orchestrating the extended workforce ecosystem\n\n13\n\n**Focus on value from outsourcing front-office and core capabilities** Investment in third-party outsourcing continues to grow, despite increased adoption of insourcing and GICs.\n\n**% of organizations outsourcing by scope area**\n\n65% Helpdesk and End-User Computing\n\n40% Business / Operations\n\n49% Procurement / Sourcing\n\n52% Supply Chain / Manufacturing\n\n54% Finance\n\n57% Human Resources (HR)\n\n60% Tax\n\n68% Legal\n\n77% IT Infrastructure Services\n\n72% Data and Analytics\n\n72% Application / Software Development\n\n69% Application Support\n\n75% Innovative Technologies (GenAI, Blockchain, IoT, etc.)\n\n77% Cybersecurity\n\nNo change\n\nDecrease moderately\n\nIncrease significantly\n\nDecrease significantly\n\nIncrease moderately\n\nBack-office outsourcing remains a fundamental component for many organizations. These functions are being outsourced at high rates, allowing businesses to focus on their core competencies while benefiting from providers’ expertise to continue driving operational efficiency.\n\nOrganizations are also outsourcing front-office and core business capabilities such as sales, marketing, customer service, and order fulfillment for additional value: 56% of organizations now outsource front-office functions, and 46% outsource research and development (R&D) activities. These figures represent an approximate 10% increase relative to our 2022 findings. The outsourcing of front-office and core capabilities is common in organizations globally and across industries—with real estate, energy, and consumer products showing the highest level of interest. This points to a continued focus on leveraging outsourced services for value creation in core business capabilities.\n\n**17%**\n\n**3%**\n\n**39%**\n\n**30%**\n\n**11%**","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/us-global-outsourcing-survey-2024-report_p13.pdf"]},"response":{"errors":[],"extract_id":"extract_7316b919863ed533c0bd203f90cda3fd","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**Global Outsourcing Survey 2024 |** Multidimensional sourcing: Orchestrating the extended workforce ecosystem\n\n13\n\n**Focus on value from outsourcing front-office and core capabilities** Investment in third-party outsourcing continues to grow, despite increased adoption of insourcing and GICs.\n\n**% of organizations outsourcing by scope area**\n\n65% Helpdesk and End-User Computing\n\n40% Business / Operations\n\n49% Procurement / Sourcing\n\n52% Supply Chain / Manufacturing\n\n54% Finance\n\n57% Human Resources (HR)\n\n60% Tax\n\n68% Legal\n\n77% IT Infrastructure Services\n\n72% Data and Analytics\n\n72% Application / Software Development\n\n69% Application Support\n\n75% Innovative Technologies (GenAI, Blockchain, IoT, etc.)\n\n77% Cybersecurity\n\nNo change\n\nDecrease moderately\n\nIncrease significantly\n\nDecrease significantly\n\nIncrease moderately\n\nBack-office outsourcing remains a fundamental component for many organizations. These functions are being outsourced at high rates, allowing businesses to focus on their core competencies while benefiting from providers’ expertise to continue driving operational efficiency.\n\nOrganizations are also outsourcing front-office and core business capabilities such as sales, marketing, customer service, and order fulfillment for additional value: 56% of organizations now outsource front-office functions, and 46% outsource research and development (R&D) activities. These figures represent an approximate 10% increase relative to our 2022 findings. The outsourcing of front-office and core capabilities is common in organizations globally and across industries—with real estate, energy, and consumer products showing the highest level of interest. This points to a continued focus on leveraging outsourced services for value creation in core business capabilities.\n\n**17%**\n\n**3%**\n\n**39%**\n\n**30%**\n\n**11%**"],"full_content":"**Global Outsourcing Survey 2024 |** Multidimensional sourcing: Orchestrating the extended workforce ecosystem\n\n13\n\n**Focus on value from outsourcing front-office and core capabilities** Investment in third-party outsourcing continues to grow, despite increased adoption of insourcing and GICs.\n\n**% of organizations outsourcing by scope area**\n\n65% Helpdesk and End-User Computing\n\n40% Business / Operations\n\n49% Procurement / Sourcing\n\n52% Supply Chain / Manufacturing\n\n54% Finance\n\n57% Human Resources (HR)\n\n60% Tax\n\n68% Legal\n\n77% IT Infrastructure Services\n\n72% Data and Analytics\n\n72% Application / Software Development\n\n69% Application Support\n\n75% Innovative Technologies (GenAI, Blockchain, IoT, etc.)\n\n77% Cybersecurity\n\nNo change\n\nDecrease moderately\n\nIncrease significantly\n\nDecrease significantly\n\nIncrease moderately\n\nBack-office outsourcing remains a fundamental component for many organizations. These functions are being outsourced at high rates, allowing businesses to focus on their core competencies while benefiting from providers’ expertise to continue driving operational efficiency.\n\nOrganizations are also outsourcing front-office and core business capabilities such as sales, marketing, customer service, and order fulfillment for additional value: 56% of organizations now outsource front-office functions, and 46% outsource research and development (R&D) activities. These figures represent an approximate 10% increase relative to our 2022 findings. The outsourcing of front-office and core capabilities is common in organizations globally and across industries—with real estate, energy, and consumer products showing the highest level of interest. This points to a continued focus on leveraging outsourced services for value creation in core business capabilities.\n\n**17%**\n\n**3%**\n\n**39%**\n\n**30%**\n\n**11%**","publish_date":null,"title":"Focus on value from outsourcing front-office and core capabilities Investment in third-party outsourcing continues to grow, despite increased adoption of","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/us-global-outsourcing-survey-2024-report_p13.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/us-global-outsourcing-survey-2024-report_p13.pdf"},"request":{"example_id":"chart/us-global-outsourcing-survey-2024-report_p13","product_type":"parse","source_file_path":"us-global-outsourcing-survey-2024-report_p13.pdf"},"started_at":"2026-11-04T19:45:05.480959"} +{"completed_at":"2026-11-04T19:45:08.081961","latency_in_ms":1281,"output":{"example_id":"chart/us-global-outsourcing-survey-2024-report_p14","grounded_pages":[],"layout_pages":[],"markdown":"**Global Outsourcing Survey 2024 |** Multidimensional sourcing: Orchestrating the extended workforce ecosystem\n\n14\n\n**Success with outcome-based delivery models** A key factor in the evolution of outsourcing toward core capabilities is the growing adoption of outcome- based delivery models—for example, operate services. ³ These models prioritize business outcomes and value creation rather than simply supplementing manpower.\n\nTwo years ago, we acknowledged that the traditional staff augmentation model is giving way to outcome- based delivery models. According to recent data, this trend has gained further momentum.\n\nHigh levels of satisfaction with managed and operate services are driving its continued adoption and increased investment across organizations.\n\nAlthough the adoption of mature delivery models is expected to continue increasing, and the delivery models may evolve over time, it is evident that organizations are interested in buying capabilities and outcomes, and not hours.\n\n### **67%**\n\nof executives reported adoption of managed or operate services (up from 45% two years ago)\n\n### **29%**\n\nof executives reported adoption of legacy staff-augmentation model\n\nReport meets or exceeds satisfaction levels from outsourced services\n\nReport a planned increase in investment in outsourced services\n\n**Organizations that adopt managed or operate services**\n\n88%\n\n47%\n\n**Organizations that do not adopt managed or operate services**\n\n71%\n\n35%","pages":[{"line_numbers":[],"markdown":"**Global Outsourcing Survey 2024 |** Multidimensional sourcing: Orchestrating the extended workforce ecosystem\n\n14\n\n**Success with outcome-based delivery models** A key factor in the evolution of outsourcing toward core capabilities is the growing adoption of outcome- based delivery models—for example, operate services. ³ These models prioritize business outcomes and value creation rather than simply supplementing manpower.\n\nTwo years ago, we acknowledged that the traditional staff augmentation model is giving way to outcome- based delivery models. According to recent data, this trend has gained further momentum.\n\nHigh levels of satisfaction with managed and operate services are driving its continued adoption and increased investment across organizations.\n\nAlthough the adoption of mature delivery models is expected to continue increasing, and the delivery models may evolve over time, it is evident that organizations are interested in buying capabilities and outcomes, and not hours.\n\n### **67%**\n\nof executives reported adoption of managed or operate services (up from 45% two years ago)\n\n### **29%**\n\nof executives reported adoption of legacy staff-augmentation model\n\nReport meets or exceeds satisfaction levels from outsourced services\n\nReport a planned increase in investment in outsourced services\n\n**Organizations that adopt managed or operate services**\n\n88%\n\n47%\n\n**Organizations that do not adopt managed or operate services**\n\n71%\n\n35%","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**Global Outsourcing Survey 2024 |** Multidimensional sourcing: Orchestrating the extended workforce ecosystem\n\n14\n\n**Success with outcome-based delivery models** A key factor in the evolution of outsourcing toward core capabilities is the growing adoption of outcome- based delivery models—for example, operate services. ³ These models prioritize business outcomes and value creation rather than simply supplementing manpower.\n\nTwo years ago, we acknowledged that the traditional staff augmentation model is giving way to outcome- based delivery models. According to recent data, this trend has gained further momentum.\n\nHigh levels of satisfaction with managed and operate services are driving its continued adoption and increased investment across organizations.\n\nAlthough the adoption of mature delivery models is expected to continue increasing, and the delivery models may evolve over time, it is evident that organizations are interested in buying capabilities and outcomes, and not hours.\n\n### **67%**\n\nof executives reported adoption of managed or operate services (up from 45% two years ago)\n\n### **29%**\n\nof executives reported adoption of legacy staff-augmentation model\n\nReport meets or exceeds satisfaction levels from outsourced services\n\nReport a planned increase in investment in outsourced services\n\n**Organizations that adopt managed or operate services**\n\n88%\n\n47%\n\n**Organizations that do not adopt managed or operate services**\n\n71%\n\n35%","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/us-global-outsourcing-survey-2024-report_p14.pdf"]},"response":{"errors":[],"extract_id":"extract_7840725ff658146bab9b354258bddd0a","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**Global Outsourcing Survey 2024 |** Multidimensional sourcing: Orchestrating the extended workforce ecosystem\n\n14\n\n**Success with outcome-based delivery models** A key factor in the evolution of outsourcing toward core capabilities is the growing adoption of outcome- based delivery models—for example, operate services. ³ These models prioritize business outcomes and value creation rather than simply supplementing manpower.\n\nTwo years ago, we acknowledged that the traditional staff augmentation model is giving way to outcome- based delivery models. According to recent data, this trend has gained further momentum.\n\nHigh levels of satisfaction with managed and operate services are driving its continued adoption and increased investment across organizations.\n\nAlthough the adoption of mature delivery models is expected to continue increasing, and the delivery models may evolve over time, it is evident that organizations are interested in buying capabilities and outcomes, and not hours.\n\n### **67%**\n\nof executives reported adoption of managed or operate services (up from 45% two years ago)\n\n### **29%**\n\nof executives reported adoption of legacy staff-augmentation model\n\nReport meets or exceeds satisfaction levels from outsourced services\n\nReport a planned increase in investment in outsourced services\n\n**Organizations that adopt managed or operate services**\n\n88%\n\n47%\n\n**Organizations that do not adopt managed or operate services**\n\n71%\n\n35%"],"full_content":"**Global Outsourcing Survey 2024 |** Multidimensional sourcing: Orchestrating the extended workforce ecosystem\n\n14\n\n**Success with outcome-based delivery models** A key factor in the evolution of outsourcing toward core capabilities is the growing adoption of outcome- based delivery models—for example, operate services. ³ These models prioritize business outcomes and value creation rather than simply supplementing manpower.\n\nTwo years ago, we acknowledged that the traditional staff augmentation model is giving way to outcome- based delivery models. According to recent data, this trend has gained further momentum.\n\nHigh levels of satisfaction with managed and operate services are driving its continued adoption and increased investment across organizations.\n\nAlthough the adoption of mature delivery models is expected to continue increasing, and the delivery models may evolve over time, it is evident that organizations are interested in buying capabilities and outcomes, and not hours.\n\n### **67%**\n\nof executives reported adoption of managed or operate services (up from 45% two years ago)\n\n### **29%**\n\nof executives reported adoption of legacy staff-augmentation model\n\nReport meets or exceeds satisfaction levels from outsourced services\n\nReport a planned increase in investment in outsourced services\n\n**Organizations that adopt managed or operate services**\n\n88%\n\n47%\n\n**Organizations that do not adopt managed or operate services**\n\n71%\n\n35%","publish_date":null,"title":"67% 29%","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/us-global-outsourcing-survey-2024-report_p14.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/us-global-outsourcing-survey-2024-report_p14.pdf"},"request":{"example_id":"chart/us-global-outsourcing-survey-2024-report_p14","product_type":"parse","source_file_path":"us-global-outsourcing-survey-2024-report_p14.pdf"},"started_at":"2026-11-04T19:45:06.800876"} +{"completed_at":"2026-11-04T19:45:09.544702","latency_in_ms":1246,"output":{"example_id":"chart/us-global-outsourcing-survey-2024-report_p15","grounded_pages":[],"layout_pages":[],"markdown":"**Global Outsourcing Survey 2024 |** Multidimensional sourcing: Orchestrating the extended workforce ecosystem\n\n15\n\n**Value-based relationships will define the future evolution of outsourcing** Four years ago ⁴ , 70% of organizations cited cost savings as the main reason for outsourcing. Two years ago, this figure was in the 40% to 57% range based on the type of outsourcing. Today, this figure has dropped to 34%. Organizations that leverage outcome-based delivery models are even less focused on cost savings as the primary driver—at 30%. Cost reduction, once the primary driver of outsourcing, is no longer the central focus.\n\nThe growing adoption of outcome-based delivery models is redefining the value proposition for outsourced services—from sourcing low-cost talent to sourcing value-added capabilities. Businesses today are prioritizing talent, service quality, global delivery, and agility, in addition to spend optimization to address the evolving business demands. In contrast, organizations that rely exclusively on legacy staff- augmentation models for outsourcing continue to prioritize spend optimization above other factors.\n\nSpend optimization will continue to remain a key driver for outsourcing—specially for first time outsourcers, however, finding incremental opportunities for value on an ongoing basis needs to be organizations’ strategic priority. Executives must continue to collaborate with their providers to explore beyond cost—e.g., building a robust talent pipeline, leveraging AI, driving innovation, enhancing core capabilities, increasing agility.\n\n**Primary drivers for adoption of third-party outsourced services**\n\n22% Desire for partner enabled innovation & transformation\n\n12% Complexity and speed technology changes\n\n19% Evolving risk landscape\n\n20% To gain access to new capabilities\n\n23% Need for flexible and scalable talent model\n\n42% Improved access to talent\n\n33% Improved quality / performance\n\n33% Adoption of global delivery model\n\n27% Better alignment to business strategy and operating model shifts\n\n34% Spend optimization\n\n35% Increasing customer demands\n\n2022 2024 2020\n\n**Respondents who identified ‘cost reduction’ as a primary outsourcing driver**\n\n### **48% 34% 70%**\n\n\\*2022 figure was in the 40% - 57% range based on type of outsourcing Source","pages":[{"line_numbers":[],"markdown":"**Global Outsourcing Survey 2024 |** Multidimensional sourcing: Orchestrating the extended workforce ecosystem\n\n15\n\n**Value-based relationships will define the future evolution of outsourcing** Four years ago ⁴ , 70% of organizations cited cost savings as the main reason for outsourcing. Two years ago, this figure was in the 40% to 57% range based on the type of outsourcing. Today, this figure has dropped to 34%. Organizations that leverage outcome-based delivery models are even less focused on cost savings as the primary driver—at 30%. Cost reduction, once the primary driver of outsourcing, is no longer the central focus.\n\nThe growing adoption of outcome-based delivery models is redefining the value proposition for outsourced services—from sourcing low-cost talent to sourcing value-added capabilities. Businesses today are prioritizing talent, service quality, global delivery, and agility, in addition to spend optimization to address the evolving business demands. In contrast, organizations that rely exclusively on legacy staff- augmentation models for outsourcing continue to prioritize spend optimization above other factors.\n\nSpend optimization will continue to remain a key driver for outsourcing—specially for first time outsourcers, however, finding incremental opportunities for value on an ongoing basis needs to be organizations’ strategic priority. Executives must continue to collaborate with their providers to explore beyond cost—e.g., building a robust talent pipeline, leveraging AI, driving innovation, enhancing core capabilities, increasing agility.\n\n**Primary drivers for adoption of third-party outsourced services**\n\n22% Desire for partner enabled innovation & transformation\n\n12% Complexity and speed technology changes\n\n19% Evolving risk landscape\n\n20% To gain access to new capabilities\n\n23% Need for flexible and scalable talent model\n\n42% Improved access to talent\n\n33% Improved quality / performance\n\n33% Adoption of global delivery model\n\n27% Better alignment to business strategy and operating model shifts\n\n34% Spend optimization\n\n35% Increasing customer demands\n\n2022 2024 2020\n\n**Respondents who identified ‘cost reduction’ as a primary outsourcing driver**\n\n### **48% 34% 70%**\n\n\\*2022 figure was in the 40% - 57% range based on type of outsourcing Source","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**Global Outsourcing Survey 2024 |** Multidimensional sourcing: Orchestrating the extended workforce ecosystem\n\n15\n\n**Value-based relationships will define the future evolution of outsourcing** Four years ago ⁴ , 70% of organizations cited cost savings as the main reason for outsourcing. Two years ago, this figure was in the 40% to 57% range based on the type of outsourcing. Today, this figure has dropped to 34%. Organizations that leverage outcome-based delivery models are even less focused on cost savings as the primary driver—at 30%. Cost reduction, once the primary driver of outsourcing, is no longer the central focus.\n\nThe growing adoption of outcome-based delivery models is redefining the value proposition for outsourced services—from sourcing low-cost talent to sourcing value-added capabilities. Businesses today are prioritizing talent, service quality, global delivery, and agility, in addition to spend optimization to address the evolving business demands. In contrast, organizations that rely exclusively on legacy staff- augmentation models for outsourcing continue to prioritize spend optimization above other factors.\n\nSpend optimization will continue to remain a key driver for outsourcing—specially for first time outsourcers, however, finding incremental opportunities for value on an ongoing basis needs to be organizations’ strategic priority. Executives must continue to collaborate with their providers to explore beyond cost—e.g., building a robust talent pipeline, leveraging AI, driving innovation, enhancing core capabilities, increasing agility.\n\n**Primary drivers for adoption of third-party outsourced services**\n\n22% Desire for partner enabled innovation & transformation\n\n12% Complexity and speed technology changes\n\n19% Evolving risk landscape\n\n20% To gain access to new capabilities\n\n23% Need for flexible and scalable talent model\n\n42% Improved access to talent\n\n33% Improved quality / performance\n\n33% Adoption of global delivery model\n\n27% Better alignment to business strategy and operating model shifts\n\n34% Spend optimization\n\n35% Increasing customer demands\n\n2022 2024 2020\n\n**Respondents who identified ‘cost reduction’ as a primary outsourcing driver**\n\n### **48% 34% 70%**\n\n\\*2022 figure was in the 40% - 57% range based on type of outsourcing Source","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/us-global-outsourcing-survey-2024-report_p15.pdf"]},"response":{"errors":[],"extract_id":"extract_749fc0ab17e3c476e73688b73804b226","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**Global Outsourcing Survey 2024 |** Multidimensional sourcing: Orchestrating the extended workforce ecosystem\n\n15\n\n**Value-based relationships will define the future evolution of outsourcing** Four years ago ⁴ , 70% of organizations cited cost savings as the main reason for outsourcing. Two years ago, this figure was in the 40% to 57% range based on the type of outsourcing. Today, this figure has dropped to 34%. Organizations that leverage outcome-based delivery models are even less focused on cost savings as the primary driver—at 30%. Cost reduction, once the primary driver of outsourcing, is no longer the central focus.\n\nThe growing adoption of outcome-based delivery models is redefining the value proposition for outsourced services—from sourcing low-cost talent to sourcing value-added capabilities. Businesses today are prioritizing talent, service quality, global delivery, and agility, in addition to spend optimization to address the evolving business demands. In contrast, organizations that rely exclusively on legacy staff- augmentation models for outsourcing continue to prioritize spend optimization above other factors.\n\nSpend optimization will continue to remain a key driver for outsourcing—specially for first time outsourcers, however, finding incremental opportunities for value on an ongoing basis needs to be organizations’ strategic priority. Executives must continue to collaborate with their providers to explore beyond cost—e.g., building a robust talent pipeline, leveraging AI, driving innovation, enhancing core capabilities, increasing agility.\n\n**Primary drivers for adoption of third-party outsourced services**\n\n22% Desire for partner enabled innovation & transformation\n\n12% Complexity and speed technology changes\n\n19% Evolving risk landscape\n\n20% To gain access to new capabilities\n\n23% Need for flexible and scalable talent model\n\n42% Improved access to talent\n\n33% Improved quality / performance\n\n33% Adoption of global delivery model\n\n27% Better alignment to business strategy and operating model shifts\n\n34% Spend optimization\n\n35% Increasing customer demands\n\n2022 2024 2020\n\n**Respondents who identified ‘cost reduction’ as a primary outsourcing driver**\n\n### **48% 34% 70%**\n\n\\*2022 figure was in the 40% - 57% range based on type of outsourcing Source"],"full_content":"**Global Outsourcing Survey 2024 |** Multidimensional sourcing: Orchestrating the extended workforce ecosystem\n\n15\n\n**Value-based relationships will define the future evolution of outsourcing** Four years ago ⁴ , 70% of organizations cited cost savings as the main reason for outsourcing. Two years ago, this figure was in the 40% to 57% range based on the type of outsourcing. Today, this figure has dropped to 34%. Organizations that leverage outcome-based delivery models are even less focused on cost savings as the primary driver—at 30%. Cost reduction, once the primary driver of outsourcing, is no longer the central focus.\n\nThe growing adoption of outcome-based delivery models is redefining the value proposition for outsourced services—from sourcing low-cost talent to sourcing value-added capabilities. Businesses today are prioritizing talent, service quality, global delivery, and agility, in addition to spend optimization to address the evolving business demands. In contrast, organizations that rely exclusively on legacy staff- augmentation models for outsourcing continue to prioritize spend optimization above other factors.\n\nSpend optimization will continue to remain a key driver for outsourcing—specially for first time outsourcers, however, finding incremental opportunities for value on an ongoing basis needs to be organizations’ strategic priority. Executives must continue to collaborate with their providers to explore beyond cost—e.g., building a robust talent pipeline, leveraging AI, driving innovation, enhancing core capabilities, increasing agility.\n\n**Primary drivers for adoption of third-party outsourced services**\n\n22% Desire for partner enabled innovation & transformation\n\n12% Complexity and speed technology changes\n\n19% Evolving risk landscape\n\n20% To gain access to new capabilities\n\n23% Need for flexible and scalable talent model\n\n42% Improved access to talent\n\n33% Improved quality / performance\n\n33% Adoption of global delivery model\n\n27% Better alignment to business strategy and operating model shifts\n\n34% Spend optimization\n\n35% Increasing customer demands\n\n2022 2024 2020\n\n**Respondents who identified ‘cost reduction’ as a primary outsourcing driver**\n\n### **48% 34% 70%**\n\n\\*2022 figure was in the 40% - 57% range based on type of outsourcing Source","publish_date":null,"title":"48% 34% 70%","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/us-global-outsourcing-survey-2024-report_p15.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/us-global-outsourcing-survey-2024-report_p15.pdf"},"request":{"example_id":"chart/us-global-outsourcing-survey-2024-report_p15","product_type":"parse","source_file_path":"us-global-outsourcing-survey-2024-report_p15.pdf"},"started_at":"2026-11-04T19:45:08.298487"} +{"completed_at":"2026-11-04T19:45:11.948322","latency_in_ms":2400,"output":{"example_id":"chart/us-global-outsourcing-survey-2024-report_p17","grounded_pages":[],"layout_pages":[],"markdown":"**Global Outsourcing Survey 2024 |** Multidimensional sourcing: Orchestrating the extended workforce ecosystem\n\n17\n\nOne pathway lies in transforming the traditional **Vendor Management Office (VMO)** into an **Extended Workforce Management Office (EWMO)** . This evolution will allow organizations to:\n\nHowever, building and maturing the EWMO is a long-term endeavor. Currently, less than 20% of executives report that their traditional VMO owns the extended workforce strategy, and nearly 70% of organizations acknowledge that their VMO function is not fully mature.\n\nEnhance risk management and regulatory compliance across all workforce models.\n\nCentralize ownership of extended workforce strategy and governance.\n\nDevelop and implement contractual provisions for use of AI and service integration across sourcing models.\n\nIntegrate all sourcing models into the organization’s operating model.\n\nRedefine how to measure service performance. Ensure visibility and control over external labor, including spend and performance management.\n\n**Extended Workforce Management and Governance Owner**\n\n13% Individual Functions / Business Units\n\n16% Risk organization\n\n32% Procurement / Sourcing organization\n\n19% Vendor / Supplier Management Office\n\n20% IT organization\n\n**Maturity of Vendor Management Office (VMO) and Processes**\n\nNo change\n\nDecrease moderately\n\nDecrease significantly\n\nIncrease moderately\n\n**8%**\n\n**25%**\n\n**36%**\n\n**31%**","pages":[{"line_numbers":[],"markdown":"**Global Outsourcing Survey 2024 |** Multidimensional sourcing: Orchestrating the extended workforce ecosystem\n\n17\n\nOne pathway lies in transforming the traditional **Vendor Management Office (VMO)** into an **Extended Workforce Management Office (EWMO)** . This evolution will allow organizations to:\n\nHowever, building and maturing the EWMO is a long-term endeavor. Currently, less than 20% of executives report that their traditional VMO owns the extended workforce strategy, and nearly 70% of organizations acknowledge that their VMO function is not fully mature.\n\nEnhance risk management and regulatory compliance across all workforce models.\n\nCentralize ownership of extended workforce strategy and governance.\n\nDevelop and implement contractual provisions for use of AI and service integration across sourcing models.\n\nIntegrate all sourcing models into the organization’s operating model.\n\nRedefine how to measure service performance. Ensure visibility and control over external labor, including spend and performance management.\n\n**Extended Workforce Management and Governance Owner**\n\n13% Individual Functions / Business Units\n\n16% Risk organization\n\n32% Procurement / Sourcing organization\n\n19% Vendor / Supplier Management Office\n\n20% IT organization\n\n**Maturity of Vendor Management Office (VMO) and Processes**\n\nNo change\n\nDecrease moderately\n\nDecrease significantly\n\nIncrease moderately\n\n**8%**\n\n**25%**\n\n**36%**\n\n**31%**","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**Global Outsourcing Survey 2024 |** Multidimensional sourcing: Orchestrating the extended workforce ecosystem\n\n17\n\nOne pathway lies in transforming the traditional **Vendor Management Office (VMO)** into an **Extended Workforce Management Office (EWMO)** . This evolution will allow organizations to:\n\nHowever, building and maturing the EWMO is a long-term endeavor. Currently, less than 20% of executives report that their traditional VMO owns the extended workforce strategy, and nearly 70% of organizations acknowledge that their VMO function is not fully mature.\n\nEnhance risk management and regulatory compliance across all workforce models.\n\nCentralize ownership of extended workforce strategy and governance.\n\nDevelop and implement contractual provisions for use of AI and service integration across sourcing models.\n\nIntegrate all sourcing models into the organization’s operating model.\n\nRedefine how to measure service performance. Ensure visibility and control over external labor, including spend and performance management.\n\n**Extended Workforce Management and Governance Owner**\n\n13% Individual Functions / Business Units\n\n16% Risk organization\n\n32% Procurement / Sourcing organization\n\n19% Vendor / Supplier Management Office\n\n20% IT organization\n\n**Maturity of Vendor Management Office (VMO) and Processes**\n\nNo change\n\nDecrease moderately\n\nDecrease significantly\n\nIncrease moderately\n\n**8%**\n\n**25%**\n\n**36%**\n\n**31%**","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/us-global-outsourcing-survey-2024-report_p17.pdf"]},"response":{"errors":[],"extract_id":"extract_1ba707c882d94fc19286b163ad1f0083","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**Global Outsourcing Survey 2024 |** Multidimensional sourcing: Orchestrating the extended workforce ecosystem\n\n17\n\nOne pathway lies in transforming the traditional **Vendor Management Office (VMO)** into an **Extended Workforce Management Office (EWMO)** . This evolution will allow organizations to:\n\nHowever, building and maturing the EWMO is a long-term endeavor. Currently, less than 20% of executives report that their traditional VMO owns the extended workforce strategy, and nearly 70% of organizations acknowledge that their VMO function is not fully mature.\n\nEnhance risk management and regulatory compliance across all workforce models.\n\nCentralize ownership of extended workforce strategy and governance.\n\nDevelop and implement contractual provisions for use of AI and service integration across sourcing models.\n\nIntegrate all sourcing models into the organization’s operating model.\n\nRedefine how to measure service performance. Ensure visibility and control over external labor, including spend and performance management.\n\n**Extended Workforce Management and Governance Owner**\n\n13% Individual Functions / Business Units\n\n16% Risk organization\n\n32% Procurement / Sourcing organization\n\n19% Vendor / Supplier Management Office\n\n20% IT organization\n\n**Maturity of Vendor Management Office (VMO) and Processes**\n\nNo change\n\nDecrease moderately\n\nDecrease significantly\n\nIncrease moderately\n\n**8%**\n\n**25%**\n\n**36%**\n\n**31%**"],"full_content":"**Global Outsourcing Survey 2024 |** Multidimensional sourcing: Orchestrating the extended workforce ecosystem\n\n17\n\nOne pathway lies in transforming the traditional **Vendor Management Office (VMO)** into an **Extended Workforce Management Office (EWMO)** . This evolution will allow organizations to:\n\nHowever, building and maturing the EWMO is a long-term endeavor. Currently, less than 20% of executives report that their traditional VMO owns the extended workforce strategy, and nearly 70% of organizations acknowledge that their VMO function is not fully mature.\n\nEnhance risk management and regulatory compliance across all workforce models.\n\nCentralize ownership of extended workforce strategy and governance.\n\nDevelop and implement contractual provisions for use of AI and service integration across sourcing models.\n\nIntegrate all sourcing models into the organization’s operating model.\n\nRedefine how to measure service performance. Ensure visibility and control over external labor, including spend and performance management.\n\n**Extended Workforce Management and Governance Owner**\n\n13% Individual Functions / Business Units\n\n16% Risk organization\n\n32% Procurement / Sourcing organization\n\n19% Vendor / Supplier Management Office\n\n20% IT organization\n\n**Maturity of Vendor Management Office (VMO) and Processes**\n\nNo change\n\nDecrease moderately\n\nDecrease significantly\n\nIncrease moderately\n\n**8%**\n\n**25%**\n\n**36%**\n\n**31%**","publish_date":null,"title":"One pathway lies in transforming the traditional into an Vendor Management Office (VMO) . This evolution will allow organizations to:","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/us-global-outsourcing-survey-2024-report_p17.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/us-global-outsourcing-survey-2024-report_p17.pdf"},"request":{"example_id":"chart/us-global-outsourcing-survey-2024-report_p17","product_type":"parse","source_file_path":"us-global-outsourcing-survey-2024-report_p17.pdf"},"started_at":"2026-11-04T19:45:09.547746"} +{"completed_at":"2026-11-04T19:45:12.661467","latency_in_ms":1231,"output":{"example_id":"chart/us-global-outsourcing-survey-2024-report_p18","grounded_pages":[],"layout_pages":[],"markdown":"**Global Outsourcing Survey 2024 |** Multidimensional sourcing: Orchestrating the extended workforce ecosystem\n\n18\n\n### **“Effective vendor management plays a crucial role in the success of our outsourcing programs, reflecting the** **specialized nature of vendor management capabilities to enhance agility and responsiveness, allowing our** **organization to optimize its operational efficiency while adapting to evolving market demands.”**\n\n_—Managing director / India GIC lead, global multi-brand restaurant company_\n\n### **Key questions for a** **successful sourcing transformation**\n\nTo succeed in their sourcing transformations and sustain the benefits long-term, organizations must address critical strategic questions:\n\n1\\. What is our sourcing strategy, and who owns it?\n\n2\\. What is the case for change? What are the strategic objectives?\n\n3\\. What are the capabilities / skills of the retained and digital workforce, GICs, and outsourcing vendors?\n\n4\\. How do we integrate various talent models, and who governs the extended workforce?\n\n5\\. How do we align with our partners on shared outcomes, and what are the metrics for success?\n\n6\\. How do we ensure technology and data systems provide visibility into our full workforce ecosystem?\n\n7\\. How can we mitigate risk and drive regulatory compliance across all talent models?\n\n8\\. How do we manage change and support our business partners throughout this transformation?\n\nAnswering these questions is essential for optimizing value from sourcing and extended workforce strategies, ensuring organizational resilience and adaptability while managing the sourcing complexity.\n\nDespite the challenges and complexity of sourcing, organizations that invest in a VMO are realizing benefits. This benefit can increase as the VMO matures.\n\nThese returns can further increase by expanding the VMO’s remit to include the broader extended workforce—creating an EWMO that is responsible for managing all talent models and sourcing strategies effectively (see sidebar, “Key questions for a successful sourcing transformation”).\n\n**Third-party spend reduction that is directly attributed to VMO actions**\n\nDeveloping or Emerging VMO Mature VMO\n\n**% organizations that report a reduction by 10% or more**\n\n77%\n\n90%\n\n**% organizations that report a reduction by 20% or more**\n\n22%\n\n61%","pages":[{"line_numbers":[],"markdown":"**Global Outsourcing Survey 2024 |** Multidimensional sourcing: Orchestrating the extended workforce ecosystem\n\n18\n\n### **“Effective vendor management plays a crucial role in the success of our outsourcing programs, reflecting the** **specialized nature of vendor management capabilities to enhance agility and responsiveness, allowing our** **organization to optimize its operational efficiency while adapting to evolving market demands.”**\n\n_—Managing director / India GIC lead, global multi-brand restaurant company_\n\n### **Key questions for a** **successful sourcing transformation**\n\nTo succeed in their sourcing transformations and sustain the benefits long-term, organizations must address critical strategic questions:\n\n1\\. What is our sourcing strategy, and who owns it?\n\n2\\. What is the case for change? What are the strategic objectives?\n\n3\\. What are the capabilities / skills of the retained and digital workforce, GICs, and outsourcing vendors?\n\n4\\. How do we integrate various talent models, and who governs the extended workforce?\n\n5\\. How do we align with our partners on shared outcomes, and what are the metrics for success?\n\n6\\. How do we ensure technology and data systems provide visibility into our full workforce ecosystem?\n\n7\\. How can we mitigate risk and drive regulatory compliance across all talent models?\n\n8\\. How do we manage change and support our business partners throughout this transformation?\n\nAnswering these questions is essential for optimizing value from sourcing and extended workforce strategies, ensuring organizational resilience and adaptability while managing the sourcing complexity.\n\nDespite the challenges and complexity of sourcing, organizations that invest in a VMO are realizing benefits. This benefit can increase as the VMO matures.\n\nThese returns can further increase by expanding the VMO’s remit to include the broader extended workforce—creating an EWMO that is responsible for managing all talent models and sourcing strategies effectively (see sidebar, “Key questions for a successful sourcing transformation”).\n\n**Third-party spend reduction that is directly attributed to VMO actions**\n\nDeveloping or Emerging VMO Mature VMO\n\n**% organizations that report a reduction by 10% or more**\n\n77%\n\n90%\n\n**% organizations that report a reduction by 20% or more**\n\n22%\n\n61%","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**Global Outsourcing Survey 2024 |** Multidimensional sourcing: Orchestrating the extended workforce ecosystem\n\n18\n\n### **“Effective vendor management plays a crucial role in the success of our outsourcing programs, reflecting the** **specialized nature of vendor management capabilities to enhance agility and responsiveness, allowing our** **organization to optimize its operational efficiency while adapting to evolving market demands.”**\n\n_—Managing director / India GIC lead, global multi-brand restaurant company_\n\n### **Key questions for a** **successful sourcing transformation**\n\nTo succeed in their sourcing transformations and sustain the benefits long-term, organizations must address critical strategic questions:\n\n1\\. What is our sourcing strategy, and who owns it?\n\n2\\. What is the case for change? What are the strategic objectives?\n\n3\\. What are the capabilities / skills of the retained and digital workforce, GICs, and outsourcing vendors?\n\n4\\. How do we integrate various talent models, and who governs the extended workforce?\n\n5\\. How do we align with our partners on shared outcomes, and what are the metrics for success?\n\n6\\. How do we ensure technology and data systems provide visibility into our full workforce ecosystem?\n\n7\\. How can we mitigate risk and drive regulatory compliance across all talent models?\n\n8\\. How do we manage change and support our business partners throughout this transformation?\n\nAnswering these questions is essential for optimizing value from sourcing and extended workforce strategies, ensuring organizational resilience and adaptability while managing the sourcing complexity.\n\nDespite the challenges and complexity of sourcing, organizations that invest in a VMO are realizing benefits. This benefit can increase as the VMO matures.\n\nThese returns can further increase by expanding the VMO’s remit to include the broader extended workforce—creating an EWMO that is responsible for managing all talent models and sourcing strategies effectively (see sidebar, “Key questions for a successful sourcing transformation”).\n\n**Third-party spend reduction that is directly attributed to VMO actions**\n\nDeveloping or Emerging VMO Mature VMO\n\n**% organizations that report a reduction by 10% or more**\n\n77%\n\n90%\n\n**% organizations that report a reduction by 20% or more**\n\n22%\n\n61%","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/us-global-outsourcing-survey-2024-report_p18.pdf"]},"response":{"errors":[],"extract_id":"extract_5043cee044af10ef1fd3537be4757a5c","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**Global Outsourcing Survey 2024 |** Multidimensional sourcing: Orchestrating the extended workforce ecosystem\n\n18\n\n### **“Effective vendor management plays a crucial role in the success of our outsourcing programs, reflecting the** **specialized nature of vendor management capabilities to enhance agility and responsiveness, allowing our** **organization to optimize its operational efficiency while adapting to evolving market demands.”**\n\n_—Managing director / India GIC lead, global multi-brand restaurant company_\n\n### **Key questions for a** **successful sourcing transformation**\n\nTo succeed in their sourcing transformations and sustain the benefits long-term, organizations must address critical strategic questions:\n\n1\\. What is our sourcing strategy, and who owns it?\n\n2\\. What is the case for change? What are the strategic objectives?\n\n3\\. What are the capabilities / skills of the retained and digital workforce, GICs, and outsourcing vendors?\n\n4\\. How do we integrate various talent models, and who governs the extended workforce?\n\n5\\. How do we align with our partners on shared outcomes, and what are the metrics for success?\n\n6\\. How do we ensure technology and data systems provide visibility into our full workforce ecosystem?\n\n7\\. How can we mitigate risk and drive regulatory compliance across all talent models?\n\n8\\. How do we manage change and support our business partners throughout this transformation?\n\nAnswering these questions is essential for optimizing value from sourcing and extended workforce strategies, ensuring organizational resilience and adaptability while managing the sourcing complexity.\n\nDespite the challenges and complexity of sourcing, organizations that invest in a VMO are realizing benefits. This benefit can increase as the VMO matures.\n\nThese returns can further increase by expanding the VMO’s remit to include the broader extended workforce—creating an EWMO that is responsible for managing all talent models and sourcing strategies effectively (see sidebar, “Key questions for a successful sourcing transformation”).\n\n**Third-party spend reduction that is directly attributed to VMO actions**\n\nDeveloping or Emerging VMO Mature VMO\n\n**% organizations that report a reduction by 10% or more**\n\n77%\n\n90%\n\n**% organizations that report a reduction by 20% or more**\n\n22%\n\n61%"],"full_content":"**Global Outsourcing Survey 2024 |** Multidimensional sourcing: Orchestrating the extended workforce ecosystem\n\n18\n\n### **“Effective vendor management plays a crucial role in the success of our outsourcing programs, reflecting the** **specialized nature of vendor management capabilities to enhance agility and responsiveness, allowing our** **organization to optimize its operational efficiency while adapting to evolving market demands.”**\n\n_—Managing director / India GIC lead, global multi-brand restaurant company_\n\n### **Key questions for a** **successful sourcing transformation**\n\nTo succeed in their sourcing transformations and sustain the benefits long-term, organizations must address critical strategic questions:\n\n1\\. What is our sourcing strategy, and who owns it?\n\n2\\. What is the case for change? What are the strategic objectives?\n\n3\\. What are the capabilities / skills of the retained and digital workforce, GICs, and outsourcing vendors?\n\n4\\. How do we integrate various talent models, and who governs the extended workforce?\n\n5\\. How do we align with our partners on shared outcomes, and what are the metrics for success?\n\n6\\. How do we ensure technology and data systems provide visibility into our full workforce ecosystem?\n\n7\\. How can we mitigate risk and drive regulatory compliance across all talent models?\n\n8\\. How do we manage change and support our business partners throughout this transformation?\n\nAnswering these questions is essential for optimizing value from sourcing and extended workforce strategies, ensuring organizational resilience and adaptability while managing the sourcing complexity.\n\nDespite the challenges and complexity of sourcing, organizations that invest in a VMO are realizing benefits. This benefit can increase as the VMO matures.\n\nThese returns can further increase by expanding the VMO’s remit to include the broader extended workforce—creating an EWMO that is responsible for managing all talent models and sourcing strategies effectively (see sidebar, “Key questions for a successful sourcing transformation”).\n\n**Third-party spend reduction that is directly attributed to VMO actions**\n\nDeveloping or Emerging VMO Mature VMO\n\n**% organizations that report a reduction by 10% or more**\n\n77%\n\n90%\n\n**% organizations that report a reduction by 20% or more**\n\n22%\n\n61%","publish_date":null,"title":"Key questions for a successful sourcing transformation “Effective vendor management plays a crucial role in the success of our outsourcing","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/us-global-outsourcing-survey-2024-report_p18.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/us-global-outsourcing-survey-2024-report_p18.pdf"},"request":{"example_id":"chart/us-global-outsourcing-survey-2024-report_p18","product_type":"parse","source_file_path":"us-global-outsourcing-survey-2024-report_p18.pdf"},"started_at":"2026-11-04T19:45:11.429583"} +{"completed_at":"2026-11-04T19:45:14.764300","latency_in_ms":1674,"output":{"example_id":"chart/us-global-outsourcing-survey-2024-report_p9","grounded_pages":[],"layout_pages":[],"markdown":"**Global Outsourcing Survey 2024 |** Multidimensional sourcing: Orchestrating the extended workforce ecosystem\n\n9\n\nWhat is driving organizations’ adoption of insourcing? It’s not dissatisfaction with outsourced services.\n\nThe leading drivers for insourcing include gaining better control over service quality and performance (beyond what outsourcing can provide) and growing strategic capabilities in-house, while eliminating vendor markup.\n\nOrganizations will likely continue balancing insourcing and outsourcing, making decisions about which functions to bring in-house and which to outsource. Both models will likely coexist as part of future sourcing strategies. This trend has potential implications on how organizations contract for outsourced services. Insourcing events can trigger changes to total cost of service, loss of knowledge, and change impacts to business stakeholders. Organizations should evaluate the commercial and business impact of insourcing based on current contract provisions. And by enhancing flexibility of contract provisions (e.g., changes to scope, service levels, term), organizations can be better prepared to leverage insourcing, when needed.\n\n**Of the respondents who are exploring more insourcing opportunities**\n\n### **80%**\n\nare either maintining or increasing their investment in outsourced services\n\n### **82%**\n\nreported ‘meets’ or ‘exceeds’ expectations from their outsourced services\n\n**Primary drivers for insourcing scope or services**\n\n12% For confidentiality / intellectual property (IP) reasons\n\n38% Better cultural alignment\n\n38% For regulatory and compliance reasons\n\n68% Better control over service quality and performance\n\n46% For risk and security reasons\n\n39% For improved transparency in service operations\n\n39% Have a more balanced talent sourcing ecosystem (risk diversification)\n\n56% Spend optimization\n\n64% Invest in and grow strategic capabilities in-house","pages":[{"line_numbers":[],"markdown":"**Global Outsourcing Survey 2024 |** Multidimensional sourcing: Orchestrating the extended workforce ecosystem\n\n9\n\nWhat is driving organizations’ adoption of insourcing? It’s not dissatisfaction with outsourced services.\n\nThe leading drivers for insourcing include gaining better control over service quality and performance (beyond what outsourcing can provide) and growing strategic capabilities in-house, while eliminating vendor markup.\n\nOrganizations will likely continue balancing insourcing and outsourcing, making decisions about which functions to bring in-house and which to outsource. Both models will likely coexist as part of future sourcing strategies. This trend has potential implications on how organizations contract for outsourced services. Insourcing events can trigger changes to total cost of service, loss of knowledge, and change impacts to business stakeholders. Organizations should evaluate the commercial and business impact of insourcing based on current contract provisions. And by enhancing flexibility of contract provisions (e.g., changes to scope, service levels, term), organizations can be better prepared to leverage insourcing, when needed.\n\n**Of the respondents who are exploring more insourcing opportunities**\n\n### **80%**\n\nare either maintining or increasing their investment in outsourced services\n\n### **82%**\n\nreported ‘meets’ or ‘exceeds’ expectations from their outsourced services\n\n**Primary drivers for insourcing scope or services**\n\n12% For confidentiality / intellectual property (IP) reasons\n\n38% Better cultural alignment\n\n38% For regulatory and compliance reasons\n\n68% Better control over service quality and performance\n\n46% For risk and security reasons\n\n39% For improved transparency in service operations\n\n39% Have a more balanced talent sourcing ecosystem (risk diversification)\n\n56% Spend optimization\n\n64% Invest in and grow strategic capabilities in-house","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"**Global Outsourcing Survey 2024 |** Multidimensional sourcing: Orchestrating the extended workforce ecosystem\n\n9\n\nWhat is driving organizations’ adoption of insourcing? It’s not dissatisfaction with outsourced services.\n\nThe leading drivers for insourcing include gaining better control over service quality and performance (beyond what outsourcing can provide) and growing strategic capabilities in-house, while eliminating vendor markup.\n\nOrganizations will likely continue balancing insourcing and outsourcing, making decisions about which functions to bring in-house and which to outsource. Both models will likely coexist as part of future sourcing strategies. This trend has potential implications on how organizations contract for outsourced services. Insourcing events can trigger changes to total cost of service, loss of knowledge, and change impacts to business stakeholders. Organizations should evaluate the commercial and business impact of insourcing based on current contract provisions. And by enhancing flexibility of contract provisions (e.g., changes to scope, service levels, term), organizations can be better prepared to leverage insourcing, when needed.\n\n**Of the respondents who are exploring more insourcing opportunities**\n\n### **80%**\n\nare either maintining or increasing their investment in outsourced services\n\n### **82%**\n\nreported ‘meets’ or ‘exceeds’ expectations from their outsourced services\n\n**Primary drivers for insourcing scope or services**\n\n12% For confidentiality / intellectual property (IP) reasons\n\n38% Better cultural alignment\n\n38% For regulatory and compliance reasons\n\n68% Better control over service quality and performance\n\n46% For risk and security reasons\n\n39% For improved transparency in service operations\n\n39% Have a more balanced talent sourcing ecosystem (risk diversification)\n\n56% Spend optimization\n\n64% Invest in and grow strategic capabilities in-house","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/us-global-outsourcing-survey-2024-report_p9.pdf"]},"response":{"errors":[],"extract_id":"extract_af98c185844a417fc1710a421659f8f9","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["**Global Outsourcing Survey 2024 |** Multidimensional sourcing: Orchestrating the extended workforce ecosystem\n\n9\n\nWhat is driving organizations’ adoption of insourcing? It’s not dissatisfaction with outsourced services.\n\nThe leading drivers for insourcing include gaining better control over service quality and performance (beyond what outsourcing can provide) and growing strategic capabilities in-house, while eliminating vendor markup.\n\nOrganizations will likely continue balancing insourcing and outsourcing, making decisions about which functions to bring in-house and which to outsource. Both models will likely coexist as part of future sourcing strategies. This trend has potential implications on how organizations contract for outsourced services. Insourcing events can trigger changes to total cost of service, loss of knowledge, and change impacts to business stakeholders. Organizations should evaluate the commercial and business impact of insourcing based on current contract provisions. And by enhancing flexibility of contract provisions (e.g., changes to scope, service levels, term), organizations can be better prepared to leverage insourcing, when needed.\n\n**Of the respondents who are exploring more insourcing opportunities**\n\n### **80%**\n\nare either maintining or increasing their investment in outsourced services\n\n### **82%**\n\nreported ‘meets’ or ‘exceeds’ expectations from their outsourced services\n\n**Primary drivers for insourcing scope or services**\n\n12% For confidentiality / intellectual property (IP) reasons\n\n38% Better cultural alignment\n\n38% For regulatory and compliance reasons\n\n68% Better control over service quality and performance\n\n46% For risk and security reasons\n\n39% For improved transparency in service operations\n\n39% Have a more balanced talent sourcing ecosystem (risk diversification)\n\n56% Spend optimization\n\n64% Invest in and grow strategic capabilities in-house"],"full_content":"**Global Outsourcing Survey 2024 |** Multidimensional sourcing: Orchestrating the extended workforce ecosystem\n\n9\n\nWhat is driving organizations’ adoption of insourcing? It’s not dissatisfaction with outsourced services.\n\nThe leading drivers for insourcing include gaining better control over service quality and performance (beyond what outsourcing can provide) and growing strategic capabilities in-house, while eliminating vendor markup.\n\nOrganizations will likely continue balancing insourcing and outsourcing, making decisions about which functions to bring in-house and which to outsource. Both models will likely coexist as part of future sourcing strategies. This trend has potential implications on how organizations contract for outsourced services. Insourcing events can trigger changes to total cost of service, loss of knowledge, and change impacts to business stakeholders. Organizations should evaluate the commercial and business impact of insourcing based on current contract provisions. And by enhancing flexibility of contract provisions (e.g., changes to scope, service levels, term), organizations can be better prepared to leverage insourcing, when needed.\n\n**Of the respondents who are exploring more insourcing opportunities**\n\n### **80%**\n\nare either maintining or increasing their investment in outsourced services\n\n### **82%**\n\nreported ‘meets’ or ‘exceeds’ expectations from their outsourced services\n\n**Primary drivers for insourcing scope or services**\n\n12% For confidentiality / intellectual property (IP) reasons\n\n38% Better cultural alignment\n\n38% For regulatory and compliance reasons\n\n68% Better control over service quality and performance\n\n46% For risk and security reasons\n\n39% For improved transparency in service operations\n\n39% Have a more balanced talent sourcing ecosystem (risk diversification)\n\n56% Spend optimization\n\n64% Invest in and grow strategic capabilities in-house","publish_date":null,"title":"80% 82%","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/us-global-outsourcing-survey-2024-report_p9.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/us-global-outsourcing-survey-2024-report_p9.pdf"},"request":{"example_id":"chart/us-global-outsourcing-survey-2024-report_p9","product_type":"parse","source_file_path":"us-global-outsourcing-survey-2024-report_p9.pdf"},"started_at":"2026-11-04T19:45:13.089963"} +{"completed_at":"2026-11-04T19:45:16.297310","latency_in_ms":2067,"output":{"example_id":"chart/wtr24_ch2_e_p14","grounded_pages":[],"layout_pages":[],"markdown":"B TRADE AND INCOME CONVERGENCE\n\n\\- 43 -\n\nTRADE AND INCLUSIVENESS: HOW TO MAKE TRADE WORK FOR ALL\n\n**Agriculture producer subsidies used to be concentrated in high-income economies, but in the past decade large middle-income economies have accounted for almost half of global agriculture subsidies** . There are important gaps in the available knowledge about subsidy practices; most of the data that are available for a broad scope of economies and sectors refer only to counts of measures and have important shortcomings (IMF et al., 2022). Agriculture is one exception where systematic evidence on quantities of subsidies exists, having been collected by the Organisation for Economic Co-operation and Development (OECD). At their peak in 1999, subsidies to agriculture producers in high- income economies reached 8.4 per cent of global agriculture production value. These subsidies declined rapidly in the 15 years following, and they have remained at between 3.6 and 4.2 per cent of global agriculture production value since 2014. ¹² Large emerging economies, on the other hand, saw rapid increase in agriculture production subsidies between 1999 and 2008, which have remained at 3.0 to 3.8 per cent of global agriculture production value since then.\n\n**_(ii) Some trade policies in developing economies can hold back their economic development_**\n\n**Import tariffs remain relatively high in low- income economies and LDCs, while services trade barriers are relatively high in the group of lower-middle-income economies** . As discussed\n\nabove, many of the channels through which openness fosters growth relate to imports. Higher tariffs act from this perspective as a barrier to growth. In fact, applied tariffs ¹³ have remained relatively high in low-income economies and LDCs over the past decade, at around 8 per cent on average, compared to 5 per cent in middle- income economies and 3 per cent in high-income economies. If we consider the group of diverging economies (as defined in section B.1), they apply higher average tariffs compared to other economies in their income group, and this difference has only increased in the past decade. Barriers to services imports, on the other hand, show, on average, little difference between developing and high-income economies. An exception is the lower-middle-income group, in which services trade restrictions are, on average, 17 per cent higher than in high-income economies. ¹⁴\n\n**While GSP schemes may offer developing economies access to consumers in high- income economies, reciprocal trade agreements still offer significant potential to boost trade.** Economic integration with nearby economies provides opportunities to increase market size, and thus to achieve economies of scale and attract imports and FDI. For Africa, the African Continental Free Trade Area could be an important step forward in this regard. Integration with high-income economies has the additional potential benefit of stimulating technology transfers by facilitating GVC participation and providing access to sophisticated consumers\n\nFigure B.6: Growing number of concerns about SPS and TBT measures raised by low- and middle-income members, 1995-2023\n\n0\n\n20\n\n40\n\n60\n\n80\n\n0%\n\n25%\n\n50%\n\n75%\n\n100%\n\n1995\n\n1996\n\n1998\n\n2000\n\n2002\n\n2004\n\n2006\n\n2008\n\n2010\n\n2012\n\n2014\n\n2016\n\n2018\n\n2020\n\n1997\n\n1999\n\n2001\n\n2003\n\n2005\n\n2007\n\n2009\n\n2011\n\n2015\n\n2013\n\n2017\n\n2019\n\n2023\n\n2022\n\n2021\n\nConcerns raised by at least one low- or middle-income economy\n\nNumber of low- and middle-income economies' SPS\n\nand TBT concerns\n\nShare of low- and middle-income economies' SPS\n\nand TBT concerns addressed to high-income economies\n\nShare of concerns addressed to high-income economies\n\nSource: Authors’ calculations, based on WTO data on SPS and TBT trade concerns. Note: The figure displays the evolution of the number and share of trade concerns raised by low- and middle-income members in the SPS and TBT committees over the period 1995-2023. The income groups are based on the 1995 World Bank classification.","pages":[{"line_numbers":[],"markdown":"B TRADE AND INCOME CONVERGENCE\n\n\\- 43 -\n\nTRADE AND INCLUSIVENESS: HOW TO MAKE TRADE WORK FOR ALL\n\n**Agriculture producer subsidies used to be concentrated in high-income economies, but in the past decade large middle-income economies have accounted for almost half of global agriculture subsidies** . There are important gaps in the available knowledge about subsidy practices; most of the data that are available for a broad scope of economies and sectors refer only to counts of measures and have important shortcomings (IMF et al., 2022). Agriculture is one exception where systematic evidence on quantities of subsidies exists, having been collected by the Organisation for Economic Co-operation and Development (OECD). At their peak in 1999, subsidies to agriculture producers in high- income economies reached 8.4 per cent of global agriculture production value. These subsidies declined rapidly in the 15 years following, and they have remained at between 3.6 and 4.2 per cent of global agriculture production value since 2014. ¹² Large emerging economies, on the other hand, saw rapid increase in agriculture production subsidies between 1999 and 2008, which have remained at 3.0 to 3.8 per cent of global agriculture production value since then.\n\n**_(ii) Some trade policies in developing economies can hold back their economic development_**\n\n**Import tariffs remain relatively high in low- income economies and LDCs, while services trade barriers are relatively high in the group of lower-middle-income economies** . As discussed\n\nabove, many of the channels through which openness fosters growth relate to imports. Higher tariffs act from this perspective as a barrier to growth. In fact, applied tariffs ¹³ have remained relatively high in low-income economies and LDCs over the past decade, at around 8 per cent on average, compared to 5 per cent in middle- income economies and 3 per cent in high-income economies. If we consider the group of diverging economies (as defined in section B.1), they apply higher average tariffs compared to other economies in their income group, and this difference has only increased in the past decade. Barriers to services imports, on the other hand, show, on average, little difference between developing and high-income economies. An exception is the lower-middle-income group, in which services trade restrictions are, on average, 17 per cent higher than in high-income economies. ¹⁴\n\n**While GSP schemes may offer developing economies access to consumers in high- income economies, reciprocal trade agreements still offer significant potential to boost trade.** Economic integration with nearby economies provides opportunities to increase market size, and thus to achieve economies of scale and attract imports and FDI. For Africa, the African Continental Free Trade Area could be an important step forward in this regard. Integration with high-income economies has the additional potential benefit of stimulating technology transfers by facilitating GVC participation and providing access to sophisticated consumers\n\nFigure B.6: Growing number of concerns about SPS and TBT measures raised by low- and middle-income members, 1995-2023\n\n0\n\n20\n\n40\n\n60\n\n80\n\n0%\n\n25%\n\n50%\n\n75%\n\n100%\n\n1995\n\n1996\n\n1998\n\n2000\n\n2002\n\n2004\n\n2006\n\n2008\n\n2010\n\n2012\n\n2014\n\n2016\n\n2018\n\n2020\n\n1997\n\n1999\n\n2001\n\n2003\n\n2005\n\n2007\n\n2009\n\n2011\n\n2015\n\n2013\n\n2017\n\n2019\n\n2023\n\n2022\n\n2021\n\nConcerns raised by at least one low- or middle-income economy\n\nNumber of low- and middle-income economies' SPS\n\nand TBT concerns\n\nShare of low- and middle-income economies' SPS\n\nand TBT concerns addressed to high-income economies\n\nShare of concerns addressed to high-income economies\n\nSource: Authors’ calculations, based on WTO data on SPS and TBT trade concerns. Note: The figure displays the evolution of the number and share of trade concerns raised by low- and middle-income members in the SPS and TBT committees over the period 1995-2023. The income groups are based on the 1995 World Bank classification.","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"B TRADE AND INCOME CONVERGENCE\n\n\\- 43 -\n\nTRADE AND INCLUSIVENESS: HOW TO MAKE TRADE WORK FOR ALL\n\n**Agriculture producer subsidies used to be concentrated in high-income economies, but in the past decade large middle-income economies have accounted for almost half of global agriculture subsidies** . There are important gaps in the available knowledge about subsidy practices; most of the data that are available for a broad scope of economies and sectors refer only to counts of measures and have important shortcomings (IMF et al., 2022). Agriculture is one exception where systematic evidence on quantities of subsidies exists, having been collected by the Organisation for Economic Co-operation and Development (OECD). At their peak in 1999, subsidies to agriculture producers in high- income economies reached 8.4 per cent of global agriculture production value. These subsidies declined rapidly in the 15 years following, and they have remained at between 3.6 and 4.2 per cent of global agriculture production value since 2014. ¹² Large emerging economies, on the other hand, saw rapid increase in agriculture production subsidies between 1999 and 2008, which have remained at 3.0 to 3.8 per cent of global agriculture production value since then.\n\n**_(ii) Some trade policies in developing economies can hold back their economic development_**\n\n**Import tariffs remain relatively high in low- income economies and LDCs, while services trade barriers are relatively high in the group of lower-middle-income economies** . As discussed\n\nabove, many of the channels through which openness fosters growth relate to imports. Higher tariffs act from this perspective as a barrier to growth. In fact, applied tariffs ¹³ have remained relatively high in low-income economies and LDCs over the past decade, at around 8 per cent on average, compared to 5 per cent in middle- income economies and 3 per cent in high-income economies. If we consider the group of diverging economies (as defined in section B.1), they apply higher average tariffs compared to other economies in their income group, and this difference has only increased in the past decade. Barriers to services imports, on the other hand, show, on average, little difference between developing and high-income economies. An exception is the lower-middle-income group, in which services trade restrictions are, on average, 17 per cent higher than in high-income economies. ¹⁴\n\n**While GSP schemes may offer developing economies access to consumers in high- income economies, reciprocal trade agreements still offer significant potential to boost trade.** Economic integration with nearby economies provides opportunities to increase market size, and thus to achieve economies of scale and attract imports and FDI. For Africa, the African Continental Free Trade Area could be an important step forward in this regard. Integration with high-income economies has the additional potential benefit of stimulating technology transfers by facilitating GVC participation and providing access to sophisticated consumers\n\nFigure B.6: Growing number of concerns about SPS and TBT measures raised by low- and middle-income members, 1995-2023\n\n0\n\n20\n\n40\n\n60\n\n80\n\n0%\n\n25%\n\n50%\n\n75%\n\n100%\n\n1995\n\n1996\n\n1998\n\n2000\n\n2002\n\n2004\n\n2006\n\n2008\n\n2010\n\n2012\n\n2014\n\n2016\n\n2018\n\n2020\n\n1997\n\n1999\n\n2001\n\n2003\n\n2005\n\n2007\n\n2009\n\n2011\n\n2015\n\n2013\n\n2017\n\n2019\n\n2023\n\n2022\n\n2021\n\nConcerns raised by at least one low- or middle-income economy\n\nNumber of low- and middle-income economies' SPS\n\nand TBT concerns\n\nShare of low- and middle-income economies' SPS\n\nand TBT concerns addressed to high-income economies\n\nShare of concerns addressed to high-income economies\n\nSource: Authors’ calculations, based on WTO data on SPS and TBT trade concerns. Note: The figure displays the evolution of the number and share of trade concerns raised by low- and middle-income members in the SPS and TBT committees over the period 1995-2023. The income groups are based on the 1995 World Bank classification.","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/wtr24_ch2_e_p14.pdf"]},"response":{"errors":[],"extract_id":"extract_e2ea78693a2bbf38519ccc3076c30f1d","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["B TRADE AND INCOME CONVERGENCE\n\n\\- 43 -\n\nTRADE AND INCLUSIVENESS: HOW TO MAKE TRADE WORK FOR ALL\n\n**Agriculture producer subsidies used to be concentrated in high-income economies, but in the past decade large middle-income economies have accounted for almost half of global agriculture subsidies** . There are important gaps in the available knowledge about subsidy practices; most of the data that are available for a broad scope of economies and sectors refer only to counts of measures and have important shortcomings (IMF et al., 2022). Agriculture is one exception where systematic evidence on quantities of subsidies exists, having been collected by the Organisation for Economic Co-operation and Development (OECD). At their peak in 1999, subsidies to agriculture producers in high- income economies reached 8.4 per cent of global agriculture production value. These subsidies declined rapidly in the 15 years following, and they have remained at between 3.6 and 4.2 per cent of global agriculture production value since 2014. ¹² Large emerging economies, on the other hand, saw rapid increase in agriculture production subsidies between 1999 and 2008, which have remained at 3.0 to 3.8 per cent of global agriculture production value since then.\n\n**_(ii) Some trade policies in developing economies can hold back their economic development_**\n\n**Import tariffs remain relatively high in low- income economies and LDCs, while services trade barriers are relatively high in the group of lower-middle-income economies** . As discussed\n\nabove, many of the channels through which openness fosters growth relate to imports. Higher tariffs act from this perspective as a barrier to growth. In fact, applied tariffs ¹³ have remained relatively high in low-income economies and LDCs over the past decade, at around 8 per cent on average, compared to 5 per cent in middle- income economies and 3 per cent in high-income economies. If we consider the group of diverging economies (as defined in section B.1), they apply higher average tariffs compared to other economies in their income group, and this difference has only increased in the past decade. Barriers to services imports, on the other hand, show, on average, little difference between developing and high-income economies. An exception is the lower-middle-income group, in which services trade restrictions are, on average, 17 per cent higher than in high-income economies. ¹⁴\n\n**While GSP schemes may offer developing economies access to consumers in high- income economies, reciprocal trade agreements still offer significant potential to boost trade.** Economic integration with nearby economies provides opportunities to increase market size, and thus to achieve economies of scale and attract imports and FDI. For Africa, the African Continental Free Trade Area could be an important step forward in this regard. Integration with high-income economies has the additional potential benefit of stimulating technology transfers by facilitating GVC participation and providing access to sophisticated consumers\n\nFigure B.6: Growing number of concerns about SPS and TBT measures raised by low- and middle-income members, 1995-2023\n\n0\n\n20\n\n40\n\n60\n\n80\n\n0%\n\n25%\n\n50%\n\n75%\n\n100%\n\n1995\n\n1996\n\n1998\n\n2000\n\n2002\n\n2004\n\n2006\n\n2008\n\n2010\n\n2012\n\n2014\n\n2016\n\n2018\n\n2020\n\n1997\n\n1999\n\n2001\n\n2003\n\n2005\n\n2007\n\n2009\n\n2011\n\n2015\n\n2013\n\n2017\n\n2019\n\n2023\n\n2022\n\n2021\n\nConcerns raised by at least one low- or middle-income economy\n\nNumber of low- and middle-income economies' SPS\n\nand TBT concerns\n\nShare of low- and middle-income economies' SPS\n\nand TBT concerns addressed to high-income economies\n\nShare of concerns addressed to high-income economies\n\nSource: Authors’ calculations, based on WTO data on SPS and TBT trade concerns. Note: The figure displays the evolution of the number and share of trade concerns raised by low- and middle-income members in the SPS and TBT committees over the period 1995-2023. The income groups are based on the 1995 World Bank classification."],"full_content":"B TRADE AND INCOME CONVERGENCE\n\n\\- 43 -\n\nTRADE AND INCLUSIVENESS: HOW TO MAKE TRADE WORK FOR ALL\n\n**Agriculture producer subsidies used to be concentrated in high-income economies, but in the past decade large middle-income economies have accounted for almost half of global agriculture subsidies** . There are important gaps in the available knowledge about subsidy practices; most of the data that are available for a broad scope of economies and sectors refer only to counts of measures and have important shortcomings (IMF et al., 2022). Agriculture is one exception where systematic evidence on quantities of subsidies exists, having been collected by the Organisation for Economic Co-operation and Development (OECD). At their peak in 1999, subsidies to agriculture producers in high- income economies reached 8.4 per cent of global agriculture production value. These subsidies declined rapidly in the 15 years following, and they have remained at between 3.6 and 4.2 per cent of global agriculture production value since 2014. ¹² Large emerging economies, on the other hand, saw rapid increase in agriculture production subsidies between 1999 and 2008, which have remained at 3.0 to 3.8 per cent of global agriculture production value since then.\n\n**_(ii) Some trade policies in developing economies can hold back their economic development_**\n\n**Import tariffs remain relatively high in low- income economies and LDCs, while services trade barriers are relatively high in the group of lower-middle-income economies** . As discussed\n\nabove, many of the channels through which openness fosters growth relate to imports. Higher tariffs act from this perspective as a barrier to growth. In fact, applied tariffs ¹³ have remained relatively high in low-income economies and LDCs over the past decade, at around 8 per cent on average, compared to 5 per cent in middle- income economies and 3 per cent in high-income economies. If we consider the group of diverging economies (as defined in section B.1), they apply higher average tariffs compared to other economies in their income group, and this difference has only increased in the past decade. Barriers to services imports, on the other hand, show, on average, little difference between developing and high-income economies. An exception is the lower-middle-income group, in which services trade restrictions are, on average, 17 per cent higher than in high-income economies. ¹⁴\n\n**While GSP schemes may offer developing economies access to consumers in high- income economies, reciprocal trade agreements still offer significant potential to boost trade.** Economic integration with nearby economies provides opportunities to increase market size, and thus to achieve economies of scale and attract imports and FDI. For Africa, the African Continental Free Trade Area could be an important step forward in this regard. Integration with high-income economies has the additional potential benefit of stimulating technology transfers by facilitating GVC participation and providing access to sophisticated consumers\n\nFigure B.6: Growing number of concerns about SPS and TBT measures raised by low- and middle-income members, 1995-2023\n\n0\n\n20\n\n40\n\n60\n\n80\n\n0%\n\n25%\n\n50%\n\n75%\n\n100%\n\n1995\n\n1996\n\n1998\n\n2000\n\n2002\n\n2004\n\n2006\n\n2008\n\n2010\n\n2012\n\n2014\n\n2016\n\n2018\n\n2020\n\n1997\n\n1999\n\n2001\n\n2003\n\n2005\n\n2007\n\n2009\n\n2011\n\n2015\n\n2013\n\n2017\n\n2019\n\n2023\n\n2022\n\n2021\n\nConcerns raised by at least one low- or middle-income economy\n\nNumber of low- and middle-income economies' SPS\n\nand TBT concerns\n\nShare of low- and middle-income economies' SPS\n\nand TBT concerns addressed to high-income economies\n\nShare of concerns addressed to high-income economies\n\nSource: Authors’ calculations, based on WTO data on SPS and TBT trade concerns. Note: The figure displays the evolution of the number and share of trade concerns raised by low- and middle-income members in the SPS and TBT committees over the period 1995-2023. The income groups are based on the 1995 World Bank classification.","publish_date":null,"title":"B","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/wtr24_ch2_e_p14.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/wtr24_ch2_e_p14.pdf"},"request":{"example_id":"chart/wtr24_ch2_e_p14","product_type":"parse","source_file_path":"wtr24_ch2_e_p14.pdf"},"started_at":"2026-11-04T19:45:14.229776"} +{"completed_at":"2026-11-04T19:45:16.631257","latency_in_ms":1137,"output":{"example_id":"chart/wtr24_ch2_e_p3","grounded_pages":[],"layout_pages":[],"markdown":"WORLD TRADE REPORT 2024\n\n\\- 32 -\n\n### **1\\. Globalization has led to income convergence, but some economies have been left behind**\n\n**The world has become more equal over the past three decades.** Until the 1990s, there was little evidence of poorer economies catching up with richer ones, a phenomenon known as unconditional convergence. However, since the mid-1990s, economic growth has been unambiguously characterized by unconditional convergence (Patel, Sandefur and Subramanian, 2021; Kremer, Willis and You, 2022). This has been due mainly to faster catch-up growth in poorer economies, although the simultaneous slowing of growth in richer economies has also contributed. Consequently, income inequality among economies has been decreasing for the last quarter-century. Only the shock of the COVID-19 pandemic ended this era of unconditional income convergence.\n\n**Richer economies have a lower poverty rate,** **higher levels of education and longer life expectancies.** Although income per capita is an imperfect measure of development, it is correlated with several dimensions of human development. Thus, the era of global income convergence has been accompanied by a stark reduction in the global poverty rate ¹ and has contributed to United Nations Sustainable Development\n\nGoals, such as those associated with education and health. According to data from the World Bank, the global share of people living in extreme poverty declined from 32.7 per cent in 1995 to 9 per cent in 2022, even though the global population rose by 38.7 per cent during the same period.\n\n**The period of unprecedented income convergence has been accompanied by a steep increase in trade participation** (see Figure B.1). While part of this correlation may be related to the fact that economic growth leads to more trade and that other common factors can cause an increase in both income per capita and trade, a compelling set of empirical research suggests that more trade also leads to higher incomes (see Figure B.1) (Feyrer, 2019, 2021; Cerdeiro and Komaromi, 2021). ² Comprehensive trade liberalization in developing economies has been shown to increase income growth by an average of 1.0 to 1.5 percentage points, resulting in incomes higher by 10 to 20 per cent after a decade (Irwin, 2024). Trade has been an essential component of high-growth episodes that typically involved a shift of capital and labour to the tradeable manufacturing sector and intensified imports of capital goods (World Bank, 2024a).\n\n**Low- and middle-income economies have become dominant in global merchandise trade.** The increased trade participation of low- and middle-income economies has transformed the global trade\n\nFigure B.1: Positive correlation between low- and middle-income economies’ convergence speed and trade participation, 1996-2021\n\n\\-1\n\n0\n\n1\n\n2\n\n3\n\n4\n\n5\n\n6\n\n7\n\n8\n\n40\n\n60\n\n80\n\n100\n\n120\n\n140\n\n160\n\n1996\n\n1998\n\n2000\n\n2002\n\n2004\n\n2006\n\n2008\n\n2010\n\n2012\n\n2014\n\n2016\n\n2018\n\n2020\n\n1997\n\n1999\n\n2001\n\n2003\n\n2005\n\n2007\n\n2009\n\n2011\n\n2015\n\n2013\n\n2017\n\n2019\n\n2021\n\nSpeed of income convergence\n\nSpeed of income convergence (percentage points)\n\nTrade participation (100=1996)\n\nTrade participation\n\nSource: Authors’ calculations, based on World Bank data on nominal GDP and real GDP per capita and WTO data on trade in goods and commercial services. Note: The figure displays the evolution over time of the population-weighted averages of trade participation and income convergence speed between 1996 and 2021. The trade participation index is the share of goods and commercial services trade in GDP, adjusted for country size. Speed of income convergence is expressed as the difference between the average real GDP per capita growth rate of low- and middle-income economies and the average growth rate in high-income economies. The light blue fill indicates a contribution of negative growth in high-income economies. The income groups are based on the 1995 World Bank classification.","pages":[{"line_numbers":[],"markdown":"WORLD TRADE REPORT 2024\n\n\\- 32 -\n\n### **1\\. Globalization has led to income convergence, but some economies have been left behind**\n\n**The world has become more equal over the past three decades.** Until the 1990s, there was little evidence of poorer economies catching up with richer ones, a phenomenon known as unconditional convergence. However, since the mid-1990s, economic growth has been unambiguously characterized by unconditional convergence (Patel, Sandefur and Subramanian, 2021; Kremer, Willis and You, 2022). This has been due mainly to faster catch-up growth in poorer economies, although the simultaneous slowing of growth in richer economies has also contributed. Consequently, income inequality among economies has been decreasing for the last quarter-century. Only the shock of the COVID-19 pandemic ended this era of unconditional income convergence.\n\n**Richer economies have a lower poverty rate,** **higher levels of education and longer life expectancies.** Although income per capita is an imperfect measure of development, it is correlated with several dimensions of human development. Thus, the era of global income convergence has been accompanied by a stark reduction in the global poverty rate ¹ and has contributed to United Nations Sustainable Development\n\nGoals, such as those associated with education and health. According to data from the World Bank, the global share of people living in extreme poverty declined from 32.7 per cent in 1995 to 9 per cent in 2022, even though the global population rose by 38.7 per cent during the same period.\n\n**The period of unprecedented income convergence has been accompanied by a steep increase in trade participation** (see Figure B.1). While part of this correlation may be related to the fact that economic growth leads to more trade and that other common factors can cause an increase in both income per capita and trade, a compelling set of empirical research suggests that more trade also leads to higher incomes (see Figure B.1) (Feyrer, 2019, 2021; Cerdeiro and Komaromi, 2021). ² Comprehensive trade liberalization in developing economies has been shown to increase income growth by an average of 1.0 to 1.5 percentage points, resulting in incomes higher by 10 to 20 per cent after a decade (Irwin, 2024). Trade has been an essential component of high-growth episodes that typically involved a shift of capital and labour to the tradeable manufacturing sector and intensified imports of capital goods (World Bank, 2024a).\n\n**Low- and middle-income economies have become dominant in global merchandise trade.** The increased trade participation of low- and middle-income economies has transformed the global trade\n\nFigure B.1: Positive correlation between low- and middle-income economies’ convergence speed and trade participation, 1996-2021\n\n\\-1\n\n0\n\n1\n\n2\n\n3\n\n4\n\n5\n\n6\n\n7\n\n8\n\n40\n\n60\n\n80\n\n100\n\n120\n\n140\n\n160\n\n1996\n\n1998\n\n2000\n\n2002\n\n2004\n\n2006\n\n2008\n\n2010\n\n2012\n\n2014\n\n2016\n\n2018\n\n2020\n\n1997\n\n1999\n\n2001\n\n2003\n\n2005\n\n2007\n\n2009\n\n2011\n\n2015\n\n2013\n\n2017\n\n2019\n\n2021\n\nSpeed of income convergence\n\nSpeed of income convergence (percentage points)\n\nTrade participation (100=1996)\n\nTrade participation\n\nSource: Authors’ calculations, based on World Bank data on nominal GDP and real GDP per capita and WTO data on trade in goods and commercial services. Note: The figure displays the evolution over time of the population-weighted averages of trade participation and income convergence speed between 1996 and 2021. The trade participation index is the share of goods and commercial services trade in GDP, adjusted for country size. Speed of income convergence is expressed as the difference between the average real GDP per capita growth rate of low- and middle-income economies and the average growth rate in high-income economies. The light blue fill indicates a contribution of negative growth in high-income economies. The income groups are based on the 1995 World Bank classification.","page_index":0}],"pipeline_name":"parallel_extract","task_type":"parse"},"pipeline_name":"parallel_extract","product_type":"parse","raw_output":{"markdown":"WORLD TRADE REPORT 2024\n\n\\- 32 -\n\n### **1\\. Globalization has led to income convergence, but some economies have been left behind**\n\n**The world has become more equal over the past three decades.** Until the 1990s, there was little evidence of poorer economies catching up with richer ones, a phenomenon known as unconditional convergence. However, since the mid-1990s, economic growth has been unambiguously characterized by unconditional convergence (Patel, Sandefur and Subramanian, 2021; Kremer, Willis and You, 2022). This has been due mainly to faster catch-up growth in poorer economies, although the simultaneous slowing of growth in richer economies has also contributed. Consequently, income inequality among economies has been decreasing for the last quarter-century. Only the shock of the COVID-19 pandemic ended this era of unconditional income convergence.\n\n**Richer economies have a lower poverty rate,** **higher levels of education and longer life expectancies.** Although income per capita is an imperfect measure of development, it is correlated with several dimensions of human development. Thus, the era of global income convergence has been accompanied by a stark reduction in the global poverty rate ¹ and has contributed to United Nations Sustainable Development\n\nGoals, such as those associated with education and health. According to data from the World Bank, the global share of people living in extreme poverty declined from 32.7 per cent in 1995 to 9 per cent in 2022, even though the global population rose by 38.7 per cent during the same period.\n\n**The period of unprecedented income convergence has been accompanied by a steep increase in trade participation** (see Figure B.1). While part of this correlation may be related to the fact that economic growth leads to more trade and that other common factors can cause an increase in both income per capita and trade, a compelling set of empirical research suggests that more trade also leads to higher incomes (see Figure B.1) (Feyrer, 2019, 2021; Cerdeiro and Komaromi, 2021). ² Comprehensive trade liberalization in developing economies has been shown to increase income growth by an average of 1.0 to 1.5 percentage points, resulting in incomes higher by 10 to 20 per cent after a decade (Irwin, 2024). Trade has been an essential component of high-growth episodes that typically involved a shift of capital and labour to the tradeable manufacturing sector and intensified imports of capital goods (World Bank, 2024a).\n\n**Low- and middle-income economies have become dominant in global merchandise trade.** The increased trade participation of low- and middle-income economies has transformed the global trade\n\nFigure B.1: Positive correlation between low- and middle-income economies’ convergence speed and trade participation, 1996-2021\n\n\\-1\n\n0\n\n1\n\n2\n\n3\n\n4\n\n5\n\n6\n\n7\n\n8\n\n40\n\n60\n\n80\n\n100\n\n120\n\n140\n\n160\n\n1996\n\n1998\n\n2000\n\n2002\n\n2004\n\n2006\n\n2008\n\n2010\n\n2012\n\n2014\n\n2016\n\n2018\n\n2020\n\n1997\n\n1999\n\n2001\n\n2003\n\n2005\n\n2007\n\n2009\n\n2011\n\n2015\n\n2013\n\n2017\n\n2019\n\n2021\n\nSpeed of income convergence\n\nSpeed of income convergence (percentage points)\n\nTrade participation (100=1996)\n\nTrade participation\n\nSource: Authors’ calculations, based on World Bank data on nominal GDP and real GDP per capita and WTO data on trade in goods and commercial services. Note: The figure displays the evolution over time of the population-weighted averages of trade participation and income convergence speed between 1996 and 2021. The trade participation index is the share of goods and commercial services trade in GDP, adjusted for country size. Speed of income convergence is expressed as the difference between the average real GDP per capita growth rate of low- and middle-income economies and the average growth rate in high-income economies. The light blue fill indicates a contribution of negative growth in high-income economies. The income groups are based on the 1995 World Bank classification.","num_pages":1,"request_body":{"advanced_settings":{"fetch_policy":{"max_age_seconds":600,"timeout_seconds":60},"full_content":{"max_chars_per_result":200000}},"urls":["https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/wtr24_ch2_e_p3.pdf"]},"response":{"errors":[],"extract_id":"extract_ce137655857bc0d0e7c47df79d58beaf","metadata":null,"results":[{"acquired_at":"2026-11-05T00:00:00Z","excerpts":["WORLD TRADE REPORT 2024\n\n\\- 32 -\n\n### **1\\. Globalization has led to income convergence, but some economies have been left behind**\n\n**The world has become more equal over the past three decades.** Until the 1990s, there was little evidence of poorer economies catching up with richer ones, a phenomenon known as unconditional convergence. However, since the mid-1990s, economic growth has been unambiguously characterized by unconditional convergence (Patel, Sandefur and Subramanian, 2021; Kremer, Willis and You, 2022). This has been due mainly to faster catch-up growth in poorer economies, although the simultaneous slowing of growth in richer economies has also contributed. Consequently, income inequality among economies has been decreasing for the last quarter-century. Only the shock of the COVID-19 pandemic ended this era of unconditional income convergence.\n\n**Richer economies have a lower poverty rate,** **higher levels of education and longer life expectancies.** Although income per capita is an imperfect measure of development, it is correlated with several dimensions of human development. Thus, the era of global income convergence has been accompanied by a stark reduction in the global poverty rate ¹ and has contributed to United Nations Sustainable Development\n\nGoals, such as those associated with education and health. According to data from the World Bank, the global share of people living in extreme poverty declined from 32.7 per cent in 1995 to 9 per cent in 2022, even though the global population rose by 38.7 per cent during the same period.\n\n**The period of unprecedented income convergence has been accompanied by a steep increase in trade participation** (see Figure B.1). While part of this correlation may be related to the fact that economic growth leads to more trade and that other common factors can cause an increase in both income per capita and trade, a compelling set of empirical research suggests that more trade also leads to higher incomes (see Figure B.1) (Feyrer, 2019, 2021; Cerdeiro and Komaromi, 2021). ² Comprehensive trade liberalization in developing economies has been shown to increase income growth by an average of 1.0 to 1.5 percentage points, resulting in incomes higher by 10 to 20 per cent after a decade (Irwin, 2024). Trade has been an essential component of high-growth episodes that typically involved a shift of capital and labour to the tradeable manufacturing sector and intensified imports of capital goods (World Bank, 2024a).\n\n**Low- and middle-income economies have become dominant in global merchandise trade.** The increased trade participation of low- and middle-income economies has transformed the global trade\n\nFigure B.1: Positive correlation between low- and middle-income economies’ convergence speed and trade participation, 1996-2021\n\n\\-1\n\n0\n\n1\n\n2\n\n3\n\n4\n\n5\n\n6\n\n7\n\n8\n\n40\n\n60\n\n80\n\n100\n\n120\n\n140\n\n160\n\n1996\n\n1998\n\n2000\n\n2002\n\n2004\n\n2006\n\n2008\n\n2010\n\n2012\n\n2014\n\n2016\n\n2018\n\n2020\n\n1997\n\n1999\n\n2001\n\n2003\n\n2005\n\n2007\n\n2009\n\n2011\n\n2015\n\n2013\n\n2017\n\n2019\n\n2021\n\nSpeed of income convergence\n\nSpeed of income convergence (percentage points)\n\nTrade participation (100=1996)\n\nTrade participation\n\nSource: Authors’ calculations, based on World Bank data on nominal GDP and real GDP per capita and WTO data on trade in goods and commercial services. Note: The figure displays the evolution over time of the population-weighted averages of trade participation and income convergence speed between 1996 and 2021. The trade participation index is the share of goods and commercial services trade in GDP, adjusted for country size. Speed of income convergence is expressed as the difference between the average real GDP per capita growth rate of low- and middle-income economies and the average growth rate in high-income economies. The light blue fill indicates a contribution of negative growth in high-income economies. The income groups are based on the 1995 World Bank classification."],"full_content":"WORLD TRADE REPORT 2024\n\n\\- 32 -\n\n### **1\\. Globalization has led to income convergence, but some economies have been left behind**\n\n**The world has become more equal over the past three decades.** Until the 1990s, there was little evidence of poorer economies catching up with richer ones, a phenomenon known as unconditional convergence. However, since the mid-1990s, economic growth has been unambiguously characterized by unconditional convergence (Patel, Sandefur and Subramanian, 2021; Kremer, Willis and You, 2022). This has been due mainly to faster catch-up growth in poorer economies, although the simultaneous slowing of growth in richer economies has also contributed. Consequently, income inequality among economies has been decreasing for the last quarter-century. Only the shock of the COVID-19 pandemic ended this era of unconditional income convergence.\n\n**Richer economies have a lower poverty rate,** **higher levels of education and longer life expectancies.** Although income per capita is an imperfect measure of development, it is correlated with several dimensions of human development. Thus, the era of global income convergence has been accompanied by a stark reduction in the global poverty rate ¹ and has contributed to United Nations Sustainable Development\n\nGoals, such as those associated with education and health. According to data from the World Bank, the global share of people living in extreme poverty declined from 32.7 per cent in 1995 to 9 per cent in 2022, even though the global population rose by 38.7 per cent during the same period.\n\n**The period of unprecedented income convergence has been accompanied by a steep increase in trade participation** (see Figure B.1). While part of this correlation may be related to the fact that economic growth leads to more trade and that other common factors can cause an increase in both income per capita and trade, a compelling set of empirical research suggests that more trade also leads to higher incomes (see Figure B.1) (Feyrer, 2019, 2021; Cerdeiro and Komaromi, 2021). ² Comprehensive trade liberalization in developing economies has been shown to increase income growth by an average of 1.0 to 1.5 percentage points, resulting in incomes higher by 10 to 20 per cent after a decade (Irwin, 2024). Trade has been an essential component of high-growth episodes that typically involved a shift of capital and labour to the tradeable manufacturing sector and intensified imports of capital goods (World Bank, 2024a).\n\n**Low- and middle-income economies have become dominant in global merchandise trade.** The increased trade participation of low- and middle-income economies has transformed the global trade\n\nFigure B.1: Positive correlation between low- and middle-income economies’ convergence speed and trade participation, 1996-2021\n\n\\-1\n\n0\n\n1\n\n2\n\n3\n\n4\n\n5\n\n6\n\n7\n\n8\n\n40\n\n60\n\n80\n\n100\n\n120\n\n140\n\n160\n\n1996\n\n1998\n\n2000\n\n2002\n\n2004\n\n2006\n\n2008\n\n2010\n\n2012\n\n2014\n\n2016\n\n2018\n\n2020\n\n1997\n\n1999\n\n2001\n\n2003\n\n2005\n\n2007\n\n2009\n\n2011\n\n2015\n\n2013\n\n2017\n\n2019\n\n2021\n\nSpeed of income convergence\n\nSpeed of income convergence (percentage points)\n\nTrade participation (100=1996)\n\nTrade participation\n\nSource: Authors’ calculations, based on World Bank data on nominal GDP and real GDP per capita and WTO data on trade in goods and commercial services. Note: The figure displays the evolution over time of the population-weighted averages of trade participation and income convergence speed between 1996 and 2021. The trade participation index is the share of goods and commercial services trade in GDP, adjusted for country size. Speed of income convergence is expressed as the difference between the average real GDP per capita growth rate of low- and middle-income economies and the average growth rate in high-income economies. The light blue fill indicates a contribution of negative growth in high-income economies. The income groups are based on the 1995 World Bank classification.","publish_date":null,"title":"1. Globalization has led to income convergence, but some economies have been left behind","url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/wtr24_ch2_e_p3.pdf"}],"warnings":[{"detail":null,"message":"Neither objective nor search_queries were provided, provide at least one to increase the relevance of excerpts.","type":"warning"}]},"url":"https://huggingface.co/datasets/llamaindex/ParseBench/resolve/main/docs/chart/wtr24_ch2_e_p3.pdf"},"request":{"example_id":"chart/wtr24_ch2_e_p3","product_type":"parse","source_file_path":"wtr24_ch2_e_p3.pdf"},"started_at":"2026-11-04T19:45:15.493679"}